Mechanistic Insights into the Palladium-Catalyzed Aziridination of Aliphatic Amines by C-H Activation

Article abstract of DOI:10.1021/jacs.5b05529

Detailed kinetic studies and computational investigations have been performed to elucidate the mechanism of a palladium-catalyzed C-H activation aziridination. A theoretical rate law has been derived that matches with experimental observations and has led

Full text of DOI:10.1021/jacs.5b05529

Page 1 of 12
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
Mechanistic Insights into the Palladium-Catalyzed Aziridination of
Aliphatic Amines by C–H Activation
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
Adam P. Smalley, Matthew J. Gaunt*
Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, United Kingdom
ABSTRACT: Detailed kinetic studies and computational investigations have been performed to elucidate the mechanism
of a palladium-catalyzed C–H activation aziridination. A theoretical rate law has been derived which matches with exper-
imental observations and has led to an improvement in the reaction conditions. Acetic acid was found to be beneficial in
controlling the formation of an off-cycle intermediate, allowing a decrease in catalyst loading and improved yields. Densi-
ty functional theory (DFT) studies were performed to examine the selectivities observed in the reaction. Evidence for elec-
tronic-controlled regioselectivity for the cyclopalladation step was obtained by a distortion-interaction analysis while the
aziridination product was justified through dissociation of acetic acid from the palladium(IV) intermediate preceding the
product-forming reductive elimination step. The understanding of this reaction mechanism under the synthetic condi-
tions should provide valuable assistance in the comprehension and design of palladium-catalyzed reactions on similar
systems.
coordinate to the electrophilic metal center to form coor-
dinately saturated square-planar complexes, a species
which is often unreactive to C–H cleavage (Figure 1(a)).5
As a result, C–H activation on aliphatic amines requires
their derivatization with electron-withdrawing protecting
groups or bespoke directing groups in order to achieve a
successful reaction, which reduces the overall efficiency of
the C–H transformation. Aliphatic amines are, however,
particularly important functional groups as they are
common in pharmaceuticals, synthetic building blocks,
natural products and polymeric materials. Therefore, the
development of new catalytic C–H activation modes for
aliphatic amines is an important challenge for the contin-
ual advance of chemical synthesis.
Introduction
Transition metal catalyzed activation of aliphatic C–H
bonds has emerged as an area of great potential for chem-
ical synthesis through the development of new transfor-
mations, the streamlining of complex molecule assembly
and in the late-stage modification of biologically im-
portant entities.1 Central to the activation of many of
these traditionally unreactive C–H bonds is a process
called cyclometallation, a functionalization event that
steers the metal catalyst to the point of reaction and
drives the C–H bond cleavage through proximity; coordi-
nation of the metal catalyst to a Lewis basic heteroatom
within the directing motif is assumed to lower the entrop-
ic and enthalpic costs of the C–H bond cleavage and ring
closure.2 Reaction of the resulting metallocycle with an
external species, to form a carbon-carbon or carbon het-
eroatom bond, completes the overall functionalization
process. Several functionalities can participate in this ‘di-
rected' C–H activation, such as carbonyl derivatives, aro-
matic nitrogen heterocycles and hydroxyl motifs,3 and,
arguably, palladium complexes have been most successful
at promoting catalytic C–H functionalization in aliphatic
systems. As a result, cyclopalladation has underpinned a
number of useful catalytic aliphatic C–H bond functional-
ization processes that have expanded the toolbox of reac-
tions available to synthetic chemists.
Figure 1. (a) Amine coordination to palladium(II) salts results
in stable complex formation; (b) The system under study in
this report – amine directed aliphatic C–H functionalization.6
Given the palladium-coordinating ability of aliphatic
amines, it is surprising that they are seldom used in cata-
lytic C–H activation reactions in comparison to other,
synthetically important and polar functional groups.4
However, it is precisely this strong coordination that lim-
its their utility in C–H activation. When treated with pal-
ladium(II) salts, two molecules of the aliphatic amine
As part of our overarching goal to develop new activa-
tion modes for catalytic C–H activation, our group recent-
ly reported the use of an unprotected aliphatic secondary
amine to direct a palladium-catalysed C–H functionaliza-
ACS Paragon Plus Environment

Journal of the American Chemical Society
tion (Figure 1(b)). This activation mode resulted in the
Page 2 of 12
1
2
3
4
5
6
7
8
transformation of a methyl group adjacent to the amine
motif via a four-membered ring cyclopalladation path-
way.6 In one of the two novel transformations detailed in
that initial report, we described a C–H amination to form
aziridines. A fully substituted and unsymmetrical second-
ary amine, morpholinone 1, underwent C–H activation to
form a four-membered ring palladacycle which, following
the action of a hypervalent iodine oxidant, promoted C–N
bond formation to aziridine 2. This aziridination had sev-
eral interesting features which would have been difficult
to predict with the current understanding of C–H activa-
tion processes.
9
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
Figure 2. Postulated mechanism based on our previous study6 along with intermediates characterized by X-ray crystallography.
The morpholinone 1 coordinates to the metal catalyst, but can only undergo C–H activation when there is a vacant coordination
site. The palladacycle 5 is presumably oxidized to a palladium(IV) species 7 (‘L’ is an undefined neutral ligand) which can then
reductively eliminate to give the product 2. The palladacycle 5 was observed through crystallization on addition of a phosphine
ligand to give 6.
dium(II) complex 4. On warming this complex, we had
been able to identify a putative C–H activation complex 5,
which, following treatment with a phosphine, resulted in
the mono-nuclear cyclopalladation species 6, as deter-
mined by X-ray diffraction of a single crystal. The catalyt-
While a comprehensive understanding of this reaction
mechanism would allow us to logically move towards fur-
ther reaction optimization, we also wanted to rationalize
(a) why the cyclopalladation occurred with extremely
ic cycle was believed to conclude with oxidation of the
high regioselectivity, even though there appeared to be a
palladacycle by the hypervalent iodine PhI(OAc)2 fol-
choice of sites to react and (b) why an aziridine was fur-
lowed by C–N reductive elimination of the resulting pal-
nished as the final product, unlike the acetoxylated prod-
ladium(IV) species 7.
uct often obtained under similar conditions.3a, 7
Kinetic Studies
Herein we report detailed mechanistic studies on the
palladium-catalyzed aziridination of tetramethylmorpho-
linone 1. Furthermore, density functional theory (DFT)
calculations were performed to gain additional insight
into the reaction. As a result, we were able to gain a com-
prehensive understanding of the factors that influence the
catalytic C–H activation reaction and improvements in
the reaction conditions for aziridination were realized.
Our kinetic protocol features experiments to probe
the reagent concentration dependencies and isotopic la-
belling to interrogate the mechanism of the reaction in
Figure 2. The process was monitored by aliquotting the
reaction mixture and measuring the concentration by gas
chromatography (GC) with 1,1,2,2-tetrachloroethane as an
internal standard. The original reaction conditions in-
volved the treatment of amine 1 with 5 mol% palladium
acetate, 1.5 equivalents of PhI(OAc)2, 2.0 equivalents of
acetic anhydride in a solution of toluene stirred at 70 ˚C.
However, in order to simplify the system and remove any
extraneous factors that may complicate the analysis we
monitored the reaction with just starting amine, catalyst,
Results and Discussion
At the outset of our studies, we were able to propose a
basic reaction mechanism upon which we could formu-
late more advanced studies (Figure 2). We had previously
observed that mixing two equivalents of the amine 1 with
palladium acetate resulted in the isolable bis-amine palla-
ACS Paragon Plus Environment

Page 3 of 12
Journal of the American Chemical Society
oxidant and solvent. This led to a rate profile that ap-
peared to be linear, but was only ~30% complete in 500
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
1
2
3
4
5
6
7
8
1.0 equiv amine at start
minutes (see Supporting Information). Doubling the con-
centration of catalyst to 10 mol% allowed us to follow the
whole reaction over an appreciable timescale and revealed
that the reaction rate increases during the majority of the
reaction (up to ~90% conversion) and only levels off at
the very end (Figure 3). Notably, the rate of product for-
mation is roughly equal to the rate of starting material
usage and so our analyses using initial rates are based on
starting material concentration so as to minimize any
errors in the concentration measurement.
0.8 equiv amine at start,
time adjusted
9
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
0
50
100
150
200
250
300
350
Time / minutes
Amine 1
Aziridine 2
Figure 4. Probing potential autoinduction by carrying out
two reactions at the same ‘excess’ but different initial amine
concentrations.9 The two plots overlay when time adjusted
suggesting that there is no autoinduction. Reaction condi-
tions for 1.0 equiv amine: amine 1 (0.10 M), Pd(OAc)2 (0.01
M), PhI(OAc)2 (0.15 M), toluene, 70 °C.
Therefore, we reasoned that the rate increase was due
to a negative order dependency of amine 1. By comparing
the initial rate of reaction with that at 50% conversion, we
saw that the rate had doubled, suggesting an approximate
order of -1 with respect to substrate. However, as the rate
decreases at the end, the situation is more complex than
this (vide infra).
Figure 3. Reaction profile for the above aziridination.
The increase in reaction rate over time suggested to us
that there was either a negative order dependence on
amine 18 or that the reaction was autoinductive. The auto-
induction could be a result of rate acceleration by the
action of product aziridine 2, or by the acetic acid or phe-
nyl iodide by-products. To test for autoinduction, same
‘excess’ experiments were performed.9 This analysis is
equivalent to carrying out identical reactions at two dif-
ferent starting points and is often used to probe product
inhibition and catalysis deactivation. In this case, the re-
action was started at 20% completion. If a (by-)product
catalyzes the reaction, the reaction starting at 20% com-
pletion will be slower than the reaction that started from
0% completion. Simply overlaying the plots, after adjust-
ing the 20% completion reaction for time so that the ini-
tial amine concentrations are equivalent, shows that there
are negligible differences in the rate profiles (Figure 4).
We also tested the effect of adding the (by-)products (0.2
equiv PhI, 0.2 equiv aziridine 2 or 0.4 equiv AcOH) to the
20% completion reaction. In the cases of PhI and aziridine
2 negligible differences in the rate was observed. Howev-
er, if we added 0.4 equiv AcOH to the reaction then a
marginal increase in rate was observed (see Supporting
Information). While these results suggested that the re-
sult was not autoinductive, we did note the effect of
AcOH and this is discussed in more detail later.
The order with respect to reaction of the other com-
ponents in the mixture was then investigated. Due to the
linear nature of the kinetic profile at the start of the reac-
tion, the initial rate may be determined from the gradient
of the concentration profile during this period. It was
found that the reaction is zero-order with respect to
PhI(OAc)2, suggesting that the turn-over limiting step
(TOLS) occurs before the oxidation step. Furthermore,
the linear variation of rate with catalyst concentration
revealed that the reaction was first order with respect to
Pd(OAc)2 (see Supporting Information).
Further kinetic information was obtained by meas-
urement of the kinetic isotope effect (KIE). To prevent
any possible regioselectivity issues between the two sets
of methyl groups on amine 1 that may interfere with la-
beling experiments, we opted to test a cyclohexane-
derivative 8 on the basis that competing methylene C–H
activation would be negligible.10 After confirming that this
substrate was viable in the C–H aziridination, the KIE was
determined from initial rate measurements of substrates
8 and d6-8 (Scheme 1). A value of 3.8 was obtained which
is indicative of a primary isotope effect, suggesting that
C–H bond cleavage occurs as part of the TOLS.11
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 4 of 12
Scheme 1. KIE measured from comparison of initial Scheme 2. Proposed elementary steps in the kinet-
1
2
rates of amines 8 and d6-8.
ically important region of the catalytic cycle.
3
4
5
6
7
8
9
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
Consolidation of the mechanism proposed in Figure 2
with the above kinetic data allowed us to provide a set of
kinetically-important elementary steps with which to test
the experimental data (Scheme 2). An equilibrium exists
whereby the amine 1 can coordinate to the catalyst to give
mono-coordinated intermediate 10. This intermediate can
then undergo cyclopalladation in the TOLS or coordina-
tion with a second molecule of amine 1 to give the bis-
amine complex 4. It is proposed that the bis-amine com-
plex 4 is an off-cycle intermediate and previous literature
suggests that a vacant coordination site is required for C–
H activation to occur via the assumed concerted metalla-
tion deprotonation pathway.12 We assumed that any
isomerization of the acetate ligands to provide the vacant
site cis to the amine is expected to be so fast as to be ki-
netically negligible.5
Evidence in support of this bis-amine complex being
an off-cycle intermediate was obtained by comparing the
reaction profiles of two sterically different amines. It was
found that the aziridination of a more hindered amine 12
was faster than that of the tetramethyl amine 1 (Figure
5(a)). This could be indicative of faster C–H cleavage or
an increased concentration of the active mono-
coordinated species through destabilization of the off-
cycle bis-amine complex on the basis of increased steric
interactions between the amine and palladium center. In
order to show that the rate increase was as a result of
suppressing the formation of this unproductive species, a
1:1 mixture of the amines 1 and 12 was subjected to the
catalytic conditions. Both species reacted at the same rate
to form the corresponding aziridines 2 and 13 (Figure
5(b)). Notably, the rate of formation of 2 in the competi-
tion experiment is greater than observed in isolation. This
suggests that the increased rate of reaction for the more
hindered substrate (12) was not as a result of a lower en-
ergy C–H activation, but instead by alteration of the
mono-/bis-amine equilibrium.
ACS Paragon Plus Environment

Page 5 of 12
Journal of the American Chemical Society
Scheme 3. Derivation of the Rate Law. Cat = Pd(OAc)2
1
2
3
catalyst, [Pd]total = initial concentration of Pd(OAc)2.
On-cycle
4
5
[
]
[
]
푘−1 ퟏퟎ = 푘1 cat [ퟏ]
6
7
Off-cycle
8
[
]
푘2 ퟏퟎ [ퟏ] = 푘−2[ퟒ]
9
Catalyst mass balance
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
푘−1
푘2[ퟏ]
푘−2
[
]
[
]
[ ]
[푃푑]푡표푡푎푙 = 푐푎푡 + ퟏퟎ + ퟒ = [ퟏퟎ] (
+ 1 +
)
13
[ ]
ퟏ
푘1
Rate law
푅푎푡푒 = −
푑[ퟏ] 푑[ퟏퟏ]
푘3[푃푑]푡표푡푎푙
[ ]
= 푘3 ퟏퟎ =
=
푘2[ퟏ]
푘−2
푘−1
푑푡
푑푡
(
+ 1 +
)
[ ]
ퟏ
푘1
By assuming that the amine concentration, [1], is
large, we can take the rate law derived in Scheme 3 and
obtain an approximation for the initial rate (eq. 1). This
assumption is based on the concentration of bis-amine
complex 4 dominating [10] and [cat]; this is supported by
computational calculations which suggest that the for-
mation of 4 from 10 is favored by 8.43 kcal mol-1 (see Sup-
porting Information).
2
푘−2푘3 [푃푑]푡표푡푎푙
퐼푛푖푡푖푎푙 푟푎푡푒 ≈
(1)
푘2
[ퟏ]
13
Varying the concentration of amine 1 whilst keeping
the catalyst concentration constant allows the determina-
tion of k-2k3/k2 to be 0.12 ± 0.02 M s-1 (see Supporting In-
formation). Unfortunately, attempts to directly measure
1
Figure 5. (a) The reaction rate of the more hindered amine 12
is faster than the tetramethyl amine 1; (b) If these amines are
mixed in a 1:1 ratio they both react at the same rate.
k2/k-2 by H NMR were unsuccessful due to the overlap-
ping peaks of the metal species and therefore we were
unable to determine a value for the rate constant of C–H
activation, k3.
Next we set up a mathematical description of the reac-
tion in order to allow quantitative agreement of our ex-
perimental data and qualitative insights. Because of the
observed primary KIE, the rate of reaction is proposed to
be given by the C–H activation step itself; this is also con-
sistent with the measurement of oxidant concentration
being zero order. To enable the derivation of the rate law,
we assumed that all the intermediates following the TOLS
provided a negligible contribution to the total catalyst
mass. Consideration of the elementary steps in the reac-
tion then led to the theoretical rate law (Scheme 3). This
rate law accounts for the initial pseudo-1st order of reac-
tion with respect to catalyst measured experimentally;
initially the high substrate concentration can be consid-
ered a constant. The reaction is also negative 1st order
with respect to substrate at high substrate concentrations,
accounting for the increase in rate over the majority of
the reaction, but changes to 1st order with respect to sub-
strate at the end, accounting for the ‘tailing-off’ at the end
of the reaction.
Kinetics led Reaction Optimization
We had previously found that the addition of 0.4
equivalents of acetic acid to the reaction mixture had a
small, but still noteworthy effect on the rate of reaction.
Based on this, we questioned whether a much higher con-
centration of acid could lead to a further increase in the
rate of reaction. The presence of acid in the reaction mix-
ture should lead to some protonation of the amine and
set up an equilibrium between the protonated and free-
based amine.13 If a proportion of the amine is protonated,
then it is no longer able to coordinate to the metal and
the effective concentration of free-amine is lowered. As
the reaction progresses, the acid equilibrium shifts to
provide more free-based amine and effectively delivers a
slow release of substrate. By keeping the amine concen-
tration low, less palladium is sequestered in the unpro-
ductive off-cycle bis-amine complex 4 leading to an in-
crease in the overall rate of reaction. This is conceptually
related to an organocatalytic reaction reported by Black-
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 6 of 12
mond where control of an off-cycle species lead to rate
up the reaction.1(a),17 However, from initial observations
acceleration.14
this was judged to be unlikely as formation of the Fujiwa-
ra cationic species with acetic acid has only been reported
with sp2 C–H activations and trifluoroacetate ligands on
palladium are required for sp3 C–H activation via this
method.18 Addition of trifluoroacetic anhydride or tri-
fluoroacetic acid was found to shut the reaction down
while using palladium(II) trifluoroacetate instead of
Pd(OAc)2 as the catalyst led to a slower, poor yielding
reaction.
We had previously noted6 that acetic anhydride was
beneficial to the reaction; however, unlike with acetic
acid, changing the concentration of acetic anhydride had
little effect. We suspect that the main effect of this addi-
tive is the removal of water.19 Water was found to be det-
rimental to the reaction20 and, furthermore, the removal
of water by the anhydride leads to the production of rate
accelerating acetic acid.
1
2
3
4
5
6
7
8
Hence, increasing the concentration of acid was found
to increase the rate of reaction, although at very high acid
concentrations this increase in rate was offset by degrada-
tion of the final product (Figure 6). Other acids, such as
benzoic acid and pivalic acid, were also found to give rate
acceleration. However, as the optimum acid loading was
found to be 20 equivalents, acetic acid was used as it is
cheaper and, as a liquid, its use is more practical. It is im-
portant to note that we have not ruled out the possibility
that the acetic acid is hydrogen bonding to the amine 1
rather than formally protonating the amine, however this
still provides the desired decrease in free substrate con-
centration.
9
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
0.10
0.08
Taken together, addition of 20 equivalents of acetic
acid and 2 equivalents of acetic anhydride led to an im-
proved reaction, giving an isolated yield of 85% after 2.5
hours (Figure 7).
0.06
No AcOH added
2 equiv AcOH
20 equiv AcOH
100 equiv AcOH
0.04
0.02
0.00
No additives
2 equiv Ac2O
20 equiv AcOH
0
100
200
300
400
500
Time / minutes
Figure 6. Increasing the concentration of acetic acid in the
reaction mixture increases the rate of reaction. Reaction
conditions: amine 1 (0.10 M), Pd(OAc)2 (0.005 M), PhI(OAc)2
(0.15 M), toluene, 70 °C.
2 equiv Ac2O +
20 equiv AcOH
Even though we believed that the beneficial addition
of acetic acid was a result of limiting the formation of bis-
amine intermediate 4, it is important to consider other
possible effects of acetic acid. Previous work by Ryabov5
on an amine directed sp2 C–H activation showed that ad-
dition of acid fully protonated the amine and changed the
TOLS from C–H activation to dissociation of the
Pd(OAc)2 trimer. The subsequent C–H activation on this
multi-metal complex was proposed to be faster due to the
production of species with vacant coordination sites or
more labile acetates. This explanation was unlikely here
because even though Pd(OAc)2 predominantly exists as a
trimer in the solid phase and at 37 °C in benzene, it has
been shown to exist as a monomer at 80 °C in benzene.15
Nonetheless, this possibility was still ruled out by testing
for a KIE with acetic acid present in the reaction. A pri-
mary KIE (of 3.9 vs 3.8 previously) was observed, indicat-
ing that the TOLS is still the C–H activation step. From
the addition of acetic acid, we also noted that no H-
incorporation was detected in the deuterated aziridine
product, suggesting that the C–H/D activation step is not
reversible under these conditions.
Figure 7. The effect of acetic acid and acetic anhydride addi-
tives on the reaction profile. Reaction conditions: amine 1
(0.10 M), Pd(OAc)2 (0.005 M), PhI(OAc)2 (0.15 M), toluene,
70 °C.
The generality of the newly optimized conditions were
tested with a selection of amine substrates (Figure 8).
Furthermore, the aziridination of amine 16 proceeded
smoothly under the reaction conditions. This conversion
of amine 16 to aziridine 17 rules out a general pathway for
aziridination that proceeds first via acetoxylation and
then displacement to form the 3-membered ring. It was
found that the reaction worked well using industry-
preferred ethyl acetate as the solvent.21 The catalyst load-
ing could also be lowered and reaction of amine 1 on a
gram scale with 1 mol% Pd(OAc)2 in ethyl acetate gave an
87% yield of the aziridine product.
We also considered that a Fujiwara-type cationic in-
termediate may form,16 whereby the C–H activation onto
the putative cationic palladium species is faster, speeding
ACS Paragon Plus Environment

Page 7 of 12
Journal of the American Chemical Society
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Figure 9. C–H activation through a concerted metallation-
deprotonation pathway.
Application of this CMD approach on the two possible
sites resulted in the transition state for the observed regi-
oselectivity being lower in energy by 4.89 kcal mol-1 (Fig-
ure 10), suggesting that this mechanism is an appropriate
representation of the cyclopalladation step.
Figure 8. The optimized conditions have improved the yield
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
of previous substrates.6 In addition, amine 16 undergoes
aziridination through C–H activation rather than SN2-type
displacement of acetate under the reaction conditions.
Computational Investigations
We next performed a series of density functional theo-
ry (DFT) studies to gain insights into the regioselectivity
of the C–H activation step and the chemoselectivity of the
C–N reductive elimination.22 The calculations were per-
formed using the Amsterdam Density Functional (ADF)
program23 at the relativistic ZORA-BLYP-D3/TZ2P level of
theory, previously benchmarked for palladium catalysis
with added dispersion corrections.24 The inclusion of dis-
persion corrections has been shown to be important when
bulky hydrocarbons are present as interactions which
may previously have been considered as repulsive are, in
fact, attractive.25 Solvation by toluene was computed us-
ing the COSMO solvation model and vibrational frequen-
cy analysis was performed to confirm that the structures
were either minima or transition states.
ΔΔG‡ = 4.89 kcal mol-1
We first wanted to understand the regioselectivity of
the C–H activation step. Initially, the geometry of the
mono-coordinated intermediate 10 was calculated. It was
found that there exists a hydrogen bond between one of
the acetate ligands on the metal and the N–H of the sub-
strate, similar to that seen in the crystal structure of the
bis-amine complex 4 (Figure 2).26 The other acetate lig-
and binds in a 2 mode so as to maximize bonding inter-
actions. The transition states for C–H activation at the
methyl groups either side of the nitrogen were then calcu-
lated. The cyclopalladation was proposed to occur via this
mono-coordinated intermediate 10 from the above kinetic
data and previous reports indicating the necessity of a
vacant coordination site for C–H activation.5,12(a) The un-
likelihood of an oxidative addition C–H activation, due to
the resulting high energy palladium(IV) species which
often require strong oxidants, and the large primary KIE
suggesting a linear transition state,10,27 meant that the
mode of cyclopalladation was expected to proceed via a
concerted metallation-deprotonation (CMD) pathway
(Figure 9).12b, 28
Figure 10. As expected from experiment, the transition state
for C–H activation at the methyl groups nearest the carbonyl
(TS‡ 2) is lower in energy than at the methyl groups furthest
from the carbonyl (TS‡ 1).
We speculated that the observed selectivity was either
a result of the difference in polarity of the methyl groups,
due to differing distances from the carbonyl group, or
from the change in shape of the ring caused by the ester
motif. This question was probed using a distortion-
interaction analysis, a method that has useful in under-
standing selectivity by splitting the activation energy into
a steric and an electronic component.29 Fragmentation of
the transition state along the new bonds being formed
and calculation of the interaction energy between these
fragments in the transition state geometry allows us to
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 8 of 12
determine the distortion energy required to position the The reaction after the TOLS was proposed to proceed
1
2
3
4
5
6
7
8
fragments in the required shape for reaction to occur. By
fragmenting as shown in Figure 11 to reform the substrate
and Pd(OAc)2, we can see that the disfavored pathway
actually has a lower distortion penalty (Table 1). This sug-
gests that the flattening of the ring is not responsible for
the selectivity. The increased distortion energy for the
observed regioselectivity is offset by a greater increase in
interaction energy, suggesting that it is the electronic
effects of the carbonyl which are important. This can be
explained by an increased acidity of the C–H bonds clos-
est to the carbonyl in the ring, lowering the energy re-
quired to cleave the C–H bond during the metal activa-
tion step.30 We believe that this analysis is valid, even
though we fragment back past the mono-coordinated in-
termediate, as both pathways go through the exact same
mono-coordinated intermediate 10.
via oxidative addition of the palladacyle with PhI(OAc)2
followed by C–N reductive elimination of the resulting
palladium(IV) species to provide the product (Figure 2).
The use of PhI(OAc)2 to access high-valent palladium is
well precedented, with the palladium(IV) species under-
going facile reductive elimination.2(c),31
9
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
Figure 11. Distortion-interaction model splits the activation
energy into two components. Fragmenting as shown in the
TS‡ and calculating the interaction energy between the frag-
ments gives the interaction energy (EINT). The distortion
energy (EDIST
) can subsequently be calculated from
knowledge of the activation energy (E‡).
Table 1. Distortion-interaction analysis of the two
possible C–H activation pathways. Note that the
small values for E‡ are as a result of fragmenting
back to the substrate and catalyst rather than the
(lower energy) mono-coordinated intermediate (10)*
E‡
2.70
0.08
EDIST
85.12
87.97
EINT
-82.42
-87.89
TS‡ 1
TS‡ 2
*energies reported in kcal mol-1
ACS Paragon Plus Environment

Page 9 of 12
Journal of the American Chemical Society
1
2
3
C–O RE
4
5
6
C–N reductive elimination (RE)
7
8
C–O RE
9
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
C–N by SN2
C–N RE
AcOH
Dissociation
of amine
Dissociation
of AcOH
Figure 12. Possible pathways to form either a C–O or C–N bond. Reductive elimination from the initial palladium(IV) intermedi-
ate 18 is a high energy pathway and the reaction instead proceeds through rapid deprotonation of the coordinated amine, liber-
ating acetic acid, to give intermediate 20. Reductive elimination from this species is the lowest energy pathway and favors C–N
bond formation (TS‡ 7). Even though the transition state for C–N bond formation through SN2-type attack of the free amine, TS‡
9, is similar in energy to the reductive elimination through TS‡ 7, accessing the necessary conformation is prohibitive due to the
high energy barrier (via TS‡ 8) associated with the dissociation of the nitrogen atom from the metal center.
However, although reductive elimination from palla-
dium(IV) is well known, the chemoselectivity of C–
heteroatom reductive elimination is poorly understood
where multiple competing pathways exist32 and therefore
any extra insight into this process is of high value. Reac-
tion via a bimetallic palladium(III) intermediate as char-
acterized by Ritter33 was considered. However, this was
deemed to be unlikely here due to the high energy species
formed from the hindered amines stacking on top of each
other in the palladium(III) dimer. This supposition was
borne out by calculations which suggested that the palla-
dium(IV) intermediate was more favorable.34
plex 20 where the nitrogen atom now formally acts as an
anionic ligand. Pleasingly, the C–N reductive elimination
via TS‡ 7 to form the aziridine was now lower in energy
than both the competing C–O reductive elimination via
TS‡ 6 as well as the C–O reductive elimination via TS‡ 4.
This is consistent with previous work on C–H amination
with amide directing groups;37 the nitrogen atoms are
formally deprotonated in the proposed catalytic cycles
and lead to C–N reductive elimination as the major path-
way. The possible dissociation of acetic acid from the pal-
ladium(IV) intermediate is perhaps not too surprising as
the electronegative oxygen atoms in the morpholinone
may lead to an increase in acidity of the N–H bond and
therefore lower the energy of the deprotonated pathway.
We initially tested the C–N and C–O reductive elimi-
nation pathways directly from the proposed palladi-
um(IV) intermediate 18 (Figure 12).35 Counter to our ex-
pectations, C–O bond formation via TS‡ 4 was a lower
energy process than C–N bond formation via TS‡ 3.36 This
reaction pathway is therefore not feasible because the
resulting acetoxylated product (16) has been experimen-
tally ruled out as an intermediate in the formation of the
aziridine (Figure 8). We then considered the possibility
of acetic acid dissociation from 18, via TS‡ 5, to give com-
For completeness, we also calculated the pathway for
C–N bond formation by SN2-type attack by the uncoordi-
nated amine. We found that even though the transition
state of the C–N bond forming step, TS‡ 9, was compara-
ble in energy with the lowest energy C–N reductive elimi-
nation transition state, TS‡ 7, the reactive conformation
cannot be reached. In order for the geometry of the palla-
dium(IV) species to be reactive to SN2, the amine must
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 10 of 12
dissociate from the palladium(IV) center (breaking both improve yields of reaction, as well as providing a more
1
2
3
4
5
6
7
8
the amine-palladium bond and the O–H hydrogen bond)
and rotate. It was found that the energy required to do
this was prohibitively high (TS‡ 8) and so we do not be-
lieve that the SN2 pathway is operable.
synthetically useful procedure that is more applicable for
large-scale chemistry. DFT calculations have given an
insight into the regioselectivity of cyclopalladation and
have provided a plausible explanation for the resulting
aziridine product.
As a result of our studies we are now able to propose a
more complete mechanistic picture of this unusual C–H
activation reaction (Scheme 4). The amine 1 first coordi-
nates to palladium acetate to give mono-coordinated in-
termediate 10. This species can either coordinate to an-
other amine, to form the off-cycle bis-amine intermediate
4, or can undergo the TOLS of intramolecular C–H activa-
tion to form a four-membered ring palladacycle 11. This
intermediate is now susceptible to oxidation by
PhI(OAc)2 and forms palladium(IV) intermediate 18. Dis-
sociation of acetic acid precedes C–N reductive elimina-
tion to provide the aziridine product and also reforms the
active catalyst. Furthermore, the overall concentration of
amine 1 is modulated by an acid-mediated equilibrium
which provides a slow release mechanism for the sub-
strate into the cycle and limits formation of the off-cycle
bis-amine intermediate 4.
We are currently investigating how we can use this
understanding to design ligands which can control the C–
H activation step with the overall goal of creating an
asymmetric process. We hope that these insights into the
behavior of palladium(IV) leads to the rational design of
related C–H functionalization reactions.
9
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
ASSOCIATED CONTENT
Supporting Information.
Experimental procedures, characterization data and compu-
tational details. This material is available free of charge via
the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
mjg32@cam.ac.uk
Scheme 4. Final catalytic cycle
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
We are grateful to the EPSRC and Pfizer (A.P.S.) for a stu-
dentship and the ERC and EPSRC for fellowships (M.J.G.).
We are also grateful to Dr Alex Thom and Dr David Pryde
(Pfizer) for helpful discussions. The computational work was
performed using the Darwin Supercomputer of the Universi-
ty of Cambridge High Performance Computing Service
(http://www.hpc.cam.ac.uk/), provided by Dell Inc. using
Strategic Research Infrastructure Funding from the Higher
Education Funding Council for England and funding from
the Science and Technology Facilities Council. Mass spec-
trometry data were acquired at the EPSRC UK National Mass
Spectrometry Facility at Swansea University.
REFERENCES
(1) (a) Jia, C.; Kitamura, T.; Fujiwara, Y. Acc. Chem. Res. 2001,
34, 633. (b) Godula, K.; Sames, D. Science 2006, 312, 67. (c) Da-
vies, H. M. L.; Manning, J. R. Nature 2008, 451, 417. (d) Davies, H.
M. L.; Du Bois, J.; Yu, J.-Q. Chem. Soc. Rev. 2011, 40, 1855. (e)
Yamaguchi, J.; Yamaguchi, A. D.; Itami, K. Angew. Chem. Int. Ed.
2012, 51, 8960. (f) Mkhalid, I. A. I.; Barnard, J. H.; Marder, T. B.;
Murphy, J. M.; Hartwig, J. F. Chem. Rev. 2010, 110, 890.
(2) (a) Dupont, J.; Consorti, C. S.; Spencer, J. Chem. Rev. 2005,
105, 2527. (b) Shilov, A. E.; Shul’pin, G. B. Chem. Rev. 1997, 97,
2879. (c) Neufeldt, S. R.; Sanford, M. S. Accounts Chem. Res.
2012, 45, 936.
(3) (a) Desai, L. V; Hull, K. L.; Sanford, M. S. J. Am. Chem. Soc.
2004, 126, 9542. (b) Zaitsev, V. G.; Shabashov, D.; Daugulis, O. J.
Am. Chem. Soc. 2005, 127, 13154. (c) Giri, R.; Chen, X.; Yu, J.-Q.
Angew. Chem. Int. Ed. 2005, 44, 2112. (d) Campeau, L.-C.; Schip-
per, D. J.; Fagnou, K. J. Am. Chem. Soc. 2008, 130, 3266. (e)
Wang, D.-H.; Wasa, M.; Giri, R.; Yu, J.-Q. J. Am. Chem. Soc.
2008, 130, 7190.
Conclusions
In summary, the mechanism of this novel native
amine directed sp3 C–H activation has been elucidated by
detailed kinetic studies. Suppression of the formation of
the off-cycle bis-amine complex by the addition of acetic
acid was found to significantly increase the overall rate of
reaction. The results of these studies have been used to
ACS Paragon Plus Environment

Page 11 of 12
Journal of the American Chemical Society
(4) (a) Giri, R.; Maugel, N.; Li, J.-J.; Wang, D.-H.; Breazzano, S.
of the methyl groups of 4 results in a higher energy conformer,
1
2
3
4
5
6
7
8
P.; Saunders, L. B.; Yu, J.-Q. J. Am. Chem. Soc. 2007, 129, 3510. (b)
Simmons, E. M.; Hartwig, J. F. Nature 2012, 483, 70.
(5) Ryabov, A. D.; Sakodinskaya, I. K.; Yatsimirsky, A. K. J.
Chem. Soc. Dalt. Trans. 1985, 2629.
(6) McNally, A.; Haffemayer, B.; Collins, B. S. L.; Gaunt, M. J.
Nature 2014, 510, 129.
(7) Giri, R.; Liang, J.; Lei, J.-G.; Li, J.-J.; Wang, D.-H.; Chen, X.;
Naggar, I. C.; Guo, C.; Foxman, B. M.; Yu, J.-Q. Angew. Chem. Int.
Ed. 2005, 44, 7420.
the resulting cyclopalladation is more favorable as a result of the
C–H bonds being placed proximal to the metal center.
(27) Melander, L. C. S.; Saunders, W. H.; Reaction Rates of Iso-
topic Molecules; Wiley, 1980, pp 44-45.
(28) (a) Rousseaux, S.; Gorelsky, S. I.; Chung, B. K. W.;
Fagnou, K. J. Am. Chem. Soc. 2010, 132, 10692. (b) García-
Cuadrado, D.; Braga, A. A. C.; Maseras, F.; Echavarren, A. M. J.
Am. Chem. Soc. 2006, 128, 1066. (c) Balcells, D.; Clot, E.; Eisen-
stein, O. Chem. Rev. 2010, 110, 749.
9
(8) Ferretti, A. C.; Brennan, C.; Blackmond, D. G. Inorganica
Chim. Acta 2011, 369, 292.
(29) (a) Gorelsky, S. I.; Lapointe, D.; Fagnou, K. J. Org. Chem.
2012, 77, 658. (b) Usharani, D.; Lacy, D. C.; Borovik, A. S.; Shaik,
S. J. Am. Chem. Soc. 2013, 135, 17090. (c) De Jong, G. T.; Bickel-
haupt, F. M. Chemphyschem 2007, 8, 1170. (d) Ess, D. H.; Houk,
K. N. J. Am. Chem. Soc. 2008, 130, 10187. (e) Green, A. G.; Liu, P.;
Merlic, C. A.; Houk, K. N. J. Am. Chem. Soc. 2014, 136, 4575.
(30) Zhang, Q.; Yu, H.; Fu, Y. Organometallics 2013, 32, 4165.
(31) (a) Hickman, A. J.; Sanford, M. S. Nature 2012, 484, 177.
(b) Sehnal, P.; Taylor, R. J. K.; Fairlamb, I. J. S. Chem. Rev. 2010,
110, 824. (c) Dick, A. R.; Hull, K. L.; Sanford, M. S. J. Am. Chem.
Soc. 2004, 126, 2300. (d) Yoneyama, T.; Crabtree, R. H. J. Mol.
Catal. A Chem. 1996, 108, 35. (e) Muñiz, K. Angew. Chem. Int. Ed.
2009, 48, 9412. (f) Cheng, X.-F.; Li, Y.; Su, Y.-M.; Yin, F.; Wang,
J.-Y.; Sheng, J.; Vora, H. U.; Wang, X.-S.; Yu, J.-Q. J. Am. Chem.
Soc. 2013, 135, 1236.
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
(9) (a) Blackmond, D. G. Angew. Chem. Int. Ed. 2005, 44, 4302.
(b) Baxter, R. D.; Sale, D.; Engle, K. M.; Yu, J.-Q.; Blackmond, D.
G. J. Am. Chem. Soc. 2012, 134, 4600.
(10) Jones, W. D.; Feher, F. J. Acc. Chem. Res. 1989, 22, 91.
(11) Simmons, E. M.; Hartwig, J. F. Angew. Chem. Int. Ed. 2012,
51, 3066.
(12) (a) Ryabov, A. D. Chem. Rev. 1990, 90, 403. (b) Davies, D.
L.; Donald, S. M. A.; Macgregor, S. A. J. Am. Chem. Soc. 2005, 127,
13754.
(13) (a) Cai, G.; Fu, Y.; Li, Y.; Wan, X.; Shi, Z. J. Am. Chem. Soc.
2007, 129, 7666. (b) Albrecht, M. Chem. Rev. 2010, 110, 576.
(14) Hein, J. E.; Armstrong, A.; Blackmond, D. G. Org. Lett.
2011, 13, 4300.
(15) Stephenson, T. A.; Morehouse, S. M.; Powell, A. R.; Heffer,
J. P.; Wilkinson, G. J. Chem. Soc. 1964, 3632.
(16) Postulated Fujiwara-type intermediate:
(32) Camasso, N. M.; Pérez-Temprano, M. H.; Sanford, M. S. J.
Am. Chem. Soc. 2014, 136, 12771.
(33) Powers, D. C.; Ritter, T. Nat. Chem. 2009, 1, 302.
(34) Possible palladium(III) bimetallic intermediate:
(17) Lu, W.; Jia, C.; Kitamura, T.; Fujiwara, Y. Org. Lett. 2000,
2, 2927.
(18) (a) Nakata, K.; Yamaoka, Y.; Miyata, T.; Taniguchi, Y.;
Takaki, K.; Fujiwara, Y. J. Organomet. Chem. 1994, 473, 329. (b)
Fujiwara, Y.; Takaki, K.; Watanabe, J.; Uchida, Y.; Taniguchi, H.
Chem. Lett. 1989, 1687.
(19) Explicit attempts to show that the main role of acetic an-
hydride was to dry the reaction were unsuccessful as addition of
molecular sieves led to some decomposition of the starting ma-
terial and MgSO4 shut down the reaction.
(20) Addition of 1 equiv H2O was found to shut down the reac-
tion, providing < 5% product (see Supporting Information).
(21) Kerton, F. M.; Marriott, R. Alternative Solvents for Green
Chemistry; Royal Society of Chemistry, 2013, pp 14-15.
(22) Boutadla, Y.; Davies, D. L.; Macgregor, S. A.; Poblador-
Bahamonde, A. I. Dalton Trans. 2009, 5820.
(23) (a) Te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fon-
seca Guerra, C.; van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T.
J. Comput. Chem. 2001, 22, 931. (b) Fonseca Guerra, C.; Snijders,
J. G.; te Velde, G.; Baerends, E. J. Theor. Chem. Accounts Theory,
Comput. Model. (Theoretica Chim. Acta) 1998, 99, 391. (c)
ADF2014, SCM, Theoretical Chemistry, Vrije Universiteit, Am-
sterdam, The Netherlands, http://www.scm.com.
(24) (a) Wassenaar, J.; Jansen, E.; van Zeist, W.-J.; Bickelhaupt,
F. M.; Siegler, M. A.; Spek, A. L.; Reek, J. N. H. Nat. Chem. 2010,
2, 417. (b) Wolters, L. P.; van Zeist, W.-J.; Bickelhaupt, F. M.
Chemistry 2014, 20, 11370.
(25) Lyngvi, E.; Sanhueza, I. A.; Schoenebeck, F. Organometal-
lics 2015, 34, 805.
(26) Species 4 is slightly higher in energy than the conformer
formed through ring flip which places the palladium pseudo
equatorial. However, the subsequent transition states for C–H
cleavage were found to be substantially higher in energy (see
Supporting Information). Presumably, although the steric clash
Pd(II) (10) + Pd(IV) (11) → Pd(III); ΔG = +8.4 kcal mol-1
(35) Intermediate 18 is -39.7 kcal mol-1 lower in energy than in-
termediate 10. Therefore, all species in Figure 12 are not kinet-
ically relevant as expected from the experimental data.
(36) Changing the position of the acetate ligands on the palla-
dium(IV) species gave higher energy intermediates (see Support-
ing Information). Also, note that the bidentate acetate trans to
the substrate does not stop the C–N reductive elimination (to
give TS‡ 3) from being symmetry allowed38 as during the reaction
the acetate binding changes from a 2 to a  1 coordination mode.
(37) (a) Nadres, E. T.; Daugulis, O. J. Am. Chem. Soc. 2012, 134,
7. (b) He, G.; Zhao, Y.; Zhang, S.; Lu, C.; Chen, G. J. Am. Chem.
Soc. 2012, 134, 3. (c) Wang, C.; Chen, C.; Zhang, J.; Han, J.; Wang,
Q.; Guo, K.; Liu, P.; Guan, M.; Yao, Y.; Zhao, Y. Angew. Chemie
Int. Ed. 2014, 53, 9884.
(38) Braterman, S. J. Chem. Soc. Chem. Commun. 1979, 70.
ACS Paragon Plus Environment

Journal of the American Chemical Society
Page 12 of 12
1
2
3
4
5
6
7
8
9
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
Insert Table of Contents artwork here
ACS Paragon Plus Environment