Palladium-Catalyzed Cross-Dehydrogenative Coupling of o-Xylene: Evidence of a New Rate-Limiting Step in the Search for Industrially Relevant Conditions

Article abstract of DOI:10.1002/cctc.201701973

An efficient cross-dehydrogenative coupling of o-xylene under neat conditions, which brings important industrial benefits towards the synthesis of a monomer used in polyimide resins, is reported. The catalyst based on the combination of Pd/N ligand/carboxylate=1:1:2 does not require a Cu cocatalyst and proceeds at 11 bar of O₂ pressure. Evaluation of the deuterium kinetic isotope effect (KIE) provides evidence for three different rate-determining steps, which depend on the reaction conditions (medium, temperature). Under the reported neat conditions, the dissociation of a carboxylate-bridged dimer to generate a more reactive monometallic Pd species is proposed to be the rate-limiting step.

Full text of DOI:10.1002/cctc.201701973

DOI: 10.1002/cctc.201701973
Full Papers
Palladium-Catalyzed Cross-Dehydrogenative Coupling of
o-Xylene: Evidence of a New Rate-Limiting Step in the
Search for Industrially Relevant Conditions
Yolanda ꢀlvarez-Casao,[a] Christian A. M. R. van Slagmaat,[b] Gerard K. M. Verzijl,[b]
Laurent Lefort,[b] Paul L. Alsters,*[b] and M. ꢀngeles Fernꢁndez-IbꢁÇez*[a]
An efficient cross-dehydrogenative coupling of o-xylene under
neat conditions, which brings important industrial benefits to-
wards the synthesis of a monomer used in polyimide resins, is
reported. The catalyst based on the combination of Pd/N
ligand/carboxylate=1:1:2 does not require a Cu cocatalyst and
proceeds at 11 bar of O2 pressure. Evaluation of the deuterium
kinetic isotope effect (KIE) provides evidence for three different
rate-determining steps, which depend on the reaction condi-
tions (medium, temperature). Under the reported neat condi-
tions, the dissociation of a carboxylate-bridged dimer to gener-
ate a more reactive monometallic Pd species is proposed to be
the rate-limiting step.
Introduction
The biaryl unit is an important structural motif in various prod-
ucts of high commercial importance.[1] Although Suzuki cou-
pling is used widely to access high-end, small-scale biaryls,
such as pharmaceutical intermediates,[2] efforts have been
made to develop alternative catalytic strategies for biaryl syn-
thesis to meet the more stringent cost and waste requirements
associated with large-scale chemicals, such as monomers. A
major step forward is the “direct arylation” approach, which
bypasses the need for boronic acids as a nucleophilic coupling
partner. Additionally, the cross-dehydrogenative coupling
(CDC) of arenes eliminates the need for electrophilic ArX cou-
pling partners.[3] Provided that an economically and ecological-
ly feasible terminal oxidant such as O2 is used, in principle,
CDC meets all the requirements for the low-cost, bulk-scale
manufacture of biaryls.[4]
Route B). Early attempts to achieve a sufficiently selective CDC
of o-xylene were not successful because, in contrast to phtha-
late, o-xylene suffers from competing, selectivity-eroding ben-
zylic oxidation during Pd/Cu-catalyzed aerobic CDC
(Scheme 2a).[7,8]
A significant step forward in the development of aerobic
CDC of o-xylene was achieved by Stahl et al. who found that
the regio- and chemoselectivity increase to comparable values
obtained with phthalate by the use of 2 equivalents of 2-fluo-
ropyridine as a ligand (Scheme 2b) instead of phenanthroline,
which is employed in the current industrial process.[9,10] In con-
trast to the neat conditions employed industrially for the CDC
of phthalate, 2-fluoropyridine-based catalysis is performed in
acetic acid as the solvent. Industrial applicability would benefit
significantly from aerobic catalysis that works efficiently under
solvent-free conditions. The use of neat o-xylene as the reac-
tion medium instead of acetic acid not only simplifies down-
stream processing and lowers the corrosivity, and thereby, also
the capital investment, but it also leads to a sufficient space–
time yield. As the CDC of arenes to yield biaryls is stopped at a
low conversion to suppress the over-arylation of the product,
space–time yields in the presence of a solvent tend to be
below an economically feasible level.
The Pd/Cu-catalyzed aerobic CDC of dimethyl phthalate
presents a case in point practiced industrially.[5] The target
product, that is, tetramethyl (1,1’-biphenyl)-3,3’,4,4’-tetracar-
boxylate (1), is an intermediate for 4,4’-biphthalic anhydride
(3), which serves as a monomer in the production of a high-
performance polyimide (Upilexꢂ; Scheme 1, Route A).[6] This
polymer would benefit from a significant decrease of cost if
the anhydride could be formed from 3,3’,4,4’-tetramethyl-1,1’-
biphenyl (2) generated by the CDC of o-xylene (Scheme 1,
The results of our search for an aerobic CDC catalyst system
in neat o-xylene have been published in patents filed in 2015
and 2016.[11,12] An important finding disclosed in the first
patent is the use of a palladium carboxylate catalyst in the
presence of 1 equivalent of an N-donor ligand, such as pyri-
dine. At a Pd/N ratio of 1, no Cu cocatalyst is required under
neat conditions. To ensure catalyst robustness, one should
avoid the use of carboxylate anions or N ligands that can un-
dergo cyclopalladation. Other salient features include the con-
trol of the activity and regio- and chemoselectivity by the
proper design of the electronics and sterics of the anion and
ligand, and the dependence of the optimum carboxylate on
[a] Dr. Y. ꢀlvarez-Casao, Prof. M. ꢀ. Fernꢁndez-IbꢁÇez
Van‘t Hoff Institute for Molecular Sciences
University of Amsterdam
Science Park 904, 1098 XH, Amsterdam (The Netherlands)
E-mail: m.a.fernandezibanez@uva.nl
[b] C. A. M. R. van Slagmaat, G. K. M. Verzijl, Dr. L. Lefort, Dr. P. L. Alsters
InnoSyn B.V.
Urmonderbaan 22, 6167RD, Geleen (The Netherlands)
E-mail: paul.alsters@innosyn.com
Supporting information and the ORCID identification number(s) for the
author(s) of this article can be found under https://doi.org/10.1002/
cctc.201701973.
ChemCatChem 2018, 10, 1 – 8
1
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
Scheme 1. Current and proposed alternative routes for the industrial synthesis of 4,4’-biphthalic anhydride.
Results and Discussion
Considering the promising results reported with 2-fluoropyri-
dine as a ligand in acetic acid as the solvent,[9a] we started by
exploring Pd catalysts based on 2-fluoropyridine under neat
conditions (Table 1). If we used the optimum catalyst composi-
tion reported for the aerobic CDC of o-xylene in acetic acid as
the reaction medium, that is, 0.10 mol% Pd(OAc)2, 0.20 mol%
2-fluoropyridine, 0.12 mol% CF3CO2H, and 0.1 mol% Cu(OTf)2
under neat conditions (17 h, 1 bar O2, 808C), the regio- and
chemoselectivity were much lower than those reported if the
reaction was performed in acetic acid. In addition, the yield de-
creased to a very low value (Table 1, entry 1 vs. 2). The omis-
sion of the Cu(OTf)2 cocatalyst improved the selectivities slight-
ly without affecting the yield (Table 1, entry 3). The yields and
selectivities were restored to significant values by increasing
the temperature (958C), O2 pressure (11 bar), Pd(OAc)2 loading
(0.5 mol%), and amount of CF3CO2H (1.0 mol%; Table 1,
entry 4). The selectivities eroded significantly if the ligand was
omitted under these conditions (Table 1, entry 5). However, a
decrease of the amount of ligand from 1.0 to 0.5 mol% (Pd/
N=1:1) boosted the yield significantly without compromising
the selectivities (Table 1, entry 6).
After we had identified conditions that led to useful yields
and selectivities under neat conditions, we embarked on an ex-
tensive screening of pyridines to further optimize the catalyst
performance (Figure 1). These ligands were screened under
both Pd/N=1:1 and 1:2 conditions, and the yields are shown
as dark gray (Pd/N=1:2) and light gray (Pd/N=1:1) bars. For
the catalyst with a Pd/N ratio of 1, selectivity ratios are dis-
played in this graph as orange markers. To emphasize per-
formance differences between the various ligands, selectivities
are expressed as the more sensitive ratio of the desired prod-
uct(s) to the undesired products rather than as a percentage
of the total desired plus undesired products. Regioselectivities
are expressed as the “regioratio”, that is, yield(2)/yield(BPu), in
which BPu is the sum of the undesired regiomeric 2,3,3’,4’/
2,2’,3,3’-tetramethyl-1,1’-biphenyl byproducts. Chemoselectivi-
ties are expressed as the “chemoratio”, that is, yield(2+BPu)/
Scheme 2. Developed catalyst for the CDC of arenes towards the synthesis
of 4,4’-biphthalic anhydride.
the O2 pressure. In this article, we report additional results on
the CDC of neat o-xylene with Pd/N=1:1 catalysts (Scheme 2c)
with a particular focus on a comparison with Pd/N=1:2 cataly-
sis and neat versus solvent conditions. Evaluation of the deute-
rium kinetic isotope effect (KIE) with the use of [D10]o-xylene
provided evidence for three different rate-determining steps,
which depends on the catalyst and reaction conditions.
&
ChemCatChem 2018, 10, 1 – 8
www.chemcatchem.org
2
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 1. Influence of the reaction conditions in the CDC coupling of o-xylene.
Entry
Remarks
ArÀAr yield [%][a,b]
Regioselectivity [%][b,c]
Chemoselectivity [%][b,d]
1
2
3
AcOH
No solvent
No solvent/No Cu
8.0
0.9
0.9
88
67
70
94
74
76
4
5
6
No remarks
No ligands
Ligand/Pd ratio=1
(0.5 mol%)
12.6
11.9
17.6
89
62
83
83
46
79
[a] Collective biaryl yield. [b] Determined by using GC, internal standard=n-dodecane. [c] Regioselectivity is the percentage of dimer 2 relative to all
dimers. [d] Chemoselectivity is the percentage of all dimer isomers relative to all products formed (dimers, trimers, benzylic oxidation…).
yield(Balc+Bald+Bza+trimers), in which Balc is 2-methylbenzyl
*
Electron-poor pyridines that can be expected to coordinate
alcohol, Bald is 2-methylbenzaldehyde, Bza is 2-methylbenzoic
less strongly induce significant yields to biaryls under Pd/
acid, and trimers are all of the isomers with a molecular weight
N=1:2 conditions (Figure 1, L2, L6, L8, L11, L12, L17), espe-
that corresponds to hexamethylterphenyl. Below we summa-
cially in the case of strong electron-withdrawing ortho sub-
rize some notable findings related to this screening:
stituents (Figure 1, L5, L7, L9, L10, L13, L15).
*
Those pyridines for which the collective biaryl yield under
*
Under Pd/N=1:2 conditions, pyridines that can be expected
Pd/N=1:2 conditions almost equals that under 1:1 condi-
to coordinate well to PdII (i.e., those that are sufficiently elec-
tions, tend to show low regio- and chemoratios (Figure 1, L5,
tron-rich and devoid of steric hindrance near the N atom)
L9, L10, L13).
fail to generate active catalysts (Figure 1, L1, L3, L4, L14,
L16).
Figure 1. Screening of pyridine ligands.
ChemCatChem 2018, 10, 1 – 8
www.chemcatchem.org
3
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
peratures. The regioselectivity was fairly constant over this
temperature range, but the chemoselectivity tended to de-
crease above 1158C. We judged this temperature to offer the
best compromise between optimum activity and selectivity re-
quirements.
*
Under Pd/N=1:1 conditions, the electronic effect of chang-
ing the meta or para substituents on the pyridine ligand is
relatively small. In the absence of ortho substituents, the ac-
tivity benefits from the presence of electron-withdrawing
substituents (Figure 1, L2, L17).
At 1158C/11 bar O2 and with a Pd(OAc)2 loading of 0.5 mol%
as in the above experiments, we then varied the amount of
CF3CO2H relative to added Pd(OAc)2 from 0 to 50 equivalents
(Table S6). In line with our earlier findings,[11] at this high O2
pressure, the collective biaryl yield benefits from the addition
of up to 2 equivalents of a strong carboxylic acid such as
CF3CO2H. More CF3CO2H decreases both the yield and regiose-
lectivity without affecting the chemoselectivity significantly.
With 2 equivalents of CF3CO2H, the collective biaryl yield
reached 26% with a regio- and chemoratio of 5.7 and 6.1, re-
spectively.[14]
*
*
Pyridines devoid of ortho substituents tend to show higher
regioratios and increased tendencies towards benzylic oxy-
genation (see Supporting Information for further details;
Figure 1, L1–L4, L17).[13]
In some cases, the reaction is very sensitive to minor
changes of the ligand. For example, the replacement of one
F atom by a H atom from 2-(trifluoromethyl)pyridine (L5) to
2-(difluoromethyl)pyridine (L6) leads to a threefold decrease
in the trimer formation in agreement with the fact that less
electron-withdrawing ligands form less active catalysts that
are not so prone to over-arylation (see Supporting Informa-
tion for further details).
To gain insight into the nature of the rate-determining step,
the order of the reaction on the Pd and deuterium KIEs were
examined. The order of the reaction was determined by the in-
itial rate method under the optimized Pd/N=1:1 neat condi-
tions described above. The kinetic data indicate a first-order
dependence on [Pd] (see Supporting Information).
The two horizontal gray and orange arrows on the graph
correspond to the minimum criteria to be met by a pyridine
ligand to be taken further in subsequent optimizations: collec-
tive biaryl yield >15%, regio- and chemoratio >4. Pyridines
that lead to a collective biaryl yield >15% and at least one of
the ratios >4 are those devoid of too much steric hindrance
near the N atom and of a too-strong-electron-withdrawing
substitution pattern although it is still more electron-poor than
pyridine. These are 4-(trifluoromethyl)pyridine (L2), 2-(difluoro-
methyl)pyridine (L6), 2-fluoropyridine (L7), and 3,5-bis(trifluoro-
methyl)pyridine (L17). Both the regio- and chemoratio exceed
4 only with 4-(trifluoromethyl)pyridine and 2-(difluoromethyl)-
pyridine. The latter ligand was used in additional experiments
because of the somewhat higher regioratio.
Deuterium KIEs were determined under various conditions
by measuring the conversion obtained at low reaction times
with o-xylene and [D10]o-xylene independently (Table 2).[15]
Under the optimized Pd/N=1:1 neat conditions described
above for 2-(difluoromethyl)pyridine as the ligand (0.5 mol%
Pd(OAc)2, 0.5 mol% 2-(difluoromethyl)pyridine, 1.0 mol%
CF3CO2H, 17 h, 11 bar O2, 1158C), a KIE of 1.6 was obtained for
the CDC reaction (Table 2, entry 1). If the catalyst loading was
decreased to 0.1 mol%, the KIE vanished (kH/kD =1.0; Table 2,
entry 2). After the restoration of the catalyst loading to
0.5 mol%, we measured the KIEs in the presence of three sol-
vents (50 vol%). In acetic acid, the KIE increased significantly
(kH/kD =10–11; Table 2, entry 3). If we replaced acetic acid by
non-protic but more polar propylene carbonate, the KIE de-
creased (kH/kD =2.0; Table 2, entry 4). In non-protic, nonpolar
hexafluorobenzene, a very low KIE (kH/kD =1.3; Table 2, entry 5)
close to that under neat conditions (Table 2, entries 1 and 2)
was obtained.
If we apply otherwise the same Pd/N=1:1 conditions as for
the ligand screening, a temperature optimization from 85–
1508C was performed for the CDC of o-xylene under the influ-
ence of the 2-(difluoromethyl)pyridine-based catalyst
(Table S4). The collective biaryl yield increased from 13% at
858C to 29% at 1258C and decreased to 15% upon a further
increase of the temperature to 1508C. This may be indicative
of catalyst deactivation by Pd black formation at too high tem-
Table 2. Results of the KIE evaluations.[a]
Entry
Ligand[b]
Equivalents
of ligand[c]
Pd(OAc)2
[mol%]
Equivalents
of CF3CO2H[c]
Solvent[d]
(50 vol%)
T
[8C]
PO
[bar]
Equipment[e]
KIE
2
1
2
3
4
5
6
7
8
9
F2
F2
F2
F2
1
1
1
1
1
2
2
2
2
2
0.5
0.1
0.5
0.5
0.5
0.1
0.1
0.1
0.1
0.1
2
2
2
2
No
No
AA
PC
F6
AA
AA
AA
AA
AA
115
115
115
115
115
80
80
80
115
115
11
11
11
11
11
1
A
A
A
A
A
B
A
A
A
A
1.6
1.0
10/11[f]
2.0
1.3
20
22
20/17[f]
12/14[f]
15/12[f]
F2
2
2F
2F
2F
2F
2F
1.2
1.2
1.2
1.2
1.2
1
11
11
1
10
[a] A variable in a row is of bold type if it has been changed compared to the preceding row. [b] F2 =2-(difluoromethyl)pyridine; 2F=2-fluoropyridine.
[c] Relative to Pd. [d] AA=acetic acid; F6=hexafluorobenzene; PC=propylene carbonate. [e] A=autoclave; B=balloon. [f] Duplicate experiment.
&
ChemCatChem 2018, 10, 1 – 8
www.chemcatchem.org
4
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
The foregoing KIE values with 2-(difluoromethyl)pyridine
under Pd/N=1:1 conditions at 1158C/11 bar O2 differ markedly
from those obtained by Stahl et al. with 2-fluoropyridine under
Pd/N=1:2 conditions in acetic acid at 808C/1 bar O2.[9b] An un-
usually large KIE range (kH/kD =18–25) was measured for the
latter system if it was applied at various Pd catalyst concentra-
tions. These very high KIE values were ascribed to rate-limiting
bimetallic transmetallation, which was supported by additional
kinetic measurements that showed a square dependence on
[Pd]. In addition to employing 2-fluoropyridine at a Pd/N=1:2
ratio and 808C/1 bar O2, the KIE study by Stahl on this catalyst
system also differed from our study with 2-(difluoromethyl)pyr-
idine in that the Pd/CF3CO2H ratio was approximately 1:1 in-
stead of 1:2. We decided to extend our KIE evaluations to this
2-fluoropyridine-based catalyst system, and we focused on the
study of its behavior at an elevated temperature and/or pres-
sure compared to the earlier study. From this, we hoped to
learn the cause of the near-absence of KIEs under our condi-
tions without a protic and/or polar solvent (Table 2, entries 1,
2, and 5) as compared to the very high KIE values obtained by
Stahl et al.[9b] In particular, we aimed to understand whether
the large differences in KIEs were due to the use of different
catalysts or to the change in reaction conditions.
the simultaneous presence of a vacant site (in the form of a
weakly bound functionality) for the incoming arene and a non-
bridged carboxylate that accepts the proton liberated upon
arene palladation.[19,20]
The generation of a monometallic species by the breaking
of the carboxylate bridges is facilitated by the use of coordi-
nating solvents (e.g., propylene carbonate or acetic acid),
which induce a switch to other kinetic regimes in which the
dimer dissociation is no longer rate-determining. In the case of
non-protic but polar propylene carbonate (Table 2, entry 4),
the KIE value is at the lower limit of the range observed previ-
ously for Pd-based oxidative arene coupling reactions (kH/kD =
2–25),[9b,21] which suggests rate-limiting CÀH activation (kH/kD =
2–5).[21c] For protic acetic acid (Table 2, entry 3), the KIE value
approaches the lower limit of the range expected for rate-limit-
ing bimetallic transmetallation (kH/kD =9–25).[9b]
We sought to get further evidence for the presence of di-
meric Pd species by adding various (soluble) tetrabutylammo-
nium salts, that is, the tetrafluoroborate, trifluoroacetate, and
acetate salts (Table 3). At least for the two carboxylates, it may
be anticipated that these have sufficient coordinative strength
to induce the (partial) breaking of a trifluoroacetate-bridged
dimer into two monometallic tricarboxylate palladate species
(in which the remaining coordination site is occupied by the
pyridine ligand; Scheme 3).
First, we determined the KIE values under the conditions as
employed earlier by Stahl et al.[9b] If we used either a glass
vessel connected to a balloon filled with O2 (Table 2, entry 6)
or an autoclave operated at 1 bar O2 (Table 2, entry 7) we ob-
tained very high KIE values that are fully in accordance with
those reported by Stahl et al.[9b] These high values were re-
tained in experiments performed in duplicate, but at 11 bar O2
instead of 1 bar O2 (Table 2, entry 8). If the temperature was in-
creased to the value employed in our KIE evaluations with the
2-(difluoromethyl)pyridine-based catalyst system, the KIE de-
creased markedly to a very similar value to that obtained with
this ligand in acetic acid medium (Table 2, entry 9 vs. 3). This
suggests that at least if we use acetic acid as the solvent, the
KIE is determined largely by the temperature of the reaction,
irrespective of which of the two catalyst systems is used.[16]
However, in acetic acid, the O2 pressure does not have a signif-
icant effect on the KIE (Table 2, entries 7 and 9 vs. 8 and 10).
The (near) absence of a KIE (Table 2, entries 1, 2, and 5) and
first-order dependency on Pd under neat or apolar C6F6 solvent
conditions may be explained by the rate-limiting dissociation
of a carboxylate-bridged dimer under such conditions (see
Supporting information for a full kinetic analysis that demon-
strates that such a rate-determining step indeed leads to a
first-order reaction in Pd). These apolar media disfavor the dis-
sociation of dimeric species into two monometallic species by
the solvent-assisted dissociation of carboxylate bridges, as
demonstrated for dimeric [Pd(RCO2)2PMe2Ph]2 tertiary phos-
phine complexes.[17] In the absence of solvolytic assistance, we
propose the formation of two monometallic Pd species that
contain a k2-(O,O)-bound carboxylate.[18] Arene CÀH activation
only takes place after the formation of a sufficiently reactive,
monometallic PdII species with a weakly bound solvent mole-
cule or k2-(O,O) carboxylate. This is in line with a concerted
metalation–deprotonation (CMD) mechanism as this requires
Compared to the monometallic species that are obtained by
bridge cleavage by (neutral) solvent molecules, such negatively
charged monometallic palladate species can be expected to
be much less electrophilic, and therefore, much less active in a
subsequent arene palladation.
Table 3. Effect of tetrabutylammonium salts addition.[a]
Entry
Pd source[b]
Equivalents
of CF3CO2H[c]
[Bu4N]X[d]
ArÀAr
yield [%][e]
1
2
3
4
5
6
7
8
TFA
TFA
TFA
TFA
Acetate
Acetate
Acetate
Acetate
–
–
–
–
2
2
2
2
–
18.9
8.2
1.3
0.3
22.0
7.2
À
X=BF4
X=TFAÀ
X=AcOÀ
–
À
X=BF4
X=TFAÀ
X=AcOÀ
4.4
1.5
[a] Conditions: Pd source (0.5 mol%), 2-(difluoromethyl)pyridine
(0.5 mol%), [Bu4N]X (1.0 mol%), and optional CF3CO2H (1.0 mol%) in o-
xylene (1 mL) were stirred for 17 h at 1158C/11 bar O2. [b] TFA=
À
Pd(CF3CO2)2; Acetate=Pd(CH3CO2)2. [c] Relative to Pd. [d] TFAÀ =CF3CO2
.
[e] Collective biaryl yield.
Scheme 3. Dissociation of the dimeric, carboxylate-bridged Pd complex in-
duced by tetrabutylammonium salts.
ChemCatChem 2018, 10, 1 – 8
www.chemcatchem.org
5
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
The effect of additional tetrabutylammonium salts was
tested on 2-(difluoromethyl)pyridine-based catalyst systems
under Pd/N=1:1 conditions with either Pd(CF3CO2)2 or
Pd(CH3CO2)2 as the Pd source, the latter in the presence of
2 equivalents CF3CO2H to allow the in situ generation of
Pd(CF3CO2)2. The addition of 2 equivalents of [Bu4N]BF4 to
these systems led to a significant reduction of the collective
biaryl yield after 17 h compared to the blank reactions without
this additive (Table 3, entries 1 and 5 vs. 2 and 6). These results
demonstrate that even an additive such as [Bu4N]BF4, which
has ionic constituents that are both typically regarded to be
noncoordinating towards PdII, exerts a deactivating effect on
the CDC process, at least under the solvent-free conditions
tested. A significant further reduction of the collective biaryl
yield resulted from the addition of [Bu4N]CF3CO2 (Table 3, en-
tries 3 and 7) and in particular [Bu4N]CH3CO2 (Table 3, entries 4
and 8). Clearly, the collective biaryl yield follows the coordina-
out a KIE (kH/kD ꢀ1). The dissociation of a carboxylate-bridged
dimer to generate a much more reactive monometallic Pd spe-
cies is proposed as a new rate-limiting arene CDC step to ex-
plain the near-absence of a KIE and other experimental obser-
vations. The intermediate KIE value (kH/kD =2) observed with a
Pd/N=1:1 catalyst system if propylene carbonate was used as
the solvent is in line with yet another kinetic regime, that is,
rate-limiting CÀH activation.[9b,21c] Thus, the present study un-
derlines the impact of the reaction conditions (medium, tem-
perature) besides the catalyst system on the kinetic regime. Fi-
nally, although the CDC of dimethyl phthalate is estimated to
be operated on a kiloton scale,[22] the process is performed
with remarkably low Pd turnover numbers.[6e–g] The key to ach-
ieve sufficient process economics is Pd recovery and recy-
cling.[4c,23] This needs to be addressed in subsequent process
research together with the development of an efficient proto-
col for the aerobic oxidation of 2 to 3.
À
À
tive strength of the anion inversely (CH3CO2 > CF3CO2
>
À
BF4 ), in line with the hypothesis that the anion-induced disso-
ciation of a carboxylate-bridged dimer is accompanied by re-
duced catalyst activity because of the build-up of less-reactive
palladate species. This deactivating effect of the addition of co-
ordinating anions (as their tetrabutylammonium salts) is very
similar to the deactivating effect of using twice the amount of
a well-coordinating pyridine, which also results in the forma-
tion of well-defined, but unreactive monometallic rather than
dimeric Pd species.[18]
Experimental Section
Note on safety: although we performed thousands of experiments
at elevated temperatures and O2 pressures (up to 11 bar) without
any explosion, a detailed safety analysis should be performed
before such hazardous conditions are applied. Experiments should
only be performed in proper equipment that can withstand the
pressure increase in case of an unintended explosion.
In a disposable culture tube, palladium acetate (0.041 mmol), 2-(di-
fluoromethyl)pyridine
(0.041 mmol),
trifluoroacetic
acid
(0.082 mmol), and o-xylene (1 mL) were combined. The reaction
tube was placed in a seven-well aluminum block, which was intro-
duced into the autoclave. The autoclave was purged with O2 for
2 min, and then the pressure was increased to 11 bar of O2. The au-
toclave was placed inside a preheated aluminum block, and the re-
action was stirred at 1158C for 17 h. After that time, the autoclave
was removed from the aluminum block and allowed to reach RT
before the pressure was released. n-Dodecane was added to the
reaction mixture as internal standard, and the mixture was filtered
through a Celite plug and rinsed with AcOEt. Samples were evalu-
ated by using GC.
Conclusions
The use of Pd/N=1:1 rather than Pd/N=1:2 conditions ex-
pands the range of pyridines that act as efficient ligands for
the palladium carboxylate catalyzed cross-dehydrogenative
coupling (CDC) of o-xylene without a Cu cocatalyst under neat
conditions. Both features, in particular the use of solvent-free
conditions, lead to important industrial benefits. We have also
formulated some trends with regard to the control of the ac-
tivity, regioselectivity, and chemoselectivity as a function of
electronic and steric features of the pyridine ligand. Preferred
pyridines, such as 2-(difluoromethyl)pyridine, lack too much
steric hindrance near the N atom and a too-strong-electron-
withdrawing substitution pattern. The optimum choice of the
(in situ generated) palladium carboxylate depends on the O2
pressure. High O2 pressures allow the use of 2 equivalents of
trifluoroacetic acid relative to Pd to increase the electrophilicity
of the PdII center. At lower O2 pressures, strongly electrophilic
PdII is not tolerated because reductive catalyst deactivation by
Pd black formation then outcompetes the aerobic reoxidation
of Pd0 to PdII. Evaluation of the kinetic isotope effects (KIEs) re-
vealed that the kinetics of the Pd/N=1:1 catalyst system under
neat conditions are different from those of neat arene systems
with acetylacetone as the ligand (kH/kD ꢀ2–5) or those of Pd/
N=1:2 systems in the presence of acetic acid as solvent stud-
ied previously. Although for the latter bimetallic transmetalla-
tion was proposed to be the rate-limiting step to explain the
very high KIE values (kH/kD =9–25),[9b] the present solvent-free
Pd/N=1:1 system is characterized by a rate-limiting step with-
Acknowledgements
We acknowledge financial support from NWO through a LIFT
grant (731.015.409). Furthermore, we thank Wouter Veldmate for
his valuable experimental contributions during his traineeship
within DSM before the foundation of InnoSyn. Finally, we thank
DSM for the permission to publish parts of this work.
Conflict of interest
The authors declare no conflict of interest.
Keywords: biaryls · CÀH activation · cross-dehydrogenative
coupling · industrial chemistry · palladium
&
ChemCatChem 2018, 10, 1 – 8
www.chemcatchem.org
6
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
[1] a) J. Hassan, M. Sevignon, C. Gozzi, E. Schulz, M. Lemaire, Chem. Rev.
2002, 102, 1359–1470; b) D. A. Horton, G. T. Bourne, M. L. Smythe,
Chem. Rev. 2003, 103, 893–930.
[2] For a selected review on the use of cross-coupling reactions in chemical
industry, see: C. Torborg, M. Beller, Adv. Synth. Catal. 2009, 351, 3027–
3043.
[3] For reviews on oxidative biaryl coupling, see: a) J. A. Ashenhurst, Chem.
Soc. Rev. 2010, 39, 540–547; b) C. S. Yeung, V. M. Dong, Chem. Rev.
2011, 111, 1215–1292; c) C. Liu, H. Zhang, W. Shi, A. Lei, Chem. Rev.
2011, 111, 1780–1824; d) Y. Yang, J. Lan, J. You, Chem. Rev. 2017, 117,
8787–8863.
[4] a) B.-Q. Xu, D. Sood, A. V. Iretskii, M. G. White, J. Catal. 1999, 187, 358–
366; b) A. V. Iretskii, S. C. Sherman, M. G. White, J. C. Kenvin, D. A. Schir-
aldi, J. Catal. 2000, 193, 49–57; c) S. C. Sherman, A. V. Iretskii, M. G.
White, D. A. Schiraldi, Chem. Innovation 2000, 30, 25–30; d) D. A. Schir-
aldi, S. C. Sherman, O. S. Sood, M. G. White (Arteva Technologies), Euro-
pean Patent 970745, 2000.
[5] For a recent book on liquid-phase aerobic oxidation catalysis that in-
cludes chapters on Pd- and Cu-catalyzed reactions, see: Liquid Phase
Aerobic Oxidation Catalysis: Industrial Applications and Academic Per-
spectives (Eds.: S. S. Stahl, P. L. Alsters), Wiley-VCH, Weinheim, 2016.
[6] For recent patents on the synthesis of tetramethyl (1,1’-biphenyl)-
3,3’,4,4’-tetracarboxylate, see: a) A. Ishikawa, H. Mitsui, K. Akao, T.
Komoda (Ube Industries) Japanese Patent 2000/186063, 2000; b) A. Ishi-
kawa, H. Mitsui, Y. Tanaka, K. Sasaki (Ube Industries) Japanese Patent
2001/302654, 2001; c) T. Tsuji, Y. Yamamoto, S. Yasuda (Ube Industries)
Japanese Patent 2006/036643, 2006; d) N. Kakeya, S. Yasuda (Ube In-
dustries) Japanese Patent 2008/044906, 2008; e) M. Nishio, Y. Imashima
Japanese Patent 2010/100616, 2010; f) M. Nishio, Y. Imashima (Ube In-
dustries) Japanese Patent 2011/213666, 2011; g) M. Nishio, Y. Imashima
(Ube Industries) Japanese Patent 2011/219447, 2011; h) M. Miyasaka; T.
Tsuji (Ube Industries) Japanese Patent 2016/037500, 2016. For articles,
see: i) H. Iataaki, H. Yoshimoto, J. Org. Chem. 1973, 38, 76–79; j) A. Shio-
tani, M. Yoshikiyo, H. Itatani, J. Mol. Catal. 1983, 18, 23–31; k) A. Shiota-
ni, H. Itatani, T. Inagaki, J. Mol. Catal. 1986, 34, 57–66.
both types of byproducts) vary in a more unpredictable way with the
pyridine structure than regioratios.
[14] From previous observations[9a] at lower O2 pressures, the addition of an
equivalent amount of a strong carboxylic acid is preferred. As the elec-
trophilicity of the PdII center increases with the amount of added
strong acid, and thereby its propensity to reductive catalyst deactiva-
tion by Pd black formation, the latter probably outcompetes aerobic re-
oxidation of Pd0 to PdII at low O2 pressures.
[15] E. M. Simmons, J. F. Hartwig, Angew. Chem. Int. Ed. 2012, 51, 3066–
3072; Angew. Chem. 2012, 124, 3120–3126.
[16] A decrease of the KIE value upon the increase of the temperature has
also been reported.[21c]
[17] T. R. Jack, J. Powell, Can. J. Chem. 1975, 53, 2558–2574.
[18] Earlier work on Pd(MeCO2)2/pyridine (1:1) catalyzed arene acetoxylation
has shown that carboxylate-bridged dimeric [Pd(MeCO2)2py]2 species
are generated quantitatively from Pd(MeCO2)2+1 equiv. 4-substituted
pyridine ligand in an aromatic solvent medium. Such dimeric species
were proposed as the catalyst resting state (see: A. K. Cook, M. S. San-
ford, J. Am. Chem. Soc. 2015, 137, 3109–3118). The kinetics of the ace-
toxylation were in line with a mechanism in which the dimeric precata-
lyst is in equilibrium with a monometallic species that enters the cata-
lytic cycle by inducing arene palladation, which represents the rate-de-
termining step. This previous work provides strong evidence for the in-
volvement of similar dimeric species and their dissociation under our
(neat) conditions. Further support for this is provided by the low de-
pendence on the electronic nature of the pyridine ligand in both our
arene CDC system and the previous arene acetoxylation system (for a
discussion on this, see the foregoing reference). The first-order depend-
ence on Pd and absence of a KIE under our neat conditions do not sup-
port dimer dissociation pre-equilibrium conditions before rate-limiting
arene CÀH activation. Instead, our results are in line with rate-determin-
ing dissociation of the dimeric precatalyst, followed by fast arene palla-
dation under low steady-state conditions for the Pd species, in which
the rate of the arene CÀH activation step is high because of the very
high o-xylene concentration under neat conditions (see Supporting in-
formation for rate equations).
[19] a) S. I. Gorelsky, D. L. Lapointe, K. Fagnou, J. Am. Chem. Soc. 2008, 130,
10848–10849; b) D. L. Lapointe, K. Fagnou, Chem. Lett. 2010, 39, 1118 –
1126.
[20] Instead of the breaking of the carboxylate bridges to generate mono-
metallic Pd species, the dissociation of a pyridine ligand from a PdII
center in a carboxylate-bridged dimer would also allow subsequent
arene palladation by a CMD mechanism. Preferential dissociation of the
pyridine ligand in the presence of much more labile carboxylate
bridges is, however, unlikely for a weakly oxophilic late transition metal
such as Pd.
[7] For Pd-catalyzed aerobic benzylic oxidation of alkylarenes, see: a) D. R.
Bryant, J. E. McKeon, B. C. Ream, J. Org. Chem. 1968, 33, 4123–4127;
b) S. D. George, S. C. Sherman, A. V. Iretskii, M. G. White, Catal. Lett.
2000, 65, 181–183; c) H. Liu, G. Shi, S. Pan, Y. Jiang, Y. Zhang, Org. Lett.
2013, 15, 4098–4101.
[8] For a review on undirected CÀH activation of simple arenes, see: N.
Kuhl, M. N. Hopkinson, J. Wencel-Delord, F. Glorius, Angew. Chem. Int.
Ed. 2012, 51, 10236–10254; Angew. Chem. 2012, 124, 10382–10401.
[9] a) Y. Izawa, S. S. Stahl, Adv. Synth. Catal. 2010, 352, 3223–3229; b) D.
Wang, Y. Izawa, S. S. Stahl, J. Am. Chem. Soc. 2014, 136, 9914–9917;
c) D. Wang, S. S. Stahl, J. Am. Chem. Soc. 2017, 139, 5704–5707.
[10] For representative examples on the use of pyridine ligands in aerobic
CÀH activation, see: a) T. Nishimura, T. Onoue, K. Ohe, S. Uemura, Tetra-
hedron Lett. 1998, 39, 6011–6014; b) S. R. Fix, J. L. Brice, S. S. Stahl,
Angew. Chem. Int. Ed. 2002, 41, 164–166; Angew. Chem. 2002, 114,
172–174; c) T. Nishimura, H. Araki, Y. Maeda, S. Uemura, Org. Lett. 2003,
5, 2997–2999; d) M. M. S. Andappan, P. Nilsson, M. Larhed, Chem.
Commun. 2004, 218–219; e) K. S. Yoo, C. P. Park, C. H. Yoon, S. Sakagu-
chi, J. O’Neill, K. W. Jung, Org. Lett. 2007, 9, 3933–3935; f) Y.-H. Zhang,
B.-F. Shi, J.-Q. Yu, J. Am. Chem. Soc. 2009, 131, 5072–5074.
[21] For representative examples, see: a) J. M. Davidson, C. Triggs, J. Chem.
Soc. A 1968, 1324–1330; b) R. S. Shue, J. Am. Chem. Soc. 1971, 93,
7116–7117; c) M. Kashima, H. Yoshimoto, H. Itatani, J. Catal. 1973, 29,
92–98; d) L. M. Stock, K. Tse, L. J. Vorvich, S. A. Walstrum, J. Org. Chem.
1981, 46, 1757–1759; e) L. J. Ackerman, J. P. Sadighi, D. M. Kurtz, J. A.
Labinger, J. E. Bercaw, Organometallics 2003, 22, 3884–3890.
[22] Oxidation: J. H. Teles, I. Hermans, G. Franz, R. A. Sheldon in Ullmann’s
Encyclopedia of Industrial Chemistry, Wiley-VCH, Weinheim, 2016.
[23] K. Imatani, A. Ishikawa, M. Kashibe (Ube Industries) UK Patent 1988/
2205765, 1988.
[11] P. L. Alsters (DSM), WO 2016/008931, 2016.
[12] P. L. Alsters, L. Lefort, C. A. M. R. van Slagmaat, R. P. van Summeren,
G. K. M. Verzijl (DSM), WO 2017/102944, 2017.
[13] As benzylic oxygenation and trimer formation respond differently to
the nature of the pyridine ligand, chemoratios (which take into account
Manuscript received: December 11, 2017
Revised manuscript received: March 3, 2018
Version of record online: && &&, 0000
ChemCatChem 2018, 10, 1 – 8
www.chemcatchem.org
7
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

FULL PAPERS
Y. ꢀlvarez-Casao,
Neat and clean: We report an efficient
aerobic cross-dehydrogenative coupling
of o-xylene under neat conditions,
which brings important industrial bene-
fits towards the synthesis of a monomer
used in polyimide resins.
C. A. M. R. van Slagmaat, G. K. M. Verzijl,
L. Lefort, P. L. Alsters,*
M. ꢀ. Fernꢁndez-IbꢁÇez*
&& – &&
Palladium-Catalyzed Cross-
Dehydrogenative Coupling of o-
Xylene: Evidence of a New Rate-
Limiting Step in the Search for
Industrially Relevant Conditions
&
ChemCatChem 2018, 10, 1 – 8
www.chemcatchem.org
8
ꢃ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!