New insights into the mechanism of palladium-catalyzed allylic amination

Article abstract of DOI:10.1021/ja055288c

A comparative investigation into palladium-catalyzed allylic amination of unsubstituted aziridines and secondary amines has been carried out. The use of NH aziridines as nucleophiles favors formation of valuable branched products in the case of aliphatic allyl acetates. The regioselectivity of this reaction is opposite to that observed when other amines are used as nucleophiles. Our study provides evidence for the palladium-catalyzed isomerization of the branched (kinetic) product formed with common secondary amines into the thermodynamic (linear) product. In contrast, the branched allyl products obtained from unsubstituted aziridines do not undergo the isomerization process. Crossover experiments indicate that the isomerization of branched allylamines is bimolecular and is catalyzed by Pd⁰. The reaction has significant solvent effect, giving the highest branched-to-linear ratios in THF. This finding can be explained by invoking the intermediacy of σ-complexes, which is consistent with NMR data. The apparent stability of branched allyl aziridines towards palladium-catalyzed isomerization is attributed to a combination of factors that stem from a higher degree of s-character of the aziridine nitrogen compared to other amines. The reaction allows for regio- and enantioselective incorporation of aziridine rings into appropriately functionalized building blocks. The resulting methodology addresses an important issue of forming quaternary carbon centers next to nitrogen. The new insights into the mechanism of palladium-catalyzed allylic amination obtained in this study should facilitate synthesis of complex heterocycles, design of new ligands to control branched-to-linear ratio, as well as absolute stereochemistry of allylamines.

Full text of DOI:10.1021/ja055288c

Published on Web 11/15/2005
New Insights into the Mechanism of Palladium-Catalyzed
Allylic Amination
Iain D. G. Watson and Andrei K. Yudin*
Contribution from the DaVenport Research Laboratories, Department of Chemistry,
The UniVersity of Toronto, 80 St. George Street, Toronto, Canada, M5S 3H6
Received August 12, 2005; E-mail: ayudin@chem.utoronto.ca
Abstract: A comparative investigation into palladium-catalyzed allylic amination of unsubstituted aziridines
and secondary amines has been carried out. The use of NH aziridines as nucleophiles favors formation of
valuable branched products in the case of aliphatic allyl acetates. The regioselectivity of this reaction is
opposite to that observed when other amines are used as nucleophiles. Our study provides evidence for
the palladium-catalyzed isomerization of the branched (kinetic) product formed with common secondary
amines into the thermodynamic (linear) product. In contrast, the branched allyl products obtained from
unsubstituted aziridines do not undergo the isomerization process. Crossover experiments indicate that
the isomerization of branched allylamines is bimolecular and is catalyzed by Pd0. The reaction has significant
solvent effect, giving the highest branched-to-linear ratios in THF. This finding can be explained by invoking
the intermediacy of σ-complexes, which is consistent with NMR data. The apparent stability of branched
allyl aziridines towards palladium-catalyzed isomerization is attributed to a combination of factors that stem
from a higher degree of s-character of the aziridine nitrogen compared to other amines. The reaction allows
for regio- and enantioselective incorporation of aziridine rings into appropriately functionalized building blocks.
The resulting methodology addresses an important issue of forming quaternary carbon centers next to
nitrogen. The new insights into the mechanism of palladium-catalyzed allylic amination obtained in this
study should facilitate synthesis of complex heterocycles, design of new ligands to control branched-to-
linear ratio, as well as absolute stereochemistry of allylamines.
compounds.4,5 Allylamines are also valuable synthetic interme-
diates for the preparation of R- and â-amino acids,6a,b
Introduction
Installation of an aziridine functional group into organic
molecules presents significant synthetic challenges. The aziridine
ring in its simplest NH form does not benefit from a straight-
forward, one-step protocol similar to the abundant examples of
oxygen atom transfer in epoxidation chemistry. Enantiomerically
enriched aziridines have been available via transition metal
catalysis since the early 1990s, but only in their N-protected
forms where the substituent on nitrogen is typically difficult to
remove. Conventional methods of making aziridines,1,2 including
reactions that utilize starting materials from the chiral pool, pose
functional group compatibility issues. Therefore, modification
of aziridine-containing starting materials can be viewed as a
viable route for making more complex aziridine-containing
molecules, especially if the ring installation is desired at a later
stage of synthesis. At present, there are only a few reactions
that fall into this category.3 As a possible implementation of
this strategy, we opted to explore routes from unsubstituted
aziridines to their allylated derivatives as a potentially rich
source of precursors to a wide range of allylamines. The
allylamine functionality is found in many biologically active
alkaloids,6c-g and aza-carbohydrate derivatives.6h,i The ally-
lamine group can be introduced by direct allylic amination of
olefins or by nucleophilic substitution at the allylic position.7
However, commonly used protocols that employ allyl alcohol
starting materials, such as Mitsunobu chemistry, cannot be used
for making allyl aziridines based on pKa considerations,8
whereas another commonly used route that utilizes epoxide ring
opening/ring closure sequence is not divergent. In turn, allylation
of unsubstituted aziridines using allyl bromides is complicated
by overallylation and concomitant aziridine ring opening.
Reductive amination, a typical solution to this problem in the
(4) Trost, B. M.; Crawley, M. L. Chem. ReV. 2003, 103, 2921-2943.
(5) Trost, B. M. Chem. Pharm. Bull. 2002, 50, 1-14.
(6) (a) Burgess, K.; Liu, L. T.; Pal, B. J. Org. Chem. 1993, 58, 4758-4763.
(b) Bower, J. F.; Jumnah, R.; Williams, A. C.; Williams, J. M. J. J. Chem.
Soc., Perkin Trans. 1 1997, 1411-1420. (c) Ichikawa, Y.; Ito, T.;
Nishiyama, T.; Isobe, M. Synlett 2003, 1034-1036. (d) Ito, K.; Akashi,
S.; Saito, B.; Katsuki, T. Synlett 2003, 1809-1812. (e) Paquette, L. A.;
Leit, S. M. J. Am. Chem. Soc. 1999, 121, 8126-8127. (f) Magnus, P.;
Lacour, J.; Coldham, I.; Mugrage, B.; Bauta, W. B. Tetrahedron 1995, 51,
11087-11110. (g) Johnson, T. A.; Curtis, M. D.; Beak, P. J. Am. Chem.
Soc. 2001, 123, 1004-1005. (h) Liu, H.; Liang, X.; Sohoel, H.; Buelow,
A.; Bols, M. J. Am. Chem. Soc. 2001, 123, 5116-5117. (i) Trost, B. M.;
Van Vranken, D. L. J. Am. Chem. Soc. 1993, 115, 444-458.
(7) Ricci, A. Modern Amination Methods; Wiley-VCH: Weinheim, Germany,
2000.
(1) Aziridines and Epoxides in Organic Synthesis; Yudin, A. K., Ed.; Wiley-
VCH: Weinheim, Germany, 2006.
(2) Watson, I. D. G.; Yudin, A. K. Curr. Opin. Drug DiscoVery DeV. 2002, 5,
906-917.
(8) (a) Mitsunobu, O. Synthesis 1981, 1-28. (b) Hughes, D. L. Org. Prep.
Proced. Int. 1996, 28, 127-164. (c) Wisniewski, K.; Koldziejczyk, A. S.;
Falkiewicz, B. J. Pept. Sci. 1998, 4, 1-14. (d) Nune, S. K. Synlett 2003,
1221-1222.
(3) (a) Alezra, V.; Bouchet, C.; Micouin, L.; Bonin, M.; Husson, H.-P.
Tetrahedron Lett. 2000, 41, 655-658. (b) Alezra, V.; Bonin, M.; Micouin,
L.; Policar, C.; Husson, H.-P. Eur. J. Org. Chem. 2001, 2589-2594.
9
17516
J. AM. CHEM. SOC. 2005, 127, 17516-17529
10.1021/ja055288c CCC: $30.25 © 2005 American Chemical Society

Mechanism of Palladium-Catalyzed Allylic Amination
A R T I C L E S
was heated at 60 °C for 2 h, the decomposition was complete
and no 3a was seen by GC, allowing the bis-allylated product
7 to be isolated in 84% yield (Table 2, entry 5 and Table 1,
entry 19). If N-benzyl cyclohexene imine was used in the
reaction instead of 1a (using the same conditions as Table 2,
entry 3) no new products were observed. This indicates that
the byproduct is not derived from a tetraalkyl aziridinium ion,
as is the case in aziridine alkylation with alkyl halides.1 Clearly,
the decomposition results from a buildup of acetic acid over
the course of the reaction. The acid-catalyzed opening of N-allyl
cyclohexene imine by the acetate anion results in a secondary
amine that is rapidly N-allylated. Addition of 2 equiv of K2-
CO3 to the reaction suppressed formation of the byproduct to
such an extent that even after 72 h at 60 °C no ring-opened
product was detected. K2CO3 was found to be superior to soluble
amine bases such as Et3N and DIPEA (2 equiv were used). The
latter bases slowed, but did not prevent, decomposition.
When we attempted to recover 3a (Table 1, entry 1), an
isolated yield of 45% was achieved, although both conversion
and selectivity were 100% by GC. The decrease in the isolated
yield was attributed to the volatility of the allyl aziridine product
(3a). Isolation of the allylic amination products of the reaction
of 2a with unsubstituted aziridines 1b and 1c was also
complicated by the same issue. These volatile allyl aziridine
products could be trapped by reaction with thiophenol, allowing
recovery of the resultant 1,2-aminosulfide products in good
yields (Table 1, entries 2 and 3). Allylic amination with aziridine
1d required high catalyst loading (20 mol % Pd) and additional
equivalents of 2a to achieve complete conversion and isolation
of the allyl aziridine product 3c in 79% yield (Table 1, entry
4). The difficulties in achieving full conversion using aziridine
1d were attributed to the increased steric environment around
nitrogen. As well, the chelating ability of 1d toward palladium
between the aziridine nitrogen and the carbonyl side chain
oxygen is a process that likely impedes catalysis.12 Aziridine
1g was by far the least reactive substrate used, providing only
traces of N-allyl aziridine product when 4 equiv of 2a and 20
mol % Pd are used and the reaction is heated at 40 °C for 72
h.
When monosubstituted allyl acetates were used in allylic
amination reactions, the possibility of two regioisomeric prod-
ucts arose: the unsubstituted, or linear, isomer (3) and the
substituted, or branched, isomer (4) (eq 1). When cinnamyl
acetate (2b) was reacted with 1a, the linear isomer (3d) was
favored over the branched isomer (4d) (Table 1, entry 5). The
[Pd(η3-C3H5)Cl]2/BINAP catalyst gave a 97% combined yield
and produced the best product distribution with a 92:8 ratio
favoring 3d. In the case of PPh3, the product distribution
remained constant at 81:19 linear to branched, irrespective of
the palladium source. Inclusion of the trifluoroacetate anion
through the use of Pd(CO2CF3)2 as the palladium source
decreased the regioselectivity to 75:25. Sterically demanding
phosphines such as P(o-Tol)3 and electron-deficient phosphines
such as P(C6F5)3 provided no conversion. Similar results were
achieved when 1b and 1c were used, giving the linear allyl
aziridine products 3e and 3f in 88 and 83% isolated yields,
respectively (Table 1, entries 6 and 7). When 1,3-diphenyl allyl
acetate (2f) was used, similar high yields were achieved (Table
Figure 1. Reductive amination of secondary amines versus unsubstituted
aziridines.
case of common secondary amines, is problematic with aziri-
dines because the corresponding iminium ions are highly
strained (Figure 1). Thus, transition metal catalyzed allylic
amination of allyl alcohol derivatives4,9 appears to be the most
reasonable route toward N-allyl aziridines. In the course of our
studies into the scope of this process, we uncovered a significant
deviation in the behavior of aziridine nucleophiles from that of
common secondary amines in allylic amination. The present
study investigates the course of allylic amination with aziridines
and provides a mechanistic basis for further development of
related nitrogen transfer chemistry. In particular, our results
suggest that the nature of amine can have a decisive effect on
the regiochemistry of amination. Our study also underscores
the facility with which simple aziridine-containing building
blocks can be installed in complex environments for further
elaboration.
Results
In the course of our recent studies in aziridine chemistry,10
we found that unsubstituted aziridines undergo facile palladium-
catalyzed allylic amination with allyl acetates and carbonates
(eq 1).10a A wide range of allyl aziridines has been prepared
using this procedure (Table 1).
Optimal reaction conditions were developed using cyclohex-
ene imine (1a) and allyl acetate (2a) as model substrates (eq
2). With 1 mol % [Pd(η3-C3H5)Cl]2 as the source of palladium
and 4 mol % PPh3 as ligand, full conversion of 1a to N-allyl
cyclohexene imine (3a) was achieved in THF in 30 min with
1.2 equiv of 2a (Table 2, entry 3). As expected, in the absence
of ligand (Table 2, entry 2) the reaction did not proceed. The
reaction also did not proceed if allyl alcohol was substituted
for allyl acetate.11 The reaction with allyl carbonates did continue
to completion, although at a slower rate. The reaction worked
equally well in THF (Table 2, entry 3), toluene (Table 2, entry
7), and MeCN (Table 2, entry 8) and with lower efficiency in
diethyl ether (Table 2, entry 6) and hexanes (Table 2, entry 9).
The reaction also worked with several other palladium sources
(Table 2, entries 10-12). When the reaction was left longer
than 30 min, an O-acyl N,N′-diallyl amino alcohol (trans-2-
(diallylamino)cyclohexyl acetate, 7) began to form, reducing
the selectivity (Table 2, entry 4). Heating the reaction mixture
or concentrating the solvent during workup increased the
decomposition such that only traces of 3a were isolated in these
cases. When 2 equiv of 2a were used and the reaction mixture
(9) Trost, B. M.; VanVranken, D. L. Chem. ReV. 1996, 96, 395-422.
(10) (a) Watson, I. D. G.; Styler, S. A.; Yudin, A. K. J. Am. Chem. Soc. 2004,
126, 5086-5087. (b) Sasaki, M.; Yudin, A. K. J. Am. Chem. Soc. 2003,
125, 14242-14243. (c) Siu, T.; Yudin, A. K. J. Am. Chem. Soc. 2002,
124, 530-531. (d) Watson, I. D. G.; Yudin, A. K. J. Org. Chem. 2003,
68, 5160-5167.
(11) (a) Atkins, K. E.; Walker, W. E.; Manyik, R. M. Tetrahedron Lett. 1970,
11, 3821-3824. (b) Kinoshita, H.; Shinokubo, H.; Oshima, K. Org. Lett.
2004, 6, 4085-4088.
(12) We have observed a lack of reactivity of carbonyl-containing aziridines in
other transition metal-catalyzed processes: Chen, G.; Yudin, A. K. 2004,
The University of Toronto, Toronto, Canada, unpublished results.
9
J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005 17517

A R T I C L E S
Watson and Yudin
Table 1. Synthesis of N-Allyl Aziridines by Palladium-Catalyzed Allylic Amination
9
17518 J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005

Mechanism of Palladium-Catalyzed Allylic Amination
A R T I C L E S
Table 1 (Continued)
a Reaction performed with 2 equiv of K2CO3. b Isolated after addition of 1 equiv of PhSH to in situ prepared allyl aziridine in MeCN. c Volatile products
isolated after reaction with PhSH as per note b. d Reaction performed with 2 equiv of allyl acetate at 60 °C. e Combined yield. f Linear to branched ratio.
g Isolated yield is low due to volatility of the product; GC analysis indicates that there is full conversion. h 0% de.
aziridine 1a afforded the corresponding N-allyl aziridine (3p)
in 97% ee. Gratifyingly, 1b afforded the corresponding N-allyl
aziridine (3q) in 98% ee under similar reaction conditions. To
our knowledge, these values are the highest achieved with
BINAP in allylic amination, which usually gives low enantio-
Table 2. Reaction of Cyclohexene Imine (1a) with 1.2 Equiv of
Allyl Acetate (2a)
selectivities.14
In the reaction of 1,1-dialkyl substituted allyl acetate 2d
palladium source
(2 mol % Pd)
ligand
T
t
conv selectivity
(prenyl acetate) with 1a, the branched isomer (4h) was
unexpectedly found to be the only product, isolated in 89% yield
(Table 1, entry 9). Although the reaction was sluggish with PPh3
as a ligand, BINAP provided complete conversion within 6 h
in THF at room temperature resulting in a regiochemical ratio
of 8:92 (linear to branched). The use of 2-methylbut-3-en-2-yl
acetate (2j), the isomer of 2d, resulted in a similar regiochemical
ratio (6:94) favoring the branched product (Figure 2). When
the reaction with prenyl acetate (2d) was left for longer periods
of time (up to 72 h), or heated at temperatures up to 60 °C, no
linear product was ever observed. The only compound that was
isolated under these reaction conditions was the product of ring-
opening with acetic acid (Table 1, entry 21). To exclude the
possibility of initial formation of linear product followed by its
in situ conversion into the isolated branched aziridine, we
entry
(4mol %)
solvent
(
°
C)
(h)
(%)
(%)a
1
2
3
4
5
6
7
8
9
-
-
-
THF
THF
THF
THF
rt
rt
rt
rt
60
rt
16
16
0.5 100
0
0
-
-
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2 PPh3
[Pd(η3-C3H5)Cl]2 PPh3
[Pd(η3-C3H5)Cl]2 PPh3
[Pd(η3-C3H5)Cl]2 PPh3
[Pd(η3-C3H5)Cl]2 PPh3
[Pd(η3-C3H5)Cl]2 PPh3
[Pd(η3-C3H5)Cl]2 PPh3
100
84
16
100
100
100
100
100
34
THF
2
1
1
1
1
0
93
Et2O
PhMe
MeCN
rt
rt
100
100
100
100
100
100
-
hexanes rt
10 Pd(PPh3)4
11 Pd2(dba)3‚CHCl3
12 Pd(OAc)2
-
THF
THF
THF
THF
rt
rt
rt
rt
0.5 100
0.5 100
0.5 100
PPh3
PPh3
-
13 Pd(PPh3)2Cl2
16
0
a Refers to the amount of N-allyl aziridine product vs the total amount
of all products in the reaction observed by GC.
1, entries 14-16). This chemistry can be applied to the
preparation of enantiomerically enriched aziridines13 by enan-
tioselective allylation. The use of (R)-BINAP as a ligand with
(13) (a) Lee, W. K.; Ha, H.-J. Aldrichimica Acta 2003, 36, 57-63. (b) Cardillo,
G.; Gentilucci, L.; Tolomelli, A. Aldrichimica Acta 2003, 36, 39-50.
(14) Kodama, H.; Taiji, T.; Ohta, T.; Furukawa, I. Synlett 2001, 385-387.
9
J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005 17519

A R T I C L E S
Watson and Yudin
Figure 2. Formation of branched allyl aziridine 4h from isomeric allyl
acetates.
Table 4. Effect of Solvent on Conversion and Regiochemistry
dielectric
constanta
time
(h)
conv
(%)
regioselectivity
entry
solvent
(linear/branched)
1
2
3
4
5
THF
7.6
6
14
14
14
24
100
49
49
31
0
8/92
46/54
45/55
61/39
-
PhMe
MeCN
CH2Cl2
CHCl3
38
8.9
4.8
a See ref 16.
Table 3. Variation of Ligand and Dependance of Regiochemistry
bite angle
(deg)a
conv
(%)
regioselectivity
entry
ligand
(linear/branched)
1
2
3
4
5
6
7
8
9
10
PPh3
P(OEt)3
dppe
-
-
82.6
91.6
92.8
-
97.1
98.7
100.0
104.6
15
100
82
10/90
8/92
15/85
14/86
8/92
11/89
6/94
15/85
7/93
dppp
92
Table 5. Effect of Various Ligands on the Reaction with 1c
BINAP
(S)-Tol-BINAP
dppb
dppf
(S,S)-DIOP
Xantphos
100
100
100
100
100
100
time
(h)
conv
(%)
regioselectivity
entry
palladium source
ligand
(linear/branched)
1
2
3
4
5
6
7
8
9
10
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
Pd(CO2CF3)2
BINAP
dppf
Xantphos
dppe
24
21
8
21
21
21
21
21
21
21
65
39
100
0
0
0
0
0
0
0
3/97
3/97
1/99
-
4/96
a Average P-M-P angles calculated from crystal structures retrieved
from the CSD.15
dppb
-
(S,S)-DIOP
P(OEt)3
PPh3
PPh3
BINAP
-
-
synthesized N-prenyl cyclohexene imine (7-(3-methylbut-2-
enyl)-7-aza-bicyclo[4.1.0]heptane, 3h) by alkylation of cyclo-
hexene imine with prenyl bromide. The aziridine 3h was
subjected to typical reaction conditions. No branched product
was observed, indicating that linear-to-branched interconversion
does not operate under the reaction conditions.
We examined whether changes in the steric environment
around palladium would have any effect on the observed
regioselectivity of allylic amination with aziridines. Different
phosphine ligands were employed in the reaction to assess
whether there was an effect on regiochemistry. A number of
different bidentate ligands with varying bite angles15 were used
in the reaction of cyclohexene imine (1a) with prenyl acetate
(2d) (Table 3). All the ligands employed favored the branched
product in similar ratios, ranging from 15:85 to 4:96 with
Xantphos. The latter provided the highest regioselectivity (Table
3, entry 10).
A significant solvent effect was observed in the reaction
between 1a and 2d. A high regioselectivity of 8:92 favoring
the branched product was observed in THF with BINAP as the
ligand (Table 4, entry 1). When the solvent was changed to
toluene, acetonitrile, or dichloromethane, the reaction regio-
selectivity disappeared (Table 4, entries 2-4). In the cases of
toluene and acetonitrile the linear-to-branched ratio was close
to 1:1, while in the case of dichloromethane, the linear isomer
predominated by about 6:4. In all these cases there was more
-
-
Pd(CO2CF3)2
-
than a 4-fold decrease in the reaction rate. In the case of
chloroform the reaction did not proceed (Table 4, entry 5).
Methyl aziridine (1b), methyl aziridine-2-carboxylate (1c) and
phenyl aziridine (1e) also produced the branched allyl aziridine
product when reacted with 2d in THF (Table 1, entries 10-
12). The reaction between 1b and 2d produced an allyl aziridine
that was too volatile to be directly isolated. Addition of
thiophenol into the reaction mixture after the initial allylic
amination was complete allowed the ring-opened 1,2-amino-
sulfide product to be isolated in 90% yield favoring the branched
isomer in a ratio of 71:29 (Table 1, entry 10).17 In the case of
1c, the usual reaction conditions using the BINAP ligand gave
only 6% conversion. When the amount of palladium catalyst
was increased to 5 mol %, the conversion increased, but was
still only 65% (Table 5, entry 1). A number of different
conditions and phosphine ligands were attempted with this
substrate (Table 5). Most of the conditions that were tried failed
to give any conversion (Table 5, entries 4-10). Only with the
bidentate Xantphos ligand did the reaction reach full conversion
(16) Riddick, J. A.; Bunger, W. B. Organic SolVents: Physical Properties and
Methods of Purification, 3rd ed.; Techniques of Organic Chemistry; Wiley-
Interscience: New York, 1970; Vol. II.
(17) We believe that the lower regioselectivity in this case results from some
isomerization of the ring-opened product as the ring-opening step is
performed in the presence of Pd catalyst (one-pot procedure).
(15) Dierkes, P.; van Leeuwen, P. J. Chem. Soc., Dalton Trans. 1999, 1519-
1529.
9
17520 J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005

Mechanism of Palladium-Catalyzed Allylic Amination
A R T I C L E S
Table 6. Reaction of 1a with 2e in THF
time
(h)
conv
(%)
selectivity
(%)b
regioselectivity
entry
palladium source
mol %
liganda
(linear/branched)
1
2
3
4
5
6
7
8
9
10
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
Pd(CO2CF3)2
5
5
5
1
1
1
1
1
1
1
BINAP
PPh3
PPh3
PPh3
P(OEt)3
dppb
dppe
CyPPh2
(p-MeO-C6H4)3P
P(o-Tol)3
12
72
32
48
48
16
48
48
16
48
99
100
100
79
100
100
69
79
3
0
56
52
76
65
54
41
37
22
9
19/81
17/83
12/88
12/88
29/71
19/81
25/75
3/97
Pd(CO2CF3)2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
[Pd(η3-C3H5)Cl]2
16/84
-
-
a 2:1 ligand to Pd for monodentate ligands, 1:1 ligand to Pd for bidentate ligands. b Refers to the amount of N-allyl aziridine product vs the total amount
of all products in the reaction observed by GC.
in 8 h, while the regioselectivity still favored the branched
isomer (Table 5, entry 3). Optimization of the reaction enabled
a 43% yield of 4m using 2.5 mol % [Pd(η3-C3H5)Cl]2 and 5
mol % Xantphos (Table 1, entry 11). The low yield of 4m was
attributed to the volatility of the product, which made its
isolation difficult.
3-enecarboxylate (2h) was performed.19 This substrate simplified
product analysis since the resultant π-allyl complex has a
symmetrical structure. The reaction of 2h and 1a with 1 mol %
[Pd(η3-C3H5)Cl]2 and 2 mol % BINAP in THF at room
temperature gave the allyl aziridine product (3t) in 80% isolated
yield after 16 h (Table 1, entry 18). The syn-stereochemistry
was assigned to allyl aziridine (3t) by NMR analysis (see
Supporting Information). This result shows that the stereochem-
istry present in the starting allyl acetate is retained. Verbenol
acetate (2k) failed to react with 1a under a variety of different
conditions, including high temperatures, long reaction times,
and use of both mono- and bidentate phosphine ligands.
The unusual regiochemical outcome of the reaction between
unsubstituted aziridines and prenyl type systems prompted us
to examine other amine nucleophiles (Table 7). When we
performed the allylic amination reaction between piperidine and
prenyl acetate (2d), only the linear isomer was detected after
16 h (Table 7, entry 2). As with other secondary amines such
as morpholine, 1,2,3,4-tetrahydroisoquinoline, and N-methyl-
aniline (entries 3-5, respectively), only the linear allylamine
products were isolated. Likewise, the use of benzylamine (Table
7, entry 6) resulted in exclusive formation of the linear product
as a mixture of mono- and dialkylated compounds. Sterically
encumbered amines such as 2,2,6,6-tetramethylpiperidine (entry
7) and cis-2,6-dimethylpiperidine (entry 8) did not react, even
when the reaction was heated to 60 °C.
Subsequent examination of the reaction between piperidine
and prenyl acetate (2d) by GC analysis indicated the initial
buildup of the branched product (10) followed by its decrease
as the linear product (11) began to form (Figure 3). The amount
of 10 peaked at around 1 h and then began to decrease. By 2 h,
11 was the major isomer present in the reaction mixture. The
ratio of the isomers changed throughout the reaction as the
percentage of 11 in relation to 10 steadily increased. This data
indicates that the linear and branched isomers are able to
equilibrate in the cases of amines other than aziridines. If 2
equiv of potassium carbonate or 1 equiv of Proton Sponge are
The use of geranyl acetate (2e) in the reaction with 1a also
produced the branched product 4n as the major isomer (Table
6) in THF. The reaction was more sluggish than the prenyl
acetate (2d) reaction, and higher catalyst loadings were required
to achieve full conversion. As with cinnamyl acetate (2b), P(o-
Tol)3 did not provide any conversion (Table 6, entry 10). Full
conversion and good selectivities were achieved with P(OEt)3
or PPh3 with [Pd(η3-C3H5)Cl]2 as the palladium source (Table
6, entries 5 and 2, respectively) after fairly long reaction times
(48 and 72 h, respectively). BINAP was found to be the best
ligand in combination with [Pd(η3-C3H5)Cl]2. This combination
provided a linear-to-branched ratio of 19:81 and full conversion
within 12 h. However, the reaction with geranyl acetate was
found to produce significant amounts of volatile byproducts that
reduced the selectivity to around 50%. These products were
found to be the result of a background reaction of 2e with Pd0,
not the previously observed acetate ring-opening of the N-allyl
aziridine products. The selectivity of the reaction suffered as a
result of the concurrent production of myrcene, (E)-ocimene,
and (Z)-ocimene via competing â-hydride elimination.18 The
selectivity was improved with a 1:4 Pd(CO2CF3)2/PPh3 catalyst
that provided a linear-to-branched ratio of 12:88 (Table 6, entry
4). By increasing the amount of catalyst, the reaction time was
decreased and full conversion was achieved with a selectivity
of 76% (Table 6, entry 3). Finally, when the reaction was
performed with a slight excess of 2e (1.2 equiv), a combined
yield of 80% was achieved after 32 h (Table 1, entry 13).
To determine the stereochemical course of the reaction, allylic
amination with the syn-isomer of methyl 5-acetoxycyclohex-
(18) Reaction of geranyl acetate with 5 mol % Pd(PPh3)4 in THF at 60 °C gives
full conversion to the background products in 6 h. See also the following
references: (a) Keinan, E.; Kumar, S.; Dangur, V.; Vaya, J. J. Am. Chem.
Soc. 1994, 116, 11151-11152. (b) Andersson, P. G.; Schab, S. Organo-
metallics 1995, 14, 1-2.
(19) (a) Trost, B. M.; Verhoeven, T. R. J. Am. Chem. Soc. 1980, 102, 4730-
4743. (b) Granberg, K. L.; Backvall, J. E. J. Am. Chem. Soc. 1992, 114,
6858-6863. (c) Granberg, K. L.; Backvall, J. E. J. Am. Chem. Soc. 1994,
116, 10853-10853.
9
J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005 17521

A R T I C L E S
Watson and Yudin
Figure 3. Conversion/time diagram for the reaction of prenyl acetate (2d) with piperidine.
Table 7. Reaction of Different Amines with Prenyl Acetate (2d)
Scheme 1. Synthesis of Branched Amine 13
under Standard Conditionsa
13 was added to the reaction mixture leading to its isomerization,
such that linear isomer 14 predominated (eq 7).
a 1 mol % [Pd(η3-C3H5)Cl]2, 4 mol % BINAP, THF, rt. b pKaH of ethylene
imine from ref 73. c Values taken from “Dissociation Constants of Organic
Acids and Bases”, in CRC Handbook of Chemistry and Physics, Internet
Version 2005; David R. Lide, Ed., http://www.hbcpnetbase.com; CRC
Press: Boca Raton, FL, 2005. d Reaction did not proceed even upon heating
at 60 °C.
Branched amine 13 was then subjected to different reaction
conditions to determine which components of the catalytic
system were necessary for branched-to-linear interconversion.
When 1 equiv of TFA was added to 13 in THF, no isomerization
was observed over 72 h at room temperature. Reaction of 13
with 10 mol % of either Pd(OAc)2 or Pd(PPh3)4 in THF also
gave no isomerization over 72 h. Additionally, reaction of 13
with 1 equiv of TFA and 10 mol % of Pd(PPh3)4 in THF gave
no isomerization over 72 h. Finally, reaction of 13 with 1 equiv
of prenyl acetate (2d) and 10 mol % of Pd(PPh3)4 in THF gave
no isomerization over 72 h.
Crossover experiments were performed between 13 and
piperidine or cyclohexene imine (1a). When 13 was mixed with
1 equiv of piperidine using [Pd(η3-C3H5)Cl]2/BINAP as the
catalyst, 13 disappeared such that almost none remained after
4 h at room temperature. Both the branched (10) and linear (11)
piperidine crossover products were detected, with only 11
present after 4 h (eq 8). Similarly, when 13 was mixed with 1
equiv of cyclohexene imine (1a) under the same conditions,
added to the reaction between piperidine and 2d, there is no
change in the regioselectivity at the end of the reaction; the
linear regioisomer remains the dominant product.
1,2,3,4-Tetrahydroisoquinoline behaved analogously to pip-
eridine, initially producing the branched isomer (13), which then
isomerized to the linear isomer (14) with either [Pd(η3-C3H5)-
Cl]2/BINAP or Pd(PPh3)4 in THF. To investigate the isomer-
ization, the branched amine of 1,2,3,4-tetrahydroisoquinoline
(13) was synthesized by a three-step procedure (Scheme 1)
without the use of any palladium reagents. The ability of 13 to
isomerize to 14 was then tested. First, 1,2,3,4-tetrahydroiso-
quinoline was allowed to react with prenyl acetate (2d) with 1
mol % [Pd(η3-C3H5)Cl]2 and 2 mol % BINAP in THF. Within
4 h the reaction equilibrated to linear isomer 14, and only traces
of branched isomer were detected by GC. The branched isomer
9
17522 J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005

Mechanism of Palladium-Catalyzed Allylic Amination
A R T I C L E S
crossover occurred and a small amount of branched aziridine
(4h) was detected (14% crossover). No linear aziridine was
detected. None of the original branched amine 13 remained;
instead, the linear amine (14) was detected, indicating that
isomerization of the starting material was occurring under these
conditions (eq 9).
geranyl acetate, the branched isomer was unexpectedly found
to be the major product. This result is unusual as other classes
of nitrogen nucleophiles including primary20a and secondary
alkylamines,20b azide,20c,d and amides20e all favor the linear
allylamine products under similar palladium-catalyzed condi-
tions. Indeed, when we reacted a variety of secondary amines
with prenyl acetate we only isolated linear products (Table 7).
This interesting dichotomy warranted mechanistic investigation.
The vast majority of palladium systems produce the linear
product as the major regioisomer.4 This is in contrast to metals
such as iridium,21a-d rhodium,21e-g molybdenum,21h-j and
ruthenium,21k which favor the branched product. There are only
a few other examples of palladium-catalyzed allylic aminations
of terminal allyl esters or carbonates resulting in predominant
formation of the branched isomers.22 These examples employ
special ligands to switch the regioselectivity. They differ from
our example in that in our case the same catalyst-ligand system
produces different regioisomeric products depending on whether
a common amine or aziridine nucleophile is used. The steric
environment around palladium appears to have no effect on the
regiochemical outcome of the reaction with unsubstituted
aziridines. Although a number of bidentate phosphines were
tried with a range of bite angles, the branched-to-linear ratio
remained constant at around 9:1 (Table 3).
The ability of branched amine 13 to act as an electrophile
prompted us to examine whether branched aziridine 4h would
react in a similar way. When 4h was mixed with 1 equiv of
1,2,3,4-tetrahydroisoquinoline using [Pd(η3-C3H5)Cl]2/BINAP
as the catalyst, no reaction was observed over 72 h at room
temperature (eq 10).
The possibility that a switch in the nature of the key bond-
forming step was operating with an unsubstituted aziridine
nucleophile led us to examine the stereochemical course of the
reaction. In palladium-catalyzed allylic substitution, the attack
by the nucleophile can occur externally by anti-addition to the
π-allyl complex or internally by reductive elimination (Figure
4). Hard nucleophiles such as Grignard,23a organozinc,23b
organotin,23c-e organoaluminum,23b silyl aluminum,23f disilane,23g
organoboranes (under Ni0 catalysis),23h,i carbon monoxide,23d,j
and certain hydride reagents23k are known to react internally
resulting in inversion of stereochemistry. Amine nucleophiles
are known to react externally resulting in overall retention of
stereochemistry (double inversion).24 The use of allyl acetate
2h allowed the stereochemical course of the reaction with
unsubstituted aziridines to be determined. The syn-isomer,
observed in the allylic amination reaction, provides direct
evidence that the unsubstituted aziridine attacks the palladium
complex in an external fashion. This relative stereochemistry
is identical to what is observed with other amine nucleophiles.
Moreover, when verbenol acetate (2k) was used as the substrate,
The fact that crossover was occurring prompted us to once
again attempt to isomerize 13. Using [Pd(η3-C3H5)Cl]2/BINAP
as the catalyst in THF, we added 1 equiv of TFA and 1,2,3,4-
tetrahydroisoquinoline to 13. After 16 h at room temperature,
18% of 13 and 82% of 14 remained , indicating that isomer-
ization did occur (eq 11).
Discussion
(20) (a) Trost, B. M.; Keinan, E. J. Org. Chem. 1979, 44, 3451-3457. (b)
Åkermark, B.; Vitagliano, A. Organometallics 1985, 4, 1275-1283. (c)
Mizuno, M.; Shioiri, T. Chem. Commun. 1997, 2165-2166. (d) Murahashi,
S.-I.; Taniguchi, Y.; Imada, Y.; Tanigawa, Y. J. Org. Chem. 1989, 54,
3292-3303. (e) Hutchins, R. O.; Wei, J.; Rao, S. J. J. Org. Chem. 1994,
56, 4007-4009.
(21) (a) Kiener, C. A.; Shu, C.; Incarvito, C.; Hartwig, J. F. J. Am. Chem. Soc.
2003, 125, 14272-14273. (b) Ohmura, T.; Hartwig, J. F. J. Am. Chem.
Soc. 2002, 124, 15164-15165. (c) Shu, C. T.; Leitner, A.; Hartwig, J. F.
Angew. Chem., Int. Ed. 2004, 43, 4797-4800. (d) Takeuchi, R.; Ue, N.;
Tanabe, K.; Yamashita, K.; Shiga, N. J. Am. Chem. Soc. 2001, 123, 9525-
9534. (e) Evans, P. A.; Robinson, J. E.; Moffett, K. K. Org. Lett. 2001, 3,
3269-3271. (f) Evans, P. A.; Robinson, J. E.; Nelson, J. D. J. Am. Chem.
Soc. 1999, 121, 6761-6762. (g) Evans, P. A.; Robinson, J. E.; Nelson, J.
D. J. Am. Chem. Soc. 1999, 121, 12214-12214. (h) Belda, O.; Moberg,
C. Acc. Chem. Res. 2004, 37, 159-167. (i) Trost, B. M.; Lautens, M. J.
Am. Chem. Soc. 1982, 104, 5543-5545. (j) Trost, B. M.; Lautens, M. J.
Am. Chem. Soc. 1987, 109, 1469-1478. (k) Trost, B. M.; Fraisse, P. L.;
Ball, Z. T. Angew. Chem., Int. Ed. 2002, 41, 1059-1061.
We have investigated the palladium-catalyzed allylic amina-
tion reaction using unsubstituted aziridines as nucleophiles. The
process allows for the asymmetric allylic amination of unsub-
stituted aziridines with high enantioselectivities in the case of
cinnamyl systems and creates the possibility of introducing
aziridine moieties into functionalized environments with high
levels of regioselectivity and high isolated yields. Our investiga-
tions into the allylic amination of unsubstituted aziridines have
revealed some interesting trends with respect to the regiochem-
istry of the products. In the case of cinnamyl substrates, the
linear allyl aziridines are the main products. In these cases, the
resulting regiochemistry of the product is as expected, producing
the same regioisomer that is observed when other alkylamines
are used in palladium-catalyzed allylic amination. In the reaction
of 1,1-dialkyl-substituted allyl acetates such as prenyl and
(22) (a) You, S.-L.; Zhu, X.-Z.; Luo, Y.-M.; Hou, X.-L.; Dai, L.-X. J. Am. Chem.
Soc. 2001, 123, 7471-7472. (b) Hayashi, T.; Kishi, K.; Yamamoto, A.;
Ito, Y. Tetrahedron Lett. 1990, 31, 1743-1746. (c) Faller, J. W.; Wilt, J.
C. Org. Lett. 2004, 45, 7613-7616. (d) Faller, J. W.; Wilt, J. C. Org. Lett.
2005, 7, 633-636.
9
J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005 17523

A R T I C L E S
Watson and Yudin
Figure 4. Internal versus external transfer in cyclic allyl acetate substrates.
Figure 5. Ionization of (S)-cis-verbenol acetate (2k).
Figure 7. Interconversion of isomeric allyl acetates.
aziridine such as 1a, the branched allyl aziridine product
predominates in about the same regiochemical ratio. This shows
that the position of the leaving group has no bearing on the
regiochemical outcome of allylic amination with aziridines. The
fact that either of the isomeric acetates results in the same
product ratio indicates that both isomeric allyl acetates (Figure
7, 2d and 2j) operate through the same intermediate, an η3
π-complex (B).
Once the allyl aziridine product (4h) is produced, it appears
to be stable toward isomerization. Over the course of the
reaction, the branched-to-linear ratio remains constant. The ratio
is not affected if the reaction mixture is left for extended periods
of time or if the reaction mixture is heated. Indeed, the linear
isomer is a trisubstituted olefin and should be of lower energy
than the disubstituted product. When the linear aziridine 3h was
subjected to typical reaction conditions, no branched product
was observed. Therefore, there is no interconversion between
the branched and linear allyl aziridines under any of the
conditions we tried.
Figure 6. Mechanism of allylic amination with unsubstituted aziridines.
no reaction took place. This is consistent with the gem-dimethyl
group blocking the exo-face of attack (Figure 5) and provides
further evidence that the attack is external. Thus, one may
conclude that the nucleophilic attack operates by the classical
mechanism of allylic amination in the case of aziridine
nucleophiles as depicted in Figure 6.
This behavior is in contrast to that observed with the
allylamine products when secondary amines such as piperidine
are used in the reaction. The observation of an initial branched
product in the reaction between piperidine and 2d, followed by
its disappearance while the amount of linear isomer increases,
indicates that the allylamine isomers are able to equilibrate.
Therefore, in the case of aziridines and other secondary amines
the kinetic product is the branched isomer. However, in the case
of other amines the branched product converts to the linear
isomer, while in the case of aziridines this equilibration
apparently does not occur (Figure 8).
In the rhodium-catalyzed allylic substitution, the regioselec-
tivity is determined by the position of the leaving group.25
Memory effects are also known to operate in some situations
during palladium-catalyzed allylic substitution.26 We examined
whether a similar effect was operating with aziridine nucleo-
philes in the case of palladium. When isomeric acetates 2d or
2j are used in the allylic amination reaction with an unsubstituted
(23) (a) Hayashi, T.; Konishi, M.; Yokota, K. I.; Kumada, M. J. Organomet.
Chem. 1985, 285, 359-373. (b) Matsushita, H.; Negishi, E. J. Chem. Soc.,
Chem. Commun. 1982, 160-161. (c) Kurosawa, H.; Kajimaru, H.; Ogoshi,
S.; Yoneda, H.; Miki, K.; Kasai, N.; Murai, S.; Ikeda, I. J. Am. Chem. Soc.
1992, 114, 8417-8424. (d) Sheffy, F. K.; Godschalx, J. P.; Stille, J. K. J.
Am. Chem. Soc. 1984, 106, 4833-4840. (e) Del Valle, L.; Stille, J. K.;
Hegedus, L. S. J. Org. Chem. 1990, 55, 3019-3023. (f) Trost, B. M.;
Yoshida, J.; Lautens, M. J. Am. Chem. Soc. 1983, 105, 4494-4496. (g)
Matsumoto, Y.; Ohno, A.; Hayashi, T. Organometallics 1993, 12, 4051-
4055. (h) Trost, B. M.; Spagnol, M. D. J. Chem. Soc., Perkin Trans. 1
1995, 2083-2096. (i) Kobayashi, Y.; Mizojiri, R.; Ikeda, E. J. Org. Chem.
1996, 61, 5391-5399. (j) Murahashi, S. I.; Imada, Y.; Taniguchi, Y.;
Higashiura, S. J. Org. Chem. 1993, 58, 1538-1545. (k) Keinan, E.;
Greenspoon, N. Tetrahedron Lett. 1982, 23, 241-244.
(25) (a) Evans, P. A.; Nelson, J. D. Tetrahedron Lett. 1998, 39, 1725-1728.
(b) Evans, P. A.; Leahy, D. K. J. Am. Chem. Soc. 2000, 122, 5012-5013.
(c) Evans, P. A.; Kennedy, L. J. Org. Lett. 2000, 2, 2213-2215. (d) Evans,
P. A.; Kennedy, L. J. J. Am. Chem. Soc. 2001, 123, 1234-1235.
(26) (a) Lloyd-Jones, G. C.; Stephen, S. C.; Murray, M.; Butts, C. P.; Vyskocil,
S.; Kocovsky, P. Chem. Eur. J. 2000, 6, 4348-4357. (b) Lloyd-Jones, G.
C.; Stephen, S. C. Chem. Eur. J. 1998, 4, 2539-2549. (c) Lloyd-Jones, G.
C.; Stephen, S. C. Chem. Commun. 1998, 2321-2322. (d) Vyskocil, S.;
Smrcina, M.; Hanus, V.; Polasek, M.; Kocovsky, P. J. Org. Chem. 1998,
63, 7738-7748. (e) Hayashi, T.; Kawatsura, M.; Uozumi, Y. J. Am. Chem.
Soc. 1998, 120, 1681-1687.
(24) Moreno-Manas, M.; Morral, L.; Pleixats, R. J. Org. Chem. 1998, 63, 6160-
6166.
9
17524 J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005

Mechanism of Palladium-Catalyzed Allylic Amination
A R T I C L E S
Figure 8. Differences in experimentally observed products in the course of allylic amination chemistry.
Figure 9. 1H NMR in d8-THF. (Top) [Pd(η3-C5H9)Cl]2. (Bottom) [Pd(η3-C5H9)Cl]2 with 4 equiv of PPh3.
What is the reason for the observed kinetic selectivity? The
mechanism of allylic substitution is classically shown as
proceeding through an η3 π-complex. However, this complex
is in equilibrium with the σ-complexes as shown in Figure 7.27
The extent of the equilibrium depends on many factors,
including temperature, solvent, metal, ligand, and counterion.
It is known that in THF the allyl palladium species exists as a
σ-complex with counterion coordinated to the metal to a much
greater extent than in other more polar solvents.28 The presence
of chloride, added as part of the Pd source ([Pd(η3-C3H5)Cl]2),
is also known to induce the formation of neutral σ-complexes.29
1H NMR experiments using the prenyl palladium chloride dimer
([Pd(η3-C5H9)Cl]2) and PPh3 reveal formation of a σ-complex
in d8-THF (Figure 9).28a,30 Attack by the nucleophile likely
proceeds through SN2′-type attack at the more substituted allyl
position to produce the branched product. The loss of selectivity
observed when different solvents are used (Table 4) is attributed
to the differing equilibrium existing between the η3 π-complex
and σ-complexes in solvents of different polarities. Therefore,
in these cases there is a loss of kinetic selectivity.
One can propose several different pathways through which
the isomerization of branched amines may occur. One possibility
is that the branched products isomerize in the presence of either
AcOH or PdII that accumulates in the reaction medium.31
However, 13 was shown to be stable to both TFA and catalytic
Pd(OAc)2.32 Another possibility is that the branched product is
subsequently allylated via a palladium catalyst, producing a
diallylated species that could re-ionize back to an allylamine.
No evidence for this mechanism was observed since in the
presence of both prenyl acetate and Pd0 no isomerization of 13
took place.
(27) Some σ-allyl palladium complexes have been structurally characterized:
(a) Kollmar, M.; Helmchen, G. Organometallics 2002, 21, 4771-4775.
(b) Braunstein, P.; Naud, F.; Dedieu, A.; Rohmer, M.-M.; DeCian, A.;
Rettig, S. Organometallics 2001, 20, 2966-2981.
The most likely mechanistic explanation for the isomerization
seems to be by a Pd0-catalyzed process. This could arise by
(28) (a) Åkermark, B.; Åkermark, G.; Hegedus, L. S.; Zetterberg, K. J. Am.
Chem. Soc. 1981, 103, 3037-3040. (b) Amatore, C.; Jutand, A.; Meyer,
G.; Mottier, L. Chem. Eur. J. 1999, 5, 466-473. (c) Bouquillon, S.; Muzart,
J. Eur. J. Org. Chem. 2001, 3301-3305.
(31) Overman, L. E.; Knoll, F. M. Tetrahedron Lett. 1979, 20, 321-324.
(32) Phosphines are known to reduce Pd(OAc)2 to Pd0. See the following
references: (a) Amatore, C.; Carre, E.; Jutand, A.; Mbarki, M. A.
Organometallics 1995, 14, 1818-1826. (b) Amatore, C.; Jutand, A.;
Thuilliez, A. Organometallics 2001, 20, 3241-3249.
(29) Thibault, C.; Ge´nin, E.; Giroud, C.; Meyer, G.; Jutand, A. J. Organomet.
Chem. 2003, 687, 365-376 and references therein.
(30) (a) Powell, J.; Shaw, B. L. J. Chem. Soc. A 1967, 1839-1851. (b) Kurosawa,
H. J. Chem. Soc., Dalton Trans. 1979, 939.
9
J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005 17525

A R T I C L E S
Watson and Yudin
Figure 10. Rationale behind allylamine isomerization.
two possible paths: oxidative addition or re-ionization. Oxida-
tive addition into a C-N bond has been observed in the Suzuki
cross-coupling of aryltrimethylammonium salts33a and 1-aryl-
triazenes.33b In these cases, Ni0 catalysts are required for the
reaction to proceed, whereas palladium catalysts are
unsuccessful.33a These observations make oxidative addition an
unlikely pathway. An alternative, more likely mechanism
involves the re-ionization of the allyl aziridine product to
generate an allyl palladium electrophile and an amine. The
reversibility of the C-N bond-forming step has been observed
in allylic amination when carbon nucleophiles were used in the
palladium-catalyzed reaction of allylic ammonium halides.34
However, reaction of 13 with Pd0 in THF failed to produce any
isomerization. Our experiments show that allylamine isomer-
ization requires very particular conditions. While the reaction
of 1,2,3,4-tetrahydroisoquinoline with prenyl acetate resulted
in fully isomerized product within 4 h, subjecting independently
prepared branched amine 13 to Pd0 and ligand resulted in no
isomerization. However, 13 did isomerize to linear product 14
when added to the reaction contents of the original allylic
amination (eq 11).
The fact that crossover of the prenyl moiety was observed
from branched amine 13 to other amines such as piperidine
indicated that the mechanism for isomerization is bimolecular.
The crossover experiments also clearly demonstrate the differing
abilities of branched N-allylamines and N-allyl aziridines to act
as electrophile precursors under the reaction conditions. The
branched amine 13 was able to transfer its prenyl unit to both
piperidine and cyclohexene imine (1a). In the case of piperidine
there was complete crossover and the product fully isomerized
to the linear isomer. Therefore, the crossover conditions
contained all the components necessary for isomerization. In
these experiments both a nucleophile (the amine or aziridine)
and an acid, generated from the palladium precursor, were
additionally present. When isomerization of 13 was again
attempted with Pd0, acid, and 1,2,3,4-tetrahydroisoquinoline, the
isomerization to 14 was observed. To summarize, all three
components (Pd0 catalyst, acid, and nucleophile) are required
for isomerization. This information leads us to propose a
rationale for the observed branched-to-linear interconversion
(Figure 10).
In THF, the reaction produces protonated branched amine as
the kinetic product by an SN2′ reaction with an allyl palladium
σ-complex. Recomplexation of Pd0 to the product alkene enables
ionization to either the η3 π-complex B or σ-complex A, which
are also in equilibrium (Figure 10). It is the generation of these
electrophilic components that enables crossover reactions to
occur. Attack of a nucleophile on the more prevalent σ-complex
A simply regenerates branched allylamine. However, reaction
at the η3 π-complex B should favor attack at the less substituted
carbon, resulting in formation of the linear product. The reaction
through σ-complex C is unlikely. Due to the increased steric
hindrance around palladium in C compared to A, C is likely to
be a very small component in the equilibrium. The strength of
the palladium/olefin interaction in complex D is higher than
that in complex E due to increased degree of substitution in
E.35 As a consequence, ionization of complex D should be more
facile than complex E. Therefore, the dynamic process outlined
in Figure 10 will steadily increase the amounts of the linear
product. With aziridine the formation of either A or B does not
take place, which accounts for the observed lack of branched-
to-linear isomerization in the case of N-allyl aziridines. We
considered the possibility that protonation of the allylamine
product was required for the isomerization. We reasoned that
the lower basicity of aziridines relative to that of amines (Table
7) would result in a lower equilibrium concentration of
protonated amine. Therefore, no isomerization would take place.
The lower basicity of aziridines results from bond strain which
increases the s-character of the nitrogen lone pair. For instance,
ethylene imine has a pKaH (8.0436) that is closer to an sp2
nitrogen atom in an imine (diphenyl imine pKaH ) 7.2) than to
an sp3 nitrogen atom in an amine (piperidine pKaH ) 11.12).
Further investigation of amines revealed that pKa is not likely
(33) (a) Blakey, S. B.; MacMillan, D. W. C. J. Am. Chem. Soc. 2003, 125,
6046-6047. (b) Saeki, T.; Son, E. C.; Tamao, K. Org. Lett. 2004, 6, 617-
619.
(34) (a) Doi, T.; Yanagisawa, A.; Miyazawa, M.; Yamamoto, K. Tetrahedron:
Asymmetry 1995, 6, 389-392. (b) Hirao, T.; Yamada, N.; Ohshiro, Y.;
Agawa, T. J. Organomet. Chem. 1982, 236, 409-414.
(35) For instance, monosubstituted olefins are known to be considerably more
reactive than disubstituted olefins in Wacker oxidation: Nelson, D. J.; Li,
R.; Brammer, C. J. Am. Chem. Soc. 2001, 123, 1564-1568.
(36) Searles, S.; Tamres, M.; Block, F.; Quarterman, L. A. J. Am. Chem. Soc.
1956, 78, 4917-4920.
9
17526 J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005

Mechanism of Palladium-Catalyzed Allylic Amination
A R T I C L E S
selectivities typically observed with common amines as well
as points to avenues for kinetic resolution.39
Experimental Section
Figure 11. Allyl aziridine coordinated to palladium.
General Procedures. Anhydrous acetonitrile, dichloromethane,
diethyl ether, and toluene were obtained using the method described
by Grubbs.40 Tetrahydrofuran (THF) was distilled from sodium
benzophenone under argon. Acetone was stored over 4 Å molecular
sieves. Column chromatography was carried out using Silicycle 230-
400 mesh silica gel or aluminum oxide, neutral, Brockman type 1.
Analytical thin-layer chromatography (TLC) was performed on Ma-
cherey Nagel precoated glass-backed TLC plates (SIL G/UV254, 0.25
mm) and visualized by UV lamp (254 nm), iodine and potassium
permanganate stain. 1H NMR and 13C NMR spectra were recorded on
a Varian Mercury 300, VRX-S (Unity) 400 or Unity 500 spectrometer.
All unsubstituted aziridines 1a-g were either purchased from com-
mercial sources or synthesized by literature methods. All allyl acetates
2a-k were synthesized by the literature method from their respective
allyl alcohols.
Representative Procedure for the Preparation of Allyl Aziridines
by Pd-Catalyzed Allylic Amination: 7-Allyl-7-aza-bicyclo[4.1.0]-
heptane (3a). In a 15 × 100 mm screw cap test tube, equipped with
septum and magnetic stir bar, were placed [Pd(η3-C3H5)Cl]2 (5 mg,
0.0137 mmol), PPh3 (14 mg, 0.0547 mmol), K2CO3 (379 mg, 2.74
mmol), and dry THF (2 mL). Allyl acetate (2a) (148 µL, 1.37 mmol)
and 1a (133 mg, 1.37 mmol) were added via syringe, and the solution
was stirred under a stream of argon at room temperature for 30 min,
when GC analysis showed no cyclohexene imine (1a) remaining. Water
(4 mL) was added, the organic layer was separated, and the aqueous
layer was extracted with diethyl ether (3 × 2 mL). The combined
organic layers were concentrated in vacuo, and the residue was purified
by flash chromatography (Rf 0.35, SiO2, 9:1 hexanes/EtOAc) to yield
N-allyl cyclohexene imine (3a) (85 mg, 0.62 mmol, 45%) as a clear
liquid.
to be the sole factor contributing to the observed behavior (Table
7). A variety of different amines were employed in the reaction,
having pKaH values both higher and lower than ethylene imine.
In all cases in which amines where used as nucleophiles the
only isomer that was isolated was the linear product.
The amination of allyl acetates produces 1 equiv of acetic
acid. In the absence of base and with full conversion of the
aziridine starting material the acid was expected to form an
aziridinium salt with the allyl aziridine products. In an attempt
to prevent the unwanted aziridine ring opening with acetic acid,
we employed potassium carbonate as the acid scavenger, which
led to a significant decrease in the reaction rate (by almost half).
A possible reason for this unexpected decrease in the reaction
efficiency is coordination of the allyl aziridine products to
palladium, sequestering some of the catalyst out of the catalytic
cycle and thus slowing the rate of conversion (Figure 11).37
Evidence for allyl aziridine/palladium interaction comes from
lack of crossover when allyl aziridine is mixed with 1,2,3,4-
tetrahydroisoquinoline and palladium.38 Protonation of the
aziridine nitrogen should prevent coordination to the metal.
In closing, significant deviation in behavior of aziridine
nucleophiles in palladium-catalyzed allylic amination allows
facile synthesis of useful branched amines that are difficult to
obtain using other techniques. It appears that the origin of this
behavior can be traced back to lack of isomerization of the
initially produced branched product, whereas in the case of other
amine nucleophiles a pathway for interconversion exists. The
higher s-character of the aziridine nitrogen and, hence, its lower
basicity reduce the relative amount of the protonated species
that is necessary for isomerization to take place. The quantitative
crossover experiments clearly confirm this relative lack of
nucleophilicity. These experiments also demonstrate that even
when protonated aziridine species is present, it is far less capable
of reacting with Pd0 than the common amine-derived allylated
species. This lack of reactivity is a result of the stronger C-N
bond due to, once again, higher s-character of the aziridine
nitrogen.
7-Allyl-7-aza-bicyclo[4.1.0]heptane (N-Allyl Cyclohexene Imine,
1
3a): Clear liquid, 45% yield. H NMR (CDCl3, 300 MHz): δ 5.91
(ddt, J ) 17.3, 10.4, 5.3 Hz, 1H), 5.22 (ddd, J ) 17.3, 3.7, 1.8 Hz,
1H), 5.08 (ddd, J ) 10.4, 3.5, 1.5 Hz, 1H), 2.85 (dt, J ) 5.3, 1.5 Hz,
2H), 1.86-1.71 (m, 4H), 1.50-1.48 (m, 2H), 1.39-1.13 (m, 4H); 13
C
NMR (CDCl3, 75 MHz): δ 136.1, 115.7, 63.4, 38.3, 24.7, 20.8. ESI:
m/z (%) 138 (M+ + 1); EI-MS: m/z (%) 136 (M+ - 1, 100), 123 (46),
108 (47), 96 (42), 84 (67); HR-MS: calcd for C9H15N, 137.1204; obsd,
137.1202.
(N-Allyl-3-(but-3-enyl)aziridin-2-yl)(phenyl)methanone (3c): Clear
oil, 83% yield. Peaks have been assigned by analysis of 2D NMR
1
data: COSY, HSQC, and CIGAR. H NMR (CDCl3, 500 MHz): δ
Conclusions
8.02-7.99 (m, 2H, ortho-aromatics), 7.61-7.56 (m, 1H, para-aromatic),
7.51-7.46 (m, 2H, meta-aromatics), 5.90-5.76 (m, 2H, -CHdCH2),
5.15-4.95 (m, 4H, -CHdCH2), 3.39 (d, J ) 2.9 Hz, 1H, -CHCOPh),
3.36 (dd, J ) 5.7 Hz, 1H, NCH2CHdCH2), 3.25 (dd, J ) 14.3, 6.0
Hz, 1H, NCH2CHdCH2), 2.47 (m, 1H, -CH2CH-), 2.33-2.12 (m,
In summary, we have investigated the palladium-catalyzed
allylic amination reaction using unsubstituted aziridines. The
observed regioselectivity favors valuable branched products in
the cases of aliphatic allyl acetates, underscoring the decisive
effect of the amine on the course of allylic amination. This
observed switch in regiochemistry results only from the
structural differences between amines and aziridines. Our
investigations uncovered isomerization of allylated amines under
palladium catalysis. This observation is significant for palladium
catalysis as it suggests a possible explanation for low enantio-
2H, CH2dCHCH2CH2-), 1.74-1.55 (m, 2H, CH2dCHCH2CH2-); 13
C
NMR (CDCl3, 125 MHz): δ 195.7 (CdO), 138.5 (ipso-aromatic), 137.9
(-CH2CH2CHdCH2), 135.6 (NCH2CHdCH2), 133.4 (para-aromatic),
128.8 (meta-aromatics), 128.4 (ortho-aromatics), 117.1 (NCH2CHd
CH2), 115.4 (-CH2CH2CHdCH2), 54.1 (NCH2CHdCH2), 47.5
(-CH2CH-), 43.9 (-CHCOPh), 32.5 (CH2dCHCH2CH2-), 31.7
(CH2dCHCH2CH2-). Rotamer peaks: 1H NMR (CDCl3, 500 MHz):
δ 6.08-5.96 (m, 0.25H), 5.30-5.24 (m, 0.25H), 3.50-3.44 (m, 0.25H),
3.20-3.14 (m, 0.25H), 2.79 (d, J ) 2.8 Hz, 0.25H), 2.59 (m, 0.25H),
2.00-1.91 (m, 0.2H), 1.84-1.76 (m, 0.2H); 13C NMR (CDCl3, 125
(37) (a) Complexes between palladium and C-vinyl aziridines have been
structurally characterized: BenCheikh, R.; Chaabouni, R.; Bonnet, M. C.;
Dahan, F. Polyhedron 1998, 17, 185-192. (b) Ferioli, F.; Fiorelli, C.;
Martelli, G.; Monari, M.; Savoia, D.; Tobaldin, P. Eur. J. Org. Chem. 2005,
1416-1426. (c) Tanner, D.; Andersson, P. G.; Harden, A.; Somfai, P.
Tetrahedron Lett. 1994, 35, 4631-4634. (d) Tanner, D.; Johansson, F.;
Harden, A.; Andersson, P. G. Tetrahedron 1998, 54, 15731-15738.
(38) The reverse is not the case: there is crossover when NH aziridine 1a is
added to allylated branched 1,2,3,4-tetrahydroisoquinoline (13).
(39) For recent use of Ir-catalyzed kinetic resolution of allyl acetates, see:
Fischer, C.; Defieber, C.; Suzuki, T.; Carreira, E. M. J. Am. Chem. Soc.
2004, 126, 1628-1629.
(40) Pangborn, A. B.; Giardello, M. A.; Grubbs, R. H.; Rosen, R. K.; Timmers,
F. J. Organometallics 1996, 15, 1518-1520.
9
J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005 17527

A R T I C L E S
Watson and Yudin
MHz): δ 196.6, 137.3, 137.2, 135.4, 133.2, 128.7, 128.4, 116.6, 115.9,
54.6, 47.8, 47.3, 31.7, 25.6. ESI: m/z (%) 242 (M+ + 1); EI-MS: m/z
(%) 241 (M+, 6), 200 (13), 186 (73), 136 (70), 105 (100), 91 (22), 77
(68), 67 (17); HR-MS: calcd for C16H19NO, 241.1466; obsd, 241.1468.
(Rf 0.56, SiO2, 8:2 hexanes/EtOAc) to yield 3h (145 mg, 0.88 mmol,
45%) as a clear oil.
1H NMR (CDCl3, 300 MHz): δ 5.29 (m, 1H), 2.82 (d, J ) 6.4 Hz,
2H), 1.85-1.71 (m, 4H), 1.71 (d, J ) 1.2 Hz, 3H), 1.58 (s, 3H), 1.48-
1.47 (m, 2H), 1.40-1.29 (m, 2H), 1.20-1.08 (m, 2H); 13C NMR
(CDCl3, 75 MHz): δ 132.9, 122.5, 58.8, 38.0, 25.8, 24.6, 20.6, 18.2.
ESI: m/z (%) 166 (M+ + 1).
1
7-Cinnamyl-7-aza-bicyclo[4.1.0]heptane (3d): H NMR (CDCl3,
300 MHz): δ 7.38-7.35 (m, 2H), 7.31-7.26 (m, 2H), 7.19 (m, 1H),
6.56 (d, J ) 15.9 Hz, 1H), 6.29 (dt, J ) 15.8, 5.6 Hz, 1H), 3.00 (dd,
J ) 5.6, 1.5 Hz, 2H), 1.90-1.71 (m, 4H), 1.57-1.53 (m, 2H), 1.44-
1.32 (m, 2H), 1.26-1.12 (m, 2H); 13C NMR (CDCl3, 75 MHz): δ
137.5, 130.8, 128.6, 127.7, 127.3, 126.4, 62.8, 38.3, 24.6, 20.7. ESI:
m/z (%) 214 (M+ + 1, 100), 117 (60); EI-MS: m/z (%) 213 (M+, 24),
117 (46), 96 (100), 69 (54); HR-MS: calcd for C15H19N, 213.1517;
obsd, 213.1511.
7-(1,1-Dimethylallyl)-7-aza-bicyclo[4.1.0]heptane (4h): Clear liq-
1
uid, 89% yield. H NMR (CDCl3, 300 MHz): δ 5.63 (dd, J ) 17.4,
10.7 Hz, 1H), 5.07-5.00 (m, 2H), 1.71-1.66 (m, 4H), 1.41-1.26 (m,
6H), 1.07 (s, 6H); 13C NMR (CDCl3, 75 MHz): δ 143.2, 113.7, 56.6,
34.9, 31.8, 30.2, 25.7, 25.3, 22.9, 20.8, 14.3. ESI: m/z (%) 166 (M+
+
1, 100), 98 (40); EI-MS: m/z (%) 165 (M+, 2), 150 (27), 96 (100), 84
(67), 69 (87); HR-MS: calcd for C11H19N, 165.1517; obsd, 165.1517.
Methyl 1-(3-Methylbut-2-enyl)aziridine-2-carboxylate (3k): 1H
NMR (CDCl3, 300 MHz): δ 5.32 (t, J ) Hz, 1H), 3.72 (s, 3H), 3.03
(dd, J ) 13.2, 6.9 Hz, 1H), 2.91 (dd, J ) 13.2, 6.9 Hz, 1H), 2.16 (d,
J ) 2.8 Hz, 1H), 2.10 (d, J ) 6.4, 2.8 Hz, 1H), 1.62 (d, J ) 6.4 Hz,
1H), 1.73 (s, 3H), 1.61 (s, 3H); 13C NMR (CDCl3, 75 MHz): δ 171.6,
135.5 120.5, 57.9, 52.3, 36.9, 34.3, 25.9, 18.3. EI-MS: m/z (%) 170
(M+ + 1, 1) 168 (1), 154 (31), 110 (61), 83 (23), 69 (100), 55 (29);
HR-MS: calcd for C9H16NO2, 170.1181; obsd, 170.1176.
1
7-(1-Phenylallyl)-7-aza-bicyclo[4.1.0]heptane (4d): H NMR (CDCl3,
300 MHz): δ 7.41-7.36 (m, 2H), 7.33-7.27 (m, 2H), 7.24-7.18 (m,
1H), 5.97 (ddd, J ) 17.0, 10.2, 6.6 Hz, 1H), 5.19 (ddd, J ) 17.1, 1.8,
1.3 Hz, 1H), 5.05 (ddd, J ) 10.2, 1.8, 1.0 Hz, 1H), 2.89 (d, J ) 6.6
Hz, 1H), 1.87-1.53 (m, 6H), 1.49-1.37 (m, 2H), 1.24-1.13 (m, 2H);
13C NMR (CDCl3, 75 MHz): δ 142.5, 140.6, 128.2, 127.4, 126.9, 114.6,
77.2, 38.1, 38.0, 24.7, 24.6, 20.7, 20.6. ESI: m/z (%) 214 (M+ + 1,
100) + 117 (30).
1-Cinnamyl-2-methylaziridine (3e): Clear oil, 84% yield. 1H NMR
(CDCl3, 300 MHz): δ 7.36 (d, J ) 7.0 Hz, 2H), 7.30-7.25 (m, 2H),
7.18 (t, J ) 7.2 Hz, 1H), 6.55 (d, J ) 15.9 Hz, 1H), 6.30 (dt, J ) 16.2,
5.7 Hz, 1H), 2.98 (dddd, J ) 15.0, 14.4, 5.7, 1.5 Hz, 2H), 1.52 (d, J
) 3.7 Hz, 1H), 1.43-1.36 (m, 1H), 1.26 (d, J ) 6.3 Hz 1H), 1.20 (d,
J ) 5.4 Hz, 3H); 13C NMR (CDCl3, 75 MHz): δ 137.2, 131.0, 128.5,
Methyl 1-(2-Methylbut-3-en-2-yl)aziridine-2-carboxylate (4k): 1H
NMR (CDCl3, 300 MHz): δ 5.56 (dd, J ) 17.4, 10.8 Hz, 1H), 5.17
(dd, J ) 10.8, 1.2 Hz, 1H), 5.11 (dd, J ) 17.4, 1.2 Hz, 1H), 3.72 (s,
3H), 2.24 (dd, J ) 6.3, 2.9 Hz, 1H), 1.98 (dd, 2.9, 1.2 Hz, 1H), 1.79
(dd, 6.3, 1.2 Hz, 1H), 1.21 (s, 3H), 1.19 (s, 3H); 13C NMR (CDCl3, 75
MHz): δ 171.9, 139.7, 116.1, 57.0, 52.1, 30.8, 27.7, 25.8, 25.4; ESI:
m/z (%) 170 (M+ + 1, 80), 102 (100); ESI(QStar) calcd for C9H16NO2,
170.1175; obsd, 170.1170.
1-(2-Methylbut-3-en-2-yl)-2-phenylaziridine (4m): Clear oil, 72%
yield. 1H NMR (CDCl3, 300 MHz): δ 7.30-7.22 (m, 4H), 7.20-7.14
(m, 1H), 5.68 (dd, J ) 17.3, 11.1 Hz, 1H), 5.12 (dd, J ) 17.3, 1.5 Hz,
1H), 5.10 (dd, J ) 11.1, 1.5 Hz, 1H), 2.56 (dd, J ) 6.4, 3.1 Hz, 1H),
1.85 (dd, J ) 6.4, 1.0 Hz, 1H), 1.62 (dd, J ) 3.1, 1.0 Hz, 1H), 1.20 (s,
3H), 1.19 (s, 3H); 13C NMR (CDCl3, 75 MHz): δ 141.5, 141.2, 128.3,
126.7, 126.6, 115.0, 56.9, 33.9, 30.5, 26.0, 25.7. ESI: m/z (%) 188
(M+ + 1, 100), 120 (75); EI-MS: m/z (%) 186 (M+ - 1, 2), 146 (8),
118 (94), 91 (100); HR-MS: calcd for C13H16N, 186.1282; obsd,
186.1286.
(E)-7-(3,7-Dimethylocta-2,6-dienyl)-7-aza-bicyclo[4.1.0]heptane
(3n): 1H NMR (CDCl3, 300 MHz): δ 5.33-5.29 (m, 1H), 5.08-5.13
(m, 1H), 2.85 (d, J ) 6.3 Hz, 2H), 2.17-1.99 (m, 4H), 1.85-1.72 (m,
4H), 1.68 (s, 3H), 1.61 (s, 3H), 1.57 (s, 3H), 1.50-1.49 (m, 2H), 1.40-
1.26 (m, 2H), 1.20-1.09 (m, 2H); 13C NMR (CDCl3, 75 MHz): δ
136.7, 131.6, 124.4, 122.4, 58.8, 39.8, 38.2, 26.7, 25.9, 24.8, 20.8, 17.9,
16.8. ESI: m/z (%) 234 (M+ + 1, 100), 98 (8); EI-MS: m/z (%) 233
(M+, 3), 218 (5), 190 (3), 164 (8), 150 (8), 136 (6), 121 (4), 110 (11),
96 (100), 81 (14), 69 (39), 55 (7); HR-MS: calcd for C16H27N,
233.2143; obsd, 233.2147.
7-(3,7-Dimethylocta-1,6-dien-3-yl)-7-aza-bicyclo[4.1.0]heptane
(4n): 1H NMR (CDCl3, 300 MHz): δ 5.55 (dd, J ) 11.0, 17.6 Hz,
1H), 5.15-5.09 (m, 1H), 5.09 (dd, J ) 1.8, 11.0 Hz, 1H), 5.03 (dd, J
) 1.8, 17.6 Hz, 1H), 2.08-1.99 (m, 4H), 1.70-1.65 (m, 7H), 1.60 (s,
3H), 1.52-1.43 (m, 2H), 1.43-1.34 (m, 2H), 1.19-1.06 (m, 2H), 0.98
(s, 3H); 13C NMR (CDCl3, 75 MHz): δ 141.5, 131.2, 125.4, 115.1,
59.2, 41.8, 30.1, 29.7, 25.9, 25.3, 25.2, 23.1, 21.3, 20.9, 20.8, 17.8.
ESI: m/z (%) 234 (M+ + 1, 100), 98 (20); EI-MS: m/z (%) 233 (M+,
11), 218 (12), 190 (27), 164 (11), 150 (100), 136 (29), 121 (18), 108
(24), 96 (93), 81 (45), 69 (82), 55 (27); HR-MS: calcd for C16H27N,
233.2143; obsd, 233.2149.
127.3, 127.2, 126.3, 62.7, 34.6, 34.4, 18.4. ESI: m/z (%) 174 (M+
+
1, 30), 117 (100); EI-MS: m/z (%) 173 (M+, 18), 130 (15), 117 (68),
104 (18), 91 (32), 77 (13), 56 (100); HR-MS: calcd for C12H15N,
173.1204; obsd, 173.1206.
Methyl 1-Cinnamylaziridine-2-carboxylate (3f): Clear oil, 88%
yield. 1H NMR (CDCl3, 300 MHz): δ 7.37 (d, J ) 7.5 Hz, 1H), 7.31
(t, J ) 7.5 Hz, 2H), 7.23 (t, J ) 7.3 Hz, 1H), 6.54 (d, J ) 16.1 Hz,
1H), 6.31 (dt, J ) 16.1, 6.0 Hz, 1H), 3.73 (s, 3H), 3.19 (ddd, J ) 13.7,
6.0, 1.3 Hz, 1H), 3.06 (ddd, J ) 13.7, 6.0, 1.3 Hz, 1H), 2.23 (d, J )
3.3 Hz, 1H), 2.17 (dd, J ) 6.4, 3.3 Hz, 1H), 1.69 (d, J ) 6.4 Hz, 1H);
13C NMR (CDCl3, 75 MHz): δ 171.3, 136.8, 132.6, 128.7, 127.8, 126.5,
125.7, 62.4, 52.4, 37.3, 34.5. ESI: m/z (%) 218 (M+ + 1, 20), 117
(100); EI-MS: m/z (%) 217 (M+, 7), 158 (59), 130 (74), 117 (100), 91
(42); HR-MS: calcd for C13H15NO2, 217.1102; obsd, 217.1101.
(E)-Methyl 1-(2-(Methoxycarbonyl)-3-phenylallyl)aziridine-2-car-
boxylate (3g): Clear oil, 84% yield. Compound isolated as an
inseparable mix of regioisomers. Major regioisomer: 1H NMR (CDCl3,
300 MHz): δ 7.89 (s, 1H), 7.67 (d, J ) 7.5 Hz, 2H), 7.42-7.35 (m,
3H), 3.83 (s, 3H), 3.71 (s, 3H), 3.42 (d, J ) 12.3 Hz, 2H), 3.29 (d, J
) 12.3 Hz, 2H), 2.41 (m, 1H), 2.22 (m, 1H), 1.95 (d, J ) 6.4 Hz, 1H);
13C NMR (CDCl3, 75 MHz): δ 171.3, 168.3, 143.8, 134.8, 130.1, 129.2,
128.5, 72.2, 55.0, 52.1, 52.0, 37.1, 34.8. Minor regioisomer, mix of
diasteriomers: 1H NMR (CDCl3, 300 MHz): δ 7.34-7.23 (m, 1H),
6.40 (s, 0.2H), 6.32 (s, 0.2H), 3.66 (s, 0.6H), 3.62 (s, 0.6H), 2.32 (m,
0.2H), 2.26-2.24 (m, 0.2H), 1.82 (d, 0.2H); 13C NMR (CDCl3, 75
MHz): δ 170.8, 166.4, 141.3, 140.0, 128.4, 127.7, 127.7, 126.5, 60.3,
51.8, 37.2, 35.4, 20.9, 14.2. ESI: m/z (%) 276 (M+ + 1, 100), 175
(80); EI-MS: m/z (%) 275 (M+, 9), 244 (12), 216 (92), 188 (100), 156
(20), 115 (85); HR-MS: calcd for C15H17NO4, 275.1157; obsd,
275.1150.
7-(3-Methylbut-2-enyl)-7-aza-bicyclo[4.1.0]heptane (3h): In a
flame-dried 50-mL one-neck round-bottom flask, equipped with
magnetic stir bar and septum, were placed 1a (200 µL, 1.95 mmol),
K2CO3 (540 mg, 3.90 mmol), and dry acetone (20 mL). Prenyl bromide
(227 µL, 1.95 mmol) was slowly added to the solution via syringe at
room temperature. The resulting solution was stirred under a stream
of nitrogen at room temperature for 30 min, when GC analysis showed
no remaining cyclohexene imine (1a). Solvent was removed in vacuo,
and the resultant yellow oil was purified by flash chromatography
7-(1,3-Diphenylallyl)-7-aza-bicyclo[4.1.0]heptane (3p): White solid,
97% yield (mp of racemate ) 79.5-81.0 °C), 97% ee as measured by
HPLC (AD column, 1 mL/min, 99:1 hexanes/i-PrOH, t ) 4.5 and 5.0
min. 97% ee - 4.5 min enantiomer, using (R)-BINAP). 1H NMR
(CDCl3, 300 MHz): δ 7.45 (d, J ) 6.9 Hz, 2H), 7.39-7.17 (m, 8H),
9
17528 J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005

Mechanism of Palladium-Catalyzed Allylic Amination
A R T I C L E S
6.57 (d, J ) 15.8 Hz, 1H), 6.35 (dd, J ) 15.8, 7.0 Hz, 1H), 3.08 (d, J
) 7.0 Hz, 1H), 1.94-1.58 (m, 6H), 1.50-1.40 (m, 2H), 1.26-1.12
(m, 2H); 13C NMR (CDCl3, 75 MHz): δ 142.8, 137.4, 132.4, 129.8,
128.6, 128.4, 127.6, 127.5, 127.1, 126.7, 76.8, 38.3, 38.3, 24.9, 24.8,
20.9, 20.8. ESI: m/z (%) 193 (100), 115 (10); EI-MS: m/z (%) 289
(M+, 7), 207 (41), 193 (100), 178 (27), 115 (89), 96 (68); HR-MS:
calcd for C21H23N, 289.1830; obsd, 289.1824.
1-(1,3-Diphenylallyl)-2-methylaziridine (3q): White solid, 97%
yield (mp 39-40 °C, racemate is a clear oil), 98% ee as measured by
HPLC (AD column, 1 mL/min, 95:5 hexanes/i-PrOH, t ) 5.0, 6.3,
6.9, and 7.2 min. 98% ee - 5.0 and 6.3 min enantiomers, using (R)-
BINAP). Isolated as a mixture of diastereomers. 1H NMR (CDCl3, 300
MHz): δ 7.46-7.18 (m, 20H), 6.66 (d, J ) 16.0 Hz, 1H), 6.55 (d, J
) 16.0 Hz, 1H), 6.39 (t, J ) 16.2 Hz, 1H), 6.39 (dd, J ) 16.0, 2.8 Hz,
1H), 3.05 (dd, 2H), 1.70 (d, J ) 3.2 Hz, 1H), 1.68-1.61 (m, 1H), 1.59
(d, J ) 5.3 Hz, 1H), 1.56-1.52 (m, 2H), 1.41 (d, J ) 6.3 Hz, 1H),
1.30 (d, J ) 5.6 Hz, 3H), 1.19 (d, J ) 5.0 Hz, 3H); 13C NMR (CDCl3,
100 MHz): δ 142.4, 142.0, 137.1, 137.0, 131.8, 131.6, 130.1, 129.8,
128.5, 128.5, 128.4, 128.4, 127.6, 127.5, 127.4, 127.3, 127.3, 127.2,
126.5, 126.5, 76.8, 76.7, 35.4, 35.0, 34.5, 34.4, 18.5, 18.2. ESI: m/z
(%) 193 (100), 115 (45); EI-MS: m/z (%) 249 (M+, 19), 207 (43), 193
(100), 178 (35), 115 (76); HR-MS: calcd for C18H19N, 249.1517; obsd,
249.1499.
7-(5-Methylcyclohex-2-enyl)-7-aza-bicyclo[4.1.0]heptane (3s): Clear
oil, 70% yield. Isolated as an 82:18 mixture of diastereomers. 1H NMR
(CDCl3, 300 MHz): δ 5.80-5.67 (m, 1H), 5.63 (dm, 1H), 2.05-1.99
(m, 1H), 1.92-1.71 (m, 6H), 1.70-1.57 (m, 2H), 1.55 (m, 2H), 1.41-
1.24 (m, 3H), 1.21-1.08 (m, 2H), 0.97 (d, J ) 6.3 Hz, 3H); 13C NMR
(CDCl3, 75 MHz): δ 129.1, 128.4, 67.3, 38.0, 37.4, 36.8, 34.4, 28.2,
25.0, 24.9, 22.4, 20.8, 20.7. Spectroscopic data assigned to the minor
diastereomer: 1H NMR (CDCl3, 300 MHz): δ 0.94 (d, J ) 6.7 Hz,
3H); 13C NMR (CDCl3, 75 MHz): δ 128.9, 127.6, 63.7, 37.2, 36.6,
33.8, 25.2, 24.8, 24.7, 21.4, 20.9. ESI: m/z (%) 192 (M+ + 1); EI-
MS: m/z (%) 191 (M+, 9), 149 (8), 96 (100), 69 (51), 55 (16); HR-
MS: calcd for C13H21N, 191.1674; obsd, 191.1675.
(Z)-Methyl-5-(7-aza-bicyclo[4.1.0]heptan-7-yl)cyclohex-3-enecar-
boxylate (3t): Clear oil, 80% yield. See Supporting Information for
peak assignments. 1H NMR (CDCl3, 500 MHz): δ 5.76-5.73 (m, 1H),
5.67 (dm, J ) 10.1 Hz, 1H), 3.68 (s, 3H), 2.56-2.49 (m, 1H), 2.29-
2.25 (m, 2H), 2.21 (dm, J ) 12.4 Hz, 1H), 1.96-1.93 (m, 1H), 1.81-
1.75 (m, 4H), 1.69 (dt, J ) 12.7, 10.4, 1H), 1.59 (m, 2H), 1.38-1.31
(m, 2H), 1.19-1.11 (m, 2H); 13C NMR (CDCl3, 125 MHz): δ 175.5,
129.3, 126.5, 66.2, 51.7, 38.8, 37.5, 36.8, 31.8, 28.0, 24.8, 20.6, 20.5.
ESI: m/z (%) 236 (M+ + 1, 100); EI-MS: m/z (%) 236 (M+ + 1, 8),
96 (100), 79 (28), 69 (41), 55 (10); HR-MS: calcd for C14H21NO2,
235.1572; obsd, 235.1567.
(E)-Methyl 1-(1,3-diphenylallyl)aziridine-2-carboxylate (3r): White
solid, 79% yield. Isolated as a mixture of diastereomers. 1H NMR
(CDCl3, 300 MHz): δ 7.52-7.21 (m, 17H), 6.64 (d, J ) 15.8 Hz,
1H), 6.62 (d, J ) 15.8 Hz, 0.75H), 6.46 (dd, J ) 15.8, 7.3 Hz, 0.75H),
6.44 (dd, J ) 16.0, 7.0 Hz, 1H), 3.76 (s, 3H), 3.72 (s, 2H), 3.20 (d, J
) 7.0 Hz, 2H), 2.40 (d, J ) 3.1 Hz, 1H), 2.38 (d, J ) 3.2 Hz, 0.75H),
2.28 (d, J ) 3.1 Hz, 1H), 2.26 (d, J ) 3.2 Hz, 0.75H), 1.93 (d, J ) 6.6
Hz, 0.75H), 1.78 (d, J ) 6.6 Hz, 1H); 13C NMR (CDCl3, 75 MHz): δ
171.1, 170.9, 141.0, 140.9, 136.7, 136.6, 130.9, 130.9, 130.6, 130.4,
128.7, 128.6, 127.8, 127.73, 127.68, 127.61, 127.2, 126.6, 126.6, 76.5,
76.4, 52.3, 52.2, 37.9, 37.2, 34.9, 34.2. ESI: m/z (%) 193 (100), 115
(15); EI-MS: m/z (%) 293 (M+, 12), 292 (M+ - 1, 23), 234 (63), 193
(94), 178 (31), 115 (100), 91 (48); HR-MS: calcd for C19H18NO2,
292.1337; obsd, 292.1339.
Acknowledgment. We thank NSERC, CFI, ORDCF, ACS-
PRF, and the University of Toronto for financial support.
I.D.G.W. acknowledges the J. Warren Flanagan Ontario Gradu-
ate Scholarship. Ms. Karen Konieczny is acknowledged for the
preparation of 1g and 3k. Dr. Evgeniy Blyumin and Mr. Greg
Chen are acknowledged for gifts of 1e and 1d, respectively.
We thank Professor John Powell for helpful discussions.
Supporting Information Available: Full experimental pro-
cedures and characterization data for all unknown compounds.
This material is available free of charge via the Internet at
http://pubs.acs.org.
JA055288C
9
J. AM. CHEM. SOC. VOL. 127, NO. 49, 2005 17529