You are on page 1of 20

DOI: 10.1002/fuce.

200700053

Review: Durability and Degradation

REVIEW
Issues of PEM Fuel Cell Components
F. A. de Bruijn1,2*, V. A. T. Dam1, and G. J. M. Janssen2
1
Schuit Institute of Catalysis, Eindhoven University of Technology, PO Box 513, 5600 MB Eindhoven, The Netherlands
2
Energy Research Centre of the Netherlands (ECN), PO Box 1, 1755 ZG Petten, The Netherlands

Received September 26, 2007; accepted December 28, 2007

Abstract
Besides cost reduction, durability is the most important cation or humidification cycling, temperatures of 90 °C or
issue to be solved before commercialisation of PEM Fuel higher and fuel starvation. This review paper aims at asses-
Cells can be successful. For a fuel cell operating under con- sing the degradation mechanisms of membranes, electrodes,
stant load conditions, at a relative humidity close to 100% bipolar plates and seals. By collecting long-term experiments
and at a temperature of maximum 75 °C, using optimal as well, the relative importance of these degradation
stack and flow design, the voltage degradation can be as low mechanisms and the operating conditions become apparent.
as 1–2 lV·h. However, the degradation rates can increase by
orders of magnitude when conditions include some of the Keywords: Components, Degradation, Durability, Electrodes
following, i.e. load cycling, start–stop cycles, low humidifi- Bipolar Plates, Lifetime, Membranes, PEFC, PEMFC, Seals

1 Introduction This review starts with the identification and categorisa-


tion of long-term experiments, focusing on observed degrada-
Now that power density specifications of PEM fuel cells tion rates and materials’ changes, followed by a detailed
for automotive and stationary applications are close to being description of degradation mechanisms and their measurable
met, the focus of PEMFC R&D is shifting to the other require- effects, both on fuel cell performance and on in situ or post-
ments for successful introduction of the PEMFC, i.e. cost and test diagnostics. The aim is to assess which mechanisms affect
durability. Long-term operation of the PEMFC has been pro- the lifetime of membranes, electrodes, bipolar plates and
ven, but mostly for fuel cells containing high loadings of seals, how degradation is enhanced or mitigated by opera-
noble metals, and under relatively ideal conditions. However, tional conditions and which tools are available to identify
especially for automotive application, PEMFCs need to oper- and quantify the degradation.
ate in a wide range of conditions, which may negatively affect The emphasis in this study is on PEMFCs operating in the
durability and lifetime. The issues of cost and durability are temperature range up to 120 °C and containing perfluori-
closely related since reduction of cost is associated with the nated, hydrated polymer membranes and electrodes based
use of less or cheaper materials, e.g. reduction of Pt content upon Pt/C catalysts. Reference to other materials is made in
or use of thinner perfluorinated, partially fluorinated or non- passing. This review does not cover the influence of contami-
fluorinated membranes. nants present in hydrogen and air. A recent review paper by
Insight into the mechanisms that lead to irreversible per- Cheng et al. covers this extensively [1].
formance loss and the relation between these mechanisms
and the operating conditions, the changes in materials’ prop-
erties that can be observed and the impact on fuel cell perfor- 2 Long-Term Experiments–Observed
mance is needed, before either fuel cell materials can be opti- Degradation Rates
mised further with respect to their durability, or systems’
controls and operating procedures can be optimised to pre- Long-term experiments can give a good indication of the
vent the most detrimental conditions. This has motivated sev- severity of degradation mechanisms and their relative contri-
eral recent studies on PEMFC degradation using long-term
testing as well as rapid ageing testing. –
[*] Corresponding author, debruijn@ecn.nl

FUEL CELLS 08, 2008, No. 1, 3–22 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 3
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

bution to performance loss under various conditions. Table 1 experiments, but now organised in increasing order of the
REVIEW

compiles key experiments longer than 1,000 h, or otherwise voltage decay rates. The compilations are meant to give a
relevant for this study, with all reported observations of qualitative understanding on the possible relation between
changes in component characteristics. Table 2 gives the same operating conditions and voltage decay rates, because the

Table 1 Long-term PEMFC experiments reported in literature, with materials changes observations.
Company/product Test period Anode/fuel Cathode/feed Membrane j / A cm–2 Voltage decay Observed Reference
(h )/Temp. at V/V ratea) / lVh–1 changes
Ballard 8 cell stack 6,000 X H2 100% RH Air 100% RH 1.076 at 0.62 –1 – [4]
Mark 513 2000 75 °C Reformate + air Air 0% RH 1,000 at 0.43 –24 a)

2,000 Y bleed 100% RH


Ballard 8 cell stack 11,000 X PtRh; 4 mg cm–2 Pt on Pt (4); mg cm–2 Pt Dow 0.538 at 0.82 –1.4 a),b)
[5]
Mark 513 2002 Toray on Toray pure O2 0.861 at 0.78 –1.3
pure H2
LANL 2005 3,400 Y Pt/C; 0.2 mg cm–2 on Pt/C; 0.2 mg cm–2 N 112 Constant V –33 a),c)
[6]
80 °C ETEK ELAT on ETEK ELAT 0.55 at 0.62
Pure H2, RH = 255% air, RH = 100%
d),e),f),-
Nuvera CGS 2-cell >9,000 X PtRu/C Pt/C 36 lm ? at 0.7 –6 [7]
stack 500 cm2 75 °C 75% H2/25% N2 air
g),h),i),j),k)

Daido Institute of 2,700 X Pt/C 0.5 mg cm–2 on Pt/C 0.5 mg cm–2 N112 0.3 at 0.66 –2.5 (0–1,900) d),l)
[8]
Technology 2005 75 °C Toray, H2 80% RH on Toray, air 80% –50 (1,900–2,700)
RH
Commercial MEA 1,000 X, Y Pt/C 0.5 mg cm–2 on Pt/C 0.5 mg cm–2 N112 Cyclic, MEA failure m),d),n), )
° [9]
Fuelcellstore.com 80 °C ETEK ELAT, H2 100% on ETEK ELAT, 0–1.06 A cm–2
RH air 100% RH 1.06 at 0.66 –180 a),d),n), )
°
Gore 5510 2,000 X, Y Pt/C 0.4 mg cm–2, H2 Pt/C 0.4 mg cm–2, Gore 25 lm 0.2 at 0.77 –25 d),e)
[10]
80 °C 100% RH air 100% RH OCV –20 d),e)

Gore 5621 26,000 X PtRu/C 0.45 mg cm–2 Pt/C 0.6 mg cm–2 Gore 35 lm 0.8 at 0.60 –6.4 a),b),d),i),l)
[3]
70 °C on Carbel on Carbel, air
Reformate + air 100% 100% RH
RH
H2 100% RH Air 100% RH 0.8 at 0.62
a),q)
Osaka Gas 36,000 X PtRu/C Pt/C Cyclic, –2.5 Under H2 [11]
70 °C Reformate, 10 ppm air 100% RH 0.15–0.30 A cm–2 –35 under
CO 100% RH reformate
10 ppm CO
d),j)
Asahi Glass 4,000 Y H2 50% RH Air, 50% RH Asahi Glass 0.2 at 0.75 –75 [12]
120 °C NPC
d),g),r)
Toray/Daido Insti- 900 H2 26% RH Air, 26% RH Nafion 0.3 at 0.7 –29 [13]
tute of Technology cyclic –218
0–0.3 A cm–2
Gore 5510 1,000 X H2 20% RH Air, 20% RH Gore 25 lm 0.5 at 0.59 –95 [14]
55 °C stoich = 4! cyclic, –45
0–0.7 A cm–2
Mitsubishi Electric 6,000 X Reformate, 10 ppm CO Air, 74% RH 0.25 at 0.7 –1.5 [2]
75 °C 57% RH

X, endurance test under normal conditions; Y, endurance test under accelerated conditions.
a)
May be recalculated from current density decay using polarisation curve.
b)
Loss of water removal efficiency.
c)
Deterioration of seals.
d)
Increase in cell resistance (unspecified).
e)
Loss of Pt surface area at cathode.
f)
Pt detected in membrane.
g)
Ru detected in anode gas diffusion layer.
h)
Carbon corrosion at anode.
i)
Carbon corrosion at cathode.
j)
Thinning of membrane.
k)
Thinning of electrodes.
l)
Loss of fluorine and sulphur or other components from membrane.
m)
Increase in hydrogen cross-over through membrane.
n)
Macroscopic holes in membrane.
)
° Loss of Pt surface area at anode.
p)
Fluoride ions in outlet water.
q)
Pt detected in cathode ionomer.
r)
Decrease in CO tolerance.
s)
Degradation of ionomer in electrodes.

4 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

Table 2 Voltage decay rates in long-term PEMFC experiments, in increasing order of decay rate.

REVIEW
Voltage decay Cell T RH anode RH cathode V/V J / mA cm–2 OCV in cycle Cyclic/SS Membrane Reference
rate / lVh–1 % % thickness / lm
1 75 100 100 0.62 1,076 No ss 125 [4]
1.4 75 100 100 0.82 538 No ss 125 [5]
1.3 75 100 100 0.78 861 No ss 125 [5]
1.5 75 57 74 0.7 259 No ss ?? [2]
2.5 70 100 100 Cyclic 150–300 No Cyclic ?? [11]
6 75 100 100 0.7 Not given No ss 36 [7]
6.4 70 100 100 0.62 800 No ss 35 [3]
20 80 100 100 OCV 0 Yes ss 25 [10]
25 80 100 100 0.77 200 No ss 25 [10]
28 80 26 26 0.7 300 No ss ?? [15]
29 80 26 26 0.7 300 No ss ?? [13]
33 80 255 100 0.62 550 No ss 50 [6]
2.5–50 75 80 80 0.66 300 No ss 50 [8]
45 55 20 20 0.6–0.35 0–700 Yes Cyclic 35 [14]
75 120 50 50 0.75 200 No ss ?? [16]
75–114 80 26 26 0.7 0.01–300 Yes Cyclic ?? [15]
95 55 20 20 0.59 500 No ss 35 [14]
180 80 100 100 0.66 1,060 Yes ss 50 [9]
210 80 26 26 0.7 0.01–300 No Cyclic ?? [13]
Failure 80 100 100 Cyclic Cyclic Yes Cyclic 50 [9]

tests are difficult to compare because of differences in e.g. out and flooding can have a detrimental effect. During the
materials, flow fields, start-up procedures. Note that the volt- 26,000 h test at Gore [3], a voltage loss of 110 mV was ob-
age decay rates in Tables 1 and 2 comprise irreversible as served at a current density of 800 mA cm–2. It should be
well as reversible losses. noted that superposed on this voltage loss, a saw-tooth was
Table 2 shows that for cells operated in a steady state, observed with a voltage decay rate of more than 40 lVh–1,
without OCV excursions, temperature 70–75 °C and at 100% which is explained by water management issues, poison for-
relative humidity of the inlet gases the overall voltage decay mation and platinum oxide formation.
rate is generally in an acceptable range, i.e. <6 lVh–1. The In this review, the focus is on irreversible degradation of
voltage decay rate can increase sharply when one or more of membrane, electrodes, bipolar plates and seals. The contribu-
the following conditions are applied: temperature above tion of each mechanism to overall degradation as reported in
75 °C, gases not fully humidified, cycles of load or potential Tables 1 and 2 is discussed.
(especially including OCV). These conditions seem to qualify
as non-ideal. The only exception seems to be a recent result
by Mitsubishi Electric [2]. Here the design of the flow pattern 3 Membrane Durability
and the MEA was modified to prevent the evaporation of
water in dry areas such as the air inlet region.
A relation with the material properties is not so clear from 3.1 Membrane Function and Requirements
these tables. Table 2 is headed by tests with a 125 lm thick
The electrolytic membrane serves as the conductor for pro-
membrane, but as they all refer to the same membrane, the
tons from anode to cathode, as a gas separator and as an elec-
thickness need not be the decisive property.
tronic insulator. During the lifetime of the fuel cell, these func-
Both under ideal and non-ideal conditions, many changes
tions must remain unimpeded, meaning that the proton
in materials and fuel cells characteristics were observed after
conductivity should be in the range of 0.1 S cm–1, its gas per-
these tests. A large fraction of these observations are only
meability should be minimal i.e. <10–12 mol H2 cm–1 s–1 kPa–1
indicative, and more detailed investigations are needed to
and <10–11 mol O2 cm–1 s–1 kPa–1 [17]. For the electronic resis-
further elucidate the degradation phenomena, especially by
tance, no clear specification exists, as short circuiting is gener-
applying suitable characterisation techniques. However, it is
ally not an issue, except for manufacturing failures.
apparent from this overview of long-term experiments that
loss of water removal efficiency and loss of platinum surface
area at the cathode are the most frequently observed degrada- 3.2 Mechanisms for Membrane Degradation
tions.
3.2.1 Chemical Attack by Radicals
In durability studies it must always be remembered that
part of the decay may be reversible. This is especially true in Chemical degradation is a major cause of failure of PFSA
the case of under or oversaturation of the gases. Both drying membrane [18–20]. It is thought that hydroxyl (OH) or per-

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 5
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

oxy (OOH) radicals attack polymer end groups that still con- Attack by OOH and OH also has an effect on partially or
REVIEW

tain residual terminal H-groups. Characterisation by XPS [21] non-fluorinated membranes, as was shown in Fenton’s test. It
revealed that during fuel cell operation the interactions was noticed that non-sulphonated aromatic membranes like
between carbon, fluorine and oxygen are changing. An exam- PEEK are unaffected but severe degradation was observed
ple given by Curtin et al. [22] is the attack of hydroxyradicals for the sulphonated compound. Also in this case has the use
on carboxylic end groups: of Fenton’s test been questioned, i.e. stable performance for
4,000 h has been obtained of membranes that did not survive
Rf–CF2COOH + OH → Rf–CF2 + CO2 + H2O (1) Fenton’s test [34]. It must be noted that gas cross-over rates
are usually lower in hydrocarbon membranes than in PFSA
Rf–CF2 + OH → Rf–CF2OH → Rf–COF + HF (2) membranes [35].
Hydrogenation of CF2 groups in the presence of H2 has
Rf–COF + H2O → Rf–COOH + HF (3) also been proposed as the source of degradation, but in the
light of the above observations this mechanism is less likely
In the last few years chemically modified membranes have to contribute substantially [19, 21].
been developed by DuPont, which show much lower fluoride Non-fluorinated compounds like PEEK are susceptible to
release rates and longer accelerated lifetimes [22]. In these the loss of the sulphonic acid group by hydrolysis at wet con-
modified polymers, the number of reactive end groups has ditions [34]. Also for PFSA have sulphonic acid groups been
been reduced by an alternative synthesis route. reported as degradation products [18]. However, it was
For a considerable time the hypothesis has been that the reported by LaConti that the equivalent weight was not
radicals are formed either at the cathode during the oxygen affected, i.e. polymer units with their sulphonic groups were
reduction reaction, e.g. from H2O2 that is formed in small dissolved, without desulphonation. Recent results at DuPont
quantities during the ORR [23], or from a decomposition of confirm the stability of the sulphonic acid containing side
H2O2 formed at the anode as a result of O2 crossing over from group of Nafion. At temperatures higher than 200 °C thermal
cathode to anode [18]. For the radicals to form from H2O2, the degradation can set in, starting with desulphonation [22].
presence of oxidizable metal ions such as Fe2+ is required. It
was thought that contaminating ions would induce radical
3.2.2 Mechanical Failure
formation. This has led to the use of Fenton’s test where Fe2+,
H2O2 and polymer are mixed and the product is analysed. Although the mechanical properties of most often used
Recently, however, evidence has been presented that the Nafion membranes such as tensile strength, tear strength and
incentive for the chemical attack can also be provided by a puncture resistance are quite satisfactory, the use of very thin
species (likely a radical as mentioned above) that is formed in membranes (25 lm) requires a well-controlled MEA manu-
a reaction of molecular hydrogen and oxygen in the presence facturing. Pinholes and foreign materials introduced during
of Pt, without necessary formation of intermediate H2O2 MEA manufacturing can initiate propagating cracks leading
[24–30]. In fact, in an operating fuel cell the so-called direct to strongly reduced lifetimes [18].
radical pathway is more important than the H2O2 pathway, Non-uniform contact pressure [18], high differential initial
as the activity of Pt/C as a reagent in Fenton’s reaction is lim- gas pressure over the membrane, punctures as well as fatigue
ited [9, 28]. According to direct radical pathway, favourable from stresses occurring during temperature and humidity
conditions for degradation are not only at the cathode (cross- cycling [36, 37] can lead to mechanical failure of the mem-
over from H2), anode (cross-over of O2), but also in the mem- brane. Tang et al. found that relative humidity cycling of
brane when Pt particles deposit there as a result of electrode NR111 membranes from water-soaked to 25% RH, as well as
degradation [29, 31–33]. It must be noted that LaConti et al. temperature cycling at 100% RH do lead to stresses higher
already reported the degradation of polystyrene sulphonic than the limit of 1.5 MPa that was found to lead to thinning
acids in a mixture of H2, O2 and Pt [18]. On the basis of these of the membrane and increased hydrogen cross-over and irre-
results replacements of Fenton’s test have been suggested versible elongation [38].
[24, 25]. It has been reported that reinforcement of the membrane
The exact mechanism of the direct radical pathway has not with e.g. a porous polyethylene or PTFE [37, 39] material
been established yet. Studies on chemical degradation have enhances the dimensional stability and reduces the shrinkage
shown that the conditions leading to increased radical attack stress in the membrane during drying. Reinforced mem-
in PFSAs include higher temperature (especially above branes were shown to be more stable, i.e. showed a lower
90 °C), low humidification, high gas pressure, use of pure decrease in the OCV [36] and longer lifetime at elevated tem-
hydrogen and pure oxygen and high cell voltages [20]. Some perature and low humidity [39].
of these conditions are consistent with a higher concentration Mechanical failure can be enhanced by chemical degrada-
of H2 and O2. The role of relative humidity, potential and also tion as shown by Cleghorn and Kolde [40]. The Primea 57 ser-
of membrane thickness is still a subject of study [19, 24, 26, ies failed after 12,000 wet–dry cycles (300 h) in inert N2 atmo-
27, 29]. Recent results suggest that the state of the catalyst sur- sphere, and after 2,000 cycles (50 h) in fuel cell mode, due to
face may play an important role [27]. sharp increase in hydrogen cross-over. New reinforced struc-

6 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

REVIEW
tures lasted more than 64,000 cycles (1,600 h) in inert atmo- toughness of membranes seemed to decrease somewhat as
sphere, the test not being ended by failure, and showed a ten- well as the permeability of the membrane for oxygen. It was
fold increase in lifetime in MEA. suggested that a rearrangement of the sulphonic acid groups
in the ionomer might have occurred.
3.2.3 Contamination by Ionic Species

Metal bipolar plates, humidifiers, tubing materials, air, can 3.3 Characterisation of Membrane Degradation
be sources for contaminant ions. The sulphonate sites have a
3.3.1 Fluoride Release Rates
stronger affinity to metal ions, except Li+, than for protons,
leading to exchange of the protons by the metal ion when The rate of release of fluoride ions in the product water is
present. There is agreement on the effect of such contamina- correlated to the rate of the chemical degradation of fluori-
tion, being direct and indirect [18, 41]. The direct effect is a nated membranes [see reactions (1)–(3)]. The fluoride analysis
reduction of the proton conductivity that is proportional to is generally done off-line by ion-chromatography or an ion
the decrease in proton concentration. Note that metal ions selective electrode [20, 39, 48, 49] but recently also a method
will contribute to the ionic conductivity, but not to the proton based upon a fluoride ion selective electrode has been used
conductivity that determines the membrane ohmic loss in a for continuous monitoring [50].
fuel cell. The indirect effect stems from the lower hydrophili- It was already established at the end of the 1980s at Hamil-
city of metal ions in comparison to protons, which can result ton Standard (part of UTC) that PEM fuel cell lifetime is
in partial drying out of membrane, thus, also reducing the directly related to the fluoride release rate [51], as was also
proton conductivity. It has also been suggested that metal confirmed by recent results from DuPont [36]. In a recent
ions poison the ORR catalyst [41] hence increasing the rate of study the impact of the fluoride release rate was related to
hydrogen peroxide and radical formation, resulting in chemi- the initial fluorine content of a membrane with 25 lm thick-
cal degradation [18]. nesses, which was estimated to be 3.8 mg F cm–2 (2 g cm–3,
75% F) [20]. This implies that with an average release rate of
3.2.4 Change of Nafion Morphology 0.01 lg F cm–2 h–1, which was measured for membranes at
near-ideal conditions [16, 20], 2% of the F is lost after 6,000 h.
Loss of water from the membrane is well known to lead to
A degradation rate of 3 lg F cm–2 h–1, as was obtained under
an almost immediate loss of fuel cell performance, due to the
harsh conditions, would lead to complete failure within
strong relation between membrane conductivity and the
1,200 h. In this study, it was also remarked that the pH of the
water content. This relation is generally believed to be revers-
effluent correlated very well with the pF, i.e. decrease in pH
ible in nature, albeit that rehydration of Nafion membranes to
may also indicate increased membrane degradation. How-
their original state can be difficult, due to the shrinkage of
ever, Healy et al. also report that data of the fluoride release
hydrophilic domains during drying.
rate can show considerable scatter [20].
Characterisation by XRD [21] revealed that during fuel cell
operation, the Nafion crystallinity changes, especially at rela-
3.3.2 Hydrogen Cross-Over Rates
tively dry conditions. It is believed that a high crystallinity
corresponds to open ion-channels, and dehydrated, collapsed Both chemical and mechanical degradation can lead to an
channels correspond to a decrease in crystallinity [21]. increase in gas cross-over. The hydrogen cross-over can be
Whether this is a reversible effect or not is not clear, as no measured electrochemically in situ [52]. The fuel cell cathode
rewetting experiment of the fuel cell membrane has been is purged with nitrogen, and set at a voltage of about 0.4 V
reported. Recently, it was suggested on the basis of thermo- higher than the anode. Hydrogen that diffuses to the fuel cell
mechanical analysis that Pt2+ ions dissolved in the membrane cathode is oxidised and the protons produced are transported
may induce physical cross-linking, resulting in stiffer mem- to the fuel cell anode, where they can be reduced to hydro-
branes [42]. gen. The current is directly proportional to the cross-over rate
The effect of freezing of water on membrane properties of hydrogen through the membrane.
has been investigated in several institutes [43–47]. The results The hydrogen cross-over is proportional to the hydrogen
indicate that about 90% of the water can freeze, but that some partial pressure at the anode, as is reflected by the pressure
water is non-freezable. The temperature at which freezing is dependency of the measured currents [52]. For a Nafion112
observed varied with the water content. However, it was membrane of 50 lm thickness, a cross-over current of
found that freeze–thaw cycles did not lead to irreversible 1 mA cm–2 at atmospheric conditions is reported for beginning
changes. McDonalds et al. studied N112 membranes that of life. This corresponds to 2.6 × 10–13 mol H2 cm–1 kPa–1 s–1.
underwent up to 385 freeze–thaw cycles between –40 and End-of-life conditions are considered to correspond with val-
+80 °C [46]. These materials were in a relatively dry state ues in the order of 13 mA cm–2 bar–1 [3]. Oxygen cross-over is
(only humidified by ambient conditions). These studies did usually not measured in situ. The permeability of membranes
not indicate any serious physical or chemical damage. The for oxygen is usually half of the hydrogen permeability [18, 33].

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 7
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

3.3.3 Open Circuit Voltage


REVIEW

The open circuit voltage is an easy


method to obtain an indication for increased
reactant cross-over. It is however less selec-
tive than the hydrogen cross-over current
measurement, and far less quantitative. The
OCV is a suitable measure for the state of the
membrane, which follows a negative trend
that coincides with the fluoride release rate
[12]. Under the test conditions, a failing
membrane showed an OCV decline from
1 V at start of the experiment to 0.7 V. At
the same time, a fluoride release rate of
60 lg day–1 cm–2 was measured. A stabi-
lised MEA shows a nearly stable OCV, and a
negligible fluoride release rate.

3.3.4 High Frequency (h.f.) Resistance


A substantial part of the high frequency
(h.f.) resistance of a PEMFC is due to the
membrane resistance. Additional contribu-
tions come from the GDLs and current col-
lector plates. Monitoring this resistance is
feasible by in situ impedance spectroscopy
[3, 49, 53]. Although the h.f. resistance would
certainly signal changes in the membrane
resistance that occurs upon degradation, it
was found not to be a very sensitive charac-
teristic for degradation [3, 49]. Chemical deg-
radation can lead to membrane thinning,
which would not only result in a lower resis-
tance but also in reduced proton conductiv-
ity that compensates this effect. Moreover,
corrosion products from metal plate oxida-
tion or carbon corrosion may have larger Fig. 1 Degradation conditions, mechanisms and effects for perfluorinated membranes.
effect on the total resistance increase. So, al-
though the h.f. resistance is a valuable pa-
rameter to monitor, its usefulness for diag- and the effects that can be measured by the various character-
nostics of membrane failure seems limited. isation tools. Membrane degradation as observed in fuel cell
testing, both under near-ideal and non-ideal conditions, is
3.3.5 Ex Situ Characterisation summarised.
Ex situ characterisation, microscopic analysis can indicate
3.4.1 Near-Ideal Conditions
cracks and especially membrane thinning that has occurred
during operation [37]. Moreover, EDX or chemical analysis In the 26,000 h test by Gore, under fully saturated condi-
can establish the presence of contaminants. 19F NMR has been tions and at 70 °C, severe (though unspecified) thinning of
used by various authors to establish the changes in the chemi- the 35 lm GORE_SELECT® membrane was observed, with
cal composition of the PFSA membranes and degradation ionomer loss most apparent at the cathode side [3]. This was
products [13, 20, 42]. X-ray diffraction is used to study mor- accompanied by an increase in the H2 cross-over rate from 1
phological changes that occurred during ageing [42]. to almost 10 mA cm–2 (at near-ambient pressure) in the first
22,000 h. At 26 300 h the cell reached end-of-life due to
membrane failure, which led researchers to postulate that
3.4 Observed Membrane Degradation in PEMFC Operation
10 mA cm–2 cross-over current density corresponds to end-
Figure 1 schematically represents the relations between of-life conditions. The h.f. resistance did not show significant
operational conditions, the main degradation mechanisms changes.

8 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

REVIEW
Membrane thinning was also observed under near-ideal 3.4.4 High Temperature
operating conditions by Nuvera [7], where the original thick-
The temperature of the operation appears to be another
ness of 36 lm was reduced to 22–26 lm in 9,000 h. Osaka gas
decisive factor. Low temperature mitigates the degradation,
reports a 36,000 h test on an MEA containing an unspecified
as no increase in hydrogen cross-over was measured by Plug-
perfluorosulphonic acid membrane without any increase in
Power in a study of the influence of start/stop conditions on
ohmic loss and only a slight OCV decrease (20 mV) [11]. This
fuel cell durability at 50 °C [57]. Although, the fuel cell per-
may imply that a thicker or a more chemically stable mem-
formance degradation was considerable at conditions in
brane was used than in the studies by Nuvera and Gore. A
which the MEA was allowed to lose much water, no lowering
sudden three-fold increase in F-concentration in the effluent
of the open circuit potential was observed, suggesting that no
water after 800 h has been observed by Xie et al. in a 1,000 h
increase in hydrogen cross-over occurred.
test at oversaturated conditions [49]. As mentioned pre-
On the other hand, a combination of high temperature and
viously, at near-ideal conditions fluoride release rates of
reduced humidity increases the degradation rate [20]. At GM
0.01–0.02 lg F cm–2 h–1 were measured [16, 20].
an accelerated test was performed at 95 °C, anode 75% RH,
cathode 50% RH and 3 bar gas pressure, with a current den-
3.4.2 Reduced Humidity
sity of 0.2 A cm–2. Depending on the membrane material, the
Continuous operation at a relative humidity lower than F release rate was between 0.3 and 2 lg cm–2 h–1[20]. The
100% can lead to an increase in membrane degradation. Bal- chemical degradation led in that study to a substantial thin-
lard [54] reports operation time before gas cross-over passes a ning of the membrane, to less than half of the initial thickness
certain threshold. For a non-optimised cell configuration, after 1,000 h at accelerated conditions.
3,300 h were reported in the case of a fully humidified anode In an early study by Gore, fluoride release rates of their
and a 70% humidified cathode, reduced to 1,100 h for a fully reinforced membranes during the accelerated fuel cell condi-
humidified anode and a dry cathode, and a mere 300 h at tion (90 °C, 0.8 A cm–2, 75% RH) were in the order of
complete dry operation. Yu et al. observed a rather high deg- 0.2–0.6 lg F cm–2 h–1 [39]. In these tests, no correlation could
radation rate under only slightly under-saturated conditions, be established between the fluoride release rate on one hand
amounting to ca. 60 lVh–1, in parallel with a strong increase and the hydrogen permeability and lifetime on the other.
in hydrogen cross-over (from 1.35 to 27 mA cm–2) [8]. As was According to Gore, this is because the membrane failure is
observed by Hori et al. [55], fuel cell degradation occurs pri- mechanical and quite localised. The hydrogen cross-over rate
marily in the part of the cell that is under-saturated. When in reinforced Gore membranes (generation 2001) at the be-
operating at relative dry conditions (at 50 °C, air RH 60%, ginning of life amounted to 2 mA cm–2, and increased over
hydrogen RH 100%), the fuel outlet/air inlet showed the larg- time to values of 5–13 mA cm–2. The value of 13 mA cm–2,
est drop in OCV and the highest hydrogen cross-over rate. marking end-of-life, was reached after 600–1,600 h. Loss of
Overall cell performance drop was significantly higher at air ionomer varied between 2 and 28%, and again was not corre-
humidification at 60% RH of air compared to 100%. lated with lifetime [39].
Studies of membrane stability at even more elevated tem-
3.4.3 High Cell Voltage perature (e.g. 120 °C) are often carried out at reduced humid-
ity (<50%). It is expected that the chemical degradation will
Exposing the fuel cell to low or zero current conditions has
be faster at these conditions than at ideal conditions. Hence, it
been shown to lead to higher degradation rates and increase
is encouraging that Asahi glass reports operation of a new
in hydrogen cross-over in comparison to operation at a con-
membrane during 400 h without membrane failure or even
stant current. Liu and Case [9] and Nakayama et al. [15]
significant fluoride release [12].
found an increase in hydrogen cross-over at OCV while
operation at a constant current density did not show such an
3.4.5 Freeze–Thaw Cycles
increase. The higher the OCV/load ratio, the sooner the
increase in hydrogen cross-over started in the experiments of The effect of freeze–thaw cycles was investigated for
Nakayama et al. [15]. MEAs without [58, 59] and with [60–63] purging with dry
The detrimental effect of OCV conditions has led to the gas. At wet conditions, Cho et al. found a large effect on the
use of open circuit tests as accelerated membrane lifetime ohmic resistance which was however not corroborated by a
evaluation [16, 36, 38]. Also the formation of a band of Pt change in polarisation curve [59]. In the other cases no sub-
deposited in the membrane and its effect on membrane deg- stantial effect on the membrane was observed. It should be
radation was studied in this way [32, 33]. Lee et al. showed noted that purging to the extent of drying out the membrane
that replacing H2 by air during OCV reduced the degradation may not be needed to protect the membrane from freezing
substantially, another indication of the role of molecular damage but rather to create a reservoir where water can be
hydrogen and oxygen in the degradation [56]. stored during start up at freezing conditions [64–66].

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 9
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

3.5 Outlook on Membranes access of gas, electrons, protons and water is needed. The life-
REVIEW

time issues of the electrodes, which include stability of mate-


The recent results on membrane degradation have shown
rials as well as of structure, are discussed for the different
that conditions associated with automotive applications such
phases of which the electrodes are composed.
as cycling of voltage, humidity and temperature accelerate
membrane degradation. However, good progress has been
made for PFSA membranes in the past few years by identify- 4.2 Mechanisms of Electrode Degradation
ing sources of chemical and mechanical failure and by find-
4.2.1 Platinum Dissolution and Particle Growth in PEMFC
ing solutions, both by chemical modifications and by using
reinforced membranes. Recent results by DuPont [36] show As was mentioned in Section 2, the loss of electrochemi-
considerable improvement in durability for new materials cally active surface area (ECSA) is a major source of degrada-
compared to those that were on the market when most of the tion. Several studies have observed the growth of Pt particles
tests in Tables 1 and 2 were done. However, a full evaluation as well as dissolution of Pt in the membrane phase [10, 67,
of these improved materials still has to be carried out. This 68]. Both phenomena were already described for the PAFC
will be facilitated by the fact that accelerated test protocols [69, 70], and are closely related as will be explained below.
have been developed that are more closely related to realistic Thermodynamics predict substantial Pt dissolution in acid
conditions than e.g. Fenton’s test and can, therefore, be media strongly increasing with potential in the region of
expected to give more reliable predictions of lifetimes. Addi- 0.85–0.95 V versus RHE [71], according to the reaction
tionally, this will enable a better comparison with non-fluori-
nated compounds. Pt $ Pt2‡ ‡ 2e E0 ˆ 1:188 V versus RHE (4)

Such behaviour has been observed for Pt foil in phosphoric


4 Electrodes acid at an elevated temperature of 196 °C [69]. The equilib-
rium electrochemical potential of the Pt dissolution decreases
4.1 Electrode Function and Requirements
with particle size [72]. Accordingly, higher equilibrium Pt2+
The conventional PEMFC electrode consists of several dis- concentrations were measured for Pt/C at 80 °C than pre-
tinct phases: the catalyst, composed of a high surface area car- dicted for bulk Pt [10].
bon support, loaded with nano-scale catalytic particles of Pt However, experiments on Pt thin film, Pt wire as well as on
mixed with an ionomer recast dispersion to form the catalytic nanosized Pt on C have shown that at 80 °C in the region
layer. This thin catalytic layer, with Pt loading in the order of above 0.85 V versus RHE Pt dissolution does not show the
0.4 mg cm–2, is applied between the membrane and a so- strong Nernstian potential dependence, and even has a maxi-
called backing or GDL, which consists of a macro porous car- mum around 1.15 V, as shown in Figure 3 [10, 73, 74]. This
bon cloth or carbon paper, and a thin micro porous gas diffu- was ascribed to the formation of an oxide layer, as also pre-
sion layer (MPL) (Figure 2). For a fast catalytic reaction, free dicted by a kinetic model taking into account electrochemical
formation of platinum oxide and its subse-
quent chemical dissolution [72]:

Pt ‡ H2 O $ PtO ‡ 2H‡ ‡ 2e
…5†
E0 ˆ 0:980 V versus RHE

Pt2‡ ‡ H2 O $ PtO ‡ 2H‡ (6)

The surface composition in the various


potential ranges is illustrated in Figure 4.
Surface X-ray scattering by Nagy [75] has
confirmed the formation of an oxygen skin
at potentials higher than 1.2 V, with Pt
atoms replacing oxygen atoms at even high-
er potentials. Note that this leads to disinte-
gration of the Pt surface. The formation of
oxides is also related to the observation that
potential cycling enhances the Pt dissolution.
Kawahara et al. have studied the dissolution
of Pt during cycling ex situ [76, 77]. It is
Fig. 2 Cross-section of an electrode with gas, proton and electron pathways. known from the literature of the early 1990s

10 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

REVIEW
occur significantly in catalysts in the gas phase at tempera-
tures below 500 °C is considered to be an indication that coa-
lescence is not the prevailing mechanism [10, 83, 84]. In recent
PEMFC studies it was found that both potential and
increased humidity enhance the particle growth [86, 87],
which favours the dissolution/redeposition mechanism. Still,
although both mechanisms lead to an increase in the average
particle size with an asymptotic particle size distribution
(PSD) the coalescence mechanism has a log-normal distribu-
tion (tail at large sizes) and the Ostwald ripening has a tail at
the smaller particle sizes but with a maximal particle size cut-
off [88]. In many studies a log-normal distribution of Pt parti-
cle sizes was found in virgin and used electrodes [10, 67, 83,
Fig. 3 Pt dissolution rate versus dissolution potential and temperature. The
86, 88]. However, it must be noted that this requires a good
real surface area of the electrode is 24 cm2. From ref. [73].
sampling also of small particles and results can be affected by
the fact that several mechanisms are active at the same time
[10, 83]. Moreover, carbon corrosion (see the next section) can
have an effect on Pt particle growth.

4.2.2 Carbon Oxidation

Electrochemical corrosion of the carbon support in PAFCs


has been known since the 1970s and 80s. Thermodynamically,
carbon (graphite) can be electrochemically oxidised to CO2 at
Fig. 4 Surface layer composition of platinum as function of the potential. quite low potentials:

C ‡ 2H2 O $ CO2 ‡ 4H‡ ‡ 4e E0 ˆ 0:207 V versus RHE (7)


that fast cycling leads to oxides that are difficult to reduce,
often called b-oxides [78, 79], and that the reduction of these Due to the slow kinetics of these reactions, carbon can still
b-oxides are accompanied by the dissolution of platinum [80]. be used in fuel cells [89]. A study by Jarvi et al. [90] showed
Pt ions can not only dissolve according to reaction 4 or 6 in that on Vulcan XC72 carbon surface oxidation occurs at 65 °C
the Pt2+ form but also in the Pt4+ form [68]. In the ionomer and potential >0.8 versus RHE. At these conditions a wide
they will replace H+, but it has also been suggested that they variety of carbon surface oxides was formed resulting in a
form negatively charged anions with degradation products reduced hydrophobicity. One specific species is the quinone
like F– and SO24 (products of chemical membrane degrada- group that is electrochemically active with a redox peak at
tion or of impurities in the support) or Cl– (possible residue 0.55 V versus RHE that can be identified in cyclic voltamme-
from catalyst preparation). Such a complex formation would try. Experiments at 1.2 V versus RHE potential hold suggested
also promote the dissolution of Pt and PtO [68]. a fast increase in surface oxides during the first 16 h and a
Either through diffusion (cations) or due to the electric slower increase thereafter. The decrease in corrosion rate in
field (anions) Pt will migrate towards the anode side of the time has also been observed elsewhere [87, 91, 92].
membrane [81]. However, Pt dissolved into the ionomer The presence of Pt catalyses the subsequent oxidation to
phase can also redeposit on other (larger) particles in the CO2 as verified by DEMS for carbon activated with 20% Pt
cathode, due to the higher equilibrium potential of dis- [93]. The mechanism consists of the following steps:
solution on these particles [82]. These particles need to be on
the carbon support in order to allow for the exchange of elec- C + H2O → COsurf + 2H+ + 2e– E>0.3 V versus RHE (8)
trons. The difference in dissolution potential is considered to
be related to the surface tension. The above mechanism of
Pt
dissolution and redeposit ion is an example of Ostwald ripen- COsurf + H2O ! CO2 + 2H+ + 2e– E = 0.8 V versus RHE (9)
ing, i.e. minimisation of the surface energy is the driving
force. At a potential higher than 0.3 V versus RHE, COsurf starts
In PAFC studies and in some PEMFC studies there has to form irreversibly on the carbon particle surface. COsurf is
been some debate on whether particle growth occurs by oxidised on neighbouring Pt sites to CO2 at a potential of
another mechanism, i.e. coalescence [10, 70, 83–85]. In this 0.8 V versus RHE. The amount of CO2 formed is proportional
mechanism Brownian motion is the driving force, causing to the total length of the two-dimensional grain boundaries
surface diffusion of particles with random collisions leading between the Pt particles and the carbon support [93]. These
to coalescence [85]. Usually the fact that sintering does not results were confirmed by Roen et al. [94] who in the absence

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 11
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

of Pt observed CO2 emission only at potentials above 1.1 V to be much lower [67]. The trace of the soluble ionomer is
REVIEW

versus RHE. In the presence of Pt, CO2 emission was already found in the cathode effluent water [67]. The dissolution
observed in a potential range of 0.55–0.6 V versus RHE. The becomes more severe over time.
influence of Pt on the carbon corrosion became less pro- As a result of the chemically ionomer degradation, the
nounced at temperature higher than 50 °C. By using the field interface between catalyst and ionomer is lost, consequently
emission-scanning electron microscopy (FE-SEM) to examine reducing the electrochemically active surface of the electrode.
a highly oriented pyrolytic graphite (HOPG) electrode with As the ionomer acts as a binder in the electrode, it has a
the Pt nanoparticles on top, CO2 formation is mostly found strong impact on the ionic and electronic conductivity of the
around the Pt particles, but interestingly these Pt particles are electrode. The structural instability due to ionomer swelling
still present without becoming detached, despite carbon cor- and dissolution during long-time exposure in high humidi-
rosion and blister formation [95]. In a recent paper Ball et al. fied conditions can change the electrical properties, leading to
found no effect of Pt on the corrosion. This conclusion, how- degradation of the fuel cell.
ever, was based upon corrosion current measurement and
not on the direct measurement of CO2 evolution [92].
4.2.5 Degradation of the GDL and MPL
The thermal degradation of carbon and platinum support-
ed on carbon in air at an elevated temperature has been stud- The GDL and MPL are responsible for gas phase transport,
ied by Stevens [96]. Whereas the Black Pearls carbon is stable electronic contact, heat transport and water removal. The
in air at temperatures as high as 195 °C, a high platinum load- GDL is a macroporous layer consisting of carbon fibres,
ing can lead to a loss of more than 80% of all carbon, where usually cloth or paper, that have to some extent been made
the time needed to reach the maximum amount of combusted hydrophobic by a Teflon (PTFE) coating. The MPL, posi-
carbon is determined by the temperature and the platinum tioned between the GDL and the catalyst layer consists of car-
loading. At platinum loadings used in commercial PEMFC bon black particles and Teflon as a binder. Unlike the carbon
catalysts, 40 wt% and higher, the loss of carbon at the lowest black in the catalyst layer, the carbon black in the MPL is not
experimental temperature, 125 °C, amounted to 15% after susceptible to electrochemical corrosion and contains no Pt to
1,000 h. From the relation between carbon loss and tempera- catalyse oxidation reactions, but chemical surface oxidation
ture, it was concluded that below 100 °C thermal oxidation of by water or even loss of carbon due to oxidation to CO or
platinum loaded carbon in air did not take place [96]. CO2 cannot be excluded [96, 98]. This will lead to increased
However, the humidification of air substantially enhanced hydrophilicity of the MPL. The carbon fibres of the GDL may
the thermal corrosion rate of carbon, by providing an addi- be more stable, but otherwise are susceptible to the same
tional pathway for chemical carbon oxidation through a reactions. Also decomposition of PTFE, used as a binder or
direct reaction with water [97, 98]. In these studies, the corro- hydrophobic coating has been suggested. Schulze et al. pres-
sion rate was also found to be higher for high surface area ent evidence for this on the basis of XPS data, but a mecha-
carbon blacks, which was ascribed to a finer dispersion of Pt nism has not been proposed [99].
particles on this material, i.e. more particles per unit weight. The result of these degradation mechanisms is that both
the GDL and the MPL lose their hydrophobic character [4, 6,
4.2.3 Oxygen Evolution 99], and that the pore structure of the materials changes. Both
these phenomena can have a substantial effect on the water
Electrochemical oxidation of water takes place at electrode
content of the GDL and MPL and therefore, on the mass
potentials >1.5 V versus RHE, i.e. in the same region where
transport properties. Increased liquid water content of the
carbon corrosion can occur:
GDL and MPL will impede gas phase mass transport, as
pores initially used for gas phase mass transport will be
H2O ↔ 1/2O2 + 2H+ + 2e– E0 = 1.229 V versus RHE (10)
increasingly blocked by water. The relation between micro-
structure and surface properties on the one hand and mass
At the cathode this will not be very harmful although it
transport properties on the other has been the subject of sev-
can lead to the formation of cracks. When O2 evolution occurs
eral recent experimental [100, 101] and modelling [102–105]
at the anode, reaction with residual hydrogen (not abundant
studies which indicate that indeed mass transport can be ser-
at high potentials) may result in a more serious damage.
iously affected by the hydrophobicity of the GDL and MPL as
well as by the pore size.
4.2.4 Ionomer Degradation
The GDL properties can also be changed by mechanical
The ionomer in the electrode is susceptible to many of the degradation as a result of the compression forces in a fuel cell.
same degradation mechanisms that were described for the From ex situ tests in which the material was aged under a
fuel cell membrane. In comparison to the membrane, the compression force, Lee and Mérida [106] concluded that the
ionomer is in a state that is more soluble than the Nafion compressive strain increased with the applied pressure but
membrane, which only starts to dissolve when in water at even more strongly with temperature. From this it was con-
210 °C and high pressure of 68 atm for more than 2 h. The cluded that the GDL strain was influenced by the PTFE stabil-
stability of recast Nafion ionomer in catalyst layer is reported ity. Properties like in plane electrical resistivity, surface con-

12 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

REVIEW
tact angle, bending stiffness and porosity were not affected. position. EDX probes the sample in a broader region closed to
However, it was found that convective air flow through the the surface. Raman spectroscopy can be used to characterise
GDL can lead to loss of material. the carbon surface area [13].
Methods to characterise the GDL include water and mer-
cury porosimetry. The hydrophobicity of the material can be
4.3 Characterisation of Electrode Degradation
probed by contact-angle measurements including sessile drop
4.3.1 Electrochemical Characterisation and Wilhelmy balance methods [6].
Cyclic voltammetry, carried out in situ, is used to establish
the ECSA of the electrodes. The electrode under investigation 4.4 Electrode Degradation Observed in Fuel Cell Operation
(working electrode) is flushed with nitrogen, the other elec-
A scheme of conditions enhancing electrode degradation
trode (counter and reference electrode) is flushed with hydro-
processes and the measurable effects is presented in Figure 5.
gen. The ECSA is measured from the hydrogen adsorption/
The results from degradation studies under relevant condi-
desorption region [52]. Cyclic voltammetry also gives infor-
tions are discussed in the following sections.
mation on the double layer capacitance of the electrode or on
other electrochemically active species on the surface such as
4.4.1 Near-Ideal Conditions
the carbon quinone. Yoda et al. blocked the Pt catalyst sites
by CO prior to measuring the cyclic voltammogram to elimi- The 26,000 h test by Gore at 0.8 A cm–2, at 70 °C and full
nate the effect of Pt on the double layer capacitance. This gave humidification of the gases revealed a 66% decrease in the
a clearer indication of changes at the carbon surface [107]. ECSA at the cathode, 25% decrease in the proton resistance of
Impedance spectroscopy at high frequency (1–100 kHz) the cathode and an increase in the polarisation resistance. The
can be used to determine the proton resistance of the elec- decrease in the ECSA corresponded to 34 mV voltage loss, by
trodes, both with O2 or with N2 on the working electrode the relation blog(a1/a2), with b being the kinetic Tafel slope
[53]. Impedance spectroscopy involving lower frequency and and a1 and a2 the ECSA before and after the test, respectively.
otherwise under operational conditions can be used to inves- The total electrode loss increased by 116 mV, which indicates
tigate the charge transfer as well as the mass transport behav- substantial mass transport losses. Post-test analysis revealed
iour of the cathode [3, 99, 108]. no significant changes in the backings or catalyst layer, i.e. no
Voltage–current characteristics measured under fuel cell corrosion of the carbon support or measurable change in
conditions yield, after correction for the ohmic losses and hydrophobicity of the backing. In addition, SEM did not
neglecting the anode contribution, the polarisation curve. detect Pt in the membrane or further substantial changes in
Comparison of such curves obtained under oxygen and the Pt distribution in the electrode [3]. Pt particle growth
air results in the oxygen gain, which gives indications of seems to have been the main factor for ECSA loss. A loss of
limitation of transport of oxygen and/or protons in the elec- ECSA at the Pt/C cathode was also reported by Ferreira in a
trode. 2,000 h test at 0.3 Acm–2. In this case a reduction of 43% was
measured [10]. Nuvera reported 60% loss of ECSA cathode
4.3.2 Chemical Analysis of Effluents after 9,000 h at 0.7 V [7].
Pt particle growth at the anode under near-ideal condi-
Evolution of CO2 can be monitored by gas chromatogra-
tions was observed by Xie et al. [67], with particles increasing
phy or mass spectroscopy [94, 109]. Chemical analysis of
from 1–6 to 3–15 nm after 500 h in a MEA operated at
effluent water by ICP–MS or ICP–AES can detect ionomer
1.07 Acm–2. Also Pt migration to the membrane interface and
degradation species both at the cathode and the anode, but
into the membrane seemed to have occurred. Liu and Case
results often cannot distinguish between electrode or mem-
report a larger decrease in ECSA at the anode than at the
brane degradation [49].
cathode under constant current conditions [9]. Pt particle
growth may occur by coalescence, as the mechanism of dis-
4.3.3 Ex Situ Characterisation
solution and redeposition is not favoured by the potential at
Microscopy techniques like (FEG)-SEM and TEM are com- the anode. On the other hand, it has been suggested that Pt
monly used to establish Pt particle growth, Pt deposition in particle growth can arise from Pt crossing over from cathode
the ionomer and electrode thinning [10, 67, 68, 83]. Pt particle to anode. This was observed for PAFC [70], and predicted by
size growth can also be determined by X-ray diffraction [86]. the model of Darling and Meyers [81]. Recently Guilminot et
Solid-state 19F NMR [13] can be used to analyse the chemical al. reported an apparent net migration of Pt from cathode to
composition of the ionomer in anode and cathode after opera- anode [83]. On the other hand Zhang et al. measured a con-
tion. EDX/EDS as well as ICP-AES and XPS have been used stant Pt content at the anode [33]. Due to the fast hydrogen
to study the chemical composition of catalyst layers and oxidation kinetics, the ECSA loss at the anode is not expected
GDLs [7, 68, 83, 99]. Whereas EDX and ICP-AES give infor- to lead to a decrease in cell voltage.
mation on the bulk composition of the materials, XPS has a Carbon corrosion was suggested by Nuvera for a cell that
high surface sensitivity and gives information of surface com- had operated for 5,000 h both from a reduction of the C/F

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 13
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

authors to conclude that optimal water man-


REVIEW

agement would be a key parameter to miti-


gate voltage loss from increased mass trans-
port limitations. The fact that excess liquid
water cannot be the sole cause for loss of
hydrophobicity was illustrated in a recent
paper by Schulze where it was observed that
PTFE decomposition was more pronounced
on the anode than on the cathode side [99].
Xie et al. [49] also observed washing out
and degradation of the ionomer phase in the
electrode by measuring F and S release in
the effluent water. Note that gas transport
rather than electronic or protonic transport
is the most probable cause of increased
polarisation losses not directly related to
ECSA loss. The effect of carbon corrosion
and ionomer degradation on the electronic
and protonic transport has not been estab-
lished unambiguously. Some results indicate
improved transport due to better contacts
arising from swelling of the ionomer [3, 107],
or due to thinning of the electrode. On the
other hand, loss of structural integrity of the
catalyst layer may add to the ohmic resis-
tance [49].

4.4.2 Potential Cycling and OCV Conditions

Several authors report results from poten-


tial cycling of MEAs. This can involve differ-
ent experiments, i.e. under H2/N2 [10, 31, 86,
110] or H2/air [111, 112] with a potentiostat,
by load on–load off cycles [9, 13] or by a
simulated drive cycle [86]. Also studies at
Fig. 5 Degradation conditions, mechanisms and effects for electrodes. OCV [10, 32, 86] or near OCV [107] condi-
tions were performed. In all cases a faster
decrease in the cathode ECSA was observed
atomic rate (established by EDX) and a thinning of the elec- than under constant load conditions. The results suggest that
trodes [7]. the ECSA decreases faster as a result of cycling to high volt-
Increase in mass transport losses was observed by several ages [10, 86, 110, 112] and that the reduction increases with
groups [3, 4, 6, 9, 99], and in most cases ascribed to increased the maximum applied voltage (range 0.96–1.5 V). This can be
hydrophilicity of the electrode. Note that these mass trans- illustrated by data by Ferreira et al. [10]. A 46% loss of ECSA
port losses were in some cases quite substantial and indeed was measured at 0.2 Acm–2 (0.75 V) after 2,000 h. At OCV
dominated the observed voltage loss, e.g. in the stationary (0. 95 V) conditions 75% loss was observed over the same pe-
state test by Liu and Case [9] and in the long-term study by riod. However, after 10,000 cycles between 0.87 and 1.2 V in
Gore (70% of the 116 mV) [3], although in this case mass only 100 h, 69% of the Pt ECSA was lost.
transport may have been enhanced by the increased compres- The faster decrease in ECSA is generally ascribed to faster
sion force (see Section 6). However, at Ballard only a slight dissolution of Pt at higher potential and during cycling, pro-
increase was observed in the 11,000 h test [5]. Other studies moting the dissolution/redeposition mechanism. Also carbon
at Ballard have shown that the increase in mass transport can corrosion may add to the ECSA loss [98]. In general the
be enhanced by exposure of the electrode to liquid water [4], decrease in ECSA is slower on lower surface area carbons,
which would be in agreement with the mechanism of carbon with graphitised carbon showing the slowest rate of ECSA
(surface) oxidation discussed previously. On the other hand, loss [87, 97, 98]. This may be not only due to larger initial par-
it was reported in the same paper that insufficient humidifi- ticle size, but also be related to a lower rate of carbon corro-
cation can also increase mass transport effects, which led the sion.

14 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

REVIEW
Since the dissolution is faster, Pt particles are also more 4.4.3 Start–Stop Cycles
often detected in the membrane after cycling or exposure to
OCV. A quite frequent observance is a Pt band at a certain At cell potentials of ≥1.2 V, carbon corrosion is accelerated.
distance from the cathode [10, 31–33, 110–112]. In the case of This was confirmed by Borup et al. [86] and also by Mathias
PAFC, Pt was able to move to the anode and deposit there et al. who demonstrated that a standard Pt/C catalyst lost
[70]. It was found that in PEMFC the position of this band 15% of its weight during 20 h at 1.2 V [87]. Although cell volt-
depends on the H2 and O2 partial gas pressures, with the ages higher than OCV are not expected to occur during nor-
band moving towards the anode upon a reduction of the H2 mal operational conditions, it was shown by Reiser et al. [114]
partial pressure or an increase in the O2 partial pressure [31, that high cathode potentials (versus RHE) can be expected
33, 111]. Burlatsky et al. [31] presented a model that shows during start–stop cycles, when due to leakage of air a hydro-
precipitation occuring at a location where there is a stepwise gen/air front develops at the anode. This is illustrated in Fig-
change of the electrochemical potential. In turn, such an ure 6. The presence of O2 at the right-hand side of the cell
increase occurs at Pt particles >20 nm, on which the reaction results in a drop of the electrolyte potential in the order of
of H2 and O2 is diffusion controlled, and at the location in the 0.6 V, which results in cathode potentials at the right-hand
membrane where this reaction results in complete local side in the order 1.4 V versus RHE. The electron balance is
depletion of H2 and O2. Burlatsky also observed smaller restored by the normal fuel cell reactions occurring at the left-
(5–10 nm) particles throughout the membrane. At these parti- hand side. Alternatively, the result can be considered as two
cles the potential is constant. Results were confirmed by fuel cell reactions occurring in the horizontal direction of the
Zhang et al. and Bi et al. [33, 111]. Zhang et al. remarked that fuel cell. At the right-hand side, the species to be oxidised are
the presence of a Pt band may affect the H2 concentration pro- water, carbon and platinum. The protons however are
file in the membrane, which may have consequences for the exchanged vertically, as the horizontal pathway is too long.
membrane degradation. At OCV conditions both oxygen reduction reactions occur at
There has been some doubt whether Pt particles observed the same speed. Mathias et al. estimated that a fuel cell car
in the membrane were result of oxidation and reduction of Pt would spend 100 h during its envisaged lifetime experiencing
or resulted from carbon oxidation which made them come off such harsh conditions. The present materials would therefore
the support. TEM studies have revealed that the particles appear inadequate for automotive application [87].
have a flower-like dendrite structure, more in agreement with Experiments by Reiser et al. confirmed severe carbon deg-
the first explanation [10, 68]. radation as a result of the above conditions [114]. However,
Enhanced carbon corrosion was also reported at higher cell water and Pt can also be oxidised at these potentials. Pt oxi-
voltage [86, 107], but appears for voltages up to OCV not to dation may be limited by oxide formation (see above). The
be a major factor. The corrosion of the carbon surface can lead importance of water oxidation, i.e. oxygen evolution, cannot
to a minor positive as well as to a major negative influence on be ignored. At the relevant potentials the respective rates are
the fuel cell performance. Part of the platinum surface can be of the same order of magnitude [115, 116] and will be affected
covered by carbonaceous species after electrode preparation, by e.g. the type of carbon and catalyst dispersed on it [87] or
making it inaccessible for oxygen and thus reducing the activ- by the humidity. In experiments by Ballard and Borup et al. it
ity of the fuel cell [113]. In the initial operation period, the car- was observed that carbon corrosion during start–stop cycles
bon corrosion can unblock these Pt particles and therefore was enhanced at low humidity [57, 86, 116].
increase the catalyst surface available for reaction. However, A better insight in the role of water at these conditions will
in the long run, the negative effects of carbon corrosion domi- help to develop mitigation strategies as oxygen evolution is
nate through increased hydrophilicity of the oxidised carbon considered to be less harmful than carbon corrosion, i.e. may
surface, reduced location sites and electronic contact for Pt even result in increased active area [57]. Development of stra-
particles and modified gas permeability of the gas diffusion tegies or catalysts promoting water oxidation can also be con-
layer [95]. sidered. Yu et al. assumed that at OCV conditions the ORR

Fig. 6 Reactions in a fuel cell under start-up/shutdown when hydrogen and air are both present in the anode compartment. Left-hand side contains
hydrogen, right-hand side contains air. Adapted from ref. [106].

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 15
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

reaction at the right-hand side is the rate-limiting factor for 4.4.6 Low Relative Humidity–High Temperature
REVIEW

the reaction occurring at OCV. Reduction of the anode cata-


In potential cycling experiments it was found that reduced
lyst loading to 0.05 mg Pt cm–2 seemed to reduce the carbon
humidification slows down Pt particle growth in agreement
and water oxidation rate, while not affecting anode losses
with reduced Pt dissolution rates [86, 87]. The effect of low
during normal operation [117].
relative humidity on carbon corrosion has already been dis-
cussed previously. An increase in temperature on the other
4.4.4 Fuel Starvation
hand promotes Pt particle growth [86, 87]. A similar result
Experiments at Ballard showed that during severe fuel was obtained by Yoda et al., who also reported enhanced car-
starvation, mimicked by replacing H2 by N2 at the anode, the bon oxidation at 90 °C compared to 80 °C [107].
cell voltage became negative, i.e. cell reversal occurred. The Guilminot et al. report on a stack test of 529 h at 60 °C, air
negative cell voltage corresponded to a potential >1.23 V ver- 75% RH and dry hydrogen operated at 0.18 Acm–2 (about
sus RHE at the anode allowing for water oxidation to sustain 0.67 V). In spite of the low temperature, short test duration
the imposed current density [54]. Further experiments and relatively dry conditions, a substantial redistribution of
showed that in time the potential decreased further, creating Pt in the cell was reported as well as carbon corrosion. How-
potentials favourable for carbon oxidation, which was con- ever, it is possible that the results were affected by a rest peri-
firmed by the presence of CO2 in the effluent gas [109]. od of 123 h in cell operation after 63 h. The conditions at rest
Taniguchi et al. [118] have shown that fuel starvation led were not completely clear, whereas in a first publication it
to cell potentials of –2 V, where the anode potential increased was stated that the cell was purged with H2, in a subsequent
to 2.2 V, while the cathode potential only dropped slightly. paper the rest period was used to explain the occurrence of
Three minutes of fuel starvation (utilisation in experi- carbon corrosion [68, 83].
ment = 100%) led to a 25% lower current density at 0.7 V,
and starvation for 7 min led to 50% lower current density. At 4.4.7 Freeze–Thaw Cycles
the anode, enrichment of platinum in the Pt/Ru catalyst was
Freezing of the water contained in the membrane-electrode
observed, indicating the loss of ruthenium. Determination of
assembly can lead to performance loss [59, 60]. With (partial)
the average particle size by TEM revealed that the mean par-
water removal prior to freezing, the cell voltage loss can be
ticle size increased from 2.6 to 5 nm at the anode and from
limited to the order of 0.1–0.3 mV per freeze–thaw cycle
2.8 to 4 nm at the cathode. The increase at the cathode was
[61–63, 120]. Observed physical changes in the electrode
somewhat unexpected at the time but may have been related
structure observed in these studies included faster loss of
to effects of local starvation described below.
ECSA as well as increased porosity. Effective removal of the
Reiser et al. already have described that the situation giv-
water is possible by either gas purging, or washing away
ing rise to high cathode potentials during start–stop cycles is
with antifreeze liquids [54, 120].
similar to that occurring during local fuel starvation, e.g. dur-
ing flooding of the anode compartment [114]. In more de-
tailed studies it was confirmed that when the fuel starvation 4.5 Outlook on Electrode Degradation
area is larger than a critical dimension of ca. 20 mm, then
Many factors that enhance membrane degradation, such as
again high cathode potentials arise [116, 119]. The situation is
potential cycling and elevated temperature, also accelerate
similar to that depicted in Figure 6, with the exception that
electrode degradation. The role of relative humidity is more
the oxygen at the anode now results from oxygen cross-over
ambiguous, but additional complications come from start–
to a hydrogen-depleted region at the anode. The oxygen
stop cycles and incidents of fuel starvation. The exact effect of
cross-over rate now determines the reaction rate in the fuel
most of these conditions on lifetime still has to be established,
starvation section, and hence the combined rates of carbon
but from estimates by e.g. GM it appears that state-of-the-art
corrosion and oxygen evolution at the cathode. In these
Pt/C catalysts will not meet the requirements [87]. Alterna-
experiments, severe carbon corrosion was observed.
tives for Pt have been suggested, such as PtCo alloys that
have in some tests shown to be more stable as well as more
4.4.5 Air Starvation
active [87, 110, 121, 122], but their stability and activity
When oxygen is depleted at the cathode (an event called require much further research.
air starvation) hydrogen evolution takes over from oxygen The issue of carbon corrosion seems to be of even greater
reduction by the reduction of protons. In this case, the cell concern. Graphitic carbon is more stable than the conven-
voltage is slightly negative, as both hydrogen oxidation and tionally used carbon black, but has a lower surface area. This
reduction are facile reactions. Although not directly as detri- limits the minimum metal particle size, which reduces the
mental as fuel starvation, the mixture of hydrogen and air at activity. In combination with the requirement for further
the cathode can lead to undesired heat generation [109]. In reduction of the Pt loading this is not a promising situation.
the 26 000 h test by Gore, a 24 h situation of air starvation The carbon corrosion may also be mitigated by promoting
seems to have led to a decrease in the available electrochemi- competing reactions. Development of catalysts for oxygen
cally active platinum surface area [3]. evolution has been suggested, but this concept still has to be

16 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

REVIEW
proven and may introduce new durability issues. The same issue. Davies et al. [126] report stable fuel cell performance
holds for alternative supports [121]. A new electrode concept using POCO Graphite for 1,750 h. Mueller et al. found no sig-
is based upon non-conductive organic whiskers coated by Pt nificant release of sodium and organics from SGL injection-
[123]. In this concept by 3M the support is not exposed to the molded and compression-molded polymer–graphite plates ex
electrolyte, and Pt forms a continuous structure that is much situ in diluted sulphuric acid at 85 °C and in concentrated
less susceptible to dissolution and shows a high specific activ- phosphoric acid at 200 °C after more than 2,000 h [127]. Mak-
ity towards the ORR. In spite of the low active surface area kus et al. [125] have measured, after 300 h fuel cell testing
that is obtained and anticipated mass transport problems with graphite plates, concentrations of 40 mgkg–1 membrane
[121], this may, from a durability point of view, be a superior for iron and 15 mgkg–1 membrane for nickel, without loss of
concept. fuel cell performance.
The issue of gradual performance loss due to degradation At normal operating conditions the cathode potential is
in the GDL is still largely uninvestigated. This may seem not high enough for oxide formation on the surface, and the
somewhat surprising in view of the fact that the effects can be contact resistance remains constant during fuel cell operation
more severe than the loss of ECSA that has received much [125, 126]. Although not reported, under extreme conditions
more attention recently. Development of techniques that corrosion of the graphite is conceivable. Especially under
enable investigation of GDL degradation in situ seems highly start–stop or fuel starvation conditions, electrodes can be
desired. exposed to potentials promoting carbon corrosion. In our
In spite of the recent research that has led to many new own laboratory, we have observed the complete erosion of a
insights, testing protocols and characterisation methods, POCO graphite flow pattern during an accidental exposure of
improvement of the electrodes to meet the durability targets the flow plate under cell reversal for several hours.
still appears to be a formidable task.
5.2 Metal Bipolar Plates
5 Bipolar Plate Durability For metal-based bipolar plates, the stability depends on
the nature of the metal, the potential and relative humidity.
The bipolar plate serves as a separator between two adja-
Corrosion of the bipolar plates takes place when the plate
cent cells. It conducts the electronic current between these
material is oxidised at the potential to which it is exposed,
two cells, separates the gases and contains the flow patterns
and the surface oxide layer is soluble in the medium it is situ-
both for the reactants and for the coolant, in most cases a liq-
ated in. The medium is often not well defined. When in direct
uid coolant. For this review, graphite, polymer–graphite com-
contact with the electrolytic membrane, the plate–membrane
posites as well as metal-based bipolar plates have been con-
contact can lead to direct exchange of cations with protons of
sidered.
the electrolyte. On the other hand, the water produced in the
With respect to the electric conductivity of bipolar plates,
fuel cell also contains ions, emanating from the electrodes
the contact resistance is more important than the bulk re-
and the membranes. It has been measured to have a pH
sistance, especially with regard to long-term behaviour.
around 4 [20] and thus, can be considered as a weak corrosive
Whereas a beginning of life resistance of 10 mXcm2 is sug-
medium.
gested as a specification [124], for long-term stability a com-
When using stainless steel bipolar plates, corrosion is a
bined specification for anode and cathode side of lower than
generally observed phenomenon, which varies strongly
50 mXcm2 has been used as a target specification which has
between the various stainless steel grades [125, 126, 128, 129].
to be maintained up to 5,000 h of fuel cell operation [125].
Corrosion generally takes place at the anode, where most
The release of contaminants from the bipolar plates is a
stainless steels are operated in their active region [130]. This
feature which can determine to a large extent the lifetime of
means that metals directly dissolve in the medium they are
the fuel cell. Whereas it does not so much affect the properties
exposed to, without the formation of an oxide layer. When in
of the bipolar plates itself, it leads to poisoning of the mem-
direct contact with the membrane, no suitable metals could
brane and the catalysts. For Nafion 117 membranes (thickness
be identified which were not subjected to corrosion at the
175 lm), it was calculated that when using metal bipolar
anode [125]. When direct contact between membrane and
plates, a total release of cations of 8 × 10–7 mol cm–2 during
bipolar plate is avoided, some metals could be selected with
the total lifetime (5,000 h) was acceptable, limiting the
an acceptable low corrosion rate [125]. The composition of
exchange of protons in the membrane to 5% [125]. As nowa-
alloy B of that study, which could at that time not be dis-
days thinner membranes are used, this maximum release
closed, was stainless steel 904L. The suitability of that steel
should amount to 2–3 × 10–7 mol cm–2 during the total life-
with respect to corrosion resistance has also been reported by
time.
Iversen [129]. Many other stainless steels suffer from corro-
sion, even when avoiding direct contact.
5.1 Graphite and Graphite Composite Plates
The use of SS316L stainless steel end plates was shown by
For graphite and graphite composite plates, corrosion and Pozio et al. to contribute to the performance loss of PEMFCs
release of contaminants under normal operation is not an [48]. Although the end plates were not in direct contact with

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 17
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

the MEA, the MEA is contaminated by iron through contact Dow Corning [131] considered silicone elastomers as seals.
REVIEW

with humidified oxygen and hydrogen. At a rather low cur- The stability against coolant is not seen as a problem by Dow
rent density, 141 mA cm–2, and using very low current densi- Corning for these silicone elastomers, while there is concern
ties of 9 mA cm–2 overnight and during weekend periods, a about the stability in the MEA compartment. Exposure of
steadily decreasing cell voltage was observed, amounting to sealing materials to a mixture of various acids in a 1 M con-
30 lVh–1. When using aluminium end plates, no performance centration at undefined temperature is used as ex situ ageing
loss was observed. At the same time, a low pH and fluoride test. General-purpose silicone elastomers show cracks after
release was measured at the anode and cathode. A direct link 336 h in such a test, while special fuel cell grade silicone elas-
between performance loss and materials properties was not tomers survive such a test for over a year without showing
provided. cracks.
Elemental analysis of the membrane after 5,000 h testing at In a recent study [133] the stability of various sealing mate-
Nuvera, using metallic current collector and bipolar plates, rials has been tested in a simulated environment at 60 and
did not lead to detectable amounts of foreign metal ions in 80 °C. This simulated environment consisted of solutions
the membrane, except for platinum [7]. containing HF and H2SO4, in two different concentrations. It
At the cathode side, the potential is high enough to form a was concluded that silicone S and silicone G are heavily
partly conducting passive layer on the metal surface. Al- degraded in the concentrated solutions as well as the diluted
though, it protects the surface from corrosion, this layer concentrations, although most of the data are collected in the
results in an increase in contact resistance. Many metals show concentrated solutions. Degradation reveals itself by weight
a build up of such a passive layer under long-term testing in loss, complete disintegration, as well as by leaching of Mg
a fuel cell. This is a process which can take place over thou- and Ca. The latter stem from magnesium oxide and calcium
sands of hours, thus finally exceeding a maximum tolerable carbonate, which are used as fillers for obtaining the desired
resistance [125]. Although, not directly influencing the build tensile strength, hardness and resistance to compression.
up of the passive layer, the compression force has a signifi- When these components are leached out, mechanical proper-
cant influence on the contact resistance, and can explain dif- ties will be lost, and one might expect a negative influence on
ferences in various studies where one qualifies a stainless fuel cell performance as well, as these components can
steel as 316L [126], while the other rejects it on the basis of replace protons in the membrane as well as affect the proper-
increasing contact resistance [125]. Switching gases in the ties of gas diffusion media and electrodes. An increase in
cathode compartment from air to hydrogen is an effective temperature, as well as exposure to stress accelerates the deg-
method to cause a breakdown of the passive layer at the cath- radation. The degradation mechanism is thought to involve
ode side of the bipolar plate, leading to a decrease in contact decross-linking as well as backbone scission. Materials that
resistance in a couple of hours [125]. much better survive the exposure to the solutions are ethyl-
ene-propylene-diene-monomer (EPDM) and fluoroelasto-
mers.
6 Seals Only in a couple of long-term experiments was seal degra-
dation observed, and this might have been the consequence
Fuel cell stacks contain seals on the MEA side as well as on
of an inappropriate materials selection. Silicone seals in direct
the coolant side of the bipolar plate. Not only are they meant
contact with a perfluorosulphonic acid membrane suffer from
to prevent leakage of the gases and the coolants outside their
degradation, at the anode as well as at the cathode [134]. The
containments, seals function as electrical insulation, stack
degradation is probably caused by acidic decomposition of
height control and variability control as well [131]. The degra-
the sealing material, leading to colouration of the membrane
dation phenomena connected to seals do not only refer to the
and detectable amounts of silicon on the electrodes. No fuel
loss of the functionality of the seals themselves, but also to
cell performance loss or increase in gas leakage along the seal
the leakage of seal components that could poison the MEA.
has been observed. The same observation has been made by
Degradation phenomena on seals are in general poorly
St.-Pierre and Jia after a 11,000 h test with a Ballard Mark 513
understood, and only a very few number of papers have been
stack [5]. The seals were visibly oxidised, albeit more in the
published on seal degradation [131–134]. Papers from Frisch
humidification section than in the active section. In the
[131] and Du et al. [132] report on the selection of seals that
26,000 h testing by Gore, complete degradation of the glass
meet fuel cell requirements. Plugpower used an ex situ
reinforced silicone seal has been observed [3]. It forced them
method to test the seals resistance against a specific coolant
to increase the compression force to keep the cell gas tight.
as well as against acids that mimic the fuel cell environment
This might have had an impact on the effective porosity of
[132]. Weight change and the release of contaminants are
the gas diffusion media. The silicon of the seal could be
used as indicators for the compatibility of the seals. Without
detected throughout the MEA, especially on the gas diffusion
giving hints to specific materials, the differences between var-
media.
ious materials appeared to be huge.

18 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

Table 3 Dominant degradation mechanisms at conditions that fuel cells might be exposed to during their operational life and mitigation strategies.

REVIEW
Condition Dominant degradation mechanism Mitigation strategies Factors leading to even more
severe degradation
Ideal (T≤75 °C, RH ≥ 95%, GDL loss of hydrophobicity Optimisation of flow field design
voltage ≤ 0.8 V)
Low RH (<95%, voltage ≤ 0.8 V) Membrane degradation (EOL) Chemically stable membranes High T
Reinforced membranes
Optimisation of flow field design
RH cycling Membrane mechanical properties (EOL) Reinforced membranes
Open circuit voltage Membrane degradation Chemically stable membranes Low RH
Reinforced membranes
Potential cycling Pt particle growth cathode More stable Pt alloys High RH, high T
Pt dissolution cathode
Start–stop cycles Carbon corrosion More stable (carbon) supports Low RH
Operating procedures
Fuel starvation Carbon corrosion More stable carbon supports
Catalysts enhancing water oxidation
Freezing Mechanical stability GDL Sufficient removal of water after shutdown

EOL, leading to a sudden end-of-life.

7 Conclusion the mitigation strategy for that mechanism. It does not


exclude that other mechanisms contribute to the irreversible
For a fuel cell operating under constant load conditions, at loss of performance under that condition.
a relative humidity close to 100% and at a temperature of As regards PFSA membranes, it is concluded that consid-
maximum 75 °C, using optimal stack and flow design, the erable progress in durability has been made in recent years,
voltage degradation can be as low as 1–2 lVh–1. For station- by chemical modifications and reinforcement with inert mate-
ary applications this would result in 40–80 mV performance rial. It can be expected that these improvements will not ser-
loss over 40,000 h, i.e. an efficiency and power loss of 10% iously affect cost and performance. For the electrodes, the
over 40,000 h lifetime. At these near-ideal conditions the severity of durability issues was perhaps recognised at a later
dominant cause of degradation issue is the decrease in water stage (although known from earlier PAFC studies). Conse-
removal capacity of the GDL. Other degradation issues are (i) quently, solutions are further away. State-of-the-art Pt and
chemical degradation of the PFSA membrane, which does not carbon-black support materials do not meet the requirements.
result in significant voltage degradation but can determine Alternatives are being investigated, but still have to be evalu-
end-of-life, (ii) Pt particle growth which results in a steady ated with respect to performance, cost and durability. A final
decrease in cell voltage. Bipolar plates, if made of graphite or issue is the research into the loss of water-removing capacity
graphite composite, do not present durability issues. Sealing of the GDLs, which is still in its infancy. As these water man-
materials on the other hand require further investigation as agement issues contribute a major part of the overall degrada-
there are incidents of severe deterioration. tion, they deserve more attention.
However, the degradation rates can increase by orders of
magnitude when conditions include load cycling, start–stop
cycles, low humidification or humidification cycling, tem- Acknowledgements
peratures of 90 °C or higher and fuel starvation. Such condi-
This work was part of the Dutch EOS-LT PEMLIFE project,
tions are expected for automotive applications, which means
supported by the Ministry of Economic Affairs, contract no.
that a target for automotive of 10% efficiency loss, albeit over
EOSLT01029.
only 5,000 h, cannot be met with state-of-the-art materials. It
does not seem realistic that such conditions can always be
avoided by system optimisation without making the system Reference
too complex. This implies that the materials need to be more
robust. Most of the degradation issues for the materials are [1] X. Cheng, Z. Shi, N. Glass, L. Zhang, J. Zhang, D. Song,
the same as for ideal conditions, but are much accelerated Z.-S. Liu, H. Wang, J. Shen, J. Power Sources 2007, 165,
under these harsh conditions. Moreover, the fact that they 739.
can reinforce each other has a larger impact under these con- [2] S. Yoshioka, A. Yoshimura, H. Fukumoto, O. Hiroi,
ditions. Additionally, carbon corrosion, which is not so H. Yoshiyasu, J. Power Sources 2005, 144, 146.
apparent at steady-state conditions, can be very detrimental [3] S. J. C. Cleghorn, D. K. Mayfield, D. A. Moore, J. C.
during start–stop cycles and fuel starvation. Table 3 sum- Moore, G. Rusch, T. W. Sherman, N. T. Sisofo,
marises the dominant degradation mechanism for each oper- U. Beuscher, J. Power Sources 2006, 158, 446.
ating condition, the factors that enhance the mechanism, and

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 19
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

[4] J. St-Pierre, D. P. Wilkinson, S. D. Knights, M. Bos, J. [27] V. O. Mittal, H. R. Kunz, J. M. Fenton, J. Electrochem.
REVIEW

New Mater. Electrochem. Syst. 2000, 3, 99. Soc. 2006, 153, A1755.
[5] J. St-Pierre, N. Jia, J. New Mater. Electrochem. Syst. 2002, [28] V. O. Mittal, H. R. Kunz, J. M. Fenton, Electrochem. Solid-
5, 263. State Lett. 2006, 9, A299.
[6] D. Wood, J. Davey, F. Garzon, P. Atanassov, R. Borup, [29] V. O. Mittal, H. R. Kunz, J. M. Fenton, J. Electrochem.
2005 Fuel Cell Seminar Abstracts, Courtesy Associates, Soc. 2007, 154, B652.
Palm Springs, 2005. [30] M. Inaba, Water, Steam and Aqueous Solutions for Electric
[7] Y. Shi, A. Horky, O. Polevaya, J. Cross, 2005 Fuel Cell Power: Advances in Science and Technology, Proceedings of
Seminar Abstracts, Courtesy Associates, Palm Springs, the 14th International Conference on the Properties of Water
2005. and Steam, (Eds: M. Nakahara, N. Mabubayas, M. Ueno,
[8] J. Yu, T. Matsuura, Y. Yoshikawa, N. Islam, M. Hori, K. Kasuoka and K. Watanabe), Maruzen Co. Ltd. Tokyo,
Electrochem. Solid-State Lett. 2005, 8, A156. Japan, 2005, pp. 395.
[9] D. Liu, S. Case, J. Power Sources 2006, 162, 521. [31] S. F. Burlatsky, V. Atrazhev, N. Cipollini, D. Condit,
[10] P. J. Ferreira, G. J. la O’, Y. Shao-Horn, D. Morgan, N. Erikhman, ECS Trans. 2006, 1, 239.
R. Makharia, S. Kocha, H. A. Gasteiger, J. Electrochem. [32] A. Ohma, S. Suga, S. Yamamoto, K. Shinohara, J. Electro-
Soc. 2005, 152, A2256. chem. Soc. 2007, 154, B757.
[11] O. Yamazaki, Y. Oomori, H. Shintaku, T. Tabata, 2005 [33] J. Zhang, B. A. Litteer, W. Gu, H. Liu, H. A. Gasteiger,
Fuel Cell Seminar Abstracts, Courtesy Associates, Palm J. Electrochem. Soc. 2007, 154, B1006.
Springs, 2005. [34] J. Rozière, D. J. Jones, Ann. Rev. Mater. Res. 2003, 33, 503.
[12] E. Endoh, 2005 Fuel Cell Seminar Abstracts, Courtesy As- [35] L. Zhang, C. S. Ma, S. Mukerjee, Electrochim. Acta 2003,
sociates, Palm Springs, 2005, pp. 180. 48, 1845.
[13] R. Miyoshi, Y. Sakiyama, Y. Miwa, Y. Aoki, T. Yamamo- [36] G. Escobedo, K. Raiford, G. S. Nagarajan, K. E. Schwie-
to, Y. Ueno, A. Masuda, Y. Nakagawa, G. Katagiri, bert, ECS Trans. 2006, 1, 303.
H. Nakayama, M. Hori, 2006 Fuel Cell Seminar Abstracts, [37] C. Stone, G. H. M. Calis, 2006 Fuel Cell Seminar Abstracts,
Courtesy Associates, Hawaii, 2006. Courtesy Associates, Hawaii, 2006.
[14] B. Wahdame, D. Candusso, X. Francois, F. Harel, M.-C. [38] H. Tang, S. Peikang, S. P. Jiang, F. Wang, M. Pan,
Pera, D. Hissel, J.-M. Kaufmann, Int. J. Hydrogen Energy J. Power Sources 2007, 170, 85.
2007, 32, 5493. [39] W. Liu, K. Ruth, G. Rusch, J. New. Mater. Electrochem.
[15] H. Nakayama, T. Tsugane, M. Kato, Y. Nakagawa, Syst. 2001, 4, 227.
M. Hori, 2006 Fuel Cell Seminar Abstracts, Courtesy As- [40] S. Cleghorn, J. Kolde, 2005 Fuel Cell Seminar Abstracts,
sociates, Hawaii, 2006. Courtesy Associates, Palm Springs, 2005, pp. 138.
[16] E. Endoh, S. Terazono, H. Widjaja, Y. Takimoto, Electro- [41] T. Okada in Handbook of Fuel Cells, Vol. 3 (Eds. W. Viel-
chem. Solid-State Lett. 2004, 7, A209. stich, A. Lamm, H. A. Gasteiger), Wiley, Chichester,
[17] C. Wieser, Fuel Cells 2004, 4, 245. 2003, pp. 627.
[18] A. B. LaConti, M. Hamdan, R. C. McDonald in Handbook [42] C. Iojoiu, E. Guilminot, F. Maillard, M. Chatenet, J. Y.
of Fuel Cells, Vol. 3 (Eds. W. Vielstich, A. Lamm, H. A. Sanchez, E. Claude, E. Rossinot, J. Electrochem. Soc. 2007,
Gasteiger), Wiley, Chichester, 2003, pp. 647. 154, B1115.
[19] D. P. Wilkinson, J. St-Pierre in Handbook of Fuel Cells, [43] M. Cappadonia, J. W. Erning, S. M. Niaki, U. Stimming,
Vol. 3 (Eds. W. Vielstich, A. Lamm, H. A. Gasteiger), Solid State Ionics 1995, 77, 65.
Wiley, Chichester, 2003, pp. 611. [44] M. Cappadonia, J. W. Erning, U. Stimming, J. Electroa-
[20] J. Healy, C. Hayden, T. Xie, K. Olson, R. Waldo, nal. Chem. 1994, 376, 189.
R. Brundage, H. Gasteiger, J. Abbott, Fuel Cells 2005, 5, [45] Y. S. Kim, L. M. Dong, M. A. Hickner, T. E. Glass,
302. V. Webb, J. E. McGrath, Macromolecules 2003, 36, 6281.
[21] C. Huang, K. S. Tan, J. Lin, K. L. Tan, Chem. Phys. Lett. [46] R. C. McDonald, C. K. Mittelsteadt, E. L. Thompson,
2003, 371, 80. Fuel Cells 2004, 4, 208.
[22] D. E. Curtin, R. D. Lousenberg, T. J. Henry, P. C. Tange- [47] M. Saito, K. Hayamizu, T. Okada, J. Phys. Chem. B 2005,
man, M. E. Tisack, J. Power Sources 2004, 131, 41. 109, 3112.
[23] A. Panchenko, H. Dilger, J. Kerres, M. Hein, A. Ullrich, [48] A. Pozio, R. F. Silva, M. De Francesco, L. Giorgi, Electro-
T. Kaz, E. Roduner, Phys. Chem. Chem. Phys. 2004, 6, chim. Acta 2003, 48, 1543.
2891. [49] J. Xie, D. L. Wood, III, D. M. Wayne, T. A. Zawodzinski,
[24] H. Liu, H. A. Gasteiger, A. B. LaConti, J. Zhang, ECS P. Atanassov, R. L. Borup, J. Electrochem. Soc. 2005, 152,
Trans. 2006, 1, 283. A104.
[25] M. Aoki, H. Uchida, M. Watanabe, Electrochem. Com- [50] T. A. Aarhaug, S. Kjelstrup, S. Møller-Holst, 2006 Fuel
mun. 2005, 7, 1434. Cell Seminar Abstracts, Courtesy Associates, Hawaii,
[26] R. He, R. Vohra, 2006 Fuel Cell Seminar Abstracts, Cour- 2006.
tesy Associates, Hawaii, USA, 2006.

20 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

REVIEW
[51] R. Baldwin, M. Pham, A. Leonida, J. McElroy, T. Nal- [78] L. D. Burke, D. T. Buckley, J. Electroanal. Chem. 1994,
ette, J. Power Sources 1990, 29, 399. 366, 239.
[52] S. S. Kocha in Handbook of Fuel Cells-Fundamentals, Tech- [79] L. D. Burke, K. J. O’Dwyer, Electrochim. Acta 1992, 37,
nology and Applications, Vol. 3 (Eds. W. Vielstich, H. A. 43.
Gasteiger, A. Lamm), John Wiley and Sons, Chichester, [80] V. I. Birss, M. Chang, J. Segal, J. Electroanal. Chem. 1993,
UK, 2003, pp. 538. 355, 181.
[53] R. Makharia, M. F. Mathias, D. R. Baker, J. Electrochem. [81] R. M. Darling, J. P. Meyers, J. Electrochem. Soc. 2005, 152,
Soc. 2005, 152, A970. A242.
[54] S. D. Knights, K. M. Colbow, J. St-Pierre, D. P. Wilkin- [82] A. N. Virkar, Y. Zhou, J. Electrochem. Soc. 2007, 154,
son, J. Power Sources 2004, 127, 127. B540.
[55] M. Hori, M. Kato, Y. Yoshikawa, J. Yu, T. Matsuura, [83] E. Guilminot, A. Corcella, M. Chatenet, F. Maillard,
2005 Fuel Cell Seminar Abstracts, Courtesy Associates, F. Charlot, G. Berthome, C. Iojoiu, J. Y. Sanchez, E. Ros-
Palm Springs, 2005. sinot, E. Claude, J. Electrochem. Soc. 2007, 154, B1106.
[56] S. Y. Lee, E. Cho, J. H. Lee, H. J. Kim, T. H. Lim, I. H. [84] A. Honji, T. Mori, K. Tamura, M. Hishinuma, J. Electro-
Oh, J. Won, J. Electrochem. Soc. 2007, 154, B194. chem. Soc. 1988, 135, 355.
[57] B. Du, R. Pollard, J. Elter, 2006 Fuel Cell Seminar Ab- [85] K. Kinoshita, Electrochemical Oxygen Technology, John
stracts, Courtesy Associates, Hawaii, 2006, pp. 61. Wiley & Sons, Inc., New York, 1992, pp. 203.
[58] M. S. Wilson, J. A. Valerio, S. Gottesfeld, Electrochim. [86] R. L. Borup, J. R. Davey, F. H. Garzon, D. L. Wood,
Acta 1995, 40, 355. M. A. Inbody, J. Power Sources 2006, 163, 76.
[59] E. Cho, J. J. Ko, H. Y. Ha, S. A. Hong, K.-Y. Lee, T.-W. [87] M. F. Mathias, R. Makharia, H. A. Gasteiger, J. J. Conley,
Lim, I.-H. Oh, J. Electrochem. Soc. 2003, 150, A1667. T. J. Fuller, C. J. Gittleman, S. S. Kocha, D. P. Miller, C.
[60] Q. Guo, Z. Qi, J. Power Sources 2006, 160, 1269. K. Mittelsteadt, T. Xie, S. G. Yan, P. T. Yu, Electrochem.
[61] J. Hou, H. Yu, S. Zhang, S. Sun, H. Wang, B. Yi, P. Ming, Soc. Interface 2005, 14, 24.
J. Power Sources 2007, 162, 513. [88] P. Ascarelli, V. Contini, R. Giorgi, J. Appl. Phys. 2002, 91,
[62] J. Hou, H. Yu, B. Yi, Y. Xiao, H. Wang, S. Sun, P. Ming, 4556.
Electrochem. Solid-State Lett. 2006, 10, B11. [89] K. Kinoshita, Carbon. Electrochemical and Physicochemical
[63] J. St Pierre, J. Roberts, K. Colbow, S. Campbell, A. Nel- Properties, John Wiley & Sons, Inc., New York, 1988,
son, J. New Mater. Electrochem. Syst. 2005, 8, 163. pp. 316.
[64] L. Mao, C. Y. Wang, J. Electrochem. Soc. 2007, 154, B139. [90] K. H. Kangasniemi, D. A. Condit, T. D. Jarvi, J. Electro-
[65] M. Oszcipok, A. Hakenjos, D. Riemann, C. Hebling, Fuel chem. Soc. 2004, 151, E125.
Cells 2007, 7, 135. [91] N. Giordano, P. L. Antonucci, E. Passalacqua, L. Pino,
[66] K. Tajiri, Y. Tabuchi, F. Kagami, S. Takahashi, K. Yoshi- A. S. Arico, K. Kinoshita, Electrochim. Acta 1991, 36,
zawa, C. Y. Wang, J. Power Sources 2007, 165, 279. 1931.
[67] J. Xie, D. L. Wood, III, K. L. More, P. Atanassov, R. L. [92] S. C. Ball, S. L. Hudson, D. Thompsett, B. Theobald,
Borup, J. Electrochem. Soc. 2005, 152, A1011. J. Power Sources 2007, 171, 18.
[68] E. Guilminot, A. Corcella, F. Charlot, F. Maillard, [93] J. Willsau, J. Heitbaum, J. Electroanal. Chem. Interfacial
M. Chatenet, J. Electrochem. Soc. 2007, 154, B96. Electrochem. 1984, 161, 93.
[69] P. Bindra, S. J. Clouser, E. Yeager, J. Electrochem. Soc. [94] L. M. Roen, C. H. Paik, T. D. Jarvi, Electrochem. Solid-
1979, 126, 1631. State Lett. 2004, 7, A19.
[70] J. Aragane, T. Murahashi, T. Odaka, J. Electrochem. Soc. [95] Z. Siroma, K. Ishii, K. Yasuda, Y. Miyazaki, M. Inaba,
1988, 135, 844. A. Tasaka, Electrochem. Commun. 2005, 7, 1153.
[71] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous [96] D. A. Stevens, J. R. Dahn, Carbon 2005, 43, 179.
Solutions, National Association of Corrosion Engineers, [97] D. A. Stevens, M. T. Hicks, G. M. Haugen, J. R. Dahn,
New York, 1974, pp. 379. J. Electrochem. Soc. 2005, 152, A2309.
[72] R. M. Darling, J. P. Meyers, J. Electrochem. Soc. 2003, 150, [98] M. Cai, M. S. Ruthkosky, B. Merzougui, S. Swathirajan,
A1523. M. P. Balogh, S. E. Oh, J. Power Sources 2006, 160, 977.
[73] V. A. T. Dam, F. A. de Bruijn, J. Electrochem. Soc. 2007, [99] M. Schulze, N. Wagner, T. Kaz, K. A. Friedrich, Electro-
154, B494. chim. Acta 2007, 52, 2328.
[74] X. Wang, R. Kumar, D. J. Myers, Electrochem. Solid-State [100] L. R. Jordan, A. K. Shukla, T. Behrsing, N. R. Avery,
Lett. 2006, 9, A225. B. C. Muddle, M. Forsyth, J. Power Sources 2000, 86, 250.
[75] Z. Nagy, H. You, Electrochim. Acta 2002, 47, 3037. [101] M. V. Williams, E. Begg, L. Bonville, H. R. Kunz, J. M.
[76] S. Kawahara, S. Mitsushima, K. Ota, N. Kamiya, ECS Fenton, J. Electrochem. Soc. 2004, 151, A1173.
Trans. 2006, 3, 625. [102] U. Pasaogullari, C. Y. Wang, Electrochim. Acta 2004, 49,
[77] S. Kawahara, S. Mitsushima, K. Ota, N. Kamiya, ECS 4359.
Trans. 2006, 1, 85. [103] U. Pasaogullari, C. Y. Wang, K. S. Chen, J. Electrochem.
Soc. 2005, 152, A1574.

FUEL CELLS 08, 2008, No. 1, 3–22 www.fuelcells.wiley-vch.de © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim 21
de Bruijn et al.: Durability and Degradation Issues of PEM Fuel Cell Components

[104] A. Z. Weber, R. M. Darling, J. Newman, J. Electrochem. [121] R. Borup, J. Meyers, B. Pivovar, Y. S. Kim, R. Mukun-
REVIEW

Soc. 2004, 151, A1715. dan, N. Garland, D. Myers, M. Wilson, F. Garzon,


[105] A. Z. Weber, J. Newman, J. Electrochem. Soc. 2005, 152, D. Wood, P. Zelenay, K. More, K. Stroh, T. Zawodzins-
A677. ki, J. Boncella, J. E. McGrath, M. Inaba, K. Miyatake,
[106] C. Lee, W. Merida, J. Power Sources 2007, 164, 141. M. Hori, K. Ota, Z. Ogumi, S. Miyata, A. Nishikata,
[107] T. Yoda, H. Uchida, M. Watanabe, Electrochim. Acta Z. Siroma, Y. Uchimoto, K. Yasuda, K. I. Kimijima,
2007, 52, 5997. N. Iwashita, Chem. Rev. (Washington, DC, U. S.) 2007,
[108] S.-Y. Lee, E. Cho, J.-H. Lee, H.-J. Kim, T.-H. Lim, I.-H. 107, 3904.
Oh, 2005 Fuel Cell Seminar Abstracts, Courtesy Associ- [122] H. A. Gasteiger, S. S. Kocha, B. Sompalli, F. T. Wagner,
ates, Palm Springs, 2005. Appl. Catal., B 2005, 56, 9.
[109] T. R. Ralph, M. P. Hogarth, Platinum Metals Rev. 2002, [123] M. K. Debe in Handbook of Fuel Cells-Fundamentals, Tech-
46, 117. nology and Applications, Vol. 3 (Eds. W. Vielstich, H. A.
[110] P. Yu, M. Pemberton, P. Plasse, J. Power Sources 2005, Gasteiger, A. Lamm), John Wiley and Sons, Chichester,
144, 11. UK, 2003, pp. 576.
[111] W. Bi, G. E. Gray, T. F. Fuller, Electrochem. Solid-State [124] A. Hermann, T. Chaudhuri, P. Spagnol, Int. J. Hydrogen
Lett. 2007, 10, B101. Energy 2005, 30, 1297.
[112] T. Patterson, Fuel Cell Technology: Opportunities and Chal- [125] R. C. Makkus, A. H. H. Janssen, F. A. de Bruijn,
lenges, Topical Conference Proceeding, 2002 AIChE Spring R. K. A. M. Mallant, J. Power Sources 2000, 86, 274.
National Meeting (Eds. G. J. Igwa and D. Mah), AIChE, [126] D. P. Davies, P. L. Adcock, M. Turpin, S. J. Rowen,
New York, USA, 2002, pp. 313. J. Appl. Electrochem. 2000, 30, 101.
[113] K. Kanamura, H. Morikawa, T. Umegaki, J. Electrochem. [127] A. Mueller, P. Kauranen, A. von Ganski, B. Hell, J. Power
Soc. 2003, 150, A193. Sources 2006, 154, 467.
[114] C. A. Reiser, L. Bregoli, T. W. Patterson, J. S. Yi, J. D. [128] R. F. Silva, D. Franchi, A. Leone, L. Pilloni, A. Masci,
Yang, M. L. Perry, T. D. Jarvi, Electrochem. Solid-State A. Pozio, Electrochim. Acta 2006, 51, 3592.
Lett. 2005, 8, A273. [129] A. Iversen, Corros. Sci. 2006, 48, 1036.
[115] J. P. Meyers, R. M. Darling, J. Electrochem. Soc. 2006, 153, [130] F. A. de Bruijn, R. C. Makkus, R. K. A. M. Mallant,
A1432. G. J. M. Janssen in Advances in Fuel Cells, 1 (Eds. T. S.
[116] J. Zhang, J. Owejan, M. Fay, B. A. Litteer, P. T. Yu, Zhao, K. D. Kreuer, T. Nguyen), Elsevier, Oxford, Uni-
W. Gu, H. A. Gasteiger, Prepr. Symp. – Am. Chem. Soc., ted Kingdom 2007, pp. 235.
Div. Fuel Chem. 2007, 52, 388. [131] L. Frisch, Sealing Technol. 2001, 2001, 7.
[117] P. T. Yu, W. Gu, F. T. Wagner, H. A. Gasteiger, Prepr. [132] B. Du, Q. Guo, R. Pollard, D. Rodriguez, C. Smith, J. El-
Symp. – Am. Chem. Soc., Div. Fuel Chem. 2007, 52, 386. ter, JOM 2006, 58, 45.
[118] A. Taniguchi, T. Akita, K. Yasuda, Y. Miyazaki, J. Power [133] J. Tan, Y. J. Chao, J. W. Van Zee, W. K. Lee, Mater. Sci.
Sources 2004, 130, 42. Eng. A 2007, 445–446, 669.
[119] T. W. Patterson, R. M. Darling, Electrochem. Solid-State [134] M. Schulze, T. Knöri, A. Schneider, E. Gülzow, J. Power
Lett. 2006, 9, A183. Sources 2004, 127, 222.
[120] E. Cho, J. J. Ko, S.-A. Ha, S. A. Hong, K.-Y. Lee, T.-W.
Lim, I.-H. Oh, J. Electrochem. Soc. 2004, 151, A661.

______________________

22 © 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.fuelcells.wiley-vch.de FUEL CELLS 08, 2008, No. 1, 3–22

You might also like