JOURNAL OF THE CHINESE

Article CHEMICAL SOCIETY

Failure Mode Analysis of Membrane Electrode Assembly (MEA) for PEMFC under Low
Humidity Operation

Cheng-hong Wang, Chien-ming Lai, Wei-han Huang, Jiunn-nan Lin and Li-duan Tsai*
Department of Fuel Cell Materials, Division of Energy Storage Materials & Technology, Material and Chemical
Research Laboratory (MCL), Industrial Technology Research Institute (ITRI), 195, Sec. 4, Chung Hsing Rd., Chutung,
Hsinchu, Taiwan 31040, R.O.C.
(Received: May 31, 2012; Accepted: Jun. 22, 2012; Published Online: ??; DOI: 10.1002/jccs.201200319)

The degradation mechanisms and phenomena of membrane electrode assembly (MEA) for proton ex-
change membrane fuel cell (PEMFC under low humidity operation were investigated by means of several
on-line electrochemical methods and transmission electron microscopy (TEM). Electrochemical active
surface area (ECSA) of catalyst and hydrogen crossover of MEA were evaluated by cyclic voltagram (CV)
and linear sweep voltagram (LSV), respectively. The decrease of catalytic active site and degradation of
membrane were evident by ECSA loss and hydrogen crossover rate increase after long-term discharging
test. Membrane degradation played the major role for MEAs failure under low humidity operation in this
study. The Pt clusters were also observed in the membrane, formed by Pt ions migration and reduction in
the membrane, which can react with the crossover gases to produce free radicals that attack membrane and
ionomers in the catalyst layer.Those phenomena will drive membrane to rupture and cause a short circuit
in MEA. By AST experiments and fluoride release rate results, the MEA with Pt-alloy/C catalyst can show
better performance and durability due to its high selectivity on ORR to depress the free radicals formation.

Keywords: PEMFC; MEA; Catalyst; Membrane; Failure mode; Durability.

INTRODUCTION material. Fig. 12 shows a schematic structure of the fuel cell

Fuel cell is a power generation device that has the ad- membrane electrode assembly (MEA), including both an-
vantage of high energy conversion, quiet and low pollut- ode and cathode sides. Each side includes a catalyst layer
ants emission. Proton exchange membrane fuel cells and a gas diffusion layer. Between the two sides is a proton
(PEMFCs) are considered one of the most promising fuel exchange membrane (PEM) conducting protons from the
cells for stationary, portable devices and power supply sys- anode to the cathode. The catalyst layer is located between
tem applications. However, several challenges remain and the PEM and the gas diffusion layer (GDL). Protons trans-
the cost and durability of fuel cell have been identified as fer between the catalyst layer and the PEM, and electrons
the two main challenges to fuel cell commercialization, transfer between the catalyst layer and the GDL. Both re-
particularly for stationary and automotive application. US quire good interfacial contact. In a PEM fuel cell, the CLs
Department of Energy’s (DOE) targets for commercializa- are where the electrochemical reactions occur for electric
tion at 2015 are lifetime requirements, 5000 hrs for vehi- power generation. For example, for H2/air (O2) PEM fuel
cles and 40000 hrs for stationary, and the fuel cell cost
should be less than $50/kW.1 To the purpose, intensive re-
search is needed to address and developing the durable
MEA operating under low humidity or without humidifiers
attracts much attention. This kind of MEA can reduce the
BOP (balance of plant) requirement and simplify the con-
trol strategy of the fuel cell systems, which is effective in
fuel cell cost reduction.
The layer structure of PEMFC is composed of several
components, including catalyst layer, polymer electrolyte
membrane, gas diffusion layers, bipolar plates and sealing Fig. 1. Cross section of PEMFC and full cell reaction.

Special Issue for the Electrical Energy Storage and Conversion

* Corresponding author. E-mail: [email protected]; [email protected]

J. Chin. Chem. Soc. 2012, 59, 000-000 © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1
Article Wang et al.

cells, the reactions occurring at the anode and cathode cata- When nano-scale Pt catalyst undergo the electron
lyst layers are as follows: transfer reaction, the surface Pt atom will rearrange and ox-
idize into Pt ions. The dissolved Pt ions favored deposition
Anode: H2 ® 2H+ + 2e- (1) on the surface of nearby Pt particles, leading to the particle
Cathode: O2 + 4H+ + 4e- ® H2O (2) growth, a phenomenon called Ostwald ripening,4 or dis-
Overall reaction: H2 + 1/2 O2 ® H2O (3) solve in ionomer and migrate into the proton conducting
membrane. Thus, the Pt ions will be reduced to metal Pt
The catalyst layer, also named as electrode in CCM crystal and stuck in the middle of the membrane, which
(Catalyst Coated Membrane), is composed of nano-scale Pt formed the Pt bands in the membrane.5,6 Besides, to reduce
particles catalyst, catalyst support and ionomer. The carbon the surface free energy, nano-scale Pt particles tends to be-
black, which has high surface area, can be used as support come agglomerate or oxidized, which also decreases the
for loading highly disperse nano-scale Pt or Pt-alloy parti- electrochemical active surface area.7
cles and for conducting electrons in catalyst layer. The Catalyst support, usually made by conductive carbon,
ionomer acts as electrode binders and also has the function has good electronic and thermal conductivity. The carbon
to help protons transport in electrodes. A fast electrochemi- is thermodynamically stable during steady fuel cell opera-
cal catalytic reaction needs the free access of fuel, electrons tion conditions; however, the carbon corrosion occurs dur-
and protons, and all electron transfer reactions happen at ing the dynamic fuel cell operations, such as start/stop8 and
this three phase contacts, including reaction gases, nano- the corrosion will become accelerated in the presence of Pt
scale Pt particles, and ionomers. Therefore, the fuel cell ef- catalysts,9 leading to the collapse of electrodes and deacti-
ficiency was strongly depended on the three-phase inter- vation of the catalyst.
face areas, named as electrochemical active surface area Membrane degradation
(ECSA). Each of the material in CCM may be degraded by Proton exchange membrane (PEM, also named as
electrochemical or mechanical stresses during operation, polymer electrolyte membrane) is one of key components
which results in the power output decreases or failure of the in MEA. The PEM transports protons and acts as a barrier
fuel cell system. Thus, the types of fuel cell materials deg- between anode and cathode to prevent gas permeation. The
radation and failure modes strongly depend on the opera- most commonly used membrane is composed of PFSA,
tion conditions. To date, various failure mechanisms have such as Nafion®.
been reported, such as catalyst particle coarsening, metal The membrane degradation comes from mechanical,
dissolution, Pt bands formation in membrane, carbon sup- thermal and chemical/electrochemical issues. The mechan-
port corrosion, and the membrane degradation are also pro- ical degradation, including membrane structure defects
posed in literatures.3 such as cracks and pinholes, can damage the fuel cell safety
Review of PEMFC materials degradation and life. Thermal degradation occurs when the membrane
Catalyst layer degradation operated under high temperature and low humidity condi-
The catalyst layer failure occurred as the access of tion that dehydrate the membrane to result in poor protons
fuel, the electrons conduction or protons transport path- transportation. The puncture of MEA due to membrane de-
ways were broken. The degradation rate of electrode mate- fects or CCM hot-press fabrication may accelerate the ma-
rial relies on the robustness of the selected materials, such terials damage for MEA failure in early test stage.10 The
as catalyst, catalyst support and ionomer, and also the elec- chemical degradation originates from chemical attacks by
trode structure integrity. The degradation can be classified peroxide radicals, which are commonly believed to be re-
into mechanical and electrochemical issues. The mechani- sponsible for chemical attack on the catalyst layer and
cal issues comprise the catalyst layer surface cracks or membrane. The peroxide radicals are produced by the
peeling off from the membrane after electrode processing chemical reaction on anode and cathode catalysts due to
or power discharge. The electrochemical issues include the oxygen or hydrogen crossover. Then, the radicals decom-
catalyst particle growth, migration, ionomer degradation pose ionomers and make the membrane backbone and side-
and carbon corrosion. They are considered as the most chain cracking, called membrane thinning, which produce
dominant causes of catalyst layer failure and fuel cell more radical species due to more fuel and oxidant cross-
power loss. over. This can decline membrane mechanical strength and

2 www.jccs.wiley-vch.de © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim J. Chin. Chem. Soc. 2012, 59, 000-000
 JOURNAL OF THE CHINESE
MEAs with Pt/C and Pt-alloy/C Catalyst CHEMICAL SOCIETY

ionic conductivity, leading to the cell impedance increase hydrogen fuel and air was 1.5 and 2.5, respectively. I–V
and cell performance reduction.11 plots were obtained by potentiostatically decreasing the
In the development of PEMFCs, the MEA durability cell voltage stepwise at intervals of 0.05 V. The measure-
is a vital issue for applications and cost reduction, espe- ment of the I–V curve started from 0.85 V and each succes-
cially for stationary power sources. In this work, the degra- sive step was maintained for 60 s. The measurements were
dation of the MEA under low humidity conditions was re- performed until the cell voltage reached 0.4 V.
ported. To develop durable MEAs, it is important to under- The MEA lifetime discharge experiment was con-
stand the egradation mechanism of the MEA. Therefore, ducted under the conditions as the followings: the cell tem-
applying different conditions to simulate the possible oper- perature was at 75 °C. The hydrogen fuel and air oxidant
ation environments for a practical fuel cell was studied in were humidified to 65% relative humidity and were fed to
this work. To investigate the degradation mechanism, mean- each electrode at atmospheric pressure. The current density
while, on-line measurements of hydrogen crossover rate was set at 480 mA/cm2 using the electronic loader (PLZ-
through the MEA and the electrochemical active surface 664WA, KIKUSUI corp., Japan). The stoichiometry of hy-
area of the Pt catalyst with long-term and accelerated stress drogen and air was 1.5 and 2.5, respectively.
test were carried out. The cyclic voltammograms were taken on-line (in test
fuel cell fixtures) on the cathode in order to compare the
EXPERIMENTAL change of electrochemical active surface areas (ECSA) be-
Preparation of MEA and single cell tween fresh MEA and degraded MEA. A two-electrode
Electrodes for the PEMFC were prepared from 46 potentiostatic measurement circuit was used for the on-line
wt.% carbon black supported platinum (Pt/C) or platinum cyclic voltammetry; a fully humidified nitrogen was passed
alloy (Pt-alloy/C) electrocatalyst and a Nafion® solution through the working electrode (fuel cell cathode) compart-
(20 wt.% solution, DE2020, E.I. DuPont de Nemours and ment with a flow rate of 100 sccm and a fully humidified
Company). Catalyst ink was prepared by mixing the cata- hydrogen was fed through the counter electrode (fuel cell
lyst powder with Nafion® solution in a 3D milling mixer. anode) compartment. The profiles were recorded at a scan
The resulting ink was applied on the polytetrafluoroethyl- rate of 20 mV/s using an electrochemical analyzer (Series
ene (PTFE) sheets with a doctor blade coater. The catalyst 885, Scribner Associates Inc., USA).
film was pre-dried by IR and then vacuum dried at 90 °C The linear sweep voltammetry (LSV) was used to de-
overnight to remove the solvent. termine the hydrogen crossover rate. A suitable inert gas
The three-layer CCM (Catalyst Coated Membrane) such as nitrogen is used to purge the fuel cathode while hy-
was formed by decal method, which transferred the elec- drogen is passed through the fuel cell anode. The output of
trodes to proton exchange membrane, Nafion ®, by hot- current vs. voltage can be used to derive the hydrogen
pressing. Assemble the CCM with gas diffusion layers crossover rate from Faraday’s law and diffusion limiting
(10BC, SGL carbon) into a five layer MEA. The MEA was current density is attained, which related to the hydrogen
then subjected to bipolar plates, made from graphite, and crossover rate.
gaskets to form a single cell with serpentine type flow
i
channels. N& H 2 = lim (4)
n×F
Electrochemical characterization
The fuel cell performance was evaluated by measur- where: ilim: limiting current density
ing the current density against cell voltage (I-V) character- n: equivalent mole of electron transfer
istics galvanostatically using the electronic loader (Series F: Faraday’s constant (76485C/eq.)
850E, Scribner Associates Inc., USA). For the evaluation Accelerated stress test (AST)
of the fuel cell performance, the cell was heated to 80 °C. The accelerated stress test (AST) was conducted to
The fuel (pure hydrogen) and oxidant (air) gases were hu- simulate the possible operation environments for a practi-
midified to reach a 65% relative humidification, which was cal fuel cell test and to simplify the degradation of the tar-
calibrated and corrected by an optical hygrometer get MEA components. Many AST protocols have been re-
(VIASENSOR HS-1000, LANDTEC), and fed to each ported and discussed for different MEA components.12-13 In
electrode at atmospheric pressure. The stoichiometry of this study, the AST protocols based on DOE were taken to

J. Chin. Chem. Soc. 2012, 59, 000-000 © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.jccs.wiley-vch.de 3
Article Wang et al.

verify the failure modes of the specific components, such Table 1. Catalyst accelerated stress test protocol
as catalysts and membranes, as well as to minimize con-
founding effects between different components in MEAs.
For the stressors accelerating the ripening effect on electro-
catalysts degradation, the load-cycling test protocol was
applied. The potential load cycling was performed at scan
rate of 50 mV/s from 0.6 V to 0.9 V for 30000 cycles, which
can avoid possible degradation of carbon support oxidation
induced by higher potentials, to verify the durability and
stability of electrocatalyst. For the membranes accelerating
degradation, the OCV holding test procedure was applied
to MEAs to simulate the undesirable environments and to
monitor the potential metrics as an on-line diagnosis. The
OCV holding test was an accelerated degradation stressor
to the mixed potential effects due to fuel and oxidant cross-
over, which resulted in membrane failure because of free
radicals formation. All the test protocols were summarized
Table 2. Membrane accelerated stress test protocol
in Table 1 and Table 2. On-line electrochemical character-
izations were also conducted at every appropriate time in
the whole AST protocols.
Characterization by TEM
After the long-term discharge test, the samples were
examined by TEM (Transmission electron microscope,
JEM-2100F, JEOL Ltd, Japan). The ultramicrotomy tech-
nique was employed to prepare TEM specimens for cross-
sectional observation. A piece of CCM was embedded in an
epoxy resin and sliced by an ultramicrotome (Leica
ULTRACUT UCT) using a diamond knife at room temper-
ature. The sliced specimens with thickness of about 30–80
nm were supported on a conventional Cu mesh (j = 3 mm)
with a carbon microgrid and examined through a TEM sys-
tem. During the examination, the accelerating voltage of
TEM was fixed at 100 kV, and the distinction of the anode, represented as the reversible voltage minus the activation,
cathode, and membrane was verified by means of EDX ohmic, and concentration losses. The concentration loss is
(Energy dispersion spectroscope) analysis. The particle also known as mass transport loss. We assumed the semi-
size of the catalyst was estimated from the TEM images. empirical relationship between cell potential E and current
density i to obey the following equation:
RESULTS AND DISCUSSION æ i ö æ i ö
MEA performance and characterization E = E rev - b log çç ÷ - iR - C exp ç lim ÷
÷ ç i - 1÷ (5)
è i0 ø è lim ø
The polarization curves of two different homemade
MEAs are shown in Fig. 2. The Pt-alloy/C MEA discharge Erev is the reversible potential for the cell; i0 and b are
performances under high temperature and low humidity the exchange current density and Tafel slope for oxygen re-
environment (high temperature 80 °C, the RH £ 65%) can duction reaction, respectively; R represents the resistance
reach 1100 mA/cm2 above 0.65 V. The maximum power that causes a linear variation of cell potential with the cur-
density is 900 mW/cm2 and the overall catalyst loading is rent density; C and ilim are the parameters related to mass
less than 0.5 mg Pt/cm2. In typical polarization curves, the transport limitation, the overpotential of HOR is generally
actual cell voltage E at any given current density can be much smaller than ORR and is frequently neglected. Base

4 www.jccs.wiley-vch.de © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim J. Chin. Chem. Soc. 2012, 59, 000-000
 JOURNAL OF THE CHINESE
MEAs with Pt/C and Pt-alloy/C Catalyst CHEMICAL SOCIETY

on Equation (5), the potential losses, at a current density of 10%. The Pt-alloy/C MEA showed higher initial voltage
480 mA/cm2 and low humidity condition, can be obtained and it can continuously operate over 3700 hours with a
and plotted in Fig. 2(b). It is found that the activation loss voltage drop less than 8%. At the beginning, Pt–alloy/C
of Pt-alloy/C is comparably smaller than Pt/C, while the MEA maintains a fairly stable potential and it still pos-
ohmic and concentration losses are quite close since the sessed a higher potential to the end of the test. During the
components, except for cathodic catalyst, are the same. It final few test hours, however, the potential of Pt-alloy/C
indicates that Pt-alloy/C is much efficient than Pt/C for MEA oscillates and the amplitude of oscillation amplifies
ORR at low humidity operation as many workers claimed.14 with time. Finally, then the potential abruptly plunged be-
MEA durability test results are shown in Fig. 3. The low the safety shut down potential of the fuel cell test sys-
MEA with different cathode catalysts as Pt/C and Pt-al- tem.
loy/C showed different initial performance and long term Both MEAs showed similar failure behavior of po-
durability. The Pt/C MEA showed a lower initial potential tential oscillations at the final stage of durability test. These
in the constant current discharge than that of Pt-alloy/C unstable potential oscillation phenomena can be attributed
MEA. The decay rate of Pt/C MEA is more than 50 mV/hr to the membrane break down or catalyst layer failure, which
and shows a more frequent potential oscillation phenome- is more significant as using the 25 mm thick membrane for
non during the discharge period. The Pt/C MEA was failed MEA, not shown in this study. Under the fuel cell discharge
to discharge at around 1500 hr with an overall power loss of conditions, the free radicals formation and the damages to
the membrane and catalyst layer can be illustrated in elec-
trochemical reactions (6) to (8). The formation of water
from one O 2 needs four electrons and four protons, as
shown in reaction (2). However, hydrogen peroxide forms
as only two protons and two electrons reacts with one oxy-
gen molecule due to insufficient three-phase interface ar-
eas, especially under the low humidity test conditions at
cathode side. This peroxide breaks into free radicals more
easily on Pt catalyst. The free radicals then attack the iono-
mers in the catalyst layer or membranes according to reac-
tion (8).

O2 + 2H+ + 2e- ® H2O2 (6)

H2O2 + 2Pt + 2e- ® 2Pt + 2OH- (7)
R-CF2COOH + OH- ® R-CF2- + CO2 + H2O (8)

Fig. 2. (a)Homemade MEA IV performance results at

65% relative humidity, the cell temperature
was 80 °C, and fuel stoichiometry was 1.5 and
2.5 at anode and cathode side, respectively. (b)
Simulated results of different potential losses Fig. 3. MEA durability test result at 65 %RH, cell tem-
for Pt-alloy/C and Pt/C at 480 mA/cm2 and low perature 75 °C, and fuel stoic. is 1.5 and 2.5 at
humidity operation. anode and cathode side, respectively.

J. Chin. Chem. Soc. 2012, 59, 000-000 © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.jccs.wiley-vch.de 5
Article Wang et al.

Therefore, a possible degradation of the MEAs is that ture, it is reported that Pt-alloy catalyst demonstrates a
the radicals from cathode attack the ionomers and the con- higher coverage of oxygen species than that at pure Pt cata-
ductive film, which caused the decreasing reactivity, exac- lyst, which was ascribed to the enhanced ORR activity of
erbating the hydrogen enetration and membrane thinning,15 Pt-alloy catalyst.17 This suggested that oxygen adsorption
and finally result in the MEA failure. on Pt catalyst is more important than proton adsorption for
On-line electrochemical diagnosis and TEM analysis ORR at cathode side. Thus, Pt-alloy/C MEA can really
The electrochemical active surface area (ECSA) im- show better performance than Pt/C MEA.
plies the amount of active sites which participate in the In Fig. 4(a) and Fig. 4(b), both CV curves of Pt/C
electrochemical reaction at the three-phase interface. There- MEA and Pt-alloy/C MEA after durability test showed a
fore, larger ECSA denotes higher electrochemical activity. current shift as compare to the pristine MEAs, which indi-
The reaction of interest is the electrochemical reduc- cates an increase of the hydrogen penetration rate. Also, the
tion of proton (H+) and subsequent deposition of atomic decrease of double layer capacitance of both degraded cat-
hydrogen on the surface of the platinum catalyst, alysts was observed. The loss of the double layer was
around 28% for degraded Pt/C MEA and 21% for degraded
Pt + H+ + e- ® Pt –Hads (9) Pt-alloy/C MEA. This may imply the loss of areas of car-
bon or Pt catalyst in contact with ionomers in the catalyst
The electrochemical active surface area is calculated layers.
by equation (10) as the following,16 The hydrogen crossover rate was determined by lin-
ear sweep voltammetry (LSV). In the test MEA, anode was
ECSA (cm2 Pt/g Pt) = q/(L × G) (10) fed with hydrogen, as the reference electrode and counter
electrode, and cathode was purged with the inert gas, as the
where q (C/cm2) is the integration of the coulombic charge working electrode. Fig. 5 shows the LSV curves of pristine
from the atomic hydrogen adsorption due to this reaction. G and degraded Pt-alloy/C MEA and Table 3 summarizes the
(210 mC/cm2 Pt) is the quantity of charge density to reduce data. The pristine MEA exhibited a limiting current density
a monolayer of protons on Pt catalyst and L (g Pt/cm2) rep- (ilim) of 1.1 mA/cm2 (black curve). This limiting current
resents the Pt catalyst loading. density is proportional to the hydrogen penetration rate
The ECSAs of Pt/C and Pt-alloy/C catalysts, tested by from the anode to the cathode through the membrane. The
rotation disk electrode (RDE) method, not shown in this limiting hydrogen crossover flux can be calculated by ap-
study, were about 58 m2/gPt and 63 m2/gPt, respectively. plying Faraday’s law, equation (4), and the value was ~5.9
The oxygen reduction reaction (ORR) of Pt-alloy/C cata- × 10-9 mole/s/cm2. The pristine cell also did not exhibit an
lyst also showed a higher onset potential and higher activ- internal short as evidenced by the absence of an increasing
ity than Pt/C catalyst, which is consistent with the MEA current with increasing polarization beyond 0.15 V vs. RE/
performances. The ECSAs of Pt/C MEA and Pt-alloy/C CE. However, the degraded cell (red curve) showed signifi-
MEA were shown in Fig. 4(a) and Fig. 4(b), respectively.
The initial ECSA of Pt/C MEA was about 40 m2/gPt (black
curve) and the ECSA reduced to 27 m2/gPt (red curve) after
1500 hr test, 33% reduction of original ECSA. For Pt-al-
loy/C MEA, the initial ECSA of the Pt catalyst was 37
m2/gPt (black curve) and it reduced to 22 m2/gPt (red
curve) after 3700 hr test, 41% reduction of ECSA. The ini-
tial ECSA of Pt-alloy/C MEA was smaller than that of Pt/C
MEA. The Pt-alloy/C MEA had only 59% of catalyst utili-
zation, which was less than 70% of catalyst utilization for
Fig. 4. Electrochemical surface area of (a) Pt/C and (b)
Pt/C MEA. This may suggest an inappropriate MEA fabri-
Pt-alloy/C catalyst test before and after life test.
cation for Pt-alloy/C catalyst, which resulted in less proton Cell temperature 75 °C, 100 %RH, H2/N2, 100
transportation and reaction, since RDE results showed sim- sccm/100sccm at anode and cathode side, re-
ilar ECSA for both Pt/C and Pt- alloy/C catalyst. In litera- spectively. Scan rate is 20 mV/s.

6 www.jccs.wiley-vch.de © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim J. Chin. Chem. Soc. 2012, 59, 000-000
 JOURNAL OF THE CHINESE
MEAs with Pt/C and Pt-alloy/C Catalyst CHEMICAL SOCIETY

Table 3. Hydrogen crossover rate of Pt-alloy/C MEA before and

lyst, shown in Fig. 6(c), the Pt-alloy particles were larger
after durability test
with an average particle size of 5.03 nm and the Pt-alloy
Cross over H2 cross over particle size distribution was broader, from 2 nm to 14 nm,
current (A) sccm/cm2 mol/sec/cm2 than pure Pt particles. After the durability test, shown in
Before 0.0285 8.4 ´ 10-3 ~5.9 ´ 10-9 Fig. 6(d), the Pt-alloy particles showed little change of par-
After 0.1750 5.5 ´ 10-2 ~3.9 ´ 10-8 ticle size with an average particle size of 5.65 nm and the
Pt-alloy particle size distribution became a little bit broader,
from 2 nm to 17 nm. By comparing the TEM images, the
cant hydrogen penetration flux increase through the mem-
fresh pure Pt particles are smaller than Pt-alloy particles.
brane (ilim = 7 mA/cm2, hydrogen flux ~3.9 × 10-8 mole/s/
After long term durability test, however, pure Pt particles
cm2, almost an order of magnitude increase to the pristine
showed similar average particle size to Pt-alloy particles.
MEA). The curve of degraded MEA also shows an internal
The pure nano-scale Pt particles has been reported to grow
electronic short-circuit between the two electrodes, which
into larger particles by Ostwald ripening4 under fuel cell
is indicated by the linear, positive slope on the curve. The
test conditions. In these severe electrochemical reactions,
resistance of the electrical short is estimated to be 152
the metallic Pt tends to oxidize to Pt ions, and the dissolved
W-cm2. This may imply a pinhole formation or break of the
Pt ions favored to redeposit on the surface of Pt particles,
membrane, which is ascribed to the major failure mode of
leading to the particle growth and loss of ECSAs. From a
both Pt/C MEA and Pt-alloy/C MEA. This on-line diagno-
view of geometric similarity, the ECSAs of the pure Pt par-
sis method, moreover, can also be used as a standard safety
ticles on fresh and degraded MEAs should be inversely
specification for MEA or stack test, which etermines the
proportional to the particle diameters of fresh and degraded
MEA failure when the hydrogen penetration flux is over
the set security value.
Fig. 6 showed the TEM images of Pt/C and Pt-alloy/C
catalysts from fresh and degraded MEAs. The particle size
distributions of these catalysts were also summarized in
Fig. 7. In Fig. 6(a), most of the Pt particles on fresh Pt/C
were less than 4 nm and they showed an average particle
size of 2.94 nm. After the durability test, the Pt particle size
significantly increased with an average particle size of 5.24
nm and the Pt particle size distribution became broader
than those on the fresh catalyst. For fresh Pt-alloy/C cata-

Fig. 6. TEM images of MEA. Pt/C catalyst (a) before,

(b) after test ; Pt-alloy/C catalyst (c) before, (d)
after test.

Fig. 5. Hydrogen crossover test of Pt-alloy/C MEA be-

fore and after life test. Cell temperature 75 °C,
100%RH, H2/N2, 100sccm/100 sccm at anode Fig. 7. Particle size distribution of Pt catalyst in MEAs
and cathode side, respectively. Scan rate 2 before and after test. (a). Pt/C catalyst, (b).
mV/s. Pt-alloy/C catalyst.

J. Chin. Chem. Soc. 2012, 59, 000-000 © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.jccs.wiley-vch.de 7
Article Wang et al.

Pt catalysts. The ECSA ratio of fresh Pt particles to de- more than 20 nm. These large Pt particles act as heteroge-
graded Pt particles was estimated, from the inverse ratio of neous defects and catalyst for peroxide formation in the
fresh and degraded Pt particle diameters, to be 1.7, which is membrane, which could cause the pinholes and stress con-
quite close to the real ECSA ratio of 1.5. Thus, the change centration inside the membrane to speed up the membrane
of ECSA can be a good index of particles growth ratio for degradation. Furthermore, the deformation of the catalyst
pure Pt particles. However, the power loss percentage of layer was also observed in Fig. 8(c), compared to the fresh
Pt/C MEA cannot directly correlate to the Pt particle growth catalyst layer. The strength of the CCM was decreased after
and ECSA loss. For Pt-alloy/C MEA, the power loss per- long-term operation and part of the catalyst layer was dis-
centage is about 8%, and it seems to be related to the Pt-al- connected or lost in some region. The carbon corrosion and
loy particles size increase, ~ 7%, but it also can not be as- ionomers degradation in the catalyst layer have been re-
cribed to the ECSA loss of Pt-alloy/C MEA, with 41% de- ported to increase with decreasing RH, 18 which deterio-
crease. These results suggest that the capability of oxygen rated the integrity of the catalyst layer and resulted in a de-
species formation over Pt particles is more important than crease of ECSA and performance. The double layer capaci-
the Pt particle size effects, which relates to the ECSA, an tance decrease of the degraded MEA also demonstrates the
index of proton reduction and subsequent deposition of degradation of the whole catalyst layer. In a more severe
atomic hydrogen ability. This is the case since ORR is less operation conditions, without humidification at both anode
efficient than HOR for pure Pt particles in PEMFC opera- side and cathode side fed gas, all the ECSA was lost in the
tion. failed MEA and the catalyst layer became thinner without
Fig. 8 demonstrates the TEM images of micro-struc- observing significant Pt particle growth and membrane
tures of Pt-alloy/C MEA after durability test. After long- thinning, unpublished data.
term test, the degraded MEA is very fragile, which make AST results
the TEM sample preparation difficult. Many catalyst deg- The dissolution of Pt from the carbon support was
radation phenomena can be observed in Fig. 8. Fig. 8(a) in- less favorable at low electrode potentials, which makes Pt
dicates the agglomeration of nano-scale Pt particles, re- catalysts more stable at the anode electrode than that at the
sulted in the decrease of the active site and loss of MEA cathode side, and at low RH and low temperaturem19 There-
performance. In the meanwhile, Pt ions were dissolved fore, the catalyst degradation AST test was conducted at
from catalyst and then moved to the membrane via the elec- 100%RH and 65 °C to inhibit the carbon corrosion and
trophoresis effect. Then, the Pt ions can be reduced to the ionomer degradation rate. The test protocol is listed in Ta-
metallic state by hydrogen and precipitated in the mem- ble 1. The ESCA of Pt/C MEA and Pt-alloy/C MEA were
brane, as shown in Fig. 9(b) and Fig. 9(d). The metallic
platinum particles are single Pt crystals, whose sizes are

Fig. 8. TEM cross-section images of Pt-alloy/C MEA Fig. 9. Normalized ECSA of Pt/C and Pt-alloy/C as
after life test under low humidity condition. (a) function of cycle number. Cell temperature 65
Pt particles coarsen and (b)(d)migrate into °C, 100% RH, 100 sccm/100 sccm at anode and
membrane , as well as (c) catalyst layer defor- cathode side, respectively. Scan rate is 20
mation. mV/s.

8 www.jccs.wiley-vch.de © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim J. Chin. Chem. Soc. 2012, 59, 000-000
 JOURNAL OF THE CHINESE
MEAs with Pt/C and Pt-alloy/C Catalyst CHEMICAL SOCIETY

then estimated by CV during the potential cycling test pe- steadily with time until 80 hr. Over 80 hr of OCV holding
riod. Fig. 9 demonstrates the ESCA of Pt/C and Pt-alloy/C test, both potentials dropped dramatically while hydrogen
change as a function of cycle number. The ECSAs of both crossover current densities increased highly and the fluo-
catalysts decreased with an increase of test cycle number. ride release rates kept constant. These results suggest that
The ECSA of Pt/C MEA decreased more quickly and sig- more free radicals are produced over Pt/C MEA than Pt-al-
nificantly within 1000 cycles and only 48% of the initial loy/C MEA. This is consistent with Sethuraman’s work21
ECSA was obtained after 20000 cycles test. After 30000 that the Pt-alloy catalyst has a higher selectivity to depress
cycles, the ECSA of Pt/C MEA was less than 40%, below the H2O2 formation in ORR than pure Pt catalyst, indicat-
the set target. On the other hand, the ECSA of Pt-alloy/C ing the MEA with Pt-alloys catalyst are more efficient to in-
MEA increased in the first 100 cycles to reach the initial hibit the membrane failure and the ionomers degradation in
ECSA of Pt/C MEA and it then decreased steadily with the catalyst layer.
cycle number increase. After 30000 cycles test, over 65%
of ECSA was retained on Pt-alloy/C MEA. Moreover, Pt- CONCLUSIONS
alloy/C MEA showed higher OCV value and less OCV de- In this study, MEAs with Pt/C and Pt-alloy/C catalyst
cay rate than Pt/C MEA did. These results indicates that have been continually operated in constant current mode
Pt-alloy/C is much durable than Pt/C. under low humidity condition, and the degradation mecha-
The chemical degradation of membrane was acceler- nisms were evaluated by on-line electrochemical analysis
ated when the fuel cell is operated under open circuit volt- method and TEM image analysis. The Pt-alloy/C MEA has
age (OCV) and low humidity conditions.20 Thus, the OCV superior performance and durability to Pt/C MEA. The Pt-
holding test was conducted at low humidity conditions, as alloy/C MEA has operated over 3700 hr with a power loss
summarized in Table 2, for membrane degradation test. less than 8%, and the Pt/C MEA showed only 1500 hr of
Fig. 10 shows the OCV decay trend, hydrogen crossover operation with a decay rate over 50 mV/hr, about 7% of
current density, which corresponds to the hydrogen perme- power loss. TEM images show the metallic particle growth
ation rate, and fluoride release rate as a function of time for over both Pt/C and Pt-alloy/C catalyst and Pt-alloy parti-
both Pt/C MEA and Pt-alloy/C MEA. The Pt-alloy/C MEA cles are more resistant to particle growth, known as
revealed higher OCV, lower permeation rate, and lower flu- Ostwald ripening phenomenon. For ORR in the cathode
oride release rate at beginning and every subsequent on- side, the Pt particle size is less important than the ability to
line diagnosis than Pt/C MEA did. The fluoride release rate possess a higher coverage of oxygen species over the cata-
of Pt/C MEA was four to five times the fluoride release rate lyst.
of Pt-alloy/C MEA. The OCV of both samples decreased Both MEA showed a failure of dramatic voltage drop
and incapability for discharge and an increase of hydrogen
crossover flux, meanwhile, can be observed, which indi-
cates the membrane failure. This suggests that the mem-
brane and ionomer degradation played a fatal role on MEA
durability as operated at low humidity. From the fluoride
release rate monitoring results, Pt-alloy/C catalyst shows a
better prohibition against free radicals formation than Pt/C
catalyst. That’s the reason why the Pt-alloy/C MEA is more
durable than the pure Pt/C MEA under low humidity opera-
tion.

ACKNOWLEDGEMENTS
The authors would like to thank the Bureau of En-
Fig. 10. OCV holding test of Pt/C and Pt-alloy/C MEA
ergy, Ministry of Economic Affairs of Republic of China,
as a function of operation time. Cell tempera-
ture 80 °C, 30 %RH, 100 sccm (H2)/100 sccm for financially supporting this research under Contract No.
(Air) at anode and cathode side, respectively. 101-D0212.

J. Chin. Chem. Soc. 2012, 59, 000-000 © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.jccs.wiley-vch.de 9
Article Wang et al.

REFERENCES 2005.
1. Hydrogen, Fuel Cells & Infrastructure Technology Program 12. Yuan, X.-Z.; Li, H.; Zhang, S.-S.; Martin, J.; Wnag, H.-J. J.
Multi-Year Reaserch, Development and Demonstration Power Sources 2011, 196, 9107.
Plan, August 2006 at http://www1.eere.energy.gov/ 13. Hashimasa, Y.; Matsuda, Y.; Imamura, D.; Akai, M; Sasaki,
hydrogenandfuelcells/mypp. (Accessed May 31, 2012) M. Transactions of the Japan Society of Mechanical Engi-
2. Barbir, F. PEM Fuel Cells: Theory and Practice; Elsevier neers. B 2011, 77, 147.
Academic Press: San Diego, 2005. 14. Stamenkovic, V. R.; Mun, B. S.; Arenz, M. Nat. Mater. 2007,
3. Shao, Y.; Yin, G.; Gao, Y. J. Power Sources 2007, 171, 558. 6, 241.
4. Watanabe, M.; Tsurumi, K.; Mizukami, T.; Stonehart, P. J. 15. Curtin, D. E.; Lousenberg, R. D.; Henry, T. J.; Tangeman, P.
Electrochem. Soc. 1994, 141, 2659. C.; Tisack, M. E. J. Power Source 2004, 131, 41.
5. Zhang, J.-X.; Litteer, B. A.; Gu, W.-B.; Liu, H.; Gasteiger, H. 16. Xu, H.; Kunz, R.; Fenton, J. M. Electrochem. Solid-State
A. J. Electrochem. Soc. 2007, 154, B1006. Lett. 2007, 10, B1.
6. Kim, L.; Chung, C.-G.; Sung, Y.-W.; Chung, J.-S. J. Power 17. Wakisaka, M.; Suzuki, H.; Mitsui, S.; Uchida, H.; Watanabe,
Sources 2008, 183, 524. M. J. Phys. Chem. C. 2008, 112, 2750.
7. Ascarelli, P.; Contini, V.; Giorgi, R. J. Appl. Phys. 2002, 91, 18. Borup, R. L.; Davey, J. R.; Garzon, F. H.; Wood, D. L.;
4556. Inbody, M. A. J. Power Sources 2006, 163, 76.
8. Wilkison, D. P.; St-Pierre, J. Handbook of Fuel Cells: Fun- 19. Mathias, M. F.; Makharia, R.; Gasterger, H. A.; Conley, J. J.;
damentals, Technology and Applications; John Wiley & Fuller, T. J.; Gittleman, G. J.; Kocha, S. S.; Miller, D. P.;
Sons Ltd.: Hoboken, New Jersey, 2003; Vol. 3, Chpter 47. Mittelsteadt, C. K.; Xie, T.; Yan, S. G.; Yu, P. T. Electrochem.
9. Roen, L. M.; Paik, C. H.; Jarvi, T. D. Electrochem. Solid- Soc. 2005, 14, 24.
State Lett. 2004, 7, A19. 20. Endoh, E.; Terazono, S.; Widjaja, H.; Takimoto, Y.
10. Wang, H.-J.; Li, H.; Yuan, X.-Z. International Standard Electrochem. Solid-State Lett. 2004, 7, A209.
Book Number-13; 2012; p 78. 21. Sethuramam, V. A.; Weidner, J. W.; Haug, A. T.; Pemberton,
11. Yamazaki, O.; Oomori, Y.; Shintaku, H.; Tabata, T. Fuel Cell M.; Protsailo, L. V. Elctrochem. Acta 2009, 54, 5571.
Seminar Abstracts; Courtesy Associate: Washington D.C.,

10 www.jccs.wiley-vch.de © 2012 The Chemical Society Located in Taipei & Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim J. Chin. Chem. Soc. 2012, 59, 000-000