Immobilization of ruthenium catalysts for allylations with allyl alcohol

Article abstract of DOI:10.1016/j.molcata.2010.06.027

[RuCp(PP)]⁺ complexes active for allylation of alcohols with allyl alcohol as the allylating agent were immobilized on solid supports. Two different immobilization methods have been applied: (1) via electrostatic interactions of the cationic complex on ion-exchange resins, where the anion is present on the support and (2) via a coordination bond with a ligand covalently-bound on the support. Both methods give high yields of immobilized complex through relatively simple procedures. The catalysts immobilized via ionic interactions prove to be able to allylate both 1-octanol and 4-tert-butylphenol with very low leaching of the catalyst, thus forming allyl octyl ether and C-allylated phenol, respectively. The accumulation of water in the highly hydrophilic resin precludes the O-allylation of phenol and also retards the C-allylation reaction. The catalysts immobilized via a coordination bond are not hydrophilic; with these catalysts selective O-allylation of phenols is achieved, with recycling of the catalysts over multiple runs. Leaching of the catalyst from the support is somewhat higher than for the electrostatically-bound catalyst and quarternisation (allylation) of the excess of phosphine groups present on the support plays an important role in the activity of the immobilized catalysts for the allylation reaction.

Full text of DOI:10.1016/j.molcata.2010.06.027

Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Immobilization of ruthenium catalysts for allylations with allyl alcohol
Jimmy A. van Rijn, Elisabeth Bouwman∗, Eite Drent
Leiden Institute of Chemistry, Gorlaeus Laboratories, Leiden University, P.O. Box 9502, 2300 Leiden, The Netherlands
a r t i c l e i n f o
a b s t r a c t
[RuCp(PP)]+ complexes active for allylation of alcohols with allyl alcohol as the allylating agent were
immobilized on solid supports. Two different immobilization methods have been applied: (1) via elec-
trostatic interactions of the cationic complex on ion-exchange resins, where the anion is present on the
support and (2) via a coordination bond with a ligand covalently-bound on the support. Both methods
give high yields of immobilized complex through relatively simple procedures. The catalysts immobi-
lized via ionic interactions prove to be able to allylate both 1-octanol and 4-tert-butylphenol with very
low leaching of the catalyst, thus forming allyl octyl ether and C-allylated phenol, respectively. The accu-
mulation of water in the highly hydrophilic resin precludes the O-allylation of phenol and also retards
the C-allylation reaction. The catalysts immobilized via a coordination bond are not hydrophilic; with
these catalysts selective O-allylation of phenols is achieved, with recycling of the catalysts over multiple
runs. Leaching of the catalyst from the support is somewhat higher than for the electrostatically-bound
catalyst and quarternisation (allylation) of the excess of phosphine groups present on the support plays
an important role in the activity of the immobilized catalysts for the allylation reaction.
Article history:
Received 12 May 2010
Received in revised form 25 June 2010
Accepted 29 June 2010
Available online 31 July 2010
Keywords:
Immobilization
Allylation
Allyl alcohol
Ruthenium
Phenol
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
Immobilization of homogeneous catalysts on solid supports
is a well-studied field of research. Several methods can be used
Nowadays allylations on industrial scale still depend on stoi-
chiometric salt chemistry and employ allyl halides as allyl donors.
Allyl alcohol as the allylating agent is an environmentally benign
alternative for the production of allyl ether, since only water is co-
produced as a byproduct when no other stoichiometric additives
are needed for the reaction to proceed.
Several homogeneous catalysts are known to catalyze the
reaction between an aliphatic or aromatic alcohol and allyl alco-
hol, all being precious metal-ion complexes [1–6]. The successful
ruthenium-based systems previously described by us only need
catalyst amounts of 0.1 mol% on substrate or less and do not rely
on the use of stoichiometric amounts of additives to control activ-
ity and selectivity; however, the catalytic system is present in the
same phase as the substrates and products, from which it is dif-
ficult to recover after the reaction [1–3]. It would be desirable to
find a means of recycling of the catalyst, which may be achieved by
immobilization of the active complex onto a non-soluble support.
A simple filtration after the reaction could then be performed or
the catalyst could thus be used in a continuous process by flushing
the substrates through a fixed-bed.
to attach a metal complex onto an insoluble support. One often
applied method is to covalently link one of the ligands of the reac-
tive complex onto the support. Typical supports are polystyrenes
[7] and inorganic materials [8]. A complex with a free coordination
site can be added to a resin with coordinating groups (Scheme 1a)
[9], but a more common approach is the substitution of a ligand
from the complex with a ligand present on the support (Scheme 1b)
[10,11]. The advantage of covalently-bound ligands is that the link-
age between ligand and support is very stable, but a disadvantage
is that a change of synthesis is required of at least one of the cat-
alyst ligands and thus the complex, which can be cumbersome
and expensive. It may also affect the catalyst’s structure, which is
undesirable as it may change the reactivity of the catalyst. These
disadvantages can be overcome by immobilizing the catalyst using
non-covalent interactions for which several methods are known,
ranging from strongly ionic to weak Van der Waals interactions to
confine the catalyst to a support [12]. Amongst the non-covalent
interactions, the ionic interaction is the most stable [13–15]. By
adding a cationic catalyst to an anionic ion-exchange resin, an
electrostatically-immobilized catalyst is obtained (Scheme 1c).
Advantages of this method are that the immobilization process is
often fairly easy and the active catalyst does not need to be mod-
ified. Furthermore, reloading of the support with fresh catalyst is
readily achieved using this method. A disadvantage of this method
is that it can only be applied for ionic catalysts, which have to keep
∗
Corresponding author. Tel.: +31 71 527 4550; fax: +31 71 527 4761.
E-mail address: bouwman@chem.leidenuniv.nl (E. Bouwman).
1381-1169/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcata.2010.06.027

J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
27
Scheme 1. Immobilization strategies: (a) coordination of an unsaturated complex onto a support with a coordinating residue, (b) coordination of a saturated complex onto
a support via ligand substitution with a coordinating residue and (c) absorption of a cationic complex onto an anionic ion-exchange support.
their charge throughout the whole catalytic cycle in order to remain
immobilized.
over Celite and to the filtrate was added the resin-bound triph-
enylphosphine (135 mg; 0.4 mmol phosphine residues) and stirred
(100 rpm) at room temperature for 48 h. The resin was collected
and washed with dichloromethane (3 ml× 4 ml). The efficiency of
the loading was measured in duplicate by means of measuring the
Ru-content in the combined filtrates with ICP-AES to find that 91%
of the initial Ru-complex was present on the support.
In this paper, the immobilization of a number of the complexes
described previously [1–3] is demonstrated, and their application
as catalysts in allylation reactions is explored. The ionic as well
as the covalent immobilization approach was used and the result-
ing heterogenized catalysts were tested in allylation reactions of
aliphatic alcohols and phenols with allyl alcohol as the allyl donor.
2.4. Immobilization of [RuCp(PPh3)2](OTs) onto resin-bound
triphenylphosphine (catalyst B)
2. Experimental
[RuCpCl(PPh3)2] (72 mg; 0.1 mmol) was reacted with AgOTs
(25 mg; 0.1 mmol) by refluxing in dichloromethane (5 ml) for
10 min. The reaction mixture was then filtered under argon
over Celite and to the filtrate was added the resin-bound triph-
enylphosphine (135 mg; 0.4 mmol phosphine residues) and stirred
(100 rpm) at room temperature for 48 h. The resin was collected
and washed with dichloromethane (3 ml × 4 ml). The efficiency of
the loading was measured in duplicate by means of measuring the
Ru-content in the combined filtrates with ICP-AES to find that 97%
of the Ru-complex was present on the support.
2.1. General
All manipulations were performed under an argon atmosphere
using standard Schlenk techniques. Solvents were dried and dis-
tilled by standard procedures and stored under argon. DOWEX 50
WX 4 (1.1 mequiv. H+/ml), Amberlyst (4.7 mequiv. H+/g) and Nafion
(0.8 mequiv. H+/g) ion-exchange resins were purchased from
Sigma–Aldrich. Resin-bound triphenylphosphine (resinPhPPh2)
(3 mequiv. phosphine residue/g; 2% DVB/polystyrene) and
[RuCp(ACN)3]PF6 (ACN = acetonitrile) were commercially available
and used as received.
[RuCpCl(PPh3)2] [16], [RuCpCl(dppe)] [17], [RuCpCl(dppdep)]
[2] and [RuCpCl(dppb)] [18], were prepared according to the lit-
erature procedures. The products of the allylation reactions were
described previously in Refs. [1,19]. Loading and leaching of the
ruthenium complexes was quantified by determination of the
Ru-content of the filtrates with a Varian-MPX CCD simultaneous
ICP-AES. Pictures of resin were taken with a Zeiss Axiovert 125M
microscope equipped with a colorview camera. Solid state CP MAS
31P-NMR spectra were recorded on a Bruker MSL 400 spectrometer
operating at 161.99 MHz with a rotational spin speed of 11 kHz.
2.5. Immobilization of [RuCp(ACN)3](PF6) onto resin-bound
triphenylphosphine (catalyst C)
[RuCp(ACN)3](PF6) (44 mg, 0.1 mmol) was dissolved in
dichloromethane (5 ml) and resin-bound triphenylphosphine
was added 133 mg, 0.4 mmol phosphine residues). The resulting
suspension was stirred (100 rpm) for 1 h, after which the solid was
collected by filtration, washed with dichloromethane (3 ml × 2 ml)
and dried in vacuo. The efficiency of the loading was measured in
duplicate by means of measuring the Ru-content in the combined
filtrates with ICP-AES to find that 99% of the initial Ru-complex
was present on the support.
2.2. General procedure for the immobilization of complexes on
ion-exchange resins
2.6. General procedure for catalytic reactions using catalyst
immobilized on ion-exchange resins
0.025 mmol of [RuCp(PP)Cl] and 0.05 mmol of AgOAc were
charged into a reaction vessel which was flushed with argon.
Degassed methanol was added (5 ml) and the mixture was refluxed
for 1 h. After cooling to room temperature, Celite (200 mg) was
added and the suspension was filtered under argon. The residue
was washed with methanol (3 ml× 2 ml) and to the combined fil-
trate, the resin (0.25 mmol H+ residues) was added. This mixture
was stirred (100 rpm) for 15 h at room temperature. Finally the yel-
low beads were collected by filtration and washed with methanol
(3 ml× 2 ml). The efficiency of the loadings was estimated by means
of measuring the Ru-content in the combined filtrates with ICP-AES.
To the immobilized complex (0.025 mmol) under argon atmo-
sphere, 2.5 mmol of 1-octanol or 4-tert-butylphenol was added.
Degassed and dried toluene was added (4 ml) and the mixture was
stirred for 5 min. Allyl alcohol was then added (5 mmol) and the
reaction was stirred (100 rpm) for 20 h at 80 ◦C. Samples were taken
at certain time intervals with an airtight syringe and analyzed by
gas chromatography. After the reaction, the solid was collected by
filtration, washed with methanol (3 ml × 2 ml) and dried in vacuo.
Leaching amounts were measured in duplicate by means of mea-
suring the Ru-content in the combined filtrates with ICP-AES.
2.3. Immobilization of [RuCp(PPh3)2](OTs) onto resin-bound
triphenylphosphine (catalyst A)
2.7. General procedure for catalytic reactions using catalyst
immobilized on resin-bound triphenylphosphine
[RuCpCl(PPh3)2] (72 mg; 0.1 mmol) was reacted with AgOTs
(50 mg; 0.2 mmol) by refluxing in dichloromethane (5 ml) for
10 min. The reaction mixture was then filtered under argon
To the immobilized complex (0.025 mmol) under argon atmo-
sphere, 2.5 mmol of 1-octanol or 4-tert-butylphenol and 0.05 mmol

28
J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
Scheme 2. Synthesis of cationic [RuCp(PP)]+ complexes and immobilization on anionic-exchange resin.
Table 1
of p-toluenesulfonic acid were added. Degassed and dried toluene
was added (4 ml) and the mixture was stirred for 5 min. Allyl alco-
hol was added (5 mmol) and the reaction was stirred (100 rpm)
for 3 h at 80 ◦C or 20 h at 60 ◦C. Samples were taken at certain time
intervals with an airtight syringe and analyzed by gas chromatogra-
phy. After the reaction, the solid was collected by filtration, washed
with dichloromethane (3 ml × 2 ml) and dried in vacuo. Leaching
amounts were measured in duplicate by means of measuring the
Ru-content in the combined filtrates with ICP-AES.
Efficiencies of loading of RuCp-complexes on ion-exchange resins.
Entry
[RuCp(PP)]+ PP=
Resin
Loading efficiencya (%)
1
2
3
4
5
6
7
8
dppe
dppe
dppe
dppe
DOWEX 50 WX 2
DOWEX 50 WX 4
DOWEX 50 WX 8
Amberlyst 15
Nafion NR 50
DOWEX 50 WX 4
Nafion NR 50
72
85
62
79
78
94
98
96
dppe
dppdep
dppb
(PPh3)2
Nafion NR 50
2.8. GLC method
Amount of Ru-complex initially present in solution transferred onto the resin.
a
0.025 mmol of [RuCp(PP)](OAc) was added to 0.25 mmol H+ on resin.
Quantitative gas liquid chromatography analyzes were car-
ried out on a Varian CP-3800 apparatus equipped with a VF-1 ms
(25 m × 0.25 mm) column with decane as internal standard. The
temperature gradient used was: isothermal for 5 min at 40 ◦C, heat-
ing 10 ◦C/min to 250 ◦C and finally isothermal for 5 min at 250 ◦C.
AES (inductively coupled plasma atomic-emission spectroscopy)
and were in the range of 70–85%. The Amberlyst 15 resin, also con-
taining tosylic acid residues, but with a macroreticular structure,
was used for comparison (entry 4). Loading was in the same range
as for the DOWEX resins. Finally for [RuCp(dppe)]+, the Nafion NR
50 resin, with triflic acid residues, was used as a support (entry 5).
Again, immobilization proved to be successful and a high loading
efficiency was achieved. For the other complexes, either DOWEX
50 WX 4 or Nafion NR 50 was used as the support, because with
these resins the highest immobilization efficiencies were obtained
(entries 6–8). Upon introduction of the resins to the ruthenium
solutions the color of the solutions rapidly faded with the concur-
rent coloration of the resin. For the Nafion NR 50 resin, the beads
were homogeneously colored and when cut in half, the yellow color
was also clearly present inside the bead, indicating penetration of
the complex throughout the whole resin.
Interesting is the observation that when a second batch of
resin was added to the loaded resin in solution and the mixture
was stirred for several hours at room temperature, the complex
did not migrate into the fresh resin. Also at reaction temperature
(80 ◦C), migration was not observed. Despite the use of an excess of
10 equiv. of acidic sites with respect to the Ru-complex quantitative
loadings were not achieved after 15 h of reaction time, indicating
that not all acidic sites present on the resin are accessible for the
ruthenium complex to bind. An equilibrium reaction is not play-
ing a role since increasing the amount of ruthenium complex in
solution does not significantly increase the final loading.
3. Results and discussion
3.1. Ionic immobilization of [RuCp(PP)]+
3.1.1. Catalyst synthesis
The RuCp-complexes with bidentate phosphine ligands pre-
viously used in allylation reactions required a non-coordinating
anion in order to be catalytically active. The presence of a tosy-
late (p-toluenesulfonate) anion gives very active catalysts, but also
–
other anions like triflate (trifluoromethanesulfonate) or PF6 can
be used. Many commercially available ion-exchange resins carry
tosylic and triflic acid-type residues, such as DOWEX 50 WX (tosy-
late), Amberlyst 15 (tosylate) and Nafion NR 50 (triflate). The
polystyrene scaffold is expected to be very stable and unreactive
under the reaction conditions described previously in Refs. [1,2].
Therefore these resins were chosen for immobilization. Other sup-
ports like silicates or aluminates contain hydroxyl residues and may
therefore interfere in the desired allylation reactions.
The [RuCp(PP)Cl] complexes were synthesized following the
procedure reported previously [1,2]. The chloride ion was then
exchanged for an acetate anion, by reaction of the complex with
silver(I) acetate in methanol (Scheme 2). Methanol was used as a
solvent, as it enhances swelling of the resin, making its reactive sites
more accessible. The acidic ion-exchange resin was then added to
the solution of the acetate complex.
3.1.2. Catalysis
The lower pKa of the acidic residues (<1) on the resins compared
to the pKa of acetic acid (∼3.5) favors formation of the immobilized
complex and acetic acid. A ratio acidic residues over Ru-complex of
10 was used, to ensure that enough accessible sites for the complex
to bind were available; the presence of an excess of acidic residues
was shown previously to improve activity for allylation reactions
[1–3]. Different catalysts that are used to allylate both aliphatic
alcohols and phenol were thus immobilized. An overview of the
various immobilized catalysts thus prepared and the efficiency of
loading of the various combinations is summarized in Table 1.
The precursor complex [RuCp(dppe)]+ of the most active cat-
alysts was immobilized on the commercially available DOWEX 50
WX resins. These are gel-type resins and different cross-linking per-
centages were employed (entries 1-3). Loading efficiencies were
calculated by analysis of the Ru-content of the filtrate using ICP-
The aliphatic alcohol 1-octanol was investigated for its reactivity
in the allylation with allyl alcohol (Scheme 3). The catalyst of choice
was immobilized [RuCp(dppe)]+, as this proved to be a good catalyst
in the homogeneous system [1]. For [RuCp(dppe)]+ on DOWEX 50
WX2 and DOWEX 50 WX4 (Table 1; entries 1–2), catalytic activity
was observed; however, over multiple runs irreproducible results
were obtained, possibly due to the loss of small amounts of the
relatively small, powdery resin beads during the multiple Schlenk
filtrations. The complex on DOWEX 50 WX8 resin (Table 1; entry
3) did not show any activity in the allylation reaction. This is most
likely caused by limited substrate accessibility due to high cross-
linking percentage of the polystyrene chains in this resin.
For the Nafion NR 50 support, no detectable loss of resin
occurred, since this resin has large beads (10–35 mesh), unlike
the DOWEX 50 WX resins. The results of the multiple catalytic

J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
29
Scheme 3. Allylation of 1-octanol with allyl alcohol as allylating agent, in the presence of immobilized Ru-catalyst.
Fig. 2. Allylation of 4-tert-butylphenol with allyl alcohol using immobilized
[RuCp(dppe)]+ on Nafion NR 50 over multiple runs. Reaction conditions: ratio 4-
tert-butylphenol/allyl alcohol/Ru-complex = 100/200/1, toluene, 80 ◦C. White bars
represent conversion after 6 h, black bars after 20 h. Only C-allylated products (6 + 8)
are formed. Exact conversion numbers are indicated on top of the bars. Values in
parentheses give percentage of Ru-complex leached from the resin relative to total
Ru-complex after 20 h.
Fig. 1. Allylation of 1-octanol with allyl alcohol using immobilized [RuCp(dppe)]+
on Nafion NR 50 over multiple runs. Reaction conditions: ratio 1-octanol/allyl
alcohol/Ru-complex = 100/200/1, toluene, 80 ◦C. 2.5 mmol of 1-octanol was used in
each run. White bars represent conversion after 6 h, black bars after 20 h. Exact
conversion numbers are indicated on top of the bars. Values in parentheses give per-
centage of Ru-complex leached from the resin relative to total Ru-complex present
after 20 h.
Also 4-tert-butylphenol was used as
a substrate in the
heterogeneously-catalyzed allylation reaction. Phenols can be O-
allylated as well as C-allylated, thus forming several products
(Scheme 4) and it was previously found that the structure of
the catalyst plays a crucial role in determining the selectivity
of this reaction. The results of the multiple reaction runs using
[RuCp(dppe)]+ immobilized on Nafion NR 50 as the catalyst are
shown in Fig. 2.
reaction runs using [RuCp(dppe)]+ on Nafion NR 50 are shown in
Fig. 1.
It was shown previously that the use of an apolar solvent like
toluene is essential to obtain a reasonable conversion in the ally-
lation reaction. However, the use of such an apolar solvent is not
beneficial for the swelling of the hydrophilic resins. Nonetheless,
the allylation of 1-octanol with [RuCp(dppe)]+ on Nafion NR 50 pro-
ceeded nicely with high yields and a very low level of leaching of the
catalyst from the support. The ruthenium complex is necessary for
allylation activity; the unloaded acidic resin does not show activity
for allylation under the reaction conditions used. Over five con-
secutive runs, the conversions decreased significantly; however,
this seems not to be caused by leaching of the catalyst, which is
orders of magnitude lower than the decrease in conversion. The
liquid reaction mixture which was separated by filtration from
the immobilized catalyst did not show any activity in allylation
reactions, indicating that the complex leached from the support is
not causing activity during a reaction run. The decrease in activ-
ity could be caused by the retention of water in the resin due to
the strongly hydrophilic sulfonate groups; during the multiple runs
water thus accumulates in the resin, as a result shifting the reac-
tion equilibrium towards the starting materials. When the beads
are thoroughly warmed under vacuum (at 60 ◦C for 2 h), loss of
water is observed. A second explanation could be that chemical
degradation of the catalyst occurs on the support without loss of
its cationic nature, therefore staying immobilized.
The reaction of 4-tert-butylphenol with allyl alcohol using the
catalyst [RuCp(dppe)]+ on Nafion resin proceeded again with rel-
atively high conversion and low leaching. This catalytic system
is completely selective for C-allylation; O-allylated product was
not observed, not even at very short reaction times. Unlike in the
allylation of 1-octanol, no significant deactivation was observed
over multiple runs. C-allylation of phenols is thermodynamically
favored over O-allylation and is an irreversible process [1]. In con-
trast to O-allylation, the presence of water in the hydrophilic resin
does not thermodynamically hinder the C-allylation reaction. How-
ever, the water environment inside the resin will change the local
reaction medium and thus will certainly have its effect on the catal-
ysis, for example by limiting conversion to about 72% after 20 h.
This water is also formed in the reaction of two molecules of allyl
alcohol into diallyl ether, which also is an allylating agent, as was
described previously in Ref. [1]. The hydrophobic diallyl ether will
not be easily taken up in the hydrophilic resin, which limits its use
as allyl donor.
In order to further investigate the possibilities for O-allylation,
catalysts that were found to be highly selective for O-allylation
Scheme 4. Allylation of 4-tert-butylphenol with allyl alcohol as allylating agent, in the presence of immobilized Ru-catalyst.

30
J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
Table 2
Allylation of 4-tert-butylphenol with allyl alcohol using immobilized [RuCp(PP)]+a
.
Entry
[RuCp(PP)]+ PP=
Resin
Conversion of 4 (%)
Selectivity (%)b
6 h
20 h
5
6–8
1
2
3
dppdep
dppb
(PPh3)2
DOWEX 50 WX 4
Nafion NR 50
Nafion NR 50
4
13
1
11
30
2
37
12
100
63
88
0
a
Reaction conditions: ratio 4-tert-butylphenol/allyl alcohol/Ru-complex = 100/200/1, toluene, 80 ◦C.
After 6 h.
b
were also immobilized onto the resin in a similar fashion (Table 1;
entries 6–8) and tested for their activity in the allylation of
4-tert-butylphenol (Table 2). Unfortunately, the immobilized cat-
alyst [RuCp(dppdep)]+ is far less selective for O-allylation than
the homogeneous counterpart, for which selectivities of 80% and
higher were obtained at conversions below 50% [2]. The com-
plexes [RuCp(dppb)]+ (entry 2) and [RuCp(PPh3)2]+ (entry 3), which
have been shown to be very selective in the homogeneous ally-
lation reaction [1,3], show a very different catalytic behavior
when immobilized onto an ion-exchange resin. The immobilized
catalyst [RuCp(dppb)]+ showed relatively high conversion of 4,
but selectivity for O-allylation was low. The immobilized cata-
lyst [RuCp(PPh3)2]+ showed very low activity, although the trace
amount of product formed was the desired allyl phenyl ether. The
fact that no propanal was formed indicates that the acidic residues
on the support are available for catalysis and in this case clearly
block the undesired allyl alcohol isomerization reaction. The differ-
ent behavior of these catalysts when immobilized on ion-exchange
resins is ascribed to the accumulation of water in the hydrophilic
pores of the resin. Although seemingly the catalyst in Table 2, entry
3 is not active in the allylation of phenol, diallyl ether and thus
water is formed in all cases. The O-allylation equilibrium therefore
is shifted toward the starting material side, while the irreversible
formation of C-allyl phenol is not hindered by the local high con-
centrations of water.
Fig. 3. CP MAS 31P-NMR spectra of (a) resinPhPPh2, (b) AgPPh3/
RuCp(PPh3)2(resinPhPPh2)](OTs) (catalyst A), (c) RuCp(PPh3)2(resinPhPPh2)](OTs)
(catalyst B) and (d) [RuCp(ACN)(resinPhPPh2)](PF6) (catalyst C). Signals origi-
nating from the excess of free resin-bound phosphine are marked resinPhPPh2,
whereas those originating from Ru-coordinated phosphines are marked Ru–P and
Ag-coordinated phosphines Ag–P. Signals originating from spinning side-bands are
denoted *.
3.2. Coordinative immobilization of [RuCp(PP)]+
3.2.1. Catalyst synthesis
In order to avoid the hydrophilic character of the anionic-
exchange resins, another type of linkage between the catalyst and
the support was investigated; a coordinative interaction between
a covalently-bound phosphine at the support and a Ru-complex is
used. Such an immobilization should not change the structure of the
catalyst too much, as large changes in the coordination sphere of
the metal-ion may have a dramatic effect on the catalytic behavior
of the ruthenium complex [1–3]. Therefore a covalently immobi-
lized triphenylphosphine analogue (resinPhPPh2) on a polystyrene
backbone (Merrifield-type) was chosen as the solid support [20].
The use of monodentate phosphine ligands on ruthenium, together
with an acidic promoter, has been shown to create a very active
catalyst with high selectivity for the desired O-allylated phenol [3].
The resin used has 3 mequiv. phosphine groups per gram
of resin, supported on a polystyrene resin with 2% divinyl-
depending on the extent of substitution of the triphenylphosphine
ligands. After stirring for 48 h at room temperature, 31P-NMR of
the filtrate did not show a signal corresponding to free triph-
enylphosphine, not even when the immobilized resin was washed
thoroughly with dichloromethane by means of Soxhlet extraction.
Solid state 31P-NMR of the loaded resin (Fig. 3b) indicates the
presence of multiple phosphine-containing species: the free phos-
phine on resin (resinPhPPh2) at–6.3 ppm (also present in Fig. 3a),
Ru-coordinated phosphine at +36.6 ppm, and a P species with a res-
onance at +5.2 ppm (Fig. 3b). The latter resonance is assigned to a
silver–phosphine species, that is present due to the excess of AgOTs
used in the dehalogenation step (Scheme 5a). It was confirmed that
triphenylphosphine coordinated to Ag(I)(OTs) in solution indeed
shows a resonance around +5 ppm. The fact that no free triph-
enylphosphine liberated from the Ru-complex was found in the
filtrate is thus most likely caused by the coordination of the released
phosphine ligand onto the Ag+ ion. Indeed, when [RuCpCl(PPh3)2]
is reacted with 1 equiv. of AgOTs and the resulting mixture is added
to the resin after filtration, the peak at +5.2 ppm is not present
(Fig. 3c; Scheme 5b) and the filtrate was shown to contain free
triphenylphosphine.
benzenepolystyrene cross-linking. In solid state 31P-NMR,
a
single resonance is clearly observed (Fig. 3a). An excess of
4 equiv. of resinPhPPh2 was used with respect to the Ru-
complex. [RuCp(PPh3)2](OTs), synthesized from [RuCpCl(PPh3)2]
and 2 equiv. of AgOTs, was reacted with resinPhPPh2 in the
non-coordinating solvent dichloromethane. After stirring for 48 h,
the solution was almost completely colorless and the resin had
obtained a yellow color. The efficiency of the loading was deter-
mined again with ICP-AES of the filtrate and it was found that
91% of the complex was transferred from solution onto the resin.
The catalyst on resin, however, may consist of various species,
The complex [RuCp(ACN)3](PF6) (ACN = acetonitrile) was also
immobilized onto the resin-bound phosphine (Scheme 5c). This
precursor compound is a starting material often employed for
the synthesis of various Ru(II)Cp-complexes and its coordination
behavior has been widely studied [21]. Since it is known that

J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
31
Scheme 5. Synthesis of (a) immobilized [RuCpCl(PPh3)2] + 2 AgOTs on resin-bound phoshine (catalyst A), (b) immobilized [RuCpCl(PPh3)2] + 1 AgOTs on resin-bound phoshine
(catalyst B) and (c) immobilized [RuCp(ACN)3](PF6) on resin-bound phosphine (catalyst C). ACN = acetontrile.
Fig. 4. Allylation of 1-octanol with allyl alcohol using immobilized
Fig. 5. Allylation of 4-tert-butylphenol with allyl alcohol using immobilized
[RuCp(PPh3)2](OTs) (catalyst A) on resin-bound triphenylphosphine over multiple
[RuCp(PPh3)2]+ (catalyst A) over multiple runs. Reaction conditions: ratio 4-tert-
runs. Reaction conditions: ratio 1-octanol/allyl alcohol/[Ru]/HOTs = 100/200/1/2;
butylphenol/allyl alcohol/[Ru]/HOTs = 100/200/1/2; toluene; 80 ◦C, 3 h; yield of
toluene; 80 ◦C, 60 min; exact conversion number indicated on top of bar. Values in
O-allylated product (5) represented by black bar, with exact number in white; yield
of C-allylated products (6–8) represented by the white bar, with exact number in
parentheses give percentage of [Ru] leached from the resin after 60 min.
black. Values in parentheses give percentage of [Ru] leached from the resin after 3 h.
phosphines easily displace an acetonitrile ligand from this ruthe-
nium complex [22], this compound was added to resinPhPPh2 in
dichloromethane. Indeed immediate coordination to the resin took
place and a complex with the proposed formulation [RuCp(ACN)
(resinPhPPh2)2] (PF6) was obtained. From the initial amount of
Ru-complex present in solution, 99% was loaded onto the resin.
This newly obtained immobilized catalyst was analyzed by solid
state 31P-NMR (Fig. 3d) and it was clearly observed that the
resinPhPPh2 was coordinated to ruthenium. Also the presence of
the PF6 anion is clearly indicated by the NMR spectrum. Most likely
two resinPhPPh2 moieties are coordinated to the ruthenium cen-
ter, since it has been reported that displacement of two acetonitrile
ligands with two phosphine ligands readily occurs at room temper-
Fig. 6. Allylation of 4-tert-butylphenol with allyl alcohol using immobilized
ature [22].
[RuCp(PPh3)2]+ (catalyst A) over multiple runs. Reaction conditions: ratio 4-tert-
The complexes thus immobilized are air stable when properly
dried and can be used after several weeks without a significant
butylphenol/allyl alcohol/[Ru]/HOTs = 100/200/1/2; toluene; 60 ◦C, 20 h; yield of
O-allylated product (5) represented by black bar, with exact number in white; yield
decrease in activity.
of C-allylated product (6–8) represented by the white bar, with exact number in
black. Values in parentheses give percentage of [Ru] leached from the resin after
20 h.
3.2.2. Catalysis
The catalytic activity of catalyst A was investigated in the ally-
lation of 1-octanol with allyl alcohol (Fig. 4).
lyst A showed a very high activity in the first run of the allylation
of 1-octanol. The substrate was fully converted within 60 min into
allyl octyl ether. In the following runs the conversion after 60 min
was significantly lower, while the amount of complex leached from
the support was constant. It must be noted that these conversions
after 60 min were obtained with a catalyst loading much higher
than that used in the homogeneous system (1.0 mol% vs. 0.1 mol%
The homogeneous catalyst [RuCp(PPh3)2]+ was previously
demonstrated to be only active for allylation in the presence of
strong acid [3]. For the catalysts immobilized on ion-exchange
resins described in Section 2.1, the acid required for catalytic activ-
ity was present on the support, but in this case external acid HOTs
is needed to be added and was therefore present in solution. Cata-

32
J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
Fig. 7. Allylation of 4-tert-butylphenol with allyl alcohol using immobilized
[RuCp(PPh3)2]+ (catalyst B) over multiple runs. Reaction conditions: ratio 4-tert-
butylphenol/allyl alcohol/[Ru]/HOTs = 100/200/1/2; toluene; 80 ◦C, 3 h; yield of
O-allylated product (5) represented by black bar, with exact number in white; yield
of C-allylated products (6–8) represented by the white bar, with exact number in
black. Values in parentheses give percentage of [Ru] leached from the resin after 3 h.
Fig. 9. Allylation of 4-tert-butylphenol with allyl alcohol using immobi-
lized [RuCp(ACN)3](PF6) (catalyst C) on resin-bound triphenylphosphine
over multiple runs. Reaction conditions: ratio 4-tert-butylphenol/allyl alco-
hol/[Ru]/HOTs = 100/200/1/2; toluene; 80 ◦C, 3 h; yield of O-allylated product
(5) represented by black bar, with exact number in white; yield of C-allylated
product (6–8) represented by the white bar, with exact number in black. Values in
parentheses give percentage of Ru-complex leached from the resin after 3 h.
Scheme 6. Substitution of an ACN ligand from [RuCp(resinPhPPh2)2(ACN)](PF6)
(catalyst C) with allyl alcohol.
The conversion gradually decreased over the first consecutive
four runs; in the fifth run only 10% of phenol was converted. The
amount of leached complex does not account for this rapid decrease
and therefore a change or chemical deactivation of the complex
present on the support must occur. The reaction was also per-
formed at 60 ◦C in order to investigate whether this rapid decrease
of activity could be prevented by lowering the reaction temperature
(Fig. 6).
Fig. 8. CP MAS 31P-NMR spectra of (a) RuCp(PPh3)2(resinPhPPh2)](OTs) (cata-
lyst B) and (b) catalyst B after first run in allylation of 4-tert-butylphenol with
allyl alcohol. Signals originating from the excess of free resin-bound phosphine
are marked resinPhPPh2, whereas those originating from Ru-coordinated phos-
phines are marked Ru–P and those originating from allyl phosphonium salts
resinPhP+(allyl)Ph2. Signals originating from spinning side-bands are denoted *.
The activity of catalyst A at 60 ◦C of course is lower than at 80 ◦C
and therefore the reaction was run for 20 h; again, a high selectivity
for O-allylation was observed. Intriguingly, the fifth consecutive run
shows again a large decrease in conversion; apparently deactiva-
tion of the catalyst is also taking place at this reaction temperature.
Catalyst B was also tested as catalyst for the allylation of phenol
and the results over five subsequent runs are shown in Fig. 7. The
large difference between the first and second run can be explained
by the consumption of the acid added to the reaction. Phosphines
(here immobilized) react with allyl alcohol and strong acid into
on octanol for the homogenous catalysts) [3]. Leaching of ruthe-
nium was considerably higher than for the catalysts immobilized
using electrostatic interactions. This is indicative for a lower stabil-
ity of a coordination bond compared to the electrostatic interaction.
The liquid phase reaction mixture containing the complex leached
from the support was separately tested for catalytic activity, but
showed no conversion of added 1-octanol and this proved that the
allylation activity is really due to the heterogeneous catalyst.
The challenge is of course, to perform selective O-allylation of
phenol and therefore the allylation of 4-tert-butylphenol was per-
formed using catalyst A; the results are shown in Fig. 5.
While the catalysts immobilized on ion-exchange resin
formed predominantly C-allylated product, this coordinatively-
bound catalyst selectively forms O-allylated product; only traces
of C-allylated products were observed. After 3 h, conversions
approaching 70% were reached. It must again be noted that sim-
ilar conversions were obtained after 3 h with the homogeneous
system [3] with the difference that a much higher catalyst load-
ing was used here and a higher temperature is used (80 ◦C vs. 60 ◦C
for the homogeneous system). Thus the heterogeneous system is
significantly less active, which is most likely caused by diffusion
limitations of the substrates into the solid support.
Scheme 7. Proposed substitution of triphenylphosphine with resin-bound triph-
enylphosphine.

J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
33
Scheme 8. Proposed mechanism for conversion of heterogeneous Ru-complex to homogeneous Ru-complex.
allyl phosphonium salts and water, but only in the presence of an
3.2.4. Physical deactivation of the catalysts
allylation catalyst [3]. Due to consumption of the acid propanal
is also formed by isomerization of allyl alcohol [3]. The immobi-
lized allylphosphonium groups are observed in solid state 31P-NMR
(Fig. 8), where the resonance at −6.3 ppm from the resinPhPPh2
(Fig. 8a) has disappeared with the appearance of a resonance at
19.7 ppm assigned to the allyl phosphonium species (Fig. 8b).
In the second run with catalyst B, when the free resinPhPPh2
moieties have been totally converted into phosphonium species,
the acid co-catalyst (HOTs) is not consumed, thus making the cat-
alyst active for O-allylation. In the following runs, conversions and
selectivities towards O-allylation slowly decreased, but the cata-
lyst B appeared to be more stable than catalyst A. The fact that
catalyst A shows a much higher conversion in the first run is most
likely caused by the coordination of the silver(I)tosylate to the
resinPhPPh2 moiety, making them inaccessible for quarterisation
and preventing acid consumption. This hypothesis is supported by
the observation that propanal is not formed in the first run when
catalyst A is used.
When catalyst C was used as a catalyst in the reaction between
4-tert-butylphenol and allyl alcohol in the presence of acid (Fig. 9),
in the first run only 35% of allyl phenyl ether was obtained and
propanal was formed, again indicative of the consumption of acid
and quarternisation of the remaining free phosphines. The remain-
ing molecule of acetonitrile is not playing a role during the catalysis,
since it is most likely replaced by the large excess of substrate allyl
alcohol present during the reaction (Scheme 6), creating an almost
identical complex as described earlier with two coordinated triph-
enylphosphine ligands and a non-coordinating anion. The amount
of ruthenium leached from the support is high in the first run, but
rather low in the following runs. Conversions of phenol after 3 h are
relatively stable over multiple runs and the selectivity is similar to
that in the reactions with catalyst B.
After each run the reaction mixture was cooled to room
temperature, filtered and the residual resin was washed with
dichloromethane. Leaching of ruthenium was determined by quan-
tifying the amount of Ru present in the combined filtrates after
each reaction run. For catalyst A, it appeared that the leaching is
considerably higher when the reaction mixture was filtered while
still hot (8.4% vs. 2.5% in a single run). This indicates that when the
immobilized catalyst is heated to reaction temperature, it dissoci-
ates from the support and can be active as if it were a homogenous
catalyst. The equilibrium reaction shown in Scheme 8 determines
the amount of homogeneous complex present in solution. Most
likely, the triphenylphosphine ligand still present on the silver ions
plays a crucial role to facilitate this process, because for the cat-
alysts B and C such a difference in the amount of leached Ru is
not observed and leaching of ruthenium at reaction temperature
is similar to that at room temperature. Also, after the first run a
difference between cold and warm filtration is not present any-
more for catalyst A, probably because the excess of phosphine is
converted to allyl phosphonium salt. The extra equivalent of PPh3
induces this difference in leaching, because at higher temperatures,
the weakly coordinated PPh3 on silver(I) dissociates, migrates to a
Ru-species, and displaces a resinPhPPh2 group, thus resulting in a
homogeneous complex. At room temperature, the Ag–PPh3 com-
plex is more stable and the PPh3 ligand present on Ru can migrate
back to the silver ion present on the resin.
Such a “boomerang” effect, where the catalyst dissociates from
the support during the reaction and returns after the reaction is
finished (in our case when the reaction is cooled down) has been
reported earlier for other catalytic systems [23,24]. The advantage
of such a system is that the catalyst is homogeneous during the
reaction and thus shows a high activity and selectivity. After the
reaction, the catalyst becomes heterogeneous again and is easily
separated from the reaction mixture. A disadvantage is that the
amount of complex leached is significantly higher than for a fully
heterogeneous system.
3.2.3. Chemical deactivation of the catalysts
Analysis with 31P-NMR of the filtrate after the first run in the
series with catalysts A–C shows the presence of phosphonium salts,
indicating that free triphenylphosphine was present during the
reactions, which is rapidly converted by the catalyst into allyl phos-
phonium salts (Scheme 7) as described previously in Ref. [3]. When
a triphenylphosphine ligand is released, its place will be taken by
an immobilized phosphine group (Scheme 7). However, the immo-
bilized phosphines also form immobilized phosphonium salts as
indicated by the NMR spectra in Fig. 8. Coordinated phosphines
are not prone to allylation; however, when the concentration of
the substrate decreases, it is expected that some quarternization
of coordinated phosphines occurs, thus causing deactivation of the
catalyst [3].
4. Conclusions
The use of ion-exchange resins as support has been shown to
yield stable immobilized cationic Ru-complexes with which cat-
alytic allylation reactions were successfully performed. Due to the
hydrophilicity of these resins, however, water is retained in the
resin, which is detrimental for the selective O-allylation of phe-
nol. The retention of water in the resin causes the formation of
only C-allylated products in the catalytic allylation of phenol. The
immobilization of the ruthenium catalysts by covalently-bound
phosphine onto supports without hydrophilic residues makes O-
allylation possible. Although leaching from the resin is considerably

34
J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 330 (2010) 26–34
higher than for the catalyst immobilized on ion-exchange resins,
the activity over multiple runs is relatively stable. The activity of
all heterogeneous systems described in this paper is significantly
lower than that of the homogenous catalysts [1–3].
[7] N.E. Leadbeater, M. Marco, Chem. Rev. 102 (2002) 3217–3273.
[8] C.E. Song, S.G. Lee, Chem. Rev. 102 (2002) 3495–3524.
[9] N.E. Leadbeater, K.A. Scott, L.J. Scott, J. Org. Chem. 65 (2000) 3231–3232.
[10] C.U. Pittman, S.K. Wuu, S.E. Jacobson, J. Catal. 44 (1976) 87–100.
[11] N.E. Leadbeater, J. Org. Chem. 66 (2001) 2168–2170.
[12] J.M. Fraile, J.I. Garcia, J.A. Mayoral, Chem. Rev. 109 (2009) 360–417.
[13] R. Selke, M. Capka, J. Mol. Cat. 63 (1990) 319–334.
Acknowledgements
[14] P. Barbaro, C. Bianchini, G. Giambastiani, W. Oberhauser, L.M. Bonzi, F. Rossic,
V. Dal Santo, Dalton Trans. (2004) 1783–1784.
[15] P. Barbaro, F. Liguori, Chem. Rev. 109 (2009) 515–529.
[16] M.I. Bruce, F.S. Wong, B.W. Skelton, A.H. White, J. Chem. Soc. Dalton Trans.
(1981) 1398–1405.
[17] A.G. Alonso, L.B. Reventos, J. Organomet. Chem. 338 (1988) 249–254.
[18] R.C. van der Drift, M. Gagliardo, H. Kooijman, A.L. Spek, E. Bouwman, E. Drent,
J. Organomet. Chem. 690 (2005) 1044–1055.
[19] A. Dahlen, A. Sundgren, M. Lahmann, S. Oscarson, G. Hilmersson, Org. Lett. 5
(2003) 4085–4088.
[20] M. Bernard, W.T. Ford, J. Org. Chem. 48 (1983) 326–332.
[21] C. Slugovc, E. Ruba, R. Schmid, K. Kirchner, K. Mereiter, Monatsh. Chem. 131
(2000) 1241–1251.
This work was supported by the Technology Foundation STW.
We would like to thank. Dr. R. Postma (Hexion) and Dr. J.K. Buijink
(Shell Global Solutions International BV) for fruitful discussions.
References
[1] J.A. van Rijn, M. Lutz, L.S. von Chrzanowski, A.L. Spek, E. Bouwman, E. Drent,
Adv. Synth. Catal. 351 (2009) 1637–1647.
[2] J.A. van Rijn, M.A. Siegler, A.L. Spek, E. Bouwman, E. Drent, Organometallics 28
(2009) 7006–7014.
[3] J.A. van Rijn, E. van Stapele, E. Bouwman, E. Drent, J. Catal. 272 (2010) 220–226.
[4] I.F. Nieves, D. Schott, S. Gruber, P.S. Pregosin, Helv. Chim. Acta 90 (2007)
271–276.
[5] H. Saburi, S. Tanaka, M. Kitamura, Angew. Chem. Int. Ed. 44 (2005) 1730–1732.
[6] Y. Tada, A. Satake, I. Shimizu, A. Yamamoto, Chem. Lett. 12 (1996) 1021–1022.
[22] E. Ruba, W. Simanko, K. Mauthner, K.M. Soldouzi, C. Slugovc, K. Mereiter, R.
Schmid, K. Kirchner, Organometallics 18 (1999) 3843–3850.
[23] M. Ahmed, A.G.M. Barrett, D.C. Braddock, S.M. Cramp, P.A. Procopiou, Tetrahe-
dron Lett. 40 (1999) 8657–8662.
[24] L. Jafarpour, S.P. Nolan, Org. Lett. 2 (2000) 4075–4078.