Cyclopropyl conjugation and ketyl anions: When do things begin to fall apart?

Article abstract of DOI:10.1021/ja063857q

Results pertaining to the electrochemical reduction of 1,2- diacetylcyclopropane (5), 1-acetyl-2-phenylcyclopropane (6), 1-acetyl-2-benzoylcyclopropane (7), and 1,2-dibenzoylcyclopropane (8) are reported. While 6^?- exists as a discrete species, the barrier to ring opening is very small (<1 kcal/mol) and the rate constant for ring opening is >10⁷ s^-1. For 7 and 8, the additional resonance stabilization afforded by the benzoyl moieties results in significantly lower rate constants for ring opening, on the order of 10 ⁵-10⁶ s^-1. Electron transfer to 8 serves to initiate an unexpected vinylcyclopropane → cyclopentene type rearrangement, which occurs via a radical ion chain mechanism. The results for reduction of 5 are less clear-cut: The experimental results suggest that the reduction is unexceptional, with a symmetry coefficient α ≤ 0.5, and reorganization energy consistent with a simple electron-transfer process (one electron reduction, followed by ring opening). In contrast, molecular orbital calculations suggest that 5^?- has no apparent lifetime and that reduction of 5 may occur by a concerted dissociative electron transfer (DET) mechanism (i.e., electron transfer and ring opening occur simultaneously). These seemingly contradictory results can be reconciled if the increase in the internal reorganization energy associated with the onset of concerted DET is offset by a lowering of the solvent reorganization energy associated with electron transfer to a more highly delocalized LUMO.

Full text of DOI:10.1021/ja063857q

Published on Web 03/20/2007
Cyclopropyl Conjugation and Ketyl Anions: When Do Things
Begin to Fall Apart?
J. M. Tanko,* Xiangzhong Li, M’hamed Chahma, Woodward F. Jackson, and
Jared N. Spencer
Department of Chemistry, Virginia Polytechnic Institute and State UniVersity,
Blacksburg, Virginia 24061
Received June 1, 2006; E-mail: jtanko@vt.edu
Abstract: Results pertaining to the electrochemical reduction of 1,2-diacetylcyclopropane (5), 1-acetyl-2-
phenylcyclopropane (6), 1-acetyl-2-benzoylcyclopropane (7), and 1,2-dibenzoylcyclopropane (8) are
•
-
reported. While 6 exists as a discrete species, the barrier to ring opening is very small (<1 kcal/mol) and
7
-1
the rate constant for ring opening is >10 s . For 7 and 8, the additional resonance stabilization afforded
5
by the benzoyl moieties results in significantly lower rate constants for ring opening, on the order of 10 -
6
-1
1
0 s . Electron transfer to 8 serves to initiate an unexpected vinylcyclopropane f cyclopentene type
rearrangement, which occurs via a radical ion chain mechanism. The results for reduction of 5 are less
clear-cut: The experimental results suggest that the reduction is unexceptional, with a symmetry coefficient
R e 0.5, and reorganization energy consistent with a simple electron-transfer process (one electron reduction,
•-
followed by ring opening). In contrast, molecular orbital calculations suggest that 5 has no apparent lifetime
and that reduction of 5 may occur by a concerted dissociative electron transfer (DET) mechanism (i.e.,
electron transfer and ring opening occur simultaneously). These seemingly contradictory results can be
reconciled if the increase in the internal reorganization energy associated with the onset of concerted DET
is offset by a lowering of the solvent reorganization energy associated with electron transfer to a more
highly delocalized LUMO.
Introduction
Cyclopropyl groups are unique as alkyl substituents because
of their characteristic bonding features. The Walsh/Sugden
model1 constructs cyclopropane from three sp -hybridized CH2
-3
2
2
fragments, with the sp hybrid orbitals oriented radially toward
the center of the three membered ring, and the three p orbitals
coplanar (constituting a H u¨ ckel and M o¨ bius array, respectively).
The HOMO and LUMO for cyclopropane arise from the latter
and are depicted in Figure 1. The theoretical foundation for the
cyclopropyl group as a good π-electron donor is well established
and attributable to interaction of one of the degenerate cyclo-
propyl HOMOs (ψ1) with an adjacent π system; this interaction
is maximal in the so-called bisected conformation, where the
cyclopropyl p-derived orbitals are fully aligned with the π
system.4 As a result of π-donation, the cyclopropyl group
Figure 1. Depiction of (a) the p-derived molecular orbitals of cyclopropane
and (b) the interaction of the cyclopropyl HOMO with an adjacent π system
in the bisected conformation.
The ability of the cyclopropyl group to act as a π-electron
acceptor to stabilize an electron-rich center is less clear-cut.
Experimental evidence has been presented which suggests the
,5
9
cyclopropyl group can stabilize a negative charge in some cases,
6
7
stabilizes electron-deficient species such as free radicals and
but the effect is not general. In an electrochemical approach
7
carbocations. For other π systems (e.g., carbonyl, vinyl,
to the question, the cyclopropyl group was reported to have no
discernible effect on the half-wave (reduction) potentials of
phenyl), interactions with the cyclopropyl group have been
1
0
detected by predictable geometric distortions measured by X-ray
activated olefins. Clark and Schleyer have concluded that the
cyclopropyl group can act as either a σ- or π-acceptor only in
extreme cases (i.e., with potent electron donors), invoking a
slightly more complicated molecular orbital description of the
8
crystallography, chemical shift changes measured by NMR,
7
and perturbations detected by other forms of spectroscopy.
(
(
(
(
(
(
(
1) Sugden, T. M. Nature (London) 1947, 160, 367-268.
1
1
2) Walsh, A. D. Nature (London) 1947, 159, 712-713.
Walsh model than presented here.
3) Walsh, A. D. Trans. Faraday Soc. 1949, 54, 179-190.
4) Hoffmann, R.; Davidson, R. B. J. Am. Chem. Soc. 1971, 93, 5699-5704.
5) Wiberg, K. B. Acc. Chem. Res. 1996, 29, 229-234.
6) Walton, J. C. Magn. Reson. Chem. 1987, 25, 998-1000.
7) Tidwell, T. T. In The Chemistry of the Cyclopropyl Group; Rappoport, Z.,
Ed.; John Wiley and Sons: New York, 1987; pp 565-632.
(8) Allen, F. H. Acta Crystallogr. 1980, B 36, 81-96.
(9) Perkins, M. J.; Peynircioglu, N. B. Tetrahedron 1985, 41, 225-227.
(10) Baizer, M. M.; Chruma, J. L.; Berger, P. A. J. Org. Chem. 1970, 35, 3569-
3571.
10.1021/ja063857q CCC: $37.00 © 2007 American Chemical Society
J. AM. CHEM. SOC. 2007, 129, 4181-4192
9
4181

A R T I C L E S
Scheme 1
Tanko et al.
vacuum (until evolution of water vapor ceased) prior to use. Tetra-n-
butylammonium perchlorate (TBAP) was prepared by the method of
House21 and recrystallized 4× from ethyl acetate/hexane and vacuum-
22
oven-dried before use. 1,2-Diacetylcyclopropane (5), 2-phenyl-1-
acetylcyclopropane (6),23 and 2-benzoyl-1-acetylcyclopropane (7) (all
mainly trans) were prepared according to published procedures. trans-
24
1
,2-Dibenzoylcyclopropane (8) and all of the mediators used in this
study were obtained from Aldrich.
General. GC/MS was performed on a Hewlett-Packard HP 5890
gas chromatograph interfaced to a HP 5970 low-resolution mass
spectrometer and a HP series computer. High-resolution mass spectral
data were obtained from a VG Analytical model 7070 E-HF double-
focusing magnetic sector high-resolution spectrometer using electron
impact (70 eV) ionization. GC analysis was performed on a Hewlett-
Packard 5890A gas chromatograph equipped with an FID detector and
an HP 3393A reporting integrator. Nuclear magnetic resonance spectra
Our group has been interested in the electrochemistry of
cyclopropyl ketones, particularly with regard to the rate
constants for ring opening of their corresponding radical anions
2-20
(
Scheme 1).1
Although the emphasis of these studies has
been on the ring-opening reaction, we have collected a fair
amount of thermodynamic data pertaining to the issue of the
cyclopropyl group as a substituent and its ability to stabilize a
ketyl anion. Specifically, these experiments have led to good
estimates of the reduction potentials of several cyclopropyl-
containing carbonyl compounds. The fast cyclopropane ring
opening of many of these radical anions enables the use of an
electrochemical method (homogeneous redox catalysis, vide
infra) which allows measurement of the reduction potentials of
some classes of compounds (e.g., aliphatic ketones) which could
not be measured otherwise.1
1
13
( H, C) were obtained on either a Bruker WP 270 MHz, Bruker AM
360 MHz, or a Varian Unity 400 MHz FT NMR spectrometer. Infrared
spectra were recorded on a Perkin-Elmer model 1600 FT-IR spectrom-
25
eter. Molecular orbital calculations were performed using Gaussian 03
(
running on SGI Inferno 2, a super-cluster containing 64 1.6 GHz
2
6
Itanium 2 processors and 256 GB of memory) and/or Spartan ’04
(running on a Windows XP system with an AMD 1.8 GHz XP2200+
processor and 2 GB of RAM).
2-14,19
Electrochemistry. Electrochemical measurements were performed
on an EG&G Princeton Applied Research (EG&G/PAR) model 273
potentiostat/galvanostat interfaced to an MS-DOS computer. The details
To date, we have found no evidence that the cyclopropyl
group can significantly affect the reduction potential of a ketone
or stabilize a radical anion. For example, spiro[2.5]octa-4,7-
dien-6-ones 1 have nearly the same reduction potential as
cyclohexadienone 2.13 Similarly, within experimental error,
methyl cyclopropyl ketone (3) and dicyclopropyl ketone (4) have
13,14,17
of this system were described earlier.
Briefly, a three-electrode
voltammetry cell was used with a glassy carbon working electrode
(GCE), which was fabricated from 0.5 cm diameter glassy carbon rod
(type 1, Alfa Aesar). The GC rod was cut into several 4-5 mm plugs,
which were secured into glass rods with Torrseal-Varian vacuum epoxy
resin (Varian vacuum products), and attached to a Cu brazing rod with
silver two-part conductive adhesive (Alfa Aesar). After being sanded,
the electrode surface was polished with alumina slurry (Buehler) starting
with 1.0 µm grit and decreasing to 0.3 and finally 0.05 µm until a
the same reduction potential.1
4,19
2
mirror finish was obtained. The area was 0.197 cm , determined from
the voltammetric response of ferrocene whose diffusion coefficient in
0
.1 M TBAP/DMF is known.27 The reference electrode was Ag/AgNO
3
In this paper, we report our results pertaining to the
electrochemical reduction of substituted cyclopropyl ketones 5
f 8.
(0.1 M in CH CN, 0.337 V vs SCE). For calibration purposes, ferrocene
3
oxidation occurs at +0.035 V in DMF versus this reference electrode.
A Pt wire coil was used as the auxiliary electrode. Positive-feedback
IR compensation was employed.
The following protocol was followed for the voltammetry experi-
ments: After the working electrode was thorooughly polished with fine
alumina (0.05 µm), it was subsequently rinsed with isopropyl alcohol
and sonicated for 15 min (to remove any residual alumina from the
surface). The electrode was activated by scanning over the potential
range several times at 100 mV/s, and then background voltammograms
(solvent + electrolyte-no substrate) were obtained. Adsorption of the
products formed during the reduction was a problem for all the
substrates, and it was usually necessary to repeat this cleansing process
between runs.
Experimental Section
Materials. N,N-Dimethylformamide (DMF, EM Science, 98%) was
stirred over copper(II) sulfate (Aldrich, 98%) and activated alumina
(Aldrich, neutral, Brockman activity 1) for >3 days and vacuum-
distilled immediately before use. Alumina was flame-dried under
Preparative-scale electrolyses were performed on solutions which
contained 0.2 M TBAP in DMF. A conventional H-cell with two
compartments separated by a medium glass frit was utilized. A gold
foil working electrode (4 cm ) was utilized. All electrolysis experiments
were performed at ambient temperature. Reaction progress was
(
(
(
(
(
11) Clark, T.; Spitzmagel, G. W.; Klose, R.; Schleyer, P. v. R. J. Am. Chem.
Soc. 1984, 106, 4412-4419.
12) Chahma, M.; Li, X.; Phillips, J. P.; Schwartz, P.; Brammer, L. E.; Wang,
Y.; Tanko, J. M. J. Phys. Chem. A 2005, 109, 3372-3382.
2
13) Phillips, J. P.; Gillmore, J. G.; Schwartz, P.; Brammer, L. E., Jr.; Berger,
D. J.; Tanko, J. M. J. Am. Chem. Soc. 1998, 120, 195-202.
14) Stevenson, J. P.; Jackson, W. F.; Tanko, J. M. J. Am. Chem. Soc. 2002,
1
24, 4271-4281.
(21) House, H. O.; Feng, E.; Pert, N. P. J. Org. Chem. 1971, 36, 2371-2375.
(22) Maier, G.; Sayrac, T. Chem. Ber. 1968, 101, 1354-1370.
(23) DePuy, C. H.; Dappen, G. M.; Eilers, K. L.; Klein, R. A. J. Org. Chem.
1964, 29, 2813-2815.
15) Tanko, J. M.; Brammer, L. E., Jr.; Hervas, M.; Campos, K. J. Chem. Soc.
Perkin Trans. 2 1994, 1407-1409.
(
(
(
16) Tanko, J. M.; Drumright, R. E. J. Am. Chem. Soc. 1990, 112, 5362-5363.
17) Tanko, J. M.; Drumright, R. E. J. Am. Chem. Soc. 1992, 114, 1844-1854.
18) Tanko, J. M.; Drumright, R. E.; Suleman, N. K.; Brammer, L. E., Jr. J.
Am. Chem. Soc. 1994, 116, 1785-1791.
(24) Doyle, M. P.; Buhro, W. E.; Dellaria, J. F. Tetrahedron Lett. 1979, 46,
4429-4432.
(25) Frisch, M. J.; et al. Gaussian 03, ReVision C.02.
(26) Spartan ‘04; Wavefunction, Inc.: Irvine, CA, 2004.
(27) Jacob, S. R.; Hong, Q.; Coles, B. A.; Compton, R. G. J. Phys. Chem. B
1999, 103, 2963-2969.
(19) Tanko, J. M.; Phillips, J. P. J. Am. Chem. Soc. 1999, 121, 6078-6079.
20) Tanko, J. M.; Gillmore, J. G.; Friedline, R.; Chahma, M. J. Org. Chem.
(
2
005, 70, 4170-4173.
4182 J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007

Cyclopropyl Conjugation and Ketyl Anions
A R T I C L E S
monitored by GC where necessary. Solution workup consisted of
quenching the cathodic compartment with 1 mL of 5% H
0 mL of water, and extracting with 4 × 50 mL of ether. Ether layers
were combined, washed with water, washed with saturated NaCl
solution, dried over MgSO , and concentrated. Products of bulk
2 4
SO , adding
5
4
electrolysis were all known compounds. Yields were determined by
GC vs an appropriate internal standard; mass balances were ca. 90%.
Results
Direct Electrochemistry. With direct electrochemical
techniques2
8-31
such as cyclic and linear sweep voltammetry
(
CV and LSV, respectively), substrate A (Scheme 2) is reduced
•
-
at an electrode surface generating radical anion A ; kS
represents the standard heterogeneous rate constant for this step
(
rate constant when ∆G° ) 0). This radical anion undergoes
•
-
subsequent ring opening yielding B with rate constant ko.
Either of these steps may be rate-limiting, depending on the
nature of the substrate. Kinetic data are obtained by examining
the effect of sweep rate and substrate concentration on the
electrochemical response. When the chemical step is rate-
limiting, it is possible to obtain information such as the formal
reduction potential of the substrate (E°A/A ), the rate law for
the chemical step, and the rate constant ko. When heterogeneous
electron transfer is rate-limiting, it is possible to measure kS
and the electron-transfer coefficient R. The value of R is a
measure of transition state location, and can be used to
distinguish between stepwise and dissociative electron transfer
Figure 2. Cyclic voltammogram of 1,2-dibenzoylcyclopropane (8, 0.0020
M in 0.5 M n-Bu4NClO4/DMF) at 100 mV/s.
•
-
ments such as the peak or half-peak potential. This method is
particularly useful when electron transfer is rate-limiting. A plot
of E vs the current function ln(Ilim - I(t))/i(t)) permits the
standard heterogeneous rate constant (kS) and the average value
of the transfer coefficient R to be determined (eq 3, assuming
Butler-Volmer kinetics); through eq 4, the potential dependent
heterogeneous rate constant (khet) can be calculated at any point
on the voltammogram (without the Butler-Volmer assumption),
and via differentiation (eq 5), the potential dependence of R
(
DET, vide infra).
Scheme 2
3
3
can be determined. (Note: Because the current function
becomes noisy at the extremes i(t) f 0 and I(t) f Ilim, only
data in the 10-90% range of Ilim were used in these analyses.)
For systems whose kinetics were electron-transfer-controlled,
the voltammograms were further subjected to convolution
1
/2
32
Ilim - I(t)
i(t)
analysis. In this treatment, a convolution voltammogram (plot
of I(t) vs E) is generated in accordance with eq 1, where I(t)
and i(t) are the convolutive and “normal” currents at time t.
The limiting convolutive current, Ilim, is independent of sweep
rate, eq 2 (where n, F, A, D, and C have their usual meanings).
Convolution voltammograms were generated by numerical
integration of the original voltammograms using a simple
BASIC program written in-house. Convolution voltammograms
for 5 and 6 are provided in the Supporting Information.
RT
RnF
D
RT
RnF
E ) E° -
ln
+
ln
(3)
(
)
(
)
kS
Ilim - I(t)
i(t)
1
/2
ln(khet) ) ln(D) - ln
(4)
(5)
(
)
-
∂
ln khet
∂
∆G
∆G°
RT
F
R )
) -
(
)
∂
∂E
Cyclic voltammograms for compounds 5-8 all exhibit the
following characteristics: For each substrate, the reduction wave
is not accompanied by a corresponding oxidation wave (at scan
rates up to 100 V/s), consistent with the expectation that
cyclopropane ring opening occurs rapidly on the time scale of
1
t
i(u)
I(t) )
∫
du
1/2
(1)
(2)
1
/2
0
π
(t - u)
1/2
Ilim ) nFAD C
these experiments. An oxidation wave observed at -600 f
The convolution approach essentially eliminates the mass
transport component of the electrochemical response and, in
principle, allows all the data associated with the voltammogram
to be analyzed, as opposed to relying on single-point measure-
34
-
1200 mV is attributed to oxidation of an enolate anion which
results from the ring-opening reaction. (For a representative
voltammogram, see Figure 2). In each case, the magnitude of
the peak current is consistent with a two-electron reduction,
based upon Ilim (eq 2) and estimates of the diffusion coefficient
based on molecules of similar size. Because these voltammo-
grams are irreversible, any mechanistic and kinetic inferences
(
28) Amatore, C.; Saveant, J. M. J. Electroanal. Chem. 1977, 85, 27-46.
29) Andrieux, C. P.; Sav e´ ant, J. M. In InVestigation of Rates and Mechanisms
of Reactions. Part II, 4th ed.; Bernasconi, C., Ed.; Wiley: New York, 1986;
pp 305-390.
(
(
30) Nadjo, L.; Sav e´ ant, J. M. Electroanal. Chem. 1973, 48, 113-145.
31) Parker, V. D. In Electrode kinetics: Principles and methodology; Bamford,
C. H., Compton, R. G., Eds. Elsevier: Amsterdam, 1986; Vol. 26, pp 145-
(
(33) Bard, A. J.; Faulkner, L. R.; Wiley: New York, 1980, p 236-243.
(34) Daasbjerg, K.; Pedersen, S. U.; Lund, H. In General Aspects of the
Chemistry of Radicals; Alfassi, Z. B., Ed.; Wiley: Chichester, UK, 1999;
pp 385-427.
2
02.
(32) Imbeaux, J. C.; Sav e´ ant, J.-M. J. Electroanal. Chem. 1973, 44, 169-187.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007 4183

A R T I C L E S
Tanko et al.
6).3
8-41
On this basis, E° is estimated to be -2.47 and -2.89
Table 1. LSV Data for Compounds 5-8
∂
E
p
/
∂
log
ν
|
E
p
−
E
p/
2
|
V, for 5 and 6, respectively. Pertinent plots are provided in the
Supporting Information.
a
compound
(mV/decade)
(mV)
R
b
c
d
d
5
6
7
8
-64
-57
-42
-40
119
104
72-90
54-80
0.46, 0.40, 0.43
0.51, 0.46, 0.49
b
c
1
2
F
R )
+
(E - E°)
-
(6)
e
e
0.5f
8
∆Go
0.5f
a
b
c
∂
Ep/∂ log CA ≈ 0. Calculated from ∂ Ep/∂ log ν. Calculated from
Indirect Electrochemistry. With indirect electrochemical
d
e
Ep - Ep/2. Obtained from convolution analysis (eq 3). Increases steadily
with increasing sweeprate. Assumed.
42-46
methods (i.e., homogeneous redox catalysis),
an electron-
f
transfer mediator or catalyst M rather than substrate A is reduced
at the electrode surface (Scheme 3). For this to occur, the
mediator must be more easily reduced than the substrate, (i.e.,
the peak potential of the substrate must be more positive than
are based upon the peak potential (Ep) and peak width (the
difference between the peak and half-peak potential, Ep - Ep/2.)
If the chemical step (ring opening) is rate limiting, the
expected variation of Ep with sweeprate (ν) and concentration
•
-
E°M/M ,) and the reduction of M must be reversible. Reduction
of the substrate occurs in solution (homogeneous) via electron
(
CA) are ∂ Ep/∂ log ν ) -29.6 and ∂ Ep/∂ log CA ) 0 mV/
•-
transfer from the reduced form of the mediator (M .) Effects
decade. The peak width, Ep - Ep/2, is expected to be around 50
mV and invariant with sweeprate. In contrast, if electron transfer
is rate-limiting, ∂ Ep/∂ log ν ) -29.6/R and ∂ Ep/∂ log CA ) 0
mV/decade, where R is the transfer coefficient discussed above.
Another characteristic feature of rate-limiting electron transfer
is that the peak widths are notably greater than 50 mV, with Ep
•
-
of substrate addition on the M/M couple are manifested
experimentally by an increase in peak current and a loss of
reversibility (if catalysis is occurring.) Kinetic control may be
governed by either the homogeneous electron-transfer step (k1)
or the chemical step (ko, Scheme 3.) If the rate of the chemical
step is faster than back electron transfer (ko > k-1 [M],) then
the electron-transfer step is rate-limiting and k1 can be deter-
mined.
-
Ep/2 ) 1.85RT/RF.30
For all the compounds studied, Ep did not vary significantly
with substrate concentration; the other significant features of
the voltammetry are summarized in Table 1; representative
voltammograms are provided in the Supporting Information.
Compounds 5 and 6 are clearly in the regime where the
kinetics is governed by electron transfer. For each compound,
the calculated (average) values of the transfer coefficient R are
similar whether obtained from ∂ Ep/∂ log ν, Ep - Ep/2, or
convolution analysis (vide infra); the significance of these values
will be discussed later. In contrast, compounds 7 and 8 appear
to be under mixed kinetic control: ∂ Ep/∂ log ν falls between
the expected values for a rate-limiting chemical step (30 mV/
decade) and that for rate-limiting electron-transfer (60 mV/
decade, for R ) 0.5).
Scheme 3
If the chemical step is slow relative to back electron transfer
(ko < k-1 [M],) the chemical step is rate-limiting with the
electron-transfer step as a rapid pre-equilibrium. Under these
conditions, the composite rate constant kobs ) kok1/k-1 ) koK1
can be determined; if the reduction potential of the substrate
The voltammograms were subjected to convolution analysis;
for 5, electrode fouling was particularly problematic at low scan
rates, so data analysis was restricted to experiments in the 600-
1
000 mV/s regime. For 6, solvent/electrolyte discharge clearly
•
-
distorted both the conventional and convolution voltammograms,
complicating the determination of Ilim. For this system, Ilim was
obtained by fitting the I vs E curves to a sigmoidal function, as
described by Maran et al.3 Convolution voltammograms for
(E°A/A ) is known, ko can be extracted through application of
eq 7.
5,36
E°M/M•- - E°A/A•-
log K ) - F
(7)
1
(
)
5
and 6 are provided in the Supporting Information.
2.303RT
Plots of E vs the current function ln((Ilim - It)/i(t)) in
The key experimental observable in this method is the ratio
of the voltammetric peak current of the mediator in the presence
and absence of substrate (ip/ipd) at a particular value of γ (the
ratio of substrate to mediator concentration). Elucidation of the
accordance with eq 3 yields R ) 0.43 and 0.49, for 5 and 6,
respectively. Through application of eq 4, heterogeneous rate
constants for electron transfer were determined as a function
of electrode potential; a double-layer correction was not applied
to the data. The ln(khet) vs E data were curve-fit, and first
derivatives calculated (at ca. 10 mV intervals, depending on
scanrate) using Table Curve 2D37 and leading to plots revealing
the potential dependence of R (eq 5). In accordance with Marcus
theory, these R-E plots were linear, and extrapolation to R )
(
38) Antonello, S.; Benassi, R.; Gavioli, G.; Taddei, F.; Maran, F. J. Am. Chem.
Soc. 2002, 124, 7529-7538.
(39) Antonello, S.; Maran, F. J. Am. Chem. Soc. 1997, 119, 12595-12600.
(
40) Antonello, S.; Maran, F. J. Am. Chem. Soc. 1999, 121, 9668-9676.
(41) Donkers, R. L.; Workentin, M. S. J. Phys. Chem. B 1998, 102, 4061-
063.
4
(42) Andrieux, C. P.; Blocman, C.; Dumas-Bouchiat, J. M.; M’Halla, F.; Sav e´ ant,
J. M. J. Electroanal. Chem. 1980, 113, 19-40.
0
.5 (corresponding to ∆G° ) 0) allows an estimate of E° (eq
(
43) Andrieux, C. P.; Hapiot, P.; Sav e´ ant, J.-M. Chem. ReV. 1990, 90, 723-
738.
(
35) Donkers, R. L.; Maran, F.; Wayner, D. D. M.; Workentin, M. J. Am. Chem.
(44) Andrieux, C. P.; Sav e´ ant, J.-M. J. Electroanal. Chem. 1986, 205, 43-59.
(45) Nadjo, L.; Sav e´ ant, J. M.; Su, K. B. J. Electroanal. Chem. 1985, 196, 23-
34.
(46) Sav e´ ant, J. M.; Su, K. B. J. Electroanal. Chem. 1985, 196, 1-22.
Soc. 1999, 121, 7239-7248.
(
(
36) Antonello, S.; Maran, F. J. Am. Chem. Soc. 1998, 120, 5713-5722.
37) TableCurVe 2D, 3.0 ed.; Jandel Scientific Software: San Rafeal, CA.
4184 J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007

Cyclopropyl Conjugation and Ketyl Anions
A R T I C L E S
Figure 3. Mediated reduction of 1,2-diacetylcyclopropane (5) by 9-phenylanthracene (DMF, GCE, 0.5 M TBAP, ν ) 0.1 - 2.0 V/s, γ ) 1).
Table 2. Rate Constants for Homogeneous Electron Transfer (k1)
between the Reduced Form of the Mediator (M ) and
Table 3. Rate Constants for Homogeneous Electron Transfer (k1)
between the Reduced Form of the Mediator (M ) and
•
-
•-
1,2-Diacetylcyclopropane (5)
1-Acetyl-2-phenylcyclopropane (6)
mediator (M)
E
°
M/
M
•−
a
k
1
(M-1s-1)
F
mediator (M)
E
°
M/
M
•−
a
k
1
(M-1s-1)
F
2
2
3
3
3
3
2
2
2
3
4
benzo[a]pyrene
-2.227
-2.253
-2.292
-2.340
-2.342
-2.359
1.85 × 10
4.41 × 10
1.29 × 10
3.43 × 10
6.56 × 10
4.32 × 10
0.00
0.00
0.10
0.00
0.15
0.00
methyl benzoate
ethyl benzoate
benzonitrile
-2.623
-2.644
-2.688
-2.692
-2.728
1.20 × 10
2.00 × 10
1.30 × 10
4.40 × 10
1.29 × 10
0.90
0.90
0.50
0.60
0.35
9
9
,10-diphenylanthracene
-phenylanthracene
anthracene
methyl 2-methylbenzoate
2-tolunitrile
9
9
,10-dimethylanthracene
•-a
b
-methylanthracene
E°A/A
-3.07 ( 0.06
E°A/A•-
a,b
-2.62 ( 0.21
25 ( 5
b
a
Volts vs 0.1 M AgNO /Ag. b Estimated on the basis of Marcus theory
λ (kcal/mol)
3
and eq 8.
a
Volts vs 0.1 M AgNO3/Ag. b Estimated on the basis of Marcus theory
and eq 8.
data points were in the counter-diffusion-controlled region43,47
of the Marcus plot (i.e., the third term of eq 8 dominates) and
a reliable value of the reorganization energy (λ) could not be
obtained). Values of E° obtained in this manner were consistent
with, though slightly lower than, those obtained via convolution
analysis.
rate-limiting step is straightforward because, at a given scan
rate and γ, ip/ipd varies with substrate concentration if electron
transfer is rate-limiting.4
2-46
This point is illustrated in Figure
3
a for the mediated reduction of 5 with 9-phenylanthracene:
A plot of ip/ipd vs log(1/ν) at two different concentrations yields
two discrete lines. When substrate concentration is corrected
for (i.e., a plot of ip/ipd vs log(C°M/ν); Figure 3b), the data points
merge to form a single curve. These plots graphically demon-
strate that homogeneous electron transfer is the rate-limiting
1
1
1
)
+
+
2
]
k1 kd
-λ
4RT
K Z exp
(1 + ∆G°/λ)
d
[
1
(8)
4
2-46
step. Fitting of the data to appropriate working curves
as
kd exp(- ∆G°/RT)
described previously1
3,14
yields the rate constant for electron
transfer between the substrate and M . It should be noted that
•
-
Terephthalonitrile proved to be an effective catalyst for the
the kinetics of these systems are also complicated by a
mediated reduction of 8, and mixed kinetic control was
•
-
48,49
competing bimolecular reaction between M and the ring-
opened product B• involving addition (kadd) and a second
electron transfer (ket) which is accounted for in the working
curves (Scheme 3). The parameter F is the rate constant ratio
ket/(kadd + ket).45
observed. A procedure outlined by Sav e´ ant,
plotting 1/kapp
-
vs mediator concentration in accord with eq 9 (where kapp is
the apparent rate constant assuming rate-limiting electron
transfer), allowed the pertinent rate constants to be resolved:
6
-1 -1
5
-1
k ) 4.68 × 10 M
1
s
and k ) 7.14 × 10 s .
o
For the reduction of 5 and 6 studied using several mediators,
electron transfer was found to be rate-limiting in every instance
using the diagnostic criteria described above. The results are
summarized in Tables 2 and 3; pertinent plots are provided in
the Supporting Information.
1
1
k -1
)
+ 0.33
C°M
(9)
(
)
kapp k1
k k
1 o
Reconciliation of these results with the LSV results, which
also were subject to mixed kinetic control, allowed this system
to be solved completely. In Figure 4, the simultaneous variation
of peak potential (Ep) and peak width (Ep - Ep/2) as a function
of sweeprate (ν) are reconciled to theoretical working curves
(assuming R ) 0.5) via the floating parameters C1 and C2,
•
-
The reduction potential of the substrate (E°A/A ) was
estimated using a Marcus-based treatment involving a non-
linear least-squares fitting of k1 as a function of the reduction
•
-
47
potential of the mediator (E°M/M ) according to eq 8, (with
-
1
11
s-1) as described
∆G° is the free energy change for the reaction
+ A f M + A , and is related to k1, k-1, and the redox
Kd ) 0.16 M
and Z ) 6 × 10
5
0
13,14
defined in eqs 10 and 11, with C1 ) -0.4778 V and C2 )
previously.
M
•
-
•-
(
48) Amatore, C.; Combellas, C.; Pinson, J.; Oturan, M. A.; Robveille, S.;
Sav e´ ant, J.-M.; Thi e´ bault, A. J. Am. Chem. Soc. 1985, 107, 4846-4853.
49) Andrieux, C. P.; Combellas, C.; Kanoufi, F.; Sav e´ ant, J.-M.; Thi e´ bault, A.
J. Am. Chem. Soc. 1997, 119, 9527-9540.
potentials of A and M according to eq 7. (Note: For 6, the
(
(
47) Eberson, L. Electron Transfer Reactions in Organic Chemistry; Springer-
(50) Andrieux, C. P.; Sav e´ ant, J.-M.; Tallec, A.; Tardivel, R.; Tardy, C. J. Am.
Chem. Soc. 1997, 119, 2420-2429.
Verlag: Berlin, 1987; Vol. 25.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007 4185

A R T I C L E S
Tanko et al.
Figure 4. Variation of Ep and Ep - Ep/2 as a function of sweep rate for the reduction of 8. Solid lines represent theoretical working curves.
-
1.949 V. Because ko is now known (from the homogeneous
In contrast, the product arising from the bulk electrolysis of
redox catalysis experiments), eqs 10 and 11 can be combined
to give eq 12. On this basis, E° for 8 was determined to be
-
8 is remarkable and deserves further comment. In a process
which is catalytic in electrons (<0.2 equiv of electrons
transferred), constant current electrolysis of 8 (30 mA) yielded
furan 12 in 65% isolated yield. The proposed mechanism for
this reaction is the chain reaction depicted in Scheme 5,
2.07 V.
2
F
2RT
koD
•-
C ) log
(10)
involving ring opening of 8 , subsequent cyclization of distonic
1
(
4
)
k
•-
5
S
radical ion 9 (in a ∆ -hexenyl manner), and electron transfer
•
-
from 10 to 8 to complete the chain. Air oxidation of
RT
F
koD
dihydrofuran 11 during workup produces 12. The electron
C ) E° +
ln 10 log
(11)
(12)
•-
2
2
transfer from 10 to 8 should be favorable, as the reduction
(
)
( k
)
S
potential of dihydrofuran 11 is expected to be similar to other
Ph(CdO)-containing compounds, ca. -2.2 V. (Furan 12 exhibits
a reversible reduction wave with E° ) -2.00 V, more positive
because of the extended conjugation afforded by the furan ring).
As noted, on a preparative scale, this reaction is catalytic in
electrons. However, the voltammetry of 8 (see Figure 2) gave
no indication of an electron-transfer catalyzed process; the
observed currents were consistent with a two-electron process,
RT ln 10
F
2RT
E° ) C -
log k + C - log
2
(
o 1
)
2F
For 7, the mediated reduction could be achieved with
terephthalonitrile (E°M/M•- ) -1.949 V) and ring opening was
rate-limiting with kobs ) (k1/k-1)ko )10.6 s . Fitting of the
LSV results to the mixed kinetic control model as described
above yielded C1 ) 3.163 V and C2 ) -1.931 V. Reconciling
eq 12 with kobs yields E° ) -2.13 V and ko )1.0 × 10 s .
Product Studies. As expected, preparative scale electrolyses
of 5 f 7 (constant potential, two electrons transferred) gave
rise solely to cyclopropane ring-opened products. In the case
of 5 and 7, the initially produced diketone products underwent
further reaction (Robinson annulation), presumably under the
basic conditions associated with the electrolysis, or upon workup
to yield cyclohexenones; the products and presumed mechanism
for their formation are summarized in Scheme 4.
-
1
•-
suggesting that ring-opened radical anion 9 is simply reduced
to the dianion under these conditions.
5
-1
The transformation 8 f 11 is essentially a radical anion
variant of the electron-transfer catalyzed “vinylcyclopropane f
51
cyclopentene rearrangement.” The intramolecular reaction
•-
•-
between an enolate and enoyl radical (9 f 10 ) has also
52
been previously proposed by Bauld et al. to account for an
unusual reduction of tethered bis(enones) to produce Diels-
•-
•-
Alder-type product (15 f 16 , Scheme 6).
Discussion
Cyclopropyl Ketones: LUMO Energies and Properties.
Calculations at the Hartree Fock 6-311G* level were performed
on compounds 5 f 8, as well as numerous other cyclopropyl
ketones for which reduction potentials have been previously
Scheme 4
Figure 5. LUMO of cyclopropyl ketones 5, 7, and 8.
4186 J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007

Cyclopropyl Conjugation and Ketyl Anions
A R T I C L E S
Scheme 5
Scheme 6
reported. (A list of these compounds, their reduction potentials,
and LUMO energies are available in the Supporting Information
for this paper). Inspection of the LUMO indicates that for 5, 7,
and 8, the LUMO is derived from the two π* CdO of the
carbonyl groups, and the p-orbital-based LUMO of cyclopropane
Figure 6 presents a plot of the reduction potential of these
compounds vs the calculated LUMO energy. The observed
correlation suggests that, for most substrates, the LUMO
energies of the neutral ketones are a reasonable predictor of
reduction potential. It is noteworthy, however, that 5 shows the
greatest deviation from the line and is about 400 mV more easily
reduced than expected on the basis of its LUMO energy. (This
point is addressed in more detail later).
(ψ3, Figure 1) as illustrated below in Figure 5. This analysis
suggests that the corresponding radical anions of 5 f 8 might
be highly delocalized, and in the case of 5, 7, and 8, charge
and spin may be delocalized over both carbonyl groups through
the cyclopropyl group.
Scheme 7
DET: Concerted or Stepwise? A fundamental question in
modern electron transfer theory in situations where electron
transfer and bond cleavage occurs is whether the two events
5
3-55
occur simultaneously or in a stepwise manner.
For the
reduction of cyclopropyl ketones, this point is illustrated in
Scheme 7 where path (a) is the concerted pathway, and path
(b) is stepwise. Only in the latter scenario does the radical anion
exist as a discrete intermediate.
(
51) Dinnocenzo, J. P.; Conlon, D. A. J. Am. Chem. Soc. 1988, 110, 2324-
326.
52) Roh, Y.; Jang, H.-Y.; Lynch, V.; Bauld, N. L.; Krische, M. J. Org. Lett.
2
(
2002, 4, 611-613.
Figure 6. Observed reduction potentials vs ELUMO (HF/6-311G*) for
cyclopropyl ketones.
(53) Sav e´ ant, J. M. Acc. Chem. Res. 1993, 26, 455-461.
(54) Sav e´ ant, J.-M. AdV. Phys. Org. Chem. 2000, 35, 117-192.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007 4187

A R T I C L E S
Tanko et al.
The calculated barriers for ring opening of 7•- and 8•- are
7
.7 and 6.1 kcal/mol, respectively, consistent with experiment:
•
-
The rate constant for ring opening of 8 is about 1 order of
magnitude greater than that for 7 . The calculations also provide
•-
•
-
insight into why 7 does not undergo the “vinylcyclopropane
f cyclopentene-type rearrangement” depicted in Scheme 5. The
calculations suggest that the distonic, ring-opened radical anion
•
-
•-
9
(formed from 7 ) has the negative charge on the acetyl
•-
moiety. As a consequence, cyclized product 13 does not enjoy
the same resonance stabilization afforded to 10 (formed by
the cyclization of 8 ).
•
-
•
-
Does 5 Fit Known Theoretical Models for Concerted
DET? At first glance, no. In principle, the question of concerted
vs stepwise DET can be addressed by considering the voltam-
metric behavior of the substrates. Specifically, when the electron
transfer is rate-limiting, the electron-transfer coefficient (R) is
a sensitive diagnostic probe of mechanism. The electron-transfer
coefficient is directly related, in the context of Marcus theory
*
(
eq 13), to the intrinsic barrier (∆Go , the free energy barrier
at zero driving force) for the electron transfer (eq 14). If bond
breaking is occurring in the transition state, as would be the
case for a concerted pathway, considerable structural reorga-
nization is occurring (i.e., high intrinsic barrier) and R is
Figure 7. Energy of 1,2-diacetylcyclopropane radical anion (5•-) as a
function of C1-C2 distance, calculated at the UHF/6-31+G* level.
The issues of radical anion structure, charge/spin delocaliza-
tion, and the possibility of concerted DET were probed by
performing molecular orbital calculations on the ring-closed and
ring-opened radical anions derived from 5 f 8. All of these
calculations were performed at the UHF/6-31+G* level; for 6
f 8, spin contamination proved problematic, so the energies
were obtained from a single-point ROHF calculation, using the
UHF-derived geometries. It should be mentioned that the results
at these levels of theory were significantly different from what
was obtained from density functional calculations (B3LYP/6-
38-40,57-67
characteristically low (<0.5).
2
-
-
o
∆
G°
∆
G ) ∆G 1 +
(13)
(14)
-
(
)
4
∆Go
-
∂
∂
∆G
∆G
1
2
∆G°
R )
)
1 +
o
-
(
)
4∆Go
For 5 and 6, R values obtained three different ways are sum-
marized in Table 1. While electron transfer is rate-limiting in
both cases, the derived R values for 5 are slightly less than 0.5
but certainly not as low as is often found for a concerted disso-
ciative electron-transfer process. In contrast, for 6, R ) 0.5 is
consistent with a stepwise pathway, and certainly all the all the
results for 7 and 8 clearly indicate electron transfer and ring
opening occur in discrete steps and the corresponding radical
anions have a finite lifetime (path (b), Scheme 7). The longer
31+G*), which predict that both the ring-closed and ring-opened
forms of symmetrical radical anions such as 5 and 8 are fully
delocalized. In 1996, Bally noted that DFT calculations fail to
accurately predict the dissociative behavior of symmetrical
56
radical ions. Accordingly, HF methods were used in this study
because they do not seem to suffer from this problem.
At the UHF/6-31+G* level, we were unable to locate a
•-
•-
stationary structure corresponding to the cyclopropane ring-
lifetime for 7 and 8 can clearly be attributed to resonance sta-
bilization provided by the benzoyl group in the latter (Scheme 8).
Returning to the question of concerted DET for 5. According
to Marcus theory, the free energy of activation for electron
•
-
closed form of 5 . The only structure obtained was a cyclo-
propane ring-opened structure, with a C1-C2 bond length of
about 2.5 Å, as opposed to the 1.5 Å normally associated with
a cyclopropyl C-C bond. Attempts to map out the potential
energy surface, by incrementally closing the C1-C2 bond
revealed that the interaction between C1 and C2 was purely
repulsive (Figure 7). Thus, the calculations very much suggest
*
transfer (∆G ) as a function of driving force (∆G°) is described
*
by eq 13, where ∆Go ) (λ + λ )/4; λ and λ represent the
i
o
i
o
(
57) Andrieux, C. P.; Gallardo, I.; Sav e´ ant, J.-M.; Su, K.-B. J. Am. Chem. Soc.
1986, 108, 638-647.
•
-
that 5 has no significant lifetime and that electron transfer to
(58) Andrieux, C. P.; Robert, M.; Saeva, F. D.; Sav e´ ant, J.-M. J. Am. Chem.
Soc. 1994, 116, 7864-7871.
(
5
may occur simultaneously with ring opening.
59) Andrieux, C. P.; Sav e´ ant, J.-M.; Tardy, C. J. Am. Chem. Soc. 1998, 120,
4167-4175.
In contrast, the cyclopropane ring-closed forms of the radical
(60) Borsari, M.; Dallari, D.; Fontanesi, C.; Gavioli, G.; Iarossi, D.; Piva, R.;
anions generated from 6, 7, and 8 all reside at potential energy
Taddei, F. J. Chem. Soc., Perkin Trans. 2 1997, 1839-1843.
•
-
(61) Cardinale, A.; Isse, A. A.; Gennaro, A.; Robert, M.; Sav e´ ant, J.-M. J. Am.
Chem. Soc. 2002, 124, 13533-13539.
minima. Ring opening of 6 is highly exothermic, and the
calculated barrier for ring opening is small, <1 kcal/mol. This
result is certainly consistent with the fact that both the direct
and indirect electrochemistry of 6 are characterized by rate-
limiting electron transfer (suggesting a rate constant for ring
(62) Christensen, T. B.; Daasbjerg, K. Acta Chem. Scand. 1997, 51, 307-317.
(63) Daasbjerg, K.; Jensen, H.; Benassi, R.; Taddei, F.; Antonello, S.; Gennaro,
A.; Maran, F. J. Am. Chem. Soc. 1999, 121, 1750-1751.
(64) Houmam, A.; Hamed, E. M.; Still, I. W. J. J. Am. Chem. Soc. 2003, 125,
7258-7265.
(
65) Lexa, D.; Sav e´ ant, J.-M.; Sch a¨ fer, H. J.; Su, K.-B.; Vering, B.; Wang, D.
L. J. Am. Chem. Soc. 1990, 112, 6162-6177.
7
-1
opening .10 s ).
(
66) Workentin, M. S.; Donkers, R. L. J. Am. Chem. Soc. 1998, 120, 2664-
2665.
(
55) Maran, F.; Workentin, M. S. Interface 2002, 44-49.
(67) Workentin, M. S.; Maran, F.; Wayner, D. D. M. J. Am. Chem. Soc. 1995,
117, 2120-2121.
(
56) Bally, T.; Sastry, G. N. J. Phys. Chem. A 1997, 101, 7923-7925.
4188 J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007

Cyclopropyl Conjugation and Ketyl Anions
A R T I C L E S
Scheme 8
inner and outer reorganization energies. The heterogeneous rate
constants obtained from the convolution analysis of 5 and 6
can be treated in the context of Marcus theory through the
Eyring equation (eq 15), where Z for heterogeneous electron
transfer is obtained from eq 16 (kB is Boltzmann’s constant; m
is the molecular mass). The results of this analysis are provided
in Figure 8, and described below.
khet
-
RT
F
∆
G ) -
ln
(15)
(16)
(
)
Z
1/2
k T
B
Zechem
)
(2πm
)
Figure 8. Plots of ln(khet) vs ∆G° for the reduction of 1,2-diacetylcyclo-
propane (5) and 1-acetyl-2-phenylcyclopropane (6) reconciled to electron
transfer theory.
Fitting the heterogeneous kinetic data for the reduction of 5
and 6 to eq 13 leads to reorganization energies (λ) ) 1.49 and
1
.32 eV, respectively. If it is assumed that 6 undergoes stepwise
electron transfer, then the reorganization energy for 5 is only
.17 eV (4 kcal/mol) higher, and at first glance, this small
stabilization energy of ca. 6 kcal/mol for CH3(CdO). The C-C
bond strength in cyclopropane is 61 kcal/mol, two CH3(CdO)
0
71
difference is difficult to rationalize with a concerted DET
process for 5.
Sav e´ ant has developed theory53,54,68 pertaining to concerted
DET where an additional term is added to the inner reorganiza-
tion energy to account for the strength of the bond (DR) which is
groups would lower this to 49 kcal/mol, reasonably consistent
with the DFT approach.
*
cleaved during electron transfer; ∆Go ) (DR + λo)/4. Because
6
undergoes stepwise electron transfer, inferring little or no
The assumption that λ for 5 is the same for as for 6 and that
o
bond-breaking occurs in the transition state, it may be reasonable
R
D ) 1.95 eV leads to the solid (bottom) curve in Figure 7
to assume that most of λ is attributable to solvent reorganization
clearly showing that the activation/driving force relationship for
(λo). In DMF, the solvent reorganization energy can be estimated
5 does not fit the concerted DET model.
by λo (eV) ) 3/a (Å), where a is the radius of the electron
The concerted DET model assumes there is no significant
interaction between the radical and anion after bond cleavage.
However, coulombic interactions have been detected in some
6
9
acceptor. Assuming that this radius can be approximated by
the size of the CH3CO group in 6 (i.e., the charge is localized
as indicated by the MO calculations) leads to an estimated λo
of 1.27 eV, very close to the observed value of 1.32 eV.
38,61,64,72-78
systems
and are the result of attractive anion-dipole
interactions between the negatively charged leaving group and
neutral radical. These interactions have the net effect of
offsetting the importance of the DR term because, in essence,
the bond is not completely broken. This situation is illustrated
On this basis, the characteristics of the ln(khet) vs ∆G° plot
for 5 can be constructed on the basis of concerted DET theory.
The C1-C2 bond strength of 5 is estimated to be 45 kcal/mol
•
-
in Figure 9, where concerted DET to R-X results in a (R X )
(1.95 eV), based upon the hypothetical hydrogenation reaction
caged pair, which subsequently diffuses apart to the free radical
depicted in eq 17, using literature values for strengths of the
H-H bond and C-H bonds, and ∆Hf°’s obtained via DFT
(
(
(
71) Berson, J. A.; Pedersen, L. D.; Carpenter, B. K. J. Am. Chem. Soc. 1976,
1
3
98, 122-143.
calculations (B3LYP/6-31G*) as described previously. This
estimate of the C-C bond strength for 5 is quite reasonable on
the basis of group additivity. The C-C bond strength in ethane
is 90 kcal/mol, vs 84 kcal/mol for acetone,70 suggesting a radical
72) Pause, L.; Robert, M.; Sav e´ ant, J.-M. J. Am. Chem. Soc. 2000, 122, 9829-
9835.
73) Costentin, C.; Robert, M.; Sav e´ ant, J.-M. J. Am. Chem. Soc. 2003, 125,
10729-10739.
(74) Chang, J.; Goddard, J. D.; Houmam, A. J. Am. Chem. Soc. 2004, 126,
8076-8077.
(75) Costentin, C.; Robert, M.; Sav e´ ant, J.-M. J. Am. Chem. Soc. 2004, 2004,
(
(
(
68) Sav e´ ant, J. M. In AdVances in Electron Transfer Chemistry; Mariano, P.
16834-16840.
S., Ed.; JAI Press: Greenwich, 1994; Vol. 4, pp 53-116.
(76) Isse, A. A.; Gennaro, A. J. Phys. Chem. A 2004, 108, 4180-4186.
(77) Costentin, C.; Louault, C.; Robert, M.; Teillout, A.-L. J. Phys. Chem. A
2005, 109, 2984-2990.
(78) Costentin, C.; Robert, M.; Sav e´ ant, J.-M. J. Am. Chem. Soc. 2003, 125,
105-112.
69) Pause, L.; Robert, M.; Sav e´ ant, J.-M. J. Am. Chem. Soc. 2001, 123, 11908-
1
1916.
70) Kerr, J. A. In Handbook of Chemistry and Physics, 84th ed.; CRC Press:
Boca Raton, FL, 2003-2004; pp 9-52-9-75.
J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007 4189

A R T I C L E S
Tanko et al.
computationally. An energy minimum (no imaginary frequen-
cies) was found for a structure which placed the oxygen of the
enolate anion ∼3.75 Å from the carbonyl carbon of the enoyl
radical. In Figure 10, energy is plotted as a function of the
distance between these two atoms, suggesting an interaction
energy around 9.6 kcal/mol (0.42 eV) in the gas phase. It is
expected that the magnitude of this interaction would be reduced
in DMF solvent and in the presence of electrolyte. In addition,
•
-
the ring-opened form of 5 does not appear to have the
flexibility to achieve a conformation such as that depicted in
structure 18, which would allow the enolate oxygen to directly
interact with the carbonyl group. Thus, although the results seem
to fit the “sticky” concerted DET model, the derived value of
DP ) 0.98 eV is completely unreasonable.
Figure 9. Concerted dissociative electron transfer to R-X with and without
•
-
interaction between the products R and X .
Does 5 Relax to a “Loose” Radical/Anion Pair Upon
Addition of an Electron? In some ways, the results observed
for the reduction of 5 are reminiscent of those presented for
the one-electron reduction of various disulfides (RSSR) by
Maran et al. In these cases, R was low, indicative of a larger
than normal intrinsic barrier to electron transfer. For the
disulfides, it was suggested that electron transfer occurred
concurrently with a lengthening of the S-S bond to form a
•
-
loose radical/ion pair (RS / SR), and the large intrinsic barrier
to electron transfer is interpreted as arising from this structural
reorganization. In essence, the bond is not completely ruptured,
and a similar argument might explain the kinetics of the
reduction of 5.
To address this issue, the structure of the ring-opened
distonic) radical anion was probed further (UHF/6-31+G*).
-
•
Figure 10. Energy of CH3(CO)CH2 /CH3(CO)CH2 as a function of the
distance between the enolate oxygen and carbonyl carbon of the enoyl
radical.
(
The lowest energy structure for this species finds the two
CHCOCH3 groups of the enolate and enoyl radical parallel to
each other, with the CH’s separated by ∼2.5 Å; “cisoid” form
and anion. In this situation, eq 18 pertains where DP refers to
the magnitude of the interaction between the fragments, and
the ∆G°SP term refers to the difference between the standard
1
9 is lower in energy that “transoid” 20. The geometric and
electronic structure of the ring-opened radical anion is not
consistent with the characterization of this structure as a loose
radical/ion pair: The distance between the two CHs is identical
to the distance between C1 and C3 of propane, which is
obviously “ring-opened.” Moreover, the radical anion is not
delocalized (i.e., most of the negative charge is localized on
one of the CHCOCH3 groups, and most of the spin on the other.)
It is interesting to note that other conformations such as 21,
which in principle suffer from less torsional strain, are higher
in energy and do not exist at potential energy minima, and this
may indicate an (small) ion-dipole interaction between the
enolate and enoyl portions of the radical ion.
free energies of the separated and caged products (mainly
attributable to entropy).72
2
2
(
D - D ) + λ
∆G° - ∆G°SP
x
R
x
P
o
-
∆
G )
1 +
2
4
[
]
(
D - D ) + λ
x
R
x
P
o
(
18)
In 5, an anion/dipole interaction may be occurring between
the enolate anion portion and the CdO group of the enoyl
radical portion of the distonic radical anion, illustrated with 18.
In applying the “sticky” concerted DET model to this system
o
(
eq 18), we assume ∆G SP ≈ 0 because for these cyclopropane
cleavages, the radical and anion fragments cannot diffuse apart
because they are tethered by the CH2 group. While the
experimental results for 5 can be fit to the “sticky” concerted
DET model (dotted line in Figure 7), the magnitude of the
interaction (DP ) 0.98 eV ) 23 kcal/mol) seems unreasonably
high.
One-Electron Reduction of 5: Resolution of the Di-
chotomy? As discussed, MO calculations suggest 5• does not
exist as a discrete intermediate, yet the kinetic results for the
reduction of 5 fit neither the normal nor “sticky” concerted DET
-
To verify this conjecture, potential interactions between an
enolate anion (CH3COCH2 ) and enoyl radical (CH3COCH2 ),
analogous to that proposed in structure 18 were probed
-
•
4190 J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007

Cyclopropyl Conjugation and Ketyl Anions
A R T I C L E S
Figure 11. Variation in the energy of the LUMOs for 1,2-diacetylcyclopropane (5) and methyl cyclopropyl ketone (22) as a function of C1-C2 distance;
the LUMO of 5.
modelssassuming that 5 has the same solvent reorganization
energy as 6 and is localized. The geometric and electronic
structure of the (distonic) ring-opened radical anion is not
consistent with a relaxed C-C bond, as might be anticipated
for a loose radical/anion pair-type structure, as has been
suggested for reduction of RSSR. Rather, the kinetic results fit
rather well with the classical Marcus model, with a reorganiza-
tion energy (λ) only slightly (0.17 eV) higher than for compound
proximated by the size of the OdC-CH-CH-CdO frame-
work, then the estimated solvent reorganization energy in DMF
(λo ) 3/a, where a is the radius in Å of the acceptor) is 1.07
eV. It is somewhat satisfying to note that the total reorganization
energy λ ) λi + λo based upon this analysis (1.46 eV) is close
to the experimentally observed value of 1.49 eV.
Finally, does electron transfer theory need to be modified to
accommodate these results? Probably not. The estimate of 0.39
eV for the internal reorganization energy associated with the
reduction of 5 is approaching one-fourth of the estimated
strength of the C1-C2 bond strength, as predicted by the
Sav e´ ant concerted DET model. Indeed, refitting the kinetic data
to this model (again using DR ) 1.95 eV) and solving for the
solvent reorganization energy leads to λo ) 1.01 eVsclose to
the estimated value of 1.07 eV, assuming electron transfer to a
more delocalized system. Given the uncertainties associated with
the experimental data, the nature of some of these assumptions,
and the number of parameters involved (and their compensatory
relationship), it is probably unwise to attach too much quantita-
tive significance to these values. However, qualitatitively at least,
the experimental results for the reduction of 5 can be accom-
modated by the concerted DET model, and the resulting
parameters are of reasonable magnitude.
6swhich, consistent with both theory and experiment, certainly
undergoes stepwise DET (electron transfer followed by ring
opening).
The conflicting results obtained for 5 may be rationalized by
considering the unique bonding properties of the cyclopropyl
group. As noted earlier (Figure 5), the LUMO of 5 is derived
from the two π* orbitals of CdO, and the p-orbital-based
LUMO of the cyclopropyl group. Calculations indicate that the
LUMO energy of 5 is Very sensitive to C-C bond length, much
more so than is the case for methyl cyclopropyl ketone (22).
For both these compounds, stretching the C-C bond from 1.5 Å
(normal cyclopropyl C-C bond length) to 1.7 Å “costs” the
samesabout 9 kcal/mol (0.39 eV). However, while the LUMO
energy of 22 does not change significantly over this range, the
LUMO energy of 5 decreases dramatically (from 3.6 to 2.8 eV)
as the bond is lengthened. This observation makes sense because
there is a significant antibonding interaction between C1 and
C2 in the LUMO of 5 (Figure 11).
Conclusions
The one-electron reduction of 5 may proceed via a concerted
DET process; however, the only evidence for this statement is
derived from the results of MO calculations. The reorganization
energy obtained from experiment is only slightly higher than
expected on the basis of the behavior of other cyclopropyl
ketones and at first glance, is consistent with a normal (stepwise)
electron-transfer process. However, the MO calculations also
suggest that a low reorganization energy may be quite reasonable
for the reduction of this compound: A modest, “low-cost”
internal reorganization (lengthening of the C1-C2 bond) may
occur which lowers the LUMO energy sufficiently for electron
transfer to occur; the estimated total reorganization energy based
upon this hypothesis (assuming a more delocalized structure,
consistent with the properties of the LUMO) matches well with
experiment. The energetic consequences of this internal reor-
With classical Marcus theory, the internal component of the
reorganization energy encompasses geometric distortions needed
to bring the LUMO energy close to the energy of the donor so
that electron transfer can occur. In the case of 5, a 0.2 Å
lengthening of the C1-C3 bond may be sufficient to do this,
and at little cost (ca. 9 kcal/mol or 0.39 eV) in the context of
internal reorganization energy (λi). (Perhaps not so coinciden-
tally, this geometric distortion would be sufficient to bring 5
back onto the line in the plot of E° vs LUMO energy for the
various cyclopropyl ketones, Figure 6). Also, since the LUMO
encompasses both carbonyl groups and the cyclopropane, the
effective size of the electron acceptor is larger than for a simple
cyclopropyl ketone (i.e., the charge is transferred to a more
delocalized system). Assuming that the radius can be ap-
J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007 4191

A R T I C L E S
Tanko et al.
ganization are also of the magnitude predicted by the concerted
DET model, again, with the assumption of a more delocalized
structure. Thus, what makes this possible case of concerted DET
particularly difficult to detect from an experimental standpoint
is that two compensating factors are at play: The onset of
concerted DET raises the internal reorganization energy, but
that is offset by a lower solvent reorganization associated with
formation of a more delocalized system (compared to other
cyclopropyl ketones). We suspect that the reason that this
sort of issue has not come up previously is that well-
documented examples of concerted DET involve electron
transfer to a rather localized framework, typically a σ* anti-
bonding orbital.
In contrast, the radical anion of 6 exists for a finite lifetime,
and hence, stepwise DET is observed. However, the barrier to
ring opening is very small. Substituents at the 2-position of
methyl cyclopropyl ketones play a critical role in determining
whether the DET will be concerted vs stepwise. It seems likely
that additional examples of concerted DET may be found in
these (and related) systems, but for the reasons discussed, the
unambiguous experimental detection of concerted vs stepwise
DET will be problematic.
these compounds are truly “hypersensitive” mechanistic probes.
Finally, although an unsubstituted cyclopropyl group does not
stabilize a radical anion, electron-withdrawing substituents at
C2 significantly enhance the electron-withdrawing properties
of the cyclopropyl group. The energy of the LUMO of the
neutral ketone is (in general) a good predictor of the reduction
potential of these compounds. (The exception is 5, whose
reduction potential is more favorable than predicted, possibly
because concerted DET is operative and the reduction is partly
driven by the relief of cyclopropane ring strain.) There is no
•
-
•-
•-
evidence which suggests that 6 , 7 , or 8 are fully delocal-
ized (symmetrically, in their ring-closed forms) through the
cyclopropyl group, as might be predicted on the basis of the
properties of the LUMO of the neutral ketone.
Acknowledgment. Financial support from the National Sci-
ence Foundation (CHE-0108907) and (CHE-0548129) is ac-
knowledged and appreciated. We also thank Profs. Mark
Anderson, Paul Carlier, T. Daniel Crawford, and Diego Troya
for enlightening and helpful discussions which aided us in the
preparation of this manuscript and several anonymous reviewers
whose helpful comments and criticisms proved invaluable in
the evolution of this manuscript.
For 7 and 8, a stepwise mechanism is preferred, and the rate
constant for ring opening of the corresponding radical anions
is lower. The difference is clearly attributable to the resonance
stabilization afforded by the benzoyl group for the radical anions
generated from 7 and 8.
Supporting Information Available: Voltammetry data and
plots pertaining to the electrochemistry of 5-8, table of
reduction potentials and LUMO energies of several cyclopropyl-
containing ketones used to construct Figure 6, the full Gaussian
’03 citation, and absolute energies and optimized geometries
Cyclopropyl ring openings and fragmentations of this type
are often used as mechanistic probes for electron transfer. The
fact that 5 may undergo concerted dissociated electron transfer,
of all calculated structures. This material is available free of
charge via the Internet at http://pubs.acs.org.
•
-
and 6 has a very low barrier for ring opening, means that
JA063857Q
4192 J. AM. CHEM. SOC.
9
VOL. 129, NO. 14, 2007