Generation of acyloxyl spin adducts from N-tert-butyl-a-phenylnitrone (PBN) and 4,5-dihydro-5,5-dimethylpyrrole 1-oxide (DMPO) via nonconventional mechanisms

Article abstract of DOI:10.1039/a701479a

The reaction between N-tert-butyl-α-phenylnitrone (PBN) and carboxylic acids has been studied Two mechanisms are discernible: the generation of PBN⁺ by oxidation of PBN with a photochemically produced excited state [from either 2,4,6-tris(4-methoxyphenyl)pyrylium ion 2⁺ or tetrachlorobenzoquinone 4], followed by reaction with RCOOH, or the addition of RCOOH to PBN to give a hydroxylamine derivative, followed by thermal oxidation by a weak oxidant. The latter sequence is the Forrester-Hepburn mechanism. In this mechanism, neither 2⁺ nor 4 is effective as an oxidant, whereas bromine could be used. Thus only oxidants with redox potentials ≥ 0.1 (SCE) are reactive enough to oxidize the intermediate hydroxylamine. This behaviour is in agreement with the redox reactivity of hydroxylamines. For the cylic nitrone, 4,5-dihydro-5,5-dimethylpyrrole 1-oxide (DMPO), acyloxyl spin adducts have been prepared by the photochemical route. The reaction between dibenzoyl peroxide and PBN to give PhCOO-PBN is not cataysed by added PhCOOH. It could be shown that the rate of formation of PhCOO-PBN is compatible with the rate of thermal decomposition of dibenzoyl peroxide. Thus dibenzoyl peroxide does not support the Forrester-Hepburn mechanism, in agreement with its redox potential of ca. -0.2 V.

Full text of DOI:10.1039/a701479a

View Article Online / Journal Homepage / Table of Contents for this issue
Generation of acyloxyl spin adducts from N-tert-butyl-á-phenyl-
nitrone† (PBN) and 4,5-dihydro-5,5-dimethylpyrrole 1-oxide (DMPO)
via nonconventional mechanisms
Lennart Eberson and Ola Persson
Department of Chemistry, Lund University, PO Box 124, S-221 00 Lund, Sweden
The reaction between N-tert-butyl-á-phenylnitrone (PBN ) and carboxylic acids has been studied. Two
mechanisms are discernible: the generation of PBN ϩ by oxidation of PBN with a photochemically
᝽
produced excited state [from either 2,4,6-tris(4-methoxyphenyl)pyrylium ion 2ϩ or tetrachlorobenzo-
quinone 4], followed by reaction with RCOOH , or the addition of RCOOH to PBN to give a
hydroxylamine derivative, followed by thermal oxidation by a weak oxidant. The latter sequence is the
Forrester–H epburn mechanism. In this mechanism, neither 2ϩ nor 4 is effective as an oxidant, whereas
bromine could be used. Thus only oxidants with redox potentials у 0.1 V (SCE) are reactive enough to
oxidize the intermediate hydroxylamine. This behaviour is in agreement with the redox reactivity of
hydroxylamines.
For the cyclic nitrone, 4,5-dihydro-5,5-dimethylpyrrole 1-oxide (D M PO), acyloxyl spin adducts have
been prepared by the photochemical route.
ؒ
The reaction between dibenzoyl peroxide and PBN to give PhCOO–PBN is not catalysed by added
ؒ
PhCOOH . It could be shown that the rate of formation of PhCOO–PBN is compatible with the rate of
thermal decomposition of dibenzoyl peroxide. Thus dibenzoyl peroxide does not support the Forrester–
H epburn mechanism, in agreement with its redox potential of ca. ؊0.2 V.
adduct being too low for EPR spectral detection if the rate of
trapping of acetoxyl is taken to be the same as that of trapping
Introduction
ؒ
Acyloxyl radicals RCOO undergo decarboxylation with rate
the benzoyloxyl radical, 5 × 106 dm3 molϪ1 Ϫ1. This rate con-
s
constants in the range of 104–1012 Ϫ1, depending upon the
s
stant is presumably a maximal one; a lower estimate4 gives
4 × 105 dm3 molϪ1 sϪ1 at 40 ЊC.
structure of R [eqn. (1)].1 The large variation in rate constants
ؒ
ؒ
Yet acetoxyl radical spin adducts of PBN have been des-
cribed. A number of oxidizing systems, such as Pb(OAc)4 ϩ a
trace of HOAc, Me3PbOAc–hν, Hg(OAc)2–hν, AgOAc ϩ Br2,
MeCOOOH–HOAc or KMnO4–HOAc, in benzene gave AcO–
RCOO → R ϩ CO2
(1)
R = Substituted vinyl 106 Ϫ 107 sϪ1
R = Substituted Ph (2–200) × 104 sϪ1
R = Substituted ethynyl (2–5) × 105 sϪ1
R = Alkyl (1–11) × 109 sϪ1
ؒ
PBN , which was considered evidence for trapping of the
acetoxyl radical.5 Somewhat later,6 it was shown that AcO–
R = Ph 2 × 106 sϪ1
R = Ar2C(OH) (1–8) × 1011 sϪ1
ؒ
PBN was produced under mildly oxidizing conditions, oxid-
ation by dioxygen of a solution of PBN in acetic acid contain-
ing some potassium acetate. This experiment excludes acetoxyl
makes acyloxyl radicals and their spin adducts important diag-
nostic tools for spin trapping mechanisms. In such reactions,
transient radicals X react with spin traps, for example N-tert-
butyl-α-phenylnitrone (PBN, 1) and can be characterized by
EPR spectral examination of the persistent aminoxyl radicals
formed [eqn. (2)]. Second-order rate constants for spin trapping
radical as an intermediate, since dioxygen [EЊ(O2/O2 Ϫ) = Ϫ0.4
᝽
ؒ
V vs. saturated calomel electrode (SCE)] cannot possibly oxidize
Ϫ
ؒ
acetate ion [EЊ(AcO /AcO ) = 2.2 V vs. SCE] to acetoxyl rad-
ical.7 Instead, a nucleophilic addition–oxidation mechanism
was proposed, shown in its general form in eqns. (3)–(5), HA =
t
t
ؒ
ؒ
Ϫ ϩ PBN
A–PBNϪ ϩ Hϩ
A–PBNϪ
(3)
(4)
Ph᎐CH᎐N(O)Bu ϩ X → Ph᎐CH(X)᎐N(O )Bu (2)
᎐
1
A
X = PhCOO (PhH) 5 × 106 dm3 molϪ1 sϪ1
ؒ
ؒ
A–PBN(H)
2 × 105 dm3 molϪ1 sϪ1
1 × 104 dm3 molϪ1 sϪ1
ؒ
ؒ
X = Ph (PhH)
ؒ
ؒ
ؒ
A–PBN(H) ϩ Ox → A-PBN ϩ Hϩ ϩ Red (5)
ؒ
X = Me3C (PhH)
X = Cl3C (CCl4)
7 × 104 dm3 molϪ1 sϪ1
.
ؒ
HOAc and Ox = a general one-electron oxidant. The inter-
mediate hydroxylamine is oxidized even by weak oxidants and
thus it was concluded that ‘results obtained from spin trapping
experiments should be used with caution whenever anionic spe-
cies are present in the reaction mixture’.6
reactions by PBN at room temperature generally fall in the
region between 1 × 104 and 107 dm3 molϪ1 sϪ1 for radicals of
interest in the context of acyloxyl radical chemistry.2
An analysis of the kinetics of the putative trapping of acyl-
oxyl radicals shows that it should be possible to trap with PBN
the benzoyloxyl radical, generated by, for example, the thermal
or photolytic decomposition of dibenzoyl peroxide, before it
has had time to undergo decarboxylation. On the other hand,
the decarboxylation rate of acetoxyl radical, 2 × 109 sϪ1, is too
high for trapping to be feasible,3 the concentration of spin
Acetoxyl radical, as well as other aliphatic acyloxyl radicals,
could also be formally ‘trapped’ when solutions of the appro-
ϩ
priate tetrabutylammonium carboxylates and PBN [EЊ(PBN
/
᝽
PBN) = 1.49 V) were oxidized by strong oxidants, such as tris-
(4-bromophenyl)ammoniumyl or hexachloroosmate() [both
with EЊ above 1.1 V vs. SCE]; in such cases, the mechanism
proposed involved the oxidation of PBN to its radical cation,
followed by reaction with a carboxylate ion [eqns. (6)–(7)].8
† IUPAC name: N-(benzylidene)-tert-butylamine N-oxide.
J. Chem. Soc., Perkin Trans. 2, 1997
1689

View Article Online
ϪeϪ
ϩ
PBN
PBN
(6)
(7)
᝽
PBN ϩ ϩ RCOOϪ → RCOO–PBN
ؒ
᝽
We have recently shown 9 that acids HA can act as promotors
or autocatalysts for the formation of spin adducts in the pres-
ence of oxidants of the general formula R–X, where the corre-
Ϫ
sponding radical anion R–X undergoes fast cleavage. The
᝽
initiating steps again involve formation of a hydroxylamine, as
Ϫ
Ϫ
ؒ
in eqns. (3) and (4), and RX cleaves to R and X thereby
᝽
generating HX [eqn. (8)] which can be entered into eqn. (3), etc.
If HX = HA, the autocatalytic case is at hand.10
ϩ
Ϫ
ؒ
ؒ
A–PBN(H) ϩ R–X → A-PBN ϩ H ϩ R ϩ X
(8)
In this context, carboxylic acids were used in the role of HA
and various weak oxidants explored. We then became intrigued
by the fact that 2,4,6-tris(4-methoxyphenyl)pyrylium ion [2ϩ;
ϩ
ؒ
EЊ(2 /2 ) = Ϫ0.6 V vs. SCE; this compound is an efficient
photoelectron transfer oxidant in its triplet state T2ϩ* with
T
ϩ
EЊ( 2 */2 ) = 1.8 V vs. SCE]11,12 could seemingly be used to oxi-
ؒ
ؒ
Fig. 1 Development of the CH3COO–PBN signal from solutions of
PBN (0.10 mol dmϪ3) and CH3COOH (0.35 mol dmϪ3) in dichlo-
romethane, prepared in a completely darkened room. The oxidant was
added after 20 min, and exposure to laboratory light was made at the
points indicated. Circles, 2ϩ (1 mmol dmϪ3); triangles, 4 (20 mmol
dmϪ3); squares, Br2 (50 mmol dmϪ3).
dize mixtures of PBN and various RCOOH to give RCOO–
PBN , thus providing easy access to these aminoxyl radicals.
ؒ
We now report the results of these studies.
Results
Reaction of PBN, acetic acid and 2؉; effect of light
Our initial experiments involved preparation of the PBN–
RCOOH solution in an EPR tube, deaeration by argon bub-
bling and final addition of the 2ϩ salt. These operations were
carried out in laboratory light and the problem of a possibly
photo-initiated reaction during the short interval (ca. 60 s)
between the addition of 2ϩ and the introduction of the sample
Me
OMe
N
O–
3
Me
O
Cl
Cl
Cl
Cl
O
–
BF4
MeO
OMe
O
+ BF4
4
–
2
ؒ
Fig. 2 Development of the CH3COO–PBN signal from a solution of
PBN (0.10 mol dmϪ3) and CH3COOH (0.35 mol dmϪ3) in dichloro-
methane, prepared under LLE conditions, upon irradiation and
renewed treatment with dioxygen
tube into the EPR cavity therefore must be addressed. The reac-
tion with acetic acid was chosen for a closer study.
Mixing PBN (0.10 mol dmϪ3) and CH3COOH (0.35 mol
dmϪ3) in dichloromethane with rigorous exclusion of oxygen in
a completely darkened environment gave an EPR-silent solu-
tion (Fig. 1, circles). After 20 min, 2ϩ (ca. 1 mmol dmϪ3) was
added to the solution and the paramagnetic activity monitored
at intervals during 0.5–1 h. No paramagnetic activity was
detected. The sample tube was then exposed to laboratory light
for 20 s which led to the development of a strong signal of
ؒ
a strong signal for CH3COO–PBN even in complete darkness
(Fig. 1, squares). This spectrum decayed rapidly with for-
mation of N-benzoyl-N-tert-butylaminoxyl radical (PBNOx,
not shown). Finally, the strong oxidant, 2,3-dichloro-5,6-
dicyanobenzoquinone DDQ [EЊ(DDQ/DDQ Ϫ) = 0.54 V], was
᝽
employed in a similar experiment but now the only radical spe-
cies detected within 2 min from mixing was DDQ in high
Ϫ
᝽
ؒ
CH3COO–PBN . Thus we conclude that the mixing of samples
concentration.
under conditions where the sample tube is exposed to labora-
tory light for a short period, allows for photogeneration of
Reaction of PBN, acetic acid and oxygen
The results above raise the question why dioxygen [EЊ(O2/O2
ϩ
Ϫ
)
ؒ
CH3COO–PBN , and that 2 is not active as a thermal oxidant
in eqn. (5), A = CH3COO.
᝽
ca. Ϫ0.4 V] can act as an oxidant toward PBN–HOAc but not
4. Therefore we repeated the original Forrester–Hepburn
experiment,6 mixing PBN (0.08 mol dmϪ3) and KOAc (0.090
mol dmϪ3) in acetic acid in laboratory light with access of air
for a short period before deaeration by argon. A weak signal of
A similar experiment was performed with the stronger oxi-
dant, chloranil [tetrachlorobenzoquinone, 4, EЊ(4/4 Ϫ) =
᝽
Ϫ0.02 V vs. SCE], with similar results. No spin adduct was
detected with complete exclusion of light, whereas exposure to
ؒ
ؒ
laboratory light for 10 min developed a signal of AcO–PBN ,
AcO–PBN was obtained (Fig. 2). This signal was unaffected
although not as rapidly as in the case of 2ϩ (see Fig. 1, triangles).
by irradiation with light of λ > 400 nm but decreased to zero on
irradiation with UV light.
A still stronger oxidant, bromine [EЊ(Br2/Br2 Ϫ) ca. 0.1 V] gave
᝽
1690
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
Table 1 Hfs constants for spin adducts formed in the reaction between PBN and RCOOH (5–18) in the presence of 2ϩ BF4Ϫ under LLE conditions
or with irradiation by light of λ > 400 nm. For pK values, see ref. 33
R in
RCOOH
pK of
RCOOH
No.
Conditionsa
aN /mT
aH /mT
5
5
5
5
5
5
5
5
6
7
7
7
7
8
8
9
9
H
H
H
H
H
H
H
H
D
H3C
H3C
H3C
3.75
DCM–LLE
DCM–hν
1.33
1.33
1.32
1.32
0.144b
0.143b
0.147b
0.149b
No signal
0.15e
AN–LLE c
AN–hνd
HFP–LLE
HFP–hν (UV)c
Benzene–LLE c
Benzene–hν
DCM–LLE
DCM–LLE g
HFP–LLE
1.43
1.31
1.31
1.35
1.36
0.14b
0.138b
0.147f
0.164
No signal
0.205
0.205
0.161
0.161
0.149
0.149
No signal
0.182
0.148
0.148
0.129
0.135
0.160
0.160
No signal
0.206
0.166
0.282
—
0.156
0.24
0.156
4.76
HFP–hν
1.47
1.47
1.35
1.35
1.34
1.34
H3C
HFP–hν (UV)
DCM–LLE
DCM–hνd,h
DCM–LLE i
DCM–hνd,h
HFP–LLE
Me2CH
Me2CH
Me3C
Me3C
Me3C
Me3C
CF3
CF3
Cl3C
Cl3C
Ph
4.84
5.03
9
9
HFP–hν (also UV)c
DCM–LLE j
DCM–hν
1.45
1.29
1.29
1.28
1.28
1.35
1.35
10
10
11
11
12
12
12
12
13
14
14
15
16
17
17
18
18
0.52
0.52
4.20
DCM–LLE
DCM–hν
DCM–LLE k
DCM–hν
Ph
Ph
Ph
HFP–LLE
HFP–hν (also UV)c
DCM–hνd
1.47
1.34
1.41
—
1.35
1.49
1.35
4-NO2-Ph
4-MeO-Ph
4-MeO-Ph
4-F-Ph
4-NH2-Ph
4-MeO-naph
4-MeO-naph
Ph2C(OH)
Ph2C(OH)
3.44
4.47
DCM–hνh
HFP–LLE or hν
DCM–LLE
DCM–hνc
4.14
4.89
4.31
DCM–LLE c,l
DCM–hνd,h
DCM–LLE m
DCM–hνc
3.05
1.33
1.31
0.131
0.132
a
LLE means exposure of the sample tube to laboratory light for ca. 60 s. DCM = dichloromethane, AN = acetonitrile, HFP = 1,1,1,3,3,3-
hexafluoropropan-2-ol. b Coupling to two H (the second one to H-COO). c Weak signal. The signal from Bu 2NO also appeared (a = 1.58 mT).
e Doublet (the coupling to H-COO was 0.20 mT). f No further resolution due to D-COO was seen. g Lit.8b: 1.36, 0.17 mT; in benzene34 1.27–1.31,
0.17–0.20 mT; 1.36, 0.18 mT; in acetic acid 34 1.40, 0.20 mT. h Broad signal centred around g = 2.003 also seen. i Lit.,8b: 1.36, 0.15 mT. j Lit.8a
1.33, 0.14 mT. k Lit.17: 1.35, 0.15 mT; in benzene34: 1.26–1.29, 0.13–0.15 mT. l Lit.35 : 1.29, 0.14 mT. m Second 3 × 2 lines signal: aN = 1.29, aH = 0.104
mT; the assignment of the two signals might be the reverse one.
d
t
N
ؒ
The EPR-silent solution was then bubbled with dioxygen for
30 s, purged with argon and the EPR activity monitored. No
signal whatsoever was detected after this treatment.
Reaction of PBN, RCOOH and 2ϩ in laboratory light
Table 1 lists spin adducts obtained from the LLE (LLE = ‘labor-
atory light exposed’ being defined as ‘exposed to laboratory
light for р60 s during mixing of the sample’) reaction between
PBN and RCOOH (5–18) in the presence of 2ϩ (ca. 1 mmol
dmϪ3). Most reactions were performed in dichloromethane, but
other solvents were tried for specific purposes, for example
1,1,1,3,3,3-hexafluoropropan-2-ol (HFP) in which nucleophilic
reactivity is known to be strongly curtailed.16 In general, EPR
spectra with the expected hyperfine splitting (hfs) constants for
acyloxyl adducts of PBN were recorded (i.e. showing typically
low values of both aN and aH ) which in the appropriate cases
agreed with literature values.
It should be noted first that none of the LLE experiments in
HFP gave any paramagnetic signal from spin adducts, in line
with the strong attenuating effect exerted by HFP on the
reactivity of nucleophiles. Thus the mechanisms involving
nucleophilic attack [eqns. (3)–(4) and (7)] will be suppressed in
this solvent, as found previously.17 No such effect was noted
with dichloromethane, benzene or acetonitrile.
Redox reactivity of hydroxylamines
Literature reports on the electrochemistry of hydroxylamines
give widely differing redox potentials for the oxidation of
hydroxylamines under various conditions, as exemplified here
by N-phenylhydroxylamine. In aqueous solution at pH 13, the
ؒ
redox active species must be the anion, E1/2(PhNHO /
PhNHOϪ) being Ϫ0.48 V vs. SCE at the Hg anode.13a Similarly,
what was denoted ‘Eox(AϪ)’ = Ϫ0.75 V was determined by cyclic
voltammetry at a Pt anode in dimethyl sulfoxide-Et4NPF6;13b
Ϫ
ؒ
this again is Epa(PhNHO /PhNHO ). On the other hand, the
peak potential measured by cyclic voltammetry of PhNHOH in
acetonitrile–NaClO4 at a glassy carbon electrode must refer
to Epa(PhNHOH ϩ/PhNHOH), a value of 0.45 V being
᝽
reported.14 Addition of a base, pyridine, gave an extra peak at a
lower potential, 0.15 V, ascribed to oxidation of a ‘partially
ionized’ hydroxylamine (pKa of RNHOH estimated to be 12–
13, see ref. 15).
Formic acid 5 gave a previously unknown spin adduct with a
1:2:1 triplet splitting due to two hydrogens with identical hfs
constants, one from the α-H of PBN and the second from
HCOO. The triplet structure persisted in both benzene and
acetonitrile. With D-COOD (6) a doublet was seen since the
coupling to D-COO could not be resolved.
In view of these discrepancies, we have determined
ϩ
Ϫ
ؒ
Epa(PhNHOH /PhNHOH) and Epa(PhNHO /PhNHO ) in
᝽
acetonitrile–Bu4NBF4 by cyclic voltammetry at a Pt anode, and
obtained 0.61 and Ϫ0.75 V, respectively. The base used in the
latter measurement was butyllithium (in deficit with respect to
PhNHOH).
Acetic acid (7) gave a strong signal, as already mentioned.
J. Chem. Soc., Perkin Trans. 2, 1997
1691

View Article Online
Table 2 Hfs constants for spin adducts formed in the reaction between
PBN and ROH (19–22) in dichloromethane in the presence of 2ϩ BF4
Ϫ
under LLE conditions or with irradiation by light of λ > 400 nm. For
pK values, see ref. 33
No. R in ROH
pK of ROH Conditions aN /mT aH /mT
19
Ph2CH
15.4
LLE a
1.35
0.156
(PhCH2OH)
19
20
20
21
21
22
22
Ph2CH
Ph2CCO2CH3
Ph2CCO2CH3
Et
hνb,c,d
LLE a
hνe
—
—
—
1.32
1.32
1.401
1.399
1.44
—
0.16
0.16
0.225
0.227
0.21
—
15.9
19.2
LLE a,f
hν
Et
But
But
LLE a,g
hνb,c,d
a
t
N
Weak signal. b The signal from Bu 2NO also appeared (a = 1.58 mT).
ؒ
c Broad signal centred around g = 2.003 also seen. d The 3 × 2 lines
signal disappeared while the signals referred to in
appeared. e The
b,d
3 × 2 signal increased in intensity for a short period and then decayed
upon prolonged irradiation. f Lit.,34 in ethanol: 1.44, 0.26 mT. g Lit.,34
in benzene: 1.44, 0.19 mT.
ؒ
Both stronger (trifluoro- and trichloro-acetic acid 10 and 11)
and weaker aliphatic acids (dimethyl- and trimethyl-acetic
acid 8 and 9) gave signals assigned to the corresponding acyl-
oxyl adducts. The signal from 11 was shortlived and only lasted
for ca. 5 min after mixing. Benzoic acid 12 and 4-fluorobenzoic
acid 15 reacted smoothly, whereas 4-nitro- (13), 4-methoxy- (14)
and 4-aminobenzoic acid (16) did not react. Benzilic acid (18),
corresponding to the least stable acyloxyl radical known [eqn.
(1)],1f reacted with formation of two spin adducts, of which one
was assigned to the acyloxyl type and the second one to the
alkoxyl type. Separately, it was shown that a few alcohols (19–
22), including methyl benzilate 20, undergo the same type of
reaction (Table 2) under similar conditions.
Fig. 3 Time development of the EPR signal of PhCOO–PBN in the
reaction of dibenzoyl peroxide (0.135 mol dmϪ3) and PBN (0.10 mol
dmϪ3) in dichloromethane at 22 ЊC. Triangles: no additive; circles:
PhCOOH (0.064 mol dmϪ3) added.
ؒ
Formation of PhCOO–PBN by the spontaneous reaction
between PBN and dibenzoyl peroxide
The reaction between dibenzoyl peroxide and PBN is known to
give PhCOO–PBN in a thermal reaction, assumed to proceed
by homolysis of the O᎐O bond of the peroxide and trapping of
the benzoyloxyl radical by PBN.4 However, since dibenzoyl
peroxide is a weak electron transfer (ET) oxidant {it has a ther-
ؒ
Ϫ
ؒ
mochemically estimated EЊ[(PhCOO)2/(PhCOO PhCOO )] of
0.6 V in water and Ϫ0.2 V in a solvent like acetonitrile;19
a
Photochemical reaction of PBN, RCOOH and 2ϩ
kinetic study of the reaction between dibenzoyl peroxide
and a series of hydroquinones in acetonitrile,20 in combination
Deliberate irradiation with light of λ > 400 nm to ensure that
only 2ϩ was excited (λmax in acetonitrile 422 nm,11 in HFP 408
ؒ
with the Marcus theory, gave EЊ[(PhCOO)2/(PhCOO
ؒ
PhCOOϪ)] = Ϫ0.17 V; an irreversible reduction potential of
Ϫ0.1 V in benzene-methanol has been determined;21 see also
below} and is likely to contain and/or form small amounts of
benzoic acid or peroxybenzoic acid upon dissolution, it cannot
be excluded that the HA catalysed mechanism [eqns. (3)–(5),
with HA = PhCOOH and/or PhCO3H and Ox = (PhCOO)2] is
nm) produced stronger signals of RCOO–PBN from reactions
performed in dichloromethane, acetonitrile or benzene, typic-
ally by at least one order of magnitude. Sometimes, other sig-
nals were detectable, such as the persistent triplet from
t
Bu 2NO , a known product from the photolysis of PBN,12 and a
ؒ
broad unresolved feature centred around g = 2.003 which pre-
sumably was due to 2 ,18 one of the products of the redox pro-
ؒ
ؒ
the pathway for formation of PhCOO–PBN .
cess initiated by the photochemical reaction [eqns. (9)–(10)]. It
ؒ
Fig. 3 shows the result from runs in which [PhCOO–PBN ]
was monitored in a solution of dibenzoyl peroxide (0.135 mol
dmϪ3) and PBN (0.10 mol dmϪ3) without any additive (tri-
angles) or with benzoic acid present (0.064 mol dmϪ3; circles).
The addition of benzoic acid in a sizeable concentration
increased the initial rate of the reaction, but only by ca. 50%.
When a similar reaction was carried out in HFP, no PhCOO–
hν >400 nm
2ϩ
2ϩ*
(9)
ϩ
ϩ
ؒ
2 * ϩ PBN → 2 ϩ PBN
(10)
᝽
ؒ
should be noted that RCOO–PBN was formed by photo-
chemical activation also in HFP, presumably an effect of the
ؒ
PBN was detected during the same period as the reactions
higher rate of generation of PBN ϩ under these conditions. It is
shown in Fig. 3.
The PhCOO–PBN signal was calibrated against a solution
᝽
ؒ
ؒ
likely that RCOO–PBN originates from the reaction between
PBN ϩ and RCOOH [eqn. (11)]. In another context,17 we have
of a stable aminoxyl radical, 2,2,6,6-tetramethylpiperidin-1-
oxyl, TEMPO,4 establishing that one unit (Int) on the spectral
intensity axis of Fig. 1 corresponds to a radical concentration
᝽
ϩ
᝽
ϩ
ؒ
PBN ϩ RCOOH → RCOO–PBN ϩ H
(11)
of 7 × 10Ϫ10 mol dmϪ3. Thus the initial rates of PhCOO–
Ϫ10
found that PBN ϩ does react with certain nucleophiles in HFP,
especially neutral ones which are not so strongly solvated as are
anions by HFP.
The increased rate of production of RCOO–PBN in the
HFP–hυ reaction made possible the resolution of the coupling
PBN formation could be estimated at 4 × 10 mol dmϪ3 sϪ1
ؒ
᝽
(‘uncatalysed’ run; triangles) and 6 × 10Ϫ10 mol dmϪ3 sϪ1
(‘catalysed’ run; circles).
ؒ
In order to get a measure of the reactivity of a typical
aminoxyl radical toward dibenzoyl peroxide, the EPR signal
from a solution of TEMPO (0.27 mmol dmϪ3), in dichlo-
romethane in the presence of an excess of dibenzoyl peroxide
(200 mmol dmϪ3) was monitored. The rate constant of disap-
pearance of the TEMPO signal was 9.1(2) × 10Ϫ3 minϪ1, cor-
responding to a second-order rate constant of 7.7 × 10Ϫ4 dm3
ؒ
to the two hydrogens in HCOO–PBN , the coupling constant to
the formyl hydrogen being increased from 0.14 to 0.20 mT due
to the high polarity of HFP.17
ؒ
The CH3COO–PBN signal built up to a fairly high level
during photolysis. When the light was turned off, the signal
decayed with a half-life of ca. 1 min.
molϪ1 sϪ1
.
1692
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
the presence of 2ϩ did not give any spin adduct, nor did irradi-
Table 3 Formation of radical cations from the thermal reaction
between dibenzoyl peroxide (0.01–0.02 mol dmϪ3) and ArH (0.04–0.05
mol dmϪ3) in HFP at 22 ЊC
ation with light of λ > 400 nm. Irradiation of DMPO and
PhCOOH in dichloromethane with chloranil (4) present gave
an aminoxyl species with the expected EPR spectrum
[aN = 1.224, aH = 0.963, aH = 0.087 (2 H) mT, obtained from
treatment of DMPO with dibenzoyl peroxide in benzene; found
here: see Table 4]. In fact, further resolution of the spectrum
was possible, the couplings to both the two 3-hydrogens (0.078,
0.111 mT and the six methyl hydrogens (0.018 mT) being
obtained (Fig. 4). A similar experiment with C6D5COOH gave
an identical spectrum.
Similarly, acetic acid and DMPO gave rise to an EPR spec-
trum of similarly high resolution (Table 4), the assignment
being supported by the observation of an identical spectrum
from CD3COOD. Formic acid gave a signal with a spectrum
with only the couplings to the 3-hydrogens resolved, in addition
to that of the formyl hydrogen. The latter disappeared when
DCOOD was employed. Trimethyl- and trifluoro-acetic acid
also gave acyloxyl adducts by this procedure.
ϩ
/
EЊ(ArH
ArH)/V
vs. SCE
Radical
cation
formed
᝽
ArH
Comment
ϩ
4-ButC6H4NMe2
Ph3N
2,3-Me2-1,4-(MeO)2-
Benzene
3,3Ј,4,4Ј-Me4-1,1Ј-Bi- 1.36
naphthalene
0.65
0.92
1.18
ArH
Strong signal
Strong signal
Strong signal
᝽
ϩ
Ar-Ar
᝽
ϩ
ϩ
ϩ
ArH
ArH
ArH
᝽
᝽
᝽
Weak signal
Strong signal
Dibenzo-1,4-dioxine
Pentamethylanisole
1.37
1.44
No signal
ArH ϩ detected
᝽
upon UV
irradiation
ؒ
Formation of PhCOO-DMPO by the spontaneous reaction
between DMPO and dibenzoyl peroxide
For spectral comparison, the thermal reaction between DMPO
and dibenzoyl peroxide in dichloromethane was investigated.
The same EPR spectrum as in Fig. 4 was obtained, confirming
that the small couplings involve the 5,5-methyl groups.
Discussion
Formation of acyloxyl spin adducts under laboratory light
exposure
ؒ
Regardless of structure and decarboxylation rate of RCOO , a
carboxylic acid RCOOH reacts with PBN and 2ϩ in dichlo-
romethane, benzene or acetonitrile under LLE conditions to
ؒ
give an acyloxyl adduct RCOO–PBN (Table 1). As shown in
the experiments with PBN–acetic acid-2ϩ which were carried
out in complete darkness, the formation of RCOO–PBN is
ؒ
ؒ
Fig.
4
EPR spectrum of PhCOO-DMPO . (a) Generated by
photolysis of DMPO (0.10 mol dmϪ3), PhCOOH (0.05 mol dmϪ3) and
dependent on the photooxidation of PBN even during short
exposure to the relatively weak laboratory light. The reaction is
chloranil (0.03 mol dmϪ3) in dichloromethane. The light was filtered
(λ > 400 nm). The singlet in the middle is from Cl4Q Ϫ. (b) Generated
᝽
assumed to proceed via ET oxidation of PBN by excited 2ϩ
by the thermal reaction between DMPO (0.10 mol dmϪ3) and
dibenzoyl peroxide (0.05 mol dmϪ3) in dichloromethane.
ϩ
[eqns. (9)–(10)] and reaction of PBN
with RCOOH [eqn.
᝽
(11)].17 Irradiation by light of λ > 400 nm intensified the signals
and thus also made possible the generation of RCOO–PBN in
ؒ
One-electron oxidizing properties of dibenzoyl peroxide
HFP. We note that a neutral nucleophile like RCOOH retains
some reactivity in HFP, similarly to other cases studied
previously.17
A classical problem in ET chemistry is the mechanism of the
one-electron oxidation of easily oxidizable compounds, such as
N,N-dimethylaniline, by dibenzoyl peroxide or other diacyl
peroxides.19,20,22 The strongly increased stability of radical
cations in HFP,17 in part caused by the vastly attenuated reactiv-
ity of nucleophiles due to strong hydrogen bonding, makes pos-
sible a semiquantitative estimate of the oxidizing power of
dibenzoyl peroxide, as done previously for TlIII oxidants,23
bromine,24 iodine chloride,24 N-fluorodibenzenesulfonamide,25
nitrogen dioxide26 and chlorotricyanomethane.27
The failure of DMPO to give acyloxyl adducts upon LLE
treatment with RCOOH and 2ϩ or chloranil can have several
explanations. We suggest that the most likely one is that the
activity of light under LLE conditions is not sufficient to create
ؒ
a detectable concentration of RCOO–DMPO due to the lower
redox reactivity of DMPO compared to PBN (∆Epa = 0.2 V).
ؒ
Moreover, there are indications that RCOO–DMPO is more
ؒ
reactive toward further oxidation than RCOO–PBN . Ami-
A series of aromatic compounds ArH was allowed to react
with dibenzoyl peroxide [eqn. (12)] and the development of the
ϩ
ؒ
noxyl radicals have EЊ(R2N᎐O /R2N᎐O ) in the region of 0.6–
0.8 V (SCE)28 and an important pathway for their disappear-
ance should be ET oxidation. As shown 29 previously for the
case of HA = benzotriazole, DMPO reacts much faster than
PBN according to the Forrester–Hepburn mechanism [eqns.
(3)–(5), with tetrabutylammonium 12-tungstocobalt()ate as
the oxidant], but the N-benzotriazolyl adduct of DMPO disap-
pears more rapidly.
primary radical cation ArH ϩ (or possibly a secondary radical
᝽
cation) was monitored by EPR spectroscopy. The benzoate ion
formed in eqn. (12) is virtually unreactive and thus normally
HFP
ϩ
Ϫ
ؒ
(PhCOO)2 ϩ ArH
ArH ϩ PhCOO ϩ PhCOO
(12)
᝽
ϩ
reactive ArH will persist. Table 3 shows the results of such
᝽
It should be noted that there does not seem to be any correl-
ation between the pK of the acid and its ability to form spin
adducts according to eqns. (9)–(11). Tables 1,2 and 4 list acids
in the pK range of 0–19, with no obvious difference between
the extremes except that the spin adducts of strong acids are
more short-lived, not surprising in view of the stronger leaving
group tendencies of anions corresponding to strong acids.30
experiments; the limit of EЊ(ArH ϩ/ArH) for which it was just
᝽
ϩ
possible to observe an EPR spectrum of ArH was ca. 1.4 V
᝽
(SCE).
Photochemical reaction of DMPO, RCOOH and chloranil
The LLE reaction between DMPO (3, 4,5-dihydro-5,5-
dimethylpyrrole 1-oxide) and PhCOOH in dichloromethane in
J. Chem. Soc., Perkin Trans. 2, 1997
1693

View Article Online
Table 4 Hfs constants for spin adducts formed in the reaction between
DMPO and HA in dichloromethane by photolysis together with
chloranil (light of λ > 400 nm)
The thermal reaction between PBN and dibenzoyl peroxide
The thermal reaction between PBN and dibenzoyl peroxide to
ؒ
give PhCOO–PBN , assuming 100% trapping efficiency and
ؒ
inertness of PhCOO–PBN under the reaction conditions, has
Other hfs
constants/mT
aN /mT
1.26
aH /mT
1.03
been investigated kinetically by an EPR spectral study in ben-
HA
zene at 38–48 ЊC and it was concluded that the spin adduct was
ؒ
CH3COOH (7)
CD3COOD
HCOOH (5)
0.018 (3 H), 0.036 (3H),
0.109 (1H)
formed by trapping of PhCOO formed by homolytic decom-
position of dibenzoyl peroxide [eqns. (14) and (15)].4 At these
1.26
1.03
0.018 (3 H), 0.036 (3 H),
0.109 (1 H)
ؒ
(PhCOO)2 → 2 PhCOO
(14)
(15)
1.26
1.01
0.074 (1 H), 0.097 (1 H),
0.190 (1 H)
0.074 (1 H), 0.097 (1 H)a
0.080 (1 H), 0.105 (1 H)
ؒ
ؒ
DCOOD (6)
1.26
1.27
1.40
1.26
1.01
0.94
1.24
0.97
PhCOO ϩ PBN → PhCOO–PBN
Me3CCOOH (9)b
CF3COOH (10)
PhCOOH (12)
temperatures, the half-life of the reaction falls in the range of
5000–1000 h.
0.078 (1 H), 0.114 (1 H),
0.018 (6 H)c
The Forrester–Hepburn mechanism of eqns. (3)–(5), applied
to the reaction between PBN and dibenzoyl peroxide with
adventitiously formed PhCOOH as the catalyst HA, takes the
form of eqns. (16) and (17). It should be noted that the putative
C6D5COOH
1.27
1.26
0.98
0.97
0.084 (1 H), 0.120 (1 H),
0.017 (6 H)
(PhCOO)2
0.111, 0.081, 0.018 (6 H)c
d
a
c
Coupling to DCOO not resolved. b Weak signal. Lit.,34 (in PhH):
1.22; 0.96; 0.087 (2 H). d Thermal reaction.
PhCOOH ϩ PBN → PhCOO–PBN(H)
(16)
2 PhCOO–PBN(H) ϩ (PhCOO)2 →
Formation of acyloxyl spin adducts under irradiation
Photochemical activation of 2ϩ leads to the triplet state T2ϩ*
ؒ
2 PhCOO–PBN ϩ 2 PhCOOH (17)
T
ϩ
ؒ
which is a one-electron oxidant with EЊ( 2 */2 ) = 1.8 V vs.
SCE, capable of rapidly removing an electron from PBN with
EЊ(PBN ϩ/PBN) = 1.49 V with formation of PBN ϩ. The latter
ET oxidation of the hydroxylamine intermediate by dibenzoyl
ؒ
᝽
᝽
peroxide gives a molecule of RCOO which is assumed to be
reacts with nucleophiles to give spin adducts, as exemplified by
eqn. (7). In this particular case, neutral carboxylic acids
RCOOH act as nucleophiles [eqn. (11)].
trapped with 100% efficiency in eqn. (15). Thus each peroxide
molecule gives rise to two molecules of PhCOO–PBN , as in the
ؒ
homolytic mechanism of eqns. (14) and (15).
The reaction between T2ϩ* and DMPO-RCOOH did not give
Various literature reports19–21 place the redox potential of the
dissociative electron transfer of dibenzoyl peroxide,
ؒ
any EPR-detectable concentration of RCOO–DMPO . This
might be due to a less efficient production of DMPO ϩ due to
Ϫ
ؒ
᝽
EЊ[(PhCOO)2/(PhCOO PhCOO )], at ca. Ϫ0.2 V (SCE) in a
dipolar aprotic solvent like acetonitrile or dichloromethane.
Our study in HFP [eqn. (12), Table 3], a strongly polar solvent,
showed that dibenzoyl peroxide can oxidize ArH with redox
the higher oxidation potential of DMPO, and/or a lower
ؒ
stability of RCOO–DMPO , as already discussed above. With
a stronger and more efficient oxidant, the excited state of
chloranil (4*), the reaction works well.
potentials up to 1.4 V to give persistent ArH ϩ. By analogy
᝽
with the redox properties of Br2 under similar conditions,24 one
Ϫ
ؒ
Thermal reaction of CH3COOH–PBN–oxidant under
completely dark conditions
would then put EЊ[(PhCOO)2/(PhCOO PhCOO )] in HFP at
ca. 0.4 V which is compatible with the value of Ϫ0.2 V above,
considering the large difference in solvent polarity. As discussed
above, the redox potential of Ϫ0.2 V is too low for the ET
oxidation of PhCOO–PBN(H) according to eqn. (17), as
judged from the results with 2ϩ or 4 as oxidants. It is thus
understandable why no catalysis by benzoic acid of the
Forrester–Hepburn mechanism can be detected in the
dibenzoyl peroxide–PBN system.
As shown above, the pyrylium ion 2ϩ is remarkably efficient as a
ؒ
photoelectron oxidant and thus gave RCOO–PBN also upon
short exposure to laboratory light. In order to elucidate the
scope of the Forrester–Hepburn mechanism [eqns. (3)–(5)],
experiments were performed under completely darkened condi-
tions. It then became clear that 2ϩ is not a thermal oxidant in
eqn. (5), A = CH3COO, nor indeed is chloranil 4. Only with
ؒ
bromine as the oxidant was CH3COO–PBN obtained by the
The reaction between PBN and (PhCOO)2 in dichlorometh-
ane proceeded thermally at 22 oC, the initial rate of production
thermal reaction, whereas the even stronger oxidant, DDQ,
ؒ
of PhCOO–PBN being 4 × 10Ϫ10 mol dmϪ3 sϪ1. A rate constant
ؒ
presumably gave CH3COO–PBN in a fast reaction but equally
of 3.5 × 10Ϫ9 sϪ1 for the homolytic decomposition of (PhCOO)2
at 22 ЊC was obtained by extrapolation of data from the tem-
perature range of 50–80 ЊC; strictly, these data were from the
decomposition in benzene,4 but the variation of rate of homo-
lytic bond cleavage between different solvents is generally
small.31 At [(PhCOO)2] = 0.14 mol dmϪ3, this rate constant
rapidly destroyed it, the net result being the appearance of a
Ϫ
very strong signal of DDQ
.
᝽
These results can be understood in terms of the redox chem-
istry of hydroxylamines13,14 and aminoxyl radicals,28 as sum-
marized in eqn. (13). Under the conditions normally employed
Epa ca. 0.6 V
Epa = 0.6–0.8 V
would correspond to an initial rate of production of [PhCOO–
ؒ
RCOO–PBN(H)
RCOO–PBN
PBN ] of 2 × 0.14 × 3.5 × 10Ϫ9 = ca. 1.0 × 10Ϫ9 mol dmϪ3 sϪ1
,
ؒ
RCOO–PBNϩ (13)
ؒ
assuming that the trapping efficiency of PhCOO is 100% and
ؒ
PhCOO–PBN does not undergo further reactions. With these
for generating spin adducts via the Forrester–Hepburn mechan-
ism, hydroxylamines exist as such and are oxidized at Epa
ca. 0.6 V. This redox potential decreases under basic condi-
tions, reaching a limit of ca. Ϫ0.7 V when the hydroxylamine
proton has been fully removed. When the redox potential of the
oxidant reaches 0.1 V (bromine) the thermal oxidation of the
hydroxylamine becomes feasible; at the same time the further
oxidation of the aminoxyl radical becomes fast. With DDQ
(0.52 V) both reactions will be fast and no spin adduct will be
detected.
assumptions and the experimental uncertainty of the cali-
bration of [PhCOO–PBN ], the experimental initial rate can be
explained as being due entirely to the homolytic decomposition
of dibenzoyl peroxide [eqns. (14) and (15), as was found previ-
ously with benzene as solvent.
ؒ
However, if a fast mechanism for the disappearance of
ؒ
PhCOO–PBN might exist, the conclusion may change. In
order to get a measure of the reactivity of an aminoxyl radical,
ϩ
ؒ
the thermal reaction of TEMPO [EЊ(R2NO /R2NO ) = 0.61 V
(SCE)] with dibenzoyl peroxide was used as a model. Most
1694
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
likely this reaction proceeds according to an ET mechanism
< 10Ϫ18 dm3 molϪ1 Ϫ1. Yet a puzzling observation remains to
s
[eqn. (18)]. It proceeded with a second-order rate constant of
be explained: why is the thermal decomposition of dibenzoyl
ؒ
peroxide not producing PhCOO–PBN in HFP? In itself, the
ؒ
ؒ
photochemical generation of PhCOO–PBN in HFP is feasible,
R2NO ϩ (PhCOO)2 →
both under the conditions given in Table 1 and by photolysis of
Ϫ
R2NOϩ ϩ PhCOO ϩ PhCOO
(18)
ؒ
(PhCOO)2.17
7.7 × 10Ϫ4 dm3 molϪ1
ical formed in eqn. (18) is not trapped by TEMPO [eqn. (19)];
s
Ϫ1, assuming that the benzoyloxyl rad-
The role of dioxygen in the Forrester–Hepburn mechanism
Finally we need to address the problem of dioxygen acting as an
oxidant in the Forrester–Hepburn mechanism, in spite of the
fact that EЊ(O2/O2 Ϫ) is Ϫ0.4 V in water or even lower in
ؒ
ؒ
R2NO ϩ PhCOO → R2N(O)OCOPh
(19)
᝽
organic solvents. Dioxygen has been implied as the oxidant in
ؒ
eqn. (5) not only in the formation of RCOO–PBN but also in
with 100% trapping efficiency, the rate constant should be
halved to 3.9 × 10Ϫ4 dm3 molϪ1 sϪ1
12
29
ؒ
ؒ
the formation of F-PBN and N-heteroaryl-PBN , and it is
therefore of interest to find an explanation for this effect which
does not involve its redox reactivity. As shown in Fig. 2, the
.
Application of the Marcus theory, using the redox potentials
quoted above (0.61, Ϫ0.2 V), a reorganization energy of 40 kcal
molϪ1 (1 cal = 4.184 J) and a distance between the reacting
centres of 6 Å in the transition state provides a calibration of
this experimental rate constant. The calculation gives log
kET = Ϫ2.0, to be compared with the experimental value of
Ϫ3.1 or Ϫ3.4. Considering the uncertainty in the dibenzoyl
peroxide redox potential, this difference is hardly significant (it
corresponds to a potential change of ca. 0.06 V).
ؒ
concentration of CH3COO–PBN was rather low after oxid-
ation by dioxygen. Once the signal had disappeared, it was not
re-formed after a short treatment by dioxygen.
The latter behaviour is exactly what one would expect from
its redox potential, and we therefore have considered an altern-
ative explanation in terms of some credible dioxygen-dependent
low-level impurity in PBN. One might for example envisage a
very slow, possibly light-promoted reaction between solid PBN
and dioxygen, taking place during storage and leading to the
formation of a cycloadduct 23 [eqn. (24)].32 Attack by a
Aminoxyl radicals undergo oxidation with reversible poten-
tials in a rather narrow range, 0.6–0.8 V (SCE). Among spin
ؒ
adducts of PBN, only Ph-PBN (0.7 V) appears to have been
nucleophile at the ring carbon of 23 and loss of O2 Ϫ from the
investigated.28a The introduction of the electron-withdrawing
᝽
ؒ
intermediate 24 [eqn. (25)] would produce a spin adduct. If this
PhCOO group in PhCOO–PBN is expected to decrease the
reactivity toward ET oxidation, and it is therefore safe to
ؒ
assume that PhCOO–PBN will react more slowly with diben-
O–
O
O
zoyl peroxide than TEMPO. Thus, assuming for a moment that
the same rate constant applies to the reaction between PhCOO–
PBN and dibenzoyl peroxide as for TEMPO, the initial rate of
+
O2
(24)
N
O
C
H
But
Ph
ؒ
N
But
ؒ
disappearance of the PhCOO–PBN signal at [(PhCOO)2] =
0.14 mol dmϪ3 and [PhCOO–PBN ] = 10 mol dmϪ3 (which is
Ϫ6
ؒ
23
ؒ
the estimated concentration of PhCOO–PBN after ca. 20 min
–O
O
N
in the uncatalysed reaction of Fig. 1) would be (0.14 × 10Ϫ6
×
O
O•
• –
7.7 × 10Ϫ4) ≈ 1 × 10Ϫ10 mol dmϪ3 Ϫ1, which is ca. four times
s
–O2
Ph CH
Nu
Ph CH
Nu
N
23
+
Nu–
(25)
slower than the calculated initial rate of dibenzoyl peroxide
decomposition. Thus, from considerations based upon the
TEMPO–(PhCOO)2 reaction, it seems that the rate of form-
But
But
24
ؒ
ation of PhCOO–PBN would be controlled by the rate
of homolytic decomposition of dibenzoyl peroxide also in
dichloromethane, as found previously for benzene.
reaction is 100% effective, the level of 23 in PBN would need to
be of the order of 0.001% to produce the observed concen-
tration of spin adduct.
Unfortunately, the rate of the reaction between PhCOO–
ؒ
PBN and (PhCOO)2 cannot be measured directly, since
ؒ
each molecule of PhCOO–PBN oxidized gives rise to a new
ؒ
molecule of PhCOO [eqns. (20)–(22)]. Thus the net rate of
Experimental
Materials
ؒ
PhCOO–PBN ϩ (PhCOO)2 →
PBN (1) and DMPO (3) were purchased from Aldrich and used
PhCOO–PBNϩ ϩ PhCOO ϩ PhCOO
(20)
(21)
Ϫ
ؒ
as
received.
4-MeOPBN,
2,3-dimethyl-1,4-dimethoxy-
benzene, 3,3’,4,4’-tetramethyl-1,1’-binaphthalene, dibenzo-1,4-
dioxine and pentamethylanisole were available from earlier
studies.17,23,24 All other chemicals and solvents used were of
ؒ
ؒ
PhCOO ϩ PBN → PhCOO–PBN
highest commercial quality available. The sensitizer 2ϩ BF4
Ϫ
Net reaction:
was a gift from Professor E. Steckhan, University of Bonn.
PBN ϩ (PhCOO)2 → PhCOO–PBNϩ ϩ PhCOOϪ (22)
EPR spectral experiments
ؒ
These were performed as described previously.12,17,23 The ‘com-
pletely darkened’ environment had as the only source of light in
the laboratory a dimmed computer screen at a distance of 2 m
from the site of sample preparation. LLE conditions were
equivalent to day-light from a northern window in November–
February, 5 m from the site of sample separation, and one 30 W
neon light at 70 cm distance.
disappearance of PhCOO–PBN will only reflect a trapping
efficiency < 100% and not the desired reaction of eqn. (20).
Thus again it seems that the thermal decomposition of
dibenzoyl peroxide can account for the formation of PhCOO–
PBN . An ET mechanism involving PBN and dibenzoyl per-
oxide [eqn. (23), followed by eqn. (7), R = Ph] can be ruled out
ؒ
(PhCOO)2 ϩ PBN →
PhCOOϪ ϩ PhCOO ϩ PBN
(23)
ϩ
ؒ
Cyclic voltammetry
᝽
This was carried out in acetonitrile–Bu4NPF6 at a Pt anode,
due to its high endergonicity (ca. 1.7 eV), corresponding to kET
using the BAS-100 instrument. The sweep rate was 100 mV sϪ1
.
J. Chem. Soc., Perkin Trans. 2, 1997
1695

View Article Online
15 P. A. S. Smith, The Chemistry of Open-Chain Organic Nitrogen
Acknowledgements
Compounds, Benjamin, New York, 1966, vol. 2, p. 3.
16 L. Eberson, M. P. Hartshorn, O. Persson and F. Radner, Chem.
Commun., 1996, 2105.
Financial support from the Swedish Natural Science Research
Council and the Knut and Alice Wallenberg Foundation is
gratefully acknowledged. We thank Professor E. Steckhan,
17 L. Eberson, M. P. Hartshorn and O. Persson, J. Chem. Soc., Perkin
Trans. 2, 1996, 141.
University of Bonn, for a gift of 2ϩ BF4
Ϫ.
ؒ
18 The EPR spectrum of 2 has been denoted ‘broad structureless’:
H. Kawata and S. Niizuma, Bull. Chem. Soc. Jpn., 1989, 62, 2279.
19 L. Eberson, Chem. Scr., 1982, 20, 39.
20 W. Adam and A. Schönberger, Chem. Ber., 1992, 125, 2149.
21 E. J. Kuta and F. W. Quackenbush, Anal. Chem., 1960, 32, 1069;
V. L. Antonovskii, Z. S. Frolova, T. T. Shleina and M. M.
Buglanova, Zh. Obshch. Khim., 1969, 39, 368.
22 A short summary is given in ref. 7, pp. 138–141. See also: G. B.
Schuster, Acc. Chem. Res., 1979, 12, 366; S. Srinivas and K. G.
Taylor, J. Org. Chem., 1990, 55, 1779; X.-K. Jiang, C.-X. Zhao and
Y.-F. Gong, J. Phys. Org. Chem., 1991, 4, 1; L. Grossi, Res. Chem.
Intermed., 1993, 19, 697.
References
1 (a) J. Chateauneuf, J. Lusztyk and K. U. Ingold, J. Am. Chem. Soc.,
1988, 110, 2877, 2886; (b) J. Chateauneuf, J. Lusztyk, B. Maillard
and K. U. Ingold, J. Am. Chem. Soc., 1988, 110, 6727; (c)
H. G-Korth, J. Chateauneuf, J. Lusztyk and K. U. Ingold, J. Org.
Chem., 1991, 56, 2405; (d) D. E. Falvey and G. B. Schuster, J. Am.
Chem. Soc., 1986, 108, 7419; (e) J. W. Hillborn and J. A. Pincock,
J. Am. Chem. Soc., 1991, 113, 2683; ( f ) T. M. Bockman, S. M.
Hubig and J. K. Kochi, J. Am. Chem. Soc., 1996, 118, 4502.
2 E. G. Janzen and D. L. Haire, in Advances in Free Radical Chem-
istry, Vol. 1, ed., D. D. Tanner, JAI Press, London, 1990, p. 253.
3 W. A. Pryor, C. K. Govindan and D. F. Church, J. Am. Chem. Soc.,
1982, 104, 7563.
23 L. Eberson, M. P. Hartshorn and O. Persson, J. Chem. Soc., Perkin
Trans. 2, 1995, 1735.
24 L. Eberson, M. P. Hartshorn, F. Radner and O. Persson, Chem.
Commun., 1996, 215.
25 L. Eberson and O. Persson, J. Chem. Soc., Perkin Trans. 2, 1997, in
press.
4 E. G. Janzen, C. A. Evans and Y. Nishi, J. Am. Chem. Soc., 1972, 94,
8236.
26 L. Eberson, M. P. Hartshorn, O. Persson and F. Radner, Res. Chem.
Intermediat., 1996, 22, 799.
5 E. G. Janzen and B. J. Blackburn, J. Am. Chem. Soc., 1969, 91, 4481.
6 A. R. Forrester and S. P. Hepburn, J. Chem. Soc. (C), 1971, 701.
7 L. Eberson, Electron Transfer Reactions in Organic Chemistry,
Springer-Verlag, Heidelberg, 1987.
8 (a) L. Eberson and M. Nilsson, Acta Chem. Scand, 1993, 47, 1129;
(b) L. Eberson, J. Chem. Soc., Perkin Trans. 2, 1992, 1907.
9 L. Eberson, J. J. McCullough and O. Persson, J. Chem. Soc., Perkin
Trans. 2, 1997, 133.
27 L. Eberson and J. J. McCullough, manuscript in preparation.
28 (a) A. J. Bard, J. C. Gilbert and R. D. Goodin J. Am. Chem. Soc.,
1974, 96, 620; (b) W. Sümmerman and U. Deffner, Tetrahedron,
1975, 31, 593; (c) J. M. Bobbitt and C. L. Flores, Heterocycles, 1988,
27, 509.
29 A. Alberti, P. Carloni, L. Eberson, L. Greci and P. Stipa, J. Chem.
Soc., Perkin Trans. 2, 1997, in press.
30 M. J. Davies, B. C. Gilbert, J. K. Stell and A. C. Whitwood, J. Chem.
Soc., Perkin Trans. 2, 1992, 333.
10 As demonstrated for HA = benzotriazole and RX = N-chloro-
benzotriazole, see: P. Carloni, L. Eberson, L. Greci and P. Stipa,
J. Chem. Soc., Perkin Trans. 2, 1996, 1297.
31 As found for the thermal decomposition of azoisobutyronitrile:
C. Walling, Free Radicals in Solution, Wiley, New York 1957.
32 Nitrones are substrates in 1,3-dipolar cycloadditions: P. N. Con-
falone and E. M. Huie, Org. React. (NY), 1988, 36, 1.
33 G. Kortüm, W. Vogel and K. Andrussow, Dissociation Constants
of Organic Acids in Aqueous Solution, Butterworths, London, 1961;
E. P. Sarjeant and B. Dempsey, Ionisation Constants of Organic
Acids in Aqueous Solution, Pergamon Press, Oxford, 1979.
34 G. R. Buettner, Free Radical Biol. Med., 1987, 3, 259.
35 L. Eberson, M. P. Hartshorn, F. Radner and J. O. Svensson, Acta
Chem. Scand., 1996, 50, 885.
11 M. Martiny, E. Steckhan and T. Esch, Chem. Ber., 1993, 126, 1671.
12 Earlier attempts to use TAP as sensitizer for the photooxidation of
spin traps in the presence of strong anionic nucleophiles were only
partially successful (L. Eberson, J. Chem. Soc., Perkin Trans. 2,
1994, 171) due to the competing reaction between TAPT and the
nucleophiles (for a review, see: K. Ohkata and K.-Y. Akiba, Adv.
Heterocycl. Chem., 1996, 65, 283).
13 (a) P. E. Iversen and H. Lund, Anal. Chem., 1969, 41, 1322; (b)
F. G. Bordwell and W.-Z. Liu, J. Am. Chem. Soc., 1996, 118, 8777;
the experimental description of these measurements can be found
in: F. G. Bordwell and M. J. Bausch, J. Am. Chem. Soc., 1986, 108,
8777.
Paper 7/01479A
Received 3rd March 1997
Accepted 18th March 1997
14 S. Ozaki and M. Masui, Chem. Pharm. Bull. (Tokyo), 1978, 26, 1364;
H. Sayo, S. Ozaki and M. Masui, Chem. Pharm. Bull. (Tokyo), 1973,
21, 1988.
1696
J. Chem. Soc., Perkin Trans. 2, 1997