Pd nanoclusters in C-C coupling reactions: Proof of leaching

Article abstract of DOI:10.1002/anie.200504321

A simple experiment based on exclusion of Pd nanoclusters was used to identify the true catalytic species in cluster-catalyzed Heck cross-couplings. A special reactor was used in which two compartments are separated by a nanoporous membrane that allows the passage of Pd atoms and ions but not of Pd nanoclusters (see picture). The results show that the real catalysts in the Heck coupling of n-butyl acrylate with iodobenzene are species leached from the Pd nanoclusters. (Figure Presented).

Full text of DOI:10.1002/anie.200504321

Communications
Cluster Catalysis
DOI: 10.1002/anie.200504321
À
Pd Nanoclusters in C C Coupling Reactions:
Proof of Leaching
Mehul B. Thathagar, Johan E. ten Elshof, and
Gadi Rothenberg*
Carbon–carbon coupling reactions are ubiquitous in organic
[5,6]
synthesis. Heck,[1,2] Suzuki,[3,4] andSonogashira couplings
occupy a special place among such reactions due to their mild
reaction conditions. The coupling products are mostly used as
intermediates for polymers, natural products, and bioactive
compounds.[7,8]
The most common catalyst precursors for these reactions
are palladium(ii) complexes,[7,9,10] usually with phosphane
ligands.[11] However, most of these ligands are air- and
moisture-sensitive, involve multistep syntheses, andare
difficult to separate from the final product. This makes
“ligand-free” Pd catalysis an attractive option.[12–20] PdII salts
form Pd0 clusters under the reaction conditions, and these can
catalyze coupling reactions. However, the “naked” Pd clusters
[15,19,21,22]
aggregate to inactive Pdblack.
aggregation, stabilizedPdclusters
To prevent this
[23]
are increasingly being
[20,21,24–32]
À
usedas C C coupling catalysts.
These clusters may
be goodcatalysts, but we still do not know (in contrast to
heterogeneous Pd[33]) whether catalysis occurs on the cluster
[15,19,22,35,36]
surface or by leached[34] Pdspecies.
Although much work has been done in this area, there is
no single definitive experiment that reveals the detailed
[35]
mechanism of cluster catalysis in the liquidphase.
From
studies of cluster-size effects in Heck and Suzuki reactions, it
was proposedthat catalysis occurs at defect sites on the
cluster surface.[27,37,38] Conversely, de Vries et al.[15] suggested
an equilibrium between Pdclusters anda monomeric or
dimeric moiety which is the actual active species. Recently,
using transmission electron microscopy (TEM) andkinetic
[26,40]
studies, we[39] andothers
also demonstrated that species
leachedfrom the cluster surface are probably responsible for
catalysis in these types of reactions. However, the conclusions
were basedon circumstantial evidence. Here we report the
first direct and unambiguous test proving that leached Pd
À
species are the true catalysts in Pd-cluster-catalyzed C C
coupling reactions.
[*] Dr. M. B. Thathagar, Dr. G. Rothenberg
van’t Hoff Institute for Molecular Sciences
University of Amsterdam
Nieuwe Achtergracht 166, 1018 WV Amsterdam (The Netherlands)
Fax : (+31)20-525-5604
E-mail: gadi@science.uva.nl
Dr. J. E. ten Elshof
Faculty of Science and Technology
University of Twente
P.O. Box 217, 7500 AE Enschede (The Netherlands)
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
2886
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 2886 –2890

Angewandte
Chemie
For this experiment, we built a special U-tube permeation
cell[41] membrane reactor (Figure 1, top) consisting of two
stainless steel compartments separatedby a membrane
(henceforth referredto as sides A and B). The membrane
inert gas
attachment
19 nm
19 nm
45
40
35
30
25
Frequency/ % 20
membrane
stainless
steel
15
10
5
membrane
with 5 nm pores
0
12
13
14
15
16
17
Particle size/ nm
Figure 2. Transmission electron micrograph (top) and corresponding
size distribution of the Pd clusters (bottom, based on 85 particles
counted).
reaction mixture
without catalyst
size was 12 nm. Moreover, due to the amorphous structure of
the membrane, all the paths through it are tortuous worm-
holes, andno path consists of only maximum-size pores. Thus,
this membrane also retains particles smaller than 11 nm.
In the first test reactions, we carriedout Sonogashira
coupling of phenylacetylene (1) with iodobenzene (2)
Side B
Side A
Pd clusters (15 nm)
leached Pd species diffused
through the membrane
cannot pass through the
membrane
[Eq. (1)]. We placedthe Pdcluster suspension on side
A,
Figure 1. Photograph (top) and schematic (bottom) of the two-com-
partment membrane reactor and the cluster-exclusion concept.
physically separates the Pdclusters andthe reaction mixture
in the reactor. The concept is simple: the suspension of Pd
clusters is placedon one side of the membrane, andthe
reaction mixture on the other (Figure 1, bottom). The
membrane is designed[42] with 5-nm pores that allow the
diffusion of leached Pd species, but not of Pd clusters. The
clusters cannot pass through the membrane because they are
too large (14.5 Æ 2.5 nm, see below). Thus, monitoring the
reaction over time on both sides gives direct information on
the true catalyst.
The Pd clusters were synthesized by reducing Pd(NO3)2
with tetraoctylammonium glycolate (TOAG), usedboth as
reducing and stabilizing agent.[43] This methodgives clusters
with an average size of 14.5 nm anda narrow particle size
distribution (Figure 2). An a-alumina-supportednanoporous
and 1, 2, andtetrabutylammonium acetate (TBAA) as base
on side B. The experiments were performedin DMF at 70 8C.
Note that the base is soluble in this case. After 3 days, we
observed72% yieldof coupling product 3 on side B, where
there was originally no catalyst. However, as the reactants and
base were not evenly distributed on both sides, they may have
diffused from side B to the lower concentration on side A. It
cannot be excluded that the reaction occurred in compart-
ment A and the product then diffused from side A to side B.
We did observe 9% yield of 3 on side A, which couldbe
attributed to the product coming from side B or forming on
side A. Consequently, this experiment did not give a clear
answer to our question.
g-alumina membrane was synthesizedaccoridng to the
[42]
reportedproceudre.
The pore size distribution of the
membrane was determined by permporometry.[44,45] Most of
the pores were around4.9 nm andfew pores were larger than
5 nm, with a maximum size of 11 nm. The minimum Pdcluster
To reveal the actual catalytic species, it was necessary to
prevent diffusion of the reactants through the membrane and
Angew. Chem. Int. Ed. 2006, 45, 2886 –2890
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2887

Communications
to use an insoluble base. We therefore investigatedthe Heck
100
90
80
70
60
50
40
30
20
10
0
reaction in the presence of NaOAc, which is insoluble in
DMF. The coupling of n-butyl acrylate (4) with iodobenzene
(2) to give n-butyl cinnamate (6) was usedas a model reaction
[Eq. (2)].
COOBu
side B
66
(base, no catalyst)
Yield / %
side A
(catalyst, no base)
0
30
60
90
120
t / h
Figure 4. Time-resolved profile for the yield of 6. The product observed
on side A has diffused from side B.
Before performing the coupling reaction, we examined
the time-dependent transport of 4, 2, and 6 through the
membrane. Pure 4, 2, and 6 in DMF were placedon side B,
andonly DMF on side A. Figure 3 shows that indeed all three
4.9% of the product was observed that may have diffused
from side B. Duplicate experiments showedgoodreproduci-
bility. No aggregation was observedafter the reaction on
either side of the reactor.
X-ray photoelectron spectroscopy was performedon the
membrane after the reaction to determine whether Pd enters
the pores (see Supporting Information for details). There
were no significant differences between the elemental com-
positions of the surface andsubsurface layers. Traces of Pd
were foundon all spots on or below the membrane surface,
but the concentration was in all cases within the range of 0.02–
0.07 atom%. As the specific surface area[47] of this g-alumina
is about 285 m2 gÀ1, this is equivalent to a surface concen-
COOBu
10
8
I
6
Amount
transported/mol%
4
COOBu
2
2
0
tration of roughly one Pdatom per 20 nm .
0
30
60
90
120
The above results prove that the reaction involves leached
Pdspecies, for which the nanosizedPdclusters act as
reservoirs. We still do not know whether these species are
Pd0 atoms releasedfrom the defect sites or soluble Pd II species
formed after oxidative addition of PhI.[36,39] In theory, the Pd0
t / h
Figure 3. Time-dependent transport of molecules through the mem-
brane from side B to side A.
II
compounds were transported from side B to side A. Note that
no pressure was appliedto the system; transport was purely
due to concentration gradients. The transport rate was size-
dependent: 4 > 2 > 6. After 100 h, 10.5 mol% of 4, 6.5 mol%
of 2, and4.5 mol% of 6 were transportedto side A. This blank
experiment gives a measure of the backgrounddiffusion of
these specific compounds through the membrane. Similar
transport behavior can be expectedduring the reaction.
We then ran the Heck reaction using 1.5:1 molar mixtures
of 4 and 2 on both sides of the membrane to avoid
concentration gradients. NaOAc (1.5 equiv) was then added
to side B, andthe suspension of Pdclusters (0.01 equiv) was
introduced into side A at 1008C. Note that the base is
necessary for closing the catalytic cycle. As there is no base on
side A, no product can form there. Thus, any product
observedon side A must come from side B.[46]
andPd species could“recluster” on the other side of the
membrane, but in practice there is no stabilizer or reducing
agent, both of which are crucial for cluster formation, so this
option is extremely unlikely. In any event, these experiments
prove the hypothesis of de Vries et al. regarding ligand-free
Pdcatalysts in Heck reactions. [15,48]
In summary, using a simple approach basedon physical
exclusion, we prove here that the Pd-cluster-catalyzed Heck
reaction between n-butyl acrylate andioodbenzene (and
[39]
probably also other Heck andSonogashira reactions)
involves soluble Pdspecies releasedfrom the surface of the
clusters. Moreover, the general new technique we described
here permits the testing of many individual reactions. Further
investigations into the mechanism of other cross-coupling
reactions will be the subject of future research.
Figure 4 shows the yieldof n-butyl cinnamate (6) versus
time. No reaction was observedfor the first 5 h because the
releasedPdspecies from side A must first diffuse through the
nanoporous membrane to initiate the catalytic cycle on side B.
When the leachedPdspecies reach side B the reaction rate
increases andthen slows as 2 is consumed. After 120 h, the
yieldof 6 on side B was 88%. We did not observe any diffused
product 6 during the first 12 h on side A. However, after 120 h
Experimental Section
Membrane preparation: The alumina membrane consists of a macro-
porous a-alumina support anda thin nanoporous g-alumina layer.
The a-alumina supports were made by colloidal filtration of well-
dispersed 0.3-mm a-alumina particles (AKP-30, Sumitomo) in demin-
eralizedwater. The 50-wt% dispersion was stabilizedby peptizing
with 0.02m HNO3. After drying at 258C for 24 h, the filter compact
2888 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 2886 –2890

Angewandte
Chemie
was sinteredat 1100 8C in air for 2 h. Flat disks (1 39 mm, 2.0 mm
thickness) were obtainedby machining andpolishing. The final
porosity of these supports is about 30%, andthe average pore size 80–
120 nm.[49] The nanoporous g-alumina layer was preparedby dip-
coating the a-alumina supports in a boehmite-baseddip sol. The
boehmite sol was prepared by heating 70.0 mol of double-distilled
water to 958C, and then adding 0.50 mol of Al(OsBu)3 (97% purity,
Acros) dropwise under a nitrogen flow to avoid premature hydrolysis.
The mixture was stirredvigorously throughout the synthesis. After all
the Al(OsBu)3 was added, the mixture was kept at 958C for 3 h to
evaporate the butanol. The mixture was cooledto 60 8C, peptizedwith
HNO3 (65%, Merck) to pH 2.5, andthen heatedat reflux at 90 8C for
[17] B. M. Choudary, S. Madhi, N. S. Chowdari, M. L. Kantam, B.
Sreedhar, J. Am. Chem. Soc. 2002, 124, 14127.
[18] M. B. Thathagar, J. Beckers, G. Rothenberg, Green Chem. 2004,
6, 215.
[19] M. T. Reetz, J. G. de Vries, Chem. Commun. 2004, 1559.
[20] M. T. Reetz, E. Westermann, Angew. Chem. 2000, 112, 170;
Angew. Chem. Int. Ed. 2000, 39, 165.
[21] M. B. Thathagar, J. Beckers, G. Rothenberg, Adv. Synth. Catal.
2003, 345, 979.
[22] A. Biffis, M. Zecca, M. Basato, J. Mol. Catal. A 2001, 173, 249.
[23] For
a general overview on metal nanoclusters andtheir
applications in catalysis, see: a) J. D. Aiken III, R. G. Finke, J.
Mol. Catal. A 1999, 145, 1; b) J. S. Bradley, E. Hill, M. E.
Leonowicz, H. Witzke, J. Mol. Catal. 1987, 41, 59; c) H.
Bꢀnnemann, R. M. Richards, Eur. J. Org. Chem. 2001, 2455.
[24] For a recent review on the application of metal nanoclusters in
20 h (pH increases to 3.5) to yielda homogeneous andstable 0.5
m
boehmite sol. The dip sol was prepared by mixing 30 mL of the
boehmite sol with 20.0 mL of a solution of poly(vinyl alcohol) (PVA,
Fluka, 98% purity, M = 72000 gmolÀ1) made by adding 150 g of
0.05m HNO3 in water to 4.5 g of PVA andstirring for 2 h at 80 8C. The
a-alumina supports were dip-coated with the g-alumina sol under
À
C C bond-formation reactions, see: M. Moreno-Maꢁas, R.
Pleixats, Acc. Chem. Res. 2003, 36, 638.
class 1000 clean-room conditions at a dip-coating speed of 1.1 cmsÀ1
.
[25] M. B. Thathagar, J. Beckers, G. Rothenberg, J. Am. Chem. Soc.
2002, 124, 11858.
The coatedsupports were driedfor 3 h (25
8C, relative humidity
40%), andthen calcinedfor 1 h (600 8C, air, heating/cooling rates
0.58CminÀ1). The deposition/calcination cycle was repeated. This
resultedin a flat nanoporous g-alumina layer with a layer thickness of
about 3 mm.
Heck coupling: Identical 1.5:1 mixtures of 4 (3.0 mmol, 0.38 g)
and 2 (2.0 mmol, 0.40 g) in 50 mL of DMF were placedon both sides
of the membrane reactor. NaOAc (3.0 mmol, 0.41 g) was then added
to side B, andthe Pdcluster suspension (3 mL, 10 m m, 1.0 mol%) was
added to side A. The reactor was heatedto 100 8C, andsamples were
taken from both compartments andanalyzedby GC (pentadecane
internal standard). After each experiment, the membrane was washed
with acetone (4 ” 10 mL) andextractedfor 24 h with EtOH.
[26] A. Gniewek, A. M. Trzeciak, J. J. Ziolkowski, L. Kepinski, J.
Wrzyszcz, W. Tylus, J. Catal. 2005, 229, 332.
[27] R. Narayanan, M. A. El-Sayed, Langmuir 2005, 21, 2027.
[28] M. Beller, H. Fischer, K. Kuhlein, C. P. Reisinger, W. A.
Herrmann, J. Organomet. Chem. 1996, 520, 257.
[29] M. T. Reetz, G. Lohmer, Chem. Commun. 1996, 1921.
[30] M. T. Reetz, R. Breinbauer, K. Wanninger, Tetrahedron Lett.
1996, 37, 4499.
[31] Y. Li, X. M. Hong, D. M. Collard, M. A. El-Sayed, Org. Lett.
2000, 2, 2385.
[32] A. R. Gholap, K. Venkatesan, R. Pasricha, T. Daniel, R. J.
Lahoti, K. V. Srinivasan, J. Org. Chem. 2005, 70, 4869.
[33] a) F. Y. Zhao, B. M. Bhanage, M. Shirai, M. Arai, Chem. Eur. J.
2000, 6, 843; b) R. L. Augustine, S. T. Oleary, J. Mol. Catal. A
1995, 95, 277; c) A. Biffis, M. Zecca, M. Basato, J. Mol. Catal. A
2001, 173, 249; d) B. M. Choudary, S. Madhi, N. S. Chowdari,
M. L. Kantam, B. Sreedhar, J. Am. Chem. Soc. 2002, 124, 14127;
e) S. S. Prockl, W. Kleist, M. A. Gruber, K. Kꢀhler, Angew.
Chem. 2004, 116, 1917; Angew. Chem. Int. Ed. 2004, 43, 1881;
f) K. Kꢀhler, R. G. Heidenreich, J. G. E. Krauter, M. Pietsch,
Chem. Eur. J. 2002, 8, 622.
Received: December 5, 2005
Revised: February 3, 2006
Publishedonline: March 23, 2006
Keywords: cross-coupling · Heck reaction ·
.
homogeneous catalysis · nanoparticles · palladium
[34] For clarity, we use here the term “leaching”, which is common in
this context. In practice, oxidative addition is most likely the first
event leading to detachment of PdII ions from the cluster.
[35] J. A. Widegren, R. G. Finke, J. Mol. Catal. A 2003, 198, 317.
[36] A. Biffis, M. Zecca, M. Basato, Eur. J. Inorg. Chem. 2001, 1131.
[37] Y. Li, E. Boone, M. A. El-Sayed, Langmuir 2002, 18, 4921.
[38] J. Le Bars, U. Specht, J. S. Bradley, D. G. Blackmond, Langmuir
1999, 15, 7621.
[39] M. B. Thathagar, P. J. Kooyman, R. Boerleider, E. Jansen, C. J.
Elsevier, G. Rothenberg, Adv. Synth. Catal. 2005, 347, 1965.
[40] C. C. Cassol, A. P. Umpierre, G. Machado, S. I. Wolke, J.
Dupont, J. Am. Chem. Soc. 2005, 127, 3298.
[41] K. B. Jirage, J. C. Hulteen, C. R. Martin, Science 1997, 278, 655.
[42] S. R. Chowdhury, J. E. ten Elshof, N. E. Benes, K. Keizer,
Desalination 2002, 144, 41.
[43] J. S. Bradley, B. Tesche, W. Busser, M. Maase, M. T. Reetz, J. Am.
Chem. Soc. 2000, 122, 4631.
[44] Permporometry is a characterization technique for determining
pore size distributions in membranes. It is based on the
controlledblocking of the pores by capillary condensation and
the simultaneous measurement of the inert-gas diffusion flux
through the remaining open pores. When a condensable vapor
(e.g., cyclohexane) is introduced at low vapor pressure, first a
molecular adsorption layer (the t layer) is formed on the inner
surface of the pores. The pore size (d [nm]) is defined by d =
2(rK+t), where rK is the Kelvin radius, and t the layer thickness
[1] T. Mizoroki, K. Mori, A. Ozaki, Bull. Chem. Soc. Jpn. 1971, 44,
581.
[2] V. Farina, Adv. Synth. Catal. 2004, 346, 1553.
[3] N. Miyaura, T. Yanagi, A. Suzuki, Synth. Commun. 1981, 11, 513.
[4] F. Bellina, A. Carpita, R. Rossi, Synthesis 2004, 2419.
[5] K. Sonogashira, Y. Tohda, N. Hagihara, Tetrahedron Lett. 1975,
16, 4467.
[6] P. Siemsen, R. C. Livingston, F. Diederich, Angew. Chem. 2000,
112, 2740; Angew. Chem. Int. Ed. 2000, 39, 2633.
[7] Metal-Catalyzed Cross-Coupling Reactions (Eds.: F. Diederich,
P. J. Stang), Wiley-VCH, Weinheim, 1997.
[8] Transition Metals for Organic Synthesis; Building Blocks and
Fine Chemicals (Ed.: M. Beller), Wiley-VCH, Weinheim, 2004.
[9] A. Zapf, M. Beller, Top. Catal. 2002, 19, 101.
[10] R. B. Bedford, C. S. J. Cazin, D. Holder, Coord. Chem. Rev. 2004,
248, 2283.
[11] D. H. Valentine, J. H. Hillhouse, Synthesis 2003, 2437.
[12] Q. W. Yao, E. P. Kinney, Z. Yang, J. Org. Chem. 2003, 68, 7528.
[13] Y. Mori, M. Seki, J. Org. Chem. 2003, 68, 1571.
[14] A. Alimardanov, L. Schmieder-van de Vondervoort, A. H. M.
de Vries, J. G. de Vries, Adv. Synth. Catal. 2004, 346, 1812.
[15] A. H. M. de Vries, J. Mulders, J. H. M. Mommers, H. J. W.
Henderickx, J. G. de Vries, Org. Lett. 2003, 5, 3285.
[16] A. H. M. de Vries, F. J. Parlevliet, L. Schmieder-van de Vonder-
voort, J. H. M. Mommers, H. J. W. Henderickx, M. A. M. Walet,
J. G. de Vries, Adv. Synth. Catal. 2002, 344, 996.
Angew. Chem. Int. Ed. 2006, 45, 2886 –2890
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 2889

Communications
(ca. 0.3 nm). For more details, see: a) A. Mey-Marom, M. G.
Katz, J. Memb. Sci. 1986, 27, 119; b) M. G. Katz, G. Baruch,
Desalination 1986, 58, 199.
[45] N. Benes, A. Nijmeijer, H. Verweij in Recent Advances in Gas
Separation by Microporous Ceramic Membranes (Ed.: N. K.
Kanellopoulos), Elsevier, Amsterdam, 2000.
[46] One can quantify the number of moles of diffused 6 on the
clusters side by comparing the diffusion rates. However, very
small amounts of 6 form since in the initial stage of the Heck
reaction the concentration gradient is very low. This results in
lower diffusion rates of 6 (3.5 mol% after 100 h) comparedto
those observedin the control experiments with pure
(4.5 mol% after 100 h, see Figure 3).
6
[47] S. R. Chowdhury, K. Keizer, J. E. ten Elshof, D. H. A. Blank,
Langmuir 2004, 20, 4548.
[48] J. G. de Vries, Dalton Trans. 2006, 421.
[49] For a detailed overview of this type of synthesis, see: N. Benes,
A. Nijmeijer, H. Verweij in Recent Advances in Gas Separation
by Microporous Ceramic Membranes (Ed.: N. K. Kanellopou-
los), Elsevier, Amsterdam, 2000, pp. 335 – 372.
2890 www.angewandte.org
ꢀ 2006 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2006, 45, 2886 –2890