Structure and deformations of Pd-Ni core-shell nanoparticles

Article abstract of DOI:10.1021/jp040473i

Homogeneous collections of Pd-Ni core-shell nanoparticles have been prepared by decomposition of metal-organic compounds and studied by several electron microscopy techniques: transmission electron microscopy (TEM), energy-dispersive X-ray spectroscopy (E

Full text of DOI:10.1021/jp040473i

342
J. Phys. Chem. B 2005, 109, 342-347
Structure and Deformations of Pd-Ni Core-Shell Nanoparticles
S. Sao-Joao,† S. Giorgio,*,† J. M. Penisson,§ C. Chapon,† S. Bourgeois,‡ and C. Henry†
CRMCN-CNRS, Campus de Luminy, Case 913, 13288 Marseille Ce´dex 9, France, UMR5613 CNRS/UniVersite´
de Bourgogne, UFR Sciences et Techniques, BP 47870 Dijon Ce´dex, France, and DRFM/SP2M/ME Bat C5,
435, CEA-G, 17 Rue des Martyrs, 38054 Grenoble Ce´dex 9, France
ReceiVed: June 30, 2004
Homogeneous collections of Pd-Ni core-shell nanoparticles have been prepared by decomposition of metal-
organic compounds and studied by several electron microscopy techniques: transmission electron microscopy
(TEM), energy-dispersive X-ray spectroscopy (EDS), high-resolution transmission electron microscopy
(HRTEM), energy-filtered microscopy (EFTEM), and by X-ray photoelectron spectroscopy (XPS). The physical
and chemical properties of the Pd shell are supposed to depend on its electronic properties, which are influenced
by the presence of the Ni core and by the deformation in the Pd lattice. Here, the interfacial structure of
Pd/Ni and the lattice deformations in the core and the shell are studied in detail. The catalytic properties of
the pure metal and the bimetallic particles, toward CO oxidation, have been investigated.
1. Introduction
interfaces Pd/Ni and Ni/MgO are investigated by HRTEM and
the coexistence of Ni2+/Ni by XPS. The activity of these
particles toward CO oxidation has been studied as well as that
of pure Pd and Ni clusters.
Bimetallic particles with the core-shell structure represent
a new class of materials to improve or tune the catalytic activity
or selectivity of catalytic reactions. The tuning of the catalytic
properties is connected to the variation of the electronic structure
of the surface atoms which are influenced by core and shell
atoms.1-5 Indeed, at the particle surface, the heteronuclear
metal-metal bonds induce a shift of the valence band center
in the first monolayers. Aside from this effect, variations in the
coordination number of the surface atoms and strains also move
the valence band center. In a core-shell cluster, the outside
layer can be different from the bulk structure because of the
influence of the core and also its small size. A very large number
of publications exists in the domain of core-shell clusters from
colloid solutions. We have already prepared Pd-Cu core-shell
clusters.6
The reactivity of surfaces of Pd-Ni systems has been
investigated toward the hydrogenation of butadiene, in the case
of the (110) surface of the Pd8Ni92(110) alloy7 and in the case
of thin Pd layers deposited on a Ni(110) surface.8,9 Four
monolayers of Pd have been found to be 40 times more active
than a Pd(110) single crystal, which shows the highest activity
for this reaction.
From a comparison between these studies and ab initio
calculations,5 it was concluded that the catalytic properties of
the Pd-Ni bimetallic system is related both to the changes in
the electronic properties and in the atomic structure. Complex
structures were evidenced. In extended Ni/Pd interfaces, periodic
vacancies are produced from the stress induced by Ni.10
The morphology and structure at the interface of Ni particles
or thin Ni layers supported on MgO have been studied by
grazing incidence X-ray scattering.11
2. Preparation Techniques, Composition of Pd-Ni
Particles
Pure Ni particles were prepared by the decomposition of
Ni(acac)2 on MgO oxide powders, then the Pd shells were grown
on the Ni particle surfaces from solution, following the
previously described techniques.11,12 The Ni(acac)2 (Fluka, 40-
90 mg) was dissolved in acetonitrile (200 mL). MgO clean
powders (Fluka, 200 mg) with a surface specific area of 30 m2
g-1 were mixed into the solution and stirred at 80 °C until
complete evaporation of the solvent. After drying at 80 °C in
air, the powders were slowly annealed in a vacuum (10-4 Torr),
until 450 °C, then reduced under H2 at 400 Torr during 3 h at
the same temperature. At this stage, the core was prepared.
Different amounts (8-50 mg) of Pd(acac)2 (Aldrich) were
dissolved in toluene (200 mL). The oxide powders with Ni
particles on their surfaces were mixed with the solution and
heated under a flux of bubbling H2 until complete evaporation
of the solvent. In the solution at high temperature (until 119
°C) and with H2, the decomposition of Pd(acac)2 occurs, and
then, the Pd atoms directly crystallize on the Ni clusters. If the
temperature is too low, and without H2 in solution, the
decomposition of Pd(acac)2 does not occur, and then, the
compound is directly adsorbed on MgO, and Pd crystallizes
separately from Ni. Then, the samples were annealed once again
in a vacuum (10-4 Torr) until 450 °C before alloy formation.
The particles were analyzed by EDS in an electron micro-
scope (JEOL 2000 FX). Their composition was determined
either on isolated clusters in convergent beam mode or on
collections of clusters in standard analysis. For structural studies,
other microscopes were used (JEOL 4000EX and JEOL 3010).
The experimental Ni loading in Ni /MgO clusters, measured
by EDS, is plotted in Figure 1 as a function of the mass of
Ni(acac)2 for 200 mg of clean MgO. If we compare the
experimental Ni loading to the ratio of Ni over MgO mass
concentration in the solution, we conclude that about one-half
In this paper, we elaborate on Pd-Ni core-shell particles in
epitaxy on MgO microcrystals, by decomposition of Pd and Ni
acetyl acetonates (acac). The lattice deformations at the
* Corresponding author.
† CRMCN-CNRS, Campus de Luminy. Associated to Universities of Aix-
Marseille 2 and 3.
‡ UMR5613 CNRS/Universite´ de Bourgogne.
§ DRFM/SP2M/ME Bat C5.
10.1021/jp040473i CCC: $30.25 © 2005 American Chemical Society
Published on Web 12/15/2004

Pd-Ni Core-Shell Nanoparticles
J. Phys. Chem. B, Vol. 109, No. 1, 2005 343
Figure 3. Small Ni particle limited by 4 (111) faces, seen in profile
view along [110].
Figure 1. (b) Ni loading in weight in the sample, according to the
quantity of Ni(acac)2 in solution. (Ã) Ni particle size.
Figure 4. Ni particle limited by 4 (111) faces and truncated at the top
by a (001) face. The limited number of defects at the interface despite
a large misfit (17%) is certainly due to NiO formation in the first metal
layers.
of the same size.13 Here, the Ni lattice is completely accom-
modated to the MgO substrate and dilated by 17%. The distances
between the atomic columns were measured along the directions
parallel to the interface, with the MgO taken as an internal
calibration. The lattice of Ni is completely accommodated to
the lattice of MgO, at least in the first six layers from the
interface. For larger particle sizes (Figure 4), the shape is a top-
truncated half-octahedron; few local defects are seen at the
interface. The ratio between the height and the diameter is h/D
) 0.7, which is larger than in the case of small Ni particles.
If the particle has the equilibrium shape, the aspect ratio is
related to the adhesion energy between the deposit and the
substrate.
Figure 2. Pd loading in weight in the Pd/Ni/MgO sample, according
to the quantity of Ni(acac)2 in solution. The data were obtained from
Ni/MgO samples with a loading of 3% by weight.
of the Ni is lost during the impregnation (we noticed that a
colored deposit is visible on the glass vessel).
The atomic concentration of Pd in the core-shell particles
directly increases with the quantity of Pd(acac)2 in the solution.
As seen in Figure 2 for a Ni/MgO with 3% of Ni, the Pd loading
in the Pd/Ni/MgO sample is roughly proportional to the Pd-
(acac)2 concentration in the solution. Interestingly, this allows
control of the thickness of the shell.
For equivalent sizes, the observed Ni particles on MgO are
more truncated at the interface than the Pd particles.14 Moreover,
with σ100 Ni being larger than σ100 Pd,
14,15 one expects from the
Wulf Kaishev theorem less truncations for Ni particles.
In the Ni/MgO samples, by XPS, both Ni0 and NiII are clearly
detected.
The good accommodation of the metal to the substrate is
certainly due both to deformations and to the partial formation
of oxide, near the interface. Indeed, the outside lateral surfaces
of the particles are of (111) type, and the equilibrium morphol-
ogy of NiO particles of any size corresponds to cubes, just
limited by (100) faces at all the edges. During the observation
in the microscope, the pure Ni particles are transformed in NiO
despite the reducing effect of the electrons. In Figure 5a,b,c,d,
a series of images of the same 4-nm-sized particle are taken
3. Particle Size, Morphology, Composition, and
Crystalline Structures
All of the samples were observed in TEM, either by EFTEM
(with a JEOL 3010 equipped with a Gatan imaging filter) or in
HRTEM (with a JEOL 4000EX, Cs ) 0.5 mm, and a JEOL
3010, Cs ) 0.6 mm).
The size of the pure Ni particles grows with the Ni (acac)2
concentration in solution, as seen on Figure 1. For quantities
higher than 0.06 g of Ni(acac)2 for 200 mg of MgO in the
standard preparation, the size distribution of the particles
becomes much broader than for smaller depositions.
Figure 3 is an HRTEM image of a pure Ni particle of about
3 nm in size, (001) epitaxially oriented on MgO, limited by 4
(111) faces and truncated at the top by a (001) face. The ratio
between the height and the diameter is h/D ) 0.4. The shape is
close to the equilibrium shape found in the case of Pd particles
with intervals of irradiation doses of about 2 × 104 C m-2
.
In Figure 5a, the particle limited by (111) faces starts to be
truncated at the edges and the top by (100) faces. As in the
previous cases, a dislocation can be seen at the interface at the
right side, due to the misfit with MgO. The aspect ratio h/D is
about 0.57, but it is not clear at all if the particles still have the
equilibrium shape. For an oxide (NiO), the conditions to get
the equilibrium are certainly more demanding (very high

344 J. Phys. Chem. B, Vol. 109, No. 1, 2005
Sao-Joao et al.
Figure 5. (a, b, c, d) Evolution of a Ni particle during the observation in the electron beam. (e) Evolution of the misfit between the particle and
MgO near the interface, in the volume and near the top of the particles (a, b, c).
temperature) than for metals. In Figure 5b, the shape is mainly
limited by (100) faces, the dislocation still exists exactly at the
middle part of the interface, and the aspect ratio is higher than
in the preceding case. Then, in Figure 5c, the shape is only
limited by (100) faces with sharp edges, and the interface is
quasiperfect without dislocations. After that step, the shape no
longer evolves (Figure 5d), and the interface stays perfect.
The distance between the (200) lattice fringes in the particles,
perpendicular to the interface, are measured by numerical
analysis and compared to the spacing in MgO taken as the
internal calibration. The misfit between bulk Ni and MgO is
high, 17%.
irradiation of 2 × 104 C m-2, the lattice is more expanded, but
the misfit still exists with MgO, by 2% near the interface and
more than 3% in the volume. After the same irradiation doses
(c), the 2% misfit remains the same near the interface and in
the volume. The shape of the particle is that expected for a NiO
particle only exposing (100) facets. However, the lattice of the
particle is still contracted relative to bulk NiO. The misfit
between NiO and MgO is 0.7%. A possible explanation could
be that the particle is not stoichiometric.
The transformation from Ni to NiO is clearly a microscope
artifact, and this is the evidence that as soon as the clusters are
limited by (111) faces, they are not yet transformed in NiO in
the volume, even if NiO starts to grow near the interface.
It can be observed that pure NiO in an electron microscope
can be easily reduced in Ni,16 but in our case, the Ni clusters
The results are reported in Figure 5e. In (a), the (200)
distances in the cluster are dilated, but the misfit with MgO is
about 4% in all of the volume. In the same particle (b) after an

Pd-Ni Core-Shell Nanoparticles
J. Phys. Chem. B, Vol. 109, No. 1, 2005 345
Figure 7. (a) Small Pd-Ni particle seen in profile view along a [110]
direction. (b) Large Pd-Ni core-shell particle seen in profile view
along a [100] direction.
are epitaxially oriented on MgO and more influenced by the
substrate. We believe that the oxidation of Ni starts at the Ni-
MgO interface in order to minimize the interfacial energy.
The Pd/Ni core-shell particles are homogeneous in composi-
tion. The energy-filtered images (Figure 6) are able to show
the Ni core, from the peak L3, at 855 eV (Figure 6c) and the
Pd shell, from the peak M 45, at 335 eV (Figure 6b) in each
cluster seen in Figure 6a. The particle sizes in Figure 6a,b are
equal, and the particle sizes in Figure 6c are smaller. It can be
deduced that all of the large clusters are bimetallic. However,
some very small Ni clusters can be detected on the surface.
According to the XPS analysis of these samples, the Ni is found
as Ni0 and NiII.
The general morphology and crystalline structure of the Pd-
Ni core-shell particle looks like the particles seen in Figure
7a,b.
Figure 7a shows a smal Pd/Ni particle of about 3 nm. The
Pd shell is made of 5 layers. The particle is seen in a [110]
direction, parallel to the (111) lateral faces. The difference of
contrast between the center and the shell is low, because of the
small particle size, but the resolution in the image is improved
in the external parts which do not contain Ni in the projection.
Below the interface, some high-resolution fringes are seen in
the MgO with a strong contrast, which can be due to local
deformations in the substrate. Such deformations in MgO were
never observed in previous studies of Pd/ MgO, but in these
cases, the misfits were weak (8% with Pd).
Figure 7b is a larger particle with a Pd shell made of about
10 layers. The Ni core is perfectly oriented (001) on MgO. The
Pd shell is epitaxially oriented on the core. Despite the particle
size, no dislocations are visible at the interface Ni/ MgO. The
quasiperfect accommodation at the interface is once again
interpreted by the partial oxidation of Ni interface atoms.
Slight deviations of the (200) fringes normal to the interface
of metal/oxide are seen at the interface of Pd/Ni. However, these
deformations are too weak to be measured quantitatively by
numerical analysis of the HRTEM images, by the technique of
the intensity profiles.17 The lattice deformations were investi-
gated from the phase images of Figure 8a by using the geometric
phase method.18
Figure 6. (a) Bright field image of Pd/Ni core-shell particles. (b)
Energy-filtered image with the peak Pd M 45 at 335 eV. (c) Energy-
filtered image with the peak Ni L3 at 855 eV.

346 J. Phys. Chem. B, Vol. 109, No. 1, 2005
Sao-Joao et al.
Figure 9. Plot of the calculated ratio between the elastic energy and
the energy of dislocation, according to the number of Pd layers
crystallized on bulk Ni. Eel/E ) 2. (1 + νme´t/1 - νme´t)δ2‚h/[(Ln(h/b )
+ 1)/2π(1 - νme´t)], with νme´t ) Poisson coefficient of Pd (νme´t ) 0.39),
δ2 ) misfit between Pd and Ni, b ) Burgers vector ) (1/2) at Pd-
[100] or (1/2) [110], according to the direction of dislocations, and h )
thickness of the Pd layer.
simple calculation of the ratio between the elastic energy and
the energy of dislocations in Pd.19 The ratio between the elastic
energy and the energy of dislocations (Eel/Edisl) can be calcu-
lated.
Dislocations are formed when Eel/Edisl > 1. For increasing
thicknesses between 1 and 3 nm, similar to the thickness of the
Pd layer in Figure 7, this ratio is calculated and plotted in Figure
9 for both directions of dislocations [100] and [110]. It can be
seen that below 2.7 nm, which is similar to the thickness of the
Pd layers in the shells, in both directions, the elastic energy is
lower than the energy of dislocations, and then the system is
thermodynamically more stable with deformations instead of
dislocations.
Figure 8. (a, b) Pd-Ni core-shell particle (a) with the Power
spectrum, displacement along the [100] direction (b).
If the reciprocal lattice vector in the image of the shell differs
from the reference lattice vector (g) of the core by ∆g, then the
phase can be written as a function of the position: Pg(r) )
2π∆gr.
4. Catalytic Test
The catalytic activity of these preparations has been tested
on the oxidation of CO.
The displacement field u(r) is determined by combining two
sets of lattice fringes (200) and (020), shown in the power
spectrum (in Figure 8a) of the square image (Figure 8a). The
vectors a1 and a2 represent the lattice in the real space, in the
[100] and [010] directions, and then the displacement vector is
The sample of MgO powders covered with the Pd/Ni, Pd, or
Ni particles is introduced at room temperature in a quartz closed
reactor (130 cm3). After evacuation down to 10-6 mbar, CO
and O2 were introduced at the same partial pressure of 2 mbar.
The reactor is linearly heated, and the gas composition in the
reactor is continuously recorded by sampling the gas through a
leak valve connected to a UHV chamber equipped with a
quadripol mass spectrometer. The leak valve is regulated to
maintain a pressure of 1 × 10-8 mbar in the vacuum chamber.
The total pressure inside the reactor is continuously recorded
by a capacitance manometer.
In Figure 10, the evolution of the reaction rate for CO2
production is plotted as a function of time in three cases. The
estimated temperature inside the reactor is indicated along the
time axis. Three samples were tested, each containing the same
atomic amount of metal, for a total loading of 7% in weight in
u(r) ) -(1/2π)[Pg1(r)a1 + Pg2(r)a2]
The displacement field is clearly visible near the Pd/Ni interface
in Figure 8b where the phase image changes. However, the
displacement field at the level of Ni/MgO does not follow the
interface. Thus, we conclude that the Pd layer in the shell is
locally strained without dislocations and that the Ni-MgO
interface is not homogeneously distorted. The critical thickness
corresponding to the formation of misfit dislocations in a
continuous Pd layer, in epitaxy on Ni, can be estimated by a

Pd-Ni Core-Shell Nanoparticles
J. Phys. Chem. B, Vol. 109, No. 1, 2005 347
In the case of Pd/Ni nanoparticles, the coexistence of Ni and
NiO in the core allows a perfect lattice accommodation with
MgO (001) in the “cube-on-cube orientation”. The NiO forma-
tion at the interface of Ni/MgO can be explained by the ability
of Pd to dissociate O2, then by the diffusion of atomic oxygen
through the shell.
The outer shape of pure Ni particles observed on MgO(001)
is typical from better wetting than for Pd/MgO, which is
unexpected for pure Ni. The corresponding increase of the
adhesion energy is also explained by the existence of interfacial
NiO that decreases the misfit with MgO.
The adhesion energy calculated21 for various transition metals
on Mg and on O indicates a weak interaction which normally
leads to a more isotropic shape, with reentrance angles at the
interface. So, it seems reasonable to believe that NiO exists at
the interface, which increases the adhesion energy of the cluster.
For the average size (5-7 nm) of our core-shell clusters,
the catalytic activities toward CO oxidation are similar to the
case of pure Pd clusters. As Pd is on the external faces, it
represents an interesting savings of expensive metal. However,
we do not observe an enhancement of reactivity for the core-
shell clusters for the oxidation of CO, in contradiction with the
experiments on four monolayers (ML) of Pd on Ni(110) in the
case of butadiene hydrogenation.7,8 It is interesting to note that
for this reaction, no increase of reactivity was observed on PdNi
alloy particles,22 although the bulk alloys show an increase of
reactivity. It has been recently shown that the enhancement of
catalytic activity was due to a local strain in the Pd top layer in
the case of four ML of Pd on Ni(110).10 A possible explanation
for the absence of enhanced activity of small particles could be
the presence of edges. By HRTEM, it is impossible to quantity
the deformations on the first external layer.
Figure 10. Rate of CO2 production in the reactor, as a function of the
time, during heating from room temperature to 100 °C.
TABLE 1: Reaction Rate at 50% Conversion and
Maximum Reaction Temperature for Each Sample, for Two
Cycles of Reaction
Ni/MgO Ni/MgO Pd/Ni/MgO Pd/Ni/MgO
1
2
1
2
Pd/MgO
0.31
reaction rate
at 50%
0.07
88
0.24
70
conversion
(Torr/min)
maximum
reaction rate
temperature
(°C)
63
References and Notes
(1) Rodriguez, J. A; Goodman, D. W. Science 1992, 257, 897.
(2) Hammer, B.; Norskov, J. K. AdV. Catal. 2000, 45, 1.
(3) Holmblad, M.; Larsen, J. H.; Chorkendorff, I.; Pleth Nielsen, L.;
Besenbacher, F.; Stensgaard, I.; Laegsgaard, E.; Kratzer, P.; Hammer, B.;
Norskov, J. K. Catal. Lett. 1996, 40, 131.
the catalyst. Sample a contains 100% of Pd atoms, b 15% of
Pd and 85% of Ni, and c 100% of Ni.
The respective quantities used for the elaboration of each
sample are given. Sample a was elaborated with 150 mg of
MgO and 39.28 mg of Pd(acac)2. Sample b was elaborated with
150 mg of MgO, 5.55 mg of Pd(acac)2, and 28.3 mg of
Ni(acac)2. Sample c was elaborated with 150 mg of MgO and
32.9 mg of Ni(acac)2.
It is clearly seen that a and b catalysts present similar
behaviors for the oxidation of CO, while c is poorly active and
only at higher temperature. Table 1 clearly shows that deactiva-
tion does not occur in the Pd-Ni catalyst. So, it is clear that
the core-shell structure is stable during the catalytic reaction
and does not alloy in the particles.
(4) Ruban, A.; Hammer, B.; Stoltze, P.; Skriver, H. L.; Norskov, J. K.
J. Mol. Catal. A: Chem. 1997, 115, 421.
(5) Filhol, J. S.; Simon, D.; Sautet, P. Phys. ReV. B 2001, 64, 85412.
(6) Giorgio, S.; Henry, C. R. Eur. Phys. J. Appl. Phys. 2002, 20, 23.
(7) Michel, A. C.; Lianos, L.; Rousset, J. L.; Deliche`re, P.; Prakash,
N. S.; Massardier, J.; Jugnet, Y.; Bertolini, J. C. Surf. Sci. 1998, 416, 288.
(8) Hermann, P.; Guigner, J. M.; Tardy, B.; Jugnet, Y.; Simon, D.;
Bertolini, J. C. J. Catal. 1996, 163, 169.
(9) Porte, L.; Phaner-Goutorbe, M.; Guigner, J. M.; Bertolini, J. C.
Surf. Sci. 1999, 424, 262.
(10) Filhol, J. S.; Saint-Lager, M. C.; De-Santis, M.; Dolle, P.; Simon,
D.; Baudoing-Savois, R.; Bertolini, J. C.; Sautet, P. Phys. ReV. Lett. 2002,
14, 8914.
(11) Barbier, A.; Renaud, G.; Robach, O. J. Appl. Phys. 1998, 84, 4259.
(12) Giorgio, S.; Chapon, C.; Henry, C. R. Langmuir 1997, 13, 2279.
(13) Graoui, H.; Giorgio, S.; Henry, C. R. Philos. Mag. B 2001, 81,
1649.
(14) Tyson, W. R.; Miller, W. A. Surf. Sci. 1977, 62, 267.
(15) Alde´n, M.; Skriver, H. L.; Mirbt, S.; Johansson, B. Surf. Sci. 1994,
315, 157.
As the reactivities of pure Pd and Pd-Ni particles are very
close, it is evident that the use of core-shell clusters with Pd
outside allows us to save important amounts of Pd.
5. Discussion
(16) Aronin, A.; Abrosimova, G.; Bredikhin, S.; Matsuda, K.; Maeda,
K.; Awano, M. J. Ceram. Soc. Jpn. 2002, 110 (8), 722.
(17) Giorgio, S.; Chapon, C.; Henry, C. R.; Nihoul, G. Philos. Mag. B
1993, 67, 773.
(18) Hytch, M. J.; Snoeck, E.; Kilaas, R. Ultramicroscopy 1998, 74,
131.
(19) Matthews, J. W.; Jackson, D. C.; Chambers, A. Thin Solid Films
1975, 74, 129.
(20) Rellinghaus, B.; Stappert, S.; Wassermann, E. F.; Sauer, H.;
The chemical structure and geometrical deformations in Ni-
Pd core-shell particles supported on MgO have been ex situ
studied by XPS and various techniques of TEM.
XPS indicates the partial oxidation of the Ni in the particles.
This oxidized Ni is at least due to the air transfer, but the partial
oxidation of Ni atoms at the Ni/MgO interface is another
possible explanation.
After air transfer, bimetallic Ni50Pt50 catalysts and pure Ni
supported on SiO220 show the existence of NiO as observed by
HRTEM.
Spliethoff, B. Eur. Phys. J. D 2001, 16, 249.
(21) Goniakowski, J.; Noguera, C. Interface Science 2004, 12, 93.
(22) Renouprez, A.; Faudon, J. F.; Massardier, J.; Rousset J. L.;
Delichere, P.; Bergeret, G. J. Catal. 1997, 170, 181.