with most of the Lewis bases 3-7,16 furnished the corre-
sponding chlorohydrin in 85% ee (see footnote 25 in ref 15a).
This prompted us to investigate PINDOX in the asymmetric
opening of a series of cyclic epoxides 9a-e (Scheme 3).
The opening of epoxides 9a-e with SiCl4, catalyzed by
PINDOX 8 (10 mol %), was carried out in CH2Cl2 at -90
°C in the presence of (i-Pr)Et2N (3 equiv) for 48 h. To ensure
the accuracy and consistency of the results, the reaction
mixtures were quenched with an aqueous saturated 1:1 KF/
K2HPO4 solution, known to be highly effective in preventing
the large quantities of acid released during the hydrolysis of
Si-Cl bonds from entering the organic solution and causing
a nonselective opening of the remaining unreacted epoxide.8
system 10c. Note that to attain the highest enantioselectivity
it is crucial to carry out the reaction at low temperature. Thus,
opening of 9c at -60 °C gave 10c in 75% ee (a ∼15% drop
in ee compared to -90 °C), whereas at -35 °C the product
was racemic, as shown by chiral GC analysis. It is also
pertinent to note that PINDOX (8) seems to be unique as a
catalyst for these transformations. Its close analogues, such
as the corresponding N,N′-dioxide15 or METHOX,18 were
inefficient, giving practically racemic products.19
The tricyclic exo-norbornene oxide 9f afforded the syn-
exo-chloroalcohol 11 as the major product (53% yield and
90% ee), rather than the vicinal chlorohydrin (Scheme 4).
The latter result is not surprising as the tendecy of 9f to
undergo Wagner-Meerwein rearrangements in the presence
of hydrochloric or hydrobromic acids is well docu-
mented.20,21
Scheme 3
The formation of 11 suggests that the epoxide opening
does not strictly follow a bimolecular SN2-type manifold but
involves a substantial degree of the C-O bonds ionization
in the initially formed complex E, generating the carbocation
intermediate F/G. According to this scenario, the enantiose-
lectivity of the process is determined by the selection of one
of the two C-O bonds of the epoxide for ionization. It seems
likely that the silicate counterion in the intermediate complex
B, incorporating the chiral Lewis base (Scheme 2), exerts
significant influence on the enantiodifferentiation.22
The syn relationship between the hydroxyl and chloride
groups in the syn-exo-chlorohydrin 11 is noteworthy. Intrinsic
preference for the nucleophile to form an exo-adduct, coupled
with the close proximity of the hydroxy group, suggests that
the delivery of the chloride ion may occur from the silicon
atom coordinated to the oxygen, rather than from the silicate
ion.
In conclusion, we have demonstrated that PINDOX 8 can
catalyze an enantioselective formation of chlorohydrins from
(18) (a) Malkov, A. V.; Bell, M.; Castelluzzo, F.; Koc˘ovsky´, P. Org.
Lett. 2005, 7, 3219. (b) Malkov, A. V.; Kabeshov, M. A.; Barłog, M.;
Koc˘ovsky´, P. Chem.sEur. J. 2009, 15, 1570.
The highest levels of selectivity (up to 90% ee) were
attained for derivatives with the ring size of eight carbons
or more (10c and 10e).17 Reduction of the ring size resulted
in a dramatic drop in selectivity: thus, cycloheptane oxide
9b afforded the corresponding chlorohydrin 10b in good
yield (87%) but with only 49% ee, whereas cyclohexane
oxide 9a furnished racemic product 10a. The introduction
of a double bond into the eight-membered ring system also
had a negative effect, with 10d showing a drop to 57% ee
compared to the 90% ee for the corresponding saturated
(19) The reaction rate observed for PINDOX (8) is equally unique, since
all other catalysts (Figure 1) react considerably slower. In principle, the
fast conversion of 9c into chlorohydrin 10c in the case of 8 could be
attributed to the cleavage of the unreacted oxirane ring during the workup
by the HCl released from SiCl4. However, this process could hardly be
expected to be enantioselective. Furthermore, the epoxide opening during
the workup would not account for the variation of enantioselectivity with
the reaction temperature. Note also that the reaction was quenched with a
saturated aqueous solution of KF and K2HPO4 (1:1) to buffer the mixture
and prevent the organic phase from becoming acidic (vide supra8).
(20) (a) Gerteisen, T. J.; Kleinfelter, D. C. J. Org. Chem. 1971, 36,
3255. (b) Loreto, M. A.; Pellacani, L.; Tardella, P. A. Synth. Commun.
1981, 11, 287. (c) Gargaro, G.; Loreto, M. A.; Pellacani, L.; Tardella, P. A.
J. Org. Chem. 1983, 48, 2043. For a general discussion on the behavior of
norbornyl cations, see the following: (d) Lowry, T. H.; Richardson, K. S.
Mechanism and Theory in Organic Chemistry; Harper and Row: New York,
1987; p 448. (e) Anslyn, E. V.; Dougherty, D. A. Modern Physical Organic
Chemistry; University Science Books: Sausalito, CA, 2006; p 662.
(21) Recently, Ready reported on the opening of epoxide 9f with SiCl4,
catalyzed by an axially chiral allene-derived phosphine oxide, which
occurred with 50% ee.13c However, he formulated the product as a vicinal
chlorohydrin, which is apparently incorrect, as revealed by his NMR spectra
that are compatible with those for 11 (as reported here and by Waegell:
Chauvet, F.; Heumann, A.; Waegell, B. J. Org. Chem. 1987, 52, 1916.) but
differ from those recorded for the vicinal chlorohydrin (also described by
Waegell)
(16) For a correction of erroneous reports, see: Denmark, S. E.; Wynn,
T.; Jellerichs, B. Angew. Chem., Int. Ed. 2001, 40, 2255.
(17) (a) The absolute configuration of chlorohydrins 10 is unknown;
the formulae shown here merely represent an arbitrarily chosen enantiomer.
(b) The enantioselectivities shown in Scheme 3 were established by 19F
NMR analysis of the corresponding Mosher’s esters, as most of the
chlorohydrins 10 (except 10c) proved difficult to separate by chiral
chromatography. (c) The Mosher esterication was quantitative, as revealed
by TLC analysis and by the 1H NMR spectra of the crude products, which
excludes a possible ee amplification during this derivatization. (d) The
enantiopurity of Mosher’s acid was independently verified to be >99% ee
(see the Supporting Information). (d) Chiral GC analysis of the trifluoro-
acetate derived from 10c revealed 87% ee, which is in good agreement
with the the value obtained by the Mosher method (90% ee).
(22) Hamilton, G. L.; Kanai, T.; Toste, F. D. J. Am. Chem. Soc. 2008,
130, 14984.
5392
Org. Lett., Vol. 11, No. 23, 2009