Achieving control over the branched/linear selectivity in palladium-catalyzed allylic amination

Article abstract of DOI:10.1021/jo3025253

Palladium-catalyzed reaction of unsymmetrical allylic electrophiles with amines gives rise to regioisomeric allylic amines. We have found that linear products result from the thermodynamically controlled isomerization of the initially formed branched products. The isomerization is promoted by protic acid and active palladium catalyst. The use of base shuts down the isomerization pathway and allows for the preparation and isolation of branched allylic amines. Solvent plays a key role in achieving high kinetic regioselectivity and in controlling the rate of isomerization. The isomerization can be combined with ring-closing metathesis to afford the synthesis of exocyclic allylic amines from their endocyclic precursors.

Full text of DOI:10.1021/jo3025253

Subscriber access provided by University of Sussex Library
Article
Achieving control over the branched/linear
selectivity in palladium-catalyzed allylic amination
Igor Dubovyk, Iain D. G. Watson, and Andrei K Yudin
J. Org. Chem., Just Accepted Manuscript • DOI: 10.1021/jo3025253 • Publication Date (Web): 18 Jan 2013
Downloaded from http://pubs.acs.org on January 20, 2013
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
The Journal of Organic Chemistry is published by the American Chemical Society.
1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
Achieving control over the branched/linear selectivity in palladium-catalyzed allylic
amination
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Igor Dubovyk, Iain D. G. Watson, and Andrei K. Yudin*
Davenport Research Laboratories, Department of Chemistry, The University of Toronto,
80 St. George Street, Toronto, Canada, M5S 3H6
ayudin@chem.utoronto.ca
Solvent-dependent selectivity:
R3 R4
R1
N
H R2
1 2
R R NH
X
R3 R4
PdLn
PdLn
R1
+
+
R3
N
R3
CH Cl
H R2
2
2
THF
R1
X
R4
N
R2
R4
OAc
H
X
NO selectivity
HIGH selectivity
R3
Application to cyclic systems:
Pd0L
R4
R3 R4
Pd0Ln
TFA
n
R1
TFA
R1
N
N
NR
2
CH Cl2
THF
R
R
Abstract
Palladiumꢀcatalyzed reaction of unsymmetrical allylic electrophiles with amines gives
rise to regioisomeric allylic amines. We have found that linear products result from the
thermodynamically controlled isomerization of the initiallyꢀformed branched products.
The isomerization is promoted by protic acid and active palladium catalyst. The use of
base shuts down the isomerization pathway and allows for the preparation and isolation
of branched allylic amines. Solvent plays a key role in achieving high kinetic
1
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 2 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
regioselectivity and in controlling the rate of isomerization. The isomerization can be
combined with ringꢀclosing metathesis to afford the synthesis of exocyclic allylic amines
from their endocyclic precursors.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Introduction
Controlling regioselectivity has been an important goal in transition metalꢀ
1
,2
catalyzed allylic substitution. Branched allylic amines have attracted particular interest
due to the frequent occurrence of these structural fragments among natural products and
pharmaceuticals. In addition, branched allylic amines are valuable building blocks in
3
chemical synthesis. The challenge of branched selectivity has been approached using
4
5
metals such as iridium and rhodium. The corresponding catalysts allow for the selective
formation of branched products from the corresponding amines and allylic acetates or
carbonates. The use of palladium catalysts in the amination of allylic acetates and
carbonates typically leads to the formation of the more thermodynamically stable linear
1
products.
Branched product formation with palladium has been observed and reported in the
literature since 1981. Åkermark and coꢀworkers were the first to describe amination of
crotyl and prenyl palladium chloride complexes to be reversible, suggesting that the
linear product was formed as a result of isomerization of the kinetic branched product
6
(Scheme 1).
2
ACS Paragon Plus Environment

Page 3 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
Scheme 1
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The original explanation suggested
σꢀcomplexes as reactive species, but
7
8a
8b
subsequent studies provided no evidence supporting this claim. Hou, Hayashi, and
8c
Faller hypothesized that certain bidentate ligands direct amines to the more substituted
site on the palladium πꢀallyl complex, which ultimately results in the irreversible
formation of branched products. In addition to catalyst modifications, there were also
instances when the reactants themselves were biased to give high branched selectivities.
Thus, Trost and coworkers have shown that ring size could be used to control selectivity
9
in intramolecular allylic amination.
Branched 2ꢀvinyl pyrrolidines form kinetically faster and are more
thermodynamically stable than the corresponding linear allylic tetrahydroazepines.
Certain substituents on the allylic substrates such as trifluoromethyl group may also favor
1
0
the formation of branched isomers. Our own investigations in allylic amination started
with a finding that, unlike typical branched allylic amines, branched allylic aziridines are
stable against branchedꢀtoꢀlinear (b/l) isomerization. Despite the fact that the origins of
this stability are still not well understood, it became clear that the regioselectivity can be
1
1
controlled by the selection of an appropriate nucleophile. Subsequently, Hartwig and
coꢀworkers found that hydrazine and hydroxylamine derivatives can also be used to form
1
2
the corresponding branched products. We later discovered that a wider range of amines
can be allylated with high b/l ratios provided that the pathway for the protonꢀdriven
1
3
isomerization of the branched allylic amine to the linear isomer is shut down. The
3
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 4 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
present contribution documents the scope of this process and its mechanistic foundation.
In the course of our studies aimed at delineating the role of base additives, we have
stumbled upon a curious effect of THF in this chemistry, which further highlights
palladium’s versatility. This effect also allowed us to optimize reaction conditions for an
intramolecular rearrangement of cyclic allylic amines.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Results
In the course of our studies in the area of palladiumꢀcatalyzed allylic amination
we discovered that the presence of base can alter the regioselectivity of this reaction,
channeling it towards the branched products. At the outset, we noted that during the
amination of prenyl acetate with two equivalents of piperidine, one equivalent of this
amine acted as the base, giving piperidinium acetate salt precipitate and the
corresponding branched allylated product (4:1 branched/linear selectivity). This effect
was limited to piperidine and relied on the removal of acid from the reaction mixture by
precipitation. Starting with an equimolar mixture of 1a and prenyl acetate gave full
conversion with the corresponding linear allylic amine as the major product (Table 1,
entry 1), which indicates that the presence of base in solution is critical to achieving high
11
b
branched selectivity. With 2 equivalents of 1a (Table 1), the corresponding linear
t
product was formed exclusively (Table 1, entry 2), whereas 2,6ꢀdiꢀ butylpyridine, a
14
commonly used proton scavenger, gave a very low conversion (Table 1, entry 3). We
4
ACS Paragon Plus Environment

Page 5 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
carried out an extensive search for base additives that could lead to high branched
selectivity without concomitant base allylation. Pyridine and Hunig’s base (Table 1,
entries 4 and 5) had almost no effect on the reaction. These additives preferentially gave
the linear product, whereas NꢀMethylmorpholine gave linear product with 3:1 selectivity
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
t
(Table 1, entry 6). Other bases such as triethylamine, DABCO, TMEDA and BuOK
(
Table 1, entries 7ꢀ12) gave the branched product with a low 2:1 selectivity even when a
tenfold excess of base was used. The reactions with inorganic bases such as K CO , NaH
2
3
and Bu NOH (Table 1, entries 13ꢀ15) produced very little allylic amine product because
4
of the background reaction with the acetate. Finally, DBU (Table 1, entry 16) was found
to give the branched product with a 19:1 selectivity. DBU has a pKaH of 16.6 in THF,
1
5
which is substantially higher than that of other organic bases such as triethylamine
t
(
pK 12.5). DBN and P ꢀ Buꢀtrisꢀ(tetramethylene)phosphazine (Table 1, entries 17 and
aH
1
1
8) showed the same selectivity for the branched product as did DBU, although with
t
DBN the reaction took longer to reach completion, whereas with P ꢀ Buꢀtrisꢀ
1
(
tetramethylene)phosphazine the reaction was not as clean, probably due to the
decomposition of this base. Optimal conditions were developed using 1a (1 equivalent)
and prenyl acetate (1 equivalent). Full conversion was reached after 17 hours using 1 mol
3
%
of [(η ꢀallyl)PdCl] as the source of palladium, 4 mol % of P(OEt) as the ligand, and 1
2
3
equivalent DBU as the base (Table 1 entry 16).
Table 1. Optimization of base additives
Entry
Base
Equivalents
Time, h
GC Conv %
Regioselectivity
b/l)
(
5
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 6 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
1
2
3
4
5
6
7
8
9
ꢀ
ꢀ
24
24
24
24
24
24
24
96
96
24
24
24
24
24
24
24
24
24
100
100
20
0:100
1:99
ꢀ
tetrahydroisoquinoline
3
1
1
1
1
1
1
1
1
1
1
2
2
1
1
1
1
t
2,6ꢀdiꢀ Bupyridine
Pyridine
100
100
100
100
50
1:99
1:99
1:3
2:1
2:1
2:1
2:1
2:1
2:1
ꢀ
Hunig’s Base
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
NꢀMethylmorpholine
Et
Et
Et
3
3
3
N
N
N
50
1
0
DABCO
TMEDA
100
100
15
11
t
1
1
1
1
2
3
4
5
BuOK
K
2
CO
NaH
Bu NOH
3
100
100
100
100
91
ꢀ
4
ꢀ
16
DBU
DBN
19:1
19:1
19:1
1
7
t
18
Phosphazene base P
1
ꢀ Buꢀtrisꢀ
100
(
tetramethylene)
1
6
Our system was found to exhibit a strong solvent effect. Table 2 shows that the
branched products are favored in THF, whereas the selectivity drops substantially in less
polar solvents such as dichloromethane. While 2ꢀmethyl THF (Table 2, entry 3) was
found to provide similar selectivity to the reaction in THF, 2,5ꢀdimethyl THF (Table 2,
entry 4) showed a decrease in the b/l selectivity. The reaction in less polar THP displayed
1
7
a preference for the branched product, albeit with lower selectivity (Table 2, entry 5).
Table 2. Solvent effects on regioselectivity in palladium-catalyzed allylic amination
6
ACS Paragon Plus Environment

Page 7 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
Entry
Solvent
THF
b/l
1
2
3
4
5
95:5
1:1
2 2
CH Cl
2ꢀmethyl THF
2,5ꢀdimethyl THF
THP
96:4
78:22
85:15
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Table 3 shows the scope of reactivity of secondary as well as primary amines with
prenyl acetate. Primary amines exhibited different reactivity depending on whether or not
DBU was present in the system. With no DBU added, the allylation of benzylamine with
prenyl acetate gave mixtures of linear monoꢀ and bisꢀallylated products, whereas with
DBU present, only the monoallylated branched product 3ad (Table 3, entry 4) was
observed even when a twofold excess of prenyl acetate was used (Scheme 2). Most other
amines showed high regioselectivity in the presence of DBU, except for aniline (Table 3,
entry 6). The use of an electronꢀdefficient P(OEt) as a ligand made the reaction tolerant
3
to functional groups such as aryl bromide (Table 3, entry 15). Switching to more
electronꢀpoor amines (Table 3, entries 17 – 27) resulted in full recovery of prenyl acetate
°
even at 50 C. Surprisingly, amine 1q gave no conversion.
Table 3. Substrate scope of amination of prenyl acetate
entry
1
amine
product
%yield entry
amine
product
%yield
(b/l)
a
a
(b/l)
d
92
15
85
(19:1)
(19:1)
ꢀ
7
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 8 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
d
b
b
2
3
80
16
17
ꢀ
(99:1)
b,c,d
81
ꢀ
ꢀ
(
19:1)
d
b
b
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
5
90
18
19
ꢀ
ꢀ
ꢀ
(9:1)
b,d
85
ꢀ
(
19:1)
b,d
b
b
6
7
71
(
20
21
ꢀ
ꢀ
ꢀ
4:1)
b,d
80
(
ꢀ
9:1)
b,d
b
b
8
9
87
22
23
ꢀ
ꢀ
ꢀ
(19:1)
b,d
87
ꢀ
(
99:1)
b,d
b
1
0
1
S
S
61
24
25
ꢀ
ꢀ
ꢀ
ꢀ
(4:1)
N
3aj
b,d
b
1
76
(9:1)
b
b
12
ꢀ
ꢀ
26
27
ꢀ
ꢀ
ꢀ
ꢀ
b
b
1
3
71
(
24:1)
b
14
54
N
(12:1)
H
O
O
3
an
a
b
o
c
isolated yields of branched products. b/l obtained by GC. The reaction was carried out at 50 C. NMR yield. The product could only
be purified by distillation, and was contaminated with P(OEt) . These entries have been previously reported in our communication.
3
d
13
8
ACS Paragon Plus Environment

Page 9 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
We examined the lack of over allylation with primary amines (Table 3). It
transpired that substituents next to the nitrogen atom produced a profound effect on the
reaction outcome. Thus, sterically congested cisꢀ2,6ꢀdimethyl piperidine 1l (Table 3,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
°
entry 12) showed no conversion even at 50 C. Neither was any conversion observed
when the branched allylic amine 3ad (Table 3, entry 16) was reꢀsubjected to the reaction
conditions in the presence of DBU. The fact that without DBU primary amines can
produce bisꢀallylated products such as 3ad’, while amines with substituents next to
nitrogen are completely unreactive, suggests that the kinetic branched product has to
undergo isomerization before another allylation can occur (Scheme 3). When the
monoallylated branched product cannot isomerize, it is too hindered to react with another
molecule of prenyl acetate leading to the monosubstituted branched product (Scheme 3).
Scheme 2
Scheme 3
We later decided to explore the reactivity of allylic carbonates. Allyl enol
1
8
19
carbonates were previously explored by Stoltz and Trost towards allylation of
9
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 10 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
enolates. The formation of stable palladium enolates took place upon extrusion of CO2
from the carbonate (Scheme 4). The formation of CO and palladium enolate is the
2
driving force for the decarboxylation step. When we replaced prenyl acetate 2a with the
corresponding carbonate, there was no change in selectivity for the branched product.
Similarly, when no DBU was used, only the linear product was observed. One
explanation for this result is that ethyl carbonic acid that forms in situ does not rapidly
decompose into ethanol and water, but rather remains in solution long enough to promote
product protonation and the b/l isomerization in the absence of DBU (Scheme 5).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Alternatively, if ethyl carbonic acid does decompose to release CO , ethanol could
2
activate the branched allylic amine to ionization via hydrogen bonding. This latter system
has been described by Zhang in palladiumꢀcatalyzed allylation of ketones with allylic
20
amines in protic solvents.
Scheme 4
Scheme 5
While we were able to control regioselectivity with prenyl acetate, achieving high
levels of selectivity with disubstituted substrates was not as straightforward. The reaction
10
ACS Paragon Plus Environment

Page 11 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
outcome strongly depended on the nature of the ligand and the acetate. When hexꢀ2ꢀ
enylacetate 2d was reacted with amine 1a in the presence of DBU and P(OEt) , a 1:4 b/l
3
ratio was detected (Table 4, entry 1). This is in contrast to the reaction between 2d and
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
prenyl acetate 2a, where P(OEt) afforded 19:1 branched selectivity under the same
3
reaction conditions. In order to test whether the inferior selectivity with 2d was kinetic in
origin, we monitored the reaction progress using GC. In the presence of DBU and
P(OEt) , the b/l ratio of 1:4 remained constant throughout the experiment, indicating that
3
isomerization is not occurring under these conditions. This finding suggests that the
presence of DBU only prevents the erosion of the initial regioselectivity, whereas the
electronic effects of the allylic acetate determine this selectivity.
Replacing P(OEt) with other ligands showed a strong variation in selectivity
3
(Table 4). The bidentate ligands (Table 4, entries 3ꢀ7) gave the linear product almost
exclusively. Most monodentate ligands also favoured the linear product, although with
lower selectivity. The highest selectivity for the branched product was obtained with (oꢀ
21
biphenyl)dicyclohexyl phosphine (Table 4, entry 16).
Table 4. Ligand Screening for amination of disubstituted allylic acetates
Entry
Ligand
% Conversion
b/l
1
2
3
4
5
6
P(OEt)
3
100
100
100
100
100
100
1:4
a
P(OEt)
3
0:1
(racꢀBINAP)
dppf
1:32
1:19
1:24
1:24
(S,S)ꢀDIOP
Xantphos
11
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 12 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
7
8
dppp
100
1:24
ꢀ
P(OPh)
3
0
100
9
PPh
3
1:3
1:4
1:19
1:3
ꢀ
10
Pd(PPh
3
)
4
100
1
1
Ph
2
P(Phꢀoxazoline)
CyPPh
PCy
100
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
12
2
100
1
1
1
1
3
4
5
6
3
0
t
( Bu)
2
P(oꢀbiphenyl)
100
1:6
2:1
2:1
t
P Bu
3
40 (2 days)
80 (2 days)
2
Cy P(oꢀbiphenyl)
2
Allyl acetate (1 eq), nucleophile (1 eq), DBU (1 eq), [(πꢀallyl)PdCl] 1 mole %, Monodentates 4 mole%, Bidentates 2 mole %
a
THF 0.5 M, 24 h.
No DBU was used.
The reactions with transꢀhexꢀ2ꢀenylacetate 2d were very slow, and therefore
crotyl acetate 2b was used instead. (oꢀBiphenyl)dicyclohexyl phosphine significantly
improved the selectivity, giving greater than 99:1 b/l ratio for this challenging allylic
acetate (Table 5, entry 6). P(OEt) , the ligand of choice for the trisubstituted substrates,
3
gave inferior 6:1 b/l selectivity with 2b (Table 5, entry 2). When the methyl group on the
crotyl substituent was replaced with a phenyl substituent, however, the conjugated linear
product predominated over the branched regioisomer with a 2.5:1 selectivity (Table 5,
entry 5).
Table 5. Substrate scope of amination with trans-disubstituted allylic acetates
a
entry
1
amine
acetate
product
%yield (b/l)
d
82
(13:1)
c,e
2
100
(6:1)
12
ACS Paragon Plus Environment

Page 13 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
3
4
5
6
83
(6:1)
Et
OAc
2
c
c,d
80
(
2:1)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
66
1:2.6)
(
b,d
84
(99:1)
b,d
7
8
82
(
19:1)
b,d
81
(
9:1)
a
b
o
c
Combined isolated yields of mixtures. b/l obtained by GC. The reaction was carried out at 50 C. Conversion determined by GC.
These results have been previously reported in our communication.
oꢀbiphenyl)PCy
d
13
e
The reaction was performed with P(OEt)3 instead of
(
2
.
We had mentioned earlier that the nature of the reaction solvent was crucial in
establishing high branched regioselectivity (Table 2). In addition to this kinetic effect of
THF, we also discovered the solvent influence on the rate of branched to linear
isomerization. Thus, the product isomerization in the presence of DBU in THF was slow,
and only 30% of product isomerized after one week. In contrast, with DBU in CH Cl ,
2
2
branched allylic amines fully isomerized to form linear products within four days.
Inspired by such observations, we envisioned this isomerization to be enabling in paving
2
2
access to cyclic amines by skeletal isomerizations of allylic amine scaffolds.
Table 6 shows the condition optimization for the isomerizationꢀdriven ring
construction. We were encouraged that in dichloromethane 50% of tetrahydroazepine 4a
was converted to the corresponding 2ꢀprenyl pyrrolidine 5a after 8 hours (Table 6 Entry
2
). 1 equivalent TFA was employed to activate the amine. Interestingly, the addition of
morpholine (10 mol%) gave further boost and pushed the reaction to completion (Table 6
13
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 14 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Entry 3). With no ligand present, the rearrangement of 4a did not occur (Table 6 Entry 4).
In addition, uncatalyzed olefin isomerization was not observed when either 4a or 5a were
used as starting materials. Switching the solvent to THF with or without the addition of
morpholine gave no conversion (Table 6 Entries 1, 5). Reaction with only 5 mol % TFA
gave no conversion with or without morpholine added. This methodology can be
strategically applied to lateꢀstage modifications of complex amines by using amineꢀ
containing fragments simultaneously as nucleophiles and as leaving group precursors.
Once combined with olefin ring closing metathesis, our methodology was shown to find
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
23
immediate application in combinatorial chemistry.
Originally, morpholine was added with the intention to help form the active
catalyst from allyl palladium precursor. Nucleophilic morpholine attacks the πꢀallyl
ligand on palladium (II) leading to the formation of Nꢀallylꢀmorpholine, and a
2
4
catalytically active palladium(0) complex. However, the reaction did achieve 50%
conversion in the absence of morpholine, which suggests that morpholine is not essential
for catalyst activation. For instance, a fraction of starting tertiary cyclic allylic amine can
reduce allylpalladium (II) complex in a similar fashion, forming a quaternary amine as a
result. Another possibility is that morpholine acts as a nucleophilic catalyst in the
reaction. To test if this was the case, morpholine was replaced with Nꢀmethylmorpholine.
Such modification gave full conversion (Table 6, entry 8), ruling out the catalytic role of
morpholine. In addition, morpholineꢀcontaining intermediates have never been isolated or
observed spectroscopically during our studies. Alternatively, the combination of
morpholine and TFA likely forms a buffer, in which morpholinium trifluoroacetate acts
as an active acid to allow for the controlled protonation of the starting tetrahydroazepine.
14
ACS Paragon Plus Environment

Page 15 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
Such careful protonation is necessary to avoid the undesirable protonation of palladium
catalyst or the ligand. Indeed, even in the presence of morpholine, the reaction fails when
a phosphite ligand is replaced by a more basic phosphine (Table 6, entry 9). The substrate
scope is shown in Table 7. All of the starting materials were prepared using metathesis
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
2
5
protocols and other standard reactions.
Table 6. Condition screening for the Aza-allylic rearrangement
a
Entry
Ligand
Solvent
THF
Additive
% Conv
(
10 mol %)
1
2
3
4
5
6
7
8
9
P(OEt)
P(OEt)
P(OEt)
ꢀ
3
ꢀ
ꢀ
0
50
100
0
3
3
CH
CH
CH
2
2
2
Cl
Cl
Cl
2
2
2
morpholine
morpholine
morpholine
ꢀ
P(OEt)
P(OEt)
P(OEt)
P(OEt)
3
3
3
3
THF
0
b
CH
CH
CH
CH
2
2
2
2
Cl
2
Cl
2
Cl
2
Cl
2
0
b
morpholine
Nꢀmethylmorpholine
morpholine
0
100
0
Trost ligand
a
1
b
Conversion was monitored by HꢀNMR. 5mol% acid was used.
Table 7. Reaction scope of the Aza-allylic rearrangement
entry
1
reactant
product
yield
entry
13
reactant
product
S.M.
yield
d
b
97
ꢀ
15
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 16 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
b
2
S.M.
ꢀ
14
S.M.
ꢀ
d
b
b
3
4
94
15
16
S.M.
S.M.
ꢀ
ꢀ
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(E/Z=3.6
1)
:
d
92
(
E/Z=4.6
:
1)
d
d
d
5
6
93
17
18
78
d
97
83
(
1:1)
b
7
S.M.
ꢀ
19
S.M.
ꢀ
b,d
b
8
9
53
20
21
ꢀ
(
2:1)
d
b
71
S.M.
ꢀ
(
3:1)
b,d
c,d
1
0
50
22
96
(2:1)
b,d
1
1
1
2
68
(
1:1)
b,d
S.M.
ꢀ
a
b
Isolated yields. Pdꢀcatalyst/ligand/morpholine/substrate/TFA
=
1/4/10/40/40 in CH
2
Cl
2
(0.5 M) reflux, overnight. Pdꢀ
c
catalyst/ligand/morpholine/substrate/TFA
=
1/4/10/20/20 in DCE (0.5 M) reflux overnight. Pd catalyst/ligand/
d
morpholine/substrate/TFA = 1/4/10/100/ 100 in CH
Cl
2 2
(0.2 M) rt, overnight. These results have been previously reported in our
22
communication.
16
ACS Paragon Plus Environment

Page 17 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
Discussion
We have achieved control over regioselectivity in palladiumꢀcatalyzed allylic
amination by suppressing the protonꢀassisted b/l isomerization, which is responsible for
the observed linear selectivity. This linear selectivity is thermodynamic in nature, and can
be reversed in the presence of DBU. In addition to controlling the b/l isomerization with
base, there are factors that influence the kinetic formation of branched products.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Kinetic solvent effects
Solvent is a key parameter that is responsible for the observed kinetic
regioselectivity in allylic amination. As can be seen from Table 2, the formation of the
branched product is kinetically favoured in THF, whereas in other solvents lower or no
selectivity is observed. To explain the formation of both branched and linear products,
the presence of both πꢀ and σꢀmetal complexes may be considered. πꢀComplexes are
26
known to be in equilibrium with their σꢀisomers, and the extent of this equilibrium
2
7
depends on the nature of the metal, ligand, solvent, and counteranion.
2
8
3
Computational studies performed earlier on prenyl η ꢀallyl palladium complexes
showed that the difference in the activation barriers for the attack at the tertiary and
primary carbons is less than 1 kcal/mol, which is too small of a difference to account for
3
the observed branched kinetic selectivity based on the η ꢀallyl complex alone. Despite the
fact that the computational study did not take into account the nature of solvent, and PH3
ligands were used for simplification, this result suggests a possibility of the presence of
17
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 18 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
another catalytically relevant palladium intermediate, which would account for high
2
9
branched selectivity.
The dependence of regioselectivity on the nature of the solvent (Table 2)
prompted us to examine the solvent effect on the equilibrium between allylpalladium
intermediates. When prenylpalladium chloride dimer was mixed with PPh in THFꢀd ,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
8
1
1a, 30
the corresponding σꢀcomplex was the only species observed by NMR (Scheme 6).
This observation, combined with the fact that high branched selectivity is only observed
in THF, led us to examine whether the involvement of palladium
THF was a possible reason for the high branched selectivity.
σꢀallyl intermediate in
Scheme 6
R
PPh3 (4 equiv)
R
L
Pd
2
Pd
L
X
d8-THF
Cl
R = Me, H
L = phosphine ligand
X = Cl
Table 2 shows a correlation between the solvent nucleophilicity and branched
selectivity in allylic amination. One can conclude that THF alters the reactivity of allyl
3
1
palladium intermediates. Such behavior would explain the selectivity trend in other
ethereal solvents. When THF is replaced by 2,5ꢀdimethylmethylꢀTHF, the oxygen center
becomes less nucleophilic (Table 2, entry 4), and the selectivity for the branched product
drops. Similar result was observed with the lessꢀcoordinating THP. To test this
hypothesis we monitored branched selectivity in THF/CH Cl mixtures of different
2
2
composition (Figure 1). In the presence of P(OEt) , it takes as low as 10% THF in
3
solution to maintain branched selectivity of 9:1. On the other hand, with BINAP, almost
9
0% THF is required to give rise to the branched product. The results of this study
18
ACS Paragon Plus Environment

Page 19 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
support the notion that THF is able to compete for palladium as a ligand. We believe that
3
1
by coordinating to palladium, THF can promote the η to η isomerization, thus altering
3
2
the b/l selectivity of the reaction. Since bidentate ligands stabilize πꢀallyl complexes
much better than monodentate ligands, it is not surprising that with BINAP at least 90%
of THF in dichloromethane is required in order to reach the selectivity observed with
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
P(OEt) . The fact that THF cannot be used in substoichiometric quantities to dictate the
3
selectivity does not necessarily revoke the possibility of coordination of THF to
palladium. Such large amounts of THF may be needed due to the fact that compared to
phosphines, acetate, or chloride, THF is a weak ligand, and high concentrations of it are
3
3
required to make it competitive in terms of binding.
Impact of THF on the Selectivity
1
00
9
8
0
0
70
6
5
4
0
0
0
P(OEt)3
BINAP
30
2
1
0
0
0
0
20
40
60
80
100
%
THF in the THF / CH2Cl2 mixture
Figure 1 Impact of THF on selectivity in the presence of P(OEt) and BINAP ligands.
3
Upon conducting a more detailed study, however, we observed exclusive σꢀ
complex formation in dichloromethane, in which there is no kinetic selectivity at all. This
finding suggests that there is no correlation between the abundance of the σꢀcomplex in
solution and branched selectivity. Therefore, the presence of the σꢀcomplex cannot alone
explain the observed branched selectivity. Even though the πꢀcomplex is the minor of the
19
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 20 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
two allylꢀpalladium intermediates in solution, it is expected to be more reactive, by virtue
3
4
of it being charged. This ionic nature of the πꢀcomplex makes its reactivity strongly
dependent on the polarity of the solvent. Given that dichloromethane has a higher
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
35
dielectric constant than THF, it is conceivable that dichloromethane can preferentially
make the relevant cationic πꢀcomplex more reactive by increasing the extent of its ion
pair dissociation. The nucleophilic attack on this charged intermediate would then give
poor kinetic regioselectivity, which is consistent with our results, and predicted by
computations. On the other hand, in THF the ionic πꢀcomplex exists as a tighter ion pair,
and therefore is expected to be less reactive when compared to the πꢀcomplex in
3
6
dichloromethane.
Scheme 7
Scheme 7 shows possible palladium intermediates that may explain the observed
kinetic selectivity. Intermediate B can convert into A through an intimate ion pair A’,
which upon dissociation would form a solvent separated ion pair A. Amine attack on
intermediate A is expected to yield both branched and linear products because the termini
28
of the πꢀcomplex are not well differentiated. On the other hand, intermediate B is
susceptible to a nucleophilic attack proceeding in an S 2’ fashion at the more substituted
N
terminus, which would lead to the branched allylic amine. In dichloromethane, the
20
ACS Paragon Plus Environment

Page 21 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
interconversion between A and A’ is expected to be fast comparing to the nucleophilic
attack on B. Since intermediate A is charged, and is therefore more reactive towards
nucleophiles than B, the reaction is more likely to proceed through A. Given the
negligible difference in activation energy between the attacks on either terminus of A, the
resulting reaction outcome in dichloromethane suggests that both branched and linear
products in fact come from A. On the other hand, in THF, even though the attack on the
intermediate A is still faster than that on B, the interconversion between A and A’ must
be slow compared to the nucleophilic attack on B. In the presence of DBU, the charged
newlyꢀformed branched kinetic product is irreversibly deprotonated to form the neutral
kinetic product, which is unable to reꢀenter the catalytic cycle. In the absence of DBU the
protonated branched product reenters the catalytic cycle to form complex A’. Therefore,
the solvent dictates kinetic regioselectivity by controlling the accessibility of the
dissociated charged cationic πꢀallyl complex.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The thermodynamic solvent effect
In addition to the ability of the solvent to control kinetic regioselectivity by
controlling the rate of dissociation of reactive palladium intermediates, solvent can also
affect the rate of isomerization by modulating the relative strength of acid in solution.
The fact that in the presence of DBU in THF only 30% of the branched product
isomerizes to give the corresponding linear product after one week, whereas under the
same conditions in dichloromethane full isomerization is observed after only four days,
and, likewise, the fact that azaꢀallylic rearrangement proceeds in dichloromethane, but
+
ꢀ
not in THF, both suggest that the relative acid strength of DBUH OAc and
21
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 22 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
morpholinium trifluoroacetate, respectively, is higher in dichloromethane than in THF. In
THF each acid may exist as an ion pair shielded by the solvent cage, which prevents it
from achieving the proton transfer to the cyclic allylic amine. On the other hand, in
dichloromethane, ionꢀpair dissociation and the proton transfer occur to a much greater
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
7
extent.
The kinetic ligand effect
Table 4 shows the effect of different ligands on kinetic selectivity of amination of
a transꢀdisubstituted allylic acetate in THF. The results indicate that in the presence of
DBU all the bidentate ligands preferentially favour the formation of the linear
regioisomer, whereas the corresponding branched product was given preference with
bulky monodentate ligands. Whereas it is still unclear why all bidentate ligands favour
high kinetic linear selectivity, the trend with monodentate ligands can be explained by
examining the corresponding palladium intermediates shown on Scheme 8.
Scheme 8
Bulky monodentate ligands are more likely to arrange themselves in the transꢀ
fashion favouring the σꢀcomplex, the attack on which is expected to give high branched
selectivity. Bidentate ligands, on the other hand, can only arrange themselves in the cisꢀ
fashion, which would make the formation of the σꢀcomplex improbable, and the reaction
22
ACS Paragon Plus Environment

Page 23 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
would likely proceed through the πꢀcomplex. Even though computations mentioned
earlier predict that the attack on a πꢀcomplex should give both regioisomers with equal
distribution, such computations have not been performed with bidentate ligands, and,
therefore, do not exclude the possibility that with bidentate ligands the attack on the lessꢀ
substituted terminus of the πꢀcomplex is preferred.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
This ligand effect is much more pronounced with disubstituted allylic acetates
than it is with the trisubstituted ones. This is because in the πꢀcomplex derived from
prenyl acetate one of the methyl substituents will always be placed in the antiꢀposition
imposing additional energy cost due to the 1,3ꢀallylic strain. In contrast, the πꢀcomplex
derived from a disubstituted acetate can avoid such cost by placing its largest substituent
in the syn position. As a result, the formation of a πꢀcomplex is more feasible with
disubstituted acetates, and therefore the nature of the ligand becomes an important factor
in imposing selectivity (Table 5, entries 1 and 2).
1
In addition, the size of the substituent R on a disubstituted allylic acetate
1
(Scheme 8) also has some impact on the regioselectivity. Table 5 shows that as R
1
becomes larger than R = Me, the regioselectivity drops and finally becomes reversed
1
when R = Ph (Table 5, entries 1,3,4,5). Additionally, the rate of reaction also slows
1
down. Both of these factors suggest that when R becomes large, the rate of attack on the
σꢀcomplex becomes smaller, while the rate of attack on the πꢀcomplex becomes more
significant, and eventually overrides the effects of the ligand.
The original position of the leaving group
23
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 24 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Testing our system for the presence of the memory effect has supported the
involvement of the πꢀallyl complex. When acetate 2a was replaced with its isomer 2l, no
change in selectivity took place. When the reaction of 2l was monitored by GC, the
formation of the acetate 2a was observed (Scheme 9). Furthermore, without the
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
8
nucleophile, acetate 2l isomerized into prenyl acetate 2a.
Scheme 9.
In addition, when the deuterated allyl acetate was used, our system showed
deuterium scrambling in the product (Scheme 10). This observation is indicative of the
3
9
absence of a memory effect that is known to occur with carbonꢀbased nucleophiles. The
absence of the memory effect always suggests that there is a common intermediate, and is
consistent with the presence of a symmetrical πꢀcomplex, which again supports the fact
that the πꢀcomplex is involved even though it is not observed spectroscopically.
Scheme 10
The driving force in the isomerization of cyclic allylic amines
24
ACS Paragon Plus Environment

Page 25 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
We have turned back to our test substrate in attempts to figure out which of the
two factors (the size of the ring or the substitution on the alkene), was tipping the balance
in favour of the product. Subjecting tetrahydroazepines 4c and 4e to the reaction
conditions yielded the corresponding 2ꢀvinylic pyrrolidines with diꢀ and monosubstituted
alkenes, respectively, in high yields (Table 7, entries 3 and 5). Similar yields were
observed when analogous hexahydroazocines 4d and 4f were used in place of
tetrahydroazepines (Table 7 entries 4 and 6). In cases where disubstituted olefins were
obtained, transꢀisomers predominated. The fact that products containing the lessꢀ
substituted alkenes formed in high yields suggests that the rearrangement is primarily
driven by the stability of the ring. The preference to yield the more stable ring also
overrides the formation of a conjugated alkene, as can be seen from the attempted ringꢀ
expansion of a tetrahydroisoquinoline 4g (Table 7, entry 7).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Steric effects in the isomerisation of cyclic allylic amines
We became interested in whether it is be possible to control the
diastereoselectivity of the rearrangement. To test this, we subjected substituted sevenꢀ
and eightꢀmembered precursors (Table 7, entries 8ꢀ11) to allylpalladium chloride and
P(OEt) in dichloromethane in the presence of morpholine and TFA. None of the
3
subjected allylic amines gave the rearranged products except for 4i, which rearranged to
the corresponding piperidine 5i in good yield with a 3:1 selectivity for the transꢀisomer.
When dichloromethane was replaced with dichloroethane and the reaction temperature
was increased, tetrahydroazepines 4h, 4j, and 4k (Table 7, entries 8, 10, 11) gave the
corresponding ringꢀcontracted products in modest yields with very low selectivity for the
25
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 26 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
transꢀproduct. Such poor selectivities are probably thermodynamic in nature, and reflect
the difference in relative stability between the transꢀ and the cisꢀ isomers. Moreover, the
resulting chiral centres are not contiguous, and are less likely to affect each other.
Tetrahydroazepines 4l, 4m, and 4n with substituents in the 4ꢀpositions (Table 7, entries
12ꢀ14) were designed such that in case the rearrangement did occur, the products would
contain two contiguous chiral centres. Unfortunately, none of these substrates underwent
the rearrangement, probably due to the excessive steric interactions between the
substituents that are placed in close proximity during cyclization. Likewise, 4o and 4p
gave no rearranged product, most likely for the same reason (Table 7, entries 15ꢀ16).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Electronic effects in the isomerisation of cyclic allylic amines
Another class of substrates that were subjected to the rearrangement reaction
contained an extra heteroatom (Table 7, entries 2, 17ꢀ20). When subjected to the reaction
conditions, tetrahydrodiazocine 4q gave the corresponding tetrahydroquinoxaline 5q in
good yield (Table 7, entry 17). Similarly, dihydrooxazocine 4r gave the expected
dihydrobenzooxazine 5r along with its isomer 5r’, which was formed from the ionization
of the phenol ether (Table 7, entry 18). Phenoxide is a good leaving group, and has been
4
0
utilized as a leaving group in allylic alkylation.
There is no control over
chemoselectivity of this system, and both products form in an equimolar ratio.
Interestingly, the homologated version of this substrate, 4s, showed no conversion to the
expected tetrahydrobenzooxazepine (Table 7, entry 19). This is most likely due to the
reduced basicity of 4s, which lacks the electronꢀdonating alkoxy substituent on the
aromatic ring. As a result, effective protonation of the weaker aniline in the alreadyꢀ
26
ACS Paragon Plus Environment

Page 27 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
buffered system is quite low. Substrate 4s is more electron rich, which explains why
product 5s is formed in high yield. Similarly, amide 4b did not undergo the
rearrangement due to its low bacisity, when compared to 4a. Heterocycle 4t underwent
full hydrolysis under the reaction conditions due to the presence of the acidꢀsensitive
carbanolamine functionality (Table 7, entry 20).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Stereoelectronic effects in the isomerisation of cyclic allylic amines
Finally, stereoelectronic effects were found to be very significant when we
considered the rearrangement of bicyclic allylic amines. Vinyl azetidine 4w rearranged to
give tetrahydropyridine 5w (Table 7, entry 22). We became interested in whether we
could use strain to drive the rearrangement of two rings simultaneously in a fused
bicyclic system. To test this out we have prepared bicycle 4u (Table 7, entry 21) and
subjected it to the reaction conditions. Surprisingly, it did not show any conversion even
in DCE at reflux. Such difference in behavior between the simple vinyl azetidine 4w and
bicycle 4u can be attributed to the limited ability of the latter to achieve the reactive
conformation, with πꢀsystem and the C–N bond being placed orthogonal to one another
either during ionization or ringꢀclosure steps. Such condition is necessary to ensure the
favourable overlap between the πꢀorbital of the alkene and the σ* orbital of the C–N
bond.
Conclusions
27
ACS Paragon Plus Environment

The Journal of Organic Chemistry
Page 28 of 62
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
In summary, we have investigated the factors that control regioselectivity in
palladium catalyzed allylic amination. We have shown that the branched regioselectivity
can be first set by choice of the appropriate solvent and ligand that determine the extent
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
3
of equilibrium between the competing η and η allyl palladium intermediates. Mixed
solvent studies showed that the highest regioselectivity was observed in THF in the
presence of P(OEt) . This branched selectivity can be further maintained by preventing
3
acidꢀpromoted isomerization by using DBU. In addition, solvent was shown to have a
thermodynamic effect on the reaction regioselectivity by either favouring or slowing
down the branchedꢀtoꢀlinear isomerization. This thermodynamic effect of solvent formed
the basis for developing the rearrangement of cyclic amines. In addition to the solvent
effects and basicity effects, steric as well as stereoelectronic effects dictated the
feasibility of the azaꢀallylic ring rearrangements.
Experimental Section.
General procedure A for the preparation of allylic acetates.
In a flameꢀdried, 250 mL oneꢀneck round bottom flask, equipped with a magnetic stir bar,
were placed allylic alcohol (98.4 mmol), triethylamine (35 mL, 250 mmol), and dry
dichloromethane (75 mL) via syringe. The resulting solution was stirred under nitrogen at
room temperature for 30 min, after which DMAP (0.60 g, 4.92 mmol) was added. The
flask was cooled in an ice bath and acetic anhydride (23.5 mL, 250 mmol) was added
dropwise via syringe. The resulting solution was stirred under a stream of nitrogen at
room temperature overnight, when GC analysis showed no remaining starting material.
The reaction mixture was washed with saturated NaHCO (2 x 100 mL). The combined
3
28
ACS Paragon Plus Environment

Page 29 of 62
The Journal of Organic Chemistry
1
2
3
4
5
6
7
8
9
organic fractions were then washed with brine (100 mL) and dried over Na SO . Solvent
2
4
was removed in vacuo and crude product was obtained as slightly yellow oil, which was
subjected to Kugelrohr distillation to yield the corresponding allylic acetate as a clear oil.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
41
Prenyl acetate (2a)
1
Procedure A. GC retention time of 2a: T = 6.7 min. Yield = 88%, 11.96g. H NMR
(
CDCl , 300 MHz): δ 5.35 (t, J = 6.3Hz, 1H), 4.57 (d, J = 6.3 Hz, 2H), 2.05 (s, 3H), 1.76
3
(
s, 3H), 1.71 (s, 3H).
42
Crotyl acetate (a mixture of trans and cis) (2b)
1
Procedure A. GC retention time of 2b: T = 3.65 min. Yield = 63%, 8.7g. H NMR
(CDCl , 300 MHz): δ 5.86ꢀ5.72 (m, 1H), 5.65ꢀ5.54 (m, 1H), 4.50(d, J = 6.3 Hz, 2H), 2.06
3
(
s, 3H), 1.72 (d, J = 7.9 Hz, 3H).
4
3
(
E)-pent-2-enyl acetate (2c)
1
Procedure A. GC retention time of 2c: T = 6.5 min. Yield = 51%, 6.93g. H NMR
(CDCl , 300 MHz): δ 5.89ꢀ5.78 (m, 1H), 5.56 (dt, J = 15.3, 6.5 Hz, 1H), 4.5 (d, 6.5 Hz,
3
2
H), 2.11ꢀ2.04 (m, 2H), 2.06 (s, 3H), 1.01 (t, J =7.5 Hz, 3H).
4
4
(E)-hex-2-enyl acetate (2d)
1
Procedure A. GC retention time of 2d : T = 8.8 min. Yield = 88%, 10.47g. H NMR
CDCl , 400 MHz): δ 5.81ꢀ5.72 (m, 1H), δ 5.61ꢀ5.53 (m, 1H), 4.51(d, J = 6.5 Hz, 2H),
(
3
2
.06 (s, 3H), 2.04 (q, J = 6.4 Hz, 2H), 1.46ꢀ1.37 (m, 2H), 0.91 (t, J = 7.2 Hz, 3H).
29
ACS Paragon Plus Environment