Cationic intermediates in the Pd-catalyzed negishi coupling. Kinetic and density functional theory study of alternative transmetalation pathways in the Me-Me coupling of ZnMe2 and trans -[PdMeCl(PMePh2) 2]

Article abstract of DOI:10.1021/ja204256x

The complexity of the transmetalation step in a Pd-catalyzed Negishi reaction has been investigated by combining experiment and theoretical calculations. The reaction between trans-[PdMeCl(PMePh₂)₂] and ZnMe₂ in THF as solvent was analyzed. The results reveal some unexpected and relevant mechanistic aspects not observed for ZnMeCl as nucleophile. The operative reaction mechanism is not the same when the reaction is carried out in the presence or in the absence of an excess of phosphine in the medium. In the absence of added phosphine an ionic intermediate with THF as ligand ([PdMe(PMePh₂)₂(THF)]⁺) opens ionic transmetalation pathways. In contrast, an excess of phosphine retards the reaction because of the formation of a very stable cationic complex with three phosphines ([PdMe(PMePh₂)₃]⁺) that sequesters the catalyst. These ionic intermediates had never been observed or proposed in palladium Negishi systems and warn on the possible detrimental effect of an excess of good ligand (as PMePh₂) for the process. In contrast, the ionic pathways via cationic complexes with one solvent (or a weak ligand) can be noticeably faster and provide a more rapid reaction than the concerted pathways via neutral intermediates. Theoretical calculations on the real molecules reproduce well the experimental rate trends observed for the different mechanistic pathways.

Full text of DOI:10.1021/ja204256x

ARTICLE
pubs.acs.org/JACS
Cationic Intermediates in the Pd-Catalyzed Negishi Coupling. Kinetic
and Density Functional Theory Study of Alternative Transmetalation
Pathways in the MeÀMe Coupling of ZnMe2 and
trans-[PdMeCl(PMePh2)2]
Max García-Melchor,† Beatriz Fuentes,‡ Agustí Lledꢀos,† Juan A. Casares,*,‡ Gregori Ujaque,*,† and
Pablo Espinet*,‡
†Química Física, Edifici C.n., Universitat Autꢁonoma de Barcelona, E-08193 Bellaterra, Catalonia, Spain
‡IU CINQUIMA/Química Inorgꢀanica, Facultad de Ciencias, Universidad de Valladolid, E-47071 Valladolid, Spain
S Supporting Information
b
ABSTRACT: The complexity of the transmetalation step in a Pd-
catalyzed Negishi reaction has been investigated by combining
experiment and theoretical calculations. The reaction between
trans-[PdMeCl(PMePh2)2] and ZnMe2 in THF as solvent was ana-
lyzed. The results reveal some unexpected and relevant mechanistic
aspects not observed for ZnMeCl as nucleophile. The operative
reaction mechanism is not the same when the reaction is carried out
in the presence or in the absence of an excess of phosphine in the
medium. In the absence of added phosphine an ionic intermediate with
THF as ligand ([PdMe(PMePh2)2(THF)]+) opens ionic transmetalation pathways. In contrast, an excess of phosphine retards the
reaction because of the formation of a very stable cationic complex with three phosphines ([PdMe(PMePh2)3]+) that sequesters the
catalyst. These ionic intermediates had never been observed or proposed in palladium Negishi systems and warn on the possible
detrimental effect of an excess of good ligand (as PMePh2) for the process. In contrast, the ionic pathways via cationic complexes
with one solvent (or a weak ligand) can be noticeably faster and provide a more rapid reaction than the concerted pathways via
neutral intermediates. Theoretical calculations on the real molecules reproduce well the experimental rate trends observed for the
different mechanistic pathways.
’ INTRODUCTION
different isomer (trans or cis, respectively) of the coupling
intermediate [PdRfMe(PPh3)2].17 The study revealed also the
existence of secondary undesired transmetalations (methyl by
aryl exchanges, eqs 1 and 2), which could eventually produce
homocoupling products. The same phenomenon had been noted
by van Asselt and Elsevier,18 and after our report was also
observed by A. Lei et al. on related reactions with aryl zinc
derivatives.19
The Negishi reaction is a wide scope most reliable cross-
coupling process that can be applied to every possible combina-
tion of carbon type (sp, sp2, or sp3), and tolerates many different
functions in the reagents.1À3 The reaction can be carried out
using either ZnR2 or ZnRX as the nucleophile, and the more
accessible reagent is usually chosen. While the choice of orga-
nozinc does not seem to alter the cross coupling products in the
Negishi coupling, the rate and mechanism of the transmetalation
step might be affected. However, very little is known about this
transmetalation, compared to the transmetalation mechanisms in
the Stille4À8 or Suzuki9À12 reactions, in spite of the fact that the
transmetalation between organozinc and palladium complexes is
also involved in other synthetically useful processes such as the
hydroalkylation of styrenes,13 the asymmetric allylation of aryl
aldehydes,14 the coupling of propargylic benzoates and aldehy-
des,15 or the double-transmetalation oxidative cross-coupling
reaction.16
Recently, we have published kinetic experimental and theore-
tical studies of the transmetalation between ZnMeCl and
trans-[PdMeCl(PMePh2)2].20 That study showed that the
In an initial mechanistic work on the Negishi reaction we met a
remarkable behavior in the transmetalation of trans-[PdRfCl-
(PPh3)2] (Rf = 3,5-dichloro-2,4,6-trifluorophenyl) with ZnMe2
or with ZnMeCl: apparently each methylating reagent afforded a
Received: May 9, 2011
Published: July 27, 2011
r
2011 American Chemical Society
13519
dx.doi.org/10.1021/ja204256x J. Am. Chem. Soc. 2011, 133, 13519–13526
|

Journal of the American Chemical Society
ARTICLE
Scheme 1. Simplified Reaction Mechanism of the Negishi
Scheme 2
Coupling with ZnMeCl
the more stable cis-[PdMe2(PMePh2)2]. It was demonstrated
that the latter transformation is not a direct trans/cis isomeriza-
tion, which is slow, but occurs via retrotransmetalation to the
initial reagents followed by transmetalation to cis-[PdMe2-
(PMePh2)2].21 The catalytic cycle in Scheme 1 summarizes
those results. No other Pd complex was observed in that study.20
In this paper we study the transmetalation using ZnMe2 as the
transmetalating nucleophile. A priori only minor differences with
the results with ZnMeCl were expected. However, this study has
revealed unknown and unexpected aspects of the Negishi
process.
’ RESULTS AND DISCUSSION
1. Experimental Studies. The reaction of 1 with ZnMe2 (1:20
ratio simulating catalytic conditions with 5% Pd, in THF at room
temperature) gives cis-[PdMe2(PMePh2)2] (2) as the only
observable product. In these conditions complex 2 slowly
decomposes to give the Negishi coupling product ethane.22,23
This behavior is in coincidence with the synthetic results
observed using ZnMeCl at room temperature.20 However,
important mechanistic differences between the two Zn reagents
were observed when the reactions were monitored by 31P NMR
at 203 K (at which temperature the coupling rate to give ethane is
negligible), with and without added phosphine (Figure 1).
1.1. Occurrence of Cationic Pd Species. The first unexpected
observation (Figure 1A) was that the rate of consumption of
trans-[PdMeCl(PMePh2)2] (1) with ZnMe2 seemed too high,
compared to our previous study with ZnMeCl,20 even consider-
ing that ZnMe2 is a stronger nucleophile than ZnMeCl and some
rate acceleration was to be expected. Addition of a small amount
of PMePh2 markedly decreased the rate of consumption
(Figure 1B), bringing the reaction rate into the range convenient
for kinetic studies by 31P NMR. Interestingly, further increases in
PMePh2 concentration affected only slightly the rate of con-
sumption (Figure 1CÀE) and the profiles of the other curves in
the graphics (these are discussed later). This behavior suggests
the existence, in the initial conditions of reaction 1A, of a minute
proportion of a nonobserved catalytic intermediate that opens a
much faster reaction pathway; this intermediate would be
quenched by addition of just a small amount of phosphine.
A most plausible intermediate is the cation trans-[PdMe-
(PMePh2)2(THF)]+ (THF = tetrahydrofuran, 4+), formed in
the presence of a large amount of ZnMe2 acting as ClÀ scavenger
(Scheme 2). As a precedent for this proposal, the role of ZnBr2 as
a Lewis acid, facilitating halide abstraction from Ni complexes,
was hypothesized by Buchwald et al.,24 who found an accelerating
effect of ZnBr2 as cocatalyst in a process of enantioselective
R-arylation of R-substituted γ-butyrolactones requiring a step of
Figure 1. Concentration/time data, obtained by 31P NMR, for the
reaction of trans-[PdMeCl(PMePh2)2] (1) with ZnMe2 in THF at
203 K. Starting conditions: [1]0 = 1.0 Â 10À2 M, [ZnMe2]0 = 0.21 M.
Added phosphine: (A) [PMePh2] = 0 M, (B) [PMePh2] = 6.0 Â 10À4
M, (C) [PMePh2] = 2.0 Â 10À3 M, (D) [PMePh2] = 5.0 Â 10À3 M, (E)
[PMePh2] = 1.0 Â 10À2 M.
reagents (ZnMeCl + trans-[PdMeCl(PMePh2)2] (1)) are more
stable than the products (ZnCl2 + cis-[PdMe2(PMePh2)2] (2)),
and the latter are more stable than the alternative products
(ZnCl2 + trans-[PdMe2(PMePh2)2] (3)). The Gibbs energy
differences are such that, in catalytic conditions (using a large
excess of ZnMeCl), the equilibria are shifted toward the most
stable products ZnCl2 + cis-[PdMe2(PMePh2)2], but the kinetics
of the reactions favor the faster formation of trans-[PdMe2-
(PMePh2)2], which is seen to appear and then fade out in favor of
13520
dx.doi.org/10.1021/ja204256x |J. Am. Chem. Soc. 2011, 133, 13519–13526

Journal of the American Chemical Society
ARTICLE
would not be formed in solutions very rich in species containing
terminal chloro donor atoms, which will coordinate palladium in
preference to THF.27
The proposal of intermediate 4+ was reinforced by the ob-
servation of a cationic complex (5+ in Scheme 2) when PMePh2
was added (Figure 1BÀE). It was identified as [PdMe(PMePh2)3]+
by comparison with an authentic sample prepared independently
(eq 3).
The observation of 5+, in variable concentration depending on
the amount of phosphine added, confirms unequivocally that, in
the presence of added phosphine, a second cationic species, this
time observable, is made available from the very beginning of the
reaction, opening a plausible additional alternative transmetala-
tion pathway. In summary, depending on the reaction conditions
the transmetalation could take place on 1, on 4+, or on 5+
(Scheme 2). It looks that the reaction via 4+ is faster than on 1,
while the presence of 5+ does not accelerate the transmetalation.
1.2. Kinetic Studies of the Transmetalation with ZnMe2,
without Added PMePh2. The high transmetalation rate observed
in Figure 1A (no added PMePh2) leads to the complete trans-
formation of the starting complex trans-[PdMeCl(PMePh2)2]
(1) into a mixture of the transmetalated products trans- and
cis-[PdMe2(PMePh2)2] (3 and 2, respectively) where the trans
isomer is very major. This takes place in about one minute, which
is very fast considering that the working temperature is 203 K.
Then a very slow isomerization of 3 to 2 follows.
1.3. Kinetic Studies of the Transmetalation with ZnMe2 and
Added PMePh2. In order to observe the products involved in
catalysis, these reactions were carried out at 203 K to quench the
fast exchange of coordinated and uncoordinated phosphine that
already occurs at 223 K (Figure 2).
Figure 2. 31P NMR of a mixture of 1, ZnMe2, and PMePh2 in THF at
203 and 223 K, after 15 min, close to the end of the transmetalation.
Starting conditions: [1]0 = 0.033 M, [ZnMe2]0 = 0.16 M, [PMePh2] =
0.008 M. The coalescence of the signals at 223 K is due to exchange of
coordinated and uncoordinated phosphine. The signal of the free
phosphine appears at À29.5 ppm at 203 K, and is not observed at 223 K.
The reactions in the presence of added PMePh2 (Figure 1BÀD)
show the fast formation of the cationic species 5+, andamuchslower
rate of transmetalation than without added phosphine, in agreement
with the absence of 4+. The graphics show that, as usual, trans isomer
3 is formed faster than cis isomer 2. In order to decide whether the
transmetalation takes place on 1, on 5+, or competitively on both
species, the experimental data collection was fitted to two kinetic
models: (a) supposing that the transmetalation operates only on the
neutral complexes 1 (eqs 4 and 5); and (b) considering also the
competitive transmetalation operating on the cationic complex
[PdMe(PMePh2)3]+ (5+) (eqs 6 and 7). The results are given in
Tables 1 and 2. The fitting using model b affords negligible values for
halide abstraction from [(S)-BINAP]Ni(Ar)(X), and also by
Majundar and Cheng in the Ni(II)/Zn-mediated chemoselective
arylation of aromatic aldehydes.25 Actually, the formation of
[Ni(η3-Bz)(diphosphine)][ZnBr3(THF)] complexes from
[Ni(σ-Bz)Br(diphosphine)] and ZnBr2 in THF has been re-
ported by Anderson and Vicic.26
To the best of our knowledge this instrumental effect of ZnMe2
acting as ClÀ scavenger has never been observed nor proposed in
palladium Negishi systems where, probably due to the different
hardness of the metal center, the behavior is somewhat different
from nickel. A THF complex similar to 4+, trans-[Pd(C6F5)-
(PPh3)2(THF)](OTf), in equilibrium with trans-[Pd(C6F5)-
(OTf)(PPh3)2] in THF,6,7 has been reported, but TfOÀ is a
very good leaving ligand. Moreover, treating trans-[PdMeCl-
(PMePh2)2] in THF solution at À20 °C with AgBF4, to force
the extraction of ClÀ (see Supporting Information for details),
k+ ; thus, the competitive transmetalation pathway via the tris-
3
phosphine cationic complex 5+ can be discarded. Consistently, almost
identical values are obtained for k3 and for k4 in models a and b. The
two rate constants, with average values k3 ≈ 2.8 Â 10À2 minÀ1, and
k4 ≈ 6.2 Â 10À3, differ in less than 1 order of magnitude.
[PdMe(PMePh2)2(THF)](BF4) (4 BF4) is formed quantita-
3À
tively. When the counteranion is Cl an equilibrium between
trans-[PdMeCl(PMePh2)2]andtrans-[PdMeCl(PMePh2)2(THF)]+
(4+) seems to be established to a small extent, but only in the
presence of ZnMe2 as ClÀ acceptor. This accelerating effect was
not detected in the reactions where the reagent was the better
acceptor ZnMeCl,20 but this has some logic: intermediate 4+
13521
dx.doi.org/10.1021/ja204256x |J. Am. Chem. Soc. 2011, 133, 13519–13526

Journal of the American Chemical Society
ARTICLE
Table 1. Rate Constants (minÀ1 molÀ1L) and ΔG‡ values (kcal molÀ1) at 203 K, Using Model Aa
b
‡
‡
% added PMePh2
k3
ΔG3
k4
ΔG4
6
3.0((0.1) Â 10À2
3.3((0.1) Â 10À2
1.9((0.1) Â 10À2
2.2((0.1) Â 10À2
14.80 ((0.07)
14.77 ((0.07)
14.99 ((0.08)
14.93 ((0.08)
7.5((0.2) Â 10À3
7.7((0.2) Â 10À3
4.1((0.1) Â 10À3
5((1) Â 10À3
15.36 ((0.08)
15.37 ((0.08)
15.61 ((0.08)
15.53 ((0.08)
20
50
100
a Sandstr€om, J. Dynamic NMR Spectroscopy; Academic Press: London, 1982. The fittings were made using the values k5 = 9.8 minÀ1 molÀ1L; kÀ5 = 0.002
minÀ1 molÀ1 L for the formation of 5+ (eq 6). These were determined in the experiment in Figure 1E, where the experimental values to be measured for
formation of 5+ are larger, and hence more reliable. This also applies to Table 2. b Standard deviations in parentheses.
Table 2. Rate Constants at 203 K, Using Model ba
trans-[PdMeCl(PPh2Me)2], reaching equilibrium (1:2 ≈ 43:57) in
200 min.
(%) added
PMePh2
k3
k4
k+
3
2. DFT Studies of the Transmetalation. In order to support
the proposals based on the experimental data and obtain a more
detailed picture of the transmetalation process, all the proposed
transmetalation pathways were computationally investigated by
density functional theory (DFT) calculations using the M06
dispersion-corrected functional, which has been shown to pro-
vide good results for both organometallic systems and noncova-
lent interactions.29 The theoretical calculations were performed
on the very catalyst used for the experiments (i.e., with the real
ligands). The solvent, THF, was introduced by a discrete-
continuum model: two THF molecules were explicitly included
in the calculations as potential ligands, and the effect of the
bulk solvent was taken into account with a continuum model
(ε = 7.4257). This methodology allowed us to consider tetrahedral
Zn species coordinated with THF along the studied pathways,
rather than the unrealistic linear dicoordinated Zn species often
used that lead to calculated species of unlikely existence in THF
solution.30 Gibbs energies in THF solution (ΔGTHF) were
calculated adding the gas-phase Gibbs energy corrections of the
solute to the energies in solution. All the energies collected in the
text are ΔGTHF, calculated at 203 K for further approximation to
the experimental conditions.31 With these DFT calculations we
expected to reproduce the trends of energy barriers found
experimentally,32 which are the following: ionic to trans (PMePh2)
> concerted to cis > concerted to trans > ionic to trans (THF).
2.1. Ionic Mechanism via an Undetectable Cationic Inter-
mediate [PdMe(PMePh2)2(THF)]+ (4+). The experimental stud-
ies suggest the formation of an undetected cationic intermediate
[PdMe(PMePh2)2(THF)]+ (4+) (with [ZnMe2Cl(THF)]À as
the counterion), which is the one undergoing the fastest trans-
metalation process. The computational analysis of a plausible
transmetalation via 4+ afforded the reaction profile and opti-
mized transition state structures shown in Figure 3. The reaction
pathway starts with a chloride abstraction from the initial Pd
complex (1), assisted by a THF molecule and ZnMe2 (which is in
large excess). This results in the formation of IT2 (correspond-
ing to the proposed intermediate 4+ in the experimental nomen-
clature) through the transition state TsIT1.33 Interestingly, the
position of IT2 in the Gibbs energy profile is 5.3 kcal molÀ1
higher than the initial reactants, which makes its observation by
NMR extremely unlikely.34 Subsequently, the transmetalation
takes place from IT2 via TsIT2, yielding the last intermediate
IT3. The latter transition state has the highest energy value
(10.3 kcal molÀ1) within the overall reaction profile thus
becoming the rate-determining step.
(minÀ1 molÀ1 L)
(minÀ1 molÀ1 L)
(minÀ1
)
6
3((>3) Â 10À2
3.4((1) Â 10À2
2.7((2) Â 10À2
2((>2) Â 10À2
8((>8) Â 10À3
8((2) Â 10À3
4((1.6) Â 10À3
6((>6) Â 10À3
2((>2) Â 10À16
1((>1) Â 10À4
3((>3) Â 10À15
2((>2) Â 10À16
20
50
100
a See note text in Table 1. Standard deviations in parentheses.
In summary, the experimental results show the following: (1)
In the absence of added phosphine the observed rate is unusually
fast for the transmetalation that should take place on
trans-[PdMeCl(PMePh2)2],28 and this suggests that it must be
going through a nonobserved cationic intermediate 4+. This
proposal is later supported by DFT calculations. (2) In the
presence of some added phosphine the reactions go via
trans-[PdMeCl(PMePh2)2]. The presence of a new cationic
intermediate 5+ is kinetically of little relevance (except for the
detrimental effect associated to the percentage of Pd scavenging),
indicating that it is not an intermediate toward a faster reaction
pathway. (3) At variance with the system using ZnMeCl, where
the reaction equilibrium is displaced toward the reagents (1 +
ZnMeCl, see Figure 1),20 in the system 1 + ZnMe2 the reaction is
displaced toward the products (2 + ZnMeCl, and 3 + ZnMeCl),
reflecting the higher nucleophilicity of ZnMe2. Methyl for
chloride exchange is an exergonic process for ZnMe2 and an
endergonic one for ZnMeCl, reflecting the relative stability of the
Zn complexes. Since it is already known that 2 is about 2 kcal
molÀ1 more stable than 3,20 it follows that the reaction equilib-
rium with ZnMe2 is strongly displaced toward 2. (4) The transme-
talation to trans (3) is clearly faster than the transmetalation to cis (2),
in coincidence with the reaction with ZnMeCl.20 (5) The subsequent
isomerization of 3 to 2 is much slower than in the reaction with
ZnMeCl. This was expected from our previous study, which showed
that the rate of the trans/cis uncatalyzed isomerization is negligible at
low temperatures, while the isomerization via retrotransmetalation to
trans-[PdMeCl(PMePh2)2] is catalyzed by the presence of ZnCl2
and, less efficiently, by ZnMeCl.20 Additional experiments (at 223 K,
in order to increase the isomerization rates) show that ZnMe2 does
not catalyze the isomerization: after 13 h at 223 K a solution 0.01 M of
3 and 0.21 M of ZnMe2 contains only 1% of the cis isomer 2; in
contrast, the same experiment performed with ZnMeCl gives, after 10
min, a mixture of about 25% cis-[PdMe2(PMePh2)2] and 75%
2.2. Concerted Mechanisms. The only operative transmetala-
tion pathways for the transmetalation reaction with ZnMeCl involve
concerted Cl for Me exchanges.20 A similar concerted mechanism
13522
dx.doi.org/10.1021/ja204256x |J. Am. Chem. Soc. 2011, 133, 13519–13526

Journal of the American Chemical Society
ARTICLE
Figure 4. Gibbs energy profile in THF (in kcal molÀ1) calculated for
the concerted transmetalation to the trans product (3). The optimized
structure for the transition state TsCT is shown. Phenyl rings simplified
for clarity.
Figure 3. Gibbs energy profile in THF (in kcal molÀ1) calculated for
the transmetalation to the trans product (3) via Pd ionic intermediate
with THF. The optimized structures for the transition states TsIT1
(left) and TsIT2 (right) are shown. Phenyl rings simplified for clarity.
stable, as expected from its NMR observation. For the ionic pathway
through 5+ the calculated transmetalation barrier TsIT2 increases to
15.5 kcal molÀ1. This is more than 3 kcal molÀ1 higher than that
calculated for the concerted processes (11.8À10.7 kcal molÀ1, see
above), which in terms of reaction rates means more than 3 orders
of magnitude slower. In other words, this pathway is clearly the most
disfavored one for the Negishi transmetalation studied, in agreement
with the experimental observations.
was found for ZnMe2. Figures 4 and 5 depict the Gibbs energy
profiles of the concerted transmetalation pathways on trans-
[PdMeCl(PMePh2)2] (1) to trans- and cis-[PdMe2(PMePh2)2],
respectively. Both mechanisms share the initial dissociation of a THF
molecule from the Zn reagent with concomitant chloride coordina-
tion to Zn, yielding intermediates CT1 and CC1, respectively.35 The
concerted transmetalation to the trans product 3 (Figure 4) takes
place through one transition state (TsCT), with an energy barrier
of10.7 kcalmolÀ1. Atvariance, the transmetalation tocis product2
(Figure 5) displays two transition states of almost identical energy
(TsCC1 =11.5; TsCC2 = 11.8 kcal molÀ1). TsCC1 corresponds
to the substitution of the phosphine ligand by a methyl group from
the Zn reactant, giving rise to a high energy intermediate CC2,
unobservable by NMR. In TsCC2, the phosphine ligand replaces
the chloride group, which ends up bound to Zn.
From the calculated barriers, the concerted transmetalation to
trans is slower than the transmetalation through the ionic
intermediate 4+, and faster than the concerted transmetalation
to cis. On the other hand, the cis isomer 2 is 2.3 kcal molÀ1 more
stable than the trans isomer 3.
2.3. Ionic Mechanism via a Detected Cationic Intermediate
[PdMe(PMePh2)3]+ (5+). Finally, in order to unravel the observed
decrease of the reaction rates for the transmetalation process with
added phosphine, and further support the observation of the species
[PdMe(PMePh2)3]+[ZnMe2Cl(THF)]À (5+), we computed the
transmetalation pathway via this species (Figure 6). The pathway
is analogous to that going through 4+, but very different energies
are involved. The calculations indicate that intermediate IT2
(corresponding to 5+ in the experimental nomenclature) is very
3. Comparing Theoretical and Experimental Results. The
number of computational studies of transition metal catalyzed
reactions has increased exponentially in recent years,36 but only
rarely are experimental energy values (especially transition
barriers) available for direct comparison with the theoretical
results. When they are, it is found that the quantitative matching
is not always as good as hoped.37À39 It has been stated that, at
present, it cannot be excluded that good matches stem from error
cancellation of different contributions to the calculated values.38
The experimental ΔG values derived from the kinetic experi-
mental studies are estimated to have an error of (0.08 kcal mol-1.40
Aiming at obtaining theoretical results to compare with the
experimental values, we have used in this paper an optimum
chemical model for calculations. The reactant molecules were
not substituted by a simplified model: the complete catalyst with
the real ligands and realistic tetracoordinated Zn species includ-
ing explicitly solvent molecules were used. Moreover, a well
balanced computational cost/quality method was selected by
using a dispersion-corrected functional with a large basis set and
including a discrete + continuum solvent scheme (the new
implemented SMD solvation model) in the calculations.
Simplified experimental profiles for the concerted transmeta-
lations to trans and to cis using ZnMe2 are plotted in black in
Figure 7. The values for the concerted mechanisms are the
quantitative kinetic results, whereas for the ionic mechanisms the
13523
dx.doi.org/10.1021/ja204256x |J. Am. Chem. Soc. 2011, 133, 13519–13526

Journal of the American Chemical Society
ARTICLE
Figure 7. Comparison, for the reactions with ZnMe2, of the relative
Gibbs energy values at 203 K (kcal molÀ1) of the rate determining
transition states and the products. Experimental values in black, calcu-
lated values in red.
Scheme 3. Simplified Reaction Mechanism of the Negishi
Coupling with ZnMe2
Figure 5. Gibbs energy profile in THF (in kcal molÀ1) calculated for
the concerted transmetalation to the cis product (2). The optimized
structures of the transition states TsCC1 (left) and TsCC2 (right) are
shown.
values plotted are the estimated maximum (for L = THF) or
minimum (for L = PMePh2) value compatible with the experi-
mental observations. The data in red correspond to the compu-
tationally obtained values. First of all, it is worth noting that the
calculations reproduce qualitatively the experimental observa-
tions. From the kinetic point of view, the order of the energy
barriers for the transmetalation process through the different
pathways is as follows: ionic to trans (PMePh2) > concerted to
cis > concerted to trans > ionic to trans (THF). As far the
thermodynamics is concerned, the order of stability of the
reagents and products in equilibrium (as represented by the Pd
complex contained in the corresponding combination of
compounds) is 2 (cis) > 3 (trans) > 1.
From a strictly quantitative perspective, we can compare
theoretical and experimental values only for the concerted
mechanisms, for which we have specific experimental values
(for the ionic mechanisms specific values could not be obtained
from experiments, and only lower (for 4+) or higher (for 5+)
limits can be estimated). On the basis of these results it seems
that our theoretical calculations underestimate the reaction barriers
by about 4 kcal molÀ1. However, the calculations nicely reproduce
the energy difference between both concerted mechanisms
(the difference is 0.6 kcal molÀ1 for the experimental values and
1.1 kcal molÀ1 for the theoretical values), which is important, and
sufficient in this case to compare reaction rates and elucidate which
Figure 6. Gibbs energy profile in THF (in kcal molÀ1) calculated for
the transmetalation to the trans product (3) via ionic intermediate with
PMePh2. Gibbs energies are given in kcal molÀ1. Phenyl rings simplified
for clarity.
13524
dx.doi.org/10.1021/ja204256x |J. Am. Chem. Soc. 2011, 133, 13519–13526

Journal of the American Chemical Society
ARTICLE
pathway is preferred. On the other hand, the relative stability for
the products 2 and 3 could not be experimentally quantified in
this paper, but we had determined in a previous study that 2 is about
2.0 kcal molÀ1 more stable than 3.20 In this study the calculated
difference is 2.3 kcal molÀ1, which is a very satisfactory fit with the
experiment.
propitiate ionic transmetalation mechanisms. In reactions with
aryls, a less donating Ar group on Pd, instead of Me, is expected
to be somewhat less prone to produce ionic species, whereas
heavier halogens will propitiate ionic Pd intermediates better
than Cl. Concerning solvents, those more coordinating solvents
than THF will also be more favorable for ionic mechanisms. All
these circumstances can now be considered, in the framework of
our reaction scheme of competitive pathways, when planning
new synthesis.
It is worth recalling that this study concerns the transmetala-
tion step only. The oxidative addition or the reductive elimina-
tion steps are required for a successful cross coupling, but can
turn out to be rate determining. This should not be overlooked.42
’ CONCLUSIONS
The experimental study of the transmetalation step in a Pd-
catalyzed Negishi reaction to produce MeÀMe coupling using
ZnMe2, in THF as solvent, shows that in all cases the reactions
are faster than with ZnMeCl. The concerted transmetalation to
trans is faster than the transmetalation to cis, and the trans-to-cis
isomerization (which is required to produce the coupling
product) is slower than in the case of ZnMeCl as nucleophile,
because ZnMe2 is not a good isomerization catalyst.
’ ASSOCIATED CONTENT
S
Particularly interesting is the observation that, in addition to
the expected concerted pathways previously found for ZnMeCl
as nucleophile, alternative ionic mechanisms can operate,
through ionic intermediates [PdMe(PMePh2)2L]+[ZnMe2Cl-
(THF)]À (L = THF, PMePh2). For L = THF this pathway is
faster than the concerted mechanisms, while for L = PMePh2 this
pathway is much slower than the concerted mechanisms; in fact,
the corresponding intermediate [PdMe(PMePh2)3]+ becomes a
trap of part of the Pd catalyst. The observation of these ionic
intermediates warns on the possible detrimental effect of using
excess of good ligands (as PMePh2) in the Negishi process, but is
also suggestive of the possibility to use solvents or weak ligands
more coordinating than THF in order to promote the existence,
in higher concentration than for THF, of ionic intermediates that
could still produce further acceleration of the transmetalation
step. All these conclusions are summarized in Scheme 3.
Concerning the DFT results, the theoretically calculated free
energy values reproduce fairly well the experimental trends
(which depend on the differences of calculated values for several
pathways). Nevertheless, in spite of all the computational efforts,
the calculated values do not match the experimental ones. In fact,
matching experimental relative Gibbs energies in solution with
theoretical calculations for a complex multistep reaction system
like ours is, particularly regarding Gibbs activation barriers, a
harsh work, even using good chemical and methodological
models. There is still a long way in front before such accuracy
is reached. In particular, the computation of entropic contribu-
tions in solution and the description of the solvent effects require
further improvements.41 Any progress in this problem will need
more interactive work where experiments and calculations are
put and discussed together.
Finally, following the suggestion of a reviewer, it is worth
putting into perspective the significance of this work. It provides
proof of the existence, operation, and effects of competitive
transmetalation pathways, some of which had not been invoked
before but should be considered from now on when planning or
discussing Negishi syntheses. The basic transmetalation scheme
proposed here is expected not to change dramatically for reac-
tions involving aryls instead of Me, although obviously the
specific values and energetic barriers determining the fastest
pathway will depend on each change in the reacting system (R, X,
ligands, solvent). For instance, the RX reagent in a general
synthesis could probably be ArX (X = Br, I or OAc, OTs, etc.)
rather than MeCl, and afford as starting point for transmetalation
[PdArXL2], or [PdArL2(solvent)]+XÀ in the case of weakly
coordinating anions; the latter groups will undoubtedly
Supporting Information. Details of experimental stud-
b
ies, computational details, relative energies for all the computed
structures, optimized structures, and absolute energies in gas
phase and in solution. This material is available free of charge via
the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION
Corresponding Authors
casares@qi.uva.es; gregori@klingon.uab.es; espinet@qi.uva.es
’ ACKNOWLEDGMENT
We thank the Ministerio de Ciencia y Tecnología (CTQ2010-
18901, CTQ2008-06866-CO2-01), Consolider Ingenio 2010
(CSD2006-0003, CSD2007-00006), and the Junta de Castilla y
Leꢀon (VA373A11-2 and VA281A11-2) for financial support, and
for studentships to M.G.-M. (UAB PIF) and B.F. (MCyT).
’ REFERENCES
(1) For reviews see: (a) Negishi, E. Bull. Chem. Soc. Jpn. 2007,
80, 233–257. (b) Negishi, E.; Hu, Q.; Huang, Z.; Qian, M.; Wang, G.
Aldrichimica Acta 2005, 38, 71–87. (c) Phapale, V. B.; Cꢀardenas, D. J.
Chem. Soc. Rev. 2009, 38, 1598–1607.
(2) (a) Handbook of Organopalladium Chemistry for Organic Synth-
esis, Vol. 1, part III; Negishi, E., Ed; Wiley-Interscience: New York,
2002. (b) Negishi, E.; Zeng, X.; Tan, Z.; Qian, M.; Hu, Q.; Huang, Z. In
Metal-Catalyzed Cross-Coupling Reactions; de Meijere, A., Diederich, F.,
Eds.; Wiley-VCH: New York, 2004; Chapter 15.
(3) (a) Zhou, J.; Fu, G. C. J. Am. Chem. Soc. 2003, 125, 12527–
12530. (b) Zhou, J.; Fu, G. C. J. Am. Chem. Soc. 2003, 125, 14726–14727.
(c) Tan, Z.; Negishi, E. Angew. Chem., Int. Ed. 2006, 45, 762–765.
(d) Jones, G. D.; Martin, J. L.; McFarland, C.; Allen, O. R.; Hall, R. E.;
Haley, A. D.; Bandon, R. J.; Konovalova, T.; Desrochers, P. J.; Pulay, P.;
Vicic, D. A. J. Am. Chem. Soc. 2006, 128, 13175–13183. (e) Jones, G. D.;
McFarland, C.; Anderson, T. J.; Vicic, D. A. Chem. Commun. 2005,
4211–4213.
(4) Espinet, P.; Echavarren, A. M. Angew. Chem., Int. Ed. 2004,
43, 4704–4734.
(5) Casado, A. L.; Espinet, P. J. Am. Chem. Soc. 1998, 120, 8978–8985.
(6) Casado, A. L.; Espinet, P.; Gallego, A. M. J. Am. Chem. Soc. 2000,
122, 11771–11782.
(7) Nova, A.; Ujaque, G.; Maseras, F.; Lledꢀos, A.; Espinet, P. J. Am.
Chem. Soc. 2006, 128, 14571–14578.
(8) Pꢀerez-Temprano, M. H.; Gallego, A. M.; Casares, J. A.; Espinet, P.
Organometallics 2011, 30, 611–617.
(9) Goossen, L. J.; Koley, D.; Hermann, H. L.; Thiel, W. J. Am. Chem.
Soc. 2005, 127, 11102–11114.
13525
dx.doi.org/10.1021/ja204256x |J. Am. Chem. Soc. 2011, 133, 13519–13526

Journal of the American Chemical Society
ARTICLE
(10) Braga, A. A. C.; Morgon, N. H.; Ujaque, G.; Maseras, F. J. Am.
Chem. Soc. 2005, 127, 9298–9307.
(11) Braga, A. A. C.; Ujaque, G.; Maseras, F. Organometallics 2006,
25, 3647–3658.
(39) (a) Zhao, Y.; Truhlar, D. G. Org. Lett. 2007, 9, 1967–1970.
(b) Zhao, Y.; Truhlar, D. G. J. Chem. Theory Comput. 2009, 5, 324–333.
(40) At 203 K, assuming as safe estimations 10% error in the rate
constant and 1 degree in the determination of T, the error in ΔG is only
(12) Jover, J.; Fey, N.; Purdie, M.; Lloyd-Jones, G. C.; Harvey, J. N.
0.08 kcal molÀ1
.
J. Mol. Catal. A: Chem. 2010, 324, 39–47.
(13) Urkalan, K. B.; Sigman, M. S. J. Am. Chem. Soc. 2009, 131,
18042–18043.
(41) (a) Harvey, J. N. Faraday Discuss. 2010, 145, 487–505.
(b) McMullin, C. L.; Jover, J.; Harvey, J. N.; Fey, N. Dalton Trans.
2010, 39, 10833–10836.
(14) Howell, G. P.; Minnaard, A. J.; Feringa, B. L. Org. Biomol. Chem.
2006, 4, 1278–1283.
(15) Marshall, J. A. Chem. Rev. 2000, 100, 3163–3185.
(16) (a) Jin, L.; Zhao, Y.; Zhu, L.; Zhang, H.; Lei, A. Adv. Synth. Catal.
2009, 351, 630–634. (b) Chen, M.; Zheng, X.; Li, W.; He, J.; Lei, A.
J. Am. Chem. Soc. 2010, 132, 4101–4103.
(42) For oxidative addition, see refs 4 and 5; for concurrence of
competitive pathways and the occasional need of isomerization, see
reference 8; for CÀC coupling, see: Pꢀerez-Rodríguez, M.; Braga,
A. A. C.; García-Melchor, M.; Pꢀerez-Temprano, M. H.; Casares, J. A.;
Ujaque, G.; de Lera, A. R.; Alvarez, R.; Maseras, F.; Espinet, P. J. Am.
Chem. Soc. 2009, 131, 3650–3657.
ꢀ
(17) Casares, J. A.; Espinet, P.; Fuentes, B.; Salas, G. J. Am. Chem. Soc.
2007, 129, 3508–3509.
(18) van Asselt, R.; Elsevier, C. J. Organometallics 1994, 13, 1972–
1980.
(19) Liu, Q.; Lan, Y.; Liu, J.; Li, G.; Wu, Y. D.; Lei, A. J. Am. Chem.
Soc. 2009, 131, 10201–10210.
(20) Fuentes, B.; García-Melchor, M.; Casares, J. A.; Ujaque, G.;
Lledꢀos, A.; Maseras, F.; Espinet, P. Chem.—Eur. J. 2010, 16, 8596–8599.
(21) By “retrotransmetalation” we mean the reverse of the transme-
talation reaction. See: Pꢀerez-Temprano, M. H.; Nova, A.; Casares, J. A.;
Espinet, P. J. Am. Chem. Soc. 2008, 130, 10518–10519.
(22) Gillie, A.; Stille, J. K. J. Am. Chem. Soc. 1980, 102, 4933–4941.
(23) (a) Ozawa, F.; Ito, T.; Nakamura, Y.; Yamamoto, A. Bull. Chem.
Soc. Jpn. 1981, 54, 1868–1880. (b) Ozawa, F.; Kurihara, K.; Yamamoto,
T.; Yamamoto, A. J. Organomet. Chem. 1985, 279, 233–243.
(24) (a) Spielvogel, D. J.; Davis, W. M.; Buchwald, S. L. Organome-
tallics 2002, 21, 3833–3836. (b) Spielvogel, D. J.; Buchwald, S. L. J. Am.
Chem. Soc. 2002, 124, 3500–3501.
(25) Majumdar, K. K.; Cheng, C.-H. Org. Lett. 2000, 15, 2295–2298.
(26) Anderson, T. J.; Vicic, D. A. Organometallics 2004, 23, 623–625.
(27) The exact structure of the anion represented as “[ZnMe2ClÀ]”
in Scheme 2 was not determined, but it is very reasonable to assume a
chloro-bridged dimeric structure [Me2Zn(μ-Cl)2ZnMe2]2-, which sa-
tisfies a tetrahedral coordination for Zn. This anion is not good at
coordinating, and THF can compete for the vacant Pd site. In contrast, if
[ClMeZn(μ-Cl)2ZnMeCl]2- is formed by ClÀ abstraction when using
ZnMeCl, the anionic dimer produced contains terminal chlorides that
will still be better coordinating than THF, and will occupy the Pd site
forming PdÀClÀZn bridges.
(28) Although the kinetic data for this process are not so good, we
estimate that the value of ΔG‡transmetal for the reaction via 4+ should be at
most 13.6 kcal molÀ1 to the trans and 14.6 kcal molÀ1 to the cis, or lower.
(29) Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215–
241.
(30) Weidenbruch, M.; Herrndorf, M.; Schafer, A.; Pohl, S.; Saak, W.
J. Organomet. Chem. 1989, 361, 139–145.
(31) For further computational details, see Supporting Information.
(32) As discussed below, because of the difficult treatment of
noncovalent interactions, matching quantitative values in transition
metal containing molecules is unlikely.
(33) This transition state TsIT1 appears 0.6 kcal molÀ1 lower in
energy than the following intermediate IT2 in the Gibbs energy surface,
indicating a very easy backward process to regenerate the reactants.
(34) A Gibbs energy difference larger than 3 kcal molÀ1 compared to
the starting materials should be sufficient to make its equilibrium
concentration low enough to be nonobservable by NMR.
(35) The very small energy difference between these two intermedi-
ates (0.4 kcal molÀ1) is due to the different conformations of the
tetrahedral Zn.
(36) Sakaki, S.; Ohnishi, Y.-Y.; Sato, H. Chem. Rec. 2010, 10,
29–45.
(37) Bartlett, K. L.; Goldberg, K. I.; Borden, W. T. J. Am. Chem. Soc.
2000, 122, 1456–1465.
(38) Sieffert, N.; B€uhl, M. Inorg. Chem. 2009, 48, 4622–4624.
13526
dx.doi.org/10.1021/ja204256x |J. Am. Chem. Soc. 2011, 133, 13519–13526