Regiochemical effect of an α-trimethylsilyl group on epoxide reactions with non-nucleophilic bases

Full text of DOI:10.1021/ja972256j

J. Am. Chem. Soc. 1997, 119, 11689-11690
11689
opening of the epoxide ring. From the DATMP reaction, 3 was
isolated in 83% yield. It was distinguished from the alternate
regioisomer, R-hydroxysilane 4, by the NMR spectra.5 Epoxy-
silane 1 was less reactive than epoxysilane 2 to R2NMgBr4c
(81% remained after 34 h) and to DATMP4d (45% remained
after 18 h), and no products were isolated.
Reactions of (trimethylsilyl)cyclohexene oxide (7) with
Et2NLi/Et2O and with R2NMgBr/THF gave mixtures of â-hy-
droxysilane 86 and diene 10.7 The alternate regioisomer,
R-hydroxysilane 9,8 was present to the extent of only a few
percent (identified by GC and GC/MS comparison). When 7
was treated with Et2NLi/Et2O, allylic alcohol 8 was the major
initial product, but later diene 10 predominated.
Regiochemical Effect of an r-Trimethylsilyl Group
on Epoxide Reactions with Non-nucleophilic Bases
Paul F. Hudrlik,* Laykea Tafesse, and Anne M. Hudrlik
Department of Chemistry, Howard UniVersity
Washington, D.C. 20059
ReceiVed July 8, 1997
Reactions of epoxides with strong, non-nucleophilic bases
have been widely used for the synthesis of allylic alcohols.1
Bases used to effect these rearrangements include R2NLi,1a,b
R2NMgBr,1a,c diethylaluminum 2,2,6,6-tetramethylpiperidide
(DATMP),1a,d and i-Pr2NLi/KO-t-Bu (LIDAKOR).1e We report
here that a silyl substituent can affect the regiochemical course
of these rearrangements.
Ring opening reactions of R,â-epoxysilanes show a strong
preference for cleavage of the C-O bond R to silicon by a wide
variety of nucleophilic reagents, and a number of years ago,
we introduced a synthesis of olefins and heteroatom-substituted
olefins based on these reactions.2 The reasons for this regio-
chemical preference are not well understood, although a number
of possibilities have been suggested, including initial coordina-
tion of the nucleophile with both silicon and carbon and
coordination with silicon followed by a 1,2-rearrangement.2d,3
In order to broaden our understanding of the regiochemistry of
the reactions of R,â-epoxysilanes, we have studied the influence
of an R-silyl group on the reactions of epoxides with non-
nucleophilic bases, in which prior coordination with silicon
would not be expected.
We have prepared epoxysilanes 1, 2, 7, and 14 and have
studied their reactions with Et2NLi/Et2O, i-Pr2NLi/HMPA,
R2NMgBr/THF, LIDAKOR/THF, and (for 1 and 2) DATMP.
To the extent allylic alcohols are formed, the reactions are
regioselective, and the allylic alcohols are those from R opening
of the epoxide ring. The R2NMgBr reagent gave allylic alcohols
with all the epoxysilanes except 1, which was unreactive.
Et2NLi was slower but did give an allylic alcohol from 7.
DATMP, which has been primarily used for acyclic and large
ring epoxides, was the reagent of choice for epoxide 2. The
use of HMPA resulted in alternate reaction pathways with the
epoxysilanes, as discussed below, and LIDAKOR did not give
any major products cleanly with any of the epoxysilanes.
Allylic alcohol 9 was not the precursor of diene 10. When
9 was treated with Et2NLi/Et2O (room temperature (rt) for 10
min, then 42 °C), it was not converted to 10 (although reaction
in the presence of an internal standard indicated that it very
slowly disappeared, e.g., 63% remained after 12 h). However,
allylic alcohol 8 was converted to diene 10 under the reaction
conditions.
Treatment of epoxysilane 7 with i-Pr2NMgBr in THF/Et2O
or with c-Hex(i-Pr)NMgBr in THF produced allylic alcohol 8
and diene 10 in ratios more favorable to allylic alcohol 8, with
traces of R-hydroxysilane 9. In a preparative experiment using
c-Hex(i-Pr)NMgBr,4e allylic alcohol 8 and diene 10 were
obtained in 64% and 22% yields, respectively.
To gain insight on the effect of steric factors, we have studied
reactions of 1-tert-butylcyclohexene oxide (11). Epoxide 11
yielded allylic alcohol 13,9 the opposite regiochemistry to that
found with silyl epoxide 7. Treatment of 11 with Et2NLi/Et2O
and with c-Hex(i-Pr)NMgBr/THF produced 13 in a slow
reaction. Treatment of 11 with i-Pr2NLi/HMPA (rt, 15 h)
1
produced 13 in 80% yield. In the H NMR spectrum of the
crude product from the reaction with c-Hex(i-Pr)NMgBr,4f the
multiplets at δ 5.86 and 5.80 due to 13 were by far the major
peaks in the olefin region; a small signal at δ 5.62 (crude triplet)
suggests a maximum of 7% of regioisomer 12.10
When (trimethylsilyl)cyclooctene oxide (14) was treated with
c-Hex(i-Pr)NMgBr in THF, â-hydroxysilane 15 was formed in
(4) Reaction conditions: (a) DATMP/benzene/hexanes, 0 °C/4 h, rt/12
h. (b) c-Hex(i-Pr)NMgBr/THF, 0 °C/30 min, rt/20 min, then 40 °C/30 h.
(c) c-Hex(i-Pr)NMgBr/THF, 0 °C/3 h, rt/10 h, then 45 °C. (d) DATMP/
benzene/hexanes, 0 °C/2 h, then rt. (e) c-Hex(i-Pr)NMgBr/THF, 0 °C/2 h,
rt/3 h, 55°C/28 h. (f) c-Hex(i-Pr)NMgBr/THF, 0 °C/3.5 h, rt/12 h, 40 °C/
92 h; product had a 1:1.5 ratio of 11:13 by GC.
Treatment of epoxysilane 2 with DATMP4a or with
c-Hex(i-Pr)NMgBr4b gave only the â-hydroxysilane 3, from R
(1) (a) Crandall, J. K.; Apparu, M. Org. React. 1983, 29, 345-443. (b)
Rickborn, B.; Thummel, R. P. J. Org. Chem. 1969, 34, 3583-3586. (c)
Corey, E. J.; Marfat, A.; Falck, J. R.; Albright, J. O. J. Am. Chem. Soc.
1980, 102, 1433-1435. (d) Yasuda, A.; Tanaka, S.; Oshima, K.; Yamamoto,
H.; Nozaki H. J. Am. Chem. Soc. 1974, 96, 6513-6514. Yasuda, A.;
Yamamoto, H.; Nozaki, H. Bull. Chem. Soc. Jpn. 1979, 52, 1705-1708.
Marshall, J. A.; Audia, V. H. J. Org. Chem. 1987, 52, 1106-1113. (e)
Mordini, A.; Ben Rayana, E.; Margot, C.; Schlosser, M. Tetrahedron 1990,
46, 2401-2410.
(2) (a) Hudrlik, P. F.; Peterson, D.; Rona, R. J. J. Org. Chem. 1975, 40,
2263-2264. (b) Hudrlik, P. F.; Hudrlik, A. M.; Rona, R. J.; Misra, R. N.;
Withers, G. P. J. Am. Chem. Soc. 1977, 99, 1993-1996. (c) Hudrlik, P. F.;
Hudrlik, A. M.; Kulkarni, A. K. Tetrahedron Lett. 1985, 26, 139-142. (d)
For review, see: Hudrlik, P. F.; Hudrlik, A. M. In AdVances in Silicon
Chemistry; Larson, G. L., Ed.; JAI Press: Greenwich, CT, 1993; Vol. 2,
pp 1-89.
(3) In the ring opening with methylorganocopper reagents, we have
shown that the pentacoordinate intermediate [RSiR3Me]- is not involved:
Hudrlik, P. F.; Ma, D.; Bhamidipati, R. S.; Hudrlik, A. M. J. Org. Chem.
1996, 61, 8655-8658.
(5) (a) Close examination of the 1H NMR spectrum suggested the
presence of two isomers (presumably E and Z) in an approximate ratio of
7:1. (b) Regioisomer 4 is not known, but a small multiplet about δ 5.3-
5.4 in the 1H NMR spectra of the crude products may have been due to
about 2.5% (DATMP reaction) and <1% (R2NMgBr reaction) of 4.
(6) Fristad, W. E.; Bailey, T. R.; Paquette, L. A. J. Org. Chem. 1980,
45, 3028-3037.
(7) Paquette, L. A.; Daniels, R. G.; Glelter, R. Organometallics 1984,
3, 560-567.
(8) A comparison sample of 9 was prepared by treatment of cyclohex-
enone with Al(SiMe3)3‚Et2O: Altnau, G.; Ro¨sch, L.; Jas, G. Tetrahedron
Lett. 1983, 24, 45-46.
(9) The structure of allylic alcohol 13 was confirmed by comparison
with the product from treatment of cyclohexenone with tert-butyllithium:
(a) Still, W. C.; Mitra, A. Tetrahedron Lett. 1978, 2659-2662. (b)
Schumann, H.; Genthe, W.; Hahn, E.; Pickardt, J. J. Organomet. Chem.
1986, 306, 215-225.
(10) (a) D’Haenens, L.; Van de Sande, C. C.; Tavernier, D.; Vandewalle,
M. Bull. Soc. Chim. Belg. 1986, 95, 273-281. (b) Corona, T.; Crotti, P.;
Ferretti, M.; Macchia, F. J. Chem. Soc., Perkin Trans. 1 1985, 1607-1616.
S0002-7863(97)02256-7 CCC: $14.00 © 1997 American Chemical Society

11690 J. Am. Chem. Soc., Vol. 119, No. 48, 1997
Communications to the Editor
a slow reaction. After 0 °C/3 h, rt/12 h, and then 43 °C/38 h,
the product mixture contained 59% of 14 and 37% of 15 by
GC. Chromatography gave a pure sample of 15. The 1H NMR
Figure 1. LUMO of 22 (left); LUMO of 24 (right).
Table 1. AM1 Coefficients
spectra of the crude sample showed that a triplet at δ 5.96 and
a doublet of doublets at δ 4.9 were the only peaks in the vinyl
region, indicating that R-hydroxysilane 16 was not present to
any measurable extent.
epoxide
Σc2 R-C
Σc2 â-C
Σc2 Si or CR3
7 (Me3Si)
11 (t-Bu)
22 (Me3Si)
23 (t-Bu)
24 (Me)
0.228
0.364
0.232
0.400
0.413
0.002
0.371
0.002
0.415
0.412
0.651
0.024
0.648
0.037
0.025
Reactions of the tert-butyl analog 18 with base have been
reported to yield isomeric allylic alcohols 19 and 20 in
comparable amounts using Et2NLi/PhH/reflux and using t-BuLi/
pentane/reflux; in the former case, dienes were also formed.11
The reactivity of the epoxysilanes to Et2NLi/Et2O was in some
cases similar to that of their non-silyl analogs and in some cases
lower. Epoxysilane 7 (which gave allylic alcohol 8) had a
similar reactivity to epoxide 5 (which gave 6): when a mixture
of 5 and 7 was treated with Et2NLi/Et2O/45 °C in the presence
of an internal standard, both epoxides disappeared at similar
rates. On the other hand, silyl epoxides 1, 2, and 14 gave slow
reactions, with no major products observed. The corresponding
non-silicon epoxides are known to react (4-octene epoxide f
allylic alcohol; cyclooctene oxide f allylic alcohol + 17).1a
The use of HMPA resulted in different reaction pathways
with the epoxysilanes. Reactions with lithium amide bases in
solvents such as ether are believed to proceed Via a base-
epoxide complex and a syn â-elimination-type pathway.1a
HMPA is believed to break up the complex and has been found
to facilitate the reactions and to affect the selectivity.1a,12,13
With epoxysilanes 1, 2, and 14, the use of HMPA resulted
in desilylation; treatment of 14 with i-Pr2NLi in HMPA (rt, 4
h) gave cyclooctene oxide in 81% yield. Treatment of epoxy-
silane 7 with i-Pr2NLi/HMPA yielded the silanol 21 in 72%
yield in a relatively fast reaction (<2 h at room temp). The
metalation, which would be facilitated by the stabilization of
the allylic anion by silicon.
The regioselectivity of the reactions of epoxysilane 2 (to form
3) can be rationalized by prior formation of the least hindered
complex between the aluminum or magnesium amide and the
epoxide oxygen.1a,d However, the regiochemistry of the ring
openings of epoxysilanes 7 and 14 is remarkable. Although
the C-Si bond is longer than the C-C bond, the comparison
of the reactions of 7 and 14 with those of the tert-butyl epoxides
11 and 18 is noteworthy. A conformational search15a to examine
the role of ground state conformational preferences for epoxides
7 and 11 yielded comparable results: a number of low-energy
conformations of similar energy, the two expected chair
conformations with different rotational conformers of the silyl
and tert-butyl groups.
We believe the R opening of epoxysilanes with non-
nucleophilic bases described here is related to the well-known
nucleophilic R opening of epoxysilanes. We have found that
the preference for R opening of epoxysilanes correlates with
the character of the LUMO.15 For the epoxysilanes 7 and
(trimethylsilyl)ethylene oxide (22), the AM1 LUMO isosurface
is large at the Si and R carbon side of the molecule, and small
at the â-carbon. For the alkyl-substituted epoxides 11, tert-
butylethylene oxide (23), and propylene oxide (24), the iso-
surface is large at both the epoxide carbons. A similar trend
was found for ab initio calculations15 (HF/6-31G*) on silyl-
ethylene oxide (25) and propylene oxide (24). The AM1 LUMO
isosurfaces for (trimethylsilyl)ethylene oxide (22) and for
propylene oxide (24) are shown in Figure 1, and the AM1
coefficients (Σc2) are given in Table 1.
product can be rationalized by a pathway involving deproton-
ation of a methyl group on silicon followed by intramolecular
epoxide opening.14 The disparate pathways taken in the
reactions of epoxysilane 7 and the tert-butyl analog 11 (above)
with i-Pr2NLi in HMPA are most likely related to the stabiliza-
tion of anions in the position R to silicon.
The order of reactivity of epoxysilanes 7 > 14 > 1 can be
rationalized by ease of formation of the base-epoxide complex.
The large amounts of diene (10) formed from reactions of 7 is
unusual for this type of reaction (contrast 5 f 6). The
dehydration of 8 (but not 9) to give diene 10 may involve allylic
R,â-Epoxysilanes exhibit a regiochemical preference for R
opening of the epoxide ring not only with nucleophiles, but with
strong, non-nucleophilic bases as well.
Acknowledgment. Financial support from the National Science
Foundation (CHE-9505465) is gratefully acknowledged.
Supporting Information Available: Experimental details (8 pages).
See any current masthead page for ordering and Internet access
instructions.
(11) Crandall, J. K.; Lin, L.-H. C. J. Org. Chem. 1968, 33, 2375-2378.
(12) Apparu, M.; Barrelle, M. Tetrahedron 1978, 34, 1541-1546.
(13) Morgan, K. M.; Gajewski, J. J. J. Org. Chem. 1996, 61, 820-821.
(14) A similar mechanistic pathway was recently suggested in reactions
of 1,1-bis(trimethylsilyl)epoxides with i-Pr2NLi/Et2O to give silanols:
Hodgson, D. M.; Comina, P. J. Tetrahedron Lett. 1996, 37, 5613-5614.
JA972256J
(15) Calculations were carried out using HyperChem 4 and 5.01 with
ChemPlus 1.5. (a) The conformational search was carried out using MM+,
and the low-energy minima were refined using AM1.