Bis(pyridyl)siloxane-Pd(II) complex catalyzed oxidation of alcohol to aldehyde: Effect of ligand tethering on catalytic activity and deactivation behavior

Article abstract of DOI:10.1016/j.apcata.2010.09.008

A series of oligomeric methylsiloxane compounds functionalized with pyridyl groups was synthesized and used as ligands in the aerobic Pd(OAc) ₂-catalyzed oxidation of benzyl alcohol to benzaldehyde at 353 K. The effect of tethering two pyridine ligands at the terminal positions of linear siloxanes of varying length was systematically investigated, as was the effect of the attachment point of the pyridine ring (meta or para). It was found that meta-substituted pyridylsiloxanes were generally more effective in protecting the catalyst against Pd agglomeration. For this purpose, the optimal meta-pyridylsiloxane ligands had 4-6 silicon atoms or 10 silicon atoms for para-pyridylsiloxane ligands. The metal-ligand binding properties were used to interpret the catalytic behavior, and the ability of the catalyst stability correlated with ability of the bis-pyridyl ligand to form a mononuclear cyclic complex with Pd.

Full text of DOI:10.1016/j.apcata.2010.09.008

Applied Catalysis A: General 391 (2011) 297–304
Contents lists available at ScienceDirect
Applied Catalysis A: General
journal homepage: www.elsevier.com/locate/apcata
Bis(pyridyl)siloxane–Pd(II) complex catalyzed oxidation of alcohol to aldehyde:
Effect of ligand tethering on catalytic activity and deactivation behavior
Michael N. Missaghi, John. M. Galloway, Harold H. Kung∗
Department of Chemical and Biological Engineering, Northwestern University, Evanston, IL 60208, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
A series of oligomeric methylsiloxane compounds functionalized with pyridyl groups was synthesized
and used as ligands in the aerobic Pd(OAc)2-catalyzed oxidation of benzyl alcohol to benzaldehyde at
353 K. The effect of tethering two pyridine ligands at the terminal positions of linear siloxanes of varying
length was systematically investigated, as was the effect of the attachment point of the pyridine ring (meta
or para). It was found that meta-substituted pyridylsiloxanes were generally more effective in protecting
the catalyst against Pd agglomeration. For this purpose, the optimal meta-pyridylsiloxane ligands had 4–6
silicon atoms or 10 silicon atoms for para-pyridylsiloxane ligands. The metal–ligand binding properties
were used to interpret the catalytic behavior, and the ability of the catalyst stability correlated with ability
of the bis-pyridyl ligand to form a mononuclear cyclic complex with Pd.
Received 27 February 2010
Received in revised form 3 August 2010
Accepted 16 September 2010
Available online 29 September 2010
Keywords:
Partial oxidation
Pd(II) catalysis
Ligand tethering
Pyridylsiloxane
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
improve the stability of this catalyst is to protect the sensitive
Pd0 intermediate against agglomeration, while maintaining its
Palladium acetate Pd(OAc)2-catalyzed aerobic oxidation of alco-
hols to carbonyl compounds, first reported in 1998, is an appealing
method to replace environmentally unfriendly oxidants with
molecular oxygen in processes of commercial importance [1].
Among the Pd(OAc)2 systems studied, the Pd(OAc)2/pyridine (py)
catalyst system developed by Uemura and coworkers were found
to be practical for preparative synthetic work [2]. The many deriva-
tive studies based on this catalyst system have been summarized in
several reviews [3–5]. Of particular interest are approaches which
enable the catalysis to proceed at room temperature and/or under
1 atm of air [6–8], and successful efforts to introduce enantiose-
lectivity to the active site using a chiral ligand [6–8]. A common
concern regarding these systems is the stability of the catalyst –
in particular, the tendency of soluble Pd to aggregate into metallic
nanoparticles with loss of activity.
Mechanistic studies on this catalyst system suggest that the
alcohol oxidation takes place through a palladium alkoxide inter-
mediate which reductively eliminates, along with two equivalents
of AcOH, to give a (py)2Pd0 intermediate [9–10]. This zero-valent
Pd species is readily attacked by O2 to form either a peroxo or a
hydroperoxide species which reacts with AcOH to regenerate the
cationic Pd resting state, releasing H2O2 [11]. However, it could
also form a Pd–Pd bond with another complex, release the py
ligands, and form Pd agglomerates. Therefore, one approach to
reactivity towards O2. Typically this can be done by using excess
py ligand to shift the dissociation equilibrium towards the bound
state. While effective, the presence of excess py substantially
decreases the observed catalytic activity, and also necessitates the
removal of the excess ligand from the reaction products.
Tethering two pyridyl ligands with a flexible spacer offers an
alternative to the use of excess pyridine, since an increased local
concentration of pyridyl species may stabilize the active Pd center
without resorting to using excess ligands. In addition, such ligands
can be designed to tune the active site environment by adjusting
the length and flexibility of the tether. While a variety of oligomeric
spacers are available for this purpose, we have investigated lin-
ear methylsiloxanes in this work due to the unique conformational
properties of the Si–O–Si linkage. In particular, we suppose that
the highly flexible Si–O backbone can accommodate the rigid and
sterically demanding trans square planar geometry of coordination
complexes of PdII [12].
We investigated the dependence of the catalytic properties on
the separation distance between the pyridyl groups in the ligands,
the attachment point of Si to the pyridine ring, the Pd concentration,
the ligand–metal ratio, and the substrate–catalyst ratio. The results
of our investigation are reported here.
2. Experimental
2.1. Ligands
∗
Scheme 1 shows the ligands used and their designation.
Details of the preparation and purification of the bis-(meta-
Corresponding author.
E-mail address: hkung@northwestern.edu (H.H. Kung).
0926-860X/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2010.09.008

298
M.N. Missaghi et al. / Applied Catalysis A: General 391 (2011) 297–304
Scheme 1. Structures of pyridylsiloxane (spy) ligands studied in this work.
pyridyl)siloxane ligands have been reported [13]. Those for the
bis-(para-pyridyl)siloxane ligands can be found in the Appendix.
They are low- to medium-viscosity oils at room temperature,
and were found by 1H NMR and elemental analysis to be free of
detectable impurities. In order to distinguish siloxane-substituted
pyridines from unsubstituted pyridine py, the former will be
labeled spy. In this paper, the ligand–metal ratios py:Pd or spy:Pd
were calculated on the basis of total pyridyl groups, noting that
each bis(pyridyl)siloxane ligand contributes two pyridyl groups.
were removed from the heating block and examined visually for
coloration or darkening associated with Pd black formation. In the
majority of cases, the reaction mixture was clear and colorless to
light yellow after reaction; exceptions are noted in the text.
2.3. Determination of effect of stirring rate on reaction rates
The potential effect of O2 dissolution on reaction kinetics was
examined by investigating the effect of stirring rate using a reac-
tion mixture containing 0.1 M BnOH, 5 mM Pd(OAc)2, and pyridine
(py:Pd = 2) at 353 K. Pyridine was used since its complexes were
the most active. It was found that doubling the stirring rate from
400 to 800 rpm increased the reaction rate at short reaction times
(2–6 min) by less than 5%. However, halving the stirring rate from
400 to 200 rpm decreased the conversion at 2 min by nearly a quar-
ter and at 6 min by half. At the same time, Pd black appeared in the
reaction mixture within 1.5 min. It was also found that vials placed
farther from the center of the stirring plate experienced better mix-
ing at all stirring rates, due to eccentric motion of the stirring bar,
but the conversion varied by a maximum of 10% for different vial
positions under the conditions listed above. Since most of the kinetic
experiments had a much smaller peak oxygen demand, these obser-
vations suggested that the reaction rates were at most marginally
affected by O2 dissolution rate.
2.2. Catalysis experiments
Catalytic reactions were carried out in the batch mode in the
liquid phase at 353 K under an atmospheric pressure of O2. Freshly
prepared (<1 week old) stock solutions of 25 mM Pd(OAc)2 (Aldrich,
99.9% grade) in toluene, stored in the dark, was used as the Pd
source. Stock solutions of benzyl alcohol (BnOH) in toluene (0.250,
1.00, and 4.00 M), stored in tightly closed vials, and stock solutions
of ligand (0.8 N of spy in toluene) were also prepared. The latter was
prepared by weighing the mass of ligand and assuming a ligand
density of 1 g/mL.
The reactors were 4 mL glass vials, each equipped with a septum
and a 10 mm × 3 mm PTFE-coated magnetic stirring bar. The vials
were placed in tightly fit wells in an aluminum block which was
heated by a temperature-controlled hotplate. Typically, 12 exper-
iments were conducted simultaneously in 12 different vials, and a
thermocouple was placed in the thirteenth vial, charged with the
same volume of solvent and a stirring bar, for temperature con-
trol. The oxygen atmosphere (∼7 kPa gauge pressure) above the
liquid in the vials was maintained by an oxygen distributor, con-
structed with 12 Luer-Lok needle connection ports and 21-gauge
1.5 in. needles that punctured through the septa.
In a typical experiment, the reaction vials were charged with
the required volumes of Pd(OAc)2 solution, toluene solvent, and
ligand solution to achieve the desired system concentration and
ligand–metal ratio. For example, to achieve a 2:spy–metal ratio and
a final Pd concentration of 5 mM, 0.40 mL of 25 mM Pd(OAc)2 solu-
tion (0.010 mmol), 0.025 mL of 0.8 N ligand solution (0.020 mmol
spy ligand), and 1.37 mL of toluene would be added by pipette.
The vials were placed in the Al block, tightly closed, vented with
short 23-gauge needles, and connected to the oxygen distribu-
tor. An oxygen atmosphere was established by bubbling O2 at
room temperature for 10–15 min while the solution was stirred
(400 rpm unless otherwise noted), after which the vent needles
were removed and the oxygen supply needles were raised above
the liquid level to prevent siphoning, but remained inserted into the
vial to resupply O2 consumed during reaction. The reaction block
was heated to 353 0.5 K. After the temperature reading was stabi-
lized, the catalyticreactionwas commencedbyinjecting0.2–0.4 mL
of 0.25–4.0 M BnOH in toluene into the vials. At regular intervals, a
0.1 mL sample was withdrawn by a syringe and diluted with 0.7 mL
of CDCl3 in a standard 5 mm NMR tube. The combination of dilu-
tion and cooling to room temperature effectively quenched the
catalytic reaction. At the end of the experiment, the reaction vials
2.4. Determination of initial rates
The conversion
X of benzyl alcohol was calculated from
the peak areas of the 1H NMR resonances of benzyl alcohol
(ı 4.8 ppm) and benzaldehyde (ı 10.2 ppm), using the formula
X = (2IPhCHO)/(2IPhCHO + IPhCH2OH), where Ix is the peak area for
species x. The factor of 2 was due to the different number of protons
in those molecules at those resonances. The concentration of ben-
zyl alcohol CBnOH,0(1 – X) decreased with reaction time in a manner
that followed approximately first-order kinetics in BnOH for low
(<30%) conversions, but which deviated significantly from first-
order kinetics at medium and high conversions, primarily due to
product inhibition. The available conversion data were fitted with
a modified first-order rate law that takes into account product inhi-
bition with an adjustable parameter b (Eq. (1), where CBnOH,0 is the
initial concentration of BnOH):
2
d(CBnOH
)
−kCBnOH
=
(1)
dt
CBnOH + b(CBnOH,0 − CBnOH
)
This rate law reduces to first-order kinetics for t → 0 for any
value of b (i.e. when no product was present) and also for
b → 0. b(=0.30) was obtained by minimizing the average least-
squares residual over the entire kinetic data set. The best-fit
values of k to experimental data were determined by solving
for CBnOH(t) numerically and minimizing the sum of squares of
the residuals. Initial rates were then computed with the formula
d(CBnOH)/dt|t=0 = −kCBnOH,0. The magnitude of the correction in ini-
tial rate obtained with b = 0.30 versus b = 0 averaged +10%, with a
maximum of +26% and a minimum of 0%. While this procedure

M.N. Missaghi et al. / Applied Catalysis A: General 391 (2011) 297–304
299
b
a
0.012
0.008
0.004
0.000
1m
py
1p
2p
0.016
2m
3m
4m
0.012
0.008
0.004
0.000
0
2
4
6
8
10
0
2
4
6
8
10
py:Pd Ratio
py:Pd Ratio
0.010
0.008
0.006
0.004
0.002
0.000
d
c
0.010
0.008
0.006
0.004
0.002
0.000
5m
6p
8p
6m
10p
7m
10m
0
2
4
6
8
10
0
2
4
6
8
10
py:Pd Ratio
py:Pd Ratio
Fig. 1. Effect of ligand (py or spy)-metal ratio on initial rate of BnOH oxidation (353 K, toluene, 5 mM Pd(OAc)2, 20:1 BnOH:Pd). The concentrations of py or spy groups were
used to compute the py:Pd ratios.
accounted for deactivation due to product inhibition, it did not
account for deactivation due to Pd agglomeration. Thus, initial rates
estimated by this method would reflect the effect of Pd agglomer-
ation, which typically occurred in the first few minutes of reaction
before the first samples were taken. They could be used to compare
ligands for their ability to stabilize the Pd complexes.
more than for 1p. For short-chain bis(meta-pyridyl)siloxanes 2m
and 3m, the activity was lower relative to 1m at spy:Pd = 1.2, but
was slightly higher at spy:Pd = 8. This trend of weaker dependence
of activity on the spy/Pd ratio continued for medium-chain ligands
4m and 5m as well as longer ones. For 6m and 7m, the activity
decreased by a factor of only 1.5 between spy:Pd of 1.2:1 and 8:1.
The decrease was slightly more pronounced for 10m over the same
spy:Pd range.
3. Results
3.1. Effect of ligand–metal ratio on initial rate
3.2. Effect of ligand on stability of Pd catalysts
The effect of varying the ligand to metal ratio spy/Pd between
1.2:1 and 8:1 on the initial rate was studied for py and the spy
shown in Scheme 1 (reaction conditions: 353 K, 20:1 BnOH:Pd,
5 mM Pd(OAc)2). Fig. 1 shows representative data of initial rates
for different ligands. Although for all ligands, increasing the ligand
concentration suppressed the activity, the sensitivity of the activity
to the ligand concentration depended strongly on the nature of the
ligand.
For pyridine, the initial rate decreased by a factor of 4 upon
increasing the py:Pd ratio by a factor of 4 (from 2:1 to 8:1).
Similar, suppression, but of somewhat different magnitudes, was
observed for mono(para-pyridyl)siloxane 1p and for bis(para-
pyridyl)siloxanes 2p, 6p, 8p, and 10p (Fig. 1b and d). The
quantitatively similar activity of 1p to pyridine suggested that the
two ligands are broadly similar in their binding properties. For the
tethered ligands, the activity increased with chain length in the
order 2p < 6p < 8p < 10p. A low spy:Pd ratios, the differences among
6p, 8p and 10p were less significant than at higher ratios.
For mono(meta-pyridyl)siloxane 1m, the activity decreased by
a factor of 10 as the spy:Pd ratio was raised from 1.2:1 to 8:1, much
The effect of ligand structure on the catalyst stability was inves-
tigated under a range of reaction conditions. The variables studied
included Pd concentration and BnOH:catalyst ratio. In these exper-
iments, the spy(or py):Pd ratio was fixed at the stoichiometric value
of 2 to remove the effect of excess or deficient ligands.
The active site concentration was varied by changing [Pd(OAc)2]
while holding the BnOH:catalyst ratio at 20:1. For Pd concentra-
tions of 5 mM (the condition used in Fig. 1) or less, the reaction
mixture remained clear throughout the experiment for all ligands
studied, suggesting that the catalysts were stable. At the higher Pd
concentration of 10 mM, agglomeration of Pd0 occurred for some
ligands (Table 1 and Fig. 2), which was apparent within 2 min of
the start of the reaction. At the highest Pd concentration of 20 mM,
Pd0 aggregates were observed for all ligands tested. The data in
Table 1 indicated a general trend that the meta-pyridylsiloxane lig-
ands were more effective in preventing Pd agglomeration, whereas
the para-pyridylsiloxanes were similar to or, at best, slightly better
than pyridine.
Under conditions used here and when the solution remained
clear, i.e. at lower Pd concentrations, the initial reaction rates were

300
M.N. Missaghi et al. / Applied Catalysis A: General 391 (2011) 297–304
Fig. 2. Appearance of reaction mixture after 10 min of reaction (353 K, toluene, 20:1 BnOH:Pd, 2:spy:Pd. 10 mM Pd). Ligand used were: a, py; b, 1m; c, 2m; d, 3m; e 4m; f,
5m; g, 6m; h, 7m; i, 10m; j, 1p; k, 2p; l, 6p; m, 8p; n, 10p. For vial a, black precipitates settled on the bottom.
a
b
0.030
0.025
0.020
0.015
0.010
0.005
0.000
0.030
0.025
0.020
0.015
0.010
0.005
0.000
1m
2m
3m
4m
py
1p
2p
0
5
10
15
20
25
0
5
10
15
20
25
[Pd(OAc)2] (mM)
[Pd(OAc)2] (mM)
0.050
0.040
0.030
0.020
0.010
0.000
0.030
0.025
0.020
0.015
0.010
0.005
0.000
c
d
6p
5m
8p
6m
10p
7m
10m
0
5
10
15
20
25
0
5
10
15
20
25
[Pd(OAc)2] (mM)
[Pd(OAc)2] (mM)
Fig. 3. Dependence of initial reaction rate on Pd concentration (353 K, spy(or py)/Pd = 2, BnOH/Pd = 20). Since the ratios of metal, ligand, and substrate were fixed in these
experiments, the x-axis corresponds to the state of dilution of the reaction system.
expected to show a dependence of [Pd]3/2 (see Section 4). This
seemed to be the case within experimental uncertainties, which
also indicated that the reaction kinetics was not affected by the
O2 dissolution rate at low Pd concentrations (Fig. 3). Except for
4m, 5m, and 10p, the rate was substantially slower at higher Pd
concentrations (10 mM or higher) than predicted by the 3/2 order
dependence, consistent with loss of catalyst due to formation of less
active Pd aggregates. This deviation was particularly noticeable for
py, 1p, 2p, 6p, 1m, 2m, and 3m.
The long-term stability of the catalysts was also investigated
using a reactant concentration of either 25 or 250 mM BnOH and a
low Pd concentration of 1.25 mM. This Pd concentration was much
lower than those used for Figs. 1 and 2, and was chosen such that
there was no detectable formation of Pd black for any of the ligands.
Table 2 summarizes the results. All of the ligands tested exhib-
ited good long-term stability, and high turnover numbers were
obtained. Thus, at low Pd concentrations, there was no adverse
effect with spy ligands. As reported in the literature, buildup of
the side product benzoic acid had a negative effect on the activ-
ity. As a result, at the higher substrate–catalyst ratio studied, the
oxidation activity became strongly suppressed by the side product
at long reaction times, such that the reactions were not complete
even after 18 h.
4. Discussion
The experimental results show that siloxane-modified pyridine
(spy) ligands fulfill the mechanistic role of py in Pd-catalyzed aer-
obic oxidation. Broadly speaking, the activities of complexes with
spy ligands were within a factor of 3–4 that of py, and, as with py,
they stabilized the Pd complex from agglomeration into Pd black.
However, there are significant variations of catalyst activity and
stability among the bidentate siloxane ligands with respect to the
chain length and the attachment point of silicon to the pyridine ring.
These structure–activity trends give further insights into the role
of pyridine in the catalytic mechanism, as well as the most impor-
tant factors affecting the stability of the monomeric Pd active site,
particularly at higher Pd concentrations.

M.N. Missaghi et al. / Applied Catalysis A: General 391 (2011) 297–304
301
2
i
i-2
R
i-1
1
R
1
2
R
2
i
R
i
C
L,i
C
Pd,i
R
i
spy:Pd > 2 2 > spy:Pd > 1
Coordination rings
Coordination chains
Dimeric Pd rings
Scheme 2. Schematic representation of some of the various coordination oligomers of Pd(OAc)2 and spy present in dilute solution. Pd centers are represented as filled dots,
and bifunctional ligands as line segments. The complexes denoted by 2Ri represent rings containing a Pd–Pd bond.
Table 1
Table 3
Effect of system concentration on catalyst stability (353 K, toluene, 20:1 BnOH:Pd,
Experimental effective molarities of cyclic spy–Pd ring complexes Ri (353 K, 5.7 mM
in toluene-d8, 2:spy:Pd), as calculated from 1H NMR measurements of the ring-chain
distribution.
2:spy:Pd).
Ligand
[Pd], mMa
Ligand
Effective molarity (EMi) for Ri (mol L−1
)
1.25, 2.5, 5b
10b
20c
EM1
EM2
EM3
EM4
Pyridine
1m
2m
3m
4m
5m
6m
7m
10m
1p
C
C
C
C
C
C
C
C
C
C
C
C
C
C
P
C
C
C
C
C
C
C
C
B
Y
B
B
Y
B, P
B
P
B
B, P
B
B
B
B
P
2m
3m
4m
5m
6m
7m
10m
6p
0
0
0
0
0.0030
0.0032
(0.0063)
(0.0046)
(0.0037)
(0.0031)
(0.0019)
n.d.
(0.0063)
(0.0041)
(0.0031)
(0.0022)
(0.0018)
(0.0015)
(0.0009)
n.d.
0.0027
0.0075
0.0091
(0.0103)
(0.0086)
(0.0054)
0.0071
0.0043
0.0058
0.0041
0.0129
0.0162
0.0142
0
8p
10p
0
n.d.
n.d.
n.d.
n.d.
2p
6p
8p
10p
P
P
P
P
0.003
The much weaker dependence of activity on spy:Pd ratio for
complexes of ligands 5m to 10m arise primarily from their abil-
ity to form monomeric cyclic coordination complexes R1 with
Pd(OAc)2. Since Pd can normally bind two equivalents of pyri-
dine, and a bis(pyridyl)siloxane can bind two Pd centers, linear
and cyclic coordination oligomers can be formed (Scheme 2). The
observed ring-chain distribution depends on the ligand structure
as well as the concentrations of spy and Pd(OAc)2 [14–15]. Only Ri
are observed for spy/Pd = 2 due to the strong N–Pd bond and the
resulting chain termination mechanism from N–Pd bond dissocia-
tion. For spy/Pd > 2, linear oligomers CL,i terminated by semibound
ligand are formed alongside the ring species, whereas for spy/Pd < 2,
CPd,i terminated by coordinatively unsaturated Pd are formed.
The data in Fig. 1 were collected for spy/Pd = 2. Thus, only ring
complexes were expected in these solutions. Among the ring com-
plexes, monomeric R1 could be formed only if the bidentate ligand
is sufficiently long, such that the monomeric ring can be formed
with little ring strain. Computational results using the DFT method
(B3LYP basis set) suggested that for the bis(meta-pyridyl)siloxanes,
R1 formed with 5m or higher oligomers are essentially strainless,
whereas for the para-pyridylsiloxanes, oligomers longer than 8p
are required [16]. These computational results were substantiated
by experiments. By analyzing the 1H NMR spectra of spy–Pd(OAc)2
at various concentrations and ligand–metal ratios, we found that
R1 was present for bis(meta-pyridyl) ligands 5m and longer and
for bis(para-pyridyl) ligands longer than 8p [13]. These results are
summarized in Table 3 in terms of effective molarities for R1 (EM1).
For the R1 complexes of 5m to 10m and 10p, effective molari-
ties of 3–14 mM were found, which were comparable to the Pd
concentrations of 1.25–20 mM used.
Reactions were run for 80, 40, 20, 10, and 5 min for Pd concentrations of 1.25, 2.5, 5,
10, and 20 mM, respectively in order to achieve roughly equivalent conversions.
a
Clear solution (C), intensely yellow but clear solution (Y), black suspension (B),
black precipitates (P).
b
At 400 rpm stirring rate.
At 800 rpm stirring rate.
c
The data in Fig. 1 show that the activities of all of the para-
substituted siloxanes, as well as meta-pyridylsiloxane 1m and
the short-chain bis(meta-pyridyl)siloxanes 2m, 3m, and 4m, were
inversely dependent on the total pyridyl concentration in the
system, similar to unsubstituted pyridine. In fact, 1p behaved
practically identically to pyridine. However, for longer bis(meta-
pyridyl)siloxanes 5m to 10m, the dependence of activity on [spy]
was diminished, with the dependence for 6m and 7m being the
weakest.
Table 2
Turnover numbers (TONs) at various reaction times using pyridine and ligands 1–10
(80 ◦C, toluene, 2:1 py:Pd or spy:Pd, 20:1 or 200:1 BnOH:Pd).
Ligand
TON at sampling time
BnOH:Pd = 20a
BnOH:Pd = 200b
60 min 120 min
116
24 min
48 min
80 min
1080 min
Pyridine
1m
1p
2m
2p
3m
4m
5m
6m
6p
16.5
13.8
16.3
7.0
11.8
8.6
8.6
10.2
6.5
10.3
6.9
10.3
6.5
19.5
15.7
19.5
10.9
14.9
12.3
12.6
14.1
9.8
14.6
9.6
14.6
9.6
20.0
18.9
20.0
15.8
14.7
14.7
15.1
15.9
13.2
15.9
11.2
18.6
10.8
17.4
132
102
128
93
179
165
186
149
156
145
136
137
110
166
101
159
114
155
79
106
68
75
64
56
60
46
71
42
72
47
67
100
87
73
75
59
92
54
89
60
Although binding of spy to Pd is important for stabilizing Pd
from deactivation (to be discussed later), Pd complexes of these
ligands are less active than Pd–py complexes due to ligand com-
petition with BnOH for coordination sites, especially at high spy
concentrations. For these bidentate ligands, because both pyridyl
groups are tethered to the siloxane chain, they are never far from
the Pd atom even after dissociation from the metal center, since
the second pyridyl group is still bound. That is, the effective con-
centration of the pyridyl groups at the Pd center would be higher
7m
8p
10m
10p
13.9
15.5
85
a
1.25 mM Pd, 25 mM BnOH.
1.25 mM Pd, 250 mM BnOH.
b

302
M.N. Missaghi et al. / Applied Catalysis A: General 391 (2011) 297–304
than the average concentration in solution. This effect is expected
to be more prominent for shorter ligands that can form R1. Thus,
the activities of 5m and 6m are the most suppressed, and they
also exhibit the weakest dependence on [spy] (Fig. 1). For longer
chains, a spy group dissociated from Pd could diffuse farther away,
such that the local concentrating effect becomes diluted. Con-
sequently, ligands 8m and 10p behave closer to pyridine. For
ligands that cannot form R1, their behavior are like pyridine, as
expected.
Steric hindrance also contributes to the suppression of activity.
The siloxane tether of a R1 complex is expected to block access
to the Pd center along certain directions. This effect is expected
to be more important for R1 complexes and of shorter ligands for
which the siloxane chain is closer to the Pd center, consistent with
the observations. Unlike the steric and local concentrating effects,
electronic effects due to the siloxane substituents are expected to
be negligible. Chvalovsky and coworkers have reported Hammett
substituent constants of −0.01 and 0.00 for the meta- and para-
(trimethylsiloxy)dimethylsilyl group, respectively, indicating that
the inductive effect of a methylsiloxane substituent on an aromatic
ring is negligible [17]. Competitive binding studies conducted by us
also showed that 1m and 1p bind Pd(OAc)2 with the same affinity
as pyridine [13].
It should be mentioned that O2 mass transfer may play a role
in these observations. The solubility of O2 in toluene at 353 K is
rather low (ca. 20 mM). Since the initial concentration of substrate
is much higher than this, oxygen is consumed most rapidly at the
start of the reaction. The drop in O2 concentration would lead to
buildup of Pd0 species, and eventually Pd black formation. This is
expected to be more severe at higher Pd concentrations and with
more active catalysts. Therefore, the suppression of activity by the
siloxane ligands helps preserve the catalytic activity, in addition to
their protective effect by more effective ligation.
5. Conclusion
We have demonstrated that bidentate, bis(pyridyl)siloxane lig-
ands form effective catalysts of Pd complexes. Those ligands that
can form monomeric ring complexes result in a catalytic system
that exhibits a much weaker dependence of activity on ligand/Pd
ratio than those that cannot, and improved catalyst stability. The
results demonstrate the importance of effective molarity of lig-
ands on the observed catalytic properties, which are results of
steric effect of the siloxane chains, effective local concentration
of pyridyl groups, and ability to stabilize Pd0 complexes against
metal agglomeration. The stabilizing effect of bis(pyridyl)siloxane
ligands may be useful in catalytic systems where using an excess
of pyridine is undesirable or impossible, such as supported Pd
complexes. The observations also lead to the possibility of using
multi-functional ligands to further tune the catalytic properties
by introducing additional and designed interactions between the
ligands and the substrate.
The differences in binding properties among the spy ligands
correspond to clear trends in the deactivation behavior. Visual
observation of the reaction mixtures (Table 1 and Fig. 2) sug-
gested that some spy ligands are more effective in stabilizing
Pd from agglomeration than pyridine. In fact, all of the meta-
pyridylsiloxanes appeared to be superior, in line with the higher
effective molarities of pyridyl groups achieved with these lig-
ands. In order to quantify this observation, the apparent initial
reaction rates were determined at each of the concentrations
shown in Table 1; the results are shown in Fig. 3. In the absence
of rapid deactivation, the initial rate is expected to be propor-
tional to [BnOH][Pd(OAc)2]1/2 when the py:Pd ratio is 2 [9]. In
our experiments, the metal–ligand and substrate–catalyst ratios
were unchanged among different ligands; therefore, the initial rate
should be proportional to [Pd]3/2, where [Pd] is the total Pd concen-
tration. Indeed, at low Pd concentrations (5 mM or less) when no
metal agglomeration was detected visually, the expected depen-
dence on [Pd] was observed.
Acknowledgements
Support by the U.S. Department of Energy, Basic Energy Sciences,
DE-FG02-01ER15184 is gratefully acknowledged. NMR analyses
were performed at the IMSERC facility at Northwestern Univer-
sity, using a 400 MHz NMR spectrometer acquired under NSF Grant
CHE-9871268 (1998).
Appendix A.
At higher Pd concentrations (10 mM or above), the apparent
initial rates dropped off from the expected values due to rapid
deactivation of the Pd species, which was essentially complete
prior to the time when the first sample was taken. The extent of
deviation differed among the spy ligands, and can be used as a qual-
itative indication of the extent of metal agglomeration. Thus, the
meta-substituted ligands with a siloxane backbone longer than 3
Si–O units (i.e., 4m and longer) improve stability of the Pd cata-
lyst from agglomeration, although none of the ligands was able to
prevent agglomeration at the high concentration of 20 mM. Para-
substituted ligands are also effective, but they require longer spacer
lengths (8–10 Si–O units). These correlate well with the ability of
the ligands to form R1 complexes – those that can form R1 com-
plex can stabilize the Pd complex more effectively than those that
cannot.
The exception to the above is ligand 4m. Computational results
suggest that a R1 complex would not be formed with this ligand due
to ring strain. We propose that, in this case, a ring complex similar
to 2R1 could be formed with this ligand, and the additional Pd atom
help relieve the ring strain. Analogous to Pd nanoclusters which
have been reported to have some activity in this reaction [18],
these 2R1 complexes would also be active for alcohol oxidation,
although possibly less than the mononuclear active site. Evidence
for Pd–Pd bond formation in similar complexes have been reported
by Komano et al. using MALDI mass spectrometry [19].
The series of bis(para-pyridyl)methylsiloxane ligands was syn-
thesized with materials purchased from either TCI America, Gelest,
or Sigma-Aldrich, which were used as received. Manipulations
of air- and moisture-sensitive chemicals were performed using a
Schlenk line under N2. Spectroscopic and HRMS data were collected
in the IMSERC facility at Northwestern University. NMR spectra
were collected on a Varian INOVA 400 MHz spectrometer with a
broadband probe.
4-Pyridyldimethylsilane (1a): 5.00 g of 4-bromopyridine
hydrochloride (TCI America, 0.0257 mol) was weighed into a
250 mL Erlenmeyer flask, and a magnetic stir bar was added.
Approximately 125 mL of a 5% aqueous solution of Na3CO3 were
added and stirred until the product dissolved, bubbling stopped,
and the pH was greater than 8.5 (<5 min). The reaction mixture was
washed with 2× 50 mL portions of water and 1× 50 mL brine. The
organic phase was dried with MgSO4, and solvents were removed
by rotary evaporation to obtain the free base (3.3 g yield, 0.021 mol).
The free base was injected into a 100 mL Schlenk flask containing
25 mL of anhydrous THF, and equipped with a magnetic stir bar
and rubber septum under N2. To this solution, 12 mL of a 2.0 M
solution of isopropylmagnesium chloride in THF were added by
syringe over 1 min, causing the reaction mixture to turn orange.
The reaction was allowed to proceed for 1.5 h at room tempera-
ture. Finally, 2.8 mL (0.025 mol) of chlorodimethylsilane was added
slowly. The reaction mixture warmed and the orange color slowly

M.N. Missaghi et al. / Applied Catalysis A: General 391 (2011) 297–304
303
0.41 ppm. 29Si NMR (80 MHz, CDCl3): ı −3.2, −6.6, −18.0 ppm.
HRMS (ESI) calcd for ([M + H]+): m/z 286.1111; found 286.1109.
1-(4-Pyridyl)-1,1,3,3,5,5-hexamethyltrisiloxan-3-ol (1f): Pre-
pared in the same manner as 1d, from 1.23 g (5.4 mmol) of
1e. Evolution of H2 was complete within 4 h. The product 1f
was isolated as 1.1 g (4.9 mmol, 91%) a turbid oil. 1H NMR
discharged. Reaction was allowed to proceed for 1 h. The reaction
mixture was diluted in 2 parts of diethyl ether, washed with 2×
50 mL portions of distillated water and 1× 50 mL of brine, and
dried over Na2SO4. Solvents were removed by rotary evaporation
to yield a free-flowing yellow oil in good yield (2.6 g, 0.019 mol,
91% from free base). Further purification was accomplished by
vacuum distillation (20 Torr, 75 ◦C) to give 1.95 g of 1a as a clear
colorless liquid (0.014 mol, 75%). 1H NMR (400 MHz, CDCl3): ı 8.54
ꢀ
(400 MHz, CDCl3): ı 8.49 (dd, JAB = 4.3 Hz, JAA = 1.1 Hz, 2H), 7.42
ꢀ
(dd, JBA = 4.4 Hz, JBB = 1.3 Hz, 2H), 5.32 (bs, 1H), 0.34 (s, 6H), 0.10
(s, 6H), 0.08 (s, 6H). 13C NMR (100 MHz, CDCl3): ı 150.81, 148.15,
127.98, 1.25, 0.52, 0.41 ppm. 29Si NMR (80 MHz, CDCl3) ı −3.35,
−12.64, −19.74 ppm. HRMS (ESI) calcd for C7H11NOSi ([M + H]+):
m/z 302.1058; found 302.1059.
ꢀ
(d, JAB = 4.4 Hz, 2H), 7.38 (dd, JBA = 4.2, JBB = 1.4 Hz, 2H), 4.38 (sept,
JD = 3.8 Hz, 1H), 0.34 (JC = 3.8 Hz, 6H) ppm. 13C NMR (100 MHz,
CDCl3): ı 149.02, 148.88, 129.04, −4.45 ppm. 29Si NMR (80 MHz,
CDCl3): ı −19.8 ppm. HRMS (ESI) calcd for C7H11NSi ([M + H]+): m/z
138.0734; found 138.0733.
(4-Pyridyl)pentamethyldisiloxane (1p): Prepared according to
the synthesis of 1b on the 0.139 g (0.91 mmol) scale, except that
chlorotrimethylsilane (2 mmol) is used in the condensation step.
Purified by column chromatography. Yield 0.151 g (1.14 mmol, 74%
from 1b). 1H NMR (400 MHz, CDCl3): ı 8.57 (dd, J = 4.3 Hz, 1.1 Hz,
2H), 7.38 (dd, J = 4.3 Hz, 1.1 Hz, 2H), 0.32 (s, 6H), 0.09 (s, 9H) ppm.
13C NMR (100 MHz, CDCl3): ı 149.7, 148.9, 127.8, 2.04, 0.61 ppm.
29Si NMR (80 MHz, CDCl3): ı 6.11, −7.21 ppm. HRMS (ESI) calcd for
4-Pyridyldimethylsilanol (1b): A 50 mL Schlenk flask equipped
with magnetic stir bar and rubber septum was connected to an
oil bubbler and purged with a stream of N2, then charged with
20 mg of Pearlman’s catalyst (20 wt% Pd(OH)2/C), 19 mL of tetrahy-
drofuran, and 1 mL of water. The reaction was stirred while 1.0 g
(6.5 mmol) of 1a was added by syringe. Evolution of H2, as mon-
itored with bubbler, was complete within 4 hours. The reaction
was then syringe-filtered to remove the catalyst and evacuated
(final pressure ca. 50 mTorr) to remove solvents and water. The
product 2p was isolated as a turbid oil (when impure) or in crys-
talline form. Isolated yields of greater than 90% were achieved
with less than 10% being the homofunctional condensation prod-
uct bis(4-pyridyl)-tetramethyldisiloxane. Product was found to be
marginally stable in the pure state at room temperature, with a
half-life of greater than 24 h. 1H NMR (400 MHz, CDCl3): ı 8.41 (dd,
C
10H19NOSi2 ([M + H]+): m/z 226.1078; found 226.1083.
1,3-Bis(4-pyridyl)tetramethyldisiloxane
(2p):
Prepared
according to the synthesis of 1c, except that 1.0 g (6.5 mmol) of
4-pyridyldimethylsilanol (1b) was isolated and warmed to 100 ◦C
under vacuum for 2 h. Purified by column chromatography on
silica gel. Yield 0.27 g (0.94 mmol, 29%). 1H NMR (400 MHz, CDCl3):
ꢀ
ı 8.58 (dd, JAB = 4.1 Hz, JAA = 1.1 Hz, 2H), 7.37 (dd, JBA = 3.9 Hz,
JBB = 1.2 Hz, 2H), 0.36 (s, 12H) ppm. 13C NMR (100 MHz, CDCl3):
ꢀ
ı 153.7, 150.7, 140.8, 134.2, 123.5, 0.96 ppm. 29Si NMR (80 MHz,
CDCl3): −0.1 ppm. HRMS (ESI) calcd for C14H20N2Si2O ([M + H]+):
m/z 289.1187; found 289.1191.
ꢀ
ꢀ
JAB = 4.2 Hz, JAA = 1.5 Hz, 2H), 7.46 (dd, JBA = 4.3 Hz, JBB = 1.5 Hz, 2H),
6.40 (bs, 1H), 0.38 (s, 6H) ppm. 13C NMR (100 MHz, CDCl3): ı 151.10,
148.12, 128.41, 0.08 ppm. 29Si NMR (80 MHz, CDCl3): ı 3.74 ppm.
HRMS (ESI) calcd for C7H11NOSi ([M + H]+): m/z 154.0683; found
154.0674.
1,11-Bis(4-pyridyl)dodecamethylhexasiloxane (6p): Pre-
pared from 1.0 g (6.5 mmol) of 1b and 1.16 g (3.3 mmol)
1,7-dichlorotoctamethyltetrasiloxane. Purified by chromatog-
raphy on silica gel. Yield 0.66 g (1.1 mmol, 51%). 1H NMR
1-(4-Pyridyl)-1,1,3,3-tetramethydisiloxane (1c): Prepared in
the same manner as 1b, but rather than isolating the product, the
silanol (6.9 mmol) was redissolved with anhydrous diethyl ether
in a 50 mL Schlenk flask which was previously connected to an oil
bubbler and purged with N2. Into this flask was injected 2.15 mL
(15.4 mmol) of triethylamine, followed immediately by 1.55 mL
(14 mmol) of chlorodimethylsilane. The reaction mixture turned
into a voluminous white precipitate accompanied by a significant
amount of fuming. The reaction mixture was washed with 2× 50 mL
portions of water, followed by 1× 30 mL of brine, and volatile com-
ponents were removed by rotary evaporation to give 1.23 g of 1c as
a clear liquid (5.4 mmol, 78%). 1H NMR (400 MHz, CDCl3): ı 8.57 (dd,
ꢀ
ı 8.59 (dd, JAB = 4.2 Hz, JAA = 1.4 Hz, 2H),
(400 MHz, CDCl3):
ꢀ
7.42 (dd, JBA = 4.2 Hz, JBB = 1.5 Hz, 2H), 0.36 (s, 12H), 0.09 (s, 12H),
0.05 (s, 12H). 13C NMR (100 MHz, CDCl3): ı 148.9, 137.7, 127.8,
1.3, 1.2, 0.46 ppm. 29Si NMR (80 MHz, CDCl3): ı −7.1, −23.8,
−25.6 ppm. HRMS (ESI) calcd. for C22H44N2O5Si6 ([M + H]+): m/z
585.1939; found 585.1929.
1,15-Bis(4-pyridyl)hexadecamethyloctasiloxane (8p): Pre-
pared from 1.0 g (4.4 mmol) of 1d and 0.77 g (2.2 mmol) 1,7-
dichlorotoctamethyltetrasiloxane. Purified by chromatography on
silica gel. Yield 0.77 g (1.1 mmol, 48%). 1H NMR (400 MHz, CDCl3):
ꢀ
ꢀ
JAB = 4.4 Hz, JAA = 1.2 Hz, 2H), 7.40 (dd, JBA = 4.3 Hz, JBB = 1.3 Hz, 2H),
4.75, (sept, J = 2.8 Hz, 1H), 0.35 (s, 6H), 0.19 (d, J = 2.8 Hz, 6H) ppm.
13C NMR (100 MHz, CDCl3): ı 147.0, 146.4, 125.5, −1.44, −2.07 ppm.
29Si NMR (80 MHz, CDCl3) ı −0.82, −3.86 ppm. HRMS (ESI) calcd for
C9H18NOSi2 ([M + H]+): m/z 212.0921; found 212.0920.
ꢀ
ı 8.58 (dd, JAB = 4.3 Hz, JAA = 1.2 Hz, 2H), 7.41 (dd, JBA = 4.2 Hz,
ꢀ
JBB = 1.4 Hz), 0.35 (s, 12H), 0.09 (s, 12H), 0.06 (s, 12H), 0.05 (s, 12H).
13C NMR (100 MHz, CDCl3): ı 149.4, 148.8, 127.8, 1.2, 1.1, 0.43 ppm.
29Si NMR (80 MHz, CDCl3): ı −2.8, −19.5, −21.3, −21.4 ppm. HRMS
(ESI) calcd. for C26H56N2O7Si8 ([M + H]+): m/z 733.2314; found
733.2305.
1-(4-Pyridyl)-1,1,3,3-tetramethyldisiloxanol (1d): Prepared
in the same manner as 1b from 1.23 g (5.4 mmol) of 1c. Evolution
of H2 was complete within 4 h. The product 1d was isolated as 1.1 g
(4.9 mmol, 91%) of a colorless oil. 1H NMR (400 MHz, CDCl3): ı 8.41
(dd, JAB = 4.3 Hz, JAA = 1.4 Hz, 2H), 7.40 (dd, JBA = 4.3 Hz, JBB = 1.5 Hz,
2H), 5.58 (bs, 1H), 0.36 (s, 6H), 0.14 (s, 6H) ppm. 13C NMR (100 MHz,
CDCl3): ı 150.2, 148.0, 128.0, 0.63, 0.39 ppm. 29Si NMR (80 MHz,
CDCl3) ı −2.82, −10.40 ppm. HRMS (ESI) calcd for C9H17NO2Si2
([M + H]+): m/z 228.0871; found 228.0870.
1-(4-Pyridyl)-1,1,3,3,5,5-hexamethyltrisiloxane (1e): Pre-
pared in the same manner as 1c, from 6.9 mmol of 1d. After
washing and removal of solvents 0.55 g of 1e was obtained
(1.9 mmol, 83%). 1H NMR (400 MHz, CDCl3): ı 8.57 (dd, JAB = 4.3 Hz,
1,19-Bis(4-pyridyl)icosamethyldecasiloxane
(10p):
Pre-
pared from 1.0 g (3.3 mmol) of 1f and 0.58 g (1.7 mmol)
1,7-dichlorotoctamethyltetrasiloxane. Purified by chromatog-
raphy on silica gel. Yield 0.64 g (1.1 mmol, 44%). 1H NMR (400 MHz,
ꢀ
ꢀ
ꢀ
ı 8.59 (dd, JAB = 4.2 Hz, JAA = 1.4 Hz, 2H), 7.42 (dd,
CDCl3):
ꢀ
JBA = 4.2 Hz, JBB = 1.5 Hz, 2H), 0.36 (s, 12H), 0.09 (s, 12H), 0.05
(s, 12H). 13C NMR (100 MHz, CDCl3): ı 149.5, 148.9, 1.2, 0.5 ppm.
29Si NMR (80 MHz, CDCl3): ı −2.7, −19.4, −21.2, −21.4 ppm. HRMS
(ESI) calcd. for C30H68N2O9Si10 ([M + H]+): m/z 881.2690; found
881.2698.
ꢀ
ꢀ
JAA = 1.2 Hz, 2H), 7.41 (dd, JBA = 4.2 Hz, JBB = 1.2 Hz, 2H), 4.68 (sept,
JCD = 2.8 Hz, 1H), 0.35 (s, 6H), 0.15 (d, J = 2.8 Hz, 6H), 0.07 (s, 6H)
ppm. 13C NMR (100 MHz, CDCl3): ı 149.4, 148.8, 127.8, 0.74,
References
[1] S.S. Stahl, Science 309 (2005) 1824–1826.

304
M.N. Missaghi et al. / Applied Catalysis A: General 391 (2011) 297–304
[2] T. Nishimura, T. Onoue, K. Ohe, S. Uemura, Tetrahedron Letters 39 (1998)
6011–6014.
[12] Ta. Stephens, Sm. Morehous, A.R Powell, J.P. Heffer, G. Wilkinso, Journal of the
Chemical Society (1965) 3632.
[3] M.S. Sigman, D.R. Jensen, Accounts of Chemical Research 39 (2006) 221–229.
[4] J.H. Liu, F. Wang, X.L. Xu, Progress in Chemistry 19 (2007) 1718–1726.
[5] T. Nishimura, S. Uemura, Synlett (2004) 201–216.
[13] M.N. Missaghi, J.N. Galloway, H.H. Kung, Organometallics 29 (2010) 3769–3779.
[14] H. Jacobson, W.H. Stockmayer, Journal of the Chemistry and Physics 18 (1950)
1600–1606.
[6] D.R. Jensen, J.S. Pugsley, M.S. Sigman, Journal of the American Chemical Society
123 (2001) 7475–7476.
[15] G. Ercolani, L. Mandolini, P. Mencarelli, S. Roelens, Journal of the American
Chemical Society 115 (1993) 3901–3908.
[7] J.A. Mueller, D.R. Jensen, M.S. Sigman, Journal of the American Chemical Society
124 (2002) 8202–8203.
[16] I. Konstantinov, J.M. Galloway, M.N. Missaghi, L.J. Broadbelt, H.H. Kung, Journal
of Physical Chemistry C (submitted for publication).
[8] S.K. Mandal, D.R. Jensen, J.S. Pugsley, M.S. Sigman, Journal of Organic Chemistry
68 (2003) 4600–4603.
[9] B.A. Steinhoff, I.A. Guzei, S.S. Stahl, Journal of the American Chemical Society
126 (2004) 11268–11278.
[10] B.A. Steinhoff, S.S. Stahl, Organic Letters 4 (2002) 4179–4181.
[11] B.V. Popp, S.S. Stahl, Chemistry-A European Journal 15 (2009) 2915–2922.
[17] Z. Plzak, F. Mares, J. Hetflejs, J. Schraml, Z. Papouskova, V. Bazant, E.G. Rochow,
V. Chvalovsky, Collection of Czechoslovak Chemical Communications 36 (1971)
3115–3127.
[18] K. Wada, K. Yano, T. Kondo, T.A. Mitsudo, Catalysis Letters 112 (2006) 63–67.
[19] T. Komano, T. Iwasawa, M. Tokunaga, Y. Obora, Y. Tsuji, Organic Letters 7 (2005)
4677–4679.