A Review of The Main Parameters Influencing Long-Term Performance and Durability of PEM Fuel Cells
A Review of The Main Parameters Influencing Long-Term Performance and Durability of PEM Fuel Cells
A Review of The Main Parameters Influencing Long-Term Performance and Durability of PEM Fuel Cells
com
Review
Abstract
This paper presents an overview of issues affecting the life and the long-term performance of proton exchange membrane fuel cells based
on a survey of existing literature. We hope that this brief overview provides the engineers and researchers in the field with a perspective of the
important issues that should be addressed to extend the life of next-generation fuel cells. Causes and fundamental mechanisms of cell degradation
and their influence on long-term performance of fuel cells are discussed. Current research shows that main causes of short life and performance
degradation are poor water management, fuel and oxidant starvation, corrosion and chemical reactions of cell components. Poor water management
can cause dehydration or flooding, operation under dehydrated condition could damage the membrane whereas flooding facilitates corrosion of
the electrodes, the catalyst layers, the gas diffusion media and the membrane. Corrosion products and impurities from outside can poison the cell.
Thermal management is particularly important when the fuel cell is operated at sub-zero and elevated temperatures and is key at cold start-ups as
well as when subjected to freezing conditions.
© 2008 Elsevier B.V. All rights reserved.
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Influence of water management on fuel cell performance and life . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2.1. Fuel cell flooding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1.1. Cathode flooding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1.2. Anode flooding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2. Dehydration of the membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3. Degradation of the electrodes/electrocatalyst, membrane, gas diffusion layer and bipolar plates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.1. Corrosion of the electrodes/electrocatalyst . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.1.1. Cathode corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
3.1.2. Anode corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.2. Corrosion of the gas diffusion layer (GDL) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.3. Chemical and mechanical degradation of the membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.4. Corrosion and mechanical degradation of the bipolar plates and gaskets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4. Contamination of the cell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.1. Contamination of the electrodes/electrocatalyst . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4.2. Contamination of the membrane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
5. Reactant gas starvation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
6. Thermal management of PEM fuel cells and the impact on performance and durability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
6.1. Influence of freezing temperatures on durability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
∗ Corresponding author. Tel.: +1 864 656 4718; fax: +1 864 656 7299.
E-mail addresses: [email protected] (W. Schmittinger), [email protected] (A. Vahidi).
0378-7753/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jpowsour.2008.01.070
2 W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14
Table 1
Results of long-term durability tests under laboratory conditions (steady state)
Authors Test time (h) Degradation rate (V h−1 )
Sishtla et al. [9] (reformate fuel, 4 thermal cycles over 1200 h) 5100 6
Washington [10] (Ballard Mk5R) 4700 6
Washington [10] (Ballard Mk6000) 8000 2.2
Nakayama [11] 4000 4.3
Maeda et al. [12] (on reformate fuel) 5000 6
Fowler et al. [13] (non-continuous operation, start-stop cycling, long-term storage, dehydration, flooding) 600 120
Endoh et al. [14] 4000 2
Knights et al. [15] (short stack, methane reformate operation, 0.5 A cm−2 ) 13,000 0.5
Scholta et al. [16] (GDL Toray TGP 120, 600 mA cm−2 , Tstack = 65–70 ◦ C, H2 , O2 humidified) 2500 20
Scholta et al. [16] (GDL SIGRACET® SGL-10BB, 300 mA cm−2 , Tstack = 55 ◦ C, H2 = 1.3 bar, dry 2500 60
O2 = ambient pressure, dry)
Cheng et al. [17] (400 mA cm−2 , Tcell = 60 ◦ C, H2 O2 : RH = 100%, ambient pressure) 4000 3.1
Wang et al. [18] (∼1.0 A cm−2 , 0.64 V) 1000 54
Lightner [19,20] (first 1000 h steady state, degradation 2 V h−1 ; then accelerated cyclic testing) 4000 20
Borup et al. [21,22] (steady-state, Tcell = 80 ◦ C, RHH2 /O2 = 75%, gas pressH2 /O2 = 15 psig, 1000 12
flow rateH2 /O2 = 1.2/2.0*1.5 A cm−2 equiv.)
Cleghorn et al. [23] (single cell, 800 mA cm−2 , Tcell = 70 ◦ C, RH H2 , O2 100%) 26,300 4–6
W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14 3
nents. The longer the cell is exposed to excess water the stronger
is the degradation. Therefore a proper water balance between
water formation and water removal is required [25,27]. The
water balance depends on the water carried in or formed inside
the cell and the water removed out of the cell. Reactant gases,
which are pre-humidified or saturated and the oxygen-reduction
reaction at the cathode generate water within the cell. Water is
removed by evaporation into the gas streams, exiting humidified
gases and exiting liquid water.
reactant gases as well as liquid water injection also facilitate with a low hydration state of the fuel gas stream. If the
flooding. hydration state is not as high as at the cathode in addition
Water removal mechanisms are water back-diffusion to the to low current densities, water back-diffusion through the
anode, evaporation, water-vapor diffusion and capillary trans- GDL to the anode will surpass the electro-osmotic effect.
port of liquid water through the porous cathode backing layer Ge and Wang [36] observed, by using a camera, that water
[28]. Water back-diffusion takes place when the water content vapor coming from the anode GDL condensed at its surface.
of the cathode side of the membrane exceeds that of the anode However, the water accumulation by condensation did not
side [35]. In comparison to other mechanisms, back-diffusion seem to be much.
does not contribute much to water removal. Only at low cur- (iii) Liquid water injection for cooling and humidification
rent densities (<≈0.3 A cm−2 ) the effect of back-diffusion can together with moderate cell temperatures (lower evapo-
exceed electro-osmosis [27]. Water evaporation is facilitated by ration) can be another reason for anode flooding [27,29].
higher cell temperatures and a higher air flow rate is benefi- Liquid-water accumulation in the anode by humidification
cial to carry water out of the cell. He et al. [28] show that by was confirmed by Ge and Wang [36].
increasing the cell temperature from 40 to 50 ◦ C the cathode
pressure can drop from 3 to 2 kPa and hence the flooding level 2.2. Dehydration of the membrane
drops within 15 min (air flow rate, 2.0 A cm−2 ; H2 flow rate,
2.0 A cm−2 ; p = ambient pressure; H2 , 43 ◦ C; air, 25 ◦ C). Sev- Dehydration of the membrane is more likely to occur at the
eral factors, caused by a higher temperature contribute to a higher anode side of the membrane. The main factor contributing to
water removal rate. At higher temperatures, water evaporation dehydrated condition of the membrane is probably poor water
as well as the volumetric air flow rate (if the mass flow rate of air management leading to a shortage of water. Dehydrated cell
stays the same) will increase. Also, a decrease in surface tension operation leads to instant and long-term degradation. The main
and viscosity of water makes it easier to flush water out of the consequence of dehydration is drying of the proton-conducting
cell [28]. Under the same conditions doubling of the air flow membrane. With decreasing water content the conductivity
rate to 4 A cm−2 leads to a current density increase from 0.5 decreases which leads to higher ionic resistance and higher
to around 0.75 A cm−2 within 1 h [27]. The capillary transport ohmic losses [5,24,37]. That results in a substantial drop in cell
of water through the porous cathode backing layer to the flow potential and thus a temporary power loss [24,27,35]. Although
channels also helps reducing the hydration level [28]. a temporary drop in voltage can usually be recovered by humid-
ification, dry cell operation over a long time can cause serious
2.1.2. Anode flooding and irreversible damage to the membrane. The recovery time
Since the cathode is naturally the water generating electrode, depends on the membrane thickness and the water diffusion
it takes much longer to accumulate water at the anode [36]. coefficient [24,35]. Sone et al. [37] measured the conductiv-
Although flooding at the anode happens less often than at the ity for Nafion 117 membranes in terms of relative humidities
cathode, it can have serious consequences on fuel cell operation; (RH) with the AC impedance method (Table 2). Fig. 2 shows
one being fuel starvation with subsequent carbon corrosion in an experiment performed by Le Canut et al. [24] in which a
the catalyst layer. Also, due to usually low hydrogen flow rates, cell was exposed to drying conditions for about 12 min. In this
liquid water is more likely to stay in the anode [36]. time period the initial cell voltage of around 0.8 V (current den-
sity = 0.1 A cm−2 ) dropped to about 0.75 V. After 15–20 min
(i) Anode flooding is more likely to happen at low current of rehumidification the initial value could be reached again.
densities [29]. In experiments carried out by Ge and Wang Data of Büchi and Srinivasan [38] collected in lifetests under
[36] liquid water at the anode could only be found when zero external humidification show long-term MEA performance
operating the cell at low current densities (0.2 A cm−2 ), degradation. In 1200 h of operation (observed in the interval
whereas at high current densities a higher electro-osmotic from 150 to 1350 h of operation) the current density dropped
force reduced the water content at the anode. In addition, from 170 to 130 mA cm−2 at a constant potential of 0.61 V.
lower cell temperatures and hence higher water conden- Anode dehydration is expected to be more serious at the
sation in the anode channels contribute to anode flooding inlet of the cell. That can be explained by the higher water
[36]. Pasaogullari and Wang [11] confirm the statement of back-diffusion to the anode at the bottom of the cell. Since the
Ge and Wang that anode flooding is often observed at low hydration state at the exit of the cathode is higher, caused by exit-
current densities, especially at low reactant flow rates and
lower temperatures. At the inlet of the anode, where the Table 2
proton flux is high, a strong electro-osmotic force drags Conductivity (S cm−1 ) at different relative humidities [%] for E- and N-form
the water molecules from the anode to the cathode result- Nafion 117 membranes (E-form: no heat-treatment, N-form: heat treatment at
85 ◦ C and 105 ◦ C) [37]
ing in low water content. At the exit in contrast, where the
current density is lower, the water content is higher. Two RH (%)
other (independent) mechanisms also seem to contribute to 20 60 100
anode flooding.
Proton conductivity E-form ≈2 × 10−3 ≈2 × 10−2 ≈7 × 10−2
(ii) Nguyen and White [27] show that anode flooding can be (S cm−1 ) N-form ≈3 × 10−4 ≈8 × 10−3 ≈5 × 10−2
caused by water back-diffusion from the cathode together
W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14 5
Table 3 Table 4
Potential cycling effect on the electrochemical active surface area (ECSA) [7] Pt-particle size (nm) after cycling from 0.1 to 0.96 V as a function of relative
humidity and cycling temperature [7]
# of potential cycles % initial ECSA % initial ECSA
(0.1–0.75 V) (0.1–1.2 V) RH (%) 10 50 100
Pt-particle size (nm) 2.6 3.2 3.3
300 96 60
900 90 23 Temperature (◦ C) 60 100 120
1500 83 11 Pt-particle size (nm) 2.8 4.1 4.8
(i) Cell potential cycling is one of the most important factors relative humidity (RH) of the gases, the less is the growth of
contributing to platinum agglomeration and/or oxidation the catalyst particles, especially during potential cycling.
and hence to a decrease in ECSA. Table 3 shows that with Low RH helps the lifetime of the catalyst. Table 4 shows
the number of cycles, the lifetime of the catalyst and cell the particle-size as a function of RH.
potentials strongly go down. For instance, in simulated
drive cycle tests of Borup et al. [46] a 10% decrease of the The catalyst loading, that is the density of Pt-particles, is
ECSA could be observed after operating a cell for 850 h believed to have no further influence on particle growth or cor-
under 100% RH (20-min cycles, total 2550 cycles). Other rosion [7,46,54]. However, according to the paper of Boyer et al.
experiments of Borup et al. show a loss of the ECSA of [57] the loading has at least an influence on the cell output. At a
around 40% after 1500 potential cycles from 0.1 to 1.0 V fixed cell potential of 0.7 V an increase of the loading from 0.03
[7]. to 1 mg Pt cm−2 results in a 200% increase of the current density
During cycling the Pt-particle size grows depending on to a final value of around 1.5 A cm−2 (H2 /O2 : 50 ◦ C, p = atm).
the cell potential and faster than during constant poten- Another important factor influencing the ECSA degradation
tial operation. In general, higher potential levels accelerate is the starting surface of the catalyst which highly affects the
cell degradation. Agglomeration can also be noticed during amount of dissolved platinum. For example an initially oxidized
constant cell voltages, but it is not as strong. For instance, platinum surface results in a higher amount of dissolved Pt-
the particle-size can increase up to 8 nm after 1500 cycles ions than a reduced catalyst surface. Thus, in order to obtain
from 0.1 to 1.2 V (cell temperature, 80 ◦ C; H2 RH 226%, air for instance comparable experimental data one should also
RH 100%) [7]. Wilson et al. [56] have shown that the num- take the starting condition of the Pt-catalyst layer into account
ber of Pt-particles smaller than 3 nm was reduced from 40 [18].
to 5% at the cathode after operation of a PEMFC for 2200 h Table 5 provides a quick overview of literature and operating
(operation at maximum power, overhumidified gases, 80 ◦ C conditions on electrocatalyst degradation.
cell temperature).
To maintain a large catalyst surface for a long time, 3.1.2. Anode corrosion
operation of the fuel cell at both constant voltages and The anode catalyst is much less susceptible to corrosion than
low voltage levels is preferred, that means, higher poten- the cathode catalyst. Various long-term experiments in indepen-
tial levels accelerate cell degradation. For example, if one dent works show the anode to be almost unaffected by platinum
considers automotive applications high potentials during agglomeration/sintering, dissolution and oxidation, irrespective
sitting of the stack at stoplights or during idling occur and of the conditions (constant voltage or cycling) at which the fuel
accelerate particle growth. Cell voltages during rapid shut- cell was operated [7,46,55].
down of the stack are even higher and hence more harmful Only in extended fuel cell testing after long operation periods
[46]. anode catalyst deterioration can be observed. Platinum agglom-
Particle redistribution can be explained with potential eration does not occur, although the ECSA decreases. One
cycling. It is believed that platinum-solubility is a function reason might be the detachment from the carbon layer and loss
of the potential and that there exists a particular equilib- of Pt-particles. Another reason may be the loss of ionomer in
rium voltage. When the cell is cycled above the equilibrium the catalyst (Pt-particles which are not well-bound to the car-
voltage Pt-ions are driven into the solution (water). When bon support move in the ionomer and can be lost easier than
decreasing the cycling potential again below the equilib- well-bound particles [7,58]).
rium, the platinum is forced back out of the solution onto
the catalyst surface again [7,46,55]. Step changes in the 3.2. Corrosion of the gas diffusion layer (GDL)
cell voltage may also help this process [18].
(ii) Different cell temperatures during operation also have an To this day much less research work has been done on the
impact on the ECSA. Generally, kinetics goes up with field of carbon corrosion than on the electrode catalysts. But it
increasing temperature and hence higher temperatures is known that besides the catalyst particle growth which leads
result in faster growth of Pt-particles [7,46,54]. This is to ECSA loss, carbon corrosion of the gas diffusion layer has a
shown in Table 4. negative influence on the catalyst properties and subsequently
(iii) The humidification level of the incoming reactants affects on the output voltage and performance of the cell. Since carbon
the growth of the catalyst particles as well. The lower the paper or cloth often serves as the support for the catalyst, car-
W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14 7
Table 5
Literature overview of electrocatalyst degradation
Authors Experimental setup/testing conditions Initial ECSA (%) Loss of catalyst Comments
bon corrosion directly affects lifetime of the cell [54]. With lost surface (Vulcan/Pt) area was also shown. At 0% carbon
carbon, the bound platinum can also be lost [55]. weight loss of the Pt-surface area was about 40 m2 g−1 Pt.
This value was reduced to around 15 m2 g−1 Pt at a carbon
(i) Potential cycling, particularly on a high level and high con- weight loss of about 65% [54].
stant voltages exacerbate the loss of carbon material and
therefore the cell lifetime [7,59]. 3.3. Chemical and mechanical degradation of the
(ii) Lower relative humidity of reactant gases aggravates the membrane
loss of carbon over time [7]. In experiments and subse-
quent analysis it was observed that an increase of small Degradation of the membrane is probably among the main
pore volumes happens which is believed to be caused by factors reducing the lifetime of PEM fuel cells. Chemical stabil-
loss of carbon material from the micro-porous GDL. But ity of the membrane is critical to fuel cell’s long life. Nowadays,
regarding humidification of the GDL one needs to look at DuPont’s Nafion® and Gore’s Primea® series are two of the
it accurately. In general, the water management is handled mostly used ion exchange membranes and are considered as the
by the GDL. If we look at the performance of the cell in industrial standard [60]. For example, Nafion® consists of a ther-
terms of humidity, it can be attributed to the performance of moplastic resin which, due to its perfluorinated composition, is
the GDL. New GDLs are usually treated to be hydrophobic relatively stable both chemically and thermally [61]. It has been
to facilitate water removal and improve gas diffusion, that shown that Nafion-type membranes are long-lasting while being
is, they are optimized for high relative humidity. But over used for ion exchange or electrolysis. They can reach lifetimes of
time they lose the hydrophobicity which in turn reduces the several thousands of hours [5]. However, for application in PEM
mass transport rate of the gas. Higher hydrophilicity (lower fuel cells they are more vulnerable and degrade more rapidly,
hydrophobicity) means that more water tends to stay in the especially in automotive applications during potential cycling,
GDL blocking the pores and hinders transport of reactant during start-up and shut-down phases as well as in freezing
gas molecules. This leads to a performance loss [7,46,54]. periods when exposed to sub-zero temperatures [5].
(iii) The literature is in disagreement on the influence of the It is widely understood that the degradation of the mem-
operating temperature on the GDL corrosion. The reason brane is a complicated multi-step mechanism, which can lead to
may be different experimental setups or testing conditions. catastrophic failure when operating the fuel cell for an extended
Experiments of the Los Alamos National Laboratory [7] period of time. The two major steps are (i) formation of hydroxyl
show that temperature does not affect carbon corrosion, (OH) and peroxyl (OOH) radicals stemming from hydrogen per-
however this might only apply to this certain experimen- oxide (H2 O2 ). They chemically attack the polymer end groups
tal setup. Li et al. [54] have monitored the carbon weight that are present in the membrane [53,62]. (ii) The chemical attack
loss (gas-phase corrosion) at steady fuel cell operation with along with the transient operating conditions such as poten-
humidified air at different temperatures. For example, in tial, temperature and humidification cycling causes mechanical
accelerated tests at an air temperature of 120 ◦ C a carbon degradation and a change in membrane properties [53,62–64].
weight loss of 8% and at 150 ◦ C a loss of around 36% after Many researchers are unanimous in believing that the chem-
125 h could be observed (carbon support: Vulcan). A cor- ical attack caused by radicals initiates membrane degradation
relation between carbon corrosion and loss of the catalyst [5,53,62,65]. Due to their unpaired electrons, the highly reactive
8 W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14
Fig. 4. Image of a cracked membrane/MEA. (Reproduced from [66] with per- Corrosion of the bipolar plates also impacts performance and
mission from Elsevier). life of a fuel cell. Three major degradation mechanisms have
W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14 9
producing oxygen in the anode: Nafion® -type membranes rather than the water inside the mem-
brane leads to degradation of the MEA. Due to the different
2H2 O → O2 + 4H+ + 4e− (anode)
densities of water and ice (0.9998 and 0.9168 g m−3 , respec-
Also, in the absence of hydrogen, the following anode reaction tively) the volume of freezing water expands about 9% [82].
can take place: The repetitive cycles of ice formation on the membrane surface
and melting into water can delaminate the catalyst layer from
C + 2H2 O → CO2 + 4H+ + 4e− (anode) both the membrane and the gas diffusion layer [3,82,83]. The
Similarly during oxygen starvation the reaction at the cathode resulting mechanical damage might lead to a loss of thermal and
will produce hydrogen [77]. The normal cathode reaction: electrical interfacial contact since the components within the cell
are no longer in proper contact. He and Mench [84] correlated
O2 + 4H+ + 4e− → 2H2 O the ice formation on the membrane surface with its thickness and
changes to initial water content. Since the Nafion® membranes have a large
freezing temperature depression (∼24.5 K), the contained liquid
2H+ + 2e− → H2 (cathode) [76, 77]. water can flow out of the membrane and will freeze immediately
on the surface and in the catalyst layer. The thicker the membrane
The presence of oxygen at the anode and hydrogen at the cath-
and the higher the initial water content, the thicker will be the
ode will lead to reversal of the cell potential, that is a negative
developing ice layer. For instance, in the case of Nafion® 112, the
potential difference between the anode and the cathode [26]. Cell
ice thickness is 5 m, whereas it is about 18 m for Nafion® 117
reversal accelerates corrosion of carbon components such as the
(water content of Nafion® = 20 molH2 O (molSO3 − )−1 ). Also,
backing layers with ensuing electrocatalyst corrosion and even-
cracking of fully hydrated membranes after several freeze/thaw
tually leads to damaged components [76–78]. During hydrogen
cycles (from −30 to 20 ◦ C) was observed in tests carried out
starvation, oxygen at the anode can react with the carbon present
by Plug Power [59]. However, the lower the hydration state of
in the gas diffusion and backing layers to form carbon dioxide
the membrane after shut-down, the less serious are the cracks.
(second anode reaction above) [5,76,77,79].
Cracks in the membrane lead to gas crossover and in turn to
Several factors can cause reactant starvation. A poor water
uncontrolled reaction between hydrogen and oxygen with subse-
management with flooding and a poor heat management dur-
quent pinhole formation damaging the membrane and reducing
ing sub-zero temperatures and cold start-ups with ice within the
the life of the cell [83]. This process has been discussed in Sec-
cell can block the pores of the gas diffusion layers. A poor gas
tion 3.3 of this paper. In the review part of the paper by Kim
feeding management can lead to non-uniform distribution of the
and Mench [85] some results show that cells which were dried
reactant gases resulting in partial or complete fuel and/or oxi-
during shut-down show neither observable physical damage nor
dant starvation or in sub-stoichiometric operation in individual
electrochemical losses during freezing.
cells. Also, an imperfect stack and cell design with an uneven
distribution of mass in the flow fields, a poor stack assembly as
well as quick load demands can be reasons contributing to gas 6.2. Influence of freezing temperatures on performance
starvation [5,76].
In general, performance meaning the power output of PEM
6. Thermal management of PEM fuel cells and the fuel cells decreases with decreasing temperatures, especially
impact on performance and durability when the stack is operated below 0 ◦ C [5,86]. However, when
cell temperatures reach +80 ◦ C and higher performance starts to
Thermal management is particularly important when the fuel decrease again [3,6].
cell is exposed to freezing temperatures. To this day the opera-
tion of PEM fuel cells when cycled between sub- and above-zero 6.2.1. Different experiments on performance degradation
temperatures and at elevated temperatures as well as cold starts The work of Chang et al. shows that with decreasing tem-
are not completely understood. Although a lot of research has perature of the liquid water in the cell from 80 to 30 ◦ C the
been done in recent years, more work needs to be done to bet- proton conductivity of a Nafion® membrane can decrease up to
ter understand degradation of performance and durability under 30%, depending on the thickness and the measurement method
sub-zero and elevated temperature operation [3,6,8,80,81]. [87]. Mukundan et al. [8] observed that the conductivity of
a Nafion® membrane below 0 ◦ C is higher, when the water
6.1. Influence of freezing temperatures on durability content is lower. In conductivity measurements at different tem-
peratures of Mukundan et al. Nafion® membranes showed 10
Exposure of a non-operating PEM fuel cell to freezing tem- times higher conductivity at around 25 ◦ C (0.02 S cm−1 ) than
peratures is one of the issues affecting durability. When the at −30 ◦ C (0.002 S cm−1 ) [8]. Performance of the fuel cell
fuel cell is subjected to sub-zero temperatures for an extended can be attributed to some extent to the conductivity of the
period of time the residual water contained within the cell can membrane. In the study of McDonald et al. experiments were
freeze. This leads to thermal and mechanical stress and hence to carried out to understand the physical and chemical changes in
mechanical damage of the cell components or may even cause the membrane (Nafion® 112)/MEA during freeze/thaw cycles
physical breakdown. It was observed that freezing water on [88]. After 385 cycles from 80 to −40 ◦ C over a period of 3
W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14 11
months, the membranes/MEA were analyzed on ionic conduc- 6.2.2. Start-up from freezing temperatures
tivity, structure, and ultimate strength. No catastrophic failure Another concern at sub-zero temperatures is the start-up of
was observed, but the membrane structure at the molecular level fuel cells. If the generated water in the cathode is not removed
seemed to have changed resulting in a different percent elonga- while the cell is running at sub-zero temperatures, ice will form
tion at failure and ultimate strength as well as the ability for water causing voltage drop and even shuts down the electrochemical
take up. reaction. This is more likely to happen at higher current densi-
In the open literature it is agreed that fuel cells cycled between ties. Therefore it is important that before ice completely blocks
sub- and above-zero temperatures (cell temperatures) for an the catalyst layer the cell temperature reaches above freezing
extended period of time show strong degradation with reversible [91,92]. Ge and Wang [93] studied liquid water and ice formation
as well as irreversible damage [83]. on the surface of the catalyst layer during cold start at different
Mukundan et al. [8] show performance loss and damage freezing temperatures (−5, −3 and −1 ◦ C). Before the cold start
of a fuel cell under freeze/thaw cycles. In their experiments with dry reactants the cell was purged for 2 min and the current
with Nafion® 112 and humidified reactants the performance density was set to 0.02 A cm−2 . Throughout the whole cold start
decreased with each cycle from −80 to 80 ◦ C. For example after no water drops or ice on the surface of the “catalyst-coated mem-
9 cycles the cell potential dropped from initially 0.6 to 0.5 V (at a brane” (CCM) was observed which shows that purging the cell
constant current density of 1.0 A cm−2 ). After 10 cycles the cell prior to start is beneficial during cold starts. Tajiri et al. conducted
failed completely due to physical damage of the carbon cloth. a test on isothermal start-up of fuel cells from −30 ◦ C. They
In other tests of Mukundan et al., a different fuel cell was cycled show that the membrane is the critical component for improving
from −40 to 80 ◦ C. A potential loss could also be observed. the performance during isothermal cold start [94]. Khandewal
After 40 cycles the cell potential dropped from initially ca. 0.57 et al. [95] also studied cold start behavior and the corresponding
to ca. 0.54 V at a constant current density of 0.9 mA cm−2 [8]. energy requirement of PEM fuel cells using a one-dimensional
Cho et al. [82] measured the current density of a cell subjected to thermal model. They observed that there is a range of current
thermal cycling at a constant voltage of 0.6 V. After four cycles density in which the cell can be started optimally from freez-
from 80 to −10 ◦ C (measurement at 80 ◦ C, cool down to −10 ◦ C, ing temperatures. The simulation shows that a current density
1 h at −10 ◦ C, warm up to 80 ◦ C for the next measurement) the of 0.1 A cm−2 is not enough to bring the stack from −20 to
current density dropped from initially 880 to 780 mA cm−2 . In above 0 ◦ C in order to reach the start-up condition, whereas at
the research paper of Mukundan et al. [89] polarization curves 1.0 A cm−2 the stack needed around 69 s (neither heated coolant
during freeze/thaw cycling experiments using different GDL, flow was used nor initial ice was present). The start-up time of
among other things, were obtained. It could be seen that a paper 69 s could be reduced significantly to 20 s, when the fuel was
GDL performed worse than a cloth GDL. In these tests, the stack heated from −20 to around 0 ◦ C. Interestingly, heating the reac-
was operated at 80 ◦ C with humidified reactants. After reactant tant gas at the cathode did not contribute much to a faster start-up.
shut-down it was subjected to 100 repeated freeze/thaw cycles Additionally, heated endplates of the stack is recommended to
from 40 to −40 ◦ C (slow cooling within 4 h) and from 80 to achieve a rapid start-up. Ahluwalia and Wang [91] observed that
−40 ◦ C (fast cooling within 1 h). In the slow cooling experi- the current density during cold start needs to be limited in order
ments the cells were removed to be tested at 80 ◦ C; when fast to prevent continuous voltage decrease until shut-down of the
cooled after 10 cycles. During the slow cooling, the paper GDL electrochemical reaction due to freezing water. In experiments
showed strong degradation after 40 cycles, especially at high (isothermal) and in calculations ice is formed only when the cur-
current densities in the mass transport region, whereas the cloth rent density exceeds a certain value. For example, during start-up
GDL showed no performance loss even after 100 cycles. For from −25, −20 and −10 ◦ C, the maximum current density was
the experiments with the fast cooling (new cells prepared the determined to be less than 1, 3 and 10 A cm−2 , respectively
same way) an even stronger degradation could be observed for (pressure = 1 atm, flow rate L min−1 : H2 0.66 (−25 ◦ C), 1.19
the paper GDL, while the influence on the cloth GDL was little (−20 ◦ C) and 0.79 (−10 ◦ C); O2 2.83, 9.49 and 6.33).
[89].
When operating in freezing conditions as with a fuel cell 6.3. Influence of elevated temperatures on performance and
exposed to freezing temperatures, the MEA, the backing layers lifetime
and the gas diffusion layers can be delaminated. The backing
layer fibers and the binder structure as well as the gas diffusion Operation of fuel cells under higher temperatures (>100 ◦ C)
layer itself can be damaged, too. The reason for GDL deteriora- has a few advantages [81]: the electrochemical kinetics and
tion might be the relatively high water content during operation hence the efficiency improves, the tolerance for contaminants
and after shut-down. Therefore the probability that freezing increases and water management and cooling are enhanced
water can be found within the pores is high. With delamination due to a higher temperature difference between the cell and
of various component layers the thermal and electric interfacial the coolant. Waste heat can be recovered, CO-poisoning is
contact is lost [3,83,90]. Basically, the same consequences as reduced and therefore lower quality hydrogen from reformation
mentioned in the previous paragraph can be given. Water con- can be used [81,96]. Despite the various advantages of operat-
tained inside the membrane with strong bonds with cations does ing the fuel cell at higher temperatures the degradation of cell
not freeze and therefore does not have an impact on performance components will be accelerated and longtime performance and
[3,90]. durability are expected to decrease [81,96–98].
12 W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14
While even at moderate temperatures the degradation of tion, mostly at the anode side of the membrane can occur which
the electrodes/electrocatalyst and growth and loss of its (Pt-) may lead to lower conductivity in the short term and brittleness
particles is a concern [5,7,46], at higher temperatures the chem- in the long term. Membrane cracking is possible leading again
ical (in)stability of the catalyst is of even higher importance to gas crossover, hot spots and pinhole formation, a subsequent
as redistribution, sintering and agglomeration of the particles destructive cycle leads to cell failure.
are accelerated [51,96,98]. This process was described in detail Operation or even storage of fuel cells at freezing tempera-
when we discussed corrosion of the electrodes and electrocata- tures results in internal thermal and mechanical stresses, cracks
lyst. A second issue at elevated temperature operation can arise in the membrane, delamination of the component layers and loss
if the potential at the cathode is held high. This initiates the split of electrical contact, all negatively influencing fuel cell perfor-
of the oxygen molecule into oxygen atoms, which, at a high tem- mance and life. Operation at elevated temperatures aggravates
perature, can react easily with carbon and/or water to form CO degradation mechanisms such as corrosion.
and CO2 resulting in carbon corrosion [81]. This in turn causes Since a fuel cell stack is a complex system consisting of elec-
catalyst degradation which affects lifetime [54]. trodes, membrane, gas diffusion layers and other components
At elevated temperature operation the water uptake of the fuel cell life depends on individual components as well as on
membrane can be aggravated and proton conductivity may the interaction of all parts. Degradation mechanisms are inter-
decrease, especially at low relative humidities. This leads to connected and individual degradation can influence or initiate
a significant resistive loss lowering the cell performance and further deterioration of other components. Therefore it is diffi-
efficiency [97]. For example, Song et al. studied the effect of cult to quantify durability and to rank the mechanisms. However,
Nafion® content on electrode performance under three different corrosion and change in the chemical structure of the membrane
temperatures (80, 100 and 120 ◦ C). They observed that the influ- might be among the most important issues during operation.
ence of the Nafion content at 120 ◦ C was larger than at the lower Degradation of the catalysts and of other components deterio-
temperatures, meaning that the content needs to be taken into rates performance but plays a less important role in sudden cell
account during elevated temperature operation. In the case of failure.
non-optimized Nafion® content a mass transport problem was
observed which resulted in lower performance [97].
Acknowledgements
7. Conclusions
The authors are grateful to Dr. Rui (Jim) Qiao and Dr. Pier-
This paper provided a brief overview of the main parameters luigi Pisu of the Department of Mechanical Engineering at
influencing performance and durability of PEM fuel cells. Vari- Clemson University for their positive feedback. The authors
ous interacting mechanisms contribute to loss of performance would also like to thank Dr. Steve Creager from the Chemistry
and negatively impact fuel cell durability. It is important to Department at Clemson for his input.
understand these processes and interactions well to take the nec-
essary steps to extend the life of next-generation fuel cells. We References
hope that this paper is a step towards understanding the vast but
spread-out work that has been done and reported in the litera- [1] G. Escobedo, M. Gummalla, R.B. Moore, DOE Hydrogen Program, FY
ture on fuel cell performance and durability; and help identify 2006 Annual Progress Report, 2006, pp. 706–711.
critical directions for further research. A summary of some of [2] R. Borup, M. Inbody, J. Davey, D. Wood, F. Garzon, J. Tafoya, J. Xie,
the key points is given below. S. Pacheco, DOE Hydrogen Program, FY 2004 Annual Progress Report,
2004, pp. 579–584.
Cycled operation of the fuel cell is much more detrimental to [3] Q. Yan, H. Toghiani, Y.-W. Lee, K. Liang, H. Causey, J. Power Sources
lifetime than operation at constant load. Poor water management 160 (2006) 1242–1250.
can cause flooding, while flooding reduces the electrochemical [4] http://www.mercedes-benz.de/content/germany/mpc/mpc germany
active surface area (ECSA) as well as corrosion or degradations website/de/home mpc/buses/home/bus world/whats new/Bus news
of the electrodes, the catalyst layers, gas diffusion media and 2005/World record for fuel cell buses.html, 2005.
[5] G. Hinds, NPL Report DEPC-MPE 002, 2004, pp. 25–42.
the membrane. The loss of ECSA results in lower activity of the [6] A. Faghri, Z. Guo, Int. J. Heat Mass Transfer 48 (2005) 3891–3920.
catalyst and hence in lower power output. It does not lead to [7] R.L. Borup, J.R. Davey, F.H. Garzon, D.L. Wood, M.A. Inbody, J. Power
failure of the cell. Corrosion products can contaminate the cell; Sources 163 (2006) 76–81.
however this does not seem to have a big impact on fuel cell [8] R. Mukundan, Y.S. Kim, F. Garzon, B. Pivovar, ECS Trans. 1 (2006)
performance and life. CO-poisoning can hinder or slow down 403–413.
[9] C. Sishtla, G. Koncar, R. Platon, S. Gamburzev, J. Power Sources 71 (1998)
the reaction at the anode leading to lower cell potentials, but it 249.
is a reversible process. Membrane contamination mainly results [10] K. Washington, Proceedings Fuel Cell Seminar 2000, Portland, U.S.A.,
in lower performance. 2000, p. 468.
Robustness of the membrane is critical to durability of PEM [11] U. Pasaogullari, C.Y. Wang, J. Electrochem. Soc. 152 (2) (2005)
fuel cells. In particular carbon corrosion and change of the A380–A390.
[12] H. Maeda, A. Yoshimura, H. Fukumoto, Proceedings Fuel Cell Seminar
chemical structure of the membrane can seriously compromise 2000, Portland, U.S.A., p. 379.
durability. During membrane corrosion, pinholes may be formed [13] M. Fowler, J.C. Amphlett, R.F. Mann, B.A. Peppley, P.R. Roberge, J. New
leading to gas crossover through the membrane. Also dehydra- Mat. Electrochem. Syst. 5 (2002) 255.
W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14 13
[14] E. Endoh, S. Terazono, H. Widjaja, Abstract 89, Electrochem. Soc. Meeting [50] T. Patterson, Effect of potential cycling on loss of electrochemical surface
Abstracts, Salt Lake City, U.S.A., 2002. area of platinum catalyst in polymer electrolyte membrane fuel cell, in:
[15] S.D. Knights, K.M. Colbow, J. St-Pierre, D.P. Wilkinson, J. Power Sources AICHE National Conference, New Orleans, LA, 2002.
127 (2004) 127–134. [51] M.S. Wilson, F.H. Garzon, K.E. Sickafus, S. Gottesfeld, J. Electrochem.
[16] J. Scholta, N. Berg, P. Wilde, L. Jorissen, J. Garche, J. Power Sources 127 Soc. 140 (1993) 2872–2877.
(2004) 206–212. [52] P. Ascarelli, V. Contini, R. Giorgi, J. Appl. Phys. 91 (2002).
[17] X. Cheng, L. Chen, C. Peng, Z. Chen, Y. Zhang, Q. Fan, J. Electrochem. [53] G. Escobedo, K. Schwiebert, K. Raiford, G. Nagarajan, F. Principe, Pro-
Soc. 151 (2004) A48–A52. ceedings of the Fuel Cells Durability, first ed., Washington, DC, 2006, pp.
[18] X. Wang, D. Myers, R. Kumar, Proceedings of the Fuel Cells Durability, 83–100.
first ed., Washington, DC, 2006, pp. 151–162. [54] W. Li, M. Ruthkosky, M. Balogh, R. Makharia, S. Oh, Proceedings of the
[19] V. Lightner, DOE Hydrogen Program, Record 5036, 2006. Fuel Cells Durability, first ed., Washington, DC, 2006, pp. 101–114.
[20] V. Lightner, DOE Hydrogen Program, Backup Ref. 5036a, 2006. [55] A. Horky, K. Beverage, O. Polevaya, Y. Shi, Proceedings of the Fuel Cells
[21] R. Borup, D. Wood, J. Davey, P. Welch, F. Garzon, DOE Hydrogen Review, Durability, first ed., Washington DC, 2006, pp. 133–150.
Presentation, 2006. [56] M.S. Wilson, F.H. Garzon, K.E. Sickafus, S. Gottesfeld, J. Electrochem.
[22] R. Borup, D. Wood, J. Davey, P. Welch, F. Garzon, DOE Hydrogen Review, Soc. 140 (1993) 2872.
FY 2006 Annual Progress Report, 2006. [57] C. Boyer, S. Gamburzev, O. Velev, S. Srinivasan, A.J. Appleby, Elec-
[23] S.J.C. Cleghorn, D.K. Mayfield, D.A. Moore, J.C. Moore, G. Rusch, T.W. trochim. Acta 43 (1998) 3703–3709.
Sherman, N.T. Sisofo, U. Beuscher, J. Power Sources 158 (2006) 446–454. [58] K.L. More, K.S. Reeves, 2005 DOE Hydrogen Program Review, Arlington,
[24] J. Le Canut, R.M. Abouatallah, D.A. Harrington, J. Electrochem. Soc. 153 VA, 2005.
(2006) A857–A864. [59] Z. Qi, Proceedings of the Fuel Cells Durability, first ed., Washington, DC,
[25] I. Manke, N. Kardjilov, A. Haibel, C. Hartnig, M. Strobl, A. Rack, A. 2006, pp. 163–190.
Hilger, J. Scholta, W. Lehnert, W. Treimer, S. Zabler, J. Banhart, DGZfP- [60] K. Duff, Stainless Steel alloys for Polymer Electrolyte Membrane (PEM)
Berichtsband 94-CD, 2005. Fuel Cells, presentation, 2005, http://www.chemistry.oregonstate.edu/
[26] T. van Nguyen, M.W. Knobbe, J. Power Sources 114 (2003) 70–79. courses/ch407h/CH407H%20project%20keegan%20fuel%20cell
[27] T.V. Nguyen, R.E. White, J. Electrochem. Soc. 140 (1993) 2178–2186. %20steels.ppt.
[28] W. He, G. Lin, T.V. Nguyen, AICHE J. 49 (2003) 3221–3228. [61] J. Larminie, A. Dicks, Fuel Cell Systems Explained, John Wiley & Sons
[29] D.A. McKay, W.T. Ott, A.G. Stefanopoulou, Modeling, parameter identi- Ltd., Chichester, England, 2003.
fication, and validation of reactant water dynamics for a fuel cell stack, in: [62] D.E. Curtin, R.D. Lousenberg, T.J. Henry, P.C. Tangeman, M.E. Tisack, J.
Proceedings of the IMECE, ASME Int. Mech. Eng. Congress & Exposition, Power Sources 131 (2004) 41–48.
2005. [63] F. Finsterwalder, M. Quintus, M. Schaloske, T. Guth, G. Frank, The Elec-
[30] B.A. Dipierno, M.H. Fronk, US Patent 6,103,409 (2000). trochem. Soc. 210th Meeting, Abstract 0485, Cancun, Mexico, 2006.
[31] A.Z. Weber, R.M. Darling, J. Newman, J. Electrochem. Soc. 151 (10) [64] V.O. Mittal, H.R. Kunz, J.M. Fenton, Abstract 0448, The Electrochem.
(2004) A1715–A1727. Soc. 210th Meeting, Cancun, Mexico, 2006.
[32] A. Turhan, K. Heller, J.S. Brenizer, M.M. Mench, J. Power Sources 160 [65] G. Hübner, E. Roduner, J. Mater. Chem. 9 (1999) 409–418.
(2006) 1195–1203. [66] S. Kundu, M.W. Fowler, L.C. Simon, S. Grot, J. Power Sources 157 (2006)
[33] J.J. Kowal, A. Turhan, K. Heller, J.S. Brenizer, M.M. Mench, J. Elec- 650–656.
trochem. Soc. 153 (2006) A1971–A1978. [67] V. Ramani, H.R. Kunz, J.M. Fenton, The Electrochem. Soc. Interf. 17–19
[34] J. St-Pierre, D.P. Wilkinson, S. Knights, M. Bos, J. New Mat. Electrochem. (2004) 45.
Syst. 3 (2000) 99–106. [68] R. Baldwin, M. Pham, A. Leonida, J. McElroy, T. Nalette, J. Power Sources
[35] Y. Wang, C.-Y. Wang, Electrochim. Acta 51 (2006) 3924–3933. 29 (1990) 399–412.
[36] S. Ge, C.-Y. Wang, J. Electrochem. Soc. 154 (2007) B998–B1005. [69] W. Liu, K. Ruth, G. Rusch, J. New Mat. Electrochem. Syst. 4 (2001)
[37] Y. Sone, P. Ekdunge, D. Simonsson, J. Electrochem. Soc. 143 (1996) 1254. 227–231.
[38] F.N. Büchi, S. Srinivasan, J. Electrochem. Soc. 144 (1997) 2767. [70] Q. Li, R. He, J.O. Jensen, N.J. Bjerrum, J. Am. Chem. Soc., Chem. Mater.
[39] G. Li, P.G. Pickup, Electrochem. Solid-State Lett. 9 (2006) A249–A251. 15 (2003) 4896–4915.
[40] X. Huang, R. Solasi, Y. Zou, M. Feshler, K. Reifsnider, D. Condit, S. [71] B. Kienitz, H. Baskaran, B. Pivovar, T. Zawodzinski Jr., ECS Trans. 11
Burlatsky, T. Madden, J. Polym. Sci. Part B Polym. Phys. 44 (16) (2006) (2007) 777–788.
2346–2357. [72] X. Cheng, Z. Shi, N. Glass, L. Zhang, J. Zhang, D. Song, Z.-S. Liu, H.
[41] J. Yu, T. Matsuura, Y. Yoshikawa, M.N. Islam, M. Hori, Electrochem. Wang, J. Shen, J. Power Sources 165 (2007) 739–756.
Solid-State Lett. 8 (2005) A156–A158. [73] A. Rodrigues, J.C. Amphlett, R.F. Mann, B.A. Peppley, P.R. Roberge,
[42] N.E. Vanderborgh, J.R. Huff, J. Hedstrom, IEEE CH2781-3/89/0000-163 Proceedings of the 32nd Intersociety Energy Conversion Engineering Con-
(1989) 1637–1640. ference, IECEC-97 (1997) 768–773.
[43] T. Nguyen, J. Hedstrom, N. Vanderborgh, in: R.E. White, A.J. Appleby [74] D.D. Penta, K. Bencherif, Q. Zhang, M. Sorine, Proceedings of the 2006
(Eds.), The Electrochem. Soc. Softbound Proceedings Series PV 89–14 IEEE Int. Conference on Control Applications, Munich, Germany, 2006.
(1989) 39. [75] J.J. Baschuk, X. Li, Int. J. Energy Res. 25 (2001) 695–713.
[44] T.E. Springer, J. Zawodzinski, S. Gottesfeld, in: R.E. White, M.W. Ver- [76] Z. Liu, L. Yang, Z. Maoa, W. Zhuge, Y. Zhang, L. Wang, J. Power Sources
brugge, J.F. Stockel (Eds.), The Electrochem. Soc. Softbound Proceedings 157 (2006) 166–176.
Series PV 91–10 (1991) 209. [77] D.P. Wilkinson, J. St-Pierre, PEM fuel cell durability, in: W. Vielstich,
[45] P.J. Schutz, in: R.E. White, A.J. Appleby (Eds.), The Electrochem. Soc. H. Gasteiger, A. Lamm (Eds.), Handbook of Fuel Cells – Fundamentals,
Softbound Proceedings Series PV 89–14 (1989) 87. Technology and Applications, Fuel Cell Technology and Applications, vol.
[46] R.L. Borup, J.R. Davey, F.H. Garzon, D.L. Wood, M.A. Inbody, Proceed- 3, John Wiley and Sons, 2003.
ings of the Fuel Cells Durability, first ed., Washington, DC, 2006, pp. [78] T.W. Patterson, R.M. Darling, Electrochem. Solid-State Lett. 9 (2006)
21–42. A183–A185.
[47] P. Meyers, R.M.S. Darling, The Electrochem. Soc. Meeting Abstracts, [79] M. Saito, K. Hayamizu, T. Okada, J. Phys. Chem. 109 (2005) 3112–
Abstract 1212, Paris, France, 2003. 3119.
[48] S. Motupally, T.D. Jarvi, The Electrochem. Soc. 208th Meeting, Abstract, [80] M. Sundaresan, R.M. Moore, J. Power Sources 145 (2005) 534–545.
Los Angeles, CA, 2005. [81] J. Zhang, Z. Xie, J. Zhang, Y. Tang, C. Song, T. Navessin, Z. Shi, D. Song,
[49] C. Paik, T. Skiba, V. Mittal, S. Motupally, T. D. Jarvi, The Electrochem. H. Wang, D.P. Wilkinson, Z.S. Liu, S. Holdcroft, J. Power Sources 160
Soc. 207th Meeting, Abstract 771, Quebec City, Canada, 2005. (2006) 872–891.
14 W. Schmittinger, A. Vahidi / Journal of Power Sources 180 (2008) 1–14
[82] E. Cho, J. Ko, H.Y. Ha, S. Hong, K. Lee, T. Lim, I. Oh, J. Electrochem. [91] R.K. Ahluwalia, X. Wang, J. Power Sources 162 (2006) 502–512.
Soc. 150 (2003) A1667–A1670. [92] L. Mao, C.-Y. Wang, J. Electrochem. Soc. 154 (2007) B139–B146.
[83] A. Pesaran, G.-H. Kim, J. Gonder, Proceedings of the Fuel Cells Durability, [93] S. Ge, C.Y. Wang, Electrochem. Solid-State Lett. 9 (11) (2006)
first ed., Washington, DC, 2006, pp. 205–227. A499–A503.
[84] S. He, M.M. Mench, J. Electrochem. Soc. 153 (9) (2006) A1724–A2173. [94] K. Tajiri, Y. Tabuchi, C.Y. Wang, J. Electrochem. Soc. 154 (2) (2007)
[85] S. Kim, M.M. Mench, J. Power Sources 174 (2007) 206–220. B147–B152.
[86] F. Kagami, F. Ogawa, Y. Hishinuma, T. Ghikahisa, Fuel Cell Seminar 2002, [95] M. Khandelwal, S. Lee, M.M. Mench, J. Power Sources 172 (2007)
Palm Springs, USA, 2002, p. 239. 816–830.
[87] C.H. Lee, H.B. Park, Y.M. Lee, R.D. Lee, Ind. Eng. Chem. Res. 44 (2005) [96] A.S. Arico, A. Stassi, E. Modica, R. Ornelas, I. Gatto, E. Passalacqua, V.
7617–7626, Sources 165 (2007) 739–756. Antonucci, ECS Trans. 3 (2006) 765–774.
[88] R.C. McDonald, C.K. Mittelsteadt, E.L. Thompson, Fuel Cells 4 (3) (2004) [97] Y. Song, H. Xu, Y. Wei, H.R. Kunz, L.J. Bonville, J.M. Fenton, J. Power
208–213. Sources 154 (2006) 138–144.
[89] R. Mukundan, Y.S. Kim, T. Rockward, J.R. Davey, B. Pivovar, D.S. Hussey, [98] W. Bi, T.F. Fuller, Abstract 395, The Electrochem. Soc., 212th ECS Meet-
D.L. Jacobson, M. Arif, R. Borup, ECS Trans. 1 (2007) 543–552. ing, Washington, DC, 2007.
[90] Y. Hishinuma, T. Chikahisa, F. Kagami, T. Ogawa, JSME Int. J. B 47 (2004)
235–241.