Mechanistic rationalization of unusual kinetics in Pd-catalyzed C-H olefination

Article abstract of DOI:10.1021/ja207634t

Detailed kinetic studies and novel graphical manipulations of reaction progress data in Pd(II)-catalyzed olefinations in the presence of mono-N-protected amino acid ligands reveal anomalous concentration dependences (zero order in o-CF₃-phenylacetic acid concentration, zero order in oxygen pressure, and negative orders in both olefin and product concentrations), leaving the catalyst concentration as the sole positive driving force in the reaction. NMR spectroscopic studies support the proposal that rate inhibition by the olefinic substrate and product is caused by formation of reversible off-cycle reservoirs that remove catalyst from the active cycle. NMR studies comparing the interaction between the catalyst and substrate in the presence and absence of the ligand suggest that weak coordination of the ligand to Pd prevents formation of an inactive mixed acetate species. A fuller understanding of these features may lead to the design of more efficient Pd(II) catalysts for this potentially powerful C-H functionalization reaction.

Full text of DOI:10.1021/ja207634t

Article
pubs.acs.org/JACS
Mechanistic Rationalization of Unusual Kinetics in Pd-Catalyzed
C−H Olefination
Ryan D. Baxter, David Sale, Keary M. Engle, Jin-Quan Yu, and Donna G. Blackmond*
Department of Chemistry, The Scripps Research Institute, La Jolla, California 92037, United States
S
* Supporting Information
ABSTRACT: Detailed kinetic studies and novel graphical manipulations of
reaction progress data in Pd(II)-catalyzed olefinations in the presence of
mono-N-protected amino acid ligands reveal anomalous concentration
dependences (zero order in o-CF3-phenylacetic acid concentration, zero
order in oxygen pressure, and negative orders in both olefin and product
concentrations), leaving the catalyst concentration as the sole positive driving
force in the reaction. NMR spectroscopic studies support the proposal that rate
inhibition by the olefinic substrate and product is caused by formation of
reversible off-cycle reservoirs that remove catalyst from the active cycle. NMR
studies comparing the interaction between the catalyst and substrate in the presence and absence of the ligand suggest that weak
coordination of the ligand to Pd prevents formation of an inactive mixed acetate species. A fuller understanding of these features
may lead to the design of more efficient Pd(II) catalysts for this potentially powerful C−H functionalization reaction.
Scheme 2. ortho-C−H Olefination of Phenylacetic Acids
INTRODUCTION
■
The study of Pd-catalyzed C−H activation for a diverse range
of carbon−carbon and carbon−heteroatom bond-forming
reactions has become a vibrant area of research.1 The
development of catalyst−ligand systems that can improve
reactivity2−4 and positional selectivity2a,5 for a broad range of
synthetically useful substrates has been hampered by lack of
mechanistic understanding. Complications in C−H functional-
ization reactions arise from the fact that Pd(II) in the absence
of a ligand often exists as a mixture of monomeric and higher
order species, all potentially participating in the catalytic cycle.
A recently uncovered Pd(II)/ligand system for C−H
olefination (Scheme 1) offers a catalytic platform for our
Scheme 1. C−H Activation of Arylacetic Acids
reoxidation closes the catalytic cycle with regeneration of
Pd(II). A hydrogen atom abstraction mechanism for reoxidation
with O2 is also possible.
While previous work demonstrated that addition of mono-
N-protected amino acid ligands provided higher yields in
significantly shorter reaction times, the precise role of the
ligand was not made clear. A recent computational study
examining enantioselective C−H activation using these
catalysts pointed to a pathway initiated by N−H cleavage
followed by ortho-C−H activation rather than a direct C−H
focus on the reaction kinetics to provide mechanistic clues to aid
in the design of improved catalysts and reaction protocols.
The Pd(II)-catalyzed C−H olefination of substrates such as
o-(trifluoromethyl)phenylacetic acid 1 has been proposed to
follow the basic catalytic shown in Scheme 2.2c Following
coordination of the substrate 1 to Pd(II), carboxylate-directed
ortho-C−H cleavage takes place to form a cyclopalladated
intermediate. Olefin coordination followed by 1,2-migratory
insertion forms the new C−C bond, and β-hydride elimination
releases the product. Reductive elimination followed by
Received: August 12, 2011
Published: February 10, 2012
© 2012 American Chemical Society
4600
dx.doi.org/10.1021/ja207634t | J. Am. Chem. Soc. 2012, 134, 4600−4606

Journal of the American Chemical Society
Article
activation route.6 However, none of this work provides
evidence to support the simplified mechanism presented in
Scheme 2 and the role of the ligand in these elementary steps.
We report here detailed kinetic studies on the ortho-C−H
olefination of phenylacetic acids using Pd(OAc)2 in the
presence of mono-N-protected amino acid ligands (Scheme 1),
which have previously been demonstrated to accelerate the
reaction rate.2,6 In- situ reaction monitoring using ReactIR
spectroscopy and NMR spectroscopy presents concentration
dependences, identifies the resting state of the catalyst, and
presents a modified mechanistic proposal to clarify several
features of the putative mechanism for these reactions.
Figure 2. Reaction rate measured from the slope of ReactIR
concentration versus time profiles for the reaction of Scheme 1.
Reactions carried out at different Pd (Pd:ligand = 1:2) with [1]0 =
0.080 M; [2]0 = 0.302 M. All other conditions as in Scheme 1.
RESULTS AND DISCUSSION
■
Kinetic Studies. Our kinetic protocol features experiments
probing catalyst stability and the concentration dependences of
the substrates that allow interrogation of the mechanism shown
in Scheme 2 in the presence of the ligand and provide
clarification of several mechanistic features not highlighted
previously. ReactIR spectroscopy was employed for continuous
monitoring of reaction progress.7 Figure 1 compares the kinetic
stirring speed (200−600 rpm) and oxygen pressure (15−100 psi)
reveal that the reaction rate is zero order in oxygen
concentration.7 The reaction was observed to proceed at
similar rates in NMR tubes using indigenous oxygen present in
nondegassed solvent. These observations confirm that Pd
reoxidation is not the slow step in the cycle, suggesting that a
prior step in the cycle controls the rate.
The concentration dependences and the stability of the
catalyst system may be probed using the “different excess” and
“same excess” protocols, respectively, of Reaction Progress
Kinetic Analysis (RPKA).7,8 A brief description of these
protocols follows. In a reaction as in Scheme 1 exhibiting 1:1
stoichiometry between two substrates, the “excess” [e] is
defined as the difference in concentrations of two reactants, in
this case 1, the arylacetic acid salt, and 2, the olefin (eq 2). The
parameter excess has units of concentration and is invariable
over the course of a catalytic reaction, and hence [e] is fixed by
the initial concentrations chosen for a reaction.
Figure 1. Concentration of substrate 1 as a function of time for the
reaction of Scheme 1 carried out using four different ligand/protecting
group combinations. [1]0 = 0.11 M; [2]0 = 0.44 M; [Pd] = 5.6 mM;
Pd:L = 1:2.
excess = [e] = [2] − [1] = [2] − [1]
(2)
0
0
The logic of how kinetic information is extracted from these
protocols may be illustrated by considering a simple model
reaction rate expression. The rate of most reactions is
dependent on the concentrations of the reactants in some
form. In initial studies of any new reaction or in any case where
the mechanism is uncertain, an empirical description of the
concentration driving forces may be written in what is known as
the power-law form, where x equals the order in 1 and y equals
the order in 2 (eq 3). The experimentally determined
magnitudes of x and y may then be used to provide clues
about the detailed reaction mechanism.
profiles for reactions carried out under identical conditions
using four ligand/protecting group systems. The relative
reactivities of the ligand/protecting group systems are given
in eq 1. The most reactive case differed from the least reactive
by a factor of ca. 1.5, and the basic form of the kinetic profile is
similar in all cases. These rates are considerably faster (by more
than 10-fold) than the reaction in the absence of ligand, as
reported previously.2c Since similar results were obtained for all
four ligand systems, representative results are shown here for
the Boc-Val-OH system; data for the others are found in the
Supporting Information.
x
y
rate = k·[1] ·[2]
(3)
When fit to data from catalytic or other multistep reactions,
non-integer values of reaction orders in substrate concen-
trations are often obtained, which are not directly related to the
molecularity of any individual elementary step in the
mechanism. Power-law expressions may be employed as a
simplification of the more complex algebraic form of
Michaelis−Menten catalytic kinetic rate expressions. The
relationship between the two forms, including use of the
empirical form to delineate the fundamental mechanism, were
recently demonstrated for a proline-mediated aldol reaction.9
Since phenylacetic acid 1 and olefin 2 concentrations are related
to each other through the definition of excess, the rate expression
of eq 3 may be written in terms of a single concentration variable,
as shown in eqs 4 and 5. Equation 5 thus reveals that reaction rate
reaction rate: Ac‐Ile‐OH > Ac‐Val‐OH > Boc‐Val‐OH
(1)
≈ Boc‐Ile‐OH
For all four ligand systems, the reaction was found to be first
order in [Pd], as shown in Figure 2. Since the data in Figure 1
reveal overall kinetics of nearly zero order, the rates in Figure 2
may be obtained from the slopes of the concentration profiles
of reactions carried out under identical conditions except for
varying the catalyst concentration.
Identical rates were observed using Pd:ligand concentrations
of 1:1 and 1:2,7 supporting the proposal that a single ligand
coordinates Pd in the catalytic cycle. Experiments varying the
4601
dx.doi.org/10.1021/ja207634t | J. Am. Chem. Soc. 2012, 134, 4600−4606

Journal of the American Chemical Society
Article
may be measured over the course of the reaction by following the
concentration of only one of the two substrates.
figure. Comparison of the profiles of these two reactions from a
point of identical concentrations of the two substrates is made
possible by shifting curve (a) over to the time point marked.
Two reactions under identical conditions should give identical
temporal profiles from this time point onward. However, it is
easily seen that the reaction of curve (b) proceeds faster than
that of (a) from this point, even though the two exhibit identical
concentrations at this point. The fact that the “time-adjusted”
profile from reaction (a) does not overlay onto that of reaction
(b) from the marked time point onward confirms that some
process other than the intrinsic kinetics related to the temporal
substrate concentrations [1] and [2] contributes to the
observed reaction rate.
We consider that there are two key differences between these
reactions at the point marked by the arrows shown in Figure 3,
despite the fact that substrate concentrations [1] and [2] are
identical at this point: (1) the reaction in run (a) has
undergone a number of turnovers prior to reaching
concentrations that are identical to the initial concentrations
of the reaction in run (b), which has not yet completed any
turnovers at the point marked by the arrows; and (2) the
reaction mixture for run (a) contains product from these
previous turnovers, whereas the reaction mixture in newly
initiated run (b) contains only substrates and catalyst at the
time marked by the arrows. Therefore the most likely
possibilities to rationalize the lack of overlay between the two
curves from the time-adjusted point onward in Figure 3 are
either that catalyst deactivation has occurred in run (a) or that
the catalyst is sensitive to product inhibition.
[2] = [e] + [1]
(4)
x
y
rate = k·[1] ·([e] + [1])
(5)
It may be shown that data from two experiments carried out
at two different values of [e] are mathematically sufficient for a
unique determination of the reaction orders in two substrates
(x and y), which is the theoretical basis of the “different excess”
protocol.8
Equation 5 also makes it clear that for all reactions carried
out at a specific value of [e], the same rate should be obtained
at the same concentration of 1 regardless of its initial
concentration. Thus if two reactions employing the same [e]
and different initial concentrations of 1 do not exhibit the same
rate at a given value of [1], this result provides a clear indication
that a process other than the intrinsic kinetics influences the
reaction rate. The most likely candidates include catalyst
activation or deactivation and substrate or product inhibition,
none of which is accounted for in the intrinsic kinetic rate law.
This provides the theoretical basis of the “same excess”
protocol for probing catalyst stability and process robustness.8
Commonly used kinetic methods typically measure concen-
trations as a function of time, and raw temporal data must be
manipulated mathematically in order to allow a comparison of
reaction rate as a function of substrate concentration, as dictated
by RPKA analysis in both the “same excess” and “different
excess” protocols. Differentiation of temporal concentration data
to obtain rate often results in increased noise in the data, a
drawback that potentially hampers the use of reaction progress
data to their fullest extent in many mechanistic studies. However,
as presented below, simple graphical manipulations of the
temporal profiles may be used in some cases to provide kinetic
information about catalyst stability and concentration depend-
ences without the requirement of mathematical manipulations.
We first illustrate this treatment for the “same excess”
protocol for the reaction of Scheme 1. Figure 3 shows
This latter proposition may be tested by carrying out a
reaction under the conditions of run (b) with the appropriate
quantity of product added so that the concentrations of not
only substrates 1 and 2 but also that of product 3 are matched
to those for run (a) at the time-adjusted point. Figure 4
Figure 4. Concentration of substrate 1 as a function of time for the
reaction of Scheme 1 carried out using the “same excess” protocol as in
Figure 5, with product addition: run (b) has added product to match the
product concentration of run (a) at the point of the time adjustment
(shown by arrows). All other conditions as given in Scheme 1.
Figure 3. Concentration of substrate 1 as a function of time for the
reaction of Scheme 1 carried out using the “same excess” protocol:
runs (a) and (b) have different initial concentration of substrates 1 and
2, with the difference between the two (defined as the excess = [2]0 − [1]0)
held the same. The time-adjusted profile for run (b) has the same [1]
and [2] as run (a) has at the point marked by the arrows. All other
conditions as given in Scheme 1.
compares the concentration profile for a reaction using this
product addition protocol with that of run (a) from Figure 3.
The results reveal excellent superposition of the time-adjusted
same excess experiment with the product added same excess
experiment. Overlay between the two curves shown in Figure 4
demonstrates that the two reactions exhibit the same rate,
providing clear evidence that the catalyst system is sensitive to
concentration profiles for two “same excess” experiments. The
temporal concentrations of 1 and 2 in run (a) are identical to
those in run (b) at the time point marked by the arrows in the
4602
dx.doi.org/10.1021/ja207634t | J. Am. Chem. Soc. 2012, 134, 4600−4606

Journal of the American Chemical Society
Article
product inhibition. In addition, this figure confirms the robustness
of the system to other irreversible catalyst deactivation processes,
which are clearly negligible under these conditions.
The use of temporal concentration profiles using the
“different excess” protocol8 in reaction progress kinetics is
illustrated in Figures 5−7. Again, while the standard RPKA
analysis involves consideration of the data in the form of
reaction rate as a function of concentration, we demonstrate
Figure 7. Concentration profiles for reactions probing the relative rate
suppression caused substrate 2 and product 3.7
determined qualitatively for these conditions by comparing initial
slopes of the temporal profiles (eq 6), giving y ≈ − 0.5.
⎡
log⎢
⎣
⎤
⎡
⎤
⎡
⎤
(rate)
(slope)
[2]
a
a
a
⎥ = log⎢
⎥ = y log⎢
[2]
b
⎥
⎦
(rate)
(slope)
⎦
⎣
⎦
⎣
(6)
b
b
Figure 5. Concentration of substrate 2 as a function of time for the
reaction of Scheme 1 carried out using [2]0 = 0.167 M and two
different initial concentrations of substrate 1: [1]0 = 0.11 M; [1]0 =
0.06 M. The dashed line shows where the reaction in blue circles ends
due to full consumption of 1 while the reaction in green circles
continues due to its higher initial concentration of 1. All other
conditions as given in Scheme 1.
Several further kinetic experiments allow estimation of an
empirical order describing product inhibition. Figure 7 shows
time course profiles for two runs that exhibit nearly identical
rates, carried out with different concentrations of substrate 2
and added product 3. The observed rates for run (a) and run
(b) are equal (eq 7). This allows an empirical power-law rate
expression incorporating product inhibition to be written as
shown in eq 8. Inserting concentrations from the data in Figure 7
and the previously obtained value y = −0.5, we obtain a value of
z = −0.13.
that for particular simplified cases, significant kinetic infor-
mation may be extracted directly from the temporal
concentration profiles without prior conversion to rate. Figure 5
shows the kinetic profile for runs carried out with identical [2]0
and different initial concentrations of 1. The time course of [2]
during the two reactions is identical up to the point of complete
consumption of 1 for the reaction with lower [1]0. The fact that
these profiles overlay at different values of [1] indicates that the
reaction rate is independent of [1].
(rate) = (rate)
(7)
a
b
y
z
y
z
(k′[2] ·[3] ) = (k′[2] ·[3] )
(8)
a
b
Taken together, these kinetic results present an intriguing
mechanistic puzzle. Most catalytic cycles exhibit positive order
kinetics in one or more substrate concentrations as well as in
catalyst concentration, and these quantities are generally
considered to act as driving forces for the reaction. In this
coupling reaction, however, neither of the substrates 1 or 2 nor
the [O2] required to regenerate the catalyst exhibits positive
order kinetics, and in fact, both one of the substrates as well as
the reaction product itself are found to suppress the reaction
rate. Thus the catalyst provides the only positive concentration
driving force in this reaction. The generally accepted
mechanism shown in Scheme 2 cannot account for these
observations.
Next we may transform the empirical power-law form into a
more rigorous proposed steady-state catalytic rate expression by
proposing a mathematical description of elementary steps in a
modified reaction network that takes into account all of these
kinetic observations. The basic elementary steps corresponding
to one proposed network, along with mathematical descriptions
of the intermediate species’ concentrations under steady-state
conditions, are given in Scheme 3. While the kinetic data
cannot provide proposals concerning the nature of these
intermediate species, this mathematical treatment may then be
used to inform further spectroscopic experiments probing the
identity of the proposed species.
Figure 6. Concentration of substrate 1 as a function of time for the
reaction of Scheme 1 carried out using two different initial
concentrations of substrate 2, showing a slower rate for higher [2].
All other conditions as given in Scheme 1.
Figure 6 shows two experiments carried out using the
opposite protocol, that is, with identical [1]0 and different
initial concentrations of 2. Higher [2]0 results in a slower reaction,
indicating an unusual negative dependence on the olefinic substrate
[2]. In this case the magnitude of the exponent y may be
Zero order kinetics in substrate 1 suggests a pre-equilibrium
to form a kinetically meaningful species I1 with coordination of
4603
dx.doi.org/10.1021/ja207634t | J. Am. Chem. Soc. 2012, 134, 4600−4606

Journal of the American Chemical Society
Article
Solving for [Pd] in the mass balance and substituting into the
rate law of eq 9 gives the simplified rate law of eq 11.
Scheme 3. Elementary Steps and On- and Off-Cycle
Intermediates in a Modified Reaction Network
a
k K [1][Pd]
eq,1
2
total
rate =
(1 + K [1](1 + K
eq,1
[2] + K
[3]))
eq,3b
eq,2b
(11)
Making an assumption that the concentration of I1 is much
greater than that of the free Pd within the cycle (i.e., [I1] ≫
[Pd]) simplifies the expression to that of eq 12. The validity of
this assumption is supported by the fact that this form of the
steady-state rate law reconciles the kinetic dependences
observed, that is, zero order kinetics in [1] together with
negative order kinetics in [2] and [3].
k [Pd]
total
2
rate =
(1 + K
[2] + K
[3])
eq,3b
eq,2b
(12)
The data in Figure 7 from reactions carried out with different
concentrations of 2 and 3 also allow estimation of the relative
stability of the two reservoir species 2b and 3b. Since the two
reactions exhibited equal rates, we may equate the catalytic rate
law of eq 12 written for conditions of experiment (a) equal to
that written for conditions of experiment (b) in Figure 7.
Canceling common terms provides eq 13, which may be
rearranged as eq 14. This treatment reveals that binding of the
catalyst to product to form 3b is ca. two times stronger than
binding of the catalyst to olefin to form 2b.
a
See Supporting Information for derivations.
(K
·[3] + K
·[2])
a
eq,3b
eq,2b
1 to Pd. The kinetic data indicate that such a species exists non-
fleetingly prior to C−H cleavage in a step that forms another
intermediate I2 prior to olefin coordination to form I3, which
eliminates the reaction product 3 and forms the reduced Pd
species I4. The observation of zero order kinetics in [O2]
indicates that the only steps prior to reoxidation of I4 are
potentially kinetically meaningful. Negative dependences on
olefin substrate [2] and product olefin [3] suggest off-cycle
binding interactions between these species and the catalyst
forming 2b and 3b that effectively decrease the active Pd
content within the cycle. Keq,2b and Keq,3b represent binding
constants between I1 and 2 or 3, respectively. It is important to
note that the off-cycle reservoir species 2b and 3b must also
contain 1 in order to reconcile the zero order dependence on
[1], as demonstrated in the equations below.
= (K
·[3] + K
·[2])
b
eq,3b
eq,2b
(13)
(14)
K
[2] − [2]
eq,3b
eq,2b
b
a
=
≈ 2
K
[3] − [3]
a b
We are now in a position to compare the empirical power-
law form first developed as eq 8 with the steady-state catalytic
form of eq 12 developed from the mechanistic treatment of
Scheme 3, as shown in eq 15. The usefulness of the empirical
power-law form of the rate law is 2-fold: first, the empirical
assessment of the magnitude of various concentration driving
forces aids in proposing the elementary steps to account for the
observed behavior and hence may be used as a mechanistic
tool; and second, the power-law form provides facility in
selecting conditions for practical scale-up and application of this
chemistry.
The rate of reaction may be written as the first irreversible
step in the network, which is suggested to be the C−H
activation step to form I2 (eq 9). Intermediates formed on
the cycle after this step will be present in low concentra-
tions relative to the other catalytic species (i.e., [I3], [I4] ≪
[Pd], [I1]) and may be neglected in the catalyst mass balance
(eq 10).
k [Pd]
total
2
rate =
(1 + K
[2] + K
−0.13
[3])
eq,3b
eq,2b
−0.5
≈ k′·[2]
·[3]
·[Pd]
total
(15)
The first order dependence on Pd(II) concentration shown
in either form of the rate law is consistent with involvement of a
well-defined mononuclear Pd(II)−amino acid complex as the
active catalyst. The presence of off-cycle catalyst reservoirs
implies that the active steady-state concentration of Pd within
the catalytic cycle remains low throughout the reaction. A
consequence of sequestering Pd in off-cycle reservoirs is that
the lower on-cycle exposure of Pd enables C−H activation and
subsequent steps without promoting irreversible deactivation of
Pd, for example, to form reduced Pd0 in the form of Pd black.
Thus off-cycle reservoirs that suppress rate may play a positive
rate = k [I ] = k K [1][Pd]
2 1 2 eq,1
(9)
[Pd]
total
≈ [Pd] + I + 2b + 3b
1
= [Pd](1 + K [1](1 + K
eq,1
[2]
eq,2b
+ K
[3]))
eq,3b
(10)
4604
dx.doi.org/10.1021/ja207634t | J. Am. Chem. Soc. 2012, 134, 4600−4606

Journal of the American Chemical Society
Article
role in maintaining an effective steady-state concentration of
active catalyst species.11 Further development of this type of
ligand scaffold for well-defined monomeric Pd catalysts,
including tailoring more reactive ligands through tuning the
steric and electronic properties, may help to optimize reactivity
while maintaining the beneficial aspects of competitive olefin
binding prior to the C−H cleavage step.
Spectroscopic Studies. The kinetic studies reveal the
mathematical form of the reaction rate, suggest a plausible set
of elementary steps, and provide clues to the nature of on- and
off-cycle intermediates. We turn to NMR spectroscopic probing
of this system first to probe the role of the ligand in rate
acceleration and then to propose structures for intermediate
species that are consistent with these kinetics and other
experimental and computational results.
1
Figure 9. 500 MHz H NMR spectra at 70 °C in t-AmylOH of (a) 2
(0.1 M), (b) 2 (0.1 M), and Pd(OAc)2 (0.03 M), Ac-Ile-OH (0.03 M),
and KHCO3 (0.1 M), (c) 1 (0.1 M), (d) 1 (0.1 M), Pd(OAc)2 (0.03 M),
Ac-Ile-OH (0.03 M), and KHCO3 (0.1 M), and (e) 1 (0.1 M),
Pd(OAc)2 (0.03 M), Ac-Ile-OH (0.03 M), KHCO3 (0.1 M), and 2
(0.1 M) immediately after the addition of 2.
NMR spectra of the interaction of Pd(OAc)2 and substrate 1
in the absence of ligand revealed the formation of a new
species. Figure 8 shows EXSY correlation indicating equilibrium
begins immediately upon addition of olefin to this system. No
evidence of a Pd hydride species was found under any
conditions.
These results taken together with those of recent computa-
tional studies6 help to rationalize the role of the ligand in
promoting the C−H activation reaction. A mixed acetate Pd
complex formed with substrate 1 in the absence of ligand is
inactive as a catalyst. Presence of the mono-N-protected amino
acid ligand prevents formation of this stable complex via weak
coordination of the ligand. Computational results suggest that
reaction in the presence of ligand may proceed through an N−H
activation pathway rather than through direct C−H activation.6
In the context of our kinetic scenario of Scheme 3, intermediate
species I1 becomes two kinetically indistinguishable species I1a
and I1b resulting from coordination of 1 to the catalyst followed
by N−H activation, as shown in Scheme 4. Prior to N−H
activation, the weakly coordinated species I1a may also undergo
olefin and product coordination to provide the inactive
reservoirs 2b and 3b. This proposed partitioning of the
substrate-bound intermediate I1 into two entities is supported
not only by the computational results but also helps to
rationalize deuterium isotope effects observed previously for
the mono-N-protected amino acid ligand systems.2c While the
reaction in the presence of ligand gave positive isotope effects,
the absolute magnitude of the effect varied from one the ligand
system to another and was not in keeping with the strong
normal isotope effect predicted if the C−H abstraction step is
categorically rate-determining. This effect will be modulated if
both the ligand-dependent N−H activation and the subsequent
C−H activation steps contribute to the rate. It has been
demonstrated that observations of isotope effects via global
kinetics of complex multistep reaction networks may be difficult
to attribute to a single step unless multiple isotope effect
measurements may be made.10 The overall kinetic rate law we
determined holds for the case where I1 exists as two species I1a
and I1b. We propose that subtle differences in the intrinsic rates
of ligand N−H and substrate C−H activation for the different
ligand systems rationalize the kinetic, computational, and
spectroscopic results. Incorporating these features into Scheme
2 results in the revised catalytic cycle of Scheme 4.
Figure 8. 500 MHz NOESY 2-D spectrum of 1 (0.1 M), Pd(OAc)2
(0.03 M), and KHCO3 (0.1 M) in t-AmylOH at 70 °C. Insets
highlighting chemical exchange between protons in different regions:
(a) 6.9−7.4 ppm, aromatic protons, (b) 3.4−3.7 ppm, benzylic
protons, (c) 1.5−1.8 ppm, acetyl protons.
between the free substrate and the new species. These spectra
suggest a 1:1 ratio between two different acetate species. This
suggests that displacement of one acetate from the Pd
precursor by the carboxylate of 1 results in a new mixed-
acetate species 4. This new species appears to be unaffected by
addition of olefin 2, and very little product is formed after three
hours in the absence of the amino acid ligand.
The new species 4 is not formed when the ligand is present
1
in the system. Figure 9 shows H NMR spectra of substrates 1
and 2 alone (Figure 9c and a, respectively) and in the presence
of Pd(OAc)2 and the mono-N-protected ligand Ac-Ile-OH
(Figure 9d and b, respectively). Broadening and shifting of
peaks result upon addition of 1 to the Pd−ligand system, and a
new broad peak appears below 7 ppm (Figure 9d). While no
interaction between the olefin 2 and the Pd−ligand system is
evident in the absence of substrate 1, addition of olefin to the
Pd−ligand−1 system causes significant broadening of the olefin
peaks (compare Figure 9b and Figure 9e). Reaction turnover
4605
dx.doi.org/10.1021/ja207634t | J. Am. Chem. Soc. 2012, 134, 4600−4606

Journal of the American Chemical Society
Article
Scheme 4. Proposed Reaction Mechanism Highlighting Dual
Rate-Determining Steps and Off-Cycle Catalyst Reservoirs
AUTHOR INFORMATION
Corresponding Author
blackmond@scripps.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
J.-Q.Y. gratefully acknowledges support from National Science
Foundation under the Center of Chemical Innovation in
Stereoselective C−H Functionalization (CHE-0943980). We
acknowledge D.-H. Huang and L. Pasternack (TSRI NMR
Facility) for valuable assistance with NMR spectroscopy and
Dr. J. Bures for help with NMR data analysis.
REFERENCES
■
(1) For selected reviews of Pd-catalyzed C−H functionalization, see:
(a) Campeau, L.-C.; Stuart, D. R.; Fagnou, K. Aldrichimica Acta 2007,
40, 35. (b) Satoh, T.; Miura, M. Chem. Lett. 2007, 3, 200. (c) Seregin,
I. V.; Gevorgyan, V. Chem. Soc. Rev. 2007, 36, 1173. (d) Chen, X.;
Engle, K. M.; Wang, D.-H.; Yu, J.-Q. Angew. Chem., Int. Ed. 2009, 48,
5094. (e) Ackermann, L.; Vicente, R.; Kapdi, A. R. Angew. Chem., Int.
Ed. 2009, 48, 9792. (f) Daugulis, O.; Do, H.-Q.; Shabashov, D. Acc.
Chem. Res. 2009, 42, 1074. (g) Lyons, T. W.; Sanford, M. S. Chem. Rev.
2010, 110, 1147. (h) Jia, C.; Kitamura, T.; Fujiwara, Y. Acc. Chem. Res.
2001, 34, 633.
(2) (a) Wang, D.-H.; Engle, K. M.; Shi, B.-F.; Yu, J.-Q. Science 2010,
327, 315. (b) Engle, K. M.; Wang, D.-H.; Yu, J.-Q. Angew. Chem., Int.
Ed. 2010, 49, 6169. (c) Engle, K. M.; Wang, D.-H.; Yu, J.-Q. J. Am.
Chem. Soc. 2010, 132, 14137.
(3) (a) Lu, Y.; Wang, D.-H.; Engle, K. M.; Yu, J.-Q. J. Am. Chem. Soc.
2010, 132, 5916. (b) Lu, Y.; Leow, D.; Wang, X.; Engle, K. M.; Yu,
J.-Q. Chem. Sci. 2011, 2, 967. (c) Dai, H.-X.; Stepan, A. F.; Plummer,
M. S.; Zhang, Y.-H.; Yu, J.-Q. J. Am. Chem. Soc. 2011, 133, 7222.
(d) Huang, C.; Chattopadhyay, B.; Gevorgyan, V. J. Am. Chem. Soc.
2011, 133, 12406. (e) Engle, K. M.; Thuy-Boun, P. S.; Dang, M.; Yu,
J.-Q. J. Am. Chem. Soc. 2011, 133, 18183. (f) Novak, P.; Correa, A.;
́
Gallardo-Donaire, J.; Martin, R. Angew. Chem., Int. Ed. 2011, 50,
12236.
(4) (a) Ferreira, E. M.; Stoltz, B. M. J. Am. Chem. Soc. 2003, 125,
9578. (b) Stuart, D. R.; Fagnou, K. Science 2007, 316, 1172. (c) Zhang,
Y.-H.; Shi, B.-F.; Yu, J.-Q. J. Am. Chem. Soc. 2009, 131, 5072.
(d) Izawa, Y.; Stahl, S. S. Adv. Synth. Catal. 2010, 352, 3223.
(5) (a) Hickman, A. J.; Sanford, M. S. ACS Catal. 2011, 1, 170.
(b) Jintoku, T.; Taniguchi, H.; Fujiwara, Y. Chem. Lett. 1987, 16, 1159.
(c) Eberson, L.; Jonsson, E. Acta Chem. Scand. Ser. B 1974, 28, 771.
(d) Ye, M.; Gao, G.-L.; Yu, J.-Q. J. Am. Chem. Soc. 2011, 133, 6964.
(6) Musaev, D. G.; Kaledin, A.; Shi, B.-F.; Yu, J.-Q. J. Am. Chem. Soc.
2012, 134, 1690.
CONCLUSIONS
■
In summary, detailed kinetic studies based on reaction progress
concentration profiles coupled with NMR spectroscopic studies
provide an expanded mechanistic picture of the C−H activation
reaction of Scheme 1. Suppressing formation of a stable mixed
acetate species by introduction of mono-N-protected amino
acid ligands accounts for the observed rate acceleration. Novel
graphical manipulations of the data allow information about
reaction orders to be extracted from concentration profiles
without the requirement to convert the data to reaction rate.
The empirical power-law form of the rate expression aids in
proposing a series of elementary steps. Consideration of the
steady-state rate expression in conjunction with reaction
progress data allows rationalization of the unusual rate
observations,12 as well as clarifying observed differences in
kinetic isotope effects for different ligand systems. Zero order
dependence on [1] and [O2], first order dependence on [Pd],
and negative dependences on [2] and [3] are attributed to the
presence of off-cycle reservoirs containing the substrate and
product olefin species bound to a Pd carboxylate formed from
interaction of 1 with the catalyst. While these off-cycle species
suppress reaction rate, they may also play a role in maintaining
catalyst stability by suppressing irreversible catalyst deactiva-
tion. The proposal that the rate-determining step involves an
interplay between N−H and C−H activation processes helps to
explain both the similar form of the rate expression, as well as
the observed differences in absolute magnitude of rate, for the
different amino acid ligands.
(7) See Supporting Information for details.
(8) (a) Blackmond, D. G. Angew. Chem., Int. Ed. 2005, 44, 4302.
(b) Mathew, J. S.; Klussmann, M.; Iwamura, H.; Valera, F.; Futran, A.;
Emanuelsson, E. A. C.; Blackmond, D. G. J. Org. Chem. 2006, 71, 4711.
(9) Zotova, N.; Broadbelt, L. J.; Armstrong, A.; Blackmond, D. G.
Bioorg. Med. Chem. Lett. 2009, 19, 3934.
(10) Jones, W. D. Acc. Chem. Res. 2003, 36, 140.
(11) Rosner, T.; Le Bars, J.; Pfaltz, A.; Blackmond, D. G. J. Am. Chem.
Soc. 2001, 123, 1848.
(12) Unusual concentration dependences were also observed in Pd(II)-
mediated C−H functionalization: Deprez, N. R.; Sanford, M. S. J. Am.
Chem. Soc. 2009, 131, 11234.
ASSOCIATED CONTENT
* Supporting Information
■
S
Details of the experimental procedures and kinetic analysis.
This material is available free of charge via the Internet at
http://pubs.acs.org.
4606
dx.doi.org/10.1021/ja207634t | J. Am. Chem. Soc. 2012, 134, 4600−4606