Ortho-[2,3]-wittig rearrangement of benzyl propargyl ethers: Striking preference over the competing [1,2]-wittig shift

Article abstract of DOI:10.1246/cl.2000.1394

Treatment of benzyl γ-(trimethylsilyl)propargyl ether with n-BuLi is shown to afford the rarely precedent ortho-[2,3]-Wittig product in remarkable preference to the [1,2]-Wittig product. The factors governing the periselectivity in this type of carbanion

Full text of DOI:10.1246/cl.2000.1394

1394
Chemistry Letters 2000
Ortho-[2,3]-Wittig Rearrangement of Benzyl Propargyl Ethers:
Striking Preference over the Competing [1,2]-Wittig Shift
Katsuhiko Tomooka, Manabu Harada, Takayuki Hanji, and Takeshi Nakai*
Department of Chemical Technology, Tokyo Institute of Technology, Meguro-ku, Tokyo 152-8552
(Received October 3, 2000; CL-000899)
Treatment of benzyl γ-(trimethylsilyl)propargyl ether with
n-BuLi is shown to afford the rarely precedent ortho-[2,3]-
Wittig product in remarkable preference to the [1,2]-Wittig
product. The factors governing the periselectivity in this type
of carbanion rearrangement are discussed.
First, we carried out the rearrangement of benzyl γ-
(trimethylsily)propargyl ether (1a) under various conditions
(Scheme 1, Table 1). Under the standard condition (n-BuLi,
THF, –78 °C), the ortho-[2,3]-product 3a was obtained in large
preference to the [1,2]-product 4a, unexpectedly, together with
the [1,4]-product 5a (entry 1).8 As expected, the ortho-[2,3] pref-
erence was remarkably enhanced when the reaction temperature
was lowered (entry 2), while the [1,2]-product predominated
when the temperature was raised to 0 °C (entry 3). Interestingly
enough, the use of the lithium amide base led to the predominant
formation of the [1,4]-product (entry 4). Of special interest is
that the ortho-[2,3]/[1.2] ratio (62% : 4%) observed in entry 2 is
significantly higher than that reported for the γ-methylpropargyl
analog (45% : 18%).4 Both the high ortho-[2,3] preference and
the appreciable formation of the [1,4]-product can be explained
in terms of the well-known conjugation of the propargylic lithi-
um 2a with the allenic form 2a which makes the carbanion termi-
nus more stable toward the radical cleavage leading to the [1,2]-
shift and more favorable for the [1,4]-shift. In a similar
rearrangement of the α-methylbenzyl analog (1b), the ortho-
[2,3]-shift was considerably suppressed, instead the [1,2]-Wittig
shift becoming predominant (entries 5–7),9 presumably due to the
radical-stabilizing effect of the α-methyl substituent.
The [2,3]-Wittig rearrangement of α-allyloxy organolithiums
(eq 1) currently enjoys wide application in many facets of organ-
ic synthesis.1 In principle, on the other hand, a similar [2,3]-sig-
matropic shift is conceivable for α-benzyloxy organolithiums to
give, after the 1,3-hydrogen transfer, the ortho-alkylated benzylic
alcohols (eq 2), though it is kinetically less favorable due to the
involvement of the dearomatization step. However, most of α-
benzyloxy organolithiums studied so far, such as G = Ph2 and
alkyls,3 have not afforded any detectable amount of the ortho-
[2,3]-Wittig product, instead producing the [1,2]-Wittig product
via the radical cleavage–recombination pathway (eq 3), while a
single exception has been reported where the rearrangement of
the propargylic substrate (G = C≡ CMe, R = H) affords the ortho-
[2,3]-Wittig product together with the [1,2]-Wittig product.4 In
contrast, a similar ortho-[2,3]-shift has been amply precedent in
the thia-[2,3]-Wittig rearrangement of α-benzylthio organolithi-
ums5 and the Sommelet–Hauser rearrangement of N- and S-ben-
zyl ylides6 (eq 4). These situations are rather peculiar from theo-
retical points of view, since an α-oxy carbanion, compared with
the α-thio analog and the ylide counterpart, should have the
HOMO at a higher level and hence should be more reactive as a
migrating terminus for the [2,3]-sigmatropic shift. In order to
define the factors directing the rearrangement to the ortho-[2,3]-
Wittig shift, we systematically investigated the periselectivity in
the carbanion rearrangement of various benzyl propargylic
ethers.7 Disclosed herein is that the carbanion rearrangement of
benzyl γ-(trimethylsilyl)propargyl ether exhibits remarkably
greater preference for the ortho-[2,3]-Wittig shift, compared with
the non-silylated and γ-methylated counterparts.
Copyright © 2000 The Chemical Society of Japan

Chemistry Letters 2000
1395
In a similar way, we also examined the dianion rearrange-
ment of benzyl propargyl ether where the conjugation with the
allenic form would be inhibited (Scheme 2, Table 2). As expect-
ed, when the rearrangement of 1c was performed in ether the
ortho-[2,3]-shift competed considerably with the [1,2]-shift
(entry 1).10 Rather surprisingly, however, when the rearrange-
ment was conducted in THF the α'-[1,2]-Wittig product 6c aris-
ing from lithiation at the benzyl position (α') was formed as the
major product (entry 2). That means that the second lithiation
involved occurs predominantly at the propargylic position in
ether and at the benzylic position in THF. At present, the exact
reason for the regiochemical changeover is unclear. We also car-
ried out the rearrangement of the α-methylbenzyl analog (1d)
where the lithiation at the benzylic position might be consider-
ably suppressed. Indeed, the rearrangement in ether gave the α-
lithiation-derived ortho-[2,3]- (3d) and [1,2]-product (4d) in 22%
and 51% yield, respectively (entry 3).11 In THF, nearly the same
ratio for [2,3]- vs [1,2]-product was observed, accompanied by
the formation of the α'-[1,2]-product 6d (entry 4). Furthermore,
the rearrangement of the α-methylpropargyl analog (1e) in ether
was found to result in the exclusive formation of the α-[1,2]- (4e)
and α'-[1,2]-product (6e) without any detectable formation of the
ortho-[2,3]-product.12 This observation indicates that the dilithio
species generated from the α-methylpropargyl moiety can no
longer act as a migrating terminus for the ortho-[2,3]-Wittig shift.
The overall spectrum of the periselectivity observed herein
reveals that a key structural factor for effecting the ortho-[2,3]-
Wittig rearrangement is not the HOMO level of the carbanion
terminus involved, but its kinetic stability toward the radical
cleavage leading to the competing [1,2]-Wittig shift. In other
words, the more the [1,2]-Wittig shift is suppressed, the more
the ortho-[2,3]-Wittig produst is formed. The same argument
could be extended to explain why α-benzylthio carbanions and
S- and N-benzyl ylides can easily undergo the ortho-[2,3]-shift.
In summary, we have shown that benzyl γ-(trimethylsilyl)-
propargyl ether (1a), when treated with n-BuLi, undergoes the
rarely precedent ortho-[2,3]-Wittig rearrangement in remark-
ably large preference to the competing [1,2]- and [1,4]-shifts.
Furthermore, we have pointed out the importance of the kinetic
stability of the carbanion terminus toward the radical cleavage
as a key factor directing the rearrangement to the ortho-[2,3]-
Wittig shift.
References and Notes
1
2
Reviews: a) T. Nakai and K. Mikami, Chem. Rev., 86, 885
(1986). b) T. Nakai and K. Mikami, Org. React., 46, 105
(1994).
a) G. Wittig and L. Lmann, Liebigs Ann. Chem., 550, 260
(1942). b) H. Felkin and C. Frajerman, Tetrahedron Lett.,
1977, 3485. c) K. Tomooka, K.Yamamoto, and T. Nakai,
Angew. Chem., Int. Ed., 38, 3741 (1999).
3
4
a) K. Tomooka, T. Igarashi, and T. Nakai, Tetrahedron, 50,
5927 (1994). b) K. Tomooka, H. Yamamoto, and T.Nakai,
Liebigs Ann./Recueil, 1997, 1275.
U. Schölkopf, K. Fellenberger, and M. Rizk, Liebigs Ann.
Chem., 734, 106 (1970). In this paper, any mechanistic
reason for the exceptional periselectivity has not been
given. Note that a similar [2,3]-Wittig shift has been
reported in the carbanion rearrangement of several 3-furyl-
methyl ethers: B. Cazes and S. Julia, Synth. Commun.,
1977, 113; M. Tsubuki, H. Okita, and T. Honda, J. Chem.
Soc., Chem. Commun., 1995, 2135 .
5
6
K. Brickmann and R. Brkner, Chem. Ber., 126, 1227
(1993).
Review: I. Markó, “Comprehensive Organic Synthesis,”ed.
by G. Pattenden, Pergamon Press, Oxford (1991), Vol. 3,
p. 913.
7
Note that the ortho-[2,3]-Wittig shift requires benzylic
ethers as substrates which could be lithiated on the benzyl-
oxy-bearing carbon preferentially over the benzylic carbon.
In this regard, allyl benzyl ether, e.g., is not qualified as an
substrate, since its lithiation occurs exclusively on the ben-
zylic carbon.
1
8
9
The three products are distinguishable by H NMR spectra
(CDCl3): 3a, δ 2.44 (s, o-Me) and δ 5.60 (d, CH–OH); 4a,
δ 2.95–3.06 (m, CH2Ph) and δ 4.53–4.59 (m, CH–OH); 5a,
δ 10.01 (d, CHO), and δ 6.32 (dt, =CHCHO).
1
The H NMR spectra of products 3b, 4b, and 5b are in
accord with the assigned structures.
10 The three products are distinguishable by 1H NMR spectra:
3c, δ 2.45 (s, o-Me), δ 5.63 (d, CH–OH), and δ 2.65 (d,
HC≡ ); 4c, δ 2.97–3.09 (m, CH2Ph), δ 4.59 (ddd,
CH–OH), and δ 2.50 (d, HC≡ ); 6c, δ 4.88 (t, CH–OH), δ
2.65 (ddd, CH2C≡ ), and δ 2.08 (t, HC≡ ).
1
11 The H NMR spectra of products 3d, 4d, and 6d are in
accord with the assigned structures.
12 1H NMR (CDCl3): 4e, δ 2.94, 3.00 (dd, CH2Ph) and δ
1.56 (s, Me–C–OH); 6e, δ 4.52 (d, CH–OH) and δ 1.11
(d, Me–CH).