major product (entry 1). In the reaction mixture, however, a
trace amount of â-dimethylamino ketone 7a was also
detected. This preference for the formation of the aldol
product was unchanged either by using a higher reaction
temperature (50 °C, 5 h) (entry 2) or by use of DMF as a
solvent at low temperature (-70 to 20 °C, 18 h) (entry 3).
To our surprise, however, a dramatic change of the reaction
course was observed when the reaction was carried out in
DMF at 50 °C (1.5 h). The amino ketone 7a was obtained
as the major product with a minor formation of 6 (6:7a )
1:10) (entry 4). We also observed that the corresponding
diethylamino derivative 5b favored the amino ketone forma-
tion more than 5a. Even when using Hoffmann’s conditions
(i.e., THF and low temperature), the ratio of 7b to 6 was
found to be greater than 1:2 (entry 5). Applying the optimized
conditions (DMF, 50 °C) increased the yield of 7b to 72%,
without any detectable formation of 6 (entry 6). We
confirmed that 7 and 6 are derived by independent reaction
pathways. Thus, before aqueous treatment of the reaction
mixture obtained under the reaction conditions for entry 5
(Table 2), the solvent (THF) was replaced with DMF, and
the resultant DMF solution was heated at 50 °C for 1.5 h.
Aqueous workup resulted in the major formation of the aldol
product 6 in a ca. 2:1 ratio (6:7).
We propose the following explanation for the observed
Mannich-type reaction. A direct aldol-type reaction via a
cyclic transition state may be suppressed by the unfavorable
coordination of the carbonyl group to the di(amino)boron
center as a result of its low Lewis acidity. Instead, an iminium
intermediate is generated via nucleophilic attack of the boron-
bound nitrogen to the aldehyde followed by deoxygenation,
with corresponding formation of a B-O bond (Scheme 2).
A boron enolate then attacks the iminium intermediate, for
which a cyclic transition state is unlikely. The existence of
the iminium ion pair may be supported by the crossover
experiment shown in Scheme 3. Thus, a mixture of the boron
enolates 2b and 2e (1:1) was reacted with excess benzalde-
hyde in DMF at 50 °C for 2 h. It had been confirmed that
no exchange of the amino groups between the two boron
enolates took place in the absence of the aldehyde under
otherwise identical reaction conditions. The reaction of
benzaldehyde with the mixture of the boron enolates yielded
two â-amino ketones. One of these â-amino ketones was
4ea, in which both functionalities were derived solely from
the boron enolate 2e. Unexpectedly, the other product was
Scheme 2
found to be a crossover product 4aa, whose amino and
carbonyl functionalities were derived from 2e and 2b,
respectively. No amino ketones bearing the diallylamino
group were detectable. The selective transfer of a diethyl-
amino group over a diallylamino group was also observed
in the reaction of benzaldehyde with the (diallylamino)-
(diethylamino)boron enolate of acetophenone. The boron
enolate with mixed amino groups yielded diethylamino
ketone 4aa exclusively in good yield.The formation of 4aa
suggests that the amino and enoxy groups are derived from
two independent boron components, as shown in Scheme 2.
Scheme 3
In summary, we have reported a new Mannich-type
reaction using bis(amino)boron enolates, which yield syn-
thetically useful â-amino ketones. In addition to their
synthetic importance, this demonstrates the generality of the
aminative C-C bond formations using diaminoboryl-
substituted nucleophiles. We are currently investigating the
mechanistic details to expand the scope of the reaction.
(10) The boron enolates 2a, 2b, 2d, and 2e were prepared from the
corresponding lithium enolates with chlorobis(dialkylamino)boranes and
used after distillation. The enolate 2c was prepared via amino exchange by
reaction of 2a with pyrrolidine (3-4 equiv). The boron enolates generated
by the amino exchange reaction were used after evaporation of volatile
materials in vacuo.
Acknowledgment. A.Y. thanks the Japan Society of
Promotion of Science for the fellowship support.
(11) All of the â-amino ketones reported in this work decompose
gradually on storage at room temperature, even under an inert atmosphere.
Moreover, owing to chromatographic instability, the amino ketones could
Supporting Information Available: Experimental pro-
cedures and spectral data for new compounds This material
1
not be isolated in the pure state. The yields were determined by H NMR
with an internal standard for crude â-amino ketones, which were obtained
by acid/base extraction (see Supporting Information). In all cases, a purity
>90% was achieved.
OL0497436
Org. Lett., Vol. 6, No. 7, 2004
1169