Solvent effects in the regio- and diastereselective epoxidations of acyclic allylic alcohols by dimethyldioxirane: hydrogen bonding as evidence for a dipolar transition state

Article abstract of DOI:10.1021/jo951984r

A mechanistically significant solvent effect is observed in the regioselectivity of the geraniol epoxidation by dimethyldioxirane.In hydrogen bonding solvents (MeOH), the 6,7-epoxide is preferred over the 2,3-epoxide (74:26), which reveals that the more nucleophilic 6,7 double bond (the 2,3 double bond is inductively deactivated by the allylic hydroxy group) is preferentially attacked by the electrophilic dimethyldioxirane.In MeOH, both regioisomeric dipolar transition states are equally well stabilized by interaction through intermolecular hydrogen bonding with solvent molecules.In the nonpolar CCl4, intramolecular hydrogen bonding with the allylic hydroxy functionality favors attack at the 2,3-double bond and proportionally more 2,3-epoxide is formed.Similarly, also the ?-facial selectivity in the dimethyldioxirane epoxidation of methyl-substituted cchiral acyclic allylic alcohols is controlled by intermolecular versus intramolecular hydrogen bonding.Thus, higher threo selectivities are obtained in the nonpolar CCl4 by stabilization of the diastereomeric transition state with minimal allylic strain through intramolecular hydrogen bonding with the allylic hydroxy group.The geometry of the dipolar transition state for the dimethyldioxirane epoxidations is similar to that of m-CPBA, but with apparently a slightly larger (ca. 130 deg) dihedral angle α to relieve 1,2-allylic strain.

Full text of DOI:10.1021/jo951984r

3506
J . Org. Chem. 1996, 61, 3506-3510
Solven t Effects in th e Regio- a n d Dia ster eoselective Ep oxid a tion s
of Acyclic Allylic Alcoh ols by Dim eth yld ioxir a n e: Hyd r ogen
Bon d in g a s Evid en ce for a Dip ola r Tr a n sition Sta te
Waldemar Adam and Alexander K. Smerz*
Institute of Organic Chemistry, University of Wu¨rzburg, Am Hubland, D-97074 Wu¨rzburg, Germany
Received November 8, 1995X
A mechanistically significant solvent effect is observed in the regioselectivity of the geraniol
epoxidation by dimethyldioxirane. In hydrogen-bonding solvents (MeOH), the 6,7-epoxide is
preferred over the 2,3-epoxide (74:26), which reveals that the more nucleophilic 6,7 double bond
(the 2,3 double bond is inductively deactivated by the allylic hydroxy group) is preferentially attacked
by the electrophilic dimethyldioxirane. In MeOH, both regioisomeric dipolar transition states are
equally well stabilized by interaction through intermolecular hydrogen bonding with solvent
molecules. In the nonpolar CCl4, intramolecular hydrogen bonding with the allylic hydroxy
functionality favors attack at the 2,3 double bond and proportionally more 2,3-epoxide is formed.
Similarly, also the π-facial selectivity in the dimethyldioxirane epoxidation of methyl-substituted
chiral acyclic allylic alcohols is controlled by intermolecular versus intramolecular hydrogen bonding.
Thus, higher threo selectivities are obtained in the nonpolar CCl4 by stabilization of the
diastereomeric transition state with minimal allylic strain through intramolecular hydrogen bonding
with the allylic hydroxy group. The geometry of the dipolar transition state for the dimethyldiox-
irane epoxidations is similar to that of m-CPBA, but with apparently a slightly larger (ca. 130°)
dihedral angle R to relieve 1,2-allylic strain.
In tr od u ction
oxidizing agent proceeds through a spiro rather than a
planar butterfly transition state.5 The electronic nature
of the transition state still remains unclear, i.e., whether
the epoxidation is synchronous or whether some dipolar
character applies. Also the possibility of a bona fide
diradical process, which was proposed already in early
investigations on dioxiranes,6 was again invoked.7
For some time it has been recognized that the hydroxy
group directs the π-facial selectivity in the epoxidation
of allylic alcohols for a variety of oxidizing agents,1
provided the substrate possesses allylic strain.2 Such
stereochemical studies have revealed valuable mecha-
nistic information on the transition state geometry of
oxygen transfer processes. Particularly informative have
been, for example, m-CPBA and vanadium-catalyzed
epoxidations, in which the oxidant associates in different
ways with the hydroxy functionality, i.e., through hy-
drogen bonding with m-CPBA versus metal-alkoxide bond
formation for vanadium.3 As a consequence of the
domination of 1,3-allylic strain (A1,3) for m-CPBA versus
1,2-allylic strain (A1,2) for vanadium in the highly ordered
associate, the π-facial selectivity in the epoxide formation
of a chiral allylic alcohol possessing either strain is
opposite for the two oxidants, namely, threo for m-CPBA
versus erythro for vanadium. This change has been
ascribed to differing dihedral angles R in the associate
of the allylic substrate and oxygen transfer agent in the
transition state, in particular ca. 120° for m-CPBA and
ca. 50° for the vanadium/tBuOOH system.
Recently it was found that these epoxidations can be
assisted through polar solvents, as demonstrated by
kinetic experiments.8 Partial charge separation was
proposed in the dimethyldioxirane epoxidation, and
hence, the dipolar transition state shown was suggested.
Analogous to the m-CPBA or vanadium-catalyzed epoxi-
dations, a hydroxy-directing effect was expected for DMD
in the oxidation of chiral allylic alcohols which possess
allylic strain, but in acetone/CH2Cl2 (ca. 1:1) as solvent
In this context, only little is known to date on such
hydroxy-directed stereochemical probing in dimethyl-
dioxirane (DMD) chemistry; unquestionably, such infor-
mation should be desirable for stereoselective synthesis
with this now popular oxidant.4 It has been proposed
that the attack on the double bond by this electrophilic
(4) (a) Adam, W.; Curci, R.; Edwards, J . O. Acc. Chem. Res. 1989,
22, 205-211. (b) Murray, R. W. Chem. Rev. 1989, 89, 1187-1201. (c)
Curci, R. In Advances in Oxygenated Processes; Baumstark, A. L., Ed.;
J AI: Greenwich, CT, 1990; Vol. 2, Chapter I. (d) Adam, W.; Hadjiara-
poglou, L. P.; Curci, R.; Mello, R. In Organic Peroxides; Ando, W., Ed.;
Wiley, New York, 1992; Chapter 4, pp 195-219. (e) Adam, W.;
Hadjiarapoglou, L. Top. Curr. Chem. 1993, 164, 45-62.
(5) Baumstark, A. L.; McCloskey, C. J . Tetrahedron Lett. 1987, 28,
3311-3314.
X Abstract published in Advance ACS Abstracts, May 1, 1996.
(1) Hoveyda, A. H.; Evans, D. A.; Fu, G. C. Chem. Rev. 1993, 93,
1307-1370.
(2) Hoffmann, R. W. Chem. Rev. 1989, 89, 1841-1860.
(3) (a) Rossiter, B. E.; Verhoeven, T. R.; Sharpless, K. B. Tetrahedron
Lett. 1979, 20, 4733-4736. (b) Sharpless, K. B.; Verhoeven, T. R.
Aldrichimica Acta 1979, 12, 63-74. (c) Narula, S. A. Tetrahedron Lett.
1981, 22, 2017-2020. (d) Narula, S. A. Tetrahedron Lett. 1982, 23,
5579-5582. (e) Itoh, T.; J itsukawa, K.; Kaneda, K.; Teranishi, S. J .
Am. Chem. Soc. 1979, 159-169. (f) Mihelich, E. D. Tetrahedron Lett.
1979, 20, 4729-4732.
(6) (a) Adam, W.; Bottle, S. E.; Mello, R. J . Chem. Soc., Chem.
Commun. 1991, 771-773. (b) Adam, W.; Curci, R.; Gonza´les Nu´n˜es,
M. E.; Mello, R. J . Am. Chem. Soc. 1991, 113, 7654-7658.
(7) (a) Minisci, F.; Zhao, L.; Fontana, F.; Bravo, A. Tetrahedron Lett.
1995, 36, 1697-1700. (b) Minisci, F.; Zhao, L.; Fontana, F.; Bravo, A.
Tetrahedron Lett. 1995, 36, 1895-1898. (c) Bravo, A.; Fontana, F.;
Fronza, G.; Mele, A.; Minisci, F. J . Chem. Soc., Chem. Commun. 1995,
1573-1574. (d) Bravo, A.; Fontana, F.; Fronza, G.; Minisci, F.; Serri,
A. Tetrahedron Lett. 1995, 36, 6945-6948.
(8) Murray, R. W.; Gu, D. J . Chem. Soc., Perkin Trans. 2 1993,
2203-2207.
S0022-3263(95)01984-0 CCC: $12.00 © 1996 American Chemical Society

Dimethyldioxirane Epoxidation of Acyclic Allylic Alcohols
J . Org. Chem., Vol. 61, No. 10, 1996 3507
the diastereoselectivities were generally low.9 Neverthe-
less, appreciable effects could be observed for cyclic allylic
alcohols; the diastereoselectivities were significantly
enhanced in nonpolar solvents incapable of hydrogen
bonding.10,11
Ta ble 1. Solven t Effects on th e Regioselectivity in th e
Ep oxid a tion of Ger a n iol (1)
1
Analogous studies on directing effects in the O2 ene
reaction showed that cyclic substrates behave differently
from acyclic ones12 in their propensity to conduct the
π-facial attack on the double bond. In view of this
happenstance, a detailed study on the hydroxy-directing
effect in the DMD epoxidations appeared to us to be in
order to assess the regio- and diastereoselectivity for
acyclic allylic alcohols as a function of solvent polarity.
Geraniol was chosen to probe the regioselectivity, since
it contains an allylic alcohol functionality as well as an
unfunctionalized double bond, both of the same degree
of substitution, i.e., trialkyl substituted. For this pur-
pose, the hydrogen-bonding ability of the solvent13 was
varied to assess, through changes in the regioselectivity,
the relative rate of DMD epoxidation of the allylic
hydroxy-substituted versus the unfunctionalized double
bonds in geraniol (1). In addition, the chiral, acyclic
allylic alcohols 4a -g were selected to determine the
diastereoselectivity as a function of solvent polarity. The
degree and type of methyl substitution of the simplest
chiral allylic alcohol 4a was successively varied to
investigate the importance of 1,3- versus 1,2-allylic strain
in the epoxidation with DMD as reflected by the diaster-
eomeric ratio (dr). It was anticipated that the preferred
π-facial attack (threo versus erythro) as a function of
allylic strain and hydrogen-bonding ability of the medium
should provide information on the transition state ge-
ometry of the hydroxy-directed epoxidation. Solvents
which showed the largest effect in similar investigations
on cyclic substrates were employed as cosolvent with
acetone, namely, CCl4 and MeOH (both 9:1 relative to
acetone).10,11 Unfortunately, acetone is unavoidable in
view of the preparation of DMD.14
product
distributionc
convsnb,c
entry equiva
solvent
acetone
acetone/MeOH (1:9)
acetone
acetone/CCl4 (1:9)
acetone
[%]
6,7-2 2,3-2 3d
1
2
3
4
5
1
93
27
28
27
>95
73
88
74
51
17
12
26
49
10
0.3
0.3
0.3
5
95
a
b
Equivalents of DMD relative to geraniol. Mass balances
>90%. c Determined by 1H NMR analysis directly on the crude
d
product mixture (error ( 5% of the stated values). Diastereomeric
ratio 50:50.
6,7 double bond is preferred (88:12) over the 2,3 one,
whereas in pure acetone the ratio decreases to 74:26. In
acetone/CCl4 (1:9), the least polar medium, the attack is
completely unselective, since the two regioisomers were
obtained in a 51:49 ratio.
Solvent effects on the diastereoselectivity were tested
for the set of allylic alcohols 4 and are compiled in Table
2. Clearly, the diastereoselectivity in the epoxide forma-
tion depends on the substitution pattern of the allylic
alcohol 4. In pure acetone (medium B), for substrates
without allylic strain there is essentially no diastereo-
selectivity observed (entries 1 and 5), whereas substrates
with a methyl group at the R1 or R3 position (entries 3
and 8) show a distinct threo selectivity in the epoxidation
by DMD. This selectivity is higher in the case of 1,3-
allylic (4d ) than 1,2-allylic strain (4b). Further introduc-
tion of methyl groups as in derivatives 4e (entry 11), 4f
(entry 14), and 4g (entry 16) leads in all cases to an
enhanced diastereoselectivity, even though in some cases
no increased allylic strain is evident. For instance,
comparison of the pair 4d and 4e (entries 8 and 11) and
the pair 4f and 4g (entries 14 and 16), in which only R2
is changed from hydrogen to methyl (constant allylic
strain), reveals a significantly greater diastereoselectivity
for the higher substituted allylic alcohol in both cases.
Also several derivatives of allylic alcohol 4e, which was
chosen as a representative example, were oxidized with
DMD under conditions B in Table 2. In contrast to the
parent alcohol 4e, a low erythro selectivity was displayed
in the epoxidations of methyl ether 5e (38:62), trimeth-
ylsilyl ether 6e (46:54), and the acetate 7e (42:58), for
which high conversions (>95%) and mass balances
(>90%) were obtained.
Resu lts
As already reported,9 the oxidation of geraniol (1) with
1 equiv of DMD in pure acetone yields a mixture of the
monoepoxides 6,7-2 and 2,3-2 and the bisepoxide 3 (dr
50:50) in a ratio of 73:17:10, whereas at a large excess (5
equiv) of DMD the bisepoxide is formed exclusively. The
formation of bisepoxide must be avoided since it masks
the determination of the regioselectivity. Therefore, the
oxidations of geraniol were run to only ca. 30% conversion
by employing a deficiency (0.3 equiv) of DMD. Indeed,
as the results in Table 1 show, the bisepoxide was not
detected.
A significant solvent effect is observed on the product
distribution of the two regioisomeric epoxides 2 in this
series. In acetone/MeOH (1:9), the solvent mixture of
highest hydrogen-bonding capacity, the epoxidation of the
(9) Adam, W.; Prechtl, F.; Richter, M. J .; Smerz, A. K. Tetrahedron
Lett. 1993, 34, 8427-8430.
(10) (a) Adam, W.; Smerz, A. K. 12th IUPAC Conference on Physical
Organic Chemistry, Padova, 1994, Poster 151. (b) Adam, W.; Smerz,
A. K. Tetrahedron 1995, 51, 13039-13044.
(11) Murray, R. W.; Singh, M.; Wiliams, B. L.; Moncrieff, H. M.
Tetrahedron Lett. 1995, 36, 2437-2440.
(12) Adam, W.; Prein, M. Angew. Chem. 1996, 108, 519-538.
(13) Kamlet, M. J .; Abboud, J . L.; Taft, R. W. J . Am. Chem. Soc.
1977, 99, 6027-6038.
(14) (a) Murray, R. W.; J eyaraman, R. J . Org. Chem. 1985, 50,
2847-2853. (b) Adam, W.; Hadjiarapoglou, L.; Bialas, J . Chem. Ber.
1991, 124, 2377.
In all cases in which either 1,2- or 1,3-allylic strain is
present, also a notable solvent effect can be perceived.

3508 J . Org. Chem., Vol. 61, No. 10, 1996
Adam and Smerz
Ta ble 2. Solven t Effects on th e Dia ster eoselectivities in th e DMD Ep oxid a tion s of Allylic Alcoh ols
mbb
[%]
convsnc,d
[%]
drd
entry
substr
R1
R2
R3
mediuma
threo:erythro
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
4a
4b
4b
4b
4c
4c
4d
4d
4d
4e
4e
4e
4f
H
H
H
H
H
CH3
CH3
H
H
H
CH3
CH3
CH3
H
H
H
H
H
H
H
H
H
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
CH3
B
A
B
C
B
C
A
B
C
A
B
C
A
B
C
B
97
90
73
81
99
78
100
84
98
100
96
100
99
96
100
100
89
>95
>95
>95
78e
50:50
57:43
60:40
70:30
53:47
56:44
64:36
67:33
85:15
59:41
76:24
82:18
82:18
87:13
91:9
CH3
CH3
CH3
H
H
H
H
H
H
H
>95e
86
>95
>95
>95
87
>95
>95
>95
>95
>95
H
CH3
CH3
CH3
CH3
4f
4f
4g
CH3
> 95:5
a
b
d
(A) MeOH/acetone (9:1), (B) acetone, (C) CCl4/acetone (9:1). Mass balance. c Yields >95% in all cases. Determined by 1H NMR
analysis of characteristic signals (error ( 5% of the stated values). e Epoxide/enone ca. 90:10.
Thus, for substrates 4b (entries 2-4), 4d (entries 7-9),
4e (entries 10-12), and 4f (entries 13-15) the diaste-
reoselectivity is enhanced with decreasing solvent polar-
ity, i.e., in the order acetone/MeOH (1:9) > acetone >
acetone/CCl4 (1:9). For derivative 4g, already in pure
acetone the highest possible diastereoselectivity of >95:5
was obtained. In contrast, the same diastereoselectivities
within experimental error (entries 5 and 6) were observed
for the substrate 4c without any allylic strain, when the
solvent was changed from acetone to the less polar
mixture acetone/CCl4 (1:9). Furthermore, as expected,
in all these epoxidations no cis-trans isomerization
occurred.
molecular hydrogen bonding with methanol stabilizes the
dipolar transition state sufficiently so that the allylic
hydroxy functionality does not come into the act. In
contrast, in the least polar medium, i.e., 9:1 CCl4 and
acetone, external hydrogen bonding is of no consequence
and the dipolar transition state should seek assistance
through intramolecular hydrogen bonding by the allylic
hydroxy functionality. Therefore, since internal hydro-
gen bonding is more effective for dioxirane attack on the
2,3 double bond (six-membered ring is conformationally
preferred), its epoxidation should prevail. The observed
51:49 regioselectivity (Table 1) bears out the right trend,
but the effect is not sufficient to invert the ratio in favor
of the epoxide 2,3-2. However, it should be kept in mind
that there is still 10% acetone in this medium and its
polar nature moderates the effect apparently through
association with the allylic hydroxy group and addition-
ally through possible polar stabilization of the transition
state for the 6,7-epoxide. As a matter of fact, the 74:26
regioselectivity (Table 1) in pure acetone as solvent
confirms this counteracting trend. It is for this reason
that the observed regioselectivities for geraniol (1) are
not dramatic but mechanistically significant in that they
reflect a superposition of two electronic effects, namely,
the nucleophilic reactivity, which favors the 6,7 over the
2,3 double bond, counteracted by assistance through
intramolecular (allylic hydroxy group) versus intermo-
lecular (solvent) hydrogen bonding.
Discu ssion
The present results on the regioselectivity in the DMD
epoxidation of geraniol (1) in Table 1 and on the diaste-
reoselectivities of the chiral allylic alcohols 4 (Table 2)
as a function of solvent polarity provide convincing
experimental evidence for a hydroxy-directing effect.
Thus, the hydroxy group of the substrate associates with
the incoming dioxirane oxidant in the transition state to
reconcile the observed solvent polarity trends. We pro-
pose that also here the hydrogen-bonding ability of the
solvent is the principal molecular feature in controlling
the selectivity.
Let us first take up the regioselectivity results for
geraniol (Table 1). The convenience of this substrate is
the fact that the regioselectivity in the epoxidation of the
6,7 versus the 2,3 double bond provides a relative rate
measure in terms of the ratio of the two possible mono-
epoxides 6,7-2 and 2,3-2, an intramolecular competition
experiment that obviates recourse to cumbersome abso-
lute rate studies. Since the two double bonds are
trisubstituted, the inductive electron withdrawal by the
hydroxy group will lower the nucleophilicity of the 2,3
double bond, and if this electronic property is the decisive
reactivity factor, preferential epoxidation of the 6,7
double bond would be expected. Indeed, in the most polar
medium, i.e., 9:1 methanol and acetone, the regioselec-
tivity 88:12 (Table 1) bears out this expectation. Inter-
The diastereoselectivity results for DMD epoxidations
of the chiral allylic alcohols 4 (Table 2) corroborate the
conclusions reached from the regioselectivities of geraniol
(1). Appreciable π-facial selectivities can only be expected
if intramolecular hydrogen bonding by the allylic hydroxy
functionality operates and allylic strain helps in dif-
ferentiating energywise appropriate conformations in the
transition state.
Thus, it is not surprising that the two substrates 4a
and 4c (entries 1 and 5, Table 2), which do not possess
any allylic strain, do not display any diastereoselectivity
in the dioxirane epoxidation and also no solvent effects.
In the case of substrate 4b with 1,2-allylic strain only,

Dimethyldioxirane Epoxidation of Acyclic Allylic Alcohols
J . Org. Chem., Vol. 61, No. 10, 1996 3509
the diastereoselectivity is quite low, but increases regu-
larly from methanol to CCl4 as cosolvents and for acetone
it is intermediate. Again, in the polar methanol, in-
tramolecular association of the incoming dioxirane with
the allylic hydroxy functionality is offset by external
hydrogen bonding with the solvent and the inherent 1,2-
allylic strain essentially does not come to fruition in
steering the π-facial attack; consequently, a poor dia-
stereoselectivity is observed. In the nonpolar CCl4 as
cosolvent (entry 4), the trend to higher threo diastereo-
selectivity is evident, but the effect is small.
t
F igu r e 1. Preferred transition states in the epoxidations of
chiral allylic alcohol 4f with different oxidants.
The action of 1,3-allylic strain in the substrates 4d
(entries 7-9) and 4e (entries 10-12) is more pronounced
in controlling the expected π-facial selectivity. Thus, in
the nonpolar CCl4 as cosolvent, the medium of highest
stereocontrol through intramolecular hydrogen bonding
with the allylic hydroxy group, essentially the same but
relatively high (dr ca. 85:15) diastereoselectivities for both
derivatives 4d and 4e (entries 9 and 12) are manifested.
When both threo-directing 1,2- and 1,3-allylic strain
operate in concert, as in the substrate 4f (entries 13-
15), it is not surprising that in the nonpolar cosolvent
CCl4 the π-facial diastereoselectivity is enhanced to as
much as 91:9 and a pronounced solvent effect operates.
Nevertheless, the highest diastereoselectivity is displayed
by the tetrasubstituted derivative 4g even in pure
acetone. This was unexpected since an additional methyl
group at the R2 position exerts no further allylic strain
when compared to 4f. This higher diastereoselectivity
may result from stronger hydrogen bonding to the
adjacent allylic hydroxy functionality presumably on
account of larger charge separation in the transition state
for the peroxide bond cleavage. Moreover, despite their
higher nucleophilicity, tetrasubstituted alkenes can be
less reactive than trisubstituted ones in DMD epoxida-
tions due to counteracting steric effects.5 Consequently,
for such substrates a higher demand is placed on the
adjacent allylic hydroxy group to overcome this steric
barrier through hydrogen bonding and, hence, a higher
π-facial selectvity is observed. This effect was also noted
in the diastereoselective epoxidation of cyclic allylic
alcohols.10b
Clearly, the diastereoselective epoxidations of the
chiral allylic alcohols 4 in Table 2 establish evidence for
an interaction of the OH functionality with the dipolar
transition state in the dioxirane SN2 attack, which is
stabilized through intramolecular hydrogen bonding with
the allylic hydroxy group, unless a polar solvent inter-
feres. Further support for this proposed association is
also gained from the DMD epoxidation of methyl ether
5e, trimethylsilyl ether 6e, and acetate 7e of allylic
alcohol 4e. While the free hydroxy group in 4e directs
the DMD attack to an extent of 76:24 (threo to erythro)
even in acetone as solvent, all three derivatives 5e-7e,
which due to the lack of the proton cannot involve
H-bonding, yield the corresponding epoxides in a low
(maximum 38:62) erythro selectivity. Therefore, merely
steric factors are ineffective in governing the π-facial
selectivity in such DMD epoxidations.
F igu r e 2. Diastereomeric transition states in the epoxidation
of 4b.
stereocontrol is generally lower for the DMD epoxidation.
This trend may be explained by the fact that the
symmetrical dioxirane peroxide bond is much less polar-
ized in the transition state than the already inherently
polarized peroxide bond in the peracid and the adjacent
hydroxy group hydrogen bonds considerably less ef-
fectively for DMD. This is supported by the appreciable
solvent effect for DMD but not for m-CPBA on the
diastereoselectivity.
The surprising result that the epoxidation of allylic
alcohol 4b (1,2-allylic strain) proceeds threo selectively
with DMD in all three solvents employed (entries 2-4,
Table 2), whereas the peracid yields the erythro product
but in low selectivity (45:55),15 offers an additional reason
for the similar yet differentiated diastereoselectivities
between these two oxidants. The two transition states
for the diastereomeric epoxides are shown in Figure 2
for DMD. If a dihedral angle R of ca. 120° applies, there
is no dominant steric interaction that should direct the
π-facial attack. In the threo transition state, the two
methyl groups are not close enough together to exert
appreciable 1,2-allylic strain, while for the erythro attack
the 1,3-strain is not significant since the steric interaction
between H and CH3 is not effective. A dihedral angle R
smaller than 120° will tend to bring the two methyl
groups responsible for the 1,2-allylic strain in the threo
transition state closer together, this geometry will be
disfavored, and a low erythro selectivity is expected. In
contrast, if the dihedral angle R is larger than 120°, the
1,2-allylic strain (A1,2) in the threo case will be decreased
but the small 1,3-strain (A1,3) in the erythro transition
state will be increased, which seems to be responsible
for the fine tuning in the observed diastereoselectivities.
Therefore, while the exact dihedral angle is open to
debate, the trends in the moderate but mechanistically
relevant π-facial selectivities suggest that for DMD a
larger angle applies than for m-CPBA. With the as-
sumption that for m-CPBA a dihedral angle of 120°
operates,3 we had estimated for the cyclic allylic alcohols
The question that remains concerns the preferred
transition state geometry with respect to the dihedral
angle R for optimal internal hydrogen bonding. Com-
parison with the proposed transition state structures for
m-CPBA and vanadium-catalyzed epoxidations (Figure
1) provides a clue. The preferred threo diastereoselec-
tivities obtained for DMD resemble closely those for
m-CPBA rather than those for vanadium, except that the
(15) Adam, W.; Nestler, B. Tetrahedron Lett. 1993, 34, 611-614.

3510 J . Org. Chem., Vol. 61, No. 10, 1996
Adam and Smerz
an angle R of ca. 130°,10b which is a reasonable guess also
epoxidations of cyclic allylic alcohols10,11 and cyclic allylic
ammonium salts.17 Since no cis-trans isomerizations
were observed in these epoxidations, radical-type pro-
cesses7 are unlikely.
for the acyclic ones examined herein.
Con clu sion
Altough on kinetic grounds skepticism has been ex-
pressed16 as to whether the observed directing effects in
the dioxirane epoxidations of allylic alcohols derive from
assistance by the hydroxy functionality through intramo-
lecular hydrogen bonding with the peroxide bond, our
present regioselectivities (relative rates) and stereose-
lectivities (π-facial reactivities) provide convincing ex-
perimental evidence in favor. A dipolar transition state
applies for this oxygen transfer process, which is stabi-
lized by hydrogen-bonding solvents or by hydrogen bridg-
ing with an adjacent allylic hydroxy functionality. These
results substantiate previous reports that such coordi-
native effects operate in the diastereoselective DMD
Ack n ow led gm en t . Financial support by the
Deutsche Forschungsgemeinschaft (Schwerpunktpro-
gramm “Peroxidchemie: Mechanistische und pra¨para-
tive Aspekte des Sauerstofftransfers”) and the Fonds der
Chemischen Industrie is gratefully acknowledged.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures with the characterization of all products (8 pages). This
material is contained in libraries on microfiche, immediately
follows this article in the microfilm version of the journal, and
can be ordered from the ACS; see any current masthead page
for ordering information.
J O951984R
(16) Curci, R.; Dinoi, A.; Rubino, M. F. Pure Appl. Chem. 1995, 67,
811-822.
(17) Asensio, G.; Mello, R.; Boix-Bernardini, C.; Gonza´les Nu´n˜es,
M. E.; Castellano, G. J . Org. Chem. 1995, 60, 3692-3699.