DOI: 10.1002/chem.

201204342

Hydroxypalladation Precedes the Rate-Determining Step in the Wacker

Oxidation of Ethene

Venkataramana Imandi, Sooraj Kunnikuruvan, and Nisanth N. Nair*[a]

4724  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 4724 – 4731
 FULL PAPER

Abstract: The complete reaction mech- of PdII. Herein, we report that with the experimentally observed rate
anism and kinetics of the Wacker oxi- hydroxylACHTUNGREpalladation is not the rate-de- law and rate. Moreover, this mecha-
dation of ethene in water under low termining step and is, in fact, in equili- nism is in consensus with the observed
[Cl], [PdII], and [CuII] conditions are brium. The newly proposed rate-deter- kinetic isotope effects. This report fur-
investigated in this work by using mining step involves isomerization and ther confirms the outer-sphere (anti)
ab initio molecular dynamics. These ex- follows the hydroxypalladation step. hydroxypalladation mechanism. Our
tensive simulations shed light on the The mechanism proposed herein is calculations also ratify that the final
molecular details of the associated indi- shown to be in excellent agreement product formation proceeds through a
vidual steps, along two different reac- reductive elimination, assisted by sol-
tion routes, starting from a series of vent molecules, rather than through
Keywords: isomerization · molecu-
ligand-exchange processes in the cata- b-hydride elimination.
lar dynamics · oxidation · reaction
lyst precursor PdCl42 to the final alde-
mechanisms · Wacker process
hyde-formation step and the reduction

Introduction

The Wacker reaction[1] has taken its place in graduate-level

inorganic chemistry textbooks as a classic example of homo-
genous dual-metal catalysis.[2] This reaction comprises the
oxidization of olefins into carbonyl compounds by water in
the presence of PdCl2 catalyst. Wacker-type oxidation reac-
tions, in their original and modified forms, are still employed
today in industrial and laboratory chemistry.[3] Beyond the
remarkable chemistry of redox-coupled reactions that are
involved in the Wacker process, it highlights various inter-
esting aspects of Pd chemistry and, thus, understanding the
molecular details of this reaction is still significant in today’s
research.[3e] Even after several decades of research on the ki-
netics and the mechanism of this reaction, various aspects
are yet to be resolved.[4] Although there is a consensus on
the general features of the reaction mechanism, controver-
sies persist in terms of the stereochemistry of the hydroxy-
palladation step, which involves nucleophilic attack of water
on the olefin, and in terms of the rate-determining step. The
origin of the inverse-first-order and inverse-second-order re-
lationships to [H + ] and [Cl], respectively, has to be inter-
preted with a consistent mechanism of this process.
Several kinetic and stereochemical studies have addressed
the issue of inner-sphere (syn) versus outer-sphere (anti)
modes of water attack on ethene (see steps 3.1!5.1 and
3.2!5.2, Figure 1). Pioneering work along these directions
was carried out by Henry; see Ref. [4]) for a review. Initial
kinetic studies with deuterated ethene showed no primary
kinetic isotope effects, thus indicating that intramolecular
hydrogen transfer, that is, the formation of compound 6, is
not rate determining.[5, 6] One of the possible mechanisms in-

[a] V. Imandi, S. Kunnikuruvan, Dr. N. N. Nair

Department of Chemistry
Indian Institute of Technology Kanpur
Kanpur, 208016 (India)
Fax: (+ 91) 512-259-7436
E-mail: nnair@iitk.ac.in Figure 1. Mechanism of the Wacker oxidation of ethene investigated in
Supporting information for this article is available on the WWW this work. Computed free-energy barriers (in kJ mol1) are indicated
under http://dx.doi.org/10.1002/chem.201204342. above the arrows.

Chem. Eur. J. 2013, 19, 4724 – 4731  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4725
 N. N. Nair et al.

volved the formation of a Pd-coordinated hydroxide species chemistry of the hydroxypalladation in substituted olefins,
from water, in agreement with the experimentally observed the kinetic isotope effects, and isotope scrambling within the
first-order proton inhibition, followed by slow hydroxypalla- framework of a unified mechanism also remains elusive.
dation in a syn fashion.[5] Conversely, stereochemical studies Moreover, the rate-determining step needs to be probed.
of Bckvall et al.[7] observed an anti mode of water attack. As a continuation of our efforts to unveil the molecular
First-order proton inhibition in this case was explained by details of the Wacker oxidation of ethene, herein, we ad-
considering that the hydroxypalladation is in equilibrium, dress the detailed mechanism and kinetics of the Wacker
followed by some rate-determining steps, which Bckvall process under low [Cl], low [CuCl2], and ambient aqueous
and co-workers interpreted as the elimination of Cl from conditions. We show that the mechanism established herein,
the coordination sphere.[7b] However, this mechanism failed which involves an anti-mode attack of water, complies well
to explain the isotope-scrambling experiments of Henry and with the experimentally estimated free-energy barrier and
co-workers,[8] in which the isomerization of [D2]allyl-1,1-al- the observed rate law at low [Cl].
cohol into [D2]allyl-3,3-alcohol during the Wacker process We began our investigation with catalyst precursor 1 and
was not observed, which suggested that the hydroxypallada- continued through to the final product formation (Figure 1).
tion step does not involve an equilibrium. Stereochemical The catalyst model used herein assumes no direct involve-
studies by using substituted olefins indicated that the syn ment of CuCl2 or the formation of PdCu heterometallic
route is preferred under low [Cl] and [CuCl2] conditions,[9] complexes. This assumption is reasonable at low [CuCl2]
whereas the anti pathway is preferred under high [Cl] and under aqueous conditions and is supported by the fact that
[CuCl2] conditions.[7, 10] the [CuCl2] term is absent in the rate law of the Wacker
Most of the previously reported theoretical calculations[11] process. Experimentally, oxo-bridged PdCu complexes
have focused on the stereochemistry of the hydroxypallada- have only been isolated under non-aqueous conditions.[16]
tion step and they have consistently predicted the operation Cl-bridged dinuclear Pd complex structure for the catalyst
of an anti mode over a syn mode. Some of the quantum-me- has also been a matter of debate in the literature.[17] At high
chanical calculations have also suggested that the intramo- or moderate [PdII], under non-aqueous conditions or in sol-
lecular b-hydrogen transfer step is rate determining. A vent mixtures with water, the formation of a dimer seems to
mechanism in which the simultaneous attack of cis-water on occur, which explains the different rate expression that is
the olefin and deprotonation by the solvent was proposed observed under these conditions. Recently, this result has
by Goddard and co-workers.[11b] However, only a marginal been further confirmed by MS (ESI) studies.[18] Thus, the
preference for the syn route was noted by these authors. mono-Pd catalyst used in the current work, corresponds to
They suggested an isomerization process, subsequent to the an experiment in water solvent at low [CuII], [PdII], and
hydroxypalladation step, as being the rate-determining step. [Cl].
The major issue in applying static quantum-chemical cal- To simulate the chemical reactions, we employed metady-
culations to the mechanistic study of the Wacker process is namics techniques,[19] together with AIMD, based on period-
the incorporation of finite solvent effects. In fact, the inclu- ic DFT. Metadynamics is a powerful technique to simulate
sion of explicit solvent molecules makes these calculations chemical reactions within the short timescale that is accessi-
practically difficult. Earlier works by Siegbahn have re- ble to AIMD and to elicit the reaction pathways and associ-
vealed the crucial importance of incorporating explicit water ated free-energy changes (for selected reviews, see
molecules in modeling this reaction.[11g, h] Accounting for sol- Ref. [20]). In our calculations, the accuracy of the free-
vent dynamics and, in particular, entropic effects are vital energy estimates is  5 kJ mol1, owing to the metadynamics
and, to overcome such problems, ab initio molecular dynam- simulation procedures that are followed, and about
ics (AIMD) techniques are more appropriate. Recently, 1 kJ mol1, owing to the employed PBE density functional
Comas-Vives et al.[12] have investigated the mechanistic as- (with respect to a more-accurate hybrid meta-GGA func-
pects of the hydroxypalladation step by using AIMD tech- tional; see the Supporting Information). All of the recon-
ACHTUNGREniques,[13] together with periodic density functional theory structed free-energy surfaces that are used for the estima-
(DFT). Interestingly, they observed an important role of the tion of free energies are also available in the Supporting In-
trans effect in the Pdethene complex in determining its formation.
structure, dynamics, and reactivity. The same conclusions
were also reached by ourselves[14] through an independent
study at about the same time. Importantly, both of these Results and Discussion
studies observed a preference for an anti-mode attack of
water over syn-mode attack. A more comprehensive anal- The 1!2!3.1 reaction involves ligand-exchange processes
ACHTUNGREysis of the ligand-exchange mechanisms and their associated in a stepwise manner. The trans effect of a Pdolefin com-
equilibria in the catalytic precursor complex have also been plex is found to favor the formation of structure 3.1, in the
reported recently[15] by using the same techniques. same way as observed in simulations at high [Cl][14] and
In short, a detailed connection between the rate law and low [Cl].[12, 15] In all of these cases, the ligands approach
the mechanism has yet to be established. Moreover, a from the solution and coordinate with the Pd complex in
proper justification of the experimentally observed stereo- the axial position, in tandem with the release of the trans

4726 www.chemeurj.org  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 4724 – 4731
Hydroxypalladation in the Wacker Oxidation of Ethene
FULL PAPER
ligand. Surprisingly, the trans effect of the complex was ne- only has a barrier of 35 kJ mol1. The forward free-energy
glected in the earlier work by Henry.[5] The barriers for the barrier is in quantitative agreement with previous calcula-
steps up to the formation of structure 3.1 are in reasonably tions[12, 14] by using similar techniques. The hydroxypallada-
good agreement with the results of previous AIMD simula- tion step proceeds through an anti mode of attack, which is
tions.[12, 15] The ligand-exchange processes follow an associa- the only possibility for compound 3.1, in which both cis posi-
tive pathway, that is, the incoming ligand (i.e., water) ap- tions are occupied by Cl ions. In this case, the deprotona-
proaches in an axial manner towards the square-planar com- tion of the coordinating water molecule occurs concurrently
plex, thereby forming a trigonal-bipyramidal intermediate, with CO bond formation. As reported previously,[14] the
and subsequently loses the trans-Cl group (Figure 2 a–c). slipping motion[22] of the olefin occurs prior to the water
attack, thus resulting in a tran-
sient h1-coordination.
The next step, that is, 5.1!6,
involves a b-hydrogen elimina-
tion. Not many theoretical in-
vestigations have focused on
this step and, thus, a detailed
analysis of the mechanism is re-
ported herein. It is clear that,
during the reaction, a H atom
migrates from the b-C atom to
the Pd center and takes over
the cis position, which is preoc-
cupied by Cl ions (in structure
5.1). Thus, an isomerization, by
the cis-to-trans rotation of a Cl
ligand, is essential, either prior
to or in conjunction with the b-
hydrogen transfer. However,
the dissociation of a cis-Cl
ligand would require a higher
Figure 2. Snapshots from the reactive trajectories of simulated steps 2!3.1 (a–c) and 3.1!4 (d–f): a) Axial ap-
free energy than isomerization,
proach of a water molecule to square-planar complex 2; b) trigonal-bipyramidal transition-state structure;
c) elimination of a trans-Cl ligand; d) axial approach of a water molecule to complex 3.1; e) T-shaped transi- which will be discussed below.
tion-state structure (arrows indicate the motion of the approaching water molecule and the exchanging cis-Cl We found that the reaction in-
ligand to form structure (f)). Color scheme: Pd brown, Cl yellow, C black, O red, H white, solvent water mole- volved an initial discharge of
cules gray sticks. the trans-water molecule (thus
forming structure 5.3), followed
This mechanism is anticipated for d8 square-planar com- by isomerization prior to b-hydrogen transfer (Figure 3).
plexes with strong trans-directing ligands.[21] Most significantly, the hydrogen transfer occurs spontane-
However, the difference in Gibbs free energy (DG) be- ously after the isomerization. Thus, the process with the
tween structures 1 and 2, as estimated by experiments highest free-energy requirement on going from 5.1!6 is the
(7 kJ mol1)[5] is much lower in magnitude than the com- isomerization step and not the trans-water discharge or the
puted Helmholtz free-energy difference (DF) between struc- H shift. We observed several trans-water exchanges in the
tures 1 and 2 (30 kJ mol1). This variance could be attribut- simulations, but never a b-hydrogen transfer that preceded
ed to the error in the free-energy estimates, the differences the isomerization step, thus supporting this conclusion.
in the ensemble that is used, or to finite size effects. In the Cis-to-trans rotation of a Cl ligand in structure 5.3 occurs
 [15]
previous AIMD studies at low [Cl ], DF was estimated to through a narrowing of the Cl-Pd-Cl angle along the bend-
be +20 kJ mol1. ing mode of its vibration. In the transition-state structure,
We found that the reaction 3.1!7 could proceed through all of the C, Pd, and Cl atoms lie almost in a plane. One of
two different pathways, that is, 3.1!5.1!6!7 (path A) and the bonding orbitals (the eighth state below the HOMO,
3.1!5.2!6!7 (path B; Figure 1). These two pathways HOMO8) of transient structure 5.4 has weak PdH and
differ in the position of the Cl ligand when it undergoes partial CC p-bond formation (Figure 3 g). The weak PdH
the hydroxypalladation reaction. As will be clear from the bond formation in structure 5.4 can also justify the presence
detailed discussion below, the positions of Cl ligands have of secondary isotope effects in the Wacker oxidation of
a significant effect on the overall reaction kinetics. ethene. Interestingly, the LUMO of this structure has sub-
At low [Cl], reaction 3.1!5.1, that is, the hydroxypalla- stantial contributions from CC (p), Pd (d), and H (s) orbit-
dation step, has a free energy barrier of 75 kJ mol1. Most ACHTUNGREals. This result explains the subsequent spontaneous reac-
importantly, the barrier for the reverse process, 5.1!3.1, tion, in which hydrogen atom transfers onto the Pd center,

Chem. Eur. J. 2013, 19, 4724 – 4731  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4727
 N. N. Nair et al.

brium is crucial in interpreting

the experimental rate law,
which will be discussed below.
An alternative route to struc-
ture 6 proceeds through struc-
ture 5.2. Thus, next, we looked
at the formation of structure 5.2
from structure 3.1, which could
circumvent the rate-determin-
ing isomerization step. In struc-
ture 3.1, the ligand exchange of
cis-Cl with bulk water (3.1!4)
requires about 110 kJ mol1,
which is in qualitative agree-
ment with previous calcula-
tions.[14] The trans effect of the
Pdethene complex favors
ligand exchange at the trans po-
sition;[12, 14] however, cis ligand
exchange is a slow proACHTUNGREcess. On
the other hand, the attack of
Cl ions at the trans position in
structure 4 (4!3.2) and its re-
verse process (3.2!4) are rela-
tively fast (DF = 40 and
35 kJ mol1, respectively), both
boosted by the trans effect.
With the approach of a ligand
from the axial position, a trigo-
nal-bipyramidal structure (as in
Figure 2 b) is always formed. In
this structure, the cis ligands lie
in an axial position, thereby
forming strong s bonds with the
Figure 3. Free-energy surface (a) and free-energy profile (b) of step 6!5.1; the minimum-energy pathway that
connects structures 6 and 5.1 is indicated by dotted lines. Relevant snapshots from the simulation are shown in
dz2 orbital. Thus, the weak
(c–f). Natural localized molecular orbitals of the HOMO8 and LUMO states of transient structure 5.4 are equatorial ligands are prefera-
shown in (g) and (h), respectively (isovalues  0.04 e). bly exchanged, as shown in Fig-
ure 2 a–c. During the cis ligand
thus resulting in compound 6. We also observed that the exchange along 3.1!4, we didn’t observe the formation of a
PdCl bonding orbitals are much lower in energy than the trigonal-bipyramidal structure. Instead, the cis ligand ex-
HOMO (HOMO10). Therefore, it is clear that the isom- change follows a dissociative-type mechanism, in which a T-
ACHTUNGREerization step is energetically preferred over PdCl dissocia- type intermediate is formed by the dissociation of a cis-Cl
tion. ligand from the coordination sphere, followed by the attack
In an independent simulation, in which we sampled the of an external water molecule (Figure 2 e).
Cl-Pd-Cl angle and the Pd coordination number but not ex- The resulting structure (3.2) undergoes a hydroxypallada-
plicitly the collective coordinates that described the b-hy- tion reaction, thereby forming structure 5.2. The mechanism
drogen transfer, the same mechanism was observed, that is, of this step is certainly of importance because of the contro-
the hydrogen transfer occurred spontaneously subsequent to versy that surrounds the stereochemistry of the hydroxypal-
the isomerization step. To further affirm this conclusion, we ladation. The simulation procedure that we followed was
carried out a simulation of the reverse proACHTUNGREcess, that is, 6! such that, of the two possibilities of water attack, that is,
5.1. The reconstructed free-energy surface (Figure 3) dis- anti versus syn, only the one with the lowest free-energy bar-
tinctly shows the absence of a free-energy barrier on going rier was sampled. The free-energy barrier for step 3.2!5.2
from 5.4!6 and that the rate-determining step in 5.1!6 was 75 kJ mol1, the same as that for 3.1!5.1, and it pro-
only involves isomerization and no b-hydrogen transfer. ceeded through an anti mode of water addition. This result
The barrier for step 5.1!6 (75 kJ mol1) is much higher is in quantitative and qualitative agreement with previous
than that for 5.1!3.1, thus implying the existence of the AIMD calculations,[12, 14] in which the anti mode was found
equilibrium 3.1Ð5.1 (Figure 5). The presence of this equili- to be preferred over the syn mode. Although, we didn’t per-

4728 www.chemeurj.org  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 4724 – 4731
Hydroxypalladation in the Wacker Oxidation of Ethene
FULL PAPER
cis-Cl dissociation, followed by
hydrogen transfer. A lower
free-energy barrier for the cis
ligand exchange in structure 7
compared to that in structure
3.1 can be explained by the
strong trans effect of ethene in
structure 3.1 compared to a
weak trans effect of the (CH3)
CH(OH) moiety in structure 7.
The following step, 7!8, in-
Figure 4. Snapshots from the metadynamics simulation of step 7!8: a) Starting structure 7; b) transient struc- volves the release of a proton
ture, in which a proton is transferred from the hydroxy group to a solvent water molecule; c) product structure
from the OH group of the coor-
8. In (b) and (c), the Pd, O, a-C, and Cl atoms lie in a plane.
dinated alkyl alcohol (7), thus
resulting in a coordinated ace-
taldehyde species. This step is not believed to involve a b-
hydrogen elimination,[7b, 23] but rather to proceed through a
deprotonation from the OH group to either the Cl ion[11a]
or a solvent water molecule.[11e] We carried out this reaction
in such a manner that, of these different routes, only the
one with the minimum free-energy barrier would be sam-
pled. We found that deprotonation of the OH group occur-
red to a solvent water molecule, but not to Cl or to Pd
(i.e., b-hydrogen elimination), in agreement with the results
reported in Ref. [11e] (Figure 4). This deprotonation is con-
sistent with the OPd bond formation (Figure 4 b). In the
Figure 5. Free-energy profiles based on the free-energy barriers along product (8), the acetaldehyde is p-complexed with the Pd0
paths 1!5.1!9 (A) and 1!5.2!9 (B). The reverse barrier for step 9!8
was not computed and, thus, is indicated by a dotted line.
center. The free-energy barrier for the forward (7!8) and
reverse processes (8!7) are 30 kJ mol1 and 20 kJ mol1, re-
form simulations to estimate the free-energy barrier for the spectively.
syn mode, we can conclude from our results that it should The final product-formation step, that is, 8!9, involves
be associated with a higher free-energy barrier than the anti the reductive elimination of the p-complexed acetaldehyde.
mode. The syn mode of hydroxypalladation at low [Cl] was We simulated the reaction 8!9 and the associated free-
simulated by Comas-Vives et al.,[12] who estimated the free- energy barrier was estimated to be 45 kJ mol1.
energy barrier to be about 200 kJ mol1. Now, considering the overall reaction (1!9), there are
The reverse reaction, that is, 5.2!3.2, only has to over- two possible mechanisms: Path A, which proceeds through
come a barrier of 30 kJ mol1, which is again consistent with structure 5.1, and path B, which proceeds through structure
step 5.1!3.1. The hydroxylated product (5.2) then proceeds 5.2 (Figure 5). Along path A, hydroxypalladation is not the
through a b-hydrogen elimination to form compound 6. In rate-determining step. The bottleneck is the isomerization
this step, the dissociation of cis water from the coordination by cis-to-trans ligand rotation (5.1!6). For step 5.1!6,
sphere of Pd is the rate-determining step and subsequent b-hydrogen exchange occurs in a spontaneous manner, sub-
b-hydrogen transfer takes place. The free-energy barrier for sequent to the isomerization, and, thus, primary kinetic iso-
the former process is only 50 kJ mol1, which is 25 kJ mol1 tope effects are not expected, in agreement with the experi-
lower than that for the reaction 5.1!6. mental data.[5, 6]
Attempts to simulate the reaction 6!5.2 always resulted Having obtained the forward and reverse barriers, we can
in the formation of structure 5.1, thus hinting a much higher derive the rate expression for the entire process. Along
barrier for this process. However, this reaction can be com- path A, the rate law has the form given in Equation (1),
puted, based on the relative energies of structures 5.2 and 6 where {ki} are rate constants and keff is the effective rate
with respect to 3.1, and is about 115 kJ mol1. constant.
In the subsequent step, structure 6 rearranges into struc-
ture 7 through a hydrogen transfer from the Pd center onto d½CH2 CH2 

the a-carbon of the hydroxylated olefin. After the hydrogen dt
 
transfer, the freed cis position is occupied by a solvent water ka PdCl2
4 ½CH2 CH2 
molecule. The barrier for the entire process is 35 kJ mol . 1 ¼
ð1Þ
½Cl 2 kb þ kg ½Hþ  þ kd ½Hþ 2 þ kw ½Cl  þ ke
The reverse process consists of cis ligand dissociation, fol-  
lowed by hydrogen transfer from the a-carbon atom onto keff PdCl24 ½CH2 CH2 
 þ  2
Pd. While simulating the reverse process, we observed initial ½ H  ½Cl 

Chem. Eur. J. 2013, 19, 4724 – 4731  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4729
 N. N. Nair et al.

This simplification of the rate expression is achieved by con- reaction pathway does not involve any cis ligand exchange
sidering that [Cl] ! [H2O] and by comparing the values of and, thus, anti-type attack is the only possibility. Wacker oxi-
the rate constants (evaluated by using the transition-state dation through a cis ligand exchange (path B) is much
theory; for a detailed derivation, see the Supporting Infor- slower than the route that does not involve the cis exchange
mation). This expression is in complete agreement with the (path A). The cis ligand exchange follows a dissociative
experimentally observed rate law at low [Cl].[5] mechanism, with the formation of a T-shaped intermediate.
Having derived this expression, we can now dissect the However, a trigonal-bipyramidal structure is formed when a
origins of the H+ and Cl inhibition. The presence of the water molecule approaches the complex in an axial manner,
proton-inhibition term is contributed to by the hydroxypal- facilitating trans ligand exchange. Even in path B, the anti
ladation step and the second-order chloride inhibition is due mode of hydroxypalladation is preferred, in agreement with
to the ligand-exchange steps along the process 1!3.1. Sim- the previous theoretical work with AIMD. We also found
plification of the rate expression (as shown above) takes ad- that the final product formation did not proceed through
vantage of the fact that the rate constant for step 5.1!6 is b-hydride elimination, but rather through a proton transfer
negligible compared to that of 5.1!3.1, or, in other words, to the solvent, followed by reductive elimination.
the existence of the equilibrium 3.1Ð5.1 and a subsequent For both path A and path B, the rate expressions show an
slow step are crucial for obtaining the above rate law. Ear- inverse-first-order dependence on [H+] and an inverse-
ACHTUNGRElier theoretical attempts at deriving the rate law by using second-order dependence on [Cl]. The rate-determining
quantum-chemical calculations failed to explain the second- step in path A is the isomerization step, whereas, in path B,
order Cl inhibition.[11c] By using the transition-state theory, it is the dissociation of the PdOH2 bond. Still, these paths
the free-energy barrier that corresponds to the effective rate show the same rate law, because, in both cases, the rate-de-
constant (keff) is 90 kJ mol1, which is in excellent quantita- termining step occurs after the hydroxypalladation step.
tive agreement with the experimental data (see the Support- Herein, we have confirmed that, for a mechanism that in-
ing Information). However, this agreement is surprising, volves an equilibrium hydroxypalladation step, the rate law
owing to the associated errors that are present in our free- of the entire reaction process is in full agreement with the
energy estimates (as discussed above). experimentally observed rate law and that it is not essential
From Figure 5, it is clear that path B is slower than to have a rate-determining hydroxypalladation step, as pro-
path A. In fact, path B also gives the same rate law, but has posed by Henry.[5] It is interesting to note the origin of [H+]
an effective barrier of about 135 kJ mol1 (see the Support- and [Cl] inhibition terms in the rate law, based on this pro-
ing Information). Thus, this pathway is ruled out for ethene posed mechanism. The first-order proton inhibition is a
oxidation at low [Cl]. result of the equilibrium hydroxypalladation step and the
second-order chloride inhibition is due to the Cl ligand-ex-
change steps that occur prior to the hydroxypalladation
Conclusions step. The effective reaction free-energy barrier, based on an
effective rate constant, is in qualitative agreement with the
Our calculations, by employing ab initio molecular-dynamics experimentally determined value.
simulation techniques, have resolved the mechanistic and ki- Although our proposed mechanism could explain many of
netic details of the Wacker oxidation of ethene at low [Cl], the crucial experimental observations, it is unable to clarify
starting from the ligand-exchange steps that involve the the isotope-scrambling experiments of Henry and co-work-
mononuclear PdII catalytic precursor through to the final ers.[8] The existence of the equilibrium 3.1Ð5.1 could have
product (acetaldehyde) formation. resulted in the formation of [D2]allyl-3,3-alcohol during the
We scrutinized a new rate-determining step, which in- Wacker oxidation of [D2]allyl-1,1-alcohol, but was only ob-
volves isomerization through a cis-to-trans ligand rotation. served in trace amounts. The only way that this result can
This isomerization occurs in the tri-coordinated intermedi- be explained is that the mechanism of ethene oxidation is
ate that is formed after the release of the trans water. The different than allyl alcohol, although the observed rate laws
rate-determining step is succeeded by a hydroxypalladation are identical. However, an earlier report on the oxidation of
step, but prior to b-hydrogen elimination. The latter step allyl alcohol by using static quantum-chemical calculations
occurs spontaneously after the isomerization, through a didn’t show a different mechanistic route.[11f] It is likely that
transient species that has weak a PdH bond, thus explain- the dppp bonding between Pd and the olefin is weaker
ing the absence of any primary isotope effects and the pres- compared to ethylene, owing to the electronic effects of the
ence of secondary isotope effects. CH2OH moiety. Thus, the crucial trans effect is weaker in
This work establishes the existence of an equilibrium hy- allyl alcohol, which may affect the reaction mechanism and
droxypalladation step, unlike the earlier mechanistic propos- kinetics. Alternative reaction pathways may also be active in
als by Henry and co-workers. Our findings profoundly agree this case, because alcoholic oxygen atoms can coordinate
with the mechanistic picture put forward by Bckvall with Pd or the reaction can proceed through an intramolec-
et al.,[7] except for the proposed slow step. ular hydroxypalladation step.[24]
We show that, for ethene, the question of anti- versus syn- The concerns regarding the dependence of the reaction
type hydroxypalladation is immaterial because the preferred mechanism on the structure of the olefin will be addressed

4730 www.chemeurj.org  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 2013, 19, 4724 – 4731
Hydroxypalladation in the Wacker Oxidation of Ethene
FULL PAPER
in future work, which is also crucial for interpreting the [10] a) N. Gregor, K. Zaw, P. M. Henry, Organometallics 1984, 3, 1251 –
stereoACHTUNGREchemical studies that have demonstrated a syn mode 1256; b) J. W. Francis, P. M. Henry, Organometallics 1992, 11, 2832 –
2836.
of water attack. At the same time, the effects of [Cl], [11] For a review and further references, see Ref. [4]; a) J. A. Keith, J.
[PdII], and [CuII] on the reaction, including the formation of Oxgaard, W. A. Goddard III, J. Am. Chem. Soc. 2006, 128, 3132 –
dinuclear complexes, also need to be probed. We believe 3133; b) J. A. Keith, R. J. Nielsen, J. Oxgaard, W. A. Goddard III, J.
that the details of the reaction mechanism and kinetics re- Am. Chem. Soc. 2007, 129, 12342 – 12343; c) S. A. Beyramabadi, H.
Eshtiagh-Hosseini, M. R. Housaindokht, A. Morsali, Organometal-
ported herein will lay a platform for future research along
lics 2008, 27, 72 – 79; d) B. J. Anderson, J. A. Keith, M. S. Sigman, J.
these directions. Am. Chem. Soc. 2010, 132, 11872 – 11874; e) S. A. Beyramabadi, H.
Eshtiagh-Hosseini, M. R. Housaindokht, A. Morsali, J. Mol. Struct.
THEOCHEM 2009, 903, 108 – 114; f) J. A. Keith, R. J. Nielsen, J.
Experimental Section Oxgaard, W. A. Goddard III, Organometallics 2009, 28, 1618 – 1619;
g) E. M. Siegbahn, J. Am. Chem. Soc. 1995, 117, 5409 – 5410;
h) E. M. Siegbahn, J. Phys. Chem. 1996, 100, 14672 – 14680.
Periodic DFT calculations were carried out by using the plane-wave basis
[12] A. Comas-Vives, A. Stirling, A. Lleds, G. Ujaque, Chem. Eur. J.
set and the PBE[25] exchange-correlation functional, as available in the
2010, 16, 8738 – 8747.
CPMD[26] simulation package. Only valence electrons were treated ex-
[13] D. Marx, J. Hutter, Ab Initio Molecular Dynamics: Basic Theory
plicitly, whereas the interactions between the core and the valence elec-
and Advanced Methods, Cambridge University Press, Cambridge,
trons were accounted for by ultrasoft pseudopotentials. We employed a
2009.
plane-wave cutoff of 30 Ry. Car–Parrinello simulations[13] were performed
[14] N. N. Nair, J. Phys. Chem. B 2011, 115, 2312 – 2321.
with a fictitious orbital mass of 700 and a time step of 0.145 fs. An ex-
[15] G. Kov cs, A. Stirling, A. Lleds, G. Ujaque, Chem. Eur. J. 2012, 18,
tended Lagrangian metadynamics scheme[19] was employed to simulate
5612 – 5619.
the chemical reactions. The reaction in bulk water was modeled by using
[16] a) T. Hosokawa, M. Takano, S.-I. Murashashi, J. Am. Chem. Soc.
29 water molecules, together with complex 1 in a periodic supercell of
1996, 118, 3990 – 3991; b) T. Hosokawa, M. Takano, S. I. Murashashi,
size 10.0   10.0   10.0 , roughly reproducing the density of water at
J. Organomet. Chem. 1998, 551, 387 – 389; c) Y. Kawamura, Y.
300 K. Natural localized molecular orbitals[27] of structure 5.4 were com-
Kawano, T. Matsuda, Y. Ishitobi, T. Hosokawa, J. Org. Chem. 2009,
puted for its gas-phase-optimized structure by using the Gaussian 09[28]
74, 3048 – 3053.
suite of programs with the 6-31G + ** basis set by using the PBE func-
[17] a) P. M. Henry, J. Am. Chem. Soc. 1972, 94, 4437; b) I. I. Moiseev,
tional. More technical details regarding the simulation set-up and the col-
O. G. Levanda, M. N. Vargaftik, J. Am. Chem. Soc. 1974, 96, 1003 –
lective coordinates that are used in the metadynamics simulations are
1007.
available in the Supporting Information.
[18] D. Harakat, J. Muzart, J. L. Bras, RSC Adv. 2012, 2, 3094.
[19] M. Iannuzzi, A. Laio, M. Parrinello, Phys. Rev. Lett. 2003, 90,
238302.
[20] a) B. Ensing, M. D. Vivo, Z. W. Liu, P. Moore, M. L. Klein, Acc.
Acknowledgements Chem. Res. 2006, 39, 73; b) A. Laio, F. L. Gervasio, Rep. Prog. Phys.
2008, 71, 126601; c) A. Barducci, M. Bonomi, M. Parrinello, WIREs
The authors thank the Department of Chemistry, IIT Kanpur for the use Comput. Mol. Sci. 2011, 1, 826 – 843.
of the ChemFIST computational facility and the CSIR, India, for funding [21] G. K. Anderson, R. Cross, J. Chem. Soc. Rev. 1980, 9, 185 – 215.
the project. V.I. and S.K. thank the UGC and IIT Kanpur for their Ph.D. [22] O. Eisenstein, R. Hoffmann, J. Am. Chem. Soc. 1981, 103, 4308 –
fellowships. 4320.
[23] R. F. Heck, Hercules Chem. 1968, 57, 12.
[24] T. Hosokawa, S. I. Murahashi, Acc. Chem. Res. 1990, 23, 49 – 54.
[1] J. Smidt, W. Hafner, R. Jira, J. Sedlmeier, R. Sieber, R. Rttinger, [25] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865.
H. Kojer, Angew. Chem. 1959, 71, 176 – 182. [26] CPMD, version 13.3, IBMCorp 1990 – 2012,MPI fr Festkçrper-for-
[2] a) F. A. Cotton, C. A. Murillo, M. Bochmann, Advanced Inorganic schung Stuttgart 1997 – 2001; also see http://www.cpmd.org.
Chemistry, 6th ed., Wiley-Interscience, New York, 1999; b) G. Roth- [27] A. E. Reed, F. Weinhold, J. Chem. Phys. 1985, 83, 1736.
enberg, Catalysis: Concepts and Green Applications, Wiley-VCH, [28] Gaussian 09 (Revision B.01), M. J. Frisch, G. W. Trucks, H. B. Schle-
Weinheim, 2008. gel, G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V.
[3] a) J. M. Takacs, X. Jiang, Curr. Org. Chem. 2003, 7, 369 – 396; b) J. Barone, B. Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X.
Tsuji, Synthesis 1984, 369 – 384; c) A. Heumann, M. R-glier, Tetra- Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Son-
hedron 1996, 52, 9289 – 9346; d) C. N. Cornell, M. S. Sigman, Inorg. nenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa,
Chem. 2007, 46, 1903 – 1909; e) M. S. Sigman, E. W. Werner, Acc. M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven,
Chem. Res. 2012, 45, 874 – 884. J. A. Montgomery, Jr., J. E. Peralta, F. Ogliaro, M. Bearpark, J. J.
[4] J. A. Keith, P. M. Henry, Angew. Chem. 2009, 121, 9200 – 9212; Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J.
Angew. Chem. Int. Ed. 2009, 48, 9038 – 9049. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar,
[5] P. M. Henry, J. Am. Chem. Soc. 1964, 86, 3246 – 3250. J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox,
[6] a) P. M. Henry, J. Org. Chem. 1973, 38, 2415 – 2416; b) M. Kosaki, J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E.
M. Isemura, Y. Kitaura, S. Shinoda, Y. Saito, J. Mol. Catal. 1977, 2, Stratmann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W.
351 – 359; c) Y. Saito, S. Shinoda, J. Mol. Catal. 1980, 9, 461 – 464. Ochterski, R. L. Martin, K. Morokuma, V. G. Zakrzewski, G. A.
[7] a) J. E. Bckvall, B. kermark, S. O. Ljunggren, J. Chem. Soc. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D. Daniels,
.
Chem. Commun. 1977, 264 – 265; b) J. E. Bckvall, B. kermark, Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox, Gaussian,
S. O. Ljunggren, J. Am. Chem. Soc. 1979, 101, 2411 – 2416. Inc., Wallingford CT, 2010; see also http://www.gaussian.com/..
[8] M. K. Wan, K. Zaw, P. M. Henry, Organometallics 1988, 7, 1677 –
1683. Received: December 5, 2012
[9] J. W. Francis, P. M. Henry, Organometallics 1991, 10, 3498 – 3503. Published online: March 13, 2013

Chem. Eur. J. 2013, 19, 4724 – 4731  2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org 4731