Oxidative and photochemical stability of ionomers for fuel-cell membranes

Article abstract of DOI:10.1002/hlca.200690219

To predict hydroxyl-radical-initiated degradation of new proton-conducting polymer membranes based on sulfonated polyetherketones (PEK) and polysulfones (PSU), three nonfluorinated aromatics are chosen as model compounds for EPR experiments, aiming at the identification of products of HO^.-radical reactions with these monomers. Photolysis of H₂O₂ was chosen as the source of HO^. radicals. To distinguish HO ^.-radical attack from direct photolysis of the monomers, experiments were carried out in the presence and absence of H₂O₂. A detailed investigation of the pH dependence was performed for 4,4′-sulfonylbis[phenol] (SBP), bisphenol A (-4,4′- isopropylidenebis[phenol]; BPA), and [1,1′-biphenyl]-4,4′-diol (BPD). At pH > pK_A of HO^. and H₂O ₂, reactions between the model compounds and O₂ ^. or ¹O₂ are the most probable ways to the phenoxy and 'semiquinone' radicals observed in this pH range in our EPR spectra. A large number of new radicals give evidence of multiple hydroxylation of the aromatic rings. Investigations at low pH are particularly relevant for understanding degradation in polymer-electrolyte fuel cells (PEFCs). However, the chemistry depends strongly on pH, a fact that is highly significant in view of possible pH inhomogeneities in fuel cells at high currents. It is shown that also direct photolysis of the monomers leads to 'semiquinone'-type radicals. For SBP and BPA, this involves cleavage of a C-C bond.

Full text of DOI:10.1002/hlca.200690219

2354
Helvetica Chimica Acta – Vol. 89 (2006)
Oxidative and Photochemical Stability of Ionomers for Fuel-Cell Membranes
by Svetlin Mitova), Olga Delmer (geb. Reifschneider)a)1), Jochen Kerresb), and Emil Roduner*a)
a) Institute of Physical Chemistry, University of Stuttgart, Pfaffenwaldring 55, D-70569 Stuttgart
(phone: +49-711-6856-4490; fax: +49-711-6856-4495; e-mail: e.roduner@ipc.uni-stuttgart.de)
b) Institute for Chemical Engineering, University of Stuttgart, Bçblingerstr. 72, D-70199 Stuttgart
In memoriam Professor Hanns Fischer
To predict hydroxyl-radical-initiated degradation of new proton-conducting polymer membranes
based on sulfonated polyetherketones (PEK) and polysulfones (PSU), three nonfluorinated aromatics
are chosen as model compounds for EPR experiments, aiming at the identification of products of
HOC-radical reactions with these monomers. Photolysis of H2O2 was chosen as the source of HOC radicals.
To distinguish HOC-radical attack from direct photolysis of the monomers, experiments were carried out
in the presence and absence of H2O2. A detailed investigation of the pH dependence was performed for
4,4’-sulfonylbis[phenol] (SBP), bisphenol
A (=4,4’-isopropylidenebis[phenol]; BPA), and [1,1’-
biphenyl]-4,4’-diol (BPD). At pHꢀpKA of HOC and H2O2, reactions between the model compounds
and O2Cꢁ or 1O2 are the most probable ways to the phenoxy and ꢀsemiquinoneꢁ radicals observed in
this pH range in our EPR spectra. A large number of new radicals give evidence of multiple hydroxyla-
tion of the aromatic rings. Investigations at low pH are particularly relevant for understanding degrada-
tion in polymer-electrolyte fuel cells (PEFCs). However, the chemistry depends strongly on pH, a fact
that is highly significant in view of possible pH inhomogeneities in fuel cells at high currents. It is
shown that also direct photolysis of the monomers leads to ꢀsemiquinoneꢁ-type radicals. For SBP and
BPA, this involves cleavage of a CꢁC bond.
Introduction. – In the past years, there has been a tremendous acceleration in
research devoted to nonfluorinated polymer membranes, both as competitive alterna-
tives to commercial perfluorosulfonic acid membranes operating in the same temper-
ature range and with the objective of extending the range of operation of polymer
fuel cells toward those more generally occupied by phosphoric acid fuel cells [1]. Con-
sequently, much effort has focused on the development of alternative proton-exchange
membranes for PEFCs (polymer-electrolyte fuel cells) and DMFCs (direct methanol
fuel cells), in particular with the aim of increasing their temperature of operation.
Important advantages in terms of water management, increased proton conductivity,
and CO tolerance in fuel cells can be gained by operating at higher temperature, typ-
ically at ca. 1208. This operating temperature imposes additional stringent requirements
in terms of membrane stability in a highly oxidative environment. There are few non-
fluorinated membrane materials appropriate for fuel cell applications at temperatures
above 808, and they are made up of polyaromatic or polyheterocyclic repeat units.
1
)
Present address: Max Planck Institute for Solid State Research, Heisenbergstr. 1, D-70569 Stuttgart.
ꢂ 2006 Verlag Helvetica Chimica Acta AG, Zürich

Helvetica Chimica Acta – Vol. 89 (2006)
2355
Examples include polysulfones (PSU), polyetherketone (PEK), polybenzimidazole
(PBI), polyethersulfone, polyphenylquinoxaline, polyetherimides, etc. Furthermore,
polymer blending is a potentially versatile way of tuning the properties to those desir-
able for fuel-cell application. For instance, sulfonated PEK and PSU have been used in
blends with PBI, as well as with more weakly basic components such as polyetherimine,
polyethylenimine and poly(4-vinylpyridine) [1][2]. A general problem of most
approaches for development of novel, nonfluorinated fuel-cell membranes is that
these polymers and membranes are developed via the trial-and-error principle. There-
fore, there is a strong need for the preparation of novel fuel-cell membranes following a
more fundamental approach, including subsequent investigation of the chemical and
thermal stability of the monomers, polymers, membranes, and membrane–electrode
assemblies (MEAs).
To ensure long-term durability of the new materials, investigations of possible deg-
radation mechanisms are necessary. HOC and HOOC radicals originating from oxygen
diffusion through the membrane and incomplete reduction at the fuel-cell anode are
held responsible for these degradation reactions [3][4]. Hydroxyl and hydroperoxyl
radicals are regular intermediates of the cathode reaction. This is unproblematic as
long as they remain attached to the catalyst surface. The intermediate of oxygen reduc-
tion, hydrogen peroxide, can be formed in sufficient amounts on the catalyst surface at
the cathode side of the cell [5]. In the presence of noble metals and at elevated temper-
atures, H2O2 decays into hydroxyl radicals HOC which attack the membrane [6].
Because of partial oxygen crossover through the membrane, degradation can take
place also at the anode.
On an experimental scale, the generation of HOC and HOOC radicals can be ach-
ieved by a variety of different methods, for instance by direct photochemical [7][8]
or photocatalytic degradation of H2O2 [9], also by eꢁ irradiation [10][11] or sonolysis
[12] of aqueous solutions. In the present work, we chose photolysis of an aqueous
H2O2 solution as a source of HOC radicals [13] (Eqn. 1) which react further via Eqns.
2 and 3 and other reactions as detailed in Table 1. Since aromatic monomers absorb
UV radiation as well, care has to be taken to distinguish HOC radical reactions from
direct photochemical cleavage.
H2O2 ! 2 HOC
(1)
(2)
(3)
(4)
HOC+H2O2 ! HOOC+H2O
HOOC+H2O ! O2Cꢁ +H3
A
HOC+RH ! H2O+RC
The formation of radical species from the attacked substrate, for instance RH, can
occur by H-abstraction, and the polymer-derived RC radicals can initiate a degradation
cascade (Eqn. 4). The focus of the present work is the chemical degradation which is
generally assumed to be based on radical oxidative processes. Electron paramagnetic
resonance (EPR) methods are particularly sensitive and suitable for their investigation.
In the center of interest are the reaction mechanisms which are accessible via the struc-

2356
Helvetica Chimica Acta – Vol. 89 (2006)
Table 1. Rate Constants for Reactions of HOC Radicals [22]
ꢁ1
Reaction
Rate constant k/109
M
sꢁ1
ꢁ
a: pH>11.7: HOC+HO2 ! H2O+O2Cꢁ
b: pH<11.7: HOC+H2O2 ! H2O+HO2C
c: HOC+O2Cꢁ ! HOꢁ +1O2
7.5
0.027
8
d: pH=11.0: HOC+OHꢁ ! OCꢁ +H2
A
12
e: HOC+HOC ! H
2O2
5.5
0.4
0.6
ꢁ
f: OCꢁ +HO2 ! HOꢁ +O2Cꢁ
g: O2Cꢁ +OCꢁ +H2O ! 2 HOꢁ +1O2
tures of radical intermediates, and the identification of fuel-cell operating conditions
which may lead to enhanced radical production. In the present work, these points
are addressed by exposing membrane building blocks in solutions to the above inten-
tionally produced oxidative radicals. We investigate the oxidative degradation of 4,4’-
sulfonylbis[phenol] (SBP), bisphenol
A
(=4,4’-(1-methylethylidene)bis[phe-
nol]=4,4’-isopropylidenebis[phenol]; BPA), and [1,1’-biphenyl]-4,4’-diol (BPD) in
the pH range 0.4–13.9.
The radical intermediates of the HOC-initiated degradation of aromatic-group-con-
taining membranes are of the cyclohexadienyl, benzyl, phenoxy, or ꢀortho-semiqui-
noneꢁ type. Typical rate constants for reactions with model compounds of relevance
in the present context are collated in Table 2. They show that HOC addition occurs
with rate constants close to the diffusion-controlled limit, H-abstractions are slower
by less than an order of magnitude, while OCꢁ can abstract H with a similar efficiency
but does not undergo addition. The radicals are observable by EPR and distinguishable
by means of their hyperfine splittings and g values [6]. Based on the pKA values of H2O2
N
(11.7), HOC (11.9), and HOOC (4.8), the main reactive species below pH 10 is HOC;
above pH 13, it is OCꢁ and O2Cꢁ, while at 10ꢂpHꢂ13, we have a strongly pH-dependent
mixture [6]. Relatively little quantitative information is available for HOOC, but it is
known to be significantly less reactive than HOC which reacts by H-abstraction or by
addition to unsaturated bonds. The anionic species are less efficient in abstraction reac-
tions but add readily to double bonds or aromatic rings. Under alkaline conditions, rad-
icals are formed by abstraction of HC from the side chains by OCꢁ [6]. Investigations at
low pH are particularly relevant for understanding degradation in PEFCs, but one
should keep in mind that the chemistry depends strongly on pH, a fact that is highly
significant in view of possible pH inhomogeneities in fuel cells under conditions of
fuel starvation at high currents [14] or when parts of the anode are flooded by water.

Helvetica Chimica Acta – Vol. 89 (2006)
2357
Table 2. Rate Constants for Reactions of HOC andO Cꢁ Radicals with Different Substrates [22]
Substrate
Reaction of HOC
Reaction of OCꢁ
Product
k/108
Oꢁ
ꢁ1
ꢁ1
Product
k/108
M
sꢁ1
M
sꢁ1
CH3
C6H5OH/C6
6H5CH3
A
CCH2
A
9.7
66
30
CCH2
R
7.5
6.5
N
Oꢁ
C6H5(OH)2
C6H5(OH)CH3/C6H5CCH2
adduct
C6H5OC
C
N
N
C6H5CCH2
21
Benzophenone
88
Benzenesulfonic acid
Benzenesulfonate ion
adduct
adduct
16
30
2. Experimental. – Electron Paramagnetic Resonance. EPR Spectra: Varian E-Line-X-band spec-
trometer at 2.5 mW microwave power; at r.t. g Values were calculated relative to the Brukerꢃ weak
pitch sample (g=2.0028). They are accurate to ꢃ0.0002 units. Second-order effects are within this
range of errors for the comparatively small coupling constants observed in our systems, and were
neglected. The obtained digital spectra were analyzed by using Microcal Originꢃ, Version 5.0. Hyperfine
splittings were derived from computer simulations with the WINEPR Simphonia program and assigned
by comparison with literature values of similar radical systems [15].
Photolysis. At r.t., 4 mM solns. of the monomers (unless otherwise stated) in H2O/MeOH 1:1 (v/v)
(MeOH was added for solubility reasons) were photolyzed within the microwave resonator by a 500-
W Hg (Xe) high-pressure mercury-arc lamp (Oriel 66142) in a flat quartz flow cell (0.4 mm”10
mm”50 mm). Experiments were also carried out in the presence of 40 mM H2O2. Wavelengths below
210 and above 400 nm were eliminated with a filter solution (1.14M NiSO4/0.21M CoSO4/0.01M H2SO4).
A constant flow (8–135 ml hꢁ1) of the solns. through the cell was maintained by a syringe pump. By
changing the flow, the residence time within the cavity is varied between 5 and 90 s, so that successive
reaction products could separately be focused on.
Doubly distilled H2O was used, and the pH was adjusted with H2SO4 or KOH, by using a digital
WTW pH meter (model pH 330) calibrated with commercial buffer solns. Before irradiation, solns.
were deoxygenated by bubbling with N2 for 15 min. H2O2 (30% soln.) was obtained from Merck, the
BPA, SBP, and BPD monomers from Aldrich.
3. EPR Results and Related Discussion. – 3.1. 4,4’-Sulfonylbis[phenol] (SBP). SBP
is a monomeric building block of the sulfonated PSU. The only species that was
detected both in the presence and the absence of H2O2 is the multiplet of radical 1
with five equidistant EPR lines with relative intensities corresponding to four equiva-
lent protons. Based on its g value (2.0049) and its proton hyperfine coupling constant,
there is little doubt that it corresponds to the ꢀp-semiquinoneꢁ radical anion
(=cyclohexa-2,5-diene-1,4-dione radical ion (1ꢁ)) 1 (Fig. 1, Table 3). SBP has a high
molar extinction coefficient (e296 =20600 l molꢁ1 cmꢁ1 at pH 12.2 and e260 =17600 l
molꢁ1 cmꢁ1 at pH 7.2) so that also in the presence of H2O2, the dominant fraction of
the light is absorbed directly by the monomer and leads to its cleavage. The signal
amplitude increases by a factor of five between the flow rates 8 and 135 ml hꢁ1. At
lower flow rates, no signals were detected, indicating that the radical is short lived or
the monomer gets depleted.
In the presence of H2O2, EPR signals were observed in the range between pH 7.6
and 13.9 (Fig. 2), with a maximum at pH 12.2, near the pKA of the HOC radicals
(11.9) and the pKA of H2O2 (11.7). No signals were observed under acidic or neutral
conditions.

2358
Helvetica Chimica Acta – Vol. 89 (2006)
Fig. 1. Experimental EPR spectrum of the ꢀp-benzosemiquinoneꢁ radical anion 1 obtainedwith SBP at
ꢁ1
pH 12.2 and135 ml h
flow rate
Fig. 2. Relative intensities of the EPR signals of radical 1 observedwith SBP at 135 ml hꢁ1 flow rate in
dependence of the pH value

Helvetica Chimica Acta – Vol. 89 (2006)
2359
Table 3. EPR Parameters of Intermediates in the Reaction of HOC Radicals with Sulfonated Aromatics.
Hyperfine splittings were derived from computer simulations with the WINEPR Simphonia program and
assigned by comparison with literature values of similar radical systems [15]. Due to the similarity of all
parameters, the structural assignments are somewhat tentative.
Substrate and conditions
Intermediates
Nr. structure
1
Hyperfine splittings [G]
g
SBP, pH 12.2, 135 ml hꢁ1
4 H (2,3,5,6): 2.35
2.0049
BPA, pH 12.0, 135 ml hꢁ1
1
2
see above
4 H (2,3,5,6): 2.35
2.0049
2.0048
1 H (5): 3.85
1 H (6): 2.12
3
4
5
1 H (5): 1.52
1 H (6): 3.83
2.0048
2.0048
2.0047
1 H (2): 0.64
1 H (5): 0.31
1 H (6): 3.85
BPD, pH 12.2 (10.3), 135 ml hꢁ1, no H2
A
4 H (2,6,2’,6’): 1.84
4 H (3,5,3’,5’): 0.92
BPD, pH 10.3, 135 ml hꢁ1, with H2
A
6
1 H (5): 1.22
1 H (6): 3.24
2.0044
7
8
1 H (5): 3.25
1 H (6): 1.72
2.0044
2.0044
1 H (2): 0.50
1 H (5): 0.24
1 H (6): 3.23

2360
Helvetica Chimica Acta – Vol. 89 (2006)
Table 3 (cont.)
Substrate and conditions
Intermediates
Nr. structure
9
Hyperfine splittings [G]
g
1 H (2,6): 3.13
1 H (3,5): 0.26
2.0044
10
1 H (3): 0.26
1 H (5): 0.74
2.0044
Spin trapping is considered to be a powerful tool to visualize short-lived radicals
indirectly [16]. In this technique, a diamagnetic compound, called a spin trap, reacts
with a primary radical to form a more stable radical that can accumulate to a high
enough concentration for an EPR study. In our experiments, 3,5-dibromo-4-nitrosoben-
zenesulfonic acid (DBNBS) was used as a spin trap, and the reaction that takes place is
shown in Scheme 1. The advantage of the DBNBS–spin adduct is that it can yield con-
siderable structural information since the primary radical adds directly to the N-atom
and remains, therefore, close to the unpaired electron. The applied spin-trap concentra-
tion was fixed at 10 mM, all other experimental conditions were identical with the pre-
vious measurements. As seen from Fig. 3, the observed signals at pH 0.4 consist of a
1:1:1 triplet (due to aN =13.54 G) of a 1:2 :1 triplet (due to 2 equivalent protons
with aH =8.95 G). This spectrum clearly arises from the CCH2OH-radical adduct to
DBNBS and not from SBP. The hydroxyl radicals react first with MeOH, which is to
be expected as the initial MeOH concentration (12M) is a factor of 3000 higher than
that of SBP (4 mM). In view of the high reaction rate of HOC radicals with MeOH
(see Table 2), it is clear that in the presence of solvents other than H2O one should
not expect to see products of HOC reactions with a substrate that is present only in mil-
limolar concentrations. Water-soluble monomers are, therefore, more suitable for
experiments involving hydroxyl radicals under acidic conditions. Furthermore, at the
higher flow rate (Fig. 3,b) a small triplet splitting is observed that is due to the Hmeta
of the spin trap molecule DBNBS (aH =1.02 G).
Scheme 1
3.2. Bisphenol A (BPA). BPA is a monomeric building block of the sulfonated PSU
which, together with PEK, is an alternative for the replacement of the most widely used
but expensive perfluorinated Nafionꢃ. PSU is usually sulfonated by electrophilic sub-

Helvetica Chimica Acta – Vol. 89 (2006)
2361
Fig. 3. EPR spectrum of an SBP/DBNBS/H2O/MeOH/H2O2 mixture at pH 0.4 andat a) a flow rate of
8 ml hꢁ1 and b) a flow rate of 90 ml hꢁ1
stitution at the position ortho to the ether bridge of the BPA fragment, because this
part of the molecule has a high electron density (+M effect from the ether bridge,
+I effect from the linking isopropylidene group), in contrast to the diarylsulfone por-
tion of the monomer repeating unit which has a low electron density due to the elec-
tron-withdrawing SO2 group (ꢁI and ꢁM effect).
In the absence of H2O2, the EPR spectrum shows the same multiplet as that found
with SBP, which is assigned to the ꢀp-semiquinoneꢁ radical anion 1. This multiplet is
superimposed by further lines at lower intensity which appear more strongly in the
presence of H2O2. BPA has a lower absorption coefficient at lmax
(e293 =4000 l molꢁ1
G
cmꢁ1 at pH 12.0) than SBP but absorbs strongly at shorter wavelength (e242 =16960 l
molꢁ1 cmꢁ1).
In the presence of H2O2, the EPR spectrum at pH 12 and flow rate 135 ml hꢁ1 is
quite different (Fig. 4). H2O2 or its anions are obviously photolyzed, most likely at
wavelengths near 267 nm where there is a window between the above absorption
bands. The spectrum is nearly symmetric and consists of two groups of lines, which,
based on their intensities and their variation with experimental conditions, cannot be
assigned to a single radical. From simulations and comparison of the coupling constants
with literature values [15], it was concluded that the observed spectrum consists of a
superposition of the four different radical species 1–4 with almost identical g values
(Table 3). Due to the mesomerism, the largest coupling constants are in the ortho-
and para positions relative to the phenoxy radical center.

2362
Helvetica Chimica Acta – Vol. 89 (2006)
ꢁ1
Fig. 4. a) Experimental EPR spectrum obtainedwith BPA at pH 12 andat 135 ml h
flow rate, b)
simulation of the spectrum for the sum of radicals 1–4, c) simulation of the spectrum of radical 1
(33%), d) simulation of the spectrum of radical 2 (5%), e) simulation of the spectrum of radical 3
(14%), and f) simulation of the spectrum of radical 4 (48%)
BPA is a symmetric compound; therefore, only one aromatic ring will be taken into
consideration. No coupling through the Me2
ACHTREUNG
the same ꢀp-semiquinoneꢁ radical anion as observed in the absence of H2O2.
AHCTREUNG
from the fraction of light that is absorbed directly by the monomer. For the other spe-
cies, the number of doublet splittings in the spectrum shows that there are only two H-
atoms for radicals 2 and 3, and three for radical 4. The lack of H-couplings within aro-

Helvetica Chimica Acta – Vol. 89 (2006)
2363
matic rings suggests that secondary reactions like further hydroxylation or dimerization
of the monophenoxy radicals occur. The latter can be excluded at higher flow rates,
because of the absence of the expected additional splittings from the H-atoms in the
neighboring ring. At lower flow rates, the spectra are different (Fig. 5), but in contrast
to Fig. 3 where additional structure appears at the higher flow rate, we have more and
narrower lines at low flow rate. This points to an explanation involving primary radicals
predominantly at high flow rates and secondary ones at lower rates with corresponding
extended residence time in the EPR-active zone of the cell. Due to dimerization, addi-
tional smaller coupling constants appear. Coupling of phenoxy radicals forming either
phenoxy-, phenol- or biphenyl-like radicals has previously been described [17][18], but
the possible formation of consecutive new monomeric species should nevertheless not
be excluded.
Fig. 5. EPR Spectra obtainedwith BPA at pH 12 andat different flow rates: a) 90 ml hꢁ1; b) 20 ml hꢁ1
and c) 8 ml hꢁ1
An additional contribution to the formation of higher-substituted products may be
due to the increased absorption coefficients of H2O2 solutions under alkaline conditions
(pH 0.45: e230 =42 l molꢁ1 cmꢁ1 for H2O2; pH 12.6: e230 =317 l molꢁ1 cmꢁ1 for HOOꢁ).

2364
Helvetica Chimica Acta – Vol. 89 (2006)
The signal amplitudes arising from the four observed radicals depend only slightly on
the flow rates (not shown) and do not permit a clear conclusion as regards radical-for-
mation mechanism. For the determination of the relative intensities, a clear EPR line,
not overlapping with other signals, was chosen for each individual radical. Its intensity
(1/2·DH (line width)·signal intensity) was used to determine the cumulative intensity
of all lines of the individual radical.
It is clearly seen that all four radical species 1–4 exist in a specific pH range
between pH 9 and 14 (Fig. 6). The relative concentrations of the four radicals remain
the same, and a maximum of the absolute intensity is adopted at pH 11.8. This behavior
will be discussed in Sect. 4.2.
~
!
^
*
Fig. 6. Relative intensities of the EPR signals of the radicals 1 ( ), 2 ( ), 3 ( ), and 4 ( ) observed
with BPA at 90 ml hꢁ1 flow rate in dependence on the pH value
Experiments with the DBNBS spin trap at pH 0.4 and a flow rate of 90 ml hꢁ1 led to
the same spectrum (not shown) as observed with SBP, arising from the CCH2OH radical.
At the given concentration, BPA cannot successfully compete for HOC with the MeOH
solvent. Nevertheless, a change of color of the solution demonstrates that photochem-
ical reactions do occur.
3.3. [1,1’-Biphenyl]-4,4’-diol (BPD). BPD has a strong absorption band at 286 nm
(e286 =32400 l molꢁ1 cmꢁ1 at pH 11.7) which shifts to 262 nm and somewhat lower
absorption at pH 10.35. In the absence of H2O2, the experimental EPR spectrum of 2
mM BPP consists of a multiplet with 13 equidistant lines. The intensity distribution is
compatible with two sets of four equivalent protons with coupling constant which are
a factor of two different (0.92 G and 1.84 G) but not with 12 equivalent protons. It is

Helvetica Chimica Acta – Vol. 89 (2006)
2365
conceivable that this spectrum is due to the ꢀp-dipheno-semiquinoneꢁ anion radical
(=4-(4-oxocyclohexa-2,5-dien-1-ylidene)cyclohexa-2,5-dien-1-one radical ion (1ꢁ)) 5
which may be formed by photooxidation followed by deprotonation of BPD, although
a somewhat different set of coupling constants (0.56 and 2.25 G, the sum of which is
nearly the same as that of the present observation) has been reported for this species
in the literature [19]. The spectrum is considerably weaker near pH 10 than at pH 12.2.
In the presence of H2O2, the 13-line EPR spectrum of radical 5 (Fig. 7,a) is absent,
and the observed EPR signals are strongly pH dependent (Fig. 7,b–e), in a similar
sense as seen before with BPA. Simulating the experimental EPR spectrum at pH
10.3 and flow rate 135 ml hꢁ1, and comparing the coupling constants with literature val-
ues [15], it was concluded that the observed spectrum consists of five different, simul-
taneously appearing radical species with identical g values (Table 3; details are reported
in [20]). The fractions which are determined by matching the intensities in the simula-
tion are 26% for radical 6, 14% for radical 7, 35% for radical 8, 17% for radical 9, and
ꢁ1
Fig. 7. EPR Spectra obtainedwith BPD acquiredat 135 ml h
flow rate and a) at pH 12.2 in the
absence of H2O2, b) at pH 10.3 in the presence of H2O2, c) at pH 11.3 in the presence of H2O2, d) at
pH 12.2 in the presence of H2O2 and e) at pH 13.0 in the presence of H2O2

2366
Helvetica Chimica Acta – Vol. 89 (2006)
8% for radical 10. The number of doublet splittings shows that there are only two H-
atoms for radicals 6, 7, and 10, and three for radical 8. The lack of H-couplings within
aromatic rings suggests that secondary reactions like further hydroxylation of the
monophenoxy radicals occur. Only radical 9 possesses four H-atoms.
It should be noted that there is a pronounced change in the spectra as the pH
increases from 10.3 to 13.0. The well-resolved spectrum with 27 almost equidistant
lines at pH 10.3 (Fig. 7,b) transforms into a less well resolved 16-line spectrum at pH
12.2 (Fig. 7,d). On further increase to pH 13.0, most but not all of the lines split into
0.1-G doublets. The total width and one dominant hyperfine splitting of ca. 3.2 G
remain roughly constant. It is likely that the changes reflect protonation equilibria,
but the details remain elusive.
4. Discussion. – 4.1. Aspects of Direct Photolysis. While it is clear that SBP and
related sulfones are degradable by UV light, little is known about the mechanism.
The formation of the ꢀp-benzosemiquinoneꢁ radical anion establishes that the molecule
is cleaved in a-position to the sulfone group, which is conceivable in analogy to the
known Norrish-type-II cleavage of ketones under conditions where photoreduction is
suppressed. It is less clear though why this does not work under acidic conditions
where only the spin-trap adduct of the hydroxymethyl radical is observed. These con-
ditions may point to the involvement of HOꢁ to trap the phenyl fragment and produce
the ꢀsemiquinoneꢁ. It is furthermore noteworthy but not so rare that the counter radical
from photolysis is not observed, a fact that may be related to a short lifetime.
It is known that BPA photoionizes at 308 nm to give a bis-phenoxy radical and a
solvated electron [21]. It is perhaps the shorter wavelength employed here which
leads to cleavage of the CꢁC bond in the bridge to provide the ꢀp-benzosemiquinoneꢁ
radical. Again, the latter is not observed under acidic conditions, suggesting that an
analogous mechanism as suggested for SBP is operative.
The BPD does not cleave but photo-oxidizes to the ꢀp-dipheno-semiquinoneꢁ rad-
ical anion. There can be little doubt about the nature of this species, and it is unclear
why the spectrum is different from the one reported in [19].
4.2. Formation of Phenoxy- andꢀSemi-quinoneꢁ Radicals andthe Origin of Its pH
Dependence. The pH dependence shown by the HOC radical in reactions with substi-
tuted aromatic hydrocarbons is related to its deprotonation (pKA 11.9) under alkaline
conditions [22]. The resulting OCꢁ radical possesses only negligible tendency to add to
the aromatic ring, so that at pH values exceeding the pKA of HOC, H-abstraction from
the substituents leading to phenoxy radicals is expected to be the predominating proc-
ess.
As detailed in Sect. 3.1, the presence of MeOH as a solvent scavenges HOC
efficiently so that no addition to the aromatic rings is observed.
It is remarkable that the maximum formation of phenoxy radicals from BPA and
SBP occurs in the pH range close to the pKA of the HOC radicals (11.9) and of H2O2
1
(11.7). This has been shown to be due to the rates of formation of O2Cꢁ and of O2
which are proportional to the products of the rate constants and the pH dependent con-
centrations of the reactants [6]. H-Abstraction from the deprotonated form HO2ꢁ is ca.
300 times faster than from H2O2 (see Table 1, Reactions a and b) [22]. The resulting O2Cꢁ
undergoes oxidation by HOC or OCꢁ, so that oxygen is formed in-situ in the reaction sol-

Helvetica Chimica Acta – Vol. 89 (2006)
2367
utions (see Table 1, Reactions c and g) [22]. Oxygen formed by this reaction has been
reported to be of a singlet character (1O2) [23]. This is obviously true, since 3O2 would
lead to an extensive spin exchange and thus to the loss of hyperfine structure in the
EPR spectra, in contrast to our observations (see, e.g., Fig. 7,e). The evolution of oxy-
gen is apparent by the bubble formation in the photolysis cell at high pH.
O2Cꢁ plays a determining role for the production of all radicals formed at high pH.
Moreover, OCꢁ and O2Cꢁ are radicals with nucleophilic nature that leads to H abstraction
from phenolic OH substituents, thus producing the primary phenoxy radicals. O2Cꢁ can
also attach directly to the aromatic ring. A suggested reaction mechanism is displayed
in Scheme 2. As may be evidenced by mesomeric structures, there is a higher probabil-
ity that the unpaired electron is at the ortho or para position to the primary oxy group.
The O2Cꢁ attacks the ortho position (the para position is sterically hindered), and the
proton is readily split off due to rearomatization under alkaline conditions. The
OꢁO bond is broken by irradiation with UV light, and an intermediate is formed
which reacts very fast via the same reaction pathway and cannot be detected at the
used flow rate. The formation of other radicals may be rationalized by a similar mech-
anism.
Scheme 2. SuggestedFormation Mechanism of Radical 2 (R=HOC6H4C(Me)2) [26]
Alternatively, radicals may be formed by reaction with singlet oxygen. 1O2 is known
to have a pronounced tendency to undergo [2+2] and [2+4] cycloadditions to organic
C=C bonds [24][25]. Thus, the addition of 1O2
directly to the aromatic ring leads to the
G
formation of the observed dioxy-substituted radicals. The O2Cꢁ radical is formed accord-
ing to Reactions a and f (Table 1), and in contrast to its protonated form (HOOC, pKA
4.8), it also shows tendencies to add directly to the aromatic rings and hence must be
considered as an intermediate to phenoxy-radical formation (Scheme 3).
Strongly deactivating substituents such as SOꢁ3 and SO2 which exert a ꢁI/ꢁM effect
direct nucleophilic substitution to the meta position. An oxy substituent which is

2368
Helvetica Chimica Acta – Vol. 89 (2006)
andFurther Dioxy-SubstitutedSpecies
Scheme 3. Possible Formation Mechanism of Radical
6
(R=HOC6H4) [13][24][25]
strongly activating, with a +I/+M effect, favors further substitution at ortho and para
positions.
4.3. Relevance for Fuel-Cell Membranes. The proton-exchange membrane is a key
component of the PEFC. In the current state of technology, perfluorinated membrane
materials such as Nafionꢃ (DuPont, USA) and Flemionꢃ (Asahi Kasei, Japan) are used
predominantly due to the attractive conductivity and chemical stability of these mate-
rials. However, for market introduction of fuel-cell products to take place, cost-compet-
itive membrane technology has to be available.
The present work investigates monomeric membrane building blocks under condi-
tions which are not in every aspect those in fuel cells. However, when high currents are
drawn, this can give rise to significant pH inhomogeneities as a consequence of local
variations in catalytic activity and in proton conductivity. Oxidative degradation of pol-
ymers is usually a radical process which involves attack of the hydroxyl and the hydro-
peroxyl radicals, HOC and HOOC. Both radicals are intermediates of the oxygen reduc-
tion in a fuel-cell process, and since hydrogen peroxide has been detected as a trace by-
product of the fuel-cell reaction, it is plausibly assumed that attack of these two radicals
is an important degradation pathway of the proton-conducting polymer membrane. At
pHꢀpKA of HOC and H2O2, reactions between the model compounds and O2Cꢁ or 1O2
are the most probable ways to the phenoxy and ꢀsemiquinoneꢁ radicals observed in this
pH range in our EPR spectra. It was shown in previous work [6] that the dominating
reaction is the addition of HOC to the aromatic rings, preferentially in the ortho-position

Helvetica Chimica Acta – Vol. 89 (2006)
2369
to alkyl and RO substituents (since in sulfonated PEK, PSU, PS, and ETFE or FEP-g-
PSA, the para position is substituted and thus blocked). The present study focuses on
the case in which O2Cꢁ or 1O2 are the damaging species above pH 11.7, and in which the
dominating reaction is that with the aromatic ring.
It is normally essential for the EPR technique that the samples are deoxygenated.
3O2 reacts with radicals by spin exchange, but in particular it adds and thus forms dioxy
radicals which are more difficult to see. Under fuel-cell operating condition there is
3
1
plenty of O2 present, at least near the cathode, but O2 is not expected to form.
1
Only under local pH inhomogeneities, O2 formation should also be considered as a
possible reaction pathway especially at pHꢀ11.7. In the presence of both HOC radicals
and oxygen, complete degradation of the aromatic rings can be achieved within a few
hours [13]. In view of this, the perfluorinated Nafionꢃ which is much more inert has an
inherent advantage over the new membranes based on aromatic hydrocarbons.
It is known that the sulfone group has a deactivating role. Thus, in comparison with
the other monomeric building blocks, SBP has an advantage in terms of chemical sta-
bility towards HOC addition. Furthermore, blocking of the aromatic ring by suitable
substitution, for example by F-atoms, limits the HOC addition reactions. SOꢁ3 Substitu-
ents are introduced to give rise to proton conductivity, but as an additional benefit, they
reduce the activity of the aromatic ring towards HOC hydroxyl radical addition (see
Table 2). Another possibility, as briefly mentioned in the introduction, is the polymer
blending. In DMFC, the most promising fuel-cell performance is obtained on mem-
branes based on sPEKs and their blends, and recent results indicate power density as
good as the ones obtained with Nafionꢃ, with the difference that polyaromatic mem-
branes can be restarted over several weeks without degradation at temperature
>1008 [1].
In addition, future in situ EPR experiments by using a complete MEA in an oper-
ating fuel cell are planned to investigate the membrane under realistic and forced run-
ning conditions. Here, radicals are no longer generated photolytically. Instead, it will be
the key aim of these investigations to see whether the same radicals which are inter-
mediates of the oxygen reduction can be liberated from the catalyst surface and
cause their havoc on the membrane. Furthermore, spin-trap agents will be used to
make elusive radicals visible. The novel method for a direct investigation of radical
processes in a running fuel cell is already developed in our group, and first results
have been reported [27].
5. Conclusions. – The 4,4’-sulfonylbis[phenol] (SBP) and bisphenol A (BPA) can be
cleaved by UV photolysis in aqueous methanol solution at pH>10 to give ꢀp-benzose-
miquinoneꢁ radical anions, whereas [1,1’-biphenyl]-4,4’-diol (BPD) does not cleave but
gives the ꢀp-dipheno-benzosemiquinoneꢁ radical anion. If in addition hydrogen perox-
ide is present, this opens new reaction pathways. At pH<10, H2O2 is photolyzed to give
HOC radicals which react more readily with the MeOH solvent than with the substrate.
The hydroxymethyl-radical intermediate can, however, been trapped by spin-trap mol-
ecules. Since both H2O2 and HOC deprotonate near pH 11.7, the new reactive species is
OCꢁ which can abstract H-atoms but shows little tendency for addition, but in particular,
O2Cꢁ or 1O2 are formed which are held mainly responsible for the formation of phenoxy-
and semiquinone-type aromatic polyoxy radicals. For fuel cells with polymer-mem-

2370
Helvetica Chimica Acta – Vol. 89 (2006)
brane electrolytes, this means that high pH values which may occur under conditions of
fuel starvation and large inhomogeneities are very damaging and lead to rapid degra-
dation of the membrane.
We greatly acknowledge financial support by the Deutsche Forschungsgemeinschaft (RO 1249/4-1)
and the Sino-German Center in Beijing. The Varian EPR spectrometer on which the present work was
performed was donated by Prof. H. Fischer.
REFERENCES
[1] J. Roziere, D. Jones, Annu. Rev. Mater. Res. 2003, 33, 503.
[2] J. A. Kerres, J. Membr. Sci. 2001, 185, 3.
[3] H. Wang, G. A. Capuano, J. Electrochem. Soc. 1998, 145, 780.
[4] F. N. Büchi, B. Gupta, O. Haas, G. G. Scherer, Electrochim. Acta 1995, 40, 345.
[5] W. Liu, D. Zuckerbrod, J. Electrochem. Soc. 2005, 152, A1165.
[6] G. Hübner, E. Roduner, J. Mater. Chem. 1999, 9, 409.
[7] S. Lunak, P. Sedlak, J. Photochem. Photobiol., A 1992, 68, 1.
[8] N. Jacob, I. Balakrishnan, M. P. Reddy, J. Phys. Chem. 1977, 81, 17.
[9] E. J. Wolfrum, D. F. Ollis, Aquat. Surf. Photochem. 1994, 2, 451.
[10] P. Neta, M. Z. Hoffmann, M. Simic, J. Phys. Chem. 1972, 76, 847.
[11] K. Eiben, R. W. Fessenden, J. Phys. Chem. 1971, 75, 1186.
[12] K. Makino, M. M. Mossoba, P. Riesz, J. Phys. Chem. 1983, 87, 1369.
[13] H. Kaczmarek, L. A. Linden, J. F. Rabek, Polym. Degrad. Stab. 1995, 47, 175.
[14] E. Roduner, S. Schlick, ꢀDegradation of Polymer Membranes Used in Fuel Cell Applicationsꢁ, in
ꢀAdvanced EPR Methods in Polymer Researchꢁ, Ed. S. Schlick, John Wiley & Sons, New York,
2006, Chapt. 8.
[15] ꢀLandolt-Boernstein, New Ser. IIꢁ, Springer-Verlag, Berlin, 1977.
[16] E. G. Janzen, Acc. Chem. Res. 1971, 4, 31.
[17] M. Ye, H. Schuler, J. Phys. Chem. 1989, 93, 1898.
[18] H. I. Joschek, S. I. Miller, J. Am. Chem. Soc. 1973, 88, 3273.
[19] F. R. Hewgill, M. C. Pass, Aust. J. Chem. 1985, 38, 537.
[20] S. Mitov, B. Vogel, E. Roduner, J. Kerres, M. Hein, D. Xing, F. Schçnberger, H. Zhang, X. Zhu, V.
Gogel, L. Jçrissen, in preparation.
[21] C. M. R. Clancy, M. D. E. Forbes, J. Phys. Chem. A 2001, 105, 9869.
[22] G. V. Buxton, C. L. Greenstock, W. P. Helman, A. B. Ross, J. Phys. Chem. Ref. Data 1988, 17, 513.
[23] R. Ruggiero, A. E. H. Machado, A. Castellan, S. Grelier, J. Photochem. Photobiol., A 1997, 110, 91.
[24] I. Saito, S. Kato, T. Matsuura, Tetrahedron Lett. 1970, 11, 239.
[25] D. R. Kearns, Chem. Rev. 1971, 71, 395.
[26] C. Went, ꢀIonic Organic Mechanismsꢁ, in ꢀDimensions of Scienceꢁ, Ed. J. J. Thompson, Macmillan
Education Ltd., London, 1986.
[27] A. Panchenko, H. Dilger, J. Kerres, M. Hein, A. Ulrich, T. Kaz, E. Roduner, Phys. Chem. Chem.
Phys. 2004, 6, 2891.
ReceivedMay 2, 2006