Tandem formation and [2,3] rearrangement of methylene ammonium ylides derived from amines and the Simmons-Smith reagent

Article abstract of DOI:10.1021/ol034404i

(Matrix presented) Zinc-complexed methylene ammonium ylides are formed from tertiary amines and the Simmons-Smith reagent. These stable entities can be activated with n-BuLi to allow reactions typical of ammonium ylides such as [2,3] rearrangements. In th

Full text of DOI:10.1021/ol034404i

ORGANIC
LETTERS
2003
Vol. 5, No. 10
1757-1760
Tandem Formation and [2,3]
Rearrangement of Methylene Ammonium
Ylides Derived from Amines and the
Simmons−Smith Reagent
,†
Varinder K. Aggarwal,* Guang yu Fang,† Jonathan P. H. Charmant,† and
Graham Meek‡
Department of Chemistry, UniVersity of Bristol, Cantock’s Close, Bristol, UK,
and Dowpharma, Chirotech Technology, Ltd., A Subsidiary of the Dow Chemical
Company, Cambridge Science Park, Milton Road, Cambridge, UK
V.aggarwal@bristol.ac.uk.
Received March 7, 2003
ABSTRACT
Zinc-complexed methylene ammonium ylides are formed from tertiary amines and the Simmons−Smith reagent. These stable entities can be
activated with n-BuLi to allow reactions typical of ammonium ylides such as [2,3] rearrangements. In the case of oxazolidine 12, ylide formation,
activation, and subsequent [2,3] rearrangement was highly efficient and occurred with very high diastereoselectivity.
Over 40 years ago, Wittig described the reaction of tri-
methylamine with a variation of the Simmons-Smith reagent
(ClCH2)2Zn generated from CH2N2/ZnCl2, which on aqueous
workup gave the quaternary ammonium salt 1 (Scheme 1).1
that this chemistry has remained completely dormant since
this initial report, especially because the generation of
ammonium ylides2 without additional anion-stabilizing groups
can be troublesome. Two methods have previously been
employed to generate such ylides: fluoride-induced desily-
lation of R-trimethylsilylammonium salts3 and the trans-
metalation of R-trimethylsilylammonium salts with n-BuLi.4
We felt that Wittig’s zinc-complexed ammonium ylide
offered some potential for generating unsubstituted am-
monium ylides. However, Wittig’s observations also caused
us concern about the hurdles in attempting to exploit this
process. Ammonium ylides are generally highly reactive
intermediates that readily undergo Stevens rearrangements,5
but the zinc-complexed ylide 3 was reported to be stable.
Scheme 1
Alternatively, quenching with iodine furnished the iodo-
methylammonium salt 2. These reactions presumably occur
via the zinc-complexed ammonium ylide 3. It is surprising
(2) For a review on ammonium ylide, see: Marko´, I. E. In ComprehensiVe
Organic Synthesis; Trost, B. M., Fleming, I., Eds.; Pergamon Press:
Elmsford, NY, 1991; p 913. For a recent review on asymmetric ylide
chemistry, see: Aggarwal, V. K.; Li, A. H.; Dai, L. X. Chem. ReV. 1997,
97, 2341.
* To whom correspondence should be addressed.
† University of Bristol.
(3) Vedejs, E.; West, F. G. Chem. ReV. 1986, 86, 941.
(4) Maeda, Y.; Sato, Y. J. Org. Chem. 1996, 61, 5188.
(5) Stevens, T. S.; Creighton, E. M.; Gordon, A. B.; Macnicol, M. J.
Chem. Soc. 1928, 3193.
‡ Dowpharma, Chirotech Technology, Ltd.
(1) Wittig, G.; Schwarzenbach, K. Liebigs Ann. Chem. 1961, 650, 1.
10.1021/ol034404i CCC: $25.00 © 2003 American Chemical Society
Published on Web 04/16/2003

Was this due to the known lower propensity of the metal-
free ammonium ylide related to 3 to undergo the Stevens
rearrangement,6 or was this due to the stability associated
with the formation of a zinc complex? In this paper, we show
that the latter explanation is correct and describe a method
for converting the stable zinc-complexed ammonium ylide
into a reactive one.
participates in these reactions, whereas ammonium ylides,
being much less stable,12 remain tightly bound to the zinc
cation, resulting in an insignificant amount of the free ylide
(Scheme 3).
Scheme 3
We began our studies by repeating Wittig’s reaction,
although we used the more user-friendly Furukawa protocol7
for generation of the Simmons-Smith reagent and employed
quinuclidine rather than trimethylamine, and some useful
observations were made. Treatment of quinuclidine 4 with
the Simmons-Smith reagent in diethyl ether resulted in the
formation of a white precipitate after a few minutes at -10
°C. Interestingly, if the reaction was worked up at this stage,
quinuclidine was regenerated, indicating that amine-zinc
complex 5 had been formed. Workup of the reaction mixture
after 2 days at 0 °C gave the methylammonium salt 7, which
is presumably formed from the zinc-complexed ammonium
ylide 6 (Scheme 2).
We reasoned that formation of an ate complex should
render the organozinc reagent more reactive, as dissociation
to form a zinc free ylide now seemed more feasible (Scheme
3). A number of additives were tested,13 and it was found
that n-BuLi was effective. The product of [2,3] sigmatropic
rearrangement 9a was formed upon treatment of the zinc-
complexed ammonium ylide 10a with n-BuLi, albeit in low
yield (Scheme 4).
Scheme 2
Scheme 4
To test whether the lack of rearrangement of the zinc-
complexed ammonium ylide was related to the stability
associated with the [2.2.2] bicyclic amine, an alternative
amine with a high propensity for both [1,2] and [2,3]
rearrangements was sought. Thus, unsaturated amine 8a was
selected and treated with the Simmons-Smith reagent.
However, no rearrangement products or cyclopropanes8 were
formed, but as before, the N-methylated ammonium salt was
isolated instead. This indicates that the zinc-complexed
ammonium ylide is completely unreactive. We were surprised
at the complete lack of reactivity of these ylides, especially
because sulfides are known to react with the Simmons-
Smith reagent to give zinc-complexed sulfonium ylides,
which readily participate in [2,3] rearrangements,9 epoxida-
tions,10 and aziridinations.11 Presumably, in the sulfide case,
there is a significant amount of the free ylide, which
During optimization of the reaction conditions, our first
observation was that the choice of the Simmons-Smith
reagent was crucial. When the reagent generated from 1 equiv
(10) (a) Aggarwal, V. K.; Ali, A.; Coogan, M. P. J. Org. Chem. 1997,
62, 8628. (b) Aggarwal, V. K.; Coogan, M. P.; Stenson, R. A.; Jones, R.
V. H; Fieldhouse, R.; Blacker, J. Eur. J. Org. Chem. 2002, 2, 319.
(11) Aggarwal, V. K.; Stenson, R. A.; Jones, R. V. H.; Fieldhouse, R.;
Blacker, J. Tetrahedron Lett. 2001, 42, 1587.
(6) Meada, Y.; Sato, Y. J. Chem. Soc., Perkin Trans. 1 1997, 1491.
(7) (a) Furukara, J.; Kawabata, N.; Nishimura, J. Tetrahedron Lett. 1968,
3495. (b) Furukara, J.; Kawabata, N.; Nishimura, J. Tetrahedron 1969, 25,
2647.
(8) Allylic amines bearing a non-nucleophilic nitrogen readily undergo
cyclopropanation with the Simmons-Smith reagent; see: Wipf, P.; Kendall,
C.; Stephenson, C. R. J. J. Am. Chem. Soc. 2003, 125, 761.
(9) (a) Kozarych, Z.; Cohen, T. Tetrahedron Lett. 1982, 23, 3019. (b)
Charette, A. B.; Marcoux, J. F.; Molinaro, C.; Beauchemin, A.; Brochu,
C.; Isabel, E. J. Am. Chem. Soc. 2000, 122, 4508.
(12) pKa values of Et3N+CH2Ph and Bu2S+CH2Ph are 30.8 and 18.8,
respectively: Cheng, J. P.; Liu, B; Zhao, Y. Y.; Sun, Y.; Zhang, X; Lu, Y.
J. Org. Chem. 1999, 64, 604.
(13) We have tried MeONa and pseudo-ephedrine salts made from 1
equiv of (1R,2R)-pseudo-ephedrine/1 equiv of LiHMDS or NaHMDS and
1 equiv of (1R,2R)-pseudo-ephendrine/2 equiv of n-BuLi as additives in
THF solution at 0 °C. In all cases, no neutral amines were obtained after
workup.
1758
Org. Lett., Vol. 5, No. 10, 2003

of ZnEt2 and 2 equiv of ICH2Cl was treated with the tertiary
amine 8a, a small amount of cyclopropane byproduct was
observed in addition to the rearrangement product. When
the Simmons-Smith reagent was generated from 1 equiv
of ZnEt2 and 1 equiv of ICH2Cl or 1 equiv of CH2I2 and
treated with amine 8a, low conversion to 9a was observed.
However, the species generated from 1 equiv of ZnEt2 and
2 equiv of CH2I2 had the desired balance of reactivity and
selectivity. With 1.1 equiv of of this reagent (generated from
1.1 equiv of ZnEt2 and 2.2 equiv of CH2I2), we observed
that the amine was totally consumed and the rearrangement
product was the only product obtained. This was important,
as it was difficult to separate the rearrangement product from
the starting material or any other byproducts. TBME, diethyl
ether, benzene, and toluene were superior to other solvents,
and reaction in diethyl ether gave the highest yield.
These optimized conditions were then applied to a series
of tertiary allylamines (Table 1).
Scheme 5
Treatment of oxazolidine 12 with our standard optimized
procedure furnished the ring-expanded product 13, derived
from [2,3] sigmatropic rearrangement in 72% yield as a
single diastereoisomer. A small amount of the morpholine
derivative 14 was also isolated, again, as a single diastere-
oisomer (Scheme 6).
Scheme 6
a
Table 1.
substrate
R1
R2
R3
yield (%)
8a
8b
8c
8d
Ph(CH2)2
Ph(CH2)2
Ph(CH2)2
PhCH2
H
CH3
H
H
70
50
76
30b
CH3
Ph
H
The relative stereochemistry of the eight-membered ring
was determined by X-ray crystallography, and the relative
stereochemistry of the [1,2] rearrangement product was
determined by NOESY NMR.
H
a Stoichiometry of the process is ZnEt2:CH2I2:amine:nBuLi ) 1.1:2.2:
1:3.9. b Starting material (24%) was recovered; a small amount of Som-
melet-Hauser rearrangement product was observed.
The formation of 13 can be rationalized by considering
the possible TSs for rearrangement and the possible diaster-
eomers (15/16) generated upon ylide formation. Of the two
possible diastereomers, ylide 15 is expected to be the major
isomer due to the known preference for alkylation/ylide
formation of five-membered ring heterocycles to occur cis
to adjacent substituents as a result of the favored conforma-
tion of the ring (Scheme 7).17,18c Subsequent [2,3] sigmatropic
rearrangement of ylide 15 has to occur on the (s)-cis
conformer of the alkene to furnish the (Z)-oxazocine 13.18
Although the product from 8a could potentially arise from
either a [1,2] or [2,3] rearrangement, the product from
substrates 8b and 8c proved that [2,3] sigmatropic rearrange-
ment occurred exclusively. The benzylamine derivative 8d
gave a lower yield of the [2,3] rearrangement product 9d.
In addition to recovery of a small amount of starting material,
a small amount of Sommelet-Hauser rearrangement14
product was also observed.
Oxazolidine derivatives provided additional opportunities
to probe the scope and the stereochemical consequences of
the novel rearrangement process. Oxazolidine 12 was
prepared in a facile manner from (1R,2R)-pseudo-ephedrine
and cinnamaldehyde in the presence of Na2SO4. A single
diastereomer was formed in high yield,15 the relative stere-
ochemistry of which was assigned by analogy with previous
reports on related substrates (Scheme 5).16
Scheme 7
(14) Sommelet, M.; Hauser, C. R. Hebd. Seances Acad. Sci. 1937, 205,
56.
(15) (a) Khruscheva, N. S.; Loim, N. M.; Sokolov, V. L.; Makhaev, V.
D. J. Chem. Soc., Perkin Trans. 1 1997, 2425. (b) Huche, M.; Aubouet, J.;
Pourcelot, G.; Berlan, J. Tetrahedron Lett. 1983, 24, 585. (c) Hassan, A.;
Rene, G.; Robert, C. Tetrahedron Lett. 1982, 23, 503. (d) Mangeney, P.;
Alexakis, A.; Normant, J. F. Tetrahedron, 1984, 40, 1803.
(16) Hussain, A.; Wyatt, P. B. Tetrahedron 1993, 49, 2123.
Org. Lett., Vol. 5, No. 10, 2003
1759

The (s)-trans conformer of 15 places the alkene too far away
from the ylide carbon to effect cyclization. [2,3] Sigmatropic
rearrangement of ylide 16 in the (s)-trans conformation (the
(s)-cis conformer places the groups too far apart) would lead
to formation of the (E)-isomer of the eight-membered ring
compound.18a,c However, such rings are highly strained and
their formation must be accompanied by a higher activation
barrier. In this case, an alternative pathway for rearrangement
emerges: the Stevens rearrangement to give morpholine 14
(Scheme 7).19
Scheme 8
While morpholine 14 could indeed have arisen from a
Stevens rearrangement of ylide 16,20 the stereochemistry of
the product caused us concern, as such intramolecular
rearrangements usually proceed with retention of stereo-
chemistry.21 In our case, clean inversion was observed. It is
possible that the neighboring oxygen atom and cinnamyl
groups confer added stability to the radical intermediate 17,
allowing time for bond rotation to occur prior to ring closure.
Alternatively, it is possible that the intermediate ylide
undergoes ring opening to give oxonium ion 1822, which is
then trapped by the carbanion to give compound 14. This
mechanism does involve an intermediate non-stabilized
carbanion, but this may be stabilized by the zinc salts present
in solution (Scheme 8).
In conclusion, we have shown that amines react with the
Simmons-Smith reagent to give zinc-complexed ammonium
ylides, which are highly stable. Even amines bearing allyl
groups showed no indication of rearrangement. This indicates
that Wittig’s inference of the formation of a zinc-complexed
ammonium ylide was primarily due to the stability of the
metal-complexed ylide and not due to the reduced propensity
for rearrangement. In addition, we have described a protocol
for converting the easily accessible, stable zinc-complexed
ammonium ylides into reactive species. This involves addi-
tion of n-BuLi, which converts the zinc complex into a much
more reactive zincate complex that is capable of undergoing
[2,3] rearrangement.
(17) Glaeske, K. V.; West, F. G. Org. Lett. 1999, 1, 31.
(18) For other examples of rearrangements of ammonium ylides to give
eight-membered rings, see: (a) Clark, J. S.; Hodgson, P. B. Tetrahedron
Lett. 1995, 36, 2519. (b) Wright, D. L.; Weekly, R. M.; Groff, R.; McMills,
M. C. Tetrahedron Lett. 1996, 37, 2165. (c) Vedejs, E.; Hagen, J. P.; Roach,
B. L.; Spear, K. L. J. Org. Chem. 1978, 43, 1185.
(19) For another example in which one diastereoisomer of an ylide
undergoes [2,3] rearrangement and the other diastereomer undergoes a [1,2]
rearrangement, see ref 18b.
(20) Ylide 15 would not be expected to undergo Stevens rearrangement,
as [2,3] sigmatropic rearrangements are generally faster processes than [1,2]
rearrangements. Mageswaran, S.; Ollis, W. D.; Sutherland, I. O.; Thebt-
aranonth, Y. J. Chem. Soc., Chem. Commun. 1971, 1494. For a review on
Stevens rearrangements, see ref 2.
Acknowledgment. We thank Chirotech for financial
support.
(21) (a) Chelucci, G.; Saba, A.; Valenti, R.; Bacchi, A. Tetrahedron:
Asymmetry 2000, 11, 3449. (b) Naidu, B. N.; West, F. G. J. Am. Chem.
Soc. 1994, 116, 8520.
(22) For other examples of ylides derived from acetals/thioacetals
undergoing effectively a [1,2] rearrangement, see: (a) Doyle, M. P.; Griffin,
J. H.; Chinn, M. S.; Leusen, D. V. J. Org. Chem. 1984, 49, 1917. (b)
Trauner, D.; Porth, S.; Opatz, T.; Bats, J. W.; Giester, G.; Mu¨lzer, J.
Synthesis 1998, 653. (c) Mu¨lzer, J.; Trauner, D.; Bats, J. W. Angew. Chem.,
Int. Ed. Engl. 1996, 35, 1970.
Supporting Information Available: Spectroscopic data
and experimental procedures for compounds 8a-d, 9a-c,
and 12-14 and X-ray crystallography of compound 13. This
material is available free of charge via the Internet at
http://pubs.acs.org.
OL034404I
1760
Org. Lett., Vol. 5, No. 10, 2003