Comparative experimental and EXAFS studies in the mizoroki-heck reaction with heteroatom-functionalised N-heterocyclic carbene palladium catalysts

Article abstract of DOI:10.1002/chem.200601278

A study on the Mizoroki-Heck coupling of selected aryl bromides with acrylates catalysed by a series of Pd complexes of bidentate pyridyl-, picolyl-, diphenylphosphinoethyl- and diphenylphosphinomethyl-functionalised N-heterocyclic carbene (NHC) is reported. The observed activity is dependent on the type of solvent and base used and the nature of the classical donors of the mixed-donor bidentate ligand and its bite angle. A mechanistic model is presented for the pyridine-functionalised NHC complexes based on an in situ EXAFS study under dilute catalyst conditions (2 mM Pd). The model involves pre-dissociation of the pyridine functionality and oxidative addition of ArBr in the early stages of the reaction, as well as formation of monomeric and dimeric Pd species at the time of substrate conversion.

Full text of DOI:10.1002/chem.200601278

DOI: 10.1002/chem.200601278
Comparative Experimental and EXAFS Studies in the Mizoroki–Heck
Reaction with Heteroatom-Functionalised N-Heterocyclic Carbene
A
C
H
T
R
E
U
N
G
Palladium Catalysts
[
b, d]
[a, c]
[d]
[d]
[a]
Steven G. Fiddy,*
John Evans,
Thomas Neisius, Mark A. Newton,
[
a]
[a]
Nikolaos Tsoureas, Arran A. D. Tulloch, and Andreas A. Danopoulos*
Abstract: A study on the Mizoroki–
Heck coupling of selected aryl bro-
mides with acrylates catalysed by a
series of Pd complexes of bidentate
pyridyl-, picolyl-, diphenylphosphi-
noethyl- and diphenylphosphinometh-
yl-functionalised N-heterocyclic car-
bene (NHC) is reported.The observed
activity is dependent on the type of sol-
vent and base used and the nature of
the “classical” donors of the mixed-
donor bidentate ligand and its bite
angle.A mechanistic model is present-
ed for the pyridine-functionalised NHC
complexes based on an in situ EXAFS
study under dilute catalyst conditions
(2 mm Pd).The model involves pre-dis-
sociation of the pyridine functionality
and oxidative addition of ArBr in the
early stages of the reaction, as well as
formation of monomeric and dimeric
Pd species at the time of substrate con-
version.
Keywords:
carbene ligands
·
EXAFS spectroscopy
·
Heck
reaction · homogeneous catalysis ·
palladium
[2]
Introduction
ladacycles which generate the catalytic species in situ.
0
Well-defined Pd complexes have also been used for this
purpose with variable activity, occasionally inferior to sys-
tems formed in situ.In contrast, the development of cata-
lysts for the Mizoroki–Heck reaction has proceeded at a
slower pace.In particular, the general coupling of aryl chlor-
ides with alkenes under mild conditions has still not been re-
alised.In addition, many questions on the mechanism of cat-
alysis remain unanswered, including the nature of the active
species, the involvement of palladium nanoparticles, the role
Research into the discovery of new catalysts for CꢀC bond
formation has given active systems for Suzuki–Miyaura,
Kumada and Negishi coupling, even with unreactive aryl
[1]
chlorides as substrates. Some of the successful catalysts are
0
formed by a combination of Pd precursors and trialkylphos-
phines, phosphites, N-heterocyclic carbenes (NHCs) and pal-
[
a] Prof.J.Evans, Dr.N.Tsoureas, Dr.A.A.D.Tulloch,
Dr.A.A.Danopoulos
[3]
of the halide ions as ligands and so on. Palladium com-
plexes with simple monodentate or bidentate NHC ligands,
either pre-formed or generated in situ from imidazolium
salts and palladium acetate, were first introduced as Mizoro-
School of Chemistry
University of Southampton
Highfield, Southampton, SO171BJ (UK)
Fax : (+44)238-059-6805
[4]
ki–Heck catalysts by Herrmann et al. Catalysts formed
E-mail: ad1@soton.ac.uk
II
from Pd and mixed-donor bidentate ligands with NHC and
[
[
b] Dr.S.G.Fiddy
[5]
phosphine donors were predicted to be superior catalysts
CCLRC Daresbury Laboratory
Warrington, WA44AD (UK)
Fax : (+44)192-560-3124
E-mail: s.g.fiddy@dl.ac.uk
by virtue of the hemilability of the phosphine donor of the
heterobifunctional ligand system.Preliminary experimental
data did not confirm this prediction, even though the cata-
lysts showed good activity with aryl bromides, while they
[
c] Prof.J.Evans
Diamond Light Source, Chilton, OX11 0DE (UK)
[6]
were inactive with aryl chlorides. Catalysts formed from
d] Dr.S.G.Fiddy, Dr.T.Neisius, Dr.M.A.Newton
The European Synchrotron Radiation Facility
II
Pd and bidentate ligands with mixed NHC and pyridine
donors showed high activity with aryl iodides and bromides
as substrates under high dilution, but conversions with unac-
tivated or deactivated aryl chlorides was low even under
6
rue Jules Horowitz, BP220, 38043 Grenoble (France)
Supporting information for this article is available on the WWW
under http://www.chemeurj.org/ or from the author.
3652
ꢀ 2007 Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim
Chem. Eur. J. 2007, 13, 3652 – 3659

FULL PAPER
[7]
harsh conditions. The poten-
tial hemilability of the pyridine
donor of the bidentate ligand
has been considered as a possi-
ble reason for the observed
high activity, even though
direct evidence (mainly spec-
troscopic) of its occurrence
was difficult to obtain.Recent-
ly, a few reports claimed that,
as a consequence of the forc-
ing reaction conditions em-
ployed in the high-temperature
Mizoroki–Heck reaction, the
catalytic species responsible
for the transformation are Pd
Table 1.Comparative activity data in the coupling of various aryl bromides and acrylates catalysed by pyri-
dine- and picoline-functionalised N-heterocyclic carbene palladium complexes.Solvent was NMP in all reac-
tions.
[
a]
[b]
Entry
Aryl halide
Br
4-CH OC
4-NO
4-CH
Alkene
Cat.
T [8C]
t [h]
Base
NEt
Et
NaOAc
NaOAc
Yield [%]
TON
1
2
3
4
5
6
7
8
9
C
6
H
5
MA
BA
BA
1a
1a
1a
1a
1a
1c
2c
1a
1c
2c
140
140
140
140
140
140
140
140
140
140
18
18
18
18
18
18
18
75
80
80
3
20
4
100
100
60
66
74
35
40
2900
4300
200
200
850
3
6
H
4
Br
Br
COC
3
N
2
C
6
H
4
3
6
H
4
Br
BA
4-CH COC H Br
3
3
3
6
6
6
4
4
4
MA
MA
MA
MA
MA
MA
Et N
3
4-CH
4-CH
COC
COC
H
H
Br
Br
Et
3
Et
3
Et
3
Et
3
Et
3
N
N
N
N
N
950
1050
5000
5700
12900
C
C
C
6
6
6
H
H
H
5
5
5
Br
Br
Br
10
48
ꢀ4
[a] BA: butyl acrylate, MA: methyl acrylate.[b] Catalyst concentration was 3 5. ”10
m in all reactions.
Table 2.Comparative activity data for the coupling of aryl bromides with acrylates catalysed by phosphine-
and pyridine-functionalised N-heterocyclic carbene palladium complexes and bis(N-heterocyclic carbene) pal-
ladium catalysts.Solvent was NMP in all reactions except for entries 17–21, for which the solvent was N,N-di-
methylacetamide.
nanoparticles
rather
than
metal complexes operating in
[3]
solution.
Here we compare the activi-
ty in the Mizoroki–Heck reac-
tion of functionalised NHC
complexes of Pd that we have
[a]
[b]
Entry
Aryl halide
Alkene
Cat.
T [8C]
t [h]
Base
Yield [%]
TON
1
2
3
4
5
6
7
8
9
4-CH
4-CH
3
3
COC
COC
6
6
H
H
4
4
Br
Br
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
MA
3c
3b
4c
5c
3c
3b
4c
2c
5c
3c
3b
4c
2c
5c
3c
3b
3c
3b
4c
2c
5c
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
120
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
NEt
NEt
3
92
67
100
100
5
185
134
200
200
10
110
65
200
200
20
45
35
70
105
32
3
4-CH COC H Br
3
3
3
3
3
6
6
6
6
6
4
4
4
4
4
NEt3
NEt
4-CH
4-CH
4-CH
4-CH
COC
COC
COC
COC
H
H
H
H
Br
Br
Br
Br
3
recently
Scheme 1).
We also describe efforts to
prepared
[7a,8,9]
Cs
2
Cs
2
Cs
2
CO
CO
CO
3
3
3
3
3
(
55
32
100
100
10
22
17
35
52
16
18
61
28
90
45
95
probe the nature of the dis-
solved species in a catalytic so-
lution obtained from 2c or 2d
using EXAFS in solution.
4-CH COC H Br
Cs CO
3
6
4
2
4-CH
3
COC
6
H
4
Br
2
Cs CO
10
11
12
13
14
15
16
17
18
C
C
C
C
C
C
C
C
C
6
6
6
6
6
6
6
6
6
H
H
H
H
H
H
H
H
H
5
5
5
5
5
5
5
5
5
Br
Br
Br
Br
Br
Br
Br
Br
Br
NEt
NEt
NEt
NEt
NEt
3
3
3
3
3
Cs
2
Cs
2
Cs
2
Cs
2
CO
CO
CO
CO
3
3
3
3
3
3
3
Results and Discussion
35
122
56
180
90
190
Selected catalytic data for the
Mizoroki–Heck reaction of
phenyl bromides are given in
Tables 1 and 2.All of the com-
19
20
C H Br
6
6
6
5
5
5
Cs CO
2
C
C
H
H
Br
Br
Cs
Cs
2
CO
CO
21
2
ꢀ
2
plexes catalyse the coupling of
[a] MA: methyl acrylate.[b] Catalyst concentration was 5”10 m in all reactions.
activated and unactivated aryl
bromides with electron-defi-
cient alkenes.Prolonged reaction times, high temperatures
and low catalyst loadings under dilute conditions lead to
high turnover numbers, especially with the pyridine-func-
tionalised ligands.This manifests the good thermal stability
of the active species under catalytic conditions.In all runs in
which a pyridine- or picoline-functionalised NHC complex
was used as catalyst precursor,
there is no visible indication of
Pd black formation during the
reaction;
however,
small
amounts of Pd black were ob-
served with the phosphine-
functionalised ligands.With
the picoline-functionalised li-
gands there is no strong influ-
ence of the nature of the NHC
substituents
on
activity.
II
(Table 1, entries 5 versus 6 and
8 versus 9); however, over
Scheme 1.Pd complexes used as catalytic precursors in the coupling of aryl bromides with acrylates. a: R=
2 6 3
tBu; b: R=mesityl; c: R=DiPP (Dipp = 2,6-iPr C H ); X=Br, OTf (only for 2d), Y=Me.
Chem. Eur. J. 2007, 13, 3652 – 3659
ꢀ 2007 Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim
www.chemeurj.org
3653

A.A.Danopoulos, S.Fiddy et al.
long reaction times the pyridine-functionalised system is
slightly more active than the picoline analogue (Table 1, en-
tries 9 versus 10).Further comparison of the picoline- and
pyridine-functionalised complexes is possible by plotting
conversion versus reaction time (see Figure 1).
the unactivated bromobenzene is lower than for 4-bromoa-
cetophenone, as expected.Interestingly, the activity of the
phosphine-based 3c and 4c towards bromobenzene with
Cs CO as base is superior to that of the pyridine-based
2
3
system if dimethylacetamide is used as solvent.
EXAFS studies on the pyridine-functionalised NHC palladi-
um catalytic system: Initially, a spectrum of precatalyst 2c in
the solid state and under non-catalytic conditions (solution
in N-methylpyrrolidone, NMP) was recorded at room tem-
perature and compared to the crystallographic model previ-
[7a,9]
ously published.
This was conducted to 1) show that the
EXAFS data collected in this study provide a reasonable fit
to the crystallographic model and 2) discover whether any
structural changes could be detected in solution under dilute
conditions.Pd K-edge standard EXAFS analysis for the
pure isolated complex in the solid state (Figure 2a, Table 1
in the Supporting Information) gave the expected three-
shell model of a first coordination shell of two carbon atoms
at 2.01 Š, a second coordination shell of one nitrogen atom
at 2.21 Š and a third coordination shell of one bromine
atom at 2.48 Š. Splitting of the first carbon shell into two
separate shells did not lead to a significant improvement of
the data.Importantly, EXAFS analysis provides proof of co-
ordination of Npyridine to Pd.Although distinguishing be-
tween C and N is virtually impossible in this situation, there
Figure 1.Comparison of the conversion of the coupling of bromoaceto-
phenone with methyl acrylate in the presence of triethylamine with 1c
and 2c as pre-catalysts.Amounts: Aryl bromide 5 mmol, alkene/amine
ꢀ
4
6
mmol, catalyst 3.5”10 mmol.
Interestingly, the initial rates of the system based on pre-
catalyst 1c are higher than those of 2c.The reasons for this
behaviour are not fully under-
stood, although it may be relat-
ed to different rates of the pre-
activation process involving
dissociation of the pyridine
donor from the metal atom.
This has also emerged as plau-
sible initiation step, based on
the interpretation of the solu-
tion EXAFS data (see below).
Finally, the best base for the
reactions described in Table 1
is triethylamine.
The activities of the phos-
phine-functionalised
NHC
complexes with five- and six-
membered chelate rings and of
the monodentate dicarbene
complex 5c are similar when
triethylamine is used as base
(
Table 2, entries 1, 3 and 4).
However, for the coupling of
-bromoacetophenone, re-
placement of triethylamine by
Cs CO results in a drop of the
4
3
2
3
Figure 2.Pd K-edge k -weighted EXAFS and Fourier transform, phase-shift-corrected for carbon, of a Pd bro-
mine carbene catalyst (2c, Scheme 1) a) in the solid-state form, b) 2 mm in NMP and during the Mizoroki–
Heck reaction (2 mm Pd) after c) 353 K, d) 5 mins at 403 K and e) 20 mins at 403 K.Black and grey lines rep-
resent the experimental and theoretical fits derived from spherical wave analysis in EXCURV98, respectively.
The time represents the total time at the reaction temperature of 403 K; however, the catalytic solution was
quenched to room temperature at each of these time intervals to allow six scans to be recorded on the catalyt-
ic solution.Once this was completed, the solution was then heated back to 403 K for the required amount of
time before quenching to room temperature at the next time interval.
activity for the systems based
on the phosphine-functional-
ised ligands 3c and 4c as com-
pared to the pyridine-function-
alised 2c and dicarbene 5c.
The activity of all catalysts for
3654
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim
Chem. Eur. J. 2007, 13, 3652 – 3659

EXAFS Studies in the Mizoroki–Heck Reaction
FULL PAPER
1
3
is a good match between the PdꢀNpyridine distance in the crys-
tallographic data and the EXAFS model.Other distances
obtained from the EXAFS analysis are also comparable to
both the k -weighted and k -weighted experimental data (a
reduction in R factor of approximately 25% was observed
1
when this extra shell was added to the k -weighted data).
[9]
the crystallographic data (in the solid the average Pdꢀ
This suggests that complex 2c is stable in solution at room
temperature and the stated concentration and that Npyridine
of the ligand remains bound even in a polar solvent.A fur-
ther shell of one nitrogen could also be successfully fitted at
2.83 Š for the k -weighted data (2.88 Š for the k -weighted
data), and this causes a reduction in the R factor.This dis-
CNHC, PdꢀC
, PdꢀN
and PdꢀBr distances are 1.97,
methyl
pyridine
2
.03, 2.17 and 2.45 Š, respectively). Furthermore, addition
of a fourth shell of about one nitrogen atom at 2.87 Š de-
creases the R factor (from 26.2% to 24.3%) and improves
the fit of the data.However, if the coordination number of
the fourth, nitrogen shell is increased from one to two, the
R factor increases by about 9%, which is consistent with the
crystallographic model.Incorporation of the NHC function-
ality in a chelate ring results in unequal distances between
the two N atoms of the imidazole ring and the Pd centre:
1
3
tance is consistent with the shorter PdꢀN
distance fitted
NHC
for the single-crystal data.
There is still little known about the mechanism operating
[3]
with carbene-type catalysts. The EXAFS spectroscopic
study was intended to probe the following questions: 1) Are
Pd nanoparticles formed during the reaction? 2) Is there
the PdꢀN
distance for the N atom bonded to the pyri-
NHC
0
dine moiety is significantly shorter than that for the N atom
attached to the DiPP ring (2.82 Š versus 3.15 Š, respective-
ly).Attempts were also made to incorporate extra carbon
atoms to model the carbon backbone between the imidazole
evidence for the generation of Pd by reductive elimination
of 2-methylimidazolium salts from the starting precatalyst?
3) Does the pyridine arm of the bidentate ligand detach
from Pd during the catalysis? 4) Is there any observable loss
of bromide during the reaction which could offer an alterna-
tive possibility for alkene coordination?
Hence, ultradilute XAS studies of the Pd/carbene system
were then conducted under the conditions for the coupling
of n-butyl acrylate with p-bromoacetophenone in order to
identify any species present during the catalytic reactions
and possibly determine the structure of the catalysts in solu-
tion under dilute conditions.After a solution of complex 2c,
butyl acrylate and p-bromoacetophenone in NMP had been
injected into the cell, it was heated gradually (10 Kmin ) to
the reaction temperature (403 K) and, at certain intervals,
the cell was rapidly cooled to 263 K to quench the reaction,
thereby providing a “snapshot” of the intermediate species
present in solution.The progress of the reaction was moni-
tored under the same temperature regime by taking aliquots
and analysing quantitatively the reaction mixture by gas
chromatography (Figure 3).
Initially, the catalytic solution at room temperature before
heating provided a structure with almost identical parame-
ters to those observed for 2c in the non-catalytic NMP solu-
tion, with a positive indication from the EXAFS analysis
that a shell of one nitrogen atom could be incorporated to
model any pyridine interaction.A second measurement was
made after the catalytic solution had reached 353 K and had
been quenched to room temperature.Even after heating,
the Pd pre-catalyst remained intact and no significant
changes in structure were seen.After cooling, EXAFS indi-
cated that the best theoretical model of the data consisted
of two carbon atoms at 2.03 Š, one nitrogen atom at 2.19 Š
and one bromine atom at 2.49 Š (Table 2 in the Supporting
Information).Even though the shell of nitrogen at 2 1. 9 Š
could be fitted significantly to the experimental data, the
Debye–Waller factor for this shell was relatively high.Hy-
[10]
and pyridine rings.Evans et al.
cases, the Ni···Cbackbone carbon atoms could be detected for a
series of tris(diamine) complexes, in particular for the cyclo-
showed that, in some
A
C
H
T
R
E
U
N
G
hexanediamine complexes in which the methylene carbons
in the 3,6-positions were forced to be equatorial.With the
present materials, detection of the carbon backbone is hy-
pothesised to be very difficult, due to the unsymmetrical
chelation and the lack of carbon atoms involved in the back-
bone.
ꢀ
1
Furthermore, the solid-state data reported so far were an-
ꢀ
1
alysed in a k range of 2.8–15 Š , but analysis of the com-
plex dilute conditions (0.002m Pd) in solution is only realis-
ꢀ
1
tically possible between 2.8 and 11 Š .To determine
whether the smaller k range causes any significant problems,
the solid-state data were also analysed between 2.8 and
ꢀ
1
1
1 Š (Table 1b in the Supporting Information).With the
smaller k range, negative Debye–Waller factors appeared
for the short-range nitrogen shell and distances varied signif-
icantly compared to the crystallographic data.However, by
1
3
using k -weighted data (as well as k -weighted data) for this
smaller k range, the parameters determined became more
consistent with the crystallographic distances and therefore,
for the EXAFS studies under dilute conditions, the data
shown here represent parameters obtained from analysis of
3
k -weighted data (but with significant consideration of the
1
k -weighted data and parameters).Both forms of the
weighted data provide valuable information in the analysis,
1
as the k -weighted data enhance the contribution of the
3
lighter scatterers, where as the k -weighted data enhance the
contribution of bromine and other more distant shells of
scatterers.
Ultradilute Pd K-edge EXAFS analysis of an NMP solu-
tion of complex 2c (0.002m Pd) at room temperature could
be best fitted to a first co-ordination sphere for the carbene
with about two light backscatters (C) at 2.01 Š and a second
coordination sphere of one bromine atom at 2.48 Š. Also at
this stage, a shell of one nitrogen atom at a distance of
potheses that might explain this are that the PdꢀN
pyridine
bond is weakened, and consequently the nitrogen atom is
[11]
able to move with a higher degree of freedom or some of
the bidentate ligand acts as monodentate (cf.complex 5c).
A further shell of one nitrogen atom at 2.93 Š can also be
2.19 Š can also be fitted successfully and significantly to
Chem. Eur. J. 2007, 13, 3652 – 3659
ꢀ 2007 Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim
www.chemeurj.org
3655

A.A.Danopoulos, S.Fiddy et al.
the EXAFS data.Consequent-
ly, although no conclusive evi-
dence can be drawn, our analy-
sis suggests that the loss of pyr-
idine “coordination” at this
temperature may provide
free site for co-ordination of
the olefin (Reaction (ii),
a
Scheme 2) before insertion
into the PdꢀC s bond.Analy-
sis of the longer PdꢀN distance
also provides further indirect
evidence that complete dissoci-
ation of the pyridine functional
group from the Pd centre may
occur.The analysis can be best
fitted now to a shell of two ni-
trogen atoms at 2.90 Š, consis-
tent with dissociation of the
pyridyl group and the ligand
acting as monodentate from
Figure 3.Reaction profile for the Mizoroki–Heck reaction between 4-bromoacetophenone and butyl acrylate.
Red: pre-catalyst 2c, green: precatalyst 2c (repeat), blue: pre-catalyst 2d; circles denote product, triangles
denote reactant (4-bromoacetophenone); negative times represent temperature between 287 and 403 K while
ꢀ
1
solution was heated at 5 Kmin .Catalyst concentration =0.002 m Pd.
[9]
added to the model, which improves the fit and decreases
the R factor by about 7%.This nitrogen atom is proposed
to be the NHC nitrogen atom, and this indicates that at this
temperature the carbene ligand remains intact and bound to
the Pd centre.
Refinement of the model for the experimental data of the
solution that had been heated to 403 K showed a significant
change (Table 2c in the Sup-
the NHC end (see structures 10–12 in ref. or complex 5c
in this paper).
After a further 5 min at 403 K (and subsequent quenching
to room temperature), another significant change was
noted: addition of a shell of one palladium atom at 2.74 Š
improved the R factor and provided a better fit to the Fouri-
er transform; attempts to fit more than one palladium atom
porting Information).In this
case a shell of one N atom at
about 2.19 Š could not be suc-
1
cessfully fitted for both the k -
3
and k -weighted data and did
not provide a further improve-
ment in the fit.This is consis-
tent with previous measure-
ments in which the high
Debye–Waller factors suggest-
ed greater lability of the pyri-
dine nitrogen atom and/or
complete loss of the Pdꢀ
Npyridine bond for some catalyst
at 353 K.The absence of a Pd ꢀ
N distance at this stage is inter-
esting if not wholly unexpect-
ed, as the hemilabile nature of
the pyridine ligand will pre-
sumably hinder detection of
the pyridine N shell.It may
also indicate that, at 403 K, the
PdꢀN
interaction be-
pyridine
comes labile, and possibly a co-
ordination site is left free.At-
tempts to fit further shells to
model any solvent interactions
did not improve the fitting of
Scheme 2.Proposed mechanism for the Mizoroki–Heck reaction of bromoacetophenone with butyl acrylate
catalysed by 2c (observed species characterised by XAS indicated by boxes; the remainder are all plausible
transient species).
3656
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim
Chem. Eur. J. 2007, 13, 3652 – 3659

EXAFS Studies in the Mizoroki–Heck Reaction
FULL PAPER
to the data led only to deterioration in the fit.This points to
several possibilities: 1) Br-bridged dimer species were
formed (this is thought to be highly unlikely, as the PdꢀPd
planation for this is that the catalyst has now formed a new
dimeric species containing bridging bromide (reaction (viii),
Scheme 2).In the scheme we propose an equilibrium be-
tween the dimeric species (VI) and the oxidative-addition
product (I).A closely related mechanistic model was al-
distance is too short in comparison with those observed for
bromine-bridged palladium dimers, which range from 3.50
[
12]
[15]
to 3.75 Š),
formed,
2) direct PdꢀPd dimers and trimers are
ready postulated by Evans et al. in which a dimeric spe-
[
13]
3) small soluble colloidal particles are formed,
cies acts as the starting catalyst.The formation of a dimeric
species is also consistent with the reaction profile and the
model postulated by De Vries, in which the species involved
in the actual catalytic cycle are monomeric and dimeric spe-
cies and not soluble palladium colloids.This hypothesis is
also borne out by further EXAFS spectra taken after total
heating times of 30 and 50 min, for which the same dibro-
mide model can be fitted successfully to the EXAFS data
and high conversion is shown by the reaction profile/yield
plot.
either by loss of the NHC ligand (as imidazolium) or by loss
of HBr and stabilisation of the colloids towards formation
of Pd black by the NHC ligands (reaction (vii), Scheme 2) or
4) a small percentage of the palladium has been converted
to large palladium particles or “Pd black”.However, if the
last possibility is correct, the PdꢀPd interaction must be
caused by a small percentage of these large particles due to
the low coordination number of the palladium shell.For ex-
ample, for a fcc spherical Pd particle (diameter ca.40 Š)
consisting of about 2171 atoms, the first shell coordination
To confirm whether oxidative addition had occurred when
a temperature of 353 K was reached, a supplementary ex-
periment was conducted with precatalyst complex 2d, an an-
[14]
number from XAS would be approximately 10. However,
as the PdꢀPd coordination number observed for this model
ꢀ
ꢀ
was only 1, it means that only about 10% of the palladium
could be in the form of large palladium particles.In combi-
nation with the appearance of a palladium interaction in the
EXAFS data, it is also noted that the yield of product is low
at this stage and the reaction profile appears to show an in-
duction period in the reaction, which could be related to the
formation of these soluble Pd colloids.
alogue of 2c in which Br has been replaced by CF SO .
3 3
The occurrence of oxidative addition could easily be proved
since the light oxygen scatterer associated with the triflate
ion would be substituted by the heavier scatterer Br.When
the reaction sequence described above was carried out it
was found that a model consisting of only mainly light scat-
1
3
terers gave the best fit to both the k and k data at room
temperature (Figure 4, Table 4 in the Supporting Informa-
tion).However, a small percentage of Br in the model (Br
coordination number of about 0.3) was found to improve
the R factor significantly, that is, even at room temperature
some oxidative addition may occur.However, after a tem-
perature of 353 K was reached, the best fit to the data also
contained a heavier scatterer, which was best fitted to one
bromine atom, which suggests that oxidative addition had
taken place.The final model to the 353 K experimental data
fitted very well to the main experimental data collected for
the precatalyst 2c (Figure 2, Table 2 in the Supporting Infor-
mation) with a NpyridineꢀPd interaction persistent at least
Moreover, after a further 5 min at 403 K (10 min total
time at 403 K), the best theoretical model that could be
fitted to the experimental EXAFS data does not contain a
Pd shell.The best fit to the data consists of a simple two-
shell model of two carbon atoms at 2.04 Š and one bromine
atom at 2.49 Š. When a shell of palladium is fitted, the R
factor does not decrease at all.At this stage, it is unclear
why the shell of palladium would appear then disappear.
Possible hypotheses include: 1) rapid conversion of the di-
meric and trimeric species (which are initially formed) to
ꢀ
monomeric species when excess Br becomes available or
2) conversion of palladium colloids to the catalytically active
monomeric or possibly dimeric species.This is consistent
until a temperature of 353 K.Finally, analysis of the catalyt-
[3b]
with the findings of De Vries that monomeric and dimeric
species (which are the active species in the mechanism pro-
posed) are formed from stabilised Pd nanoparticles.Howev-
er, if a shell of two nitrogen atoms is added at a distance of
ic solution after heating to 403 K indicated loss of the Pdꢀ
N
interaction, but the stability of the PdꢀNHC interac-
pyridine
tion is demonstrated by the significance of the fitting of two
nitrogen atoms at 2.84 Š. Therefore from this experiment it
can be concluded that oxidative addition is an early step in
the reaction mechanism and does, as expected, occur by a
temperature of 353 K.However, it is unclear exactly how
the oxidative addition mechanism operates, as there is no in-
2.88 Š, the significant reduction in the R factor suggests that
the carbene ligand remains attached to a large majority of
the palladium at this stage of the reaction, probably either
stabilising the Pd against Pd black formation or as one of
the intermediates in the proposed mechanism, that is, the
product of oxidative addition (note: when a shell of one ni-
trogen atom is added, negative Debye–Waller factors per-
sist).
After another 10 min of heating (20 min total), a signifi-
cant change in the EXAFS envelope occurs with an associ-
ated change in the Fourier transform and a significant in-
crease in the product yield/reactant loss.The best fit to the
data now consists of a model containing two carbon atoms
at 2.00 Š and two bromine atoms at 2.42 Š. A possible ex-
0
dication of Pd formation either in the metallic form or as a
metal complex (such as the species formed in reaction (v),
Scheme 2), and the carbene ligand appears to be present in
the 293/353/403 K spectra (as indicated by the two short
PdꢀC distances of 2.00 Š). The reductive elimination of
[16]
0
methylimidazolium as a possible step to Pd formation is
therefore an unlikely pathway.
The following conclusions can be drawn from the above-
discussed dilute EXAFS data: 1) The PdꢀNHC functionality
in the catalyst is stable in both catalytic and non-catalytic
Chem. Eur. J. 2007, 13, 3652 – 3659
ꢀ 2007 Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim
www.chemeurj.org
3657

A.A.Danopoulos, S.Fiddy et al.
of monomeric and dimeric spe-
cies, which are likely to be cat-
alytically active.
In view of these findings, the
activity difference observed for
1
c and 2c may be due to dif-
ferent formation rates of spe-
cies of type I due to the struc-
tural differences of the chelat-
ing ligand.Complex 1c with a
six-membered chelate ring may
undergo pyridine dissociation/
oxidative addition faster than
the more robust and stable 2c.
The EXAFS data collected
do not help us understand the
reactivity differences observed
with phosphine-functionalised
NHC ligands.Future work is
aimed at clarifying these subtle
differences, which in the long
term may lead to the develop-
ment of more active catalysts.
3
Figure 4.Pd K-edge k -weighted EXAFS and Fourier transform, phase-shift-corrected for carbon, of a Pd car-
bene triflate catalyst (2d, Scheme 1) during the Mizoroki–Heck reaction (2 mm Pd) after heating to a) 293 K,
b) 353 K and c) 403 K.Black and grey lines represent the experimental and theoretical fits derived from spher-
ical wave analysis in EXCURV98 respectively.
Experimental Section
Instrumentation and materials: All
reactions were carried out under inert atmosphere.Gas chromatograms
were obtained on a Varian 3400GC equipped with a Varian 8100 auto-
sampler, flame-ionisation detector.The column used was Phase DBWAX
from J & W Scientific.Complexes 1–5 were prepared as previously de-
solution.There is good evidence that oxidative addition is
the first stage in this reaction and occurs by a temperature
of 353 K.However, from our study we cannot suggest which
[
6a,7a,8,9]
scribed.
Complex 2d has not been reported before and was pre-
0
Pd species is formed prior to oxidative addition.Reductive
pared by mixing 2c with an equivalent amount of silver trifluorometha-
nesulfonate, stirring at room temperature for 24 h, removing the precipi-
tated AgBr by filtration through Celite and evaporating the organic
elimination of 2-methylimidazolium from 2c can be ruled
out, as there is always evidence for coordinated NHC at Pd.
The possibility of partial involvement of reversible methyl-
or phenylimidazolium formation is also highly unlikely since
it would require CꢀC bond cleavage, which has not been re-
phase to dryness.The off-white product 2d was formulated as Pd
(CH )(Br) based on satisfactory analytical and spectroscopic data.All
solvents used in the catalytic reactions were dried, distilled and stored
over 4 Š molecular sieves under N in sealed glass vessels.All other ma-
A
C
H
T
R
E
U
N
G
(ligand)-
A
H
R
U
G
3
2
terials were obtained from commercial suppliers and used without further
purification.
ported for relevant metal complexes.
2) On further heating, two important features are detect-
General procedure for the Mizoroki–Heck reaction: A 50-mL ampoule
equipped with a Young valve was charged with the corresponding aryl
halide, alkyl acrylate, base, diethylene glycol dibutyl ether (500 mL, inter-
nal standard) and solvent (5 mL).A quantity of standard solution of the
catalyst in the same solvent was added to the reaction mixture, and the
whole was heated to the specified reaction temperature.After the de-
sired time the reaction mixture was rapidly cooled to room temperature
and quenched with water (2 mL) and dichloromethane (5 mL).The or-
ganic layer was analysed by gas chromatography.The data reported in
Tables 1 and 2 are the average of two separate runs, which typically
agree within ꢁ2%.
ed: Firstly, the loss of NpyridineꢀPd coordination by 403 K in
both 2c and 2d could point to a mechanism by which olefin
coordination can take place and suggests that, at a certain
stage/temperature in the reaction, the pyridyl group be-
comes highly labile and dissociates from the palladium
centre.At no stage of the reaction was there any observable
loss of Br, and this provides indirect evidence that coordina-
tion of the olefin proceeds via Npyridine and not Br loss.Sec-
ondly, the detection (and disappearance) of a small amount
of PdꢀPd distances suggests the possible presence of palladi-
EXAFS measurements: X-ray absorption spectra were recorded on the
undulator beamline ID26 of the European Synchrotron Radiation Facili-
ty based in Grenoble, France (operating with an average current of ca.
um dimers and trimers or soluble Pd colloids stabilised by
the NHC ligand.However, it is proposed that these are not
the active catalytic species but form monomeric and dimeric
species which are involved in the actual catalytic cycle.
200 mA in a uniform filling mode).A fixed exit Si
was used, providing an energy resolution of ca.0 8. eV for XANES and
.6 eV for EXAFS at the Pd K-edge. For all spectra, a metallic Pd refer-
A
H
R
U
G
1
ence foil was used to provide an energy calibration for the monochroma-
tor.Two Cr mirrors were used for harmonic rejection of the incident X-
ray beam.EXAFS data were recorded in standard scanning mode with
each scan taking 20 min.The spectra were acquired at room temperature
3) Finally, EXAFS analysis suggests that dibromine-bridg-
ed dimeric species are formed when the conversion to the
product is at its maximum; this further evidence in support
3658
www.chemeurj.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim
Chem. Eur. J. 2007, 13, 3652 – 3659

EXAFS Studies in the Mizoroki–Heck Reaction
FULL PAPER
in fluorescence mode by a solid-state 13-element germanium detector
tuned to the PdKa fluorescence peak.The solution sample (ca.4 mL) was
injected into a custom-built XAS cell (see Supporting Information) con-
structed from stainless steel.In addition, at the front of the cell, alumini-
um windows (20 mm thick) were mounted to permit the entrance of X-
rays into the sample and allow fluorescence to be collected.This cell was
mounted at about 458 to the direction of the incoming beam with the de-
tector at 908 in the horizontal plane to maximise the fluorescence/scatter
ratio.In addition, the stainless steel cell was mounted to an FC 64 flange
to allow the cell to be positioned inside a six-way cross to provide a
vacuum sleeve and thereby prevent any problems caused by formation of
ice crystals on the cell windows.Typically, six scans were collected at the
Pd K-edge and summed prior to curve fitting.Background-subtracted
[3] a) N.S.Phan, M.van der Sluys, C.W.Jones,
Adv. Synth. Catal. 2006,
348, 609; b) J.G. de Vries, Dalton Trans. 2006, 421; c) M.Moreno-
Manas, L.Pleixtas, Acc. Chem. Res. 2003, 36, 638.
[4] W.A. Herrmann, M. Elison, J. Fischer, C. Kocher, G.J.R. Artus,
Angew. Chem. 1995, 107, 2602; Angew. Chem. Int. Ed. Engl. 1995,
34, 2371.
[5] K.Albert, P.Gisdakis, N Rosch, Organometallics 1998, 17, 1608.
[6] a) N. Tsoureas, A.A. Danopoulos.A.A.D. Tulloch, M.E. Light,
Organometallics 2003, 22, 4750; b) C.Yang, H .M .Lee, S .P .Nolan,
Org. Lett. 2001, 3, 1511.
[7] a) A.A.D.Tulloch, A.A.Danopoulos, R.P.Tooze, S.M.Cafferkey,
S.Kleinhenz, M B. .Hursthouse,
Chem. Commun. 2000, 1247;
b) D.S.McGuiness, K.J.Cavell, Organometallics 2000, 19, 741.
[
17]
EXAFS data were obtained by using the programme PAXAS. Spheri-
[8] A.A.Danopoulos, N.T.Tsoureas, S.MacGregor, C.Smith,
metallics, 2007, 26, 253.
[9] A.A.D.Tulloch, S.Winston, A.A.Danopoulos, G.Eastham, M.B.
Hursthouse, Dalton Trans. 2003, 699.
Organo-
[
18]
cal wave curve fitting analysis was executed in EXCURV98, using ab
initio phase shifts and backscattering amplitudes calculated by using
Von-Barth ground state potentials and Hedin–Lundqvist exchange poten-
tials.Fourier transforms of the EXAFS spectra were used to obtain an
approximate radial distribution function around the central transition
metal atom; the peaks of the FT can be related to shells of surrounding
backscattering atoms, which are characterised by atom type, number of
atoms in the shell, the absorber–scatterer distance and the Debye–Waller
factor, a measure of both the thermal motion between the absorber and
scatterer and the static disorder of the absorber–scatterer distances.The
[10] J.Evans, W.Levason, R.J.Perry,
3691.
J. Chem. Soc. Dalton Trans. 1990,
[11] Principles, Applications and Techniques of EXAFS, SEAXFS and
XANES, (Eds.: D. C. Koningsberger, R. Prins), Wiley, New York,
1991.
[12] a) J.M. Vila, M. Gayoso, M.T. Pereira, A. Romar, J.J. Fernandez,
M.Thornton-Pett, J. Organomet. Chem. 1991, 401, 385; b) J.Barro,
3
k -weighted parameters are presented in Table 1 in the Supporting Infor-
J.Granell, D.Sainz, J.Sales,
J. Organomet. Chem. 1993, 456, 147;
mation.The accuracy of bonded and non-bonded interatomic distances is
c) A.Crispini, M.Ghedini, F.Neve, J. Organomet. Chem. 1993, 448,
241; d) N.Gul, J.H.Nelson, Organometallics 2000, 19, 91.
[
19]
considered to be 1.4 and 1.6% respectively. Precision in 1st shell coor-
dination numbers is estimated to be about 5–10% and between 10 and
[13] a) C.Tubaro, A.Biffis, C.Gonzato, M.Zecca, M.Basato,
J. Mol.
T
E
3
2
[
0% for non-bonded shells.The R factor is defined as(s
A
H
R
U
G
Catal. A 2006, 248, 93; b) M.Tromp, J.R.A.Sietsma, J.A.van Bok-
hoven, G.P.F. van Strijdonck, P.W.N.M. van Leeuwen, D.C. Ko-
ningsberger, Chem. Commun. 2003, 128.
E
3
T
E
c ]k dk)”100% where c and c are the theoretical and experimental
EXAFS and k is the photoelectron wave vector.
[
[
14] A.Jentys, Phys. Chem. Chem. Phys. 1999, 1, 4059.
15] J.Evans, L.OꢁNeill, V .L .Kambhampati, G.Rayner, S.Turin, A.
Genge, A.J.Dent, T.Neisius, J. Chem. Soc. Dalton Trans. 2002, 10,
2
207.
Acknowledgements
[
[
[
16] D.S.McGuiness, N.Saendig, B.F.Yates, K.J.Cavell,
J. Am. Chem.
Soc. 2001, 123, 4029.
17] N.Binsted, PAXAS Programme for the analysis of X-ray absorption
spectra, University of Southampton, 1988.
We thank Johnson-Matthey Catalysts for support of part of this project
and generous loan of Pd salts, the EPSRC for support (to N.T. and
A.A.D.), the staff of the ESRF for their assistance and the Directors for
access to the ESRF.
18] a) S.J. Gurman, N. Binsted, I. Ross, J. Phys. C: Solid State Phys.
1
984, 17, 143; S.J. Gurman, N. Binsted, I. Ross, J. Phys. C: Solid
State Phys. 1986, 19, 1845; b) N.Binsted, EXCURV98, CCLRC
Daresbury Laboratory computer programme, 1998.
19] a) J.M. Corker, J. Evans, J. Chem. Soc. Chem. Commun., 1994,
[
1] Reviews: a) Metal-Catalyzed Cross-Coupling Reactions (Eds.: F.
Diederich, P.J.Stang), Wiley-VCH, Weinheim, 1998; b) Cross-Cou-
pling Reactions (Ed.: N. Miyaura), Springer, New York, Topics Curr.
Chem. 2002, 219, 1.
[
1
027; b) J.M.Corker, J.Evans, H.Leach, W.Levason,
J. Chem. Soc.
Chem. Commun. 1989, 181.
[
2] a) R.B. Bedford, Chem. Commun. 2003, 1787; b) A.Zapf, M.
Received: September 5, 2006
Revised: November 24, 2006
Beller, Chem. Commun. 2005, 431; c) A.F.Littke, G.C.Fu,
Angew.
Chem. 2002, 114, 4350; Angew. Chem. Int. Ed. 2002, 41, 4176.
Published online: February 16, 2007
Chem. Eur. J. 2007, 13, 3652 – 3659
ꢀ 2007 Wiley-VCH Verlag GmbH & Co.KGaA, Weinheim
www.chemeurj.org
3659