Palladium nanoparticles in allylic alkylations and heck reactions: The molecular nature of the catalyst studied in a membrane reactor

Article abstract of DOI:10.1002/adsc.200800424

A series of palladium nanoparticles stabilized by five chiral sugar-based oxazolinyl-phosphite ligands, containing various substituents at the oxazoline and phosphite moieties has been synthesized. They were characterized by transmission electron microsco

Full text of DOI:10.1002/adsc.200800424

FULL PAPERS
DOI: 10.1002/adsc.200800424
Palladium Nanoparticles in Allylic Alkylations and Heck
Reactions: The Molecular Nature of the Catalyst Studied in a
Membrane Reactor
Montserrat Diꢀguez,a,* Oscar Pꢁmies,a Yvette Mata,a Emmanuelle Teuma,b
Montserrat Gꢂmez,b,* Fabrizio Ribaudo,c and Piet W. N. M. van Leeuwenc,d,*
a
Departament de Quꢃmica Fꢃsica i Inorgꢁnica. Universitat Rovira i Virgili. Campus Sescelades, C/Marcel·lꢃ Domingo, s/n.
43007 Tarragona, Spain
Fax : (+34)-97-755-9563; e-mail: montserrat.dieguez@urv.cat
b
Laboratoire Hꢀtꢀrochimie Fondamentale et Appliquꢀe, UMR CNRS 5069. Universitꢀ Paul Sabatier. Bꢄt 2R1, 2iꢅmeꢀtage,
118, route de Narbonne. 31062 Toulouse cedex 9, France
Fax : (+33)-56-155-8204; e-mail: gomez@chimie.ups-tlse.fr
vanꢆt Hoff Institute for Molecular Sciences, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam,
c
The Netherlands
Fax : (+34)-97-792-0224; e-mail: pvanleeuwen@iciq.es
Institute of Chemical Research of Catalonia (ICIQ), Av. Paꢇsos Catalans 16, 43007 Tarragona, Spain
d
Received: July 9, 2008; Revised: September 22, 2008; Published online: November 4, 2008
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/adsc.200800424.
Abstract: A series of palladium nanoparticles stabi- observations, classical poisoning experiments, and ki-
lized by five chiral sugar-based oxazolinyl-phosphite netic measurements have been carried out. Conclu-
ligands, containing various substituents at the oxazo- sive evidence of the nature of the species involved in
line and phosphite moieties has been synthesized. the use of PdNPs in asymmetric catalytic reactions
They were characterized by transmission electron has been obtained. The CFMR experiments proved
microscopy (TEM), X-ray powder diffraction the molecular nature of the true catalysts and all
(XRD), infrared spectroscopy (IR) and elemental conversions can be justified by the amount of molec-
analysis. These nanoparticles were applied in Pd-cat- ular palladium that leached as measured by ICP-
alyzed asymmetric allylic alkylation and Heck cou- AES.
pling reactions. A detailed study to elucidate the
nature of the active species using a continuous-flow Keywords: active species; asymmetric catalysis; C C
membrane reactor (CFMR), accompanied by TEM coupling; nanocatalysts; structure elucidation
À
Introduction
and not unexpectedly this remains the area of homo-
geneous catalysis. Notable examples of MNPs as
The chemistry and catalytic properties of metal col- asymmetric catalysts include: (a) the asymmetric Pt-
loids have attracted increasing interest during the last cinchonidine system for the hydrogenation of ethyl
decade.[1] One important application of metallic nano- pyruvate,[2] (b) the Pd-BINAP chiral system for the
particles (MNPs) may be found in catalysis because hydrosilylation of styrene,[3] and (c) the enantioselec-
of their large surface area and concomitant higher po- tive Pd-diphosphite systems for the allylic alkylation
tential activity and because of their recyclability, both of racemic substrates.[4]
of which represent a breakthrough at the frontier be-
Thus, while the first examples of the use of MNPs
tween homogeneous and heterogeneous catalysis.[1] In as enantioselective catalysts or catalyst precursors
this context, MNPs have proven to be efficient and have been reported, more research needs to be done
selective catalysts for several types of catalytic reac- for which two main directions can be distinguished.
À
tions, such as olefin hydrogenation and C C cou- The first goal, not different from molecular homoge-
pling.[1g] Despite the striking developments of nanoca- neous systems, is the systematic evaluation of the ef-
talysts, their use as enantioselective catalysts is rare fectiveness of new chiral ligands to study the potential
Adv. Synth. Catal. 2008, 350, 2583 – 2598
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2583

Montserrat Diꢀguez et al.
FULL PAPERS
of nanocatalysts in asymmetric catalysis. Such a task
becomes significantly more facile if modular chiral li-
gands are readily at hand. Carbohydrate-based li-
gands are one class that is particularly useful for ad-
dressing this need.[5] The second research target in
asymmetric catalysis using MNPs is to learn whether
or not the MNPs are responsible for the catalytic ac-
tivity, or that molecular species leached from the
MNP are the actual catalyst.[6] In general, the nature
Figure 1. General structure for oxazolinyl-phosphite ligands
used in this paper.
of the active species has been proposed based on cir- series have provided excellent activities and selectivi-
cumstantial evidence such as poisoning of the MNPs ties (regio- and enantioselectivities) for both process-
(i.e., CS2, Hg, etc.), TEM observations, kinetic meas- es in a wide range of substrate types.[10] In MNPs it is
urements, reaction selectivity and, in some cases,[7] re- assumed that the phosphite groups of the ligands co-
activity patterns.[4,6a,h]
ordinate firmly to the surface atoms of the nanoparti-
The possible involvement of dynamic equilibria be- cles and, in addition, the oxygen atoms of the sugar
tween solid palladium (or PdNPs) and dissolved mo- moieties are able to interact weakly with this sur-
lecular palladium species in palladium-catalyzed cou- face.[4] We also present a detailed study to provide
pling reactions has been the subject of much debate. conclusive evidence of the nature of the species in-
De Vries studied his ligand-free catalyst for the Heck volved in the catalytic reaction in liquid phase by in-
reaction extensively using ES-MS to establish the troducing the use of a continuous-flow membrane re-
nature of the catalyst.[8] In addition he reviewed a actor (CFMR), combined with TEM observations,
large number of publications on this matter and con- classical poisoning and kinetic measurement experi-
cluded that at least for reactions conducted at high ments. The use of the continuous-flow membrane re-
temperatures (>1308C) such equilibria are invol- actor (CFMR) has been crucial to elucidate the true
ved.[6b] Jones and co-workers scrutinized in their nature of the catalysts. The CFMR system has a cut-
review over 500 reports on palladium precursors used off molecular weight of 700 Dalton and thus it will
in Heck and Suzuki coupling reactions and their con- retain the MNP species, but the mononuclear metal
clusions agree with those of the De Vries as regards species will pass the membrane.[11] The CFMR system
the solid/colloid/nanoparticle precursors, which leaves has been used previously for the study of catalyst
the situation undecided for the majority of reported leaching from dendrimeric catalysts as it retains the
protocols.[6a]
dendrimeric species.[12] Especially allylic palladium
Therefore, it is important to develop a practical ex- substitution reactions were found to be prone to
periment that reveals definitively the nature of the metal leaching. The CFMR offers a more accurate
true catalyst in the liquid phase. Recently, Rothen- yardstick to establish the involvement of molecular
berg and co-workers have developed a two-cell mem- species than the two-compartment reactor mentioned
brane batch reactor to probe if molecular palladium above. Furthermore, the reactions in the present
species generated by leaching or palladium nanoparti- study were carried out at a low temperature (T=
cles are the true catalysts in the ligand-free Heck and 238C, see below).
Suzuki reactions; indications are that metal leaching
in the ligand-free system is responsible for the catalyt-
ic activity.[9] In the experiment with the two-compart- Results and Discussion
ment reactor of Rothenberg et al. nanocolloid parti-
cles of 15 nm were used and the reactor contains an Synthesis and Characterization of Pd-Nanoparticles
alumina membrane having a pore diameter of 5 nm. Stabilized by Chiral Oxazolinyl-Phosphites
No enforced flow was applied and movement of the
particles from one compartment to the other relied Pd1–Pd5 nanoparticles were prepared from
on diffusion; experiments took several days at 1008C.
ACHTUGTNRENG[NU Pd2ACHTUNGTREN(NUNG dba)3] under hydrogen pressure (3 bar) in the
In view of the size of the nanopores of 5 nm it would presence of the appropriate chiral ligand 1–5 (ligand-
seem that palladium nanoparticles as used in this to-palladium ratio 0.2) at room temperature in THF,
study (2.5 nm) will also pass the alumina membrane, based on the methodology previously described
albeit slowly.
(Scheme 1).[13] They were isolated by precipitation,
This paper describes the use of sugar-based oxazo- washed with dry pentane and dried under reduced
linyl-phosphite ligands (Figure 1) as a new type of sta- pressure. They were found to be stable, both in solu-
bilizers for PdNPs in order to facilitate the catalyst re- tion and in the solid state. Pd1–Pd5 were analyzed by
covery and recycling in the enantioselective Pd-cata- transmission electron microscopy (TEM), X-ray
lyzed allylic substitution and Heck coupling reactions. powder diffraction (XRD), infrared spectroscopy
Molecular catalytic precursors containing this ligand (IR) and elemental analysis.
2584
asc.wiley-vch.de
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 2583 – 2598

Palladium Nanoparticles in Allylic Alkylations and Heck Reactions
FULL PAPERS
Scheme 1. Synthesis of Pd1–Pd5 nanoparticles.
The size and morphology of PdNPs were investigat- posed by Moiseev and co-workers. [The diameter of
ed by TEM analysis (Figure 2). These particles dis- the MNPs together with the packing structure deter-
played a small spherical shape showing a tendency to mines the number of layers and the amount of metal
give inter-particle organization and in some cases ag- atoms in the cluster. So for instance, a 2.4 nm PdNP
glomeration regions were observed. The nanoparticle with fcc packing structure corresponds to a nanoclus-
size distribution was estimated from the measurement ter that contains 561 Pd atoms in five layers, while a
of about 300 particles, assuming a spherical shape, 3.15 nm PdNP corresponds to a seven-shell cluster
found in an arbitrary chosen area in enlarged micro- that contains 1515 Pd atoms. See for instance ref.[1b]
]
photographs (Figure 3). The mean size found for all This full-shell nanocluster contains ~561 Pd atoms in
PdNPs prepared was ca. 2.5 nm with a narrow size five layers with an icosahedron arrangement and a
distribution (Table 1). The concentration of the mean diameter of 2.4 nm.[15,16] Therefore, the outer
medium reaction had no effect on the particles size or layer contains ~252 palladium atoms surrounded by
dispersion (dilution range from 5ꢉ10À4 to 2.5ꢉ10À3 60 ligands coordinated in a bidentate fashion on the
molꢉlÀ1, keeping the ligand-to-palladium ratio con- palladium atoms located at the vertices (12) and the
stant).
edges (~48 only, for steric reasons out of a total of
In all cases, the X-ray powder diffraction (XRD) 120) of the icosahedron, in a stacked fashion.[17] The
showed the face-centered cubic (fcc) packing mode of present oxazolinyl-phosphite ligand may also coordi-
palladium atoms within the nanoparticle metal core nate to atoms in edge and vertex positions in a bi-
(see Supporting Information). The diameter of the dentate manner. The oxazolinyl-phosphite ligand is
different nanoparticles could also be estimated by bulkier than phenanthroline, yet our PdNPs show a
means of the Debye–Scherrer equation (using the pa- slightly higher ligand-to-palladium ratio than the
rameters obtained with Rietveldꢆs refinements).[14] Moiseev cluster (75 vs. 60 bidentate ligands per clus-
The calculated diameters were between 2.2 and ter of the same size).[18] The reason for this higher
2.5 nm, in good agreement with TEM observations ligand-to-palladium ratio is that oxazoline coordi-
(Table 1). Palladium oxide was not observed in any nates weakly to zerovalent palladium. Therefore,
case.
some or all oxazoline-phosphite ligands may be coor-
IR spectra of PdNPs Pd1–Pd5 showed bands at dinated as monodentate ligands to palladium atoms
1594–1649 cmÀ1 and 1019–1029 cmÀ1 corresponding to located at the faces, the edges and even the vertices
À
À
C N and P O stretching bands, which are shifted of the icosahedron with their phosphorus donor
compared to the free ligands (Table 1). This confirms groups.
the presence of the oxazolinyl-phosphite ligands at
the surface of the particles (see Supporting Informa-
À
tion for spectra).
Application of PdNPs to Asymmetric C C Bond
The elemental analyses, C, H, N, P and Pd, of the Formation Reactions
PdNPs prepared (Pd1–Pd5) are shown in Table 2.
These analyses indicated a similar Pd/P ratio of ca. In this section, we present the application of these
15/2 for all the nanoparticles. These elemental analy- new PdNPs as catalysts in the asymmetric allylic alky-
sis together with the diameter data (ca. 2.4 nm, lation and Heck coupling reactions. The interest of
À
Table 1) and packing structure (fcc) of the nanoparti- these asymmetric C C bond formation processes is
cle core are in agreement with a structure similar to found in their usefulness in organic synthesis allowing
that of the [Pd561ACHTUNGTRNEN(UG OAc)180ACHUTNTGREN(NUGN Phen)60] giant cluster pro- the formation of enantioselective carbon-carbon and
Adv. Synth. Catal. 2008, 350, 2583 – 2598
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2585

Montserrat Diꢀguez et al.
FULL PAPERS
Figure 2. TEM micrographs of PdNPs synthesized in THF from [Pd
Pd1; (b) Pd2; (c) Pd3; (d) Pd4 and (e) Pd5.
2ACHTUNGTRENUN(NG dba)3] and in the presence of 0.2 mol% of ligand: (a)
carbon-heteroatom bonds under mild reaction condi-
tions.[19]
Asymmetric Pd-Catalyzed Allylic Alkylation of
rac-3-Acetoxy-1,3-diphenyl-1-propene
of ligands carrying different donor groups.[10a] Here,
we used as starting reaction conditions those that pro-
Pd1–Pd5 nanoparticles were used as catalytic precur- vided the best results with palladium molecular cata-
sors in the Pd-catalyzed asymmetric allylic alkylation lysts (i.e., dichloromethane as solvent and room tem-
of rac-3-acetoxy-1,3-diphenyl-1-propene S1, using di- perature).[10a] The results are summarized in Table 3.
methyl malonate under basic conditions [Eq. (1)].[20]
In a first set of experiments, the effect on catalysis
In a previous work, we studied this catalytic reaction of the added ligand was studied using Pd1 nanoparti-
using the corresponding molecular catalytic systems cles (Table 3, entries 1–4). The results indicated that
and reaction conditions were optimized for a variety without the addition of free ligand, the catalytic
2586
asc.wiley-vch.de
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 2583 – 2598

Palladium Nanoparticles in Allylic Alkylations and Heck Reactions
FULL PAPERS
Figure 3. Size distributions obtained from TEM micrographs of PdNPs: (a) Pd1; (b) Pd2; (c) Pd3; (d) Pd4 and (e) Pd5.
Table 1. Diameter and IR data for Pd1-Pd5 nanoparticles and IR data for ligands 1-5.
Pd1–Pd5 nanoparticles
(XRD) (nm)
Free ligand
IR (cmÀ1)[a]
Ligand
d
(TEM) (nm)
dR
IR (cmÀ1)[a]
1649 (C=N), 1025 (P O)
À
À
1
2
3
4
5
2.6Æ0.4
2.6Æ0.3
2.4Æ0.3
2.5Æ0.3
2.6Æ0.3
2.55Æ0.04
2.41Æ0.03
2.41Æ0.03
2.35Æ0.05
2.21Æ0.05
1650 (C=N), 1015 (P O)
1651 (C=N), 1014 (P O)
1650 (C=N), 1018 (P O)
1666 (C=N), 1019 (P O)
À
À
1594 (C=N), 1019 (P O)
À
À
1636 (C=N), 1021 (P O)
À
À
1630 (C=N), 1029 (P O)
À
À
1647 (C=N), 1027 (P O)
1651 (C=N), 1017 (P O)
IR spectra recorded in 4000–400 cmÀ1 range as KBr pellets. Only the two more representative absorptions are indicated.
[a]
system is practically inactive (Table 3, entry 1), while tioselectivity (Table 3, entries 2 and 3 vs. 1), as ob-
the addition of small amounts (5–10%) of ligand in- served for other related PdNP catalytic systems.[4]
creases the reaction rate, having little effect on enan- This effect may be due to the agglomeration produced
Adv. Synth. Catal. 2008, 350, 2583 – 2598
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2587

Montserrat Diꢀguez et al.
Idealized formula
FULL PAPERS
Table 2. Elemental analysis and calculated formula for Pd1–Pd5.
Nanoparticle
% C
% H
% N
% P
% Pd
Formula
Pd1
Pd2
Pd3
Pd4
Pd5
36.92
36.74
36.74
38.71
37.45
4.09
3.74
4.89
4.75
4.32
1.02
0.70
0.96
0.78
0.84
2.09
1.69
1.85
1.84
1.85
43.55
47.60
46.30
45.20
45.42
Pd32(L)
Pd32(L)
Pd32(L)
Pd32(L)
Pd32(L)
U
Pd561(L)
Pd561(L)
Pd561(L)
Pd561(L)
Pd561(L)
88A
68A(THF)140
75A(THF)173
77A(THF)173
75A(THF)149
E
E
N
G
Table 3. Asymmetric Pd-catalyzed allylic alkylation of S1 using nanoparticle and molecular catalytic systems.
Nanoparticle precursor[a]
% Conv.[c] (h)
Molecular precursor[b]
% Conv.[c] (h)
Entry
L
S1/PdL/L
% ee 6[d]
S1/Pd/L
% ee 6[d]
1
2
3
4
5
6
7
8
1
1
1
1
2
3
4
5
20/1/–
<5 (72)
33 (72)
60 (72)
62 (72)
51 (72)
32 (72)
78 (72)
100 (24)
–
100/1/0.9
–
–
95 (0.5)
–
–
86 (S)
–
–
20/1/0.05
20/1/0.1
20/1/0.2
20/1/0.1
20/1/0.1
20/1/0.1
20/1/0.1
70 (S)
71 (S)
70 (S)
80 (S)
60 (S)
46 (S)
20 (S)
–
–
–
100/1/0.9
100/1/0.9
100/1/0.9
100/1/0.9
100 (0.5)
100 (0.25)
92 (0.5)
67 (0.5)
92 (S)
85 (S)
86 (S)
45 (S)
[a]
5 mol% Pd, 0.5 mmol of S1, 1.5 mmol of BSA, 1.5 mmol of dimethyl malonate, CH2Cl2 as solvent, T=238C.
0.5 mol% [Pd
[b]
((h3-C3H5)]2, 0.9 mol% of ligand, 0.5 mmol of S1, 1.5 mmol of BSA, 1.5 mmol of dimethyl malonate,
ACHTUNGTRENNUNG(m-Cl)ACHTUGNTRENNUGN
CH2Cl2 as solvent, T=238C.
Conversion percentage determined by H NMR.
Enantiomeric excesses determined by HPLC on a Chiralcel-OD column. Absolute configuration shown in parentheses.
[c]
[d]
1
under catalytic conditions (corroborated by TEM which contains tert-butyl groups at the ortho and para
analysis after catalysis, see Figure 5b) by ligand de- positions of the biphenyl phosphite moiety, provided
sorption from the metallic surface when not enough better enantioselectivity and slightly lower activities
ligand is present in the medium, which leads to a cata- than Pd1, containing trimethylsilyl ortho-substituted
lytically inactive heterogeneous species. When free biphenyl phosphite moiety. The same effect was ob-
ligand is added (see Section: Elucidation of the served for the analogous molecular systems. In sum-
Active Species) this leads to the formation of molecu- mary, good enantioselectivities (ees up to 80%) with
lar, catalytically active palladium species by metal low activities were obtained with Pd1 and Pd2 nano-
leaching. It should be noted that the addition of particle catalytic systems. If we compare the results
larger amounts of ligand (more than 10 mol%) did obtained with the application of the nanoparticles
not increase the reaction rates further (Table 3, with the ones obtained with the molecular systems,
entry 4 vs. 3). Therefore, addition of 10 mol% of we observed that the nanoparticles are less active and
ligand was applied for the subsequent reactions.
less enantioselective than their corresponding homo-
We then tested the remaining PdNPs, Pd2–Pd5. We geneous catalysts (Table 3).
found an effect of the size of the oxazoline substituent
For the best PdNPs with regard to asymmetric in-
on the catalytic performance (activity and enantiose- duction, Pd1 and Pd2, we next studied catalyst recy-
lectivity). Therefore, activities were higher when cling (Figure 4).[21] Interestingly, the nanoparticles
more sterically demanding substituents were present could be reused three times maintaining the enantio-
(Table 3, entries 3, 5–8), the most active being system selectivity, but with decreases in activity.[22] This obser-
Pd5 which contains a tert-butyloxazoline substituent vation may point to metal leaching from the nanopar-
(entry 8, Table 3). However, enantioselectivities de- ticles (see Section: Elucidation of the Nature of the
crease when steric hindrance increases (Table 3, en- Active Species) or to blocking of the solid catalyst by
tries 3, 5–8). Both catalytic findings contrast with the strongly adsorbing species.
results of the analogous molecular system (Table 3).
This fact can point to a different nature of catalytic
species or a different extent of metal leaching (see Asymmetric Pd-Catalyzed Phenylation of
Section: Elucidation of the Nature of the Active Spe- 2,3-Dihydrofuran
cies). Regarding the biphenyl phosphite moiety, we
found a slight but distinct effect on activity and enan- Pd1–Pd5 were also used as catalyst precursors in the
tioselectivity. Therefore, Pd2, stabilized by ligand 2 asymmetric Heck reaction of 2,3-dihydrofuran S2 and
2588
asc.wiley-vch.de
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 2583 – 2598

Palladium Nanoparticles in Allylic Alkylations and Heck Reactions
FULL PAPERS
phenyl triflate [Eq. (2)] as a model reaction. For this
process, not only does the enantioselectivity of the
process need to be controlled but regioselectivity is
also a problem because a mixture of regioisomers can
be obtained. The reactions were carried out by adding
10 mol% of the corresponding ligand to the nanopar-
ticle suspension under the reaction conditions that
provided the best results with the molecular system
[i.e., THF as solvent, 508C and (i-Pr)2NEt as
base].[10b,c] The results are summarized in Table 4.
Pd1–Pd5 systems show low activity (conversions up
to 23% after 72 h of reaction), but they provided
good regio- (up to 82%) and enantioselectivities (up
to 68%). We also found an effect of the size of the ox-
azoline substituent and the substituents at the biphen-
yl phosphite moiety in the catalytic performance (ac-
tivity and selectivity); when the size of the group on
the oxazoline decreased, the regio- and enantioselec-
tivity of the catalyst increased (Table 4, entries 1–5).
Moreover, the presence of trimethylsilyl ortho-substi-
tuted biphenyl phosphite moiety has a positive effect
on activity and enantioselectivity (Table 4, entries 1
vs. 2). In contrast to that observed in the allylic alky-
lation, these results follow the same trend as that ob-
served for the comparable molecular systems. The
best results were obtained with Pd1 and Pd3 nanopar-
ticle catalysts. Pd3 was reused once without signifi-
cant loss in activity and selectivity (Table 4, entries 3
and 6). As observed for the allylic alkylation of S1,
Figure 4. Catalyst recycling for the asymmetric allylic alkyla-
tion of S1 using: (a) Pd1 and (b) Pd2.
Figure 5. TEM micrographs showing Pd5 nanoparticles: (a) before catalysis, (b) after the allylic alkylation of S1 and (c)
after allylic alkylation of S1 and washed five times with THF.
Adv. Synth. Catal. 2008, 350, 2583 – 2598
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2589

Montserrat Diꢀguez et al.
FULL PAPERS
Table 4. Asymmetric Pd-catalyzed phenylation of S2 using nanoparticle and molecular catalytic systems.
PdNP precursor[a]
Molecular precursor[b]
Entry
Ligand
% Conv.[c] (h)
7/8[c]
% ee 7[c]
% Conv.[c] (h)
7/8[c]
% ee 7[c]
1
2
3
4
1
2
3
4
5
3
23 (72)
17 (72)
19 (72)
15 (72)
8 (72)
81/19
76/24
82/18
64/36
52/48
82/18
65 (R)
63 (R)
68 (R)
54 (R)
55 (R)
68 (R)
100 (15)
98 (24)
94 (24)
80 (24)
12 (24)
–
97/3
87/13
97/3
71/29
65/35
–
99 (R)
97 (R)
99 (R)
84 (R)
83 (R)
–
5
6[d]
15 (72)
[a]
10 mol% Pd, 2 mmol of S2, 0.5 mmol of PhOTf, 1 mmol of (i-Pr)2NEt, THF as solvent, T=508C, 10 mol% of ligand
added.
[b]
[c]
[d]
1.25 mol% of [Pd2ACHTNUGTRENUNG(dba)3], 2 mmol of S2, 0.5 mmol of PhOTf, THF as solvent, 1 mmol of (i-Pr)2NEt, T=508C.
Conversion percentages based on PhOTf and enantiomeric excesses measured by GC.
Recycling experiment.
again the nanoparticle catalytic systems provided ty. However, metal leaching cannot be ruled out since
lower activities and lower enantioselectivities than the Pd(0) molecular species tend to rejoin to the clusters
corresponding molecular systems (Table 4).
after catalysis as has been shown for the ligand-free
Pd-catalyzed Heck reaction at high temperatures.[6a,b]
Kinetic measurements: For the homogeneous mo-
lecular systems containing ligands 1 and 5, the catalyst
concentration was decreased in order to obtain reac-
tion rates comparable with those of the nanoparticle
ones. We therefore studied the molecular system at
Elucidation of the Nature of the Active Species
Allylic Alkylation
In order to establish the nature of the catalyst, several different palladium/substrate ratios: 1/500, 1/1,000, 1/
experiments were carried out using the molecular and 5,000 and 1/10,000. For the catalytic systems contain-
nanoparticle catalytic systems. For these experiments ing ligand 1, at Pd/S1=1/1,000 and Pd/S1=1/5,000,
we chose Pd1 and Pd5 nanoparticles since they show the molecular system (Mol.1) was much faster than
very different catalytic activity (see Table 3). There- the nanoparticle one, while the system Pd/S1=1/
fore, while Pd1 showed low activity (62% conversion 10,000 gave a similar order of magnitude of rate as
after 72 h), Pd5 was the fastest (100% conversion the Pd1 one (Figure 6a). However, for the catalytic
after 24 h).
system containing ligand 5, the Pd/S1=1/500 provided
TEM and elemental analysis after catalysis: In a rate of similar order of magnitude as Pd5 (Fig-
general, the shape and size distribution of PdNPs is ure 6b). In addition, the kinetic study showed that the
no different from that of the samples before catalysis reaction rate profiles for Pd1 and Pd5 and the corre-
(Figure 5). This contrasts with the decrease in particle sponding molecular systems, at Pd/S1=1/10,000 and
size during the reaction when Pd atoms and/or ions Pd/S1=1/500, respectively, are similar. Unlike the ki-
leach from the colloidal clusters as observed by Roth- netic resolution observed for other PdNP systems sta-
enberg and co-workers.[6d] Although the TEM graphs bilized by chiral non-labile diphosphites,[4] in the pres-
of the unwashed PdNPs after catalysis have more ag- ent ligand systems substrate S1 remains almost race-
glomeration regions (Figure 5b) than the PdNPs mic for both molecular and nanoparticle catalytic sys-
before catalysis (Figure 5a), the washed recovered tems. This kinetic resolution of substrate S1 has been
PdNPs had the same degree of agglomerated regions one of the main clues to confirm that the PdNPs sta-
as before catalysis (Figure 5c).[7b] The higher agglom- bilized by chiral diphosphites are the species truly re-
eration degree observed in Figure 5b is probably due sponsible for the catalytic activity. The catalytic be-
to the presence of large amounts of organic material haviour of the oxazolinyl-phosphite ligand systems
(products and starting compounds) and the ligand may therefore suggest that molecular species leached
leaching from the PdNPs, because of the weak inter- from the nanoparticles used as catalyst precursors.
actions with the metal surface. Moreover, the P and However, it should be pointed out that for the diphos-
Pd elemental analysis of the recovered nanoparticles phite-systems, no control experiments have been done
after catalysis indicates that the Pd/P ratio is the same for the molecular catalysts under the same conditions
as before the catalysis (Pd/P ratio of ca. 15/2).[23] This as in the PdNP catalysis (a huge excess of diphosphite
together with no modification in the shape and size and an absence of chloride). Thus, the nature of the
distribution of PdNPs during the reaction may suggest active species with our PdNPs remains uncertain.
that the PdNPs are responsible for the catalytic activi-
2590
asc.wiley-vch.de
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 2583 – 2598

Palladium Nanoparticles in Allylic Alkylations and Heck Reactions
FULL PAPERS
Figure 6. Kinetic profiles for the Pd-catalyzed allylic substitution of S1: (a) using Pd1 and molecular system (Mol.1) at S1/
Pd=10,000 and (b) using Pd5 and molecular system (Mol.5) at S1/Pd=500.
Poisoning tests: In order to provide further evi- to indicate the presence of Pd(0) species but they do
dence about the nature of the catalyst, we studied the not provide conclusive evidence about the nature of
effect of classical poisons (CS2 and Hg) for both cata- the active species (nanoparticle or molecular).[6a]
lysts, molecular and PdNPs. The results are summar-
Continuous-flow membrane reactor experiments:
ized in Table 5 and Table 6. As expected, for the mo- For these experiments we used a home-made continu-
lecular system (Mol.1 and Mol.5) the presence of mer- ous-flow membrane reactor (see Experimental Sec-
cury does not modify the catalytic behaviour, but tion for a more detailed description and set-up of the
carbon sulfide decreases the activity of the system. reactor) equipped with a stirring bar and a membrane
However, the fact that the catalytic reaction stops for placed in the bottom of the reactor (total volume=
the Pd1 and Pd5 systems in the presence of either 6 mL). The reactive solution is pumped through the
CS2 or Hg (added after 4 h or 24 h of reaction), seems reactor with a flow rate of 6 mL/h (Figure 7). We
Table 5. Poisoning tests for the asymmetric Pd-catalyzed allylic alkylation of S1 using Pd1 and Mol.1.
Nanoparticle precursor Pd1[a]
S1/Pd1/1/poison
Molecular precursor Mol.1[b]
S1/Pd/1/poison
Entry
Poison
% Conv.[c] (h)
% ee 6[d]
% Conv.[c] (h)
% ee 6[d]
1
2
3
4
5
–
20/1/0.1/–
60 (72)
<5 (72)
24 (72)
<5 (72)
45 (72)
71 (S)
–
70 (S)
–
100/1/0.9/–
100/1/0.9/1
100/1/0.9/1
100/1/0.9/100
100/1/0.9/100
95 (0.5)
12 (0.5)
43 (0.5)
90 (0.5)
93 (0.5)
86 (S)
83 (S)
85 (S)
85 (S)
86 (S)
[e]
CS2[f]
20/1/0.1/0.3
20/1/0.1/0.3
20/1/0.1/100
20/1/0.2/100
CS2
Hg[e]
Hg[f]
70 (S)
[a]
5 mol% Pd, 0.5 mmol of S1, 1.5 mmol of BSA, 1.5 mmol of dimethyl malonate, CH2Cl2 as solvent, T=238C.
0.5 mol% [PdCl
[b]
ACHTUNGTRENNUNG
((h3-C3H5)]2, 0.9 mol% of 1, 0.5 mmol of S1, 1.5 mmol of BSA, 1.5 mmol of dimethyl malonate, CH2Cl2 as
solvent, T=238C.
Conversion percentage determined by H NMR.
Enantiomeric excesses determined by HPLC on a Chiralcel-OD column. Absolute configuration shown in parentheses.
Poison added at the beginning of the reaction.
Poison added after 24 h reaction for the colloidal system and after 10 min for the molecular system
[c]
[d]
[e]
[f]
1
Adv. Synth. Catal. 2008, 350, 2583 – 2598
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2591

Montserrat Diꢀguez et al.
FULL PAPERS
Table 6. Poisoning tests for the asymmetric Pd-catalyzed allylic alkylation of S1 using Pd5 and Mol.5.
Nanoparticle precursor Pd5[a]
S1/Pd5/5/poison
Molecular precursor Mol.5[b]
S1/Pd/5/poison
Entry
Poison
–
% Conv.[c] (h)
% ee 6[d]
% Conv.[c] (h)
% ee 6[d]
1
2
3
4
5
20/1/0.1/–
100 (24)
<5 (24)
27 (24)
<5 (24)
57 (24)
20 (S)
–
21 (S)
–
100/1/0.9/–
100/1/0.9/1
100/1/0.9/1
100/1/0.9/100
100/1/0.9/100
67 (0.5)
<5 (0.5)
18 (0.5)
59 (0.5)
64 (0.5)
45 (S)
42 (S)
43 (S)
44 (S)
45 (S)
[e]
CS2[f]
20/1/0.1/0.3
20/1/0.1/0.3
201/0.1/100
20/1/0.2/100
CS2
Hg[e]
Hg[f]
20 (S)
[a]
5 mol% Pd. 0.5 mmol of S1, 1.5 mmol of BSA, 1.5 mmol of dimethyl malonate, CH2Cl2 as solvent, T=238C.
0.5 mol% [PdCl
[b]
ACHTUNGTRENNUNG
((h3-C3H5)]2. 0.9 mol% of 5. 0.5 mmol of S1. 1.5 mmol of BSA. 1.5 mmol of dimethyl malonate. CH2Cl2 as
solvent. T=238C.
Conversion percentage determined by H NMR.
Enantiomeric excesses determined by HPLC on a Chiralcel-OD column. Absolute configuration shown in parentheses.
Poison added at the beginning of the reaction.
Poison added after 4 h reaction for the colloidal system and after 10 min for the molecular system
[c]
[d]
[e]
[f]
1
used a Koch/SelRO MPF-50 nanofiltration membrane ysis of the solution collected during the first two reac-
(molecular weight cutoff=700 Dalton) which is de- tor volumes pumped through the reactor shows a con-
signed to allow the diffusion of leached molecular version of 10% with a 19% enantioselectivity. We ob-
species, but not of Pd nanoparticles.
served that the catalytic activity stopped after six re-
Firstly, we studied the alkylation of S1 using Pd5. actor volumes (Figure 8). To the solution from the
The membrane, the stirring bar and KOAc were first two reactor volumes we added 2 mg of KOAc
placed at the base of the reactor. They were first and allowed this sample to stir for a further 4 h. We
flushed with dry dichloromethane overnight and then observed that conversion increased until 32% while
with a solution containing the substrate S1, dimethyl maintaining the same enantioselectivity. This is in
malonate and BSA during 2 h (approx. two reactor agreement with the fact that the active species are
volumes). The reaction was started by transferring the molecular complexes that arise from leaching from
catalyst solution (5 mg of Pd5 and 20 mol% of 5 in the nanoparticle. After 8 reactor volumes pumped
2 mL of dichloromethane) into the membrane reactor. through, we added to the reactor 10 mol% of ligand
We collected and analyzed the total volume of the so- in 1 mL of dichloromethane and we observed that the
lution pumped through the reactor every two reactor system was again active (6% conversion after 2 reac-
volumes. The data are collected in Figure 8. The anal- tor volumes, shown also in Figure 8). This clearly indi-
cates that the presence of ligand is necessary to
induce the leaching from the PdNPs and therefore ex-
clude the possibility of “naked” molecular Pd(0) spe-
cies being formed. We also quantified the amount of
Pd leached in the solutions coming out of the reactor
using ICP-AES. After the first two reactor volumes,
0.6 mg of Pd was found in solution. The molecular
system using 0.6 mg of Pd precursor provides a con-
version (15% after 2 h, Figure 6b) comparable to that
observed for Pd5 nanoparticle catalysts after two re-
actor volumes. Therefore, we conclude that the mo-
lecular species are the only ones responsible for the
catalytic activity. In addition, the lower enantioselec-
tivity achieved using PdNPs compared with those of
the molecular system (at L/Pd=0.9, Table 3) is due to
the formation of monomeric molecular Pd-allyl spe-
cies containing two oxazolinyl-phosphite ligands coor-
dinated in a monodentate fashion through the phos-
(k1-P-L)2]+ (L=1–5). These
phite moiety, [PdACTHNUTRGENN(GU allyl)ACHUTNGTRENNUGN
species have been observed in the molecular system
when excess of ligand is present as is the case in the
catalysis using PdNP where the resulting ligand-to-
metal ratio is high.[24,25]
Figure 7. Schematic presentation of a membrane reactor.
2592
asc.wiley-vch.de
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 2583 – 2598

Palladium Nanoparticles in Allylic Alkylations and Heck Reactions
FULL PAPERS
&
^
Figure 8. Conversion ( ) and enantioselectivity ( ) obtained from the use of Pd5 in a continuous-flow membrane reactor.
We then studied the alkylation of S1 using Pd1. molecular catalyst.[4] The system differs from the pres-
Since the activity of Pd1 is lower than that of Pd5, we ent one in that the particles were much larger
performed the experiment in the membrane reactor (4.2 nm, or ~Pd3000) and contained only ~11 diphos-
but in a different way. After the conditioning of the phites per cluster (idealized as Pd2869, 9 shells icosahe-
membrane as previously described, a solution contain- dron, carrying the bulky diphosphites perhaps at the
ing all the reactive compounds (S1, dimethyl malo- vertices). The enormous excess of ligand in solution
nate and BSA), Pd1 and 20 mol% of ligand 1 were can perhaps account for the kinetic resolution ob-
placed in the reactor. The reaction mixture was al- served.
lowed to stir in a batch mode for 2.5 days. After this
time, we pumped out the solution of the reactor
through the membrane. Analysis of the resulting solu- The Asymmetric Heck Reaction
tion indicated a 63% conversion with an ee of 74%.
To this sample, we then added 2 mg of KOAc. After In contrast to the allylic alkylation reaction, the TEM
24 h, we observed that conversion increased until analysis after the reaction shows larger particles than
85% maintaining the same enantioselectivity. This those found in the TEM analysis before the reaction
clearly shows again that the active species are molecu- (Figure 9). This indicates that the PdNPs are not
lar species that originate from metal leaching from stable under the reaction conditions and tend to form
the nanoparticle. The amount of Pd in the filtered larger nanoparticles, probably by ligand loss.[26] In ad-
sample is 0.06 mg and this fully accounts for the activ- dition, also the poisoning experiments (Table 7) pro-
ity if we assume that all the leached Pd atoms form vide conclusive evidence about the nature of the spe-
molecular palladium/ligand species.
cies responsible for the catalytic activity. They agree
In summary, the continuous-flow membrane reactor with a leaching of molecular species being responsible
is crucial in the elucidation of the nature of the active for the catalytic activity. In summary, these experi-
species. It proves that the leached molecular palladi- ments in combination with the fact that the nanoparti-
um species containing oxazolinyl-phosphite ligands cle system shows the same catalytic trends as the mo-
are the true catalysts in the Pd-nanocatalyzed allylic lecular system (Table 4) indicates that our Pd nano-
alkylation reaction. In addition, this study indicates particles act as a reservoir of molecular species, analo-
that the behaviour of the Pd nanoparticles depends gous to that proposed for Heck free-ligand coupling
on their stability; Pd1 is more resistant to leaching of using Pd nanoparticles as catalysts.[6b,9,27] The fact that
molecular species than Pd5 and therefore the latter is lower enantioselectivities were achieved using Pd
more active.
nanoparticles compared with those of the homogene-
The previous study on palladium nanoparticles in ous system (Table 4) is probably due to the presence
allylic allylation using diphosphites as the ligands de- of an excess of ligand when using Pd nanoparticles as
viated from the molecular catalysts in that kinetic res- has been observed for the allylic alkylation reac-
olution was obtained, which was not observed for the tion.[24]
Adv. Synth. Catal. 2008, 350, 2583 – 2598
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2593

Montserrat Diꢀguez et al.
FULL PAPERS
Figure 9. TEM micrographs showing Pd1 nanoparticles: (a) before and (b) after Pd-catalyzed Heck reaction of S2.
Table 7. Poisoning test for the asymmetric Pd-catalyzed phenylation of S2 using Pd1 and Mol.1.
Nanoparticle precursor Pd1[a,c]
Molecular precursor Mol.1[b,c]
Entry
Poison
–
S2/Pd1/1/poison
% Conv. (h)
% 7
% ee 7
S2/Pd/1/poison
% Conv. (h)
% 7
% ee 7
1
2
3
4
5
10/1/0.1/–
23 (72)
<3 (72)
7 (72)
<3 (72)
18 (72)
81
–
79
–
65 (R)
–
64 (R)
–
40/1/2.2/–
100 (15)
21 (24)
34 (24)
87 (24)
100 (24)
97
97
95
95
96
99 (R)
96 (R)
96 (R)
94 (R)
97 (R)
[d]
CS2[e]
10/1/0.1/0.3
10/1/0.1/0.3
10/1/0.1/100
10/1/0.1/100
100/1/0.9/1
100/1/0.9/1
100/1/0.9/100
100/1/0.9/100
CS2
Hg[d]
Hg[e]
80
64 (R)
[a]
10 mol% Pd, 2 mmol of S2, 0.5 mmol of PhOTf, 1 mmol of (i-Pr)2NEt, THF as solvent, T=508C.
1.25 mol% of [Pd2ACHTNUTRGNE(NUG dba)3], 2 mmol of S2, 0.5 mmol of phenyl triflate, THF as solvent, 1 mmol of (i-Pr)2NEt, T=508C.
Conversion percentages and enantiomeric excesses measured by GC.
Poison added in the beginning of the reaction.
[b]
[c]
[d]
[e]
Poison added after 24 h reaction for the colloidal system and after 4 h for the molecular system.
Conclusions
phosphite ligands are the actual catalysts. We propose,
as in dendrimer catalysis, that CFMR experiments
We successfully synthesized a series of palladium should be conducted routinely to reveal the nature of
nanoparticles Pd1–Pd5 stabilized by five chiral sugar- the catalyst in the liquid phase for catalytic reactions
based oxazolinyl-phosphite ligands, containing several using MNPs. Without further study, our results cannot
substituents at the oxazoline and phosphite moieties. be extrapolated to other systems. Cross-coupling reac-
They were characterized by transmission electron mi- tions, however, require three coordination sites at a
croscopy (TEM), X-ray powder diffraction (XRD), divalent, hence square-planar, palladium ion – the
infrared spectroscopy (IR), and elemental analysis. presence of ees shows that the chiral ligands are coor-
Their structure is akin to that of [Pd561ACTHNUGTRNEG(UN OAc)180(L)60] dinated – and it is hard to imagine how a surface
(L=phenantroline, bipyridine), analogous to the atom, even one on the edges, could provide sufficient
giant clusters described by Moiseev and co-workers. space to carry out this reaction via the commonly ac-
The new nanoparticles were applied in Pd-catalyzed cepted mechanism. Nevertheless, the findings can be
asymmetric allylic alkylation and Heck coupling reac- made useful; when an excess of oxidizing aryl halide
tions. The results obtained by using a continuous-flow is used part of the palladium used will remain in solu-
membrane reactor (CFMR), TEM, poisoning and ki- tion, but an excess of nucleophile will result in a pal-
netic measurements show that the occurrence of mono- ladium(0) complex at the end of the reaction, which
metallic, molecular species can account for the cata- could well precipitate on the MNPs, thus facilitating
lytic activity observed. Especially the CFMR experi- separation.[6e,8,28] Interpretation of such recycling ex-
ments proved to be crucial to elucidate that the leach- periments remains tricky though, as it is known that
ed molecular palladium species containing oxazolinyl- very low concentrations of palladium catalyst can ac-
2594
asc.wiley-vch.de
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 2583 – 2598

Palladium Nanoparticles in Allylic Alkylations and Heck Reactions
FULL PAPERS
lowed by BSA (370 mL, 1.5 mmol) and KOAc (2 mg). The
mixture was stirred at room temperature and samples were
taken off. Each sample was then diluted with diethyl ether
(5 mL), filtered over Celite and washed with a saturated
aqueous solution of NH4Cl (3ꢉ20 mL) and water (2ꢉ
20 mL). The organic phase was dried over MgSO4, filtered
off, and solvent removed under reduced pressure. The con-
count for the activity of Suzuki reactions and Heck
reactions,[29] and thus the recycling experiments with-
out measurable loss of activity reported frequently
are not very informative. For ligand-modified cata-
lysts most likely the most reactive catalyst (precursor)
is an arylpalladium halide ligand complex as was
shown for the Heck reaction[30] and for the Buch-
1
version was measured by H NMR. To determine the enan-
À
wald–Hartwig N C cross-coupling reaction, for which
tioselectivities by HPLC (Chiralcel-OD, 0.5% 2-propanol/
hexane, flow 0.5 mLminÀ1), a sample was filtered over basic
alumina using dichloromethane as eluent.[35]
turnover frequencies of 140 hÀ1 were obtained at
room temperature.[31] Enantioselective catalysis by
bulk metal particles surface-modified by chiral addi-
tives is an established technique,[32] which remains in- Procedure for Allylic Alkylation using Palladium
teresting for MNPs as well, but perhaps one should Molecular Systems
search for new reactions, more common to metallic
A degassed solution of [PdCl
and the appropriate
ACHTUNGTRENNUNG
catalysts, such as hydrogenation and oxidation,[15,17,33]
or reactions less common for molecular species, for
example photogeneration of H2.[34]
(0.0055 mmol) in dichloromethane (0.5 mL) was stirred for
30 min. Subsequently, a solution of rac-3-acetoxy-1,3-diphen-
yl-1-propene (126 mg, 0.5 mmol) in dichloromethane
(1.5 mL), dimethyl malonate (171 mL, 1.5 mmol), N,O-bis-
AHCTUNGTREG(NNNU trimethylsilyl)-acetamide (370 mL, 1.5 mmol) and KOAc
Experimental Section
(2 mg) were added. The reaction mixture was stirred at
room temperature. After the desired reaction time, the reac-
tion mixture was diluted with Et2O (5 mL) and saturated
NH4Cl (aqueous) (25 mL) was added. The mixture was ex-
tracted with Et2O (3ꢉ10 mL) and the extract dried over
MgSO4. The solvent was removed and the conversion was
General Considerations
All reactions were carried out under argon using standard
Schlenk tube, Fischer–Porter bottle and vacuum line tech-
niques, or in a glove-box. Ligands 1–5 were synthesized as
previously described.[10] [Pd
2ACHTNUTRGENNG(U dba)3] was used as commercial-
1
measured by H NMR. To determine the enantioselectivities
ly available. Solvents were purified and dried by standard
procedures. All reagents and solvents were degassed before
use by using three freeze-pump-thaw cycles. Elemental anal-
yses were carried out by the “Servei Cientꢃfic-Tꢅcnics” of
the Universitat Rovira i Virgili and by the “Service Central
d’Analyses du CNRS” in Lyon.
by HPLC (Chiralcel-OD, 0.5% 2-propanol/hexane, flow
0.5 mLminÀ1), a sample was filtered over basic alumina
using dichloromethane as eluent.[35]
Procedure for Heck Reaction using Palladium
Nanoparticles
To a suspension of nanoparticles (10 mg) in THF (5 mL) the
desired amount of the corresponding ligand was added. 2,3-
Dihydrofuran (151 mL, 2.0 mmol), phenyl triflate (81 mL,
0.50 mmol) and (i-Pr)2EtN (174 mL, 1.0 mmol) were added
to the catalyst solution. The solution was stirred at 508C
under argon. After the desired reaction time, the mixture
was diluted with additional diethyl ether, filtered and
washed with water, dried over MgSO4 and evaporated. Con-
version and selectivity were determined by GC.[36]
Synthesis of Palladium Nanoparticles
ACHTUNGTRENNUNG[Pd2ACHTUNGTRENNUNG(dba)3·CHCl3] (64.7 mg, 0.06 mmol) was dissolved under
argon in a solution of THF (50 mL) containing the corre-
sponding ligand (0.0125 mmol) in a closed pressure Fischer–
Porter bottle. After pressurization at room temperature
under H2 (3 bar) for 30 min, the initial purple solution
became grey in a few minutes. The vigorous magnetic stir-
ring and the pressure of H2 were maintained for 18 h. The
hydrogen pressure was then eliminated, and a drop of the
colloidal solution was deposited under argon on a carbon-
covered copper grid for microscopy analysis. Evaporation of
the solvent gave a black precipitate which was washed with
pentane (3ꢉ10 mL) and dried at low pressure. The colloids
were characterized by TEM, elemental analysis, IR spectros-
copy and XRD. Pd1: yield: 21 mg, 78%; Pd2: yield: 231 mg,
83%; Pd3: yield: 20 mg, 64%; Pd4: yield: 19 mg, 69%; Pd5:
yield: 21 mg, 73%.
Procedure for Heck Reaction using Palladium
Molecular Systems
(dba)3] (12 mg, 1.25ꢉ10À2 mmol) and the
A mixture of [Pd2ACTHNUTRGNEUNG
corresponding chiral ligand (2.8ꢉ10À2 mmol) in dry de-
gassed THF (3.0 mL) was stirred under argon at room tem-
perature for 15 min. 2,3-Dihydrofuran (151 mL, 2.0 mmol),
phenyl triflate (81 mL, 0.50 mmol) and (i-Pr)2EtN (174 mL,
1.0 mmol) were added to the catalyst solution. The solution
was stirred at 508C under argon. After the desired reaction
time, the mixture was diluted with additional diethyl ether
and washed with water, dried over MgSO4 and evaporated.
Conversion and selectivity were determined by GC.[36]
Procedure for Allylic Alkylation Catalyzed using
Palladium Nanoparticles
To a suspension of nanoparticle (5 mg) in dichloromethane
(3 mL) the desired amount of the corresponding ligand was
added.
rac-3-Acetoxy-1,3-diphenyl-1-propene
(126 mg,
0.5 mmol) and dimethyl malonate (171 mL, 1.5 mmol) dis-
solved in dichloromethane (2 mL) were then added, fol-
Adv. Synth. Catal. 2008, 350, 2583 – 2598
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2595

Montserrat Diꢀguez et al.
FULL PAPERS
TEM Analysis
ICP-AES Analysis
Samples for TEM analysis were prepared in a glove-box by
slow evaporation of drops of the colloidal solution deposited
onto a carbon-covered copper grid. TEM analyses were per-
formed on a JEOL 1011 electron microscope operating at
100 kV. The size distributions were determined through a
manual analysis by measuring 300 particles on a given grid
to obtain a statistically size distribution and a mean diame-
ter.
The corresponding solution of the reactor volumes pumped
through was evaporated under reduced pressure and dis-
solved in HCl. The ICP-AES measurements were performed
with a sequential Jarrell Ash upgraded (Model 25) Atom-
scan model 2400 ICP scanning monochromator and a
Perkin-Elmer Optima 3000 XL instrument. The measured
atomic line of Pd was 340.458 nm.
XRD Analysis
Acknowledgements
Powder X-ray diffraction patterns of the different samples
were obtained with a Siemens D5000 diffractometer using
nickel-filtered Cu Ka radiation. The patterns were recorded
over a range of 2q angles from 108 to 908 and crystalline
phases were identified using the Joint Committee on
Powder Diffraction Standards (JCPDS) files.
We thank Erik Berteling, University of Amsterdam, for his
valuable assistance. We thank the Spanish Government [Con-
solider Ingenio 2010 (Grant CSD2006–0003), CTQ2004–
04412/BQU, CTQ2007–62288/BQU and the Ramꢀn y Cajal
fellowship
to
O.P.),
the
Catalan
Government
(2005SGR007777, 2006BE00291 to O.P. and Distinction to
M.D.), the University of Amsterdam, the Universitꢁ Paul Sa-
batier and CNRS for financial support.
IR Analysis
IR spectra were recorded on a Perkin–Elmer Spectrometer
GX (FT-IR). The samples were prepared by deposition of a
drop of concentrated solution onto KBr pellets.
References
Continuous-Flow Membrane Reactor Experiments
[1] See, for instance: a) G. Schmid, in: Nanoparticles, from
Theory to Application, VCH, Weinheim, 2004; b) G.
Schmid, Chem. Rev. 1992, 92, 1709; c) N. Toshima, T.
Yonezawa, New J. Chem. 1998, 1179; d) J. D. Aiken III,
R. G. Finke, J. Mol. Catal. A: Chem. 1999, 145, 1;
e) M. A. El-Sayed, Acc. Chem. Res. 2001, 34, 257; f) H.
Bçnnemann, R. M. Richards, Eur. J. Inorg. Chem. 2001,
2455; g) D. Astruc, F. Fu, J. R. Aranzaes, Angew.
Chem. Int. Ed. 2005, 44, 7852; h) P. Migowski, J.
Dupont, Chem. Eur. J. 2007, 13, 32.
[2] a) M. Studer, H.-U. Blaser, C. Exner, Adv. Synth. Catal.
2003, 345, 45; b) H. Bçnnemann, G. A. Braun, Angew.
Chem. Int. Ed. Engl. 1996, 35, 1992; c) H. Bçnnemann,
G. A. Braun, Chem. Eur. J. 1997, 3, 1200; d) X. Zuo, H.
Liu, D. Guo, X. Yang, Tetrahedron 1999, 55, 7787;
e) J. U. Kçhler, J. S. Bradley, Catal. Lett. 1997, 45, 203;
f) J. U. Kçhler, J. S. Bradley, Langmuir 1998, 14, 2730.
[3] M. Tamura, H. Fujihara, J. Am. Chem. Soc. 2003, 125,
15742.
[4] a) S. Jansat, M. Gꢂmez, K. Phillipot, G. Muller, E.
Guiu, C. Claver, S. Castillꢂn, B. Chaudret, J. Am.
Chem. Soc. 2004, 126, 1592; b) I. Favier, M. Gꢂmez, G.
Muller, M. R. Axet, S. Castillꢂn, C. Claver, S. Jansat, B.
Chaudret, K. Philippot, Adv. Synth. Catal. 2007, 349,
2459.
[5] See for example: a) M. Diꢀguez, O. Pꢁmies, C. Claver,
Chem. Rev. 2004, 104, 3189; b) M. Diꢀguez, O. Pꢁmies,
A. Ruiz, Y. Dꢃaz, S. Castillꢂn, C. Claver, Coord. Chem.
Rev. 2004, 248, 2165; c) M. Diꢀguez, A. Ruiz, C.
Claver, Dalton Trans. 2003, 2957; d) G. J. H. Buisman,
M. E. Martin, E. J. Vos, A. Klootwijk, P. C. J. Kamer,
P. W. N. M. van Leeuwen, Tetrahedron: Asymmetry
1995, 6, 719; e) S. C. Gilbertson, C. Chang, J. Org.
Chem. 1995, 60, 6226; f) H. B. Kagan, T. P. Dang, J.
Am. Chem. Soc. 1972, 94, 6429; g) H. Brunner, W. Pier-
onczyk, J. Chem. Res. (S) 1980, 76; h) T. V. RajanBabu,
T. A. Ayers, A. L. Casalnuovo, J. Am. Chem. Soc. 1994,
Description of the reactor: A continuous-flow membrane
reactor developed previously was used for the elucidation of
the nature of the catalytic species. It consists of a stainless
steel autoclave in which the catalyst solution can be injected
via a two-way valve. The membrane is located at the bottom
of the autoclave and kept in place by a Vitonꢊ O-ring. The
Koch/SelRO MPF-50 nanofiltration membrane (MWCO=
700 Dalton)[37] was chosen for the filtration experiments
since it has the correct size to allow the diffusion of leached
molecular Pd species, but not of Pd nanoparticles.[38]
A
vessel containing the substrate solution (under inert atmos-
phere) is connected to the membrane reactor via an HPLC
pump that pushes the substrate solution into the reactor and
through the membrane.
Reactor set-up: A piece of Koch/SelRO MPF-50 mem-
brane (stored in EtOH) was cut to the correct size for the
reactor, stored in acetone for one night, and then stored in
methanol (for at least one night). After that it was trans-
ferred into the membrane reactor. The membrane and
KOAc (2 mg) were first flushed overnight with CH2Cl2 and
then with substrate solution (approximately two reactor vol-
umes, 12 mL). The substrate solution was prepared by
mixing S1 (512 mg, 2 mmol), dimethyl malonate (0.68 mL,
6 mmol) and oBSA (1.48 mL, 6 mmol) in CH2Cl2 (60 mL).
This solution was pumped through the reactor with a flow
rate of 6 mLhÀ1. The reaction was started by transferring
the catalyst solution (5 mg of the corresponding nanoparti-
cle and 20 mol% of ligand in 2 mL of CH2Cl2) into the
membrane reactor. Samples of the solution coming out of
the reactor were collected and analyzed by 1H NMR and
HPLC and the palladium content was measured after the
appropriate work-up by ICP-AES.[38b,39]
2596
asc.wiley-vch.de
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 2583 – 2598

Palladium Nanoparticles in Allylic Alkylations and Heck Reactions
FULL PAPERS
116, 4101; i) R. Selke, React. Kinet. Catal. Lett. 1979,
10, 135.
votortsev, A. V. Kvit, A. L. Chuvilin, J. Cryst. Growth
1996, 163, 377; b) M. K. Starchevsky, S. L. Hladiy, Y. A.
Pazdersky, M. N. Vargaftik, I. I. Moiseev, J. Mol. Cat.
A: Chem. 1999, 146, 229; c) I. I. Moiseev, M. N. Vargaf-
tik, in: Perspectives in Catalysis, (Eds.: K. I. Thomas,
K. I. Zamaraev), Blackwell, Oxford, 1992, pp 91–123;
d) V. Y. Shevchenko, M. I. Samoilovich, A. L. Talis,
A. E. Madison, Glass. Phys. Chem. 2005, 31, 259.
[16] Schmid and co-workers have found that usually the
ideal icosahedron arrangement is distorted and forms a
cubooctahedron geometry. See for instance: a) G.
Schmid, M. Harms, J.-O. Malm, J.-O. Bovin, J. van Rui-
tenbeck, H. W. Zandbergen, W. T. Fu, J. Am. Chem.
Soc. 1993, 115, 2046; b) G. Schmid, B. Morun, J.-O.
Malm, Angew. Chem. Int. Ed. Engl. 1989, 28, 778;
c) ref.[1a]
[6] See for instance: a) N. T. S. Phan, M. van der Sluys,
C. W. Jones, Adv. Synth. Catal. 2006, 348, 609; b) J. G.
de Vries, Dalton Trans. 2006, 421; c) I. W. Davies, L.
Matty, D. L. Hughes, P. J. Reider, J. Am. Chem. Soc.
2001, 123, 10139; d) M. B. Thathagar, P. J. Kooyman, R.
Boerleider, E. Jansen, C. J. Elsevier, G. Rothenberg,
Adv. Synth. Catal. 2005, 347, 1965; e) J. P. Simeone,
J. R. Sowa Jr. , Tetrahedron 2007, 63, 12646; f) G. Bu-
droni, A. Corma, H. Garcia, A. Primo, J. Catal. 2007,
251, 345; g) D. Astruc, Inorg. Chem. 2007, 46, 1884;
h) J. A. Widegren, R. G. Finke, J. Mol. Cat. A: Chem.
2003, 198, 317; i) R. Chinchilla, C. Nꢋjera, Chem. Rev.
2007, 107, 874; j) L. Durꢋn Pachꢂn, G. Rothenberg,
Appl. Organomet. Chem. 2008, 288.
[7] The use of reactivity patterns is hampered due to the
difficulty to find a reaction that only works with hetere-
geneous catalysts or viceversa (see ref.[6h]). For recent
examples, see: a) J. A. Widegren, R. G. Finke, J. Mol.
Catal. A: Chem. 2003, 191, 187; b) S. Jansat, D. Picurel-
li, K. Pelzer, K. Philippot, M. Gꢂmez, G. Muller, P. Le-
cante, B. Chaudret, New J. Chem. 2006, 30, 115.
[8] A. H. M. de Vries, F. J. Parlevliet, L. Schmieder-van
de Vondervoort, J. H. M. Mommers, H. J. W. Hender-
ickx, M. A. N. Walet, J. G. de Vries, Adv. Synth. Catal.
2002, 344, 996.
[9] a) M. B. Thathagar, J. E. ten Elshof, G. Rothenberg,
Angew. Chem. Int. Ed. 2006, 45, 2886; b) A. V. Gaik-
wad, A. Holuigue, M. B. Thathagar, J. E. ten Elshof, G.
Rothenberg, Chem. Eur. J. 2007, 13, 6908.
[10] a) Y. Mata, O. Pꢁmies, M. Diꢀguez, C. Claver, Adv.
Synth. Catal. 2005, 347, 1943; b) Y. Mata, M. Diꢀguez,
O. Pꢁmies, C. Claver, Org. Lett. 2005, 7, 5597; c) Y.
Mata, O. Pꢁmies, M. Diꢀguez, Chem. Eur. J. 2007, 13,
3296.
[17] a) M. N. Vargaftik, I. I. Moiseev, D. I. Kochubei, K. I.
Zamaraev, Faraday Discuss. 1992, 92, 13; b) V. P. Olesh-
ko, J. Mol. Catal. A: Chem. 2006, 249, 4.
[18] This is meant as a rough calculation; also, the clusters
are different in nature, as the Moiseev cluster is
charged and the present ones contain Pd(0).
[19] For recent reviews on Pd-catalyzed allylic substitution,
see: a) J. Tsuji, Palladium Reagents and Catalysis, in:
Innovations in Organic Synthesis, Wiley, New York,
1995; b) B. M. Trost, D. L. van Vranken, Chem. Rev.
1996, 96, 395; c) M. Johannsen, K. A. Jorgensen, Chem.
Rev. 1998, 98, 1689; d) A. Pfaltz, M. Lautens, in: Com-
prehensive Asymmetric Catalysis, (Eds.: E. N. Jacobsen,
A. Pfaltz, H. Yamamoto), Springer-Verlag, Berlin,
1999, Vol. 2, Chapter 24; e) B. M. Trost, M. L. Crawley,
Chem. Rev. 2003, 103, 2921; for recent reviews on Pd-
catalyzed Heck reactions, see: f) L. F. Tietze, H. Ila,
H. P. Bell, Chem. Rev. 2004, 104, 3453; g) L. X. Dai, T.
Tu, S. L. You, W. P. Deng, X. L. Hou, Acc. Chem. Res.
2003, 36, 659; h) C. Bolm, J. P. Hildebrand, K. MuÇiz,
N. Hermanns, Angew. Chem. Int. Ed. 2001, 40, 3284;
i) M. Shibasaki, E. M. Vogl, in: Comprehensive Asym-
metric Catalysis, (Eds.: E. N. Jacobsen, A. Pfaltz, H. Ya-
mamoto), Springer, Heidelberg, 1999; j) O. Loiseleur,
M. Hayashi, M. Keenan, N. Schmees, A. Pfaltz, J. Or-
ganomet. Chem. 1999, 576, 16; k) M. Beller, T. H. Rier-
meier, G. Stark, in: Transition Metals for Organic Syn-
thesis, (Eds.: M. Beller, C. Bolm), Wiley-VCH, Wein-
heim, 1998.
[20] B. M. Trost, D. J. Murphy, Organometallics 1985, 4,
1143.
[21] The recycling was done adding in every run 10 mol%
of ligand.
[22] The recycling experiments have been performed fol-
lowing the methodology described in ref.[4a] Therefore,
the final catalytic suspension was cooled down to
À708C to favour the separation of the nanoparticles
from the solution. Then, the resulting solution was re-
moved. The remaining solid was washed with dichloro-
methane under argon and subsequently used as catalyst
precursor.
[11] Several complexes that may be involved have a molec-
ular mass slightly above 700, but this does not seem to
influence the results.
[12] a) D. de Groot, E. B. Eggeling, J. C. de Wilde, H. Kooij-
man, R. J. van Haaren, A. W. van der Made, A. L.
Spek, D. Vogt, J. N. H. Reek, P. C. J. Kamer, P. W. N. M.
van Leeuwen, Chem. Commun. 1999, 1623; b) N. J.
Hovestad, E. B. Eggeling, H. J. Heidbꢌchel, J. T. B. H.
Jastrzebski, U. Kragl, W. Keim, D. Vogt, G. van Koten,
Angew. Chem. Int. Ed. 1999, 38, 1655; c) N. Brinkmann,
D. Giebel, G. Lohmer, M. T. Reetz, U. Kragl, J. Catal.
1999, 183, 163; d) D. de Groot, B. F. M. de Waal,
J. N. H. Reek, A. P. H. J. Schenning, P. C. J. Kamer,
E. W. Meijer, P. W. N. M. van Leeuwen, J. Am. Chem.
Soc. 2001, 123, 8453.
[13] a) C. Pan, K. Pelzer, K. Philippot, B. Chaudret, F. Das-
senoy, P. Lecante, M.-J. Casanove, J. Am. Chem. Soc.
2001, 123, 7584; b) K. Philippot, B. Chaudret, C. R.
Chimie 2003, 6, 1019.
[14] a) H. M. Rietveld, J. Appl. Cryst. 1969, 2, 65; b) C. W.
Scheeren, G. Machado, S. R. Teixeira, J. Morais, J. B.
Domingos, J. Dupont, J. Phys. Chem. B. 2006, 110,
13011; c) A. J. Bruss, M. A. Gelesky, G. Machado, J.
Dupont, J. Mol. Cat. A: Chem. 2006, 252, 212.
[23] For recovered Pd1 we found 42.59% of Pd and 2.11%
of P. For recovered Pd5 we found 46.45% of Pd and
1.81% of P.
[15] a) V. V. Volkov, G. van Tendeloo, G. A. Tsirkov, N. V.
[24] Y. Mata, PhD Thesis, Universitat Rovira i Virgili, 2007.
Cherkashina, M. N. Vargaftik, I. I. Moiseev, V. M. No-
Adv. Synth. Catal. 2008, 350, 2583 – 2598
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
2597

Montserrat Diꢀguez et al.
FULL PAPERS
[25] The in situ NMR study indicates that [Pd
(allyl)
N
[30] G. P. F. van Strijdonck, M. D. K. Boele, P. C. J. Kamer,
J. G. De Vries, P. W. N. M. van Leeuwen, Eur. J. Inorg.
Chem. 1999, 1073.
L)2]+ (L=1–5) are in equilibrium with species [Pd-
G
ACHTUNGTRENNUNG
(k2-N,P-L)]+ (L=1–5) at a ligand-to-palladium
[31] a) Y. Guari, G. P. F. van Strijdonck, M. D. K. Boele,
J. N. H. Reek, P. C. J. Kamer, P. W. N. M. van Leeuwen,
Chem. Eur. J. 2001, 7, 475; b) Y. Guari, D. S. van Es,
J. N. H. Reek, P. C. J. Kamer, P. W. N. M. van Leeuwen,
Tetrahedron Lett. 1999, 40, 3789.
[32] T. Mallat, E. Orglmeister, A. Baiker, Chem. Rev. 2007,
107, 4863.
[33] R. A. T. M. van Benthem, H. Hiemstra, P. W. N. M. van
Leeuwen, J. W. Geus, W. N. Speckamp, Angew. Chem.
Int. Ed. 1995, 34, 457.
G
ACHTUNGTRENNUNG
E
ACHTUNGTRENNUNG
(k1-P-
palladium ratio of 0.9, the formation of [PdACTHNUTRGENNU(G allyl)ACHTUGNTRENNUGN
1)2]+ is not observed.
[26] See, for instance: a) A. Howard, C. E. J. Mitchell, R. G.
Egdell, Surf. Sci. 2002, 515, L504; b) R. Narayanan,
M. A. El-Sayed, J. Am. Chem. Soc. 2003, 125, 8340;
c) J. Li, H. C. Zeng, J. Am. Chem. Soc. 2007, 129,
15839.
[34] N. Toshima, K. Nakata, H. Kitoh, Inorg. Chim. Acta
1997, 265, 149.
[35] O. Pꢁmies, G. P. F. van Strijdonck, M. Diꢀguez, S. Deer-
enberg, G. Net, A. Ruiz, C. Claver, P. C. J. Kamer,
P. W. N. M. van Leeuwen, J. Org. Chem. 2001, 66, 8867.
[36] S. R. Gilbertson, Z. Fu, Org. Lett. 2001, 3, 161.
[37] a) Koch Membrane Systems, Wilmington, USA, http://
www.kochmembrane.com; b) C. Linder, M. Nemas, M.
Perry, R. Katraro, U. S. Patent 5,265,734, 1992.
[38] This set-up has previously given good results with den-
dritic ligands in transition metal catalysed reactions.
See, for instance ref.[9a] and: a) D. de Groot, P. G. Em-
merink, C. Coucke, J. N. H. Reek, P. C. J. Kamer,
P. W. N. M. van Leeuwen, Inorg. Chem. Commun. 2000,
3, 711; b) D. de Groot, J. N. H. Reek, P. C. J. Kamer,
P. W. N. M. van Leeuwen, Eur. J. Org. Chem. 2002,
1085; c) J. N. H. Reek, D. de Groot, G. E. Oosterom, R.
van Heerbeek, P. C. J. Kamer, P. W. N. M. van Leeuwen,
Polym. Mater. Sci. Eng. 2001, 84, 158; d) G. E. Ooster-
om, S. Steffens, J. N. H. Reek, P. C. J. Kamer,
P. W. N. M. van Leeuwen, Top. Catal. 2002, 19, 61.
[39] A. Buhling, J. W. Elgersma, S. Nkrumah, P. C. J.
Kamer, P. W. N. M. van Leeuwen, J. Chem. Soc.,
Dalton Trans. 1996, 2143.
[27] C. C. Cassol, A. P. Umpierre, G. Machado, S. I. Wolke,
J. Dupont, J. Am. Chem. Soc. 2005, 127, 3298.
[28] a) R. Najman, J. K. Cho, A. F. Coffey, J. W. Davies, M.
Bradley, Chem. Commun. 2007, 5031; b) F. Fernꢋndez,
B. Cordero, J. Durand, G. Muller, F. Malbosc, Y. Kihn,
E. Teuma, M. Gꢂmez, Dalton Trans. 2007, 5572; c) R.
Soler, S. Cacchi, G. Fabrizi, G. Forte, L. Martin, S. Mar-
tinez, E. Molins, M. Moreno-Manas, F. Petrucci, A.
Roig, R. M. Sebastian, A. Vallribera, Synthesis 2007,
3068: d) Y. Zhu, S. C. Peng, A. Emi, Z. Su, Z. Monali-
sa, R. A. Kemp, Adv. Synth. Catal. 2007, 349, 1917;
e) Y. Tsuji, T. Fujihara, Inorg. Chem. 2007, 46, 1895;
f) M. A. R. Meier, M. Filali, J.-F. Gohy, U. S. Schubert,
J. Mater. Chem. 2006, 16, 3001; g) J. K. Cho, R.
Najman, T. W. Dean, O. Ichihara, C. Muller, M. Brad-
ley, J. Am. Chem. Soc. 2006, 128, 6276.
[29] a) A. Alimardanov, L. Schmieder-van de Vondervoort,
A. H. M. de Vries, J. G. de Vries, Adv. Synth. Catal.
2004, 346, 1812; b) R. B. Bedford, Chem. Commun.
2003, 1787.
2598
asc.wiley-vch.de
ꢈ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2008, 350, 2583 – 2598