Communications
well as 3 were examined (Table 1, entry 2).[6] Interestingly, the
use of either resulted in full recovery of starting material,
indicating that CoIII was not suitable for the reaction under
the examined conditions. Examination of complexes derived
from other diamine backbones only led to poor results (see
Supporting Information for compilation). We also screened a
number of ligand variants differing in the substitution pattern
on the salicyl groups. Examination of the latter set of ligand
analogs revealed the importance of a bulky alkoxy group at
the C-3/C-3’ positions (compare entries 3–5 in Table 1), from
which the corresponding cobalt catalysts resulted in increased
conversion and afforded products with enhanced selectivity. It
is interesting to note that steric congestion alone is not
determinant for enantioselectivity, as the complex derived
from the ligand incorporating isopropoxy groups at C-3 and
C-3’ (Table 1, entry 6) furnished product with inferior ee and
conversion, making evident the existence of a steric “sweet
spot” at these positions. The screening also revealed that the
substituents at C-5/C-5’ positions were critical to the gener-
ation of catalysts affording products with improved enantio-
selectivity, with a preference for electron-withdrawing groups.
Particularly noteworthy were the results obtained with
complex 8, which includes the C-3/C-3’ di-iBuO and C-5/C-
5’/C-6/C-6’ tetrachlorinated ligand (Table 1, entry 7). These
observations are interesting when considered in light of the
reported electronic effects for metal–salen catalysts in a
variety of transformations.[6f,10] In the asymmetric epoxidation
of Jacobsen et al. a clear preference for electron-rich ligands
has been noted with respect to the enantioselectivity of the
process.[10a,b] Similarly, in the cyclopropanation reaction of
styrene with diazoacetates and CoIII–salen complexes Katsuki
et al. have reported greater selectivity with electron-rich
complexes.[6f] The electronic effects have been suggested to
correlate to the position of the transition state along the
reaction coordinate. However, a mechanistic study by Gro-
nert et al. on the reaction of CoIII–salen complexes with
diazoacetates indicates that reactions are faster with ligands
bearing substituents with positive s values.[10c] Consequently,
the combination of electron donating and withdrawing group
in the optimal ligand scaffold in 8 is unique.
Scheme 2. Preparation of 8 and the X-ray crystal structure of the
solvate, 9.[11] NCS=N-chlorosuccinimide.
buffer with excess NaNO2 (3.6 equiv) (Table 1, entries 11 and
12). It is important to note that in order to carry out the
reaction at À158C (Table 1, entries 9–12), it was necessary to
use 20% aqueous NaCl solution as solvent to prevent freezing
of the aqueous reaction media. Interestingly, although NMI is
commonly employed in related processes and has been
suggested to exert its influence on the catalyst as an axial
auxiliary ligand, Ph3As has not been previously examined in
this capacity.[12] The use of a co-solvent (DCM or toluene)
resulted in decreased conversion, consistent with our previous
suggestion that the transformation proceeds on water.[13]
Having identified the optimal catalyst and conditions, we
examined the scope of the reaction (Table 2). The method is
widely applicable to a range of styrenes, including electron-
rich and -poor as well as o-substituted substrates. Electron-
rich substrates proved to be the best, leading to excellent
enantioselectivity (up to 94% ee), diastereoselectivity (up to
180:1 d.r.) and yield (up to 95%). Electron-poor substrates
were slightly inferior but nonetheless afforded products in up
to 90% ee. The diastereoselectivity is in most cases excellent,
with impressive preference for the trans isomer (up to 180:1).
It was possible to convert the product of 4-bromostyrene to a
crystalline derivative by its subjection to Cu-catalyzed
amidation (Scheme 3), allowing determination of the abso-
lute configuration by X-ray analysis.[14]
The preparation of optimal catalyst 8 is conveniently
carried out in three steps from 2,3-dihydroxybenzaldehyde
(Scheme 2). These include alkylation with iBuBr in DMSO
(68%), chlorination with NCS in acetic acid (89%), and
condensation with (S,S)-1,2-cyclohexyldiamine in the pres-
ence of Co(OAc)2 (80%). As shown in Scheme 2, we have
obtained an X-ray structure of the catalyst as the methanol/
water solvate (see 9) by recrystallization from CH2Cl2/MeOH.
We next investigated the effect of other experimental
parameters including temperature, solvent, and additives
(Table 1). This included the conditions under which the
diazoalkane is generated in situ, such as addition of an
aqueous solution of NaNO2 employing a syringe pump
(Table 1, entries 1–9) versus addition of solid NaNO2 in one
portion (Table 1, entries 10–12). In contrast to what we had
reported for the Fe-catalyzed process, addition of NaNO2 in
one portion proved optimal for conversion. Additional
benefits were noted when the reaction was conducted at
À158C in the presence of 20 mol% Ph3As and H2SO4/NaOAc
Scheme 3. Determination of the absolute configuration by X-ray analy-
sis of an amide derivative.
The novel transformation we have described can be
extended to 1,1-disubstituted styrene derivatives with addi-
tional functional groups.[15] For example, the cyclopropana-
tion reaction proceeded smoothly for the acetophenone-
derived enolacetate to afford adduct in 65% yield, 9:1 d.r.,
and 87% ee (Scheme 4). We could also show that an allylic
acetate was smoothly converted into the trifluoromethyl-
substituted cyclopropane in 2:1 d.r. and excellent enantiose-
1102
ꢀ 2011 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2011, 50, 1101 –1104