Advanced Review

A review of PEM fuel cell

durability: materials degradation,
local heterogeneities of aging and
possible mitigation strategies
Laetitia Dubau,1 Luis Castanheira,1 Frédéric Maillard,1∗ Marian
Chatenet,1† Olivier Lottin,2 Gaël Maranzana,2 Jérôme Dillet,2 Adrien
Lamibrac,2 Jean-Christophe Perrin,2 Eddy Moukheiber,1 Assma
ElKaddouri,1 Gilles De Moor,1 Corine Bas,1 Lionel Flandin1 and
Nicolas Caqué3

Through a tight collaboration between chemical engineers, polymer scientists, and

electrochemists, we address the degradation mechanisms of membrane electrode
assemblies (MEAs) during proton exchange membrane fuel cell (PEMFC) oper-
ation in real life (industrial stacks). A special attention is paid to the heteroge-
neous nature of the aging and performances degradation in view of the hard-
ware geometry of the stack and MEA. Macroscopically, the MEA is not fuelled
evenly by the bipolar plates and severe degradations occur during start-up and
shut-down events in the region that remains/becomes transiently starved in hydro-
gen. Such transients are dramatic to the cathode catalyst layer, especially for the
carbon substrate supporting the Pt-based nanoparticles. Another level of hetero-
geneity is observed between the channel and land areas of the cathode catalyst
layer. The degradation of Pt3 Co/C nanocrystallites employed at the cathode can-
not be avoided in stationary operation either. In addition to the electrochemical
Ostwald ripening and to crystallite migration, these nanomaterials undergo severe
corrosion of their high surface area carbon support. The mother Pt3 Co/C nanocrys-
tallites are continuously depleted in Co, generating Co2+ cations that pollute the
ionomer and depreciate the performance of the cathode. Such cationic pollution
has also a negative effect on the physicochemical properties of the proton-exchange
membrane (proton conductivity and resistance to fracture), eventually leading
to hole formation. These defects were localized with the help of an infrared
camera. The mechanical fracture-resistance of various perfluorosulfonated mem-
branes further demonstrated that polytetrafluoroethylene-reinforced membranes
better resist hole formation, due to their high resistance to crack initiation and
propagation. © 2014 John Wiley & Sons, Ltd.

How to cite this article:

WIREs Energy Environ 2014. doi: 10.1002/wene.113

∗ Correspondence to: [email protected]

1 Laboratoire d’Electrochimie et de Physico-chimie des Matériaux et des Interfaces, UMR 5279 CNRS/Grenoble-INP/Université de
Savoie/Université Joseph Fourier, 38402 Saint Martin d’Hères Cedex, France
2 LEMTA, UMR CNRS 7563, CNRS/Université de Lorraine, 54518 Vandoeuvre-lès-Nancy, France

3 AXANE, 2 rue de Clémencière, BP 15, 38360 Sassenage, France

† Chatenet is a member of the French University Institute (IUF).

Conflict of interest: The authors have declared no conflicts of interest for this article.

© 2014 John Wiley & Sons, Ltd.

 Advanced Review wires.wiley.com/wene

INTRODUCTION The present contribution addresses changes in

structure and composition of the various MEA con-
P roton exchange membrane fuel cells (PEMFC)
convert the chemical energy of a fuel into
electricity with minimal pollutant generation,
stituents caused by PEMFC operation in real life (com-
mercial materials operated in industrial stacks) at the
macro and micrometer scale. In particular, the con-
and are therefore regarded amongst the most sequences of the long-term operation of the PEMFC
promising electrical generators for stationary and stack in terms of physical-chemistry of the materials,
mobile applications.1,2 Their core materials are electrochemistry as well as mass-transport and heat
platinum-based carbon-supported nanoparticles, to management are dealt with. The results are empha-
accelerate the rate of the sluggish oxygen-reduction sized at the stack, MEA and materials scales, and a
reaction (ORR) and enable the hydrogen oxidation particular focus is made to bridge the materials (of the
reaction (HOR), and a perfluorosulfonated membrane catalytic layer, CL and proton exchange membrane,
(usually Nafion), to electronically and chemically sep- PEM) degradation to the MEA/system performances
arate the anode from the cathode whilst enable degradations. The heterogeneous nature of the aging
protons conduction. Thanks to nearly five decades and performances degradation is also addressed in
of scientific and technological improvements of these view of the hardware geometry of the stack and MEA.
core materials, their assembly, and the architecture
of the overall system, the PEMFC initial perfor-
mances have reached sufficient levels for industrial
THE DEGRADATION OF THE PEMFC
deployment. Nevertheless, their durability in real
life is still open to question, and remains essential
MEA MATERIALS AND OF THE
for widespread industrial commercialization. The ELECTRICAL PERFORMANCES ARE
aging of the PEMFC system (Figure 1(a)), mainly HETEROGENEOUS
depends on the membrane-electrode assembly (MEA, The operating conditions in a PEMFC are by nature
Figure 1(b)), itself related to the PEM and electro- heterogeneous because the concentration in oxygen
catalyst materials (Figure 1(c)). The consequences and hydrogen diminishes along the channels while
of the PEMFC materials degradation are multiple. gases enrich in water. Furthermore, depending on the
They firstly affect the system gross performance, its quality of the cooling circuit, the temperature may not
fuel consumption per kWh generated, and therefore be uniform over the MEA surface. Although model-
its competitiveness to present delocalized genera- ing can put forward some local variations in current
tors (e.g., thermal engines). Secondly, it can affect density, gas concentration, and liquid water content
the reliability and the availability of the systems in PEMFC,16,17 both the diversity and the complexity
(in other words, the maintenance cost). Thirdly, it of the physical phenomena taking place in membrane
can yield safety issues, related to the use of H2 . electrodes assemblies (not to mention the gas-liquid
There is extensive literature dealing with durability flows in the bipolar plate channels) makes experi-
of PEMFC materials and systems (see for example mental investigations indispensable. Numerous visu-
Refs 2–11), and many point out toward the strong alization techniques have been used to monitor water
influence of operating conditions (load cycle, start-up in a running PEMFC, from optical photography18,19
and shut-downs, fluid and temperature management, and magnetic resonance imaging20 (MRI) to neutron21
etc.) on the nature and extent of the degradations. and synchrotron X-ray imaging.22 However, visualiza-
Recently the heterogeneous nature of the aging tion techniques mostly provide qualitative informa-
was also put forth. This is easily understandable if tion about the temporal and spatial distribution of
one considers that PEMFC are complex structures water inside a running fuel cell. In contrast, quanti-
involving complex materials (Pt-based nanopar- tative information on local operation inside an ele-
ticles, carbon, ionomer, membrane, PTFE-carbon mentary cell may be obtained using segmented cells.
gas-diffusing media, bipolar plates) yielding differ- Many groups have developed their own tools23 but
ent local values of electrocatalyst utilization and fundamentally, a segmented cell is an ordinary sin-
effectiveness factors, and that PEMFC operation gle cell with one of the electrodes divided into smaller
induces local heterogeneities in fluidics (presence of electrodes, where various electrochemical signals (e.g.,
water and gas fluids), performances and therefore current density, impedance, or even potential and
aging.12,13 As a consequence, the determination of active surface area), can be measured independently
the mechanisms at stake during PEMFC long-term of the other segments (Figure 2). Although it remains
degradation in real operation is extremely difficult to an invasive technique, segmented cells are increasingly
assess. used for in situ investigation of PEMFC operation, to

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

(a) Auxiliary commands Power electronics

and control systems (conversion)

Frame Fuel cell stack

Air compressor

Humidifier
Battery Gas distribution

(b) (c)

25 μm
Signal A = BSD Signal B = InLens Date: 16 Jun 2009 INPG
WD = 8.9 mm Mixing = Off Time :17:06:37 CMTC 20 nm
Signal = 1.000

FIGURE 1 | (a) Exploded view of a PEMFC system (without the H2 tank). Courtesy of AXANE fuel cell systems (http://www.axane.fr/). (b) SEM
image of a commercial membrane-electrode assembly, Reproduced with permission from Ref 14. Copyright 2010, The Electrochemical Society. (c)
Transmission electron microscopy (TEM) image of the Pt-Co/C cathode electrocatalyst. (Reproduced with permission from Ref 15. Copyright 2013,
Elsevier)

design flow fields, or to optimize the electrical perfor- conditions were initially presented by Reiser et al.30
mances of the system.25–27 In some cases, segmented and, to the best of our knowledge, Siroma et al.28 and
cells can be used in parallel with a visualization tech- Maranzana et al.29 were the first to succeed in mea-
nique, for instance to study the impact of channel suring internal currents in PEMFC. An illustration of
flooding on the local current density.19 reverse currents during start-up of a PEMFC is given
The use of segmented cells recently enabled in Figure 3.
to understand the origin of local degradations in Internal currents, such as those presented in
PEMFC in occasions these measurements were further Figure 3, are transient phenomena that occur when
completed by post mortem analyses2,12,13,28,29 — and the fuel cell is in open circuit and when hydrogen
allowed bridging them with the local operating condi- is injected in an anode compartment initially filled
tions. One of the most spectacular phenomenon they with air or nitrogen: the cell is temporally divided
contributed to unveil is the reverse currents that set into an active part and a passive part, delimited by
up during cell start-up or shut-down under open cir- the hydrogen front. The active part operates normally,
cuit conditions. The main mechanisms at work in these with hydrogen oxidation at the anode and oxygen

© 2014 John Wiley & Sons, Ltd.

 Advanced Review wires.wiley.com/wene

FIGURE 2 | Segmented cell with five straight parallel channels used in Ref 24. The flow field plate is divided into 20 segments and it can be used
either on the cathode side or on the anode side; the other plate is not segmented. The water circuit used to control the cell temperature flows through
the aluminium end-plate located under the gold-plated flow field. (Reproduced with permission from Ref 24. Copyright 2013, Elsevier)

1.2 1000 Figure 3 highlights that the intensity of reverse

First segment
currents remains small in absolute value; this is espe-
1.0 800

Local current density / mA cm−2

supplied with H2 cially true when compared to the magnitude of the
PEM fuel cell voltage / V

0.8 600
(forward) currents measured when the cell is in nor-
mal operation. However, the reverse currents can be
Last segment
0.6 supplied with H2 400 very damaging to the membrane-electrodes assembly,
because they are associated to high potential excur-
0.4 200
sions on the cathode side, especially in the segments
located near the hydrogen outlet in start-up or the
0.2 0
hydrogen inlet in shut-down (i.e., those staying for
0.0 −200 the longest amount of time in the presence of air in
0 1 2 3 4 the anode compartment). As shown in Figure 5, one
Time / s
half of the anode compartment (region A) is filled with
FIGURE 3 | PEMFC voltage (blue curve) and current densities hydrogen, while the other is still filled with air (region
through each of the 20 segments of the segmented cell presented in B) during a start-up in air. According to Reiser et al.,30
Figure 2, as a function of time. The fuel cell was started-up by injecting three potentials can be expressed with reference to the
10 standard liters per hour (slph) of hydrogen in the anode standard hydrogen electrode:
compartment initially filled with air.
a
• The anode metal potential Vm .
c
• The cathode metal potential Vm .
reduction at the cathode (Figure 4). However, in the
passive part, i.e., the part not yet supplied with hydro- • The electrolyte potential V e (𝜙 in Figure 5). V e is
gen, reverse currents are generated. Similar phenom- assumed constant in the direction perpendicular
ena occur during the fuel cell shut-down, as well as to the electrolyte.
when injecting oxygen or air in a nitrogen-filled cath-
ode compartment.31 It must be noted that connecting Owing to the high conductivity of the catalyst layer
the fuel cell to a load while injecting hydrogen enables and gas diffusing layer materials at the anode and
a c
limiting the intensity of negative currents and there- cathode, Vm and Vm are uniform and equal to approx-
fore mitigating the irreversible degradations occurring imately 0 V versus the reversible hydrogen electrode
during such transients. Nevertheless, this solution is (RHE) and 0.85 V versus RHE (typical values corre-
generally avoided when operating a fuel cell stack, sponding, for instance, to those measured at t = 3.5 s
because it would require that the hydrogen enters at in Figure 3) in both regions, respectively. However,
the same instant in every individual cell: if not, some the electrolyte being highly resistive in the direction
cells would be (fully or partially) active while some of the gas flow (in the plane of the MEA, with char-
others would still be passive; since the (average) cur- acteristic dimensions on the order of the centimeters
rent density flowing through the stack is identical in or tens of centimeters, compared to a few tens μm
all cells, passive cells would suffer temporarily high perpendicular to the MEA surface), its potential is
oxidative stress because of the presence of air in both nonuniform. Upstream the hydrogen front (active part
the anode and cathode compartments: in other words in Figure 4 or region A in Figure 5), V e stays close
a
the passive cell would suffer complete inversion and to Vm , while downstream (passive part in Figure 4 or
therefore large extent of degradation. region B in Figure 5), V e drops to −0.593 V versus

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

(a) H2 inlet V Air outlet

(b) H2 inlet V Air outlet

Active part Passive part

C + 2H2O → CO2 + 4H+ + 4e−
2H2 → 4H+ + 4e− O2 + 4H+ + 4e− → 2H2O O2 + 4H+ + 4e− → 2H2O Pt + H2O → PtO + 2H+ + 2e−
Pt + H2O → PtO + 2H+ + 2e− Pt → Pt2+ + 2e−
Other reactions? 2H2O → 4H+ + 4e− + O2
Other reactions?

C + 2H2O → CO2 + 4H+ + 4e−

O2 + 4H+ + 4e− → 2H2O Pt + H2O → PtO + 2H+ + 2e−
Pt → Pt2+ + 2e− 2H2 → 4H+ + 4e− O2 + 4H+ + 4e− → 2H2O
PtO + 2H+ + 2e− → Pt + H2O
2H2O → 4H+ + 4e− + O2
Other reactions? Other reactions?

Passive part Active part

H2 H2
H2 outlet Air Air inlet H2 outlet Air Air inlet

FIGURE 4 | Schematics of the active and passive part of a cell during (a) start-up and (b) shut-down, and possible electrochemical reactions
occurring during the transients. In that case the heterogeneities generated are at the level of the centimeters to tens of centimeter scale. (Reproduced
with permission from Ref 24. Copyright 2013, Elsevier)

x C + 2H2O -> CO2 + 4H+ + 4e−

O2 + 4H+ + 4e− -> 2H2O e− 2H2O -> O2 + 4H+ + 4e−
Cathode
H+
Region A Region B Membrane
+
H Anode
H2 -> 2H+ + 2e− e− O2 + 4H+ + 4e− -> 2H2O
0.9 Vmo = 0.850 V
Region A Region B
0.7
0.5 Vmc − Φ = 0.836 V Vmc − Φ = 1.443 V
Potential (V)

0.3
0.1
Vma = 0.000 V
−0.1
Vma − Φ = 0.002 V Vma − Φ = 0.593 V
−0.3
−0.5
Φ = −0.593 V
−0.7
0.00 0.02 0.04 0.06 0.08 0.10
Distance from anode inlet (cm)

FIGURE 5 | Potential distribution in a cell during a start-up in air with reference to the RHE. (Reproduced with permission from Ref 30. Copyright
2005, The Electrochemical Society)

RHE, which results in a high interfacial potential dif- diffuse under the land. During this time, a transient
ference between the cathode metal and the electrolyte. nonuniformity in gas concentration sets up within the
Such a potential difference in the cathode acceler- catalyst layer, that can lead to internal currents at the
ates carbon and catalyst corrosion, which leads to millimeter scale36 (i.e. between channels and lands)
some of the degradations described in Section Degra- and subsequently, to another (much smaller) level
dation of the Catalytic Layers. Verifying experimen- of aging heterogeneity.13 Using physicochemical and
tally the numerical results of Reiser et al.30 requires electrochemical techniques, Durst et al.13 did put into
introducing one or several reference electrode(s) in a evidence the much larger extent of degradation of
running cell. Using this methodology, Shen et al.,32 the cathode catalyst layer materials occurring under
Hinds and Brightman33 and Lamibrac et al.34,35 the land (facing the land of the anode bipolar plate).
experimentally confirmed the numerical predictions of These harsh degradations do result from the increased
Reiser et al.30 fuel starvation transients in the land region, thereby
Another issue to be mentioned about start-up experimentally attesting the dramatic local degrada-
and shut-down procedures concerns the time needed tions induced by the internal currents measured at the
by hydrogen (at start-up) or oxygen (at shut-down) to millimeter scale by Schneider et al.36

© 2014 John Wiley & Sons, Ltd.

 Advanced Review wires.wiley.com/wene

There are currently two main perspectives in in larger modules (stack). Of course, in addition to
the use of instrumented and segmented cells. The these local heterogeneities, one must consider that
first one is to explore the degradation mechanisms in the MEA is a very complex assembly of materials
other situations than start-up and shut-down events, (membrane, ionomer, hydrophobic binder, electrocat-
for instance when a fuel cell is operated in dead-end alyst nanoparticles, carbon powder and fiber, etc.),
mode.37 Dead-end mode operation may lead to local which can evidently not undergo the same degrada-
fuel starvation due to the excessive accumulation tion phenomena upon use. This induces ‘materials’
of liquid water and possibly nitrogen and oxygen heterogeneities; the next two sections address this
(because of membrane crossover) in the anode com- extremely important issue.
partment therefore yielding significant rise of the
anode (and thus cathode) potentials. More gener-
ally, segmented cells can be used to analyze the local DEGRADATION OF THE CATALYTIC
operation of a cell during any kind of transient or LAYERS
steady-state operation. The main challenge in these
cases consists of finding reliable markers of the degra- The CLs of PEMFC are multicomponent/multifunc-
tional media. They consist of the tailored assembly
dation of the electrolyte membrane: up to now, instru-
of (1) carbon-supported electrocatalysts nanoparti-
mented cells were mostly used to monitor the degra-
cles: Pt-M/C to promote both the electrochemical reac-
dation of the electrodes.
tions (Pt-M, where M is an alloying element of Pt,
The second perspective concerns the investiga-
e.g., Co, Ni, Cu, etc.) and the electronic conductiv-
tion of the temperature over the MEA surface and
ity to/from the current collector (C), (2) ionomer:
through its thickness. Contrary to local variations in
Nafion, to enable the conduction of the required pro-
gas concentration, the impact of temperature hetero-
tons to the reaction and (3) porosity, mandatory for
geneities is far from being well understood, know-
mass-transport of the reactants and products to/from
ing all MEA components (membrane and electrode
the active sites. Not only these materials must be opti-
materials) degradation phenomena, reaction mech-
mized individually to reach sufficient performances,
anisms/kinetics and mass-transport phenomena are
their coexistence/interplay must also be tightly con-
in one way or another dependent on that param-
trolled to reach/maintain performances compatible
eter. Nevertheless, some recent experimental results
with on-site PEMFC operation. While the initial per-
show a positive correlation between the local cur-
formances of PEMFC have now proved to be suffi-
rent densities and temperatures.29,38,39 Temperature
cient for their widespread deployment, their durability
variations are also significant in the direction perpen-
has been under intense focus since the 2000s.3,4,6–9
dicular to the MEA flat surface, with a significant
Indeed, upon real and long-term operation, the CLs
impact on water transport, which now starts to be bet-
of PEMFC MEA are degraded, such degradation
ter understood. Djilali and Lu40 numerically demon-
being multiple in scales, materials affected and con-
strated that the temperature difference between the
sequences, as detailed in the following sections. For-
bipolar plate and the cathode could reach 1 to 5 K,
mer studies revealed that the cathode CLs based on
depending on the current density and thermo-physical
Pt/C electrocatalysts were the most prone to degra-
properties of the materials. Such temperature dif-
dation (compared to anode CLs), owing to the more
ferences were confirmed experimentally by Vie and
severe oxidizing conditions encountered and the wider
Kjelstrup.41 In addition, Weber and Newman,42 Wang
range of potential variation in operation.7–10,46–49 For
and Wang43 and Eikerling44 showed that evapo-
that reason, the focus will be made on the cathode
ration/condensation (heat pipe effect) through the
in the present section. Most of the results presented
porous layers could contribute significantly to the
herein were obtained for MEAs operated at I = 50 A
evacuation of water from the electrodes toward the
(i.e., j = 0.60 A cm−2 ), T = 70∘ C, 65% RH after var-
bipolar plates, which was confirmed experimentally
ious durations of operation in PEMFC 16-cell stack
by Thomas et al.45
designed and operated by Axane fuel cell systems. The
This selected literature review demonstrated cathode material was a commercial (and proprietary)
that a PEMFC MEA cannot be considered as a Pt-Co/C electrocatalyst.
medium that behaves and ages homogeneously in
steady-state continuous or load-cycle operation.
Indeed, heterogeneities set up locally in the assembly, At the Micrometer Scale
due to the ‘open nature’ of the fuel cell, i.e., the Electrode Thicknesses
fact that a fuel cell bares an inlet and an outlet for Figure 6 shows representative images of the MEAs in
each single cell, and that these might be assembled their initial state and after various aging times up to

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

3422 h recorded with a scanning electron microscopy The degradation of the cathodic electrocata-
equipped with a field emission gun (FEG-SEM). The lyst is tightly linked to their thermodynamic instabil-
cathode suffers drastic and continuous thinning over ity. Firstly, carbon is only metastable above 0.207 V
time, ultimately yielding depletion or detachment from versus RHE, i.e., seriously corrodes at the cathode
the PEM in some regions. By comparison, the anode CL on the long-term.15,50,53,54 Secondly, the corro-
remains essentially unaffected, except at t = 3422 h for sion of the carbon support promotes the motion of
which it is no longer attached to the PEM (but this the metal nanoparticles, until they stabilize (upon
likely originates from the severe MEA degradation collision/agglomeration with other nanoparticles and
at that stage and not to degradations specific to the possibly their coalescence). Thirdly, Pt, although a
anode). The cathode thickness, initially of 13.5 μm for noble metal, is prone to corrosion,55 especially in
the fresh MEA, decreases by nearly a factor 3 (down the environment of an operating cathode in PEMFC
to 5 μm) and 4 (down to 3.5 μm) after 1163 h and around OCP (Open Circuit Potential) or small cur-
2259 h of operation, respectively. The severe aging rent load.47 The nanometer-scale of the metal crys-
after 3422 h of operation is demonstrated by the local tallites emphasizes such degradation: indeed due to
absence or severe cracking of the cathode, which pre- the Gibbs-Thomson equation, smaller nanoparticles
vents any accurate measurement of its thickness at corrode more easily into metal ions than larger ones,
that stage. It is wise noting that both the thinning and which are more stable.49 This can yield coarsening of
creation of defects (like cracks in the cathode CL or the metal nanoparticles within the cathode CL (as very
decohesion between the cathode CL and the PEM) often observed experimentally46–49 ) or, fourthly, to the
seem linked and emphasized at long durations of the redistribution of the Ptz+ ionic species (z = 2 or 4) into
PEMFC operation. These macroscopic effects are the the whole MEA, including the PEM where Pt crys-
consequence of a microscopic phenomenon, i.e., car- tallites can form when cross-over H2 is present,47,48
bon corrosion, as detailed later (see Section Carbon as detailed in the previous section. Although Pt-M/C
corrosion). The corrosion of the carbon support is electrocatalysts were initially considered more sta-
strongly enhanced by start-up/shut-down events dur- ble than Pt/C,1,2,11 these processes also happen for
ing which oxygen (air) and hydrogen are simultane- Pt-M/C-based CLs14,56,57 (see e.g., the case of Pt-Cu/C
ously present in the anode compartment,30,34,50 there- cathodic electrocatalysts, for which Cu2+ ions can
fore yielding local heterogeneities of aging (the air cross over the PEM and be reduced in the PEM or at
inlet/H2 outlet is much more degraded than the air the anode).58
outlet/H2 inlet after several tens of events).
Carbon Corrosion
Presence of a Pt-band in the PEM
The degradation of the carbon support of the Pt-M
Another feature that is evident in Figure 6 is the
nanoparticles greatly participates to the performances
presence of a Pt particles band in the PEM over
losses of a PEMFC. The carbon corrosion induces:
long-term operation. These particles originate from
(1) the loss of electrochemical surface area (ECSA),
the corrosion of the nanometer-sized electrocatalysts
following the detachment/agglomeration of the metal
of the cathode CL (Pt/C or Pt-M/C), which produces
nanoparticles,50 (2) the decrease of the hydrophobicity
metal ions (e.g., Pt2+ , Co2+ ) that are transported into
of the CL (because carbon corrosion initiates by the
the PEM and then reduced by the H2 molecules that
formation of oxygen surface groups at its surface,
cross-over from the anode side.46–48 The particles in
that mostly are hydrophilic53 ), affecting the water
the PEM are only composed of metallic Pt (no Co is
management of the whole MEA, and (3) losses in
present in the solid form in the PEM when Pt-Co/C
porosity of the CL through the collapsing of the
electrocatalyst is used in the cathode CL), because the
carbon structure, which hinders mass-transport of gas
enthalpy of mixing of Pt-Co alloys (approximately
reactants toward the active sites and water removal
−10 kJ mol−1 , see Ref 51) is too low to significantly
from the electrode. All these issues do emphasize the
upshift the Co/Co2+ redox potential. Here again, it is
need to efficiently survey carbon corrosion in CL
worth noting that the microscale observation of the Pt
materials.
band in the PEM is related to phenomena that proceed
at the nanoscale in the cathode CL, as detailed below. The Raman spectra of Figure 8(a) present typi-
cal characteristics of carbon black materials: the weak
band at 1595 cm−1 is ascribed to the ordered graphitic
At the Nanometer Scale structure (G band) while the disorder-induced bands
Figure 7 provides a schematic vision of the mecha- (D at 1355 cm−1 and D′ at 1615 cm−1 ) are attributed
nisms of degradation of Pt/C electrocatalysts, which to defects in the graphite structure.50,59–61 Although
are currently admitted in the literature.5,7,9,10,47–49 these characteristic bands are present for the fresh

© 2014 John Wiley & Sons, Ltd.

 Advanced Review wires.wiley.com/wene

Fresh MEA t = 1163 h

Cathode side Cathode side

(thickness 13.5 μm) (thickness 5.0 μm)
10 μm Mag = 1.00 K X Date :14 Nov 2008 INPG 10 μm Grenoble INP - CMTC
Signal A = QBSD Signal B = InLens Mag = 1.00 K X Signal A = QBSD
EHT = 20.00 kV WD = 8 mm Mixing = Off Time :15:50:12 CMTC EHT = 15.00 kV WD = 7.9 mm Date :3 Feb 2012
Signal = 1.000

t = 2259 h t = 3422 h

Cathode side Cathode side

(thickness 3.5 μm) (thickness not quantifiable)
10 μm Mag = 1.00 K X Signal A = QBSD Grenoble INP - CMTC 10 μm Mag = 1.00 K X Signal A = QBSD Grenoble INP - CMTC
EHT = 15.00 kV WD = 6.3 mm Date :3 Feb 2012 EHT = 15.00 kV WD = 7.6 mm Date :3 Feb 2012

FIGURE 6 | FEG-SEM images obtained in back-scattered electrons for MEAs operated at j = 0.60 A cm−2 , T = 70∘ C, 65% RH after various
durations of operation in PEMFC 16-cell stack. (Reproduced with permission from Ref 15. Copyright 2013, Elsevier)

carbon support, mainly is the intensity of the D increasingly composed of graphitic domains over time,
band depreciated during PEMFC operation. This sug- but the trend is mildly reversing at t = 3422 h, as a
gests that disordered carbon domains corrode pref- result of the carbon amorphization mentioned above.
erentially, emphasizing the proportion of ordered
graphitic domains over time. In particular, the high Crystallite Migration and Agglomeration
frequency band narrows after 2259 h, following the Figure 9 presents particle-size distributions and rep-
near-complete extinction of the D′ band (Figure 8(a)). resentative TEM images of the Pt-Co/C cathode cat-
Nevertheless, the enhanced stability of graphite is not alyst after PEMFC operation. The center of the
a thermodynamic warranty (it is just the result of Pt-Co/C particle size histogram shifts toward larger
the slowest kinetics of degradation of graphite versus particle sizes by approximately 1.5 nm upon aging
amorphous carbon) and the G band also decreases in and a long tailing toward larger particle sizes con-
intensity after 3422 h of operation, as confirmed by comitantly develops. Also, the fraction of aggre-
the attenuation of the intensity of the (002) reflection gated Pt-Co/C nanoparticles progressively increases,
of graphite at 2𝜃 ≈ 26.1∘ in X-ray diffraction patterns the aged Pt-Co/C nanoparticles featuring complex
(not shown for brevity). These results suggest the grad- shapes, suggesting particle overlap, coalescence and/or
ual (and irreversible) conversion of ordered graphitic particle necking by Ptz+ ions redepositing between two
domains into disordered domains, a process denoted individual crystallites.
as amorphization by More et al.62 The density of isolated and aggregated parti-
The structural changes of the high surface area cles estimated from the TEM images, plotted as a
carbon support are fast during the first 1163 h of function of the PEMFC operation time (Figure 10),
operation and then slow down, as reflected by the statistically confirms this trend. The density of iso-
variation of the mean crystallite size (La ) versus the lated Pt-Co/C particles severely decreases during the
PEMFC operation time, La being derived from the first approximately 1200 h of operation (two third
I(D) /I(G) ratio using the Knight and White formula.60 of the isolated Pt-Co/C particles are lost during this
Whatever the operating conditions of the PEMFC, the period of time), whereas the process slows down for
La value increases upon operation (until t = 2259 h, the following 2200 h. Assuming the comparable trend
Figure 8(b)), indicating that the remaining carbon is observed for the regime of carbon corrosion, this result

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

Pt Carbon corrosion
Carbon

Pt Pt crystallite migration and

coalescence
Carbon

Ptz+

Pt dissolution and
Pt redeposition: electrochemical
Ostwald ripening
Carbon
ze−
H2

Ptz+ Pt redeposition in the

PEM/ionomer
Pt

Carbon

FIGURE 7 | Schematics of the four main processes of Pt/C nanoparticles degradation during PEMFC operation. (Reproduced with permission from
Ref 52. Copyright 2010, The Electrochemical Society)

highlights the strong interaction of the metal nanopar- number of nanoparticles (Figure 10(a)). At the same
ticles with the disordered domains of the carbon sup- time, the extent of agglomerated nanoparticles dra-
port, most likely via the oxygen-containing groups matically increases, which reflects both the carbon
(the fraction of which is proportional to the degree corrosion-induced crystallite migration (as observed in
of disorder of the carbon black nanoparticles).53,63 Refs 14 and 48) and the redeposition of Pt between
Interestingly, similar decrease of the density of iso- two neighboring Pt-Co nanoparticles, that can literally
lated Pt-Co/C nanoparticles was observed over time, bridge them.
which confirms that crystallite migration and agglom- The Pt-Co/C nanoparticles suffer large mod-
eration is a potential-independent phenomenon (at ifications of their chemical composition upon
least in the potential region experience by the PEMFC aging, because Co (the non-noble element of
cathode). the pristine Pt3 Co/C nanoalloy) is preferentially
leached-out, without possibility to redeposit in the
Dissolution/Redeposition of Metal PEMFC environment.14,51,64 This has a dramatic but
Nanoparticles (Electrochemical Ostwald nonetheless complex impact on their electrocatalytic
Ripening) properties. On the one hand, homogeneous Pt-Co/C
As stated in Section At the Nanometer Scale for nanoparticles depleted in Co display depressed ORR
Pt/C, the stability of Pt-Co/C nanoparticles depends activity, because they lose the beneficial strain and
on the catalyst particle size, pH and temperature. ligand effects of the Co atoms.14,15,51,56,57 On the
Figures 9 and 10 evidence the progressive increase other hand, the remaining nanoparticles are on aver-
of the surface-averaged mean particle diameter upon age larger (because of the electrochemical Ostwald
aging. Its smaller magnitude (+1.2 nm in 3422 h of sta- ripening and related Pt redeposition on the remaining
tionary operation) than that observed in Ref 48 may crystallites), which is favorable for the ORR activity,
be related to the moderate reduction of the chemi- and Pt ‘hollow’ nanoparticles (that is containing a
cal potential for crystallites of approximately 4–5 nm, cavity in their center) with dramatically improved
with respect to nanocrystallites smaller than 3 nm in ORR activity may form in certain conditions,15,64
diameter.49 Such particle size increase is accounted both effects positively balancing the former one.
for by the occurrence of the electrochemical Ost- To these modifications of the intrinsic ORR
wald ripening, and explains why the fraction of the activity of the Pt-Co/C nanoparticles, the severe alter-
smallest nanoparticles decreases (they are irreversibly ation of the morphology of the cathode CL also
corroded) at the expense of the larger ones, ulti- induces changes to its apparent ability to efficiently
mately yielding non-negligible decrease of the overall reduce oxygen. Indeed, the inter-particle distance,

© 2014 John Wiley & Sons, Ltd.

 Advanced Review wires.wiley.com/wene

(a) the PEM, detrimentally affect the PEM stability (see

t=0h Section Degradation of the Membrane).67
D band G band
t = 1163 h D′ band
t = 2259 h Influence of Metal Cations on the Performance
t = 3422 h of the Cathode Catalytic Layer
The fate of Ptz+ ions generated by the corrosion of
Pt/C or Pt-Co/C nanoparticles is multifold. They can
either redeposit within the cathode CL (electrochemi-
cal Ostwald ripening), or be transported away from
the cathode CL, e.g., to the PEM, where they are
likely reduced by cross-over H2 . Therefore, the exis-
tence of Ptz+ ions is limited in an operating PEMFC
MEA, even if it still exhibits detrimental influences
to the PEM properties.67 On the contrary, owing to
the thermodynamic impossibility to redeposit Co2+
ions in the PEMFC environment, and to the fact that
1000 1200 1400 1600 1800
Co2+ ions are likely in the form of cations, Co2+
Raman shift / cm−1
species are stable at the cathode of an operating
PEMFC.68 Unfortunately, their presence is harmful to
(b) j = 0.6 A cm−2 - stationary the physico-chemical properties of the cathode CL.68
4.0 j = 0.6 A cm−2 - start/stop
j = 0.24 A cm−2 - stationary 16 Indeed, these species strongly interact with the sul-
3.5 14
fonic acid groups of the perfluorosulfonated mem-
brane, which yields a decrease of its proton transport
3.0 12
properties.69,70 They also enhance the mass-transport
I(D) / I(G)

La / nm

2.5 10 limitation of O2 within the ionomer (Figure 11). The

2.0 8 latter is clear from the sharp decrease of the lim-
6
iting current plateau measured for polluted mem-
1.5
branes (Figure 11(a) and (c)). In addition, Figure 11(d)
4
1.0 demonstrates that O2 mass-transport limitation is
0.5
2 prevalent versus that of protons,68 which somewhat
0 mitigates usual conclusions of the literature toward
0 500 1000 1500 2000 2500 3000 3500 the effect of cation contamination.69,70 A slight effect
PEMFC operation time / h in the kinetic current is also witnessed (Figure 11(b))
FIGURE 8 | (a) Raman spectra of the fresh/aged Pt-Co/C cathode and the ORR mechanism is slightly more directed
electrocatalysts before/after operation at j = 0.60 A cm−2 , T = 70∘ C, toward the two-electron pathway in the presence of
65% RH for various durations. Reproduced with permission from Ref 15. cationic pollution, but these are not prevalent.
Copyright 2013, Elsevier. (b) Variation of the I (D) /I (G) ratio and the mean Finally, the effect of the cations is also severe
crystallite size (La ) versus the PEMFC operation time in various to the PEM properties and aging,67,71,72 as will be
experimental conditions (j = 0.24 or 0.60 A cm−2 in stationary detailed in Section Degradation of the Membrane.
conditions or j = 0.60 A cm−2 in 1 h start/1 h stop, T = 70∘ C, 65% RH),
as detailed in Refs 14 and 15.
DEGRADATION OF THE MEMBRANE
which varies as the reciprocal of the average particle The polyelectrolyte membrane (PEM, usually of per-
density over the carbon substrate, constantly increases fluorosulfonic acid type, PFSA) is an essential com-
during operation (Figure 10(a)). Knowing the ORR is ponent of the proper PEMFC operation; it should
a stepwise process that involves generation of hydro- firstly enable the mass-transport of protons from the
gen peroxide in a two-electron reaction, followed anode catalytic sites to the cathode.73 The proton
by its disproponation or further reduction on neigh- conductivity is related to the physicochemical prop-
boring sites,65,66 aged electrocatalysts may generate erties of the PEM and in particular to its ion-exchange
increased fraction of H2 O2 , which (1) lower the effi- capacity (Section IEC and Cationic Contamination).
ciency of O2 conversion, and assuming H2 O2 is a In that regards, knowing the PEM is not selective
strong oxidant and a well-known source of radical (2) to protons, the degradation of the PEM conductivity
may further accelerate the degradation of the CL, or (Section Conductivity Measurements) can gradually
(3) if the H2 O2 species or radicals are released within occur in the course of the PEMFC operation, where

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

0.20 0.20

0.15 0.15

N / Ntotal

N / Ntotal
0.10 0.10

0.05 0.05

0.00 0.00
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
d / nm d / nm

20 nm Fresh MEA 20 nm t = 1163 h

0.20 0.20

0.15 0.15

N / Ntotal
N / Ntotal

0.10 0.10

0.05 0.05

0.00 0.00
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
d / nm d / nm

20 nm t = 2259 h 20 nm t = 3422 h

FIGURE 9 | Particle size distributions and representative TEM images of the Pt-Co/C cathode electrocatalyst before/after operation at
j = 0.60 A cm−2 , T = 70∘ C, 65% RH for various durations. (Reproduced with permission from Ref 15. Copyright 2013, Elsevier)

metal cations are generated. In addition, the PEM form using infrared spectroscopy or the analysis of
must act as a barrier to the reactant and electrons.1,2 batch-EW in its acid form using titration. However,
This last property is not less important, since PEMFC these measurements will statistically scatter with some
failure mostly occurs as a result of the PEM failure, or standard deviation.76 The means to precisely deter-
in other words in the loss of its barrier properties.74 mine the IEC were recently reviewed.76 A minimal
This likely happens because of dramatic changes of the absolute error of ±0.03 meq g−1 could be obtained
PEM mechanical properties upon aging (Section Mem- with the proper treatment of thermal gravimetric
brane Toughness Characterization), eventually result- analysis (TGA) and conductivity measurements.
ing in hole formation within the membrane75 (Section Surprisingly the Fourier-transformed infrared spec-
Toward Membrane Holes Localization). The present troscopy (FTIR) and titration methods performed
section addresses these issues and details how fast, rather poorly,76 revealing large errors in the deter-
efficient, and robust tools were designed to charac- mination of the IEC. The latter may in addition be
terize these properties and localize defects (holes or largely altered, even under normal operating condi-
short-circuits) in the PEM. tions, with cations contamination,77 originating from
the system materials degradation (catalytic layers,
bipolar plates) or from the reactants. Indeed, the
IEC and Cationic Contamination cations substitute the protons in the membrane and
The ionic exchange capacity (IEC) defines a PFSA impact the end-use properties.
membrane, for both the initial performances and The cationic contamination obviously induces
their durability. Membrane manufacturers measure a reduction of the proton conductivity78 (Section
their equivalent weight (EW) using different methods, Membrane Toughness Characterization), but also a
namely the analysis of the batch-EW in its precursor drop in the mechanical toughness79 (Section Towards

© 2014 John Wiley & Sons, Ltd.

 Advanced Review wires.wiley.com/wene

Density of NPs / μmcarbon

(a) j = 0.60 A cm−2 - stationary

−2

16000 j = 0.24 A cm−2 - stationary 0

12000 j = 0.60 A cm−2 - start/stop (a)
−1 gCo2+ = 0
8000

j / mA cmgeo
−2
−2 gCo2+ = 0.4
4000
gCo2+ = 0.7
0 −3
gCo2+ = 0.9
(b) −4
6.5

6.0 0.4 0.5 0.6 0.7 0.8 0.9 1.0

ds / nm

E versus RHE / V
5.5 1000
(b)
5.0
100

jk / μA cmPt
−2
0 1000 2000 3000
PEMFC operation time / h 10

FIGURE 10 | (a) Density of isolated (full symbols) and aggregated 1

(open symbols) Pt-Co/C nanoparticles (NPs) and (b) surface-averaged
diameter (d S ) statically estimated by measuring the size of isolated
0.80 0.82 0.84 0.86 0.88 0.90 0.92
Pt-Co/C particles versus PEMFC operation time. The PEMFC short stacks
E versus RHE / V
were operated in various experimental conditions (j = 0.24 or 1.25
0.60 A cm−2 in stationary conditions or j = 0.60 A cm−2 in 1 h start/1 h (c)
stop, T = 70∘ C, 65% RH), as detailed in Refs 15 and 64. 1.00
Normalized jD

0.75
Membrane Holes Localization). It also affects the
0.50
overall oxygen reduction reaction process when the
ionomer is contaminated (Section Influence of Metal 0.25 Kienitz et al. (Cs+ ions)
SSC (Co2+ ions)
Cations on the Performance of the Cathode Catalytic
0.00 Liquid electrolyte (Co2+ ions)
Layer). The propensity of the cations to exchange with
protons within the membrane and ionomer is essen- 0.00 0.25 0.50 0.75 1.00
gM+
tially controlled by their Lewis Acid Strength, as pro-
posed by Brown.80 The amplitude of the cationic con- (d)
tamination is defined as the ratio between the con- 0.00E+000 2H+ + 2e− → H2
taminant charge and the number of sulfonate groups
6.25E–009
present in the membrane.72,81 The nature and amount
of cationic contaminants within a aged PFSA mem-
j / mA cmgeo

1.25E–008
−2

O2 + 4H+ + 4e− → 2H2O

brane can be determined by performing FTIR anal-
1.88E–008
yses, preferentially in transmission mode. According gCo2+ = 0
to Figure 12, which gathers a broad series of results 2.50E–008 gCo2+ = 0.7
from the literature,71,81,82 the position of the peaks
3.13E–008
at 𝜐 = 970 and 1060 cm−1 in a FTIR spectrum usu-
ally attributed to SO3 − and COC moieties, respectively 3.75E–008
gives enough information to target the possible cations 0.0 0.2 0.4 0.6 0.8 1.0
E versus RHE / V
that induced a contamination of the membrane during
PEMFC operation.77,81 FIGURE 11 | (a) Linear sweep ORR voltammograms measured for
Pt/C in contact with fresh or Co2+ -contaminated Nafions 117 PEM, (b)
corresponding Tafel plots of the mass-transport corrected kinetic
Conductivity Measurements current, and (c) normalized diffusion limited current for fresh and
In direct connection with the IEC, the proton con- Co2+ -contaminated Nafion 117 PEM. Temperature = 22∘ C, sweep
ductivity (𝜎) is evidently among the most impor- rate = 0.001 V s−1 . (d) Linear sweep ORR voltammograms for Pt/C in
tant parameters for membrane selection. The plot contact with a noncontaminated (𝛾 CO2+ = 0, black curves) and a Nafion
in Figure 13 shows a unique relationship between 117 PEM contaminated with Co2+ ions (𝛾 CO2+ = 0.7, red curves) and
𝜎 and the IEC for a large series of membrane with linear sweep voltammograms in argon-saturated solution into the
dissimilar chemistry: it emphasizes that the proton hydrogen evolution region. (Reproduced with permission from Ref 68.
conductivity is not much depending on the length Copyright 2012, Royal Society of Chemistry)
of the side chain,83 but rather on the IEC. In the

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

Li+ 100
974 Alkaline Na+ R+/−
1.0 0.5 0.0
2+
Ca
972
Cs+
νCOC / cm−1

Alkaline earth
Sr2+ 10−1

s / S cm−1
970
Transition metal
968
Ca2+
K+
Various 10−2
R+/−
966 Cr3+ Fe3+ Fe
2+

Satured

1055 1060 1065 1070

νSO3− / cm−1 10−3
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
FIGURE 12 | Absolute position of the 970 cm−1
and 1070 cm−1 IEC / meq g−1
bands from FTIR experiments performed on dried membranes.
FIGURE 13 | Dependence of the proton conductivity of membranes
Significantly different behaviors are observed depending on the nature
with the IEC, for various chemical structures. The conductivity is
and amount of the cations that was added on purpose to replace the
governed by the IEC, following percolation behavior with a threshold
protons in the PFSA.
close to 0.60 meq g−1 . The controlled addition of cations (R+/− is the
number of cations divided by the number of SO3 − in the polymer)
literature, the conductivity is described as a func- induces the gradual decrease of the proton conductivity, as if the IEC
tion of the water volume fraction, according to the was reduced76 regardless of the nature of the cations.72 R+/− is also
percolation law for ion-exchange membranes.84,85 A denoted as 𝛾 M+ in the literature68 and in Section Degradation of the
similar equation may be employed for the depen- Catalytic Layers, with M+ the metal cation contaminant.
dence of the conductivity on IEC75 : 𝜎 ≈ (IEC − IECc)t
with t close to unity. The value of the critical
The two systems performed very similarly on regu-
IECc suggests a significant change in the morphol-
lar tensile tests,86 but the larger resistance to tearing
ogy of the PFSA at IEC = 0.63 ± 0.03 meq g−1 . This
of the composite membrane could be clearly revealed
morphology transition resulting in a discontinuous
with double edge notch tension (DENT) geometry.86
increase in the conductivity with increasing IEC has
The two extracted parameters,87,88 corresponding to
already been demonstrated for sulfonated styrene and
the initiation (we ) and propagation (𝛽wp ) of the holes,
copolymers.76 Below IEC = 0.63 meq g−1 , the degree
are significantly better for the three-layer membranes.
of connectivity between ionic domains is not sufficient
In the case of membranes with anisotropic behaviors,
for protons and water transport through the mem-
the propagation of the cracks is easier in the direction
brane, resulting in excessively depressed proton con-
of the polymeric chains than perpendicular to it. Since
ductivity (Figure 13). This threshold makes clear why
water and gas channels are plausible sources to the
no PFSA membrane with IEC below 0.6 meq g−1 is
appearance of defects or fractures in the membranes
commercially available. Membrane further submitted
due to their sharp edges under the compression of
to cationic exchange of the protons, display dramati-
the stack, the orientation of polymer chains in com-
cally decreased conductivity (secondary X-axis, on the
parison to channels should be an important param-
top of Figure 13). The contamination with cationic
eter to take into account during the manufacturing
moieties thus induces a reduction of the conductivity,
and the processing of MEA. The best configuration
as if the IEC was reduced, in line with the conclusions
would be to align the PEM so that the gas channels
of Iojoiu et al.67
remain perpendicular to the orientation of the poly-
mer chains of the membrane, thereby resulting in opti-
Membrane Toughness Characterization mized tear resistance. Here again the chemical struc-
To prevent the system shutdowns resulting from mem- ture of the PEM matrix might slightly alter its ability to
branes failure, many attempts have been made to perform. Although the proton conductivity is tightly
define the most relevant mechanical characterization linked to the IEC (Figure 13), the latter has less impact
of the PEM. The tempting tensile tests are not related on the tearing properties (we and 𝛽wp ). Figure 14
to the capacity of the membrane to resist tearing.86 indeed demonstrates that the crack initiation gradu-
In contrast, measuring the essential work of fracture ally decreases with the ionic exchange capacity, mainly
was found more discriminative.86 A good example to because of an increase in the water content,89 whereas
describe this is the comparison between nonreinforced the resistance to crack propagation remains almost
membranes and similar ones composed of three-layer. unaffected by IEC variations. These series of data do

© 2014 John Wiley & Sons, Ltd.

 Advanced Review wires.wiley.com/wene

11 10 the problem from the system level (macroscale) down

10
the membrane defect level (microscale). Here, more
8 insight will be provided on their localization.
9 Firstly the gas leakage test74,91 is a pragmatic,
We / kJ m−2

b.wp / MJ m−3
6 efficient, and widely employed way to probe the for-
8
mation of holes both in and out the stack. Depending
7 4
on the nature and magnitude of the pressure drop,
6 one may even separate a membrane from a gasket
2
5
failure,92 and a harmless hole from a severe one that
0
might trigger the shutdown of the entire system for
4 safety reasons. The set of defective membranes may
0.9 1.0 1.1 1.2 1.3
also be established through passive electrical mea-
IEC / meq g−1
surements (Figure 15(a)) performed in the absence of
FIGURE 14 | Required energy to initiate (left) and propagate (right) both gas reactants. The measurement of the membrane
the formation of crack in PFSA as a function of the IEC of various resistance to electronic current is determined through
membranes. The chemical structure does not alter much this behavior the so-called relaxation technique.93 The comparison
measured with the essential work fracture (EWF) method. Layered of the two macroscopic measurements already fur-
structures do present significantly higher toughness according to this nishes interesting hints on the severity of the dam-
method (not shown on this graph) up to 23 kJ m−2 and 10 MJ m−3 for ages. In the most critical holes the membrane degrada-
w e and 𝛽 w p , respectively.86 The substitution of protons by any cations
tion was accompanied by the corrosion of the carbon
from a salt, induces a dramatic plasticity loss, as revealed for instance by
a 𝛽 w p value close to zero (blue dot). Lines were added to guide the eye.
within the active layer, resulting from the intense (and
local) crossover of the hydrogen and/or oxygen. From
that viewpoint, the interest to define the most probable
not primarily depend on the chemical structure of the localization of the hole induced by PEMFC operation
membrane, and similar results were observed for simi- is obvious. This can be achieved with the use of a ther-
lar structures.89 Only the three-layer structure exhibits mal camera (Figure 15(b)) to reveal the hydrogen leak
a significantly better behavior than the monolith coun- in an open fuel cell.74,94,95 In the present case it was
terpart, regardless of the IEC. In addition, the cationic observed that almost all critical holes were located in
pollution of these membranes induces an impressive the hydrogen inlet and air outlet position. This may
reduction in its ability to resist tearing propagation, be the result either from the presence of dry hydro-
regardless of the nature of inorganic cations.72 In par- gen combined with wet air that mechanically submit
ticular the 𝛽wp parameter almost dropped to zero,
the membrane to large strains,96–99 or from the forma-
meaning that nonreinforced membranes subjected to
tion of H2 O2 -derived radicals that degrade the mem-
cationic contamination would not resist crack propa-
brane through a Fenton-like mechanism.100 In any
gation. This was typically observed for earlier versions
case, the actual morphology of the severe holes and
of PEM, that could not withstand PEMFC operation
the chemical properties of the membrane at its vicin-
for longer than several hundred hours,48,67 resulting in
ity should be studied. After removing the active layers
premature hole formation and related PEMFC failure.
the membrane could be observed more directly. In the
damaged zones underneath the membrane was found
Toward Membrane Holes Localization perforated by holes with rather long aspect ratios as
Electrochemical impedance spectroscopy (EIS) is con- shown in Figure 15(c). From that prospect, the term
sidered one of the most reliable techniques to probe ‘pinhole’, which is massively used to describe these
MEA performances during aging. EIS is mostly per- holes, is not relevant: herein, the term ‘hole’ is sim-
formed on single cell,34,90 and sometimes on the ply used, although ‘knife-hole’ might be more rele-
whole stack. However, the complexity of the extracted vant. Several hundred MEAs originating from on-site
parameters makes it difficult to develop robust and PEMFC stack operation were tested following the pro-
on-site applications. When a membrane failure is sus- cedure described in Figure 15. In every circumstance
pected, there is a vivid need to understand the orig- the critical holes were located in the hydrogen inlet
inal reason for it. Sections IEC and Cationic Con- area, found highly oriented in the direction of the
tamination to Membrane Toughness Characterization gas flow, and with a large aspect ratio. This indicates
demonstrated that the reasons for membrane hole for- that the mechanical contribution was important to the
mation upon PEMFC operation were numerous and final perforation, and is evidently controlled by the
Figure 15 highlights that a simple series of experimen- local toughness of the MEAs (cf. Section Membrane
tation may be employed one after the other to track Toughness Characterization), itself related to the local

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

FTIR
(a)
(c) Tensile test
Pressure drop (b) SEM
relaxometry
IR thermography

50
0
μm
Stack Cations
contamination
Chains
orientation
Pinholes
Flaws
Cell

FIGURE 15 | Methodology developed to define both 1-the position of the holes (a) in the stack, (b) in the MEA and (c) in the membrane, and
2-the causes for failure, which might be related to mechanical issues, such as chain orientation or chemical contamination of the sulfonic groups with
cationic foreign moieties.

membrane physicochemical properties (IEC, proton the decrease in ECSA and the corrosion of the carbon
conductivity, cation contamination, cf. Sections IEC support is decaying with time (the most corrodible ele-
and Cationic Contamination and Conductivity Mea- ments/groups at the carbon surface vanish first, result-
surements). ing in gradual predominance of more stable struc-
tures over time, cf. Section Degradation of the Cat-
alytic Layers). This however cannot be used to predict
LINKS BETWEEN THE OBSERVED the end-of-life, since the PEM degradation is at stake
DEGRADATIONS AND THE MEA there (it occurs by hole formation, which is indeed
impacted by this gradual aging of the active layer,
PERFORMANCES IN FC SYSTEMS but not directly depends on it, cf. Section Degrada-
Understanding the previously described degradation tion of the membrane). Post mortem analyses of failed
phenomena of the catalyst layer and the membrane MEAs confirmed this hypothesis, as PEM holes were
at a local scale has enabled dramatic improvements clearly identified on defective cells (by pressure drop
of the MEA durability in PEMFC systems. As an and IR camera characterization) and as performances
example, fuel starvation issues have been mitigated by of nonperforated MEA were still satisfactory for sys-
reducing the time of presence of the hydrogen front tem operation. The two consequences of MEA degra-
in the anode bipolar plate channels, or by avoiding dation (performance decay, mainly linked to catalyst
air injection in these channels when filled with H2 , layer degradation, and MEA failure, mainly linked to
resulting in enhanced durability on the field (e.g., up PEM holes formation) still need some mitigation to
to approximately 11700 h and 2400 start/stops for respectively hold a satisfactory H2 consumption and
the data presented in Figure 16(a)). However, some delay the stack end-of-life.
phenomena linked to gas concentration and temper-
ature heterogeneities or most commonly to material
physico-chemical changes are more complicated to CONCLUSION
avoid and still result in performances degradation and PEMFCs, despite now being ready for commercial-
failure. In most cases, the apparent (overall) perfor- ization, remain limited by their insufficient durabil-
mance loss is related to local performance degrada- ity. The latter is often regarded as homogeneous,
tions (Figure 16(b) and (c)), themselves related to although stationary operation like load cycles induce
the MEA materials (see Sections The Degradation of evident heterogeneity in the unit cell operation and
the PEMFC MEA Materials and Performances Are therefore, in their materials and performance degra-
Heterogeneous, Degradation of the Catalytic Lay- dations. Through a tight collaboration between chem-
ers, and Degradation of the membrane). Since 2010, ical engineers, polymer scientists and electrochemists,
the performances of Axane systems deployed on the the present contribution demonstrated that MEAs
field have been followed and degradations rates mon- do not age homogeneously during PEMFC opera-
itored. Figure 16 reveals the clear influence of the tion, should it be in accelerated stress test (repetitive
aging time on the measured degradation rate. The rate start-up and shut-down procedures) on segmented sin-
of the cell-voltage loss decreases over time, mainly gle cells of various geometries (parallel linear versus
because these losses are associated to the (local) multiple U-shaped channels) and long-duration sta-
change of morphology of the Pt-M/C nanoparticles: tionary operation in short-stack.

© 2014 John Wiley & Sons, Ltd.

 Advanced Review wires.wiley.com/wene

(a) 60

Degradation rate per unit cell / μV h−1

Number of start/stop events

1000

40 100

10
20

0
0.1
0 2000 4000 6000 8000 10000 12000
Total operation time on the field / h
(b) (c)

Jsegment at 0.5 A cm−2 global / A cm−2

EOL
0.70 1.0
PEM fuel cell voltage / V

0.65 0.8
BOL
0.6
0.60
0.4
0.55 BOL
0.2
2 slph H2 in air
0.50 10 slph H2 in air
0.0 EOL
0 5 10 15 20
0 50 100 150 200 250 300
Segment #1 Segment number Segment #20
Number of start-up events H2 inlet/air outlet H2 outlet/air inlet

FIGURE 16 | (a) Variation of the degradation rate per unit cell versus the operation time on the field for Axane systems. The number of start/stop
procedures undergone by the stacks is also given for information. The nature of the MEA operated in these stacks is not identical and proprietary. In
most cases, the customer stopped exploitation of the system before any failure. (b) Example of cell voltage drop monitored for a segmented cell
during repetitive start-up/shut-down procedures (the cell was operated 10 min at j = 0.50 A cm−2 between each start-up and shut-down events); (c)
the overall voltage drop of 190 mV, measured between the beginning of life (BoL) and the end of life (EoL) after 272 start-up with 2 slph of H2 in an
air-filled anode, results from very local degradations.

Macroscopically, the MEA is not fuelled evenly the Pt electroactive area, the Pt3 Co/C electrocatalyst
by the bipolar plates and severe degradations occur undergoes severe carbon corrosion. The amorphous
during start-up and shut-down events in the region carbon (defects) is attacked first, but graphite is only
that remains/becomes transiently starved in hydro- metastable from a thermodynamic view point: it will
gen. Such transients are particularly dramatic for eventually be corroded, through surface functionaliza-
the cathode catalyst-layer, with special emphasis for tion, amorphization and consecutive CO2 generation,
the carbon-substrate of the Pt-based electrocatalyst overall resulting in Pt-Co nanocrystallite migration,
nanoparticles, which undergoes irreversible corrosion loss of hydrophobicity of the cathode catalyst layer
into CO2 at the high potentials (>1 V) monitored and larger mass-transport hindrance because of the
in such sequences. Another level of heterogeneity is layer gradual collapse. Finally, the composition of the
witnessed between the channel and land areas of mother alloy is continuously depressed in Co (sacri-
the cathode catalyst layer, because the hindrance for ficial metal), generating Co2+ cations which pollute
gas-diffusion under the land areas is larger. In the the ionomer and hinders the whole performances of
cathode catalyst layer, the degradation of Pt3 Co/C the cathode catalysts layer. Such cationic pollution
electrocatalyst cannot be avoided in stationary opera- is detrimentally affecting the main usage properties
tion either. Apart from the well-documented Ostwald of the ionomer and membrane materials (proton
ripening and crystallite migration, both detrimental to conductivity and resistance to fracture), eventually

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

leading to hole formation. These defects were local- various PFSA membranes can nevertheless further
ized on used PEMFC MEA with the help of an Infrared be improved by using PTFE-reinforced membranes,
camera, mainly close to the gases inlet/outlet, in agree- which better resist to holes formation due to their high
ment with the findings derived from the segmented resistance to crack initiation and propagation.
cell analyses. The mechanical fracture-resistance of

ACKNOWLEDGEMENTS
Funding by Oseo A2I in the frame of the Horizon Hydrogen Energy (H2E) project is gratefully acknowledged.
MC thanks the french IUF for its support.

REFERENCES
1. Vielstich W, Lamm A, Gasteiger HA. Handbook of and non-Pt oxygen reduction catalysts for PEMFCs.
Fuel Cells, vol. 1–4. Chichester, England: Wiley; 2003. Appl Catal B 2005, 56:9–35.
2. Gasteiger HA, Vielstich W, Yokokawa H. Handbook 12. Dubau L, Durst J, Maillard F, Chatenet M, Guétaz L,
of Fuel Cells, vol. 5–6. John Wiley & Sons Ltd: André J, Rossinot E. Heterogeneities of aging within a
Chichester, England; 2009. PEMFC MEA. Fuel Cells 2012, 12:188–198.
3. Wang Y, Chen KS, Mishler J, Cho SC, Adroher XC. 13. Durst J, Lamibrac A, Charlot F, Dillet J, Castanheira
A review of polymer electrolyte membrane fuel cells: LF, Maranzana G, Dubau L, Maillard F, Chatenet
technology, applications, and needs on fundamental M, Lottin O. Degradation heterogeneities induced
research. Appl Energy 2011, 88:981–1007. by repetitive start/stop events in proton exchange
4. Zhang SS, Yuan XZ, Wang HJ, Merida W, Zhu H, membrane fuel cell: inlet vs. outlet and channel vs.
Shen J, Wu SH, Zhang JJ. A review of accelerated land. Appl Catal B 2013, 138–139:416–426.
stress tests of MEA durability in PEM fuel cells. Int
14. Dubau L, Maillard F, Chatenet M, Guétaz L, André
J Hydrogen Energy 2009, 34:388–404.
J, Rossinot E. Durability of Pt3 Co/C cathodes in a
5. Zhang SS, Yuan XZ, Hin JNC, Wang HJ, Friedrich 16 cell PEMFC stack: macro/microstructural changes
KA, Schulze M. A review of platinum-based catalyst and degradation mechanisms. J Electrochem Soc 2010,
layer degradation in proton exchange membrane fuel 157:B1887–B1895.
cells. J Power Sources 2009, 194:588–600.
15. Dubau L, Lopez-Haro M, Castanheira L, Durst J,
6. Wu JF, Yuan XZ, Martin JJ, Wang HJ, Zhang JJ,
Chatenet M, Bayle-Guillemaud P, Guétaz L, Caqué
Shen J, Wu SH, Merida W. A review of PEM fuel cell
N, Rossinot E, Maillard F. Probing the structure,
durability: degradation mechanisms and mitigation
the composition and the ORR activity of Pt3 Co/C
strategies. J Power Sources 2008, 184:104–119.
nanocrystallites during a 3422 h PEMFC aging test.
7. Yu XW, Ye SY. Recent advances in activity and Appl Catal B 2013, 142–143:801–808.
durability enhancement of Pt/C catalytic cathode in
PEMFC—part II: degradation mechanism and dura- 16. Berning T, Lu DM, Djilali N. Three-dimensional
bility enhancement of carbon supported platinum cat- computational analysis of transport phenomena in a
alyst. J Power Sources 2007, 172:145–154. PEM fuel cell. J Power Sources 2002, 106:284–294.
8. Shao YY, Yin GP, Gao YZ. Understanding and 17. Chupin S, Colinart T, Didierjean S, Dube Y, Agbossou
approaches for the durability issues of Pt-based cat- K, Maranzana G, Lottin O. Numerical investiga-
alysts for PEM fuel cell. J Power Sources 2007, tion of the impact of gas and cooling flow con-
171:558–566. figurations on current and water distributions in
9. de Bruijn FA, Dam VAT, Janssen GJM. Durability and a polymer membrane fuel cell through a pseudo-
degradation issues of PEM fuel cell components. Fuel two-dimensional diphasic model. J Power Sources
Cells 2008, 8:3–22. 2010, 195:5213–5227.
10. Borup R, Meyers J, Pivovar B, Kim YS, Mukundan 18. Yang XG, Zhang FY, Lubawy AL, Wang CY. Visu-
R, Garland N, Myers D, Wilson M, Garzon F, Wood alization of liquid water transport in a PEFC. Elec-
D, et al. Scientific aspects of polymer electrolyte fuel trochem Solid State Lett 2004, 7:A408–A411.
cell durability and degradation. Chem Rev 2007, 19. Dillet J, Lottin O, Maranzana G, Didierjean S, Con-
107:3904–3951. teau D, Bonnet C. Direct observation of the two-phase
11. Gasteiger HA, Kocha SS, Sompalli B, Wagner FT. flow in the air channel of a proton exchange membrane
Activity benchmarks and requirements for Pt, Pt-alloy, fuel cell and of the effects of a clogging/unclogging

© 2014 John Wiley & Sons, Ltd.

Advanced Review wires.wiley.com/wene

sequence on the current density distribution. J Power 32. Shen Q, Hou M, Liang D, Zhou Z, Li X, Shao Z, Yi
Sources 2010, 195:2795–2799. B. Study on the processes of start-up and shutdown in
20. Bedet J, Maranzana G, Leclerc S, Lottin O, Moyne C, proton exchange membrane fuel cells. J Power Sources
Stemmelen D, Mutzenhardt P, Canet D. Magnetic res- 2009, 189:1114–1119.
onance imaging of water distribution and production 33. Hinds G, Brightman E. In situ mapping of electrode
in a 6 cm2 PEMFC under operation. Int J Hydrogen potential in a PEM fuel cell. Electrochem Commun
Energy 2008, 33:3146–3149. 2012, 17:26–29.
21. Morin A, Xu F, Gebel G, Diat O. Influence of PEMFC 34. Lamibrac A, Maranzana G, Lottin O, Dillet J, Mainka
gas flow configuration on performance and water J, Didierjean S, Thomas A, Moyne C. Experimen-
distribution studied by SANS: evidence of the effect of tal characterization of internal currents during the
gravity. Int J Hydrogen Energy 2011, 36:3096–3109. start-up of a proton exchange membrane fuel cell. J
22. Hartnig C, Manke I, Kuhn R, Kleinau S, Goebbels Power Sources 2011, 196:9451–9458.
J, Banhart J. High-resolution in-plane investigation of 35. Lamibrac A, Maranzana G, Dillet J, Lottin O, Didier-
the water evolution and transport in PEM fuel cells. J jean S, Durst J, Dubau L, Maillard F, Chatenet M.
Power Sources 2009, 188:468–474. Local degradations resulting from repeated start-ups
23. Pérez LC, Brandão L, Sousa JM, Mendes A. Seg- and shut-downs in proton exchange membrane fuel
mented polymer electrolyte membrane fuel cells—A cell (PEMFC). Energy Procedia 2012, 29:318–324.
review. Renewable Sustainable Energy Rev 2011, 36. Schneider IA, von Dahlen S. Start-stop phenomena in
15:169–185. channel and land areas of a polymer electrolyte fuel
24. Dillet J, Spernjak D, Lamibrac A, Maranzana G, cell. Electrochem Solid State Lett 2011, 14:B30–B33.
Mukundan R, Fairweather J, Didierjean S, Borup RL, 37. Abbou S, Dillet J, Spernjak D, Mukundan R, Fair-
Lottin O. Impact of flow rates and electrode specifi- weather J, Borup RL, Maranzana G, Didierjean S, Lot-
cations on degradations during repeated startups and tin O. Time evolution of local potentials during PEM
shutdowns in polymer-electrolyte membrane fuel cells. fuel cell operation with dead-ended anode. ECS Trans
J Power Sources 2014, 250:68–79. 2013, 58:1631–1642.
25. Freunberger SA, Reum M, Evertz J, Wokaun A, Büchi 38. Zhang G, Guo L, Ma L, Liu H. Simultaneous mea-
FN. Measuring the current distribution in PEFCs surement of current and temperature distributions in a
with sub-millimeter resolution: I. Methodology. J Elec- proton exchange membrane fuel cell. J Power Sources
trochem Soc 2006, 153:A2158–A2165. 2010, 195:3597–3604.
26. Reum M, Freunberger SA, Wokaun A, Büchi FN. Mea- 39. Zhang G, Shen S, Guo L, Liu H. Dynamic characteris-
suring the current distribution with sub-millimeter res- tics of local current densities and temperatures in pro-
olution in PEFCs: II. Impact of operating parameters. ton exchange membrane fuel cells during reactant star-
J Electrochem Soc 2009, 156:B301–B310. vations. Int J Hydrogen Energy 2012, 37:1884–1892.
27. Reum M, Wokaun A, Büchi FN. Measuring the current 40. Djilali N, Lu D. Influence of heat transfer on gas and
distribution with submillimeter resolution in PEFCs: water transport in fuel cells. Int J Therm Sci 2002,
III. Influence of the flow field geometry. J Electrochem 41:29–40.
Soc 2009, 156:B1225–B1231.
41. Vie PJS, Kjelstrup S. Thermal conductivities from
28. Siroma Z, Fujiwara N, Ioroi T, Yamazaki S-i, Senoh temperature profiles in the polymer electrolyte fuel cell.
H, Yasuda K, Tanimoto K. Transient phenomena in Electrochim Acta 2004, 49:1069–1077.
a PEMFC during the start-up of gas feeding observed
with a 97-fold segmented cell. J Power Sources 2007, 42. Weber AZ, Newman J. Coupled thermal and water
172:155–162. management in polymer electrolyte fuel cells. J Elec-
trochem Soc 2006, 153:A2205–A2214.
29. Maranzana G, Lottin O, Colinart T, Chupin S,
Didierjean S. A multi-instrumented polymer exchange 43. Wang Y, Wang C-Y. A nonisothermal, two-phase
membrane fuel cell: observation of the in-plane model for polymer electrolyte fuel cells. J Electrochem
non-homogeneities. J Power Sources 2008, 180: Soc 2006, 153:A1193–A1200.
748–754. 44. Eikerling M. Water management in cathode catalyst
30. Reiser CA, Bregoli L, Patterson TW, Yi JS, Yang layers of PEM fuel cells: a structure-based model. J
JDL, Perry ML, Jarvi TD. A reverse-current decay Electrochem Soc 2006, 153:E58–E70.
mechanism for fuel cells. Electrochem Solid State Lett 45. Thomas A, Maranzana G, Didierjean S, Dillet J,
2005, 8:A273–A276. Lottin O. Measurements of electrode temperatures,
31. Didierjean S, Lamibrac A, Geneston T, Rako- heat and water fluxes in PEMFCs: Conclusions about
tondrainibe A, Maranzana G, Rozier E, Beille F, transfer mechanisms. J Electrochem Soc 2013, 160:
Lottin O. Internal currents in response to a load F191–F204.
change during fuel cell start-up. Int J Hydrogen 46. Ferreira PJ, la O’ GJ, Shao-Horn Y, Morgan D,
Energy 2012, 37:6798–6807. Makharia R, Kocha S, Gasteiger HA. Instability of

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

Pt/C electrocatalysts in proton exchange membrane a Cu based case study using in situ electrochemical
fuel cells—a mechanistic investigation. J Electrochem X-ray absorption spectroscopy. J Phys Chem C 2013,
Soc 2005, 152:A2256–A2271. 117:4585–4596.
47. Guilminot E, Corcella A, Charlot F, Maillard F, 59. Tuinstra F, Koenig JL. Raman spectrum of graphite. J
Chatenet M. Detection of Ptz+ ions and Pt nanopar- Chem Phys 1970, 53:1126–1130.
ticles inside the membrane of a used PEMFC. J Elec- 60. Knight DS, White WB. Characterization of diamond
trochem Soc 2007, 154:B96–B105. films by Raman spectroscopy. J Mater Res 1989, 4:
48. Guilminot E, Corcella A, Chatenet M, Maillard 385–393.
F, Charlot F, Berthome G, Iojoiu C, Sanchez J-Y, 61. Cancado LG, Takai K, Enoki T, Endo M, Kim YA,
Rossinot E, Claude E. Membrane and active layer Mizusaki H, Jorio A, Coelho LN, Magalhaes-Paniago
degradation upon PEMFC steady-state operation: I. R, Pimenta MA. General equation for the deter-
Platinum dissolution and redistribution within the mination of the crystallite size La of nanographite
MEA. J Electrochem Soc 2007, 154:B1106–B1114. by Raman spectroscopy. Appl Phys Lett 2006,
49. Shao-Horn Y, Sheng W, Chen S, Ferreira P, Holby E, 88:163106–163103.
Morgan D. Instability of supported platinum nanopar- 62. Liu ZY, Zhang JL, Yu PT, Zhang JX, Makharia R,
ticles in low-temperature fuel cells. Topics Catal 2007, More KL, Stach EA. Transmission electron microscopy
46:285–305. observation of corrosion behaviors of platinized car-
50. Zhao Z, Castanheira L, Dubau L, Berthomé G, bon blacks under thermal and electrochemical condi-
Crisci A, Maillard F. Carbon corrosion and platinum tions. J Electrochem Soc 2010, 157:B906–B913.
nanoparticles ripening under open circuit potential 63. Zhao Z, Dubau L, Maillard F. Evidences of the
conditions. J Power Sources 2013, 230:236–243. migration of Pt crystallites on high surface area carbon
51. Chen S, Gasteiger HA, Hayakawa K, Tada T, supports in the presence of reducing molecules. J
Shao-Horn Y. Platinum-alloy cathode catalyst degra- Power Sources 2012, 217:449–458.
dation in proton exchange membrane fuel cells: 64. Dubau L, Durst J, Maillard F, Guétaz L, Chatenet
Nanometer-scale compositional and morphological M, André J, Rossinot E. Further insights into the
changes. J Electrochem Soc 2010, 157:A82–A97. durability of Pt3 Co/C electrocatalysts: formation of
52. Vion-Dury B, Chatenet M, Guétaz L, Maillard F. "hollow" Pt nanoparticles induced by the Kirkendall
Determination of aging markers and their use as a effect. Electrochim Acta 2011, 56:10658–10667.
tool to characterize Pt/C nanoparticles degradation 65. Ruvinskiy PS, Bonnefont A, Pham-Huu C, Savinova
mechanism in model PEMFC cathode environment. ER. Using ordered carbon nanomaterials for shedding
ECS Trans 2011, 41:697–708. light on the mechanism of the cathodic oxygen reduc-
53. Kinoshita K. Carbon, electrochemical and physico- tion reaction. Langmuir 2011, 27:9018–9027.
chemical properties. New York: John Wiley & Sons; 66. Schneider A, Colmenares L, Seidel YE, Jusys Z, Wick-
1988. man B, Kasemo B, Behm RJ. Transport effects in the
54. Hara M, Lee M, Liu C-H, Chen B-H, Yamashita Y, oxygen reduction reaction on nanostructured, planar
Uchida M, Uchida H, Watanabe M. Electrochemical glassy carbon supported Pt/GC model electrodes. Phys
and Raman spectroscopic evaluation of Pt/graphitized Chem Chem Phys 2008, 10:1931–1943.
carbon black catalyst durability for the start/stop 67. Iojoiu C, Guilminot E, Maillard F, Chatenet M,
operating condition of polymer electrolyte fuel cells. Sanchez J-Y, Claude E, Rossinot E. Membrane and
Electrochim Acta 2012, 70:171–181. active layer degradation upon PEMFC steady-state
55. Pourbaix M. Atlas d’équilibres électrochimiques. operation: II. Influence of Ptz+ on membrane proper-
Paris: Gauthier-Villars & Cie; 1963. ties. J Electrochem Soc 2007, 154:B1115–B1120.
56. Dubau L, Maillard F, Chatenet M, André J, Rossinot 68. Durst J, Chatenet M, Maillard F. Impact of metal
E. Nanoscale compositional changes and modification cations on the electrocatalytic properties of Pt/C
of the surface reactivity of Pt3 Co/C nanoparticles nanoparticles at multiple phase interfaces. Phys Chem
during proton-exchange membrane fuel cell operation. Chem Phys 2012, 14:13000–13009.
Electrochim Acta 2010, 56:776–783. 69. Kienitz B, Baskaran H, Zawodzinski T, Pivovar B. A
57. Maillard F, Dubau L, Durst J, Chatenet M, André J, half cell model to study performance degradation of
Rossinot E. Durability of Pt3 Co/C nanoparticles in a a PEMFC due to cationic contamination. ECS Trans
proton-exchange membrane fuel cell: direct evidence 2007, 11:777–788.
of bulk Co segregation to the surface. Electrochem 70. Kienitz BL, Baskaran H, Zawodzinski TA Jr. Model-
Commun 2010, 12:1161–1164. ing the steady-state effects of cationic contamination
58. Jia QY, Ramaker DE, Ziegelbauer JM, Ramaswamy on polymer electrolyte membranes. Electrochim Acta
N, Halder A, Mukerjee S. Fundamental aspects 2009, 54:1671–1679.
of ad-metal dissolution and contamination in low 71. Bas C, Reymond L, Danerol AS, Albérola ND,
and medium temperature fuel cell electrocatalysis: Rossinot E, Flandin L. Key counter ion parameters

© 2014 John Wiley & Sons, Ltd.

Advanced Review wires.wiley.com/wene

governing polluted nafion membrane properties. acid ionomers: why they perform better in PEM fuel
J Polym Sci Part B Polym Phys 2009, 47:1381–1392. cells. J Power Sources 2008, 178:499–509.
72. Bas C, Albérola ND, Flandin L. Effects of contaminant 84. Elabd YA, Napadensky E, Sloan JM, Crawford DM,
on thermal properties in perfluorinated sulfonic acid Walker CW. Triblock copolymer ionomer membranes:
membranes. J Memb Sci 2010, 363:67–71. Part I. Methanol and proton transport. J Memb Sci
73. Di Noto V, Zawodzinski TA, Herring AM, Giffin 2003, 217:227–242.
GA, Negro E, Lavina S. Polymer electrolytes for a 85. Hsu WY, Barkley JR, Meakin P. Ion percolation and
hydrogen economy. Int J Hydrogen Energy 2012, insulator-to-conductor transition in Nafion Perfluo-
37:6120–6131. rosulfonic Acid membranes. Macromolecules 1980,
74. De Moor G, Bas C, Charvin N, Moukheiber E, 13:198–200.
Niepceron F, Breilly N, André J, Rossinot E, Claude E, 86. Moukheiber E, Bas C, Flandin L. Understanding the
Albérola ND, et al. Understanding membrane failure formation of pinholes in PFSA membranes with the
in PEMFC: comparison of diagnostic tools at different essential work of fracture (EWF). Int. J. Hydrogen
observation scales. Fuel Cells 2012, 12:356–364. Energy 2014, 39:2717–2723.
75. Huang XY, Solasi R, Zou Y, Feshler M, Reifsnider 87. Barany T, Czigany T, Karger-Kocsis J. Application
K, Condit D, Burlatsky S, Madden T. Mechanical of the essential work of fracture (EWF) concept for
endurance of polymer electrolyte membrane and PEM polymers, related blends and composites: a review.
fuel cell durability. J Polym Sci Part B Polym Phys Prog Polym Sci 2010, 35:1257–1287.
2006, 44:2346–2357. 88. Garnier G, Chehab B, Yrieix B, Brechet Y, Flandin
76. Moukheiber E, De Moor G, Flandin L, Bas C. Inves- L. On the essential work of fracture in polymer-metal
tigation of ionomer structure through its dependence multilayers. J Mater Sci 2009, 44:5537–5543.
on ion exchange capacity (IEC). J Memb Sci 2012, 89. Moukheiber E, Planes E, Bas C, Flandin L. The effect
389:294–304. of water and temperature on the formation of pinholes
77. Bas C, Flandin L, Danerol AS, Claude E, Rossinot E, in PFSA membranes. Submitted to Fuel Cells—from
Albérola ND. Changes in the chemical structure and Fundamentals to Systems (special issue).
properties of a perfluorosulfonated acid membrane 90. Mainka J, Maranzana G, Dillet J, Didierjean S, Lottin
induced by fuel-cell operation. J Appl Polym Sci 2010, O. Effect of oxygen depletion along the air channel
117:2121–2132. of a PEMFC on the Warburg diffusion impedance. J
78. Kelly MJ, Fafilek G, Besenhard JO, Kronberger H, Electrochem Soc 2010, 157:B1561–B1568.
Nauer GE. Contaminant absorption and conductivity 91. Martin JJ, PART I. In situ diagnostic tools—proton
in polymer electrolyte membranes. J Power Sources exchange membrane fuel cell testing. In: Wang HH, Li
2005, 145:249–252. H, Yuan X-Z, eds. PEM Fuel Cell Diagnostic Tools.
79. Jia R, Han B, Levi K, Hasegawa T, Ye J, Dauskardt Boca Raton: CRC Press; 2012, 3–13.
RH. Effect of cation contamination and hydrated pres- 92. Husar A, Serra M, Kunusch C. Description of gasket
sure loading on the mechanical properties of pro- failure in a 7 cell PEMFC stack. J Power Sources 2007,
ton exchange membranes. J Power Sources 2011, 169:85–91.
196:3803–3809. 93. Flandin L, Danerol AS, Bas C, Claude E, De-Moor
80. Brown ID. Chapter 14. The bond-Valence method: an G, Albérola N. Characterization of the degradation
empirical approach to chemical structure and bonding. in membrane electrode assemblies through passive
In: O’Keffe M, Navrotsky A, eds. Structure and Bond- electrical measurements. J Electrochem Soc 2009,
ing in Crystals, vol. II. New York: Academic Press, Inc.; 156:B1117–B1123.
1981, 1–30. 94. Zhang S, Yuan X-Z, Hin JNC, Wang H, Wu J,
81. Moukheiber E. Compréhension par établisse- Friedrich KA, Schulze M. Effects of open-circuit oper-
ment de courbes d’étalonnage de la structure des ation on membrane and catalyst layer degradation in
membranes perfluorées sulfoniques pour pile à proton exchange membrane fuel cells. J Power Sources
combustible. Matériaux, Mécanique, Génie Civil, 2010, 195:1142–1148.
Electrochimie 2011. Vol. Ph.D., p. 188. Available at: 95. Vengatesan S, Fowler MW, Yuan X-Z, Wang H.
http://tel.archives-ouvertes.fr/tel-00873025. Diagnosis of MEA degradation under accelerated
82. Moukheiber E, Bas C, Albérola ND, Flandin L. relative humidity cycling. J Power Sources 2011,
Infrared and thermal behaviour of proton exchange 196:5045–5052.
membrane (PEM) after cationic contamination. J 96. Burlatsky SF, Gummalla M, O’Neill J, Atrazhev VV,
Memb Sci 2013, 431:105–112. Varyukhin AN, Dmitriev DV, Erikhman NS. A math-
83. Kreuer KD, Schuster M, Obliers B, Diat O, Traub ematical model for predicting the life of PEM fuel cell
U, Fuchs A, Klock U, Paddison SJ, Maier J. membranes subjected to hydration cycling. J Power
Short-side-chain proton conducting perfluorosulfonic Sources 2012, 215:135–144.

© 2014 John Wiley & Sons, Ltd.

 WIREs Energy and Environment A review of PEM fuel cell durability

97. Kusoglu A, Karlsson AM, Santare MH, Cleghorn S, 99. Tang Y, Karlsson AM, Santare MH, Gilbert M,
Johnson WB. Mechanical response of fuel cell mem- Cleghorn S, Johnson WB. An experimental investi-
branes subjected to a hygro-thermal cycle. J Power gation of humidity and temperature effects on the
Sources 2006, 161:987–996. mechanical properties of perfluorosulfonic acid mem-
98. Kusoglu A, Karlsson AM, Santare MH, Cleghorn S, brane. Mater Sci Eng A 2006, 425:297–304.
Johnson WB. Mechanical behavior of fuel cell mem- 100. Liu H, Gasteiger HA, Laconti A, Zhang J. Fac-
branes under humidity cycles and effect of swelling tors impacting chemical degradation of perfluori-
anisotropy on the fatigue stresses. J Power Sources nated sulfonic acid ionomers. ECS Trans 2006, 1:
2007, 170:345–358. 283–293.

© 2014 John Wiley & Sons, Ltd.