Silica-supported imine palladacycles - recyclable catalysts for the Suzuki reaction?

Article abstract of DOI:10.1016/S0022-328X(01)01080-4

Silica-supported, imine-based palladacyclic catalysts have been synthesised and the crystal structure of complex 9, the triphenylphosphine adduct of the pre-supported precursor complex 8, has been determined. The solid-supported catalysts show considerably lower activity in the Suzuki reaction than their homogeneous counterparts. Poor recyclability of the silica-immobilised catalysts and the presence of active catalysts in solution indicate that imine-based palladacyclic catalysts are unstable with respect to liberation of zero-valent palladium species. Whilst the solid-supported complexes are not useful as catalysts, they do function as excellent mechanistic probes. Studies on model complexes give further information on the processes that cause the liberation of zero-valent species not only from the solid-supported catalysts, but also from homogeneous systems. In all cases it appears that a reductive-elimination event occurs to generate the active catalyst.

Full text of DOI:10.1016/S0022-328X(01)01080-4

Journal of Organometallic Chemistry 633 (2001) 173–181
www.elsevier.com/locate/jorganchem
Silica-supported imine palladacycles—recyclable catalysts for the
Suzuki reaction?
Robin B. Bedford a,*, Catherine S.J. Cazin a, Michael B. Hursthouse b,
Mark E. Light b, Kevin J. Pike a, Stephen Wimperis a
a School of Chemistry, Uni6ersity of Exeter, Exeter EX4 4QD, UK
b Department of Chemistry, EPSRC National Crystallography Ser6ice, Uni6ersity of Southampton, Highfield Road, Southampton SO17 1BJ, UK
Received 18 May 2001; accepted 20 June 2001
Abstract
Silica-supported, imine-based palladacyclic catalysts have been synthesised and the crystal structure of complex 9, the
triphenylphosphine adduct of the pre-supported precursor complex 8, has been determined. The solid-supported catalysts show
considerably lower activity in the Suzuki reaction than their homogeneous counterparts. Poor recyclability of the silica-immo-
bilised catalysts and the presence of active catalysts in solution indicate that imine-based palladacyclic catalysts are unstable with
respect to liberation of zero-valent palladium species. Whilst the solid-supported complexes are not useful as catalysts, they do
function as excellent mechanistic probes. Studies on model complexes give further information on the processes that cause the
liberation of zero-valent species not only from the solid-supported catalysts, but also from homogeneous systems. In all cases it
appears that a reductive-elimination event occurs to generate the active catalyst. © 2001 Published by Elsevier Science B.V.
Keywords: Suzuki reaction; Palladacycles; Catalysis; Silica supported
1. Introduction
There has recently been considerable interest in the
synthesis of new, high-activity palladium-based cata-
lysts that can be used in low concentration in the
Suzuki reaction (Scheme 1) since such catalysts have
the potential to be used in industrial systems. In partic-
ular, palladacyclic catalysts in which a ligand coordi-
nates to the metal centre through both a donor atom
and metallated carbon have shown considerable
promise. Beller et al. demonstrated that complex 1
shows good activity [1] whilst we have shown that the
palladated triarylphosphite complex 2, the phosphinite
complexes 3 and the bis(phosphinite) PCP-pincer com-
plexes 4 show excellent activity [2–4]. High activity is
not limited to metallated phosphorus donor systems—
Weissman and Milstein have shown that the metallated
imine complex 5a shows excellent activity [5], whilst
Zim et al. have shown that the metallated thioether
complexes 6 can also be used [6].
* Corresponding author. Fax: +44-1392-263434.
E-mail address: r.bedford@ex.ac.uk (R.B. Bedford).
0022-328X/01/$ - see front matter © 2001 Published by Elsevier Science B.V.
PII: S0022-328X(01)01080-4

174
R.B. Bedford et al. / Journal of Organometallic Chemistry 633 (2001) 173–181
and their methodology was adapted for this work. The
triethoxysilyl-modified imine 7 was synthesised in 91%
yield by the reaction of 3-aminopropyltriethoxysilane
with 2-bromobenzaldehyde (Scheme 2). Imine 7 reacts
with [Pd(dba)2] in toluene to give the dimeric palladacy-
cle 8 in good yield. Complex 8 readily reacts with
triphenylphosphine to give the mononuclear adduct 9,
the structure of which was determined by single crystal
X-ray analysis. There are two molecules in the asym-
metric unit that differ significantly in the conformation
of the triethoxysilyl group. One of the two independent
molecules (molecule A) is shown in Fig. 1 and selected
data are given in Table 1. The palladium adopts an
approximately square planar geometry with the
triphenylphosphine ligand cis to the metallated carbon.
Both the average PdꢀC and PdꢀN bond lengths are
significantly higher than those reported for complex 5a
[5], presumably as a result of differences in the trans
influence of the co-ligands.
Scheme 1. The Suzuki biaryl coupling reaction.
The 31P{1H}-NMR spectrum of 9 shows a single
peak at l 43 ppm confirming that only one isomer is
present in solution. The 31P{1H}-MAS NMR spectrum
of 9 shows a singlet at l 39 ppm with a slight broaden-
ing on the low frequency side. This broadening is likely
to be due to imperfect crystallisation and/or
polymorphism.
The immobilisation of 8 was performed by heating it
,
with mesoporous silica (Merck Kieselgel, 100 A) in a
ratio of ca. 0.60 mmol Pd per gram of silica in toluene
for 17 h, to give the modified silica product 10. Pd
analysis was performed on a sample of 10 which
showed a palladium loading of 0.31 mmol g−1 indicat-
ing that ca. 52% of complex 8 has been immobilised,
the rest was recovered from the combined supernatant
solution and washings. In order to establish that the
structure of the complex had not altered on immobilisa-
tion, 10 was treated with a solution of triphenylphos-
phine to generate the immobilised complex 11. The
31P{1H}-MAS NMR spectrum of 11 shows a singlet at
l 42 ppm with a distinct shoulder to low frequency.
The shift of the main peak is very similar to those
obtained in the 31P spectra of 9 both in solution and in
the solid state. This indicates that the immobilisation
has not perturbed the coordination sphere of the palla-
dium to any great extent.
In order to evaluate the extent to which structural
features other than immobilisation affect the activity of
the catalyst 10 compared with the reported complex 5a,
the model complex 5b was synthesised in 89% yield by
the reaction of the imine 12 with [Pd(dba)2] (Scheme 3).
The activity of the supported catalysts 10 and 11 and
the model complex 5b in the Suzuki reaction (Scheme
1) was then investigated and the results are summarised
in Table 2. The model complex 5b shows excellent
activity; only slightly lower than that reported with 5a
[5], indicating that a change in the structure of the
Scheme 2. (i) H2N(CH2)3Si(OEt)3, EtOH, molecular sieves; (ii)
[Pd(dba)2], toluene, 60 °C; (iii) PPh3, CH2Cl2; (iv) silica, toluene,
reflux temperature.
Given the high activity and longevity of palladacyclic
complexes, they would appear to be ideal candidates
for recycling protocols. In particular the imine pallada-
cyclic complexes are reported as having high stability
and they should be easily modified to incorporate func-
tionalities that will allow them to be supported [7,5].
For these reasons we wished to investigate the immobil-
isation of analogues of 5a on mesoporous silica and see
whether the resultant species are viable as recyclable
catalysts in the Suzuki reaction.
2. Results and discussion
2.1. Synthesis and catalytic acti6ity of silica-supported
palladacyclic complexes
Silica-immobilised imine ligands have been reported
previously in the synthesis of recyclable chromium-
based oxidation catalysts by Clark and co-workers [8]

R.B. Bedford et al. / Journal of Organometallic Chemistry 633 (2001) 173–181
175
,
Fig. 1. One of the two independent molecules of 9 (molecule A). Selected bond lengths (A) and angles (°) for molecule A (shown): N1ꢀPd1
2.102(3); C1ꢀPd1 2.036(4); Pd1ꢀP1 2.2567(10); Br1ꢀPd1 2.5001(5); C7ꢀN1 1.282(5); C1ꢀC6 1.424(5); C6ꢀC7 1.439(6); and molecule B (not shown):
N1%ꢀPd1% 2.102(3); C1%ꢀPd1% 2.036(4); Pd1%ꢀP1% 2.2567(10); Br1%ꢀPd1% 2.5001(5); C7ꢀN1 1.282(5); C1ꢀC6 1.424(5); C6ꢀC7 1.439(6).
although no mechanism for the liberation of the palla-
dium nanoparticles was given [10].
imine ligand and/or replacement of the TFA ligand
with bromide has only a small effect on the activity. By
contrast, when the immobilised catalyst 10 is used
considerably lower activity is observed on the first
catalytic run.
By contrast with the findings that the polystyrene-
supported systems show little or no activity after the
first run [10], we find that when mesoporous silica is
used as a support considerable activity is observed over
several runs. Whether the silica in some way stabilises
the putative palladium nanoparticles, thus increasing
their longevity and hence their recyclability, or whether
it acts merely as a gradual-release supply of colloidal
palladium is not clear. Either way, some active catalyst,
regardless of its precise nature, is removed from the
In all cases where recycling of 10 was attempted a
decrease in activity was observed from one run to the
next. Looking more closely at the coupling of 4-bro-
moanisole (entry 3) it can be seen that a fairly steady
fall in activity occurs during runs 1–7 after which a
more substantial drop in activity is observed. Interest-
ingly, this coincides with a marked colour change of the
silica. During runs 1–6 the silica has a deep brown–red
colouration compared with unused 10 which is pale
yellow. By the end of run 7, the silica was grey–black,
suggesting the presence of a substantial quantity of
bulk palladium metal. One plausible explanation for the
drop in activity between runs that agrees with the
observed colour changes is that palladium nanoparticles
are being formed, which are deep red–brown, and it is
these that are the true catalytically active species. At the
end of the useful life of these nanoparticles they decom-
pose to finely divided palladium metal, which is black.
Palladium nanoparticles have previously been shown to
be highly active catalysts in the Heck reaction [9], and
a similar decomposition of a polystyrene-supported
imine palladacycle to generate active palladium colloids
in the Heck reaction has very recently been reported,
Table 1
,
Selected bond lengths (A) and bond angles (°) for 9
Molecule A
Molecule B
Bond lengths
C1ꢀPd1
N1ꢀPd1
Br1ꢀPd1
Pd1ꢀP1
2.036(4)
2.102(3)
2.5001(5)
2.2567(10)
2.030(4)
2.096(3)
2.4905(6)
2.2528(11)
Bond angles
C1ꢀPd1ꢀN1
C1ꢀPd1ꢀP1
N1ꢀPd1ꢀP1
C1ꢀPd1ꢀBr1
N1ꢀPd1ꢀBr1
P1ꢀPd1ꢀBr1
81.05(14)
94.33(11)
174.12(10)
165.71(11)
93.44(10)
91.88(3)
80.67(15)
94.75(11)
174.54(10)
167.67(10)
93.04(10)
92.00(3)

176
R.B. Bedford et al. / Journal of Organometallic Chemistry 633 (2001) 173–181
phosphine complexes have previously been implicated
as highly active homogeneous catalysts in the Suzuki
reaction and in catalytic aminations [11,12].
Obviously the solid-supported catalysts 10 and 11
show neither high enough activity nor recyclability for
them to be of any practical use in catalytic processes,
but a study of their modes of decomposition may give
vital information on the nature of the true active spe-
cies when high activity, palladacyclic species are used in
homogeneous catalysis and this is addressed below.
Scheme 3. (i) [Pd(dba)2], toluene, 70 °C; (ii) [Pd(TFA)2], THF,
50 °C.
2.2. Mechanistic studies
In order to establish the mechanism by which zero-
valent palladium species can be released from imine
palladacycles, the reactions of the model complexes 14
and 15 with the individual substrates of the Suzuki
reaction were examined. Whilst it showed no reactivity
with 4-bromoacetophenone, heating 14 with excess
phenylboronic acid and potassium carbonate in toluene
for 18 h led to substantial deposition of palladium
black. Hydrolysis of the product mixture followed by
GC and GC–MS analysis showed the formation of
2-phenylbenzaldehyde 16, and its decarbonylation
product, biphenyl, in a combined yield of ꢀ18% [13].
By contrast, when the same reaction was performed
with the triphenylphosphine adduct 15, substantial de-
position of palladium was observed even at room tem-
perature and the combined yield of 16 and biphenyl at
the end of the reaction was ꢀ51%. A plausible mecha-
nism that accounts for the presence of 16 in these two
reactions is shown in Scheme 4. Nucleophilic attack by
the phenyl group from the boronate species on the
palladium centres of 14 and 15 in a manner essentially
identical to that invoked in a ‘classical’ Suzuki mecha-
nism [14], generates the putative intermediates 17. Re-
ductive elimination of the metallated imine ligand with
the phenyl group generates the coupled imine 18–which
silica, demonstrated by the observation that filtered
solutions of reactions catalysed by 10 still show consid-
erable activity. For instance when the coupling of 4-
bromoanisole with phenylboronic acid catalysed by 10
is stopped after 6 h, 69% conversion to 4-methoxy-
biphenyl is observed; removal of the silica by filtration,
addition of more phenylboronic acid and base and
heating for a further 19 h gives an extra 11% of the
coupled product. The fact that active catalytic species
are removed from the silica support demonstrates that
far from being stable, the metallated imine function is
highly reactive and labile.
Initially the catalyst 11 shows much higher activity
than 10. However, in contrast to the results obtained
with 10, attempts to recycle the triphenylphosphine-
modified catalyst 11 were unsuccessful with essentially
complete loss of activity being observed between the
first and second runs (entry 7). This implies that the
active catalyst in this process is completely removed
from the silica in the first catalytic run, whereas the
silica 10 seems to show a gradual release of active
catalyst over several runs. One explanation for this is
that very low coordinate, homogeneous species are
liberated from 11, stabilised by the coordination of
triphenylphosphine. Very low coordinate palladium
Table 2
Suzuki coupling of aryl bromides with phenylboronic acid catalysed by imine palladacycles
Entry Substrate
Catalyst (mol% Pd) Method [run] Conversion (%) a
TON (mol product/mol
Pd)
1
2
4-Bromoacetophenone 5b (0.0001)
4-Bromoacetophenone 10 (0.0017) b
A
A
54
540 000
Run 1: 27 650; total:
54 100
[1] 47, [2] 8, [3] 12, [4] 5, [5] 15, [6] 3, [7] 2
3
4
5
6
4-Bromoanisole
10 (0.31)
A
B
B
A
[1] 85, [2] 66, [3] 58, [4] 38, [5] 27, [6] 24, [7] 17, [8] 4 Run 1: 274; total: 1029
4-Bromoacetophenone 10 (0.031) c
91
2935
4-Bromoacetophenone 11 (0.0013) b
4-Bromoanisole
100
75 400
11 (0.0013) b
[1] 63, [2] B0.5
Run 1: 48 500
Reaction conditions, method A: 2 mmol aryl bromide, 3 mmol PhB(OH)2, 4 mmol K2CO3, 20 ml toluene, 130 °C, 17 h. Method B: 10 mmol aryl
bromide, 11.5 mmol PhB(OH)2, 15 mmol K2CO3, 30 ml toluene, 130 °C, 17 h.
a Determined by GC, based on aryl bromide.
b Catalyst diluted with silica-total mass 0.05 g.
c Catalyst diluted with silica-total mass 1.00 g.

R.B. Bedford et al. / Journal of Organometallic Chemistry 633 (2001) 173–181
177
enhancement of the rate of reductive elimination
brought about by the phosphine ligand, as seen in the
reactions of the model complexes 14 and 15 with Ph-
B(OH)2. This helps explain why catalyst 11 shows no
recyclability—all of the active catalyst species are liber-
ated into homogeneous solution during the first run.
3. Conclusions
In summary, whilst the silica-supported catalysts 10
and 11 show only poor recyclability, their modes of
decomposition give vital information on the likely na-
ture of the true active catalysts in the Suzuki reaction
catalysed by all high-activity imine palladacyles. In all
cases it seems that Pd(0) species are formed via a
reductive-elimination process, preceded by nucleophilic
attack of the phenyl boronate complex at the palladium
centre [14]. We would expect this to be favoured by
increased steric bulk around the metal centre and this is
indeed observed. The presence of large amounts of
reductively eliminated organic by-products indicates the
concomitant production of an equal amount of palla-
dium(0) species which almost certainly act as the true
active catalyst(s). This mechanism obviously applies
equally well to the previously reported complex 5a [5].
Preliminary findings in our group indicate that the
orthopalladated phosphinite complexes 3 may also un-
dergo a similar initiation process [4], whilst Hartwig has
shown a related process probably operates in Stille and
amination reactions catalysed by 1 [16]. Combined,
these results suggest that all high-activity palladacyclic
complexes may in fact act as precursors to either low-
coordinate palladium species or palladium nanoparti-
cles, depending on the coordination strength of the
donor group left after reductive elimination of the aryl
function. These results go some way to resolving the
current debate as to whether CꢀC coupling reactions
catalysed by palladacycles proceed via a Pd(0)/Pd(II) or
a Pd(II)/Pd(IV) pathway [17].
Scheme 4. (i) PPh3, CH2Cl2; (ii) five equivalents PhB(OH)2, 7.5
equivalents K2CO3, toluene, reflux temperature; (iii) HCl(aq), reflux
temperature.
subsequently undergoes hydrolysis during work–up
and palladium(0) species.
It is highly likely that the phenylation and reductive-
elimination steps outlined in Scheme 4 also occur as
initiation steps in the catalytic cycles of Suzuki reac-
tions catalysed by imine palladacycles and that the
substantial quantities of palladium(0) species generated
during this process act as the true active catalysts. One
consequence of such a process acting as an initiation
step would be that the eliminated imine ligand would
be left on the silica wall. Subsequent hydrolysis of these
imines in the presence of adventitious water [15] would
result in ‘dangling’ amines which could quaternarise to
give ammonium salts that may help to stabilise palla-
dium nanoparticles in the silica, since alkylammonium
salts are known to stabilise catalytically active palla-
dium colloids [9].
Whilst they may be active, the putative silica-sta-
bilised nanoparticles cannot be the only active catalysts
as activity is observed in filtered solutions of reactions
stopped part way through a catalytic run. The possibil-
ity also exists that any palladium nanoparticles formed
are not stabilised by the silica, but instead are produced
directly as a colloidal suspension at a steady rate by the
decomposition route outlined above, however, this does
not account for the observed colour of the silica 10
during recycling.
4. Experimental
4.1. General
All procedures were performed under an atmosphere
of dry nitrogen unless stated otherwise using standard
Schlenk line techniques. Solvents were distilled from
appropriate drying reagents before use. NMR spectra
were recorded on a Bruker AC300 or on a Bruker
DRX400 Avance spectrometer. 31P{1H}-MAS NMR
spectra were recorded on Bruker MSL400. GC analysis
was performed on a Varian GC 3800 fitted with a 25 m
CP-SIL 5CB column. GC–MS was performed on a
ThermoQuest trace GC–MS fitted with 15 m Restek
By contrast when 11 is used, it is far more likely that
very low-coordinate palladium–phosphine species enter
solution after reductive elimination of the organic func-
tions from the metal centre. Such species have been
implicated previously as catalysts in coupling reactions
[12]. The liberation of active catalysts from 11 would be
expected to be fast compared with 10 due to the

178
R.B. Bedford et al. / Journal of Organometallic Chemistry 633 (2001) 173–181
solution and addition of EtOH, followed by cooling to
−10 °C led to the precipitation of complex 9 as a
yellow powder, which was then recrystallised from
RTX-5MS column, operating in EI mode. [Pd(dba)2]
[18], [Pd(TFA)2] [19], N-(2-bromo)benzylidene
propylamine (12) [20], N-benzylidene isopropylamine
(13) [21], were prepared according to literature meth-
ods. All other materials were purchased and used as
received.
1
CH2Cl2–EtOH (0.362 g, 56%). H-NMR (CDCl3): l=
3
0.60 (br m, 2H, SiꢀCH2), 1.23 (t, J(HH)=7.0 Hz, 9H,
CH3), 2.03 (br m, 2H, CH2ꢀCH2ꢀCH2), 3.84 (q,
3J(HH)=7.0 Hz, 6H, OꢀCH2), 4.09 (br m, 2H,
3
NꢀCH2), 6.37 (br dd, 1H, H3), 6.53 (ddd, J(H4H3)=
4.2. Syntheses
7.6 Hz, 3J(H4H5)=7.5 Hz, 4J(H4H6)=1.5 Hz, 1H,
3
3
H4), 6.91 (ddd, J(H5H4)=7.5 Hz, J(H5H6)=7.4 Hz,
4.2.1. N-(2-Bromo)benzylidene(3-triethoxysilyl)-
propylamine (7)
To a mixture of 3-aminopropyltriethoxysilane (10.0
4J(H5H3)=1.1 Hz, 1H, H5), 7.26 (dd, J(H6H5)=7.4
3
Hz, J(H6H4)=1.5 Hz, 1H, H6), 7.35, 7.74 (m, 15H,
4
PPh3), 8.13 (br d, 4J(PH)=7.5 Hz, 1H, NꢁCH).
31P{1H}-NMR (CDCl3): l=43.3 (s). 31P{1H}-NMR
MAS: l=39 (br s). Anal. Found: C, 53.4; H, 5.4; N,
1.7. Calc. for C34H41BrNO3PPdSi: C, 53.9; H, 5.5; N,
1.85%.
ml, 42.73 mmol) in EtOH (30 ml), 2-bromobenzalde-
,
hyde (4.96 ml, 42.81 mmol) and 4 A molecular sieves (5
g) were added. The reaction mixture was stirred at
room temperature (r.t.) overnight and then filtered
through a plug of celite. The solvent was removed in
vacuo giving 7 as a pale yellow oil which was suffi-
ciently pure for subsequent reactions (15.06 g, 91%).
1H-NMR (CDCl3): l=0.68 (m, 2H, SiꢀCH2), 1.22 (t,
3J(HH)=7.0 Hz, 9H, CH3), 1.83 (m, 2H,
4.2.4. Silica-immobilised catalyst (10)
A mixture of 8 (0.594 g, 0.60 mmol) and mesoporous
,
silica (Merck Kieselgel, 100 A, 2.002 g) was heated in
3
4
CH2ꢀCH2ꢀCH2), 3.66 (dt, J(HH)=6.9 Hz, J(HH)=
toluene (25 ml) at reflux temperature for 17 h. The
solution was allowed to cool to r.t. and the supernatant
liquid was removed with a syringe. The pale yellow,
modified silica product 10 was washed with toluene
(7×20 ml) and CH2Cl2 (2×20 ml) and dried in vacuo.
(Palladium analysis found: 3.30% m/m Pd, RSD=
5.77%.)
3
1.1 Hz, 2H, NꢀCH2), 3.82 (q, J(HH)=7.0 Hz, 6H,
OꢀCH2), 7.21 (ddd, J(H4H3)=8.0 Hz, J(H4H5)=7.6
3
3
Hz, J(H4H6)=2.0 Hz, 1H, H4), 7.31 (m, 1H, H5), 7.54
4
(dd, J(H3H4)=8.0 Hz, J(H3H5)=1.4 Hz, 1H, H3),
3
4
7.99 (dd, J(H6H5)=7.7 Hz, J(H6H4)=2.0 Hz, 1H,
H6), 8.62 (br s, 1H, NꢁCH). Anal. Found: C, 49.0; H,
6.7; N, 3.5. Calc. for C16H26BrNO3Si: C, 49.5; H, 6.75;
N, 3.6%.
3
4
4.2.5. Silica-immobilised catalyst (11)
A mixture of 10 (0.935 g, 0.290 mmol Pd) and
triphenylphosphine (0.127 g, 0.48 mmol) in CH2Cl2 (30
ml) was stirred at r.t. for 14 h. The supernatant liquid
was removed with a syringe and the resultant pale
yellow silica 11 was washed with CH2Cl2 (5×20 ml)
and then dried in vacuo. (Palladium analysis found:
2.79% m/m Pd, RSD=2.64%); 31P{1H}-NMR MAS:
l=42 (br s).
4.2.2. [{Pd(v-Br){s2-C,NꢀC6H4-2-(CHꢁNC3H6Si-
(OEt)3)}}2] (8)
A solution of [Pd(dba)2] (1.314 g, 2.28 mmol) and 7
(1.552 g, 4.00 mmol) in 40 ml of toluene (40 ml) was
stirred at 60 °C for 17 h, allowed to cool to r.t. and
then filtered through celite. The solvent was removed in
vacuo and the residue was recrystallised from CH2Cl2–
hexane to give 8 in ꢀ80% yield contaminated with dba
(ꢀ20%). Exhaustive washing with hexane yielded pure
4.2.6. [{Pd(v-Br){s2-C,NꢀC6H4-2-(CHꢁNPr)}2] (5b)
A mixture of N-(2-bromo)benzylidene propylamine
(0.978 g, 4.32 mmol) and [Pd(dba)2] (2.258 g, 3.93
mmol) in toluene (60 ml) was warmed to 70 °C for 5 h,
the resultant solution was filtered through celite and
then the solvent was removed in vacuo. The resultant
solid was purified by column chromatography (silica,
hexane–EtOAc 3:1) to give 5b as a pale yellow solid
(1.167 g, 89%). 1H-NMR (CDCl3): l=0.95 (t,
3J(HH)=7.5 Hz, 3H, CH3), 1.89 (m, 2H, CH3ꢀCH2),
3.65 (t, 3J(HH)=7.0 Hz, 2H, NꢀCH2), 7.04 (br m, 2 H,
H4, H5), 7.22 (br m, 1H, H6), 7.56 (br m, 1H, H3), 7.84
(br s, 1H, NꢁCH). Anal. Found: C, 36.3; H, 3.6; N, 3.9.
Calc. for C20H24Br2N2Pd2: C, 36.1; H, 3.6; N,
4.2%.
1
8 (0.169 g, 15%). H-NMR (CDCl3): l=0.64 (br m,
3
2H, SiꢀCH2), 1.23 (t, J(HH)=7.0 Hz, 9H, CH3), 2.00
3
(m, 2H, CH2ꢀCH2ꢀCH2), 3.69 (t, J(HH)=6.8 Hz, 2H,
3
NꢀCH2), 3.84 (q, J(HH)=7.0 Hz, 6H, OꢀCH2), 7.04
(m, 2H, H4, H5), 7.21 (dd, 3J(H6H5)=6.9 Hz,
4J(H6H4)=2.0 Hz, 1H, H6), 7.56 (br d, 1H, H3), 7.86
(s, 1H, NꢁCH). Anal. Found: C, 38.2; H, 5.2; N, 2.6.
Calc. for C32H52Br2N2O6Pd2Si2: C, 38.8; H, 5.3; N,
2.8%.
4.2.3. [{PdBr{s2-C,NꢀC6H4-2-(CHꢁNC3H6Si-
(OEt)3)}(PPh3)] (9)
A
solution of 8 (0.425 g, 0.43 mmol) and
triphenylphosphine (0.229 g, 0.87 mmol) in CH2Cl2 (25
ml) was stirred at r.t. for 16 h. Concentration of the

R.B. Bedford et al. / Journal of Organometallic Chemistry 633 (2001) 173–181
179
4.2.7. [{Pd(v-TFA){s2-C,NꢀC6H4-2-(CHꢁNiPr)}2]
(14)
A solution of N-benzylidene isopropylamine (1.316 g,
4.3.2. Catalysis with 10 and 11
4.3.2.1. Method A. The reactions were performed in a
modified three-necked round bottom flask where one of
the side necks had the female joint removed and re-
placed by a fritted male joint. The modified flask was
charged with the appropriate amount of 10 or 11 either
undiluted, or mixed with an appropriate amount of
8.93 mmol) and [Pd(TFA)2] (2.125 g, 6.39 mmol) in
THF (80 ml) was stirred at 50 °C for 20 min, allowed
to cool to r.t. and then filtered through celite. The
solvent was removed on rotary evaporator and the
residue was recrystallised from CH2Cl2–EtOH to give
1
14 as a green solid (1.762 g, 75%). H-NMR (CDCl3):
,
silica (Merck Kieselgel, 100 A — see Table 2 for
l=0.70 (d, 3J(HH)=6.6 Hz, 3H, CH3), 1.24 (d,
3J(HH)=6.6 Hz, 3H, CH3), 3.30 (m, 1H, CHꢀCH3),
6.92 (m, 1H, aromatic proton), 7.08 (m, 3H, aromatic
protons), 7.39 (s, 1H, NꢁCH). Anal. Found: C, 39.25;
H, 3.2; N, 3.7. Calc. for C24H24F6N2O4Pd2: C, 39.4; H,
3.3; N, 3.8%.
dilutions). The flask was flushed with nitrogen, then the
appropriate aryl bromide (2.00 mmol), phenylboronic
acid (0.366 g, 3.00 mmol), K2CO3 (0.553 g, 4.00 mmol)
and toluene (20 ml) were added. The mixture was
heated at 130 °C (external temperature) under nitrogen
for 17 h. The reaction mixture was cooled in an ice-
bath, then filtered off through the frit and the catalyst-
containing residue washed with CH2Cl2 (20 ml),
distilled water (3×20 ml), acetone (3×20 ml), CH2Cl2
(20 ml) and then dried in vacuo to give the catalyst
ready for re-use. The solvent was removed from the
combined organic extracts on a rotary evaporator and
then toluene (10 ml) was added and the conversion was
determined by GC analysis (hexadecane standard).
4.2.8. [{Pd(TFA){s2-C,NꢀC6H4-2-(CHꢁNiPr)}(PPh3)]
(15)
A
solution of 14 (0.500 g, 0.68 mmol) and
triphenylphosphine (0.398 g, 1.5 mmol) in CH2Cl2 (30
ml) was stirred at r.t. for 30 min and then filtered
through celite. Addition of EtOH (50 ml) and concen-
tration on rotary evaporator gave a solid which was
recrystallised from CH2Cl2–EtOH to give 15 as a pale
yellow solid (0.712 g, 83%). 1H-NMR (CDCl3): l=1.37
(d, 3J(HH)=6.6 Hz, 6H, CH3), 4.07 (br m, 1H,
4.3.2.2. Method B. A Schlenk tube was charged with the
appropriate amount of either 10 or 11 mixed with an
appropriate amount of silica (Merck Kieselgel, 100
3
CHꢀCH3), 6.45 (br dd, 1H, H3), 6.56 (ddd, J(H4H3)=
7.7 Hz, 3J(H4H5)=7.5 Hz, 4J(H4H6)=1.1 Hz, 1H,
H4), 6.94 (ddd, 3J(H5H4)=7.5 Hz, 3J(H5H6)=7.4,
,
A — see Table 2 for dilutions). The tube was evacu-
ated and then flushed with nitrogen three times, then
4-bromoacetophenone (1.990 g, 10.00 mmol), phenyl-
boronic acid (1.402 g, 11.50 mmol), K2CO3 (2.764 g,
20.00 mmol) and toluene (30 ml) were added. The
mixture was then heated at 130 °C (external tempera-
ture) under nitrogen for 17 h. The reaction mixture was
cooled in an ice-bath and then filtered through celite
which was washed with toluene (5×20 ml) and EtOAc
(2×20 ml). The solvents were removed on a rotary
evaporator. To the resulting solid, water was added
(100 ml) and the mixture was stirred for 30 min. Then,
the solid was collected by filtration, washed with water
(5×20 ml) and dissolved in toluene (40 ml). The solu-
tion was dried over MgSO4 and then analysed by GC
(hexadecane standard).
3
4J(H5H3)=0.8 Hz, 1H, H5), 7.29 (dd, J(H6H5)=7.4
4
Hz, J(H6H4)=1.1 Hz, 1H, H6), 7.43 (m, 9H, PPh3),
7.79 (m, 6H, PPh3), 8.15 (dd, J=0.9 Hz, J=7.9 Hz,
1H, NꢁCH). 31P{1H}-NMR (CDCl3): l=41.15 (s).
Anal. Found: C, 56.7; H, 4.2; N, 2.1. Calc. for
C30H27F3NO2PPd: C, 57.4; H, 4.3; N, 2.2%.
4.3. Catalysis
4.3.1. Catalysis with 5b
A solution of 5b (2.0×10−6 mmol) in toluene (2 ml),
obtained by multiple dilutions, was put into a Schlenk
tube, the solvent removed in vacuo and the tube was
evacuated and then flushed with nitrogen three times.
Then 4-bromoacetophenone (0.398 g, 2.00 mmol),
phenylboronic acid (0.366 g, 3.00 mmol), K2CO3 (0.553
g, 4.00 mmol) and toluene (20 ml) were added. The
mixture was then heated at 130 °C (external tempera-
ture) under nitrogen for 17 h, then cooled in an ice-bath
and poured into brine (40 ml) and toluene (30 ml) was
added. The two phases were separated and the organic
layer was washed with brine (2×40 ml). The aqueous
phase was extracted with CH2Cl2 (3×30 ml). The
organic phases were combined, dried over MgSO4,
filtered through celite and the solvent removed on a
rotary evaporator. Toluene (10 ml) was added and the
conversion was determined by GC analysis (hexadecane
standard).
4.3.3. Demonstration of presence of acti6e catalyst
remo6al from silica 10
A mixture of 10 (0.020 g, 0.0062 mmol Pd), 4-bro-
moanisole (0.25 ml, 2.00 mmol), phenylboronic acid
(0.366 g, 3.00 mmol) and K2CO3 (0.553 g, 4.00 mmol)
were heated at 130 °C (external temperature) in toluene
(20 ml) for 6 h. The solution was cooled with an
ice-bath, filtered through celite which was washed with
toluene (30 ml) and hexadecane (2 ml, internal stan-
dard) was added to the combined solution and wash-
ings. GC analysis of the solution showed a 68%
conversion of the 4-bromoanisole to 4-methoxy-

180
R.B. Bedford et al. / Journal of Organometallic Chemistry 633 (2001) 173–181
biphenyl. To the solution more phenylboronic acid
(0.366 g, 3.00 mmol) and K2CO3 (0.553 g, 4.00 mmol)
were added and the solution was heated at 130 °C
(external temperature) for a further 19 h. The solution
was cooled in an ice-bath and then poured into HCl
(1.18 sg, 40 ml). The organic layer was separated,
washed with water (3×40 ml), dried over MgSO4 and
then filtered through celite. GC analysis revealed an
tions. The final R1(F2) was 0.0419 [F2\2|(F2)] and
wR(F2) was 0.1045 (all data).
5. Supplementary material
Crystallographic data for the structural analysis have
been deposited with the Cambridge Crystallographic
Data Centre, CCDC no. 153753 for compound 9.
Copies of this information may be obtained free of
charge from The Director, CCDC, 12 Union Road,
Cambridge CB2 1EZ, UK (Fax: +44-1223-336033;
e-mail: deposit@ccdc.cam.ac.uk or www: http://www.
ccdc.cam.ac.uk).
80%
conversion
of
4-bromoanisole
to
4-
methoxybiphenyl.
4.3.4. Decomposition studies
A mixture of either complex 14 or 15 (0.31 mmol
Pd), phenylboronic acid (0.194 g, 1.59 mmol) and
K2CO3 (0.330 g, 2.39 mmol) in toluene (25 ml) was
heated at 130 °C (external temperature) for 18 h. The
solution was cooled in an ice-bath, poured in HCl (1.18
sg, 30 ml) and then stirred vigorously at 70 °C for 3 h.
The two phases were separated and the organic layer
was washed with water (3×30 ml), dried over MgSO4,
filtered through celite and the solvent was removed on
rotary evaporator. Toluene (4 ml) and mesitylene (1 ml,
internal standard) were added and the mixture analysed
by GC–MS and GC. Both techniques indicated the
presence of 2-phenylbenzaldehyde and biphenyl by
comparison with authentic samples. The combined
yields of 2-phenylbenzaldehyde and biphenyl were ca.
18% (for 12) and 51% (for 13).
Acknowledgements
We thank the University of Exeter for a student
bursary (C.S.J.C.) and Johnson Matthey Chemicals for
palladium analysis. The MAS NMR measurements
were supported by EPSRC grant no. GR/M12209.
References
8
[1] M. Beller, H. Fischer, W.A. Herrmann, K. Ofele, C. Brossmer,
Angew. Chem. Int. Ed. Engl. 34 (1995) 1848.
[2] D.A. Albisson, R.B. Bedford, S.E. Lawrence, P.N. Scully, Chem.
Commun. (1998) 2095.
4.4. Crystal structure determination of complex 9
[3] R.B. Bedford, S.M. Draper, P.N. Scully, S.L. Welch, New J.
Chem. (2000) 745.
Suitable crystals of 9 were grown from CH2Cl2–
EtOH. Cell dimensions and intensity data were
recorded at 150 K, using a Nonius KappaCCD area
detector diffractometer mounted at the window of a
rotating anode operating at 50 kV, 90 mA with a
[4] R.B. Bedford, S.L. Welch, Chem. Commun. (2001) 129.
[5] H. Weissman, D. Milstein, Chem. Commun. (1999) 1901.
[6] D. Zim, A.S. Gruber, G. Ebeling, J. Dupont, A.L. Monteiro,
Org. Lett. 2 (2000) 2881.
[7] M. Ohff, A. Ohff, D. Milstein, Chem. Commun. (1999) 357.
[8] I.C. Chisem, J. Rafelt, M.T. Shieh, J. Chisem, J.H. Clark, R.
Jachuck, D.J. Macquarrie, C. Ramshaw, K. Scott, Chem. Com-
mun. (1998) 1949.
[9] M.T. Reetz, E. Westermann, Angew. Chem. Int. Ed. Engl. 39
(2000) 165.
[10] M. Nowotny, U. Hanefeld, H. Van Konigsveld, T. Maschmeyer,
Chem. Commun. (2000) 1877.
,
molybdenum anode (u (Mo–Ka)=0.71073 A). The
crystal-to-detector distance was 30 mm and ƒ- and
d-scans with a k offset were employed to access a
complete set of unique reflections (2° increments, 10 s
exposure time). Data collection and processing were
carried out using DIRAX [22], DENZO [23] and MAXUS
[24] run from within Nonius COLLECT [25] program and
an empirical absorption correction was applied using
SORTAV [26,27]. The structure was solved via direct
methods [28] and refined by full-matrix least-squares
[29] on F2. Non-hydrogen atoms were refined an-
isotropically and hydrogen atoms were treated using a
riding model. Crystal data: C34H41BrNO3PPdSi, M=
[11] J.P. Wolfe, R.A. Singer, B.H. Yang, S.L. Buchwald, J. Am.
Chem. Soc. 121 (1999) 9550.
[12] (a) F. Paul, J. Patt, J.F. Hartwig, J. Am. Chem. Soc. 116 (1994)
5969;
(b) J.F. Hartwig, F. Paul, J. Am. Chem. Soc. 117 (1995) 5373.
[13] Shorter reaction times led to higher 16:biphenyl ratios, demon-
strating that the presence of biphenyl was due to a palladium-
mediated decarbonylation of 16. Further, heating 16 with a
catalytic amount of palladium acetate led to catalytic decarbony-
lation giving biphenyl.
[14] This step is usually, misleadingly described as ‘trans-metallation’.
[15] Removal of all the water from the silica without destruction of
the supported complex is not possible. In addition, substantial
quantities of water are introduced during washing after each
catalytic run.
757.05, monoclinic, a=26.5989(6), b=14.9407(3), c=
3
,
,
17.5024(5) A, i=102.531(4)°, U=6789.9(3) A ,
T=150 K, space group P21/c (no. 14), Z=8 (the two
molecules in the asymmetric unit differ significantly in
the conformation of the triethoxysilyl group), v (Mo–
Ka)=1.84 mm−1, 32 235 reflections measured, 11 859
unique (Rint=0.0529) which were used in all calcula-
[16] J. Louie, J.F. Hartwig, Angew. Chem. Int. Ed. Engl. 35 (1996)
2359.

R.B. Bedford et al. / Journal of Organometallic Chemistry 633 (2001) 173–181
181
[17] (a) G.T. Crisp, Chem. Soc. Rev. 27 (1998) 427 (and references
therein);
raphy, Part A, vol. 276, Academic Press, New York, 1997, pp.
307–326.
(b) B.L. Shaw, S.D. Perera, E.A. Staley, Chem. Commun. (1998)
1361.
[18] M.F. Rettig, P.M. Maitlis, Inorg. Synth. 28 (1990) 110.
[19] T.A. Stephenson, S.M. Morehouse, A.R. Powell, J.P. Heffer, G.
Wilkinson, J. Chem. Soc. (1965) 3632.
[20] C.M. Anderson, M. Crespo, M.C. Jennings, A.J. Lough, G.
Ferguson, R.J. Puddephatt, Organometallics 10 (1991) 2672.
[21] J.-M. Me´lot, F. Texier-Boullet, A. Foucaud, Synthesis (1998)
558.
[22] A.J.M. Duisenberg, J. Appl. Crystallogr. 25 (1992) 92.
[23] Z. Otwinowski, W. Minor, Processing of X-ray diffraction data
collected in oscillation mode, in: C.W. Carter Jr., R.M. Sweet
(Eds.), Methods in Enzymology. In: Macromolecular Crystallog-
[24] S. Mackay, C.J. Gilmore, C. Edwards, M. Tremayne, N. Stuart,
K. Shankland, MAXUS: a Computer Program for the Solution
and Refinement of Crystal Structures from Diffraction Data,
University of Glasgow, Scotland, UK, Nonius BV, Delft, The
Netherlands and MacScience Co. Ltd, Yokohama, Japan,
1998.
[25] R. Hooft, B.V. Nonius, COLLECT: Data Collection Software,
1998.
[26] R.H. Blessing, Acta Crystallogr. A 51 (1995) 33.
[27] R.H. Blessing, J. Appl. Crystallogr. 30 (1997) 421S.
[28] G.M. Sheldrick, Acta Crystallogr. A 46 (1990) 467.
[29] G.M. Sheldrick, University of Go¨ttingen, Go¨ttingen, Germany,
1997.
.