Substituted allyl diphenylphosphine oxides as radical allylating agents

Article abstract of DOI:10.1002/anie.200601556

(Chemical Equation Presented) Radically useful: The radical addition of dithiocarbonates to branched allyl diphenylphosphine oxides can be followed by elimination of a diphenylphosphinoyl radical, thereby giving rise to allylated products (see scheme; Phth = phthalimido). Highly functionalized structures can thus be rapidly assembled under mild conditions and from cheap and readily available substrates and reagents.

Full text of DOI:10.1002/anie.200601556

Communications
Allylation
DOI: 10.1002/anie.200601556
Substituted Allyl Diphenylphosphine Oxides as
Radical Allylating Agents**
Gilles Ouvry, BØatrice Quiclet-Sire, and Samir Z. Zard*
Following its discovery nearly a quarter of a century ago,[1] the
radical allylation using allylstannanes has proved to be a very
powerful synthetic tool. Numerous applications attest to the
[*] Dr. G. Ouvry, Dr. B. Quiclet-Sire, Prof. S. Z. Zard
Laboratoire de Synth›se Organique AssociØ
CNRS (UMR 7652), DØpartement de Chimie
Ecole Polytechnique, 91128 Palaiseau Cedex (France)
Fax: (+33)1-6933-3851
E-mail: zard@poly.polytechnique.fr
[**] G.O. thanks the Ecole Polytechnique for a fellowship.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
5002
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 5002 –5006

Angewandte
Chemie
efficiency and compatibility of this methodology with a wide
number of functional groups.[2] Nevertheless, the reaction
suffers from some limitations, especially in respect to the
substitution pattern around the allyl group. g-Substituted
allylstannanes rapidly rearrange under the reaction condi-
tions into their more stable a-isomers [Eq. (1)], and these
were found to react mostly through abstraction of the allylic
hydrogen atom [Eq. (2)] rather than by the desired addition
fragmentation (Scheme 1).[2c,3,4]
idea of the rate of b-elimination.[17] We also chose to use the
higher-boiling chlorobenzene as the solvent to favor the
fragmentation pathway, which has a large, positive entropy of
activation. In the event, heating adduct 3 with lauroyl
peroxide in chlorobenzene resulted in the formation of 4
and the reduced derivative 5 as the major products. No
elimination of the phosphonate group was observed
(Scheme 2).
Scheme 1. Some side reactions during allylations with substituted
allyltriorganotin reagents.
Several systems using allylcobalt,[5] allylgallium,[6] allyl-
halogen,[7] or allylsulfur[8] derivatives have been examined in
an attempt to overcome these limitations. Nevertheless, most
of these processes require a stoichiometric amount of a tin-
based promoter[9] (or other heavy-metal derivatives) and have
usually been applied to the introduction of simple allyl
moieties.[10]
Our method to devise a new, tin-free allylation process
was initially to separate the addition and fragmentation steps.
This approach would eliminate the possibility of rearrange-
ment of the allylating agent and thus obviate the shortcomings
of the previous methods.[11] We thus needed an allylating
agent substituted with a relatively poor leaving group (in the
radical sense), which would undergo intermolecular addition
and transfer of a dithiocarbonate unit and b-elimination in the
second step.[12] In this respect, a phosphorus-centered leaving
group appeared suitable. The reversible addition of phos-
phines to olefins is well known.[13] More recently, we and other
groups found that a diethylphosphinyl group could be
expelled when adjacent to nitrogen- or oxygen-centered
radicals.[14] Other related observations include that of Clive
and Kang, who reported the elimination of an aryl phosphinyl
radical from a cyclohexadienyl system,[15] and the isolation of
a minor (8% yield) by-product that arises from an apparent b-
elimination of a methylphenylphosphinoyl radical located on
a tertiary carbon atom, as reported by Malacria and co-
workers.[16] All cases these were especially favorable, but did
not foretell the generality of the elimination process in
ordinary situations.
Scheme 2. Negative experiments with allyl phosphonate reagents.
DCE=1,2-dichloroethane.
Weakening the carbon–phosphorus bond by substitution
of the phosphonate group was the next logical step. Obtaining
substituted derivatives was also more interesting synthetically
and the purpose of our study in the first place. The radical
addition of the same dithiocarbonate 1 to allylphosphonate 6
occurred fairly efficiently to give adduct 7 in 60% yield, but
when this product was heated in 1,2-dichloroethane with
lauroyl peroxide, neither elimination nor ring closure to the
tetralone was observed. Instead, hydrogen abstraction from
the side chain occurred to ultimately give compound 8 in 36%
yield (Scheme 3). Steric hindrance in this case had clearly
slowed down tetralone formation, and a Thorpe–Ingold effect
had facilitated the 1,5-hydrogen shift. Switching to a higher-
boiling solvent was obviously not going to overcome this
unwanted radical translocation.
Decidedly, the phosphonate entity was too poor a leaving
group for our objectives. It remained for us to examine the
influence of the substituents around the phosphorus center on
its leaving ability. The use of a diphenylphosphine oxide
group seemed the most judicious choice as, in addition to
ready availability, the bond we wanted to break would now
acquire some benzylic character and perhaps be weakened
sufficiently to undergo b-scission. Strain relief upon cleavage
of such a bulky group should provide an extra driving force.
Radical addition of dithiocarbonate 9 to allyl diphenylphos-
phine oxide 10 took place in high yield, and exposure to
We initially envisaged the use of a simple diethyl
phosphonate as the departing group and chose to study the
behavior of phosphonate 3 derived from the addition of p-
bromophenacyl dithiocarbonate 1 to diethyl allylphosphonate
2. The choice was dictated by the possibility of the inter-
mediate radical undergoing closure to the aromatic ring to
give tetralone 4, and this process, although unwanted in the
present case, should act as an internal clock and give us an
Angew. Chem. Int. Ed. 2006, 45, 5002 –5006
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 5003

Communications
amounts of the peroxide (0.5 equiv) were needed. It seems
that the properties that make the diphenylphosphinoyl radical
a good “leaving group” cause it to be a poor chain-propagat-
ing agent. All our attempts at determining the fate of the
phosphorus species failed.
No elimination occurred in the absence of peroxide, so a
purely thermal mechanism can be discarded. Furthermore,
the mixture of the two diastereoisomers of 15 could be
separated, and only one of the two was in fact subjected to the
fragmentation process. Monitoring by thin layer chromatog-
raphy (tlc) indicated that a rapid equilibration occurred
between the two diastereoisomers 15a and 15b before
elimination of the diphenylphosphinoyl group. Thus, the
relative stereochemistry of the initial addition product will
have no consequence on the elimination process. Finally, it
turns out that it is not even necessary to isolate the addition
product 15, as the allylation can be accomplished directly by
heating dithiocarbonate 9 and phosphine oxide 14 in chlor-
obenzene at reflux using the much longer lasting di-tert-butyl
peroxide instead of lauroyl peroxide. The yield of enone 16
was 69%, based on 9, thus representing a significant improve-
ment on the two-step procedure.
Scheme 3. First successful allylations with allyl diphenylphosphine
oxides.
peroxide in refluxing chlorobenzene gave the corresponding
tetralone 12 and the desired elimination product 13, both in
low yield. The formation of ally-
We next explored the scope of this reaction by varying the
dithiocarbonate and allylic diphenylphosphine oxide. The
lated derivative 13 was neverthe-
less a good start, as it was expected
that substitution would make the
elimination step more efficient.
Indeed, when substituted diphenyl-
phosphine oxide 14 was employed
as the radical trap, the intermolec-
ular addition proceeded quite sat-
isfactorily; more pleasing, it was
observed that smooth elimination
of the diphenylphosphinoyl group
took place upon heating 15 with
lauroyl peroxide in refluxing chlor-
obenzene to give unsaturated
ketone 16 in 63% yield as a
mixture of geometric isomers with
the expected preponderance of the
E isomer.[18]
The diphenylphosphine oxide
group thus possessed the requisite
properties we sought, and the
desired allylation could at last be
achieved. In principle, the process
can be a chain reaction, propagated
by the reaction of the diphenyl-
phosphinoyl radical with the
dithiocarbonate moiety. The addi-
tion–fragmentation of phosphorus-
centered radicals on thiocarbonyl
derivatives is well documented in
the pioneering studies of Barton
et al.[19] In practice, however, this
approach did not turn out to be
efficient, and nearly stoichiometric
Table 1: Examples of allylations with branched allyl diphenylphosphine oxides.
Dithiocarbonate Phosphine oxide Allylation product
Yield [%][a]
58
9
9
17
19
18
20
52
21
23
23
19
17
25
22
24
26
72
68
67 (9:1)
27
27
27
25
30
30
28
31
32
70
66
71 (83:17)
[a] Ratio of the E/Z isomers is given in brackets.
5004 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 5002 –5006

Angewandte
Chemie
results are summarized in Table 1. It can be seen that
prenylation is readily accomplished with phosphine oxide
17, whereas allyl and homoallyl acetates can be introduced
with the corresponding phosphine oxides 19, 25, and 30. The
dithiocarbonate moiety itself can bear a number of useful
functional groups. Besides the initial phenacyl derivative 9,
substrates can contain a lactone (as in 21), a protected
aldehyde ester (as in 23), or, perhaps most interestingly, a
masked a-aminoketone (as in 27; Phth = phthalimido). a-
Aminoketones are at the centre of several classical syntheses
point, access to numerous substituted allyl diphenylphosphine
oxides can be accomplished by direct reaction of the anion
derived from the simplest member 10 with various electro-
philes (alkylating agents, epoxides, aldehydes, and so
forth).[20] Another powerful route is through the Arbuzov–
Tripett rearrangement starting from allylic alcohols.[21] The
radical reaction itself is flexible, convergent, and takes place
under mild neutral conditions.
of heteroaromatic rings, such as pyrroles and pyridines, and Experimental Section
Typical procedure for the radical allylation: Di-tert-butyl peroxide (a
are not always readily accessible. Moreover, the introduction
of many of these allylic fragments would not be trivial by the
more common ionic processes, especially with the more
functionalized substrates.
few drops, ca. 100 mg) was added to a solution of the dithiocarbonate
(1.0 mmol) and allyl phosphine oxide (2.0 mmol) in refluxing
degassed chlorobenzene (10 mL) in a nitrogen atmosphere. A few
more drops (ca. 100 mg) of di-tert-butyl peroxide were added after 4 h
at reflux if the reaction is not yet complete (tlc). The reaction mixture
was then cooled to room temperature, concentrated in vacuo, and
purified by flash chromatography.
Our initial attempt to extend this approach to the
À
formation of C C bonds at the anomeric position of
carbohydrates, such in the 2-deoxyglucose derivative 33, was
frustrated by the premature elimination of the dithiocarbon-
ate group at the temperature of refluxing chlorobenzene to
Received: April 20, 2006
À
give glucal 34 as the major product (Scheme 4). The C Sbond
Keywords: b-elimination · allylation · dithiocarbonates ·
phosphane oxides · radical reactions
.
[1] a) J. Grignon, M. Pereyre, J. Organomet. Chem. 1973, 61, C33 –
C35; b) M. Kosugi, K. Kurino, K. Takayama, T. Migita, J.
Organomet. Chem. 1973, 56, C11 – C13.
[2] a) G. E. Keck, J. B. Yates, J. Am. Chem. Soc. 1982, 104, 5829 –
5831; b) G. E. Keck, E. J. Enholm, D. F. Kachensky, Tetrahedron
Lett. 1984, 25, 1867 – 1870; c) G. E. Keck, E. J. Enholm, J. B.
Yates, M. R. Wiley, Tetrahedron 1985, 41, 4079 – 4094; d) G. E.
Keck, D. F. Kachensky, E. J. Enholm, J. Org. Chem. 1985, 50,
4317 – 4325.
[3] J. E. Baldwin, R. M. Adlington, D. J. Birch, J. A. Crawford, J. B.
Sweeney, J. Chem. Soc. Chem. Commun. 1986, 1339 – 1340.
[4] For exceptions, see: a) G. A. Russell, L. L. Herold, J. Org. Chem.
1985, 50, 1037 – 1040; b) T. Migita, K. Nagai, M. Kosugi, Bull.
Chem. Soc. Jpn. 1983, 56, 2480 – 2484; c) H. Fliri, C.-P. Mak, J.
Org. Chem. 1985, 50, 3438 – 3442; d) C. J. Easton, I. M. Scharf-
billig, J. Org. Chem. 1990, 55, 384 – 386; e) Y. Watanabe, T.
Yoneda, T. Okumura, Y. Ueno, T. Toru, Bull. Chem. Soc. Jpn.
1993, 66, 3030 – 3033; f) Y. Yoshida, N. Ono, F. Sato, J. Org.
Chem. 1994, 59, 6153 – 6155; g) I. Ryu, S. Kreimerman, T.
Niguma, S. Minakata, M. Komatsu, Z. Luo, D. P. Curran,
Tetrahedron Lett. 2001, 42, 947 – 950.
[5] M. D. Johnson, Acc. Chem. Res. 1983, 16, 343 – 349.
[6] S.-I. Usugi, H. Yorimitsu, K. Oshima, Tetrahedron Lett. 2001, 42,
4535 – 4538.
[7] C. C. Huval, D. A. Singleton, Tetrahedron Lett. 1993, 34, 3041 –
3042.
[8] a) G. E. Keck, J. H. Byers, J. Org. Chem. 1985, 50, 5442 – 5444;
b) A. Yanagisawa, Y. Noritake, H. Yamamoto, Chem. Lett. 1988,
1899 – 1902; c) P. Breuilles, D. Uguen, Tetrahedron Lett. 1990, 31,
357 – 360.
[9] For emphasis on the problems linked with tin-mediated radical
chemistry, see: P. A. Baguley, J. C. Walton, Angew. Chem. 1998,
110, 3272 – 3283; Angew. Chem. Int. Ed. 1998, 37, 3072 – 3082; for
a review of tin-free allylations using allylsulfones, see: F.
Bertrand, F. Le Guyader, L. Liguori, G. Ouvry, B. Quiclet-Sire,
S. Seguin, S. Z. Zard, C. R. Acad. Sci. Ser. II 2001, 4, 547 – 555.
[10] One notable exception is the elegant use of a branched
allylsulfone in the total synthesis of (+)-pseudomonic acid C;
Scheme 4. Two special cases. Ts=p-toluenesulfonyl.
is weakened by an anomeric effect of the lone pair of
electrons on the oxygen atom and thus introduces an element
of fragility into the substrate. This complication could be
circumvented in a large measure by reverting to the lower-
boiling 1,2-dichloroethane as the reaction solvent. In this
unprecedented anomeric prenylation of a carbohydrate, the
phosphine oxide is tertiary and therefore the addition–
fragmentation occurs readily at 808C.
A preliminary reaction that involved a simple secondary
dithiocarbonate moiety gave rather disappointing results.
Thus, the attempted allylation of dithiocarbonate 36 gave
mostly reduced piperidine 37 and only a low yield of the
normal product 38. Hydrogen abstraction could take place
from the solvent or from the benzylic position of the
phosphine oxide reagent 30. Further studies are needed to
ascertain the source of the hydrogen atom.
These preliminary results represent a promising approach
to a generalized allylation process and highlight the impor-
tance of the substitution around the phosphorus center in
determining its leaving-group ability. From a synthetic stand-
Angew. Chem. Int. Ed. 2006, 45, 5002 –5006
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 5005

Communications
see: G. E. Keck, A. M. Tafesh, J. Org. Chem. 1989, 54, 5845 –
5846.
Lett. 2000, 41, 9815 – 9818; c) A. Liard, B. Quiclet-Sire, R. N.
Saicic, S. Z. Zard, Tetrahedron Lett. 1997, 38, 1759 – 1762.
[18] The results reported herein are taken from chapter 4 of PhD
thesis of G. Ouvry, which was defended on December 12th, 2002
(G. Ouvry, Synthꢀses d’OlØfines par Voie Radicalaire, Thꢁse,
Ecole Polytechnique, 2002); the use of the diphenylphosphinoyl
entity as a “new” radical leaving group was recently reported by
Malacria et al. (D. Leca, L. Fensterbank, E. Lacꢂte, M. Malacria,
Angew. Chem. 2004, 116, 4316 – 4318; Angew. Chem. Int. Ed.
2004, 43, 4220 – 4222). One of the senior authors (L. Fenster-
bank) was a principal examiner of the thesis of G. Ouvry, yet no
reference is made to the thesis in this paper, nor was there any
acknowledgement in the full paper, which appeared shortly
afterwards (D. Leca, K. Song, M. Albert, M. Grangeio Gon-
ꢃalves, L. Fensterbank, E. Lacꢂte, M. Malacria, Synthesis 2005,
1405 – 1420) despite a prior exchange of correspondence with the
authors on the matter.
[11] It is possible to combine an atom- or group-transfer radical
addition to allylsilanes with a fluoride-induced elimination; see:
a) N. A. Porter, G. Zhang, A. D. Reed, Tetrahedron Lett. 2000,
41, 5773 – 5777; b) L. Chabaud, Y. Landais, P. Renaud, Org. Lett.
2002, 4, 425; c) M. E. Briggs, S. Z. Zard, Synlett 2005, 334 – 336.
[12] For reviews on dithiocarbonate chemistry, see: a) B. Quiclet-
Sire, S. Z. Zard, Top. Curr. Chem. 2006, 264, 201 – 236; b) S. Z.
Zard in Radicals in Organic Synthesis, Vol. 1 (Eds.: P. Renaud,
M. P. Sibi), Wiley-VCH, Weinheim, 2001, pp. 90 – 108; c) B.
Quiclet-Sire, S. Z. Zard, Phosphorus Sulfur Silicon 1999, 153–
154, 137 – 154; d) B. Quiclet-Sire, S. Z. Zard, J. Chin. Chem. Soc.
1999, 46, 139 – 145; e) S. Z. Zard, Angew. Chem. 1997, 109, 724 –
737; Angew. Chem. Int. Ed. Engl. 1997, 36, 672 – 685.
[13] J. Pellon, J. Am. Chem. Soc. 1961, 83, 1915 – 1916.
[14] a) S. Kim, C. H. Cho, C. J. Lim, J. Am. Chem. Soc. 2003, 125,
9574 – 9575; b) B. Quiclet-Sire, S. Z. Zard, H. Zhang, J. Organo-
met. Chem. 2002, 643–644, 404 – 408; c) H. Zhang, PhD Thesis,
Universitꢀ Paris-Sud, Orsay, 2000.
[19] a) D. H. R. Barton, D. O. Jang, J. Jaszberenyi, J. Org. Chem.
1993, 58, 6838 – 6842; b) D. O. Jang, D. H. Cho, J. Kim, Synth.
Commun. 1998, 28, 3559 – 3565.
[15] D. L. J. Clive, S. Kang, J. Org. Chem. 2001, 66, 6083 – 6091.
[16] S. Bogen, M. Gulea, L. Fensterbank, M. Malacria, J. Org. Chem.
1999, 64, 4920 – 4925.
[20] J.-K. Erguden, E. Schaumann, Synthesis 1996, 707 – 710; J. Ukai,
Y. Ikeda, N. Ikeda, H. Yamamoto, Tetrahedron Lett. 1983, 24,
4029 – 4032.
[17] For examples of a-tetralone formation, see: a) A. Cordero
Vargas, B. Quiclet-Sire, S. Z. Zard, Org. Lett. 2003, 5, 3717 –
3719; b) N. Legrand, B. Quiclet-Sire, S. Z. Zard, Tetrahedron
[21] For leading references, see: a) M. P. Savage, S. Tripett, J. Chem.
Soc. C 1966, 1842 – 1844; b) S. Demay, K. Harms, P. Knochel,
Tetrahedron Lett. 1999, 40, 4981 – 4984.
5006 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 5002 –5006