Hydroesterification Reactions with Palladium-Complexed PAMAM Dendrimers Immobilized on Silica

Article abstract of DOI:10.1021/jo0301853

Highly active, recyclable catalytic systems for the hydroesterification reaction of olefins with methanol and carbon monoxide were prepared by complexing various palladium species to generation zero through four PAMAM dendrimers immobilized on silica. The silica-dendrimer-Pd(PPh₃) ₂ complexes were the most facile recyclable catalysts and could be recycled four to six times by filtration under air. These catalysts show selectivity for the linear reaction product.

Full text of DOI:10.1021/jo0301853

Hyd r oester ifica tion Rea ction s w ith P a lla d iu m -Com p lexed P AMAM
Den d r im er s Im m obilized on Silica
J an P. K. Reynhardt and Howard Alper*
Centre for Catalysis Research and Innovation, Department of Chemistry, University of Ottawa,
1
0 Marie Curie, Ottawa, Ontario, Canada K1N 6N5
halper@uottawa.ca
Received J une 2, 2003
Highly active, recyclable catalytic systems for the hydroesterification reaction of olefins with
methanol and carbon monoxide were prepared by complexing various palladium species to
generation zero through four PAMAM dendrimers immobilized on silica. The silica-dendrimer-
Pd(PPh
3 2
) complexes were the most facile recyclable catalysts and could be recycled four to six
times by filtration under air. These catalysts show selectivity for the linear reaction product.
In tr od u ction
SCHEME 1. Hyd r oester ifica tion Rea ction
From the outset, the aim of this research was to find
an effective recyclable catalyst for the hydroesterification
of olefins. Metal-complexed dendrimers immobilized on
silica have been used previously as catalysts for the
1
2
hydroformylation of olefins, for the Heck reaction, and
3
for the carbonylation of aryl iodides. It was thought that
modification of these systems could lead to effective
recyclable catalysts for the hydroesterification reaction.
The hydroesterification reaction (Scheme 1) is a valu-
able transformation as it is atom efficient, and the ester
products (1, 2, Scheme 1) are attractive for numerous
industrial and pharmaceutical applications. For instance,
branched esters derived from the hydroesterification of
functionalized styrene molecules (2a , Scheme 1) can be
This paper describes the identification, optimization,
and evaluation of effective recyclable palladium-com-
plexed dendrimer-based catalysts immobilized on silica.
Considerable effort was also spent on developing char-
acterization methods for the dendrimers both in the
normal and phosphonated state.
Resu lts a n d Discu ssion
hydrolyzed to the corresponding acids that are potent
nonsteroidal anti-inflammatory agents4 and terminal
As with any catalytic system, the complex loaded into
the reactor usually represents only the precatalyst. The
active catalyst can be significantly different. After normal
deactivation, the active catalyst is transformed into a
resting state, which can be reactivated under the right
conditions to regenerate the active species. With sup-
ported dendrimer catalysts the situation becomes more
complicated. In conventional heterogeneous catalysis the
deactivated catalyst must be reactivated by some exter-
nal process like calcination, reduction, or oxidation. Some
catalysts must be treated with an activating substance
long-chain esters (1b, Scheme 1) are excellent industrial
5
lubricants and surfactants. Palladium-based catalysts
are preferred, but these have not found widespread
application industrially due to the challenges involved
with catalyst and product separation and the high cost
6
associated with palladium-based systems. An effective
recyclable catalyst for this reaction is therefore highly
desirable.
Only two examples of palladium-based recyclable
catalysts for hydroesterification reactions have been
reported previously. One example was based on a poly-
9
before the reaction. In the case of the supported den-
7
drimer catalyst with highly active metals such as rhod-
ium, this problem does not seem to be a significant factor.
But as soon as more complex reactions with less active
metals are attempted, like the hydroesterification reac-
tion with palladium, for instance, the resting state
becomes very important. It is essential that the resting
state of the catalyst be reactivated under the reaction
conditions to ensure high catalytic activity in subsequent
cycles.
mer support, and the other, based on modified mont-
8
morillonite clay, was prepared in this laboratory. Both
these catalysts could not be recycled satisfactorily.
(
1) Bourque, S. C.; Maltais, F.; Xiao, W.; Tardif, O.; Alper, H.; Arya,
P.; Manzer. L. E. J . Am. Chem. Soc. 1999, 121, 3035-3038.
2) Alper, H.; Arya, P.; Bourque, S. C.; J efferson, G. R.; Manzer. L.
E. Can. J . Chem. 2000, 78, 920-924.
3) Antebi, S.; Arya, P.; Manzer. L. E.; Alper, H. J . Org. Chem. 2002,
7, 6623-6631.
4) (a) Alper, H.; Hamel, N. J . Am. Chem. Soc. 1990, 112, 2803-
(
(
6
(
2
804. (b) Alper, H. Aldrichchim. Acta 1991, 24, 3-7.
(
(
(
5) Boyde, S. Green Chem. 2002, 4, 793-307.
(8) Lee, C. W.; Alper, H. J . Org. Chem. 1995, 60, 250-252.
(9) Thomas, J . M.; Thomas, W. J . Principles and practice of
heterogeneous catalysis; VCH: Weinheim, 1997.
6) Kiss, G. Chem. Rev. 2001, 101, 3435-3456.
7) Pittman, C. U., J r.; Ng, Q. Y. U.S. Patent no. 4258206, 1981.
1
0.1021/jo0301853 CCC: $25.00 © 2003 American Chemical Society
Published on Web 09/27/2003
J . Org. Chem. 2003, 68, 8353-8360
8353

Reynhardt and Alper
Palladium catalysts are difficult to characterize due
in part to the fact that palladium is not NMR active.
Previous attempts to elucidate the mechanism of this
reaction have been hindered by this fact. However by
studying the accepted proposed mechanisms it is possible
and therefore, more than likely they resemble highly
branched polymers instead of dendrimers. This in turn
leads to a smaller amount of free amines available for
phosphonation. (3) Higher generations of dendrimers
grafted to the silica surface should be subject to crowding
at the periphery leading to further steric congestion and
possible back-folding, making fewer free amines available
to predict what possible resting states the catalyst would
form.6
,10,11,21
The greater part of this research was devoted
1
4
to the search for a pre-catalyst and reaction conditions,
under which a resting state that could be reactivated in
subsequent cycles is generated.
for phosphonation.
A solid-state 13C NMR method was developed to assess
the extent of growth of the dendrimers. The process is
quite straightforward and is derived from a method
developed by Mayoral18 for his phosphorus-containing
dendrimers. If the dendrimer grows perfectly, the ratio
of the integral of the aliphatic carbons to the integral of
the carbonyl carbon in the PAMAM backbone will be a
function of the generation, with the ratio declining over
the generations. As we cannot determine the absolute
value of the aliphatic carbons associated with the ami-
nopropyl silica gel alone (there is no carbonyl carbon in
the structure of aminopropyl silica gel), we have to correct
the theoretical value for G(1), with the experimental
values of G(1). The method indicates a trend for the
growth of the G(2)-G(4) dendrimers. The correction
factor was determined to be 13.6. This is most likely due
to the methyl capping of the aminopropylsilica gel. Data
for the higher generations was then calculated with this
value as a correction in the G(1) theoretical value.
To clarify the results further, we also calculated a 50%
growth column based on the assumption that only one
branch grows for each branching point (and not two as
in the perfect dendrimer). The results are reported in
Table 2.
P r ep a r a tion a n d Ch a r a cter iza tion of P a lla d iu m -
Com plexed P AMAM-SiO
ine (PAMAM) dendrimers, up to the fourth generation
Figure 1), on commercial aminopropyl silica gel (0.9
2
Den dr im er s. Polyamidoam-
(
mmol (0.1 amine groups/g), were prepared using litera-
ture methods (aminopropyl silica gel obtained from
1
Fluka). The dendrimers were phosphonated using diphen-
ylphosphine and paraformaldehyde by modification of
1
2
literature methods. The double phosphinomethylation
of the terminal amine groups of the dendrimers was
achieved by reacting the dendrimers with diphenylphos-
phinomethanol prepared in situ from diphenylphosphine
and paraformaldehyde in toluene (110 °C, 48 h). The
resulting phosphonated dendrimers were characterized
3
1
13
by solid-state P and C NMR; e.g., a chemical shift of
-
27 ppm in the 31P NMR spectrum compares well with
1
previously reported systems. The phosphonated den-
drimers were readily complexed on treatment with the
appropriate palladium complexes in toluene (rt, 1-2 h
under argon). The silica would turn deep orange to blood
red, and decolorization of the supernatant solvent was
used as an indication of the extent of complexation.
Due to the heterogeneous nature of the catalysts, usual
dendrimer characterization methods such as MS and
GPC could not be employed, but 13C and P NMR were
used to elucidate the structure, and palladium and
phosphorus ICP analysis was done to quantify the
amount of palladium and the extent of phosphonation of
the catalysts. A solid-state NMR internal standard
method (methyltriphenylphosphonium bromide as the
internal standard) for the rapid quantification of the
extent of phosphonation was also developed. The phos-
phorus NMR method was used as a quick estimation to
determine the amount of palladium required during the
complexation step, and the ICP analyses (carried out
externally) was done to improve the level of accuracy
As the growth of the dendrimers was not complete,
another characterization method was required to verify
the ICP and NMR results.
To quantitatively determine the degree of growth of
the dendrimer, a function D, defined as follows, is
proposed:
31
D ) M(d)/M (d)
t
where D is the degree of growth of the dendrimer, M(d)
is the mass of the dendrimer per gram, and M (d) is the
t
theoretical mass of the dendrimer per gram.
To assess D, a thermal gravimetric analyses (TGA)
study was undertaken. The results are summarized in
Table 3.
(Table 1).
It is clear from Table 1 that the extent of phosphona-
As can be seen in Table 3, the degree of growth of the
dendrimers is about 0.3-0.6 for the different generations.
Therefore, about 30% growth is detected for the higher
generations of the dendrimers. This is in accordance with
tion differed from the theoretical expectations, and
several reasons for this are plausible: (1) Some of the
amino groups on the aminopropyl silica gel used for the
synthesis of the dendrimers might reside in the pores of
the support and would therefore lead to ineffective
growth of the dendrimers due to steric hindrance and
would then result in ineffective phosphonation. (2) It is
statistically probable that the growth of the dendrimers
(14) (a) Nayler, A. M.; Goddard, W. A., III; Kiefer, G. E.; Tomalia,
D. A.; J . Am. Chem. Soc. 1989, 111, 2339-2342. (b) J ansen, J . F. G.
A.; de Brabander-van den Berg, E. M. M.; Meijer, E. W. Science 1994,
2
66, 1226-1229.
(15) Amatore, C.; J utand, A. Coord. Chem. Rev. 1998, 178-180,
511-528.
(
whether inside the pores or outside) is not complete,13
(16) Newkome, G. R.; Moorefield, C. N.; Vogtle, F. Dendrimers and
Dendrons; Wiley-VCH: Weinheim, 2001.
(
10) Kawana, M.; Nakamura, S.; Watanabe, E.; Urata, H. J .
Organomet. Chem. 1997, 542, 185-189.
11) Scivanti, A.; Beghetto, V.; Campagna, E.; Zanato, M.; Mateoli,
U.; Organometallics 1998, 17, 630-635.
12) Reetz, M. T.; Lohmer, G.; Schwikkardi, R. Angew. Chem., Int.
Ed. Engl. 1997, 36, 1526-1529.
13) Tsubokawa, N.; Ichioka, H.; Satoh, T.; Hayashi, S.; Fujiki, K.;
React. Funct. Polym. 1998, 37, 75-82.
(17) Reman, W. G.; De Boer, G. B. J .; Van Langen, S. A. J .;
Nahuijsen, A.; Eur. Patent no. EU 0411721A2, 1990.
(18) (a) Slaney, M.; Bardaji, M.; Casanove, M.; Caminade, A.;
Majoral J .; Chaudret, B. J . Am. Chem. Soc. 1995, 117, 9764-9765. (b)
Bardaji, M. Kustos, M.; Caminade, A.; Majoral, J .; Chaudret, B.
Organometallics 1997, 16, 403-410. (c) Slaney, M.; Bardaji, M.;
Caminade, A.; Majoral J .; Chaudret, B. Inorg. Chem. 1997, 36, 1939-
1945.
(
(
(
8
354 J . Org. Chem., Vol. 68, No. 22, 2003

Hydroesterification Reactions with Pd-Complexed PAMAM Dendrimers
F IGURE 1. Proposed structure of the phosphonated generation zero through four (G(0)-G(4)) PAMMAM-silica dendrimer ligands
(C6-linker).
the NMR and ICP analyses. It is well documented that
the divergent growth approach to the synthesis of den-
drimers can lead to flawed dendrimer species. Statisti-
cally, only a small fraction of the higher generations will
J . Org. Chem, Vol. 68, No. 22, 2003 8355

Reynhardt and Alper
TABLE 1. ICP a n d Qu a n ta ta tive 31P NMR An a lysis of th e Differ en t Gen er a tion s of th e P d (P P h 3)2-P AMAM-Silica
Ca ta lysts
PICP
(mass %)
PNMR
(mass %)
Ptheory
(mass %)
PdICP
(mass %)
Pdused
(mass %)
entry
generationb
PICP/Ptheory
1
2
3
4
5
6
7
8
9
G(0)
G(0)c
G(1)
G(1)c
G(2)
G(2)c
G(3)
G(3)c
G(4)
G(4)c
G(0)
G(4)
1.11
1.57
1.38
1.96
1.90
1.19-1.44
0.69-0.99
1.13-1.25
0.56-0.87
0.75
4.1
0.27
0.96
0.74
0.96
0.99
0.88
1.29
0.79
1.32
1.31
1.29
5.5
0.28
0.33
0.44
0.41
4.15
4.43
4.59
1
1
1
0
1
2
a
a
1.74
1.37
1.70-2.00
0.95-1.43
4.10
4.59
0.42
0.30
a
Previously prepared dendrimer samples. b Subscript c refers to Pd-complexed dendrimers.
TABLE 2. Qu a n ta ta tive 13C NMR Stu d y of th e P AMAM-Silica Den d r im er s
integral
(aliphatic)
integral
(carbonyl)
theoretical
ratio
corrected
theoretical
50%
growth
differencea
13.6
b
generation
ratio
1
2
3
4
23.1
14.9
14.2
10.9
1
1
1
1
23.1
14.9
14.2
10.9
9.5
8.5
8.2
8.1
23.1
10.8
9.2
23.1
13.6
11.3
9.8
8.6
a
Excess carbon due to methyl capping of the silanol groups on the surface of the support. b Determined on the basis of the assumption
that only one branch, instead of two, grows at each branching point (thus 50% growth).
TABLE 3. Degr ee of Gr ow th As Deter m in ed by TGA An a lyses
organic mass
(TGA) (%)
mass theoretical
difference
corrected TGA
mass M(d) (%)
degree of growth
entry
generation
Mt(d) (%)
(correction factor)a
M(d)/Mtd
1
2
3
4
5
0
1
2
3
4
7.03
18.36
19.24
23.84
26.05
5.22
27.40
50.58
69.80
83.04
1.81
1.81
1.81
1.81
1.81
5.22
16.55
17.43
22.03
24.24
N/A
0.60
0.35
0.32
0.30
a
Excess carbon due to methyl capping of the silanol groups on the surface of the support; adsorbed water was excluded as a cause by
only using the part of the TGA associated with the organic mass.
TABLE 4. Su r fa ce Ar ea a n d P or e Volu m e Da ta for th e
Differ en t Gen er a tion s)
becomes pronounced in the higher generations (especially
as we are using the hexamethylenediamine linker).
Although the characterization results would suggest
a highly branched polymeric species as opposed to a
perfect dendrimer on the surface, the different genera-
tions could still influence the catalytic activity and
selectivity, with the higher generations having a more
pronounced influence. An evaluation of the catalytic
activity was undertaken.
Eva lu a tion of Ca ta lyst P r ecu r sor s. The generation
zero complexes were first synthesized and tested to find
appropriate precatalysts and to optimize the reaction
conditions, both for catalytic activity and recyclability
(Figure 2.)
surface
total pore
average pore
2
3
entry generation area (m /g) volume (cm /g) diameter (nm)
1
2
3
4
5
0a
1
2
3
4
305.59
248.06
225.16
190.32
147.92
0.70
0.51
0.46
0.39
0.33
7.68
7.04
7.05
6.90
6.98
a
Determined for commercial aminopropyl silica gel.
be perfect.16 In this instance, a further complication
results due to the fact that we are working on a solid
support, and therefore separation of imperfect dendrim-
ers is not feasible. Furthermore, some dendrimers might
reside inside the pores and, therefore, not grow perfectly,
also leading to lower D values. To test this hypothesis,
BET surface areas, pore volumes, and pore diameters of
the different generations were also determined. The
results are presented in Table 4. It is interesting to note
that the surface area of the support decreases with
increasing generation of the dendrimer and that the pore
volume and pore diameter also decrease with increasing
generation. This would support our first assumption that
some of the dendrimer growth takes place within the
pores of the support. Therefore, the steric interference
2
G(0)PdCl (3, Figure 2) was synthesized and tested
with an acidic solution. The complex showed no activity
without added phosphine. Although this is unusual, the
montmorillonite catalyst prepared earlier also showed
7
this behavior. It could be that a “release and capture”
mechanism is at work. This observation was consistent
with all the other catalysts tested. An initial crude
optimization showed strong temperature, pressure, and
acid concentration dependence, but a full optimization
was postponed until recyclability could be afforded. The
G(0)PdCl
2 2
(3, Figure 2) and G(4)PdCl systems were
evaluated for recyclability, but the catalysts showed little
8
356 J . Org. Chem., Vol. 68, No. 22, 2003

Hydroesterification Reactions with Pd-Complexed PAMAM Dendrimers
TABLE 6. Effect of p-Tolu en esu lfon ic Acid /P d Ra tio on
th e Con ver sion of 1-Decen ea
entry
acid/Pd
conversion (%)
1
2
3
4
5
6
7
3
4
0
0
10
14
26
37
51
17
23
39
71
36
a
Conditions: 30 mg of catalyst (4 µmol of Pd), PPh3 (25 equiv),
.150 mL of 1-decene (0.8 mmol), THF/MeOH (1:1 mixture, 5 mL),
0
PCO ) 150 psi, 22 h.
F IGURE 2. Generation zero palladium catalysts.
highest conversions but that hydrochloric acid gives the
best selectivity. Also, strongly coordinating anions such
as iodide (from HI) form very stable Pd complexes and
all catalytic activity is lost immediately (Entry 4, Table
TABLE 5. Effect of Va r iou s Acid s on th e
a
Hyd r oester ifica tion of 1-Decen e
entry
acid
conversion (%)
L/B
1
2
3
4
5
6
p-TsOH
HCO2H
F3CCO2H
HI
HCl
MeSO3H
77
0
0
0
73
82
1.36
5
). For the G(0)Pd(OAc)
acid and trifluoroacetic acid do not activate the catalyst,
although these acids will activate the PdCl systems but
MeSO H/p-TsOH
2
catalyst (4, Figure 2), formic
2
3.6
1.3
give lower activity than the Pd(OAc)
systems.
2
2
a
Conditions: 30 mg of catalyst (4 µmol of Pd), PPh3 (25 equiv),
The molar ratio of the acid to palladium is of great
importance, and an effort was made to optimize this ratio.
The acid optimization shows that no catalyst activity can
be detected at ratios of acid/palladium of less than 5.
Above ratios of 5 there exists a linear relationship
between acid concentration and conversion up to about
a ratio of 35, after which the acid concentration is high
enough to lead to detrimental side-reactions such as
isomerization. A ratio of acid to palladium of 35 was
selected for subsequent studies (Table 6)
acid (35 equiv), 0.150 mL of 1-decene (0.8 mmol), THF/MeOH (1:1
mixture, 5 mL), PCO ) 150 psi, 22 h.
or no activity in subsequent cycles with some leaching
observed for the PdCl
G(0)Pd(OAc) analogue (4, Figure 2) gave comparable
activity to the G(0)PdCl catalysts (3, Figure 2), and
2
systems. Initial trials with the
2
2
these could be recycled under very specific conditions. The
G(0)Pd(dba) (5, Figure 2)16 analogue also was very active,
Op tim iza tion of th e Am ou n t of Ad d ed P h osp h in e.
The amount of added phosphine was also optimized by
varying the ratio of P to Pd between 2 and 100. The
catalyst shows no activity up to a P/Pd ratio of 10, and
then the optimum is reached at a ratio of 25. At ratios of
more than 25 the activity decreases, suggesting that
stable Pd complexes are formed, inhibiting catalytic
activity. It is common for the activity of hydroesterifica-
tion catalysts to be highly sensitive to acid and phosphine
concentrations. In a related patent, Reman et al. reported
and this showed that a Pd(II) species as pre-catalyst is
not essential for the reaction. As the G(0)Pd(OAc)
2
complex (4, Figure 2) showed the most promise in initial
trials, this catalyst was chosen as the test catalyst of
choice for the optimization reactions. To understand the
reaction parameters better a full optimization of the
reaction conditions were undertaken
Op tim iza tion of th e Rea ction Con d ition s. In the
preliminary runs, several parameters were identified that
affect the activity of the catalyst, and these were opti-
mized separately. Thirty milligrams of the catalyst (4,
Figure 2, 4.0 µmol of Pd), was used at 115 °C, with 160
equiv of 1-decene in the optimization reactions.
P r essu r e Op tim iza tion . Optimization reactions were
conducted at pressures ranging from 100 to 1000 psi of
CO. The 1-decene conversion increases linearly with the
pressure up to a pressure of 150 psi, after which the
change in conversion is less sensitive to pressure with a
maximum conversion reached at 250 psi. At pressures
higher than 250 psi the 1-decene conversion decreases
linearly with increasing pressure. It is important to note
that although the highest conversion was obtained at 250
psi, the catalyst gives more than 90% of the optimum
conversion between 150 and 480 psi, thus enabling a
window of operation of more than 300 psi over the
preferred pressure (150 psi). The optimum pressure was
accordingly chosen as 150 psi.
2 3 2
that for the homogeneous Pd(OAc) (PPh ) -acid system,
the catalyst requires high acid and phosphine ratios for
optimum activity.17
3
Up to this point, only PPh was tested as an added
phosphine ligand, and so other phosphine ligands were
tested, to assess the effect of the added phosphine on the
catalyst. Of the ligands tested, only PPh
activity.
3
gave any
Solven t Op tim iza tion . Solvents with different polari-
ties were employed, and the catalyst performed best in
solvents with low polarity although intermediate polarity
solvents also gave acceptable results. Almost no activity
is observed with very polar solvents such as NMP (entry
6, Table 7). It became apparent that lower polarity
solvents are superior in recycle reactions (Tables 9-11).
Tem p er a tu r e Op tim iza tion . The catalyst activity
was evaluated between 60 and 150 °C, and although
activity was observed over the whole temperature range,
the optimum conversion was observed between 100 and
120 °C with 115 °C chosen as the temperature of choice
in subsequent recycle tests.
Op tim iza tion of Acid Ad d itive. Various acid addi-
tives were tested in an effort to improve the selectivity
and conversion. It is clear from the results in Table 5
that methanesulfonic and p-toluenesulfonic acid give the
J . Org. Chem, Vol. 68, No. 22, 2003 8357

Reynhardt and Alper
TABLE 7. Solven t Op tim iza tion for th e
prepared grafted homogeneous catalysts in MCM-41 and
Carbosil to study the effect of the pore walls on the
selectivity of the catalyst. They found that the catalyst
supported inside the pores of MCM-41 showed completely
different selectivity than both the carbosil-supported
catalyst and the homogeneous catalyst.20 We are cur-
rently attempting to secure evidence in support of this
hypothesis. The catalyst was also tested for activity
without added phosphine or acid. In both cases the
catalyst showed no activity, but the catalyst could be
a
Hyd r oester ifica tion of 1-Decen e
entry
solvent
hexane
toluene
dimethyl ether
THF
dielectric constant
conversion (%)
1
2
3
4
5
6
1.9
2.379
4.335
7.6
9.08
32
90
89
86
78
81
0
CH2Cl2
NMP
a
Conditions: 30 mg of catalyst (4 µmol of Pd), ligand (25 equiv),
p-TsOH (35 equiv), 0.150 mL of 1-decene (0.8 mmol), solvent/
MeOH (1:1 mixture, 5 mL), PCO ) 150 psi, 22 h.
3
activated by adding acid (p-TsOH) and phosphine (PPh )
to the reactor after the preliminary tests were done.
The ability of the catalyst to be recycled was very
promising, and the next step was to synthesize the
generation one through four analogues of this complex
Recycle Stu d ies. From the first attempts at recycling
with the PdCl
that these could not be recycled without pretreatment
with carbon monoxide and PPh3. Therefore, the Pd(OAc)
functionalized dendrimers were tested. As with the
PdCl -functionalized catalysts, the optimum recyclability
2
functionalized dendrimers it was apparent
(Figure 1). The G(1)-G(4) analogues of complex 6 were
2
tested for activity without added phosphine, but like the
G(0) catalyst no activity was detected. The catalysts were
then used for recycle tests under the same conditions as
the G(0) catalyst but leaching was observed, and the
catalysts showed little or no activity in subsequent runs
2
was achieved after the catalyst was pretreated. After
pretreatment, the catalyst could be used for six consecu-
tive runs. Because the palladium is effectively reduced
during the pretreatment step, it was thought that start-
ing off with a palladium(0) precursor could eliminate the
necessity of a pretreatment step, and a Pd(dba)-function-
alized catalyst (5, Figure 2) was prepared and tested.
Although this catalyst was highly active, it could not be
satisfactorily recycled.
(Table 10). However, by changing the solvent system to
a mixture of hexane and methanol, activity could be
afforded up to at least the third cycle and in most of the
cases up to five cycles (Table 11). Lower initial activity
was observed here, possibly due to the low solubility of
methanol in hexane. The G(1)-G(4) catalysts were also
evaluated with the addition of a small amount of toluene
to afford mixing of the two main solvents. This improved
the initial activity without changing the recycling sig-
nificantly (entries 3, 4, 5, 8, Table 11).
We purposefully exposed a large quantity of the
catalyst to the reducing environment of the reaction,
without the addition of substrate. In this manner, we
were able to generate a sample (200 mg) for solid-state
An interesting observation is that the higher genera-
tions, although having higher activity initially, do not
recycle as well as the G(0) catalysts (Table 11).
3
1
NMR analyses. A peak at 3.6-5 ppm in the P NMR
was attributed to PPh bound to a palladium(0) species
attached to the dendrimer.
P d (P P h a s Com p lexa tion Agen t. It is clear from
previous results that PPh stabilizes the catalyst and is
3
3 2
Various generations of the Pd(PPh ) -complexed den-
3 4
)
drimers were selected for representative tests to deter-
mine the scope of the reaction, and the ratio of substrate
to palladium was increased to at least 1000 to gauge the
activity of the catalysts. The results are shown in Table
3
crucial for catalytic activity. It is therefore likely that the
resting state of the catalyst would be a complex where
two coordination sites are occupied by the phosphines of
1
2.
the bidentate moiety on the dendrimer and two by PPh
The G(0)Pd(PPh catalyst (6, Figure 2) was then
synthesized and characterized by P and C NMR (entry
3
.
Internal olefins were not catalytically active (entries
3 2
)
3
1
13
10 and 13, Table 12). To determine if the internal olefins
deactivate the catalyst, or if the catalyst is just inactive
toward internal olefins, premixed 1-decene and 2-decene
were exposed to the catalyst. The catalyst was able to
react with the 1-decene selectively without any discern-
ible reaction with 2-decene. Also, styrene containing an
electron-withdrawing group shows completely different
activity, favoring oligomerization and polymerization at
the reaction temperature (entries 3, 5, and 6, Table 12).
The reaction with p-chlorostyrene was repeated at 85 °C
in an effort to eliminate polymerization, but the conver-
sion was substantially lower (entry 3b, Table 12). The
most surprising observation, however, was that even for
styrene, the linear product is favored. This is a unique
observation as the branched product is usually formed
in a highly regioselective manner. For o-methylstyrene
4
, Table 8).
The first test showed comparable activity to the
Pd(OAc) -complexed catalyst, and for the first time the
2
complex could be recycled effectively without pretreat-
ment. The catalyst could be recycled for three runs at
35 °C in THF (low yields observed during second and
third cycles). By changing the solvent to toluene and
effecting the reactions at 115 °C, the catalyst showed good
activity for up to 5 cycles and 11% conversion in the sixth
1
(
Table 9). The ratio of linear to branched products
improved during the recycle runs. This could be due to a
different reactivity of the complexes grafted on the inside
surface as opposed to the outside surface. One would
expect from the crowding inside the pores that the
catalysts residing there would favor the linear products.
It is also conceivable that the outside complexes would
be more susceptible to leaching when compared to the
complexes residing inside the pores. The catalyst there-
fore evolves by the leaching of outside grafted palladium
molecules to a more selective system where the catalyst
sites are protected inside the pores. J ohnson et al.
(entry 2, Table 12), remarkable selectivity was observed
for the linear product with the ratio of linear to branched
(
(
19) Tolman, C. A. Chem. Rev. 1977, 77, 315-348.
20) J ohnson, B. F. G.; Raynor, S. A.; Shephard, D. S.; Mashmeyer,
T.; Thomas, J . M.; Sankar, G.; Bromley, S.; Oldroyd, R.; Gladden, L.;
Mantle, M. D. Chem. Commun. 1999, 1167-1168.
8
358 J . Org. Chem., Vol. 68, No. 22, 2003

Hydroesterification Reactions with Pd-Complexed PAMAM Dendrimers
TABLE 8. 13C a n d 31P NMR Ch a r a cter iza tion of th e Ca ta lysts
entry
complex
catalyst
13C (ppm)
31P (ppm)
28, 13.0
28.0, 15.8,-27.6
28.0
27.7, 7.5, -26.1
26.23, 10.29, -26.22
25.9, 8.7, -27.4
27.7, 13.62, -25.3
28.51, 11.32, -28.97
ratio P/Pd
1
2
3
4
5
6
7
8
3
4
5
6
G(0) PdCl2
2
2
2
3
3
3
3
3
G(0) Pd(OAc)2
G(0) Pd(dba)
G(0) Pd(PPh3)2
G(1) Pd(PPh3)2
G(2) Pd(PPh3)2
G(3) Pd(PPh3)2
G(4) Pd(PPh3)2
131.1
130
130.5
129.6
128.33
127.8
TABLE 9. Recycle Test of G(0)P d (P P h 3)2-Ca ta lyzed
conditions. The catalysts do, however, show high activity
TON of up to 1200) especially as 1-2 mol % of palladium
a
Hyd r oester ifica tion of 1-Decen e
(
entry
run no.
conversion (%)
L/B
(TON of 50-100) is usually employed in homogeneous
reactions of higher olefins.
1
2
3
4
5
6
1
2
3
4
5
6
78
52
62
71
36
11
1.35
2.60
2.36
3.20
4.50
P r op osed Mech a n ism . To understand the different
parameters involved, it is important to look at a proposed
mechanism for this reaction. This mechanism has been
deduced from literature data and observations. It il-
lustrates the different parameters that were optimized
to afford maximum activity.
a
Conditions: 50 mg of catalyst (4.5 µmol of Pd), PPh3 (25 equiv),
p-TsOH (35 equiv), 1-decene (160 equiv), toluene/MeOH (1:1
mixture, 5 mL), PCO ) 150 psi, T ) 115 °C, 22 h.
From literature considerations, there exist two possible
initial steps, paths A and B. Most likely, a combination
of the two initiates the cycle, but neither has been
TABLE 10. Recycle Test of G(1)P d (P P h 3)2 a n d
a
6,8,10,11,21
G(2)P d (P P h 3)2 in Tolu en e/MeOH
confirmed experimentally.
For clarity, the ligands
L refers to either phosphine or CO ligands, but at least
one phosphine is assumed to be present during the entire
catalytic cycle (Figure 3).
entry
generation
run no.
conversion (%)
L/B
1
2
3
4
1
1
2
2
1
2
1
2
76
30
86
trace
3.41
2.23
After the initial reaction, following path A, the cat-
ionic-hydride complex I is formed. This is followed by
an attack by the olefin, most likely forming a five-coor-
dinate square-pyramidal intermediate that rearranges
to complex II. This attack is believed to proceed by an
associative mechanism, as the concentration of the olefin
affects the rate of this step. Complex III is generated from
II in much the same fashion, but by attack of CO. The
CO inserts into the palladium-carbon bond, and after
attack by the alcohol (in this case methanol), reductive
elimination occurs to generate the Pd(0) species (IV). The
cycle is completed by oxidative addition (simple proto-
nation) of the acid to complex IV regenerating species I.
It is clear that the overall reaction rate will depend on
the olefin concentration, the acid concentration, and the
methanol concentration. The rate will also depend on the
electrophilic nature of the metal center, and the nucleo-
philic nature of the alcohol. Based on the assumption that
a cationic palladium species is the key intermediate, it
follows that the solvent polarity would also have a
significant effect on the overall rate. The CO pressure
will also affect the rate of the reaction. The CO pressure
was optimized, and as predicted by the proposed mech-
anism, the reaction rate increases with pressure until a
threshold pressure is reached, at which point the CO
starts to block the vacant sites, and consequently reduces
the reaction rate. The same relationship, although more
pronounced, is observed for the acid concentration. At
higher acid ratios, side reactions such as isomerization
and ether formation are favored. The phosphine ligand
concentration, although not directly evident from the
mechanism, is also pivotal to the effectiveness of the
catalyst. This is due to the electronic interaction of the
a
Conditions: 50 mg of catalyst (4.5 µmol of Pd), PPh3 (25 equiv),
p-TsOH (35 equiv), 1-decene (160 equiv), toluene/MeOH (1:1
mixture, 5 mL), PCO ) 150 psi, T ) 115 °C, 22 h.
TABLE 11. Recycle Test of G(1)-G(4) P d (P P h 3)
Ca ta lysts in Hexa n e/MeOH
a
conversion (%)
entry
generation
run 1
run 2
run 3
run 4
run 5
44
1
2
1
2
1
2
3
3
4
4
87
87
>99
>99
>99
83
73
78
53
83
86
84
72
75
61
55
33
57
29
36
53
73
55
26
b
b
b
3
4
5
6
7
8
35
89
92
37
45
b
a
Conditions: 50 mg of catalyst (4.5 µmol of Pd), POdPh3 (25
equiv), p-TsOH (35 equiv), 1-decene (160 equiv), hexane/MeOH
1:1 mixture, 5 mL), PCO ) 150 psi, T ) 115 °C, 22 h. b Hexane/
(
MeOH (1:1 mixture, 5 mL), 1 drop of toluene.
isomers being 12.5:1. This large difference between the
selectivity of styrene and o-methylstyrene is mainly due
to the presence of a methyl group in the 2-position on
the ring. This shows the sensitivity of the catalyst system
to steric factors. p-tert-Butylstyrene (entry 4, Table 12)
also afforded full conversion, with 56% selectivity for the
ester products (L/B ratio of 2). The other 44% was
converted to ether-type products formed by the direct
addition of methanol to the double bond. (A side reaction
that was also observed due to the acid.) 1-Octene and
1
-hexene (entries 8 and 9, Table 12) showed lower activity
than 1-decene. Carbonyl-containing substrates such as
vinyl benzoate and vinyl acetate (entries 11 and 12, Table
(21) Cavinato, G.; Vavasori, A.; Tonolio, L.; Benetollo, F. Inorg.
1
2) proved to be completely unreactive under the reaction
Chim. Acta 2003, 343, 183-188.
J . Org. Chem, Vol. 68, No. 22, 2003 8359

Reynhardt and Alper
TABLE 12. Oth er Su bstr a tes Tested w ith G(0)-G(4)P d (P P h 3)2a
entry
generation
substrate
conversion (%)
TON
TOF (h-1)
L/B
1
2
1
2
4
2
1
2
2
4
1
1
4
1
2
2
styrene
o-methylstyrene
p-chlorostyrene
>99
1100
1200
829
232
1478
50
53
38
10
67
2.4
12.5
2.2
-
>99
3
3
4
5
6
7
8
9
0
1
2
3
a
b
>99 (73)
20
b
t
p- Bu-styrene
>99 (56% esters)
polymerized
oligomerized
>99
75%
63%
no reaction
no reaction
no reaction
no reaction
2
p-vinylbenzoic acid
p-methoxystyrene
1-decene
1-octene
1-hexene
1243
1108
931
56
50
42
1.9
3.2
2.47
c
1
1
1
1
2-decene
vinyl acetate
vinyl benzoate
cyclohexane
a
Conditions: 20 mg of catalyst (1.8 µ mol of Pd), PPh3 (25 equiv), pTsOH (35 equiv), olefin (2 mmol, 1000 equiv), toluene/MeOH (1:1
mixture, 5 mL), PCO ) 150 psi, T ) 115 °C, 22 h. 85 °C. c 300 psi.
b
supported dendrimers. A Fourier transform infrared spec-
trometer was used to verify the IR spectra of the products.
GC-MS was used to determine the molecular masses of the
products. Phosphorus and palladium ICP along with TGA and
BET surface characterization were determined by commercial
laboratories.
Gen er a l Exp er im en ta l P r oced u r e for th e Hyd r oes-
ter ifica tion of Olefin s w ith Meth a n ol. To a 50 mL stainless
steel autoclave equipped with a pressure gauge, a glass liner,
and a stirring bar were added 20 mg of a palladium containing
catalyst (4.0 µmol Pd), a toluene/methanol mixture (1:1 v/v, 5
3
mL), 25 equiv of PPh , 35 equiv of acid, and 1000 equiv of
olefin. The autoclave was flushed three times with carbon
monoxide, pressurized to 150 psi, and placed in a preheated
oil bath at 115 °C. The reaction was left for 22 h under vigorous
stirring, and after cooling, ether (2 mL) was added and the
contents were filtered through 0.45 µm membrane filter and
washed with ether. The conversion and the branched-to-linear
ratio was determined by GC (filtration for recycle studies was
done both under inert atmosphere and under air, with no
discernible difference).
F IGURE 3. Proposed mechanism for the hydroesterification
reaction.
Con clu sion s
We have developed a new recyclable catalyst for the
hydroesterification of olefins, based on palladium-com-
plexed dendrimers immobilized on silica. The catalysts
show high activity for styrene derivatives and linear long-
chain olefins (TON of up to 1200) and favor the linear
product of the reaction even for styrene. The catalysts
can be recycled up to six times by simple filtration in air.
We have also found through the characterization of the
dendrimers that some of the growth takes place within
the pores of the silica support, and therefore, a function
D (for degree of growth) was defined and determined to
be around 0.6 (60%) for the lower generations and around
phosphine and the metal center. At low concentrations,
the phosphine does not effectively stabilize the interme-
diates such as the palladium(0) species. Higher concen-
trations of the phosphine block the activity by stabilizing
the intermediates too much and thereby slowing down
the cycle. It appears that the other phosphorus ligands
do not possess the right electronic characteristics to
stabilize both the cationic palladium species and the
other intermediates.
Exp er im en ta l Section
0
.3 (30%) for the higher generations.
Ma ter ia ls. All olefins were used as received or percolated
through a short column of activated alumina. All solvents were
dried, distilled, and in some cases degassed before use. All
other chemicals were used as received without further puri-
fication.
Ack n ow led gm en t. We thank Dr. Glenn Facey for
the recording the solid-state NMR spectra, Dr. Mike
Green for valuable discussions, and SASOL Technology
(R&D) for financial support. We also thank Dr. Hum-
phrey Dlamini and his group at SASOL Technology
R&D) for the TGA and surface analytical determi-
nations.
In str u m en ts. Gas chromatography was used for the quan-
tification of conversions (a direct retention factor quantification
method was used). A 300 MHz NMR instrument was used for
(
1
13
the H, and C, analyses of reagents and products. A 200 MHz
solid-state NMR instrument was used to characterize the
J O0301853
8
360 J . Org. Chem., Vol. 68, No. 22, 2003