Efficient homogeneous radical-anion chain reactions initiated by dissociative electron transfer to 3,3,6,6-tetraaryl-1,2-dioxanes

Article abstract of DOI:10.1002/chem.200902023

A series of 3,3,6,6-tetraaryl1,2-dioxanes (TADs) have been investigated at an inert electrode by using cyclic voltammetry, constant potential electrolysis and digital simulations. The series consists of the phenyl-substituted TAD (la), p-methoxy-aryl TADs (lb, Ic) and the p-methoxy/nitro-bearing TAD (Id). The heterogeneous electron-transfer (ET) reduction is dissociative, causing rupture of the oxygenoxygen bond, which generates a distonic radical-anion that reacts competitively either by β-scission fragmentation or ET Fragmentation of the distonic radical anion yields an alkene, a substituted benzophenone, and a benzophenone radical anion. The benzophenone radical-anion propagates an efficient homogeneous ET-fragmentation chain reaction that accounts for the potential dependence of the product ratios and the low charge consumption observed in the controlled potential electrolysis experiments. Digital simulation of the experimental cyclic voltammograms allowed for estimates of the rate constants of the heterogeneous ET to the 0-0 bond, and for the rate constants for the β-scission fragmentation of the distonic radical-anions. Density functional theory calculations corroborate the differences in the heterogeneous kinetics of the initial dissociative ET The endoperoxides 1a-1c react predominantly by a concerted dissociative ET mechanism, although the data suggests a stepwise dissociative pathway is also competitive. Bearing a nitro-aryl substituent, Id provides a rare example of an endoperoxide that proceeds by a stepwise dissociative ET mechanism. Irrespective of the initial mechanistic details, we find a propagating radicalanion cycle is a general mechanistic feature.

Full text of DOI:10.1002/chem.200902023

DOI: 10.1002/chem.200902023
Efficient Homogeneous Radical-Anion Chain Reactions Initiated by
Dissociative Electron Transfer to 3,3,6,6-Tetraaryl-1,2-dioxanes
Donald L. B. Stringle,[a] David C. Magri,[b] and Mark S. Workentin*[a]
Abstract: A series of 3,3,6,6-tetraaryl-
1,2-dioxanes (TADs) have been investi-
gated at an inert electrode by using
cyclic voltammetry, constant potential
electrolysis and digital simulations. The
series consists of the phenyl-substituted
TAD (1a), p-methoxy-aryl TADs (1b,
1c) and the p-methoxy/nitro-bearing
TAD (1d). The heterogeneous elec-
tron-transfer (ET) reduction is disso-
ciative, causing rupture of the oxygen–
oxygen bond, which generates a diston-
ic radical-anion that reacts competitive-
ly either by b-scission fragmentation or
ET. Fragmentation of the distonic radi-
cal anion yields an alkene, a substituted
benzophenone, and a benzophenone
radical anion. The benzophenone radi-
cal-anion propagates an efficient ho-
mogeneous ET–fragmentation chain
reaction that accounts for the potential
dependence of the product ratios and
the low charge consumption observed
in the controlled potential electrolysis
experiments. Digital simulation of the
experimental cyclic voltammograms al-
lowed for estimates of the rate con-
stants of the heterogeneous ET to the
for the b-scission fragmentation of the
distonic radical-anions. Density func-
tional theory calculations corroborate
the differences in the heterogeneous
kinetics of the initial dissociative ET.
The endoperoxides 1a–1c react pre-
dominantly by a concerted dissociative
ET mechanism, although the data sug-
gests a stepwise dissociative pathway is
also competitive. Bearing a nitro-aryl
substituent, 1d provides a rare example
of an endoperoxide that proceeds by a
stepwise dissociative ET mechanism.
Irrespective of the initial mechanistic
details, we find a propagating radical-
anion cycle is a general mechanistic
feature.
ꢀ
O O bond, and for the rate constants
Keywords: cyclic voltammetry
·
electron transfer · endoperoxides ·
fragmentation · radical ions
Introduction
workers has provided evidence that this assumption is not
necessarily valid.[3–6] They demonstrated by electrochemical-
ly generating a series of cycloalkylketone radical anions that
both charge and spin are governing factors in the reactivity
of radical-anion rearrangements. The radical anions in these
studies were observed to undergo a carbon–carbon bond
fragmentation yielding a single species containing a spatially
separated radical and anion, which is otherwise known as a
distonic radical anion [Eq. (1)]. Recently, we provided evi-
dence supporting these earlier findings, based on the frag-
mentation of distonic radical anions, which were shown to
yield a localized radical-anion and a neutral molecule
[Eq. (2)].[7,8]
Radical anions have long been recognized as an important
class of reactive intermediate in synthetic organic and bio-
organic chemistry.[1] They result from the transfer of a single
electron to a neutral molecule to yield an intermediate pos-
sessing both a charge and radical character.[2] A long-stand-
ing misconception has been that the reactivity of radical
anions, in particular for fragmentation and rearrangement
reactions, could be inferred from the reactivity of analogous
neutral radicals. However, a body of work by Tanko and co-
[a] Dr. D. L. B. Stringle, Prof. M. S. Workentin
Department of Chemistry
The University of Western Ontario
London, Ontario, N6A 5B7 (Canada)
Fax : (+1)519-661-3022
E-mail: mworkent@uwo.ca
[b] Dr. D. C. Magri
Department of Chemistry, Acadia University
Wolfville, Nova Scotia, B4P 2R6 (Canada)
Distonic radical anions can be generated by an electron-
transfer (ET) reaction to an endoperoxide.[9,10] These com-
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200902023.
178
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 178 – 188

FULL PAPER
ꢀ
pounds possess an oxygen–oxygen (O O) bond within a
electrochemically-generated radical-anion donor), and the
alkoxides protonated to yield diol [Eq. (4)].[18,19,23]
cyclic molecular framework. An ever increasing number of
endoperoxides, many with intriguing structures and signifi-
cant bioactivity, are being isolated from marine and terres-
trial sources on a routine basis.[11] A high-profile example is
the potent antimalarial drug, artemisinin.[12] The endoperox-
With the alkyl-substituted endoperoxides, the reactivity of
the distonic radical anion was limited to reduction followed
by protonation to yield the cis-diol in quantitative yield.[18,19]
The study of 9,10-diphenyl-9,10-epidioxyanthracene (DPA-
O2) provided a glimpse into other competing reactions avail-
able to the distonic radical-anion.[22] A 1,2-phenyl (O-neo-
phyl-type) rearrangement was observed to occur in competi-
tion with reduction of the oxygen-centered radical and the
follow-up chemistry depended on the reduction potential of
the electron donor or electrode.[23] The investigation of the
ET reduction of 3,3,6,6-tetraphenyl-1,2-dioxane (TAD; 1a)
provided another reaction pathway for the distonic radical
anion.[24] In our initial report, we proposed ET resulted in
ꢀ
O O bond cleavage generating a distonic radical anion that
reacts by a b-scission fragmentation in competition with ET
reduction of the oxygen-centered radical. We proposed the
intermediates generated by fragmentation were responsible
for initiating a homogeneous radical-anion chain reaction
that accounted for the uncommon features observed in the
cyclic voltammograms, notably an oxidative dip after the
dissociative wave, and the pronounced potential dependence
on the product ratios and low charge consumption values
from the electrolysis studies.
Recently, we reported two other examples of homogene-
ous propagating radical-anion chain reactions initiated by
dissociative ET reduction of diphenyl-substituted bicyclic en-
doperoxides.[7,8] Just as with 1a, the distonic radical anion
ides yingzhaosu C and 10,12-peroxycalamenene are unique
for in addition to displaying antimalarial activity, they con-
tain aromatic moieties.[13] Other endoperoxides with inter-
esting molecular structures include chondrillin and muqubi-
lin, which have been evaluated to process antitumour, anti-
microbial and antiviral properties.[14] A primary aim in the
study of many of these compounds has been with respect to
their isolation, synthesis, and bioactivity. While considerable
attention has been given to the study of artemisinin,[12] very
little attention has been given, in comparison, to the rela-
tionship between the molecular structure, mechanism of
action, and observed biological response of many other en-
doperoxides. An important consideration in the bioactivity
of many of these endoperoxides is if the mechanism could
be initiated by an outer-sphere dissociative-type ET reaction
ꢀ
resulting from reduction of the O O bond was observed to
undergo a competing b-scission fragmentation with direct
reduction from the electrode. Using various electrochemical
techniques, we evaluated previously unknown kinetic and
thermochemical data, delineated the reduction mechanism,
and provided insight into the fragmentation chemistry of
neutral biradicals and analogous distonic radical anions.[7,8]
In addition to expanding on our original study on the
electrochemical reduction of 1a,[24] we extend our discussion
to include three other derivatives: the methoxy-aryl-substi-
tuted TAD 1b, a bicyclic analogue 1c, and the nitro/me-
thoxy-aryl-substituted TAD 1d (Scheme 1). We report a
thorough study of the ET-initiated reduction by employing
cyclic voltammetry, constant potential electrolyses, DFT cal-
culations, and digital simulation of the experimental CVs in
order to obtain a better understanding of the ET mechanism
and to evaluate kinetic and thermodynamic information, in
particular, the rate constant for b-scission fragmentation of
the distonic radical anions. Collectively, the techniques em-
ployed allow for the evaluation of the influence aryl sub-
ꢀ
to the O O bond.
Our previous studies[7,8,15–25] and those of others[26–33] have
examined in detail the ET-initiated reduction mechanism of
many organic peroxides and endoperoxides. These studies
have provided valuable thermodynamic and kinetic informa-
tion using various electrochemical techniques and applica-
tion of Savꢁantꢂs theory of dissociative ET.[34,35] The reduc-
tion of alkyl-substituted endoperoxides was shown to occur
by a concerted dissociative ET mechanism followed by sub-
sequent reduction and protonation as in Equations (3) and
(4).[18,19] By this mechanism, ET is concomitant with O O
ꢀ
bond cleavage to form a distonic radical anion [Eq. (3)].
Subsequently, the alkoxyl radical portion is reduced, either
heterogeneously (by an electrode) or homogeneously (by an
Chem. Eur. J. 2010, 16, 178 – 188
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
179

M. S. Workentin et al.
Scheme 1. Synthesis of the 3,3,6,6-tetraaryl-1,2-dioxanes from the corre-
sponding alkene or diene (SbCl5 =antimony pentachloride, DCA=9,10-
dicyanoanthracene).
Figure 1. Cyclic voltammograms showing the reduction of a 2.2 mm solu-
tion of 1a in 0.1m TEAP/DMF at 0.1 Vsꢀ1: a) scanning to ꢀ1.55 V (solid)
and ꢀ1.95 V (dotted) and ꢀ2.45 V (dashed). The arrow highlights the ox-
idative dip (see text). b) scanning to 1.95 V and then repetitively cycling
between ꢀ1.70 and ꢀ1.95 V.
stituents have on the reactivity of the distonic radical anions
ꢀ
as well as the initial ET reduction of the O O bond in cyclic
peroxides. The phenyl and methoxyphenyl endoperoxides
1a–1c react mainly by a concerted dissociative ET mecha-
nism, although the data suggests a stepwise dissociative con-
tribution, where as 1d is a rare example of an endoperoxide
that undergoes a solely stepwise dissociative ET mecha-
nism.[25] Irrespective of the aryl substituent and the initial
mechanistic details, we find a propagating radical-anion
cycle is a general mechanistic feature.
Results and Discussion
Synthesis: The endoperoxides were synthesized by irradiat-
ing an oxygen-saturated solution of the corresponding dia-
rylethylene (tetraarylhexadiene for 1c) in dry dichlorome-
thane or acetonitrile in the presence of a light-activated oxi-
dizing agent, such as antimony pentachloride (SbCl5) or
9,10-dicyanoanthracene (DCA), according to known litera-
ture methods (Scheme 1).[36–39] The alkene or diene precur-
sors were commercially available (in the case of 1a) or syn-
thesized. In the final step of the syntheses, the endoperox-
ides precipitated out of solution and were purified by either
recrystallization or column chromatography. Full experimen-
tal and characterization data are provided in the Supporting
Information.
Figure 2. Representative CVs of a 1.59 mm solution of 1b in 0.1m TEAP/
DMF at 0.1 Vsꢀ1: a) scanning to ꢀ2.77 V (solid) and ꢀ1.77 V (dashed)
and b) scanning to ꢀ2.27 V and repetitive cycling between the potential
window of ꢀ1.07 to ꢀ2.27 V.
Cyclic voltammetry of 1a and 1b: The cyclic voltammetry
(CV) and controlled potential electrolysis (CPE) of the
TADs were studied in N,N-dimethylformamide (DMF) and
acetonitrile (CH3CN) with 0.10m tetraethylammonium per-
chlorate (TEAP) as the supporting electrolyte at 258C using
a 3 mm glassy carbon working electrode. Representative
CVs of 1a and 1b are shown in Figures 1and 2 and the rele-
vant electrochemical data are given in Table 1. As seen in
Figure 1, on the initial cathodic scan of 1a, the CV exhibits
a broad, irreversible wave with a peak potential (Ep) of
ꢀ1.43 V versus SCE at 0.1 Vsꢀ1. The Ep shifts negatively by
110 mV per decade on increasing the scan rate. The peak
width, defined as jEpꢀEp/2 j, increases from 135 to 152 mV
between 0.1 and 10 Vsꢀ1. The transfer coefficient, a defined
in Table 1, is less than 0.40, suggesting based on these crite-
ria that the heterogeneous ET is the rate-determining step
180
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 178 – 188

Radical Ions
FULL PAPER
Table 1. Cyclic voltammetry data for the TADs in DMF and CH3CN in 0.10m TEAP at 258C at a glassy
carbon electrode.
peak, rendering the measure-
ment of the Ep intractable
above 1.0 Vsꢀ1. Scanning to
n [Vsꢀ1
]
1a[a]
1b[a]
1c[a,b]
1d[c]
Ep [V vs. SCE]
0.1
1.0
10
0.1
1.0
10
ꢀ1.43
ꢀ1.52
ꢀ1.64
135
144
152
ꢀ1.53[d]
ꢀ1.29
ꢀ1.39
ꢀ1.50
98
105
127
ꢀ1.36[d]
ꢀ1.80
nd[e]
nd[e]
223
ꢀ0.918
ꢀ0.914
ꢀ0.953
76
64
58
ꢀ0.887[d]
more
negative
potentials
ꢀ1.65[d]
ꢀ1.78[d]
152[d]
ꢀ1.48[d]
ꢀ1.60[d]
121[d]
nd[d,e]
nd[d,e]
82[d]
beyond the redox couple of 4c
reveals the irreversible wave
due to the reduction of 4c to
its dianion.
DEp/2 = jEp/2-Ep j [mV]
174[d]
145[d]
nd[e]
nd[e]
24
nd[d,e]
nd[d,e]
81[d]
197[d]
172[d]
Ip/Cn1/2 [mAs1/2 Vꢀ1/2 mmꢀ1
a=1.857RT/FDEp/2
]
25
24[d]
39
31[d]
40
[f]
Some differences between
the CVs of 1a and 1c are
worthy of further mention. At
scan rates where the peak po-
tential is observed, the Ep for
1c occurs at much more nega-
tive potentials than for either
1a or 1b. The peak heights of
1c are also much lower than
the peak heights of its monocy-
0.1
1.0
10
0.35
0.33
0.31
0.26
ꢀ0.11
0.31[d]
0.27[d]
0.24[d]
0.22[d]
ꢀ0.13[d]
0.49
0.45
0.37
0.28
ꢀ0.11
0.39[d]
0.33[d]
0.28[d]
0.22[d]
ꢀ0.12[d]
0.21
nd[e]
nd[e]
0.28
ꢀ0.10
0.63
0.74
0.82
0.87
ꢀ0.03
0.58[d]
nd[d,e]
nd[d,e]
0.49[d]
ꢀ0.06[d]
a[g]
@Ep/@ACHTUNGTRENNUNG(logn) [V]
[a] CVs in DMF/0.10m TEAP. [b] CVs in acid display two amalgamated waves, initial peak potential obscured
at all scan rates. [c] CVs in CH3CN/0.10m TEAP; CVs in DMF with and without acid were unreliable. [d] In
the presence of excess weak acid the initial wave becomes completely irreversible. [e] Peak obscured by fol-
lowing wave; unable to determine peak potential; nd=not determined. [f] Peak current normalized for con-
centration and scan rate. [g] a=1.15RT/F(dEp/dlogn).
by a concerted dissociative mechanism (vide infra).[34] At all
scan rates examined, the initial peak remains irreversible.
The CVs of 1b are similar to those of 1a in most respects. A
noticeable difference is that the initial cathodic peak of 1b
is considerably sharper with the DEp/2 increasing from
98 mV at 0.1 Vsꢀ1 to 127 mV at 10 Vsꢀ1. The Ep is located at
ꢀ1.29 V versus SCE at 0.1 Vsꢀ1, and also shifts negatively
by 110 mV per decade with increasing scan rate.
Scanning in the cathodic direction beyond the initial
wave, both 1a and 1b exhibit a sharp, oxidative current dip
at lower scan rates about ꢀ1.8 V. This dip becomes less pro-
nounced with increasing scan rates ultimately disappearing
at scan rates over 1.0 Vsꢀ1.[7,8] The current dip directly pre-
cedes the reversible wave of the product resulting from the
initial ET reduction step. By repetitive cycling voltammetry
(Figures 1b and 2b), the standard reduction potentials of the
redox couples were determined to be Eo =ꢀ1.77 and
ꢀ1.95 V versus SCE from the CVs of 1a and 1b, respective-
ly. These reversible waves are due to the reduction of benzo-
phenone (4a) and 4,4’-dimethoxy benzophenone (4b) to the
radical anions 5a and 5b, respectively, as verified by com-
parison with authentic samples. Following the reversible
redox couple in each CV is an irreversible reduction wave
(with Ep =ꢀ2.40 and ꢀ2.45 V vs. SCE) in the CVs of 1a and
1b, which are due to the reduction of the corresponding
benzophenone radical-anion to its dianion as confirmed
from the CVs of authentic 4a and 4b, respectively.
Figure 3. Representative CVs of a 1.50 mm solution of 1c in 0.1m TEAP/
DMF at 0.1 Vsꢀ1: a) scanning to ꢀ2.76 V (solid) and ꢀ2.21 V (dashed)
and b) scanning to ꢀ2.21 V and repetitively cycling between a potential
window of ꢀ1.36 and ꢀ2.21 V.
clic analogues 1b and characterization of the dissociative
wave of 1c is hindered by the close proximity of the reversi-
ble wave 4c, which also contributes to the smaller peak
height. Precautions were taken to ensure that the differen-
ces between the CVs of 1b and 1c were real and not due to
adsorption effects on the electrode. As we will discuss in
greater detail, digital simulation of the experimental voltam-
mograms provides evidence that these divergences are cor-
related to differences in the heterogeneous reduction kinet-
ics.
Cyclic voltammetry of 1c: Cyclic voltammograms of 1c are
shown in Figure 3. A broad, irreversible dissociative wave is
located at a Ep =ꢀ1.80 V versus SCE at 0.1 Vsꢀ1. The peak
is observed to shift negatively by 100 mV per decade over a
scan rate range of two log decades. The dissociative wave is
followed by a current oxidative dip that precedes the rever-
sible wave due to the reduction of 4c, which is apparent
from the repetitive cycling experiments. At higher scan
rates, the dissociative wave converges with the 4c reduction
Cyclic voltammetry of 1d: For the TADs 1a–1c, the disso-
ciative peak shape and the variations of peak breadth and
Chem. Eur. J. 2010, 16, 178 – 188
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
181

M. S. Workentin et al.
height are suggestive of some degree of a concerted disso-
peak at ꢀ1.2 V in the CVs of 1d. Further support for this as-
signment is obtained from the controlled potential electroly-
sis experiments.
ꢀ
ciative process in which heterogeneous ET and O O bond
cleavage occur simultaneously. However, the voltammetry
of 1d is intriguing in that a broad, irreversible peak is not
evident at any examined scan rate. Figure 4 shows represen-
Controlled potential electrolyses (CPE): CPE experiments
with 1a, 1b, and 1d were performed in dimethylformamide
(DMF) and acetonitrile (CH3CN) with the aim of producing
sufficient quantities of the reduction products for isolation,
identification, and quantification. The results of the CPE ex-
periments are contained in Table 2and the structures of the
products are shown in Scheme 2. The data show the product
ratios are dependent on the applied potential during the
electrolyses and the presence of a weak acid. Electrolyses of
1a at a potential near its Ep in DMF and CH3CN yields
1,1,4,4-tetraphenyl-1,4-butanediol (3a) in a 89:11 ratio over
benzophenone (4a). Similarly, reducing 1b at its Ep produ-
Table 2. Product ratios and charge consumption data obtained from the
controlled potential electrolysis of TADs in DMF and CH3CN in pres-
ence of 0.10m TEAP at a rotating disk electrode with a 12 mm glassy
carbon tip.
Endoperoxide
Electrolysis
potential[a]
Solvent
Product
ratio (3:4)
n[f]
1a
1a
1a
1a
1a
1a
1b
1b
1b
1b
1d
1d
Ep
DMF
DMF
89:11[c]
2:98[c]
1.9
0.08
2.0
1.9
0.4
1.9
1.9
0.2
1.9
0.08
0.3
2.0
<Eo4 a=5 a
<Eo4 a=5 a
Ep
DMF[b]
CH3CN
CH3CN
CH3CN[b]
DMF
99:1[c]
Figure 4. Representative CVs for 1d and 4d in 0.1m TEAP/CH3CN when
scanning at 0.1 Vsꢀ1: a) 1.2 mm 1d scanning to ꢀ1.05 V (dotted), ꢀ1.22 V
(dashed) and ꢀ1.82 V (solid). The arrow indicates the reduction peak as-
signed to 3d (see text). b) 2.4 mm 4d scanning to ꢀ1.19 V (dashed) and
ꢀ1.69 V (solid).
89:11[c]
46:54[c]
97:3[c]
<Eo4 a=5 a
<Eo4 a=5 a
Ep
>100:1[d]
3:97[d]
<Eo4 b=5 b
Ep
DMF
CH3CN
CH3CN
CH3CN
CH3CN[b]
>100:1[d]
3:97[d]
<Eo4 b=5 b
Ep
9:91[e]
tative CVs of 1d in CH3CN rather than in DMF as the CVs
were not readily reproducible. At lower scan rates, a should-
er is observed on the initial wave at about ꢀ0.8 V, which we
interpret as the beginning of the dissociative wave. As the
scan rate is raised, the dissociative wave blends into the re-
versible couple, which is assigned to the ET reduction of 4-
methoxy-4’-nitrobenzophenone
Ep
78:22[e]
[a] Eo4 a=5 a =ꢀ1.77, Eo4 b=5 b =ꢀ1.95 V vs. SCE. [b] In the presence of 20 mm
2,2,2-trifluoroethanol.[c] Determined by GC. [d] Determined by HPLC.
[e] Determined by 1H NMR spectrum integration. 3=diol, 4=benzo-
phone, 5=benzophenone radical anion. [f] Number of electron equiva-
lents consumed during electrolysis in Fmolꢀ1
.
(4d). The CV of 4d in CH3CN
replicates most of the CV fea-
tures of 1d. For example, 4d
exhibits two reversible reduc-
tion waves with EoR1 =ꢀ0.87 V
and EoR2 =ꢀ1.32 V versus SCE.
In the CVs of 1d, a small peak
between the two redox cou-
ples, as indicated by the arrow
in Figure 4a, is attributed to
the reduction of 1,4-bis(4-me-
thoxyphenyl)-1,4-bis(4-nitro-
phenyl)butane-1,4-diol
(3d).
This assignment is based on
the voltammetry of 4-methoxy-
4’-nitrobenzhydrol, which dis-
plays
similar
voltammetry
characteristics to the middle
Scheme 2. The catalytic radical-anion chain mechanism proposed for the 3,3,6,6-tetraaryl-1,2-dioxanes 1a–1d.
182
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 178 – 188

Radical Ions
FULL PAPER
ces 1,1,4,4-(4-methoxyphenyl)-1,4-butanediol (3b) in near
quantitative amount with only a trace amount of 4,4’-di-
ACHTUNGTRENNUNGmethoxybenzophenone (4b) in the product mixture. In both
cases, nearly two electron equivalents of charge are required
for complete reduction of the endoperoxides.
negative with respect to Eo4=5, any 4 produced by fragmenta-
tion of 2 is reduced to its anion, 5. This is a key aspect in
the mechanism. At these negative potentials, for every dis-
tonic radical anion that undergoes fragmentation, two equiv-
alents of 5 are generated, one directly from the b-scission
fragmentation and the other indirectly from the single heter-
ogeneous ET reduction of 4. This results in a substantial
concentration of 5 in the vicinity of the electrode surface.
As more 1 diffuses toward the electrode, the endoperoxide
is intercepted by 5 and is reduced homogeneously rather
than heterogeneously at the electrode surface. This triggers
the propagating ET radical-anion chain reaction. As species
2 is formed further and further away from the electrode sur-
face, heterogeneous ET reduction cannot as readily compete
due to the exponential distance dependence on the ET so
that fragmentation (path C) becomes favored over reduction
(path B). As a consequence, the homogeneous reduction of
1 by 5 produces more 2, which generates at least another
equivalent of 5. Therefore, at <Eo4 a=5 a, the charge necessary
for complete reduction of 1 is the amount needed to gener-
ate enough 5 near the electrode surface to initiate the frag-
mentation-homogeneous ET radical-anion chain process.
Once initiated, the chain reaction spreads into the bulk solu-
tion and consumes the rest of 1 without requiring additional
charge. Hence the low charge consumption values ranging
from 0.4 to 0.08 Fmolꢀ1.
At potentials corresponding to the Ep, which is positive to
the Eo4 a=5 a, the radical-anion propagating chain mechanism
still operates, but it is not charge efficient. The charge con-
sumption is stoichiometric, rather than catalytic, as nearly
2 Fmolꢀ1 of charge is required at these potentials. Products
resulting from fragmentation are observed, but to a much
lesser extent. In this instance, only one equivalent of 5 is
possible from each fragmentation event, as 4 cannot be di-
rectly reduced by the electrode. Thus, the amount of 4 at
the electrode interface does not readily accumulate, and
hence, the concentration of 4 is much lower such that the bi-
molecular reduction 1 and 2 is no longer as competitive as
the heterogeneous ET reduction.
The proposed mechanism also accounts for the oxidative
current dip observed in the voltammetry of 1a–1c, as indi-
cated by the arrow in Figure 1. The value of the current is
representative of the concentration of the electroactive spe-
cies 5 present at the electrode surface. The dip occurs at po-
tentials at which an equilibrium exists between 4 and 5. As
the concentration of 5 near the electrode increases, we see a
dramatic drop in current as a result of there being less 1 to
reduce at the electrode. The rapid, sudden drop in current
reflects the amount of 1 that is not reaching the electrode,
and instead being reduced homogeneously by 5, as a result
of the radical-anion chain process.
As mentioned earlier, the presence of excess weak acid
has a significant effect on the product yields. According to
Scheme 2, acid could suppress the homogeneous propagat-
ing mechanism in two ways. First, acid could protonate the
distonic radical anion 2. In the presence of excess TFE, an
almost quantitative amount of diol was isolated in the CPE
However, when the applied potential is set negative to
the standard potential of the benzophenone species E4o=5
(e.g., Eo =ꢀ1.77 and ꢀ1.95 V versus SCE), the product
ratios are inverted. Under these conditions in DMF, the
benzophenone compound is the primary product in greater
than 96% with little diol formed.
Most significantly, at an applied potential negative to E4o=5
,
no more than 0.4, and as little as 0.08, electrons per mole-
cule are needed for complete conversion of starting material
to products. In contrast, the results from the CPE experi-
ments of 1d are quite different than those of 1a and 1b.
CPE of 1d at the foot of the reduction wave results in for-
mation of benzophenone in a ratio of 91:9 over diol with a
modest charge consumption of 0.3 Fmolꢀ1. This is counterin-
tuitive based on the observations from 1a and 1b.
Electrolysis experiments of 1a and 1d were also carried
out in the presence of 2,2,2-trifluoroethanol (TFE) and the
results of these experiments are also summarized in Table 2.
CPE of 1d in CH3CN in the presence of excess TFE at an
applied E=ꢀ0.90 V yields predominantly diol 3d, but with
a significant amount of benzophenone 4d with a charge con-
1
sumption of 2 Fmolꢀ1. The product ratio by H NMR spec-
troscopy was a 4:1 ratio of 3d over 4d.[40] From the results
in Table 2, it can be generalized that the presence of an acid
results in an increased amount of diol.
Reaction mechanisms of 1a–1c: Our proposed reaction
mechanism for 1a–1c, illustrated in Scheme 2, accounts for
the observed voltammetry, the isolated products, the poten-
tial dependence of the product ratios, and the charge con-
sumption. It consists of four main steps. The first step (la-
beled path A) is the initial ET to the endoperoxide, which
occurs by a dissociative mechanism resulting in cleavage of
ꢀ
the O O bond to generate the distonic radical anion 2. Re-
duction of the alkoxy radical center of 2 to yield the di-
ACHTUNGTRENNUNGanion, followed by protonation results in the diol 3 (through
path B).[41] In competition with reduction of 2, b-scission
fragmentation results in the corresponding benzophenone 4,
its radical anion 5, and an alkene 6 (path C). The production
of 5 is the crucial step in the propagating radical-anion
chain mechanism. Since the standard potential of 4a (E4oa=5 a
)
is over 30 mV negative with respect to the Ep of 1a, and
Eo4 b=5 b is over 60 mV negative with respect to the Ep of 1b,
ET from 5a to 1a and 5b to 1b are both thermodynamically
favorable processes. Hence, the fourth and final step in
Scheme 2 is the homogeneous reduction of 1 by 5 to gener-
ate 4 and another species of 2 (path D). The isolation of 4a
and 4b from the reduction of 1a and 1b, respectively, sup-
ports b-scission fragmentation of 2 as a competitive pathway.
In fact, considering that 2 is produced near the electrode
surface, the fragmentation must occur rapidly in order to
compete with reduction. When the electrode potential is
Chem. Eur. J. 2010, 16, 178 – 188
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
183

M. S. Workentin et al.
experiments of 1a in comparison to only 89% in the ab-
sence of acid. The presence of acid causes a loss in chain re-
activity due to the protonation of 2, which yields the alkoxyl
radical species 2aH. Under these circumstances, the frag-
mentation of 2aH could produce benzophenone, ethylene,
and a diphenylhydroxymethyl radical. None of these frag-
ments are capable of acting as a homogeneous ET donor to
reduce 1a, so the chain process ceases to operate efficiently.
The second way excess acid could potentially affect the re-
duction mechanism is by protonating the benzophenone rad-
ical anion to yield a ketyl radi-
tially capable of also behaving as homogeneous ET donors
including the diol 3d.
Digital simulations of the cyclic voltammograms: The CVs
of the four TADs were simulated using Digisim 3.0 software.
The mechanism employed to simulate the cyclic voltammo-
grams and to determine unknown kinetic and thermody-
manic information is based on Scheme 2. A summary of the
most influential elementary steps included in the simulations
are shown in Scheme 3. In both the concerted and stepwise
cal intermediate, which on fur-
ther reduction and protonation
would result in benzhydrol.
This second possibility results
if the b-scission fragmentation
of 2aH is competitive with
electrode reduction. In our ex-
periments, no benzhydrol was
detected from the CPE of 1a
in the presence of excess TFE.
The experimental evidence
suggests excess weak acid
favors protonation of 2, which
hinders the b-scission fragmen-
tation and favors the forma-
tion of 3 by path B, thus inhib-
iting the radical-anion chain
reaction through path C. This
interpretation is consistent
with our previous studies that
the fragmentation rates of dis-
tonic radical anions are differ-
ent from the analogous radical
species.[7,8]
Reaction mechanism of 1d:
Scheme 3. A summary of the most influential elementary steps in the digital simulations of the 3,3,6,6-tetraar-
yl-1,2-dioxanes 1a–1d. Refer to Scheme 2 for substituents key for Ar and R.
Figure 4 highlights the lack of
a broad and irreversible wave
normally associated with ET
reduction of an endoperoxide.
Rather, the dissociative wave of 1d at ꢀ0.8 V is the small
shoulder in Figure 4, on the left side of initial reversible
redox couple due to the reduction of the nitrobenzene units.
Unlike the CVs for the other three endoperoxides, the Ep of
1d occurs very close to the reduction potential of the benzo-
phenone 4d (Eo4 d=5 d), whereas the reduction potentials of
4a–4c are significantly negative with respect to the dissocia-
tive waves of 1a–1c. The consequence is that the reduction
of 1d essentially occurs under conditions that induces the
homogeneous ET-fragmentation radical-anion chain reac-
tion from the onset. As soon as 1d is reduced at the elec-
trode, the homogeneous reduction very rapidly dominates
the heterogeneous ET reduction. In the case of 1a–1c, the
homogeneous ET donor is the radical anion 5a–5c. Howev-
er, the inclusion of the nitro substituent in 1d facilitates the
formation of a number of products and intermediates poten-
mechanisms, there are at least five rate constants of perti-
o
o
ꢀ
nence: 1) the heterogeneous ET to the O O khetC or khetS; 2)
ꢀ
fragmentation of the O O bond kdC, or kdS; 3) the heteroge-
neous ET to the distonic radical-anion, khet2; 4) the b-scis-
sion fragmentation of 2, kfrag; and 5) the homogeneous re-
duction of 1 from 5 khomET. The parameters resulting from
the simulations are reported in Table 3. Initial input values
for many of the parameters, including the standard potential
EoC, the heterogeneous rate constants khoetC and the diffusion
coefficient D were taken from reported values.[17,23,26] Tradi-
tionally, this information has been conveniently attained by
convolution potential sweep voltammetry (CPSV).[35a,42]
However, the non-Cottrell behavior (i.e., the oxidative dip)
in the voltammetry due to the homogeneous ET–fragmenta-
tion cycle hindered the quality of the data, even in the pres-
ence of excess weak acid. Notably, the convoluted curves
184
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 178 – 188

Radical Ions
FULL PAPER
Table 3. Key parameters for the heterogeneous and homogeneous ET re-
duction as determined by digital simulation of the cyclic voltam-
can be found in Table 3. In these simulations, emphasis was
placed on achieving parameter constants that provided pre-
cise simulations at a range of scan rates from 0.1 to 10 Vsꢀ1.
Prudence was placed, in particular, on accurately simulating
the peak current at all scan rates investigated. Simulations
were also done at two or more concentrations for each
TAD.
Overall, the differences in the parameters for 1a–1c are
rather subtle with the exception of the kohetC values. There is
a 10000-fold difference in the rate constants between 1b
and 1c, which is attributed to the more rigid structure of 1c.
The kohetC for 1a is in the middle of the range with a value of
2.4ꢃ10ꢀ6 cmsꢀ1. The diffusion coefficients decrease slightly
from 5.3ꢃ10ꢀ6 for 1a to 4.4ꢃ10ꢀ6 cm2 sꢀ1 for 1c in keeping
with the increasing sweep radius.
ACHTUNGTRENNUNG
mograms.[a,b]
Parameters
1a
1b
1c
1d
EoC [V]
ꢀ0.96
2.4ꢃ10ꢀ6
1.2
ꢀ1.15
ꢀ0.89
4.7ꢃ10ꢀ8
1.1
n/a[c]
n/a[c]
n/a[c]
-1.05[d]
0.1[d]
0.4[d]
n/a[c]
1ꢃ108
kohetC [cmsꢀ1
]
4.2ꢃ10ꢀ4
1.2
l+BDE, [V]
EoS [V]
n/a[c]
ꢀ1.15[d]
3.5ꢃ10ꢀ4[d]
0.45[d]
n/a[c]
kohetS [cmsꢀ1
]
n/a[c]
n/a[c]
a
n/a[c]
n/a[c]
kdC [sꢀ1
]
]
1ꢃ1013
n/a[c]
1ꢃ1013
1ꢃ1013
1ꢃ1013
n/a[c]
kdS [sꢀ1
Eo2 [V]
ꢀ0.1
0.1
kohet2 [cmsꢀ1
]
khomET [mꢀ1 sꢀ1
kfrag [sꢀ1
D [cm2 sꢀ1
]
3.7ꢃ105
1.4ꢃ106
5.3ꢃ10ꢀ6
3.5ꢃ106
1.1ꢃ106
4.8ꢃ10ꢀ6
1.5ꢃ106
8.3ꢃ106
4.4ꢃ10ꢀ6
7.3ꢃ105
4.9ꢃ104
1ꢃ10ꢀ6
]
]
[a] The subscripts C and S identify parameters specific to a concerted dis-
sociative and stepwise dissociative ET reaction, respectively. [b] Listed
values are averages from many simulations at various scan rates. [c] n/a=
not applicable. [d] Simulations employed Butler–Volmer electrode kinet-
ics for initial ET.
A primary objective of the simulations was to obtain rate
constants for the b-scission fragmentation kfrag and for the
homogeneous ET to the endoperoxide khomET. The value of
khomET was determined from simulation of the observed oxi-
dative dip. It was found that the shape of the dip was highly
sensitive to khomET, and thus used as a diagnostic parameter
in the simulations. The driving force for the homogeneous
ET reduction of 5a to 1a is calculated to be approximately
ꢀ17 kcalmolꢀ1. The homogeneous ET between 5d and 1d is
not expected to follow the same activation-driving force re-
lationship as for other peroxides, as it is a stepwise reduc-
tion.
The ET reduction of 1d clearly occurs by a stepwise disso-
ciative mechanism. The Eo value for reduction of 1d was es-
timated based on the Eo determined for 4-methoxy-4’-nitro-
benzhydrol (ꢀ1.1 V vs. SCE), which served as a reasonable
model. The Eo value assigned to 1d is more negative than
the Eo of 4d, because the additional electron of its radical
anion 5d can be delocalized onto the carbonyl moiety. This
effect is seen in the Eo values of nitrobenzene (ꢀ1.12 V vs.
SCE) and tert-butyl 4-nitrobenzoate (ꢀ0.89 V vs. SCE), in
which the ester function located para to the nitro group re-
sults in a significant decrease in the Eo through the en-
hanced stabilization of the radical anion.
were found to be unreliable as the limiting current was not
scan-rate independent and a constant plateau current was
not obtained. Therefore, average values were initially taken
from bicyclic endoperoxide CPSV studies; for example, the
EoC range from ꢀ0.6 to ꢀ1.2 V, khoetC and D average 6ꢃ10ꢀ7
and 6ꢃ10ꢀ6 cm2 sꢀ1.[7,8,23] Approximately constant values of
1.2 V, 0.1 V, and 0.1 cmsꢀ1 were used for the BDE+l, Eo2,
and kohet2, the last two parameters being specific to the heter-
ogeneous reduction of the distonic radical anion.
Examples of high-quality reproductions of the CVs of 1a–
1d can be found in Figure 5 overlapping the experimental
CVs at a scan rate 0.1 Vsꢀ1 and the simulation parameters
The initial heterogeneous ET to 1d has a rate constant of
0.1 cmsꢀ1. This value is much larger than the rate constants
for the concerted mechanisms for 1a–1c, which have values
smaller than 10ꢀ3 cmsꢀ1. In addition, the rate constant for
8
ꢀ1
ꢀ
dissociation of the O O radical anion is 1ꢃ10 s , which is
five orders of magnitude lower than for the concerted mech-
anism. Another indication that the reduction mechanism for
1d is indeed stepwise dissociative is the fact that the CVs
shift by only 30 mV per tenfold increase in scan rate.
For comparison, simulations of 1d were also carried out
assuming a concerted dissociative ET mechanism. The same
Eo value as for 1a was used, as the electron-withdrawing
ꢀ
effect of the nitro–aryl groups on the O O bond is expected
to offset the electron-donating effect of the methoxy–aryl
groups. The kohet value was taken from the results employed
in the simulations of 1b, which has equatorial 4-methoxy-
phenyl rings similar to 1d. At 0.1 Vsꢀ1, the simulation was
found to be similar as in the stepwise mechanism; however,
Figure 5. Comparison of simulated CVs (circles) with the experimental
CVs (line) at 0.1 Vsꢀ1: a) a 1.95 mm solution of 1a in 0.1m TEAP/DMF;
b) a 1.59 mm solution of 1b in 0.1m TEAP/DMF; c) a 1.50 mm solution
of 1c in 0.1m TEAP/DMF; and d) a 1.13 mm solution of 1d in 0.1m
TEAP/CH3CN.
Chem. Eur. J. 2010, 16, 178 – 188
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
185

M. S. Workentin et al.
with increasing scan rate, the initial peak becomes broader
and shifts increasingly to more negative potentials in excess
of 60 mV per log decade, thus ruling out a purely concerted
mechanism.
The fragmentation rate constants for the distonic radical
anions from the digital simulations range from 104 to 106 sꢀ1.
Our estimates for bicyclic diphenyl-substituted endoperox-
ides had a lower limit of 3ꢃ107 sꢀ1.[7,8] A possible explana-
tion for the difference may be due to the conformation of
the endoperoxide on acceptance of the incoming electron.
The endoperoxides 1a–1d are already in a chair conforma-
tion on accepting the electron, in contrast to the bicyclic di-
phenylendoperoxides, which are initially in a rigid, boat con-
formation with the lone pairs on the oxygen atoms eclipsing.
Hence, a greater reorganization energy should be expected
for the [2,2,1] and [2,2,2]bicyclic endoperoxides.
Additional insight into the reasons for the low intrinsic
barrier for 1a–1d comes from recent studies on the ET re-
duction of diaryl disulfides.[44,45] In these studies, the authors
suggested the ET proceeds by a mechanism somewhere be-
tween the two mechanistic extremes of stepwise and con-
ꢀ
the distonic radical anion upon O O bond cleavage. The bi-
cyclic diphenyl-substituted endoperoxides are anticipated to
be in a strained conformation with the lone pairs on the
¼
ꢀ
oxygen atoms eclipsing one another. Cleavage of the O O
certed. Notably, the DGo was found to be too large for a
relieves strain energy and facilitates the b-scission fragmen-
typical stepwise process, and upon investigation it was con-
ꢀ
tation. In contrast, the TAD endoperoxides 1a–1d are prob-
cluded that there is a significant contribution from S S
ꢀ
ably in a chair conformation upon rupture of the O O
bond lengthening to the radical anion. In hindsight, a relat-
ed situation may have been serendipitously observed during
bond, so that there is a lesser degree of strain. Furthermore,
in a chair conformation the electron lone pairs are staggered
so there is also a lesser degree of charge repulsion than as-
sociated within bicyclic endoperoxides.
¼
the ET reduction of DPA-O2, in which the DGo was lower
than expected for a purely concerted ET mechanism.[23] The
¼
lower values for the DGo of 1a–1d, although admittedly
Laser flash photolysis of 1a provided a lower limit for the
rate constant of fragmentation of the biradical from homo-
puzzling to us at first, lead us to consider a competitive con-
certed–stepwise ET dichotomy.
ꢀ
lytic O O bond cleavage. The transient spectrum showed
To investigate our hypothesis we employed DFT calcula-
tions. The calculations involved a geometry optimization of
the TAD structure by using semi-empirical PM3 calculations
followed by a single-point energy DFT calculation using the
B3LYP//6-31G** basis set to calculate the LUMO. A depic-
tion of the calculated LUMOs for 1a, 1b, and 1d are shown
in Figure 6.
only the decay of triplet benzophenone. Given that the laser
pulse is 20 ns suggests that b-scission fragmentation of the
dialkoxy radical to benzophenone occurs with a rate con-
stant of at least 108 sꢀ1. Previous studies from the photolysis
and thermolysis of 3,3,6,6-tetramethyl-1,2-dioxanes have
suggested a concerted fragmentation process.[43] The impact
of the additional charge on the distonic radical anion 2a is
significant, compared to its biradical analogue, as it results
in a 100-fold decrease in the rate of fragmentation of the
charged intermediate. Discrepancies between the rates con-
stants for fragmentation of radical-anions and their neutral
radical counterparts are only recently becoming
known.[3,4,7,8] For the fragmentation of 2a, it is suggested
that the alkoxyl anion is more readily stabilized by the aryl
rings than the alkoxyl radical center. This assessment agrees
with the trend in the fragmentation rate constants as 2d has
the lowest value as the 4-nitrophenyl rings provide the
greatest stabilizing effect on the alkoxide anion.
The equilibrium conformer of 1a adopts a chair confor-
ꢀ
mation with the non-bonding orbitals of the O O atoms
staggered. The equatorial phenyl rings are aligned such that
the p orbital of the ipso-carbon atom directly overlaps with
ꢀ
the C O bond, which through hyperconjugation allows for
ꢀ
electronic interaction between the O O bond and the aryl
rings. The degree of this interaction may determine the step-
wise character of the ET reduction mechanism. For example,
in 1a and 1b the interaction is strong and the result is an
Density functional theory calculations: In the digital simula-
tions, we used a value of 1.2 V or 27.7 kcalmolꢀ1 for the sum
of BDE+l. These values correspond to an intrinsic barrier
of 6.9 kcalmolꢀ1 (divided by 4). Assuming a concerted disso-
ciative ET mechanism for 1a, a simple arithmetic calcula-
tion yields a BDE of 17.7 kcalmolꢀ1, if the lowest estimated
value for l is 10 kcalmolꢀ1.[7,8,19] Our studies on the hetero-
geneous ET of bicyclic diphenylendoperoxides has consis-
tently resulted in intrinsic barriers about 10 kcalmolꢀ1,
which we previously attribute to a concerted ET mechanism
[7,8]
ꢀ
involving cleavage of a weak O O bond.
Although a
lower limit of 18 kcalmolꢀ1 for the BDE is in line with pre-
vious studies, the l value in these studies was estimated to
be 20 kcalmolꢀ1. This difference in the l values may be rec-
onciled, but only partly, by considering the conformation of
Figure 6. Depiction of the LUMOs calculated for 1a (top left), 1b (top
right) and 1d (bottom center).
186
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 178 – 188

Radical Ions
FULL PAPER
ET reduction mechanism that falls between the concerted
dissociative and stepwise dissociative mechanisms. The same
calculations indicate that the p systems of the equatorial ar-
by a mechanism that is neither distinctly stepwise nor
AHCTUNGTRENNGcUN oncerted.
ꢀ
omatic rings are not aligned perpendicular to the C O bond
in the structure of 1c, which decreases the potential for elec-
Conclusions
ꢀ
tronic interaction between the aryl rings and O O bond.
This has important implications on the mechanism of ET, as
will be discussed below.
The ET reduction of endoperoxides 1a, 1b, and 1c proceeds
by a mechanism that has characteristics of concerted and
stepwise dissociative ET pathways. The CVs and digital sim-
ulations suggest that the heterogeneous ET is rate-determin-
ing, which is generally considered a strong indication that a
concerted dissociative mechanism is in operation. The
degree of stepwise character is reflected in the standard het-
erogeneous ET rate constant, kohet although the rigidity of
the molecular structure also contributes to differences in
kohet. Compound 1c has the lowest khoet, which is attributed to
the molecular backbone restricting the optimal conforma-
tion for ET.
Compound 1d is a mixture of cis and trans isomers, and
the PM3 equilibrium conformer was determined for both.
Both isomers adopt a chair configuration as for 1a and 1b.
The trans isomer has both of the 4-nitrophenyl rings in the
axial position. This was the lowest energy conformer even
when the initial structure began with both of the 4-nitro-
phenyl rings in the equatorial positions. One of the 4-me-
thoxyphenyl rings is aligned for optimal overlap of the p
ꢀ
system with the C O bond as was seen for 1a and 1b, while
the other is completely perpendicular with the geminal 4-ni-
trophenyl ring. In the cis isomer only one 4-nitrophenyl ring
can occupy an axial position at any given time, and the 4-ni-
trophenyl ring that sits in the equatorial position does not
The reorganization energy for the ET reduction of 1a–1c
is smaller than typically observed for the reduction of perox-
ides and endoperoxides, suggesting, with the assistance of
ꢀ
ꢀ
align its p system with the C O bond, while the 4-methoxy-
phenyl ring on the other side of the molecule does. Using
these conformers the LUMO of each compound was
DFT calculations, that the O O bond does not rapidly
cleave during the rate-determining step, and thus has step-
wise character. The LUMO was found not to be localized
on the cleaving bond, but rather distributed over the equa-
torial aryl rings, which contributes electron density to the
ACHTUNGTRENNUNG
ꢀ
ꢀ
ers, were nearly identical and in both cases the LUMO was
present on the equatorial aryl rings and spreading onto the
O O bond. This results in an increase in the O O bond
strength and the delocalization of the additional charge, pos-
sibility contributing to the formation of putative radical-
anion species. It is suggested that separation of the LUMO
from the cleaving bond increases stepwise character. This
concept is evidenced by the ET reduction of 1d. The ET re-
duction occurs by a stepwise dissociative mechanism as the
LUMO for 1d is primarily associated with an axial 4-nitro-
phenyl ring. The additional charge is initially associated with
ꢀ
ꢀ
C O bond and across the O O bond. Alkyl-substituted en-
doperoxides are considered to have the LUMO mainly asso-
ꢀ
ciated with the O O bond; however, this does not appear to
be the case for 1a and 1b. Our findings support the asser-
tion that ET occurs by a mechanism that is neither distinctly
stepwise nor concerted. The singly occupied molecular orbi-
tal (SOMO) of the putative radical anion for 1a was calcu-
ꢀ
ꢀ
lated and found to be completely localized over the O O
bond.
this ring prior to an intramolecular ET leading to O O
bond cleavage. This view of the ET reduction of 1d is sup-
ported by evidence from CV, digital simulation, and DFT
calculations. Collectively, the results suggest that a continu-
um of dissociative ET mechanisms exists for 1a–1d, and
that the two mechanisms do not exist solely in competition
with one another. Though it was expected that the nitro-sub-
stituted analogue would proceed by a stepwise mechanism,
the results suggest that a prediction of the mechanistic be-
havior can be made on the basis of the calculated LUMO.
Although not depicted in Figure 6, the LUMO of 1c is
also delocalized over the O O bond and the equatorial aryl
rings, despite the lesser degree of overlap of the p system
and the C O bond. For 1d the LUMO of both the cis and
trans isomer are the same, being associated with an axial 4-
nitrophenyl ring. This indicates that indeed the ET reduction
of 1d proceeds through a stepwise dissociative mechanism.
Furthermore, the second lowest unoccupied orbital in 1d,
LUMO+1, correlates with the p system of other 4-nitro-
phenyl ring. Hence, with 1d the antibonding orbital of the
ꢀ
ꢀ
ꢀ
Following ET and O O bond cleavage, the distonic radi-
cal-anion species undergoes fragmentation for all TADs in-
vestigated. Furthermore, a propagating radical-anion chain
mechanism is observed in all cases. The rate of fragmenta-
tion, estimated from digital simulation, is dependent on sub-
stituent. The distonic radical anions bearing electron-donat-
ing substituents fragment faster than 1d with nitro substitu-
ents. This correlates with the basicity of the alkoxide frag-
ment of the distonic radical anion. Furthermore, fragmenta-
tion of the distonic radical anion was observed to be at least
two orders of magnitude slower than the fragmentation of
the corresponding neutral biradical. The results from the
ꢀ
O O bond is not involved in the heterogeneous ET from
the electrode.
These results strengthen the argument that increasing the
separation of the LUMO and the atoms of the cleaving
bond increases stepwise character. While this assertion ap-
pears self-evident, the implication is that there exists species
where the LUMO is not completely isolated from the cleav-
ing bond, spatially or electronically, yet it is not localized di-
rectly over the cleaving bond either. This is the situation en-
countered with 1a and 1b, and the result is that ET occurs
Chem. Eur. J. 2010, 16, 178 – 188
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
187

M. S. Workentin et al.
[18] M. S. Workentin, R. L. Donkers, J. Am. Chem. Soc. 1998, 120, 2664.
[19] R. L. Donkers, M. S. Workentin, Chem. Eur. J. 2001, 7, 4012.
[20] D. C. Magri, R. L. Donkers, M. S. Workentin, J. Photochem. Photo-
biol. A 2001, 138, 29.
[21] D. C. Magri, M. S. Workentin, Org. Biomol. Chem. 2003, 1, 3418.
[22] R. L. Donkers, J. Tse, M. S. Workentin, J. Chem. Soc. Chem.
Commun. 1999, 135.
[23] R. L. Donkers, M. S. Workentin, J. Am. Chem. Soc. 2004, 126, 1688.
[24] D. L. B. Stringle, R. N. Campbell, M. S. Workentin, J. Chem. Soc.
Chem. Commun. 2003, 1246.
CV simulations indicate that the presence of charge can
have a dramatic effect on the reactivity of a radical species.
Though the charge and the spin are spatially and electroni-
cally separated, they have a combined influence on the over-
all reactivity. Further investigations into the reactivity of dis-
tonic radical-anions should continue to provide insight that
is difficult to obtain from merely studying either the radical
or anionic counterparts.
[25] F. Najjar, C. Andrꢁ-Barrꢄs, C. Lacaze-Dufaure, D. C. Magri, M. S.
Workentin, T. Tzꢄdakis, Chem. Eur. J. 2007, 13, 1174.
[26] S. Antonello, M. Musumeci, D. D. M. Wayner, F. Maran, J. Am.
Chem. Soc. 1997, 119, 9541.
Experimental Section
[27] S. Antonello, F. Maran, J. Am. Chem. Soc. 1997, 119, 12595.
[28] S. Antonello, F. Maran, J. Am. Chem. Soc. 1999, 121, 9668.
[29] S. Antonello, F. Formaggio, A. Moretto, C. Toniolo, F. Maran, J.
Am. Chem. Soc. 2001, 123, 9577.
The experimental details, synthesis and characterizations of the endoper-
oxides and products from the constant potential electrolysis studies can
be found in the Supporting Information.
[30] S. Antonello, M. Crisma, F. Formaggio, A. Moretto, F. Taddei, C.
Toniolo, F. Maran, J. Am. Chem. Soc. 2002, 124, 11503.
[31] S. Antonello, F. Formaggio, A. Moretto, C. Toniolo, F. Maran, J.
Am. Chem. Soc. 2003, 125, 2874.
[32] F. Polo, S. Antonello, F. Formaggio, C. Toniolo, F. Maran, J. Am.
Chem. Soc. 2005, 127, 492.
Acknowledgements
This work was financially supported by the Natural Sciences and Engi-
neering Research Council of Canada (NSERC), the Government of On-
tario (PREA) and the University of Western Ontario. D.L.B.S. thanks
NSERC for a postgraduate scholarship. D.C.M. thanks the Ontario Gov-
ernment for an OGSST postgraduate scholarship. Doug Hairsine is
thanked for performing the mass spectroscopic measurements.
[33] R. Improta, S. Antonello, F. Formaggio, F. Maran, N. Rega, V.
Barone, J. Phys. Chem. B 2005, 109, 1023.
[34] Reviews: a) J.-M. Savꢁant in Advances in Physical Organic Chemis-
try, Vol. 26 (Ed.: D. Bethel), Academic Press, New York, 1990, p. 1;
b) J.-M. Savꢁant, Acc. Chem. Res. 1993, 26, 455; c) J.-M. Savꢁant in
Advances in Electron Transfer Chemistry, Vol. 4 (Ed.: P. S. Mariano),
JAI Press, Greenwich, 1994, p. 53; d) J.-M. Savꢁant in Advances in
Physical Organic Chemistry, Vol. 35 (Ed.: T. Tidwell), Academic
Press, New York, 2000, p. 117; e) C. Costentin, M. Robert, J.-M. Sa-
vꢁant, Chem. Phys. 2006, 324, 40; f) C. Costentin, L. Donati, M.
Robert, Chem. Eur. J. 2009, 15, 785.
[1] Z. V. Todres, Organic Ion Radicals: Chemistry and Applications,
Marcel Dekker, NewYork, 2003.
[2] L. Eberson, Electron Transfer Reactions in Organic Chemistry,
Springer, Heidelberg, 1987.
[35] Reviews: a) F. Maran, D. D. M. Wayner, M. S. Workentin in Advan-
ces in Physical Organic Chemistry, Vol. 36 (Ed.: T. T. Tidwell), Aca-
demic Press, New York, 2001, p. 85; b) S. Antonello, F. Maran,
Chem. Soc. Rev. 2005, 34, 418; c) R. A. Rossi, A. B. Pierini, A. B.
Penenory, Chem. Rev. 2003, 103, 71; d) A. Houmam, Chem. Rev.
2008, 108, 2180.
[36] R. K. Haynes, M. K. S. Probert, I. D. Wilmot, Aust. J. Chem. 1978,
31, 1737.
[37] K. Gollnick, A. Schnatterer, Tetrahedron Lett. 1984, 25, 185.
[38] K. Gollnick, A. Schnatterer, G. Utschick, J. Org. Chem. 1993, 58,
6049.
[3] J. P. Stevenson, W. F. Jackson, J. M. Tanko, J. Am. Chem. Soc. 2002,
124, 4271.
[4] J. M. Tanko, J. P. Phillips, J. Am. Chem. Soc. 1999, 121, 6078.
[5] M. Chahma, X. Li, P. Phillips, P. Schwartz, L. E. Brammer, Y. Wang,
J. M. Tanko, J. Phys. Chem. A 2005, 109, 3372.
[6] J. M. Tanko, X. Li, M. Chahma, W. F. Jackson, J. N. Spencer, J. Am.
Chem. Soc. 2007, 129, 4181.
[7] D. C. Magri, M. S. Workentin, Org. Biomol. Chem. 2008, 6, 3354.
[8] D. C. Magri, M. S. Workentin, Chem. Eur. J. 2008, 14, 1698.
[9] E. L. Clennan, C. Foote in Organic Peroxides (Ed.: W. Ando),
Wiley, New York, 1992.
[39] K. Mizuno, T. Tamai, I. Hashida, Y. Otsuji, Y. Kuriyama, K. Toku-
maru, J. Org. Chem. 1994, 59, 7329.
[10] K. J. McCullough, M. Nojima, Curr. Org. Chem. 2001, 5, 601.
[11] a) D. A. Casteel, Nat. Prod. Rep. 1999, 16, 55; b) D. A. Casteel, Nat.
Prod. Rep. 1992, 9, 289.
[40] In the 1H NMR spectra it was noted that there were two unidenti-
fied resonances, comprising less than 10% of the product mixture.
These may be due to arylmethoxy protons and aryl protons of a p-
substituted aromatic moiety. However, the compounds were not iso-
lated or identified.
[12] a) S. R. Meshnick, T. E. Taylor, S. Kamchonwongpaisan, Microbiol.
Rev. 1996, 60, 301; b) A. Robert, B. Meunier, Chem. Soc. Rev. 1998,
27, 273; c) C. W. Jefford, Curr. Med. Chem. 2001, 8, 1803; d) A.
Robert, O. Dechy-Cabaret, J. Cazelles, B. Meunier, Acc. Chem. Res.
2002, 35, 167; e) P. M. OꢂNeill, G. H. Posner, J. Med. Chem. 2004, 47,
2945; f) P. M. OꢂNeill, G. H. Posner, Acc. Chem. Res. 2004, 37, 397.
[13] C. Thebtaranonth, Y. Thebtaranonth, S. Wanauppathamkul, Y. Yu-
thavong, Phytochemistry 1995, 40, 125.
[41] One possibility that we eluded to in references [7] and [8] is that in
the presence of a weak acid, protonation of the alkoxide portion
could first occur, followed by reduction of the alkoxyl portion and
subsequent protonation to yield diol 3.
[42] a) J. C. Imbeaux, J.-M. Savꢁant, J. Electroanal. Chem. 1973, 44, 169;
b) J.-M. Savꢁant, D. Tessier, J. Electroanal. Chem. 1975, 61, 251.
[43] W. Adam, J. Sanabia, J. Am. Chem. Soc. 1977, 99, 2735.
[44] S. Antonello, R. Benassi, G. Gavioli, F. Taddei, F. Maran, J. Am.
Chem. Soc. 2002, 124, 7529.
[14] a) S. Sperry, F. A. Valeriote, T. H. Corbett, P. Crews, J. Nat. Prod.
1998, 61, 241; b) K. El Sayed, M. T. Hamann, N. E. Hashish, T.
Shier, M. Kelly, A. A. Khan, J. Nat. Prod. 2001, 64, 522; c) Y.-C.
Shen, C. V. S. Prakash, Y.-H. Kuo, J. Nat. Prod. 2001, 64, 324;
d) D. T. A. Youssef, W. Y. Yoshida, M. Kelly, P. J. Scheuer, J. Nat.
Prod. 2001, 64, 1332.
[45] S. Antonello, K. Daasbjerg, H. Jensen, F. Taddei, F. Maran, J. Am.
Chem. Soc. 2003, 125, 14905.
[15] M. S. Workentin, F. Maran, D. D. M. Wayner, J. Am. Chem. Soc.
1995, 117, 2120.
[16] R. L. Donkers, M. S. Workentin, J. Phys. Chem. B 1998, 102, 4061.
[17] R. L. Donkers, F. Maran, D. D. M. Wayner, M. S. Workentin, J. Am.
Chem. Soc. 1999, 121, 7239.
Received: July 21, 2009
Published online: November 27, 2009
188
www.chemeurj.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2010, 16, 178 – 188