refluxing dichloromethane (Table 2, entries 1ꢀ7). Each of
these catalysts exhibited a high degree of diastereocontrol,
but Yb(OTf)3 provided the highest level of diastereoselec-
tivity. Alternative catalysts such as La(OTf)3,9c,11
Zn(OTf)2,9a,3,4 Ni(OTf)2,12 and Mg(OTf)213 showed little
or no activity for this transformation under the reaction
conditions.
had simply discovered conditions for both the kinetically
and thermodynamically controlled processes; however
extensive investigations have revealed that the aldol ad-
ducts themselves are resistant to equilibration between
their syn- and anti-forms under any of the optimized
reaction conditions. When either purified 13a or 14a was
resubjected to the standard reaction conditions (Table 2,
entries 4 and 11), no interconversion of these products was
observed. Interconversion was only observed after 7 days
in refluxing CDCl3; however, interconversion under these
more forcing conditions was accompanied by a significant
amount of byproduct formation (>30% relative to 13a
Table 2. Optimization of the Intramolecular Aldol Reaction of
2,3,7-Triketoester 8a
1
and 14b based on H NMR). The absence of product
interconversion under the standard reaction conditions
suggests that this unique stereochemical divergence be-
tween Lewis and Brønsted acid catalysis is the result of a
kinetically controlled process. In this scenario Lewis acids
alter the course of the reaction to give the less common
anti-aldol product. We believe that the stereochemical
outcomes of the Brønsted and Lewis acid catalyzed aldol
reactions can be rationalized by analyzing the conforma-
tional intermediates en route to the respective products
(Scheme 4). Activation of the central carbonyl by a
Brønsted acid presumably results in the dipole minimized
orientation of the 1,2-diketo unit, which can then be
attacked by the tethered Z-enol via the lower energy
intermediate 16 thus providing 14a. Lewis acid activation,
however, has the ability to coordinate the 1,2-diketo unit
thus locking the carbonyls in a syn-orientation. The Z-enol
then reacts with the central carbonyl via intermediate 17 to
give 13a.
entrya
catalystb
anti/sync
% conversionc
d
1
Yb(OTf)3
N.D.
95:5
<5
e
2
Yb(OTf)3
65
f
3
Yb(OTf)3
95:5
>95
>95
93
4
Yb(OTf)3
Sn(OTf)2
Sc(OTf)3
Cu(OTf)2
95:5
5
85:15
87:13
89:11
25:75
32:68
30:70
30:70
93:7
6
>95
>95
75
7
g
8
H2PO4
9
p-TsOH
84
10
11
12i
Mes-SO3H
Mes-SO3Hh
Yb(OTf)3
90
>95
>95
a Reaction performed substrate 8a. b Reaction performed with
10 mol % catalyst. c Determined by 1H NMR spectroscopy (see Support-
ing Information). d 10 mol % catalyst at room temp for 24 h. e 5 mol %
catalyst. f 20 mol % catalyst. g 4 equiv of acid used. h 30 mol % of the acid
used. i The stereochemistry of the process was confirmed by a reaction
with 8c using 10 mol % Yb(OTf)3, and product 13c was analyzed by
X-ray (see Supporting Information).
Scheme 4. Stereochemical Rationale for the Diastereoselectivity
of Lewis and Brønsted Acids
We were surprised to find a significant divergence in
stereocontrol between Lewis acid and Brønsted acid cata-
lysts. The use of a Lewis acid gave the 1,2-anti product 13a
predominantly, whereas Brønsted acids greatly favored
formation of the 1,2-syn product 14a. Furthermore, given
the prevalence of syn-selective aldol reactions of 1,n-dicar-
bonyl compounds under both Brønsted and Lewis acid
catalyzed conditions,7 the fact that the anti-aldol product
13 is formed with high selectivity is notable. Our initial
thought regarding the observation that Lewis and
Brønsted acids provided divergent reactivity was that we
Expanding on these initial results, we applied the opti-
mized conditions for the Lewis and Brønsted acid cata-
lyzed processes to representative substrates (Table 3). Both
electron-withdrawing and -donating aromatic rings were
tolerated, as were sterically demanding aromatic substitu-
ents. In the case of the o,o-disubstitued substrates 13e and
13f (Table 3, entries 5 and 6), however, a significant
deterioration in anti/syn selectivity was observed under
Lewis acid catalyzed conditions. In fact, in the mesityl
example (Table 3, entry 6) selectivity was inverted, and the
syn-diastereomer was formed preferentially. This anomaly
seems to highlight the fact that the anti-transition state is
inherently more sterically demanding; thus, when the steric
(10) (a) Evans, D.; Downey, W. C.; Hubbs, J. J. Am. Chem. Soc.
2003, 125, 8706–8707. (b) Fossey, J.; Matsubara, R.; Kiyohara, H.;
Kobayashi, S. Inorg. Chem. 2008, 47, 781–783.
(11) (a) Corey, E. J.; Ishihara, K. Tetrahedron Lett. 1992, 33, 6807–
6810. (b) Ji, J.; Barnes, D.; Zhang, J.; King, S.; Wittenberger, S.; Morton,
H. J. Am. Chem. Soc. 1999, 121, 10215–10216.
(12) (a) Jung, M. E.; Johnson, T. W. J. Am. Chem. Soc. 1997, 119,
12412–12413. (b) Bouillon, J. P.; Portella, C. Eur. J. Org. Chem. 1999,
1571–1580. (c) Bouillon, J. P.; Portella, C.; Bouquant, J.; Humbel, S.
J. Org. Chem. 2000, 65, 5823–5830. Baik, T. G.; Luis, A. L.; Wang, J. C.;
Krische, M. J. J. Am. Chem. Soc. 2001, 123, 5112–5113.
(13) The stereochemistry of 13i was determined by X-ray crystal-
lography (Supporting Information).
Org. Lett., Vol. XX, No. XX, XXXX
C