C.A. de Souza et al. / Electrochimica Acta 132 (2014) 118–126
125
Table 4
support a carbanionic intermediate D. However, the effect is
less pronounced than in other nucleophilic additions and a
similar stabilization of radicals E, G and H cannot be excluded.
9. The reactivity of the carbonyl substrates used in this work fol-
lows the order generally observed in nucleophilic additions,
decreasing from aromatic aldehydes 12a-i to cinnamaldehy-
des 15a-c, 3-methyl-2-butenal (18), 3-phenylpropanal (20) and
acetophenones 23a-c, and thus supports an intermediate D.
However, as in the foregoing point, a similar order can be
attributed to decreasing stability of radical intermediates and
no preference for one of the three mechanisms is given here.
10. Pinacol formation (I) is not an important side reaction except
at potentials close to the reduction peak of benzaldehyde. This
makes a pathway via radicals G and H improbable for the cou-
pling reaction to F and supports both mechanisms starting with
A.
11. On pure graphite, reductive dimerization of A to J is observed
only with the tertiary bromide 1 in a few reactions with less
reactive carbonyl compounds such as 12d, 15a and 20. This is
probably due to the high stability of a tertiary radical C, whose
homocoupling cannot compete with the addition to more reac-
tive aldehydes. This observation is another strong argument in
favor of a radical mechanism via C. On silver doped graphite,
higher amounts of dimeric ester J were also found in several
experiments and probably cannot be explained by a carbanionic
intermediate D as postulated in point 4. The reason is rather a
dimerization of organic halides [41]. A mechanism via radicals
G and H can also be excluded because in no case both dimers I
and J were detected in the same experiment.
Compatibility of experimental observations and possible reactive intermediates.
Entry Intermediates
C + E
D
H + C
1
Reactivity order tertiary > secondary > primary
+
−
+
+
2
Reduction potentials of halide and carbonyl substrate
Applied potential
+
−
3
+
−
−
−
+
−
4
Silver catalysis in 1 and 5a
+
−
5
Electrolyte composition
+
−
6
Excess of halide
+
+
7
Leaving halogen in 5a,b,c
+
−
+
−
8
Substituents in the aromatic ring
Reactivity of carbonyl substrates
Absence of pinacol formation
Reductive dimerization of haloester
+/−
+/−
+
+/−
+/−
−
9
+
10
11
+
+
−
−
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
Leaving halogen in 9a,b
−
−
+
+
−
−
13
Silver catalysis in 9a
best results with halides 1 and 5a were all obtained at −0.8 or
−1.0 V and more negative potentials resulted in lower yields of
F and increasing formation of pinacol I. This gives additional
support to the pathway via C and E.
4. The presence of catalytic silver on the cathode turns the peak
tive one-electron transfer more difficult and favoring the direct
formation of D. This should enhance the coupling by a carban-
ionic mechanism as observed recently in the case of allylic and
benzylic halides [27,29]. However, in the case of bromoesters
1 and 5a, the effect of silver catalysis was clearly adverse in all
experiments, indicating that no carbanion D is involved in the
absence of silver, but rather a radical C. Intermediates G and H
are also improbable because no effect of silver was observed in
the voltammograms.
5. The rather weak effect of changes in the electrolyte composition
points also in the direction of radical C. Especially the addition
of acetic acid should affect severely a process based on highly
basic carbanions such as D. On the other hand, formation of
radical H and pinacol I even should be favored by acid, but this
was not observed.
Summarizing these arguments (Table 4), we can conclude that
all experimental observations are best conciliated with a pre-
dominant mechanism via radicals C and E, because none of the
eleven points is found contrary to it and only two can be con-
sidered ambiguous. A carbanionic intermediate D is compatible
only with five points, but incompatible with six. A pathway start-
ing from B via radicals G and H is even less supported with only
two points compatible, two ambiguous and seven incompatible.
It must be emphasized that this analysis refers essentially to the
results with the tertiary bromoester 1 and its less reactive sec-
ondary homologue 5a. The very low yielding primary halides 9a,b
provided some contradictory results which have to be discussed
separately.
12. In sharp contrast to 5a,b, the chloroacetate 9a gives better
yields than the bromo compound 9b. This cannot be accounted
data show that the C–Br bond is cleaved at a less negative poten-
tial and with a much higher peak current. On the other hand,
chlorides are reported to undergo directly a two − electron
reduction to carbanion D, whereas bromides can stop at the rad-
ical stage C [40]. Therefore, the higher reactivity of the chloride
9a can be explained by preferential formation of a carbanion D
and the low stability of a primary radical C. This interpretation
is reinforced by the better stabilization and lower basicity of
primary enolates in comparison to secondary or tertiary ones.
13. In the same direction points the now beneficial effect of sil-
ver catalysis with the chloride 9a. As already stated before, in
the presence of silver the reduction potential is shifted to less
cathodic values and the peak current increases significantly,
thus favoring the carbanion D.
6. Higher excess of halide as in entries 20 − 22 caused some
improvement in the yield of F, but this can be expected for
all three mechanisms because of the participation of interme-
diates derived from A.
7. The superiority of bromo- and iodoesters 5a and 5c over their
cathodic cleavage of the C–Cl bond is more difficult to occur,
but also to stop at the radical C. C–Br and C–I bonds have a bet-
ter chance for a monoelectronic process producing the radical C
[40]. A mechanism starting with the radicals G and H should not
be affected by the leaving halogen of A and can also be rejected.
8. Our coupling experiments with 1 and substituted benzalde-
hydes 12a-f show exactly the trend expected for nucleophilic
additions: increase of reactivity by electronegative substituents
and decrease by electron donors. Also the activating effect
of F − substituents observed in acetophenones 23a-c can be
attributed to increased electrophilicity. Both trends seem to
From the last two observations we must conclude that haloac-
etates 9a,b prefer an anionic mechanism in the coupling reaction
with benzaldehyde. Although this contrasts with the radical