EPR spectroscopic investigation of radical-induced degradation of partially fluorinated aromatic model compounds for fuel cell membranes

Article abstract of DOI:10.1039/b817070c

EPR spectroscopic investigations of reactions between monomeric model compounds representing typical structural moieties of poly(aryl) ionomers and photochemically generated hydroxyl radicals are reported. Deoxygenated solutions of the model compounds (in

Full text of DOI:10.1039/b817070c

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/pccp | Physical Chemistry Chemical Physics
EPR spectroscopic investigation of radical-induced degradation of
partially fluorinated aromatic model compounds for fuel cell membranes
a
a
b
b
¨
Frank Schonberger,w* Jochen Kerres, Herbert Dilger and Emil Roduner
Received 29th September 2008, Accepted 1st April 2009
First published as an Advance Article on the web 12th May 2009
DOI: 10.1039/b817070c
EPR spectroscopic investigations of reactions between monomeric model compounds representing
typical structural moieties of poly(aryl) ionomers and photochemically generated hydroxyl
radicals are reported. Deoxygenated solutions of the model compounds (in a water/methanol
mixture) containing hydrogen peroxide at defined pH values were exposed to UV light in the flow
cell within the cavity of an EPR spectrometer. Spectra were analyzed by computer simulation and
the formed radicals were assigned by comparing their g-factors and hyperfine coupling constants
(hfccs) with those from the literature and from density functional theory (DFT) calculations. The
relevance for polymer electrolyte membrane fuel cells (PEMFCs) and alkaline-anion exchange
membrane fuel cells (AAEMFCs) is discussed.
also be evolved at the anode side after oxygen has permeated
1. Introduction
through the membrane.12 Although a certain percentage of the
formed H2O2 might be washed out by the product water and
the flowing gas, another fraction might diffuse into the
membrane and decay into HOꢀ by catalysis of bivalent metal
ions at the elevated fuel cell operating temperature.6
Polymer electrolyte membrane fuel cells (PEMFCs) are—in
principle—considered as promising energy conversion devices
although many barriers are still in the way of their widespread
commercialization. The most serious obstacles are the high
system costs and the need for significant advances in hydrogen
production and storage, as well as the performance and the
durability of fuel cells under typical operating conditions.
These conditions include extreme or cyclic change in load as
well as the presence of impurities and inhomogeneities in the
fuel and oxidant supply, the catatytic reaction, the proton
conduction or the relative humidity.1 Especially, the degradation
of the membrane itself limits the lifetime of PEMFCs and
needs to be fully understood on a molecular scale in order to
minimize it by appropriate countermeasures. Besides physical
and mechanical degradations (such as thinning and pinhole
formation as a result of hydrolytic decomposition at elevated
temperature and induced stresses),2,3 membrane degradation
by hydroxyl (HOꢀ) and hydroperoxyl radicals (HOOꢀ) is
discussed in the literature.2,4–8 HOꢀ and HOOꢀ are regular
intermediates in the catalytic reduction of oxygen at the
cathode3,9 and might be responsible for membrane degradation
after desorption from the catalyst surface.8 Another often
discussed source for the formation of HOꢀ and HOOꢀ is based
on the evolution of small amounts of hydrogen peroxide as a
side product,6 especially at low coordinated platinum atoms
(e.g. edge and corner atoms).10,11 Hydrogen peroxide is mainly
formed at the cathode side of the MEA but in principle it can
For example, hydroxyl radicals are held responsible for
initiating depolymerization starting at the carboxylic acid end
groups of poly(perfluoroalkyl)sulfonic acids according to the
following reactions (also known as backbone unzipping).13
ꢀ
Rf–CF2COOH + HOꢀ - Rf–CF2 + CO2 + H2O (1)
ꢀ
ꢀ
Rf–CF2 + OH - Rf–CF2OH - Rf–COF + HF (2)
Rf–COF + H2O - Rf–COOH + HF (3)
Such carboxylic acid endgroups are formed in the presence of
water via perfluorocarbinol intermediates from sulfuric acid
ester end-groups which result from the polymerisation initiator
(K2S2O8). More recent studies on the chemical stability of
model compounds for perfluorinated sulfonic acid polymers
also suggest side-chain scission and HOꢀ radical attack of the
sulfonic acid groups as possible degradation modes.14,15
The investigation of HOꢀ and related radical induced
degradation reactions of membranes and membrane building
blocks by EPR spectroscopy has often been documented in the
literature and recently reviewed by Schlick and Roduner.16
Both in situ methods,5,6,17 where an MEA is placed into a
miniaturized fuel cell, and ex situ methods,17,18 where the
membrane is exposed to Fenton’s reagent, hydrogen peroxide
solution or vapor within the cavity of an EPR spectrometer,
was applied. In order to get a molecular understanding of
possible sites of attack for HOꢀ and oxygen radicals in
nonfluorinated poly(arylene ether) based ionomers a set of
hydrogen peroxide containing aqueous solutions of monomeric
model compounds was photolyzed within the cavity of the
EPR spectrometer.4,8,12,14,17 While these investigations were
focussed on nonfluorinated aromatic4,8,12,17 or fluorinated
a Institute of Chemical Process Engineering, University of Stuttgart,
Boblinger Strasse 72, 70199 Stuttgart, Germany.
¨
E-mail: frankschoenberger@gmx.net; Fax: +49 (0)711 685 85 242;
Tel: +49 (0)711 685 85 255
b Institute of Physical Chemistry, Pfaffenwaldring 55, 70569 Stuttgart,
Germany. E-mail: e.roduner@ipc.uni-stuttgart.de;
Fax: +49 (0)711 685 64 495; Tel: +49 (0)711 685 64 490
w Present address: Department of Chemistry and Biochemistry,
University of South Carolina, 631 Sumter Street, Columbia, SC
29208, USA
ꢁc
This journal is the Owner Societies 2009
5782 | Phys. Chem. Chem. Phys., 2009, 11, 5782–5795

View Article Online
initiates the photolytical cleavage of hydrogen peroxide into
HOꢀ radicals. Wavelengths below 210 nm and above 400 nm
are absorbed by an appropriate filter solution (1.14 M NiSO4;
0.21 M CoSO4; 0.01 M H2SO4).4 All measurements were
carried out at room temperature.
The solutions were prepared from doubly distilled water and
spectrophotometric grade methanol ( Z 99.9%, Aldrich). The
pH values of the solutions were adjusted with concentrated
sulfuric acid and caustic potash solution, respectively, by using
a digital WTW pH meter (model pH 330) calibrated with
commercial buffer solutions. All the solutions were thoroughly
deoxygenated by bubbling with nitrogen (for at least 15 min)
before injection into the flow system.
Hydrogen peroxide solution (30%) was purchased from
Merck and the model compounds A, C and D from Aldrich.
The model compound B was synthesized from A by electro-
philic aromatic substitution as follows: 40.00 g (118.97 mmol)
2,2-bis(4-hydroxyphenyl)hexafluoropropane (A) and 80 ml
concentrated sulfuric acid were stirred at 40 1C for 18 h. The
reaction mixture was then poured onto ice cubes (ca. 200 g).
After filtration, the product was precipitated by adding 300 g
sodium chloride. Finally, it was recrystallized several times
from methanol/water mixtures (9/1, v/v) and dried at 75 1C in
a vacuum oven.
Fig. 1 Overview of the investigated model compounds A–D.
non-aromatic14 model compounds, the present contribution
takes partially fluorinated poly(arylene ether) based ionomers
into account.19–32
1H NMR (250 MHz, D2O, d): 7.83 (s, 1H), 7.41
3
3
(d, JH–H = 8.2 Hz, 1H); 7.06 (d, JH–H = 8.8 Hz, 1H).
13C NMR (50 MHz, D2O, d): 156.87; 137.51; 132.09; 130.47;
1
The model compounds A and B have Csp3–F bonds, C
represents a poly(aryl) structure with Csp2–F bonds, and D is
a widely used building block with a C–S–C bridge (cf. Fig. 1).
Aromatic polymers with thioether instead of ether bridges in
their backbone have also been discussed as PEM materials33–35
which might be due to their oxidizabiliy to the sulfone group.
Such an oxidation process could be effective as an ‘internal
antioxidant’ and could prevent the ionomer from molecular
weight degradation caused by radical induced processes up to
a certain point.36,37
126.60 (q, JC–F = 287 Hz); 126.28; 120.10; 65.75 (septet,
2JC–F = 25.7 Hz). Elemental analysis for C15H6O8F6S2Na4
{experimental (theoretical) values in [%]}: C: 31.23 (30.83);
H: 2.04 (1.04); Cl: 0.85 (0.00).
The ratio between the concentrations of the model
compound and of hydrogen peroxide was first estimated from
their UV spectra in such a way that the main part of photons is
absorbed by hydrogen peroxide according to the following
inequation:
e(H2O2) ꢂ c(H2O2) c ei ꢂ ci
(4)
2. Experimental and theoretical methods
Herein, e(H2O2) and ei are the molar extinction coefficients of
hydrogen peroxide and of the model compound i, c(H2O2) and
ci describe their concentrations. However, the extinction
coefficients of the model compounds are relatively high
compared to that of hydrogen peroxide (cf. for example:
2.1 Sample preparation and EPR measurements
The model compounds A–D were exposed to HOꢀ radicals and
their reaction products with water and hydrogen peroxide,
respectively (for details: cf. section: Overview of the reaction
mixture) in a quartz flow cell (0.4 ꢂ 10 ꢂ 50 mm) within the
cavity of a Varian E-Line X-Band EPR spectrometer. For
this purpose deoxygenated methanolic–aqueous solutions or
aqueous solutions of the model compound and hydrogen
peroxide (with well-defined pH values) are circulated through
the cavity by a syringe pump (with a constant flow rate which
can be varied from 8 to 135 ml hꢃ1 and which corresponds
to residence times in the flow cell between 90 and 5 s,
respectively). Methanol was necessary as a cosolvent because
of the hydrophobic nature of the nonsulfonated model
compounds A, C and D. In order to exclude the existence of
any solvent-derived radicals, control experiments without the
model compound were carried out. A 500 W high-pressure
mercury arc-lamp (Oriel 66142) focussed on the flow cell
e(H2O2)
= ; eC = ;
98 Mꢃ1 cmꢃ1 3904 Mꢃ1 cmꢃ1
eA = 30 088 Mꢃ1 cmꢃ1 at 230 nm and pH = 12; determined
by UV spectroscopy) so that their concentrations must be
significantly smaller than that of hydrogen peroxide.
Therefore, the final composition of the sample solutions was
identified from preliminary EPR experiments in the presence
and absence of hydrogen peroxide. The ideal concentration
ratio was found when the EPR spectrum in the presence of
hydrogen peroxide exhibits an optimal signal-to-noise-ratio
while no or only minor EPR signals in the absence of hydrogen
peroxide are detectable. Thus it can be assumed that the
observed EPR signals in the presence of hydrogen peroxide
primarily result from the reaction of HOꢀ radicals with
the model compound and that they are not caused by any
solvent-derived radicals.
ꢁc
This journal is the Owner Societies 2009
Phys. Chem. Chem. Phys., 2009, 11, 5782–5795 | 5783

View Article Online
2.2 Analysis and interpretation of EPR spectra
Table 1 Rate constants for the reactions of HOꢀ with H2O2, H2O and
CH3OH and with its dissociation products. Values are taken from
ref. 54–56 if not otherwise stated
The EPR spectra were analyzed by Microcal Origins
Version 7.0. Hyperfine coupling constants (hfcc) were derived
from computer simulations using WINSIMs.38 EPR signals
were then assigned by comparing either with literature values39
(if available for the expected radicals) or with the results of
quantum mechanical calculations (using Gaussian 03).40
Density functional theory (DFT) based on the unrestricted
Kohn–Sham (UKS) methods41 has often been used for the
calculation of hyperfine coupling constants in the literature.42–45
Primarily, the spin population r(rX) at the nucleus X is
computed, which is proportional to the hyperfine coupling
constant aX (the Fermi contact interaction):46
Rate
constant/
pH
Reaction
109 Mꢃ1 sꢃ1
a
b
c
2–4
HOꢀ + H2O2 - H2O + HOOꢀ
0.04257
7.5
12
ꢃ
ꢀ
7.7–13
11.0
6–7
HOꢀ + HOOꢃ - H2O + O2
ꢃ
HOꢀ + HOꢃ - H2O + Oꢀ
d
HOꢀ + CH3OH - ꢀCH2OH + H2O
0.9758
HOꢀ+CH3OH - CH3Oꢀ+H2ꢃO
ꢃ
ꢃ
ꢀ
e
f
g
h
i
13
Oꢀ + HOOꢃ - HOꢃ + O2
0.4
0.75
5.5
9.4
r20
13.1–13.6 Oꢀ + CH3OH - H2O + ꢀCH2Oꢃ
7.0
HOꢀ ꢀOH - H2O2
+
2.74–6.75 HOꢀ + O2ꢃ - HOꢃ + O2
ꢃ
1
ꢀ
412
13–14
HOꢃꢀ + Oꢀ ꢃ - HOOꢃ
O2 + Oꢀ + H2O - 2 HOꢃ + O2 0.6
ꢀ
ꢀ
ꢁ
1
j
2m0 ge
aX
¼
gX bNrðrX Þ
ð5Þ
3
g
They lead to the formation of further radical species like
the oxygen radical anion (Oꢀꢃ), the hydroꢀperoxyl radical
(HOOꢀ), and the superoxide radical anion (O2 ꢃ). The course
of the reaction strongly depends on the pH value in the solution
and can be estimated qualitatively by means of the acid
constants of the corresponding acid–base pairs H2O2/HOOꢃꢃ
Herein, ge/g describes the ratio of the isotropic g-factors
of the radical and of the electron (approximated with 1 in
the DFT calculations), m0 = 4p ꢂ 10ꢃ7 T2 Jꢃ1 m3 is the
vacuum permeability, gX is the nucleus g-factor of X and
bN = 5.0508 ꢂ 10ꢃ7 J Tꢃ1 the nuclear magneton. The spin
population r(rX) is computed as the expectation value of the
ꢃ
(pKa = 11.7),8 HOꢀ/Oꢀ (pKa = 11.5),59 and HOOꢀ/O2
ꢀ
ˆ
(pKa = 4.7),60 as well as by means of the rate constants listed
in Table 1. In the presence of CH3OH that serves as a cosolvent
for the nonsulfonated model compounds A, C, and D, hydro-
xymethyl (ꢀCH2OH) or methoxy radicals (CH3Oꢀ) can be
formed through H abstraction by HOꢀ radicals. In none of
the cases reported here the hydroxymethyl radical, which is the
reaction product of HOꢀ radicals with methanol, could be
detected. If it was present, it could be distinguished easily from
the intermediates observed because of its larger hyperfine
coupling constant and lower g-values. In the absence or in case
of a deficiency of methanol, HOꢀ radicals preferably react with
spin operator Sz over the electronic wave function:
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
ꢂ
*
+
X
rðrX Þ ¼ Szꢃ1
c
dðr ꢃ r ÞS ðnÞ c
^
z
ꢂ
n
X
ꢂ
ð6Þ
n¼1
¼ raðrnÞ ꢃ rbðrX Þ
The index n runs over all electrons and Sz is the quantum number
which belongs to the projection of the total electron spin on the
z axis (1/2 for radicals). Any single point calculations in this work
were done with the hybrid functional B3LYP47–52 which turned
out to be the most appropriate for the calculation of hyperfine
coupling constants.53 EPR-III44,52 was used as a basis set if the
corresponding radical only contained elements of the first two
periods of the table of elements. For all the other radicals
(e.g. with sulfur atoms) the basis sets 6-31G(d) and TZVP were
applied.42 These basis sets were used in all cases for the optimiza-
tion of radical geometry before the single point calculations.
ꢃ
H2O2 to HOOꢀ and with HOO to O2 ꢃ. The latter reactions
proceed much faster in alkaline than in acid environment
(reactions a and b). HOꢀ radicals abstract hydrogen from
hydroxide ions in a diffusion-controlled reaction at pH 4 11
(reaction c) and form the oxygen radical anion (Oꢀꢃ). At very
ꢀ
ꢃ
high pH values (pH 4 13), the concentration of Oꢀ is mucꢃh
larger than that of its corresponding acid (HOꢀ) so that Oꢀ
might be the reactive species (reactions e and f). Radical
recombinations (such as reactions g to j) can also occur in the
examined reaction solutions.
3. Results and discussion
3.1 Overview of the reaction mixture
To summarize, it can be said that at pH o 10 the HOꢀ and
According to eqn (4) the absorption of photons by H2O2 must
be larger than that by the model compounds. Depending on
the quantum yield a certain amount of hydroxyl radicals is
formed by photolysis.
ꢃ
HOOꢀ radicals are dominant, while at pH 4 13 the Oꢀ and
O2 ꢃ species exist predominantly. In the pH range 10 r pH r 13
ꢀ
the reaction ꢃsolution consists of a complex mixture of HOꢀ,
Oꢀꢃ and O2 . The concentration of HOOꢀ can be neglected in
ꢀ
this range (pKa(HOOꢀ/O2 ꢃ) = 4.7). Beside possible reactions
ꢀ
hn
H2O2 ꢃ! 2HOꢀ
ð7Þ
of the aromatic ꢃmodel compounds with the four reactive
species (HOꢀ, Oꢀ , HOOꢀ, O2 ꢃ) it has to be considered that
ꢀ
However, these HOꢀ radicals are only detectable at very low
they can be involved directly in any photophysical processes or
photochemical reactions.61
temperatures or indirectꢃly by using spin traps (the same is true
ꢃ
ꢀ
for Oꢀ , HOOꢀ and O2 ). The primarily formed HOꢀ radicals
are highly reactive and can react with further hydrogen
peroxide or solvent molecules or their dissociation products
(HOꢃ, HOOꢃ), as well as with the aromatic modꢃel
compounds. First of all, reactions of HOꢀ with H2O2, HOO ,
HOꢃ and CH3OH will be considered (cf. Table 1, Reactions a–d).
3.2 2,2-Bis(4-hydroxyphenyl)hexafluoropropane (A)
Most of the EPR spectra recorded under various experimental
conditions (different pH values [pH 4 10.5, no signals were
detectable below pH = 10.5], flow rates and concentration
ꢁc
This journal is the Owner Societies 2009
5784 | Phys. Chem. Chem. Phys., 2009, 11, 5782–5795

View Article Online
ratios between A and H2O2) appear more or less chaotic,
which makes a clear interpretation impossible. Therefore
further examinations were carried out with the sulfonated
form of A (cf. section 3.3) which also represents a typical
structural moiety in partially fluorinated poly(arylene ether)
based ionomers.19,20,25
Table 2 EPR parameter of the radical B-1 formed during the reaction
of B with H2O2 under UV irradiation at pH = 9.45 and at a flow rate
of 35 ml hꢃ1. Hyperfine coupling constants (hfcc) are taken from
computer simulation using WINSIM and compared with values from
DFT calculation results
hfcc [G] from
Simulation
DFT
g-Factor
Proposed structurec
3.3 Tetranatrium-3,30(1,1,1,3,3,3-hexafluoropropane-2,2-
diyl)bis(6-oxidobenzenesulfonate) (B)
1 H (1):
1.49a (1.52)b
1 H (2):
1 H (1): 1.43
1 H (2): 3.74
2.0048
The model compound B forms radicals under UV irradiation
both in the absence and in the presence of H2O2 at pH = 9.45
(Fig. 2a and b). The singlet with g = 2.0033 observed under
2.37a (2.39)b
1 H (3):
1 H (3): 0.58
6 F (4): 0.85
1.32a (1.30)b
ꢃ
ꢀ
6 F (4):
1.22a (1.20)b
H2O2-free conditions could be assigned to the SO3 radical in
good accordance with literature.62 This radical is generated by
direct photolysis of the C–S bond. The lack of the simultaneously
formed phenyl radical indicates an extremely short lifetime of this
species so that its stationary concentration might be below the
detection limit of about 10ꢃ9 M.63 In the presence of H2O2 the
a
b
B3LYP/TZVP//B3LYP/TZVP. B3LYP/6-31G(d)//B3LYP/6-31G(d).
R = C6H3(m-SO3ꢃ)(p-Oꢃ).
c
ꢃ
singlet caused by the ꢀSO3 radical disappears. Instead, another
results of density functional theoretical calculations. Especially
the radical geometry, electron correlation and the used basis
set influence the quality of the theoretical prediction of EPR
parameters in comparison to the experimental ones. In fact, the
calculation of isotropic hyperfine coupling constants is one of the
most demanding issues in theoretical chemistry.42 The hyperfine
coupling constant is proportional to the spin population (eqn (5))
which denotes the difference between the contributions due to
electrons having spin a and b (eqn (6)). Small calculated hyperfine
coupling constants always imply large relative errors as the
calculated value is a small difference of two very high numbers.
Here, the experimentally found values for the hyperfine
coupling constants often lie in the range around 0 which
explains the observed differences between the experimental
and theoretical values. Furthermore, the size of the radical
which consists of 37 atoms and the presence of sulfur complicate
the absolute calculation of hyperfine coupling constants
further.44
signal can be detected which is most probably caused by the
phenoxyl radical B-1 (cf. Fig. 2c.) and Table 2). Unfortunately,
there are no g-factors and hyperfine coupling constants available
in the literature for this radical. The structure assignment is thus
based on a comparison with similar radicals39 and with the
The formation of the phenoxyl radical B-1 could be
explained by reaction of the corresponding bisphenolate anion
with HOꢀ radicals either by a one-electron mechanism or by
an addition/elimination mechanism via a hydroxylcyclo-
hexadienyl radical.64,65 The lack of the latter (intermediate)
radicals in the EPR spectra does not necessarily mean that
they are not formed, but that their concentration might be
below the detection limit. Therefore, no definite information
on the formation mechanism of B-1 can be given on the basis
of these data.
Fig. 3 compares the EPR spectra of the model compound B
at various pH values in the presence of H2O2 under UV
irradiation. While no signals are detectable at very low pH
values (pH = ꢃ0.80) under the chosen reaction conditions,
there are additional small signals in the range between pH = 3.10
and 5.30 besides the singlet caused by the ꢀSO3ꢃ. Although no
reliable simulation of these signals was possible due to the
poor signal-to-noise ratio, it can be supposed that they are
caused by traces of B-1. In the pH range between pH = 7.3
and pH = 11.4, B-1 seems to be the only species. The loss of
hyperfine structure with increasing pH value is attributed to
the presence of triplet oxygen (3O2) which causes spin
Fig. 2 (a) Experimental EPR spectrum of B in the presence of 85 mM
H2O2 at pH = 9.45 in aqueous solution under UV irradiation
(12.8 mM B, pH = 9.45, flow rate 35 ml hꢃ1). (b) Control experiment
under the same conditions, but in the absence of H2O2. (c) Simulation
of the phenoxyl radical B-1.
ꢁc
This journal is the Owner Societies 2009
Phys. Chem. Chem. Phys., 2009, 11, 5782–5795 | 5785

View Article Online
3.4 2,20,3,30,5,50,6,60-Octafluoro-4,40-biphenol (C)
EPR signals (Fig. 4) could be detected under the chosen
conditions both in an acidic (0 r pH r 3.5) and a basic
medium (11 r pH r 14). While the signals detected at low pH
values (pH r 3.5, depicted for pH = 0 in Fig. 4 as an
example) are most probably generated by a photophysical or
photochemical path, the signals at higher pH values might
result from the reaction of HOꢀ and related radicals with the
model compound C. But first, the spectra in the absence of
hydrogen peroxide will be discussed. A singlet with a very
low g-factor (g = 2.0007) and a relatively low intensity is
observed both at pH = 0 and at pH = 12. Definite structure
assignments of the species responsible for this singlet cannot
be given here. The energy of a photon emitted by the radiation
source (in its maximum at 230 nm) would be sufficient
for the homolytic cleavage of bonds with a maximum
dissociation energy of ca. 520 kJ molꢃ1. Thus a cleavage of
any C–F bonds (ED(C–F) = 477–644 kJ molꢃ1 70
bonds (ED(Car–Car) E 518 kJ molꢃ1 71
could be possible
)
or C–C
)
under these conditions. However, fluorine atoms formed by
photolysis of any C–F bonds might react further with water
(eqn (8)).72
Fꢀ + H2O - HF + HOꢀ
(8)
Although the low g-factor of 2.0007 argues for a s radical, it is
rather unlikely that the signal is caused by a perfluorinated
phenyl radical (according to reaction (1) or (2) in Fig. 5), since
one should expect further hyperfine splittings due to the
remaining fluorine atoms. However, it cannot be excluded
that such radicals might be present as very short-lived
intermediates since their spin concentration could be below
the detection limit of the EPR spectrometer. Eventually, the
singlet at g = 2.0007 could be explained by a photoisomerization
of one of the perfluorinated benzene rings to the corresponding
Dewar benzene ring (reactions (3) and (4) in Fig. 5) which
passes through an EPR active intermediate.73–75
Fig. 3 EPR signals of the model compound B in the presence of H2O2
(85 mM) under UV irradition and in dependence of the pH value
(conditions: 12.8 mM B; flow rate = 35 ml hꢃ1). Plain water was used
as a solvent in all experiments.
exchange in the presence of any free radicals, leading to a
dramatic line broadening.
Since all solutions were thoroughly deoxygenated before
injecting into the EPR spectrometer triplet oxygen can only be
formed in situ. This is also in accord with the observation of
gas evolution in the sample cell especially at higher pH values.
Oxygen bubbles out of the solution only when saturation is
reached at a concentration of ca. 0.3 mM, which is higher than
the concentration of any EPR active species in the solution. It
is known from the literature that oxygen generated photo-
chemically from hydrogen peroxide exists in a singlet state.66,67
The lifetime of the so-formed singlet oxygen in water is
reported to be 2 ms.68 However, the residence time of the
injected solution in the flow cell of the EPR spectrometer
amounts to 20 s at a flow rate of 35 ml hꢃ1. A part of the
evolved singlet oxygen might be embedded in the bubbles,
where its lifetime is considerably higher than in water
(7 to 12 s).69 However, another part of the singlet oxygen
might diffuse into the solution, transform into the triplet state,
and cause the observed line broadening.
Fig. 4 EPR spectra of the model compound C (4.4 mM C, 85 mM
H2O2 in acidic (flow rate: 10 ml hꢃ1) and alkaline (flow rate: 55 ml hꢃ1
)
methanolic–aqueous solutions (4.4 M CH3OH) (recorded with the
same modulation amplitude of 0.4 G).
ꢁc
This journal is the Owner Societies 2009
5786 | Phys. Chem. Chem. Phys., 2009, 11, 5782–5795

View Article Online
Fig. 5 Possible photolytic cleavage of C (reactions (1) and (2) and
photoisomerization to the corresponding dewar benzene derivative via
excited (possibly EPR active) intermediates (reactions (3) and (4)).
The photolytic cleavage-off of Fꢀ and the subsequent
formation of HOꢀ according to eqn (8) could also be the
reason for the similar EPR signals of C at pH = 0 both in the
absence and in the presence of hydrogen peroxide (cf. C
(pH = 0) and C + H2O2 (pH = 0) in Fig. 4). At this pH value,
the addition of HOꢀ radicals to C under the formation of
hydroxycyclohexadienyl radicals with subsequent HF elimination
should be expected. Unfortunately, the EPR signals of C
(pH = 0) and C + H2O2 (pH = 0) could not be resolved
further and thus no details on possible hyperfine coupling
constants could be revealed.
When the pH value is increased above pH = 3.5, the signals
disappear. Only above pH = 11 are other signals detectable
in the presence of hydrogen peroxide under UV irradiation
(C + H2O2 (pH = 12) in Fig. 4). In the absence of hydrogen
peroxide, however, only the narrow singlet at g = 2.0007
can be observed (C (pH = 12) in Fig. 4). As depicted in
Fig. 6 and summarized in Table 3 the EPR spectrum in the
presence of hydrogen peroxide could be interpreted as a
superposition of several fluorinated radicals of the phenoxyl
and benzosemiquinone type.
Fig. 6 (a) Experimental EPR spectrum of 2 mM C in the presence of
85 mM H2O2 at pH = 12 and at a flow rate of 55 ml hꢃ1 (C + H2O2
(pH = 12) in Fig. 4. (b) Superposition of the radicals C-1 to C-5.
Simulation of (c) C-1 (18%); (d) C-2 (26%); (e) C-3 (20%); (f) C-4
(13%); (g) C-5 (11%); (h) Superposition of the radicals C-1 to C-7.
Simulation of (i) C-6 (6%); (j) C-7 (6%).
The experimental EPR spectrum of C is shown again in
Fig. 6a and compared with the simulated superposition of the
radicals C-1 to C-5 (cf. Fig. 6b–g and Table 3) which may be
considered as the key contribution. However, the experimental
spectrum in Fig. 6a clearly contains further species to a minor
extent. When the experimental (Fig. 6a) and the simulated
(Fig. 6b) spectra are compared, it is obvious that the outer part
of the simulated spectrum does not match very well the
experimental one. Although a definite conclusion cannot be
drawn at this point because of the low intensity of any
minor radical species, the EPR pattern is compatible with the
existence of the two additional species C-6 and C-7
(cf. Fig. 6h–j). The formation of C-6 and C-7 would further
be understandable from a chemical point of view. Their
formation and also that of C-1 to C-5 cannot be explained
by the addition of superoxide radical anions according to
Fig. 7 or of singlet oxygen according to Fig. 8 as reported in
ꢁc
This journal is the Owner Societies 2009
Phys. Chem. Chem. Phys., 2009, 11, 5782–5795 | 5787

View Article Online
Table 3 Experimental and calculated (B3LYP/6-31G(d)//B3LYP/
EPR-III) EPR parameter of the radicals formed in the reaction of C
with H2O2 under UV irradiation; R = C6F4Oꢃ
the literature for nonfluorinated radicals. The latter two
reaction mechanisms pass through keto-enol tautomerisms
which require the presence of Car–H bonds (with Car
=
aromatic carbon atom). However, the perfluorinated model
compound C has only Car–F bonds. As the oxygen radical
anion (Oꢀꢃ) is significantly less electrophilic than the hydroxyl
radical (HOꢀ), but is in excess at pH 4 pKa(HOꢀ/Oꢀꢃ) = 11.5,
an alternative mechanism according to Fig. 9 is conceivable.
In the first reaction step, the phenoxyl radical C-1 is formed
by one-electron oxidation of the corresponding bisphenolate
hfcc [G] from
Simulation
DFT
g-factor
Proposed structure
4 F (3,30,5,50):
4.13
4 F (2,20,6,60):
1.14
4 F (3,30,5,50):
5.57
4 F (2,20,6,60):
2.00
2.0061
ꢃ
by Oꢀ and HOꢀ radicals, respectively. The remaining
fluorine atoms of C-1 can then be substituted by HOꢃ in
highly alkaline solution. Such SNAr reactions, for example,
serve for the synthesis of the model compound C from
decafluorobiphenyl76 and are known from degradation reactions
of perchlorinated aromatics.77
1 F (3): 3.10
1 F (6): 1.90
1 F (3): 2.84
1 F (6): 1.92
2.0066
2.0062
2.0062
3.5 4,40-Thiodiphenol (D)
4,40-Thiodiphenol (D) causes EPR signals both in the presence
and in the absence of hydrogen peroxide at high pH values
(pH = 12) and low flow rates (8 ml hꢃ1); the spectra are shown
in Fig. 10. In the absence of hydrogen peroxide (under UV
irradiation) a triplet (g = 2.0051 G, a = 0.90 G) can be
observed which also contributes to the superposition of the
simulated radicals in the presence of hydrogen peroxide
(D-4; Fig. 11 and Table 5). 4-Oxidophenyl- and (4-oxidophenyl)-
sulfanyl (Table 4) radicals could be formed by photolytic
cleavage of the weakest (C–S) bond in the molecule. Their
EPR spectra could eventually suggest a triplet if the hyperfine
interaction to the protons H-3 and H-5 was sufficiently small
not be resolved.
1 F (5): 0.88
1 F (5): 0.39
1 F (6): 3.12
1 F (6): 5.80
As the values in Table 4 prove, this photochemical reaction
can be excluded. However, the observed triplet could be
explained by the radical D-4 (cf. Table 5). The formation of
similar oxidosubstituted benzosemiquinone radicals has been
observed in previous investigations,4,54,62 but the proposed
reactionꢃmechanism can only occur in the presence of photons
1
ꢀ
2.0068
and O2 radicals or O2 (cf. Fig. 7 and 8). In principle, the
observed radicals D-2 to D-6 could be formed by such
a mechanism; however, the origin of radical D-4 under
H2O2-free conditions cannot be traced back on this reaction
pathway.
A possible explanation is given in Fig. 12. Photoexcited
4,40-thiodiphenol molecules can ionize into the corresponding
phenoxyl radicals and hydrated electrons.79–87 The lifetime
of these hydrated electrons is mainly limited by reaction m
(Table 6) which might explain the lack of EPR signals of the
free electron.88 Table 6 further lists the reaction rate of phenol
with hydrated electrons as the value for 4,40-thiodiphenol
could not be found in the literature, but the latter is expected
to be in the same order of magnitude. The possible reaction
of hydrated electrons with the cosolvent methanol can be
excluded as well, as the rate constant is lower by several orders
of magnitude. The primarily formed phenoxyl radicals (from
reaction step (2) in Fig. 12) could not be detected by EPR
spectroscopy (a = 2.13 G (4 H)),89 which might be traced back
to a low concentration and/or short lifetime. The reason
for this could be the univalent dehydrogenation of water by
2 F (2,6): 6.73
2 F (2,6): 4.78
2.0065
1 F (6): 12.14
1 F (2): 1.94
1 F (5): 1.80
1 F (6): 6.74
1 F (2): 2.97
1 F (5): 0.62
2.0070
ꢁc
This journal is the Owner Societies 2009
5788 | Phys. Chem. Chem. Phys., 2009, 11, 5782–5795

View Article Online
Fig. 7 Proposed reaction mechanism for the formation of oxidosubstituted benzosemiquinone radicals in the presence of superoxide radical
anions under UV irradiation.54
Fig. 8 Proposed reaction mechanism for the formation of oxidosubstituted benzosemiquinone radicals in the presence of singlet oxygen under
UV irradiation.54
ꢃ
the formed phenoxyl radicals under the reformation of
4,40-thiodiphenolate and the release of HOꢀ radicals. Because
of the released protons in the acid–base reactions (4) and (5),
the initially adjusted pH value could slightly drop. These HOꢀ
and Oꢀ radicals might be responsible for the formation of
D-4, even under hydrogen peroxide free conditions.
Radical D-1 has been detected in a previous study as
degradation product of 2,2-bis(4-hydroxyphenyl)sulfone. Its
ꢁc
This journal is the Owner Societies 2009
Phys. Chem. Chem. Phys., 2009, 11, 5782–5795 | 5789

View Article Online
Fig. 9 Possible reaction mechanism for the formation of the radicals
C-1 to C-7 in alkaline medium.
Fig. 11 (a) Experimental EPR spectrum of D in the presence of
hydrogen peroxide at pH = 12 and with a flow rate of 8 ml hꢃ1
(measured in 1.3 M CH3OH in H2O). (b) Superposition of simulations
of the radicals D-1 to D-6. Simulation of (c) D-1 (34%); (d) D-2 (8%);
(e) D-3 (9%); (f) D-4 (38%); D-5 (8%) and (h) D-6 (3%).
Fig. 10 EPR spectra of 4,40-thiodiphenol (D) in the presence and
absence of hydrogen peroxide (conditions: 6.4 mM D; pH = 12; flow
rate 8 ml hꢃ1) in aqueous methanol solution (1.3 M CH3OH).
D-6 only contributes by 3% at low flow rate (8 ml hꢃ1) while
it contributes by 9% to the simulation at high flow rate
(150 ml hꢃ1). One should expect at first look that D-6 would
be a secondary radical formed from D-2, D-3 and D-4,
respectively. It should thus be observed in higher amounts
at low flow rates, which is obviously not the case. Providing
the radical structure assignments are correct, this would
mean that D-6 is not derived from D-2, D-3 and D-4,
respectively. This could further suggest that the formation of
D-2 to D-5 on the one hand and the formation of D-6 on the
other hand proceed via competing reactions. However, no
mechanistic details can be given here. In the last part of
this section, the radical composition will be examined with
dependence on the pH value. While no EPR signals could be
detected below pH = 10.2 (Fig. 14), there is a more or less
pronounced pH dependence of the intensity ratio in the range
10.5 o pH o 13.2 (Fig. 15). Especially at very high pH values
formation could be initiated by the addition of a HOꢀ radical
to the aromatic carbon next to the sulfur atom (reaction step
(1) in Fig. 13). The subsequent deprotonation (2) under the
basic conditions favors the elimination of 4-sulfidophenolate
(3) by the enhanced +M effect of the oxido substituent.
According to Table 5, an increase in the flow rate at a constant
pH value of 12 particularly influences the ratio of the radicals
D-1 and D-6. The intensity of the signal caused by D-1
decreases, while that caused by D-6 increases. An explicit
explanation for it cannot be given here, but it might be
possible that the addition of HOꢀ radicals at the carbon next
to the sulfur atom is kinetically more hindered than the
formation of phenoxyl radicals which serve for the reaction
cascade in Fig. 7. The intensities of the other radicals D-2 to
D-5 do not seem to be influenced significantly by the flow rate.
ꢁc
This journal is the Owner Societies 2009
5790 | Phys. Chem. Chem. Phys., 2009, 11, 5782–5795

View Article Online
Table 4 EPR parameters of 4-oxidophenyl- and (4-oxidophenyl)-
sulfanyl radicals (calculated by Gaussian 03 and taken from the
literature if available)
Table 5 Experimental and calculated EPR parameter of the radicals
formed in the reaction of D with H2O2 under UV irradiation
(pH = 12, flow rate 8 ml hꢃ1); R = C6H3(m-SO3ꢃ)(p-Oꢃ)
hfcc [G]
g-Factor
Radical structure
hfcc [G] from
Simulation
2 H (3,5): 5.96 Ga (4.90 G)78
2.0022678
(literature)a
DFTb
g-Factor Proposed structure
4 H (2,3,5,6):
2.34 (2.35)
4 H (2,3,5,6): 2.32 2.0049
calculated with
b3lyp/6-31G(d)//
b3lyp/epr-iii
2 H (2,6): 0.60 Gb (0.40 G)c
2 H (3,5): 2.35 Gb (3.14 G)c
1 H (3): 1.14
1 H (5): 1.94
1 H (6): 2.98
1 H (3): 1.04
1 H (5): 2.56
1 H (6): 3.18
2.0049
B3LYP/6-31G(d)//B3LYP/EPR-III. b B3LYP/TZVP//B3LYP/TZVP.
B3LYP/6-31G(d)//B3LYP/6-31G(d).
a
c
(pH = 13.2) the percentage of the p-benzosemiquinone radical
D-1 and of the lower oxidosubstituted radicals D-2 and D-3
decreases for the benefit of the higher oxidosubstituted ones
D-4 to D-6.
1 H (3): 1.17
1 H (6): 0.37
1 H (3): 1.06
1 H (6): 0.38
2.0056
At pH 4 13 the radicals Oꢀꢃ and O2 ꢃ are prevalent while a
ꢀ
ꢃ
ꢃ
ꢀ
complex mixture of HOꢀ, Oꢀ and O2 exists in the rꢀange
10 r pH r 13. Caused by the higher concentration of O2 ꢃ at
pH = 13.2, the probability of a bimolecular collision with an
already hydroxylated species (e.g. reaction steps (1) and (4) in
Fig. 7) is higher. The composition at pH = 12 is essentially
determined by the radicals D-1 and D-4. The percentage of the
higher hydroxylated radicals D-5 and D-6 in the superposition
of the various radicals at pH = 10.5 is relatively high.
However, the molecular process responsible for this could
not be clarified on the basis of the data presented.
2 H (3,5): 0.86 2 H (3,5): 1.26
2.0051
3.6 Comparison between the various model compounds
The model compounds were chosen in such a way that they
represent different C–F bond types: Csp3–F in the model
compounds A and B, Csp2–F in the model compound C. All
model compounds yield phenoxyl and benzosemiquinone
radicals at slightly to strongly alkaline pH values under the
chosen reaction conditions. As the conditions for obtaining
analyzable EPR signals had to be adapted individually, a
quantitative comparison cannot be made here. It can be stated
that phenoxyl and benzosemiquinone radicals at elevated pH
values are formed independently of the type and number
of C–F bonds. In case of Csp2–F bonds (as in the model
compound C) the fluorine atoms might be substituteꢃd by
oxido substituents. Because of the lower tendency of Oꢀ for
an electrophilic attack to arenes in comparison to that of HOꢀ,
the following mechanism seems likely. After one-electron
oxidation by Oꢀꢃ, a perfluorinated phenoxyl radical is formed
whose fluorine atoms might subsequently be exchanged by
HOꢃ in the alkaline medium. However, if Csp3–F bonds are
present in the model compound (like in B) they do not seem to
be attacked. It might also be interesting for any further
polymer and ionomer development to note that the sulfonated
model compound B yields a phenoxyl radical already at
pH = 7.3 (and most probably already at pH = 5.3). On the
1 H (5): 2.17
1 H (5): 1.52
2.0054
1 H (6): 0.65
1 H (6): 0.32
2.0050
a
b
If available. Calculated with B3LYP/TZVP//B3LYP/TZVP
(Gaussian 03) if not otherwise stated.
ꢁc
This journal is the Owner Societies 2009
Phys. Chem. Chem. Phys., 2009, 11, 5782–5795 | 5791

View Article Online
Fig. 12 Formation of hydroxyl radicals and hydrated electrons by photoionisation of 4,40-thiodiphenol.
pH values. Although differences in the concentration might be
jointly responsible for the observed pH dependency, it is likely
that a six-membered transition state between the hydroxyl and
sulfonic acid group could facilitate the formation of the
corresponding phenoxyl radicals in the case of the sulfonated
model compound B.
Table 6 Overview of possible reactions and their reaction rates of
hydrated electrons55
Rate constant/
109 Mꢃ1 sꢃ1
pH
11
Reaction
k
l
m
eꢃ + CH3OH - Hꢀ + CH3Oꢃ
eꢃaq + C6H5Oꢃ - C6H5O2ꢃ
o0.00001
0.004
5.0
11–13 eꢃaq + eꢃ + 2 H2O - H2 + 2 HOꢃ
In the case of the aqueous-methanolic test solutions
(A, C and D), no evidence for the formation of methylol
radicals (according to reactions d and f in Table 1) could be
found. Considering the difference in concentrations for the
various monomers and for methanol (Table 7), one can
aq
aq
other hand, the nonsulfonated model compounds A, C and D
form phenoxyl and benzosemiquinone radicals only at higher
Fig. 13 Possible reaction mechanism for the formation of the p-benzosemiquinone radicals D-1.
ꢁc
This journal is the Owner Societies 2009
5792 | Phys. Chem. Chem. Phys., 2009, 11, 5782–5795

View Article Online
aqueous(-methanolic) solution under UV irradation have been
examined within the cavity of an EPR spectrometer.
As the reaction conditions in the test solution are not
comparable in every respect to those in a working fuel cell, a
critical analysis is necessary. Due to the chosen experimental
setup for the generation of HOꢀ radicals, it cannot be excluded
that the high-energy UV radiation causes any photophysical
processes or photochemical reactions of the tested monomer
directly. Beside the excitation into a higher electronic state from
which any reaction with other species could be facilitated,61
a
direct photochemical dissociation (such as the cleavage-off of
ꢃ
the ꢀSO3 radical in case of the model compound B) or a
photoionization (as in the case of D) are worth mentioning.
Such photoinduced processes—which play an important role in
the ageing and degradation of polymers in general90—are not
relevant for membrane degradation within a fuel cell stack.
However, the sulfonic acid group of the aromatic model
compound does not seem to react with HOꢀ radicals under
the chosen conditions.
Fig. 14 Dependence of the EPR signals from the pH value (conditions:
6.4 mM D; 42.5 mM H2O2; flow rate: 8 ml hꢃ1; 1.3 M CH3OH). As no
signals could be detected at pH values below 10.2, only the EPR
spectrum at pH = 9.3 is depicted in this diagram to represent the lower
pH range.
A further difference between the test solution and the fuel
cell membrane lies in their chemical composition (monomer in
the first case, polymer in the second case) and in their physical
state (solution or multiphase system, respectively). This means
that every model compound possesses two functional groups
(e.g. hydroxyl groups) while there are only two functional
end-groups per poly(aryl) ionomer. However, the importance
of end-groups for the HOꢀ radical induced degradation should
not be underestimated as was shown earlier for the example of
poly(perfluorosulfonic aicd)s.13,91 Furthermore, HOꢀ radical4
or acid2 induced ether cleavages can enhance the number of
potential end groups in the case of sulfonated poly(arylene
ether)s.
Another aspect of this discussion concerns the pH value of
the test solutions in comparison to that in a real polymer
electrolyte membrane. Most of the identified radicals could be
observed only at elevated pH values. This does not necessarily
mean that no radicals are formed at lower pH values, but their
concentration could be below the detection limit (especially in
case of short-lived radicals).63 Indications for any kind of
reaction (such as colour change, gas evolution) could be
observed for all the investigated model compounds at
various pH values so that some reaction (not necessarily a
radical-involved one) is likely.
Fig. 15 Radical composition of the solution in dependence of the pH
value.
Table 7 Overview of the investigated model compounds A to D and
of the pH range in which radicals of the phenoxyl and benzosemi-
quinone type could be observed
Model Concentration monomer/ Flow rate/
compound CH3OH (H2O2)/mM
ml hꢃ1
pH range
The pH value in a swollen proton-exchange membrane is in
the range of pH E 0. During fuel cell operation local pH
inhomogeneities could arise close to the catalyst particles
especially under high loading.18 These inhomogeneities could
then induce the same or similar reactions as those presented in
this paper. In addition, the presented model compounds might
also be relevant as building blocks for alkaline anion-exchange
membrane fuel cells (AAEMFCs)92,93 which naturally work at
pH E 14. This type of fuel cell could be of interest for future
research because the electrode kinetics are faster in an alkaline
medium94 which allows the use of platinum-free catalysts.95
The reaction conditions applied for the investigation of
these model compounds are therefore not completely comparable
to those in a working fuel cell. However, the following
conclusions for further polymer and ionomer development
may be drawn:
Aa
B
C
D
3.2/1300 (170)
12.8/0 (85)
2.0/4400 (85)
6.4/1300 (42.5)
8
35
35
8–150
E12
(5.3) 7.3–11.4
11.2–14.0
10.5–13.2
a
Complex reaction mixture, oscillating radical concentration.
conclude a significantly preferred reactivity of HOꢀ and
related radicals with the aromatic model compounds than
with the competing methanol solvent.
Conclusions
Reactions between monomeric model compounds (representing
typical structure moieties of poly(aryl) ionomers) and H2O2 in
ꢁc
This journal is the Owner Societies 2009
Phys. Chem. Chem. Phys., 2009, 11, 5782–5795 | 5793

View Article Online
ꢀ Alkaline conditions facilitate the formation of phenoxyl
and benzosemiquinone radicals, i.e. local inhomogeneities in
pH should be avoided when such building blocks are
present. The application of such aromatic building blocks in
alkaline–anion exchange membrane fuel cells holds the risk of
an enhanced susceptibility to HOꢀ and related radicals,
especially when hydroxyl end-groups are present.
20 J. A. Kerres, D. Xing and F. Schonberger, J. Polym. Sci., Part B:
¨
Polym. Phys., 2006, 44(16), 2311.
21 C. Hamiciuc, M. Bruma and M. Kapper, J. Macromol. Sci. A,
2001, 38(7), 659.
22 P. Xing, G. P. Robertson, M. D. Guiver, S. D. Mikhailenko and
S. Kaliaguine, Macromolecules, 2004, 37, 7960.
23 B. Liu, G. P. Robertson, M. D. Guiver, Y.-M. Sun, Y.-L. Liu,
J.-Y. Lai, S. Mikhailenko and S. Kaliaguine, J. Polym. Sci.,
Part B: Polym. Phys., 2006, 44, 2299.
ꢀ The number of potentially reactive end-groups (especially
of hydroxyl groups and fluorine atoms) should be minimized
(e.g. by endcapping of the polymers with a monofunctional
aromatic monomer).
24 W. L. Harrison, F. Wang, J. B. Mecham, V. A. Bhanu, M. Hill,
Y. S. Kim and J. E. McGrath, J. Polym. Sci., Part A: Polym.
Chem., 2003, 41, 2264.
25 F. Schonberger, M. Hein and J. Kerres, Solid State Ionics, 2007,
¨
178, 547.
26 N. Y. Arnett, W. L. Harrison, A. S. Badami, A. Roy, O. Lane,
F. Cromer, L. Dong and J. E. McGrath, J. Power Sources, 2007,
172, 20.
27 K. B. Wiles, C. M. de Diego, J. de Abajo and J. E. McGrath,
J. Membr. Sci., 2007, 294, 22.
28 Z. Bai, J. A. Shumaker, T. D. Dang, M. Yoonessi and
M. F. Durstock, Polym. Preprints, 2006, 47(1), 140.
29 Y. S. Kim, M. J. Sumner, W. L. Harrison, J. S. Riffle,
J. E. McGrath and B. S. Pivovar, J. Electrochem. Soc., 2004,
151(12), A2150.
ꢀ The ether bridge of poly(arylene ether)s has to be
stabilized against HOꢀ radical, acid or base induced cleavage
by appropriate electron-withdrawing substituent in ortho- or
para-position in order to reduce the probability of chain
cleavage during fuel cell operation.32 Any cleaved ether
bridges might activate a cascade of (further) radical induced
reactions (especially in a locally alkaline medium).
30 M. Sankir, Y. S. Kim, B. S. Pivovar and J. E. McGrath, J. Membr.
Sci., 2007, 299, 8.
31 T. L. Norsten, M. D. Guiver, J. Murphy, T. Astill, T. Navessin,
S. Holdcroft, B. L. Frankamp, V. M. Rotello and J. Ding, Adv.
Funct. Mater., 2006, 16, 1814.
Acknowledgements
The authors thank Dr Svetlin Mitov and Dr Stefan Jagiella for
their help and useful discussions.
¨
32 F. Schonberger, A. Chromik and J. Kerres, Polymer, 2009, 50,
2010.
33 Z. Bai, M. F. Durstock and T. D. Dang, J. Membr. Sci., 2006, 281,
508.
References
34 Z. Bai and T. D. Dang, Macromol. Rapid Commun., 2006, 27, 1271.
35 J. K. Lee and J. Kerres, J. Membr. Sci., 2006, 294, 75.
36 G. Xiao, G. Sun and D. Yan, Polym. Preprints, 2003, 44(1), 1235.
37 M. Schuster, K.-D. Kreuer, H. T. Anderson and J. Maier, Macro-
molecules, 2007, 40, 598.
1 L. Zhang and S. Mukerjee, J. Electrochem. Soc., 2000, 153(6),
A1062.
2 J. Kerres and F. Schonberger, Fundamental Aspects of Membrane
Degradation, International Workshop on Degradation Issues in
Fuel Cells, Hersonissos, Crete, (Greece), September 19th–21st,
2007.
3 N. Ramaswamy, N. Hakim and S. Mukerjee, Electrochim. Acta,
2008, 53, 3279.
¨
38 WINSIMs, National Institute of Environmental Health Science,
http://www.niehs.nih.gov.
39 Landolt-Bornstein, Group II: Molecules and Radicals, Springer-
¨
Verlag, 2006.
40 http://www.gaussian.com.
41 H. Haken and H. C. Wolf, Molekulphysik und Quantenchemie, 5.
¨
4 G. Hubner and E. Roduner, J. Mater. Chem., 1999, 9, 409.
¨
5 A. Panchenko, H. Dilger, E. Moller, T. Sixt and E. Roduner,
¨
Auflage, Springer Verlag, 2006.
J. Power Sources, 2004, 127, 325.
42 L. Hermosilla, P. Calle, J. M. Garcı
J. Phys. Chem. A, 2005, 109, 1114.
´
a de la Vega and C. Sieiro,
6 A. Panchenko, H. Dilger, J. Kerres, M. Hein, A. Ullrich, T. Kaz
and E. Roduner, Phys. Chem. Chem. Phys., 2004, 6, 2891.
7 A. Panchenko, J. Membr. Sci., 2006, 278(1–2), 269.
8 S. Mitov, O. Delmer, J. Kerres and E. Roduner, Helv. Chim. Acta,
2006, 89, 2354.
43 B. T. Nguyen, S. Creve, L. A. Eriksson and
L. A. Vanquickenborne, Mol. Phys., 1997, 91(3), 537.
44 T. H. Dunning, Jr, J. Chem. Phys., 1989, 90(2), 1007.
45 N. Rega, M. Cossi and V. Barone, J. Chem. Phys., 1996, 105(24),
11060.
9 A. B. Anderson and T. V. Albu, J. Electrochem. Soc., 2000, 147(1),
4229.
46 C. Adamo, V. Barone and A. Fortunelli, J. Phys. Chem., 1994, 98,
8648.
10 U. A. Paulus, T. J. Schmidt, H. A. Gasteiger and R. J. Behm,
J. Electroanal. Chem., 2001, 492, 134.
47 P. A. M. Dirac, Proc. R. Soc. London, Ser. A, 1929, 123, 714.
48 J. C. Slater, Phys. Rev. A, 1951, 81, 385.
49 S. H. Vosko, L. Wilk and M. Nusair, Can. J. Phys., 1980, 58, 1200.
50 A. D. Becke, Phys. Rev. A, 1988, 38, 3098.
51 C. Lee, W. Yang and R. G. Parr, Phys. Rev. B, 1988, 37, 785.
52 A. D. Becke, J. Chem. Phys., 1993, 98, 5648.
53 A. R. Rakitin, D. Yff and C. Trapp, J. Phys. Chem. A, 2003, 107,
6281.
54 S. Mitov, PhD Thesis, University of Stuttgart, 2007.
55 G. V. Buxton, C. L. Greenstock, W. P. Helman and A. B. Boss,
J. Phys. Chem. Ref. Data, 1988, 17(2), 513.
11 V. O. Mittal, H. R. Kunz and J. M. Fenton, Electrochem. Solid
States, 2006, 9(6), A299.
12 S. Mitov, B. Vogel, E. Roduner, H. Zhang, X. Zhu, V. Gogel,
L. Jorissen, M. Hein, D. Xing, F. Schonberger and J. Kerres, Fuel
¨
Cells, 2006, 6, 413.
¨
13 D. E. Curtin, R. D. Lousenberg, T. J. Henry, P. C. Tangeman and
M. E. Tisack, J. Power Sources, 2004, 131, 41.
14 M. Danilczuk, F. D. Coms and S. Schlick, Fuel Cells, 2008, 8(6),
436.
15 C. Zhou, M. A. Guerra, Z.-M. Qiu, T. A. Zawodzinski and
D. A. Schiraldi, Macromolecules, 2007, 40(24), 8695.
16 E. Roduner and S. Schlick, ESR Methods for Assessing the
Stabiltity of Polymer Membranes Used in Fuel Cells, in Advanced
ESR Methods in Polymer Research, ed. S. Schlick, John Wiley &
Sons, 2006, ch. 8.
17 B. Vogel, E. Aleksandrova, S. Mito, M. Krafft, A. Dreizler,
J. Kerres, M. Hein and E. Roduner, J. Electrochem. Soc., 2008,
155(6), B570.
56 B. H. J. Bielski, D. E. Cabelli, R. L. Arudi and A. B Boss, J. Phys.
Chem. Ref. Data, 1985, 14(4), 1985, 1041.
57 X.-Y. Yu and J. R. Barker, J. Phys. Chem. A, 2003, 107, 1313.
¨
58 K.-D. Asmus, H. Mockel and A. Heinglein, J. Phys. Chem., 1973,
77, 1218.
59 G. A. Poskrebyshev, P. Neta and R. E. Huie, J. Phys. Chem. A,
2002, 106(47), 11488.
60 A. D. N. J. De Grey, DNA Cell Biol., 2002, 21(4), 251.
61 G. Wypych, Handbook of Material Weathering, ChemTec Publishing,
3rd edn, 2003.
18 A. Bosnajakovic and S. Schlick, J. Phys. Chem. B, 2004, 108, 4332.
19 H. C. Lee, H. S. Hong, Y.-M. Kim, S. H. Choi, M. Z. Hong,
H. S. Lee and K. Kim, Electrochim. Acta, 2004, 49, 2315.
ꢁc
This journal is the Owner Societies 2009
5794 | Phys. Chem. Chem. Phys., 2009, 11, 5782–5795

View Article Online
62 G. Hubner, PhD Thesis, University of Stuttgart, 1999.
¨
63 J. A. Weil, J. R. Bolton and J. E. Wertz, Electron Paramagnetic
Resonance, John Wiley & Sons, 1994.
64 J. L. Ferry and M. A. Fox, J. Phys. Chem. A, 1998, 102, 3705.
65 S. Steenken, J. Chem. Soc., Faraday Trans. 1, 1987, 83, 113.
66 A. E. da Hora Machado, R. Ruggiero and M. G. Neumann,
J. Photochem. Photobiol., A, 1994, 81, 107.
79 M. Cocivera, M. Tomkiewicz and A. Groen, J. Am. Chem. Soc.,
1972, 20, 6598.
80 L. I. Grossweiner, G. W. Swenson and E. F. Zwicker, Science,
1963, 141(3583), 805.
81 J. Feitelson and E. Hayon, J. Phys. Chem., 1973, 77(1), 10.
82 J. Feitelson, E. Hayon and A. Treinin, J. Am. Chem. Soc., 1972,
95(4), 1025.
67 E. A. Almeida, S. Miyamoto, G. R. Martinez, M. H. G. Medeiros
and P. D. Mascio, Anal. Chim. Acta, 2003, 482, 99.
68 T. Kajiwara and D. R. Kearns, J. Am. Chem. Soc., 1973,
95, 5886.
83 G. Kohler, G. Kittel and N. Getoff, J. Photochem., 1982, 18, 19.
84 A. S. Jeevarajan and R. W. Fessenden, J. Phys. Chem., 1992, 96,
1520.
85 Y. Kajii and R. W. Fessenden, Res. Chem. Intermed., 1999, 25(6),
567.
86 A. Bussandri and H. v. Willigen, J. Phys. Chem. A, 2002, 106, 1524.
87 A. Bussandri and H. v. Willigen, J. Phys. Chem. A, 2001, 105, 4669.
88 G. Brunton, B. C. Gilbert and R. J. Mawby, J. Chem. Soc., Perkin
Trans. 2, 1976, 11, 1297.
¨
69 J. Feitelson and E. Hayon, J. Phys. Chem., 1973, 77(1), 10.
70 J. Pola, M. Urabnova
´
V. Vorlıcek, J. Mater. Chem., 1998, 8(1), 187.
´
´
, Z. Bastl, Z. Plzak, J. Subrt, I. Gregora and
71 P. W. Atkins, Physikalische Chemie, 2. Auflage, Wiley-VCH, 1996.
72 V. A. Benderskii, A. U. Goldschleger, A. V. Akimov,
E. Y. Misochko and C. A. Wight, Mendeleev Commun., 1996, 6,
245.
89 G. Brunton, B. C. Gilbert and R. J. Mawby, J. Chem. Soc., Perkin
Trans. 2, 1976, 11, 1297.
73 J. Pola, M. Urbanova
´
, Z. Bastl, Z. Plza
´
k, J. Subrt, I. Gregora and
90 D. M. Wiles, Polym. Eng. Sci., 1973, 13(1), 74.
91 T. Kallio, K. Jokela, H. Ericson, R. Serimaa, G. Sundholm,
P. Jacobsson and F. Sundholm, J. Appl. Electrochem., 2003, 33,
505.
92 J. R. Varcoe, R. C. T. Slade and E. L. H. Yee, Chem. Commun.,
2006, 1428.
93 J. R. Varcoe and R. C. T. Slade, Fuel Cells, 2005, 5(2), 187.
94 G. F. McLean, T. Niet, S. Prince-Richard and N. Djilali, Int. J.
Hydrogen Energ., 2002, 27, 507.
V. Vorlıcek, J. Mater. Chem., 1998, 8(1), 187.
´
74 I. Haller, J. Am. Chem. Soc., 1966, 88, 2070.
75 G. Camaggi, F. Gozzo and G. Cevidalli, Chem. Commun., 1996,
313.
76 D. J. Byron, A. S. Matharu, R. C. Wilson and J. W. Brown, Mol.
Cryst. Liq. Crys., A, 1995, 258, 95.
77 D. H. Sarr, C. Kazunga, M. J. Charles, J. G. Pavlovich and
M. D. Aitken, Environ. Sci. Technol., 1995, 29, 2735.
78 H. Zemel and R. W. Fessenden, J. Phys. Chem., 1975, 79, 1419.
95 H. Meng and P. K. Shen, Electrochem. Commun., 2006, 8(4), 588.
ꢁc
This journal is the Owner Societies 2009
Phys. Chem. Chem. Phys., 2009, 11, 5782–5795 | 5795