With phosphinophosphonic acids to nanostructured, water-soluble, and catalytically active rhodium clusters

Article abstract of DOI:10.1002/anie.200603402

(Chemical Equation Presented) Pressure-sized clusters: Aqueous solutions of rhodium complexes stabilized by phosphinophosphonic acid ligands turn ink-black under a hydrogen atmosphere. Depending on the hydrogen pressure applied, rhodium clusters of 2-6 nm

Full text of DOI:10.1002/anie.200603402

Communications
DOI: 10.1002/anie.200603402
Cluster Catalysis
With Phosphinophosphonic Acids to Nanostructured, Water-Soluble,
and Catalytically Active Rhodium Clusters
Julija Glöckler, Stefan Klützke, Wolfgang Meyer-Zaika, Armin Reller, F.Javier García-García,
Hans-Henning Strehblow, Petra Keller, Eva Rentschler, and Wolfgang Kläui*
Nanostructured metal clusters[1] are of great interest as
catalysts for organic as well as inorganic reactions and
especially as electrocatalysts for fuel cells. As components
for materials with special electronic, optical, or magnetic
properties they are regarded as a promising basis for entirely
new branches of industry.
mean diameter is in the range 2–6 nm, depending on the
reaction conditions. These clusters are water soluble, stable in
air, stable in aqueous solution for months, and can be isolated
as redispersible solids. The aqueous cluster solutions show
high activity in the biphasic hydrogenation of olefins and are
able to hydrogenate aromatic compounds. Complexes with
sulfonated phosphane ligands are predominantly used as
water-soluble rhodium catalysts.[6] We observed that the
introduction of phosphonate substituents (PO32ꢀ) leads to
significantly higher water solubility than the introduction of
sulfonate groups (SO3ꢀ). For example, the phosphane ligand
Na2[Ph2PCH2CH2PO3] (Na2-1) with just one phosphonate
group is as water-soluble (ca. 1100 gL)ꢀ1 as tppts, a triphe-
nylphosphine with three sulfonate groups. The synthesis of
Na2-1 is outlined in Scheme 1.
Metal clusters in the range of 1–4 nm are of special
interest since this range covers the continuous transition of
non-metallic compounds to the bulk metal.[1d,e] In this range
the particles show size-dependent catalytic properties. The
smaller the metal particles are, the higher the ratio of surface
atoms to the total number of atoms. Therefore, small clusters
often show especially high catalytic activity.[1f,2] Among other
features, the ratio of the different types of surface atoms
changes: with increasing particle size the proportion of
terrace atoms increases whereas the proportion of atoms on
edges (step atoms) and especially on vertices (kink atoms)
decreases.[3] Monodisperse metal clusters in the range of 1–
4 nm are ideally suited for a detailed analysis of the size
dependence and selectivity of catalytic reactions. However,
small clusters of this size tend to aggregate owing to their high
specific surface and can generally only be stabilized in
aqueous solution in the presence of surfactants or poly-
mers.[1g,2,4,5] We have found a route leading to rhodium
clusters with a very narrow particle size distribution. Their
[*] Dipl.-Ing. J. Glöckler, Dr. S. Klützke, Prof. Dr. W. Kläui
Institut für Anorganische Chemie und Strukturchemie
Lehrstuhl I: Bioanorganische Chemie und Katalyse
Heinrich-Heine-Universität Düsseldorf
Universitätsstrasse 1, 40225 Düsseldorf (Germany)
Fax : (+49)8112287
E-mail: klaeui@uni-duesseldorf.de
Homepage: http://www.chemie.uni-duesseldorf.de/Faecher/
Anorganische_Chemie/AC1/Klaeui/
Dr. W. Meyer-Zaika
Institut für Anorganische Chemie
Universität Duisburg-Essen, 45117 Essen (Germany)
Prof. Dr. A. Reller, Dr. F. J. García-García
Lehrstuhl für Festkörperchemie, Universität Augsburg
Universitätsstrasse 1, 86159 Augsburg (Germany)
Scheme 1. Outline of the route that leads to phosphinophophonic
acid-stabilized rhodium clusters (represented, in this case, as a five-
shell “magic number” cluster consisting of 561 rhodium atoms).
Prof. Dr. H.-H. Strehblow, Dr. P. Keller
Institut für Physikalische Chemie und Elektrochemie 2
Heinrich-Heine-Universität Düsseldorf
Universitätsstrasse 1, 40225 Düsseldorf (Germany)
Upon reaction of [{Rh(cod)Cl}2] (cod = cyclooctadiene)
with Na2-1, the water-soluble rhodium(I) complex 2 is formed
with the phosphane ligand 1 acting as P,O-chelate ligand.
Because of the low affinity of rhodium for oxygen-donor
ligands we assumed that 1 could act as “hybrid ligand”,[7]
“hemilabile ligand”,[8] or as “dangling ligand”[9] and therefore
Prof. Dr. E. Rentschler
Institut für Anorganische Chemie und Analytische Chemie
Johannes-Gutenberg-Universität Mainz
Duesbergweg 10–14, 55128 Mainz (Germany)
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
1164
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 1164 –1167

Angewandte
Chemie
would be ideally suited to provide an easily accessible
coordination site in a catalytic cycle. However, preliminary
experiments did not show any reaction of 2 with olefins in the
presence of hydrogen gas. Instead we observed that after
3 days an aqueous solution of the rhodium complex 2 turned
ink-black under atmospheric hydrogen pressure at room
temperature. No rhodium metal precipitated, no rhodium
mirror formed on the glass wall. A black solid is isolated upon
evaporation of the solvent and can be redissolved in water
yielding again a black solution. At higher hydrogen pressure
in an autoclave black solutions are obtained after just a few
hours. High-resolution TEM (HRTEM) images (Figure 1–3)
show that these solutions contain rhodium clusters of
Figure 3. TEM image (left) of rhodium nanocrystals (2.5 nm) sup-
ported on amorphous carbon film and particle size distribution (right)
of the clusters: (2.5ꢁ0.4) nm.
remarkable narrow particle size distribution. The mean
diameter depends upon the hydrogen pressure applied in
the synthesis. At atmospheric pressure the mean diameter is
6 nm (Figure 1), at 30 bar 3 nm (Figure 2a–c), and at 60 bar of
hydrogen 2.5 nm (Figure 3). A diameter of approximately
2.4 nm corresponds to a five-shell “magic number” cluster of
561 rhodium atoms.
We tried to further characterize the clusters. It seems
reasonable to assume that ligands 12ꢀ located on the surface of
the clusters prevent coagulation through strong electrostatic
repulsion.[10] Interestingly, the 31P NMR spectrum of an
aqueous cluster solution shows no signal for excess non-
coordinated ligand 12ꢀ but exclusively one for the corre-
sponding phosphane oxide [Ph2P(O)CH2CH2PO3]2ꢀ. It is
clear that the phosphane ligand is oxidized upon cluster
formation despite strict exclusion of oxygen and despite the
presence of a hydrogen atmosphere! Patin et al. investigated
the reduction of RhCl3·3H2O by tppts and showed that water
is the source of oxygen in the phosphane oxidation reac-
tion.[11]
Figure 1. TEM image (left) of rhodium nanocrystals (6 nm) supported
on amorphous carbon film and particle size distribution (right) of the
clusters: (6.2ꢁ0.4) nm.
Why in our case the phosphane ligand is converted into
the phosphane oxide is not yet clear. Even if the reduction of
the rhodium(I) complex 2 to give elemental rhodium is
coupled to a stoichiometric oxidation of the phosphane ligand
12ꢀ, excess free phosphane ligand should still be observed. It is
also remarkable that no phosphane ligands residing on the
cluster surfaces are detected in the 31P NMR spectrum. We
therefore took a closer look at the magnetic properties of the
clusters. The calculations and measurements made on the
paramagnetism of “naked” clusters are contradictory. To our
knowledge, no measurements on ligand-stabilized discrete
clusters have been performed to date. In contrast to some 3d
elements, no spontaneous magnetic ordering is observed with
4d or 5d elements. The magnetic behavior of the heavier
homologues of the 3d elements, however, is determined by
high spin-orbit coupling contributions. Thus such elements—
once magnetized—could represent an especially interesting
class of compounds with high magneto-crystalline anisotropy.
Rhodium shows extraordinary magnetic properties within this
group of elements. In the solid state it is nonmagnetic, yet in
mono-atomic layers and discrete Rhn clusters (n = 6–43) it
shows astonishingly high magnetic moments. Depending on
cluster size, magnetic moments of more than 1 mB per atom
are reported, for example, for n = 13.[12] A consistent inter-
pretation of the experimental data and theoretical calcula-
Figure 2. a) HRTEM image of rhodium nanocrystals supported on
amorphous carbon film. The image shows that the nanocrystals do not
agglomerate. b) HRTEM image (left) of a single 3-nm rhodium nano-
crystal and particle size distribution (right) of the clusters:
(3.1ꢁ0.4) nm. The measured distances between the crystallographic
planes are in accord with the Rh–Rh distances reported for rhodium
metal. c) HRTEM image of a rhodium nanocrystal oriented in the [111]
direction. In this case as well, the measured distances between
contrast maxima and minima can be interpreted in terms of the
structural data of metallic rhodium.
Angew. Chem. Int. Ed. 2007, 46, 1164 –1167
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1165

Communications
tions has not yet been reached owing to experimental
elevated pressure. The results show that under the same
conditions small clusters exhibit significantly higher catalytic
activity than lager ones. The turn-over frequency (TOF) of
the hydrogenation of 1-hexene catalyzed by 6-nm clusters at
20 bar of hydrogen is about 2100 hꢀ1. Under the same
conditions, 3-nm clusters catalyze the hydrogenation of 1-
hexene at a TOF of 3600 hꢀ1. The same trend is also seen in
the hydrogenation of benzene. Nitrobenzene is reduced to
aniline (TOF = 600 hꢀ1 at 608C, 80 bar H2-pressure, 3-nm
clusters).
Herein we have shown that the water-soluble rhodium(I)
complex 2 can be reduced by hydrogen. In this reaction,
rhodium clusters of narrow particle size distribution are
formed which can be isolated and are redispersible in water
without coagulation. The outstandingly high stability is
caused by the electrostatic repulsion of the double negative
charge of the phosphonate ligands covering the clusters. The
mechanistic role of the phosphinophosphonic acid ligands in
the reduction of rhodium is not fully understood, since more
phosphane is oxidized to phosphane oxide than would be
needed for the formal reduction of rhodium(I) to rhodium(0).
The elucidation of the reaction mechanism and stabilization
of water-soluble clusters of other metals by phosphinoposph-
onates are subject of current research.
difficulties. The 2-nm diameter rhodium clusters with their
narrow particle size distribution that we have obtained could
be suitable particles for studying magnetic properties. None of
the samples, however, gave an indication of superparamag-
netic behavior. The temperature dependence of the measured
magnetic moments is dominated by diamagnetic contribu-
tions and contributions from spin-orbit coupling (see Sup-
porting Information). In the group of clusters investigated the
solid-state character clearly dominates over the surface
properties. We are now looking for ways to get ligand-
stabilized clusters in the subnanometer range.
The clusters presented herein are air stable and their
catalytic hydrogenation activity is virtually independent of
the presence of air. To date we have no information on the
oxidation state of the rhodium surface atoms and whether
oxide ions are present on the cluster surface. The character-
ization of cluster surfaces is a general problem which is very
often not addressed. X-ray photoelectron spectroscopy (XPS)
is a suitable method for analyzing surfaces. XPS provides
information on the oxidation states of the detected elements
when suitable reference standards are available. We used
metallic rhodium and RhCl3·3H2O as references und deter-
mined the position of the Rh 3d5/2 line (see Supporting
Information). The energies determined are in good agree-
ment with the values found in the literature. The Rh 3d5/2 line
of 6-nm rhodium clusters was determined to be at EB =
308.2 eV, it is shifted by DE = 1 eV relative to metallic
rhodium. Clearly, the electronic structure of clusters, even
of this size, is different from that of rhodium metal.[13] The
reason for not being able to get any distinct information on
the surface atoms of the clusters using XPS lies in the
dimensions of the clusters. A wavelength of l = 3.29 nm is
calculated for the photoelectrons of the Rh 3d5/2 signal with
Experimental Section
1H and 31P NMR spectra were recorded on a Bruker spectrometer
DRX200, 13C NMR spectra on a Bruker DRX500. For HRTEM
imaging electron microscopes JEOL 2000F and Philips TEM CEM-
200 FEG were available. In the Figures we have given the mean
values and standard deviations of the cluster diameters. The HRTEM
experiments were carried out using an acceleration voltage of 200 kV.
The element specific analyses were performed using a built-in EDX
system (EDAX). The images were taken using a CCD camera
(Gatan). Susceptibilities were measured in the temperature range 4 to
pffiffiffiffiffiffiffiffi
[14]
E
kin = 1179.2 eV by using the relation l = 0.096 Ekin
.
This
result means that XPS does not characterize the cluster
surface but measures the electronic state of the whole cluster!
The HRTEM and electron diffraction data show that the
rhodium clusters virtually exist as single crystals. The
interplanar spacing according to the contrast maxima and
electron diffraction diagrams are in accord with the structural
data for metallic rhodium. Experiments using locally resolved
element-specific energy-dispersive X-ray (EDX) analysis
(EDAX system) give evidence for the presence of a phos-
phorus-containing ligand shell surrounding the rhodium
nanoclusters.
300 K using
a MPMS-XL5-SQUID magnetometer (Quantum
Design), the applied magnetic field was 1 T. Data were corrected
for the diamagnetic contribution of the sample holder. XPS measure-
ments were performed on an ESCALAB 200x. AlKa radiation (hn =
1486.6 eV) was used as excitation source.
The synthesis of the phosphane ligand was carried out under a dry
nitrogen atmosphere using Schlenk technique according to the
procedure outlined by Roundhill and co-workers.[15] The detailed
synthesis procedure as well as NMR spectroscopic data are given in
the Supporting Information. The reaction vessels used for cluster
synthesis were cleaned using aqua regia to eliminate any metallic
impurities.
2: A solution of Na2-1·2H2O (0.30 g, 0.80 mmol) in methanol
(20 mL) was added to a stirred suspension of [{Rh(cod)Cl}2] (0.20 g,
0.40 mmol) in methanol (20 mL). The mixture was stirred for 20 min.
The rhodium complex dissolves and NaCl precipitates causing a slight
clouding. The precipitate was removed by filtration using a glass frit.
The resulting solution was evaporated in vacuo to yield 0.40 g (90%)
of a powdery yellow solid. 31P{1H}-NMR (81 MHz, D2O): [ABX]-spin
The data on the catalytic activity of the clusters is shown in
Table 1. All the experiments on biphasic catalysis were
performed in a stainless steel autoclave with glass insert at
Table 1: Hydrogenation of 1-hexene and benzene at 208C using rhodium
clusters of varying size as catalysts.
system d = 23.3 (3JP, P = 11.4 Hz, PV), 22.2 ppm (3JP, P = 11.4 Hz, PIII
;
1-Hexene
H2 pressure
[bar]
Benzene
H2 pressure
[bar]
1JP, Rh = 154.8 Hz, PIII); H NMR (500 MHz, CD3OD): d = 7.8–7.3 (m,
10H; C6H5), 5.5–5.0 (m, 2H; CH trans to P), 2.9–2.7 (m, 2H; CH trans
to O), 2.5–1.7 ppm (m, 12H; CH2); IR (KBr): n˜ = 3421br, 3054w,
2918w, 1632br, 1483s, 1435m, 1159s, 1099m, 1041s, 744m, 696 cmꢀ1 s;
MS (FAB+, NBA): m/z 527 (73.6, [M+H]+), 549 (7.5% [M+Na]+.
Rhodium cluster (6 nm): 2 (0.16 g, 0.30 mmol) was dissolved in
water (15 mL). The solution is stirred at ambient temperature for 2–
1
Cluster size
[nm]
TOF
[hꢀ1
TOF
]
[hꢀ1
]
6.2
3.1
2.0
20
20
20
2100
3600
5700
20
20
20
1
50
80
1166
www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 1164 –1167

Angewandte
Chemie
3 days under atmospheric hydrogen pressure. The initially yellow
solution turned brown after a few hours and then black. The rhodium
clusters were isolated as a black solid after evaporation of the
solution. To remove excess ligands, the black solid is resuspended in
ethanol (10 mL) by using an ultrasonic bath. Over night the clusters
precipitate. After centrifugation and drying the rhodium clusters were
isolated as a black shiny powder which readily dissolved in water. This
work-up procedure was repeated until no more ligand signals were
seen in the 31P NMR spectrum of the centrifugate.
Coord.Chem. 1991, 24, 43 – 67; d) P. Kalck, F. Monteil, Adv.
Organomet.Chem. 1992, 34, 219 – 284; e) W. A. Herrmann,
C. W. Kohlpaintner, Angew.Chem. 1993, 105, 1588 – 1609;
Angew.Chem.Int.Ed.Engl. 1993, 32, 1524 – 1544.
[7] J. Podlahova, B. Kratochvil, V. Langer, Inorg.Chem. 1981, 20,
2160 – 2164.
[8] J. C. Jeffrey, T. B. Rauchfuss, Inorg.Chem. 1979, 18, 2658 – 2666.
[9] O. Krampe, C.-E. Song, W. Kläui, Organometallics 1993, 12,
4949 – 4954.
[10] As expected on the basis of the pKa values of the phosphonic
acid, the clusters precipitate when the pH value of the solution
sinks below 6. The clusters also coagulate upon addition of tap
water. When edta (ethylenediamine tetraacetate) solution is
added, the clusters redissolve. Phoshonic acids form sparingly
soluble salts with divalent cations, such as magnesium and
calcium.
Received: August 19, 2006
Published online: December 27, 2006
Keywords: biphasic catalysis · clusters · phosphonic acid ·
.
rhodium · transmission electron microscopy
[11] C. Larpent, R. Dabard, H. Patin, Inorg.Chem. 1987, 26, 2922 –
2924.
[1] a) Nanoparticles.From Theory to Application (Ed.: G. Schmid),
Wiley-VCH, Weinheim, 2003; b) Clusters and Colloids: From
Theory to Application (Ed.: G. Schmid), VCH, Weinheim, 1994;
c) B. F. G. Johnson, Coord.Chem.Rev. 1999, 190–192, 1269 –
1285; d) G. Schmid, Chem.Rev. 1992, 92, 1709 – 1727; e) G.
Schmid, Mater.Chem.Phys. 1991, 29, 133 – 142; f) B. C. Gates,
Chem.Rev. 1995, 95, 511 – 522; g) A. Roucoux, J. Schulz, H.
Patin, Chem.Rev. 2002, 102, 3757 – 3778.
[12] a) S. Burnet, T. Yonezawa, J. J. van der Klink, J.Phys.Condens.
Matter 2002, 14, 7135 – 7145; b) C. Barreteau, R. Guirado-López,
D. D. Spanjaard, M. C. Desjonqu›res, A. M. OlØs, Phys.Rev.B
2000, 61, 7781 – 7794; c) Z.-Q. Li, J.-Z. Yu, K. Ohno, Y. Kawazoe,
J.Phys.Condens.Matter
1995, 7, 47 – 53; d) A. J. Cox, J. G.
Louderback, S. E. Apsel, L. A. Bloomfield, Phys.Rev.B 1994,
49, 12295 – 12298.
[13] For a discussion upon the various factors determining XPS
energies of metallic nanoparticles see I. Lopez-Salido, D. C. Lim,
R. Dietsche, N. Bertram, Y. D. Kim, J.Phys.Chem.B 2006, 110,
1128 – 1136, and references therein.
[14] M. P. Seah, W. A. Dench, Surf.Interface Anal. 1979, 1, 2 – 11.
[15] S. Ganguly, J. T. Mague, D. M. Roundhill, Inorg.Chem. 1992, 31,
3500 – 3501.
[2] L. N. Lewis, Chem.Rev. 1993, 93, 2693 – 2730.
[3] W. F. Maier, Angew.Chem. 1989, 101, 135 – 146; Angew.Chem.
Int.Ed.Engl. 1989, 28, 135 – 145, and references therein.
[4] J. Schulz, A. Roucoux, H. Patin, Chem.Commun. 1999, 535 –
536.
[5] H. Bönnemann, W. Brijoux, Adv.Catal.Nanostruct.Mater. 1996,
165 – 196, and references therein.
[6] a) E. G. Kuntz, Chemtech 1987, 17, 570 – 575; b) T. G. Southern,
Polyhedron 1989, 8, 407 – 413; c) M. Barton, J. D. Atwood, J.
Angew. Chem. Int. Ed. 2007, 46, 1164 –1167
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
1167