Dynamics of bond breaking in ion radicals. Mechanisms and reactivity in the reductive cleavage of carbon-fluorine bonds of fluoromethylarenes

Article abstract of DOI:10.1021/ja971094o

The reductive cleavage mechanism and reactivity of the carbon-fluorine bonds in fluoromethylarenes are investigated, in liquid ammonia and in DMF, by means of cyclic voltammetry and/or redox catalysis as a function of the number of fluorine atoms and of the structure of the aryl moiety. The reduction of the trifluoro compounds, eventually leading to complete defluorination, involves the di- and monofluoro derivatives as intermediates. Carbenes do not transpire along the reaction pathway. Application of the intramolecular dissociative electron transfer model allows the quantitative rationalization, in terms of driving force and intrinsic barrier, of the variation of the cleavage reactivity of the primary anion radical with the number of fluorine atoms and of the structure of the aryl moiety as well as with the solvating properties of the medium. When, related to the structural factors thus uncovered, the primary anion radical generates the di- and monofluoro intermediates far from the electrode surface, their reduction occurs homogeneously giving rise to an apparently direct six-electron process according to an internal redox catalysis mechanism. Conversely, with rapid cleavages, the reduction of the di- and monofluoro intermediates takes place at the electrode surface and the stepwise expulsion of the fluorides ions transpire in the cyclic voltammetric patterns.

Full text of DOI:10.1021/ja971094o

J. Am. Chem. Soc. 1997, 119, 9527-9540
9527
Dynamics of Bond Breaking in Ion Radicals. Mechanisms and
Reactivity in the Reductive Cleavage of Carbon-Fluorine
Bonds of Fluoromethylarenes
Claude P. Andrieux,1a Catherine Combellas,1b Frede´ric Kanoufi,1b
Jean-Michel Save´ant,*,1a and Andre´ Thie´bault*,1b
Contribution from the Laboratoire d’Electrochimie Mole´culaire de l’UniVersite´ Denis Diderot,
Unite´ Associe´e au CNRS No 438, 2 place Jussieu, 75251 Paris Cedex 05, France, and the
Laboratoire de Chimie et d’Electrochimie des Mate´riaux Mole´culaires, ESPCI, 10 Rue Vauquelin,
75231 Paris Cedex 05, France
ReceiVed April 7, 1997X
Abstract: The reductive cleavage mechanism and reactivity of the carbon-fluorine bonds in fluoromethylarenes are
investigated, in liquid ammonia and in DMF, by means of cyclic voltammetry and/or redox catalysis as a function
of the number of fluorine atoms and of the structure of the aryl moiety. The reduction of the trifluoro compounds,
eventually leading to complete defluorination, involves the di- and monofluoro derivatives as intermediates. Carbenes
do not transpire along the reaction pathway. Application of the intramolecular dissociative electron transfer model
allows the quantitative rationalization, in terms of driving force and intrinsic barrier, of the variation of the cleavage
reactivity of the primary anion radical with the number of fluorine atoms and of the structure of the aryl moiety as
well as with the solvating properties of the medium. When, related to the structural factors thus uncovered, the
primary anion radical generates the di- and monofluoro intermediates far from the electrode surface, their reduction
occurs homogeneously giving rise to an apparently direct six-electron process according to an internal redox catalysis
mechanism. Conversely, with rapid cleavages, the reduction of the di- and monofluoro intermediates takes place at
the electrode surface and the stepwise expulsion of the fluorides ions transpire in the cyclic voltammetric patterns.
Fluorocarbons are compounds of wide interest in view of their
chemical and biological applications and of the importance of
perfluoro polymers. For this reason, the electrochemical
cleavage of carbon-fluorine bonds in fluorinated organic
molecules has been the object of early interest.2 In the case of
perfluoro polymers, electron transfer cleavage of C-F may be
envisaged as a route to a large variety of carbon materials as
well as a first step of the derivatization of such polymers.3
Previous work has shown that the reduction of the three
carbon-fluorine bonds often occurs simultaneously, the only
product being the methyl derivative. In a few instances
however, the CF2H and CFH2 intermediates have been found
among the electrolysis products. Perusal of the available data
leaves the impression that the simultaneous Versus stepwise
character of the reductive cleavage of the three carbon-fluorine
bonds is a complex function of molecular structure and of the
nature of the solvent. Electron-donating substituents on the
phenyl ring of benzotrifluoride makes the reduction potential
too negative for the C-F reductive cleavage to be investigated.
This is the reason that the available data concern only benzo-
trifluorides with electron-withdrawing substituents or benzo-
trifluoride itself. Simultaneous cleavage of the three carbon-
fluorine bonds occurs in the reduction of 4-carboxymethyl-
benzotrifluoride in methanol.4 Cyclic voltammetric data indicate
that the same is true in liquid ammonia for this compound and
for the para-cyano derivative.5 The CF2H and CFH2 intermedi-
ates have been identified among the reduction products of
benzotrifluoride in N,N′-dimethylformamide (DMF) together
with toluene. The respective yields vary with the electrolysis
potential in a manner that is consistent with successive breaking
-
of the C-F bonds. The intermediacy of the C6H5CF2
carbanion in the reduction in DMF is confirmed by its trapping
by electrophiles such as CO2, acetone, or DMF itself.6 The
same is true for 4-fluorobenzotrifluoride in the same solvent.6
The CFH2 derivative was also found among the reduction
product of RRR-trifluoroacetophenone in DMF,7 whereas, in
liquid NH3, the cyclic voltammetric data point to a dimerization
of the initially formed anion radical.5
So far, quantitative kinetic data are not available that would
allow one to unravel the mechanism of the reductive cleavage
of the three C-F bonds in trifluoromethylarenes. In the work
reported below we have investigated in detail the direct or
mediated electrochemical reduction of the 4-cyanotoluenes, 1′,
1′′, and 1′′′ where the R-carbon bears, one, two, and three
fluorine atoms, respectively (see Chart 1). An important issue
in the discussion of the reductive cleavage mechanism of
trifluoromethyl compounds is to know whether it involves the
intermediacy of the corresponding difluoro- and monofluoro-
methyl intermediates or whether it may involve carbene
intermediates as previously observed with gem-polychloro and
polybromo compounds.8 Besides the reduction mechanism by
(4) (a) Coleman, J. P.; Gilde, H. G.; Utley, J. H. P.; Weedon, B. C. L.
J. Chem. Soc., Chem. Commun. 1970, 738. (b) Coleman, J. P.; Naser-ud-
din; Gilde, H. G.; Utley, J. H. P.; Weedon, B. C. L.; Eberson, L. J. Chem.
Soc., Perkin Trans. II 1973, 1903.
(5) Combellas, C.; Kanoufi, F.; Thie´bault, A. J. Electroanal. Chem. 1996,
407, 195.
X Abstract published in AdVance ACS Abstracts, September 15, 1997.
(1) (a) Universite´ Denis Diderot. (b) ESPCI.
(2) (a) Elving, P. J.; Leone, J. T. J. Am. Chem. Soc. 1957, 79, 1546. (b)
Lund, H. Acta Chem. Scand. 1959, 13, 192. (c) Cohen, A. I.; Keeler, B. T.;
Coy, N. H.; Yale, H. L. Anal. Chem. 1962, 34, 216.
(3) (a) Kavan, L. Electrochemical Carbonization of Fluoropolymers. In
Chemistry and Physics of Carbon; Thrower, Ed.; Marcel Dekker: New
York, 1991; pp 70-163. (b) Combellas, C.; Kanoufi, F.; Marzouk, H.;
Thie´bault, A. French Patent 9509726, 1995.
(6) Saboureau, C.; Troupel, M.; Sibille, S.; Pe´richon, J. J. Chem. Soc.,
Chem. Comm. 1989, 1138.
(7) Stocker, J. H.; Jenevein, R. M. J. Chem. Soc., Chem. Commun. 1968,
934.
S0002-7863(97)01094-9 CCC: $14.00 © 1997 American Chemical Society

9528 J. Am. Chem. Soc., Vol. 119, No. 40, 1997
Andrieux et al.
Chart 1
Figure 1. Reduction of 1′ in DMF + 0.1 M n-Bu4BF4 at 20 °C (a, b)
liquid NH3 + 0.1 M KBr + 3 mM NH4+ at -38 °C and (a′, b′) in the
absence (a, a′) and presence (b, b′) of an acid. [1′] ) 1 mM (a, b), 3
mM (a′, b′). Scan rate: 0.2 V/s. [PhOH] ) 10 mM (b), [NH4Br] ) 3
mM (b′). The potentials are referred to SCE in DMF and to 0.01 M
Ag+/Ag in liquid NH3.
reductive cleavage reactivity with the help of the model
mentioned above. The preliminary results previously obtained
for the reductive cleavage of the first C-F bond in 5 and 95
will also be rediscussed and used in the general discussion of
the relationships between molecular structure and C-F cleavage
reactivity.
which the successive reductive cleavage of several C-F bonds
occurs, the analysis of the kinetic results provides the standard
potentials for the formation of each of the three anion radicals
and their cleavage rate constants. The cleavage of an ion radical
may be viewed as an intramolecular dissociative electron
transfer. Based on this picture, a model of the dynamics of the
reaction has been developed as an extension of the theory of
dissociative electron transfer.9 The model has been successfully
applied to the cleavage of various series of frangible anion
radicals.9a,10 It is interesting to examine whether it is also able
to rationalize the variation of the reductive cleavage reactivity
of benzylic C-F bonds as a function of the number of fluorine
atoms borne by the functional carbon and thus to explain the
mechanism by which the three fluorine atoms are successively
expelled from the molecule. All experiments were carried out
both in N,N′-dimethylformamide (DMF) at 20 °C and in liquid
ammonia at -38 °C in the purpose of examining the role of
the nature of the solvent and of temperature in the reductive
cleavage mechanism.
Results
Compound 1′. Figure 1 shows typical cyclic voltammo-
grams obtained for the reduction of 1′ in DMF and liquid NH3.
The cleavage of the C-F bond is expected to produce the
4-cyanotoluene carbanion which may react with acids present
in the medium such as residual water leading to 4-cyanotoluene.
It may also react with the starting material thus decreasing the
electron stoichiometry below 2 and possibly generating several
products which reduction is seen at more negative potentials.
These side reactions may be eliminated by addition of an acid
stronger than water as attested by the modification of the
We have also measured (in liquid ammonia at -38 °C) the
cleavage rate constants and standard potentials for the reductive
cleavage of one C-F bond in a series of other trifluoromethyl
derivatives where the aryl moiety is varied (2-4 and 6-8 in
Chart 1) in the purpose of examining the variation of the
+
voltammograms upon addition of NH4 in liquid NH3 and of
phenol in DMF (Figure 1). The electron stoichiometry then
reaches 2, and the wave appearing at more negative potentials
is that of 4-cyanotoluene. In liquid NH3, the addition of a
stoichiometric amount of NH4Br is sufficient to raise the electron
stoichiometry at the first wave up to 2. Under these conditions,
the 4-cyanotoluene wave is a one-electron reversible wave. The
(8) Petrosyan, V. E.; Niyazimbetv, M. E. Russian Chem. ReV. 1989, 58,
644.
(9) (a) Save´ant, J. M. J. Phys. Chem. 1994, 98, 3716. (b) Save´ant, J. M.
Acc. Chem. Res. 1993, 26, 455. (c) Save´ant, J. M. Dissociative Electron
Transfer. In AdVances in Electron Transfer Chemistry; Mariano, P. S., Ed.;
JAI Press: New York, 1994; Vol. IV, pp 53-116. (d) Save´ant, J. M.
Tetrahedron 1994, 50, 10117-10165.
(10) (a) Adcock, W.; Andrieux, C. P.; Clark, C. I.; Neudeck, A.; Save´ant,
J.-M.; Tardy, C. J. Am. Chem. Soc. 1995, 117, 8285. (b) Andersen, M. L.;
Mathivanan, N.; Wayner, D. D. M. J. Am. Chem. Soc. 1996, 118, 4871. (c)
Andrieux, C. P.; Save´ant, J.-M.; Tallec, A.; Tardivel, R.; Tardy, C. J. Am.
Chem. Soc. 1996, 118, 9788. (d) Andrieux, C. P.; Save´ant, J.-M.; Tallec,
A.; Tardivel, R.; Tardy, C. J. Am. Chem. Soc., in press.
+
depletion of NH4 at the electrode surface caused by the
protonation of the carbanion produced at the first wave thus
leaves a concentration of acid which is insufficient to perturb
the stability of the anion radical formed upon electron transfer
to 4-cyanotoluene at the level of the second wave. In DMF,
the addition of larger amounts of phenol is required to raise the
electron stoichiometry at the first wave up to 2. Under these

Dynamics of Bond-Breaking in Ion Radicals
J. Am. Chem. Soc., Vol. 119, No. 40, 1997 9529
Scheme 2
and to 75.2 mV at -38 °C.11 Both in DMF and in liquid NH3,
the value of these parameters lie in between the limiting values
with, as expected, a clear tendency of passing from cleavage to
electron transfer control upon raising the scan rate.11 The
competition between cleavage and electron transfer depends on
a dimensionless parameter p which is defined by the following
eq when R ) 0.5.
2
Figure 2. Variations of the peak potential (a, a′) and peak width (b,
b′) for the reduction of 1 mM 1′ in DMF + 0.1 M n-Bu4BF4 + 10 mM
PhOH at 20 °C (a, b) and of 3 mM 1′ in liquid NH3 + 0.1 M KBr +
kS
2RT
p )
x FV
x
kD
+
3 mM NH4 at -38 °C (a′, b′). The potentials are referred to SCE in
DMF and to 0.01 M Ag+/Ag in liquid NH3. The solid lines represent
the fitting of the experimental data.
From the theoretical variations of the peak potential and of
the peak width with the parameter p,11b one may fit the
experimental data in DMF (taking D ) 10-5 cm2 s-1) and in
Scheme 1
2
liquid NH3 (taking D ) 3 × 10-5 cm2 s-1) with kS /k1/2 ) 1.8
2
× 10-5 cm2 s-2.5 in the first case and kS /k1/2 ) 2.4 × 10-5
cm2 s-2.5 in the second. These fittings also provide the value
of E0 + (RT/2F) ln k, -1.82 V vs SCE for DMF and -1.365
1′
V vs 0.01 M Ag+/Ag for liquid NH3. Thus, if E01′ or k1′ could
be known independently, the above relationships would provide
the values of all three constants, E01′, k1′, and kS,1′. We recourse
to redox catalysis to obtain the lacking information.11c The
reduction is observed indirectly by means of a reversible couple,
P/Q, with a standard potential less negative than the reduction
potential of the compound under investigation (Scheme 2). The
conditions, there is enough proton donor present in the diffusion
layer to rapidly protonate the anion radical formed upon electron
transfer to 4-cyanotoluene at the level of the second wave and
thus to make it a two-electron irreversible wave.
The first cyclic voltammetric wave remains irreversible up
to scan rates as large as 1000 V/s. The variations of the peak
potential, Ep, and of the peak width, Ep/2 - Ep, with the scan
rate, V, (Figure 2) reveal that the reaction is jointly governed
by the cleavage step (rate constant: k1′) and the electrode
electron transfer step (E01′, kS,1′ R1′, standard potential, standard
rate constant, transfer coefficient, respectively) as shown in
Scheme 1.
A seen later on, the interference of reaction (2′) may be
neglected. In other words, the “ECE” pathway (0 + 1 + 2)
prevails over the disproportionation pathway (0 + 1 + 2′).11a,e
If the “ECE” reaction were to be controlled by the cleavage
step (1), with reaction (0) acting as a pre-equilibrium, ∂Ep/∂logV
would be equal to 29.1 mV at 20 °C and to 23.3 mV at - 38 °C,
while Ep/2 - Ep would be equal to 46.9 mV at 20 °C and to
37.6 mV at - 38 °C.11
Conversely, if the reaction were to be controlled by the
electron transfer step, for a transfer coefficient equal to 0.5, ∂Ep/
∂logV would be equal to 58.2 mV at 20 °C and to 46.7 mV at
-38 °C while Ep/2 - Ep would be equal to 93.8 mV at 20 °C
0
variations of the catalytic increase of the current, ip/ip , upon
addition of 1′ to a solution of P (ip and ip0 are the peak currents
in the presence and absence of the substrate, respectively) with
concentrations and scan rate provide the desired information.
As seen from Scheme 2, there are two limiting kinetic regimes
in the control of the catalytic response. If k1′ . k-0[P], the
catalytic current is governed by forward reaction (0′). Working
0
curves relating ip/ip to (RT/F)(k0[P]/V) for each value of the
excess parameter, [substrate]/[P], are available,11a,c allowing the
derivation of k0 from the experimental data (see in Supporting
0
Information how the rate constant was derided from the ip/ip
data). Conversely, when k1′ , k-0[P], the catalytic current is
governed by reaction (1), reaction (0′) then acting as a pre-
0
equilibrium. Working curves relating ip/ip to (RT/F) (k0k1′/
k-0V) for each value of the excess parameter, [substrate]/[P],
are available,11a,c allowing the derivation of k0k1′/k-0 from the
experimental data.
In DMF, with all four mediators indicated in Figure 3, the
catalytic process is under the control of forward reaction 0′ as
soon as [P] < 2.5 mM. The variation of log k0 with, the
standard potential of the mediator couple is close to 1/58.2
(mV-1) indicating that the reverse reaction is under diffusion
control reaction is under diffusion control. A procedure for
determining the standard potential E0 ensues as depicted in
1′
(11) (a) Andrieux, C. P.; Save´ant, J.-M. Electrochemical Reactions. In
InVestigations of rates and Mechanisms of Reactions; Bernasconi, C. F.,
Ed.; Wiley: New York, 1986; Vol. 6, 4/E, Part 2, pp 305-390. (b) Nadjo,
L.; Save´ant, J.-M. J. Electroanal. Chem. 1973, 48, 113. (c) Andrieux, C.
P.; Hapiot, P.; Save´ant, J.-M. Chem. ReV. 1990, 90, 723. (d) Amatore, C.;
Combellas, C.; Oturan, M. A.; Pinson, J.; Robvieille, S.; Save´ant, J.-M.;
Thie´bault, A. J. Am. Chem. Soc. 1985, 107, 4846. (e) Amatore, C.; Save´ant,
J.-M. J. Electroanal. Chem. 1977, 85, 27.
Figure 3a. From the value thus found, -2.02 V vs SCE, one
thus obtains, by use of the preceding peak potential and peak
width data, k1′ ) 7 × 106 s-1 and kS,1′ ) 0.2 cm s-1
.
In liquid NH3, with a mediator such as 4-cyanopyridine, the
0
variations of ip/ip with the mediator concentration (see Sup-
porting Information) indicate that there is mixed kinetic control

9530 J. Am. Chem. Soc., Vol. 119, No. 40, 1997
Andrieux et al.
Figure 3. Redox catalyzed reduction of 1′ in DMF at 20°C (a) and in
liq NH3 at -38 °C (b). In a, the mediators are, from left to right,
phthalonitrile (scan rates from 0.1 to 0.5 V/s), perylene (scan rates from
0.2 to 2 V/s), 4-cyanopryridine (scan rates from 2 to 20 V/s),
1-naphthonitrile (scan rates from 5 to 50 V/s), excess factor: from 0.5
to 5.3, mediator concentration: from 2 to 4.3 mM. In b, the mediator
is 4-cyanopryridine, excess factor: from 0.7 to 4.3, mediator concentra-
tion: from 2 to 4.3 mM, scan rates from 0.1 to 1 V/s.
by reactions (0′) and (1). The value of the apparent rate constant
kap defined as
k-0
1
1
)
+ 0.33
[P]
[1]
kap k0
k0k1′
0
can be derived from the forced treatment of the ip/ip data by
the working curves corresponding to pure kinetic control by
forward reaction (0).11d It is indeed found that the values of
1/kap thus determined vary linearly with the mediator concentra-
tion (Figure 3b). The validity of such an approximation rests
on the fact that the shapes of the working curves for both
limiting controls are almost the same. Replacing k0 by 3k0k1′/
k-0 in the first of these working curves gives a curve which is
practically identical to the second working curve. The intercept
of the linear plot (Figure 3b) provides k0. Assuming that reverse
reaction (0′) is under diffusion control thus leads to E01′ ) -1.54
V vs 0.01 M Ag+/Ag. The slope of the linear plot finally leads
to k1′ ) 2.2 × 107 s-1. Thus, the derivation of E01′ and k1′, i.e.,
the parameters of main interest, does not require the use of the
peak potential and peak width data. It is however interesting
to check that the values thus obtained are consistent with the
Figure 4. Reduction of 1 mM 1′′ in DMF + 0.1 M n-Bu4BF4 at 20
°C (a) in the absence and (b) presence of acid (10 mM PhOH). Scan
rate: (a) 0.2 V/s and (b) from bottom to top: 0.2, 0.5, 1, 2, V/s. Full
lines: experimental curves, open circles: simulation according to
Scheme 3. Reduction of 3 mM 1′′ in liquid NH3 + 0.1 M KBr + 3
+
mM NH4 at -38°C at 0.2 V/s (a′). Variations of the peak potential
(b′) and peak with (c′) with the scan rate.
as attested by the reversible one-electron wave following the
first wave (Figures 4a,a′). In DMF, the addition of an acid such
as phenol triggers a modest but distinct increase of the first
wave, while the second wave becomes irreversible and increases
up to a two-electron stoichiometry. In liquid NH3, the presence
of an equimolar amount of NH4+ leaves the second wave, which
corresponds to the reduction of 4-cyanotoluene formed at the
first wave, reversible as in the case of 1′. In DMF, the first
wave exhibits, at all investigated scan rates (0.1-10 V/s), a
shoulder at potential ca. 70 mV more positive than the peak
which is itself located at a potential very slightly more positive
(ca. 30 mV) than the peak of 1′ (Figures 4a, b). Under these
conditions, the peak height is not a straightforward indication
of the electron stoichiometry. This is however clearly above 2
by comparison to the height of the one-electron reversible wave
of 4-cyanotoluene. These observations points to the interme-
diacy of 1′ in the reductive cleavage of 1′′ and thus to the
reaction mechanism depicted in Scheme 3. In liquid NH3, there
is no splitting of the first wave. However, the electron
stoichiometry is also clearly above 2 and the peak width is large
(Figure 4a′,c′) pointing to the intermediacy of 1′ in the reductive
cleavage of 1′′ in this case too. We may thus assume that the
reaction mechanism still follows Scheme 3.
value of E0 + (RT/2F) ln k (-1.365 V vs 0.01 M Ag+/Ag)
1′
derived from these data. This is indeed the case since the value
of the same potential found from the redox catalysis data is
-1.369 V vs 0.01 M Ag+/Ag. The value of kS,1′ can then be
obtained, kS,1′ ) 0.34 cm s-1
.
As a test of consistency, the voltammograms obtained at 0.2
V/s, in the presence of an acid (Figure 1), have been simulated
using the values of the various parameters that we just derived.
As seen in Figure 1, the agreement between simulation and
experiment is excellent. It should be however emphasized that
redox catalysis experiments and systematic determination of the
variation of the peak potential and peak width with the scan
rate were necessary to obtain the exact values of E01′, k1′, and
kS,1′. The assumption, made at the start of this kinetic analysis
that an “ECE” process is followed rather than a “DISP” process
may be justified as follows. The parameter governing the
competition between the two pathways is (k2′/k1′3/2) [1′] (FV/
RT)1/2 11a,e
,
where k2′ is close to the diffusion limit, 5 × 109 M-1
s-1 in DMF and 3 × 1010 M-1 s-1 in liquid NH3. The value of
the parameter ranges from 9.7 × 10-4 to 1.4 × 10-2 in DMF
and 1.9 × 10-3 to 2.8 × 10-2 in liquid NH3 when the scan rate
varies from 0.2 to 2 V/s as in the experiments shown in Figure
2 indeed pointing to a large predominance of the ECE pathway.
Compound 1′′. The reduction of 1′′ also leads to the
formation of 4-cyanotoluene, both in liquid NH3 and in DMF,
Additional information may be derived from redox catalysis
experiments. Using the same mediators as already used with
1′, with the same scan rates, we found that the catalytic
efficiency is larger than the maximal efficiency predicted for a
two-electron stoichiometry of the reductive cleavage of 1′′ by
the reduced form of the mediator. Taking account of the fact

Dynamics of Bond-Breaking in Ion Radicals
J. Am. Chem. Soc., Vol. 119, No. 40, 1997 9531
Scheme 3
Scheme 4
We may thus combine this relationship with the shape and
location of the cyclic voltammograms at various scan rates to
derive the values of E0 , k1′′, and of the heterogeneous electron
1′′
transfer rate constant, kS,1′′. The voltammograms simulated
according to this procedure in the framework of Scheme 3 are
shown together with the experimental curves in Figure 4b. It
was thus found that E01′′ ) -1.89 V vs SCE, k1′′ ) 2 × 105 s-1
,
kS,1′′′ ) 0.35 cm s-1. It is worth noting that the results of the
simulations do not change appreciably when the homogeneous
electron transfer steps (2′), (4′), and (6′), where the electron
donor is the anion radical of 1′′, are not included in the
simulation. This observation points to the conclusion that
•
electron transfer to ArCHF•, ArCH2F, and ArCH2 takes place
predominantly at the electrode surface according to reactions
(2), (4), and (5). In other words the various follow-up electron
transfer steps take place according to an “ECE” process rather
than to a “DISP” process.11a,e The parameter governing the
competition between the two pathways is (k2′/k1′′3/2)[1′′](FV/
RT)1/2 where k2′ is close to the diffusion limit, 5 × 109 M-1
s-1. The value of the parameter in the present case ranges from
0.15 to 0.5 when the scan rate varies from 0.2 to 2 V/s as in the
experiments shown in Figure 4b indeed pointing to the
predominance of the ECE pathway. In other words, the
homogeneous “internal” redox catalysis of the reduction of 1′
by the anion radical of 1′′ is inefficient in the present case
because of its instability toward fluoride cleavage.
that the reduction of 1′′ takes place at a potential more positive
than the reduction potential of 1′, this observation confirms the
intermediacy of 1′ in the reductive cleavage of 1′′ thus pointing
to the catalytic mechanism depicted in Scheme 4. In order to
obtain information pertaining solely to the reduction of 1′′, we
used scan rates large enough for the catalytic response of 1′ to
be negligible as anticipated from the experiments previously
performed with 1′ alone. In DMF, it was found that the catalytic
response obtained under these conditions, with the three
mediators, 4-cyanopyridine (E0PQ ) -1.71 V vs SCE), phthalo-
The data for liquid NH3 (see Supporting Information) were
treated in the same way as for DMF. Using 4-cyanopyridine
as a mediator, it was again found that, provided [P] > 2 mM,
the catalytic response is governed by reaction (1) with (0′) acting
as a pre-equilibrium. Over the whole range of excess factors
(0.5-3.3), mediator concentrations (2-5.8 mM), and scan rates
(2-10 V/s), it was found that (k0,1′′/k-0,1′′)k1′′ ) 37 s-1, and
thus, since EP0Q ) -1.25 V vs 0.01 M Ag+/Ag, E01′′ + RT/F
ln(k1′′) ) -1.18 V vs 0.01 M Ag+/Ag. The simulation of the
cyclic voltammetric responses, using this relationship, are shown
in Figures 4b′,c′, as variations of the peak potential and peak
width with the scan rate. We thus obtained E01′′ ) -1.45 V vs
nitrile (E0PQ ) -1.57 V vs SCE), and quinoxaline (E0PQ
)
-1.525 V vs SCE), is governed by reaction 1 with 0′ acting as
a pre-equilibrium (Scheme 4), provided [P] > 2 mM (see
Supporting Information). The value of (k0,1′′/k-0,1′′)k1′′ could
thus be derived with each catalyst (excess factor ranging from
0.7 to 4.1, mediator concentration ranging from 2 to 5.5 M).
Since
k0,1′′
RT
F
RT
F
ln
k1′′ ) E01′′ - E0PQ
+
ln(k1′′)
(
)
0.01 M Ag+/Ag, k1′′ ) 1 × 106 s-1, kS,1′′ ) 0.6 cm s-1
.
k-0,1′′
Compound 1′′′. The preceding results indicate that the anion
radical cleavage rate constant decreases from 1′ to 1′′. We may
thus expect a further decrease when going to 1′′′. We indeed
observed, both in DMF and in liquid NH3, that the reduction
wave of 1′′′ becomes reversible upon raising the scan rate
(Figure 5), thus offering a new strategy for determining the
it was found that
RT
F
E01′′
+
ln(k1′′)
) -1.53 V vs SCE in average.

9532 J. Am. Chem. Soc., Vol. 119, No. 40, 1997
Andrieux et al.
Table 1. Standard Potentials and Cleavage Rate Constants of
4-Cyanofluorotoluenes
medium
DMF at 20 °C
liquid NH3 at -38 °C
c
c
compd
E0 a
kcleav
E0 b
kcleav
1′
1′′
1′′′
-2.020
-1.890
-1.785
7 × 106
4 × 105
3.8 × 101
-1.540
-1.450
-1.345
2.2 × 107
3.5 × 102
1 × 106
a V vs SCE. b V vs 0.01 M Ag+/Ag. c In s-1
.
Table 2. Standard Potentials and Cleavage Rate Constants of the
Trifluoromethylarenes in Liquid NH3 at -38 °C
b
b
compd
E0 a
kcleav
compd
E0 a
kcleav
1′′′
2
3
4
5
-1.345
-1.480
-1.385
-1.330
-1.405
3.5 × 102
1 × 101
1.5 × 102
7.5
6
7
8
9
-1.835
-2. 180
-1.650
-2.075
1.1 × 104
2.8 × 108
2 × 107
1 × 107
2.5 × 101
a V vs 0.01 M Ag+/Ag. b In s-1
.
The standard potentials and cleavage rate constants for 1′,
1′′, and 1′′′ are listed in Table 1.
Other Compounds. We now describe the cyclic voltam-
metric behavior of the other trifluoro-compounds (see Chart 1),
for which the corresponding difluoro- and monofluoro-
compounds were not available, with the aim of determining the
standard potential for the formation of the anion radical and its
cleavage rate constant in liquid ammonia. Among them,
compounds 2-5 resemble compound 1′′′. At low scan rate,
they likewise exhibit a large irreversible wave, with a stoichi-
ometry corresponding to six electrons per molecule, followed
by a reversible one-electron wave corresponding to the ArCH3
compound formed at the first wave upon reductive cleavage of
the three fluorine atoms (Figure 6). Also, like with 1′′′, the
cleavage rate constant is not very large, and thus, reversibility
does not require a very high scan rate to be reached (Figure 6).
The ensuing values of the standard potential are listed in Table
2. We thus expect the reductive cleavage of these four
compounds to take place according to the internal redox catalysis
mechanism. That this is indeed the case results from the good
fit of the curves simulated according to this mechanism with
the experimental curves (Figure 6), thus leading to the values
of the cleavage rate constant listed in Table 2. With 4, the
second wave is large, poorly defined and irreversible unlike the
second wave of 1′′′, 2, 3, and 5. It nevertheless corresponds to
the reduction of the 4-carboxymethyl toluene as checked with
an authentic sample of the latter compound. The reason for
this particular behavior of the 4-carboxymethyl toluene in liquid
ammonia (in DMF it gives rise to a one-electron reversible
wave) is not known for the time being. We may finally estimate
the value of the parameter that governs the competition between
the internal redox catalysis mechanism and the ECE mechanism,
(k2′/k1′′3/2)[1′′](FV/RT)1/2, for 2, 3, 4, and 5, 9 × 106, 2 × 105,
107, 7 × 104, respectively, thus showing that the former
mechanism largely overruns the latter in all four cases.
Figure 5. Reduction of 1′′′ in DMF + 0.1 M n-Bu4BF4 at 20 °C (a,
b) + 10 mM PhOH (c), and in liquid NH3 + 0.1 M KBr + 3 mM
NH4Br] at -38 °C (a′, b′, c′). [1′′′] ) 1 mM (a, b, c), 3 mM (a′, b′, c′).
Scan rate: 0.2 (a, a′, c, c′), 50 (b), and 300 (b′) V/s. The potentials are
referred to SCE in DMF and to 0.01 M Ag+/Ag in liquid NH3. Solid
lines: experimental data. Open circles: simulation according to Scheme
5. Open triangles: simulation according to the “internal” redox catalysis
mechanism.
mechanism and characteristics of the reductive cleavage of this
compound. At low scan rates, the first wave, which is then
irreversible, corresponds to the exchange of ca. six electrons
per molecule, pointing to the mechanism depicted in Scheme
5.
Once the standard potential E01′′′ has been determined as the
midpoint between the cathodic and anodic peak of the one-
electron reversible wave obtained upon raising the scan rate,
the value of k1′′′ and kS,1′′′ can be derived from the simulation
of the irreversible cyclic voltammograms obtained at low scan
rates according to Scheme 5, taking into account the previously
derived values of E01′′, k1′′, kS,1′′ and of E01′, k1′, kS,1′
.
Examples of the fit between simulations and experiments are
shown in Figure 5c,c′. We thus found E0 ) -1.785 V vs
1′′′
SCE, k1′′′ ) 38 s-1, kS,1′′′ ) 0.1 cm s-1 in DMF and E0
)
1′′′
-1.345 V vs 0.01 M Ag+/Ag, k1′′′ ) 350 s-1, kS,1′′′ ) 0.25 cm
s-1 in liquid NH3.
Since the cleavage rate constants are small in both media,
we expect the homogeneous electron transfer steps (2′), (4′),
(6′), (8′), and (10′) to prevail over their heterogeneous coun-
terparts. The parameter (k2′/k1′′′3/2) [1′′′](FV/RT)1/2 is indeed
equal to 1.2 × 105 in DMF and to 4.3 × 104 in liquid NH3 in
the experiments shown in Figure 5 (parts c and 5c′ respectively).
Accordingly, simulation taking only into account the homoge-
neous electron transfer steps produces curves that were found
to be in good agreement with the experimental curves as well
as with the theoretical curves involving the complete set of
reactions in Scheme 5. “Internal” redox catalysis of the
reduction of 1′′ and 1′ by the anion radical of 1′′′ thus completely
overcomes their direct reduction at the electrode.
The cyclic voltammetry of 6 has in common with that of 1′′′
and 2-5 that it exhibits an irreversible wave with a ca. six-
electron stoichiometry at low scan rate (Figure 7). There are
however two important differences. One is that there is no
second wave corresponding to the reduction of the defluorinated
product formed at the first wave. The lack of a second wave is
due to the fact that the 4-methyl-benzoate anion does not exhibit,
because of the negative charge it bears, any reduction wave prior
the discharge of the supporting electrolyte. The second differ-
ence is that the 6e-wave remains irreversible over the available

Dynamics of Bond-Breaking in Ion Radicals
J. Am. Chem. Soc., Vol. 119, No. 40, 1997 9533
Scheme 5a
a The homogeneous electron transfer steps are assumed to involve only the anion radical of ArCF3 as electron donor ((2′), (4′), (6′), (8′), (10′))
and not the anion radical of ArCHF2 and anion radical of ArCH2F in accordance with the fact that the latter species cleave much more rapidly than
the former.
Figure 6. Cyclic voltammetry of 2, 3, 4, and 5 (3 mM) in liquid NH3 + 0.1 M KBr + 3 mM NH4Br] at -38 °C. Open circles: simulation
according to the “internal” redox catalysis mechanism.
range of scan rates (0.2-200 V/s) indicating that the cleavage
of the anion radical of ArCF3 is faster. It follows that the
standard potential cannot be derived from cyclic voltammetric
experiments, thus preventing the application of the procedure
used with 1′′′ and 2-5. The lacking piece of information can
however be obtained from redox catalysis as described below.
8 and 9 also give rise (Figure 7) to a single wave which
remains irreversible over the available range of scan rates. As
with 6, the totally defluorinated products do not exhibit reduction
waves within the available potential range. There is however
an important difference with 6, namely that the reduction wave
is a composite wave pointing to an ECE mechanism in which
the di- and monofluoro products formed from the initial trifluoro
compound are successively reduced at the electrode surface. In
these two cases the instability of the primary anion radical
prevents the internal redox catalysis mechanism to compete with
the ECE mechanism. With 6, the reduction wave does not show
any splitting leading to the conclusion that the internal redox
catalysis mechanism overruns the ECE mechanism in this case.
We will see, once the cleavage rate constants will be determined,
that the cleavage of the anion radical of 6, albeit faster than
with 1′′′ and 2-5, is indeed slower than with 8 and 9 thus giving

9534 J. Am. Chem. Soc., Vol. 119, No. 40, 1997
Andrieux et al.
Figure 8. Figure 8. Redox catalysis of the reduction of compounds
7, 8, and 9 in liquid NH3 + 0.1 M KBr at -38 °C. Variation of the
apparent rate constant (eq [1]) with the mediator concentration.
0
a
point
compd
7
mediator
benzonitrile
4-toluonitrile
dimethylquinoxaline
4,4′-bipyridine
benzonitrile
-EPQ
9
0
b
O
2
1.785
1.870
1.300
1.330
1.785
8
9
a V vs 0.01 M Ag+/Ag.
between E0 and k derived from redox catalysis finally lead to
E0 ) -1.830 V vs 0.01 M Ag+/Ag, k ) 9 × 103 s-1, kS ) 0.1
cm s-1. This value of k is consistent with the fact that the redox
catalysis is under the kinetic control of reaction 1 with 0′ acting
as a pre-equilibrium. Indeed, k-0[P]/k ) 104 (k-0 is certainly
close to the diffusion limit, 3 × 1010 M-1 s-1). We may also
estimate the value of the parameter that governs the competition
between the internal redox catalysis mechanism and the ECE
mechanism in the direct electrochemical reduction of 6, (k2′/
k1′′3/2)[1′′](FV/RT)1/2. This is equal to 3 × 102, which confirms
that the former mechanism largely predominates over the later.
Redox catalysis needed also to be used in the case of 8 and
9 for which k is expected to be even larger than with 6. This
is also the case with 7, for which, in addition, self-inhibition
heavily perturbs the cyclic voltammetric curves so as to render
the extraction of kinetic information untractable. The variation
of the catalytic responses (see Supporting Information) with the
mediator concentration showed that catalysis is under mixed
kinetic control by reactions (0′) and (1). We may therefore treat
the data by means of eq [1] as summarized in Figure 8 leading
to the values of E0 and k listed in Table 1.
In the reductive cleavage of the di- and trifluoro derivatives,
we did not consider so far another mechanistic possibility
namely the intermediacy of carbenes. Such mechanisms would
involve the replacement of reactions (3), (4), (4′), and (5) in
Scheme 3 by reactions (3), (4), (4′), and (5) in Scheme 7.
Formation of carbenes as intermediate in the electrochemical
reduction of gem-dichloro, dibromo, and chlorobromo com-
pounds has precedency.8 The possibility of such a reaction with
our di- and trifluoro derivatives should therefore be examined,
even if it has never been reported so far. Reduction potentials
of carbenes are not known. However, being electron-deficient
species they may be easier to reduce (eqs [4] and [4′]) than the
starting fluoro compound. The carbene anion radical thus
formed would then be protonated (reaction (5)) giving rise to a
radical which is easier to reduce than the starting fluoro
compound yielding eventually the defluorinated product. If the
carbene mechanism were to be followed, the monofluorinated
derivative would not be an intermediate in the reductive cleavage
of the difluoro compounds, and the difluorinated derivative
would not be an intermediate in the reductive cleavage of the
trifluoro compounds. The carbene mechanism may thus be
ruled out in the case of 1′′ since the reduction wave of 1′ appears
in its cyclic voltammogram. It may likewise be ruled out in
Figure 7. Cyclic voltammetry of 6, 8, and 9 (3 mM) in liquid NH3 +
0.1 M KBr at -38 °C. Open circles: simulation according to the
“internal” redox catalysis mechanism.
more chance to the internal redox catalysis mechanism to
compete with the ECE mechanism. We may thus simulate the
low scan rate cyclic voltammograms of 6 according to the
internal redox catalysis mechanism. However, since the stan-
dard potential could not be derived from cyclic voltammetry,
we need additional information to derive the values of the
standard potential and of the cleavage rate constant. This can
be obtained from redox catalysis. As a mediator we used
pyridazine. Because its standard potential (-1.565 V vs 0.01
M Ag+/Ag) is rather positive, the catalytic increase of the current
is rather modest. Under these conditions, the scan rate could
be adjusted so as to render negligible the interference of electron
transfer from the reduced form of the catalyst, Q, to the difluoro
and monofluoro products, thus attempting to use the same
strategy that was successfully employed in the case of 1′′.
However, even if Q is unable to reduce the difluoro and
monofluoro products, the anion radical of 6, produced by
reaction with Q, may well reduce these products as we know
from the observation that the direct reduction of 6 occurs
predominantly through the internal redox catalysis mechanism.
We are thus led to treat the redox catalysis data in the framework
of the mechanism depicted in Scheme 6. It was found that the
catalytic increase of the current is independent of the catalyst
concentration showing that the kinetics are governed by the
factor k k0/k-0. Its value, derived from the appropriate working
curves,12 is 0.0017. (RT/F) ln(k k0/k-0) ) E0 - EP0Q + (RT/F)
ln k. Thus E0 + (RT/F) ln k ) -1.647 V vs 0.01 M Ag+/Ag.
Potentially, the cyclic voltammogram of 6 (Figure 7) contains
two relationships between E0, k, and kS. Simulation of the
voltammogram in Figure 7 according to the internal redox
catalysis mechanism, taking into account the relationship
(12) (a) The working curves for a given excess factor, [ArCH3]/[P], are
the same as the working curves pertaining to a simple EC mechanism for
an excess factor of 6 [ArCH3]/[P].10a,11b For 1′′, 7, 8, and 9, the working
curves for a given excess factor [ArCH3]/[P] are the same as the working
curves pertaining to a simple EC mechanism for an excess factor of 2
[ArCH3]/[P],10a,11b for mediators giving rise to a small catalytic effect. (b)
Andrieux, C. P.; Dumas-Bouchiat, J.-M.; Save´ant, J.-M. J. Electroanal.
Chem. 1978, 87, 55.

Dynamics of Bond-Breaking in Ion Radicals
J. Am. Chem. Soc., Vol. 119, No. 40, 1997 9535
Scheme 6
Scheme 7
the case of 8 and 9 since both the difluoro and monofluoro
compounds appear in their voltammograms. All other trifluoro
compounds show a single 6e wave. Their reductive cleavage
may go through the di- and monofluoro derivatives in the
framework of the internal redox catalytic mechanism discussed
earlier. However, it may also a priori follow the carbene
mechanism. A first argument against the latter possibility is
that there no obvious reasons that it would be followed with
1′′′ and 2-6 and not with 1′′, 8, and 9.
We investigated the possible occurrence of the carbene
mechanism in more details in the case of 1′′′. Phenylfluoro-
carbene has been previously obtained by deprotonation of R,R′-
bromofluorotoluene. Because of its electrophilic character it
reacts quantitatively with the electron rich olefin, 2,3-dimethyl-
2-butene. Since 4-cyanophenylfluorocarbene should be even
more electrophilic, we have selected the same olefin to attempt
a trapping experiment (reaction 8 in Scheme 7). 1′′′ was
electrolyzed (see Experimental Section) in the presence of a
large excess of 2,3-dimethyl-2-butene. The only product found
was 4-toluonitrile. The lack of reaction may be explained by
the observation that no reaction occurs when 1′′ is deprotonated
by t-BuO- in the presence of the same olefin (see Experimental
Section), in exactly the same conditions as those of the
experiment described above where quantitative carbene addition
is observed with bromofluorotoluene as the substrate. We may
thus conclude that the carbene does not form from the base of
1′′ presumably because the C-F bond is much stronger than
the C-Br bond. These observations confirm that the carbene
mechanism is not followed in the reductive cleavage of the
trifluoro compounds we have investigated.

9536 J. Am. Chem. Soc., Vol. 119, No. 40, 1997
Andrieux et al.
Discussion
The mechanism of the reductive cleavage of the compounds
investigated has been established in the preceding section.
Besides its own interest, the knowledge of the mechanism was
necessary to derive correct values of the standard potential and
the cleavage rate constant of the primary anion radical. These
are gathered in Tables 1 and 2.
We first discuss the reactivity-structure relationships for the
reductive cleavage of the three 4-cyanofluorotoluenes. A first
observation is that, in both solvents, the cleavage rate constant
decreases when the number of fluorine increases. The intramo-
lecular dissociative electron transfer model of anion radical
cleavage9 may be applied to explain this observation. The
predictions of the model are contained in the following eqs
RT
F
RT kT
ln - ∆Gq
ln kcleav
)
[2]
F
h
(in the following, energies are expressed in eV and potentials
in V)
Figure 9. Correlation between the activation free energy of the anion
radical cleavage of (from left to right) 1′, 1′′, and 1′′′ with the standard
2
∆G0
∆G0
2
∆Gq ) ∆Gq0 1 +
≈ ∆Gq0 +
[3]
potential E0
(a, a′) and the function 0.75DFC + 0.5E0
(b,
RX/RX•
RX/RX•
q
(
)
4∆G0
b′), in DMF at 20 °C (b) and liquid NH3 at -38 °C (9).
The standard potential E0RX/RX•- becomes more and more
positive from 1′ to 1′′ and 1′′′ (Table 1) as a consequence of
the increasing electron-withdrawing effect of the fluoromethyl
group on the π* orbital of the aromatic moiety. The driving
force of the cleavage reaction decreases accordingly (eq [4]).
This is a first reason why the cleavage rate constant decreases
from 1′ to 1′′ and 1′′′. If the C-F bond dissociation energy
and the intrinsic barrier were to remain constant in the series,
the linearized version of eqs [3] and [4] would imply that the
activation free energy should vary linearly with the driving force
with a slope of 0.5:
The second, approximate, version of eq [3] is valid as long
as the driving force of the reaction, -∆G0, is small as compared
to the total reorganization energy, 4∆Gq0. The driving force
may be expressed by eq [4] as a function of the bond
dissociation energy of the starting compound, DRX, of the
standard potential for the formation of the anion radical,
E0RX/RX and the standard potential for the oxidation of the
•-
leaving anion X-, (the s’s are the molar entropies of the
subscript species).
∆G0 ) DRX + ER0 X/RX•- - E0 - - T(s • + s • - sRX) [4]
•
X /X
R
X
The intrinsic barrier free energy, ∆Gq0 is the sum of two
terms. One, ∆Gq0,intra, deals with the intramolecular nuclear
rearrangement accompanying the cleavage of the C-X bond.
It may be expressed by eq [5] as a function of the bond
dissociation energy of the starting compound, DRX, of the
standard potential for the formation of the anion radical,
E0
∆Gq ) RX/RX•- + Cst
2
As seen in Figure 9a,a′, this prediction is not correct in both
solvents. The experimental activation free energy varies more
than predicted from the mere variation of the standard potential.
There is evidence that the C-F bond dissociation energy varies
with the number of fluorine atoms borne by the functional
carbon. Quantitative data are available for gem-fluoroalkanes,
e.g., for gem-fluoroethanes (4.58, 4.70, 4.86 eV for -CH2F,
-CHF2, and -CF3, respectively).14 The bond dissociation
energy interferes both in the driving force (eq [4]) and in the
intrinsic barrier. Using the linearized version of eq [3], we may
thus attempt a correlation of the experimental activation free
E0RX/RX and the standard potential for the oxidation of R-
•-
4∆Gq0,intra ) DRX•- ) DRX + E0RX/RX•- - ER0 •/R-
+ T(sRX - sRX•- + s - - s )
•
R
R
) DRX + (hRX - hRX•-) - (h • - h )
[5]
-
R
R
(the h’s are the molar enthalpies of the subscript species).
The other, ∆Gq0,solv, involves the solvent reorganization that
arises from the displacement of the negative charge from the
aromatic portion of the molecule to the leaving fluoride ion.
∆Gq0,solv may thus be estimated from eq [6]
energy with 0.75DFE + 0.5E0RX/RX (where the DFE’s are the
•-
bond dissociation energies of the gem-fluoroethanes). Such an
attempt implies several assumptions: (i) the increments in the
BDEs of 1′, 1′′, 1′′′ are the same as for the corresponding
fluoroethanes; (ii) the entropic term in eq [4] is the same for
the three compounds (it essentially corresponds to the formation
of two particles out of one and should not therefore vary much
e02
1
1
1
1
1
∆Gq0,solv
)
+
-
-
X
-
[6]
(
)(
)
4 2aRX
2a ,
d D op D s
•-
•
R
•-
•
-
in the series); and (iii) (hRX - hRX ) - (hR - hR
+
e0 is the electron charge, aRX is the radius of the sphere
∆Gq0,solv is about constant. The first term increases from 1′ to
1′′′ because of the increasing electron-withdrawing effect of the
fluoromethyl group on the LUMO energy. The second term
increases in a parallel fashion because the increasing electron-
•-
approximating the region of the starting anion radical where
•
-
the charge is initially located, and aR ,X , the radius of the sphere
approximating the location of the charge on the leaving group
when the bond is broken. d is the distance between the centers
of charge. The D ’s are the optical and static dielectric constants,
respectively.
(13) (a) Moss, R. A.; Przybyla, J. R. Tetrahedron 1969, 25, 647. (b)
Moss, R. A. Acc. Chem. Res. 1980, 13, 58.
(14) Joshi, R. M. J. Macromol. Sci.-Chem. 1974, A8, 861

Dynamics of Bond-Breaking in Ion Radicals
J. Am. Chem. Soc., Vol. 119, No. 40, 1997 9537
electron-withdrawing group, 1′′′, 3, 4, and 5, fall below this
line with the two compounds bearing an ester group as para
substituent being located on another 0.5 slope straight line
situated 0.08 eV below the main line. The compound with a
para cyano substituent is located 0.16 eV below the line and its
ortho isomer, 0.12 eV. The pyridine compound, 8, is located
further below (at a distance of 0.22 eV from the main line).
It has been shown previously that the presence of electron-
withdrawing groups in the para or ortho position of benzyl
bromide and chloride decreases the bond dissociation energy
substantially (by 0.25-0.30 eV) while electron-donating groups
have little influence.15 This effect has been interpreted as the
result of a destabilization of the starting molecule (rather than
of a stabilization of the benzyl radical) due to the opposing
polarizations introduced by the electron-withdrawing substituent
and the CH2X group when they are located in resonance
interacting positions.
We may therefore assign the downward location of 1′′′, 3, 4,
and 5 in Figure 10a to the same destabilizing effect. The ortho
cyano derivative 3 is located slightly above the para derivative
1′′′ because of through-space effects in the former case. 4 and
5 are located above 1′′′ because the electron-withdrawing power
of ester substituent is less than with a cyano group. The strong
electron-withdrawing power of the pyridine ring, explaining why
8 is located below all other compounds, matches the greater
reducibility of the pyridine ring as compared to the benzene
ring. Application of the linearized version of eqs [3] and [4]
leads to estimating the decrease of the BDE of these compounds
compared to that of the unsubstituted compound 7, as 4/3 of
the vertical distance between the two straight lines, i.e., 0.20,
0.16, and 0.10 eV, respectively. The same explanation holds
for 8 with a larger distance, 0.22 eV, between the data point
and the main line leading to a decrease of the BDE of 0.30 eV.
In this framework, the alignment of 2 and 9 with the
unsubstituted compound 7 derives from the lack of resonance
interaction between the electron withdrawing group (CN and
pyridine) and the CF3 group. In the case of 6, the para effect
observed with 1′′′, 3, 4, and 5 is counteracted by the negative
charge borne by the carboxylate group.
Figure 10. Cleavage of the anion radical of the trifluoromethylarenes
in liquid NH3 at -38 °C. Correlation between the activation free energy
with the standard potential (in V vs 0.01 M Ag+/Ag) (a) and the reaction
standard free energy (b).
withdrawing effect of the fluoromethyl group renders the
oxidation of the anion R- more and more difficult. It may thus
be considered that the variation of the first term is compensated
by the variation of the second. In the expression of ∆G0q,solv
,
the most important factor in the first parenthesis is the second
term because the radius of the fluoride anion is smaller than
the radius pertaining to the anion radical and the center-to-center
distance. The third term in ∆Gq0 may therefore be regarded as
approximately constant. As seen in Figure 9b,b′, there is indeed
a satisfactory linear correlation between the experimental
activation free energy and 0.75DFE + 0.5E0RX/RX•-. A more
quantitative analysis of the activation vs driving force relation-
ship will be presented later on, after examination of the rate
data pertaining to the trifluoro compounds in liquid NH3 (Table
2). Another striking feature of the cleavage kinetics of the anion
radicals of 1′, 1′′, and 1′′′, is that the activation free energies
are smaller in liquid NH3 than in DMF (Figure 9) in spite of a
50 °C difference in temperature. As discussed later on, in more
quantitative terms, this larger cleavage reactivity in liquid NH3
derives essentially from the stabilization of F- by NH3 as
compared to DMF, resulting in a driving force advantage (eq
[4]) which is not compensated by the decrease in temperature.
NH3 is indeed a more protic solvent than DMF, hence leading
to a H-bonding stabilization of anions such as F-.
The quantitative application of the intramolecular dissociative
electron transfer model requires the estimation of the standard
free energy of the cleavage reaction. This can be done, using
eq 4, for the unsubstituted compound, 7. The first term can be
obtained as follows. We first estimated the bond dissociation
energy of PhCH2-F from
DPhCH -F ) ∆fHPhCH • + ∆ H - ∆ HPhCH F ) 4.06 eV
•
F
f
f
2
2
2
(the ∆fH’s are the enthalpies of formation of the subscript
species) using literature data.16
We then assume that the increase of the BDE when going to
the CHF2 and CF3 derivatives is the same as in gem-fluoral-
kanes.14 Thus, DPhCHF-F ) 4.18 eV and DPhCF -F ) 4.34 eV.
2
The last term in eq 4 represents the entropy increase corre-
sponding to the formation of two particles out of one. Thus,
∆S is approximately 0.6 meV K-1 10d and the entropic term is
0.14 eV. The second term in eq 4 was determined experimen-
tally. The last factor to be estimated is the standard potential
for the oxidation of F- in liquid NH3 at 235 K. The standard
potential at 298 K was derived according to the same procedure
We now examine the rate data pertaining to the anion radical
cleavage in the whole series of trifluoromethylarenes in liquid
NH3 at -38 °C (Table 2). Figure 10a shows that there is a
rough correlation between the activation free energy and the
standard potential for the formation of the anion radical (the
more negative the later the smaller the former) as expected from
the attending variation of the driving force (eq [4]). It is
remarkable that the data points of 2, 6, 7, and 9 sit on a single
straight line with a slope of 0.5, indicating that the C-F bond
dissociation energy is closely the same for these four com-
pounds. It is also noteworthy that the data points for the
compounds substituted in the para or ortho position by an
(15) (a) Clark, K. B.; Wayner, D. D. M. J. Am. Chem. Soc. 1991, 113,
9363. (b) Andrieux, C. P.; Le Gorande, A.; Save´ant, J. M. J. Am. Chem.
Soc. 1992, 114, 6892.
(16) (a) Benson, S. N. Thermochemical kinetics; Wiley: New York, 1976.
(b) Handbook of Chemistry and Physics, 72nd ed.; CRC: Cleveland, OH,
1991-1992.

9538 J. Am. Chem. Soc., Vol. 119, No. 40, 1997
Andrieux et al.
Table 3. Standard Free Energies and Intrinsic Barrier Free
Energies for the Cleavage of the Anion Radicals of
4-Cyanofluorotoluenes
intrinsic barrier free energy may then be derived from the
application of the linearized version of eq 3 (Table 3). It is
practically constant over the three compounds, 1′, 1′′, and 1′′′.
The constancy of this parameter points to the conclusion that
the solvent reorganization energy is about constant and that the
variations of the standard potentials for the oxidation of the
carbanion (which both shift in the positive direction from 1′ to
1′′ and 1′′′) compensate each other.
medium
DMF at 20 °C liquid NH3 at -38 °C
∆G0 a ∆Gq0
∆G0 a ∆Gq0
a
a
compd
BDEa
1′
1′′
1′′′
3.86
3.98
4.14
-0.93
-0.68
-0.41
0.85 -1.35
0.91
0.88
0.92
0.90
(0.02
0.75 -1.14
0.805 -0.88
0.805
A similar analysis can be carried out for the results obtained
in DMF with the same compounds. The standard potential for
the oxidation of F-, EF0,2/F98,DKMF ) 2.59 V vs SCE, was obtained
(0.05
•
-
a In eV.
from the available free energy of transfer from water to
DMF.17c,d The values of the reaction standard free energies and
intrinsic barrier free energies ensue (Table 3). The intrinsic
barrier free energy is again ca. constant over the series pointing
to the same conclusions as in liquid NH3.
as already used for other anions.17a It was thus found that
0,298 K
F /F ,NH3
E
) 3.33 V vs 0.01 M Ag+/Ag. The value at 235 K
•
-
0,235 K
was then found to be E
Thus, E
) 3.53 V vs 0.01 M Ag+/Ag.17b
•
-
F /F ,NH3
0,235 K
PhCF3,NH3
) -1.51 eV. We may therefore recast the
We also note that the intrinsic barrier free energies are not
very different when going from liquid NH3 at -38 °C to DMF
at room temperature. It follows that the increased cleavage
reactivity observed in the former case as compared to the latter
is essentially caused by an increase of the driving force. This
is itself mainly due to a better solvation of the leaving fluoride
ion by the protic solvent NH3 as compared to the aprotic solvent
DMF. For the following reasons, the intrinsic barrier is however
not strongly affected by this variation of the anion solvating
properties. Both E0RX/RX and ER0 /R in eq 5 are expected to
shift in the negative direction when passing from liquid NH3 to
DMF, the latter factor more than the former (because the
negative charge is more concentrated in R- than in RX•-). The
ensuing increase of the intrinsic barrier is however compensated
by a decreased solvent reorganization energy.18
activation free energies of 2, 6, 7, and 9 in a diagram having
the reaction standard free energy as horizontal coordinate as
shown in Figure 10b. Extrapolation of the 0.5 slope straight
line to zero driving force thus provides the common value of
the intrinsic barrier free energy, for these four compounds,
∆Gq0 ) 0.96 eV.
The data point for 1′′′ is located 0.16 eV below the main
correlation line in Figure 10a. As discussed earlier, this means
that the bond dissociation energy of 1′′′ is 0.20 eV smaller than
that of the unsubstituted compound, 7, i.e., D1′′′ ) 4.14 eV. Thus
•
•
-
from the application of eq [4], ∆G0 ) -0.88 eV and, from
1′′′
the linearized version of eq [3], ∆Gq0 ) 0.91 eV. The
contribution of the intramolecular dissociative electron transfer
to the intrinsic barrier free energy may be estimated from eq
[5] in the following manner. The anion of 1′′, produced in situ
by deprotonation by t-BuO-, gives rise, at a concentration of 3
mM and a scan rate of 0.2 V/s, to an irreversible cyclic
voltammetric peak at 0.43 V vs 0.01 M Ag+/Ag. In view of
the values found for the heterogeneous rate constants found for
the reduction of 1′, 1′′, and 1′′′, and of the fact that the charge
is more concentrated in the anion of 1′′ than in the anion radicals
of the preceding compounds, 0.1 cm s-1 is a reasonable estimate
of the heterogeneous rate constant for the oxidation of the anion
of 1′′. Assuming that the radical thus produced undergoes a
rapid dimerization, the standard potential for this oxidation is
very close to the peak potential (0.47 instead of 0.43 V vs 0.01
M Ag+/Ag.).11b
Conclusions
1. The reductive cleavage of the trifluoromethylarenes leads
to completely defluorinated products. It goes through the
intermediate formation of the difluoro and monofluoromethyl
derivatives. The monofluoromethyl derivative is likewise an
intermediate in the reduction of aryldifluoromethyl compounds.
Unlike gem-dichloro, dibromo, or chloro-bromo compounds,
carbenes are not formed along the reaction pathway. The
reaction involves fluoride expulsion from the primary anion
radical, followed by reduction of the neutral radical thus formed.
The resulting carbanion rapidly protonates, thus producing the
hydrogenolysis product, rather than cleaving off a fluoride ion
to form a carbene. The difference with the chloro and bromo
analogues is presumably caused by the fact that carbon-fluorine
bonds are much stronger than with the other halogens
2. The standard potential for the formation of the anion
radical increases from the mono to the di- and the trifluoro
derivatives. The three standard potentials are however closely
spaced, thus opening the possibility of an internal redox catalysis
leading to an apparent direct six-electron reduction of the three
C-F bonds. The occurrence of such a mechanism depends
upon the stability of the primary anion radical toward cleavage.
It is followed, as demonstrated on quantitative grounds, when
cleavage of the primary anion radical is slow enough for the
resulting radical to be formed far from the electrode surface.
Then, the product in which one fluorine has been replaced by
one hydrogen is itself formed far from the electrode surface
and undergoes homogeneous electron transfer from the primary
It follows that ∆Gq0,intra and therefore that ∆Gq0,solv
. The
possibility of an important contribution of solvent reorganization18a
to the dynamics of intramolecular dissociative electron transfer
has recently been demonstrated in the cleavage of anion radicals
of R-substituted acetophenones.10c,d
Turning back to the di- and monofluoro compounds in the
same series (Figure 9), we may estimate the reaction standard
free energy in liquid NH3 from the bond dissociation energies
(Table 3) and the standard potentials (listed in Table 1). The
(17) (a) See Table 2 in ref 9a and the references quoted therein. (b)
E
- E
) (298 - 235)(S0F - SF0
- S0Ag
+ S0Ag) (the
0,235 K
F /F ,NH3
0,298 K
F /F ,NH3
•
-
•
-
•
-
+
,NH3
,NH3
S0’s being the entropies of formation). SF0• ) 1.65, SA0 g ) 0.44 meV K-1 16b
.
The entropies of formation of F- and Ag+ in liquid NH3 were derived from
their values in water (-0.14 and 0.76 meV K-1 respectively16b) by means
of: S0F
) S0F
+ ∆trSF
•
-
-
+
-
+
or Ag+,H2OfNH3
(the ∆trS are the
or Ag ,H2O
entropieosr Aogf ,tNrHan3 sfer). ∆trSAg
) -0.26 meV K-1 17c For F-, we
.
+,H2OfNH3
used the following procedure in the absence of directly available data. There
is a good linear correlation between the entropies of transfer of Cl-, Br-,
and I- from H2O to NH3 and from H2O to HCONH2: ∆trSX
)
-,H2OfNH3
-,H2OfHCONH2
1.52∆trSX
- 1.18 meV K-1 17d
.
Extrapolating for F- we
) -0.12 meV K-1 17d
∆trSF )
-,H2OfNH3
(18) Eq 6, being based on a Born model of solvation, is likely to be a
rather poor representation of the differences between a H-bonding solvent
and an aprotic solvent although the predictions deriving from the size of
the charge location volume and to the center-to center distant are most likely
correct.
-,H2OfHCONH2
obtained, using ∆trSF
,
-1.36 meV K-1. Since SF0-,H2O ) -0.14 meV K-1 17d S0F-,NH3 ) -1.50
,
meV. (c) Marcus, Y. Pure Appl. Chem. 1985, 51, 1103. (d) Marcus, Y.;
Kamlet, M. J.; Taft, R. W. J. Phys. Chem. 1988, 92, 3613.

Dynamics of Bond-Breaking in Ion Radicals
J. Am. Chem. Soc., Vol. 119, No. 40, 1997 9539
anion radical before having time to diffusing back and being
reduced at the electrode surface. For these reasons, this
“disproportionation” or internal redox catalysis mechanism
occurs with trifluoro derivatives bearing electron-withdrawing
groups on the aromatic moiety. Conversely, an “ECE” mech-
anism, where the defluorinated intermediate being formed close
to the electrode surface undergoes an electrode electron transfer,
occurs with trifluoro derivatives bearing an electron-donating
substituent or no substituent at all as well as with the difluoro
derivatives.
3. The variations of the cleavage reactivity with the number
of fluorines on the functional carbon and with the structure of
the aromatic moiety were derived from cyclic voltammetric and/
or redox catalysis experiments. They could be successfully
rationalized by the intramolecular dissociative electron transfer
model. Application of the model involves relating both the
driving force (the opposite of the standard free energy of the
reaction) and its intrinsic barrier to structural and medium factors
in the reactants and products.
4. The rate of anion radical cleavage increases rapidly from
the trifluoro to the difluoro and to the monofluoro derivative.
In terms of driving force, an important factor is the standard
potential for the generation of the anion radical (E0) which
measures the LUMO energy. The higher the later factor, the
larger the driving force, and thus, the faster the reaction. The
acceleration of the reaction indeed parallels a shift of the
standard potential in the negative direction. This is however
insufficient to account quantitatively for the magnitude of the
acceleration. One should take into account the variation of
another factor, namely the homolytic bond dissociation energy
of the starting molecule (D) which decreases as the number of
fluorines diminishes. This is a second reason why the driving
force increases from 3 to 2 and to 1 fluorines. This factor also
influences the intrinsic barrier, which decreases as the number
of fluorines diminishes. On total, the activation free energy
correlates, as predicted by the intramolecular dissociative
electron transfer model, with E0/2 + 3D/4. This correlation
implies that the sum of the other ingredients of the intrinsic
barrier remains constant in the series. This constancy derives
from the approximate constancy of the solvent reorganization
energy and from the fact that the variations of the standard
potentials for the formation of the anion radical and for the
oxidation of the defluorinated carbanion compensate each other
while shifting in the negative direction upon passing from 3 to
2 and to 1 fluorines.
5. The standard potential for the formation of the anion
radical of the trifluoro derivatives varies substantially upon
changing the aryl moiety. This is the main factor which makes
the cleavage rate constant vary in the series through the ensuing
variation of the driving force. The bond dissociation energy
of the starting molecule also interferes both in the driving force
and in the intrinsic barrier. This is the reason that the presence
of an electron-withdrawing substituent, when located in a
resonant interacting position with the CF3 group, accelerates
the cleavage reaction by means of the attending weakening of
the C-F bond.
mately the same in both solvents results from the compensation
of two effects. Passing from DMF to liquid NH3 shifts the
standard potentials for the formation of the anion radical and
for the oxidation of the defluorinated carbanion both in the
positive direction; the later more than the former because the
negative charge it bears is concentrated over a smaller volume.
The resulting decrease of the intrinsic barrier is however
compensated by an increase of the contribution of solvent
reorganization.
Experimental Section
Cyclic Voltammetry. In liquid NH3, the working electrode was a
0.5 mm-diameter gold disk, frequently polished with alumina.
A
platinum wire was used as counter electrode, and the reference electrode
was a 0.01 M Ag+/Ag electrode.19a The potentiostat, equipped with a
positive feedback compensation and current measurer,19b was used
together with a function generator (Tacussel TPPRT), a storage
oscilloscope (Nicolet), and an X-Y recorder (Sefram TGM164). The
cyclic voltammetry experiments were run in an electrochemical cell19c
filled with 80 mL of liquid ammonia, and potassium bromide was used
as supporting electrolyte. The temperature was maintained at -38 °C
with a cryostat (Bioblock Scientific).
In DMF, the working electrodes were 1 or 3 mm diameter carbon
disks. They were carefully polished and ultrasonically rinsed with
ethanol before each voltammogram. The counter electrode was a
platinum wire and the reference electrode an aqueous SCE electrode.
The potentiostat, function generator, and recorders were the same as
above. The experiments were carried out at 20 °C in an electrochemical
cell equipped with a double-wall jacket allowing circulation of water
filled with 5 mL of DMF and n-Bu4NBF4 as supporting electrolyte.
Chemicals. N,N′-Dimethylformamide (Fluka puriss absolute) and
the supporting electrolyte n-Bu4NBF4 (Fluka puriss) were used as
received.
All trifluoro compounds, with the exception of 5 and 6, were
purchased from Aldrich or Maybridge.
5 was prepared according to literature20a from the corresponding
benzoic acid chloride and t-BuOH. A white solid is obtained after
separation of the reacting mixture is separated on a silica column. with
a 70-30 cyclohexane-dichloro methane mixture as eluant. Yield:
1
56%; H NMR (CDCl3) δ 1.62 (s, 9 H), 7.71-8.14 (AA′BB′, JAB
)
9Hz, 4H); 13C NMR (CDCl3) δ 28.1 (CH3), 80.8 (C), 123.5 (CF3, q,
JCF ) 272 Hz), 125.3 (CH, q, JCF ) 4 Hz), 129.8 (CH), 133.2 (C),
134.3 (C, q, JCF ) 32.5 Hz), 165.7 (C).
6 was prepared in situ by addition of a stoichiometric quantity of
t-BuOK to the acid.
To obtain 1′′,20b a mixture of 4-cyanobenzaldehyde (5 mmol) and
diethylaminosulfur trifluoride (5.2 mmol) was heated carefully till the
start of the exothermic reaction. It was then maintained at 60 °C for
30 min. The resultant mixture was dissolved in dichloromethane (15
mL), and the solution was poured into ice (20 mL) to remove the
diethylamino-N-sulfinyl fluoride formed in the reaction. The organic
layer is dried with magnesium sulfate. The products are eluted with
dichloromethane on a silica column. A pale yellow liquid is obtained
1
after evaporation of the eluant. Yield: 79%; H NMR (CD3COCD3)
δ 6.70 (t, JHF ) 56 Hz, 1 H), 7.65-7.78 (AA′BB′, JAB ) 8.3 Hz, 4 H);
MS: m/z 153 (M), 152 (M - 1), 134, 103.
To prepare 1′,20c 5.5 mmol of 4-bromomethylbenzonitrile were added
to 10 mL of a THF solution of NBu4F (1 M). The mixture was stirred
under an inert atmosphere at ambient temperature during 20 h. The
resultant mixture was poured in pentane (15 mL) and washed 3 times
with water (10 mL). The organic layer was dried with magnesium
sulfate, and the products separated on a silica column with CH2Cl2 as
eluant). A pale yellow liquid was finally obtained (mp ≈ 25 °C).
6. The activation free energy of cleavage is smaller in liquid
NH3 at -38 °C than in DMF at room temperature. This
difference in reactivity is essentially of thermodynamic origin.
It derives from a better solvation by NH3 (H-bonding solvent)
than by DMF (non H-bonding solvent). In both solvent the
major contribution to the intrinsic barrier comes from the nuclear
reorganization attending the intramolecular dissociative electron
transfer. However solvent reorganization is far from negligible.
Its contribution to the intrinsic barrier is ca. one-third of the
total. The fact that the intrinsic barrier free energy is approxi-
(19) (a) Herlem, M. Bull. Soc. Chim. Fr. 1967, 1687. (b) Garreau, D.;
Save´ant, J.-M. J. Electroanal. Chem. 1972, 35, 309. (c) Combellas, C.; Lu,
Y.; Thie´bault, A. J. Appl. Electrochem. 1993, 23, 841.
(20) (a) Furniss, B. S.; Hannaford, A. J.; Smith, P. W. G.; Tatchell, A.
R. Vogel’s Textbook of Practical Organic Chemistry, 5th ed.; Longman,
Ed.; Essex UK, 1989; pp 1073-1081. (b) Markovskij, L. N.; Pashinnik,
V., E.; Kirsanov A. V. Synthesis 1973, 787. (c) Cox, D. P.; Terpinski, J.;
Lawrynowicz, W. J. Org. Chem. 1984, 49, 3216.

9540 J. Am. Chem. Soc., Vol. 119, No. 40, 1997
Andrieux et al.
1
Yield: 85%; H NMR (CD3COCD3) δ 5.63 (d, JHF ) 45 Hz, 2H),
Experiment 2.13a 1′′ (0.7 mmol) and t-BuOK (0.9 mmol) were
added, under controlled atmosphere, to 2,3-dimethyl-2-butene (8.5
mmol). The mixture rapidly turned brown, and it was stirred for 24 h.
The mixture was then dissolved in dichloromethane and poured into
water. The organic layer was dried over magnesium sulfate and then
analyzed by GC/MS. Starting products were recovered without
formation of cyclopropanic product.
7.63-7.84 (AA′BB′, JAB ) 9 Hz, 4H); MS: m/z 135 (M), 134 (M -
1), 110, 108, 107.
Attempted Trapping of Carbene Intermediates. Experiment 1.
Compound 1′′′ (3 mmol) and 2,3-dimethyl-2-butene (20 mmol) were
introduced into an electrochemical cell containing 20 mL of acetonitrile
and NBu4BF4 (0.1 M) as supporting electrolyte. The cathode (a 4 cm2
stainless steel foil) and the anode (1 cm diameter magnesium rod) were
then introduced in the reactor, and the electrolysis was performed under
controlled current density (0.7A dm-2). It was stopped after the
consumption of 3 Faradays per mol of substrate. The products were
extracted with dichloromethane, thoroughly washed with water, and
then analyzed by GC/MS. Analysis of products gave 39% 4-toluonitrile
and 61% unreacted 1′′′.
Supporting Information Available: Figures S1-S4 sum-
0
marize the ip/ip data and the procedure for extracting the rate
constants from them (2 pages). See any current masthead page
for ordering and Internet access instructions.
JA971094O