Palladium-catalyzed allylic transposition of (allyloxy) iminodiazaphospholidines: A formal [3,3]-aza-phospha-oxa-cope sigmatropic rearrangement for the stereoselective synthesis of allylic amines

Article abstract of DOI:10.1021/ja054161k

The synthesis of N-protected allylic amines has been achieved utilizing a palladium(II)-catalyzed, [3,3]-rearrangement of (allyloxy) iminodiazaphospholidines. This [3,3]-aza-phospha-oxa-Cope sigmatropic rearrangement reaction is thermodynamically driven by a P=N to P=O interconversion and is an alternative to the Overman rearrangement. The overall process involves the nucleophilic displacement of an allylic alcohol onto a P(III) precursor, followed by a Staudinger reaction to generate the (allyloxy) iminodiazaphospholidine precursors. Pd(II)-catalyzed [3,3]-aza-phospha-oxa-Cope rearrangement then gives a phosphoramide, which is readily hydrolyzed under acidic conditions to yield allylic amine derivatives. Pd(II) catalysis is believed to occur in a fashion analogous to that of the rearrangement of allylic imidates. The scope of racemic, diastereoselective, and enantioselective variants of this rearrangement is described. The use of chiral diamine auxiliaries in diastereoselective rearrangements is reported. Rearrangement of chiral N,N′-dimethyl cyclohexanediamine derived diazaphospholidines gives rise to phosphoramides with moderate diastereoselectivities (up to 3.5:1 dr). The same major diastereomeric product in these rearrangements was prepared irrespective of the starting allylic alcohol geometry. An enantioselective variant of the reaction was demonstrated for the rearrangement of cis-(allyloxy) iminodiazaphospholidines with cobalt oxazoline palladacycle (COP-X) catalysts (5 mol %) in high yield and enantioselectivity (up to 96% ee).

Full text of DOI:10.1021/ja054161k

Published on Web 10/04/2005
Palladium-Catalyzed Allylic Transposition of (Allyloxy)
Iminodiazaphospholidines: A Formal
[3,3]-Aza-phospha-oxa-Cope Sigmatropic Rearrangement for
the Stereoselective Synthesis of Allylic Amines
Ernest E. Lee and Robert A. Batey*
Contribution from the Department of Chemistry, UniVersity of Toronto,
Toronto, Ontario, Canada
Received June 23, 2005; E-mail: rbatey@chem.utoronto.ca
Abstract: The synthesis of N-protected allylic amines has been achieved utilizing a palladium(II)-catalyzed,
[3,3]-rearrangement of (allyloxy) iminodiazaphospholidines. This [3,3]-aza-phospha-oxa-Cope sigmatropic
rearrangement reaction is thermodynamically driven by a PdN to PdO interconversion and is an alternative
to the Overman rearrangement. The overall process involves the nucleophilic displacement of an allylic
alcohol onto a P(III) precursor, followed by a Staudinger reaction to generate the (allyloxy) iminodiaza-
phospholidine precursors. Pd(II)-catalyzed [3,3]-aza-phospha-oxa-Cope rearrangement then gives a
phosphoramide, which is readily hydrolyzed under acidic conditions to yield allylic amine derivatives. Pd(II)
catalysis is believed to occur in a fashion analogous to that of the rearrangement of allylic imidates. The
scope of racemic, diastereoselective, and enantioselective variants of this rearrangement is described.
The use of chiral diamine auxiliaries in diastereoselective rearrangements is reported. Rearrangement of
chiral N,N′-dimethyl cyclohexanediamine derived diazaphospholidines gives rise to phosphoramides with
moderate diastereoselectivities (up to 3.5:1 dr). The same major diastereomeric product in these
rearrangements was prepared irrespective of the starting allylic alcohol geometry. An enantioselective variant
of the reaction was demonstrated for the rearrangement of cis-(allyloxy) iminodiazaphospholidines with
cobalt oxazoline palladacycle (COP-X) catalysts (5 mol %) in high yield and enantioselectivity (up to 96%
ee).
Introduction
Efficient methods for the formation of allylic amines are
useful for the preparation of many nitrogen-containing com-
pounds, such as amino acids, amino sugars, alkaloids, and other
complex natural products.1 One general approach to their
synthesis utilizes aza-analogues of the Cope [3,3]-sigmatropic
rearrangement, including aza-, aza-oxa-, aza-thia, and diaza-
Cope rearrangements.2 Undoubtedly the most important member
of this family is the aza-oxa-Cope rearrangement of allylic
imidates 1 into allylic amides 2, originally developed by
Overman (Figure 1).3 The aza-oxa-Cope or Overman rearrange-
ment is a widely utilized reaction, and on hydrolysis of the
resultant amide 2 provides a convenient route to the synthesis
of allylic amines. It has been utilized in a number of syntheses,4
with the rearrangement usually occurring with excellent regio-
Figure 1. Aza-oxa-Cope and aza-phospa-oxa-Cope rearrangements.
chemical control, geometrical-control, and transfer of stereo-
chemistry when starting with stereodefined allylic imidates.
We envisaged an analogous [3,3]-sigmatropic rearrangement
of 3 into 4, in which allylic transposition is driven by a P(V)d
N to P(V)dO interconversion (Figure 1). Our initial com-
munication outlined the conversion of iminodiazaphospholidines
3 into phosphoramides 4 (R4 ) NR2),5 under Pd(II)-catalyzed
conditions. This conversion was the first example of an aza-
phospha-oxa-Cope rearrangement. Almost simultaneous with our
initial study on the Pd(II)-catalyzed variant of this reaction,
Mapp and Chen reported a similar thermal rearrangement of
(1) Johannsen, M.; Jørgensen, K. A. Chem. ReV. 1998, 98, 1689-1708.
(2) (a) Ritter, K. In Houben-Weyl. StereoselectiVe Synthesis, E 21e; Helmchen,
G., Hoffmann, R. W., Mulzer, J., Schaumann, E., Eds.; Thieme: Stuttgart,
1996; pp 5677-5699. (b) Lutz, R. P. Chem. ReV. 1984, 84, 205-247.
(3) (a) Overman, L. E. J. Am. Chem. Soc. 1976, 98, 2901-2910. (b) Overman,
L. E. Acc. Chem. Res. 1980, 13, 218-224. (c) Overman, L. E. Angew.
Chem. 1984, 96, 565-573; Angew. Chem., Int. Ed. Engl. 1984, 23, 579-
586. (d) Schenck, T. G.; Bosnich, B. J. Am. Chem. Soc. 1985, 107, 2058-
2066. (e) Nishikawa, T.; Asai, M.; Ohyabu, N.; Isobe, M. J. Org. Chem.
1998, 63, 188-192. (f) Savage, I.; Thomas, E. J.; Wilson, P. D. J. Chem.
Soc., Perkin Trans. 1 1999, 3291-3303.
(4) Overman reports that over 180 publications utilize allylic imidate rear-
rangement for allylic amine synthesis (see ref 14c).
(5) Lee, E. E.; Batey, R. A. Angew. Chem. 2004, 116, 1901-1904; Angew.
Chem., Int. Ed. 2004, 43, 1865-1868.
9
10.1021/ja054161k CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 14887-14893
14887

A R T I C L E S
Scheme 1
Lee and Batey
Figure 2. Palladium-catalyzed cyclization-induced rearrangement of allylic
imidates.13,14
phosphorimidates 3 (R4 ) OR).6 Our investigations were
inspired by the knowledge that reactions in which products
containing P(V)dO bonds are formed are often thermodynami-
cally favorable. There are many examples of reactions employ-
ing the conversion of P(III) to P(V)dO species, such as the
Wittig and Mitsunobu reactions. Similarly, the [2,3]-sigmatropic
rearrangement of allyl phosphites is a valuable approach to the
synthesis of allylphosphonates.7 Furthermore, thermal rear-
rangements of phosphorimidates to phosphoramides in which
formal [1,3]-alkyl migration from oxygen to nitrogen concomi-
tant with P(V)dN to P(V)dO interconversion have also been
described.8 The thermodynamic driving force of the phospho-
lidine-phosphoramide interconversion of 3 into 4 can be
estimated by considering the model system of (allyl-O)-
(NH2)2PdNH, which on [3,3]-sigmatropic rearrangement is
converted into (allyl-NH)(NH2)2PdO.9 The energy change for
this transformation is estimated at -24.4 kcal mol-1, calculated
at the B3LYP/6-31G* level. By comparison, the thermodynamic
driving force for the rearrangement of 1 into 2, as for the
transformation of the imidate (allyl-O)CHdNH into the amide
phoramidite can then readily undergo a room-temperature
Staudinger reaction11 with electron-deficient azides to generate
a phospholidine 3a. Metal-catalyzed [3,3]-sigmatropic 3-aza-
2-phospha-1-oxa-Cope12 rearrangement of 3a then generates
phosphoramide 4a, which can be deprotected under mild, acidic
conditions to give the transposed allylic amine 7. In contrast to
allylic imidates, this system contains two variable substituents
on phosphorus that could be used to control the reactivity and
selectivity of the rearrangement. In addition, the use of azides
in the Staudinger reaction allows the introduction of a range of
different nitrogen substituents.
We were particularly interested in the development of a metal-
catalyzed variant of the aza-phospha-oxa-Cope rearrangement.
Metal catalysts including Pd(II) and Hg(II) have been demon-
strated by Overman to catalyze the [3,3]-sigmatropic rearrange-
ment of allylic imidates.3 Metal-catalyzed reaction of 1 is
believed to occur in a stepwise fashion, with Pd(II) first
activating the alkene to nucleophilic attack by the N-atom of
the imidate in a 6-endo-trig fashion, leading to a metal
intermediate 9 that then undergoes rapid breakdown to yield 2
(Figure 2). An analogous metal-catalyzed process for the
rearrangement of 3 into 4 would allow the use of lower reaction
temperatures and permit the use of more sensitive substrates.
Moreover, the use of a chiral metal catalyst would, in principle,
allow the development of an asymmetric variant of the reaction.
Indeed, at the outset of our investigations, there had been
increasing interest in the use of chiral Pd(II) catalysts for
enantioselective allylic imidate Overman rearrangements.13 Early
catalysts, although high yielding and highly selective, were
plagued by limitations of the substrates to N-aryl imidates. The
enantioselective rearrangement of substrates containing readily
deprotected N-functionality has only recently been accomplished
with various chiral cobalt oxazoline palladacycles (COP-X)
10, a culmination of several generations of axially chiral
(allyl-NH)CHdO, is estimated at -17.1 kcal mol-1
.
We now report a full study on the development of a metal-
catalyzed aza-phospha-oxa-Cope rearrangement of 3 into 4,
detailing the scope of the reaction, auxiliary-based diastereo-
selective rearrangement, and the use of chiral catalysts to
accomplish enantioselective rearrangement. The overall strategy
employs allylic alcohols 5 as substrates, compounds that are
commonly used as synthetic precursors, and for which there
are numerous methods developed for their synthesis.10 Nucleo-
philic substitution of 5 on an appropriate P(III) reagent then
gives phosphoramidite 6 (Scheme 1). The electron-rich phos-
(6) (a) Chen, B.; Mapp, A. K. J. Am. Chem. Soc. 2004, 126, 5364-5365. (b)
Chen, B.; Mapp, A. K. J. Am. Chem. Soc. 2005, 127, 6712-6718.
(7) Janecki, T.; Bodalski, B. Synthesis 1990, 799-801.
(8) (a) Challis, B. C.; Frenkel, A. D. J. Chem. Soc., Chem. Commun. 1972,
303-308. (b) Challis, B. C.; Challis, J. A.; Iley, J. N. J. Chem. Soc., Perkin
Trans. 2 1978, 813-818. (c) Mastryukova, T. A.; Mashchenko, N. V.;
Odinets, I. L.; Petrovskii, P. V.; Kabachnik, M. I. Russ. J. Gen. Chem.
1988, 58, 1756-1761. (d) Cabrita, E. J.; Afonso, C. A. M.; Gil de Oliveira
Santos, A. Chem.-Eur. J. 2001, 7, 1455-1467.
(9) Geometry optimizations, single-point energies, and vibrational analysis were
calculated at the B3LYP/6-31G* level. Calculations were performed using
Spartan’04 Version 1.0, Wave function Inc., Irvine, CA.
(11) Gololobov, Y. G.; Kasukhin, L. F. Tetrahedron 1992, 48, 1353-1406.
(12) Classification based on the allylic imidate rearrangement described in:
Vo¨gtle, F.; Goldschmitt, E. Chem. Ber. 1976, 109, 1-40.
(10) There are numerous synthetic methods for the formation of enantioenriched
allylic alcohols. Resolution of allylic alcohols: (a) Drauz, K.; Waldmann,
H. Enzyme Catalysis in Organic Synthesis: A ComprehensiVe Handbook,
2nd ed.; Wiley-VCH: Weinheim, Germany, 2002; Vols. I-III. (b) Pa`mies,
O.; Ba¨ckvall, J.-E. Chem. ReV. 2003, 103, 3247-3261. (c) Carlier, P. R.;
Mungall, W. S.; Schro¨der, G.; Sharpless, K. B. J. Am. Chem. Soc. 1988,
110, 2978-2979. (d) Choi, J. H.; Choi, Y. K.; Park, E. S.; Kim, E. J.;
Kim, M.-J. J. Org. Chem. 2004, 69, 1972-1977. Asymmetric reduction
of enones and ynones: (e) Noyori, R. Angew. Chem., Int. Ed. 2002, 41,
2008-2022. (f) Itsuno, S. Org. React. 1998, 395-576. Synthesis of
enantioenriched propargylic alcohols and subsequent reduction: (g) Anand,
N. K.; Carreira, E. M. J. Am. Chem. Soc. 2001, 123, 9687-9688. (h)
McKew, J. C.; Kurth, M. J. Org. Proc. Proc. Int. 1993, 25, 125-130.
Asymmetric addition to unsaturated aldehydes: (i) Soai, K.; Niwa, S. Chem.
ReV. 1992, 92, 833-856.
(13) (a) Calter, M.; Hollis, T. K.; Overman, L. E.; Ziller, J.; Zipp, G. G. J. Org.
Chem. 1997, 62, 1449-1456. (b) Hollis, T. K.; Overman, L. E. Tetrahedron
Lett. 1997, 38, 8837-8840. (c) Uozumi, Y.; Kazuhiko, K.; Hayashi, T.
Tetrahedron: Asymmetry 1998, 9, 1065-1072. (d) Cohen, F.; Overman,
L. E. Tetrahedron: Asymmetry 1998, 9, 3213-3222. (e) Jiang, Y.;
Longmire, J. M.; Zhang, X. Tetrahedron Lett. 1999, 40, 1449-1450. (f)
Donde, T.; Overman, L. E. J. Am. Chem. Soc. 1999, 121, 2933-2934. (g)
Leung, P.-H.; Ng, K.-H.; Li, Y.; White, A. J. P.; Williams, D. J. Chem.
Commun. 1999, 2435-2436. (h) Kang, J.; Yew, K. H.; Kim, T. H.; Choi,
D. H. Tetrahedron Lett. 2002, 43, 9509-9512. (i) Kang, J.; Kim, T. H.;
Yew, K. H.; Lee, W. K. Tetrahedron: Asymmetry 2003, 14, 415-418. (j)
Anderson, C. E.; Donde, Y.; Douglas, C. J.; Overman, L. E. J. Org. Chem.
2005, 70, 648-657.
9
14888 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

Aza-phospha-oxa-Cope Rearrangement
A R T I C L E S
Table 1. Preparation of (Allyloxy) Iminodiazaphospholidines
palladacycle catalysts described by Overman and co-workers.14
This COP-X class of palladacycles has also proven to be the
most effective catalyst for the (allyloxy) iminodiazaphospho-
lidine rearrangement, with high enantioselectivities (up to 96%)
and yields obtained.15
Results and Discussion
The feasibility of the (allyloxy) iminodiazaphospholidine
rearrangement was demonstrated using 11 and 12 as substrates.
These compounds were cleanly prepared in a one-pot process
by the sequential treatment of allylic alcohols with phospholidine
13,16 as described by Alexakis and co-workers,17 followed by
reaction with tosyl azide and diphenylphosphoryl azide (DP-
PA),18 respectively (Table 1).19 The reaction was monitored by
31P NMR to ensure complete conversion of the intermediate
phosphoramidite 14 (δ ≈ 130 for 14, whereas δ ≈ 24 for 11
and δ ≈ 24 and 10 ppm for 12) (Figure 3). The iminodiaza-
phospholidines were then purified by silica gel chromatography,
using Et3N as an additive to prevent acid promoted decomposi-
tion.
Alternatively, the substrates could be prepared by treatment
of PCl3 sequentially with the diamine and allylic alcohol in the
presence of Et3N to generate the phosphorus(III) intermediate
15 (Scheme 2). Reaction of 15 with an azide then gives the
desired substrates 11 and 12. Although this route gives
comparable yields, the procedure is less convenient because of
the need for lower temperatures and filtration of the Et3N‚HCl
salts generated.
Initial attempts at the thermal rearrangement of 11 were
unsatisfactory, leading to products arising from both the desired
[3,3] and the formal [1,3]-sigmatropic rearrangement pathways.
For example, thermal rearrangement of diazaphospholidine 11c
at 130 °C in xylenes led to the [3,3]-product 16c and [1,3]
product 17c in a 3.5:1 ratio (Scheme 3). Furthermore, thermal
rearrangement of 11m only yielded the [1,3]-product 17m,
whereas reaction of 11s only yielded the [3,3]-product 16s. The
thermal rearrangements of 11m and 11s give the thermodynami-
cally more stable allylic phosphoramides, suggesting that
ionization of the diazaphospholidine substrates 11 and subse-
quent recombination is competitive under thermal conditions,
particularly for those substrates that lead to more stable allylic
carbocation intermediates.20 Interestingly, the thermal aza-
phospha-oxa-Cope rearrangements of phosphorimidates (in
refluxing xylenes), reported by Mapp and Chen, give the [3,3]-
products more cleanly.
a Isolated yields, 0.6 mmol scale. b These compounds could not be
purified by silica gel column chromatography and were used crude in
subsequent transformations. c Reaction conducted in benzene-d6. d Substrates
not prepared.
(14) (a) Overman, L. E.; Owen, C. E.; Pavan, M. M.; Richards, C. J. Org. Lett.
2003, 5, 1809-1812. (b) Anderson, C. E.; Overman, L. E. J. Am. Chem.
Soc. 2003, 125, 12412-12413. (c) Kirsh, S. F.; Overman, L. E.; Watson,
M. P. J. Org. Chem. 2004, 69, 8101-8104. (d) Kirsh, S. F.; Overman, L.
E. J. Org. Chem. 2005, 70, 2859-2861.
(15) Lee, E. E.; Batey, R. A. 229th ACS National Meeting, San Diego, 2005,
abstract 637.
Figure 3. 31P NMR of intermediates involved in the stepwise conversion
of allylic alcohols into allylic amines via an aza-phospha-oxa-Cope
rearrangement.
(16) Prepared as described in: Hanessian, S.; Bennani, Y. L.; Leblanc, Y.
Heterocycles 1993, 35, 1411-1424.
(17) Alexakis, A.; Mutti, S.; Mangeney, P. J. Org. Chem. 1992, 57, 1224-
1237.
(18) Although we have experienced no problems with either of these azides,
appropriate safety measures should be taken. For discussion on the hazards
associated with azides, see: Prudent Practices in the Laboratory: Handling
and Disposal of Chemicals; National Academy Press: Washington, DC,
1995.
(19) (a) Bellan, J.; Sanchez, M.; Marre-Mazie`res, M. R.; Beltran, A. M. Bull.
Chim. Soc. Fr. 1985, 3, 491-495. (b) Marre, M. R.; Sanchez, M.; Brazier,
J. F.; Wolf, R.; Bellan, J. Can. J. Chem. 1982, 60, 456-468.
(20) Mapp and Chen (ref 6a) have also reported mixtures of [3,3]-rearranged
and formal [1,3]-rearranged products being formed under thermal conditions
for phenyl substituted phosphorimidates.
The lack of selectivity observed in the thermal rearrangements
of the diazaphospholidines 11 can be overcome through the use
of Pd(II) catalysis. A variety of Pd(II) catalysts were screened
for the rearrangement of 11c into 16c, but only PdCl2(MeCN)2
was found to be an active catalyst.21 In the presence of 5 mol
(21) The following catalysts resulted in no reaction with complete recovery of
11c: PdCl2, PdCl2(PPh3)2, PdCl2(PCy3)2, Pd2Cl2(allyl)2, PdCl2(COD).
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14889

A R T I C L E S
Scheme 2
Lee and Batey
Table 3. Pd-Catalyzed [3,3]-Sigmatropic Rearrangement of More
Highly Substituted (Allyloxy) Iminodiazaphospholidines and Their
Subsequent Hydrolysis
Scheme 3
Table 2. Pd-Catalyzed [3,3]-Sigmatropic Rearrangement of
(Allyloxy) Iminodiazaphospholidines and Their Subsequent
Hydrolysis
a Isolated yields, 0.6 mmol scale. b Conducted in the presence of 4 Å
molecular sieves. c Conducted at 45 °C in toluene. d Only starting material
was observed by 31P NMR. e 1 M HCl in MeOH.
γ-positions of the allylic group. Notably, the reaction worked
well for substrates substituted at the â position (11i and 12i),
as previous reports of metal-catalyzed allylic imidate rearrange-
ments for these substrates have had mixed success.23 Substrates
derived from secondary alcohols, 11j, 12j, 11l, and 12l, all
rearranged in good yields to afford only their respective
E-isomers 16j, 18j, 16l, and 18l. Reactions of the 2-cyclohexenyl
substrates were far more sluggish, with 11k requiring heating
at 45 °C and 12k yielding only trace amounts of products after
48 h at 80 °C. A similar trend was observed with substrates
11m-o and 12m-o. It is apparent that the rearrangement works
very well when unhindered substrates 11a-h (Table 2) are used,
while the more sterically demanding substrates 11m-o (Table
3) react far more slowly. The steric limitations of this reaction
are further emphasized by the lack of reactivity in the γ,γ-
disubstituted substrates 11p and 12p and by the trend of lower
reactivity for the bulky DPPA derived substrates.
a Isolated yields, 0.6 mmol scale. b Conducted in the presence of 4 Å
molecular sieves. c 1 M HCl in MeOH.
% PdCl2(MeCN)2, the rearrangement of both 11 and 12
proceeded smoothly at room temperature yielding only the [3,3]-
products 16 and 18, respectively (Table 2). In the reactions of
the DPPA derived substrates 12, the addition of 4 Å molecular
sieves was required to ensure complete conversion into 18. The
rearrangements were conveniently monitored using 31P NMR
(δ ≈ 20 for 16 and δ ≈ 20 and -4 ppm for 18) (Figure 3). The
phosphoramides 16 and 18 were deprotected under acidic
conditions, yielding allylic tosylamides 19 and the HCl salts of
allylic amines 20, respectively.22
Transposition of the enantioenriched E-substrates 11q and
12q produced only the E-phosphoramides 16q and 18q with
clean transfer of chirality. The Z-substrates 11r and 12r
A variety of substitution patterns are tolerated on the allylic
substrates 11 and 12, including substitution at the R-, â-, and
(23) Depending on the allylic imidate used, either prolonged reaction time was
required or no reaction was observed: Metz, P.; Mues, C.; Schoop, A.
Tetrahedron 1992, 48, 1071-1080 and references cited.
(22) Mizrahi, V.; Modro, T. A. J. Org. Chem. 1983, 48, 3030-3037.
9
14890 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

Aza-phospha-oxa-Cope Rearrangement
A R T I C L E S
Table 4. Preparation of (Allyloxy) Iminodiazaphospholidines
Incorporating a N,N′-Dialkyl Cyclohexanediamine Auxiliary
R1
R2
geometry
product
yielda (%)
Me
Me
Me
Me
iPr
iPr
Bn
Bn
Ts
trans
trans
cis
24a
24b
24c
24d
24e
24f
85
73
75
81
80
78
85
82
Figure 4. Proposed mechanism for palladium-catalyzed aza-phospha-oxa-
Cope rearrangement.
OP(OPh)2
Ts
OP(OPh)2
Ts
OP(OPh)2
Ts
OP(OPh)2
cis
rearranged to the E-products 16r and 18r, albeit, with slightly
lower enantioenrichment.24 The [3,3]-sigmatropic rearrangement
presumably proceeds via intramolecular attack of the nitrogen
lone pair on a palladium coordinated olefin 21, followed by
rearrangement of the resulting phosphonium intermediate 22
(Figure 4). The intermediacy of a π-complex 21 (R1 ) Me and
R2 ) Et) and phosphonium ion 22 is analogous to the
mechanism proposed for the rearrangement of allylic imidates
(Figure 2).3a,c,d The absolute stereochemistry and olefin geometry
of the products in both cases are consistent with this mechanism.
Comparison of the ambient temperature Pd(II)-catalyzed reac-
tions with the thermal rearrangement of 11 demonstrates the
advantages of metal catalysis to achieve efficient [3,3]-sigma-
tropic rearrangement with these diazaphospholidine systems.
trans
trans
trans
trans
24g
24h
a Isolated yields, 0.6 mmol scale.
Table 5. Diastereoselective Rearrangements of (Allyloxy)
Iminodiazaphospholidines Incorporating an N,N′-Dialkyl
Cyclohexanediamine Auxiliary
Diastereoselective Rearrangements. Inherent in the design
and synthesis of the (allyloxy) diazaphospholidine structure was
the potential to readily incorporate chiral auxiliaries in place of
the diamine subunit. To prevent the potential complication of
diastereomeric mixtures at the phosphorus center, our investiga-
tion was restricted to C2 symmetric auxiliaries. The synthesis
of tartrate derived substrates was unsuccessful as the azide
Staudinger reaction required heating, resulting in a complex
mixture of products. Utilizing N,N′-dialkyl cyclohexanediamine
auxiliaries was more successful as the preparation and purifica-
tion of phospholidine 24 could be accomplished in a fashion
similar to that of the N,N′-dimethylethylenediamine derived
substrates 11 and 12. The use of chiral cyclohexanediamines is
also convenient as the resolution25 and substitution of trans
cyclohexanediamine26 has been well developed. Preparation of
the phospholidine 23 was accomplished by heating the ap-
propriate diamine with hexamethylphosphorus triamide17 at 100
°C. The one-pot reaction of 23 with allylic alcohols and then
azides yielded the desired chiral (allyloxy) diazaphospholidines
24 in good yields (Table 4).
yield (%)
precursor
R1
R2
geometry
drb
25
24a
24b
24c
24d
24e
24f
Me
Me
Me
Me
iPr
iPr
Bn
Bn
Ts
trans
trans
cis
3.5:1
1.7:1
3.5:1
1.5:1
2.5:1
1.5:1
1:1
83c
27d
81c
35d
80
77
73
85
OP(OPh)2
Ts
OP(OPh)2
Ts
OP(OPh)2
Ts
OP(OPh)2
cis
trans
trans
trans
trans
24g
24h
1:1
a Isolated yields, 0.6 mmol scale. b Determined by 1H and 31P NMR
integration. c Hydrolysis formed 19c (56% ee in favor of the S enantiomer
as determined by chiral HPLC; see Experimental Section in Supporting
Information for absolute configuration assignment). d Formal [1,3]-rear-
rangement was the major side reaction.
and 1.5:1 dr, respectively). Deprotection of the chiral auxiliary
with 1 M HCl yielded the allylic tosylamide 19c in 56% ee in
favor of the S enantiomer. A decrease in diastereoselectivity
was observed for the N-isopropyl substrates substrates 24e and
24f (2.5:1 and 1.5:1 dr, respectively). Reaction of the N-benzyl
substituted substrates 24g and 24h afforded the products as a
1:1 mixture of diastereoisomers.
We believe that the R,R-cyclohexanediamine adopts a con-
formation 26 similar to that described by Hanessian and co-
workers in models explaining their asymmetric conjugate
additions and cyclopropanations of crotyl and allyl phosphora-
mides.27 Palladium(II) coordination to the two faces of the
olefin3d and subsequent cyclization would produce the chairlike,
cationic intermediates 27 or 28 (Scheme 4). The chairlike
intermediate 27 leading to the minor diastereomer in this model
is disfavored because of the steric interaction between the
Stirring the trans and cis substrates 24a and 24c at room
temperature in the presence of 5 mol % PdCl2(MeCN)2 led to
clean conversion to the desired products 25a in 83% and 81%
yield, respectively (Table 5). Interestingly, the same major
diastereomer was formed in a ratio of 3.5:1 for both substrates.
The analogous DPPA derived substrates 24b and 24d afforded
the products in decreased yields and diastereoselectivities (1.7:1
(24) 11q and 12q (95% ee) were rearranged to give 16q and 18q with 91% ee,
whereas 11r and 12r (95% ee) were rearranged to give 16r and 18r with
70% ee (determined by chiral HPLC).
(25) Resolution on 160 g scale: Larrow, J. F.; Jacobsen, E. N. J. Org. Chem.
1994, 59, 1939-1942.
(26) (a) Bennani, Y. L.; Hanessian, S. Chem. ReV. 1997, 97, 3161-3195. (b)
Denmark, S. E.; Su, X.; Nishigaichi, Y.; Coe, D. M.; Wong, K.-T.; Winter,
S. B. D.; Choi, J. Y. J. Org. Chem. 1999, 64, 1958-1967.
(27) (a) Hanessian, S.; Gomstyan, A. J. Org. Chem. 1993, 58, 5024-5032. (b)
Hanessian, S.; Andreotti, D.; Gomtsyan, A. J. Am. Chem. Soc. 1995, 117,
10393-10394.
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14891

A R T I C L E S
Scheme 4
Lee and Batey
Table 7. Counterion Effect on COP-Cl-Catalyzed
Rearrangements
16ca
17ca
29a
11ca
entry
T
AgY
(%)/(ee)
(%)
(%)
(%)
1
2
3
4
5
6
7
90
90
90
90
50
50
90
Ag3PO4
AgF
9
16
12
5
9
28
40
7
0
7
30
34
12
86
68
30
10
50
65
64
AgSbF6
AgOTf
AgOTs
AgNO3
AgTFA
16
Table 6. Enantioselective Rearrangement Using COP-X
Catalysts
31/(81)
31/(82)
35/(79)
4
tr
a Ratio determined by 31P NMR integration.
entries 3 and 5). Interestingly, reaction of 11c in the presence
of powdered 4 Å molecular sieves and COP-Cl at room
temperature led to the formation of the formal [1,3]-product
17c in 13% yield. Under the same conditions at 50 °C in toluene,
78% conversion to 17c was observed. Heating the reaction of
11c with COP-Cl at 100 °C in toluene for 36 h resulted in low
conversion (33%) to the desired product 16c in 70% ee. Under
the same conditions, the deleterious effect of 4 Å molecular
sieves is again clearly evident, as the conversion to product 16c
is lowered to 24% and the conversion to the formal [1,3]-product
17c is increased to 61%. Heating the reaction at 120 °C, or at
140 °C under microwave irradiation, led to a decrease in the
observed enantioselectivity of 16c, presumably as a result of
competing thermal rearrangement, as well as ionization of the
substrate to the byproduct 29.
As the nature of the catalyst counterion has been demonstrated
to greatly affect the reactivity of palladium catalysts used in
the asymmetric allylic imidate rearrangements, we modified the
COP-Cl catalyst by addition of a variety of silver(I) salts.
Interestingly, the counterion of the silver salt plays a significant
role in the formation of the two undesired products 17c and
29. The addition of Ag3PO4 or AgF to the reaction was the
most detrimental as only small quantities of any products were
observed (Table 7, entries 1 and 2). The addition of AgSbF6
and AgOTf increased the conversion of starting material mostly
to the undesired formal [1,3]-product 17c and ionized product
29 (Table 6, entries 3 and 4). Greater success was realized with
the use of AgOTs and AgNO3 as additives. At 50 °C in both
cases, minimal side products were formed, and the desired
rearrangements were achieved in 31% yield with 81% and 82%
ee, respectively. The best result was obtained with AgTFA at
90 °C, giving the desired product 16c in 35% yield with 79%
ee, and with only a trace amount of the formal [1,3]-product
17c observed.
16ca
17ca
11ca
entry
X
solvent
T
(%)/(ee)
(%)
(%)
1
2
3
4
5
6
7
10
11
12
OAc
OAc
Cl
CH2Cl2
Tol
CH2Cl2
CH2Cl2
Tol
Tol
Tol
Tol
Tol
room temp
90
room temp
room temp
50
100
99
99
87
73
22
49
15
25
62
trace
trace
Cl/4 Å MS
13
Cl
27
Cl/4 Å MS
50
100
100
78
7b
61
8c
8
Cl
33/(70)
24
47/(69)
30/(57)
Cl/4 Å MS
Cl
Cl
120
Tol
140d
a Ratio determined by 31P NMR integration. b 11% of the ionized product
29 was observed. c 20% of the ionized product 29 was observed. d Heated
for 1.5 h in a microwave reactor.
nitrogen substituent R (tosyl or phosphoryl) and the pseudo-
axial methyl substituted nitrogen of the chiral diamine. The cis
substrate presumably reacts through a boatlike intermediate to
alleviate 1,3-diaxial interactions in the cyclic intermediate, thus
giving rise to the same diastereomer.
Enantioselective Rearrangements. Initial attempts at the
enantioselective rearrangement of substrates 11c and 11d proved
disappointing, with most palladium(II) catalyst/ligand combina-
tions resulting in no reaction. Not surprisingly, as reported for
the allylic imidate rearrangement,13a phosphine-containing ligands
such as (R)-BINAP, or (R,S)-Josiphos resulted in no reaction,
and only recovery of starting material. Early generation ligands
successful in allylic imidate rearrangements such as the oxazo-
lidinyl-phosphine, bis-oxazoline, or diamine-based ligands also
showed negligible reactivity. Most surprisingly was the lack of
reactivity when utilizing the chiral cobalt oxazoline palladacycles
(COP-X) successfully utilized in the enantioselective rear-
rangement of allylic imidates.14 Reaction of substrate 11c in
the presence of 5 mol % COP-OAc in CH2Cl2 at room
temperature or at 90 °C in toluene showed only starting material
when monitored by 31P NMR (Table 6, entries 1 and 2). For
reaction of 11c using the chloride analogue COP-Cl, no
reaction was observed at room temperature, while a 27%
conversion to product 16c could be effected at 50 °C (Table 6,
Increasing the substrate concentration to 0.8 M and stirring
at 70 °C for 40 h increased the yield of 16c to 60% and the
enantiomeric excess to 82% (Table 8). At a concentration of
0.8 M, lowering the temperature to 50 °C and room temperature
resulted in an increase in enantioselectivity to 84% and 86%
ee, respectively. Unfortunately, the increase in selectivity
achieved at lower temperatures was accompanied by a drastic
decrease in the yield of 16c to 43% and 16% respectively. To
9
14892 J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005

Aza-phospha-oxa-Cope Rearrangement
A R T I C L E S
Table 8. Enantioselective Rearrangements with COP-Cl and
faster than the trans substrate 11c at 45 °C, as nearly complete
conversion of 11d is accomplished in 10 h. In contrast, the trans
substrate 11c reaches only 45% conversion in the same amount
of time. The difference in initial reaction rates of the two
substrates 11d and 11c was estimated to be 14 times faster for
the reaction of the cis-substrates 11d.
AgTFA
T
[11]
t
yielda
(%)
eeb,c
(%)
sub
R
(
°
C)
(M)
(h)
11c
11c
11c
11d
11a
11b
11c
11d
11d
11e
11f
trans-Et
trans-Et
trans-Et
cis-Et
trans-nPr
cis-nPr
trans-Et
cis-Et
cis-Et
(E)-CH2OTBS
(Z)-CH2OTBS
trans-Me
cis-Me
70
50
rt
50
45
45
45
45
rt
45
45
45
45
0.8
0.8
0.8
0.8
2.0
0.8
2.0
0.8
0.8
2.0
2.0
2.0
2.0
40
40
40
40
60
40
60
10
40
60
60
60
40
16c
60
43
16
97
55
90
60d
92d
88
44
82
45
86
82/S
84/S
86/S
92/R
86/S
96/R
84/S
93/R
94/R
86/S
92/R
86/S
91/R
16c
16c
16c
16a
16a
16c
16c
16c
16e
16e
16g
16g
Conclusion
In conclusion, a novel palladium(II)-catalyzed rearrangement
of (allyloxy) iminodiazaphospholidines has been developed for
the synthesis of allylic amines and tosylamides. The overall
process involves displacement on phospholidine 13 with an
allylic alcohol to give a phosphoramidite followed by Staudinger
reaction and Pd(II)-catalyzed aza-phospha-oxa-Cope rearrange-
ment. Diastereoselective rearrangements of 24 were carried out
in good yields with diastereoselectivities as high as 3.5:1.
Interestingly, in these rearrangements, the same major diaste-
reomer was formed from both the trans and the cis starting
substrates. The first enantioselective rearrangements of imino-
diazaphospholidines were also achieved by the use of the cobalt
oxazoline palladacycles (COP-X) class of catalysts. There is
a very dramatic dependence on the catalyst counterion and olefin
geometry for the COP-Cl catalyst system. With reasonably low
catalyst loading at 45 °C, cis substrates of 11 underwent
rearrangement in high yields (up to 97%) and enantioselectivities
(up to 96% ee). These results demonstrate the synthetic utility
of the COP-X family of catalysts for enantioselective [3,3]-
sigmatropic rearrangements other than the Overman rearrange-
ment.
11g
11h
a Isolated yields. b Determined by deprotection to 16 and chiral HPLC
analysis. c Refer to Supporting Information for absolute configuration
assignment. d Similar results are obtained when the reaction is carried out
in CH2Cl2 in a sealed tube.
our surprise, a significantly improved result was obtained for
the cis substrate 11d. The cis substrate 11d was cleanly
rearranged in 97% yield and 92% ee, when stirred for 40 h at
50 °C. The improved reactivity of cis substrates over trans
substrates was observed for a series of alkyl substituents on
the olefin. The highest enantioselectivity of 96% was observed
for reaction of cis-nPr substrate 11b, formed in 90% yield. The
cis-CH2OTBS 11f and cis-Me 11h substrates were also obtained
in good yields (82% and 86%) and with high enantioselectivities
of 92% and 91%, respectively.
In comparison to the corresponding cis analogues, reaction
of the trans-substrates (11a, 11c, 11e, 11f) occurred in slightly
lower enantioselectivities (82-86% ee), but in significantly
lower yields, even at an increased reaction concentration of 2.0
M. Intriguingly, Overman has reported an opposite trend for
the COP-Cl promoted rearrangements of cis-allylic imidates,
which rearrange in lower yields and enantioselectivities than
their trans analogues.28 In all cases, only unreacted starting
material 11 was observed as monitored by 31P NMR of the crude
reaction mixtures. When the cis and trans substrates 11d and
11c were periodically monitored by 31P NMR over a 100 h span,
it was clear that the rate of reaction of the cis substrate 11d is
Acknowledgment. The Natural Science and Engineering
Research Council (NSERC) of Canada funded this research.
E.E.L. thanks NSERC for funding in the form of a PGS
scholarship. R.A.B. gratefully acknowledges receipt of a
Premier’s Research Excellence Award. We also thank T.
Housmann for contributing to preliminary work on the diaste-
reoselective rearrangements.
Supporting Information Available: Experimental procedures
for substrate preparation, substrate rearrangement, characteriza-
tion, spectra (1H, 13C NMR), and HPLC traces used to determine
enantiopurity for all new compounds. This material is available
free of charge via the Internet at http://pubs.acs.org.
(28) For rearrangement of the imidate derived from trans-2-hexenol and
trichloroacetonitrile, 99% yield and 95% ee was obtained. For the cis-2-
hexenol derived imidate, 17% yield and 71% ee was obtained (see ref 14b).
JA054161K
9
J. AM. CHEM. SOC. VOL. 127, NO. 42, 2005 14893