C-H activation and C-C coupling of 4-methylpyridine using palladium supported on nanoparticle alumina

Article abstract of DOI:10.1016/j.molcata.2008.01.008

The C-H activation and C-C coupling of 4-methyl pyridine to 4,4′-dimethyl-2,2′-bipyridine was studied over palladium catalysts supported on nanoparticle alumina [nano-Al₂O₃(+)]. Even though Pd/Al₂O₃ catalysts are traditionally poor catalysts in this reaction, the Pd/nano-Al₂O₃(+) catalysts prepared via the precipitation method give the highest yield observed to date for this reaction system. The catalytic activity is very dependent on the catalyst preparation as Pd/nano-Al₂O₃(+) catalysts prepared via the wet impregnation method exhibit poor activities in the reaction. Additionally, a catalyst prepared using another nanoparticle alumina with larger particle sizes [nano-Al₂O₃(-)] does not have a significant activity. It was shown that using a commercial bimodal γ-Al₂O₃ support can result in an active catalyst if prepared via the precipitation method, but compared to the best performing catalyst, Pd/nano-Al₂O₃(+), the yields obtained from Pd/bimodal-γ-Al₂O₃ are lower and less reproducible. Pd surface area measurements indicate that the reaction is structure sensitive as there is no correlation between the Pd surface area and the catalytic activity. The reaction is also very sensitive to reactant quality and the 4-methyl pyridine must be distilled over KOH to ensure reproducible yields. Additional experiments indicate that this reaction requires a solid phase for catalysis.

Full text of DOI:10.1016/j.molcata.2008.01.008

Available online at www.sciencedirect.com
Journal of Molecular Catalysis A: Chemical 284 (2008) 141–148
C–H activation and C–C coupling of 4-methylpyridine using
palladium supported on nanoparticle alumina
∗
Luke M. Neal, Helena E. Hagelin-Weaver
Department of Chemical Engineering, University of Florida, Gainesville, FL 32611, United States
Received 3 October 2007; received in revised form 1 January 2008; accepted 3 January 2008
Available online 16 January 2008
Abstract
The C–H activation and C–C coupling of 4-methyl pyridine to 4,4 -dimethyl-2,2 -bipyridine was studied over palladium catalysts supported
ꢀ
ꢀ
on nanoparticle alumina [nano-Al (+)]. Even though Pd/Al catalysts are traditionally poor catalysts in this reaction, the Pd/nano-Al (+)
catalysts prepared via the precipitation method give the highest yield observed to date for this reaction system. The catalytic activity is very
dependent on the catalyst preparation as Pd/nano-Al (+) catalysts prepared via the wet impregnation method exhibit poor activities in the
reaction. Additionally, a catalyst prepared using another nanoparticle alumina with larger particle sizes [nano-Al
(−)] does not have a significant
activity. It was shown that using a commercial bimodal ␥-Al support can result in an active catalyst if prepared via the precipitation method,
but compared to the best performing catalyst, Pd/nano-Al (+), the yields obtained from Pd/bimodal-␥-Al are lower and less reproducible.
2
O
3
2
O
3
2 3
O
2
O
3
2
O
3
2
O
3
2
O
3
2 3
O
Pd surface area measurements indicate that the reaction is structure sensitive as there is no correlation between the Pd surface area and the catalytic
activity. The reaction is also very sensitive to reactant quality and the 4-methyl pyridine must be distilled over KOH to ensure reproducible yields.
Additional experiments indicate that this reaction requires a solid phase for catalysis.
©
2008 Published by Elsevier B.V.
Keywords: Pd/Al2O3; Nanoparticle oxide; Pyridine; C–H activation; Aromatic coupling
1
. Introduction
emitting diodes [8] and in chemiluminescence detection systems
9]. Bipyridines are also commonly used as ligands to metals
[
Bipyridines are receiving increasing attention in the litera-
in various catalyst systems. For example, many copper-based
atom transfer radical polymerization (ATRP) catalysts contain
bipyridine units [10]. Iron and cobalt complexes of bi- or ter-
pyridines have been shown to catalyze the reduction of CO2 and
O2 [11]. Palladium bipyridine complexes have been used as cat-
alyst in several reactions, such as oxidative carbonylation [12],
the Kumada-Corriu reaction [13], and the Suzuki cross-coupling
ture due to their ability to coordinate to transition metal cations
and form complexes with interesting properties [1]. The impor-
tance of metal complexes containing bipyridine ligands has been
revealed in several reviews that have been published in the past
fifteen years on the synthesis of bipyridines, as well as on proper-
ties of metal-bipyridine complexes and their applications [2–6].
Of the transition metals, bipyridine complexes of ruthenium are
by far the most commonly studied systems [2,5]. This is due to
their unique photo- and electrochemical properties. Ruthenium-
bipyridine complexes can absorb photons in the visible light
region and have unique redox properties that can lead to elec-
tron transfer and chemiluminescence, which make them suitable
for application in solar energy conversion (e.g. in solar cells
and artificial photosynthesis systems [6,7]), in organic light-
ꢀ
ꢀ
reaction [14]. Of the bipyridines, 4,4 -dimethyl-2,2 -bypyridine
is of particular interest, since this compound can easily be mod-
ified by reactions with the methyl groups in the 4-positions.
However, due to the poor yields in the coupling reaction of 4-
ꢀ
ꢀ
methylpyridine, the production of 4,4 dimethyl-2,2 -bipyridine
is prohibitively expensive for large scale processes. A kilogram
quantity sells for more than $4,500 [15], while smaller quantities
sell for significantly higher prices per unit weight.
Bipyridines can be formed via a number of pathways. Some
include building the second pyridine ring from a substituted
pyridine, while other methods rely on the coupling of halo-
genated pyridines using transition metal catalysts [16–18].
∗
Corresponding author. Tel.: +1 352 392 6585; fax: +1 352 392 9513.
E-mail address: hweaver@che.ufl.edu (H.E. Hagelin-Weaver).
1
381-1169/$ – see front matter © 2008 Published by Elsevier B.V.
doi:10.1016/j.molcata.2008.01.008

1
42
L.M. Neal, H.E. Hagelin-Weaver / Journal of Molecular Catalysis A: Chemical 284 (2008) 141–148
Scheme 1. Oxidative coupling of 4-methylpyridine using a palladium catalyst.
The disadvantages with these processes are the low yields of
multi-step processes, the cost of the halogenated precursors
and the environmental impact of the waste streams of halide
salts and other byproducts that result from these reactions.
Furthermore, these processes require a high level of subse-
quent purification for applications that are sensitive to halogens,
such as catalysis. Consequently, a one-step process in which
the bipyridine is formed directly from the pyridine reactant
is desirable. The oxidative coupling of 4-methylpyridine to
idation of Pd in these solution experiments, since the catalyst
after exposure to the reaction conditions is in a reduced form
[21]. The maximum isolated yield reported for a 5% Pd/C is
∼2 g product per gram of catalyst, i.e. 40 g/g Pd [26]. However,
yields of 1.5–2 g/g for a 10% Pd/C catalyst are more common
[21,27,28]. Comparing these product yields to reactions using
a homogeneous catalyst complex or halogenated precursors in
solution is difficult. The yields reported as converted reactant
tend to be higher for reactions in solution with halogenated
precursors [17,18,29], while the yield per gram of palladium
generally is higher in reactions where 4-methylpyridine is the
reactant and no solvent is used (See summary of literature yields
in Table 1). The higher conversion of reactant for the reactions
using solvent and halogenated precursors can be outweighed by
the lower yields per gram of palladium, the formed byproducts
(halide salts) and the use of solvent. Furthermore, the unreacted
4-methylpyridine can easily be recovered in the reactions with
no solvent and a heterogeneous catalyst, and there is a poten-
tial for recycling and regeneration of the heterogeneous catalyst.
ꢀ ꢀ
4
,4 -dimethyl-2,2 -bipyridine using palladium on carbon as the
catalyst meets this criterion (Scheme 1). This reaction requires
only a catalyst plus the reactant and the only by-products of the
reaction are water and the terpyridine (Scheme 1). Furthermore,
compared to its halogenated derivatives (the most active bromo-
derivative is available through reaction with commercially
available 2-amino-4-methylpyridine at ∼$200/kg [18,19]), 4-
methylpyridine is relatively inexpensive (less than $40/kg [20])
with lower environmental impact. The disadvantages of this
reaction are the slow reaction rate and the deactivation of the
catalyst [21].
ꢀ
Therefore, there are significant advantages of synthesizing 4,4 -
ꢀ
ꢀ
Early research has shown that 2,2 -bipyridines can be formed
dimethyl-2,2 -bipyridine using a heterogeneous catalyst and no
via coupling of pyridine derivatives over catalysts such as Raney
nickel and palladium on carbon (Pd/C) [22–25]. The results
from these early experiments reveal that while palladium on
carbon is a reasonable catalyst in the coupling reaction of pyri-
dine derivatives, palladium on alumina exhibits poor activity
solvent, particularly if the yields can be increased. Naturally, the
converted reactant yield can also be increased in reactions with
no solvent by simply increasing the catalyst-to-reactant ratio,
although more terpyridine byproduct will be formed in these
cases.
[
22]. No explanation as to why alumina is an inferior sup-
Our hypothesis is that palladium supported on nanoparticle
oxide supports has potential to be a very efficient catalyst. This
hypothesis is based on the fact that nanoparticles have a large
surface area compared to their bulk analogues. Furthermore,
nanoparticles have a high degree of low coordination sites, such
as corners and edges [30]. These low coordination sites may
cause a stronger interaction between the support oxide and the
active metal deposited onto the nanoparticles compared with
more conventional supports. In addition to potentially higher
dispersions of the active metal, these interactions may result in
port has been given in the literature. The early research also
indicated that low-valent palladium is the active form of the
Pd/C catalysts. However, more recent results revealed that the
active catalyst actually contains a Pd(II) species [21] and that
the variation in catalytic activity between batches of Pd/C cat-
alysts could be reduced by simply oxidizing the catalyst before
reaction. However, despite previous improvements of the cata-
lyst, the reaction is slow and suffers from catalyst deactivation.
One of the major limitations of the reaction appears to be reox-
Table 1
ꢀ
ꢀ
Reported literature yields for 4,4 -dimethyl-2,2 -bipyridine forming reactions
Starting material
Amount reactant (mol)
Catalyst
Amount catalyst (g)
Yield (%)
Yield (g/g Pd)
Reference
−
−
−
−
3
3
2
2
2
2
2
2
4
4
4
4
-Br-4-methylpyridine
-Br-4-methylpyridine
-Br-4-methylpyridine +
-SnBu3-4-methylpyridine
-Br-4-methylpyridine
-Br-4-methylpyridine
-Methylpyridine
-Methylpyridine
-Methylpyridine
-Methylpyridine
2 × 10
2 × 10
23 × 10
27 × 10
5% Pd/C, Zn
5% Pd/C, Zn
Pd(PPh3)4
0.3
0.11
0.95
19
19
67
2.3
6.6
32.5
[17]
[17]
[18]
3
3
−
−
3
3
5 × 10
5 × 10
1.03
NiBr2
NiBr2
10% Pd/C
5% Pd/C
10% Pd/C
5% Pd/C
0.07
0.3
4
8.93
28
47
93
7
9
6
10.5 [Ni]
4.9 [Ni]
17.3
40.0
14.3
[29]
[29]
[27]
[26]
[28]
This work
2.06
7.19
75 × 10−
3
0.7
26
52

L.M. Neal, H.E. Hagelin-Weaver / Journal of Molecular Catalysis A: Chemical 284 (2008) 141–148
143
unique catalytic properties. It is possible that this is the reason for
the high activity observed at low temperatures in the palladium-
catalyzed oxidation of methane when nanoparticle oxides are
used as supports [31]. Consequently, not only the large surface
area, but also the intrinsic properties of nanoparticle oxides can
result in unique catalytic activities of nanoparticle-supported
catalysts.
The main objective in this work is to determine if palladium
supported on nanoparticle alumina can be an efficient catalyst
in the coupling of pyridines despite previous research showing
that commercial palladium on alumina is a poor catalyst in this
reaction. Part of the objective is also to determine the effects
of catalyst preparation on the catalytic activity and if the active
species is likely to be a dissolved homogeneous complex instead
of heterogeneous palladium surface species.
under reflux for 72 h. After a complete reaction the flask con-
tents were filtered using a glass micro-fiber filter and washed
with chloroform to dissolve the product. The chloroform, water
and unreacted 4-methylpyridine were removed using a rotary
evaporator.
Selected samples were purified via sublimation. These exper-
iments indicate that at least 75% of the raw yield is the desired
product. This, however, is a low number since only about 85%
of the original sample mass is recovered in the sublimate and
residue due to the difficulty in removing the product from the
sublimation apparatus. Consequently, the yields are reported as
raw yields in the paper. The sublimate, residue, and raw products
were characterized using NMR. The only significant product
found was the bipyridine. The residue did not redissolve well
into the chloroform. The soluble fraction of the sublimation
residue showed little evidence of organic compounds other than
the product. Based upon NMR it is likely that the major portion
of the non-sublimated impurities is inorganic residues from the
catalysts.
2
. Experimental
2
.1. Catalyst preparation
The catalysts were prepared using commercially available
2.3. Catalyst characterization
alumina nanoparticles [32]. Two catalyst preparation meth-
ods were used; wet impregnation and precipitation. In the wet
impregnation method the support powders were dispersed in an
aqueous solution of palladium nitrate (Fluka or Alfa Aesar). The
water was then boiled off until a paste consistency was achieved.
Brunauer–Emmett–Teller (BET) surface area measurements
were performed on a Quantachrome NOVA 1200 instrument.
Fresh catalysts were outgassed under vacuum for 3 h at room
temperature before the measurements. Catalysts that had been
stored for longer times were outgassed at 105 C for at least 1 h
before the BET analysis. The N2 adsorption was performed over
five isotherms, which gave roughly linear fits.
◦
◦
This paste was dried in a muffle furnace at 105 C overnight. The
◦
dried samples were ground and then calcinated at 450 C for 3 h
to decompose the palladium nitrate and form palladium oxide
on the support.
To determine the dispersion of Pd on the catalysts carbon
monoxide chemisorption experiments were performed using a
ChemBET 3000 instrument. The fresh catalysts were reduced
In the precipitation method, the support was dispersed into
a solution of palladium nitrate. The mixture was then titrated
with a NaOH solution, which formed Pd(OH)2 on the sup-
port [33]. The amount of NaOH used in these experiments
corresponds to 50% stoichiometric excess. The resulting mix-
ture was aged overnight at room temperature before it was
filtered. The recovered catalyst was rinsed by stirring in water
overnight, followed by another filtration. As for the catalysts
prepared via the wet impregnation method, the precipitated sam-
◦
for 2 h at 170 C using a 5% hydrogen in nitrogen gas mixture.
The samples were then out-gassed for 1 h in helium before the
◦
pulse titration experiments with CO at 25 C and 1 atm. The low
reduction temperature was used to avoid excessive sintering or
spalling of the Pd particles during the reductive treatment and
was chosen to be close to reaction temperature.
◦
◦
ples were dried over night at 105 C and calcinated at 450 C
for 3 h.
3. Results and discussion
Several commercial catalyst supports and catalysts were also
used for catalyst preparation or used as received, and tested
for activity in the coupling reaction of 4-methylpyridine. These
3.1. Effects of catalyst preparation method
In the first set of experiments 10% (by weight) of pal-
ladium was deposited onto the Nanoactive aluminum oxide
plus support [Pd/nano-Al2O3(+)] using the wet impregnation
method. This catalyst was tested for activity in the coupling
reaction of 4-methylpyridine. As expected for a palladium on
alumina catalyst, this nano-particle supported catalyst did not
exhibit any significant activity in the reaction (Table 2, entry
2
include 5% Pd/C (Alfa Aesar, surface area (SA): 695 m /g), 5%
2
Pd/Al2O3 (Alfa Aesar Pd/␥-alumina, SA: 155 m /g), ␥-Al2O3
2
(
Alfa Aesar high surface area bimodal, SA: 260 m /g) and acti-
2
vated carbon (Calgon F 400, SA: 765 m /g).
2
.2. Reaction conditions
1). When the precipitation method was used, however, a 10%
The 4-methylpyridine (Aldrich or Across) was distilled over
Pd/nano-Al2O3(+) catalyst yielded a significant amount of the
ꢀ
ꢀ
KOH or NaOH prior to use. In a typical reaction run 1 g of
catalyst was placed in a round bottom flask along with 10 g
of the distilled 4-methylpyridine. The reaction mixture was
evacuated and an oxygen atmosphere introduced before it was
4,4 -dimethyl-2.2 -bipyridine (Table 2, entries 2 and 3). In fact,
the 20–25 g product per g of palladium corresponds to 2–2.5 g
product per gram of catalyst, which is equal to or higher than
the maximum yield reported from a palladium on carbon cata-
lyst [21]. It was also shown that the Pd content can be reduced
◦
heated to the boiling point (145 C). The reaction proceeded

1
44
L.M. Neal, H.E. Hagelin-Weaver / Journal of Molecular Catalysis A: Chemical 284 (2008) 141–148
Table 2
Product yields and catalyst surface areas of various catalysts prepared and tested in the coupling reaction of 4-methylpyridine
Catalyst descriptiona
4-Methyl-pyridine b
Raw product (g/g Pd)c
Specific surface area (m /g)
2
Entry
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
10% Pd impregnated on nano-Al2O3(+)
10% Pd precipitated on nano-Al2O3(+)
10% Pd precipitated on nano-Al2O3(+)
5% Pd precipitated on nano-Al2O3(+)
5% Pd/C commercial
1
1
1
1
1
1
2
2
2
2
3
3
3
3
3
3
3
4
4
4
4
4
4
4
5
5
5
<1d
155
205
165
180
695
155
170
170
695
155
165
695
165
155
200
25.4
20.0
48.6
16.4
2.6
1.4
18.2
7.6
2.4
52.6
14.4
55.2
8.2
4.2
2.4
2.6
14
44
5.8
24
15.1
6.6
34
d
5% Pd/Al2O3 commercial
d
5% Pd impregnated on nano-Al2O3(+)
5% Pd precipitated on nano-Al2O3(+)
5% Pd/C commercial
5% Pd/Al2O3 commercial
5% Pd precipitated on nano-Al2O3(+)
5% Pd/C commercial
5% Pd precipitated (KOH) on nano-Al2O3(+)e
5% Pd precipitated on nano-Al2O3(−)
5% Pd precipitated on commercial ␥-Al2O3f
5% Pd Precipitated on activated carbon
5% Pd Impregnated on nano-Al2O3(+) + KOH
5% Pd Precipitated on nano-Al2O3(+) (Alfa)
5% Pd Precipitated on nano-Al2O3(+)
5% Pd Precipitated on nano-Al2O3(−)
5% Pd Precipitated on Commercial ␥-Al2O3f
5% Pd precipitated on activated carbon
5% Pd impregnated on activated carbon
5% Pd precipitated on commercial ␥-Al2O3f
5% Pd/C commercial
d
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
g
h
36
52
31
5% Pd precipitated on nano-Al2O3(+)
5% Pd Precipitated on nano-Al2O3(+) + H2O
i
Reaction conditions are given in the Section 2 (Experimental section).
a
Nano-Al2O3(+): nanoparticle alumina (Nanoactive aluminum oxide plus), nano-Al2O3(−): nanoparticle alumina (Nanoactive aluminum oxide). Unless otherwise
stated the palladium(II) nitrate source is Fluka.
b
4-Methylpyridine distillates: 1: 4-methylpyridine distilled over KOH, 2: 4-methylpyridine distilled over NaOH, 3: 4-methylpyridine treated with KOH over night
before distilling over the same KOH, 4 and 5: 4-methylpyridine treated with KOH over night then decanted and distilled over fresh KOH.
c
d
e
f
ꢀ
ꢀ
ꢀ
ꢀꢀ
ꢀ
ꢀ ꢀꢀ
Product: 4,4 -dimethyl-2,2 -bipyridine. Raw yield contains a small amount of the terpyridine: 4,4 ,4 -trimethyl-2,2 ,6 2 -terpyridine as the only byproduct.
ꢀ
ꢀ
There is very little 4,4 -dimethyl-2,2 -bipyridine product in these runs. The solids recovered appear to be mostly catalyst and organic byproducts.
Catalyst prepared via precipitation using KOH instead of NaOH.
Alfa Aesar high surface area bimodal ␥-alumina.
.09 g of KOH was added to the reaction mixture.
Palladium(II) nitrate source was Alfa Aesar not Fluka.
.3 g of H2O was added to the reaction mixture.
g
h
i
0
0
to 5% Pd on alumina without substantially reducing the product
yield (Table 2, entry 4). In other words, the yield per gram of
palladium can be doubled by going from a 10% to a 5% Pd/nano-
Al2O3(+). Consequently, a yield of 50 g raw product per gram
palladium can be obtained with this catalyst. The 5% Pd/nano-
Al2O3(+) catalyst is more active than the commonly used Pd/C
catalyst despite the inactivity reported for commercial Pd/Al2O3
catalysts. It is also interesting to note that the catalytic activity
is very dependent on the catalyst preparation method. While the
precipitation method yields a catalyst with the highest observed
catalytic activity to date, the impregnation technique results in
a very low activity or an inactive catalyst. Both the impregnated
and the precipitated catalysts have specific surface areas in the
catalysts are more reproducible than the yields obtained from
different runs on the same batch of commercial Pd/C (Table 2,
Entries 5 and 25). This is likely due to variations in the PdO
concentration on the “as received” Pd/C catalysts [21]. Since
our catalysts are calcined in air and not reduced before reaction,
the amount of PdO on the Pd/nano-Al2O3(+) catalysts is almost
certainly more constant compared to the commercial “Pd(0)”/C.
To test whether or not the differences in the catalytic activity
between catalysts prepared by the impregnation and precipita-
tion methods only applies to alumina supports, impregnated and
precipitated palladium on activated carbon catalysts where pre-
pared. The precipitated catalyst gave a modest yield (15.1 g/g Pd,
entry 22, Table 2), which is comparable to the commercial Pd/C
catalyst, while the impregnated Pd/C catalyst gave a low yield
(6.6 g/g Pd, entry 23, Table 2). This indicates that the precipita-
tion method is superior to the impregnation method regardless
of the support used, although the difference is more drastic for
the nanoparticle alumina support.
2
range of 150–200 m /g (Table 2, entries 1–4). Therefore, the
support surface areas cannot explain the differences in activities
between these catalysts. As can be seen in Table 2, the results are
reproducible (entries 4, 11 and 26). In fact, the yields obtained
from different preparations of precipitated Pd/nano-Al2O3(+)

L.M. Neal, H.E. Hagelin-Weaver / Journal of Molecular Catalysis A: Chemical 284 (2008) 141–148
145
Table 3
alumina despite the fact that the surface areas of these supports
are similar. The catalysts prepared via precipitation onto these
BET surface areas of the supports used in the catalyst preparations
2
Catalyst support
Surface area (m /g)
three alumina supports all have surface areas between 150 and
2
2
05 m /g. These results indicate that neither the surface area
Nanoactive alumina plus
Nanoactive alumina
Commercial ␥-alumina bimodal (Alfa Aesar)
Activated carbon (Calgon F 400)
695
275
260
765
of the bare support nor the final surface area of the prepared
catalysts is the sole determining factor of the catalytic activity.
A further indication that the initial support surface area is not
the main factor in determining the catalytic activity can be seen
when comparing alumina and carbon supported catalysts. While
a catalyst prepared via precipitation of palladium onto Nanoac-
3
.2. Catalyst support effects
As expected, the commercial 5% Pd/Al2O3 catalyst exhib-
2
tive aluminum oxide plus (695 m /g, Table 3) gives a product
yield of ∼50 g/g Pd, a catalyst prepared using the same method
2
but with an activated carbon support (surface area: 765 m /g,
ited poor activity and resulted in little, if any, product under
the reaction conditions of the experiments. This is most likely
due to the fact that the commercial 5% Pd/Al2O3 catalyst con-
sists mainly of Pd(0), in contrast to commercial Pd/C catalysts
which appear to have a relatively high, albeit varying, surface
Pd(II) content. The yield of the commercial 5% Pd/C catalyst
agrees with previous results [21]. It is lower than the high-
est yield observed using this catalyst, due to the fact that the
catalyst was used as received without oxidation treatment. To
determine if it is the nature of the nanoparticle alumina sup-
port or the preparation method that is responsible for the high
catalytic activity, a 5% Pd/Al2O3 catalyst was prepared with
a commercial alumina support using the precipitation method.
The yield for this catalyst varied widely (Table 2, entries 15,
Table 3) only gives a yield of 15 g/g Pd. Thus, despite the slightly
higher surface area of the activated carbon it does not result in
a catalyst as active as the one supported on nano-Al2O3(+).
Aside from giving the best activity of any of the supports
tested, the nano-Al2O3(+) catalysts exhibited better repro-
ducibility than any of the catalysts including the commonly used
commercial palladium on carbon. The only major activity vari-
ations seen for the catalysts supported on nano-Al2O3(+) can be
attributed to reactant quality.
3.3. Palladium surface areas
The palladium surface areas were determined after reduction
of the PdO on the catalyst surfaces using CO adsorption and the
volume of adsorbed CO is given in Table 4 for selected catalysts.
Performing CO titration measurements on reduced catalysts do
notnecessarilyresultinagoodmeasureofthecatalyticallyactive
surface area since the active phase on these catalysts is PdO
and not Pd metal. This is particularly the case when the cata-
lyst before reduction consists of Pd metal or a mixture of PdO
and Pd, as in the case of the commercial Pd/C and Pd/Al2O3
catalysts. However, in cases where the original catalysts con-
sist solely of PdO (as is the case for the nanoparticle-supported
catalysts), if the same mild reduction conditions are used for
all catalysts, it should be possible to observe trends and obtain
qualitative results from the Pd surface area measurements. CO
chemisorption measurements on supported Pd catalysts are fur-
ther complicated by the dependence of the Pd:CO stoichiometry
on the dispersion since CO can adsorb in linear, bridge and hol-
2
1 and 24). On average the catalyst does exhibit a fair activity,
but it is not as active as the nanoparticle-supported Pd/Al2O3
catalyst. Another catalyst was prepared using the precipitation
method and a second nanoparticle alumina sample [Nanoactive
aluminum oxide, nano-Al2O3(−)] as support. The surface area
2
of this alumina sample (275 m /g) is lower than the Nanoactive
2
aluminum oxide plus sample (695 m /g), but on the same order
as the commercial ␥-alumina support (260 m /g) (see Table 3).
The catalytic activity of this catalyst is significantly lower than
that obtained on the commercial Pd/C catalyst and the palla-
dium on the Nanoactive aluminum oxide plus or commercial
2
␥
-alumina supports. Consequently, there is a significant differ-
ence in the catalytic activities of the catalysts prepared with the
two Nanoactive aluminum oxide supports. It is interesting to
note that the catalyst supported on the nano-Al2O3(−) gives a
lower yield than the catalyst prepared using the commercial ␥-
Table 4
Results from CO chemisorption measurements of selected catalysts prepared and tested in the coupling reaction of 4-methylpyridine
a
2
2
Catalyst
CO adsorbed (␮l/g cat)
Pd surface area (m /g catalyst)
Product yield (g/m Pd)
Nano-Al2O3(+)
0
8480
4550
1530
280
1880
2530
310
0
N/A
0.08
0.17
0.06
0.29
0.02
0.20
0.68
0.41
3.4
1
5
5
5
5
5
5
5
5
0% Pd precipitated on nano-Al2O3(+)
% Pd precipitated on nano-Al2O3(+)
% Pd impregnated on nano-Al2O3(+)
% Pd precipitated on nano-Al2O3(−)
% Pd/Al2O3 commercial
29.4
15.8
5.3
1.0
6.5
8.8
1.1
0.8
0.5
% Pd/C commercial
% Pd precipitated on activated carbon
% Pd Impregnated on activated carbon
% Pd precipitated on commercial ␥-Al2O3
240
140
f
a
The Pd surface area has been calculated assuming a Pd:CO stoichiometry of 2:1 and a surface atom density of 1.42 × 10 atoms/cm2.
15

1
46
L.M. Neal, H.E. Hagelin-Weaver / Journal of Molecular Catalysis A: Chemical 284 (2008) 141–148
low binding modes on the surface [34]. Literature data indicate
that a stoichiometry of two Pd surface atoms per CO molecule
adsorbed is appropriate for high dispersions and a surface atom
3.4. Impact of reactant quality
As mentioned, the catalysts prepared using the nanoparticle
alumina(+) support for the most part exhibited good repro-
ducibility. There were slight variations in the specific surface
areas and yields for repeated samples as seen in Table 2 (Entries
4, 11, 13 and 26). However, the results are very sensitive to
the quality of the 4-methylpyridine used. The 4-methylpyridine
must be distilled over KOH (Distillates 1 and 3–5 in Table 2)
to give good results, preferably doubly distilled. The product
yields obtained with a reactant that had been distilled over a
small amount of NaOH (Distillate 2) were markedly lower than
the yields for reactants distilled over KOH (compare entries 8
and 11 in Table 2). It was noticed that distillate 2 discolored
more quickly over time compared with the other distillates. In
addition, distillate 3, which was treated with KOH over night
and then distilled over the same KOH, turned cloudy or tur-
bid over time. This may explain the lower yields obtained for
entries 15 and 16 in Table 2, since those were the last two
runs using distillate 3. The best and most reproducible results
are obtained if the 4-methylpyridine is treated with KOH over
night then decanted and distilled over fresh KOH. This treatment
results in a clear liquid, which will not discolor or turn opaque
over a reasonable time (on the order of months). These results
tend to indicate that an impurity is present in the lower qual-
ity reactants (distillate 2 and 3) and that this impurity inhibits
the reaction. As the primary purpose of adding KOH to the
sample during distillation is for drying, it seems possible that
water is the impurity. To test this hypothesis, parallel reactions
were run with the same catalyst preparation [5% Pd precipi-
tated onto nano-Al2O3(+)] and the same reactant distillate. To
one of these reactions 0.3 g of water per gram of catalyst was
added to the reactant. The water-free specimen gave a typi-
cal yield of 52 g/g Pd (entry 26, Table 2). The sample with
the added water yielded 31 g/g catalyst (entry 27, Table 2).
While this reveals that water exhibits an inhibiting effect on
the reaction, it does not explain the considerably lower yields
(18.2 g/g Pd) obtained from distillate 2 (cf entries 8 and 27
in Table 2). Furthermore, at a product yield of 50 g/g Pd the
amount of water formed is slightly lower than the added 0.3 g/g
catalyst. Consequently, the presence of water cannot be solely
responsible for the inhibiting effects seen in the lower quality
distillates.
15
2
density (Cm) of 1.42 × 10 /cm based on the cubo–octahedral
geometry is reasonable for small Pd crystallites [35]. While the
exact Pd:CO stoichiometry may deviate slightly from the 2:1
ratio used here, the assumptions made in calculating the Pd
surface area should give sufficient accuracy for comparisons
between the different catalysts in the study, particularly con-
sidering that the active phase is PdO rather than Pd on these
catalysts.
The 5% precipitated nano-Al2O3(+) has a very high Pd sur-
face area compared to the other catalysts in this study (Table 4).
In fact, the Pd surface area of this catalyst is more than two times
that ofthe commercial5% Pd on activated carboncatalyst, which
is the commercial catalyst with the highest Pd surface area. The
commercial 5% Pd/Al2O3 catalyst also has a high dispersion but,
as mentioned, this catalyst exhibits a low activity due to a low
PdO concentration rather than a lack of Pd surface area. The 5%
impregnated nano-Al2O3(+) has a surprisingly high Pd disper-
sion considering its low activity. The high yields per Pd surface
area for the impregnated palladium on carbon and the palladium
precipitated on nano-alumina(−) catalysts are likely an artifact
of impurities in the product and the low dispersions of the cat-
alysts. For low yields (∼0.1 g for 0.7 g catalyst or 2.9 g/g Pd)
trace impurities from the reactant and catalyst can be a signifi-
cant portion of the raw product mass, and due to the nanoparticle
supports used, it is difficult to separate these impurities from the
products. Even though the same is partly true for the precipitated
palladium on carbon catalyst, i.e. the reported yield per surface
area is unrealistically high it appears that this catalyst does give
a decent yield despite the low dispersion. The Pd surface areas
of the precipitated Pd/C and Pd/nano-Al2O3(−) are on the same
order, but the yield is higher for the Pd/C catalyst. In fact, the
data suggest that the palladium on the surface of the precipitated
Pd/C catalyst is more active than the palladium on the surface of
the commercial Pd/C. Comparing the data in Tables 2 and 4 it
is evident that the modest yields obtained from the precipitated
Pd/nano-Al2O3(−) and the precipitated and impregnated Pd/C
catalysts can be explained in part by low dispersions. Another
surprising result is the relatively high Pd surface area of the 10%
precipitated nano-Al2O3(+). It has roughly twice the metal sur-
face area of the 5% catalyst but exhibits no increase in yield. The
most striking result, however, is that the Pd surface area of the
The palladium nitrate source also had an effect on the cat-
alytic activity. Catalysts prepared using the palladium nitrate
from Fluka resulted in higher catalytic activities than catalysts
prepared using the palladium nitrate from Alfa Aesar. During the
catalyst preparation it was observed that the palladium nitrate
from Alfa Aesar did not dissolve as well in the deionized water
as the Fluka Pd(NO3)2. In fact, the residual solids from the Alfa
Aesar palladium nitrate did not dissolve even after addition of
acid (HNO3 to a pH of 1.0). XRD analysis of both materials indi-
cated the presence of a palladium oxide or related phase in the
palladium nitrate from Alfa Aesar. The Fluka Pd(NO3)2 did not
contain these phases. Consequently, it is important to check the
catalyst precursor quality before preparing supported palladium
catalysts.
5
% Pd precipitated on commercial ␥-Al2O3 is very low despite
the relatively good yield. The low dispersion of this catalyst is
the main reason for the very high yield reported per palladium
surface area. These results strongly indicate a structure sensitive
reaction, i.e. only a portion of the surface Pd atoms are active.
Consequently, the catalytic activity does not correlate with the
measured Pd surface areas. This can be a result of the fact that
the surface areas are determined on reduced catalysts, while
the active catalyst is the unreduced form. While the support is
important to give catalysts with high Pd surface areas, it is evi-
dent that this is not a sufficient criterion for an active catalyst.
The Pd, or rather PdO, on the surface must also have the correct
structure.

L.M. Neal, H.E. Hagelin-Weaver / Journal of Molecular Catalysis A: Chemical 284 (2008) 141–148
147
3
.5. Homogeneous versus heterogeneous catalysis
This is significantly less product compared to a fresh catalyst,
however, the combined product yield from the 24 h and the
72 h reactions is consistent with an uninterrupted 72 h experi-
ment (there is very little product formed after the initial 72 h
at the reaction conditions). While the results clearly show that
a recovered catalyst is still active, it is evident that the recov-
ered catalyst has a lower activity compared to a fresh catalyst.
One of the main reasons for the lower activity of a catalyst
after exposure to the reaction conditions for 24 h or more, is
probably reduction of the active Pd(II) species. However, it is
also possible that some leaching does occur, even though any
palladium in solution does not appear to be an active cata-
lyst. From the experiments on commercial Pd/C it is evident
that the 24 h product yield is higher (∼30 g/g Pd) than for the
Pd/nano-Al2O3(+) catalyst (∼20 g/g Pd), even though the 72 h
yield is considerably higher for the nano-Al2O3(+) supported
catalyst. However, the recovered catalyst is much less active
(Table 5). This indicates that the reaction is faster on Pd/C, but
the catalyst also deactivates faster than the Pd/nano-Al2O3(+)
catalyst.
With palladium–pyridine systems there is a natural question
of whether supported catalysts are truly heterogeneous, or if the
actual active species is dissolved into the solution. Palladium(II)
ions coordinate easily to the nitrogen of 4-methylpyridine and
perhaps even more strongly to the bipyridine product formed in
the reaction. Consequently, it is possible that the surface pal-
ladium is simply a precursor, or a source, to active palladium
ions in solution. Other palladium-catalyzed coupling reactions,
such as the Heck reaction, do exhibit palladium leaching when
heterogeneous catalysts are used [36,37], and there is still a
debate as to whether these heterogeneous palladium catalysts
are active catalysts or simply precursors to a dissolved active
species [38]. Therefore, a set of experiments was designed to
probe if a dissolved palladium species is active in this reac-
tion. If palladium is dissolved into the reaction mixture and this
palladium is catalytically active, it would be expected that the
reaction proceed after the heterogeneous catalyst (support) has
been removed. One reaction was taken out of the oil bath after
2
4 h and the reaction mixture was filtered hot to remove the
Even though the above results imply that no catalytically
activespeciesispresentinsolution, thereareindicationsofpalla-
dium leaching from the support. Previous results have indicated
that the palladium surface concentration is lower after reaction
[21]. In these reaction runs a palladium mirror could be observed
in some experiments. However, generally the active Pd/nano-
Al2O3(+) catalyst prepared via precipitation displayed little or
no Pd mirror on the reaction flask after a completed reaction. In
contrast, significantPdmirrorswereseenonseveralofthepoorly
performingcatalysts, suchasthePd/nano-Al2O3(+)catalystpre-
pared via impregnation. Some mirroring was observed also in
the commercial Pd/C catalyst. If the catalytic action of these cat-
alysts was reliant solely upon a dissolved palladium species, it
might be expected that palladium mirroring (which is evidence
of palladium leaching from the support) would accompany a
significant catalyst activity. In contrast, palladium dissolution is
observed mainly for catalysts with poor activity. Thus, this can
be taken as another indication that the active species is on the
surface of the heterogeneous catalyst and not dissolved into the
solution.
solid catalyst. The filtrate was then returned to reflux under oxy-
gen for an additional 72 h. The product yield recovered from
a 24 h experiment is lower than the product yield after a 72 h
reaction (17 g/g Pd versus 50 g/g Pd, entry 1, Column 3 in
Table 5). This reveals a slow reaction rate and the need for
a 72 h reaction time. Furthermore, the product recovered after
2
4 h with solid catalyst plus 72 h without solid catalyst (i.e. after
returning the filtrate to the reaction conditions) is very close to
the yield of a 24 h reaction (22 g/g Pd). This is evidence that
the reaction is heterogeneous, or at least that the presence of
a heterogeneous catalyst is necessary for the reaction to pro-
ceed.
A set of similar experiments was run by reacting a catalyst-
reactant mixture for 24 h and then use the recovered catalyst
in a subsequent reaction. The recovered catalyst from a 24 h
reaction was washed with chloroform, dried and then reloaded
with fresh 4-methylpyridine. After 72 h at the reaction condi-
tions, 24 g of product was recovered per gram of the reloaded
catalyst [Pd/nano-Al2O3(+), entry 1, Column 4 in Table 5].
Table 5
Results from experiments at different reaction times to probe homogeneous vs. heterogeneous catalysis
Catalyst description a
Raw yield (g/g Pd) 1st-day
b
Raw yield (g/g Pd) 4th-day
b
Entry
Recovered catalystc
1
2
5% Precipitated Pd/nano-Al2O3(+) (1st-day and 4th-day)
5% Pd/C Commercial (1st-day and 4th-day)
17
32
24
5
Recovered reactant mixtured
3
4
5% Precipitated Pd/nano-Al2O3(+) (1 day with and 3 days without catalyst)
5% Pd/C commercial (1 day with and 3 days without catalyst)
N/A
N/A
22
30
a
b
c
Nano-Al2O3(+): nanoparticle alumina (Nanoactive aluminum oxide plus).
Raw yield of 4,4 -dimethyl-2,2 -bipyridine, which contains a small amount of the terpyridine: 4,4 ,4 -trimethyl-2,2 ,6 2 -terpyridine as the only byproduct.
The 1st-day yield is the product recovered during filtration after the first 24 h. The 4th-day yield is the product obtained from a recovered 1st-day catalyst after
ꢀ
ꢀ
ꢀ
ꢀꢀ
ꢀ
ꢀ ꢀꢀ
reaction with fresh 4-methylpyridine for three additional days.
d
Reaction was run for 24 h with catalyst. The reaction mixture was filtered and the catalyst removed. The filtrate was then returned to the reaction conditions for
an additional 72 h.

1
48
L.M. Neal, H.E. Hagelin-Weaver / Journal of Molecular Catalysis A: Chemical 284 (2008) 141–148
3
.6. Precipitation base
References
[
[
1] C. Kaes, A. Katz, M.W. Hosseini, Chem. Rev. 100 (2000) 3553.
2] V. Balzani, G. Bergamini, F. Marchioni, P. Ceroni, Coord. Chem. Rev. 250
2006) 1254.
3] G.R. Newkome, A.K. Patri, E. Holder, U.S. Schubert, Eur. J. Org. Chem.
2004) 235.
[4] U.S. Schubert, C. Eschbaumer, Angew. Chem. Int. Ed. 41 (2002) 2892.
NaOH or KOH were used as the base to force palladium pre-
cipitation onto the support. Both these bases resulted in highly
active catalysts with similar yields (entries 11 and 13, Table 2).
To determine if the presence of trace amounts of base was
responsible for the activity of the precipitated samples, KOH
was added to a reaction with an impregnated catalyst. This cata-
lyst system showed no significant increase in activity. If the base
is crucial for the activity it is possible the Na or K is required in
the deposition stage, i.e. a closer interaction between the Pd and
the Na or K is necessary.
(
[
(
[
[
5] C.D. Clark, M.Z. Hoffman, Coord. Chem. Rev. 159 (1997) 359.
6] V. Balzani, A. Juris, M. Venturi, S. Campagna, S. Serroni, Chem. Rev. 96
(
1996) 759.
[
[
7] K. Kalyanasundaram, Coord. Chem. Rev. 46 (1982) 159.
8] R.C. Evans, P. Douglas, C.J. Winscom, Coord. Chem. Rev. 250 (2006)
2093.
9] L. G a´ miz-Gracia, A.M. Garcı ˇı a-Campa n˜ a, J.J. Soto-Chinchilla, J.F.
Huertas-P e´ rez, A. Gonz a´ lez-Casado, Trend. Anal. Chem. 24 (2005) 927.
[
4
. Conclusions
[
[
10] N.M.L. Hansen, K. Jankova, S. Hvilsted, Eur. Polym. J. 43 (2007) 255.
11] D.L. Feldheim, C.J. Baldy, P. Sebring, S.M. Hendrickson, C.M. Elliott, J.
Electrochem. Soc. 142 (1995) 3366.
While the palladium impregnated on nano-Al2O3(+) exhibits
little or no activity in the coupling reaction of 4-methylpyridine,
nano-Al2O3(+) supported catalysts prepared via the precip-
itation method give significant yields. Despite the fact that
commercial Pd/Al2O3 is not an active catalyst in this reaction,
the nano-Al2O3(+) was shown to be a viable support for the
coupling reaction. The yield obtained from the palladium pre-
cipitated onto nano-Al2O3(+) is significantly higher, the highest
reported for this reaction system, and more reproducible com-
pared to the yields obtained from commercial Pd/C catalysts,
which are the commonly used catalysts in this reaction. Addi-
tionally, it was found that traditional alumina supported catalysts
could be active for this reaction if prepared by precipitation.
However, the yields are not as reproducible or as high as those
obtained for the nano-Al2O3(+) support. The sensitivity to cat-
alyst preparation indicates that the activity of this catalyst is not
based solely upon the surface area of the support. The lack of cor-
relation between the measured Pd surface and the catalyst yields
strongly suggests a structure sensitive reaction. Furthermore,
experiments reveal that the differences in activities between cat-
alysts prepared via precipitation and impregnation is not due
to the presence of bases (NaOH or KOH) in the reaction mix-
ture. To ensure high product yields the 4-methylpyridine must
be distilled over KOH. Purification by sublimation followed by
[12] H. Ishii, M. Goyal, M. Ueda, K. Takeuchi, M. Asai, Appl. Catal. A 201
2000) 101.
(
[
[
[
13] F. Tsai, B. Lin, M. Chen, C. Mou, S. Liu, Tetrahedron 63 (2007) 4304.
14] W. Wu, S. Chen, F. Tsai, Tetrahedron Lett. 47 (2006) 9267.
15] http://gfschemicals.com/chemicals/gfschem-127.asp,
7/13/07.
[16] L.A. Summers, Adv. Heterocycl. Chem. 35 (1984) 281.
accessed
on
0
[
17] G. Cravotto, M. Beggiato, A. Penoni, G. Palmisana, S. Tollari, J. L e´ v eˆ que,
W. Bontrath, Tetrahedron Lett. 46 (2005) 2267.
[
[
18] M. Heller, U. Schubert, J. Org. Chem. 67 (2002) 8269.
19] 2-Amino-4-methylpyridine from Fluka >98% (Sigma–Aldrich), https://
www.sigmaaldrich.com/catalog/search/ProductDetail/FLUKA/08630,
accessed on 08/30/07.
[20] 4-Methylpyridine 98% available from Fluka (Sigma–Aldrich), http://www.
sigmaaldrich.com/catalog/search/ProductDetail/FLUKA/80240, accessed
08/30/07.
[
21] H. Hagelin, B. Hedman, I. Orabona, T. A˚ kermark, B. A˚ kermark, C.A. Klug,
J. Mol. Catal. A 164 (2000) 137.
[22] G.D.F. Jackson, W.H.F. Sasse, C.P. Whittle, Aust. J. Chem. 16 (1963) 1126.
[
[
[
[
[
23] P.E. Rosevear, W.H.F. Sasse, J. Heterocycl. Chem. 8 (1971) 483.
24] P.E. Rosevear, W.H. Sasse, Patent App. No. 40930/72, 1973, p. 1.
25] S. Munavalli, M. Gratzel, Chem. Ind. (1987) 722.
26] P.K. Ghosh, T.G. Spiro, J. Am. Chem. Soc. 102 (1980) 5543.
27] P.A. Adcock, F.R. Keene, R.S. Smythe, M.R. Snow, Inorg. Chem. 23 (1984)
2336.
[
[
[
28] G. Sprintschnik, H.W. Sprintschnik, P.P. Kirsch, D.G. Whitten, J. Am.
Chem. Soc. 99 (1977) 4947.
29] K.W.R. de Fran c¸ a, M. Navarro, E. L e´ onel, M. Durandetti, J.-Y. N e´ d e´ lec, J.
´
ꢀ
ꢀ
NMRmeasurementsverifiesthattheproductis4,4 dimethyl2,2
Org. Chem. 67 (2002) 1838.
30] L. Guczi, A. Beck, A. Horv a´ th, D. Horv a´ th, Top. Catal. 19 (2002) 157.
bipyridine and that the raw yields are at least 75% or more of the
desired product. Experiments also reveal that a heterogeneous
phase is necessary for the reaction to proceed at a significant rate.
[31] G.B. Hoflund, Z. Li, W.S. Epling, T. G o¨ bel, P. Schneider, H. Hahn, React.
Kinet. Catal. Lett. 70 (2000) 97.
[
32] Nanoactive aluminum oxide plus and Nanoactive aluminum oxide
available from nanoscale: http://www.nanoactive.interkan.net/content/
nanoactive materials/nanoactive home.asp, accessed on 03/07/07.
33] B. Didillon, E. Merlen, T. Pag e` s, D. Uzio, Stud. Surf. Sci. Catal. 118 (1998)
41.
Acknowledgements
[
[
Acknowledgment is made to the Donors of the American
Chemical Society Petroleum Research Fund for support of this
research. The authors are grateful to Prof. Angela Lindner
34] T. Lear, R. Marshall, J.A. Lopez-Sanchez, S.D. Jackson, T.M. Klap o¨ tke,
M. B a¨ umer, G. Rupprechter, H.-J. Freund, D. Lennon, J. Chem. Phys. 133
(
2005) 174706.
(
Environmental Engineering Sciences, EES) for a fruitful col-
[35] P. Canton, G. Fagherazzi, M. Battagliarin, F. Menegazzo, F. Pinna, N.
laboration and for providing us with experimental equipment to
initiate the experiments and to Dr. David Mazyck (EES) for pro-
viding us with an activated carbon sample. We would also like
to thank Melani Wetana and Charlotta Brodin for their initial
work on this research project.
Pernicone, Langmuir 18 (2002) 6530.
[36] R.G. Heidenreich, J.G.E. Krauter, J. Pietsch, K. K o¨ hler, J. Mol. Catal. A
82-183 (2002) 499.
1
[
[
37] F. Zhao, K. Murakami, M. Shirai, M. Arai, J. Catal. 194 (2000) 479.
38] A. Biffis, M. Zecca, M. Basato, Eur. J. Inorg. Chem. (2001) 1131.