Newcastle University ePrints

Wu X, Scott K, Xie F, Alford N. A reversible water electrolyser with porous

PTFE based OH− conductive membrane as energy storage cells.
Journal of Power Sources 2014, 246, 225-231.

Copyright:

© 2013 Elsevier B.V

DOI link to article:

http://dx.doi.org/10.1016/j.jpowsour.2013.07.081

Date deposited: 14th January 2014

This work is licensed under a Creative Commons Attribution 3.0 Unported License

ePrints – Newcastle University ePrints

http://eprint.ncl.ac.uk
 Journal of Power Sources 246 (2014) 225e231

Contents lists available at ScienceDirect

Journal of Power Sources

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

A reversible water electrolyser with porous PTFE based OH

conductive membrane as energy storage cells
Xu Wu a, b, *, Keith Scott a, *, Fang Xie b, Neil Alford b
a
School of Chemical Engineering and Advanced Materials, Newcastle University, Newcastle upon Tyne NE1 7RU, United Kingdom
b
Department of Materials, Imperial College London, South Kensington Campus, London SW7 2AZ, United Kingdom

h i g h l i g h t s

 A reversible water electrolyser was prepared with porous PTFE based OH
membrane.
 Current density was improved by adopting the PFTE composite OH
membrane.
 The peak power density in fuel cell mode achieved 0.163 W cm
2 at 45  C.
 In the water electrolyser mode, degradation rate was only 0.0379 mV h
1.

a r t i c l e i n f o a b s t r a c t

Article history: An OH
conductive anion exchange membrane was prepared by a pore-filling method with porous
Received 13 December 2012 polytetrafluoroethylene and a quaternary ammonium polymethacrylate ionomer. The composite mem-
Received in revised form brane exhibited less swelling ratios of thickness and area variation and stronger tensile strength than the
11 July 2013
pristine polymethacrylate membrane. The ionic conductivity of the composite membrane was not as
Accepted 18 July 2013
good as the pristine polymethacrylate ionomer membrane. However, the composite membrane was
Available online 2 August 2013
ultra-thin therefore it leaded to a smaller ionic resistance of the MEA and improved current densities. In
fuel cell mode, the peak power densities were respectively 0.114 and 0.163 W cm
2 at 20 and 45  C. In
Keywords:
Anion exchange membrane
water electrolyser mode, at a current density of 100 mA cm
2 the cell voltages were about 1.61 V and
Alkaline membrane 1.52 V respectively at 22  C and 50  C. In the water electrolyser mode, the degradation rate was only
Alkaline fuel cells 0.0379 mV h
1 after 120 h at a current density of 100 mA cm
2 at 22  C.
Water electrolyser Ó 2013 Elsevier B.V. All rights reserved.
Regenerative fuel cells

1. Introduction back to corresponding electrodes, it is often called a unitized

regenerative fuel cell (URFC) [5e10]. Such devices may also be called
Electrochemical energy storage cells are essential devices for reversible water electrolysers (RWE), since they exhibit both function
national grid and renewable power sources such as solar and wind of storing electricity into chemical energy as electrolysers and then
energy, of which the power is usually generated in the form of release chemical energy back to electricity with the reversed elec-
electricity [1]. Water electrolyser is one of the promising candidate trochemical process. The RWE can work in two function modes: the
devices for storing electricity into chemical energy, with H2 as the fuel cell mode and the water electrolyser mode. In the water elec-
energy carrier [2]. The H2 gas produced by electrolysers can be trolyser mode, electric energy will be stored into chemical energy via
collected and consumed as fuels when required. Or in a more effi- electrolysis of water to form H2 and O2. While in the fuel cell mode,
cient way, the energy carried by H2 gas can be released to electricity this device can generate electricity when required, with H2 and O2
by fuel cells [3]. The combination of water electrolysers and fuel cells supplied back to each electrode with producing water. There are
is often called regenerative fuel cells (RFC), which can be considered considerable works on developing the URFC in past three decades
more advanced than conventional secondary batteries [4]. If a water [5e10]. However, most URFC are based on cationic perfluorosulfonic
electrolyser can also work as a fuel cell once H2 and O2 are supplied acid (PFSA) ionomers, in which expensive noble metal catalysts and
corrosion resistant materials are necessary for good stability and
* Corresponding authors. Department of Materials, Imperial College London, cycle life [5e10]. For example in the bifunctional oxygen electrode,
South Kensington Campus, London SW7 2AZ, United Kingdom. Tel.: þ44 191
the catalysts are often based on expensive Pt with IrO2 [8]. It is of
2228771; fax: þ44 191 2225292.
E-mail addresses: [email protected], [email protected] (X. Wu), increasing interests to employ anionic OH
conductive polymer
[email protected] (K. Scott). electrolyte membranes for regenerative fuel cells, in which case

0378-7753/$ e see front matter Ó 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jpowsour.2013.07.081
226 X. Wu et al. / Journal of Power Sources 246 (2014) 225e231

non-precious-metal catalysts can be adopted in the electrodes [11]. composite membrane were compared in terms of conductivity, tensile
The configuration of the RWE in our study is schematically illustrated strength, and the swelling effect. This membrane was applied in a
in Fig. 1. The OH
conductive membrane is the core component of practical single-cell reversible water electrolyser and the current
RWE, as shown in Fig. 1. A significant advantage of using OH
density performances were evaluated in both fuel cell mode and water
conductive polymer electrolyte instead of KOH electrolyte is that only electrolyser mode. The stability of the MEA with this composite
deionised water is required for the water RWE system [11]. The membrane was evaluated by continuous operation for several days.
catalyst layers for hydrogen and oxygen electrodes are respectively
attached to opposite sides of the membrane as close as possible to 2. Experimental methods
reduce electrolytic resistances. Usually, the porous current collectors
are attached to the catalyst layers to form the membrane electrode 2.1. Membrane preparation
assembly (MEA). The current density performances as well as sta-
bility of the RWE system would often mainly depend on the pa- The porous PTFE membrane adopted in this work was obtained
rameters and methods of MEA preparation. As shown in Fig. 1, the from Membrane Solution Ltd (United States of America). The thick-
oxygen side often exhibits highly corrosive environment especially ness of this porous PTFE was 25 mm and had a porosity of 82% and
when anodic oxygen evolution reaction happens. Therefore in our pore size of 0.3e0.6 mm. The qPDTB-OH
ionomer was prepared
design a gold coated titanium mesh was employed as the current according to a published method [12]. The hydrophobic PTFE sheet
collector for the oxygen electrode. It can be seen from our previous was pre-treated for better impregnation of qPDTB, using the method
report [11] that the current densities of RWE are several times smaller developed by Scott and Li [13]. Firstly, the porous PTFE was treated
than that of typical URFC. Therefore to improve the current densities, with seven parts H2SO4 (98 wt% aqueous solution, Aldrich) and three
we have been working on improving the electrochemical activity of parts H2O2 (30 wt% aqueous solution, Aldrich) at 80  C for 1 h.
electrodes and reducing the total resistance of the electrolytic cell. Following this, the porous PTFE was rinsed with copious amounts of
Previously, we developed a quaternary ammonium Poly(DMAEMA- de-ionised (DI) water and further treated by immersion in a solution
co-TFEMA-co-BMA) ionomer (qPDTB-OH
) for the alkaline membrane containing one part aqueous NaOH (1.0 mol dm
3) solution, one part
cells [12]. However, the qPDTB-OH
ionomer was only utilized as the H2O2 (30 wt% aqueous solution), and five parts DI water at 70  C for
binder for catalyst layers in the previous study, the qPDTB-OH
30 min, followed by rinsing with copious amounts of DI water. Then
membrane exhibited remarkable swelling effects when immersed in PTFE sheet was pre-treated in boiling ethanol for 30 min. Then it was
water for a long time. Then in this study, an ultra-thin composite immersed in 5 wt% qPDTB alcohol solution for 20 min with ultra-
membrane of qPDTB-OH
ionomer was prepared by filling the pores of sonication, and then casted and dried at room temperature for
a porous polytetrafluoroethylene (PTFE) with qPDTB-OH
. The prep- 40 min to complete impregnation. The impregnation process was
aration method of composite membranes with polymer electrolyte repeated several times to achieve expected weight. This composite
and porous PTFE was developed earlier for high temperature proton membrane was changed from halide form to OH
form by immersing
exchange membranes fuel cells [13]. Yamaguchi [18] originally pro- in a 0.5 mol dm
3 KOH solution for several hours and then rinsed
posed the concept of pore-filling membranes for polymer electrolyte with deionised water several times until rinsing water pH z 7.
fuel cells. Porous or expanded PTFE were adopted in some pore-filling
composite membranes of PFSA ionomers [19e21]. Composite mem- 2.2. Membrane characterization
branes of porous PTFE and quaternary ammonium polymer were re-
ported recently for alkaline anion exchange membrane fuel cells Morphology of the membrane samples was studied by scanning
[14,26]. In this study, the pristine qPDTB-OH
membrane and the electron microscopy (SEM, JEOL JSM5300LV). Ionic conductivity of

Fig. 1. Schematic figure of reversible water electrolysers with OH
conductive membranes.
 X. Wu et al. / Journal of Power Sources 246 (2014) 225e231 227

the membranes in OH
form was determined by four point AC

impedance method, as described elsewhere [15]. Relative humidity
was controlled with a N2 purging humidifier and was monitored by
a temperatureehumidity sensor (Vaisala HMT360) of which the
probe was installed inside the conductivity measurement fixture.
Ion exchange capacity (IEC) of the membrane was measured by
acidebase back titration [16]. Tensile strength was tested with an
Instron 4505 tensile machine. The water content of the membrane
(l) was defined as average number of water molecules per the
conducting functional site, by the following equation:

ðWh
Wd Þ=Wd
l ¼
IEC  Mw

where IEC is the ion exchange capacity (i.e. OH
ion content per
gram of membrane), Wh and Wd are respectively the weight of the
membrane in dry form and after hydration, Mw is the molecular
weight of water, 18 g mol
1. The swelling ratio of thickness (HT, %)
was defined by the following equation:

dh
HT ¼  100
dd

where dd and dh are respectively the membrane thicknesses in dry

form and after hydration. To measure the swelling effects, the
membrane in OH
form was firstly immersed in room temperature
deionized water for 10 days and then the thickness variations were
gradually measured when drying the swollen membrane from fully
hydrated to totally dry.

2.3. Membrane electrode assembly preparation and measurement

The Cu0.6Mn0.3Co2.1O4 nanometre powder was adopted as

catalyst for the bifunctional oxygen electrode [11]. This
Cu0.6Mn0.3Co2.1O4 catalyst was prepared according to previously
published method [17]. The MEA was prepared by catalyst coated
on membrane method with the qPDTB-OH
ionomer as binder for
catalyst layers, according to published procedures [11]. In this
study, 20 wt% Pt/C (Alfa Aesar) was adopted in the hydrogen elec-
trode with a loading of 0.1 mgPt cm
2. To improve the stability of
Fig. 2. Morphology of (A) porous PTFE membrane and (B) PTFE þ qPDTB composite
the MEA, a cross-linker was added to qPDTB for preparing catalyst
membrane.
inks. MEA was formed by hot-pressing porous current collectors
onto the catalysts coated membrane. The MEA was immersed in
deionised water for at least 24 h before polarisation tests. When the letting qPDTB ionomer to fill the pores. However, with only one
MEA was installed in the testing system as shown in Fig. 1, the treatment the qPDTB ionomer could have filled only parts of the
oxygen electrode was connected to a potentiostat (ACM 946) as volume of the pores inside the porous PTFE membrane. Then the
working electrode and the hydrogen electrode side was adopted as repetitive impregnation processes were adopted in this study. Then
both reference and counter electrode. The polarization curves of the filler qPDTB ionomer gradually filled all the rest volumes of the
the practical MEA were obtained with a scan rate of 5 mV s
1. pores and eventually spilled to the surface of the composite
Stability of electrolyser mode was analyzed by chro- membrane. Considering the porosity of this porous PTFE mem-
nopotentiometry at a constant current density of 100 mA cm
2 for brane is about 0.82, and the density of PTFE and qPDTB are
several days. respectively about 2.2 and 1.2 g cm
3, it can be estimated that when
the weight of this sample membrane (originally 12.5 mg) increases
3. Results and discussion to 43.6 mg, the pores would ideally be fully filled by qPDTB and
little qPDTB is outside the surface of PTFE. As seen in Table 1, after 5
3.1. Membrane preparation impregnation processes, the weight of membrane increased from

It can be seen in the SEM images in Fig. 2 that the as received

PTFE membrane is highly porous. With the qPDTB ionomer
Table 1
impregnated into the pores of the matrix, the membrane became Weight of a PTFE þ qPDTB membrane sample after repetitive impregnation
non-porous. The impregnation process was repeated several times treatments.
to ensure that the pores of the PTFE are fully filled with qPDTB.
Original Treatment Treatment Treatment Treatment Treatment
Table 1 lists the variation of the weight of a sample membrane after weight 1 2 3 4 5
each impregnation treatment. After the treatment with qPDTB
12.5 mg 18.9 mg 25.1 mg 31.4 mg 37.9 mg 44.5 mg
solution, the membrane was dried to evaporate the alcohol solvent
228 X. Wu et al. / Journal of Power Sources 246 (2014) 225e231

12.5 mg to 44.5 mg. In this study, all composite membrane samples

were prepared by 5 times impregnation treatments.

3.2. Membrane properties

3.2.1. Water uptake and swelling effects

The ion exchange capacities (IEC) of pristine qPDTB-OH
mem-
brane and PTFE þ qPDTB-OH
membrane were determined by acide
base back titration method and the values were respectively
1.32  0.04 and 1.02  0.05 meq g
1. Apparently, the PTFE part of the
composite membrane does not contain any quaternary ammonium
function groups, therefore when the weight of PTFE was included the
IEC of the composite was smaller than the pristine ionomer [18].
Fig. 3 compares the influences of water content (l) on the swelling
ratios. As shown in Fig. 3A and B, after immersing the membranes in
deionised water for about seven days there was hardly any further
increase in the water contents. The effect of water swelling on this
polymethacrylate based quaternary ammonium ionomer is exces- Fig. 4. Conductivity of qPDTB-OH
and PTFE þ qPDTB-OH
membranes.
sive, compared with PFSA ionomer membranes [18]. For example in
Fig. 3C when l was about 100, the HT of qPDTB-OH
was about 185%. differences in IEC and water content. However as the porosity is
The swelling effects may not only lead to larger ionic resistance of the 82% then by a Bruggeman relationship the PTFE based membrane
membrane, but may also cause some problems of the MEA such as could have a factor of 0.821.5 ¼ 0.74 lower conductivity than the
stability [19]. However, it can be noticed that the swelling ratios of pure ionomer membrane. This value is in broad agreement with the
the PTFE reinforced composite membrane are much smaller than the ratios of conductivities of the pristine and PTFE membrane. As
pristine ionomer. The maximum thickness swelling ratio HT of shown in Fig. 4, the conductivity of PTFE þ qPDTB-OH
membrane
PTFE þ qPDTB-OH
was about 146%. with relative humidity around 100% increased from about
0.018 S cm
1 to 0.034 S cm
1 when temperature increased from
3.2.2. Conductivity 20  C to 50  C. Besides, water transport properties are interesting
The ionic conductivity of porous PTFE based composite mem- factors in the alkaline anion exchange membranes [22] and the
branes was usually not as good as the filler ionomer [18e21]. In this incorporation of PFTE may have some impacts on that point.
study, the decrease of conductivity was also observed when the However, there are rare data on the water transport properties of
qPDTB ionomer was made into composite membranes with porous porous PTFE based OH
membranes in literature. According to a
PTFE. This difference in conductivity may possibly be due to the simple model of water electrolysers [23], area specific resistance of

Fig. 3. Water content vs. immersing time and swelling ratio vs. water content of qPDTB-OH
and PTFE þ qPDTB-OH
membranes.
 X. Wu et al. / Journal of Power Sources 246 (2014) 225e231 229

the MEA is an important factor of improving current density per-

formances. The area specific resistances of the pristine qPDTB-OH

membrane (about 200 mm) and the PTFE þ qPDTB-OH
composite
membrane (about 30 mm) were respectively c.a. 0.784 U cm2 and
0.167 U cm2 at 20  C. The resistance of PTFE þ qPDTB-OH
com-
posite membrane was much smaller, which is because the com-
posite membrane is ultrathin although its conductivity was not as
good as the pristine ionomer.

3.2.3. Tensile strength

The stress strain curve of PTFE þ qPDTB-OH
is shown in Fig. 5,
from which the Young’s modulus and tensile strength could be
determined. The Young’s modulus of the PTFE þ qPDTB-OH
mem-
brane was about 0.406 GPa, which is better than the qPDTB-OH

membrane [12]. The tensile strength was also improved by incor-
porating the porous PTFE in the membrane, which was greater than
10 MPa. Good mechanic properties are necessary for these OH

conductive membranes because the membranes will be applied in
the RWE with compression to ensure good sealing of water and gas.
The enhanced Young’s modulus and tensile strength may indicate Fig. 6. Polarisation curves in fuel cell mode and water electrolyser mode.

good stability and reliability of the PTFE þ qPDTB-OH
membrane.

(from 0 to
100 mA cm
2) when the temperature was increased
from 20  C to 45  C the activation overpotential fell. The apparent
3.3. Current density performances electrolytic resistance of the MEA can be estimated from the ohmic
loss region of the polarization curves, of which the values are
The current density performances in both fuel cell mode and respectively about 1.5 U cm2 at 20  C and 1.2 U cm2 at 45  C.
water electrolyser mode are shown in Fig. 6. The open circuit Increasing temperature both improved the kinetics of electro-
voltage in fuel cell mode was about 1050 mV, which corresponds to chemical reactions and the ionic conduction. The peak power
the good ORR catalytic activity of the Cu0.6Mn0.3Co2.1O4 electrode densities at 20  C and 45  C were respectively 114 and
[11] and also indicates little gas crossover of the MEA. Crossover of 163 mW cm
2. The peak power density obtained in this study is
H2 gas to the oxygen electrode is often a problem in H2 fuel cells, comparable with results obtained with commercial Pt/C electrode
which may lead lowering of the OCV by several hundred mV and [14]. However, such power density performances are several times
may reduce the stability of the MEA. The composite membrane in smaller than that of PEMFC, of which the typical peak power den-
this study is based on a highly porous PTFE sheet therefore it is sity is 0.5e0.8 W cm
2 [13].
important to carefully fill all the pores with qPDTB to ensure good On the other hand, the overpotential in water electrolyser mode
restriction of gas crossover. was less than that in fuel cell mode. For example at 50  C the
The activation loss (exponential range of polarization curves) polarisation curve became IR control after the voltage increased
was considerable in the fuel cell mode, as shown from the rapid above about 1500 mV. It could also be observed that when tem-
decrease of current density from 1000 mV to 650 mV. The fuel cell perature increased from 22  C to 50  C the activation loss (as shown
mode polarisation curve turned to IR control from voltage was in 0e150 mA cm
2) decreased. Meanwhile, the apparent MEA
below about 650 mV. Although the content and composition of resistance decreased from about 0.896  0.061 U cm2 at 22  C to
oxygen electrode was already optimized regarding the water 0.737  0.010 U cm2 at 50  C. The current density performance in
electrolyser mode, apparently it is necessary to further improve the water electrolyser mode is better than previous results with radi-
electrode with respect to oxygen reduction. It can be seen that ation grafted membranes [12], which can be attributed to the
smaller resistance of the membrane as discussed in Section 3.2.2.
However, the current densities of water electrolysers with OH

membranes are still not as good as that of PEMWE, of which the
current density was usually around 1 A cm
2 with a voltage around
1.6e1.7 V [15,24].
The gap between the currents of the PEM system and OH

membrane system in both fuel cell mode and water electrolyser
mode may be narrowed by increasing the operation temperature or
by making larger electrode areas of RWEs with OH
membranes.
Increasing operational temperature may enhance the electrode
reactions and improve ionic conductivity. However it might also
result in faster degradation of quaternary ammonium functional
groups [25]. Increasing electrode area is practical because the ma-
terial costs of RWE with OH
membranes are much cheaper than
regenerative fuel cells based on PFSA membranes, especially with
cheap non-precious-metal catalysts [12]. Besides, as energy storage
cells the RWE will in most cases be stationary, so there is less
concern about the volume of the system. Another important
parameter for evaluating RWE is the efficiency of energy storage.
The fuel cell to electrolyser voltage ratio at a current density of
Fig. 5. Stress strain curve of PTFE þ qPDTB-OH
membrane. 100 mA cm
2 achieved c.a. 34% at 22  C. This value is better than
230 X. Wu et al. / Journal of Power Sources 246 (2014) 225e231

31.9% in our previous publication [11], which is because of

enhanced current density performances in both fuel cell mode and
water electrolyser mode with the PTFE þ qPDTB-OH
membrane.

3.4. Water electrolyser mode stability

Stability of fuel cells and water electrolysers is of vital signifi-

cance in terms of practical applications for years. For the reversible
water electrolysers, the stability in electrolyser mode may be more
critical than in fuel cell mode, because that the high voltage at
anode may facilitate degradation of components, such as corrosion
of current collectors, oxidation of ionomers, and degradation of
quaternary ammonium function groups. Therefore, a relatively low
voltage (e.g. below 2 V) is preferable for water electrolysers
although increasing voltage may increase the current density, i.e. H2
production rate. Fig. 7 shows a typical short term stability test by
chronopotentiometry at a constant current density of 100 mA cm
2
Fig. 8. Polarisation curves before and after the stability test of Fig. 7.
for several days. The voltage in Fig. 7 increased slightly in the first
60 h and then became stable around 1.75 V. Some small peaks and
valleys could be observed in the first 120 h of the curve in Fig. 6, 4. Conclusions
which might be due to the influences of randomly generated gas
bubbles. Degradation rate after 120 h was only 0.0379 mV h
1, A reversible water electrolyser was prepared with an advanced
indicating that the MEA would be stable in continuous operation for OH
conductive membrane based on porous PTFE. The PTFE þ
several thousands of hours. Polarisation curves before and after qPDTB-OH
composite membrane exhibited better swelling effects
stability test of Fig. 7 were determined and demonstrated in Fig. 8. It and mechanical strength than the pristine qPDTB-OH
membrane.
can be obtained from Fig. 8 that the apparent MEA resistances The ionic conductivity of PTFE þ qPDTB-OH
membrane was not as
before and after stability test were respectively 0.896  0.061 U cm2 good as the qPDTB-OH
membrane. However, due to this
and 0.902  0.048 U cm2, which means that there were hardly any PTFE þ qPDTB-OH
membrane could be ultra-thin (30 mm), the
degradations of the ionomers during the stability test. Besides, the resistance of the membrane was reduced and current densities were
two plots in Fig. 8 exhibited almost identical IR corrected voltages of improved compared with previous results. In the fuel cell mode, the
water electrolysis before and after stability test. This infers that the peak power density achieved 163 mW cm
2 at 45  C. In water
current density difference between the two curves in Fig. 8 was electrolyser mode at a current density of 100 mA cm
2 the voltage
probably mainly due to changes of the catalyst surface. In previous was only 1.61 V at 22  C. In future works, the water transport
studies [17], some considerable current could be observed on the properties is of interests with the influences of PTFE in the composite.
cobalt oxide electrode due to the following reaction: The current density in both fuel cell mode and water electrolyser
mode shall be further improved by further reducing ionic resistance
Co3 O4 þ H2 O þ OH
43CoOOH þ e
and improving electrode parameters. This reversible water electro-
After the stability test the current attributes to this surface re- lyser operated with only deionised water, which is unique compared
action would have declined, which corresponds to the current with KOH based fuel cell and water electrolyser systems. Besides, a
density differences between the two curves in Fig. 8 and could also non-precious-metal oxygen electrode was adopted in this study,
explain the voltage variation in the first 60 h of stability test. which is much cheaper than the PteIr based oxygen electrodes of
proton exchange membrane regenerative fuel cells. The short term
chronopotentiometry test indicated good stability in the water
electrolyser mode. From the polarisation curves before and after the
stability test it can be seen that the MEA resistance and onset voltage
did not change during the stability test. After 120 h, the degradation
rate was only 0.0379 mV h
1 at 22  C. Overall this study indicates
promising application of the reversible water electrolysers in
renewable power source systems such as the solar hydrogen system.

Acknowledgements

This study was supported by fundings from Engineering and

Physical Sciences Research Council (EPRSC) of United Kingdom,
with grants number EP/G042012/1, EP/H007962/1, EP/H019596/1,
and EP/I022570/1.

References

[1] C.A. Vincent, B. Scrosati, Modern Batteries, John Wiley & Sons, London, Great
Britain, 1984.
[2] F. Barbir, Solar Energy 78 (2005) 661e669.
Fig. 7. Short term stability test at 22  C and with a constant current density of [3] M. Korpås, C.J. Greiner, Renewable Energy 33 (2008) 1199e1208.
100 mA cm
2. [4] D. Shapiro, J. Duffy, M. Kimble, M. Pien, Solar Energy 79 (2005) 544e550.
 X. Wu et al. / Journal of Power Sources 246 (2014) 225e231 231

[5] F. Mitlitsky, B. Myers, A.H. Weisberg, Energy and Fuels 12 (1998) 56e71. [18] T. Yamaguchi, F. Miyata, S.-i. Nakao, Journal of Membrane Science 214 (2)
[6] W. Smith, Journal of Power Sources 86 (2000) 74e83. (2003) 283e292.
[7] J. Pettersson, B. Ramsey, D. Harrison, Journal of Power Sources 157 (2006) 28e34. [19] F. Liu, B. Yi, D. Xing, J. Yu, H. Zhang, Journal of Membrane Science 212 (2003)
[8] T. Ioroi, N. Kitazawa, K. Yasuda, Y. Yamamoto, H. Takenaka, Journal of Applied 213e223.
Electrochemistry 31 (2001) 1179e1183. [20] H.L. Lin, T.L. Yu, L.N. Huang, L.C. Chen, K.S. Shen, G.B. Jung, Journal of Power
[9] S.S. Dihrab, K. Sopian, A. Zaharim, WSEAS Transactions on Environment and Sources 150 (2005) 11e19.
Development 4 (2008) 1151e1160. [21] L. Wang, B.L. Yi, H.M. Zhang, Y.H. Liu, D.M. Xing, Z.G. Shao, Y.H. Cai, Journal of
[10] S. Song, H. Zhang, X. Ma, Y. Zhang, B. Yi, Progress in Chemistry 18 (2006) Power Sources 167 (2007) 47e52.
1375e1380. [22] Y.S. Li, T.S. Zhao, W.W. Yang, International Journal of Hydrogen Energy 35
[11] X. Wu, K. Scott, Journal of Power Sources 206 (2012) 14e19. (2010) 5656e5665.
[12] X. Wu, K. Scott, Journal of Power Sources 214 (2012) 124e129. [23] W. Xu, K. Scott, International Journal of Hydrogen Energy 35 (2010) 12029e
[13] M. Li, K. Scott, Electrochimica Acta 55 (2010) 2123e2128. 12037.
[14] Y.C. Cao, K. Scott, X. Wang, International Journal of Hydrogen Energy 37 [24] X. Wu, K. Scott, V. Puthiyapura, International Journal of Hydrogen Energy 37
(2012) 12688e12693. (2012) 13243e13248.
[15] W. Xu, K. Scott, S. Basu, Journal of Power Sources 196 (2011) 8918e8924. [25] G. Merle, M. Wessling, K. Nijmeijer, Journal of Membrane Science 377 (2011)
[16] H. Cheng, K. Scott, K.V. Lovell, J.A. Horsfall, S.C. Waring, Journal of Membrane 1e35.
Science 288 (2007) 168e174. [26] Y. Zhao, H. Yu, D. Xing, W. Lu, Z. Shao, B. Yi, Journal of Membrane Science
[17] X. Wu, K. Scott, Journal of Material Chemistry 21 (2011) 12344e12351. 421e422 (2012) 311e317.