Elucidation of the Electron Transfer Reduction Mechanism of Anthracene Endoperoxides

Article abstract of DOI:10.1021/ja035828a

The homogeneous and heterogeneous reductions of the endoperoxides 9,10-diphenyl-9,10-epidioxyanthracene (DPA-O₂) and 9,10-dimethyl-9,10-epidioxyanthracene (DMA-O₂) were investigated, and they were found to undergo a dissociative elec

Full text of DOI:10.1021/ja035828a

Published on Web 01/21/2004
Elucidation of the Electron Transfer Reduction Mechanism of
Anthracene Endoperoxides
Robert L. Donkers† and Mark S. Workentin*
Contribution from the Department of Chemistry, The UniVersity of Western Ontario,
London, ON, Canada, N6A 5B7
Received April 28, 2003; E-mail: mworkent@uwo.ca
Abstract: The homogeneous and heterogeneous reductions of the endoperoxides 9,10-diphenyl-9,10-
epidioxyanthracene (DPA-O2) and 9,10-dimethyl-9,10-epidioxyanthracene (DMA-O2) were investigated, and
they were found to undergo a dissociative electron-transfer reduction of the O-O bond to yield a distonic
radical anion, with no evidence for C-O bond dissociation. A number of thermochemical parameters for
each were determined using Save´ant’s model for dissociative electron transfer (ET), including E°, ∆Gqo,
and bond dissociation energies. The products of the ET are dependent on the mode of reduction, namely
heterogeneous or homogeneous, and on the electrode potential or standard potential of the homogeneous
donor, respectively. The dissociative reduction of DMA-O2 under heterogeneous and homogeneous
conditions yields the corresponding 9,10-dihydroxyanthracene DMA-(OH)2, quantitatively, in an overall two-
electron process. In the case of DPA-O2, ET reduction also yields the corresponding 9,10-dihydroxyan-
thracene DPA-(OH)2 from reduction of the distonic radical anion, but in competition with this reduction, an
O-neophyl-type rearrangement occurs that generates a carbon radical with a minimum rate constant of
5.9 × 1010 s-1. In the presence of a sufficiently reducing medium, the carbon-centered radical is reduced
(E° ) -0.85 V vs SCE) and ultimately yields 9-phenoxy-10-phenyl anthracene (PPA). The observation of
this product is remarkable. In the heterogeneous ET, the yield of DPA-(OH)2/PPA is 97:3 and allows an
estimate of the rate constant for ET to the distonic radical anion. In homogeneous reductions, the O-neophyl
rearrangement is quantitative, but the yield of PPA depends on the redox properties of the donor. A unified
mechanism of reduction of DPA-O2 is presented to account for these observations.
Introduction
decompose thermally by two distinct reaction pathways. One
pathway regenerates the parent aromatic hydrocarbon and
oxygen in either the singlet or triplet state from a formal
cycloreversion reaction, and the other generates products formed
via homolytic cleavage of the O-O bond (Scheme 1).6,12,23-40
Anthracene endoperoxides exhibit interesting thermal and
photochemical properties.1-3 Their potential for reversible
oxygen storage and as a chemical source of singlet oxygen, 1O2
(1∆g), a species responsible for causing physiological damage,
has made these compounds the subject of many fundamental
investigations.4-23 Typically, anthracenyl-type endoperoxides
(15) Jesse, J.; Comes, F. J. J. Phys. Chem. 1991, 95, 1311-1315.
(16) Jesse, J.; Markert, R.; Comes, F. J.; Schmidt, R.; Brauer, H.-D. Chem. Phys.
Lett. 1990, 166, 95-100.
(17) Jesse, J.; Comes, F. J.; Schmidt, R.; Brauer, H.-D. Chem. Phys. Lett. 1989,
160, 8-12.
† Current address: Department of Physical Chemistry, University of
Padova, via Loredan 2, 35131 Padova, Italy.
(1) Clennan, E. L.; Foote, C. S. In Organic Peroxides; Ando, W., Ed.; John
Wiley & Sons: Chichester, England, 1992; pp 225-318.
(2) ActiVe Oxygen in Chemistry; Foote, C. S., Valentine, J. S., Greenberg, A.,
Liebman, J. F., Eds.; Blackie Academic and Professional: New York, 1995;
Vol. 2.
(3) ActiVe Oxygen in Biochemistry; Foote, C. S., Valentine, J. S., Greenberg,
A., Liebman, J. F., Eds.; Blackie Academic and Professional: New York,
1995; Vol. 3.
(18) Gudipati, M. S.; Klein, A. J. Phys. Chem. A 2000, 104, 166-167.
(19) Blumenstock, T.; Comes, F. J.; Schmidt, R.; Brauer, H.-D. Chem. Phys.
Lett. 1986, 127, 452-455.
(20) Brauer, H.-D.; Schmidt, R. J. Phys. Chem. A 2000, 104, 164-165.
(21) Eisenthal, K. B.; Turro, N. J.; Dupuy, C. G.; Hrovat, D. A.; Jenny, T. A.;
Sitzmann, E. V. J. Phys. Chem. 1986, 90, 5168-5173.
(22) Ernsting, N. P.; Schmidt, R.; Brauer, H.-D. J. Phys. Chem. 1990, 94, 5252-
5255.
(23) Aubry, J.-M.; Pierlot, C.; Rigaudy, J.; Schmidt, R. Acc. Chem. Res. 2003,
36, 668-675.
(4) Turro, N. J.; Chow, M. F. J. Am. Chem. Soc. 1981, 103, 7218-7224.
(5) Stevens, B.; Small, R. D., Jr. J. Phys. Chem. 1977, 81, 1605-1606.
(6) Sigman, M. E.; Zingg, S. P.; Pagni, R. M.; Burns, J. H. Tetrahedron Lett.
1991, 32, 5737-5740.
(24) Dufraisse, C.; Rocher, H. Bull. Soc. Chim. Fr. 1935, 2235-2240.
(25) Dufraisse, C. Bull. Soc. Chim. Fr. 1936, 1847-1872.
(26) Dufraisse, C.; Horclois, R. Bull. Soc. Chim. Fr. 1936, 1894-1905.
(27) Dufraisse, C.; Le Bras, J. Bull. Soc. Chim. Fr. 1937, 5, 1037-1046.
(28) Dufraisse, C.; Velluz, L.; Velluz, M. Bull. Soc. Chim. Fr. 1937, 1260-
1264.
(7) Sitzmann, E. V.; Langan, J. G.; Hrovat, D. A.; Eisenthal, K. B. Chem.
Phys. Lett. 1989, 162, 157-162.
(8) Schmidt, R.; Drews, W.; Brauer, H.-D. J. Photochem. 1982, 18, 365-378.
(9) Schmidt, R.; Drews, W.; Brauer, H.-D. Z. Naturforsch. A 1982, 37, 55-
57.
(29) Dufraisse, C.; Priou, R. Bull. Soc. Chim. Fr. 1939, 1649-1656.
(30) Dufraisse, C.; Velluz, L. Bull. Soc. Chim. Fr. 1942, 171-187.
(31) Dufraisse, C.; Mathieu, J. Bull. Soc. Chim. Fr. 1947, 307-310.
(32) Aksnes, G.; Vagstad, B. H. Acta Chem. Scand. 1979, B33, 47-51.
(33) Bachmann, W. E.; Chemerda, J. M. J. Chem. Soc. 1939, 116-118.
(34) Etienne, A. Bull. Soc. Chim. Fr. 1937, 634-638.
(35) Etienne, A.; Staehelin, A. Bull. Soc. Chim. Fr. 1954, 748-754.
(36) Etienne, A.; Lepeley, J.-C.; Heyme`s, R. Bull. Soc. Chim. Fr. 1949, 835-
840.
(10) Schmidt, R.; Schaffner, K.; Trost, W.; Brauer, H.-D. J. Phys. Chem. 1984,
88, 956-958.
(11) Rigaudy, J.; Defoin, A.; Baranne-Lafont, J. Angew. Chem., Int. Ed. Engl.
1979, 18, 413-415.
(12) Rigaudy, J.; Breliere, C.; Scribe, P. Tetrahedron Lett. 1978, 7, 687-690.
(13) Kearns, D. R.; Khan, A. U. Photochem. Photobiol. 1969, 10, 193-210.
(14) Jesse, K. J. Chem. Phys. Lett. 1997, 264, 193-198.
9
1688
J. AM. CHEM. SOC. 2004, 126, 1688-1698
10.1021/ja035828a CCC: $27.50 © 2004 American Chemical Society

ET Reduction Mechanism of Anthracene Endoperoxides
A R T I C L E S
Scheme 1. Competitive Reaction Pathways in the Thermal and Photochemical Activation of 9,10-Epidioxyanthracences
The relative yields of these two reaction processes have been
studied as a function of structure and experimental conditions
to ascertain the mechanism and activation parameters for a
number of systems.
and an excellent account of the reversible binding of oxygen to
aromatic compounds has recently appeared.23
Photochemical reaction mechanisms and dissociative electron
transfer (ET) mechanisms are often similar in their fate. For
example, photolysis of sigma bonds of molecule A-B can result
in homolysis into A• and B•. ET reduction of A-B often results
in an analogous fragmentation into A• and B-, the only
difference being the extra charge on the B fragment. Examples
of these include the reduction of C-X where X is a halogen,43-46
or O-O bonds2,3,47-52,54-56 of peroxides and endoperoxides. In
the case of alkyl endoperoxides such as ascaridole, both
photolysis and ET lead to O-O bond fragmentation.51-53 In
the alkyl endoperoxide cases we studied using direct electro-
chemical methods, ET leads to a concerted dissociative frag-
mentation of the O-O bond to yield a distonic radical anion
•O-R-R-O- (eq 1) that is reduced in a second ET (eq 2) to
ultimately yield the cis-diol; i.e., no reactivity resulting from
the alkoxy radical fragment is observed.
The thermolysis of a prototypical aromatic endoperoxide,
namely 9,10-diphenyl-9,10-epidioxyanthracene (DPA-O2), in
chlorobenzene yields only products from the cycloreversion,
whereas when the substituents at the bridgehead are not phenyl
the amount of products from cleavage of the O-O bond and
subsequent rearrangement increases. Rigaudy and co-workers
were first to report12 the photolysis of DPA-O2. DPA-O2 has
two UV-vis absorption bands, a high energy absorption
centered at 254 nm, and less intense low-energy absorption at
wavelengths above 350 nm. They discovered that irradiation
with 254-nm light leads to cycloreversion while irradiation at
the low-energy absorption leads to products arising from O-O
homolytic cleavage, including the diepoxide and products
derived from it. The homolytic cleavage to yield the biradical
is reversible, and thus reformation of the endoperoxide competes
with other reactions of the biradical. Kearns and Khan13
predicted this dual reactivity using orbital correlation diagrams
for the 1,4-cyclohexadiene endoperoxide and generalized that
these observations should apply to the endoperoxides of all
aromatic hydrocarbons. The photochemical behavior for the
endoperoxides of aromatic molecules was investigated by Brauer
and co-workers, who observed the dual reactivity as a general
occurrence in these systems.8-10,20,22,41 Investigation of the
photocycloreversion reaction by Brauer and co-workers using
femtosecond laser flash photolysis techniques supports the
hypothesis of a two-step process. The first step in their proposed
mechanism involves the barrierless C-O cleavage to form a
biradical in competition with internal conversion. Next, the
second C-O cleavage occurs to extrude 1O2 in the rate-
determining step.7,14-17,19,22 Recent computational and experi-
mental investigations of this mechanism by Klein and co-
workers dispute Brauer assignments of the excited states
involved in the competing processes.42 This was followed with
some discussion from both sides of this issue in the literature.18,20
Currently, the assignment of the excited states is open to debate,
Because of our interest in the ET reactions of peroxides and
endoperoxides,47,49-52,54-56 we investigated the ET chemistry
of two model aromatic endoperoxides, namely, 9,10-diphenyl-
(43) Save´ant, J.-M. In AdVances in Physical Organic Chemistry; Tidwell, T.
T., Ed.; Academic Press: London, 2000; Vol. 35, pp 117-192.
(44) Save´ant, J.-M. In AdVances in Electron-Transfer Chemistry; Mariano, P.
S., Ed.; JAI Press: Greenwich, CT, 1994; pp 53-115.
(45) Save´ant, J.-M. J. Am. Chem. Soc. 1992, 114, 10595-10602.
(46) Save´ant, J.-M. J. Am. Chem. Soc. 1987, 109, 6788-6795.
(47) Workentin, M. S.; Maran, F.; Wayner, D. D. M. J. Am. Chem. Soc. 1995,
117, 2120-2121.
(48) Antonello, S.; Musumeci, M.; Wayner, D. D. M.; Maran, F. J. Am. Chem.
Soc. 1997, 119, 9541-9549.
(49) Donkers, R. L.; Maran, F.; Wayner, D. D. M.; Workentin, M. S. J. Am.
Chem. Soc. 1999, 121, 7239-7248.
(50) Magri, D. C.; Workentin, M. S. Org. Biomol. Chem. 2003, 1, 3418-3429.
(51) Workentin, M. S.; Donkers, R. L. J. Am. Chem. Soc. 1998, 120, 2664-
2665.
(52) Donkers, R. L.; Workentin, M. S. Chem.-Eur. J. 2001, 7, 4012-4020.
(53) Maheshwari, K. K.; de Mayo, P.; Wiegand, D. Can. J. Chem. 1970, 48,
3265-3268.
(37) Willemart, A. Bull. Soc. Chim. Fr. 1937, 510-517.
(38) Willemart, A. Bull. Soc. Chim. Fr. 1937, 357-363.
(39) Willemart, A. Bull. Soc. Chim. Fr. 1938, 557-563.
(40) Southern, P. F.; Waters, W. A. J. Chem. Soc. 1960, 4340-4346.
(41) Drews, W.; Schmidt, R.; Brauer, H.-D. Chem. Phys. Lett. 1980, 70, 84.
(42) Klein, A.; Kalb, M.; Gudipati, M. S. J. Phys. Chem. A 1999, 103, 3843-
3853.
(54) Donkers, R. L.; Workentin, M. S. J. Phys. Chem. 1998, 102, 4061-4063.
(55) Magri, D. C.; Donkers, R. L.; Workentin, M. S. J. Photochem. Photobiol.,
A 2001, 138, 29-34.
(56) Donkers, R. L.; Tse, J.; Workentin, M. S. J. Chem. Soc., Chem. Commun.
1999, 135-136.
9
J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004 1689

A R T I C L E S
Donkers and Workentin
9,10-epidioxyanthracene, DPA-O2 and 9,10-dimethyl-9,10-
epidioxyanthracene, DMA-O2. Our other studies indicated that
upon ET, it is the O-O bond that is the dissociating bond.
However, part of the initial interest in the ET chemistry of DPA-
O2 was to investigate whether the dual reactivity observed in
the thermal and, in particular, photochemical studies of aromatic
endoperoxides extended to ET-initiated reactions. A pertinent
example addressing this question for aromatic endoperoxides
was reported by Amatore and Brown.57 In an elegant study
emphasizing paired electrosynthesis on a femtoliter scale, they
proposed that the endoperoxide 9,10-dihydro-9,10-epidioxyan-
thracene is the initial product formed in the reaction between
the anthracene radical cation and superoxide anion. They
proposed that the product ultimately isolated, 9,10-anthracene-
dione, is formed by ET-initiated O-O fragmentation of 9,10-
dihydro-9,10-epidioxyanthracene, although they did not provide
any mechanistic details of the ET process itself.
Figure 1. Cyclic voltammogram at 0.2 V‚s-1 of 2 mmol‚L-1 DMA-O2 in
DMF with 0.1 mol‚L-1 TEAP. Dashed line is in the presence of 4 mmol‚L-1
2,2,2-trifluoroethanol.
In a previous communication,56 we reported that ET to DPA-
O2 results in a dissociative ET fragmentation of the O-O bond
to yield a distonic radical anion (•O-R-R-O-), similar to our
observations with other endoperoxides.51,52,54 However, in the
cyclic voltammogram (CV) of the reduction of DPA-O2, the
redox process of 9-phenoxy-10-phenyl anthracene (PPA) was
observed. We proposed that this product arose from an
unprecedented O-neophyl-type rearrangement of the distonic
radical anion, •O-R-R-O-, in competition with its reduction
to yield the corresponding 9,10-diol, DPA-(OH)2. The relative
yield of the products from these competing pathways, deter-
mined from preparative scale electrolyses, depended on the
potential of the reduction. However, specific details of the
mechanism were not elucidated. In this paper, we now present
the results of an extensive study of the heterogeneous and
homogeneous ET reduction of DPA-O2 and DMA-O2. The
results allow for a unified description of the mechanism of
reduction of DMA-O2 and, more specifically, DPA-O2. In
addition to providing an important new example of dissociative
ET to a biologically important class of compounds, the analysis
of the mechanism suggests that systems such as these can be
developed as kinetic probes, or “clocks”, for ET reactions of
distonic radical anions.
Figure 2. (a) Cyclic voltammogram of 2.03 mmol‚L-1 DPA-O2 in DMF
with 0.1 mol‚L-1 TEAP. (b) Voltammograms switching before reversible
product wave. (c) Voltammogram that cycles through the reversible couple
of the PPA product. (d) Voltammogram that cycles through the reversible
couple of the PPA product after holding the electrode potential at the
negative vertex -2.1 V vs SCE for 30 s. (e) Voltammogram in the presence
of 10 equiv of 2,2,2-trifluoroethanol. ALL CVs were measured at 0.2 V‚s-1
.
respectively, and for DPA-O2 the potentials are -1.06 and
-1.21 V vs SCE. As the scan rate is increased, the Ep values
shift to more negative values and the peaks become broader.
The peak heights that are normalized for scan rate also decrease
as the scan rate is increased. The transfer coefficient (symmetry
factor), R, is a measure of how the activation free energy, ∆Gq,
varies as a function of the driving force for ET, ∆GET,58 namely,
R ) ∂∆Gq/∂∆GET. Average values of R are calculated from the
shift of the peak potential with scan rate, and they are 0.25 and
Results
Cyclic Voltammetry of DPA-O2 and DMA-O2. Electro-
chemical reduction of DMA-O2 and DPA-O2 is investigated
using cyclic voltammetry at a glassy carbon working electrode
in acetonitrile (ACN) and N,N-dimethylformamide (DMF)
solutions containing 0.1 mol‚L-1 tetraethylammonium perchlo-
rate; representative voltammograms are shown in Figures 1 and
2, respectively. For DMA-O2, the peak potentials (Ep) at 0.1
V‚s-1 are -1.27 and -1.34 V vs SCE in ACN and DMF,
(58) More correctly, the R, E°, and khet values obtained are “apparent values”
because they are not corrected for the ohmic drop due to charging of the
electric double layer. For example R is related to its apparent value by R
) Rapp{1 - (dφq/dE)}. Since the double layer properties of glassy carbon
are not well-known, the correction cannot be made. However, it has been
suggested that the correction is small or negligible. Regardless, the apparent
data obtained has been shown to provide a reasonable estimate of the actual
parameters.
(57) Amatore, C.; Brown, A. R. J. Am. Chem. Soc. 1996, 118, 1482-1486.
9
1690 J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004

ET Reduction Mechanism of Anthracene Endoperoxides
A R T I C L E S
Table 1. Voltammetric Data for the Reduction of DPA-O2 and
DMA-O2 in DMF and ACN at 25 °C with 0.1 mol‚L-1 TEAP
Electrolyte Measured by Cyclic Voltammetry at a Glassy Carbon
Working Electrode
combined preparative electrolysis and voltammetric analysis are
consistent with a dissociative ET reduction of the O-O bond
to form the distonic radical anion, followed by ET reduction to
formally yield the dialkoxide dianion. The poorly defined
oxidation waves are in the potential region expected for the
oxidation of the alkoxide anions formed on the forward scan
and correspond to the oxidation of these alkoxide anions.
Determination of the Heterogeneous ET Reduction Kinet-
ics and Thermochemical Parameters. The heterogeneous ET
rate constants (khet) for the reduction of DPA-O2 and DMA-O2
are determined using the method of convolution potential sweep
voltammetry. The theoretical and experimental treatment of this
technique and its application to the determination of thermo-
chemical parameters related to dissociative ET reductions has
been described in detail elsewhere and are also provided in the
Supporting Information.48,60-63 Briefly, the experiment involves
the careful acquisition of background-subtracted linear volta-
mmetric scans of the dissociative ET waves in Figures 1 and
2b followed by evaluation of the waves according to the
convolution integral that gives the convolution current (I), which
is related to the real current (i) through the convolution integral,
versus E curves. The I-E plots have the limiting current
corresponding to the diffusion-controlled conditions. From the
sigmoid-shaped convolution data, the diffusion coefficients (D)
in ACN and DMF with 0.1 mol‚L-1 TEAP for DPA-O2 are
calculated to be 1.2 × 10-5 and 4.0 × 10-6 cm2‚s-1 and those
ν
DPA-O
DMA-O
2
2
(V‚s-1
)
ACN
DMF
ACN
DMF
Ep (V vs SCE)
0.1 -1.06 -1.21 -1.27 -1.34
1.0 -1.17 -1.37 -1.37 -1.44
5.0 -1.21 -1.41 -1.49 -1.53
10
-1.25 -1.47 -1.50 -1.61
96 122 107 117
0.32 0.25 0.26 0.27
Ep vs log ν slope (mV‚decade-1
)
R
app ) (RT/nF)(dln ν/dEp)
R ) 1.85(RT/F)(Ep/2 - Ep)
0.1
1.0
5.0
0.34
0.28
0.25
0.23
1.94
1.89
0.26
0.24
0.20
0.19
2.00
2.11
0.32
0.29
0.27
0.25
1.93
2.07
0.31
0.29
0.28
0.26
1.99
2.01
10
n (F‚mol-1) no acid
n (F‚mol-1) acid
0.27 for DPA-O2 and DMA-O2 in DMF, respectively; similar
values are found in ACN. The corresponding values calculated
from the peak widths decrease with increasing scan rate and
are consistent with the values calculated from the peak potential
shift. Values of R in this range indicate that the initial ET
controls the electrode kinetics and are used as a diagnostic
criterion to suggest a concerted dissociative ET, where the ET
and bond cleavage occur simultaneously.43,44 A summary of the
peak potentials, peak widths as a function of scan rate, and R
values along with the equations used to calculate them is
provided in Table 1.
for DMA-O2 are 1.8 × 10-5 and 8.1 × 10-6 cm2‚s-1
,
respectively.64 The generally smaller D value for DPA-O2
compared to DMA-O2 is a reflection of a significant contribution
to the molecular volume from the phenyl rings at the 9 and 10
positions. When the back electrode process is not observed (as
in the present case), the khet is related to i and I by the equation
ln khet ) ln D1/2 - ln{[Ilim - I(t)]/i(t)t}, and this khet can be
evaluated at each potential. Activation-driving force relationships
are determined from the convolution data that relate khet to the
electrode potential. The curvature of the activation-driving force
relationship is then used to quantitatively determine how the R
changes as a function of the electrode potential, where R )
-(RT/F)(d ln khet/dE).48 Figures of the convolution data
extracted from voltammetric data are provided in the Supporting
Information. The R values determined from these data are
consistently below 0.32 and decrease as potential moves toward
negative potentials, consistent with the average R values reported
in Table 1. Several thousand determinations of R are determined,
and the linear trend of these values as a function of potential
(or reaction energy) is evaluated and expressed as R ) 0.73 +
0.40E for DPA-O2 and R ) 0.67 + 0.29E for DMA-O2. With
these linear equations, the standard dissociative reduction
On the backward, positive potential scan, poorly defined
irreversible oxidation waves are observed at -0.12 and -0.19
V vs SCE at 0.5 V‚s-1 in ACN and DMF for DMA-O2. Similar
broad and undefined waves are observed for DPA-O2 in ACN
at -0.34 and -0.24 V vs SCE. These oxidation peaks
disappeared in the presence of nonnucleophilic acids such as
2,2,2-trifluoroethanol or acetanilide. Unfortunately, scanning
through these peaks results in poisoning of the electrode surface,
and the oxidation waves are unable to be quantified further.
The results of coulometric studies from preparative scale
electrolyses carried out in the potential regions of the broad
irreversible waves on a reticulated vitreous carbon working
electrode are summarized in Table 1. At these potentials,
electrolysis leads to the consumption of approximately 2
F‚mol-1, or two-electron equivalents, in either solvent in the
presence and absence of non-nucleophilic weak acids, TFE, or
acetanilide. For DMA-O2, workup and isolation of the elec-
trolysis mixture quantitatively yield the corresponding cis-diol,
namely, cis-9,10-dimethyl-9,10-dihydro-9,10-dihydoxyanthracene,
DMA-(OH)2. In the electrolysis of DPA-O2, a 97% yield of
the cis-diol, cis-9,10-diphenyl-9,10-dihydroxyanthracene, DPA-
(OH)2, is isolated. The remaining 3% of the yield is PPA. The
yield of these two products is independent of the applied
potential between -1.1 and -1.5 V vs SCE. The molecular
structure of PPA is verified using X-ray crystallography, and
the relevant data is included in the Supporting Information.
The broad irreversible waves observed for the reduction of
DMA-O2 and DPA-O2 are very similar to those previously
observed for the reduction of the endoperoxides ascaridole,
dihydroascaridole, and artemisinin studied by us51,52,54,55 as well
as acyclic peroxides studied by us47,49,50 and others.48,59 The
potentials, EoROOR/ ORRO-, for each is estimated by extrapolating
•
to the potential value at R ) 0.5 that is the value defined for
∆GEoT ) 0 for most outer-sphere or dissociative ETs; the
EoROOR/ ORRO values are reported in Table 2. Values of the
•
-
(60) Imbeaux, J. C.; Save´ant, J.-M. J. Electroanal. Chem. 1973, 44, 169-187.
(61) Save´ant, J.-M.; D, T. J. Electroanal. Chem. 1975, 65, 57.
(62) Antonello, S.; Maran, F. J. Am. Chem. Soc. 1998, 119, 12595-12600.
(63) Maran, F.; Workentin, M. S.; Wayner, D. D. M. In AdVances in Physical
Organic Chemistry; Tidwell, T. T., Richard, J. P., Eds.; Academic Press:
San Diego, CA, 2001; Vol. 36, pp 85-166.
(64) The change from ACN to DMF is expected to decrease the D value by a
value equal to the square root of the ratio of the two viscosities, which is
1.53. The ratio of D obtained with DMA-O2 is 1.49 and DPA-O2 is 1.73.
The latter value is a bit high, and we believe it is the result of the D value
in DMF being underestimated because of some interaction between the
radical with the solvent.
(59) Kjær, N. T.; Lund, H. Acta Chem. Scand. 1995, 49, 848-852.
9
J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004 1691

A R T I C L E S
Donkers and Workentin
Table 2. Summary of the Thermochemical Parameters for DPA-O2 and DMA-O2 along with Three Model Compounds: Ascaridole,
Di-tert-butylperoxide, and Bis(triphenylmethyl)peroxide
DPA-O
DMA-O
ascaridolea
(CH ) CO−OC(CH ) b,c
Ph3CO−OCPh3c
-1.13
2
2
3 3
3 3
EoROOR/ ORRO- (V vs SCE)
-0.56
-0.56
-1.3
-1.48 (DMF)
-1.55 (ACN)
-6.1
2.72
13.08
•
log kohet (cm s-1
reffd (Å)
)
-4.9
3.68
7.19
10.4
16.3e
-4.4
3.92
9.83
13.1
23.2e
-5.9
2.76
11.9
13.9
28.5
3.83
∆Gqo (kcal‚mol-1
)
λ
(Marcus)f (kcal‚mol-1
)
14.1
37.3
10.0
31.4
BDE (estimated) (kcal‚mol-1
)
a
b
c
d
e
Reference 52. References 47 and 49. Reference 48. Reference 61. Determined by using the Marcus λ and the intrinsic barrier predicted using
the Save´ant model. We believe these to be lower limits. λ ) {(e2/4πꢀo)(1/ꢀop - 1/ꢀs)(1/nreff)} where e is the charge of the electron, ꢀop and /ꢀs are the high
frequency and static dielectric constant of the solvent, reff is the effective radius, and n ) 4.
f
standard heterogeneous rate constant, kohet, are obtained by
expanded potential region, there is a second reduction that is
much smaller in current than the initial broad reduction wave.
This second reduction wave, which we label as R1, becomes
increasingly more broad and less defined and shifts to negative
potential as the scan rate is increased until it completely
disappears at a scan rate of 75 V‚s-1. On the return scan a well-
defined oxidation peak is observed; this we label as O2. In Figure
2c, the forward scan, shown with the dashed line, is the same
as the forward scan in Figure 2a. The return scan, represented
with the solid line in Figure 2c, passes through peak O2 and
then switches potential direction again. On the second scan in
the negative direction, a new reduction peak is observed, R2,
that is clear and well-defined. Qualitatively, it appears that R1
is related to the reversible couple, R2 and O2. At the lowest
scan rates investigated, the peak potential of R1 is the same as
the peak potential of R2; however, the shape of the peaks and
their behavior with increased scan rate are different. Figure 2d
shows a cyclic voltammogram that starts at the potential of
∼ -2.1 V and then scans through O2 and R2. Under this
condition, R2 and O2 are well-defined at all scan rates, and from
analysis of peak data they represent a redox couple of a single
compound with standard potentials of -1.76 and -1.80 V vs
SCE in ACN and DMF, respectively. When preparative
electrolysis is carried out at the potential equivalent to R1 or
R2, a quantitatiVe yield of PPA is isolated; the redox couple
R2/O2 corresponds to the reduction and oxidation of PPA. The
amount of the PPA formed in the cyclic voltammetry experi-
ment (estimated from the peak current) in Figure 2d depends
on the scan rate and the time the potential is held at values
more negative than ca. -1.9 V vs SCE; the amount of product
reaches a maximum value for times greater than 25 s.
extrapolation of the activation-driving force relationship to the
newly determined EoROOR/ ORRO value. Values for kohet for
•
-
concerted dissociative ET reactions are determined to be 1.3 ×
10-5 and 4.0 × 10-5 cm‚s-1 for DPA-O2 and DMA-O2,
respectively. These small values support the rate-determining
nature of the initial ET. The consistency of the kohet and E°
values are checked by digital simulation of the cyclic voltam-
mograms, which adequately reproduce the experimental curves.
Application of Save´ant’s model of concerted dissociative ET
(eq 3) to the trend of the R data with electrode potential provides
the intrinsic free-energy barrier of the reaction, ∆Gqo. This
model can be further broken down into contributions from the
bond dissociation energy (BDE) and solvating energy, λ, such
that ∆Gqo ) (λ + BDE)/4. From the slopes of the equations for
R above, the ∆Gqo values for DPA-O2 and DMA-O2 are
evaluated to be 7.14 and 9.83 kcal‚mol-1, respectively. Because
of the nature of the experiments and the inherent errors (100-
200 mV), we believe these are lower limits. There are no reliable
estimates for either the BDE or solvating energies, λ, for either
of these systems available in the literature. It is possible to
estimate the λ using the solvent cavity models of Marcus.44
A
summary of the equation used and the value determined is
shown in Table 2. The Marcus estimation is used selectively
over other models of Hush and Kojima-Bard because the latter
two are shown in other dissociative systems to significantly
overestimate λ and thus lead to underestimates of the BDE value.
DPA-O2, for example, has BDE values of 7.96 and 13.3
kcal‚mol-1, respectively, using the Hush and Kojima-Bard
values. Using the λ values of Marcus’s method, along with
∆Gqo, we calculate estimates of 16.3 and 23.2 kcal‚mol-1 for
the BDE of DPA-O2 and DMA-O2, respectively. Analysis of
the intrinsic barrier for these endoperoxides is consistent with
the differences in the BDE values and the larger calculated
solvating energies for DMA-O2. The effective solvating radius
(reff)65 for DMA-O2 is much smaller than DPA-O2, and this is
reflected in higher values of the solvating energy. The larger
reff for DPA-O2 is a result of the two outer phenyl rings that
can provide extra shielding to any charge formed.
The amount of PPA observed in the voltammogram, indicated
by the current measurement of the R2/O2 couple in the cyclic
voltammetry, is investigated in the presence of several different
(65) The solvating energy, λhet, was estimated using solvent cavity model by
Marcus and the effective radii approach for concerted dissociative ET
systems. In these expressions, the heterogeneous reorganization energy
depends only on the structure of the endoperoxide. It is crucial that the
radius entered into these equations accurately represents the amount of
volume that will require solvating upon electron transfer. The radii of DPA-
O2 and DMA-O2 were determined using the density of the compound and
the diffusion coefficients determined from the convolution experiment using
the Stokes-Einstein equation. These radii were then transformed into their
effective radii using the equation: r ) [rB(2rAB - rB)/rAB] where rAB is
the radius of the starting molecule and rB is an estimate of the radius of
the fragment receiving the charge. The developing charge in the ET upon
O-O bond cleavage is expected to develop in only part of product, and
the uncharged portion aids in screening the charge. In the case of these
endoperoxides, the uncharged fragment is quite large and therefore the
effective radii are expected to be significantly smaller than the uncorrected
radii. The size of the reduced fragment is approximated by the triphenyl-
methyl alkoxide anion for DPA-O2 and the 1,1-diphenylethyl alkoxide anion
for DMA-O2. Both are similar to the anion that is generated in the ET.
2
λ + BDE
∆GET
(λ + BDE)
∆Gq )
1 +
(3)
(
)
4
Cyclic Voltammetry of DPA-O2. The most intriguing aspect
of the cyclic voltammetry of DPA-O2 that is not seen with other
endoperoxides or peroxides we have studied to date is the
observation of a reducible product at a more negative potential
than the dissociative reduction wave (Figure 2a). In this
9
1692 J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004

ET Reduction Mechanism of Anthracene Endoperoxides
A R T I C L E S
Table 3. Summary of the Homogeneous ET Reduction of DPA-O2, the Yield of PPA product in ACN and DMF, and the amount of charge
consumed in the reduction
E° (V)c
PPA yield (%)
donor
ACN
DMF
ACN
DMF
n (F‚mol-1
)
9-phenoxy-10-phenylanthracenea
2-nitrobiphenyl
nitrobenzene
2-nitroacetophenone
2,3-dimethylanthraquinone
anthraquinone
-1.78
-1.16
-1.10
-0.96
-0.93
-0.87
-0.81
-0.77
-0.72
-0.31
-1.85
-1.19
-1.10
-0.97
-0.90
-0.84
-0.79
-0.71
-0.69
-0.26
100
95
97
92
97
97
0
29
39
0
100
91
94
92
56
38
3
9
15
0
1.99
2.07
0.95
1.02d
1.07
4-nitrobenzonitrile
benz-a-anthracene-7,12-dione
tetrakis-4-nitrophenyletheneb
11,11′,12,12′-tetracyano-anthraquinodimethaneb
a
b
c
d
Based on the peak height from cyclic voltammetry experiments compared to that of an authentic sample. Two-electron donor. V vs SCE.
Difficult to reproduce and quantify due to some loss of mediator during the experiment.
Figure 4. Yield of PPA from the homogeneous reduction of DPA-O2 as
a function of donor oxidation potential in (O) ACN and (b) DMF.
the peak current when acetanilide, water, or methanol is added
at concentrations up to 10 equiv. The peak current drops sharply
after the addition of 2 equiv of TFE and phenol, but less abruptly
for HFIP and DBP, which did not completely eliminate the
formation of PPA even after the addition of 10 equiv.
In Figure 3b, the peak current of R2 is plotted as a function
of scan rate for several different concentrations of TFE. For 1
and 2 equiv, there is no change in the peak current at any scan
rate, indicating that at these concentrations there is no effect
on the production of PPA. Once the concentration is increased
above 2 equiv, the peak current drops sharply. This suggests
that there is a competitive protonation process inhibiting the
formation of PPA. After the addition of 5 equiv of TFE there
is essentially no PPA at any scan rate, indicating trapping of
an intermediate product.
Figure 3. (a) Variation of the R2 peak height normalized for scan rate in
a cyclic voltammogram of DPA-O2 as a function of acid concentration.
(2) Acetanilide, (9) 2,6-di-tert-butyl phenol, (b) 1,1,1,3,3,3-hexafluoro-
2-propanol, (O) 2,2,2-trifluoroethanol, and (0) phenol. Water and methanol
were also used as acids and are coincident with the acetanilide data but
were left off the plot. (b) Variation of the cyclic voltammetry R2 peak height
(normalized for scan rate) as a function of scan rate in the presence of (0)
0, (b) 1, (9) 2, (2) 3, ([) 5, (1) 6, and (4) 8 equiv of 2,2,2-trifluoroethanol.
Homogeneous Reduction of DPA-O2. The yield of PPA in
the heterogeneous reduction of DPA-O2 is found to be potential-
dependent. For this reason, the reduction of DPA-O2 by a variety
of homogeneous donors of varying oxidation potentials is
studied in both ACN and DMF. This permitted the study of the
ET to be extended to a broader free-energy range and allowed
the nature of the donor to be changed. Aromatic radical anions
(D•-) and dianions (D2-) are used as outer sphere electron
donors. The homogeneous donors chosen for study along with
their standard potentials (E°) and the yield of PPA from the
product studies are summarized in Table 3. The yield of PPA
as a function of the oxidation potential of D•- is also shown
graphically in Figure 4. This figure illustrates that the formation
of PPA also depends on the oxidation potential of the
organic acids. Acids with wide-ranging pKa values and structural
characteristics are chosen to determine a structure-reactivity
relationship for the protonation reactions. The acids studied are
2,2,2-trifluoroethanol (TFE), 1,1,1,3,3,3-hexafluoro-2-propanol
(HFIP), acetanilide, methanol, water, phenol, and 2,6-di-tert-
butylphenol (DBP). The results indicate that the yield of PPA
depends on the structure, pKa, and concentration of the acid.
Figure 2e shows a voltammogram of DPA-O2 acquired in the
presence of 10 equiv of TFE. Under this condition, the initial
broad dissociative wave is unchanged compared to that shown
in Figure 2b, but the R1, R2, and O2 product waves are
essentially absent. Figure 3a shows the dependence of the peak
current of R2 measured at a scan rate of 0.2 V‚s-1 as a function
of relative concentration of these acids. There is no change in
9
J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004 1693

A R T I C L E S
Donkers and Workentin
Scheme 2. Mechanistic Pathways for the Electron Transfer Reduction of DPA-O2 and DMA-O2
Discussion
homogeneous donor. In none of the homogeneous reduction
experiments is there evidence for the 9,10-diol, DPA-(OH)2,
the major product in the heterogeneous reduction. Under
homogeneous reduction conditions at potentials more negative
than -0.90 V vs SCE, the major product isolated is PPA in
high yield. PPA, which is the minor product in the heterogen-
eous reduction, is also used as one of the homogeneous donors.
When used as a homogeneous donor, the yield of PPA formed
from DPA-O2 is quantitatively determined by the relative
increase in the current measured for the PPA couple. This has
ramifications on the mechanism of DPA-O2 reduction that will
be discussed later.
The discussion of the mechanism of reduction of DPA-O2
will focus on explaining cyclic voltammetry experiments shown
in Figure 2. The broad irreversible two-electron wave corre-
sponds to the dissociative reduction of the O-O bond, and its
shape is governed by the slow ET kinetics. When heterogeneous
electrolysis was carried out at a potential corresponding to the
broad dissociative wave, 97% of the isolated material was the
cis-diol, DPA-(OH)2. This is consistent with a second hetero-
geneous ET to the alkoxyl radical portion of the distonic radical
ion (or its protonated form) to form the dialkoxide dianion (or
the 9-hydroxyl-10-alkoxide) product that is subsequently pro-
tonated upon workup according to path A in Scheme 2. The
remaining 3% of the yield was PPA. This is in contrast to the
heterogeneous reduction of DMA-O2 that quantitatively yields
its corresponding cis-diol with no products formed that alter
the 9,10-dimethyl anthracene carbon skeleton. The formation
of PPA is consistent with an O-neophyl-type rearrangement of
the alkoxyl radical portion of the distonic radical anion to form
a carbon-centered radical (X) as depicted in path B of Scheme
2.
There are two limiting mechanisms possible for the dissocia-
tive ET reduction of the O-O bond. One is the concerted
dissociative ET process, where electron uptake and bond
breaking occur in a single step, as has been reported for a
number of alkyl peroxides and endoperoxides. An alternative
stepwise pathway involves ET to form a bound radical anion
intermediate that fragments in a second step. As we have already
stated, the voltammetric criteria are consistent with those
generally accepted for a concerted dissociative ET.
At potentials less negative than -0.81 V vs SCE, the
homogeneous reaction produces no PPA, but a number of other
products including 9-hydroxyl-9-phenylanthrone, HPA, and
other products derived from radical coupling. In this potential
region, considerable mass is lost, and the recovered yields are
always much lower than expected. Mixtures of products,
including PPA in varying yields, are observed between the two
potential regions, as summarized in Table 3 and illustrated in
Figure 4. This allowed for the determination of the number of
electrons involved in the reduction of DPA-O2; for donors with
oxidation potentials more negative than -0.95 V, the charge
consumed is 2 F‚mol-1 (two-electron equivalents), and for
donors more positive than -0.81 V, the 1 F‚mol-1 (one-electron
equivalent) process occurs.
For comparison to the electrochemical homogeneous reduc-
tions, the reduction of DPA-O2 and DMA-O2 is also studied
using ferrous chloride as a homogeneous chemical electron
donor. These experiments result in a quantitative conversion of
DPA-O2 to HPA and phenol, and DMA-O2 to 9-methyl-9-
hydroxyanthrone (HMA). A catalytic amount of ferrous chloride
is required for the conversion of DPA-O2, whereas a stoichio-
metric amount is required to convert DMA-O2.
If we assume a concerted reduction mechanism, then the
formation of PPA arises from a competitive O-neophyl-type
rearrangement of the alkoxyl radical portion of the distonic
radical anion with its reduction. By way of comparison, the
reduction potential of Ph3C-O• (E°
-) is -0.03 V vs
•
Ph3CO /Ph3CO
SCE.48 If we reasonably assume a similar reduction potential
9
1694 J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004

ET Reduction Mechanism of Anthracene Endoperoxides
A R T I C L E S
for the distonic radical anion (•O-R-R-O-, or its protonated
form •O-R-R-OH), then it is being generated at the electrode
at potentials more negative than -1.2 V, a favorable free energy
of reaction of about -27 kcal/mol. If the rearrangement of the
radical portion of the distonic radical anion intermediate is
considered to be analogous to the O-neophyl-type rearrangement
of the Ph3C-O• radical to the R-phenoxydiphenylmethyl radical
investigated by Falvey et al., then the minimum rate constant
is considered. Because the BDE values are determined from
the measured intrinsic barrier and an estimate of the λ (i.e., BDE
) 4∆Gqo - λ), the larger the estimated value of λ the lower the
calculated BDE will be. Some stepwise character in the ET
would also act to give low estimates of the BDEs using the
approach here because in general, stepwise ET rate constants
are larger and have a smaller intrinsic barrier than their concerted
counterparts. Thus, we believe the BDE values reported here
are lower limits and may be underestimated by a few kcal/mol.
However, the general trend for acyclic peroxides is that phenyl
substitution acts to lower the BDE and BDFE values.48,63 For
example, the values for bis(triphenylmethyl)peroxide are 7
kcal‚mol-1 lower compared to its non phenyl substituted
analogue, di-tert-butyl peroxide which has a BDE of 37
can be assumed to be 5.9 × 1010 s-1 66
. By using this value as
an estimate of the rate of the rearrangement, then the observation
of PPA in the heterogeneous reduction can serve as a clock for
the rate constant of the second heterogeneous electron transfer
(Scheme 2, path B). From the yield of DPA-(OH)2 and
assuming that the •O-R-R-O- is formed within the electrode
outer Helmholtz plane, which in DMF is ca. 5 × 10-8 cm, an
estimate for khet is on the order of 100 cm s-1. This rate con-
stant range is small for such a highly driven process. An
unusually high reorganization energy associated with ET to
•O-R-R-O- would slow the second ET, thus allowing the
O-neophyl-type rearrangement to compete, although there is no
apparent structural reason for this. Regardless, in the concerted
ET mechanism we propose that the small amount of PPA
produced resulted from a small amount of O-neophyl rearrange-
ment along path B in Scheme 2.
Alternatively, in the stepwise mechanism, the voltammetric
behavior can be described using an ECE mechanism (where E
is an electrochemical step and C is a chemical step) with an
initial ET step to form a bound radical anion intermediate that
rapidly fragments in a second chemical step with a rate constant
kfrag to form the distonic radical anion that can be reduced
quickly to form the dialkoxide dianion (or HO-R-R-O-). This
stepwise mechanism can also account for the formation of PPA
resulting from a competitive O-neophyl-type rearrangement of
distonic radical anion to form the carbon-centered radical if kfrag
is large enough that the majority of the distonic radical anion
is generated (•O-R-R-O-) at the electrode at the potential
where the second ET is a highly driven process and be reduced
very quickly. For the stepwise mechanism, the product distribu-
tion of DPA-(OH)2 and PPA would be an indication of the
kcal‚mol-1 47
. The previously studied endoperoxides ascaridole
and dihydroascaridole are reasonable nonphenyl analogues of
the DPA-O2 and DMA-O2 endoperoxides and a similar drop
in the bond energy values is noted in this series (Table 2). It
should also be noted that the substituent effect on differ-
ences in the E° values between (CH3)3COOC(CH3)3 and
(Ph)3COOC(Ph)3 is not unreasonable with the values of DPA-
O2 and DMA-O2 compared to ascaridole.
For DPA-O2 under homogeneous reduction conditions, the
isolated products are consistent with quantitative O-neophyl-
type rearrangement; the difference here is that the
•O-R-R-O- is not being generated at the electrode, and
therefore, its reduction is limited by diffusion. PPA is formed
quantitatively in a 2 F‚mol-1 (two-electron equivalents) process
when homogeneous donors with oxidation potentials more
negative than -0.95 V are used. This is consistent with the
carbon-centered radical, X, shown in Scheme 2, being reduced
to its carbanion in this potential region. We should note that
the intramolecular ET from the alkoxide anion to the carbon
radical is an uphill process by about 19 kcal mol-1. Donors
with oxidation potentials positive of -0.81 V undergo a 1
F‚mol-1 (one-electron equivalent) process presumably because
the reduction of the carbon-centered radical does not occur at
these potentials. Therefore, the center point of the transition
region of the data in Figure 4 provides a reasonable estimate of
the standard reduction potential of -0.85 V vs for the carbon-
centered radical.68 This value agrees well with estimates for
other similar benzylic type radicals.69,70 For example, the
reduction potential of the diphenylmethyl (DPM) radical is
-1.07 V vs SCE and that of the 9-phenoxyfluorenyl radical is
-0.57 V vs SCE.71 The slightly more positive value for X
compared to the diphenylmethyl radical may be the result of X
being more rigid, increasing delocalization, or other structure
considerations. Importantly, the fact that PPA was not formed
in the one-electron region of Figure 4 indicates that the
formation of the carbanion is essential to the formation of PPA.
In the homogeneous single-electron reduction of DPA-O2,
the O-neophyl-type rearrangement occurs quantitatively because
ratio of the rate constants for fragmentation to that of the
67
O-neophyl-type rearrangement, kfrag/kneophyl
.
Using the rate
constant of the O-neophyl rearrangement above and the
measured 97:3 DPA-(OH)2/PPA product ratio, we can estimate
kfrag to be 1.8 × 1011 s-1; a value almost on the order of a
molecular vibration and thus at least being close to that being
considered a concerted dissociative ET.
For DMA-O2, the two-electron stoichiometry and quantitative
conversion to the corresponding 9,10-diol are consistent with
what is observed with the reduction of other endoperoxides and
acyclic peroxides studied previously. Note that with this
endoperoxide, the corresponding 1,2 methyl shift or â-scission
reactions do not compete with the reduction of the distonic
radical anion. Although the rate of the â-scission reaction is
not known for Ph2MeCO•, the rate constant of the 1,2 methyl
•
the second ET reduction to O-R-R-O- to ultimately yield
DPA-(OH)2 is limited by a second diffusional encounter by
shift rate is estimated to be 4.4 × 106 s-1 66
;
this provides a
(68) Lund, H.; Daasbjerg, K.; Occhialini, D.; Pedersen, S. U. Acta Chem. Scand.
1997, 51, 135-144.
lower rate limit to compare the second heterogeneous ET.
The BDE values determined here are remarkably low,
especially when the high thermal stability of these endoperoxides
(69) Wayner, D. D. M.; McPhee, D. J.; Griller, D. J. Am. Chem. Soc. 1988,
110, 132-137.
(70) Griller, D.; Simo˜es, J. A. M.; Mulder, P.; Sim, B. A.; Wayner, D. D. M.
J. Am. Chem. Soc. 1989, 111, 7872-7876.
(66) Falvey, D. E.; Khambatta, B. S.; Schuster, G. B. J. Phys. Chem. 1990, 94,
1056-1059.
(67) Amatore, C.; Save´ant, J.-M. Electroanal. Chem. 1981, 123, 221.
(71) Daasbjerg, K.; Pedersen, S. U.; Lund, H. In General Aspects of the
Chemistry of Radicals; Alfassi, Z. B., Ed.; John Wiley & Sons: Toronto,
1999; pp 385-427.
9
J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004 1695

A R T I C L E S
Donkers and Workentin
another donor. We attempted to remove the need for the second
diffusional encounter by using the dianion (D2-) donors that
undergo a two successive oxidation reaction where the second
ET step occurs much more rapidly than the first. The dianions
of the donors chosen, 11,11′,12,12′-tetracyanoanthraquin-
odimethane (TCAQ), tetrakis-4-nitrophenylethylene (TNPE),
and tetrakis-dimethylaminoethylene (TDAE), have structural
characteristics that allow the second oxidation step to occur at
a higher driving force and, hence, faster rate than the first ET.
A more thorough explanation of potential inversion and the
structural requirements needed for this to occur is given in
several accounts by Evans and co-workers.72-75 The quantitative
presence of the rearrangement product when DPA-O2 is reduced
with these two-electron homogeneous donors (and not DPA-
(OH)2) where the second ET reaction occurs under the condi-
tions of potential inversion provides support for the short
vibrational timescale of the rearrangement.
The overall mechanism of homogeneous reduction can also
be summarized by Scheme 2, where the e- is a homogeneous
donor (e.g., D radical anion or dianion). For the donors in the
two-electron reduction region of Figure 4, PPA is produced
quantitatively through a dissociative ET, followed by rearrange-
ment to X, and then through a second ET that leads to the
formation of the dianion product (or semiprotonated equivalent)
that eventually forms PPA upon workup. For donors with
oxidation potentials in the one-electron region, ET leads to the
distonic radical anion that rearranges to the carbon-centered
radical without further reduction; i.e., X does not get reduced.
The proposed mechanism is also consistent with what is
observed in the cyclic voltammetry. On the forward scan in a
cyclic voltammogram, the R1 wave is broad and poorly defined
and moves to negative values as the scan rate is increased until
completely disappearing at 75 V‚s-1. This wave is assigned to
the reduction of PPA that is being introduced very slowly into
the diffusion layer of the electrode. This slow introduction is
indicative of a slow rate constant for production of PPA,
suggesting that there is a slow or rate-determining step in its
production. The studies of the PPA product wave in the presence
of different acids can explain the behavior of the R2 wave and
give information related to the mechanism of formation of PPA.
Under voltammetric conditions, a small proportion of the DPA-
O2 being reduced rearranges, and the resulting carbon-centered
radical is reduced to the carbanion. There are two anionic centers
in this intermediate: an alkoxide anion and the carbanion. The
data suggests that protonation of these two centers is key to
understanding the elimination mechanism and formation of
PPA.
28.2), DBP (17.7, 26.8), TFE (24.1, 34.5), acetanilide (22.3,
32.2), water (31.7, 43.8), methanol (31.5, 41.0), and HFIP (20.4,
30.7). There is no known value for pKB of the alkoxide anion
or the structurally similar triphenylmethyl alkoxide anion. Maran
and co-workers have estimated the pKA of the structurally similar
triphenylmethyl alcohol to be 31.5 in DMF and 43.5 in ACN
based on substituent effects and the acidity scale for alcohols,
and this value was used in considering the particular acids used
in this study.48 The carbanion pKB is also not known. The pKA
of 9,10-dihydroanthracene has been estimated to be 27 in
DMSO,78 and in this case the phenoxy group is expected to
increase this value. On the basis of thermodynamic grounds,
with the exception of water and methanol, the acids were chosen
to safely assume that protonation at both centers of the dianion
is expected to be quantitative.
According to the mechanism in Scheme 2, protonation of the
alkoxide anion portion is required for elimination of hydroxide
and formation of PPA. Protonation of both centers results in
trapping of an intermediate Y, and PPA is not formed directly.
With the estimated pKA values, protonation of the carbanion is
favored over the alkoxide anion, although this will lead to
trapping of the dianion intermediate and stop the formation of
PPA. However, protonation of the carbanion is expected to be
hindered sterically because of the peri interactions of the outer
rings of the anthracene backbone, and protonation requires a
change in geometry of the reacting carbon center from sp2 to
sp3 that adds to the kinetic barrier of this proton transfer
considerably.79 Evidence for the slow protonation of the
carbanion was verified with the different acids chosen, and the
results illustrated in Figure 3 indicate preferential kinetic
protonation of the alkoxide anion that leads to the formation of
PPA via an E1CB-type elimination. The small amount of PPA
that is produced in the absence of any added proton source likely
results from water present in the solvents as an impurity. Even
in the driest experiments the amount of water is still in the 0.5-1
mmol‚L-1 range, a comparable concentration to the 2 mmol‚L-1
DPA-O2. Although water is not a thermodynamically favorable
acid, it is sufficiently acidic to protonate the alkoxide anion
very slowly and allows the irreversible elimination of hydroxide
to occur. Thus, from these experiments it is concluded that the
elimination of hydroxide is the rate-determining step in the
production of PPA and is responsible for the totally irreversible
electrode behavior observed with the R1 wave.
Water, methanol, and acetonitrile are the least thermodynami-
cally favorable acids used to protonate the carbon-centered
anion. The PPA peak height in the voltammetry experiments
has the same behavior when DMF and acetonitrile are used as
the solvent, suggesting that protonation by acetonitrile is not
occurring. The auto protolysis constant of acetonitrile (43.7) is
close to the pKA of water and methanol in acetonitrile, and
therefore it is not surprising that at all concentrations of these
acids there was also no change in the R1 peak height, indicating
that protonation of the carbanion does not occur from these
acids.
To investigate this aspect, several acids were chosen so that
protonation of either center was favorable by at least five pKA
units. The pKA values in DMF and ACN for the acids used
were estimated from their pKA in DMSO76 by applying the
relationship pKA(DMF) ) 1.56 + 0.96 pKA(DMSO)77 and pKA-
(ACN) ) 7.10 + 1.17 pKA(DMSO).47 The acids used and their
corresponding pKa values in DMF and ACN are phenol (18.4,
Phenol and TFE had no effect on the PPA wave until two or
more equivalents were added, which caused the peak to
disappear. It is conceivable that the small size of TFE and phenol
(72) Evans, D. H. Chem. ReV. 1990, 90, 739-751.
(73) Evans, D. H. Acta Chem. Scand. 1998, 52, 194-197.
(74) Evans, D. H.; Hu, K. J. Chem. Soc., Faraday Trans. 1996, 92, 3983-
3990.
(75) Evans, D. H.; Lehmann, M. W. Acta Chem. Scand. 1999, 53, 765-774.
(76) Bordwell, F. Acc. Chem. Res. 1988, 21, 456-463.
(77) Maran, F.; Celadon, D.; Severin, M. G.; Vianello, E. J. Am. Chem. Soc.
1991, 113, 9320-9329.
(78) Parker, V. D.; Tilset, M.; Hammerich, O. J. Am. Chem. Soc. 1987, 109,
7905-7906.
(79) Andrieux, C. P.; Save´ant, J.-M.; Tardy, C. J. Electroanal. Chem. 1999,
476, 81-84.
9
1696 J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004

ET Reduction Mechanism of Anthracene Endoperoxides
A R T I C L E S
Scheme 3. Proposed Mechanism of Fe(II) Reduction of DPA-O2
Scheme 4. Proposed Mechanism of Fe(II) Reduction of DMA-O2
allow them to protonate the alkoxide anion easily and, in the
presence of 2 equiv, protonate the carbanion as well, shutting
off the elimination mechanism by trapping the dianion inter-
mediate Y. In contrast, the larger structural analogues of these
acids, DBP and HFIP, did not quench formation of PPA
completely. This we attribute to these larger acids being
sufficient to protonate the carbanion on thermodynamic grounds
but unable to do so kinetically. Acetanilide and TFE have nearly
identical pKA values, and the only other difference between these
two is their structure and corresponding steric bulk. Yet with
acetanilide, there is no loss of the PPA wave at any of the
concentrations investigated, suggesting that protonation of the
carbanion does not occur. The voltammetry in the presence of
these acids is consistent with kinetic protonation of the alkoxide
anion occurring over the more thermodynamically favorable
protonation of the carbanion. This allows for the irreversible
rate-determining elimination reaction of hydroxide to occur and
to form PPA before the second protonation of the carbanion.
presence of a suitable acid, the carbanion formed is trapped and
stops the E1CB type elimination of hydroxide to form PPA.
Iron(II)-induced fragmentation of the O-O bond of cyclic
peroxides has been well-studied.32,80,81 In the case of DPA-O2,
the results of Fe(II)-induced cleavage is consistent with the
O-neophyl-type rearrangement occurring quantitatively under
these homogeneous conditions. A proposed mechanism for the
reduction of DPA-O2 by ferrous ion accounting for the observed
product is shown in Scheme 3. The O-neophyl-type rearrange-
ment of the initially formed alkoxy radical leads to the carbon-
centered radical. Back ET from the carbon radical to the Fe(III)
can occur to generate a carbocation and Fe(II); this is similar
to that observed in other systems and explains why only a
catalytic amount of the FeCl2 is required. The carbocation then
can undergo a nucleophilic attack by water followed by
elimination of phenoxide (phenol) to give HPA. The Fe(II)-
mediated reduction of DMA-O2 leads to the alkoxy radical that
undergoes a â-scission reaction to extrude a methyl radical and
form the anthrone in a single step as shown in Scheme 4. The
rearrangement rate constant for the 1,2-methyl shift that would
be analogous to the O-neophyl rearrangement is not known and
is assumed not to be competitive with the â-scission reaction
The characteristics of the cyclic voltammetry of DPA-O2
illustrated in Figure 2 can now be explained in terms of the
proposed mechanism. In Figure 2a, as the potential is scanned
through the dissociative wave, a small proportion of the distonic
radical anion produced undergoes the O-neophyl-type rear-
rangement to produce the carbon-centered radical X. This radical
is reduced to the carbanion that undergoes elimination of
hydroxide on protonation of the alkoxide anion to form PPA.
As the potential is scanned to more negative values, the small
amount of PPA made at the electrode surface is converted to
the PPA radical anion, PPA•-, that catalytically reduces the
DPA-O2 that is moving toward the electrode surface. The yield
of PPA observed in the voltammogram is dependent on the rate
of formation of PPA and the amount of time spent at potentials
where it is reduced to its radical anion. This explains why the
peak O2 is larger than R1 and why the yield of PPA in cyclic
voltammetry of DPA-O2 is greater than the 3% yield observed
under preparative reduction conditions between -1.1 and -1.5
V vs SCE. Under the conditions of the experiment in Figure
2d, it is possible to obtain an essentially quantitative yield of
PPA, based on the peak height, by the homogeneous redox
catalysis by PPA•- which is an efficient catalytic ET donor
reducing the incoming DPA-O2 and effectively shut down the
heterogeneous reduction. Under purely homogeneous ET condi-
tions, the O-neophyl-type rearrangement pathway and formation
of PPA occurs quantitatively at <-0.9 V vs SCE. In the
that has a minimum rate constant of 4.4 × 106 s-1 82-85
.
In
contrast, these â-scission reactions do not occur in the hetero-
geneous reduction because the second ET to the alkoxy radical
is very rapid since it is generated at the electrode.
Conclusion
The initial ET reduction DPA-O2 and DMA-O2 follows a
dissociative mechanism with fragmentation of the O-O bond
(80) Szpilman, A. M.; Korshin, E. E.; Hoos, R.; Posner, G. H.; Bachi, M. D. J.
Org. Chem. 2001, 66, 6531-6540.
(81) Wu, W.-M.; Wu, Y.; Yao, Z.-J.; Zhou, C.-M.; Li, Y.; Shan, F. J. Am. Chem.
Soc. 1998, 120, 3316-3325.
(82) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 193-200.
(83) Griller, D.; Ingold, K. U. Acc. Chem. Res. 1980, 13, 317-323.
(84) Arends, I. W. C. E.; Ingold, K. U.; Wayner, D. D. M. J. Am. Chem. Soc.
1995, 117, 4710-4711.
(85) Avila, D. V.; Brown, C. E.; Ingold, K. U.; Lusztyk, J. J. Am. Chem. Soc.
1993, 115, 466-470.
9
J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004 1697

A R T I C L E S
Donkers and Workentin
to yield a distonic radical anion. Under the conditions of ET
and thus at the transition state whereby the O-O bond is
stretched with respect to the equilibrium length, the LUMO of
these molecules has a strong contribution of the σ*-orbital
associated with the O-O bond. The subsequent reactivity of
the distonic radical anion formed on reduction of DPA-O2 can
be controlled by using either homogeneous or heterogeneous
methods. Under homogeneous conditions, the initial dissociative
ET is followed by a rapid O-neophyl-type rearrangement that
leads to the formation of O-insertion products for DPA-O2.
Under heterogeneous reduction conditions, the O-neophyl-type
rearrangement competes with the highly driven second ET
reduction of the distonic radical anion formed at the electrode
surface that yields the corresponding 9,10-diol after protonation.
In either case, after rearrangement, the formation of the carbon-
centered radical leads to the formation of PPA through a cascade
of steps that involves its reduction followed by selective
protonation of the alkoxide anion portion of the singly reduced
intermediate. The reduction of this radical depends on the
potential of the reducing media and allows for the estimation
of its reduction potential (-0.85 V vs SCE). A unified
mechanism that accounts for the differences observed in the
homogeneous and heterogeneous ET chemistry has been
proposed.
the BDE of the O-O bond in DPA-O2 (and DMA-O2); they
represent some of the lowest of any O-O BDE yet reported.86
Therefore, a question that can arise is why does the C-O bond
fragment in preference to the weaker O-O bond under some
conditions? A possible explanation is that there is an extra
kinetic barrier to the fragmentation of the O-O bond related
to an inherent nonadiabaticity due to a failure of the Born-
Oppenheimer approximation near the transition state for O-O
cleavage.87 It is now established that ET to O-O bonded
systems is nonadiabatic,49,52,63,88,89 and evidence for this in DPA-
O2 and DMA-O2 systems is noted by the low magnitude of the
kohet values.
Acknowledgment. This work was supported financially by
the Natural Sciences and Engineering Research Council of
Canada, Canadian Foundation for Innovation, the Ontario
Research and Development Challenge Fund, a Premier’s
Research Excellence Award, and an ADF New Research and
Scholarly Initiatives Award of the University of Western
Ontario. R.L.D. is grateful for an NSERC post graduate
scholarship. Dr. M. C. Jennings performed the X-ray crystal-
lographic analysis, and Doug Hairsine performed the mass
spectroscopic analyses. Professor Flavio Maran is thanked for
his valuable scientific insights.
Under the ET conditions, reduction of DPA-O2 follows
different reaction pathways compared to either thermal or
photochemical activation. For example, there is no evidence for
C-O activation upon ET that is observed in the photochemical
and thermolysis studies of these compounds. By constructing
activation driving force relationships for the heterogeneous ET
process and applying it to the current theories of dissociative
ET, the intrinsic barrier for the bond-breaking ET process was
determined along with several other important and previously
undetermined thermochemical parameters. An interesting aspect
of this is that there appears to be significantly lower solvating
energy required for the reduction of DPA-O2 compared to
DMA-O2 that we attributed to added shielding by the adjacent
phenyl rings. Another outcome of the analysis is estimates for
Supporting Information Available: The Experimental Sec-
tion, including experimental details and spectral characterization
of all compounds and product studies, tables of crystallographic
data, and details of the convolution analysis (PDF, CIF). This
material is available free of charge via the Internet at
http://pubs.acs.org.
JA035828A
(86) Luo, Y.-R. Handbook of Bond Dissociation Energies in Organic Com-
pounds; CRC Press: Boca Raton, FL, 2003.
(87) Butler, L. J. Annu. ReV. Phys. Chem. 1998, 49, 125-171.
(88) Antonello, S.; Formaggio, F.; Moretto, A.; Toniolo, C.; Maran, F. J. Am.
Chem. Soc. 2001, 123, 9577-9584.
(89) Antonello, S.; Crisma, M.; Formaggio, F.; Moretto, A.; Taddei, F.; Toniolo,
C.; Maran, F. J. Am. Chem. Soc. 2002, 124, 11503-11513.
9
1698 J. AM. CHEM. SOC. VOL. 126, NO. 6, 2004