Reformatsky Reaction in Water
J . Org. Chem., Vol. 62, No. 26, 1997 9063
Sch em e 1
for both methyl and tert-butyl bromoacetate (entries 15
and 16). Ethyl 2-bromopropionate produced similar
results of about 50% in the presence as well as absence
of the catalyst (entries 17 and 19). Much better yields
were obtained with ethyl 2-bromoisobutyrate: 76% in the
absence and 80% in the presence of benzoyl peroxide
(entries 20 and 21). Nevertheless, with this ester too,
the addition reaction was strongly inhibited by galvinoxyl
and hydroquinone (entries 22 and 23). In all of the
previous examples (entries 16-23), solvent B was supe-
rior to solvent A; although less exothermic, the reaction
ran its course in a few minutes. For this reason, in the
following examples where the influence of the structure
of the carbonyl compound was examined, solvent A was
used only in the reactions with bromoacetate and solvent
B with bromoisobutyrate.
In the series of other unsubstituted aromatic and
heteroaromatic aldehydes, 2-naphthaldehyde gave a bet-
ter yield with ethyl bromoacetate than benzaldehyde,
furfural, and 2-thiophenecarbaldehyde (entries 24-26).
Comparing several substituted benzaldehydes (entries
27-30), increasing yields were observed, especially with
one or two electron-withdrawing groups. One of the most
attractive aspects of reactions in water is the possibility
of maintaining hydroxy groups unprotected. In fact, all
three isomeric hydroxybenzaldehydes gave moderate
yields with bromoacetate (entries 31-36); more complete
transformations were obtained with bromoisobutyrate,
but the low solubility of 3- and 4-hydroxybenzaldehyde
made necessary a large excess of bromo ester, thus
limiting its preparative interest. Unsaturated aldehydes
(entries 37-40) also produced Reformatsky adducts with
the tertiary bromo ester in good yields. In contrast to
the findings of Luche and co-workers who used haloal-
kanes in similar reactions,3 no cases of 1,4-addition were
observed. With saturated aldehydes and bromoacetate
no more adduct was detected, but bromoisobutyrate
produced still moderate yields (entries 41-44). Even in
the case of acetophenone, it was possible to detect 16%
of product 3 with this ester; substitution with an elec-
tronegative fluorine atom had little influence on the yield
(entries 46 and 47).
olefins.6 When the reaction was carried out in deuterium
oxide, however, no deuterium was incorporated in any
position of the hydroxy ester, but exclusively in the
reduction product ethyl acetate. In Scheme 1 we propose
an alternative radical chain mechanism which does not
involve hydrogen abstraction. It should predominate
when bromoacetate 4 is reacted in solvent A where zinc
is probably reactive enough to produce directly the
organometallic Reformatsky reagent 5. This will react
with water to form ethyl acetate or with a benzoyl radical
to form benzoate and radical 6 which adds to the
aldehyde 7, giving the oxyl radical 8. Reduction of the
intermediate 8 by another molecule of Reformatsky
reagent 5 produces the final adduct 9 and a new radical
6 to continue the chain. Alternatively, the initial radical
6 may be produced by bromine abstraction from 4, either
by a phenyl radical or on the zinc surface. This chain
process is probably more important in the case of bro-
moacetate in solvent A, because the primary carbanionic
intermediate 5 is more stable than a secondary or tertiary
species. In solvent B, where the reaction is slower and
less dependent on a peroxide catalyst, 6 may be formed,
at least in part, directly from 4 on the zinc surface. In
this case, there is no need for a chain process, since 8
can be reduced to 9 by the metal. When bromoisobu-
tyrate is used, this may be the main pathway, because a
tertiary radical of type 6 is more stable than a primary
one. Reactions with bromopropionate probably occupy
an intermediate position where both processes are com-
peting. In a third alternative, the aldehyde 7 may be
reduced to the radical 10 and then combined with 6 to
the final product 9. The real occurrence of radical species
is supported by the inhibitory effect of galvinoxyl and
hydroquinone (entries 10, 11, 22, and 23). Furthermore,
the intermediates 6 and 10 should be capable of dimer-
ization; indeed, careful GC/MS analyses of the crude
extracts revealed trace amounts of diethyl succinate and
1,2-diphenyl-1,2-ethanediol. Most of the intermediates
discussed here are thought to react adsorbed or close to
the metal surface. Specifically for the strongly exother-
mic and very fast reactions of bromoacetate in solvent
A, a large active metal surface is crucial as demonstrated
by markedly decreased yields when granulated zinc is
used (entry 3); in solvent B, the particle size of the metal
Discu ssion
The examples given in Table 1 demonstrate that the
Reformatsky reaction can be carried out in water with a
wide range of carbonyl substrates, including saturated
and unsaturated aldehydes and ketones where the in-
dium-promoted reaction in water was previously reported
to be uneffective.2d Preparatively interesting yields,
comparable to those of the classical procedure in anhy-
drous solvents, can be obtained from substituted benzal-
dehydes with ethyl bromoacetate and from aromatic and
unsaturated aldehydes with ethyl 2-bromoisobutyrate.
This is demonstrated by the isolated yields of selected
examples in preparative scale (entries 2, 19, 21, and 42).
From the mechanistic point of view, our results raise
new questions about pathways and intermediates of
organometallic in situ reactions in general and specifi-
cally of the Reformatsky reaction. Although the inhibi-
tion by galvinoxyl and hydroquinone fits very well with
the radical mechanism of two SET proposed by Chan,2d
the observed catalytic effect of benzoyl peroxide is
completely unexpected. As a first hypothesis, we con-
sidered a hydrogen abstraction from the bromo ester as
postulated in the benzoyl peroxide catalyzed addition to
(6) Wang, C. H.; McNair, P.; Levins, P. J . Org. Chem. 1965, 30, 3817.