MECHANISM OF PINHOLE FORMATION IN MEMBRANE ELECTRODE

ASSEMBLIES FOR PEM FUEL CELLS

Vesna Stanic
Teledyne Energy Systems, Inc., West Palm Beach, FL, 33417

Mark Hoberecht
NASA Glenn Research Center, Cleveland, OH, 44135

ABSTRACT

The pinhole formation mechanism was studied with a variety of MEAs

using ex-situ and in-situ methods. The ex-situ tests included the MEA
aging in oxygen and MEA heat of ignition. In-situ durability tests were
performed in fuel cells at different operating conditions with hydrogen and
oxygen. After the in-situ failure, MEAs were analyzed with an Olympus
BX 60 optical microscope and Cambridge 120 scanning electron
microscope. MEA chemical analysis was performed with an IXRF EDS
microanalysis system. The MEA failure analyses showed that pinholes and
tears were the MEA failure modes. The pinholes appeared in MEA areas
where the membrane thickness was drastically reduced. Their location
coincided with the stress concentration points, indicating that membrane
creep was responsible for their formation. Some of the pinholes detected
had contaminant particles precipitated within the membrane. This
mechanism of pinhole formation was correlated to the polymer blistering.

INTRODUCTION

Durability of polymer electrolyte membrane (PEM) fuel cells is one of the main
problems impeding commercialization of this technology for any application. Reliability
of power systems based on PEM fuel cell technology is mostly dependent on membrane
electrode assembly (MEA) durability. The probability of PEM fuel cells to replace other
energy conversion devices will definitively increase if longer life is verified by achieving
durability targets. This is especially important for space applications where safety and
reliability are the most important factors. In space applications PEM fuel cells operate on
pure hydrogen and oxygen, a very aggressive environment that affects durability
drastically. The minimum required life for space applications is 10,000h with other
application being much higher. This is still a challenge to be demonstrated at the system
level.

The most common failure mode of PEM fuel cells is gas crossover caused by pinhole
formation in MEAs. Possible reasons for pinhole formation are material flaws introduced
during MEA processing and conditions imposed during fuel cell operation. There are
several different explanations reported to date for pinhole causes. One of the earliest
attempts to identify them was done by Irvin et al.[l]. They identified small cracks in two
areas and correlated their formation to the local membrane shrinkage caused by drying.
The membrane drying was caused by limited water transport due to the decomposition of
This is a preprint or reprint of a paper intended for presentation at a
conference. Because changes may be made before formal
publication, this is made available with the understanding that it will
not be cited or reproduced without the permission of the author.
Dacron@ wicks. In the study performed by La Conti et al. [2] failure of MEAs was
assigned to peroxide formation in membranes. Peroxide radicals formed
electrochemically in the membrane degraded the polymer, caused its dissolution, and
failure. In addition to membrane drying and peroxide attack, the most recent work [3]
correlated the surface roughness of gas diffusion layers (GDLs) with the number of
pinholes. Stucki et al. [4] demonstrated that pinholes formed not only in PEM fuel cell,
but also in PEM electrolyzers. They explained that pinhole formation was caused by
membrane dissolution triggered by local stress.

The approach used in this paper was based on assumptions that the membrane aging,
excessive heat generated locally, fuel cell configuration (design and components), and
operating conditions mutually contribute to the pin hole formation. The objectives were
to identify the causes and failure mechanism of pinhole formation by studying the effects
of these factors individually. For this purpose the MEA mechanical degradation in
oxygen and heat of ignition were studied separately in ex-situ tests, while the effects of
operating conditions and fuel cell configurationswere evaluated in various fuel cell tests.
The MEA failure modes were analyzed by optical and scanning electron microscopes.

EXPERIMENTAL

MEAs studied

For this study MEAs with extruded and cast polymer membranes were used. They
were made from 1100EW persulfonated tetrafluoroethylene (Nafion type) polymer. The
-
membrane thicknesses ranged from 30 pm to 170 pm. The hydrogen and oxygen
electrocatalyst was platinum supported on carbon. Gas diffusion layers used were Toray
TGHP-90 carbon fiber paper. More than 120 MEAs were tested in -
30 fuel cell
durability trials.

Ex-situ test procedures

Aging of three different MEAs with cast and extruded membranes was performed in
wet and dry oxygen at 70°C and 3Opsig. They were assembled in a fuel cell and tested
without electrical load. Each experiment was carried out for 1000h. They were stopped
-
every lOOh and small MEA samples were cut out for microstructure analysis. When the
experiments were completed, the MEA bursting strength was tested [ 5 ] . This test was
performed on unsupported samples in a specially designed tool. The nitrogen gas was
pressurized from one side of the sample at a rate of 2Opsi/s until it burst.

MEA heat of ignition was determined by a cone calorimeter test [6]. This test
measured MEA properties in response to heat and flame under controlled conditions. The
properties included heat flux, time to ignite the MEA, and temperature on the MEA
surface at ignition. Two MEAs, one with a cast and a second with an extruded membrane,
with the same catalyst loadings were tested. The thickness ratio of these two MEAs was -
1:1.8.
, '.

In-situ test procedure

All MEAs selected for this durability study were tested in 4-cell fuel cell stacks with
hydrogen and oxygen reactants. More than 30 tests were performed at different operating
conditions. To complete tests within a reasonable experimental time, the conditions were
selected to accelerate the MEA failure. Assuming that reactant relative humidity is a
primary factor determining MEA life, failure location and mode, two accelerated test
matrices with different gas inlet relative humidities were generated. One set of
-
accelerated tests was performed with 100% humidified reactants, while the second set
was performed with dry gases. The stacks were tested at constant current load. Variables
identified in these tests as additional key factors for MEA durability, were changed and
implemented in long-term tests.

The stacks were periodically leak checked for gas crossover. The measurements were
performed on non-operational stacks at room temperature with nitrogen at 5 psid. When
-
the gas cross over was 10 times MEA intrinsic permeability [ 7 ] , it was considered that
the MEA failed the test.

MEA failure analysis

The MEA microstructures were evaluated after failure in fuel cell tests. Evaluation of
failure modes was done on MEA cross sections at the regions weakened during fuel cell
operation. Other samples were crushed in liquid nitrogen for cross section analysis.

The analysis of MEA failure modes was performed with an Olympus BX60 optical
microscope in different light modes and magnifications. For MEA examination at higher
magnifications, a Cambridge Instruments Stereoscan 120 scanning electron microscope
was used. Chemical analysis was done with an IXRF EDS 2004 microanalysis system.

RESULTS AND DISCUSSION

MEA Aging in Oxygen

MEA electrocatalyst microstructure changed during tests with dry and wet oxygen.
Dry oxygen caused catalyst cracking, while humidified gas increased surface roughness.
The changes that occurred during test with dry gas can be explained with MEA
shrinkage. MEA components (electrocatalyst layers and membrane) have different
shrinkage rates upon drying. Since catalysts layers are more porous, they dry much faster
than the polymer membrane. This difference may stimulate the formation and
propagation of cracks in catalyst layers. However, MEA swelling in wet oxygen increases
electrocatalyst roughness. When the membrane swells its thickness increases. Since it is
under constant compression, it fills out voids located on the surface of GDLs. This
process creates bulges and indents on the catalyst layer, resulting in rougher surface.

The results of burst tests with MEAs aged in dry and wet oxygen show that the MEA
mechanical properties did not degrade during these tests. The burst strength for MEAs
-
with 50pm thick cast and extruded membrane is 620 Wa. This result is in agreement
with once previously reported [8] that also indicates the resistance of Nafion type
polymers to the chemical attack of oxygen.

MEA heat of ignition

Cone calorimeter test results for two MEAs tested are presented in Table I. The same
heat flux was needed to ignite MEAs (1.3 W/cm2 ), even though they had different
membranes. However, time to ignition was different and proportional to their thicknesses.
Thus, the ignition heat of MEA 2 was 1.8 times higher then of MEA1.

MEA MEMBRANE THICKNESS HEAT TIME TO IGNITION TEMPERA

FLUX IGNITION HEAT TURE AT
IGNITION
(Pm) (W/cmz) (Seconds) (Ws/cmz) ("C)
1 CAST T 1.3 6.81 8.85 296

Assuming that MEA ignition heat is mostly determined by polymer membrane, and
not by catalyst layers, the polymer heat capacity was calculated from the ignition heats
(Table I) and corresponding membrane weights. The heat capacity calculated at room
temperature and ambient pressure for both cast and extruded polymer membranes was the
same 735J1g.K.

When ignition heat of MEAs is compared to the heat generated during fuel cell
operation at test conditions used, it is one to two orders of magnitude higher than the heat
generated in fuel cell. For instance, a typical heat loss of MEA 1 is 0.13 W/cm2. If there
-
is no heat loss by cooling or water evaporation, then the heat generated in fuel cell is 70
times lower than the heat necessary to ignite MEA 1. This result indicates that heat
generated locally at the MEA active sites is not high enough to create instantly a hotspot.
-
For thicker MEA 2 this heat ratio is even higher, 120.

MEA failure analysis

The results of MEA microstructure analyses after fuel cell tests showed that the MEA
failure modes were pinholes and tears. These defects were localized and usually aligned
along the channel edges. The failure location depended on operating conditions and fuel
cell configurations, and not on MEA type and thickness. When different MEAs were
tested in accelerated tests with humidified gases in fuel cells with the same configuration,
the location of defects was always the same. However, when fuel cell configuration was
changed, the failure location shifted to a different area.
 Accelerated tests vvith dry gases were cmied out with As md k e %cell
failure foeations

h ~ i ~ and ~ ~ e d
s cbse to the reactant in8etss,Since the e ~ e ~ t r ~ c hr e~~~cit ~c orate
a n~ is
t i oMEAs
~ o ~ c e ~ t r athe ~ ~ have bigher reaction rates at these
nsities and further

Figure 1: A typical pi de catalyst layer. The p i ~ o l is

e -2Opm wide
and s ~ ~ o with ~ b~a ee d dcatalyst. Lines in ~ ~ r oare~carbon r a fiber
~ ~~~ d e ~ t a t i o n s .

Tears in MEAs a e created when an MEA is taken out of a fuel cell and exposed to
e to drying in air at room temp
mder tension. This forc n~ugh to tear the MEA only at
analysis of fractured aces of teas indicated that they had aligned pinholes,
spots, cracks, md con inants. Cracks, thin spots md c o n ~ ~ iare ~ tMEA
~ the s
damages that precede the pidole genesis. The t e a analyses indicate that membrane
i o n the m e ~ b r cause
creep and c ~ n ~ i~ rne c~i ~ti ~ a twithin ~e ~ ~ ation~ ion. o ~ e
 crosc5~eof cross section af
m e ~ ~ rA)~ newe : MEA; B) MEA thickness is reduced by 40% m d n ~ ~ - ~ nafter
i f o ~
fael cell test.

M i c r o s ~ c t ~ ea ~ ~of sMEAs
i s tested with dry gases show that ~ e ~~ c k~n e sr s ~ e
is a h reduced under these c ~ n ~ ~8xrt the ~ shmd,
i ~other ~ the t h ~ c ~n~ os s~ - ~ nis ~ f ~ ~ i ~
less apparent at the m ~ c ~ ~ s cscale, n ~~f m e ~ ~ swelling
o ~ i ci n d i c ~ t ~lack r ~ e in a
cell ~ n v ~ o ~ e n t .

The results of MEA failme by creep are pinholes. The m e c h ~ of ~ ~ f o ~ a t is~ o ~

s their
affected with the ~ e ~ b r e ~ m d eh ~ ~ clevel ~ me exposed to in the fuel cell.
l i they
There are ~o m ~ c ~caused ~ ibyscreep.~ ~The first implies extruded m e m ~ r ~ate smy
~ ~mcl cast
~ ~ e ~d b rtested r ~ o~
s ~ dry gases ~ m ~ e creep
~i e with ~This type
~ of e .
creep failwe involves crac ~ e ~ e ~ owhen e ncritical
~ ~ the t ess reductio~is reached.
this critical n ~ b ise75% ~ for both exmd cast m e ~ b r ~ Cracks
e$~
de~ectedin fractured s in both exmde cast ~ e ~ b r a nare
es
on mode side. n e y agate under con
reate a le& path for the gases. Since h y ~ o g e nm d ~ x y ~ chemica~en
e ~ o ~ e(285 ~ ikJ/mo1e
c of h y d r o ~ e nhe2t
~ , produced even with small
~ y d r o ~ ec~ossovers
n can i ite the MEA. Thus, to create a pinhole in an e
~ e ~ b r a rnresented
e in Fig. TSb, a h ~ ~ r o flow
~ e nrate should be only
flow rate is cakedated based on the pinh51e size m d MEA nitio ion heat (Table 0,

ing electron ~ e r o s c ~ ~ e ed surlt2ce oftear in MEA. This MEA

mbrme and is tested wi ed gases: A) two
from mode to cathode side. The ~ e ~thickness ~ r is reduced
~ e by
A pidlole in the same MEA

The second ~ e c ~ o fapinhsle ~ s f o~~ a t i o nby creep that occurs in cast ~ e m b r ~ e s

tested with ~ ~ i ~gases i ~is caused
e d by m e ~ ~ r a thichess
ne r e ~ u c ~ i o~n ~ ~The~ ~ n g
results of ~ i c r o s t ~ c malysis
t~e showed that ~ e ~ t h ~ c ~re decrease s s~ e by ~~~

before it collapses. Spots only 5pm thick are detected in a cast ~ e ~ ~m re n~ e .

~ m e it becomes too thin, and is no longer a barrier to the mixing of gases.
~ e ~ b colla~ses
They react c ~ e ~ c a €and
l ycreate a pihole.

The failwe analysis o EAs show that tears md ~ ~ ~ o localized. ~ e s Fuel cell
e n ~ ~indi
e ~ ~ e r i m results that they appear in areas where EA is ~echanical1y
m d ~ h e ~ moverstressed.
al~~ The anisotropic c o ~ p r e s s ~ omnd heat d i s t ~ b ~ t in n cell
~ ofuel
at the ~ a c r o s ~ level,
o ~ ~caused
c ~ ~ r c o ~ i ~ ~ aand
by ~ m p r o stack ~ iwater
on ~ana~e~en~,
create MEA. areas that are more active and thus generate more heat. Any i ~ ~ e ~ e c ~ ~ o ~
that exists in the MEA 513the microscale results in ~ i ~ o~ l oe ~ a tThe ~ . for this
i ~reason
i s &e ~ccelera~ion 5f creep [lo]. Thus, locations that operate under hi er c ~ m ~ ~ e s s i 5 n
md t ~ ~ p e r a fail
t ~ faster.
e Stress ~ o ~ ~ e n ~points
~ a ~ in n MEA may be caused by
i othe
GBL surface ~ o ~ ~ ~broken e s § carbon
, fibers, or debris localized at the ~~~~B~
 In a d ~ t i o nto failure caused by ~ e creep, ~ As ~ ~S failed Q ~due to ~ e
c o n t ~ i n preci~itate~
~ t ~ within &e m e ~ ~ r The ~ e sc. eleckon ~ i ~ r o ~ a ~ h ~
cross s e c ~ ~ owith
n c ; i - a o n t ~ i nparticles.
~t The
ed surface of a tea. The micro

e ~ a ~ i c lwith
e s similar sizes, The
ter- Similar, but smaller b
en the adjacent ox

A ~ o n t ~ i ~ ~ awas
t i oa nreason for gi~hole~ o ~ a t i theo n %el
~ cell
losses were i ~ e a s ~ a %Indeed, l e ~ they d e ~ o n s ~ a t ead%I
increase of 20 p V h The volt e increase was directly ~ r o ~ to~the~ i o ~ ~ ~
r e s i s t ~ c edecay. The resist e drog~edfrom 0.983 to 0 * ~ ~ This ~ result
~ ~ m ~ '
i~dicatedthat the membr ~ n t ~ i n a nwere~ s lo in n ~ o wIEQSPeS dong the
c h a e l s where the MEA failed.

The results o f EDX c ~ e ~ i canalysis

al of c o ~ t ~particles
i ~ ~d et ~ o n s ~ athat
t e they
contain a metal oxide, ~ivalentcation I%IH~+ is leached out tpy liquid water that C Q M ~ Sinto
colltac$With a c Q n t ~ i lSOWCe. l ~ ~ The I%IH2+ laces EX+ in the membrane and m
n with -SO3-. Since the sulfonic 6-
crystal n ~ c l e a t i ~sites goups are Exed to the
~ o l ~~eo lre c ~ l ethey s , c m o t ~ a ~ i ~ iin~ the
a t crystal
e . Thus, the metal oxide
c ~ ~ t a ~c ~o ~ t e~ si ton ~ e within the ~ e m % rbeing ~ e constant~ys ~ ~ ~ lby ied
liquid water and oxyg &e gas stream, In the areas where the cation con
er, such as the inlet channels of oxygen and h y ~ o ~the e ~crystals
, are much larger
ng the ~ ~ whereathe cation
~ ~;i-aon~entration
~ ~ is d s
9

Figure 4: ~ c ~ electron
n g graphs of c o n ~ ~ iparticles
n ~ t m d blisters in
~ e al
~ e m b r created
e ~two small
m e m b r ~ c> particles in the
ole created due to ~ o n t ~p rie ~~i ~~i ~t tand
ion

The presence of co cles in the ~ e ~ causes ~ blister

r ~ f oe ~ a ~ i o n ~
~ ~ i §is ~a well-ho
e r ~ ~ ~ e~ e c ~ a B n[P
~I ~ mc m also be used to explain
the ~ ~ ~ o ~~ a t oini oNafion
~~ type e m e m ~ r a ~ gThe. icles that exist ~i~ the
e small voids ~o~~ them These voids are filled with ~ i water.
m e m ~ r ~create ~ ~ ~ d
en. the cell is heated from room t e m p e r a ~ ~
toethe he8 cell Q p e r a ~ itne~m p e r ~ the
~g~
water vapor partid. presswe increases in tbese voids, an expmds them, Since the
processes of the pmicle and blister ~~~~ are c o n ~ the ~ ~small~ blisters
o ~ ~ (Fig. 4b)
will eventually coalesce md large ones Fig, 4a). Large blisters such as this, allow
sults in their chemical reaction m d p h b k f ~ ~ a t i ~ ~
 CONCLUSIONS

Failure analyses of MEAs indicate that mechanical and chemical changes in MEA
cause pinholes. The mechanical changes are related to the membrane creep. Depending
on the membrane type and humidification level in fuel cell tests, two different
mechanisms of pinhole formation by creep are identified. The first mechanism includes
crack development and propagation that leads to the reactant gas crossover. This
mechanism is typical for extruded membranes at any humidity condition as well as cast
membranes at dry conditions. The creep of cast membranes tested with humidified gases
results in a different mechanism. It involves the reduction of membrane cross sectional
thickness until it no longer acts as a gas barrier.

Accelerated durability tests with dry and humid hydrogen and oxygen suggest that
anisotropic distribution of compression and heat across the MEA active area causes
failures. These non-uniform conditions are the result of the fuel cell configuration and
operating conditions. However, the stress concentration points localized at the
MENGDL interface cause compression and heat variation on the microscopic level.
When their effect is superimposed on the heat and compression variations on the macro
scale, then pinholes are created in localized areas of MEA.

In addition to pinholes created by creep, some are formed due to MEA chemical
failure. This failure is a consequence of proton replacement in the polymer membrane
with larger cations leached out with liquid water from the fuel cell or system components.
This mechanism of pinhole formation occurs through several steps. The first step
includes contaminant particle nucleation and growth within the membrane from cations
and oxygen provided with fluid streams coming into the fuel cell. When they begin to
form, the particles are surrounded with small voids. During fuel cell therm0 cycles, these
voids enlarge and create blisters. Blisters create a path for gases to crossover and mix.

The weak spots in membrane created by either mechanical or chemical failure mode,
such as cracks, thin spots or blisters, allow gas mixing. The chemical reaction of
hydrogen and oxygen generates enough heat to melt the polymer, bum the catalyst and
form a pinhole in MEA.

ACKNOWLEDGEMENT

The authors gratefully acknowledge the financial support of NASA Glenn Research
Center (contract # NAS 3-02093 and NAS 3-02203).

REFERENCES

1. L. F. Irvin, Evaluation of an Ion Exchange Membrane Fuel cell For Space Power,
(1974) p. 1-13.
2. A. B. La Conti, A.R. Fragala, and J. R. Boyack, in Proceedings of the Symposium on
Electrode Materials and Processes for Energy Conversion and Storage, J. D. E.
 Mcintyre, S. Srinivasan, and F. G. Will, Editors, p.354, The Electrochemical Society
Proceeding Series, Princeton, NJ (1977).
3. V. Peinecke, J. von der Osten-Fabeck, J. R. Gordon, H. R. Haas, US Patent 6,716,551,
April 6,2004.
4. S. Stucki, G. G. Sherer, S. Schlagowski, E. Fischer, J. Applied Elchem., 28, 1041
(1998).
5. ASTM D3662.
6. ASTM E-1354-97
7. J. L Larminie, A. Dicks, Fuel cell Systems Explained, p.47, John Wiley & Sons, Ltd.,
Chichester (2000).
8. S. Stucki, G. Theis, R. Kotz, H. Devantay, H.J. Christen, J. Electrochem. SOC.,132,
1041 (1985).
9. J. Scheirs, Composition and Failure Analysis of Polymers, p. 3 15, John Wiely & Sons,
Ltd., New York (2000).
10. D. R. Askeland, The Science and Engineering of Materials, p.488, PWS Publishing
Company, Boston (1994).
11. J. St-Pierre, D.P. Wilkinson, S. Knights, M.L. Boss, J. New Mat. Electrochem.
Systems, 3,99 (2000).
 . *. .
4

Keywords: MEA pinholes, MEA durability, Pinhole formation, PEM durability