Other Lewis acids were examined in an effort to favor
formation of the kinetic product 7a. Aluminum chloride was
found to be an effective catalyst, and favored formation of
7a (entry 6). Scandium triflate and tin tetrachloride were both
effective catalysts for the reaction (entries 7 and 8), but
diastereoselectivities were poor. Ferric chloride afforded only
trace amounts of products (entry 9), while zinc bromide,
ytterbium triflate, and copper(II) triflate were completely
ineffective.
Table 2. Cyclization Reactions of 6b-e
Closely related to the carbonyl ene reaction is the Prins
reaction, the addition of an aldehyde to an alkene catalyzed
by a Brønsted acid. Reports of Prins cyclizations to form
six-membered rings are less common than their carbonyl ene
counterparts, but we were intrigued by a report from Holker
and co-workers of a highly diastereoselective example
catalyzed by HCl.9 Thus, we undertook an investigation into
the diastereocontrol exerted by a small number of Brønsted
acids on the cyclization of 6a.
entry
aldehyde
acida
time (h)
7:8
yield %b
1
2
3
4
5
6
7
8
6b
6b
6c
6c
6d
6d
6e
6e
CH3AlCl2
HCl
CH3AlCl2
HCl
CH3AlCl2
HCl
CH3AlCl2
HCl
35
16
27
16
27
16
27
16
22:78
>98:2
30:70
90:10
7:93
89:11
25:75
80:20
55 (15)
86
53 (22)
71 (7)
74 (4)
72 (9)
61 (20)
60 (14)
p-Toluenesulfonic acid did not catalyze the reaction at low
temperature, while the stronger trifluoromethane sulfonic acid
was effective but the diastereoselectivity was modest (entry
10). However, to our surprise we found that addition of 3.0
equiv of concentrated hydrochloric acid to the dichloro-
methane solution of 6a at -78 °C effected quantitative
cyclization to 7a and 8a with a diastereomeric ratio of better
than 95:5 in favor of the kinetic isomer 7a (entry 11).
Frequently, a trace (<5%) of a side product, identified as
the cis chloride 9, was also produced under these conditions.
Chloride 9 was difficult to separate from 7a chromatographi-
cally, but simply stirring a solution of a mixture of 7a and
9 with silica gel10 or aqueous ammonia induced an elimina-
tion to afford pure 7a.
Reducing the amount of acid led to an unacceptably slow
reaction without improvement in diastereomeric ratio (entry
12). Although raising the temperature of the reaction lowered
the diastereoselectivity, the cyclization still favored cis
product 7a. Thus, heating 6a with 3.0 equiv of concentrated
hydrochloric acid at reflux in chloroform (entry 13) afforded
a 86:14 ratio of 7a:8a. Diastereoselectivity using a solution
of dry HCl gas in dichloromethane was essentially identical
(entry 14), although the procedure was less convenient, and
the presence of excess HCl led to an increase in the amount
of chloride side product 9.
a CH3AlCl2 refers to 1 equiv at 61 °C in chloroform. HCl refers to 3
equiv of concentrated hydrochloric acid at -78 °C in dichloromethane.
b Isolated yields of major (minor) isomers following chromatography.
detectable by NMR either before or after hydrogenation
(entry 2). Paralleling our findings for 6a, around 5% of the
chloride resulting from addition of HCl to the double bond
of 7b was also obtained.
Further extending our study, we examined the cyclization
of aldehydes 6c-e, in which the alkene is exocyclic to a
five-, six-, and seven-membered ring, respectively. Under
acidic conditions, the preference for formation of cis dia-
stereoisomers 7c-e was again marked (entries 4, 6, and 8),
although not as high as the acyclic examples. Under
equilibrating Lewis acidic conditions, all three favored the
trans diastereoisomer (entries 3, 5, and 7), a preference that
was particularly marked in the case of the cyclohexyl system
(entry 5).
As a reference point, we performed the cyclization of
citronellal under our optimal hydrochloric acid conditions.
The reaction was clearly a little more sluggish, since 25%
of the citronellal remained unchanged after the usual period
of 16 h at -78 °C. The principal products were isopulegol
and neoisopulegol, although there was evidence from the
NMR and GC-mass spectra that a number of side products
were also present resulting from dimerization (and possibly
higher oligomerization) of citronellal, presumably by aldol
chemistry. Interestingly, the cyclization process favored cis
product neoisopulegol 3, but in marked contrast with our
own system, the ratio of cis:trans products was only 75:25.
Both the acid- and the Lewis acid-catalyzed cyclizations
of 6 appear to proceed initially through the formation of the
kinetic product 7, with conversion to the thermodynamic
product 8 only proceeding at a significant rate under Lewis
acidic conditions. In contrast, the cyclization of citronellal
affords the thermodynamically more stable trans isomer
isopulegol with a variety of Lewis acids. There is no evidence
that cyclization proceeds through a kinetic intermediate that
isomerizes. Indeed Nakatani demonstrated that under the
Cyclization of the remaining aldehydes 6b-e was studied
under the optimized Lewis and Brønsted acid conditions;
the results are shown in Table 2.
Analysis of products 7b and 8b from cyclization of 6b
was complicated by the presence of E and Z double-bond
isomers, so the diastereomeric ratio was verified after
hydrogenation to the saturated products. Lewis acid-catalyzed
cyclization of 6b afforded predominantly trans product 8b
as a mixture of double-bond stereoisomers (entry 1). Under
Brønsted acid conditions, 6b exhibited a remarkable dia-
stereoselectivity of at least 98:2, with no trans product
(9) Chexal, K. K.; Holker, J. S. E.; Simpson, T. J.; Young, K. J. Chem.
Soc., Perkin Trans. 1 1975, 543-548.
(10) Iwamatsu, S. I.; Kondo, H.; Matsubara, K.; Nagashima, H.
Tetrahedron 1999, 55, 1687-1706.
Org. Lett., Vol. 4, No. 21, 2002
3729