3188 J . Org. Chem., Vol. 62, No. 10, 1997
Adam et al.
times faster than the allylic alcohol 5h with the same
substitution pattern at the double bond. This clearly
demonstrates the assistance of the homoallylic hydroxy
group in the ene diol substrate 5a to replace multidentate
ligands through coordination to the active titanium site.
Thereby an efficient catalytic cycle between the com-
plexes A and B is sustained (Scheme 5). However, for
the allylic alcohol 5h , the Ti complex F (Scheme 7) with
two tridentate â-hydroperoxy alcohol molecules accumu-
lates. As discussed under item c, oxygen transfer can
proceed, but not as effectively because the corresponding
loaded complex G is generated slower.
(e) A higher temperature is required for an effective
ligand exchange. Due to the stronger binding of the
multidentate epoxy diol product and the diol derived from
the â-hydroperoxy alcohol oxygen donor in the titanium
complex B (Scheme 5), a higher reaction temperature is
required for the effective regeneration of the loaded
complex A through faster ligand exchange. Thus, the
oxygen transfer is rather slow (ca. 17 h) at -25 °C (Table
1, entry 11), but may be substantially accelerated (ca.
0.5 h) by increasing the reaction temperature to 23 °C
(Table 1, entry 9). The practical advantage is that these
epoxidations may be performed conveniently at ambient
temperatures.
exercise steric interactions on the transition state geom-
etry of the oxygen transfer.
(h) The diastereoselectivity observed in the epoxidation
of the simple allylic alcohols 5g-n is controlled by allylic
strain. In substrates without a homoallylic hydroxy
group, i.e. the chiral allylic alcohols 5g-n (Table 3,
entries 2-9), the diastereomeric control is dictated by the
allylic strain (A1,2 and A1,3) and the dihedral angle (CdC-
C-O) of the substrate in the corresponding loaded
complex G (Scheme 7), rather than the geometrical
effects as presented for ene diols in the loaded complex
A.13 A substituent (R2) at the double bond geminal to
the chirality center provides 1,2-allylic strain, as in
derivative 5h (Table 3, enry 3), and the erythro epoxy
alcohol is the preferred product in moderate (81:19)
diastereoselectivity. In contrast to the typical case with
pronounced A1,2 strain sensitivity, namely the TBHP/VO-
(acac)2 oxidant, for which the erythro-6h is observed
essentially exclusively,2b,15 in the present titanium-
catalyzed epoxidation of 5h as much as 20% of the threo
isomer are obtained. Clearly, the CdC-C-O dihedral
angle in the transition state for the Ti-catalyzed oxygen
transfer is larger than that established for the vanadium
case (ca. 50°).13,15 Indeed, additional evidence for this is
provided by the stereochemical probe 5k with both A1,2
and A1,3 strain,13 for which a dr value of 50:50 was
obtained (Table 3, entry 6). This lack of diastereoselec-
tivity expresses that A1,2 and A1,3 strain compete, which
suggests that the dihedral angle is definitely larger than
50° (A1,2 strain dominates) but smaller than 120° (A1,3
strain prevails). The latter applies for substrates 5j,l
(Table 3, entries 5 and 7), for which the expected threo
selectivity is observed. For a substrate with a larger R5
substituent at the chirality center, i.e. tert-butyl as in
the case of the allylic alcohol 5n , essentially exclusive
erythro diastereoselectivity (entry 9, Table 3) may be
achieved through the control by the A1,2 strain.
(i) The additional coordination in the titanium template
with the homoallylic hydroxy group of the ene diols
guarantees high diastereomeric control. In the previous
item h we have seen that a sterically more demanding
R5 substituent (t-Bu versus Me) in the allylic alcohols
5h ,n (Table 3, entries 3 and 9) increases the diastereo-
meric ratio significantly through A1,2 strain. Hence, it
is tempting to explain the higher diastereomeric ratios
for the ene diols 5b,c compared to the correspondingly
substituted allylic alcohols 5h ,n similarly in terms of the
A1,2 strain caused by the 1-hydroxyethyl and phenylhy-
droxymethyl groups in the ene diols 5b (Table 1, entries
6-11) and 5c (Table 1, entries 12-16). However, if A1,2
strain were important in controlling the high erythro
selectivity, the ene diol 5d without the A1,2 strain should
display a much lower erythro selectivity than substrates
5b,c, but the experimental results unequivocally dem-
onstrate essentially equal diastereoselectivities toward
the â-hydroperoxy alcohol 2, for example entries 7, 13,
and 18 (Table 1). Moreover, comparison of the (E)-
configurated ene diol 5d (Table 1, entry 18) with the (E)-
configurated allylic alcohol 5i (Table 3, entry 4), the
former exhibits a high erythro and the latter even a slight
threo preference. Undoubtly, the observed high erythro
Dia ster eoselectivity
(f) The epoxy diols 6a -e are obtained in high to
excellent erythro diastereoselectivity in the titanium-
catalyzed epoxidation of ene diols 5a -e by â-hydroperoxy
alcohols 1-4. All ene diols 5a -e are epoxidized highly
erythro-diastereoselectively (Tables 1 and 2) without any
marked influence on the diastereomeric ratio by the
substituents in the ene diol. Comparison of all the
diastereomeric ratios in Tables 1 and 2 reveals no clear-
cut trends, but it should be kept in mind that the
variation in the diastereomeric ratio (dr) values (Table
1) is quite small, i.e. from ca. 80:20 (entries 3 and 19) to
a high of ca. 95:5 (entries 13 and 16). The low dr values
pertain to the ene diol derivatives 5a ,d with the â-hy-
droperoxy alcohol 3, the only oxygen donor with a
secondary hydroperoxy functionality (Table 1, entries 3
and 19). It seems that for sterically less encumbered ene
diol substrates such as 5a ,d (R1 or R2 ) H), a greater
steric demand is placed on the oxygen donor for high
diastereomeric control. The latter control derives from
the very rigid transition state formed by the various
Ti-O bonds, as portrayed in the loaded complex A
(Scheme 5). The substituents of the ene diol do not
interfere with the approach of the double bond along a
straight line through the activated O-O bond. Thus,
essentially irrespective of the substitution pattern of the
ene diol, the erythro epoxy diol is the favored product.
(g) The erythro and threo configurations in the diol
substrates do not effect the diastereoselectivity. It makes
no difference for the diastereomeric ratios whether the
threo diol, as in the case of the ene diols 5b-d (Table 1,
entries 6-21), or the erythro diastereomer, namely
derivative 5e (Table 2, entries 3-8), are employed. In
both cases the epoxidations are highly erythro selective.
Again, this may be readily rationalized in terms of the
proposed transition state for the oxygen transfer in the
loaded complex A (Scheme 5). Neither in the erythro (R1′
* H, R1 ) H) nor in the threo (R1′ ) H, R1 * H) isomer
are both substituents R1 and R1′ of the chiral homoallylic
functionality close enough to the reaction center to
(15) Hoveyda, A. H.; Evans, D. A.; Fu, G. C. Chem. Rev. 1993, 93,
1307-1370.
(16) The author has deposited atomic coordinates for this structure
with the Cambridge Crystallographic Data Centre. The coordinates
can be obtained, on request, from the Director, Cambridge Crystal-
lographic Data Centre, 12 Union Road, Cambridge, CB2 1EZ, UK.