A comprehensive mechanistic picture of the isomerizing alkoxycarbonylation of plant oils

Article abstract of DOI:10.1021/ja508447d

Theoretical studies on the overall catalytic cycle of isomerizing alkoxycarbonylation reveal the steric congestion around the diphosphine coordinated Pd-center as decisive for selectivity and productivity. The energy profile of isomerization is flat with diphosphines of variable steric bulk, but the preference for the formation of the linear Pd-alkyl species is more pronounced with sterically demanding diphosphines. CO insertion is feasible and reversible for all Pd-alkyl species studied and only little affected by the diphosphine. The overall rate-limiting step associated with the highest energetic barrier is methanolysis of the Pd-acyl species. Considering methanolysis of the linear Pd-acyl species, whose energetic barrier is lowest within all the Pd-acyl species studied, the barrier is calculated to be lower for more congesting diphosphines. Calculations indicate that energy differences of methanolysis of the linear versus branched Pd-acyls are more pronounced for more bulky diphosphines, due to involvement of different numbers of methanol molecules in the transition state. Experimental studies under pressure reactor conditions showed a faster conversion of shorter chain olefin substrates, but virtually no effect of the double bond position within the substrate. Compared to higher olefins, ethylene carbonylation under identical conditions is much faster, likely due not just to the occurrence of reactive linear acyls exclusively but also to an intrinsically favorable insertion reactivity of the olefin. The alcoholysis reaction is slowed down for higher alcohols, evidenced by pressure reactor and NMR studies. Multiple unsaturated fatty acids were observed to form a terminal Pd-allyl species upon reaction with the catalytically active Pd-hydride species. This process and further carbonylation are slow compared to isomerizing methoxycarbonylation of monounsaturated fatty acids, but selective.

Full text of DOI:10.1021/ja508447d

Article
pubs.acs.org/JACS
A Comprehensive Mechanistic Picture of the Isomerizing
Alkoxycarbonylation of Plant Oils
Philipp Roesle,† Lucia Caporaso,*,‡ Manuel Schnitte,† Verena Goldbach,† Luigi Cavallo,‡,§
and Stefan Mecking*,†
†Chair of Chemical Materials Science, Department of Chemistry, University of Konstanz, 78464 Konstanz, Germany
‡Department of Chemistry, University of Salerno, Via Giovanni Paolo II, 84084-Fisciano, SA, Italy
§Chemical and Life Sciences and Engineering, Kaust Catalysis Center, King Abdullah University of Science and Technology, Thuwal
23955-6900, Saudi Arabia
S
* Supporting Information
ABSTRACT: Theoretical studies on the overall catalytic cycle
of isomerizing alkoxycarbonylation reveal the steric congestion
around the diphosphine coordinated Pd-center as decisive for
selectivity and productivity. The energy profile of isomer-
ization is flat with diphosphines of variable steric bulk, but the
preference for the formation of the linear Pd-alkyl species is
more pronounced with sterically demanding diphosphines. CO
insertion is feasible and reversible for all Pd-alkyl species
studied and only little affected by the diphosphine. The overall
rate-limiting step associated with the highest energetic barrier
is methanolysis of the Pd-acyl species. Considering meth-
anolysis of the linear Pd-acyl species, whose energetic barrier is
lowest within all the Pd-acyl species studied, the barrier is
calculated to be lower for more congesting diphosphines. Calculations indicate that energy differences of methanolysis of the
linear versus branched Pd-acyls are more pronounced for more bulky diphosphines, due to involvement of different numbers of
methanol molecules in the transition state. Experimental studies under pressure reactor conditions showed a faster conversion of
shorter chain olefin substrates, but virtually no effect of the double bond position within the substrate. Compared to higher
olefins, ethylene carbonylation under identical conditions is much faster, likely due not just to the occurrence of reactive linear
acyls exclusively but also to an intrinsically favorable insertion reactivity of the olefin. The alcoholysis reaction is slowed down for
higher alcohols, evidenced by pressure reactor and NMR studies. Multiple unsaturated fatty acids were observed to form a
terminal Pd-allyl species upon reaction with the catalytically active Pd-hydride species. This process and further carbonylation are
slow compared to isomerizing methoxycarbonylation of monounsaturated fatty acids, but selective.
By contrast, with their kinetically controlled nature selective
isomerization/functionalization5c,d,6 approaches in principle can
incorporate the entire fatty acid chain. This is particularly
difficult, however, as terminal olefins are thermodynamically
strongly disfavored versus the internal double bonds of the
substrate. The state of the art of isomerization/functionalization
in terms of terminal selectivity and a lack of other undesired
side reactions like olefin hydrogenation or further reactions of
the products is isomerizing alkoxycarbonylation.7 In this
remarkable reaction, palladium(II) catalysts modified with
electron rich, sterically demanding diphosphines like 1,2-
(CH2PtBu2)2C6H4 (dtbpx)8 convert the double bond deep in
the chain of unsaturated fatty acids to a terminal ester group
with high selectivity (up to >90%) and thus fully incorporate
the fatty acid starting material into an α,ω-diester (Scheme
INTRODUCTION
■
The utilization of renewable resources as a source of chemicals
requires their efficient transformation to useful building blocks.
Fatty acids from plant oils are attractive substrates due to their
unique long-chain methylene sequences.1 Their incorporation
into linear long-chain α,ω-functionalized compounds is of
interest, for example, for the generation of semicrystalline
aliphatic polyesters,2 hydrophobic polyamides2b and hydrolyti-
cally degradable polyacetals.3 The terminal functionalization of
fatty acids is a synthetic challenge, however. Biotechnological4
as well as entirely chemical catalytic approaches5 are studied to
this end. Self-metathesis of oleate can yield 1,18-octadecan-
dioates (after subsequent hydrogenation of the double
bond).1b,5a,b However, stoichiometric amounts of the C18-
alkene are formed. Also, as an equilibrium reaction only 50%
conversion can be attained unless the product can be removed
from the reaction mixture selectively.
Received: August 25, 2014
© XXXX American Chemical Society
A
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
1).2a,b,9 Notably not only pure oleate is converted, but also
technical grade plant oils which are mixtures of fatty acids with
from a combined theoretical and experimental NMR and
pressure reactor study. Relevant pathways to different branched
byproducts are fully analyzed, which is a prerequisite for
understanding and identifying the decisive structural features
that make up a catalyst for this unique reaction. This analysis
also provides for the first time an unravelling of a successful
catalytic isomerization/functionalization scheme.
Scheme 1. Isomerizing Methoxycarbonylation of Plant Oils
Generating Linear Long-Chain Diesters
RESULTS AND DISCUSSION
■
Pressure Reactor Studies. To obtain data allowing for a
quantitative comparison of the reaction of different substrates,
preparative experiments with different olefins were performed
in a 200 mL pressure reactor using 29.5 mmol of substrate and
0.8 mol % [(dtbpx)Pd(OTf)2] as a catalyst precursor under 20
bar CO at 90 °C in neat methanol as a solvent and reagent. To
gain insights into the role of isomerization under reaction
conditions, a prior exposure of the olefinic substrate to the
catalyst in the absence of CO was avoided. The catalyst was
added into the loaded (substrate + methanol) and prepressur-
ized (2 bar CO) reactor (for details cf. Supporting
Information). Already at low conversions (e.g., ∼15% for
methyl oleate) isomerization of the substrate to the
thermodynamic distribution of all isomers is observed, as
evidenced by comparison with a genuine sample of the
substrate with added [(dtbpx)Pd(OTf)2] in methanol
(Supporting Information). Note that during the entire catalytic
transformation this distribution does not change and thus
indicates that isomerization occurs, even in the presence of
substantial amounts of CO (20 bar).
different numbers of carbon atoms and in particular multiple
double bonds, and even pure linoleate is converted to linear
diesters.
The perhaps most prominent isomerization/functionaliza-
tion reaction to date is isomerizing hydroformylation.10 For the
isomerizing hydroformylation of methyl oleate, a selectivity of
40% has been achieved.5c Other than the isomerizing
alkoxycarbonylation studied here, isomerizing hydroformylation
is mostly catalyzed by trigonal bipyramidal rhodium complexes.
The different nature and mechanism is also reflected in directly
practically relevant features. Thus, the selectivity for linear
products depends on the starting material (e.g., 1-octene results
in higher linear selectivity than 2- or 4-octene) which is not the
case for isomerizing alkoxycarbonylation (vide infra). In
addition, side reactions like olefin hydrogenation and reduction
of the generated products to alcohols are common, especially
when functional ester groups are present in the substrate. It is
worth noting that although the individual reactions of olefin
isomerization and of hydroformylation have been studied
extensively for numerous catalyst systems,11 a mechanistic
analysis of the overall catalytic cycle and reaction sequence does
not exist for a catalyst capable of selective isomerizing
hydroformylation.10
The same catalyst system as found to be efficient for
isomerizing alkoxycarbonylation is used for the industrial
production of methyl propionate by alkoxycarbonylation11 of
ethylene with CO and methanol.12 This reaction has been
deeply investigated and is well understood mechanistical-
ly.8a,b,13 Isomerization and linear versus branched selectivity
obviously do not apply here. Rates for this methoxycarbony-
lation of ethylene8a are substantially higher than rates for the
isomerizing methoxycarbonylation of methyl oleate2b and
methyl linoleate.5a
Average TOFs > 3,500 h−1 for ethylene, ca. 12 h−1 for methyl
oleate and ca. 2 h−1 for methyl linoleate,16 respectively, confirm
the aforementioned higher productivity in ethylene methox-
ycarbonylation. Comparing rates of different mono-olefinic
substrates (Figure 1) the conversion for shorter chain
substrates is faster (ethylene > octenes > octadecene).
Remarkably, the position of the double bond within the
substrate virtually does not influence the rate (1-octene versus
4-octene).
Cole-Hamilton and co-workers concluded from their study
on the isomerizing methoxycarbonylation of octenes, that a
hydride mechanism is operative and methanolysis is the rate-
determining step.7b Our direct observations of stoichiometric
isomerizing methoxycarbonylation of methyl oleate by low
temperature NMR spectroscopy showed that under these
conditions only linear product and one branched acyl, stabilized
by chelated coordination of the substrates ester group form.
Supported by DFT studies, this indicates that methanolysis is
not only rate-determining, but also responsible for the
selectivity of this remarkable transformation.14 Recent findings
suggest that steric congestion around the palladium center
enhances productivity and selectivity of isomerizing alkox-
ycarbonylation.15
Figure 1. Conversion of different substrates in (isomerizing)
alkoxycarbonylation in methanol as a solvent. Reaction conditions:
29.5 mmol olefin, 0.8 mol % [(dtbpx)Pd(OTf)2], total volume 180
mL, 20 bar CO, 90 °C. Typical composition of HO-sunflower oil
methyl ester: 92.5% methyl oleate (18:1); 2.5% methyl linoleate
(18:2); 2.5% methyl palmitate (16:0); 1.5% methyl stearate (18:0);
1.0% methyl esters of longer chain (>C18) fatty acids.
At this point, from these single pieces it is unclear how the
overall transformation of isomerizing alkoxycarbonylation can
be accounted for. Against this background, we now report a
comprehensive mechanistic picture including the role of
isomerization, and the origin of selectivity and reaction rates,
B
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
Figure 2. Energy profile (ΔG in kcal mol−1) of isomerization along the hydrocarbon chain of the model substrate methyl heptenoate with the dtbpx
(red line) and dmpx (blue line) coordinated diphosphine palladium(II) hydride species (relative to the energy of the Pd−H fragment with
coordinated methyl 4-heptenoate as a reference point, set to zero).23 “X-β-Y” indicates the respective Pd-Alkyl species,24 “TS” the respective
transition states, “BHE” the β-H elimination, and “Ins” the insertion reaction.
Selectivity for the different monoester products methyl
nonanoate (95.0%), methyl 2-methyloctanoate (3.7%), methyl
2-ethylheptanoate (0.8%), and methyl 2-propylhexanoate
(0.5%) is also identical for both octenes, thus indicating that
isomerization is not the rate-limiting step. However, the length
of the hydrocarbon chain of the substrate influences the overall
rate. Note that for all substrates the selectivity does not change
with time thus indicating that a conceivable decomposition of
the catalyst does not influence the selectivity of isomerizing
methoxycarbonylation.
Chart 1. Diphosphine Palladium(II) Complexes with
Different Steric Congestion around the Palladium Center
Used for DFT Calculations of Isomerizing
Alkoxycarbonylation
Isomerization along the Hydrocarbon Chain. As
isomerization is fast, even at temperatures of −80 °C,14
experimental data alone is not conclusive in describing the
overall energy profile of the isomerization reaction. DFT
calculations starting from the diphosphine palladium(II)
hydride fragment [(P∧P)Pd(H)]+ and methyl 4-heptenoate as
model substrate were thus performed, calculating a series of
insertion and subsequent β-hydride elimination reactions
(Figure 2 and Supporting Information). These DFT calcu-
lations were performed with the Gaussian09 package17 using
the B3LYP18 functional and the LANL2DZ ECP19 with the
associated valence basis set on the Pd-atom, and the 6-
31G(d)20 basis set on all the other atoms. All geometries were
localized in the gas phase at the B3LYP level. Minima were
localized by full optimization of the starting structures, while
transition states were approached through a linear transit
procedure starting from the coordination intermediate and then
located by a full transition state search. All structures were
confirmed as minimum or transition state through frequency
calculations.21 To gain insights into the influence of the
catalysts’ structure on catalysis, namely its steric repulsion
around the metal center, the calculations were performed with
the sterically demanding 1,2-(CH2PtBu2)2C6H4 (dtbpx)
diphosphine and the less demanding 1,2-(CH2PMe2)2C6H4
(dmpx) diphosphine (Chart 1).22
group with the metal center. The five-membered chelate (5-
m.c.) is favored by 3.3 kcal mol−1 for the dmpx coordinated
metal center, whereas in case of the sterically demanding dtbpx
diphosphine the four-membered chelate (4-m.c.) is favored by
2.6 kcal mol−1 due to steric interactions of the diphosphines
tBu-groups with the alkyl chain of the substrate thus making a
four-membered species more favorable although it may have
higher ring strain within the chelate. Considering the other
alkyl species (3-β-2 to 7-β-6)24 the linear Pd-alkyl species 7-β-6
is favored by 1.7 kcal mol−1 over the methyl branched alkyl
species 6-β-7 and over all other branched alkyl species by more
than 2 kcal mol−1 for the bulky dtpbx diphosphine. This is not
the case for the less bulky dmpx system where no preference for
the linear alkyl species is found. The 5-β-4 alkyl species of the
dmpx system is lowest in energy due to a favorable interaction
of the substrates ester group with the β-hydrogen interacting
with the metal center, thus making this alkyl species low in
energy for this particular diphosphine Pd-alkyl species (cf.
Supporting Information). Such an interaction of the substrates’
ester group with a β-hydrogen is not observed for the more
sterically demanding dtbpx system. Energy differences for the
alkyl species can be directly traced to the specific nature of the
metal center. For the bulky dtbpx system steric interactions are
more pronounced than electronic interactions and vice versa
for the less demanding dmpx system. Notably this is observed
for the 6-β-7 versus 6-β-5 alkyl species: 6-β-7 with its
interaction to a primary β-hydrogen is favored by 3.6 kcal
For both diphosphines the energetic profile of isomerization
is relatively flat with barriers around 6−12 kcal mol−1.23 For
both diphosphines, a chelated species is lowest in energy due to
the electronically favored interaction of the substrates’ ester
C
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
mol−1 for the case of dtbpx due to lower steric demand of the
substrate around the metal center, while 6-β-5 is favored by 3.9
kcal mol−1 for the less demanding dmpx ligand due to the
favored electronic interaction of a secondary β-hydrogen with
the metal center versus the primary β-hydrogen interaction in
6-β-7 (Figure 2 and Supporting Information). From these
studies of isomerization of methyl heptenoate along its
hydrocarbon chain a preference for the formation of linear
versus branched alkyl species is observed for sterically
demanding ligands (dtbpx) which is not the case for less
crowded diphosphines (dmpx). The former is in accordance
with our aforementioned studies,14 where the formation of a
linear and a four-membered chelated species was observed
when the catalytically active hydride species was exposed to
methyl oleate. Notably, these calculations suggest that isomer-
ization occurs with low barriers with either diphosphine.
Insertion of CO into Pd-Alkyl Species. CO insertion into
the Pd-alkyl species was studied for both diphosphines starting
from key alkyls, namely the four- (4-m.c.) and five- (5-m.c.)
membered chelate, the linear Pd-alkyl 7-β-6, the methyl
branched alkyl 6-β-5 or 6-β-7 (depending on which of them
is energetically favored) and the ethyl branched alkyl 5-β-4 (for
complete details cf. Supporting Information).25 Despite
somewhat differing energies of the starting Pd-alkyl species
due to chelating coordination of the substrates ester group
(vide supra), barriers for CO insertion are only 8 to 18 kcal
mol−1 and thus not prohibitive for the different linear and
branched alkyls with both diphosphines. In general CO
insertion is slightly exergonic. However, this energy released
is not very high, such that CO insertion can be considered to be
reversible. This agrees with experimental findings at low
temperature for two selected alkyls.14 The most notable
difference observed is a generally more facile CO coordination
with the less bulky diphosphine (dmpx).
Figure 3. Energy profile (ΔG in kcal mol−1) of CO insertion and
subsequent methanolysis of the sterically congested dtbpx coordinated
linear Pd-alkyl species (7-β-6; red line) and the methyl branched Pd-
alkyl species (6-β-7; blue line), relative to the energy of the linear Pd-
alkyl 7-β-6 as a reference point, set to zero.
Figure 4. Energy profile (ΔG in kcal mol−1) of CO insertion and
subsequent methanolysis of the dmpx coordinated linear Pd-alkyl
species (7-β-6; red line) and the methyl branched Pd-alkyl species (6-
β-5; blue line), relative to the energy of the linear Pd-alkyl 7-β-6 as a
reference point, set to zero.
However, both the decisive energy barriers (ΔG‡) of CO
insertion into the Pd-alkyl species and also the energetics (ΔG)
of the overall CO insertion reaction are little affected by the
diphosphine coordinated to the Pd-center and are thus in
general little affected by the steric repulsion around the Pd-
center. This was not only found by these theoretical studies but
also experimentally by insertion of CO into diphosphine Pd-
alkyl model compounds [(P∧P)Pd(Me)Cl] with variable steric
demanding diphosphines (vide infra).
Methanolysis of Pd-Acyl Species. As outlined, meth-
anolysis is the rate-determining step of isomerizing methox-
ycarbonylation. For methanolysis of the linear Pd-acyl species
[(P∧P)PdC(O)(CH2)6COOMe]+ calculations suggest that a
cluster of three methanol molecules is favored versus a single
methanol coordinated transition state (TS) for the case of the
bulky diphosphine (dtbpx).26 Conversely, for all the branched
alkyl species a single methanol coordinated transition state is
calculated with lower energy versus the respective three
methanol coordinated TS (Figure 5 and Figures S63 and
S65). This change in the mechanism decisively contributes to
the high selectivity with bulky diphosphines. In contrast for the
less demanding dmpx diphosphine the three methanol cluster
mechanism is calculated to be lower in energy for all Pd-acyl
species (Figure S64). For both systems and mechanisms the
lowest energies of methanolysis are calculated for the linear Pd-
acyl species. However, the energetic barrier for methanolysis of
the linear Pd-acyl is lower for the dtbpx coordinated species by
ΔΔG‡ = 3.0 kcal mol−1 (ΔΔG‡ = ΔG‡dmpx − ΔG‡dtbpx = 32.1−
Figure 5. Transition states of linear (three-fold MeOH coordinated;
left) versus methyl branched (single MeOH coordinated; right) Pd-
acyl species.
29.1 kcal mol−1 in Figures 4 and 3), thus indicating significantly
higher reaction rates for a more crowded Pd-center.27
For the dtbpx coordinated Pd-center, calculations indicate
that methanolysis of the methyl branched Pd-acyl is lowest
within the branched acyl species, but it is 6.1 kcal mol−1 higher
in energy as compared to the linear analogue (ΔΔG‡ in Figure
3). The highest energy was found for the methanolysis TS of
the five-membered Pd-acyl species (formed by CO insertion
D
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
into the four-membered Pd-alkyl species), 9.4 kcal mol−1 higher
than the linear Pd-alkyl (for a comparison of all pathways cf.
Figure S65).
For the dmpx coordinated Pd-center, methanolysis of the Pd-
acyl formed by insertion of CO into the five-membered Pd-
alkyl to afford dimethyl 2-alkylsuccinate is calculated to be
lowest within the branched Pd-acyl species but 1.2 kcal mol−1
higher than the TS of the linear Pd-acyl (for a comparison of all
pathways cf. Figure S64). The highest barrier is found for the
ethyl branched Pd-acyl species which is 4.8 kcal mol−1 higher as
compared to the TS of the linear Pd-acyl (Figure S64). The TS
of methanolysis of the methyl branched acyl species is
calculated 1.5 kcal mol−1 higher as compared to the TS of
the linear Pd-acyl (ΔΔG‡ in Figure 4). The difference in the
TSs energies is thus less pronounced than for the dtbpx
coordinated species. Coming along with a change of the
mechanism from a three methanol molecule cluster versus a
single methanol molecule TS, these calculations indicate that
energy differences in methanolysis TSs are less pronounced for
the less demanding dmpx diphosphine. This overall picture
suggests that the preference for the linear diester as the major
product is not unique to the crowded dtbpx, but due to these
differences the selectivity for the desired linear product is
indeed higher.
Figure 6. Energy profile (ΔG in kcal mol−1) of ester formation starting
from the Pd-hydride fragment [(dtbpx)Pd(H)]+, ethylene and the
longer chain olefinic substrate methyl 6-heptenoate (relative to the
energy of the Pd-hydride fragment as a reference point, set to zero).
Chart 2. Diphosphines Evaluated in Isomerizing
Methoxycarbonylation
Influence of the Substrate. In pressure reactor experi-
ments (vide supra), conversion is faster for shorter chain
substrates (average TOFs: > 3500 h−1 for ethylene, ca. 91 h−1
for 1-octene and 4-octene, and ca. 39 h−1 for 1-octadecene).28
Selectivity to the linear product is 95% for the octenes and 90%
for 1-octadecene and thus decreases with increasing length of
the hydrocarbon chain. When comparing rates and selectivity of
the conversion of octene versus octadecene, the overall reaction
rate is ca. only half and the amount of branched side products
twice as high for the longer chain substrate. Both observations
may be due to different concentrations of the linear Pd-acyl,
which is the starting point of the preferred further rate-
determining methanolysis. A slight preference for the linear
versus the numerous branched acyls (ΔΔG‡ ≥ 1 kcal mol−1) is
calculated, but all will be equilibrated, such that for a longer
chain substrate a larger part of the acyls will be present as less
reactive branched acyls (Figure 3 and Supporting Information).
Comparing the rate of ethylene versus octene in (isomer-
izing) methoxycarbonylation, calculations indicate the higher
rate of ethylene carbonylation is likely not only due to the
occurrence of linear acyls exclusively, but also due to a
significantly lower transition state of the rate-determining
methanolysis step by ΔΔG‡ = 9.7 kcal mol−1 (Figure 6). This
seems an intrinsic problem of the substrates’ nature as
formation of a Pd-alkyl is calculated to be more favored for
Pd−H + ethylene versus Pd−H + 1-olefin, resulting in lower
rates for (isomerizing) alkoxycarbonylation of higher olefins
versus ethylene, although barriers of methanolysis of the
respective Pd-acyls are similar for both substrates.
Alkyl bridged diphosphines (P∧P = R2P(CH2)nPR2; n = 1−4;
t
i
R= Bu, Pr, Et) were used instead of 1,2-(CH2PR2)2C6H4
diphosphines, as the former allow for more straightforward
synthesis, and tBu2P(CH2)3PtBu2 (dtbpp) is known to afford an
active catalyst for isomerizing alkoxycarbonylation (and was
also used for previous mechanistic NMR studies). Two
approaches were followed to achieve variable steric demand
around the metal center: the modification of the backbone of
the diphosphines and the modification of the substituents on
phosphorus. By increasing the number of bridging methylene
units (C1 to C4; dtbpm, dtbpe, dtbpp, dtbpb) between two di-
tert-butylphosphine moieties the space around the metal center
decreases (vide infra). Modification of the substituents on
phosphorus from tert-butyl to isopropyl and ethyl, respectively,
with constant length of the bridge (C3; dtbpp, dippp, depp),
increases the available space (vide infra). The available space
around the metal center was quantified by the opening angle at
the metal center left open by the diphosphine in the respective
Pd-complexes [(P∧P)Pd(OTf)2] (Table 1),29 calculated from
the respective X-ray crystal structures (for X-ray crystal
structure analysis cf. Supporting Information). Larger values
for the opening angle indicate less crowded metal centers. The
experimental data show the tendency of the more sterically
crowded metal centers [(dtbpp)Pd(OTf)2] and [(dtbpx)Pd-
(OTf)2] to be more productive catalysts for isomerizing
alkoxycarbonylation (Table 1).30 All the other less crowded
systems did not show any significant conversion of methyl
oleate. Surprisingly [(dtbpb)Pd(OTf)2] is also nonproductive
although its steric congestion is almost equal to [(dtbpx)Pd-
(OTf)2].
Experimental Studies on Diphosphine Structure. To
probe these theoretical findings and to gain further insights on
the influence of the diphosphines’ structure on catalysis, a series
of diphosphine palladium(II) ditriflate complexes [(P∧P)Pd-
(OTf)2] and diphosphine palladium(II) methyl chloro
complexes [(P∧P)Pd(Me)Cl] with variable steric demand
around the metal center were synthesized and evaluated
(Chart 2, for diphosphine and complex synthesis cf. Supporting
Information).
E
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
observed in the presence of CO. NMR analysis of the reaction
mixture reveals the formation of a bridged hydrido-carbonyl
species [(dtbpe)Pd(μ-H)(μ-CO)Pd(dtbpe)]+OTf− (Scheme
2) as previously observed for [(dtbpp)Pd(OTf)2].14 Addition
Table 1. Opening Angle and Catalytic Results of
Diphosphine Palladium(II) Ditriflate Complexes in
Isomerizing Methoxycarbonylation of HO-Sunflower Oil
Methyl Ester
opening
Scheme 2. Formation and X-ray Crystal Structure of the
Hydrido-Carbonyl Bridged Species [(dtbpe)Pd(μ-H)(μ-
CO)Pd(dtbpe)]+OTf− from [(dtbpe)Pd(OTf)2] in the
angle
conversion of
selectivity to
f
c
f
complex
[deg]
MO [%]
linear diester [%]
a
d
[(dtbpm)Pd(OTf)2]
213.4
192.6
175.0
<0.5
<0.5
n.d.
n.d.
82
a
a
a
Presence of Methanol and CO
[(dtbpe)Pd(OTf)2]
e
e
[(dtbpp)Pd(OTf)2]
25.4 (18 h)
70.6 (90 h)
a
[(dtbpb)Pd(OTf)2]
172.8
197.6
223.8
172.2
<0.5
<0.5
<0.5
94.8
n.d.
n.d.
n.d.
91
b
[(dippp)Pd(OTf)2]
b
[(depp)Pd(OTf)2]
[(dtbpx)Pd(OTf)2]
a
a
Reaction conditions: n(Pd) = 0.048 mmol; V(MO) = 2 mL;
V(MeOH) = 8 mL; p(CO) = 20 bar; T = 90 °C; t = 90 h (for
b
[(dtbpp)Pd(OTf)2]: 18 and 90 h, for dtbpx: 18 h). Reaction
conditions: n(Pd) = 0.024 mmol; V(MO) = 1 mL; V(MeOH) = 4
c
mL; p(CO) = 20 bar; T = 90 °C; t = 90 h. Opening angles were
determined from X-ray crystal structure data as described in ref 29.
d
As no crystals suitable for X-ray diffraction of [(dtbpm)Pd(OTf)2]
e
were obtained, data of [(dtbpm)PdCl2] was used. Previously reported
a
Hydrogen atoms (except μ2-H) and a noncoordinating triflate
data from our group was used for [(dtbpp)Pd(OTf)2] (CCDC
893845) and [(dtbpx)Pd(OTf)2] (CCDC 817578). Conversion and
selectivity determined from GC of crude reaction mixture.
f
counterion were omitted for clarity. Displacement ellipsoids are shown
at the 50% probability level.
of CO to a methanol-d4 solution of [(dtbpb)Pd(OTf)2]
resulted in quantitative formation of the deuterated diphos-
phine (tBu2DP(CH2)4PDtBu2)(OTf)2 as evidenced in 31P
With regard to their observed lack of catalytic activity,
complexes [(dtbpm)Pd(OTf)2], [(dtbpe)Pd(OTf)2] and
[(dtbpb)Pd(OTf)2] were investigated in more detail. H and
1
2
31P NMR spectra of [(dtbpm)Pd(OTf)2] and [(dtbpe)Pd-
(OTf)2] in methanol-d4 or in a mixture of CD2Cl2/MeOH did
not indicate the formation of any Pd-hydride species31 (note
that [(dtbpp)Pd(OTf)2] and [(dtbpx)Pd(OTf)2] cleanly yield
a defined hydride species upon dissolution in methanol-d4 or
CD2Cl2/MeOH).14,15 [(dtbpb)Pd(OTf)2] forms several non-
identified Pd-hydride species, also underlined by addition of 20
equiv of methyl oleate to the NMR tube resulting in rapid
isomerization (within 5 min) of the substrates’ double bond to
the equilibrium mixture of all isomers as observed for
[(dtbpp)Pd(OTf)2] previously.14 Addition of 20 equiv of
methyl oleate to the methanol solution of [(dtbpm)Pd(OTf)2]
showed no isomerization even after 12 h at room temperature.
Only after heating to 90 °C for 90 min, isomerization to the
equilibrium mixture was completed. With [(dtbpe)Pd(OTf)2]
slow isomerization of methyl oleate (20 equiv) is observed
already at room temperature, however, after 12 h at room
temperature isomerization to the equilibrium mixture is not
completed. Upon heating to 90 °C for 90 min, hydrogenation
of the substrates’ double bond to yield methyl stearate was
observed.32 These findings of olefin isomerization indicate the
formation of certain amounts of Pd-hydride species for
[(dtbpm)Pd(OTf)2] and [(dtbpe)Pd(OTf)2] although not
directly observed by NMR spectroscopy; however, entry into
the catalytic cycle may already be much less efficient for these
catalyst systems.
NMR spectra by a triplet at δ = 46.6 ppm with JPD = 70.2
Hz. Addition of methyl oleate to this solution did not result in
any isomerization. Thus, the inactivity of [(dtbpb)Pd(OTf)2]
in isomerizing alkoxycarbonylation is due to rapid decom-
position of the catalytically active species in the presence of
CO.34
To overcome the problem of less efficient hydride formation
(vide supra) of [(dtbpm)Pd(OTf)2] and [(dtbpe)Pd(OTf)2]
the methyl chloro complexes [(dtbpm)Pd(Me)Cl], [(dtbpe)-
Pd(Me)Cl] and [(dtbpp)Pd(Me)Cl] were used for isomerizing
alkoxycarbonylation of the less challenging substrate 1-octene
(Table 2). Note that these neutral diphosphine methyl chloro
Table 2. Catalytic Results of Diphosphine Palladium(II)
Methyl Chloro Complexes in Isomerizing
Methoxycarbonylation of 1-Octene
conversion of 1-
selectivity to methyl
b
b
complex
octene [%]
nonanoate [%]
a
[(dtbpm)Pd(Me)Cl]
5.7
73
92
90
a
[(dtbpe)Pd(Me)Cl]
[(dtbpp)Pd(Me)Cl]
33.8
96.1
a
a
Reaction conditions: n(Pd) = 0.051 mmol; V(1-octen) = 1 mL (6.3
mmol); V(MeOH) = 4 mL; p(CO) = 20 bar; T = 90 °C; t = 90 h.
b
Conversion and selectivity determined from GC of crude reaction
mixture.
To further probe the complexes’ ability in isomerization, the
above-mentioned experiments were performed in the presence
of 6 equiv of CO.33 [(dtbpm)Pd(OTf)2] did not show any
isomerization in the presence of CO even at a temperature of
90 °C. [(dtbpe)Pd(OTf)2] showed no isomerization in the
presence of CO at room temperature but at temperatures of 90
°C isomerization to the equilibrium mixture was observed
within 2 h. However, the formation of methyl stearate was not
complexes [(P∧P)Pd(Me)Cl] were used, as addition of AgOTf
to generate the respective cationic complexes [(P∧P)Pd-
(Me)]+(OTf)− instantaneously resulted in the formation of
Pd-black. Dissociation of the chloro ligand is expected to be
limited and thus hinders insertion reactions. The diphosphine
ligand in turn may affect the ease of chloride dissociation.
Consequently, observed rates with this system only allow for
F
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
qualitative conclusion concerning the effect of different
diphosphines on the catalytic cycle.
The conversion of 1-octene increases with increasing steric
demand of the diphosphine around the metal center which is in
accordance with the results found in theoretical studies (vide
supra). Differences in selectivity are less pronounced, in all
cases the linear ester is the major product. For dtbpm, however,
a significantly larger portion of the branched esters are formed,
∼25% versus ∼10% for the C2- and C3-backbone diphosphine.
Note that isomerizing alkoxycarbonylation of methyl oleate was
also performed with these methyl-chloro complexes (Support-
ing Information). Conversions slightly increased as compared
to the respective ditriflate complexes, however, they are lower
than for 1-octene.
The diphosphine palladium(II) methyl chloro complexes
[(dtbpm)Pd(Me)Cl], [(dtbpe)Pd(Me)Cl] and [(dtbpp)Pd-
(Me)Cl] were also used as model compounds to investigate the
CO insertion into the Pd-alkyl species: an NMR tube was
charged with the respective complex in CD2Cl2 and 1−3 equiv
of 13CO were added at 25 °C. Direct NMR spectroscopic
measurements revealed that CO is inserted cleanly and
completely within 5 min into the palladium(II) methyl species,
generating the Pd-acyl species [(dtbpm)Pd(COMe)Cl],
[(dtbpe)Pd(COMe)Cl], and [(dtbpp)Pd(COMe)Cl] which
were characterized by NMR spectroscopy (Supporting
Information). Note that also with an excess of CO the
formation of a carbonyl coordinated cationic Pd-acyl species
[(P∧P)Pd(COMe)CO]+(Cl)− was not observed.
Figure 7. Isomerizing alkoxycarbonylation of isopropyl oleate in
isopropanol ( , black), n-propyl oleate in n-propanol ( , green),
■
◆
▲
●
ethyl oleate in ethanol ( , blue), and methyl oleate in methanol ( ,
red). Reaction conditions: 29.5 mmol oleate, 0.8 mol % [(dtbpx)Pd-
(OTf)2], total volume 180 mL, 20 bar CO, 90 °C.
these preparative experiments in neat alcohol as the reaction
solvent the concentration of a lower molecular weight alcohol is
always higher in a given volume,36 resulting in higher
productivity not only due to the higher reactivity of the
alcohol, but also due to its’ higher concentration. However,
correcting the rates for the different alcohol concentrations still
resulted in higher rates for methanol > ethanol > n-propanol >
isopropanol.37 Selectivities of 91
1% were similar for
The methanolysis of the Pd-acyl species was evaluated by
generating [(dtbpm)Pd(COMe)Cl], [(dtbpe)Pd(COMe)Cl],
and [(dtbpp)Pd(COMe)Cl] from the respective [(P∧P)Pd-
(Me)Cl] complexes in CD3OD by addition of 2 equiv of 13CO
and direct observation of the methanolysis reaction after CO
addition by 1H NMR spectroscopy. [(dtbpp)Pd(Me)Cl]
directly reacted to methyl acetate D3CO(13CO)Me, the
deuterido-carbonyl bridged palladium(I) complex [(dtbpp)Pd-
(μ-D)(μ-13CO)Pd(dtbpp)]+(Cl)−, and the dichloro complex
[(dtbpp)PdCl2] within 5 min at room temperature (Supporting
Information). Observation of the Pd-acyl species was not
possible in methanol solution. By contrast [(dtbpe)Pd(Me)Cl]
reacted to the observable acyl species [(dtbpe)Pd(COMe)Cl].
This acyl species reacts slowly with methanol at room
temperature to methyl acetate D3CO(13CO)Me, the deuter-
ido-carbonyl bridged palladium(I) complex [(dtbpe)Pd(μ-
D)(μ-13CO)Pd(dtbpe)]+(Cl)−, and the dichloro complex
[(dtbpe)PdCl2] (Supporting Information). [(dtbpm)Pd(Me)-
Cl] also reacted to the acyl species [(dtbpm)Pd(COMe)Cl].
This acyl species was even less reactive toward methanol than
the respective [(dtbpe)Pd(COMe)Cl] complex. Methanolysis
was only observed at an elevated temperature of 50 °C.
These experimental findings demonstrate that, both the
generation of the catalytically active Pd-hydride species is
critical, but moreover the alcoholysis reaction of Pd-acyl species
is influenced by the variable steric demand around the metal
center which is in accordance with our theoretical findings.
Influence of the Alcohol. To study the influence of the
alcohol on isomerizing alkoxycarbonylation, pressure reactor
experiments in neat methanol, ethanol, n-propanol and
isopropanol using 29.5 mmol of the respective oleate in a
total volume of 180 mL and 0.8 mol % [(dtbpx)Pd(OTf)2] as a
catalyst precursor under 20 bar CO at 90 °C were performed
(Figure 7).35 Rates of conversion decrease within the series
methanol > ethanol > n-propanol > isopropanol. Note that in
methanol, ethanol and n-propanol and thus not significantly
influenced by the alcohol.
To study the alcoholysis reaction itself, the dtbpx
coordinated Pd-acyl species [(dtbpx)Pd(COMe)Cl] was
generated in situ in an NMR tube by addition of 2 equiv of
CO to a solution of [(dtbpx)Pd(Me)Cl] in 0.5 mL of CD2Cl2.
Ten equiv of the respective alcohol (methanol, ethanol, n-
propanol and isopropanol) were added and the alcoholysis
1
reaction was followed by H NMR spectroscopy. The rate of
alcoholysis with methanol was fastest and already complete
when the NMR tube was introduced into the probe after
methanol addition. For the other alcohols the reaction was
monitored over time showing that the rate of alcoholysis
decreases within the series methanol > ethanol > n-propanol >
isopropanol (Supporting Information). Theoretical studies on
the alcoholysis reaction from the linear Pd-acyl [(dtbpx)PdC-
(O)(CH2)6COOMe]+ with methanol and isopropanol,
respectively, indicated that the energetic barrier of alcoholysis
is higher for the sterically more demanding isopropanol versus
methanol, and suggested a single molecule TS for isopropanol
versus a three molecule cluster TS for methanol (vide supra).38
Multiple Unsaturated Fatty Acids. As outlined, plant oils
inadvertently contain multiple unsaturated fatty acids in
variable amounts. Thus, their role during catalysis must be
accounted for. To this end, experiments under pressure reactor
conditions (Figure 1 and Supporting Information) comple-
mented by mechanistic studies with NMR spectroscopic
methods were performed. As previously reported not only
monounsaturated fatty acids (e.g., methyl oleate), but also
multiple unsaturated fatty acids (e.g., methyl linoleate) are
converted to the linear diesters by isomerizing alkoxycarbony-
lation (Scheme 1).5a,39 The latter is of particular interest, as
plant oils (e.g., rapeseed, sunflower, soybean or palm oil) are
mixtures of fatty acids with different number of carbon atoms
and double bonds.1b However, rates of methyl linoleate
G
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
alkoxycarbonylation are lower as compared to methyl oleate
under identical reaction conditions as confirmed by TOFs of ca.
2 h−1 for neat linoleate versus ca. 12 h−1 for oleate (vide supra).
The selectivity to the linear diester is 45.5% and thus much
lower as compared to methyl oleate where selectivities of >90%
are observed. Compared to the isomerizing alkoxycarbonylation
of methyl oleate, not only branched ester byproducts are
formed, but also a triester, formed by double alkoxycarbony-
lation of the substrate, a methoxy substituted diester, formed by
mono alkoxycarbonylation and hydromethoxylation of the
remaining double bond of the substrate, and a C18 keto
monoester product were identified by GC, NMR spectroscopy
and ESI-MS (see also ref 39). In addition, the formation of a
terminal dimethylacetal was observed by NMR spectroscopy.
For detailed analysis and possible reaction pathways leading to
the formation of these byproducts see the Supporting
Information.
To gain mechanistic insights into the origin of the lower
(versus methyl oleate) net overall rates, the catalyst precursor
[(dtbpx)Pd(OTf)2] was dissolved in CD3OD, resulting in
quantitative formation of the catalytically active deuteride
species [(dtbpx)PdD(CD3OD)]+(OTf)− within 5 min at room
temperature. Addition of stoichiometric amounts of methyl
linoleate resulted in virtually quantitative formation of a
disubstituted/internal Pd-allyl species [(dtbpx)Pd(η3-C3H3)-
{(CH2)mCOOMe}{(CH2)nCH3}]+(OTf)− (n + m = 13).40
This internal Pd-allyl species is isomerized virtually quantita-
tively to the monosubstituted/terminal Pd-allyl species
[(dtbpx)Pd(η3-C3H4){(CH2)14COOMe}]+(OTf)− already at
room temperature (Scheme 3 and Supporting Information).
starting material had disappeared completely. 13C and 2H NMR
spectroscopy reveal the formation of deuterated dimethyl
adipate, thus indicating insertion of CO, subsequent meth-
anolysis to form methyl pentenoate, reinsertion of methyl
pentenoate, isomerization, CO insertion and methanolysis to
form dimethyl adipate (Supporting Information). These
observations reflect the high thermodynamic stability of the
allyl complexes and thus account for the lower reaction rates in
isomerizing alkoxycarbonylation of methyl linoleate versus
methyl oleate. However, as noted above small amounts (ca.
2.5%) of methyl linoleate in technical grade plant oils (Figure
1) do not significantly influence the overall rate of catalysis
under pressure reactor conditions, presumably due to a certain
entropic contribution and thus a strong influence of the
temperature on the aforementioned equilibrium.
SUMMARY AND CONCLUSION
■
From these comprehensive theoretical and experimental studies
the following picture evolves: Isomerization along the substrate
hydrocarbon chain is associated with relatively low energy
barriers for both diphosphine ligands studied to this end
(Figure 2). Linear Pd-alkyls are more stable than branched
alkyls for the very crowded metal centers. For sterically less
encumbered analogues, their energies differ less. The vicinity of
an ester group of the substrate can result in a slight stabilization
of a branched agostic alkyl by an interaction of the functional
group with a β-agostic hydrogen atom (however not to the
extent that this renders the branched alkyl more stable than the
linear alkyl for the sterically crowded metal center). More
pronouncedly, coordination of the ester group itself can
stabilize certain branched alkyls. A five-membered chelate can
form (5-m.c.). For the sterically crowded metal center, a four-
membered chelate is energetically more favorable (4-m.c.). In
both cases, this chelate formed by coordination of the plant-oil
derived ester group is more stable than all other alkyls. These
theoretical results clearly show that isomerization itself is not
the major rate-determining process. In the flat energy landscape
with low barriers for interconversion, all alkyls are readily
accessible energetically and are in equilibrium with each other.
For all the above alkyls, coordination of CO and subsequent
insertion to form the respective Pd-acyls are energetically
feasible and reversible. For the dtbpx coordinated metal center
these organometallic species are in equilibrium with each other.
Notwithstanding overall reaction rates can be influenced to a
relevant extent by the portion in this equilibrium composition
of a given species, namely, the linear Pd-acyl, which is the
starting point of the preferred further rate-determining pathway
(vide infra). This portion depends on statistics resulting from
the chain length of the substrate, and (chelating) binding of a
functional group present in the substrate. All these consid-
erations are reflected in experimental monitoring of the
progress of reaction over time for different substrates (all
resulting in the linear ester as the major product): 4-octene
reacts at virtually the same rate as 1-octene. 1-Octadecene with
its larger number of methylene groups reacts slightly slower,
and the rate of conversion of methyl oleate with its additional
ester group is further slowed down.
Scheme 3. Reactivity of Stoichiometric Amounts of Methyl
Linoleate with the Catalytically Active Deuteride Species
[(dtbpx)PdD(CD3OD)]+ (n + m = 13; ΔG in kcal mol−1)
Kinetic observation of this transformation at temperatures
between 25−55 °C yielded an overall energetic barrier of
ΔG‡
= 23.1 kcal mol−1 for this isomerization (Supporting
Information).
isom
Insertion of carbon monoxide into the terminal Pd-allyl
species was investigated using [(dtbpx)Pd(η3-C4H7)]+(OTf)−
as a model substrate (for synthesis, characterization and X-ray
crystal structure analysis of [(dtbpx)Pd(η3-C4H7)]+(OTf)− cf.
the Supporting Information). At CO pressures of up to 5 bar in
a pressure NMR tube in methylene chloride as the solvent,
neither insertion nor coordination of CO was observed. Even at
temperatures of 90 °C only decomposition of the Pd-species as
indicated by the formation of Pd-black, free diphosphine and
butadiene was observed. This is in stark contrast to the facile
insertion of CO into Pd-alkyls, which is already observed at
−80 °C.14 Note that this observation does not necessarily
reflect the CO insertion reaction rates only but also reflects the
thermodynamic stability of the Pd-allyl species. When methanol
was used as a solvent instead, under otherwise identical
conditions, no reaction of the allyl was observed at room
temperature, however after 3 days at 90 °C, the signals of the
The rate-determining step associated with the highest
energetic barrier is invariably methanolysis of the Pd-acyl
formed by carbon monoxide insertion (Figures 3 and 4).
Considering methanolysis for the linear acyls, the barrier of
methanolysis is decisively lower for a sterically encumbered
metal site compared to active sites with less bulky diphosphine
H
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
ligands. This is due to a significantly higher stability of the
methanol adducts for less bulky diphosphines. For bulky
diphosphines, this ground state which precedes the rate-
determining methanolysis step is destabilized. In fact, for bulky
diphosphine ligands, its formation from the preceding direct
product of CO insertion is not associated with a disadvanta-
geously high energy gain. For less bulky ligands, the higher
stability of the methanol adduct not only results in an effective
higher barrier, but it can also render the overall methanolysis
step thermoneutral. In addition, experimental studies suggest
that an entry into a catalytic cycle may already be less efficient
for such catalyst systems. Thus, hydrides are formed
straightforwardly from dtbpx and dtbpp coordinated Pd-
complexes in methanol, whereas this is not the case for less
demanding diphosphines.
In view of the experimentally observed very high activity of
ethylene methoxycarbonylation, and desirable enhanced rates
of isomerizing methoxycarbonylation an analysis of the origin
of these different rates is instructive. Theoretical studies
confirm the transition state of methanolysis as the rate-
determining step to be significantly lower for the case of
ethylene versus a higher olefin, considering the entire reaction
sequence starting from a Pd−H species. The overall energetic
profile is lower for the case of ethylene (Figure 6). This can be
traced back to a more facile insertion and a higher energy gain
in the first metal−carbon bond forming step for ethylene. That
is, the comparatively higher reactivity of ethylene in
methoxycarbonylation appears to result from the intrinsically
more favorable generation of an agostic ethyl from Pd−H +
ethylene versus formation of a higher alkyl from Pd−H + 1-
olefin rather than being brought about by a particular
(diphosphine) ligand environment.
Another decisive consequence of the extreme steric bulk
introduced by appropriate diphosphine ligands is a destabiliza-
tion of the transition states of methanolysis for all branched
acyls versus the linear acyl. In fact, the mechanism is calculated
to change from a three molecule cluster TS in case of the linear
Pd-acyl to a single molecule TS for all branched Pd-acyls. This
accounts for the remarkable selectivity of isomerizing
alkoxycarbonylation of methyl oleate. Even for the smallest
branched alkyl with a methyl substituent on the carbonyl-
bound carbon atom, which would result in a methyl-branched
long-chain diester, the transition state for methanolysis is
already significantly higher in energy. That the pathway to the
linear product involves a cluster of three molecules of methanol
in a concerted mechanism, while a pathway involving a single
molecule of methanol is more favorable for the branched acyls
only applies to the extremely bulky diphosphine-substituted
catalysts, in other cases a “cluster-mechanism” is predicted to be
operative also for the branched acyls.
Multiple unsaturated fatty acids, as exemplified by methyl
linoleate, insert into the Pd−H species to form a Pd-allyl
complex, which isomerizes to the terminal Pd-allyl, from which
carbonylation occurs. Other than isomerization of the Pd-alkyl
species formed from oleate, this isomerization is slow and
associated with a significant energy barrier. However, the rate of
carbonylation of this Pd-allyl is even lower, such that the major
pathway is isomerization to the terminal allyl followed by
carbonylation to the linear diester product. The remaining
internal double bond is subject to further side reaction,
alkoxycarbonylation to form a triester or hydromethoxylation.
These are comparatively slow reactions, but due to the
prolonged reaction times required to methoxycarbonylate the
allyl species they can occur on the product already formed and
thus become relevant. However, the influence of small amounts
(≤2.5%) of methyl linoleate in technical grade plant oils is
neglible. In preparative diester synthesis the use of methanol as
the alcohol is desirable as its’ concentration in a given volume is
in general higher as compared to higher molecular weight
alcohols and its’ reactivity to the Pd-acyl species is higher
compared to ethanol, n-propanol and isopropanol.
In summary, the steric demand around the Pd-center
induced by the diphosphine ligand is both responsible for
catalytic selectivity and productivity in isomerizing alkoxycar-
bonylation of plant oils. Sterically congested metal centers
result in more selective catalysts as energy differences between
the pathways leading to the linear versus the branched products
of the rate-determining methanolysis steps are higher for these
systems. Moreover these systems are more productive in
general, as the energetic barrier of this rate-determining
methanolysis step is lower as compared to less encumbered
systems, which may not be active catalysts at all for
methoxycarbonylation due to an unfavorable barrier. Concern-
ing the isomerization step, our findings suggest that this is not a
unique feature of sterically bulky diphosphines like dtbpx, but
also other diphosphine Pd−H species can promote a rapid
isomerization. However, formation of the Pd−H species from
the catalyst precursor occurs much more readily for the bulky
diphosphines, which may be a decisive factor.
From this first unraveling of a successful catalytic isomer-
ization/functionalization reaction sequence, the following
general picture emerges. The rate-determining step (in this
case methanolysis) is preceded by a diverse relatively flat energy
landscape of the various reversible reaction pathways (Curtin−
Hammett kinetics). This applies to the isomerization
sequences, but also to the first reaction steps of functionaliza-
tion. Effectively, these isomerization/functionalization steps are
in mutual equilibrium vice versa. Selectivity arises from
differentiation of pathways in the final and highest barrier
step of functionalization, here by extreme steric congestion
about the active site. These features in our opinion define the
essential prerequisite in designing isomerization/functionaliza-
tion schemes with a single type of active sites, and this picture
identified may also provide inspiration for multicomponent
catalyst systems in which equilibrium landscapes may extend
over the different cooperating types of sites.
ASSOCIATED CONTENT
■
S
* Supporting Information
Experimental and computational details, absolute energies (in
Hartrees) of all molecules whose geometries were optimized,
selected NMR spectra, selected GC traces and crystallographic
details of [(dtbpm)PdCl2] (CCDC 1010350), [(dtbpe)Pd-
(OTf)2] (CCDC 1010349), [(dtbpb)Pd(OTf)2] (CCDC
1010347), [(dippp)Pd(OTf)2] (CCDC 1010346), [(depp)Pd-
(OTf)2] (CCDC 1010345), [(dtbpe)Pd(μ-H)(μ-CO)Pd-
(dtbpe)]+OTf− (CCDC 1010348) and [(dtbpx)Pd(η3-
C4H7)]+(OTf)− (CCDC 1010351). This material is available
free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
■
Corresponding Authors
lcaporaso@unisa.it
stefan.mecking@uni-konstanz.de
I
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
J. J. R.; Damian, K.; Van Rensburg, H.; Slawin, A. M. Z.; Tooze, R. P.;
Clarke, M. L. Chem.Eur. J. 2009, 15, 10504−10513. (c) Clegg, W.;
Eastham, G. R.; Elsegood, M. R. J.; Heaton, B. T.; Iggo, J. A.; Tooze, R.
P.; Whyman, R.; Zacchini, S. J. Chem. Soc., Dalton Trans. 2002, 17,
3300−3308. (d) 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−
5539.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
P.R. gratefully acknowledges support from the Carl-Zeiss-
Foundation by a graduate fellowship. We thank Dako AG for
donation of high-oleic sunflower oils.
(14) Roesle, P.; Durr, C. J.; Moller, H. M.; Cavallo, L.; Caporaso, L.;
̈
̈
Mecking, S. J. Am. Chem. Soc. 2012, 134, 17696−17703.
REFERENCES
■
(15) Christl, J. T.; Roesle, P.; Stempfle, F.; Wucher, P.; Gottker-
̈
(1) (a) Biermann, U.; Bornscheuer, U.; Meier, M. A. R.; Metzger, J.
Schnetmann, I.; Muller, G.; Mecking, S. Chem.Eur. J. 2013, 19,
̈
O.; Schafer, H. J. Angew. Chem., Int. Ed. 2011, 50, 3854−3871.
̈
17131−17140.
(b) Chikkali, S.; Mecking, S. Angew. Chem., Int. Ed. 2012, 51, 5802−
(16) TOFs were calculated from the data shown in Figure 1, from the
data points at 2 min for ethylene, at 5 h for methyl oleate and at 6 h for
methyl linoleate, respectively.
5808.
(2) (a) Quinzler, D.; Mecking, S. Angew. Chem., Int. Ed. 2010, 49,
4306−4308. (b) Stempfle, F.; Quinzler, D.; Heckler, I.; Mecking, S.
Macromolecules 2011, 44, 4159−4166. (c) Vilela, C.; Silvestre, A. J. D.;
Meier, M. A. R. Macromol. Chem. Phys. 2012, 213, 2220−2227.
(17) 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.;
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.;
(d) Mutlu, H.; Hofsaß, R.; Montenegro, R. E.; Meier, M. A. R. RSC
̈
Adv. 2013, 3, 4927−4934. (e) Stempfle, F.; Ritter, B. S.; Mulhaupt, R.;
̈
Mecking, S. Green Chem. 2014, 16, 2008−2014.
(3) Chikkali, S.; Stempfle, F.; Mecking, S. Macromol. Rapid Commun.
2012, 33, 1126−1129.
(4) (a) Picataggio, S.; Rohrer, T.; Deanda, K.; Lanning, D.; Reynolds,
R.; Mielenz, J.; Eirich, L. D. Nat. Biotechnol. 1992, 10, 894−898.
(b) Schoerken, U.; Kempers, P. Eur. J. Lipid Sci. Technol. 2009, 111,
627−645. (c) Lu, W.; Ness, J. E.; Xie, W.; Zhang, X.; Minshull, J.;
Gross, R. A. J. Am. Chem. Soc. 2010, 132, 15451−15455.
(5) (a) Stempfle, F.; Roesle, P.; Mecking, S. ACS Symposium Series
1105 Biobased Monomers, Polymers and Materials; Gross, R. A., Smith
P. B., Eds.; American Chemical Society: Washington, D.C., 2012; pp
151−163. (b) Dinger, M. B.; Mol, J. C. Adv. Synth. Catal. 2002, 344,
671−677. (c) Behr, A.; Obst, D.; Westfechtel, A. Eur. J. Lipid Sci.
Technol. 2005, 107, 213−219. (d) Ghebreyessus, K. Y.; Angelici, R. J.
Organometallics 2006, 25, 3040−3044. (e) Deuss, P. J.; Barta, K.; de
Vries, J. G. Catal. Sci. Technol. 2014, 4, 1174−1196.
̈
Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, O.;
Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J. Gaussian 09;
Gaussian, Inc.: Wallingford, CT, 2009.
(18) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652.
(19) (a) Dunning, T. H., Jr.; Hay, P. J. In Modern Theoretical
Chemistry; Schaefer, H. F., III, Ed.; Plenum: New York, 1976; pp 1−
28. (b) Hay, P. J.; Wadt, W. R. J. Chem. Phys. 1985, 82, 270−283.
(20) (a) Ditchfield, R.; Hehre, W. J.; Pople, J. A. J. Chem. Phys. 1971,
54, 724−728. (b) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem.
Phys. 1972, 56, 2257−2261. (c) Hariharan, P. C.; Pople, J. A. Mol.
Phys. 1974, 27, 209−214. (d) Gordon, M. S. Chem. Phys. Lett. 1980,
76, 163−168. (e) Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta
1973, 28, 213−222.
(6) (a) Yuki, Y.; Takahashi, K.; Tanaka, Y.; Nozaki, K. J. Am. Chem.
Soc. 2013, 135, 17393−17400. (b) van der Veen, L. A.; Kamer, P. C. J.;
van Leeuwen, P. W. N. M. Angew. Chem., Int. Ed. 1999, 38, 336−338.
(c) Selent, D.; Hess, D.; Wiese, K.-D.; Rottger, D.; Kunze, C.; Borner,
̈
̈
A. Angew. Chem., Int. Ed. 2001, 40, 1696−1698. (d) Bronger, R. P. J.;
Bermon, J. P.; Herwig, J.; Kamer, P. C. J.; van Leeuwen, P. W. N. M.
Adv. Synth. Catal. 2004, 346, 789−799. (e) Klein, H.; Jackstell, R.;
Wiese, K.-D.; Borgmann, C.; Beller, M. Angew. Chem., Int. Ed. 2001,
40, 3408−3411.
(7) (a) Pugh, R. I.; Drent, E.; Pringle, P. G. Chem. Commun. 2001,
́
1476−1477. (b) Jimenez-Rodriguez, C.; Foster, D. F.; Eastham, G. R.;
Cole-Hamilton, D. J. Chem. Commun. 2004, 1720−1721.
(21) Note that single point energy calculations with the triple-ζ plus
polarization TZVP basis set were performed on selected Pd-alkyl
species (5-β-4, 5-β-6 and 7-β-6; cf. the Supporting Information) to
validate the chemical scenario derived with the 6-31G(d) basis set.
(22) In the isomerization study calculations on cis- and trans-isomers
of the respective olefins were performed for each coordination and
transition state. The results reported herein refer to the energetically
favored trans-isomer path for both diphosphines.
(23) The coordination energy of methyl 4-heptenoate to the Pd-H
fragment to form the olefin coordinated [(diphosphine)PdH(olefin)]+
complex is ΔG = 7.6 kcal mol−1 for the dtbpx coordinated species and
ΔG = −5.5 kcal mol−1 for the dmpx coordinated species.
(24) In the X-β-Y notation, the first number (X) denotes the carbon
atom attached to the Pd-center, the second number (Y) denotes the
carbon atom whose hydrogen atom interacts with the fourth
coordination site of the Pd-center via a β-hydrogen interaction.
(25) Energies of the four- and five-membered chelate have been
reported for both the dtbpx and dmpx coordinated diphosphine
previously by us. The energy of the linear Pd-Alkyl 7-β-6 of the dtbpx
coordinated species has been reported. CO insertion and methanolysis
has been reported for the four-membered (4-m.c.) and the linear 7-β-
6 dtbpx coordinated species (cf. ref 14).
(8) (a) Clegg, W.; Elsegood, M. R. J.; Eastham, G. R.; Tooze, R. P.;
Wang, X. L.; Whiston, K. Chem. Commun. 1999, 1877−1878.
(b) 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, 1832−1840. (c) Carr, N.; Dunne, B. J.; Orpen, A. G.;
Spencer, J. L. J. Chem. Soc., Chem. Commun. 1988, 926−928.
́
(9) (a) Jimenez-Rodriguez, C.; Eastham, G. R.; Cole-Hamilton, D. J.
Inorg. Chem. Commun. 2005, 8, 878−881. (b) Furst, M. R. L.; Le Goff,
R.; Quinzler, D.; Mecking, S.; Botting, C. H.; Cole-Hamilton, D. J.
Green Chem. 2012, 14, 472−477.
(10) Vilches-Herrera, M.; Domke, L.; Borner, A. ACS Catal. 2014, 4,
1706−1724.
(11) (a) van Leeuwen, P. W. N. M. Homogeneous Catalysis:
Understanding the Art; Kluwer Academic Publishers: Dordrecht,
2004. (b) van Leeuwen, P. W. N. M.; Chadwick, J. C. Homogeneous
Catalysts: Activity−Stability−Deactivation; Wiley-VCH Verlag GmbH
& Co. KGaA: Weinheim, 2011.
(12) (a) McCoy, M.; Tremblay, J.-F. Chem. Eng. News 2009, 87 (33),
9. (b) Harris, B. Ingenia 2010, No. 45, 18−23.
(13) (a) Eastham, G. R.; Heaton, B. T.; Iggo, J. A.; Tooze, R. P.;
Whyman, R.; Zacchini, S. Chem. Commun. 2000, 609−610. (b) Frew,
̈
(26) The methanolysis reaction of the linear and the five-membered
chelate dtbpx coordinated Pd-acyl species has been reported
previously (cf. ref 14).
(27) ΔΔG‡ = 3.0 kcal mol−1 refers to the three methanol coordinated
TSs. For the single methanol coordinated TSs ΔΔG‡ = 0.7 kcal mol−1
J
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX

Journal of the American Chemical Society
Article
is calculated, thus also indicating higher reaction rates for the more
crowded diphosphine.
(28) TOFs were calculated from the data shown in Figure 1, from the
data points at 2 min for ethylene and at 1 h for 1-octene, 4-octene and
1-octadecene, respectively.
(29) The opening angle describes a cone at the metal center, which is
defined by the space left open by the diphosphine not occupied by any
of its atoms. It amounts to twice the half-cone angle as defined in ref
15.
(30) Note that the dmpx ligand used for the DFT calculations has an
opening angle of 219.5° (calculated from the quarternary carbon
atoms of [(dtbpx)Pd(OTf)2]) and is thus similarly congesting as
[(dtbpm)Pd(OTf)2].
(31) Formation of hydride species is indicated in 1H NMR
spectroscopy by signals in the range of 0−30 ppm and in 31P NMR
spectroscopy by appearance of pairs of doublets indicating unsym-
metrical substituted square planar Pd-centers.
(32) Hydrogenation may occur via Pd-catalyzed transfer hydro-
genation with methanol as the hydrogen source.
(33) The respective complex was dissolved in methanol-d4 or
CD2Cl2/MeOH respectively, and 6 equiv of CO were added via
syringe into the NMR tube. Afterwards 20 equiv of methyl oleate were
added.
(34) Note that under preparative conditions, the protonated
diphosphine was also observed after a carbonylation experiment.
(35) Quantitative conversion of [(dtbpx)Pd(OTf)2] to the respective
[(dtbpx)PdH(R-OH)]+ species and rapid isomerization of the
respective oleates with these Pd-H species was evidenced by NMR
spectroscopy (cf. Supporting Information).
(36) In a total volume of 180 mL concentrations are 0.164 M oleate,
23.3 M methanol, 16.1 M ethanol, 12.5 M n-propanol, 12.2 M
isopropanol.
(37) Rate constants assuming first order reactions in oleate are:
kMeOH = 3.7 × 10−5 s−1; kEtOH = 2.0 × 10−5 s−1; kn‑PrOH = 1.2 × 10−5
s−1; kisoPrOH = 7.1 × 10−7 s−1.
(38) For the three molecule cluster TSs ΔG‡methanol = 29.1 kcal mol−1
and ΔG‡
= 48.4 kcal mol−1 were calculated. For the single
isopropanol
molecule TSs ΔG‡
= 34.5 kcal mol−1 and ΔG‡
= 38.2
methanol
isopropanol
kcal mol−1 were calculated.
(39) Furst, M. R. L.; Seidensticker, T.; Cole-Hamilton, D. J. Green
Chem. 2013, 15, 1218−1225.
(40) This differs from the behavior of monounsaturated olefins (e.g.,
methyl oleate) where the equilibrium between Pd-D + olefin and Pd-
alkyl is on the side of the Pd-D species under these conditions.
K
dx.doi.org/10.1021/ja508447d | J. Am. Chem. Soc. XXXX, XXX, XXX−XXX