EPR investigation of HO· radical initiated degradation reactions of sulfonated aromatics as model compounds for fuel cell proton conducting membranes

Article abstract of DOI:10.1039/a807129b

In order to predict hydroxy radical initiated degradation of new proton conducting polymer membranes based on polystyrene, polyethersulfone, polyetheretherketone, or on polymers obtained by radiation grafting of styrene on different fluoropolymers, eight sulfonated aromatics were chosen as model compounds for EPR experiments, aiming at the identification of products of HO· radical reactions with these polymers. Photolysis of H₂O₂ was employed as the source of hydroxyl radicals. A detailed investigation of the pH profile was carried out for p-toluenesulfonic acid. Besides benzyl- and hydroxy-cyclohexadienyl radicals at lower pH values, phenoxyl radicals were identified, predominating in the pH range 10.5-13.0. A large number of new radicals give evidence of multiple hydroxylation of the aromatic rings, confirming reaction mechanisms proposed on the grounds of product analysis, but no evidence of dimerisation was found. The result as regards stability of organic proton exchange membranes for fuel cells is, that all unsaturated bonds and weakly bound atoms are subject to immediate attack by HO·. Ether links open by HO· ipso addition. Strategies for the reduction of membrane degradation should focus on a minimisation of HO· formation and of its access to the interior of the membrane.

Full text of DOI:10.1039/a807129b

J O U R N A L O F
EPR investigation of HO• radical initiated degradation reactions of
sulfonated aromatics as model compounds for fuel cell proton
conducting membranes
Gerold Hubner and Emil Roduner*
¨
C H E M I S T R Y
Institut fur Physikalische Chemie, Universitat Stuttgart Pfaffenwaldring 55, D-70569 Stuttgart,
¨
¨
Germany. E-mail: Roduner@ipc.chemie.uni-stuttgart.de
Received 14th September 1998, Accepted 23rd November 1998
In order to predict hydroxy radical initiated degradation of new proton conducting polymer membranes based on
polystyrene, polyethersulfone, polyetheretherketone, or on polymers obtained by radiation grafting of styrene on
different fluoropolymers, eight sulfonated aromatics were chosen as model compounds for EPR experiments, aiming
at the identification of products of HO• radical reactions with these polymers. Photolysis of H O was employed as
2 2
the source of hydroxyl radicals. A detailed investigation of the pH profile was carried out for p-toluenesulfonic acid.
Besides benzyl- and hydroxy-cyclohexadienyl radicals at lower pH values, phenoxyl radicals were identified,
predominating in the pH range 10.5–13.0. A large number of new radicals give evidence of multiple hydroxylation
of the aromatic rings, confirming reaction mechanisms proposed on the grounds of product analysis, but no
evidence of dimerisation was found. The result as regards stability of organic proton exchange membranes for fuel
cells is, that all unsaturated bonds and weakly bound atoms are subject to immediate attack by HO•. Ether links
open by HO• ipso addition. Strategies for the reduction of membrane degradation should focus on a minimisation
of HO• formation and of its access to the interior of the membrane.
xylene-4-sulfonate were carried out in order to study radical
intermediates in photochemical electron transfer reactions in
poly(styrenesulfonate).40,41 The products of sulfonic acid
degradation have been analysed by different methods like
HPLC,33 GC–MS,35 UV spectroscopy,41 13C NMR39 and
electron paramagnetic resonance (EPR).41,42 Apart from the
intermediary benzyl and hydroxycyclohexadienyl radicals,41,42
1 Introduction
Efforts to replace the most widely used perfluorinated NafionA
as proton conducting membranes in polymer electrolyte fuel
cells (PEFC) by cheaper, ecologically more tolerable materials
are presently undertaken by several research groups engaged
in material research worldwide.1–3 Partly fluorinated polymers
like the trifluoro styrene containing membranes RAYMIONA,
the BAM3G membrane by Ballard,4 or membranes based on
fluorine back bone copolymers grafted with polystyrenesul-
fonic acid (ETFE or FEP-g-SPS type) 3,5 like R-1010, R-4010
by Pall RAI as well as polymers consisting of unfluorinated
aromatic polymers like sulfonated polyethersulfone (PSU),
polyether(ether)ketone (PEK),6,7 or the polyethylene poly-
styrenesulfonic acid copolymers (PE/SPS) by DAIS 8 have
proven almost comparable electrochemical properties.1,3,8 As
a common feature of all of these membrane materials the
building block providing the essential acidic function consists
of sulfonic acid substituted aromatic rings.
products resulting from a substitution of the SO •− group like
3
p-cresol33,35 and from a total mineralisation33 of the molecules
have been reported.
On an experimental scale, the generation of HO• or HOO•
radicals can be achieved by
a
variety of different
methods,10,11,25,43 either by direct photochemical,11,43–45 pho-
tocatalytic46 or Fe2+ or Ti3+ catalysed (Analogous to Fenton’s
reaction)43,45,47–49 degradation of H O . Moreover, HO• rad-
2 2
icals are obtained by e− irradiation10,30,50–52 or sonolysis 53 of
aqueous solutions or by photocatalytic oxidation of water on
irradiated semiconductors like TiO .54,55
2
Different radical species are formed either by addition of
To ensure long term durability of the new materials,
additional investigations are necessary about possible degra-
dation mechanisms. HO• and HO • radicals originating from
oxygen diffusion through the membrane and incomplete
reduction at the fuel cell anode are held responsible for these
HO• to the aromatic ring (acidic to neutral conditions) or by
abstraction of H• from the side chains by O•− (alkaline
conditions). Phenolate ions react with HO• by electron transfer
so that phenoxyl radicals are formed.56
By means of EPR spectroscopy, radical intermediates of the
HO• initiated degradation of the cyclohexadienyl,41,42,57–59
benzyl,41,42,60,61 or phenoxyl41,62,63 type are experimentally
distinguishable by means of their hyperfine splittings and
g-values (see Table 1).
2
degradation reactions.5,3,9 Evidence for it is the formation of
H O as detected in product water.2 Owing to their high
2 2
oxidative potentials (2.8 and 1.7 V, respectively),10,11 these
radicals are often used in industrially applied advanced oxi-
dation processes (AOPs) of organic compounds, as in waste
water treatment,12–18 or pulp lignification and bleaching 19–22
as well as for surface modifications of certain polymers.23–29
They are thought to be responsible for cancer diseases, due to
their ability to attack DNA molecules, and they play a role in
atmospheric processes.30 Degradation reactions of sulfonated
aromatics as model compounds for detergents, medicines,
agrochemicals and colouring agents have repeatedly been
investigated in water treatment processes.31–38 Sulfonated iso-
propylbenzene and tert-butylbenzene were investigated to
model the long term stability of proton conducting membranes
in oxidative media.39 Reactions of HO• radicals with p- and
m-toluenesulfonate, p-isopropylbenzenesulfonate and m-
In the present study we investigate the degradation of p-
toluenesulfonic acid in the pH range 0.1–13.9. Direct photoly-
sis of H O was chosen as the source of HO• radicals; this
2 2
method works in a wide range of different pH values, moreover
the resulting oxygen radicals are free and not bound to a
catalyst. This study is extended to further sulfonated aromatics
with ethyl, methoxy and hydroxy substituents (shown together
with the results in Table 2, which also introduces the abbrevi-
ations used below) aiming at a more complete picture of the
chemistry of HO• radicals in such systems. Methoxy substi-
tuted compounds mimic the ether bridges in the polymers with
a PEK backbone. The study also included hydroxy substituted
J. Mater. Chem., 1999, 9, 409–418
409

Table 1 EPR hyperfine splittings in Gauss and g-values for radical intermediates of the reaction of hydroxyl radicals with aromatic hydrocarbons64
Hydroxy cyclohexadienyl
radical
ortho-semiquinone
anion radical
Benzyl radical
Phenoxyl radical
1 H(1): 36
2 H(o): 9.3
2 H(m): 2.9
1 H( p): 13.4
g#2.0022–2.0026
2 H(7): 16.32
2 H(o): 5.15
2 H(m): 1.78
1 H( p): 6.2
g=2.0026
2 H(o): 6.65
2 H(m): 1.85
1 H( p): 10.2
3 H(CH ): 0.63
1 H(4): 2.85
1 H(5): 4.13
1 H(6): 0.28
3
g=2.0046
g#2.0036–2.0058
aromatics, because hydroxy groups turned out to be important
about pH 0.1, diminish at intermediate pH and recover almost
exponentially at pH ≥12.
intermediates in the degradation of TSA and the OCH
substituted compounds. Investigations at low pH are particu-
3
Between pH 1.5 and 11.0, HO• addition to the aromatic
ring in ortho position to the methyl group forming the cyclo-
hexadienyl adduct 542,60 predominated [Fig. 1(b)]. Compared
with previous studies41,42 an additional splitting of 0.62 G was
resolved. Similar splittings have been attributed to the hydroxy
coupling in hydroxycyclohexadienyl radicals derived from
other aromatic compounds.64
A third type of radical species with slightly elevated g-value
and small hyperfine splittings was detected in the pH range
10.0–13.9 [Fig. 1(c)] with a signal intensity reaching a maximum
at about pH 11.5. These species remained undetected in the
previous EPR studies.41,42 The EPR parameters show that the
signals originate from three different phenoxyl radicals 2–4 with
almost identical g-values. The signals of 4 are much weaker
and detectable only at low flow rates of about 10 ml h−1. The
small quartet splittings for 2 and 3 point to phenoxyl radicals
larly relevant for understanding degradation in polymer
electrolyte fuel cells.
2 Experimental
EPR spectra were recorded with a Varian E-Line X-band
spectrometer. Microwave power was held at 1.5 mW. The
aqueous solutions, with concentrations of 70 mmol l−1 sulfonic
acid and 40 mmol l−1 H O unless otherwise stated, were
2 2
photolysed in a flow cell (quartz, 0.4 mm×10 mm×50 mm)
at ambient temperature within the resonator. A 500 W Hg( Xe)
high pressure mercury arc lamp (Oriel 66142) was used as an
irradiation source. Wavelengths below 210 and above 400 nm
were eliminated with
a
filter solution (1.14 mol l−1
NiSO /0.21 mol l−1 CoSO /0.01 mol l−1 H SO ).65 A constant
4
4
2
4
with a CH group in the meta position. In the ortho and para
flow (10–150 ml h−1) of the solutions through the cell was
maintained by a syringe pump. By changing the flow the
residence time within the cavity is varied so that successive
reaction products can separately be focused on.
3
positions they are expected to be 7.5 and 12.7 G.64,71
Comparison of the number of doublets observed with the
number of H atoms in the aromatic ring reveals that one
coupling is missing for the radicals 2 and 3, and two for radical 4.
The lack of H couplings within aromatic rings suggests
secondary reactions like further hydroxylation or dimerisation
of the monophenoxyl radicals. Coupling of phenoxyl radicals
forming either phenoxyl-, phenol- or biphenyl-like radicals has
previously been described.19,74–76 The formation of biphenyls
however can be excluded in our systems, because of the
absence of additional splittings from the hydrogen atoms in
the attached ring.64 We therefore focus on phenoxyphenol-like
or higher hydroxylated systems. The formation of dimers in
second order reactions is expected to depend pronouncedly on
the concentration of the substrate, but reducing the TSA
concentration by a factor of ten left the signals unchanged.
An increase of the flow rate (up to 150 ml h−1) did not
appreciably change the distribution of the signals, except for
the phenoxyl radical 4 which was only present under conditions
of low flow. The formation of dioxysemiquinone-like radicals
is the more probable reason for the lack of H couplings in 2
and 3, since the hyperfine splittings agree fairly well with
g-Values were calculated relative to the BrukerA weak pitch
sample (g=2.0028). They are accurate to 0.0002 units.
Second order effects are within this range of errors for the
comparatively small coupling constants observed in our
systems, and were neglected.
Doubly distilled water was used and the pH was adjusted
with H SO or KOH, using a Schott pH-meter (CG 817)
2
4
calibrated with commercial buffer solutions. Before irradiation,
solutions and sample cell were deoxygenated by bubbling with
nitrogen. H O (30% solution) was obtained from Merck, p-
2 2
toluenesulfonic acid and 2,5-dihydroxybenzenesulfonic acid
from Sigma-Aldrich in the purest available quality. p-ethylben-
zenesulfonic acid was obtained from Aldrich, USA, and
extracted with absolute alcohol before use.66 p-Hydroxy-
benzenesulfonic acid was received as a 65% aqueous solution
from Fluka. p-Methoxybenzenesulfonic acid was obtained by
hydrolysing the corresponding sulfonylchloride in alkaline
medium.67 1-Hydroxy-4-methoxybenzene-2-sulfonic acid as
well as 3-hydroxytoluene-4- and 5-hydroxytoluene-2-sulfonic
acid were synthesised by sulfonation with concentrated
H SO .68–70 The compounds were identified by 1H and 13C
literature increments for CH substituted ortho- or para-
3
semiquinone-like radicals64,71 (see section 3.2.).
2
4
An additional contribution to the formation of higher
substituted products may be due to the increased absorption
NMR spectroscopy.71,72
coefficient of H O solutions in alkaline conditions (pH 0.2:
2 2
3 EPR results and specific discussion
e
=54 l mol−1 cm−1 for H O , pH 12.6:
e
=
230
2 2
230
332 l mol−1 cm−1 for HOO−) with a concomitant higher rad-
ical yield. This is however not quantitatively reflected by the
relative benzyl radical intensities, which at pH 13.8 are only
about 1.5 times higher than at pH 0.2 (Fig. 2). However,
3.1 para-Toluenesulfonic acid (TSA)
Three different radical species were observed with TSA, each
of which predominated in a particular pH range [Fig. 1]. The
distribution of relative signal intensities is shown in Fig. 2,
and the EPR parameters are tabulated together with those of
all the other systems in Table 2. The signal intensities of the
benzyl radicals (1 in Table 2) seem to adopt a maximum at
benzyl radicals at pH>pK (HO•)=11.9 are formed via direct
a
H abstraction by O•− radicals, and the rate constant for this
process is only one sixth of the corresponding value for HO•
addition at lower pH (Table 3).
410
J. Mater. Chem., 1999, 9, 409–418

Table 2 EPR parameters for intermediates in the reaction of HO• radicals with sulfonated aromatics. Hyperfine splittings were derived from
computer simulations using WINSIM73 and assigned by comparison with literature values of similar radical systems41,42,60,64
Substrate
TSA
Conditions
Hyperfine splittings/G
g
Proposed structure
pH<1
pH>12
2 H(CH ): 16.03 (15.89)41
2.0027
(2.00265)41
1
2
2 H(o): 5.10 (5.07)41
(Á10 ml h−1)
2 H(m): 1.77 (1.76)41
pH 10–13
1 H(4): 2.71
2.0048
2
3
(Á10 ml h−1)
3 H(CH ): 1.11
3
1 H(5): 4.50
pH 10–13
(Á10 ml h−1)
1 H(2): 3.15
2.0049
2.0049
3 H(CH ): 1.20
3
1 H(5): 2.09
pH 10–13
3 H(CH ): 2.59
4
5
3
(10 ml h−1)
1 H(5): 4.16
pH 2–10
1 H(1): 32.30 (32.5)42
1 H(OH): 0.62
2.0027
(2.0025)41
(Á10 ml h−1)
3 H(CH ): 9.60 (9.5)42
3
1 H(3): 2.78 (2.75)42
1 H(4): 12.90 (13.0)42
1 H(6): 8.50 (8.5)42
KRS
pH 5
(Á10 ml h−1)
1 H(2): 6.62
2.0050
6
3 H(CH ): 1.94
3
1 H(4): 10.41
1 H(5): 1.52
pH>13
(Á10 ml h−1)
1 H(4): 2.64
2.0049
2.0048
2
3
3 H(CH ): 1.13
3
1 H(5): 4.57
pH>13
(Á10 ml h−1)
1 H(2): 3.14
3 H(CH ): 1.20
3
1 H(5): 2.05
7<pH<13.5
(Á10 ml h−1)
3 H(CH ): 2.61
1 H(5): 4.18
2.0049
2.0046
4
7
3
pH 0.25
3 H(CH ): 3.63
3
(10 ml h−1)
1 H(3): 0.51
1H(OH): 1.41
1H(6): 3.48
KRZ
pHÁ7
(Á10 ml h−1)
1 H(3): 0.28
2.0050
2.0049
8
9
3 H(CH ): 3.78
3
1 H(6): 1.58
pHÁ7
(Á10 ml h−1)
3 H(CH ): 1.94
3
1 H(5): 3.76
1 H(6): 0.31
EBSA
pH<1
(Á10 ml h−1)
3 H(CH ): 17.52
1 H(CH): 15.92
2 H(o): 4.91
2 H(m): 1.61
2.0029
10
3
J. Mater. Chem., 1999, 9, 409–418
411

Table 2 (Continued)
Substrate
Conditions
Hyperfine splittings/G
g
Proposed structure
pH 10–12
(Á10 ml h−1)
1 H(4): 2.65
2.0048
11
2 H(CH ): 1.14
2
1 H(5): 4.59
pH 10–12
(Á10 ml h−1)
1 H(2): 3.16
2.0047
12
13
2 H(CH ): 1.20
2
1 H(5): 2.04
pH 6
(Á70 ml h−1)
Assignment difficult,
possibly formation of
additional isomer
MSA
0<pH<6
(Á10 ml h−1)
3 H(OCH ): 1.82
1 H(3): 1.83
1 H(4): 8.04
1 H(6): 4.57
2.0048
2.0025
14
15
3
pH 3–6
(Á50 ml h−1)
1 H(1): 26.16
1 H(OH): 0.47
3 H(OCH ): 0.98
3
1 H(3): 2.37
1 H(4): 12.53
1 H(6): 8.59
pH 10–12
(Á10 ml h−1)
3 H(OCH ): 0.77
1 H(4): 1.13
1 H(5): 4.88
2.0050
2.0050
16
17
3
pH 10–13.5
(100 ml h−1)
1 H(3): 0.96
1 H(5): 3.09
1 H(6): 1.35
pH 10–13.5
1 H(5): 3.55
1 H(6): 1.87
2.0048
2.0050
18
19
(10 ml h−1)
pH 10–12
1 H(2): 0.36
1 H(5): 4.18
1 H(6): 8.99
(70 ml h−1)
HMSA
pH 0–6
(>70 ml h−1)
1H(3): 0.28
2.0050
2.0050
2.0050
2.0031
20
21
21
22
1 H(OCH ): 2.18
3
1 H(5): 0.69
1 H(6): 5.35
pHÁ5
(10–150 ml h−1)
1 H(3): 1.54
1 H(5): 2.64
1 H(6): 3.03
DHSA
HSA
pHÁ5
(10–70 ml h−1)
1 H(3): 1.5
1 H(5): 2.65
1 H(6): 3.04
pH 6
(130 ml h−1)
1H(1): 28.25
1H(OH): 0.55
1H(OH): 0.55
1H(3): 2.30
1H(4): 12.36
1H(6): 8.33
pH>10
(130 ml h−1)
1 H(3): 0.97
1 H(5): 3.10
1 H(6): 1.35
2.0050
17
pH>10
1 H(5): 3.62
1 H(6): 1.92
2.0048
2.0051
18
(10 ml h−1)
pH>13
1 H(3): 1.58
1 H(5): 2.62
1 H(6): 3.08
21?
(10–150 ml h−1)
pH<1
(150 ml h−1)
1 H: 13.0
1 H: 8.8
1 H: 2.2
1.9924
23?
In fact several products of H O decomposition (HO•,
2 2
In addition to the signals originating from organic radicals
a singlet with g=2.0033 was observed in most of the spectra
HO •, O •−, O ), some of them generated by mutual reactions
2
2
2
of the primary species HO•/O•−, participate in the formation
of the phenoxyl radicals (see section 4.3). Nevertheless, as
seen in Fig. 2, the total radical yield is not a strong function
of pH, except below pH 2 where it drops off by a factor of two.
[see, for example, Fig. 1(a)], even in the absence of H O . It was
2 2
identified in the reaction with the 5,5-dimethyl-1-pyrroline-N
oxide (DMPO) spin trap as the SO •− radical,36 originating
3
from photolysis of TSA.37,38 At elevated pH values it is no
412
J. Mater. Chem., 1999, 9, 409–418

Fig. 1 The dependence of pH upon the EPR spectra obtained during
TSA degradation by HO• radicals (q=quartet, t=triplet) (a) benzyl
Fig. 3 The dependence of the molar ratio between H O and TSA (at
2
2
a fixed TSA concentration of 70 mmol l−1) upon the normalised
radical 1 and SO •− (*) at pH 0.4 (25 ml h−1), (b) hydroxycyclohexa-
3
relative intensities (1/2×DH ×signal height) of EPR signals for TSA
dienyl radical 5 at pH 4.8 (90 ml h−1) (c) phenoxyl radicals 2 and 3
pp
degradation by HO• radicals at a flow rate of 100 ml h−1 (1) benzyl
at pH 11.5 (20 ml h−1).
radical 1 at pH 0.6, (&) hydroxycyclohexadienyl radical 5 at pH 4.8,
(6) phenoxyl radical at pH 12.6 (initial pH value).
additional H O (pH 5 and also pH 0.6: TSA:
e
=
2 2
230
4865 l mol−1 cm−1, H O : e =54 l mol−1 cm−1). Obviously,
2 2 230
most of the light is absorbed by TSA, which then transfers the
energy efficiently to H O . Alternatively, one would have to
2 2
postulate photoionisation of TSA and subsequent reaction of
the radical cation with H O . In the absence of H O only the
2 2
2 2
SO •− radical is observed. The occurrence of a maximum at
3
equimolar concentrations in the case of the benzyl radical on the
other hand may have to be attributed to further reactions with
H O , leading to a drop in the steady state signal intensity (based
2 2
on the bond strengths, abstraction of H from H O by benzyl is
2 2
thermoneutral, while it is endothermic by 65 kJ mol−1 for cyclo-
Fig. 2 The dependence of pH upon the normalised relative intensities
of EPR signals obtained on TSA degradation by HO• at a flow rate
of 125 ml h−1: (1) benzyl radical 1, (&) hydroxycyclohexadienyl
radical 5, (6) phenoxyl radical (sum of radicals 2 and 3).
hexadienyl radicals). For the phenoxyl radicals (pH 12.6) no
measurements were possible above 0.14 mol l−1 H O due to
2 2
extensive gas evolution at these pH values (presumably singlet
oxygen, see section 4).
Table 3 Rate constants for reactions of HO• radicals10,20,41
3.2 3-Hydroxytoluene-4-sulfonic acid (KRS) and 5-
hydroxytoluene-2-sulfonic acid (KRZ)
Rate constant,
k/109 dm3
mol−1 s−1
KRS is a p-toluenesulfonic acid which already bears a hydroxy
group meta to the CH group. In reactions with H O at
Reaction
3
2 2
a: pH>11.7: HO•+HO −H O+O •−
7.5
0.027
8
pH 13.5, where HO• is no longer available for addition due to
2
2
2
2
b: pH<11.7: HO•+H O H O+HO •
the predominance of O•− [pH>pK (HO•)] we obtained the
2
2
2
c: HO•+O •−HO−+1O
a
2
2
phenoxyl radical EPR spectra for 2 and 3 which were also
d: HO•+4-CH C H SO −4-CH (OH)C H SO •− 6.1
3
6
4
3
3
6
4
3
observed with TSA. This is clear evidence that additional oxy
substitution can occur by mechanisms other than direct HO•
addition to the ring. At elevated pH values this occurs possibly
e: O•−+4-CH C H SO −
0.99
3
6
4
3
4-C•H C H SO −+HO−
2
6
4
3
f: HO•+HO•H O
g: O•−+HO −HO−+O •−
5.5
0.4
0.6
2
2
via the reaction of O•− with HO − (see Discussion).
2
2
2
h: O •−+O•−+H O2 HO−+1O
2
At pH 7–12 a species predominates which shows the same
2
2
hyperfine splittings as the phenoxyl radical 4 from TSA, that
was observed only at reduced flow rates. These signals probably
originate from the corresponding 1,2,4-trioxy radical. When the
pH is further decreased to about 5 a rather weak and poorly
resolved spectrum with an additional splitting is obtained. It
represents the phenoxyl radical 6, obtained by H abstraction
from KRS. At pH<1 a weak spectrum was observed which we
assign to the protonated para-semiquinone radical 7.
From KRZ, a byproduct in the synthesis of KRS, a
superposition of only two phenoxyl radicals was obtained in a
wide range of pH>7. They are attributed to the two ortho-
semiquinone radicals 8 and 9. Weak and poorly resolved spectra
were obtained with both compounds at lower pH values. This is
probably due to HO• addition to several positions of the aromatic
ring, resulting in different cyclohexadienyl radicals and several
successive products in relatively small yields.
longer observed in accord with the increased absorption
coefficient of H O relative to TSA. The corresponding phenyl
2 2
radical was not observed, its lifetime is too short so that the
stationary concentration remains below the detection limit of ca.
10−9 mol l−1 of paramagnetic centres.77 The formation of cresols
by SO -substitution 33,35 was not detected, as coupling constants
3
for the benzyl and hydroxy-cyclohexadienyl radicals agree well
with those for the SO -substituted compounds,41,42 whereas HO
3
substitution should lead to slightly different values.60,64
In another experiment the molar ratio between H O and TSA
2 2
was varied between 0.03 and 33.2 with a fixed TSA concentration
of 70 mmol l−1 at three characteristic pH values. At low H O
2 2
concentrations the intensities increase linearly, showing that HO•
is the precursor of all of them (Fig. 3). For the benzyl radicals
(pH 0.6) the relative signal intensities show a distinct maximum
for about equimolar concentrations of H O and TSA, whereas
for the cyclohexadienyl radical (pH 4.8) a saturation behaviour
3.3 para-Ethylbenzenesulfonic acid (EBSA)
2 2
is suggested when the molar ratio approaches 10. This reflects
the high optical density where no light is available to cleave the
With EBSA, radicals analogous to those observed with TSA
were obtained in about the same pH ranges. At pH 0.3, the
J. Mater. Chem., 1999, 9, 409–418
413

Fig. 5 The dependence of pH and flow rate upon the EPR spectra
obtained during HSA degradation by HO• radicals (a) 17 at pH 11.5
(130 ml h−1), (b) 18 at pH 11.5 (10 ml h−1), (c) 21 at pH 13.5
(130 ml h−1).
Fig. 4 The dependence of pH upon the EPR spectra obtained during
MSA degradation by HO• radicals, SO •− (*) (d=doublet, q=
3
quartet), (a) phenoxyl radical 14 at pH 0.4 (110 ml h−1), (b) hydroxy-
cyclohexadienyl radical 15 at pH 4.8 (110 ml h−1), (c) phenoxyl
radicals at pH 11.5 (110 ml h−1).
nates, whereas at higher pH values the radical 21 without the
OCH substituent was obtained. DHSA lead to a radical with
3
methyl substituted benzyl radical 10 was observed. At pH 11.2,
again a superposition of two phenoxyl radicals was attributed
to the CH CH substituted ortho- or para-semiquinone radicals
coupling constants identical to 21. With these compounds, H
(and CH !) loss from the oxygen substituent, can already be
achieved by UV-illumination of the solutions in the absence
3
2
3
of H O . This is in contrast to MSA, where fission of the
11 and 12. A rather complex EPR spectrum at pH 5.3 probably
belongs to the hydroxycyclohexadienyl radical 13. Owing to
too many overlapping signals we were unable to further assign
it. Similar problems were reported for the isopropyl substituted
aromatic sulfonic acid.41 The same holds for pH values above
13, where besides the signals from the phenoxyl radicals, a
variety of additional EPR lines were detected, possibly based
on a superposition of radicals due to H abstraction from the
CH and CH groups or oxy adducts, respectively.
2 2
OCH ether bridge occurred only when H O was present
3
2 2
during photolysis. With HMSA and DHSA no hydroxycyclo-
hexadienyl radicals and no higher hydroxylation products of
the aromatic ring were found.
3.6 para-Hydroxybenzenesulfonic acid (HSA)
At pH 6 HSA showed signals of an o-hydroxycyclohexadienyl
radical 22. They appear only at elevated flow rates
(130 ml h−1) and are accompanied by the above mentioned
phenoxyl radicals 17 and 18 from MSA, which predominate
at low flow rates (section 3.4). At about pH 10 and flow rates
of 130 ml h−1 the phenoxyl radical with splittings due to three
different H atoms dominates [Fig. 5(a)], whereas at reduced
flow the spectrum of another species with one splitting less
becomes more clearly visible [Fig. 5(b)]. We attribute the first
signals to the ortho oxy species 17, which was also observed
with MSA and which by longer exposure to HO• (lower flow
rate) is further hydroxylated to the corresponding trioxy
radical, which is again the same as 18 from MSA. A similar
product distribution has been reported for p-hydroxybenzoic
acid, where also ortho addition occurred.82 Omura and
Matsuura assigned the species with the three oxy groups to
the 2,3,4 and the symmetric 3,4,5 isomers. The two different
coupling constants in our case point to the asymmetric product,
which we attribute to the 1,2,3 isomer 18. These signals are
also observed in smaller amount at pH values above 13, but
here a third phenoxyl radical predominates, also consisting of
splittings due to three H atoms [Fig. 5(c)]. This species has
almost the same coupling constants as the radical 21 obtained
by H abstraction from the dihydroxy compound DHSA and
3
2
3.4 para-methoxybenzenesulfonic acid (MSA)
MSA shows a behaviour totally different from that of the
other compounds at pH<1. A phenoxyl radical 14 is obtained
with a somewhat misleading quintet splitting because one of
the H splittings on the aromatic ring has exactly the same
value as the OCH substituent. This is typical for o-OCH
3
3
substituted phenoxyl radicals,64,71 and the splitings of 1.85 G
are correct for the ortho species, whereas 0.6 and 2.0 G were
reported for the meta and para-OCH groups, respectively.
3
The radical is still observed at reduced intensity at pH 5.3, but
when the flow rate is increased to 100 ml h−1 it disappears,
and the o-hydroxycyclohexadienyl radical 15 [Fig. 4(b)] is
observed. In contrast to the reactions of p-methoxybenzoic
acid with O•−51 and the reactions using photolysed di-tert-
butyl peroxide as radical initiator78 no H• abstractions from
the methoxy group were observed in our investigation.
At pH 10–12 a superposition of several phenoxyl radicals was
obtained. With a rather small signal, radical 16 was detected,
which again lacks one H splitting compared with the phenoxyl
radical 14 at lower pH values [Fig. 4(c)]. It is probably again
due to additional hydroxylation. The splittings agree quite well
by OCH substitution in HMSA, and the question is, how a
phenoxyl radical species resulting from HSA can have the
same hyperfine splittings. One explanation is that in all of
the three molecules the SO group has been replaced by an O
with those of the corresponding OCH substituted ortho-semiqui-
3
3
none.64 In addition, intense spectra belonging to a variety of
phenoxyl radicals 17–19 appear above about pH 10 [Fig. 4(c)]
which do not show splittings due to the OCH group anymore.
3
3
group, but the resulting benzene with three oxy groups is
Identical coupling constants resulted from HSA, indicating that
reported to have different coupling constants (1H: 0.6 G, 1 H:
4.98 G, 1H: 1.34 G).64 Another explanation is that the initially
formed ortho-semiquinone radical rearranged into the more
stable para-isomer, so that finally phenoxyl radicals identical
to the ones from DHSA and HMSA result.
In contrast to an earlier investigation of HSA by Dixon and
Murphy,42 the phenoxyl radical resulting from direct H
abstraction from the OH group (2 H: 7.25 G, 2 H: 2.1 G) was
the OCH ether bridge is not stable at elevated pH values (see
3
section 3.6 and discussion).79–83
3.5 1-Hydroxy-4-methoxybenzenesulfonic acid (HMSA) and
2,5-dihydroxybenzenesulfonic acid (DHSA)
Fission of the OCH ether bridge is also observed with HMSA,
3
where in the pH range 0–6 at flow rates Á70 ml the product
20 of direct H abstraction from the OH group still predomi-
not detected. Dixon and Murphy used Ti3+–H O as the
2 2
414
J. Mater. Chem., 1999, 9, 409–418

source of HO• radicals, and the question is, whether this
system offers a sufficiently large number of free HO• radicals
for addition, or whether a fraction of them remains complex-
4.3 Formation of polyoxy aromatics
Further hydroxylation by HO• is the most direct way to
polyoxy aromatics,63,84 and it is probably responsible for the
formation of higher hydroxylated products at neutral and
acidic pH values according to Fig. 7 (I). This mechanism
bound to Ti3+ and therefore prefers to react by
H
abstraction.47,49
At acidic pH values all hydroxylated compounds lead to
rather poorly resolved EPR spectra with low line intensities.
requires the presence of O in solution, it could thus be very
2
relevant under actual fuel cell conditions. In the present work
Besides the singlet attributed to SO •− they contain triplet
3
however, the solutions were O -free.
splittings belonging to not further attributable oxy radicals
2
Owing to its electrophilic nature, the HO• radical shows
(g=2.005; 2.8 and 4.8 G) and some unassigned lines in the
coupling range 0.6–1 G (see Discussion). Here again, HSA
and DHSA lead to almost identical spectra. Sometimes the
hydroxylated compounds known from the higher pH values
are still detectable (HMSA), but with much lower resolution.
With HSA at pronouncedly increased flow rates (up to
150 ml−1) an additional species with a lower g-value of 1.9924
appears, possibly originating from an acyl type s radical 23
due to rupture of the aromatic ring. The details of its structure
are uncertain, the hyperfine splittings are somewhat large for
the structure suggested in Table 2.
preference for ring positions ortho or para to an activating
substituent. According to Hammett’s law,88,89 the rate constant
for HO• addition to an aromatic ring is about doubled when
there is an OH substituent in para (ortho) position to the site of
attack, whereas electron withdrawing substituents (SO −) con-
3
siderably reduce it. A number of representative rate constants
are given in Table 4. The highest values approach the diffusion
limit. The data show further that addition of HO• radicals to
compounds which already bear hydroxy substituents is fav-
oured.82 A further increase occurs for the deprotonated forms of
the phenols (pK 7–10).89,90 This is probably the reason why
a
with p-hydroxysulfonic acid (HSA) only higher substituted
4 Discussion
phenoxyl radicals are observed at alkaline pH values.
Oxygen has a strong tendency to add directly to cyclohexa-
dienyl-like adducts, forming peroxy radical intermediates,
which can lead to higher hydroxylated products as shown in
Fig. 7, I19,63,91 and finally to bond breaking reactions (Fig. 7,
II).23,29,33,43,92
The phenoxyl radicals derived from TSA are, however, still
observed at pH 13.9 where the formation of hydroxycyclohexa-
dienyl adducts seems no longer possible due to the predomi-
nance of the O•− radicals (see Fig. 9), they can therefore be
excluded as intermediates and other mechanisms which lead
to additional oxygen substituents on the aromatic rings have
4.1 Reasons for pH dependence
The pronounced pH dependence shown by the HO• radical in
reactions with substituted aromatic hydrocarbons has been
attributed to its deprotonation (pK =11.9)10 in alkaline con-
a
ditions. The resulting O•− radical possesses only negligible
tendency to add to the aromatic ring,10,20 so that at pH values
exceeding the pK of HO•, H• abstraction from the substituents
leading to benzyl or phenoxyl radicals is expected to be the
predominating process.
a
4.2 Formation of benzyl radicals
The appearance of benzyl radicals in acidic conditions was
explained by an acid catalysed loss of water from hydroxy-
cyclohexadienyl radicals, as given in Fig. 6 and previously
suggested by others.37,42,50 The higher extent of protonation
of the hydroxycyclohexadienyl intermediates at decreasing pH
leads to the almost linear increase of the intensity in the pH
range 1.5–0.5 (Fig. 2) and to a saturation behaviour at pH
<0.3, presuming that the formation of hydroxycyclohexadi-
enyl adducts is independent of pH. A competing water elimin-
ation from the unprotonated adducts which has been suggested
to also lead to benzyl radicals49 was not observed here, the
formation of benzyl radicals was almost negligible at higher
pH values (see Fig. 2). H O elimination at acidic pH values
2
can only arise from compounds with alkyl or hydroxy substitu-
ents with a-H atoms, like toluenes or phenols,23,29,33,43,63,84
but it is not possible from b-H atoms or from the OCH
group in MSA.
3
Radical cations have also been reported as intermediates in
the formation of benzyl radicals, when aromatics substituted
by electron donating methyl, methoxy or amino groups are
involved.19,85–87 Even though not yet observed with TSA,
possibly due to the electron withdrawing property of the SO H
group,41 they have to be considered as short-lived intermediates
in our case as well.
3
Fig. 7 Mechanism for consecutive reactions of hydroxycyclohexadi-
enyl radicals with oxygen leading to different phenolic compounds
and to bond breaking.29,33,34,43,93
Table 4 Rate constants for the para addition of OH• radicals to
substituted benzenes88
Substituent
Rate constant, k
/109 dm3 mol−1 s−1
OH+Substrate
H
3.3
1.76
3.0
3.6
5.1
SO −
3
CH
3
OCH
3
Fig. 6 Formation of benzyl radicals by acid catalysis from hydroxy-
cyclohexadienyl radicals.42,50
OH
J. Mater. Chem., 1999, 9, 409–418
415

to be considered. The fact that the same polyoxy radicals were
observed with unhydroxylated and with partly hydroxylated
reactants (TSA, KRS) is a clear indication, that these species
arise from successive reactions.
It is remarkable that the formation of phenoxyl radicals
from TSA as shown in Fig. 2 occurs just in the pH range close
to the pK of the HO• radicals (pK =11.9) and of H O
a
a
2 2
(pK =11.7). H• abstraction from the deprotonated form
a
HO − is about 300 times faster than from H O (see Table 3,
2
2 2
reactions a and b).20 The resulting O •− undergoes oxidation
2
by HO• or O•− (see Table 3, reactions c and h), so that oxygen
is formed in situ in the reacting solutions.10 This is supported
by the evolution of gas observed in this pH range at elevated
H O concentrations. Gas still evolves when H O is photo-
2 2
2 2
lysed without the addition of a substrate molecule, showing
that it does not originate from mineralisation of the organic
substrate to SO or CO as mentioned elsewhere.33,35,37,38,93
Fig. 9 Simulation of the relative rates of formation of the superoxide
radical anion (&) O •− and singlet oxygen ($) 1O according to the
2
2
2
2
Oxygen formed by this reaction is reported to be of singlet
rate constants (Table 3) dependence upon pH (assumed concentrations
character.22,93 This is obviously true, since 3O would lead to
after photolysis: HO•: 40 mmol l−1, H O : 20 mmol l−1). (1) c(O•−),
scale on the left, filled symbols to the scale on the right.
2
2
2
2
(%) c(HO•), (6) c(H O ), (×) c(HO −). Open symbols relate to the
extensive spin exchange and thus to the loss of hyperfine structure
2
2
in the EPR spectra, in contrast to our observation [Fig. 1(c)].
Because the rate constants of the reactions contributing to
asymmetry is due to the growing importance of reaction g
since the concentrations of both reactants increase with pH,
whereas reaction a is responsible for the maximum since the
opposite pH dependence of the two reactant concentrations
gives a significant contribution only at an intermediate pH of
about 11.7. This correlation suggests that reactions between
TSA (or the other compounds at pH>pK of HO•) and O •−
the formation of oxygen and O •− (see Table 3, reactions a
direct HO• addition and H• abstraction reactions (see Table 3,
2
and c) are of the same order of magnitude as the competing
reactions d and e) the resulting concentration of 1O reaches
2
quite high values. 1O is known to have a pronounced tendency
2
to undergo direct [2+2] and [2+4] cycloadditions to organic
double bonds.94,95 Hydroxycyclohexadienyl radicals are not
a
2
or 1O are the most probable ways to the phenoxyl radicals
needed as precursors since 1O can add directly to the aromatic
2
2
observed in this pH range.
ring, introducing oxygen substituents as shown in Fig. 8 (I).
Investigations of lignin model compounds showed rapid break-
down of the molecules under alkaline conditions, which was
4.4 OCH ether bridge fission
3
attributed to singlet oxygen.22 The ability of 1O to add directly
Formal ipso substitutions of the OCH groups in different
substituted methoxy compounds by HO• radicals as observed
with MSA are well known.79,81,82,83 As a mechanism leading
to the phenoxy radicals, acid catalysed79,80 or spontaneous
79,81,82 elimination of methanol from the ipso hydroxycyclo-
hexadienyl adducts was identified. With activating hydroxy
2
3
to aromatic rings may be the reason for the formation of
phenoxyl radicals at pH values where hydroxycyclohexadienyl
intermediates are no longer formed (e.g. TSA, KRS). The
O •− radical formed according to reactions a and g (Table 3)
2
in contrast to its protonated form (HO •, pK =4.8) also
2
a
shows tendencies to add directly to aromatic rings and hence
must be considered as an intermediate to phenoxyl radical
formation as shown in Fig. 8 (II).30
substituents in the ortho or para positions to OCH the yield
of methanol was 4–5 times larger than for the meta substituted
3
isomers.81 The same mechanism was proposed for phenoxy
phenols like thyronines, where the rupture of the diphenyl
ether link occurred at higher HO• concentrations.82 In the
present work with MSA, ipso substitution occurred only at
pH values above 10. This is in accord with the expected
deactivation of MSA by the SO group, so that ipso sub-
stitution is obviously not possible until activating oxy substitu-
When the formation of O •− comes about mainly by the
2
reactions a and g (Table 3), its formation should strongly
depend on the pH value. The same holds for the formation of
1O according to reactions c and h (Table 3). The relative
2
rates of formation for O •− and 1O , given by the products
2
2
3
of the rate constants and the pH dependent relative concen-
trations of the reactants show maxima for O •− at pH 11.9
ents are present in the molecule. The fact that CH photoelim-
2
and for 1O at 11.6 (Fig. 9). These species thus predominate
2
3
ination from HMSA occurs even in the absence of H O is
2 2
in the same pH range where the phenoxyl radicals occur in
probably due to the stability of the resulting para-semiquinone
Fig. 2. The curve for O •− even reproduces the asymmetric
radical, the formation of which is favoured. With DHSA the
para-semiquinone radical is also formed by photolysis. This is
reported to occur by a homolytic cleavage of the O–H bond
under basic conditions, or by an electron ejection from the
phenolate ions.76,90
2
shape of the phenoxyl radical intensity distribution. The
4.5 Semiquinone spectra at low pH values
In contrast to the well resolved benzyl radical spectra obtained
with TSA or EBSA, or of the phenoxyl radical from MSA at
acidic pH values, HSA and DHSA lead to poorly resolved
EPR spectra. The two compounds give the same EPR spectra,
which can only be explained when SO •− elimination or
3
substitution has taken place to a high extent, so that the
identical dihydroxylated aromatics result. Loss of resolution
in acidic pH ranges was ascribed to the partial protonation of
the oxy groups and to a growing tendency to exchange protons
between them.96 Only at pronouncedly higher acid concen-
trations (1.8 mol l−1 H SO ) is resolution reported to increase
Fig. 8 Proposed mechanism for reactions of aromatic hydrocarbons
with 1O and O •− leading to semiquinones.29,94,95
2
2
2
4
416
J. Mater. Chem., 1999, 9, 409–418

again when both of the oxy groups are protonated. We found
a spectrum with a lower g value at pH<1, which possibly
belongs to a s radical, originating from bond breaking of the
aromatic ring in HSA. The fact that this oxidation product is
only seen at high flow rates may be explained by the shorter
lifetime of this species compared with the radicals which still
contain aromatic rings.
5 Relevance for fuel cell membranes
The present work investigates model compounds of building
blocks of ion exchange membranes under conditions which
are not in every respect those in fuel cells. The photolysing
light absorbed by the sulfonic acid leads to the SO •− radical
3
as a direct photochemical side product, which is still observed
although not normally prominent in presence of H O .
2 2
However, the hyperfine splittings for the benzyl and cyclohexa-
dienyl radicals from TSA agree very well with the literature
values for the SO substituted radicals,41,42 indicating that no
3
SO exchange by HO• occurred.
3
It is a general belief that HO • and HO• are responsible for
the degradation process.3,5,9 This is based on the observation
2
Fig. 10 Possible sites for HO• radical attack on various polymers used
for proton conducting membranes in fuel cells.
of H O as a product of the cell reaction and on the well
2 2
known degeneration processes of polymers in gen-
eral,23,29,33,43,92 thus it is a plausible assumption for which
pH<5 so that strand breaks of PSU and of PEK may be avoided
as long as the local pH is sufficiently low. Fig. 10 surveys
positions of highest sensitivity to HO• radical attack.
there are no proofs yet. The present study focuses on the case
that HO• is the damaging species below pH 11.7. HO • is
2
formed as well via the reaction of HO• with H O (see Table 3,
2 2
It is essential for the EPR technique to work that the samples
reaction b), but under the conditions of this work the conver-
are deoxygenated. 3O reacts with free radicals by spin exchange,
2
sion is at the 1% level, and the dominating reaction of HO• is
but in particular it adds and thus forms peroxy radicals which,
that with the aromatic ring. No oxidation due to H O was
owing to their broad lines are less characteristic and more
difficult to see. Under fuel cell operating conditions there is
plenty of 3O present, at least near the cathode, but 1O is not
2 2
observed in the absence of photolysing light.
With an ion exchange capacity of the order of 1 mequiv. g−1
the pH of a water swollen membrane is of the order of zero,
but considerable local gradients across the membrane and near
the catalyst particles are expected when the cell is operated.
It is clear from the present study that the dominating reaction
is the addition of HO• to the aromatic rings, preferentially in the
ortho position to alkyl- and RO-substituents (since in sulfonated
PS, PSU, PEK and ETFE or FEP-g-PSA the para position is
substituted and thus blocked). It demonstrates the combined
ortho activation by these substituents and the meta directing effect
2
2
expected to form. We expect that 3O reacts with cyclohexadienyl
2
radicals forming peroxy radicals, which finally lead to bond
breaking within the aromatic rings (see Fig. 7).29 In the presence
of both HO• radicals and oxygen, complete degradation of the
aromatic rings can be achieved within a few hours.23,29 In view
of this, the saturated and perfluorinated NAFIONA which is
much more inert has an inherent advantage over the new
membranes based on aromatic hydrocarbons.
Since oxygen is obviously important for the degradation of
a membrane, one strategy is to prevent its diffusion into the
membrane (and across it to the other side, since for still
mysterious reasons the degradation seems to be more severe at
the anode side). This may be achieved by optimised cross
linking of the membrane, or by surface coverage with a suitable
barrier layer. Another strategy should attempt to suppress the
formation of these radicals or make them stick better to the
catalyst, but at this point it is unclear under which running
conditions they are formed in fuel cells since an in situ detection
technique is not yet available.
As a short term step to improve the stability of the aromatic
hydrocarbon based polymers we suggest to consider a blocking
of the aromatic ring by suitable substitution, so that HO•
addition is more limited, moreover, alkyl substituents with a-
H atoms, enabling the formation of benzyl radicals, should be
avoided as much as possible, if this can be done economically
and without badly influencing the other properties of the
of the SO − group in electrophilic addition reactions.
3
A crucial point is the acid catalysed water elimination from
hydroxycyclohexadienyl adducts according to Fig. 6. This pro-
cess requires the presence of labile benzylic a-H atoms, thus it
is not relevant for PSU or PEK type membranes but important
in the case of ETFE or FEP-g-PSA. In an endothermic process
this can lead to chain scission (‘reverse polymerisation’) and
thus to loss of PSA groups. Assink et al.39 found much higher
stability for the sulfonated tert-butylbenzenesulfonic acid with
no benzylic a-H atom than for the sulfonated isopropylbenzene
with one a-H atom left. They also reported that styrene grafted
membranes are much more susceptible to weight loss than the
membranes grafted with a-methylstyrene, when they are
exposed to oxidising solutions. The lack of a-H atoms certainly
contributes to the much better long term behaviour of the
RAYMIONA or the BAM3G membrane, with F in the benzylic
position as compared to PERMION 4010A, with the labile
benzylic H.2,3 When water was eliminated from cyclohexadienyl
radicals in poly(styrenesulfonate), radical cation intermediates
could be identified by UV–VIS and by EPR under acidic
conditions.40 They could also act as centres of degradation
reactions in all of the newly developed polymers.
membrane too much. SO − substituents are introduced to give
3
rise to proton conductivity, but as an additional benefit they
reduce the activity of the aromatic ring towards hydroxyl
radical addition.
Another crucial point is the loss of OCH from the methoxy
substituted compound MSA, probably initiated by ipso attack
Acknowledgments
3
of HO•. This mechanism is of relevance for the polymers with
phenoxybenzene ether bridges like PSU or PEK, possibly leading
to bond breaking within the C–O–C connections (see Fig. 10).
Fortunately, the resulting phenoxyl radical was not observed at
We are grateful for encouragement by and useful discussions
with Prof. G. Eigenberger and Drs. J. Kerres, G. Scherer and
H.-P. Brack, and we thank Dr. P. Fischer for NMR reactant
analysis.
J. Mater. Chem., 1999, 9, 409–418
417

44 N. Jacob, I. Balakrishnan and M. P. Reddy, J. Phys. Chem., 1977,
81, 17.
References
45 E. Lipczynska-Kochany, Chem. Oxid., 1994, 3, 12.
46 E. J. Wolfrum and D. F. Ollis, Aqu. Surf. Photochem., 1994, 2, 451.
47 S. Croft, B. C. Gilbert, J. R. L. Smith and A. C. Whitwood, Free
Rad. Res. Commun., 1992, 17, 21.
1
O. Savadogo and P. R. Roberge, in Proceedings of the second inter-
national symposium on new materials for fuel cell and modern battery
systems, ed. O. Savadogo and P. R. Roberge, Montre´al, 1997.
G. G. Scherer, Ber. Bunsenges. Phys. Chem., 1990, 94, 1008.
H. Wang and G. A. Capuano, J. Electrochem. Soc., 1998, 145, 780.
A. E. Steck and C. Stone, in Proceedings of the second international
symposium on new materials for fuel cell and modern battery sys-
tems, ed. O. Savadogo and P. R. Roberge, Montre´al, Que´bec,
Canada, 1997, 792.
2
3
4
48 C. Walling, Acc. Chem. Res., 1975, 8, 125.
49 T. Shiga, J. Phys. Chem., 1965, 69, 3805.
50 K. Sehested, H. Corfitzen, H. C. Christensen and E. J. Hart,
J. Phys. Chem., 1975, 79, 310.
51 P. Neta, M. Z. Hoffman and M. Simic, J. Phys. Chem., 1972, 76, 847.
52 K. Eiben and R. W. Fessenden, J. Phys. Chem., 1971, 75, 1186.
53 K. Makino, M. M. Mossoba and P. Riesz, J. Phys. Chem., 1983,
87, 1369.
54 A. Mills and S. Le Hunte, J. Photochem. Photobiol. A: Chem.,
1997, 108, 1.
55 M. A. Fox and M. T. Dulay, Chem. Rev., 1993, 93, 341.
56 J. L. Ferry and M. A. Fox, J. Phys. Chem. A, 1998, 102, 3705.
57 T. Shiga, T. Kishimoto and E. Tomita, J. Phys. Chem., 1973,
77, 330.
58 C. R. E. Jefcoate and R. O. C. Norman, J. Chem. Soc. B, 1968, 48.
59 W. T. Dixon and R. O. C. Norman, J. Chem. Soc., 1964, 4857.
60 P. Neta and R. H. Schuler, J. Phys. Chem., 1973, 77, 1368.
61 R. A. Jackson and M. Sharifi, J. Chem. Soc., Perkin Trans. 2,
1996, 775.
5
6
7
8
F. N. Buchi, B. Gupta, O. Haas and G. G. Scherer, Electrochim.
¨
Acta, 1995, 40, 345.
J. Kerres, G. Eigenberger and W. Schnurnberger, Euromembrane
’95, proceedings, 1995, vol. I, p. 284.
J. Kerres, W. Cui and S. Reichle, J. Polym. Science, Chem. Ed.,
1996, 34, 2421.
S. G. Ehrenberg, J. M. Serpico, B. M. Sheikh-Ali, T. N. Tangredi, E.
Zador and G. E. Wnek, in Proceedings of the second international
symposium on new materials for fuel cell and modern battery systems,
ed. O. Savadogo and P.R. Roberge, Montre´al, 1997, p. 828.
R. Hodgdon, J. R. Boyack and A. B. LaConti, Advance
Development and Laboratory Technical Report, No. 65DE5,
General Electric Co., West Lynn, MA, 1966.
9
10 G. V. Buxton, C. L. Greenstock, W. P. Helman and A. B. Ross,
J. Phys. Chem. Ref. Data, 1988, 17, 513.
62 B. C. Gilbert, P. Hanson, W. J. Isham and A. C. Whitwood,
J. Chem. Soc., Perkin Trans. 2, 1988, 2077.
11 R. Sapach and T. Viraraghavan, J. Environ. Sci. Health, 1997,
A32, 2355.
63 P. Neta and W. Fessenden, J. Phys. Chem., 1974, 78, 523.
64 Landolt–Bornstein, New Ser. II, ed. K.-H. Hellwege and
¨
12 R. W. Matthews, Pure Appl. Chem., 1992, 64, 1285.
O. Madelung, Springer Verlag, Berlin, 9b 66ff., 472ff., 550ff.
13 F. J. Beltra´n, G. Ovejero and B. Acedo, Water Res., 1993, 27, 1013.
(1977), 17e 34ff (1988), 9d 113 (1980).
14 R. Apak and M. Hugul, Chem. Tech. Biotechnol., 1996, 67, 221.
¨
2
65 M. Kasha, J. Opt. Soc. Am., 1948, 38, 929.
15 I. N. Martyanow, E. N. Savinov and V. N. Parmon, J. Photochem.
Photobiol. A: Chem., 1997, 107, 227.
66 A. Borthakur and V. S. B. Rao, Org. Mass. Spectrom., 1984, 19, 48.
67 M. R. Forbes and H. Maskill, J. Chem. Soc., Chem. Commun.,
1991, 13, 854.
16 R. Bauer and H. Fallmann, Res. Chem. Intermed., 1997, 23, 341.
17 F. J. Beltra´n, M. Gonza´lez and J. F. Gonza´lez, Water Res., 1997,
31, 2405.
18 M. Nahen, D. Bahnemann, R. Dillert and G. Fels, J. Photochem.
Photobiol. A: Chem., 1997, 110, 191.
19 J. Gierer, E. Yang and T. Reitberger, Holzforschung, 1992, 46, 495.
20 J. Gierer and K. Jansbo, J. Wood Chem. Technol., 1993, 13, 561.
21 D. Lachenal, N. B. Nguyen Thi, C. Chirat and L. Soria, Pap.
Timber, 1997, 79, 252.
68 J. B. Pridham and M. J. Saltmarsh, Biochem. J., 1963, 87, 218.
69 W. Baker and C. Evans, J. Chem. Soc., 1938, 372.
70 D. F. Evans and N. Iki, J. Chem. Soc., Dalton Trans., 1990, 3773.
71 G. Strandlund and P. O. Lagerstrom, Acta Chem. Scand., Ser. B,
¨
1979, 33, 261.
72 H. Cerfontain, N. J. Coenjaarts and A. Koeberg-Telder, Recl.
Trav. Chim. Pays-Bas, 1989, 108, 7.
73 NIEHS, National Inst. of Environmental Health Science, USA.
74 K. Nakagawa, S. Tero-Kubota, Y. Ikegami and N. Tsuchihashi,
J. Photochem. Photobiol. A: Chem., 1994, 60, 199.
75 M. Ye and H. Schuler, J. Phys. Chem., 1989, 93, 1898.
76 H.-I. Joschek and S. I. Miller, J. Am. Chem. Soc., 1973, 88, 3273.
77 J. A. Weil, J. R. Bolton and J. E. Wertz, Electron Paramagnetic
Resonance, John Wiley & Sons, New York, 1994.
78 A. Hudson and K. D. J. Root, J. Chem. Soc. B, 1970, 656.
79 A. Valavanidis and B. C. Gilbert, Chim. Chron., New Ser., 1995,
24, 217.
80 C. von Sonntag and H.-P. Schuchmann, in The chemistry of ethers,
crown ethers, hydroxyl groups and their sulphur analogues, Part 2,
ed. S. Patai, John Wiley & Sons, Chichester, 1980, p. 956.
81 S. Steenken and P. O’Neill, J. Phys. Chem., 1977, 81, 505.
82 K. Omura and T. Matsuura, Tetrahedron, 1975, 24, 3475.
83 T. Matsuura, T. Nagamachi and A. Nishinaga, J. Org. Chem.,
1977, 36, 2016.
84 K. Lang, J. Brodilova´ and S. Luna´k, Collect. Czech. Chem.
Commun., 1996, 61, 1729.
85 K. Sehested and J. Holcman, J. Phys. Chem., 1978, 82, 651.
86 P. O’Neill, S. Steenken and D. Schulte-Frohlinde, J. Phys. Chem.,
1975, 79, 2773.
87 D. Behar and B. Behar, J. Phys. Chem., 1991, 95, 7552.
88 M. Anbar, D. Meyerstein and P. Neta, J. Phys. Chem., 1966,
70, 2660.
89 S. Steenken, J. Chem. Soc., Faraday Trans., 1987, 83, 113.
90 H. D. Becker, in The chemistry of the hydroxyl group, ed. S. Patai;
Interscience Publishers, London, 1971, p. 894.
22 R. Ruggiero, A. E. H. Machado, A. Castellan and S. Grelier,
J. Photochem. Photobiol. A: Chem., 1997, 110, 91.
23 N. A. Weir, Eur. Polym. J., 1978, 14, 9.
24 H. Nishide, M. D. Cho, T. Kaku and Y. Okamoto,
Macromolecules, 1993, 26, 2377.
25 L. A. Linde´n, J. F. Rabek, H. Kaczmarek, A. Kaminska and
M. Scoponi, Coord. Chem. Rev., 1993, 125, 195.
26 H. Niino and A. Yabe, J. Photopolym. Sci. Technol., 1994, 7, 369.
27 P. Ulanski, E. Bothe, K. Hildenbrand, J. M. Rosiak and C. von
Sonntag, J. Chem. Soc., Perkin Trans. 2, 1996, 13.
28 H. Kaczmarek, Polym. Bull., 1995, 34, 211.
29 H. Kaczmarek, L. A. Linde´n and J. F. Rabek, Polym. Degrad.
Stab., 1995, 47, 175.
30 C. von Sonntag and H.-P. Schuchmann, Angew. Chem., 1991,
103, 1255.
31 H. Hidaka, H. Kubota, M. Gratzel and N. Serpone, Nouv.
¨
J. Chim., 1985, 9, 67.
32 J. Bandara, C. Pulgarin, P. Peringer and J. Kiwi, J. Photochem.
Photobiol. A: Chem., 1997, 111, 253.
33 B. Sangchakr, T. Hisanaga and K. Tanaka, J. Photochem.
Photobiol. A: Chem., 1995, 85, 187.
34 V. Brezova´, A. Stasko, S. Biskupic, A. Blazkova and
B. Havlinova´, J. Phys. Chem., 1994, 98, 8977.
35 V. Brezova´, M. Jankovicova´, M. Solda´n, A. Blazko´va,
M. Reha´kova´, I. Surina, M. Ceppan and B. Havlinova´,
J. Photochem. Photobiol. A: Chem., 1994, 83, 69.
36 V. Brezova´, A. Stasko and S. Biskupic, J. Photochem. Photobiol.
A: Chem., 1993, 71, 229.
37 M. Nakamura and Y. Ogata, Bull. Chem. Soc. Jpn., 1977, 50, 2396.
38 Y. Ogata, K. Takagi and S. Yamada, Bull. Chem. Soc. Jpn., 1977,
50, 2205.
39 R. A. Assink, C. Arnold, Jr. and R. P. Hollandsworth, J. Membr.
Sci., 1991, 56, 143.
40 D. Behar and J. Rabani, J. Phys. Chem., 1988, 92, 5288.
41 D. Behar, J. Phys. Chem., 1991, 95, 4342.
42 W. T. Dixon and D. Murphy, J. Chem. Soc., Perkin Trans. 2,
1974, 1630.
91 C. R. E. Jefcoate, J. R. Lindsay Smith and R. O. C. Norman,
J. Chem. Soc. B, 1969, 1013.
92 K. Okamoto, Y. Yamamoto, H. Tanaka and M. Tanaka, Bull.
Chem. Soc. Jpn., 1985, 58, 2015.
93 A. E. da Hora Machado, R. Ruggiero and M. G. Neumann,
J. Photochem. Photobiol. A: Chem., 1994, 81, 107.
94 I. Saito, S. Kato and T. Matsuura, Tetrahedron Lett., 1970, 3, 239.
95 D. R. Kearns, Chem. Rev., 1971, 71, 395.
96 I. C. P. Smith and A. Carrington, Mol. Phys., 1974, 12, 439.
43 S. Luna´k and P. Sedlak, J. Photochem. Photobiol. A: Chem., 1992,
˜
68, 1.
Paper 8/07129B
418
J. Mater. Chem., 1999, 9, 409–418