Uncovering the mechanism of homogeneous methyl methacrylate formation with P,N chelating ligands and palladium: Favored reaction channels and selectivities

Article abstract of DOI:10.1021/om500970k

The catalytic alkoxycarbonylation of alkynes via palladium and P,N ligands, studied through a prototypical reaction involving propyne methoxycarbonylation yielding methyl methacrylate, has been explored at the B3PW91-D3/PCM level of density functional theory. Four different reaction routes have been probed in detail, spanning those involving one or two hemilabile P,N ligands and either hydride or carbomethoxy mechanisms. The cycle that is both energetically most plausible and congruent with experimental data involves Pd(0) and two P,N ligands acting cocatalytically in turn to shuffle protons via both protonation and deprotonation reactions. Other mechanisms proposed in the literature can be discounted because they would lead to insurmountable barriers or incorrect selectivities. For the preferred mechanism, the P,N ligand is found to be crucial in determining the strong regioselectivity and intrinsically controls the overall turnover of the catalytic cycle with moderate barriers (ΔG^a§§ of 20.1 to 22.9 kcal/mol) predicted. Furthermore, the necessary acidic conditions are rationalized via a potential dicationic channel.

Full text of DOI:10.1021/om500970k

Article
pubs.acs.org/Organometallics
Uncovering the Mechanism of Homogeneous Methyl Methacrylate
Formation with P,N Chelating Ligands and Palladium: Favored
Reaction Channels and Selectivities
Luke Crawford, David J. Cole-Hamilton, and Michael Buhl*
̈
School of Chemistry, University of St. Andrews, North Haugh, St. Andrews, Fife, KY16 9ST, Scotland, United Kingdom
S
* Supporting Information
ABSTRACT: The catalytic alkoxycarbonylation of alkynes via
palladium and P,N ligands, studied through a prototypical reaction
involving propyne methoxycarbonylation yielding methyl meth-
acrylate, has been explored at the B3PW91-D3/PCM level of
density functional theory. Four different reaction routes have been
probed in detail, spanning those involving one or two hemilabile
P,N ligands and either hydride or carbomethoxy mechanisms.
The cycle that is both energetically most plausible and congruent
with experimental data involves Pd(0) and two P,N ligands acting
cocatalytically in turn to shuffle protons via both protonation and deprotonation reactions. Other mechanisms proposed in the
literature can be discounted because they would lead to insurmountable barriers or incorrect selectivities. For the preferred
mechanism, the P,N ligand is found to be crucial in determining the strong regioselectivity and intrinsically controls the overall
turnover of the catalytic cycle with moderate barriers (ΔG⧧ of 20.1 to 22.9 kcal/mol) predicted. Furthermore, the necessary
acidic conditions are rationalized via a potential dicationic channel.
frequency (TOF) for carbonylation.2,3 Ligands based upon
INTRODUCTION
■
3-PyPPh2, 4-PyPPh2, and PPh3 exhibit a reduced efficiency,
Alkyne alkoxycarbonylation is a transformation with 100% atom
suggesting that both the presence and location of the nitrogen
efficiency that forms acrylate esters.1−8 Methoxycarbonylation of
atom are important. 2-PyPPh2 allows the methoxycarbonylation
propyne yields methyl methacrylate (MMA), a small-molecule
of propyne to proceed under mild conditions of 45 °C, attaining a
feedstock crucial in industry due to its polymer poly(methyl
turnover frequency of 40 000 mol (mol Pd h)−1 with a selectivity
methacrylate), more commonly known as Perspex. This material
of ∼99% toward MMA.
has a wide range of uses, which include important surgical roles,9
P,N ligands are known to coordinate in a number of binding
cosmetics and coatings, and as a rigid transparent plastic
modes. While monocoordination and multiple unidentate co-
for windows, especially in transport.10 There is also a growing
ordination are typically through the softer phosphorus atom,17
demand from use in LCD screens.11 Functionalization of
propyne yielding the branched product of methoxycarbonylation
has received considerable attention,2,3,5−7 and such chemistry
has been extended to higher alkynes such as ethynyl benzene12
and alkynols.13 Transition metals are key to many industrial
processes,14−16 and in the example of MMA formation from
propyne, palladium complexes with P,N chelating ligands play
a key role in this transformation (Scheme 1).
many structures have been isolated that show chelation.18−24
2-PyPPh2 may also coordinate metal (hetero)dimers,4,25−27 and
structures involving iridium suggest that two 2-PyPPh2 ligands
should be able to coordinate in both unidentate and chelating
fashion around a single metal center.28
Contrastingly, diphosphine ligands employed in the alkoxy-
carbonylation of other polymer precursors tend to remain
bidentate in their coordination modes.29−35 The hemilability
of P,N-type ligands coupled with a wide range of coordination
modes36 is one of the reasons for their continued use in
catalysis37−40 and nanomaterials41 and as analytical molecular
sensing agents.42 This intrinsic property of 2-PyPPh2 has been
implicated in a number of proposed catalytic alkoxycarbonyla-
tion mechanisms.
Drent has reported that the presence of a P,N ligand,
2-pyridyldiphenylphosphine (2-PyPPh2), is necessary for both
high selectivity for the branched product and a high turnover
Scheme 1. Conditions2 for the Formation of MMA Involving
P,N Ligands from Propyne
The 2-PyPPh2 scaffold is the most effective for the
methoxycarbonylation of propyne.2 Through 6-methylation of
the pyridyl moiety the selectivity toward MMA over the linear
Received: September 23, 2014
Published: January 7, 2015
© 2015 American Chemical Society
438
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
product, methyl crotonate (MC), was enhanced from 98.9% to
99.95% with a 20% increase in activity under some conditions.3
More than two equivalents per Pd of a strong acid such as
methanesulfonic acid (MeSA) are needed for efficient turnover.
Such an observation suggests an additional role of the acid
beyond removing acetate and allowing coordination of ligands
and substrates. Protonation of zerovalent Pd or the nitrogen
and subsequent steps would involve CO insertion and solvolysis
with 2-PyPPh2 acting as an in situ base. This nonclassical hydride
mechanism is proposed to occur via a Pd(0) complex. These are
known to exist in mixtures of tertiary phosphines and protic
solvents,43−45 similar to conditions employed in alkoxycarbony-
lation reactions.
Dervisi et al. countered with evidence that would appear to
support a nonclassical carbomethoxy mechanism (cycle C)5
involving a dicationic complex and terminating protonolysis
supported by P,N−H+; Pd(2-PyPPh2)2(CO2CH3) was found to
be active toward branched small-molecule production under CO
pressure in methanol with propyne.
These experiments were conducted in the absence of acid with
benzene as solvent, markedly different from turnover conditions,
and do not inherently support a messenger protonolysis as in C
nor discount either C or D operating in the presence of MeSA.
Dervisi later confirmed6 the presence of the same vinyl inter-
mediate suggested by Scrivanti but could not detect the σ-vinyl
analogue expected with propyne, instead finding a phosphonium
salt.
8
of a hemilabile 2-PyPPh2 is therefore a plausible pathway for
forming the catalytically active species. Weak acids exhibit
dramatically decreased TOFs, and halide containing acids retard
the reaction. On the basis of such evidence the active catalyst
is probably a cationic Pd complex that is deactivated upon
coordination of halides.
The simplest routes for catalytic turnover of MMA involve
hydride and carbomethoxy pathways (Scheme 2). In the
Scheme 2. Classical Pd-carbomethoxy (A) and Pd-hydride (B)
Mechanisms
Recently Drent and Cole-Hamilton have provided a summary
of the experimental evidence for the proposed reaction
channels.1 They argued that the bulkier groups of propyne
being situated at the α carbon in the “nonclassical” Pd(0)
mechanism should evoke steric clashes. To overcome this,
a concerted protonation and nucleophilic attack of Pd0CO on
coordinated propyne (Scheme 4) has been proposed. This would
carbomethoxy route propyne inserts into the metal−carbonyl
bond in a 1,2 mode with the bulky substituents at the β position
relative to Pd (cycle A). During the hydride cycle, propyne
insertion into the Pd−H bond follows a 2,1 mode (cycle B). An
increased steric bulk at the 6 position, as in 2-(6-Me)PyPPh2,
should have reversed effects upon product selectivity in both
cases, with the carbomethoxy cycle encouraging MMA formation
and preference for MC being increased if a classical hydride
mechanism is operating.
Scheme 4. Alternative Mechanism for the Initial
Carbonylation of Propyne Involving a Concerted
Nucleophilic Attack and Protonation1
It has been argued that the role of 2-PyPPh2 must extend
beyond selectivity enhancement, as the nature of the ligand
is closely tied to the overall productivity of the catalytic cycle2
and therefore is likely acting as a rate-enhancing messenger for
protonolysis in a carbomethoxy cycle or as an in situ base
expediting the solvolysis of a “nonclassical” hydride mechanism
(Scheme 3).
then be followed by a methanolysis step as in mechanism D.
From this data, the exact nature of the catalytic cycle that
accounts for the observed regioselectivities remains unknown.
Support for a nonclassical mechanism arises from reaction rates
for a related transformation, which are found to increase in line
with the 2-PyPPh2:Pd ratio (until 30:1) and acid:Pd (until
66:1),46 suggesting two coordinated P,N units with one of them
likely protonated.
Scheme 3. Nonclassical Pd-carbomethoxy (C) Proposed by
Drent and Scrivanti’s Nonclassical Pd(0) Mechanism (D)
from Labeling Experiments
With these considerations we have tackled this problem using
modern density functional methods. When experiment gives
rise to ambiguous interpretation, quantum chemistry can offer
insight that allows for a more definitive answer,47−50 thereby
aiding rational catalyst improvement. Our primary aims were in
establishing an experimentally congruent route for alkyne
alkoxycarbonylation, modeled through production of MMA.
To this end, we characterized complete catalytic cycles for all
four pathways, A −D, that had been proposed. Any plausible
mechanism must have surmountable barriers compatible with
the high turnover and must be able to closely reproduce the
selectivities observed with both 2-PyPPh2 and 2-(6-Me)PyPPh2
ligands. Only one of the four proposed mechanisms, path D,
is able to achieve this. We have recently communicated the
essential features of this mechanism,51 and in the present paper we
give a full account of the results that have led to its identification.
A nonclassical hydride route operating in the formation of
MMA was suggested due to the presence of a Pd vinyl
intermediate, based on ethynylbenzene, being readily observed
through 1H NMR studies.7 Should this route (D) be operating,
the cycle would be initiated via proton transfer from 2-PyPPh2
439
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
a
Scheme 5. Methoxycarbonylation at a Pd(II) Center Involving a 2-PyPPh2 Monochelate Operating through Cycle A
a
Profile sketched according to computed ΔG values (italics). TS4-5 determines regioselectivity, and TS6-7 controls turnover.
and D using a hemilabile κ1-(2-PyPPh2) as a proton relay
(Scheme 3).
COMPUTATIONAL METHODOLOGY
■
We have used the B3PW9152−54 hybrid functional, which has been
successfully validated for a range of reactions that rely upon metals55−58
and has been used to study related (2-pyridyl)thiourea Pd(II)
complexes.59 This method benchmarks well against explicitly correlated
CCSD(T)60 when coupled with Grimme’s DFT-D3,61−63 including
Becke−Johnson damping.64,65 This postcalculation empirical correction
serves to account for dispersive forces that are not well described by
DFT yet have been shown to be essential for the reproduction of
accurate energies, especially when triphenylphosphine and similarly
bulky moieties are present.66,67
1. Classical Carbomethoxy (A). A classical methoxycarbo-
nylation pathway begins with a single chelating 2-PyPPh2
ligand and requires the prior formation of a Pd(II) center
with coordinated methoxide, such as [(2-PyPPh2)Pd(OMe)]+
(1, Scheme 5). Complex 1 would be unlikely to be present in
significant amounts with excess acid under catalytic conditions;
however we sought to exclude this route on the grounds of
selectivity and the energetics of individual steps.
The first step of this cycle is the uptake of CO (2) and
subsequent migratory insertion of the nucleophilic methoxy unit
onto carbon monoxide, affording 3.78 This process is facile, with a
free-energy barrier of 11.6 kcal/mol via TS2−3, and coordina-
tion of propyne affording 4 should be rapid. Propyne binds
in a perpendicular orientation relative to the Pd−N−P plane
(Figure 1) that shows no predisposition for regioselectivity at this
Geometries were computed at the level of B3PW91/ECP1, where
ECP1 corresponds with the 6-31G** basis set on all nonmetal atoms
and the relativistic SDD pseudopotential and the corresponding
valence electron basis set on Pd. Transition states were located at this
level either through coordinate driving and subsequent optimization to
the transition state or by using the QST3 algorithm.68 Stationary points
were confirmed by the presence of the correct number of imaginary
frequencies using the harmonic approximation. These frequencies
allowed for corrections to enthalpies and free energies to be evaluated
from standard thermodynamic expressions at 298.15 K. Transition states
were confirmed to link to the respective reactants and products using
intrinsic reaction coordinate (IRC) calculations.69,70
Energies were refined through single-point calculations employing
the same functional and an ECP2 level, i.e., the same SDD
pseudopotential and valence electron basis set for Pd but the larger
triple-ζ 6-311+G** basis set on all nonmetal atoms. Bulk solvent effects
were included through a polarizable continuum (PCM)71−74 with
methanol as the solvent. We do not include the presence of the weakly
coordinating anions arising from the protonation of Pd-coordinated
acetate. To these PCM and ECP2 corrected energies DFT-D3BJ
corrections were added to accurately account for the missing
dispersion.75 All calculations were performed using Gaussian 09,76
and structures were built by hand, guided by a small model study (see SI,
Scheme S1). A range of low-energy conformers are known to exist for
related rhodium species, and preferential orientation is impacted little by
the location of the pyridyl group.77
Figure 1. Geometries involved in Pd(II)-mediated 1,2 insertion of
COOMe and propyne. Distances are in Å. Blue = N, orange = P,
turquoise = Pd, gray = C, red = O, and white = H.
stage. Subsequent MMA formation occurs via a 1,2 insertion
through TS4−5, as opposed to MC formation, which occurs via a
2,1 insertion (for the differences inherent in these processes see
section 1.1). At this point in the reaction, coordinated propyne
rotates from the perpendicular orientation in 4 toward an in-
plane mode, decreasing the distance between bond-forming
carbon participants from 3.08 Å to 2.02 Å. Concomitantly, the
acyl moiety begins to dissociate from Pd and a stronger Pd−C
bond with propyne forms to compensate.
RESULTS AND DISCUSSION
■
We begin by exploring a carbomethoxy cycle (mechanism A in
Scheme 2) as a potential route for the classical methoxycarbo-
nylation of propyne to MMA involving one κ2-(2-PyPPh2) ligand
at a Pd(II) center. Next, a classical hydride mechanism (B) is
computed, followed by inspection of the “nonclassical” routes C
440
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
The activation energy associated with forming the branched
Pd-alkenyl intermediate, 5, is low, at 13.4 kcal/mol, relative to 4.
Due to the considerable enthalpic and entropic gain upon
insertion and the challenging kinetic barrier for terminating
protonolysis, the palladium alkenyl ester 5 rests in a deep
thermodynamic sink. From 5, TS6−7 has an imposing barrier of
38.0 kcal for solvolysis, incompatible with turnover conditions.
This step involves cleavage of a Pd−C bond and the formation of
a C−H bond, which is ultimately less favorable than retention of
the Pd−alkenyl complex (7 vs 5).
Scheme 6. Selectivity Determining Pathways for MC (Left)
and MMA (Right) Formation in the First Pd(II)
Carbomethoxy Pathway
a
However, the classic carbomethoxy mechanism is more
complex than Scheme 5 suggests. After dissociation of the
product, 1a has the methoxy group trans to phosphorus as
opposed to trans to the nitrogen of chelating 2-PyPPh2 as in 1.
This leads to a continuation of the cycle to regenerate 1 through
isomers 1a to 7a. Strictly, two MMA molecules are produced in
every turn of this full catalytic engine, and Scheme S2 in the SI
illustrates the energetic stipulations of producing MMA through
this isomeric pathway.
a
Energies are taken against 4.
The largest free-energy span along this second leg is
protonolysis between 5a and TS6a−7a at 31.2 kcal/mol, lower
than the equivalent process in Scheme 5. Dynamic studies on
Pd(II) 2-PyPPh2 systems have suggested that fluxional and
ligand exchange processes are accessible without considerable
barriers.36 If 1 and 1a could interconvert rapidly, the two
pathways presented may interweave to allow the system to
traverse over the less imposing kinetic barriers and skirt the
highest one. For instance, a path leading from 1a via 7a to 1
followed by isomerization of 1 to 1a would avoid the higher
barrier via TS6−7.
The trigonal transition state between 1 and 1a has been
located and found to be 17.5 kcal/mol relative to 1, while 1a is
11.0 kcal/mol above 1. Due to the kinetic hindrance of this route
as opposed to the rapid CO uptake by 1, it is unlikely that this
process will form the major pathway. There are other points of
such “pathway switching”, however, with 3 and 5 being potential
locales for isomerization through a trigonal transition state;
although irrespective of this, a difficult barrier of at least
31 kcal/mol must be crossed. On the basis of the energy profiles
obtained, a classical carbomethoxy route is unlikely. To rule
out this path conclusively, we now explore the MMA vs MC
selectivity of 4 → 5 and 4a → 5a.
1.1. Classical Carbomethoxy Selectivity. The selectivity-
determining step is the insertion of the perpendicular
coordinated alkyne into the Pd−C(carbonyl) bond via TS4−5.
Besides this 1,2-insertion step leading to the branched product,
an alternative linear route through a 2,1 insertion (TS4L−5L) is
possible (Scheme 6). Due to the lack of steric bulk surrounding
a single chelating 2-PyPPh2 ligand, there is little discrimination
between 4 and 4L, although a distinct kinetic difference is
observed between TS4−5 and TS4L−5L.
MMA-affording TS4−5 is 2.4 kcal/mol higher than MC-
producing TS4L−5L, leading to the surprising conclusion that
2,1 insertion is more favorable! The same is found for the
mechanistic alternative with lower overall barrier (Scheme S2),
where ΔΔG⧧ between TS4a−5a and TS4aL−5aL is 2.2 kcal/mol
(Scheme S3, SI). Thus, if such a methoxycarbonylation route
were operating under turnover conditions, MC should be
produced with a high selectivity (≈ 2−3% branched product
at 45 °C).
HCCCH3 toward the bulky carbomethoxy group, there is no
such close contact, and thus the linear-forming transition state is
easier to access. Mechanism A can therefore be safely excluded
because it affords the incorrect selectivity.
2. Classical Hydride Mechanism (B). We now turn to the
classical hydride mechanism starting with Pd(II) hydride 8
(Scheme 7). Surprisingly, no stable intermediate with side-on-
coordinated alkyne, i.e., [(κ1-(2-PyPPh2))Pd(H)(HCCCH3)]+,
could be located, irrespective of the binding mode of propyne
(parallel or perpendicular to the H−Pd-chelate plane).79 All
attempts to optimize such structures result in spontaneous
insertion, affording either the branched vinyl complex 9 or the
linear analogue 9L. The effect on selectivity is discussed below.
Scans from the vinyl intermediates 9 and 9L probing the reaction
coordinate associated with the formative C−H bond have been
performed. On such a potential energy surface movement of the
hydrogen from carbon back to palladium is observed, but no
stable minimum with a π-coordination mode of propyne can be
located (see SI, section S3). Thus, it may be assumed that uptake
of propyne would lead directly to either 9 or 9L, controlled by
approach trajectory and orientation.
Therefore, we have computed two reaction profiles, producing
either MMA or MC. As with carbomethoxy mechanism A, the
production of one MMA unit results in a different stereoisomer
of the hydride, and thus a full catalytic cycle producing two
MMA units and switching the trans orientation of P and N of the
κ2-(2-PyPPh2) chelate must be considered.
We have explored a route for interconversion between these
two half-cycles via trigonal TS8−8a, although this is found to be
47.5 kcal/mol uphill relative to 8, which, in turn, is more stable
than 8a by 15.3 kcal/mol. Such a simple path switching is thus
excluded, although we are aware that interconversion might be
possible at different stages of the reaction (e.g., at 11) or through
fluxional associative exchange of the 2-PyPPh2 ligand with one
from the bulk.36 Accordingly, we will limit our discussion to the
lowest free-energy spans across individual half-cycles (Table S1)
such as that with the most accessible methanolysis, illustrated
in Scheme 7.
The rate-limiting barrier on path B is associated with
terminating methanolysis, which has been found to be a difficult
process in prior studies.80−82 Over the whole pathway, the largest
free-energy barriers are predicted between 11a and TS12a−13a,
namely, 53.3 kcal/mol (MC-forming) and 54.2 kcal/mol (MMA-
forming, Table S1).
The unexpected 2,1 preference arises due to a clash of the
propyne methyl with the ester substituent in the branched-
forming route. In TS4L−5L, which has the terminal hydrogen of
441
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
a
Scheme 7. Classical Hydride Mechanism Beginning from an N-trans Pd−H Species
a
TS12−13 is rate-determining.
Even with path switching that would avoid this kinetic
bottleneck, free-energy barriers of at least 40.8 kcal/mol (MC
forming) and 42.5 kcal/mol (MMA forming) are predicted
(relative to the lowest preceding intermediate 11). On the basis
of such evidence, this hydride path can already be discounted on
energetic grounds.
Assessment of selectivity is difficult in this case, as there is no
barrier on the electronic surface for the irreversible 1,2 or 2,1
insertions between propyne and 8(a) on the potential energy
surface. Entropic barriers may be expected due to the associative
nature of these reactions, but according to the Bell−Evans−
Polyani principle, the relative thermodynamic driving forces
should govern these barriers and, thus, the selectivity. On the
pathway showing the most accessible terminating methanolysis,
that of Scheme 7, there is little distinction between 9 and
MC-forming 9L. The former is slightly more favorable in terms
of enthalpy, by 0.5 kcal/mol, while the latter is barely more
stable by 0.1 kcal/mol in free energy, as shown in Table S1. Little,
if any, selectivity for MMA over MC would be expected, further
disfavoring pathway B.
with presumed resting state 14. The following transition state for
concerted migratory insertion and deprotonation, TS16−17,
emerges as the rate-limiting state at 24.5 kcal/mol. Intermediate
17 is less favorable against both 14 and 16 (when considering
free energy), and the resultant uptake of propyne leaves 18
4.0 kcal/mol above 14. All subsequent steps are exergonic, with
barriers of 20.5 kcal/mol (protonolysis via TS19−20) or less.
The irreversible regioselective transition state, TS18−19,
produces a palladium-alkenyl intermediate leading to MMA.
Compared to the analogous transition states in cycles A and B,
TS18−19 incorporates the greater steric bulk of the second
phosphine ligand, the influence of which upon selectivity has
been explored (Figure 2).
Energetically, the competing transition states do not show a
difference that would be compatible with observed selectivity;
TS18−19 is only 0.2 kcal/mol (ΔG) more favorable than
TS18L−19L, corresponding to a selectivity for MMA of
approximately 73%. The change in selectivity from a 2,1 mode
in mechanism A to a 1,2 mode in C-II arises because in the 2,1
process the methyl unit of propyne is orientated toward the
pyridyl ring, invoking an unfavorable steric interaction. Despite
this, such a clash is apparently not enough to reach the 99%
selectivity observed under turnover conditions.
To conclude, the classical hydride mechanism suffers from an
essentially insurmountable intramolecular methanolysis step and
cannot reproduce the observed selectivity.
3. Nonclassical Messenger Carbomethoxy Mecha-
nisms (C and C-II). Drent’s original nonclassical carbomethoxy
mechanism (C, Scheme 3) can be excluded because the proposed
selectivity-determining transition state is the same as in cycle A.
Dervisi and co-workers proposed5,6 a mechanism for the production
of MMA from [(2-PyPPh2)2Pd(CO2R)(OAc)], which incorpo-
rates two monocoordinating phosphine ligands. Increasing the
quantity of both acid and phosphine leads to an observed zero
order in acid, and therefore any protonolysis step appears to
involve 2-PyPPh2 acting as a relay. A possible adapted messenger
carbomethoxy mechanism accounting for this experimental
nuance, which incorporates two monocoordinating phosphines,
C-II, is presented in Scheme 8.
Additional evidence against mechanism C-II emerges from
the effect of 6-methylation of the pyridyl rings. While such a
modification leads to an increase in selectivity toward MMA from
99% to 99.95% in practice, our density functional calculations
with the added 6-Me groups fail to reproduce this noticeable
change. In fact we find that the selectivity of mechanism C-II is
reversed by 6-methylation, as TS18−19 6-Me is 0.2 kcal/mol
less favorable than TS18L−19L 6-Me! A closer inspection of the
transition states given in Figure 2 provides some rationale for the
low computed selectivity. The regioselective transition states
in C-II have the pyridyl rings orientated such that the bulk (and
the methyl modification) points away from the plane spanned by
the metal and the C atoms involved in migratory insertion.
Meanwhile, the clash between the carbomethoxy moiety and
propyne methyl in the 1,2 mode is made more severe by the
additional methyl unit on the pyridyl ring.
The key difference between Drent’s original carbomethoxy
mechanism, C, and the modified cycle we have investigated, C-II,
is that the latter incorporates two 2-PyPPh2 ligands throughout
the reaction. Coordination of both methanol and carbon
monoxide (affording 16) is exothermic, but essentially isoergonic
On the grounds of insufficient and incorrect regioselectivity,83
both mechanisms C and C-II can be excluded because
442
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
a
Scheme 8. Carbomethoxy Mechanism C-II Incorporating a P,N Ligand Acting as a Proton Messenger (TS19−20)
a
The rate-determining step is the migratory insertion of methoxide into carbon monoxide (TS16−17), and the selectivity is determined via
TS18−19, which irreversibly leads to 19.
[(κ2-(2-PyPPh2))(κ1-(2-PyPPh2))Pd(H)]+ species being cata-
lytically relevant was also assessed, and calculations excluding
a route involving such an intermediate are included in the SI,
Scheme S5.
The first transformation that 21 must undergo is regioselective
ligand-assisted protonation, as detailed in Scheme 9. Protonation
of the alkyne via TS21−22 needs only 7.2 kcal/mol of activation
energy and establishes a cocatalytic role of P,N as an agent for
accessible proton transfer. We attempted to locate a transition
state such as that outlined in Scheme 4, although ultimately could
not characterize such a concerted process. Instead an alternative
pathway of 21+CO undergoing a stepwise protonation, similar
to that discussed for 21, with CO remaining coordinated
throughout became apparent. Following insertion, this route
leads directly to 24, but the activation cost of the proton transfer
transition state, TS21+CO-24, is higher (byΔΔG⧧ = 3.8 kcal/mol)
Figure 2. MC (TS18L−19L)- and MMA (TS18−19)-forming
than that offered by TS21−22. For further details on this process
transition states with 2-PyPPh2 (top) and 2-(6-Me)PyPPh2 (bottom).
see the SI, section S5.
Distances are in Å, and Ph groups have been omitted for clarity. Steric
Key intermediates and transition states of the initial carbon-
clashes are highlighted in red.
ylation steps are displayed in Figure 3. 21 displays strong back-
bonding between palladium and propyne with Pd−C distances of
2.07 Å. TS21−22 is followed by a series of energetically downhill
steps from β-agostic intermediate 22, yielding 24 (Figure 3)
following displacement of a chelating binding mode in 23 via CO
uptake. Regioselective insertion is therefore practically irrever-
sible under experimental conditions, and this is essential to
the high selectivities (discussed below) observed by preventing
dynamic communication of branched and linear intermediates.
The migratory insertion into the metal−carbon bond via
TS24−25 is easily surmountable, being only 10.3 kcal/mol uphill
from the prior intermediate, and is encouraged by formation of a
Pd-acyl complex. The now-deprotonated pyridyl acts as an inter-
mittent chelate, stabilizing intermediates that would otherwise
be coordinatively unsaturated. Wiberg bond index (WBI)87
measurements show a significant sharing of electron density
between bond-forming carbon centers (0.12) in 24, accounting
for the ease of this transformation.
experimental selectivities are not reproduced and subtle ligand
effects are incorrectly accounted for. Full C-II energetics,
including 2-(6-Me)PyPPh2, are given in the SI, Table S2.
4. In Situ Base Mechanism (D). Scrivanti’s Pd(0)
mechanism (D, Scheme 3) involves formation of a Pd-vinyl
species7 through protonation of coordinated propyne8 with
selectivity imbued during the proton transfer step involving a
monocoordinated 2-PyPPh2 ligand acting as a proton-shuffling
relay.84−86 The precursor species could be [(κ1-(2-PyPPh2))(κ1-
(2-PyHPPh2))Pd(CO)(HCCMe)]+ (21+CO), with a concerted
protonation and migratory insertion yielding a palladium-acyl
intermediate (Scheme 4).1 This complex, 21+CO, with the
expected tetrahedral geometry was identified as being slightly
exergonic relative to 21, and this led us to start the reaction
from the more stable planar complex [(κ1-(2-PyPPh2))-
(κ1-(2-PyHPPh2))Pd(HCCMe)]+. The possibility of a
443
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
a
Scheme 9. Initial Proton Transfer and Carbonylation at Pd(0) with Two Monocoordinated 2-PyPPh2 Units
a
TS21−22 imbues regioselectivity.
the second coordination sphere around 26 may play a role in
accepting and transferring the proton arising from deprotonation
away from the Pd center. This would serve to lower the free-
energy cost of the process with an entropic gain from distribution
and an enthalpic gain from additional hydrogen bonding.
However, since a 2-PyPPh2 unit must be reprotonated in order
to conduct regioselective propyne insertion at 21, we have not
explored this route further.
Displacing chelating 2-PyPPh2 and binding methanol is slightly
endergonic, though by only 0.7 kcal/mol. Reaction entropies for
such associative processes are often overestimated by our
standard protocol, and it is worth noting that binding is found
to be exothermic by 12 kcal/mol (see ΔH values in Scheme 10).
This methanol coordination preferentially occurs such that the
hydroxyl unit orientates its proton toward the basic nitrogen
of 2-PyPPh2, resulting in the nucleophilic oxygen and acyl
carbon being separated by 2.71 Å (Figure 4). Establishing this
intramolecular hydrogen bond with a close distance of 1.65 Å
between nitrogen and hydrogen is key in facilitating methanolysis
through TS26−27. With a ΔG⧧ of 22.9 kcal/mol relative to 25
this is considerably more accessible than the related solvolysis
process in mechanism B, which requires a barrier exceeding
40 kcal/mol.
Figure 3. Intermediates and TS involved in carbonylation of propyne
during mechanism D. Distances are in Å.
In this transition state coordinated methanol is deprotonated
by the cocatalytic 2-PyPPh2 ligand and the transient methoxide
attacks the Pd−C(acyl) bond. Since the MMA product
immediately dissociates away from the primary coordination
sphere (28/28b), this step is characterized as reductive
elimination. Here, the P,N ligand demonstrates its second
cocatalytic property as an in situ base, which is essential for the
high performance of methoxycarbonylation. Subsequent uptake
of propyne to re-form 21 provides the final driving force
for closing the catalytic cycle. TS26−27 also rationalizes why
2-PyPPh2 has much greater activity than 3-PyPPh2.2,3 In the
latter, the orientation of the nitrogen lone pair would not allow
for formation of the intramolecular hydrogen bond in 26, and the
ability of pyridyl to act as a cocatalyzing base is decreased.
Unlike the proton transfer steps, CO insertion is reversible with
an activation free energy of 10.3 kcal/mol and a backward cost
of 16.4 kcal/mol, notably lower than the barrier we compute for
methanolysis (Scheme 10). However, as the prior propyne
insertion step remains irreversible, the reversibility of CO insertion
will not affect the overall regioselectivity. Following carbonylation
a terminating alcoholysis step is necessary to yield MMA. Unlike
hydride mechanism B, this process does not require the costly
step of an alcoholic proton being transferred to a Pd(II) center but
can rather involve the 2-PyPPh2 ligand, reprotonating the pyridyl
moiety. As demonstrated in Scheme 10, incorporation of the
pyridyl as an in situ base offers a far more accessible route.
We note that one limitation of our computations is in the use
of a PCM to account for bulk solvation. Methanol molecules in
444
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
a
Scheme 10. Terminating Methanolysis to Yield MMA
a
Energies presented relative to 21. The rate-limiting step is solvolysis (TS26−27), while the overall cycle is thermodynamically driven.
Scheme 11. Selectivity-Determining Transition States for
Mechanism D
Figure 4. Geometries associated with MMA-producing methanolysis of
Pd-acyl species in mechanism D. Distances are given in Å.
The MC-forming methanolysis step is of little importance to
the overall regioselectivity of the methoxycarbonylation within
this reaction mechanism. Even if it is more favorable as a result
of lesser steric constraints for reductive elimination of the linear
acyl, it will not affect the ultimate product distribution, as this
is determined by the irreversible propyne insertion step, the
selectivity of which we now explore.
severe clashes; however in the MC-forming TS21L−22L this
movement introduces a more pronounced interaction between
the pyridyl ring and the methyl group of propyne, resulting in a
higher barrier. Interestingly, the irreversible propyne insertion
step can be viewed as equivalent to a Markovnikov addition of a
strong acid, [(2-PyPPh2)Pd(2-PyPPh2H)]+, to an alkyne. In this
analogy, the regioselectivity follows that of protonation at the
least substituted carbon of propyne, yielding the Markovnikov
product, while the anti-Markovnikov process is less favorable in
both kinetics and thermodynamics.
Table 1 reports the free energies governing turnover and
selectivity for ligand systems 2-PyPPh2 and 2-(6-Me)PyPPh2
within path D. Species 22 and 28/28b are not presented due
to their negligible impact on the reaction profile. We detail the
selectivities of the less favorable mechanistic alternative from
21+CO to 24 in SI Table S3, noting here that this route is
favorable for MMA production over MC. A comparison of these
energies shows that 6-methylation results in improved selectivity
for MMA against MC with ΔΔG⧧ increasing from 2.2 kcal/mol
to 4.1 kcal/mol, arising from a subtle ligand effect.
4.1. Selectivity of the in Situ Base Mechanism. The linear-
forming route through TS21L−22L was compared to the
branched-forming TS21−22 (Scheme 11). Thermodynamically,
the intermediate that leads to MMA formation (23) is favorable
over that leading to MC (23L), and furthermore the associated
saddle point (TS21−22) is more accessible than TS21L−22L.
The free-energy difference between the kinetic barriers for these
processes is 2.2 kcal/mol, a selectivity toward MMA of 98%
at 25 °C and 97% at 45 °C that is in line with experimental
observations. A concern that had led to the proposal of the
concerted pathway1 outlined in Scheme 4 was the expectation
that an alkenyl intermediate with bulk α to Pd would be less
favorable than the linear alternative. Calculations show that the
isomeric products are in fact almost equivalent in free energy and
the branched transition state does not stifle proton transfer.
Regioselectivity is controlled by steric effects, as in delivering
a proton the pyridyl moiety moves into the plane of the
π-coordinated propyne (Scheme 11). In TS21−22 there are no
4.2. Impact of Acidic Conditions. Finally, we turn to the effect
of strongly acidic conditions that are essential for achieving
445
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
Table 1. Comparative Reaction Profiles (ΔG, kcal/mol) for Methoxycarbonylation of Propyne under Differing Experimental
a
Conditions
ΔG
21
TS21−22
23
21L
TS21−22L
23L
2-PyPPh2
0.0
0.0
0.0
7.2
9.0
−17.7
−14.7
−14.2
1.8
1.2
1.3
9.4
13.1
13.4
27
−17.4
−15.4
−12.4
2-(6-Me)PyPPh2
2-PyPPh2 dicat.
10.3
24
TS24−25
25
26
TS26−27
21 + MMA
2-PyPPh2
−22.1
−20.6
−14.1
−11.8
−10.3
−5.2
−28.2
−25.1
−24.9
−27.5
−26.1
−24.3
−5.3
−3.7
−4.8
−24.9
−25.3
−24.8
−36.3
−36.3
−36.3
2-(6-Me)PyPPh2
2-PyPPh2 dicat.
a
Enthalpies are given in Table S4 in the SI.
expedient turnover. Under such conditions (and because the pKb
values of the two remote pyridine bases within the same complex
are expected to be very similar) not only one but both pyridyl
groups may be protonated.8 We have thus repeated all cal-
culations summarized in Schemes 9 and 10 for diprotonated,
dicationic complexes, and the resulting free energies are included
in Table 1 (see 2-PyPPh2 dicat. entries). According to these data,
not only does a dicationic equivalent improve selectivity with
ΔΔG⧧ between regioselective transition states increasing to
3.1 kcal/mol but, crucially, the methanolysis barrier is also
reduced to 20.1 kcal/mol.
becomes more negative (see the NPA charges in Figure 5),
facilitating methanolysis.
5. Analysis of Mechanisms A to D: Interpretation
through Energetic Span. The energetic span model of Kozuch
and Shaik88−90 allows for the calculation of turnover frequencies
from computed free-energy values of reactants, intermediates,
transition states, and products. This prescription further affords
detail of the most abundant reaction intermediates (MARIs) and
the highest energy transition state (HETS) that maximize the
energetic span along one complete cycle and exert kinetic control
on the propensity of the reaction.
Greater regioselectivity arises from protonation removing
the weak interaction between the basic pyridyl nitrogen and
acidic propyne hydrogen in TS21L−22L, which can be seen at
the left-hand side of Scheme 11. In the protonated congener of
this transition state, the previously basic nitrogen now presents
a partially positive hydrogen and the pyridyl group rotates to
avoid an unfavorable electrostatic interaction with coordinated
propyne. This protonation and subsequent reorientation removes
a polarizing interaction that reduces the effective barrier of
TS21L−22L in the monocationic 2-PyPPh2 system, and therefore
diprotonation leads to enhanced selectivity for MMA.
Excess acid also positively impacts turnover by decreasing the
rate-limiting solvolysis barrier to 20.1 kcal/mol. In a dicationic
system, the previously basic nitrogen lone pair of the non-
participating pyridyl is now replaced by an acidic N−H+
functionality, which accelerates methanolysis by forming an
intramolecular H-bond to the acyl carbon (right-hand side in
Figure 5).
While absolute TOF values are associated with the usual
uncertainties of simple transition-state theory, relative TOF
values can afford insight into the preferred reaction channels of
competing mechanisms. Table 2 reports the relative TOFs of all
Table 2. Mechanisms Relevant to the Methoxycarbonylation
of Propyne by Pd(P,N)n Systems under the Scrutiny of the
a
Energetic Span Model
TOF
energy
b
mechanism
(relative)
MARI
HETS
TS6−7
span
c
A (full 1 → 1a → 1)
1
5
38.0
31.2
54.2
42.5
24.5
22.9
22.4
20.1
d
A (partial 1a → 1a)
4.9 × 104
1.3 × 10−11
2.3 × 10−4
3.8 × 109
4.5 × 1010
1.2 × 1011
5a
11
11a
14
25
26
25
TS6a−7a
TS12−13
TS12a−13a
TS16−17
TS26−27
TS26−27
TS26−27
c
B (full 8 → 8a → 8)
d
B (partial 8a → 8a)
C-II
D
D-6Me
D-Dication
4.9 × 1012
a
b
Computed at T = 298.15 K. Free-energy difference in kcal/mol
between the MARI (most abundant reaction intermediate) and HETS
c
(highest energy transition state). Full cycle producing 2 equiv of
d
MMA. Half-cycle with lower energy span producing 1 equiv of MMA
and assuming rapid interconversion between 1 and 1a or 8 and 8a as
applicable.
computed profiles, taken against mechanism A, and additionally
identifies the MARIs and HETS for each reaction channel.
A (full) and B (full) have had their TOFs multiplied by 2, as these
cycles produce two equivalents of MMA.
In keeping with the free-energy barriers discussed previously,
the computed TOFs cover a large span, more than 23 orders
of magnitude. Compared to the highest TOF obtained for the
dicationic variant of pathway D (last entry in Table 2),
mechanisms A and B show negligible turnover. This conclusion
is relevant regardless of whether the full paths have to be
completed or shortcuts are available that could bypass the
highest barriers (cf. the discussion in sections 1 and 2 above).
Mechanism C-II, like D-Dication, a dicationic pathway, is barely
Figure 5. Transition states for the monocationic methanolysis, TS26−
27 (left), and dicationic analogue (right). NPA charges on key atoms are
included to highlight the different charge distributions.
This is confirmed by analysis of the natural population analysis
(NPA) charges and geometries at the methanolysis transition
states. On going from the mono- to the dicationic system, the
O(methanol)−C(acyl) separation increases from 1.71 Å to 1.81 Å,
consistent with an earlier transition state, and the acyl group
446
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
competitive with the latter, with relative turnover predicted to be
3 orders of magnitude lower.
sible protonation of coordinated propyne and terminating via
P,N-assisted methanolysis. With both 2-PyPPh2 and 2-(6-Me)-
PyPPh2 ligand systems, this catalytic cycle exhibits a selectivity
for the branched-forming route that is completely congruent
with experimental observations. Calculations show that the P,N
system is strikingly different from other ligands used in catalytic
carbonylation at Pd. Typically, these concern bidentate
diphosphine backbones and give high selectivities toward linear
(alkoxy)carbonylation products.35,91−93 By contrast, 2-PyPPh2
and congeners undergo an alternative “Pd−H” cycle, and it is the
unique mechanistic inclusion of the cocatalytic pyridyl moiety
in regioselective proton transfer that enables this framework to
conduct an alternative selectivity and favor branched products
over linear. We have been able to identify the subtle ligand effects
that dictate why the 6-methylated analogue exhibits enhanced
selectivity and can trace the requirement of acidic conditions back
to the involvement of at least one, but possibly two, hemilabile
2-PyPPh2 ligands protonated at the free nitrogen atoms during a
monocoordinating mode.
A diprotonated dicationic system is indicated to have the
lowest overall barrier (for rate-determining methanolysis), and
improving the accessibility of the solvolysis transition state has
been outlined as a potential target for directed design.
We are hopeful that our detailed computational insights, which
account for reactivities and selectivities at an atomistic level, will
spur the development of new generations of more active and
more selective catalyst systems for this important feedstock
molecule. We have already proposed a simple ligand modification
that should lower the turnover-determining barrier,51 and further
studies along these lines are in progress.
Pathway D, with its 10-fold increase in TOF over C-II and the
correct selectivity, is clearly the most plausible mechanism.
Crucially, D and D-6Me remain within an order of magnitude,
which is inline with experiment. The TOF of pathway D is still
2 orders of magnitude below that of D-Dication, but in practice
both will operate simultaneously, depending on the precise
position of the protonation equilibria, and an intermediate TOF
will be observed. Therefore, an in situ base mechanism is the
most likely one at this point, and the key intermediates and
transition states, prime targets for rational catalyst design, are
identified as 25/26 and TS26−27 respectively.
CONCLUSIONS
■
We have studied four of the most likely mechanisms for MMA
production at Pd catalysts with P,N ligand systems. Our
computations show that a typical carbomethoxy mechanism
(A) cannot account for selectivity or turnover. Likewise, a
hydride mechanism (B) can be excluded because it would suffer
deactivation as a result of a particularly high barrier for intra-
molecular methanolysis.
Drent’s original messenger carbomethoxy route (C) was
proposed to operate via the same selectivity-determining
transition state as that of A, which we have shown exhibits
a preference for MC over MMA, and we are confident in
excluding this route. Dervisi proposed a similar mechanism
that we have adapted to operate through a dicationic pathway.
This mechanism, C-II, always contains two 2-PyPPh2 units and
displays promising barriers and reasonable selectivity should
the Pd-OMe species be available under turnover conditions.
However, this mechanism fails to reproduce the observed
increase in selectivity for the 6-methylated analogue and can thus
be discounted.
The preferable pathway is D, a Pd(0) in situ base mechanism
with hemilabile and cocatalytic 2-PyPPh2 ligands and/or a
dicationic variant thereof (Scheme 12). While we could not
confirm a concerted protonation and migratory insertion step
that had been proposed recently,1 we were able to characterize a
stepwise pathway, beginning with a regioselective and irrever-
ASSOCIATED CONTENT
■
S
* Supporting Information
Model studies, additional energies for variants of mechanisms
A−D, XYZ coordinates of all complexes and small molecules,
and keywords employed are included. This material is available
free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: buehl@st-andrews.ac.uk.
Scheme 12. Catalytic Cycle of Mechanism D Involving an in
Situ Base, As Emerging from Our DFT Calculations
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
In memory of Prof. Dr. Paul von Rague Schleyer. The authors
would like to thank the University of St. Andrews School of
Chemistry and EaStCHEM for support and for access to a
computing facility maintained by Herbert Fruchtl. We also thank
̈
Prof. Eite Drent for key discussions and suggestions.
REFERENCES
■
(1) Cole-Hamilton, D. J.; Drent, E. Alkyne Carbonylation. In Applied
Homogeneous Catalysis with Organometallic Complexes; Cornils, B.,
Hermann, W. A., Beller, M., Paciello, R., Eds.; Wiley-VCH: Weinheim,
Germany, manuscript in preparation.
(2) Drent, E.; Arnoldy, P.; Budzelaar, P. H. M. J. Organomet. Chem.
1993, 455, 247.
(3) Drent, E.; Arnoldy, P.; Budzelaar, P. H. M. J. Organomet. Chem.
1994, 475, 57.
(4) Dervisi, A.; Edwards, P. G.; Newman, P. D.; Tooze, R. P.; Coles, S.
J.; Hursthouse, M. B. J. Chem. Soc., Dalton Trans. 1998, 1, 3771.
447
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
(5) Dervisi, A.; Edwards, P. G.; Newman, P. D.; Tooze, R. P.; Coles, S.
J.; Hursthouse, M. B. J. Chem. Soc., Dalton Trans. 1999, 3, 1113.
(6) Dervisi, A.; Edwards, P. G.; Newman, P. D.; Tooze, R. P. J. Chem.
Soc., Dalton Trans. 2000, 523.
(42) Angell, S. E.; Rogers, C. W.; Zhang, Y.; Wolf, M. O.; Jones, W. E.
Coord. Chem. Rev. 2006, 250, 1829.
(43) Portnoy, M.; Milstein, D. Organometallics 1994, 13, 600.
(44) Amatore, C.; Carre,
1995, 14, 1818.
́
E.; Jutand, A.; M’Barki, M. A. Organometallics
(7) Scrivanti, A.; Beghetto, V.; Campagna, E.; Zanato, M.; Matteoli, U.
(45) Krause, J.; Bonrath, W.; Klaus, P. Organometallics 1992, 11, 1158.
(46) Tang, C.-M.; Zeng, Y.; Yang, X.-G.; Lei, Y.-C.; Wang, G.-Y. J. Mol.
Catal. A Chem. 2009, 314, 15.
Organometallics 1998, 17, 630.
(8) Scrivanti, A.; Bertoldini, M.; Beghetto, V.; Matteoli, U.; Venzo, A. J.
Organomet. Chem. 2009, 694, 131.
(47) Sieffert, N.; Buhl, M. J. Am. Chem. Soc. 2010, 132, 8056.
(48) Das, P.; Sarmah, P. P.; Borah, M.; Phukan, A. K. Inorg. Chim. Acta
(9) Raj, P. A.; Dentino, A. R. Future Med. Chem. 2013, 5, 1635.
(10) Harris, B. Ingenia 2010, 18.
̈
(11) Moukwa, M. Chem. World 2010, 50.
2009, 362, 5001.
(12) De Pater, J. J. M.; Maljaars, C. E. P.; de Wolf, E.; Lutz, M.; Spek, A.
L.; Deelman, B.-J.; Elsevier, C. J.; van Koten, G. Organometallics 2005,
24, 5299.
(49) Zhang, Q.; Wu, C.; Zhou, L.; Li, J. Organometallics 2013, 32, 415.
(50) Haibach, M. C.; Guan, C.; Wang, D. Y.; Li, B.; Lease, N.; Steffens,
A. M.; Krogh-Jespersen, K.; Goldman, A. S. J. Am. Chem. Soc. 2013, 135,
15062.
(13) Consorti, C. S.; Ebeling, G.; Dupont, J. Tetrahedron Lett. 2002, 43,
(51) Crawford, L.; Cole-Hamilton, D. J.; Drent, E.; Buhl, M. Chem.
753.
̈
(14) Michalak, A.; Ziegler, T. Top. Organomet. Chem. 2005, 12, 145.
(15) Troegel, D.; Stohrer, J. Coord. Chem. Rev. 2011, 255, 1440.
(16) Vougioukalakis, G. C.; Grubbs, R. H. Chem. Rev. 2010, 110, 1746.
(17) Abram, U.; Alberto, R.; Dilworth, J. R.; Zheng, Y.; Ortner, K.
Polyhedron 1999, 18, 2995.
(18) Helmchen, G.; Pfaltz, a. Acc. Chem. Res. 2000, 33, 336.
(19) Dedieu, A.; Rohmer, M.; Rettig, S. J. Organometallics 2001, 20,
2966.
Eur. J. 2014, 20, 13923.
(52) Becke, A. D. J. Chem. Phys. 1996, 104, 1040.
(53) Perdew, J. P.; Chevary, J. A.; H, V. S.; Jackson, K. A.; Pederson, M.
R.; Singh, D. J.; Fiolhais, C. Phys. Rev. B 1992, 46, 6671.
(54) Perdew, J. P.; Burke, K.; Wang, Y. Phys. Rev. B 1996, 54, 533.
(55) Arnold, P. L.; Hollis, E.; Nichol, G. S.; Love, J. B.; Griveau, J.-C.;
Caciuffo, R.; Magnani, N.; Maron, L.; Castro, L.; Yahia, A.; Odoh, S. O.;
Schreckenbach, G. J. Am. Chem. Soc. 2013, 135, 3841.
(20) Felicissimo, M. P.; Batista, A. A.; Ferreira, A. G.; Ellena, J.;
Castellano, E. E. Polyhedron 2005, 24, 1063.
(56) Salman, S.; Bredas, J.-L.; Marder, S. R.; Coropceanu, V.; Barlow, S.
Organometallics 2013, 32, 6061.
́
(21) Chang, Y.; Lin, C.; Wang, S.; Liao, F. J. Organomet. Chem. 1999,
588, 160.
(57) Montero-Campillo, M. M.; Lamsabhi, A. M.; Mo,
Theor. Chem. Acc. 2013, 132, 1328.
́ ́
O.; Yanez, M.
̃
(22) Standfest-Hauser, C.; Slugovc, C.; Mereiter, K.; Schmid, R.;
Kirchner, K.; Xiao, L.; Weissensteiner, W. J. Chem. Soc., Dalton Trans.
2001, 2989.
(23) Wajda-Hermanowicz, K.; Ciunik, Z.; Kochel, A. Inorg. Chem.
2006, 45, 3369.
(58) Ignatov, S. K.; Panteleev, S. V.; Maslennikov, S. V.; Spirina, I. V.
Russ. J. Gen. Chem. 2013, 82, 1954.
(59) Wang, L.; Zhang, Y.; He, H.; Zhang, J. Synth. Met. 2013, 167, 51.
(60) Kesharwani, M. K.; Martin, J. M. L. Theor. Chem. Acc. 2014, 133,
1452.
(24) Wasserman, H. J.; Moody, D. C.; Paine, R. T.; Ryan, R. R.;
Salazara, K. V. Chem. Commun. 1984, 3, 533.
(61) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010,
132, 154104.
(62) Grimme, S.; Ehrlich, S.; Goerigk, L. J. Comput. Chem. 2011, 32,
1456.
(25) Alcock, N. W.; Moore, P.; Lampe, P. A.; Mok, K. F. J. Chem. Soc.,
Dalton Trans. 1982, 207.
(63) Risthaus, T.; Grimme, S. J. Chem. Theory Comput. 2013, 9, 1580.
(64) Becke, A. D.; Johnson, E. R. J. Chem. Phys. 2005, 122, 154104.
(65) Johnson, E. R.; Becke, A. D. J. Chem. Phys. 2006, 124, 174104.
(26) Olmstead, M. M.; Maisonnat, A.; Farr, J. P.; Balch, A. L. Inorg.
Chem. 1981, 4060.
(27) Flapper, J.; Wormald, P.; Lutz, M.; Spek, A. L.; van Leeuwen, P.
W. N. M.; Elsevier, C. J.; Kamer, P. C. J. Eur. J. Inorg. Chem. 2008, 2008,
4968.
(66) Sieffert, N.; Buhl, M. Inorg. Chem. 2009, 48, 4622.
̈
(67) McMullin, C. L.; Jover, J.; Harvey, J. N.; Fey, N. Dalton Trans.
2010, 39, 10833.
(28) Govindaswamy, P.; Carroll, P. J. J. Chem. Sci. 2006, 118, 319.
(29) Furst, M. R. L.; Le Goff, R.; Quinzler, D.; Mecking, S.; Botting, C.
H.; Cole-Hamilton, D. J. Green Chem. 2012, 14, 472.
(30) Furst, M. R. L.; Seidensticker, T.; Cole-Hamilton, D. J. Green
Chem. 2013, 15, 1218.
(31) Clegg, W.; Eastham, G. R.; Elsegood, M. R. J.; Tooze, R. P.; Wang,
X. L.; Whiston, K. Chem. Commun. 1999, 1877.
(32) Clegg, W.; Eastham, G. R.; Elsegood, M. R. J.; Heaton, B. T.; Iggo,
J. A.; Tooze, R. P.; Whyman, R.; Zacchini, S. Organometallics 2002, 21,
1832.
(33) Van Leeuwen, P. W. N. M.; Zuideveld, M. A.; Swennenhuis, B. H.
G.; Freixa, Z.; Kamer, P. C. J.; Goubitz, K.; Fraanje, J.; Lutz, M.; Spek, A.
L. J. Am. Chem. Soc. 2003, 125, 5523.
(34) Fanjul, T.; Eastham, G.; Fey, N.; Hamilton, A.; Guy Orpen, A.;
Pringle, P. G.; Waugh, M. Organometallics 2010, 29, 2292.
(68) Peng, C.; Ayala, P. Y.; Schlegel, H. B.; Frisch, M. J. J. Comput.
Chem. 1995, 17, 49.
(69) Gonzalez, C.; Schlegel, H. B. J. Chem. Phys. 1989, 90, 2154.
(70) Gonzalez, C.; Schlegel, H. B. J. Phys. Chem. 1990, 5523.
(71) Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 106, 5151.
(72) Tomasi, J.; Mennucci, B.; Cances
1999, 464, 211.
(73) Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105, 2999.
(74) Klamt, A.; Mennucci, B.; Tomasi, J.; Barone, V.; Curutchet, C.;
Orozco, M.; Luque, F. J. Acc. Chem. Res. 2009, 42, 489.
(75) We computed barriers for the reaction of 21 → TS21−22 and 25
→ TS26−27 with an alternative methodology of B3PW91-D2/ECP2/
PCM//B3PW91-D2/ECP1. The barriers optimized at the level of
B3PW91-D3 are lower by 0.3 and 0.4 kcal/mol, respectively. As such, we
do not anticipate that optimization with a dispersion inclusive functional
or method would drastically change or alter our conclusions with regard
to the overall preferred mechanism. These dispersion-corrected
geometries are provided in the SI.
(76) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb,
M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.;
Petersson, G. A.; Nakatsuji, H.; Caricato, M.; Li, X.; Hratchian, H. P.;
Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.;
Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima,
T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.;
Peralta, J. E.; Ogliaro, F.; Bearpark, M.; Heyd, J. J.; Brothers, E.; Kudin,
K. N.; Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.;
́
, E. J. Mol. Struct. (Theochem)
(35) Nunez Magro, A. A.; Robb, L.-M.; Pogorzelec, P. J.; Slawin, A. M.
́
̃
Z.; Eastham, G. R.; Cole-Hamilton, D. J. Chem. Sci. 2010, 1, 723.
(36) Liu, J.; Jacob, C.; Sheridan, K. J.; Al-Mosule, F.; Heaton, B. T.;
Iggo, J. A.; Matthews, M.; Pelletier, J.; Whyman, R.; Bickley, J. F.;
Steiner, A. Dalton Trans. 2010, 39, 7921.
(37) Kumar, P.; Singh, A. K.; Yadav, M.; Li, P.; Singh, S. K.; Xu, Q.;
Pandey, D. S. Inorg. Chim. Acta 2011, 368, 124.
(38) Guiry, P. J.; Saunders, C. P. Adv. Synth. Catal. 2004, 346, 497.
(39) Loiseleur, O.; Hayashi, M.; Keenan, M.; Schmees, N.; Pfaltz, A. J.
Organomet. Chem. 1999, 576, 16.
(40) Chelucci, G.; Orru, G.; Pinna, G. A. Tetrahedron 2003, 59, 9471.
̀
(41) Braunstein, P. J. Organomet. Chem. 2004, 689, 3953.
448
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449

Organometallics
Article
Rendell, A.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.;
Millam, J. M.; Klene, M.; Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.;
Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.;
Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.; Morokuma, K.;
Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
̈
Dapprichs, S.; Daniels, A. D.; Farkas, O.; Foresman, J. B.; Ortiz, J. V.;
Cioslowski, J.; Fox, D. J. Gaussian 09, Revision A.1, 2009.
(77) Purcell, W.; Conradie, J.; Chiweshe, T. T.; Venter, J. A.; Visser, H.
G.; Coetzee, M. P. J. Mol. Struct. 2013, 1038, 220.
(78) Both 1 and 3 may bind methanol instead of carbon monoxide and
propyne, respectively. In both cases, coordination of MeOH is found to
be less favorable than the route illustrated in Scheme 5. Against CO
uptake to form 2 from 1, binding of MeOH is 2.9 kcal/mol (ΔG
B3PW91-D3/ECP2/SDD PCM) uphill. Coordination of propyne to
form 4 from 3 has a driving force of 15.9 kcal/mol, which is greater than
MeOH coordination with 3, which has a gain of 9.1 kcal/mol. Solvent
inhibition of catalytic cycle A is therefore unlikely and has no noticeable
impact on either the selectivity- or rate-determining steps.
(79) Similar to mechanism A, complexes with MeOH coordinated in
the vacant site of 8 and 8a have been located. Such intermediates are
approximately 9 and 3 kcal/mol more stable in free energy (vs 8/8a,
respectively) though do not perturb the overall profile energetics.
Associative transition states for propyne undergoing hydride transfer
may thus be present, but we have not located such in light of the
challenging barrier for solvolysis. These solvated intermediates do not
alter predicted TOF values for cycle B.
(80) Donald, S. M. A.; Macgregor, S. A.; Settels, V.; Cole-Hamilton, D.
J.; Eastham, G. R. Chem. Commun. 2007, 562.
(81) Zuidema, E.; Bo, C.; van Leeuwen, P. W. N. M. J. Am. Chem. Soc.
2007, 129, 3989.
(82) Much debate still surrounds the exact mechanism of such
methanolysis processes,31−34,76,77,85−88 and we note that one commonly
proposed route is the loss of a proton from Pd(II)-coordinated methanol
into bulk, forming a high-lying methoxide species and allowing for
subsequent reductive elimination. Such a mechanism has been studied by
Zuidema77 and has been found to be considerably uphill. Therefore,
more efficient methanolysis routes may well be present, but these would
do little to resolve the lack of selectivity we find for mechanism B.
(83) Due to the small difference in kinetic energy between TS18−19
and TS18L−19L and the inclusion of moieties that contribute
noticeably to dispersive interactions (i.e., phenyl rings), we additionally
optimized these complexes at the level of B3PW91-D2/ECP1. At this
level, and following minimization on a dispersion-inclusive potential
energy surface, we find that the gross structure of these complexes is
retained. However, ΔΔG⧧ between these transition states drops from
1.6 kcal/mol (B3PW91/ECP1) to 0.6 kcal/mol at the aforementioned
level, with TS18−19 being more accessible in both cases. Upon
refinement at B3PW91-D2/ECP2/PCM//B3PW91-D2/ECP1, we find
that ΔΔG⧧ decreases further to less than 0.1 kcal/mol. A ΔΔG⧧ of 0.1
kcal/mol is also computed at B3PW91-D2/ECP2/PCM//B3PW91-
D2/ECP1 between TS18−19 6-Me and TS18L−19L 6-Me. This level
of theory, therefore, also discounts mechanism C-II from operating.
(84) Johnson, D. G.; Lynam, J. M.; Slattery, J. M.; Welby, C. E. Dalton
Trans. 2010, 39, 10432.
(85) Noyori, R.; Yamakawa, M.; Hashiguchi, S. J. Org. Chem. 2001, 66,
7931.
(86) Alonso, D. A.; Brandt, P.; Nordin, S. J. M.; Andersson, P. G. J. Am.
Chem. Soc. 1999, 121, 9580.
(87) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. 1985,
83, 735.
(88) Kozuch, S.; Shaik, S. J. Am. Chem. Soc. 2006, 128, 3355.
(89) Kozuch, S.; Lee, S. E.; Shaik, S. Organometallics 2009, 28, 1303.
(90) Kozuch, S.; Shaik, S. Acc. Chem. Res. 2011, 44, 101.
(91) Tijani, J.; Suleiman, R.; El Ali, B. Appl. Organomet. Chem. 2008, 22,
553.
(92) Suleiman, R.; Tijani, J.; El Ali, B. Appl. Organomet. Chem. 2009, 24,
38.
(93) Jimenez-Rodriguez, C.; Eastham, G. R.; Cole-Hamilton, D. J.
Dalton Trans. 2005, 2, 1826.
449
dx.doi.org/10.1021/om500970k | Organometallics 2015, 34, 438−449