You are on page 1of 11

Available online at www.sciencedirect.

com

Polymer Degradation and Stability 93 (2008) 214e224


www.elsevier.com/locate/polydegstab

Comparison of two accelerated Nafion degradation experiments


Sumit Kundu, Leonardo C. Simon*, Michael W. Fowler
Department of Chemical Engineering, University of Waterloo, 200 University Avenue West, Waterloo, Ontario, Canada N2L 3G1
Received 7 August 2007; received in revised form 4 October 2007; accepted 7 October 2007
Available online 12 October 2007

Abstract
This work describes a systematic investigation of the degradation of Nafion 112 membranes using a Fentons accelerated aging experiment.
Two variations of the experiment were compared: a solution method where iron ions and peroxide exist together in solution prior to the addition
of Nafion, and an exchange method where Nafion in the Fe2 form is exposed to hydrogen peroxide. Accelerated aging experiments were
conducted over 3e5 days. Weight loss, fluoride ion release, ion exchange capacity, intrinsic viscosity, morphological characteristics, and dimensional changes were measured. FTIR spectra, mechanical properties and membrane barrier properties were also investigated.
2007 Published by Elsevier Ltd.
Keywords: PEM fuel cell; Nafion; Degradation; Fentons test

1. Introduction
Polymer electrolyte membrane fuel cells (PEMFCs) fueled
with hydrogen are an emerging alternative energy technology
with possible applications ranging from cell phones to homes
and vehicles. These devices generate power by electrochemically reacting hydrogen and oxygen in the presence of a catalyst. On the anode, hydrogen separates into protons and
electrons. The protons pass through the ion conducting polymer electrolyte membrane (PEM), which makes up part of
the membrane electrode assembly (MEA), to the cathode
where they recombine with the electrons and oxygen, typically
from air, to produce electricity, water and heat. Along with the
role of conducting protons, the PEM must also act as a gas barrier between the anode and cathode and provide mechanical
strength to the MEA. A schematic of the different parts of
a fuel cell is shown in Fig. 1.
Though there are many parts of a fuel cell system that can
influence reliability and durability, degradation of the membrane electrode assembly, and in particular the polymer

* Corresponding author. Tel.: 1 519 888 4567x33301.


E-mail addresses: s2kundu@uwaterloo.ca (S. Kundu), lsimon@uwaterloo.
ca (L.C. Simon), mfowler@uwaterloo.ca (M.W. Fowler).
0141-3910/$ - see front matter 2007 Published by Elsevier Ltd.
doi:10.1016/j.polymdegradstab.2007.10.001

electrolyte membrane, is the most important. Furthermore,


failure of the membrane can lead to safety issues with the
overall system. The most common fuel cell electrolyte used today is Nafion, a perfluorinated sulfonic acid polymer whose
chemical structure is shown in Fig. 1. Nafion consists of
a hydrophobic Teflon-like backbone, which provides mechanical stability, and side chains ending in hydrophilic sulfonic
acid groups that allow proton conduction when the membrane
material is hydrated.
There are many causes and degradation modes related to
failure of polymer electrolyte membranes [1e7]. There has
recently been considerable research on the chemical degradation of the polymer electrolyte during normal fuel cell operation. It has been proposed that carboxylic end groups left over
from the Nafion manufacturing process may be susceptible
to attack by radical species generated during fuel cell reactions
[8]. The proposed mechanism is as follows:
Step 1: ReCF2COOH OH / ReCF2 CO2 H2O
Step 2: ReCF2 OH / ReCF2OH / ReCOF HF
Step 3: ReCOF H2O / ReCOOH HF
The radical species, such as hydroxyl radicals, are thought
to be formed by the decay of hydrogen peroxide which is an
intermediate of the electrochemical oxygen reduction

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

215

Fig. 1. Schematic of a polymer electrolyte membrane fuel cell.

reaction; additionally it has been proposed that oxygen may


permeate through the polymer electrolyte membrane and react at the anode to also produce peroxide species [1,8e11].
The presence of hydroxyl radicals has been confirmed by using spin trapping techniques which can detect the species in
situ [9]. There has also been evidence suggesting that peroxide may not be the sole degrading species in a fuel cell [12].
Membrane degradation, by hydroxyl radical attack, has
been studied by ex situ accelerated methods. Accelerated testing is important to perform on these materials which can last
for thousands of hours under regular operating conditions. The
overall purpose of accelerated testing is to achieve the same
degradation effects in a shorter amount of time. Ex situ accelerated chemical degradation experimentation of fuel cell ionomers most commonly employs Fentons testing. Fentons
reagents are made by combining hydrogen peroxide with
Fe2 ions in order to produce radicals as follows:
Fentons reaction : H2 O2 Fe2 / Fe3 OH OH
There have been two main methods employed to study the
degradation of Nafion membranes by Fentons type reaction.
The first method exposes the membrane to a solution of peroxide and metal ions (solution method) while the second method
exchanges the metal ions with the acid sites of the polymer
before exposure to peroxide (exchange method).
Healy et al. [11] compared degradation products from samples of Nafion degraded with a Fentons solution containing
4e16 ppm Fe2 and 29% H2O2 with degradation products
from an in situ experiment. Using F19 NMR they found that
membranes degraded in a Fentons reagent released chemical

compounds that shared many chemical signals as those


released by membranes degraded during fuel cell testing.
Not only did they identify fluoride ions in the Fentons solution water but also identified a fluorinated species with similar
characteristics as the side chains of Nafion. It was suggested
that as the fluorinated backbone of Nafion degraded it would
release the side chain components.
Wayne [13] studied the effect of Fentons testing with
Nafion 117 in peroxide solutions ranging from 12% to 24%,
iron concentrations of 2.8 mM (112 ppm) to 5.0 mM
(280 ppm), at temperatures ranging between 20  C and 70  C
with the longest immersion time being approximately 45 min
and the shortest being 10 min. No significant changes in weight
were observed, although there were significant changes to the
morphology. After degradation samples contained bubbles
which appeared to split the membrane and samples became
more opaque. The ion exchange capacity and the conductivity
of the membrane were not observed to have changed. Finally,
the membrane shrank slightly in one planar direction and elongated slightly in the opposite direction and the thickness increased from the production of bubbles.
Recently Tang and coworkers [14] performed degradation
studies with 30% peroxide and a combination of Fe, Cr, and
Ni to simulate ionic contaminants from stainless steel fuel
cell components. Testing was conducted up to 96 h between
80  C and 90  C with solution replenishment every 30 min.
They found that fluoride containing polymer fragments
entered into the degradation solution but no significant new
peaks in the FTIR spectra of the degraded membranes. They
further found that the membrane morphology changed significantly, developing bubbles with time.

216

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

Inaba and Kinumoto [10,15] performed similar work with


a Fentons reagent consisting of 30% H2O2. Instead of adding
iron and peroxide together and exposing Nafion to the mixture they first exchanged iron and other metallic cations into
samples of Nafion and then added peroxide. They also measured fluoride release and sulfate release with ion chromatography from which they estimated that almost 70% of the initial
number of CeF bonds were broken and almost 35% of SO
3
groups had degraded. Furthermore, a small amount of S]O
bonds was identified in the polymer film after degradation
using FTIR. A decrease in mechanical strength of the membrane and a loss of membrane thickness were also reported.
LaConti and coworkers [1] described how the rate of loss of
fluoride ions from Nafion was proportional to the logarithm
of peroxide concentration used in the Fentons solution.
Peroxide in the absence of iron particles has also been shown
to degrade samples of Nafion. Qiao et al. [16] observed that
exposure of Nafion to 30% H2O2 over a period of 30 days decreased its conductivity of the membrane and water uptake
characteristics. Furthermore, FTIR analysis of the membranes
showed the development of SeOeS bonds whose formation
was attributed to the oxidation of sulfonic acid groups.
To date there has not been a comparison of the differences between the solution and exchange methods of accelerated degradation. This is important when determining which method best
reflects in situ degradation and therefore is the better accelerated
degradation method. This paper reports on the ex situ degradation of Nafion 112 membrane using a Fentons reagent with
16 ppm Fe2, as well as Nafion in the exchanged Fe2
form, over multiple days and on the investigation of chemical,
morphological and mechanical changes in the membrane. The
goal is to examine any differences between the two methods
of degradation as well as develop a set of features for each
degradation method which can be compared against membranes
degraded in situ.

2.2. Fentons test


Two different styles of Fentons test were employed in this
study. For the primary test, which will be referred to as the
solution method, a solution of Fe2 and H2O2 was made to
which a sample of Nafion was added. This method is similar
to that reported by Healy et al. [11]. In the alternative method,
similar to the method reported by Inaba and Kinumoto
[10,15], Fe2 ions were exchanged into a sample of Nafion
which was then submerged in H2O2 and allowed to degrade.
The second method will be referred to as the exchange method.
The solution method used 30% H2O2 mixed with
FeCl2$4H2O to produce a 16 mg/L solution of Fe2 ions.
This solution (40 mL) was placed in a closed vial containing
the sample of Nafion. A Teflon prong (of negligible size)
was used to ensure that the sample of Nafion did not float
into the head space of the vial. Gas pressure was released
from the vessel through a needle piercing the septum in the
lid. The Nafion and solution were kept in a 72  C oven
and the solution of peroxide and iron was replaced by fresh solution changed approximately every 12 h. Chemical analysis
of this solution was therefore also done every 12 h. The solutions were stored in polyethylene bottles for fluoride analysis.
Every 24 h three samples were removed from the study, rinsed,
and saved for later study.
In the exchange method the samples of Nafion were
placed in a saturated solution of FeCl2$4H2O for 24 h. The
samples were then rinsed and placed into vials to which
40 mL of peroxide was added. The vials were kept in
a 72  C oven and replaced with fresh peroxide every 24 h.
Samples were removed periodically, rinsed and saved for later
study. Saved samples were rinsed and conditioned in 1 M
H2SO4 prior to further testing to remove any residual iron
and return the membrane to the H form.
2.3. Gravimetry

2. Experimental
2.1. Materials
The main material of study was Nafion 112 provided by
Ion Power Corporation. A 30% peroxide solution was obtained
from Caledon Laboratory Chemicals and analytical grade
FeCl2$4H2O was obtained from Fisher Scientific. Catalyst
coated membranes used in in situ experiments were supplied
by Ion Power Corporation and used Nafion 112 as the electrolyte material.
Samples of Nafion 112 were cut from a larger sheet and
had dimensions of approximately 7.5  2.5 cm. Directionality
in the membrane can be determined by swelling characteristics. The machine direction will refer to the direction of the
greatest swelling and the cross-direction will refer to the direction 90 from the machine direction. For further simplicity, the
width will describe the machine direction and the length the
cross-direction. All samples from a set were cut in the same
direction. Prior to degradation the samples were soaked in
1 M H2SO4 at 72  C for a minimum of 3 h.

Samples of Nafion were weighed by first drying the samples for 24 h at 72  C in open glass vials. The vials were capped in the oven and then samples were transported to a Metler
AE 160 balance with 0.001 g accuracy. Samples were then
removed from the vial and weighed.
2.4. Fluoride ion release
An Orion 611 Digital pH/millivolt meter was used with
a fluoride ion selective electrode to measure the concentration
of fluoride ions in sample water. Calibration curves ranged
from 0.001 M to 1 M. Both the concentration of fluoride
ions and the volume of sample water remaining after conducting the degradation experiment were used to determine the
total number of moles of fluoride released.
2.5. Ion exchange capacity (IEC)
Ion exchange capacity measurements were done by conditioning samples in 1 M H2SO4 at 72  C for a minimum of 3 h,

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

217

rinsing the samples and then placing them in a solution of 1 M


NaCl at 72  C for 24 h. The sample was then removed and the
remaining liquid was titrated with 0.01 M NaOH using an
ORION 420A digital pH meter.

were done with the Nafion samples in liquid water. Dry samples of degraded Nafion tended to be wrinkled, and in order
to obtain measurements of dimensions the samples were
gently pressed down with a glass slide.

2.6. Viscometry

2.10. Tensile testing

Samples of fresh and aged Nafion were dissolved into


solutions of 50/50 wt% ethanol/water at 250  C prior to
viscosity measurements similar to the procedure used in
Ref. [17]. This dissolution procedure used here is usual for
Nafion membranes and is not expected to cause significant
degradation. Prior to dissolution the aged Nafion samples
were conditioned in 1 M H2SO4 at 72  C, rinsed, and dried
to remove any iron and other ions from the membrane. Viscosity was measured in a constant temperature bath at 25  C using
a Cannon Ubbelohde glass viscometer. Relative viscosity of
the solutions was calculated using Eq. (1):

A Rheometrics Mini Mat 2000 tensile tester was used to


generate stressestrain data of materials. Samples of Nafion
were cut to a width of 0.6 cm and placed in the tension testing
fixtures with a gap width of 5 mm. The load cell range was
0e200 N. The strain rate was 2 mm/min at 25  C. The
Youngs modulus was estimated using the linear rise in the
stressestrain curve corresponding to elastic behavior.

hr

t  t0 1
t0 C

where t is the elution time of the polymer/solvent sample and


t0 is the elution time of the solvent. C is the concentration of
the polymer in g/dl. Concentration was determined by measurement of the mass of dry membrane dissolved into a measured mass of solvent of given density.

2.11. Dynamic mechanical thermal analysis (DMTA)


A Rheometrics DMTA V dynamic mechanical thermal analyzer was used to determine transition temperatures of the
Nafion films. Experiments were done using a rectangular
tension fixture with a gap width of 2.5 mm. The samples
were cut from the degraded samples to a width of approximately 0.6 mm as measured with a digital caliper. Experiments used a frequency of 1 Hz and 0.1 N initial static
force. The temperature ramp rate was 2  C/min at a constant
strain of 0.05%. Samples were tested from 40  C to 140  C.
2.12. Nitrogen crossover

2.7. Fourier transform infrared (FTIR)


Samples of fresh and degraded membranes were analyzed
with FTIR. Measurements were conducted on a Bruker Tensor
27 FTIR using an MVP 2 Series ATR accessory made by Herrick Scientific Products Inc. The spectrum was collected after
32 scans with a resolution of 4 cm1. Background subtraction
was used and measurements were conducted at room
temperature.
2.8. Scanning electron microscopy (SEM)
SEM analysis was carried out using an LEO SEM with field
emission Gemini Column. X-ray compositional analysis was
done using an electron dispersive spectroscopy (EDS) manufactured by EDAX. The detection limit of the EDAX system
is atoms with atomic weights equal to or larger than carbon.
Samples of Nafion or catalyst coated membrane (CCM)
were cut into squares of approximately 0.5  0.5 cm2 and
fixed to an aluminum stub with double sided conductive
tape. Cross-sections were made by freeze fracture from a strip
of sample submerged in liquid nitrogen. Once frozen, the sample was broken in half while still submerged. Samples were
also sputter coated with gold to improve conductivity.
2.9. Dimensional measurements
The dimensions of the Nafion films were measured using
digital calipers at room temperature. Swollen measurements

Nitrogen gas crossover measurements, the amount of nitrogen gas flow across the membrane with a given pressure differential (regardless of transport mechanism), were performed by
clamping Nafion samples between two 1/4 inch face sealed
fittings at room temperature. One fitting was pressurized with
nitrogen gas to 5 psi and the other fitting was attached to a bubble column at atmospheric pressure. Crossover was measured
by timing the progression of bubbles through the graduated
column. If no significant movement of the bubble (>0.1 mL)
was seen after 10 min, then the samples were said to have
no crossover.
3. Results
Samples of Nafion were degraded via two accelerated
degradation methods. In the solution method, samples were
exposed to a mixture of peroxide and Fe2 ions; in the
exchange method, samples of Nafion in the Fe2 form
were exposed to a peroxide solution.
3.1. Chemical changes in membrane structure
An initial sign of polymer degradation is weight loss over the
testing period. With both degradation methods the Nafion
membranes lost significant weight with time (Fig. 2). The
weight loss increased following and after 80 h of exposure
exceeded 20% of the original weight in both cases. This was particularly interesting since different amounts of iron are present

40.00
Degraded - Solution Method
35.00

Degraded - Exchange Method

Weight Loss

30.00
25.00
20.00
15.00
10.00
5.00
0.00

20

40

60

80

100

120

140

Time (h)

and solution renewal intervals were different between the two


methods. One explanation is that with the solution method
iron, peroxide, and polymer must be in proximity to each other
for degradation to occur. Radicals produced within the bulk solution may terminate by an alternate mechanism such as reaction
with another radical or with polymer fragments. On the other
hand with the exchange method iron is present within the polymer membrane and so a smaller amount of iron could potentially
cause the same amount of degradation in the polymer with this
method. The important item to note is that weight loss with both
methods produced similar results over the same period of time
with the procedures used.
Control samples, summarized in Table 1, established that
the observed accelerated degradation was due to the combination of iron and peroxide. Nafion samples were exposed to
water (Control 1) and a water/iron solution (Control 3). Both
these solutions showed little weight loss over the testing
time of 120 h though the sample in the water/iron solution
lost almost 5% of its original weight. A Nafion sample
was exposed to 30% peroxide/water solution (Control 2)
(without any Fe2) and showed no significant weight loss after
80 h. Finally, Teflon films (Control 4) were also exposed to
a 16 ppm Fentons solution to establish that the Teflon prongs
were stable under the test conditions. The results showed that
there was no significant change in weight of the Teflon films.
Fluoride ion release was measured with every solution
change over the course of the experiments and the cumulative
fluoride loss is shown in Fig. 3. Fluoride release is considered
Table 1
Comparison of weight loss of Nafion control samples, Teflon, and degraded
Nafion

% Weight loss
2.82
Cumulative fluoride Negligible
release (mmol)

120
Degraded - Solution Method
Degraded - Exchange Method
100
80
60
40
20
0

20

60

40

80

100

120

140

Time (h)

Fig. 2. Percent weight loss of Nafion 112 samples degraded by the solution
method and the exchange method.

Control 1
Nafion in
water
(120 h)

Average Cumulative Fluoride Loss (mol)

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

218

Control 2
Nafion 30%
HOOH
(96 h)

Control 3
Nafion
H2O Fe2
(120 h)

Control 4
Teflon
Fentons
(96 h)

1.18
0.87

5.39
Negligible

0.13
0.06

Fig. 3. Cumulative fluoride loss from membrane samples degraded by the


solution and exchange methods.

to be an indicator of membrane degradation at the molecular


level. It would be expected that for similar weight losses a similar
amount of fluoride would be released. However as can be seen in
Fig. 3 the solution method of degradation lost more fluoride ions
cumulatively than the exchange method of degradation. One
possible explanation is that small polymer fragments make up
part of the mass leaving the polymer films and entering into
the surrounding solution. In the case of the solution method
these polymer fragments can continue degrading since iron
ions and peroxide exist in the surrounding solution at high concentrations. This is not the case with the exchange method since
all the iron is bound in the polymer. The samples degraded by the
solution method for 120 h had a cumulative fluoride ion measurement of 100 mmol or 1.5% of the available fluorine in the
polymer structure. Samples degraded by the solution method
measured 34 mmol of fluoride ions cumulatively or 0.5% of
the available fluorine. As discussed, it is expected that more fluorine atoms were still bonded in chain fragments leaving the
membrane. Fluoride release from each of the control samples
was negligible (Table 1).
Ion exchange capacity (IEC) measurements and chemical
analysis using electron dispersive spectroscopy (EDS) were
performed on the degraded membranes in order to determine
if the ratio of backbone to side chain groups had been changed
upon degradation. The results showed no significant change in
the ion exchange capacity or in the atomic ratio of oxygen to
fluorine (O/F ratio) over the exposure time for both the solution and the exchange methods. The value for the IEC was
1.1 mmol/gdry-membrane for the fresh membrane and the O/F
atomic ratio was 0.18. Similar behavior was observed for sulfur/oxygen ratio and sulfur/fluorine ratio. Other report in the
literature [10,15] based on ion chromatography data suggested
that the backbone components degrade to a larger extent than
the sulfonic acid groups, thus indicating that IEC should increase with degradation. This difference is probably due to
the fact that if sulfur containing chain segments may not be detected by ion chromatography. The results presented here indicate that the polymer remaining in the membrane had a similar
average chemical composition to the original fresh polymer.

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

8
Fresh Nafion 112

Reduced Viscosity (dl/g)

Degraded - Solution Method (96h)


Degraded - Exchange Method (84h)

S-O
A

1700

1600

1500

1400

1300

1200

1100

1000

900

800

Wavenumber (cm-1)
Fig. 5. FTIR spectra for fresh Nafion membrane samples, a sample degraded
via the solution method, and a sample degraded via the exchange method.

Evidence of chemical changes in the membranes was investigated by FTIR. Typical curves of the fresh and degraded
spectra are shown in Fig. 5. The main peaks of interest are
at 1147 cm1 and 1204 cm1 which are attributed to CeF
vibration [15,22] and at 1054 cm1 which is due to SeO
stretching [16,23]. The peak at 1147 cm1 was used to normalize the curves for comparison purposes. The FTIR results
of degraded membranes and control samples reinforce IEC
and EDAX findings which showed no significant change in
the relative numbers of side chains to backbone segments in
degraded samples as compared to fresh Nafion. FTIR analysis of the membranes degraded in this experiment by both
methods as well as the control samples could not detect any
new peaks which could be correlated with exposure time despite the high level of degradation. This was also observed
by Tang and coworkers [14].
The chemical characterization of Nafion 112 membranes
degraded by a Fentons solution indicates that the only significant change to membranes degraded by OH radicals in this
way was the average molecular weight. Changes to chemical
structure, such as changes to the ratio of side chain groups
to backbone groups as well as changes in the chemical bonds
present in the membranes, were not detected here. Nevertheless, exposure to the Fentons reagent severely degraded the
membranes as seen by the loss in weight and fluoride loss data.
3.2. Changes in morphology and dimensional
characteristics

5
4
3
2
1
0

A Fresh Nafion 112


C-F
B Degraded - Solution Method 120 h C-F
C Degraded - Exchange Method 84h

Normalized Absorbance

Although the ratio of oxygen atoms in the side chains to


fluorine atoms in the backbone did not change, degradation
could have removed backbone and side chain segments in
the same proportion. Viscosity measurements were carried
out to determine if the average molecular weight of the polymer membranes were changing. Samples of fresh Nafion
112, degraded sample from the exchange method experiment
(84 h of degradation) and from the solution method (96 h of
degradation) were dissolved in 50/50 by weight mixtures of
ethanol and water. These samples were chosen because of their
high degree of degradation and similar weight loss. All samples were in the hydrogen form and reduced viscosity was calculated by Eq. (1). Plots of reduced viscosity against the
concentration of Nafion 112 are shown in Fig. 4. The plots
show an almost exponential increase in reduced viscosity with
a reduction in polymer concentration. This is a typical behavior of polyelectrolytes [18] and has been shown to be present
in sulfonated electrolytes, including Nafion, by several authors [19,20]. The curves for fresh samples lie above those
for samples degraded by the solution method (96 h). There
are two possible explanations for this difference between fresh
and aged samples. One possibility is that the structure of the
repeating units making up the polymer chains has changed
and the other possibility is that the polymer molecular weight
has decreased, leading to changes in the size of aggregates
when in solution. It has already been shown by IEC results
that this structure, in so far as the ratio of side chain components to backbone, has not changed. On the other hand, a moderate decrease in average molecular weight has been shown to
shift the viscosity curves down by Cohen and coworkers [21].
A change in molecular weight is also consistent with the suggested mechanisms of Nafion degradation found in the literature [8]. Interestingly, the reduced viscosity curves of
samples degraded by the solution method and the exchange
method are very similar, indicating that their average molecular weight is also similar even though the weight loss of the
specific samples in question was different (Fig. 2).

219

0.05

0.1

0.15

0.2

0.25

Nafion Concentration (g/dl)


Fig. 4. Viscosity data for dissolved samples of fresh Nafion 112 and samples
degraded by the solution and exchange methods. Samples were dissolved in
50/50 ethanol/water mixtures at 250  C.

Despite the weight loss and IEC trends being similar


between the two degradation methods, scanning electron
microscopy (SEM) analysis of the Nafion samples showed
significant differences in the morphology of the samples. In
the case of samples degraded by the solution method, the
surface was roughened by the presence of defects. Upon
closer examination of the membrane the defects appear to
be bubbles formed within the membrane (Fig. 6a and c) as
well as tears and bumps on the membrane surface. The bubbles were in the order of hundreds of microns in diameter

220

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

Fig. 6. Comparison of surface morphologies between Nafion degraded by the solution method and the exchange method. (a) Solution method 100 magnification,
(b) exchange method 100 magnification, (c) solution method 500 magnification, and (d) exchange method 500 magnification.

while the bumps and tears were less than 10 mm. Overall, the
number of defects increased with the time of exposure to
degradation for both methods. The progression of degradation for the solution method is shown in Fig. 7c,e,g. The
membrane splits into two halves after 24 h of exposure and
after long time samples degraded by the solution method developed large pits within the two halves of the polymer
membrane (Fig. 8). With exception of the pit which occurs
after long exposure times, close examination of the
separate halves of the membrane shows that their morphology resembles that of the fresh membrane with no extra
distinguishing features. The samples themselves were translucent in appearance.
Samples degraded with the exchange method show fewer
large bubbles on the surface as compared to the solution
method even after 84 h of exposure. The surfaces have smaller
defects and fewer holes penetrating the membrane as shown in
Fig. 6b and d. Overall, membranes degraded by this method
were far smoother to the touch than the solution method samples due to smaller number of large bubbles and tears. Crosssections of samples degraded by the exchange method show
a markedly different morphology than those degraded by the
solution method. The progression of degradation of samples
degraded by the exchange method is shown in Fig. 7b,d,f.
The surface of the membrane has the appearance of foam
with the centre of the membrane essentially remaining as
one continuous phase. This foam appearance was not observed
with solution method samples even after long time. Control
samples showed no change in morphology with time compared with new samples. The appearance of the membranes
was also different from the solution method, the membrane

changed from being transparent to being opaque and white.


The difference in morphology is attributed to differences in
how reactive species diffuse into the membrane as a result
of the differences in the experimental solutions.
A reduction in thickness is a common observation in
Nafion samples degraded in fuel cells and is usually attributed to chemical degradation [11]. A dimensional analysis of
the membranes was performed to determine the effect of degradation by Fentons reagent on the samples shape and size.
Fresh samples measured with calipers and SEM had thicknesses between 0.04 mm and 0.06 mm. The SEM was used
to measure membrane thicknesses on aged samples since
caliper measurements were affected by the rough surface features. Thickness measurements of solution method samples
excluded the empty space created by bubbles and focused on
the combined thickness of the two halves of the membranes.
For the exchange method, the degraded membranes had well
defined edges for measurement encompassing both the highly
degraded foamed edges and the less degraded centre. Such
results showed that neither method of degradation produced
a significant change in membrane thickness.
Nafion samples preferentially swell in one planar direction versus the other. Here, the sample width will be the direction characterized by a high degree of swelling and the sample
length will be the direction characterized by less swelling.
These dimensions changed significantly with degradation.
With exposure using both methods the swollen width increased by 10e15% while at the same time the swollen length
decreased by 3e5% as shown in Fig. 9 after long exposure
times. Changes in swelling in aged samples were much
more significant than changes observed in control samples

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

221

Fig. 7. Comparison of SEM cross-sections of Nafion samples degraded by the solution method and by the exchange method. (a) Fresh Nafion samples, 2000
magnification. (b, d, f) Nafion degraded by the exchange method for 24 h, 60 h, and 84 h, respectively, 2000 magnification. (c, e, g) Nafion degraded by the
solution method for 24 h, 48 h, and 120 h, respectively, 500 magnification.

though Nafion exposed to hydrogen peroxide did have an


increase in swollen width of almost 5% as shown in Table 2.
Dry dimensions followed a similar trend. Additionally, aged
samples tended to swell more than fresh samples.
As the membrane degrades the total molecular weight of
the polymer chains decreased as shown by viscosity measurements. This facilitates the unrestricted movement of the
chains. The swelling behavior already shows that when given
greater mobility, through hydration, the entangled polymer

chains will preferentially relax in one direction. Thus, as the


chains degrade the membrane will tend to gain length in one
direction over another.
3.3. Effect of degradation on mechanical properties
and barrier properties
Degradation imparts many physical changes on membrane
morphology and dimensions, hence it is expected that

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

222

Table 2
Comparison of dimensional changes in control samples

Fig. 8. Nafion membrane split by a bubble after degradation by the solution


method for 108 h.

20
15
10
5
0
-5
-10
-15
-20
20
15
10
5
0
-5
-10
-15
-20

Control 1
Nafion in
water
(120 h)

Control 2
Nafion
30% HOOH
(96 h)

Control 3
Nafion
H2O Fe2
(120 h)

Control 4
Teflon
Fentons
(96 h)

% Change in width
% Change in length

2.78
1.11

4.60
0.32

2.61
2.10

2.50
0.96

this material [24] and is known as the a-transition. This transition is associated with increased mobility of the ionic domain, which tends to agglomerate into clusters within the
polymer [25]. The storage modulus (E0 ) of degraded samples
was lower than fresh samples (Fig. 11) followed similar trends
as the Youngs modulus. Again, this is expected of polymers
with decreased molecular weights.
Fig. 11 also shows that the loss tangent curves of degraded
samples are also lower than the fresh samples. Furthermore,
degraded samples showed an increase in a-transition temperature with samples degraded by the solution method showing
the largest increase (Fig. 12).
Within a fuel cell, Nafion electrolyte membranes are not
only responsible for transporting protons across the membrane
but are also responsible for acting as a gas barrier between the
two sides. Failure of the membrane as a gas barrier can result
in lower and unstable voltages or complete failure of the fuel
cell. The gas barrier properties of the membranes degraded in
this work were evaluated as a function of time and are shown
in Fig. 13. Fresh samples exhibited nitrogen flow less than
0.0002 mL s1. Samples degraded with the solution method
had nitrogen crossover rates as high as 0.27 mL s1. The samples degraded by the exchange method all had negligible
crossover rates. The reason for this extends from the morphology of the degraded samples, those degraded by the solution
method had many holes penetrating to the centre of the membrane facilitating crossover. In contrast, the membranes
degraded by the exchange method did not have these features
and therefore could continue to act as a gas barrier.

160

Degraded - Solution Method

150

Degraded - Exchange

140
Swollen Width
Swollen Length

Modulus (MPa)

Change

Change

mechanical and barrier properties will also be affected. The


mechanical properties were investigated by tensile stresse
strain curves and by temperature scanning in dynamic mechanical measurements. The barrier properties were studied
by measuring the crossover flow.
The Youngs modulus for both degradation methods was
observed to decrease by 10e20% in relation to fresh samples
(Fig. 10). This is consistent with a reduction in average molecular weight shown by the viscosity measurements. The shorter
chains resulting from OH radical degradation also result in less
chain entanglements. This increases the mobility of the chains
when stressed and therefore a lower Youngs modulus.
Transition temperatures were measured using the loss tangent (tan d) in the dynamic mechanical thermal analysis
(DMTA). Typical curves for fresh and degraded Nafion samples are shown in Fig. 11. A loss tangent peak occurred between 120  C and 130  C in all cases which is typical for

Sample

130
120
110
100
90
80

Swollen Width
Swollen Length
0

20

70
40

60

80

100

120

140

Time (h)
Fig. 9. Changes in length and width for samples degraded by the solution
method (a) and the exchange method (b).

60

20

40

60

80

100

120

140

Time (h)
Fig. 10. Youngs modulus results for samples degraded by the solution and
exchange methods.

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224


0.30

Fresh Nafion 112


Degraded - Exchange Method, 84h.
Degraded - Solution Method, 120h

100

0.6
0.5
0.4

60
0.3

Tan ( )

E' (MPa)

80

40
0.2
20

0.1

0
50

60

70

80

90

100

Temperature

110

120

130

140

0
150

0.25
0.20
0.15
0.10
0.05
0.00

20

(oC)

40

60

80

100

120

140

Time (h)

Fig. 11. Storage modulus (E0 ) and loss tangent (tan d) of fresh Nafion, sample degraded by the solution method for 120 h, and sample degraded by the
exchange method for 84 h.

4. Conclusions
This work has presented a comparison of Nafion membranes degraded by two ex situ accelerated degradation
methods promoting degradation by OH radical attack using
a Fentons type test. The two methods of Fentons testing
that were compared were a solution method which comprises
introducing Nafion membrane to a solution of peroxide and
iron and an exchange method where a sample of Nafion in
the Fe2 form was exposed to a peroxide solution. It was
found that the two degradation methods significantly degraded
the membrane over days of exposure time with samples loosing over 20% of their original weight over the testing time.
The weight loss was accompanied with fluoride release and
no change in the ratio of side chains to backbone. Viscosity
measurements of dissolved membranes suggest that the average molecular weight decreased with degradation. Furthermore, FTIR analysis also indicated that both methods of
degradation did not produce any new chemical groups on
the membranes.
128

-transition temperature (oC)

Nitrogen Crossover Flow (ml s-1)

0.7

120

223

127

Fig. 13. Crossover results for membranes degraded by the solution method.
Samples degraded by the exchange method showed negligible crossover.

Despite there being no significant chemical differences


between the two degradation methods, differences in morphology of the degraded samples were observed. Analysis
of the morphology showed that membranes degraded by the
solution method had many holes, tears, and large bubbles
on the surface and cross-sections revealed that the bubbles
originated at the centre of the membrane, splitting it into
two. Membranes degraded by the exchange method did not
split into two but instead areas close to the surface appeared
foamy. The thickness of the membranes did not change appreciably in both cases though with degradation, the swollen
width increased while the membrane shrank slightly in
length.
Finally, mechanical and gas barrier properties were examined for the degraded membranes. In both degradation
methods the modulus decreased with exposure time which
was attributed to the reduction in molecular weight of the
polymer chains. This was also observed in dynamic mechanical results of the storage modulus. The peak height of the loss
tangent curve in dynamic mechanical experiments was lower
with increased degradation. This may indicate that ionic clusters in the membrane, made up of sulfonic acid groups, shrank
as the polymer was degraded. The transition temperature however increased with exposure to the Fentons reagent which
may indicate that the smaller ionic clusters were also more
stable.

126

Acknowledgements

125

The authors would like to acknowledge Ion Power Inc. for


supporting this study and the Natural Sciences and Engineering Research Council (NSERC) of Canada for financial
support.

124
123
Solution Method
Exchange Method

122
121

20

40

60

80

100

120

140

References

Time (h)
Fig. 12. a-Transition temperatures for membranes degraded by the solution
and exchange methods.

[1] LaConti AB, Hamdan M, McDonald RC. In: Vielstich W, Gasteiger H,


Lamm A, editors. Handbook of fuel cells: fundamentals, technology
and applications, vol. 3. UK: Wiley-Interscience; 2003. p. 647e63.

224

S. Kundu et al. / Polymer Degradation and Stability 93 (2008) 214e224

[2] Paik CH, Skiba T, Mittal V, Motupally S, Jarvi TD. Membrane degradation studies under accelerated conditions in PEMFC. In: 207th Meeting
of the electrochemical society e meeting abstracts; 2005. p. 771.
[3] Mathias MF, Makharia R, Gasteiger HA, Conley JJ, Fuller TJ,
Gittleman CJ, et al. Electrochem Soc Interface 2005;14(3):24e36.
[4] Guo Q, Qi Z. J Power Sources 2006;160(2):1269e74.
[5] Endoh E, Terazono S, Widjaja H, Takimoto Y. Electrochem Solid State
Lett 2004;7(7):A209e11.
[6] St Pierre J, Wilkinson DP, Knights S, Bos ML. J New Mater Electrochem
Syst 2000;3(2):99e106.
[7] Kundu S, Fowler MW, Simon LC, Grot S. J Power Sources
2006;157(2):650e6.
[8] Curtin DE, Lousenberg RD, Henry TJ, Tangeman PC, Tisack ME. J
Power Sources 2004;131(1e2):41e8.
[9] Panchenko A, Dilger H, Kerres J, Hein M, Ullrich A, Kazc T, et al. Phys
Chem Chem Phys 2004;(11):2891e4.
[10] Inaba M. Degradation mechanism of polymer electrolyte membrane fuel
cells. In: Conference proceedings of the 14th international conference on
the properties of water and steam in Kyoto; 2005. p. 395.
[11] Healy J, Hayden C, Xie T, Olson K, Waldo R, Gasteiger H, et al. Fuel
Cells 2005;5(2):302e8.
[12] Mittal VO, Kunz HR, Fenton JM. Electrochem Solid State Lett
2006;9(6):299e302.

[13] Wayne RJ. Durability studies on polymer electrolyte membrane fuel


cells. MSc thesis. Case Western Reserve University, Cleveland, OH,
USA; 2004.
[14] Tang H, Peikang S, Jiang SP, Wang F, Pan M. J. Power Sources
2007;170(1):85e92.
[15] Kinumoto T, Inaba M, Nakayama Y, Ogata K, Umebayashi R, Tasaka A,
et al. J Power Sources 2006;158(2):1222e8.
[16] Qiao JL, Saito M, Hayamizu K, Okada T. J Electrochem Soc
2006;153(6):A967e74.
[17] Martin CR, Rhoades TA, Ferguson JA. Anal Chem 1982;54(9):1639e41.
[18] Bohdanecky M, Kovar J. Viscosity of polymer solutions. Amsterdam:
Elsevier Scientific Publishing Company; 1982. p. 108e15.
[19] Aldebert P, Gebel G, Loppinet B, Nakamura N. Polymer
1995;36(2):431e4.
[20] Moore RBI, Martin CR. Macromolecules 1988;21(5):1334e9.
[21] Cohen J, Priel Z, Rabin Y. J Chem Phys 1988;88(11):7111e6.
[22] Hung Y, El Khatib KM, Tawfik H. J Appl Electrochem 2005;35(5):445e7.
[23] Ludvigsson M, Lindgren J, Tegenfeldt J. Electrochim Acta
2000;45(14):2267e71.
[24] Kundu S, Simon LC, Fowler M, Grot S. Polymer 2005;46(25):
11707e15.
[25] Bauer F, Denneler S, Willert-Porada M. J Polym Sci, Part B Polym Phys
2005;43(7):786e95.

You might also like