Palladium-diphosphine complexes as catalysts for allylations with allyl alcohol

Article abstract of DOI:10.1016/j.molcata.2010.06.023

Several palladium complexes with bidentate phosphine ligands were tested for their activity in the O-allylation of phenols with allyl alcohol. The use of C₃-bridged bidentate phosphine ligands results in very high selectivity for O-allylation. The reactions do not require stoichiometric amounts of additives to control the chemoselectivity. Especially, catalysts with gem-dialkyl substituted C₃-bridged bidentate phosphine ligands perform very well, resulting in a (equilibrium) conversion of ～50% of phenol with a selectivity of 99% for O-allylation. The use of diallyl ether as the allylating agent results in a significant increase in phenol conversion while maintaining high selectivity for O-allylation. Apart from Pd(OAc)₂ as catalyst precursor, Pd(dba)₂ was also employed, making it possible to use other types of phosphine or phosphite ligands. With the palladium catalytic system not only phenol, but also aliphatic alcohols can be allylated, as well as aromatic and aliphatic amines.

Full text of DOI:10.1016/j.molcata.2010.06.023

Journal of Molecular Catalysis A: Chemical 329 (2010) 96–102
Contents lists available at ScienceDirect
Journal of Molecular Catalysis A: Chemical
journal homepage: www.elsevier.com/locate/molcata
Palladium–diphosphine complexes as catalysts for allylations with allyl alcohol
Jimmy A. van Rijn, Angela den Dunnen, 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
Article history:
Received 13 May 2010
Received in revised form 18 June 2010
Accepted 23 June 2010
Available online 1 July 2010
Several palladium complexes with bidentate phosphine ligands were tested for their activity in the O-
allylation of phenols with allyl alcohol. The use of C3-bridged bidentate phosphine ligands results in
very high selectivity for O-allylation. The reactions do not require stoichiometric amounts of additives
to control the chemoselectivity. Especially, catalysts with gem-dialkyl substituted C3-bridged bidentate
phosphine ligands perform very well, resulting in a (equilibrium) conversion of ∼50% of phenol with a
selectivity of 99% for O-allylation. The use of diallyl ether as the allylating agent results in a significant
increase in phenol conversion while maintaining high selectivity for O-allylation. Apart from Pd(OAc)2
as catalyst precursor, Pd(dba)2 was also employed, making it possible to use other types of phosphine or
phosphite ligands. With the palladium catalytic system not only phenol, but also aliphatic alcohols can
be allylated, as well as aromatic and aliphatic amines.
Keywords:
Catalysis
Allylation
Palladium
Allyl alcohol
Phosphines
Phenol
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
Apart from the observation that selectivity of the reaction is time-
dependent, the role of the ligand appeared to be of great importance
In the context of the development of an environmentally benign
catalytic route to epoxy resins, O-allylation of phenols is a highly
desirable reaction [1]. Allyl phenyl ethers can be epoxidized to
obtain glycidyl ethers, which are currently produced in the epoxy
resin industry via the conventional epichlorohydrin route with the
co-production of stoichiometric amounts of chloride-containing
waste. Allylation reactions of phenols are mostly performed with
allylating agents such as allyl chloride or allyl acetate and sto-
ichiometric amounts of base are added to induce selectivity for
O-allylation [2,3]. Numerous allylation reactions, generally known
as the Tsuji–Trost reaction, using palladium catalysts have been
reported [4–6]. Regarding atom efficiency, it would be desirable to
use allyl alcohol as the allylating agent. Several palladium systems
have been reported to be able to use allyl alcohol as an allylating
agent [7–11]. However, in the cases where phenols are used as the
substrate, stoichiometric amounts of base are necessary to induce
chemoselectivity towards O-allylation, inevitably resulting in inor-
ganic waste [9]. In the absence of base, C-allylation occurs [10], a
feature also observed with ruthenium-based systems [12]. We have
developed a catalytic system based on ruthenium that catalyzes
both O- and C-allylation of phenols; we have shown that the O-
allylated products are reversibly formed, while C-allylated products
are produced irreversibly [13]. For this ruthenium-based system a
large variety of bidentate phosphine ligands has been used [13,14].
for the activity as well as the selectivity; it was shown that with
only minor changes in the ligand-structure dramatic effects on
both activity and selectivity were accomplished. Restricted coor-
dination space at the ruthenium center favors the formation of the
O-allylated product, which could be achieved using ligands that
either have a large bite angle [13] and/or form kinetically stable
chelates [14]. For palladium however, the use of Pd(OAc)2 for the
allylation of phenol-type substrates has only been reported with
monodentate phosphine ligands [8–10].
In this paper a palladium system with a selection of bidentate
phosphine ligands is reported (Fig. 1); the activity and selectivity
for O-allylation as a function of the ligand is discussed.
2. Experimental
2.1. General
All reactions were performed under an argon atmosphere using
standard Schlenk techniques. Solvents were dried and distilled
using standard procedures and were stored under argon. The phos-
phine ligands with phenyl substituents (dppm, dppe, dppp, dppb
and PPh3) and triphenylphosphite were commercially available
and used as received. Pd(OAc)2 and Pd(dba)2 were purchased
from Strem Chemicals. The ligands dppdmp [15] and dppdep [16]
were earlier described in literature. The ligands o-MeOdppp [17],
o-MeOdppdmp [18], p-MeOdppp [19], o-Medppp [20] and trisani-
sylphosphine [21] were a gift from Shell Global Solutions and used
as received. Allyl-1,1-d2 alcohol was synthesized as reported [22].
∗
Corresponding author. Tel.: +31 71 527 4550; fax: +31 71527 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.2010.06.023

J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 96–102
97
(CDCl3): ı 133.5 ( CH), 129.2 (ArH), 117.6 ( CH2), 114.2 (ArH), 68.8
(OCH2), 31.5 (tBu). MS (ESI) m/z = 193.37 [M+H]+
2.3. General procedure for catalytic reactions with Pd(dba)2
5 mmol of 4-tert-butylphenol, 5 ␮mol of Pd(dba)2 and 20 ␮mol
of bidentate phosphine ligand (or 40 ␮mol of monodentate lig-
and) were charged into the reaction vessel which was then flushed
with argon. Degassed and dried toluene (5 ml) was added and
the mixture was stirred for 5 min. Allyl alcohol (10 mmol) was
added and the reaction mixture was heated to 100 ◦C. Sam-
ples were taken at certain time intervals and analyzed by gas
chromatography.
2.4. Procedure for mercury test
The general procedure for catalytic reactions with Pd(OAc)2 was
followed with the difference that after 1 h reaction time, 100 mg
(0.5 mmol) of mercury was added. The reaction mixture was stirred
vigorously and the reaction was continued at a temperature of
100 ◦C. Stirring was temporarily halted while taking samples to
prevent sampling of mercury droplets.
Fig. 1. Overview of the bidentate phosphine ligands employed in this study and
their abbreviations.
1
1H NMR spectra (300 MHz), 13C NMR (75.5 MHz) and 31P{ H} NMR
spectra (121.4 MHz) were measured on a Bruker DPX-300 spec-
trometer. Chemical shifts are reported in ppm. The spectra were
taken at room temperature. Mass spectrometry was performed on
a Finnigan MAT 900 spectrometer equipped with an electrospray
interface.
3. Results and discussion
3.1. Catalysis with Pd(OAc)2 as catalyst precursor
2.2. General procedure for catalytic reactions with Pd(OAc)2
The palladium precatalysts were formed in situ by addition of a
bidentate phosphine ligand to Pd(OAc)2. This results in the initial
formation of the species [Pd(II)(OAc)2(PP)], but it has been reported
that these complexes can be reduced in the presence of an excess
of phosphine ligand to form Pd(0) complexes, acetic anhydride and
phosphine oxide [25,26]. This process is slower for bidentate lig-
ands than for monodentate ligands and does not occur without an
excess of phosphine ligand. Immediate stabilization of the formed
Pd(0) species with ligand or substrate, or subsequent uptake into
the catalytic cycle is highly beneficial for the reduction reaction to
occur [26]. Surprisingly it was found that the reduction of palla-
dium(II) requires the presence of phosphine ligands, despite the
presence of a large excess of allyl alcohol, usually also considered
as a reducing agent.
5 mmol of 4-tert-butylphenol (or an aliphatic alcohol or an
amine), 5 ␮mol of Pd(OAc)2 and 20 ␮mol of bidentate phosphine
ligand (or 40 ␮mol of PPh3) were charged into the reaction vessel
which was then flushed with argon. Degassed and dried toluene or
n-heptane (5 ml) was added and the mixture was stirred for 5 min.
The allyl donor (10 mmol) was added and the reaction mixture was
heated to 100 ◦C. Samples were taken at certain time intervals and
were analyzed by gas chromatography.
The spectroscopic data of the compounds 3–6 [13], allyl 1-octyl
ether, allyl cyclohexyl ether [23], N-allylaniline, N,N-diallylaniline,
N-allyloctan-1-amine and N,N-diallyloctan-1-amine [24] corre-
sponded with the data reported in literature. Spectroscopic data
for allyl-1,1-d2 4-tert-butylphenyl ether and allyl-3,3-d2 4-tert-
butylphenyl ether (13 + 14); mixture of products: 1H NMR (CDCl3):
ı 7.28 (d, 4H, J = 9 Hz, ArH), 6.85 (d, 4H, J = 9 Hz, ArH), 6.05–6.02 (m,
2H, H-allyl), 5.39 (dd, 1H, J = 2 and 17 Hz, H-allyl), 5.27 (dd, 1H, 2 and
9 Hz, H-allyl), 4.51 (d, 1H, J = 6 Hz, OCH2), 1.28 (s, 9H, tBu). 13C NMR
A palladium over diphosphine ligand ratio of 1/4 was found to
give optimal activity for the allylation reaction shown in Table 1.
The use of Pd(OAc)2 to bidentate phosphine ratios of 1/2 or 1/1
results in a lower stability of the catalyst; the conversions are sig-
nificantly lower and precipitation of metallic palladium (plating) is
Table 1
Reaction of 4-tert-butylphenol (1) with allyl alcohol (2) catalyzed by different Pd–diphosphine complexes.a
.
Entry
Pd(OAc)2 + PP
PP =
Conversion of 1 (%)
Selectivity (%)
3
4–6
3
4–6
3
4–6
1 h
3 h
6 h
1 h
3 h
6 h
1
2
3
4
5
dppm
dppe
dppp
dppb
2 PPh3
2
0
15
21
28
6
0
20
35
44
13
2
26
53
56
100
–
100
100
65
0
–
0
0
35
100
–
100
87
29
0
–
0
13
71
100
100
100
72
0
0
0
28
84
16
a
Reaction conditions: ratio 4-tert-butylphenol/allyl alcohol/Pd(OAc)2/PP = 1000/2000/1/4, toluene, 100 ◦C.

98
J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 96–102
Table 2
Reaction of 4-tert-butylphenol (1) with allyl alcohol (2) catalyzed by different Pd–diphosphine complexes starting from Pd(OAc)2.a
Entry
Pd(OAc)2 + PP
PP =
Solvent
Conversion of 1 (%)
Selectivity (%)b
3
4–6
3
4–6
1 h
6 h
22 h
6 h
22 h
1
2
dppp
Toluene
Toluene
Toluene
Toluene
n-Heptane
n-Heptane
n-Heptane
15
16
14
0
11
9
26
41
38
0
31
23
43
38
58
50
0
54
56
80
100
99
99
–
100
99
99
0
1
1
–
0
1
1
94
95
99
–
99
97
85
6
5
1
–
1
dppdmp
dppdep
dppdmp
dppdmp
dppdep
dppdep
3
4c
5
6
3
15
7d
4
a
Reaction conditions: ratio 4-tert-butylphenol/allyl alcohol/Pd(OAc)2/PP = 1000/2000/1/4, toluene, 100 ◦C.
For all entries the selectivity after 1 h 100% for 3.
p-Toluenesulfonic acid was added (10 eq. on Pd).
b
c
d
Diallyl ether was used as the allyl donor (ratio 4-tert-butylphenol/diallyl ether = 1/1).
more pronounced. The results of the use of different phosphine lig-
ands with Pd(OAc)2 in the catalytic allylation of 4-tert-butylphenol
(1) are summarized in Table 1.
and in preventing the formation of palladium black. The results of
the catalytic reactions using the gem-dialkyl ligands are shown in
Table 2.
The addition of dppm as a bidentate phosphine ligand to
Pd(OAc)2 (entry 1) results in the formation of a selective catalyst
for O-allylation of phenol with allyl alcohol, but, the conversion of
1 is relatively low. The solution turned intensely red, indicating the
formation of Pd(I) dimers [27,28]. When dppe is used as the lig-
and (entry 2), the resulting catalytic activity is very low. Formation
of [Pd(dppe)(OAc)2] is slow, due to the relatively high stability of
a [Pd(dppe)2](OAc)2 intermediate [29]. This phenomenon clearly
slows down the formation of the active catalyst. However, when
[Pd(dppe)(OAc)2] is synthesized ex situ and used as a catalyst in the
presence of three equivalents of dppe (as the fourth equivalent is
already present in the complex) to obtain a similar catalytic system
as the in situ formed catalytic system, activity for allylation remains
very low.
Indeed, when the bridge length of the bidentate ligand is
increased to a C3-fragment in dppp (entry 3) the conversion
increases, with maintenance of the very high selectivity. Only O-
allylation is observed, making this the first Pd-based catalyst that
is selective for O-allylation without the need for any stoichiometric
additives.
When using the C4-bridged ligand dppb (entry 4), the con-
version of 1 is even higher, but the selectivity for O-allylation is
decreased. Finally, the catalyst with the monodentate ligand triph-
enylphosphine (entry 5) has a low selectivity for the desired O-allyl
ether; after 6 h mainly C-allylated products are obtained, in agree-
ment with earlier studies with similar substrates [10].
A dramatic effect on both activity and selectivity is thus
observed by simply changing the bridge length of the bidentate
ligand. A drying agent is not required and high catalytic activity
is observed even with a low catalyst concentration of 0.1 mol% of
Pd on 1. However, the stability of these palladium catalysts is dif-
ferent from the Ru-complexes previously reported by us [13,14].
For the ruthenium-based catalysts high turnover numbers could
be achieved when using high substrate over catalyst ratios. For the
Pd-based catalysts described here increasing the substrate over cat-
alyst ratio does not lead to significantly higher turnover numbers
and a maximum TON of ∼800 is reached. During all reactions some
plating of palladium metal was visible, which most likely is the
major deactivation pathway of the catalyst.
The conversion of 1 by the catalytic system with dppp increases
when longer reaction times (22 h) are used (entry 1). For the cat-
alyst with dppdmp (entry 2) as the bidentate phosphine ligand,
initial conversion after 1 h is very similar to that of the reaction
with dppp as the ligand. This indicates a similar rate constant and
thus a comparable activity of these catalysts. Higher conversions
after 6 and 22 h are observed for the catalyst with dppdmp, indi-
cating the formation of a more stable catalyst compared to Pd(OAc)2
with dppp. The use of the gem-diethyl substituted phosphine lig-
and dppdep (entry 3), also leads to higher conversion of 1. When
a yield of about 50% of 3 is obtained, equilibrium is reached and
further conversion of 1 into 3 is halted. In contrast to the cationic
ruthenium-based system, the addition of a strong acid to the reac-
tion mixture does not increase the rate of the reaction, but actually
inhibits the catalyst for allylation (entry 4). When the acid is added
after 1 h reaction time, the reactivity of the catalyst is completely
halted from that moment on, including conversion of O-allylated
products into the thermodynamically favored C-allylated products.
The equilibrium reaction of O-allylation is governed by the
amount of water that is soluble in the reaction mixture. The use
of the apolar solvent toluene limits the solubility of this water in
the reaction mixture and the water forms a separate phase. The
more apolar solvent n-heptane can also be used as a solvent (entries
5–6); this results in a lower conversion after 1 and 6 h, but after
22 h conversions are very similar to those obtained for the reac-
tions in toluene. Apparently, activation of the catalyst proceeds
less efficient in n-heptane than in toluene. The use of dppdmp or
dppdep in this reaction medium results very similar selectivities for
O-allylation after 22 h. Unlike most of the reported studies, which
need to remove water from the reaction mixture, a water scavenger
is not necessary as in this apolar system the water forms a separate
phase. However, complete water removal from the system would
probably be beneficial to obtain higher yields of 3 in a batch pro-
cess and higher conversion per pass in a continuous process. When
diallyl ether is used as the allylating agent (entry 7), the conversion
of 1 and the yield of 3 after 22 h is indeed increased significantly in
agreement with the lower quantity of water produced relative to
that using allyl alcohol as allylation agent. However, at high phe-
nol conversion and very long reaction time, C-allylated product 4
is also formed.
It has been found previously that the use of kinetically sta-
ble chelating bidentate phosphine ligands in cationic ruthenium
complexes results in very selective O-allylation of phenols with
allyl alcohol [14]. Although the selectivity of the palladium sys-
tem is already high when dppp is used as the ligand, the use
of kinetically more stable chelating ligands might aid in further
stabilization of the intermediate zerovalent palladium complexes
3.2. Catalysis with Pd(dba)2 as catalyst precursor
The use of phosphine ligands that are not oxidized easily, such as
o-anisylphosphines previously used for Pd-catalyzed olefin–carbon
monoxide copolymerization [30], or the use of triphenylphos-

J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 96–102
99
Table 3
If such steric hindrance would pose a limiting factor on activation,
the catalyst with o-Medppp would be less active compared to the
catalyst with dppp, which is not the case. Most likely coordination
of the methoxy group, as already reported for similar complexes
[31], hampers activation of allyl alcohol at the palladium centre
and thus results in lower catalytic activity. The use of electron-
withdrawing triphenylphosphite as the ligand (entry 7) for catalytic
allylation also results in very low conversion of 1. From these results
it is clear that addition of unsubstituted phenylphosphine ligands,
in particular the bidentate ligands dppp, dppdmp and dppdep, to
either a source of Pd(II) or Pd(0) yields the most active and selective
catalysts for the allylation of phenols.
Reaction of 4-tert-butylphenol (1) with allyl alcohol (2) catalyzed by different
Pd–diphosphine complexes starting from Pd(dba)2.a
Entry
Pd(dba)2 + PP
Conversion of 1
after 22 h (%)
Selectivity after
22 h (%)
PP =
3
4–6
1
2
3
4
5
6
7
dppdmp
49
10
12
17
38
72
11
96
4
0
0
0
0
o-MeOdppp
o-MeOdppdmp
2 P(o-An)3
p-MeOdppp
o-Medppp
2 P(OPh)3
100
100
100
100
71
29
0
100
a
Reaction conditions: ratio 4-tert-butylphenol/allyl alcohol/Pd(dba)2/
PP = 1000/2000/1/4, toluene, 100 ◦C.
3.3. Allylation of other nucleophiles
Apart from 4-tert-butylphenol, other nucleophiles have been
explored for their reactivity in the presence of the Pd(OAc)2 with
dppdmp and the results are summarized in Table 4. The aliphatic
alcohol 1-octanol is efficiently allylated (Table 4; entry 1), while the
secondary alcohol cyclohexanol is much less reactive (entry 2); only
low conversion towards the allyl ether is observed. Not only alco-
hols are readily allylated, but also nitrogen-containing substrates
such as aniline can be efficiently allylated with allyl alcohol. After
20 h, conversion is complete when an excess of allyl alcohol is used
and mainly the N,N-bisallylated product is obtained (entry 3). After
3 h the conversion already is quite high, and mostly monoally-
lated aniline is present (entry 4). The high conversion after only
3 h is illustrative for the much higher reactivity of aniline com-
pared to hydroxyl-containing substrates, which need considerably
longer reaction times to achieve similar conversions. By reduc-
ing the amount of allyl alcohol, N-monoallylated aniline is formed
selectively, although the conversion is considerably lower (entry 5).
Finally, apart from aromatic amines, also alkylamines can be ally-
lated, as shown for n-octylamine, which is less reactive than aniline
(entry 6).
phite in combination with Pd(OAc)2 does not yield active catalysts
for allylation. In combination with the Pd(0) precursor Pd(dba)2,
(dba = dibenzylideneacetone) however, the use of these ligands
does result in an active catalytic system and the results are shown
in Table 3.
When dppdmp is added to Pd(dba)2 (entry 1), a similar conver-
sion with high selectivity is reached compared to the reactions in
which Pd(OAc)2 is used as precursor, indicating that in situ the same
catalyst is obtained. For o-MeOdppp (entry 2) and o-MeOdppdmp
(entry 3), the anisyl analogues of dppp and dppdmp, only a very low
conversion is observed, albeit with high selectivity for 3. The addi-
tion of monodentate tris(o-anisyl)phosphine as the ligand (entry
4) results in a slightly higher conversion of 1, but when compared
to its unsubstituted analogue triphenylphosphine in combination
with Pd(II) (Table 1; entry 5), the resulting catalyst reaches consid-
erably lower conversion. In order to investigate steric vs. electronic
effects of the different ligands on the catalytic results, p-MeOdppp
was used as ligand (entry 5). In this case the conversion of 1 is
considerably higher than for o-MeOdppp and comparable to that
observed for the combination of Pd(OAc)2 with dppp as the ligand
(Table 1; entry 3); therefore an electronic effect can be excluded.
The addition of o-Medppp (entry 6) results in a considerably higher
conversion than when o-MeOdppp is used. The increase of steric
bulk around the palladium centre therefore seems not to be the
sole reason for the low activity of the catalyst with o-anisyl lig-
ands, since an o-tolyl group is considered to cause similar steric
hindrance around the metal centre compared to an o-anisyl group.
3.4. Homogeneous complexes vs. heterogeneous nanoparticles
When using zerovalent transition metal complexes as catalysts,
the question arises whether the active catalyst is truly a homoge-
neous metal complex or whether catalytic activity is caused by the
formation of heterogeneous nanoparticles or clusters. Especially in
the case of Pd(0) catalysts this should be investigated, since numer-
Table 4
Reaction of nucleophilic substrates with allyl alcohol (2) catalyzed by Pd–diphosphine complexes.a
.
Entry
Reaction
Nucleophilic substrate
Reaction time
Conversion of 7–8 or 11–12 (%)
Selectivity in reaction II (%)
13–14
15–16
1
2
3
I
I
II
II
II
II
1-Octanol
Cyclohexanol
Aniline
Aniline
Aniline
20
20
20
3
3
20
94
11
100
85
16
81
–
–
25
77
100
75
–
–
75
23
0
4
5b
6
n-Octylamine
25
a
Reaction conditions: ratio substrate/allyl alcohol/Pd(OAc)2/dppdmp = 1000/2000/1/4, toluene, 100 ◦C.
Ratio substrate/allyl alcohol/Pd(OAc)2/dppdmp = 1000/500/1/4.
b

100
J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 96–102
in formation of the deuterated allyl products indicates that a metal
␲-allyl species must be present somewhere in the catalytic cycle.
The dissociation of a phosphine moiety is also believed to play
a key role in the mechanism: a strong ligand-structure effect is
observed and especially the use of phosphine ligands that do
not form kinetically stable chelates on Pd(II), in this case dppb
and monodentate triphenylphosphine, results in lower selectiv-
ity for O-allylation. In analogy with the Ru-based catalytic system
previously reported [13,14], it is proposed that (at least partial)
dissociation of one phosphine moiety of the diphosphine is needed
to accommodate the transition state for intramolecular attack of
O–CH moiety of the phenolate by the allyl fragment, and thus
for C-allylation to occur. The proposed catalytic cycle is shown in
Scheme 2.
After formation of the Pd(0) species by means of phosphine oxi-
dation, allyl alcohol coordinates to form species A. It is thought
that despite the use of excess of (di-)phosphine over Pd, the active
organo-Pd species will contain one chelating di-phoshine ligand.
Although bis-(bidentate phosphine) Pd(0) complexes will certainly
exist as resting states, the pseudo zero-order kinetics in ligand con-
centration suggests that in the applied ligand to Pd ratio of 1–4
and large excess of the allylic substrate, one diphosphine is eas-
ily displaced by the allylic substrate, thus reflecting the relatively
high “back donation” binding energy of olefin (relative to that of a
phosphine moiety) to Pd(0) species. Dissociation of the first diphos-
phine ligand is expected to be much less energy demanding than
dissociation of the single remaining diphosphine ligand at Pd.
Oxidative addition takes place to initially form the ␴-allyl Pd(II)
species B, which is in equilibrium with the isomeric ␲-allyl inter-
mediates C and D. Due to exchange of the anion via an acid–base
reaction, the phenolate Pd(II) species E, F and G are formed. The
reductive elimination towards O-allylated products is believed to
take place from intermediate G, in agreement with the Tsuji–Trost
mechanism, in which it is proposed that hard nucleophiles, such as
phenolate, coordinate to the metal centre prior to reductive elimi-
nation [6]. After this step, an intermediate in which the allyl phenyl
ether product is bound to Pd(0), (H) is formed. As indicated, C-
allylated products are most likely formed via a (mono)phosphine
dissociation step (F). Finally, the product is replaced with a
molecule of allyl alcohol to complete the catalytic cycle.
The catalysts with the highest selectivity have ligands with
C3-based bridging groups, being dppp, dppdmp and dppdep, for
which stable chelation is expected in the Pd(II) intermediates E
and G. This is most likely caused by the fact that the natural
bite angle of these ligands is close to 90◦, which is also the opti-
mal angle required for cis-coordination in a square-planar Pd(II)
complex, making the chelate ring free of strain and relatively
stable under the reaction conditions. The introduction of alkyl
substituents at C2 of the C3-bridging group of the ligand is less
important for selectivity, as the unsubstituted dppp ligand already
gives highly selective O-allylation. Note, however that bulkier di-
Et C2-backbone substituents yield a small but measurably higher
selectivity for O-allylation. However, more distinctly, the stabil-
ity of the complex improves by the use of gem-dialkyl substituted
ligands. Deactivation via plating to Pd-black will be related to phos-
phine dissociation of the ligand and thus due to a more stable
chelation, catalyst deactivation is prevented. The ligands which cre-
ate a large P–Pd–P angle (>90◦), such as dppb or two monodentate
PPh3 ligands, result in a catalyst with a relatively low selectivity.
Since the P–Pd–P angle deviates from the preferred 90◦, chelation
of this type of ligands is weaker than that of ligands with C3-based
bridging groups. Intermediate F most likely is lower in energy and
will be more abundant. C-allylation requires sufficient coordina-
tion space on the Pd(II)(allyl) intermediate in order to activate the
ortho-position of the phenolate-anion and in species F this space is
provided.
Scheme 1. Reaction of 1 with deuterated allyl alcohol 12 in the presence of a Pd(PP)
catalyst.
ous reactions have been reported to be catalyzed by heterogeneous
nanoparticles, such as hydrogenation [32], Heck-reactions [33,34]
as well as allylic alkylations [33,34].
The strong effect of the ligand on the activity and selectivity of
the palladium catalysts as described above indicates that a homo-
geneous complex is responsible for catalytic activity, although
ligand-dependent nanoparticle formation cannot be excluded. The
use of heterogeneous Pd(0) on carbon as the catalyst does not
result in conversion of allyl alcohol, giving another indication that
a homogeneous complex is the active catalyst. Finally, when mer-
cury is added to the reaction mixture after 1 h, the catalytic system
remains active, indicating that truly a homogeneous catalyst is
responsible for the observed catalytic activity. The addition of mer-
cury is often used to indicate the presence of active heterogeneous
Pd(0) particles, as it leads to the formation of an amalgam with the
surface of a heterogeneous catalyst, thereby blocking any catalytic
activity [35].
3.5. Mechanistic considerations
The excess of bidentate phosphine ligand (4 equivalents on Pd;
8 equivalents of P on Pd) added to the reaction mixture is neces-
sary to prevent plating of metallic palladium, which leads to loss
and therefore deactivation of the catalyst. One equivalent of biden-
tate phosphine ligand is consumed in the reduction of Pd(II)(OAc)2
to the active Pd(0) species and one bidentate ligand is present on
the Pd centre throughout the catalytic cycle. This means that the
remaining two equivalents of bidentate ligand will assist in keep-
ing the Pd(0) species in the homogeneous phase when allyl alcohol,
diallyl ether or allyl phenyl ether is not coordinated, forming most
likely a tetrakisphosphine palladium(0) compound. Allylation of
the phosphine groups can occur, but it has been reported to be
a reversible process in the presence of Pd(0) catalysts [36,37]. For
the reactions with Pd(dba)2 as catalyst precursor, a Pd(0) species
is already present and consumption of a phosphine ligand for acti-
vation does not take place. The phosphine ligands replace the dba
ligands, and one phosphine ligand then needs to be replaced by
allyl alcohol to form the active Pd(0) catalyst.
Several mechanisms have been proposed for allylation reac-
tions with allyl alcohol as the allylating agent [7–9]. We propose
the formation of an initial Pd(II)(␴-allyl) species immediately after
oxidative addition of allyl alcohol, which rapidly isomerises to
a ␲-allyl species with either phosphine or anion dissociation to
maintain a stable 16 e Pd(II) species, depending on the chelate sta-
bility of the bidentate phosphine ligand. It has been reported that
␴-allyl species are indeed formed in the presence of phosphines
and coordinating anions [37]. In order to investigate if a ␲-allyl
species is present at some point in the catalytic cycle, the reac-
tion of 1 was performed with allyl-1,1-d2 alcohol 17 (Scheme 1).
This resulted in an approximate 1/1 mixture of allyl-1,1-d2 4-
tert-butylphenyl ether and allyl-3,3-d2 4-tert-butylphenyl ether
(Scheme 1; products 18 and 19). The observation of scrambling

J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 96–102
101
Scheme 2. Proposed mechanism for Pd-catalyzed allylation of phenols.
Addition of acid at any time during the reaction immedi-
4. Conclusions
ately inhibits the allylation reaction. It has been reported that
Pd(II)-hydride species are formed from a Pd(0) species and p-
toluenesulfonic acid [38]; this would prevent the catalyst to return
to the Pd(0) state, which is the active species.
Pd-phosphine complexes show good catalytic activity in ally-
lation reactions with allyl alcohol as the allylating agent in the
absence of additives. Complexes with a chelating phosphine lig-
and with C3-based bridging groups show high selectivity towards
O-allylation of phenols. Introduction of gem-dialkyl substituents
on C2 of the ligand backbone results in an increase of the conver-
sion of the reaction, most likely due to an enhanced stability of the
Pd(0) species. The use of diallyl ether as allylating agent results in an
increase in the conversion while maintaining the high selectivity for
O-allylation. Both Pd(II)(OAc)2 as well as Pd(0)(dba)2 can be used
as catalyst precursor in combination with phosphine ligands. Cata-
lysts having phosphine ligands with non-substituted phenyl rings
show higher activity compared to catalysts with phosphine ligands
with ortho-methoxy substituted phenyl rings, possibly due to coor-
dination of the methoxy groups. Apart from 4-tert-butylphenol,
also aliphatic alcohols are efficiently allylated, as well as aromatic
or aliphatic amines.
The nucleophilicity of the alcoholate/phenolate formed plays
an important role in the reactivity of substrates with hydroxyl
groups. This is illustrated by the large increase in conversion
observed when 4-tert-butylphenol is replaced with 1-octanol.
When instead of 4-tert-butylphenol, a more acidic phenol is used,
such as p-nitrophenol or pentafluorophenol, conversion is not
observed due to the low nucleophilicity of the formed phenolate.
A unique feature of the Pd catalysts – distinguishing them from
the ruthenium catalysts that we investigated before – is the ben-
efit that stronger nucleophiles, such as primary and secondary
amines, can also be efficiently allylated. This again can be seen
as a consequence of the strong (back donating) binding energy
of the olefinic moiety of allyl alcohol to Pd(0), relative to that
of the N-coordination of amines. This allows an easy approach
of allyl alcohol to the Pd(0) centre without much competition
by the nitrogen compound. In fact, strong nitrogen coordination
is thought to be the reason why strongly Lewis acidic cationic
Ru catalysts [13,14], are unsuitable for allylation of amines:
strong coordination compounds between Ru and the basic amine
exist.
Acknowledgement
This work was supported by the Technology Foundation STW.
We would like to thank Shell Global Solutions International BV,

102
J.A. van Rijn et al. / Journal of Molecular Catalysis A: Chemical 329 (2010) 96–102
in particular Dr. K. Almeida Lenero and Mr. L. Schoon for supplying
phosphine ligands. Dr. R. Postma (Hexion) and Dr. J.K. Buijink (Shell
Global Solutions International BV) are acknowledged for fruitful
discussions.
[18] R.v. Ginkel, A.v.d. Made, J.d. With, W. Eilenberg, Patent US 6548708B (2003).
[19] E. Drent, Patent US5091587 (1992).
[20] P.W. Clark, B.J. Mulraney, J. Organomet. Chem. 217 (1981) 51–59.
[21] L. Brandsma, H.D. Verkruijsse, Synth. Commun. 20 (1990) 2273–2274.
[22] R.C. van der Drift, M. Vailati, E. Bouwman, E. Drent, J. Mol. Catal. A: Chem. 159
(2000) 163–177.
[23] A. Dahlen, A. Sundgren, M. Lahmann, S. Oscarson, G. Hilmersson, Org. Lett. 5
(2003) 4085–4088.
References
[24] H.M.S. Kumar, S. Anjaneyulu, B.V.S. Reddy, J.S. Yadav, Synlett. (1999)
551–552.
[25] C. Amatore, A. Jutand, M.A. Mbarki, Organometallics 11 (1992) 3009–3013.
[26] C. Amatore, A. Jutand, A. Thuilliez, Organometallics 20 (2001) 3241–3249.
[27] C.L. Lee, Y.P. Yang, S.J. Rettig, B.R. James, D.A. Nelson, M.A. Lilga, Organometallics
5 (1986) 2220–2228.
[28] C.T. Hunt, A.L. Balch, Inorg. Chem. 20 (1981) 2267–2270.
[29] A. Marson, A.B. van Oort, W.P. Mul, Eur. J. Inorg. Chem. (2002) 3028–3031.
[30] W.P. Mul, A.W. van der Made, A.A. Smaardijk, E. Drent, in: A. Sen (Ed.), Catalytic
Synthesis of Alkene–Carbon Monoxide Copolymers and Cooligomers, Kluwer
Academic Publishers, Dordrecht, 2003, pp. 87–140.
[1] A.T. Au, J.L. Nafziger, Patent WO 9620232 (1996).
[2] P.A. Evans, D.K. Leahy, J. Am. Chem. Soc. 122 (2000) 5012–5013.
[3] C. Bruneau, J.L. Renaud, B. Demerseman, Chem. Eur. J. 12 (2006) 5178–5187.
[4] G. Consiglio, R.M. Waymouth, Chem. Rev. 89 (1989) 257–276.
[5] B.M. Trost, M.L. Crawley, Chem. Rev. 103 (2003) 2921–2943.
[6] B.M. Trost, D.L. VanVranken, Chem. Rev. 96 (1996) 395–422.
[7] H. Bricout, J.-F. Carpentier, A. Montreux, J. Mol. Catal. A: Chem. 136 (1998)
243–251.
[8] E. Kuntz, A. Amgoune, C. Lucas, G. Godard, J. Mol. Catal. A: Chem. 244 (2006)
124–138.
[9] T. Satoh, M. Ikeda, M. Miura, M. Nomura, J. Org. Chem. 62 (1997) 4877–4879.
[10] Y. Tada, A. Satake, I. Shimizu, A. Yamamoto, Chem. Lett. (1996) 1021–1022.
[11] Y. Kayaki, T. Koda, T. Ikariya, J. Org. Chem. 69 (2004) 2595–2597.
[12] I.F. Nieves, D. Schott, S. Gruber, P.S. Pregosin, Helv. Chim. Acta 90 (2007)
271–276.
[31] J.S. Sun, C.E. Uzelmeier, D.L. Ward, K.R. Dunbar, Polyhedron 17 (1998)
2049–2063.
[32] A. Molnar, A. Sarkany, M. Varga, J. Mol. Catal. A: Chem. 173 (2001) 185–221.
[33] M. Dieguez, O. Pamies, Y. Mata, E. Teuma, M. Gomez, F. Ribaudo, P. van Leeuwen,
Adv. Synth. Catal. 350 (2008) 2583–2598.
[13] J.A. van Rijn, M. Lutz, L.S. von Chrzanowski, A.L. Spek, E. Bouwman, E. Drent,
Adv. Synth. Catal. 351 (2009) 1637–1647.
[14] J.A. van Rijn, M.A. Siegler, A.L. Spek, E. Bouwman, E. Drent, Organometallics 28
(2009) 7006–7014.
[15] C.S. Kraihanzel, J.M. Ressner, G.M. Gray, Inorg. Chem. 21 (1982) 879–887.
[16] C. Bianchini, H.M. Lee, A. Meli, S. Moneti, F. Vizza, M. Fontani, P. Zanello, Macro-
molecules 32 (1999) 4183–4193.
[34] L.X. Yin, J. Liebscher, Chem. Rev. 107 (2007) 133–173.
[35] J.A. Widegren, R.G. Finke, J. Mol. Catal. A: Chem. 198 (2003) 317–341.
[36] J.M. Basset, D. Bouchu, G. Godard, T. Karame, E. Kuntz, F. Lefebvre, N. Legag-
neux, C. Lucas, D. Michelet, J.B. Tommasino, Organometallics 27 (2008)
4300–4309.
[37] C. Amatore, A. Jutand, M.A. M’Barki, G. Meyer, L. Mottier, Eur. J. Inorg. Chem.
(2001) 873–880.
[17] P.H.M. Budzelaar, J.A. Vandoorn, N. Meijboom, Recl. Trav. Chim. Pays-Bas-J. Roy.
Neth. Chem. Soc. 110 (1991) 420–432.
[38] A. Seayad, S. Jayasree, K. Damodaran, L. Toniolo, R.V. Chaudhari, J. Organomet.
Chem. 601 (2000) 100–107.