Selective butadiene hydrogenation by Pd nanoparticles deposed onto nano-sized oxide supports by CVD of Pd-hexafluoroacetylacetonate

Article abstract of DOI:10.1016/j.ica.2011.10.057

The organometallics chemical vapor deposition (OM-CVD) technique, using Pd(hfac)₂ as a precursor, was employed for the preparation of a series of heterogeneous catalysts based on Pd nanoparticles supported on nanometric oxides (CeO₂, ZnO and TiO₂). The properties of metal nanoparticles were tuned by proper pretreatment of the oxidic support. Complete decomposition of Pd precursor during formation of the metal phase by hydrogen reduction was investigated by TPRD-mass spectrometry analysis, and the obtained nanoparticles were characterized by CO-DRIFT spectroscopy and HR-TEM microscopy. Butadiene selective hydrogenation to butene, a well known structure-sensitive reaction, was used to test the performances of the systems prepared. Pd/CeO₂ catalyst showed the best selectivity to butene with the lowest amount of undesired butane, result ascribable to the morphology of palladium nanoparticles.

Full text of DOI:10.1016/j.ica.2011.10.057

Inorganica Chimica Acta 380 (2012) 216–222
Contents lists available at SciVerse ScienceDirect
Inorganica Chimica Acta
journal homepage: www.elsevier.com/locate/ica
Selective butadiene hydrogenation by Pd nanoparticles deposed onto
nano-sized oxide supports by CVD of Pd-hexafluoroacetylacetonate
b
a
a
Vladimiro Dal Santo a, , Alessandro Gallo , Alberto Naldoni , Laura Sordelli
⇑
a CNR – Istituto di Scienze e Tecnologie Molecolari, Via C. Golgi 19, Milano 20133, Italy
b CNR – Istituto di Scienze e Tecnologie Molecolari, Via Fantoli 16/15, Milano 20128, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 29 October 2011
The organometallics chemical vapor deposition (OM-CVD) technique, using Pd(hfac)2 as a precursor, was
employed for the preparation of a series of heterogeneous catalysts based on Pd nanoparticles supported
on nanometric oxides (CeO2, ZnO and TiO2). The properties of metal nanoparticles were tuned by proper
pretreatment of the oxidic support.
Young Investigator Award Special Issue
Complete decomposition of Pd precursor during formation of the metal phase by hydrogen reduction
was investigated by TPRD-mass spectrometry analysis, and the obtained nanoparticles were character-
ized by CO-DRIFT spectroscopy and HR-TEM microscopy.
Butadiene selective hydrogenation to butene, a well known structure-sensitive reaction, was used to
test the performances of the systems prepared.
Keywords:
Palladium
Cerium oxide
Zinc oxide
Titanium dioxide
1,3-Butadiene hydrogenation
Nanoparticles
Pd/CeO2 catalyst showed the best selectivity to butene with the lowest amount of undesired butane,
result ascribable to the morphology of palladium nanoparticles.
Ó 2011 Elsevier B.V. All rights reserved.
1. Introduction
they were in the past, some of them being largely used in the
microelectronics industry [6].
In the field of heterogeneous catalysis operated by supported
metal nanoparticles (hereinafter referred to as NPs) it is well estab-
lished that the catalytic properties exerted by NPs are size depen-
dent [1], and moreover, when structure sensitive reactions are
considered, performances, as activity and selectivity to desired
products, are strongly dependent on the morphology of supported
NPs [2].
We have recently reported that CVD allows the easy tuning of
structural properties of metal nanoparticles [7] and that can be
successfully applied to the preparation of NPs supported on oxidic
materials with different properties [8]. In particular, multistep
methodology provides supported well formed, crystalline, homo-
geneously dispersed nanoparticles that show superior catalytic
performances if compared to ‘‘standard’’ catalysts [9].
Thus, size, composition, morphology and crystal structure of the
metal NPs can largely influence the rate of several elementary
steps and control the selectivity to desired products, as well as to
undesired ones.
Tuning the shape, i.e. the ratio between edges, corners and dif-
ferent exposed crystal planes (facets) and keeping the size in the
nanometric range is one of the key point in the development of
supported NPs based catalysts.
Chemical vapor deposition (CVD) techniques [3] can be fruit-
fully exploited for the preparation of ad hoc materials, as they nor-
mally yield quite narrow particle size distribution and cleaner
precursor decomposition pathways [4]. CVD based methodologies
have found many applications in the field of nanoparticles and
films deposition on powder supports [5]. Volatile metal precursors
are nowadays products commercially available and less exotic than
Partial hydrogenation of dienes and alkynes is a reaction of
great industrial importance [10]. Butadiene is an impurity in C4 al-
kenes produced by cracking and butane/butadiene mixtures have
to be purified before butene transformations, such as polymeriza-
tion or alkylation processes.
Butene purification is actually performed by selective hydroge-
nation of butadiene. A good process will consequently work in such
a way that the complete hydrogenation of butadiene into butenes
takes place, while avoiding any butane formation. Palladium is still
the reference catalytic metal for butadiene hydrogenation. How-
ever, it has been shown that its activity and selectivity can be
strongly influenced by many factors like the metal dispersion,
the nature of the support, and the preparation method. The
improvement of selectivity still remains an important issue.
Butadiene selective hydrogenation is sensitive to the surface
structure; the hydrogenation rate has been found to be more rapid
with the more open (111) Pd face than with the close-packed
(100) surface [11].
⇑
Corresponding author. Tel.: +39 02 50314428; fax: +39 02 50314405.
E-mail address: v.dalsanto@istm.cnr.it (V. Dal Santo).
0020-1693/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2011.10.057

V. Dal Santo et al. / Inorganica Chimica Acta 380 (2012) 216–222
217
For these reasons this transformation can be considered as a
good test reaction due to its sensitivity to the size and the shape
of supported metal nanoparticles and as an important reaction in
its industrial applications at the same time.
Recently was reported the catalysts design, i.e. the optimized
preparation, of palladium supported nanoparticles exposing the
desired (111) faces, however such nanoparticles were in the tens
of nanometers dimensional range [12].
formed under H2 5% by vol. in He (flow = 20 mL minÀ1) with a
heating rate of 5 °C minÀ1 from RT to 550 °C. The samples (20 mg
ca., diluted in quartz powder) were held on the glass frit of a U-
shaped Pyrex reactor and selected mass channels (HF (m/z = 20);
methane (m/z = 15); H2O (m/z = 18); CF3 (m/z = 69); 2-propanone
(m/z = 58); CF3CO fragment (m/z = 97); CO (m/z = 28); CO2 (m/
z = 44); CF3COCH3 (m/z = 112); 2,4-pentandione (m/z = 200)) were
monitored by an on-line mass spectrometer.
Here, we wish to report about the development of Pd catalysts by
Pd(hfac)2 OM-CVD as an easy preparation method for palladium
nanoparticles, falling in the nanometric dimensional range, sup-
ported on different nanometric oxides, namely ZnO, CeO2 and
TiO2. Appropriate thermal pretreatments were used to tune the sur-
face properties of the oxides with the aim of control the dispersion
and, in some extent, the morphology of the supported Pd NPs. Sur-
face and structural properties of the NPs were then characterized
by means of infrared spectroscopy of adsorbed CO, a typical surface
sensitive technique and by HR-TEM, which is instead a bulk tech-
nique. The catalysts were then tested in butadiene traces selective
hydrogenation, a typical surface-sensitive reaction, in order to re-
veal the different properties of supported metal nanoparticles
(dimension, morphology, i.e. exposed facets, etc.). Finally some
structure-properties relationships were drawn and discussed.
CO chemisorption experiments were made according the fol-
lowing procedure: before chemisorption, the samples (20 mg ca.
deposed on a KBr bed) were reduced in situ at 300 °C for 30 min
under 20 mL of hydrogen flow. A mass spectrometer-monitored
pulsed CO chemisorption was performed at RT until the saturation
of carbonyl adsorbed on Pd bands was reached (injected volume
50
l
L, CO 50%, Ar 50%) [8].
The morphology and size distribution of the supported metal
particles were evaluated by HR-TEM micrographs. The powder
samples have been further grinded and dispersed in isopropanol
in an ultrasound bath. A drop of the suspension has been deposited
on the carbon grid which, after solvent evaporation under vacuum,
has been inserted in the column of the JEOL Jem-2010 EX high res-
olution transmission electron microscope.
Pictures were taken at 250,000–800,000Â magnifications, span-
ning wide regions of several support grains in order to provide a
well representative map of the catalyst system.
2. Experimental
Distribution histograms of metal particles numbers and vol-
umes versus diameters were evaluated from about 300 to 550
counts per sample.
2.1. Materials
The nanometric oxides (Nanoactive™ TiO2, ZnO, and CeO2) were
used as received (NanoScale Materials Inc.). Palladium hexafluoro-
acetylacetonate (hereinafter denoted as Pd(hfac)2) was purchased
from Strem Chemicals and used as received. All the gases were pur-
chased from SIAD S.p.A. (Bergamo, Italy) and were research grade
(at least 99.995% purity), gases were used as received, only He
was purified by passing through a gas-purifier (Supelco).
2.2.1. 1,3-Butadiene hydrogenation catalytic tests
Catalysts, 10 mg ca., dispersed in 450 mg quartz powder, were
loaded in a U-shaped Pyrex reactor, and reduced at 300 °C under
H2 flow for 1 h (H2 flow 20 mL/min, heating rate 5 °C/min). Catalytic
tests were conducted on the reduced catalysts by feeding a reaction
mixture of 1000 ppm of H2 and 500 ppm of 1,3-butadiene diluted in
He. The reaction was monitored from RT to 400 °C (heating rate 5 °C/
min) by an online quadrupole mass spectrometer (HPR-20 QIC gas
analysis mass spectrometer system/Hiden Analytical Ltd.) con-
nected downstream to the reactor. The QMS was calibrated with a
mixture of 1,3-butadiene, 500 ppm; 1-butene, 500 pmm; n-butane,
500 ppm diluted in He. Since MS does not allow to distinguish the
three butene isomers (1-butene, cis-2-butene and trans-2-butene)
hereinafter we will generally refer to ‘‘butene’’.
2.2. Catalysts preparation and characterization
Around 500 mg of support (TiO2, ZnO, CeO2) were calcined 1 h
at 300 °C in O2 flow, and then cooled down to RT in a inert gas flow.
These materials were denoted as TiO2_, ZnO_, CeO2_A samples.
On the other hand TiO2_, ZnO_, CeO2_B samples were simply
evacuated (turbo-molecular pump and liquid N2 trap, final vacuum
10À5 mbar) at RT for 2 h in order to remove most of physisorbed
water.
3. Results and discussion
Specific surface area after heat treatments was measured by
single point method nitrogen physisorption under flow conditions
with a PulseChemisorb2700 apparatus (Micromeritics).
CVD of Pd(hfac)2 was conducted in static mode as already re-
ported elsewhere [8]: the appropriate amount of Pd precursor
(namely 24.5 mg) was thoroughly mixed, by a magnetic stirring
bar, with the activated supports (0.5 g) in a Schlenk flask under
Ar atmosphere at RT, then the sealed flask was kept at 80 °C over-
night in a oven.
All the samples were then transferred in a U-shaped Pyrex reac-
tor and reduced at 300 °C under H2 flow (flow of 50 mL minÀ1) for
1 h, then cooled down to RT in Ar flow (50 mL minÀ1).
Pd loading was determined by ICP-MS (X Series II, Thermo Elec-
tron) after microwave digestion of reduced catalysts in 3:1 HCl/
HNO3 mixture. The measurements largely confirmed the nominal
1% wt. load (real Pd loadings are comprised between 0.9% and
1.1% wt.).
3.1. Catalysts preparation
Two series of three supported Pd catalysts, one on activated
TiO2, ZnO and CeO2 supports (A series) and one on non-activates
oxidic supports (B series), were prepared by the OM-CVD of the
volatile Pd precursor Pd(hfac)2. A general scheme of the experi-
mental set-up has been described elsewhere [13].
The OM-CVD was carried out at 80 °C overnight, this tempera-
ture was chosen in order to ensure the complete sublimation of
Pd precursor avoiding any residue to be left in the support powder.
As a result of the interaction (adsorption) of Pd precursor the sup-
port materials originally white (TiO2 and ZnO, whereas CeO2 is pale
yellow) turned to a vivid yellow color. Finally, the adsorbed Pd pre-
cursor on oxide powder was converted in supported metal nano-
particles by a reductive treatment at 300 °C under H2 flow and at
the end of this treatment all the catalysts turned to gray color
due to the formation of supported Pd NPs.
TPRD analysis and CO chemisorption tests were performed
using a home-made reactor stand/DRIFTS reaction chamber. Appa-
ratus details are described elsewhere [13]. TPRD analysis was per-
This preparation method was simple and effective, since the use
of solvents and operations such as filtration, washings typical of

218
V. Dal Santo et al. / Inorganica Chimica Acta 380 (2012) 216–222
Table 1
Selected properties of Pd catalysts, oxide pretreatment temperature and Pd loading
(measured by ICP-MS).
Catalyst
Pretreatment T of support (°C)
Pd loading (wt.%)
Pd/TiO2_A
Pd/ZnO_A
Pd/CeO2_A
Pd/TiO2_B
Pd/ZnO_B
Pd/CeO2_B
300
300
300
RT
RT
RT
1.0
0.9
1.1
1.1
1.0
0.9
the preparations by traditional ‘‘wet methods’’ were avoided. The
final properties of the catalysts obtained are summarized in
Table 1.
3.2. TPRD characterization
The decomposition of adsorbed Pd(hfac)2 was investigated by
TPRD experiments, by means of an on-line QMS, in order to follow
the evolution of adsorbed molecules and of gas phase decomposi-
tion products. The aim was to find out temperature and conditions
suitable in order to obtain supported palladium metal particles
starting from the adsorbed Pd precursor.
QMS analysis of evolved gas products (see Fig. 1, only the
decomposition profiles of Pd(hfac)2/ZnO_A was reported as an
example) shows, as the main feature, a series of intense peaks of
several decomposition products fragments located between 200
and 300 °C. More in detail the following peaks were evident: meth-
ane (m/z = 15) and CF3 fragment (m/z = 69) show two peaks at 205
and 265 °C; 2-propanone (m/z = 58) and CF3CO fragment (m/z = 97)
show only one peak at 265 °C. CO (m/z = 28) and CO2 (m/z = 44)
evolution occurs with a very intense peak located respectively at
225 and 219 °C with weak shoulders at 270 °C. Minor peaks can
be detected at lower temperature (<150 °C) during the initial
stages of ligand decomposition. It should be stressed that above
300 °C all the mass channel profiles monitored became flat and
the evolution of any decomposition products is totally absent, thus
confirming that ligands decomposition results to be complete just
below this temperature. For these reasons the reduction under
hydrogen flow at 300 °C is an appropriate method to obtain clean
supported Pd metal (nano)particles, since no residues of the hfac
ligand are left on support and/or metal surface that can be further
eliminated at higher temperatures. Usually, in such cases, an evo-
Fig. 2. HR-TEM micrograph and particle size distribution histogram of Pd/TiO2_A
sample.
lution of methane at higher temperatures, resulting from hydroge-
nation of carbon containing adsorbed compounds, occurs [14].
3.3. HR-TEM characterization
Samples of the A series, which are expected to have the smaller
Pd NPs, were analyzed by HR-TEM microscopy.
The registered micrographs show that the support oxidic mate-
rials are present as aggregates of nanometric particles with differ-
ent dimension and structure: titania consists in crystalline grains
with irregular shape and a mean diameter of 10 nm; ZnO is again
in the form of agglomerated crystalline grains but of spherical
shape and mean diameter of 13 nm; finally cerium oxide is in
the form of agglomerates of well formed nanocrystallites with
mean diameter as small as 5 nm.
Supported Pd NPs fall in the nanometer range for all the inves-
tigated samples and, more in detail, Pd/TiO2_A sample shows the
presence of Pd NPs with a mean diameter of 1.2 nm (see Fig. 2),
whereas in Pd/ZO_A sample NPs are even smaller having a mean
diameter of 0.8 nm (see Fig. 3). For Pd/CeO2_A sample Pd NPs were
not detectable by standard HR-TEM measurements, because of the
high electron density of CeO2 as already reported [15].
3.4. CO-DRIFTS characterization
Infrared spectroscopy of adsorbed CO is a powerful technique
for the identification of almost all the metal surface sites present
on the catalysts surface: CO adsorbed on Pd0 metallic particles or
on Pdn+ oxidized sites show distinctive absorption bands whose
Fig. 1. TPRD profiles of Pd(hfac)2/ZnO_A sample; curve with open squares: m/
z = 15(CH3); curve with open triangles: m/z = 58 (CH3CO); curve with open circles:
m/z = 69 (CF3); black curve with no markers: m/z = 97 (CF3CO).

V. Dal Santo et al. / Inorganica Chimica Acta 380 (2012) 216–222
219
Table 2
General attribution of absorption bands of CO adsorbed on Pd sites.
Frequency
Assignment
Reference
(cmÀ1
)
2080
2060
1980–1994
1864–1936
1967–1977
<1960
Linear CO on exposed (111) Pd faces
Linear CO on Pd defects
[16]
[17]
Twofold bridged CO on (100) Pd faces
Twofold bridged CO on (111) Pd faces
Twofold bridged CO on (100) Pd faces
Hollow bonded CO(111) Pd faces
Threefold hollow and bridge bonded CO(111) Pd
faces
1932
2091
2085
1986
1937
Linear CO on defects Pd sites
CO linearly bonded to Pd atoms
Bridge-bonded CO on Pd (100) planes
Bridge-bonded CO on Pd (111) planes
[18]
second band is located at 1973 cmÀ1 and it is typical of CO ad-
sorbed on highly defective Pd (111) planes, as commonly found
in TiO2 supported palladium sites.
The third band, located at 1898 cmÀ1, starts to develop only
after the third CO pulse and can be ascribed to three-folded CO.
The first two bands undergo a small blue shift to higher wavenum-
bers increasing the CO coverage (from the 1st to the 5th CO pulse)
passing, respectively, from 2107 to 2112 cmÀ1 and from 1963 to
1973 cmÀ1
.
Pd/TiO2_B sample shows similar bands (see Fig. 4(b)), with some
exceptions: the high frequency band is red-shifted to 2067 cmÀ1
,
and low frequency bands are more convoluted, but two peaks can
be recognized and are located at 1934 and 1870 cmÀ1. It should be
noted that CO absorption bands in this sample do not suffer any
relevant change during CO pulses (wavenumbers shifts and/or
intensity changes). Such data agree withthe presenceof Pd nanopar-
ticles with more extended planes (i.e. larger nanoparticles) where
lateral interaction between adsorbed CO are less important, result-
ing in negligible frequency shifts of the bands [19].
Fig. 3. HR-TEM micrograph and particle size distribution histogram of Pd/ZnO_A
sample.
Ceria supported Palladium NPs show a more complex CO
absorption bands pattern: high frequency band is broader and in
one sample two components are clearly present, also in the low
frequency region at least three or four bands are present.
Pd/CeO2_A sample shows a peculiar behavior in the high fre-
quency region (see Fig. 5(a)): at low CO coverage (1st CO pulse)
an intense band located at 2045 cmÀ1 readily appears and progres-
sively blue-shifts to higher wavenumbers up to 2067 cmÀ1. In the
meanwhile a shoulder, located at 2042 cmÀ1, becomes apparent
at high CO coverage (after 4th pulse).
In the low frequency region of two- to three-folded adsorbed CO
two bands are present, one located at 1922 cmÀ1 ascribable to two-
folded CO and another one at 1845 cmÀ1 ascribable to three-folded
CO. The 1745 cmÀ1 band can be attributed to bridged carbonates
adsorbed on oxidic support [20]. According to the literature [21]
the band at higher wavenumbers can be attributed to linear CO
on Pd nanoparticles and the band located at 2042 cmÀ1 to multi-
bonded CO on Pd atoms [22,23] that are formed only at high
coverage.
Pd/CeO2_B sample shows a quite different bands profile (see
Fig. 5(b)): the high frequency band shifts from 2061 (1st CO pulse)
to 2081 cmÀ1 (4th CO pulse), while the low frequency ones are
strongly convoluted and at least two components of similar inten-
sity can be found and are located at 1945 cmÀ1 and around
1865 cmÀ1. The different ratio between the CO adsorbed species
with respect to those found for Pd/CeO2_A sample accounts for a
different morphology of palladium nanoparticles (i.e. different
relative amounts of Pd sites of different exposed planes, steps, cor-
ners and/or different particles size). In particular the lower ratio
(two-folded CO)/(three-folded CO) can be related to the presence
attribution can be easily performed according to the existing liter-
ature as reported in Table 2 [16–18].
The exposed faces and low coordination sites (corners and
edges) of metal nanoparticles, and also their defectivity can be de-
rived from the analysis of the pattern (bands shape and wavenum-
bers) of adsorbed CO spectra.
As a general consideration, in the spectra of CO adsorbed on pal-
ladium nanoparticles, four main components can be revealed and
they can be assigned on the basis of the existing literature (passing
from high to low wavenumbers): (a) linear carbonyls on Pd(111)
faces and/or on defects; (b) twofold bridged carbonyls on
Pd(100) faces; (c) twofold bridged carbonyls on Pd(111) faces;
(d) threefold bridged carbonyls on Pd(111) faces.
Due to the different pretreatments of the oxidic supports, the de-
gree of interaction between Pd precursor and the support can vary
between the A and B series, and, in turn, also the final structural
and morphological properties of supported Pd nanoparticles are ex-
pected to be different. A similar effect was reported for Pd/MgO cat-
alysts, prepared by impregnation of Pd(acac)2 of differently
pretreated magnesia. In that case Pd nanoparticles on activated
MgO showed uniform size distribution in the sub-nanometer range,
whilst the nanoparticles supported on non-activated magnesia were
larger and with a broader size distribution [14].
The HR-TEM characterization, reported above, substantially
confirms the presence of (sub)-nanometric Pd NPs in the samples
obtained from activated oxidic support materials (the A series).
Pd/TiO2_A shows three main bands (see Fig. 4 (a)), two of them
appearing just after the first CO pulse: one, very sharp and intense,
is located at 2112 cmÀ1 and can be ascribed to terminal CO the

220
V. Dal Santo et al. / Inorganica Chimica Acta 380 (2012) 216–222
Fig. 4. CO-DRIFTS spectra of Pd/TiO2_A (a) and Pd/TiO2_B sample (b) recorded after pulses of CO (numbers indicate the corresponding nth pulse).
Fig. 5. CO-DRIFTS spectra of Pd/CeO2_A (a) and Pd/CeO2_B (b) sample recorded after pulses of CO (numbers indicate the corresponding nth pulse).
of less extended exposed planes. Finally the general shift to higher
wavenumbers of the whole band system accounts for a more
electropositive character of the supported nanoparticles (less
back-donation from metal to CO anti-bonding orbitals) or can be
explained by geometrical effects (less dipole–dipole coupling due
to larger particles).
When measuring ZnO supported samples a strong absorption
was noticed over the whole spectral range after in situ reduction
of samples, also resulting in a low S/N ratio. In particular for Pd/
ZnO_B sample to record spectra of absorbed CO was not possible,
due to the very low S/N ratio and/or to the low intensity of CO
absorption bands.
Pd/ZnO_A sample showed three bands (see Fig. 6): linear CO
one located at 2073 cmÀ1 and the second broad one, centered at
1886 cmÀ1 falling in the region of three-folded CO. The high fre-
quency band can be attributed to linear CO on Pd(111) faces more
than to CO on defective Pd sites, since the low wavenumbers band
falls in the region typical of twofold bridged CO on (111) Pd faces
[16].
CO(linear) to CO(bridged) ratio can be considered an important
parameter for the estimation of particles dispersion and/or mor-
phologies in supported metal nanoparticles of different metals
[9,17,24] and in general a good agreement between CO-FTIR and
HRTEM data was found.
Fig. 6. CO-DRIFTS spectra of Pd/ZnO_A recorded after pulses of CO (numbers
indicate the corresponding nth pulse).
CO(lin.)/CO(br.) ratio higher than the corresponding ceria based A
sample, thus suggesting a higher dispersion and/or a different
shape of supported palladium nanoparticles. Since the more
dispersed supported metal nanoparticles are usually the ones
derived from precursor adsorbed on activated oxide [14], the
CO-DRIFTS data suggest a different shape of Pd NPs.
It should be stressed that, due to the different kind of bands
present on the different supports, a rigorous comparison between
all the catalysts cannot be performed. However, it is possible to
compare A and B samples, supported on the same oxide, drawing
some reliable conclusions.
In our case A series samples show, in general, the higher
CO(lin.)/CO(br.) ratio accounting for more dispersed Pd nanoparti-
cles. The only exception is Pd/CeO2_B sample which shows a
3.5. Butadiene hydrogenation performances
Butadiene hydrogenation is a consecutive reaction that, accord-
ing to reaction conditions (temperature, butadiene and hydrogen
partial pressures, ratio, etc.) can show very different behavior in
terms of conversion and selectivity to butene (the selective hydro-
genation product) or to butane (the total, unselective, hydrogena-
tion byproduct) (see Scheme 1). However, if reaction parameters

V. Dal Santo et al. / Inorganica Chimica Acta 380 (2012) 216–222
221
butadiene selective hydrogenation: almost all of them are active
at low temperatures (namely 50 °C) and butene selectivity is al-
ways higher than 50% (see Fig. 7).
Hydrogenation performances versus reaction temperature
show a general trend for all the catalysts studied: (i) a sudden in-
crease of butadiene conversion at low temperatures, above 100 °C
almost total conversion is reached by all systems investigated; (ii)
a low selectivity to butene (i.e. total hydrogenation to butene
occurring) around 100 °C followed by an increase in butene selec-
tivity that reaches 100% above 200 °C.
+
+
H2
H2
butane
1,3-butadiene
butene
Scheme 1. 1,3-Butadiene hydrogenation reaction scheme.
The different structural and, in some cases, morphological prop-
erties of the supported Pd NPs influence the selectivity of the
hydrogenation reaction.
In general A series samples show higher selectivity towards
butene, the selective hydrogenation product, coupled to low
amounts of butane, the unselective hydrogenation product, at least
at low reaction temperatures (below 100 °C). On the other hand B
are kept constant, as in our case, butadiene hydrogenation can be
considered a typical surface sensitive reaction and the selectivity
to butene can mainly be related to palladium nanoparticles
morphology (i.e. exposed planes) [12].
During our butadiene hydrogenation tests all the A and B series
catalysts synthesized show good activity and selectivity in
Fig. 7. Selected catalytic performances at different reaction temperatures of Pd/TiO2_A (a); Pd/TiO2_B (b); Pd/CeO2_A (c); Pd/CeO2_B (d); Pd/ZnO_A (e); Pd/ ZnO_B (f).
Butadiene conversion: very dark gray bar; butene selectivity: light gray bar; butane selectivity: dark gray bar.

222
V. Dal Santo et al. / Inorganica Chimica Acta 380 (2012) 216–222
series catalysts show lower butene selectivities and corresponding
higher butane selectivities at the same reaction temperatures. This
difference in the catalytic behavior of the two catalysts series can
be explained by the structural/morphological evidences obtained
from the characterization data. The higher butene selectivities ob-
served for the A series systems are in good agreement with the
presence of more dispersed Pd nanoparticles in A series samples
with respect to the B series ones. Indeed CO-DRIFTS and HRTEM
data revealed the presence of highly dispersed palladium nanopar-
ticles in A series catalysts as the result of the better interaction be-
tween Pd precursor and the activated oxide supports [14].
Moreover, the high butene selectivity can be related also to the
morphology of palladium nanoparticles exposing the (111) faces
that were reported to be very selective in butadiene hydrogenation
to butene [12]. In our case the presence of Pd NPs exposing the
(111) faces was clearly revealed for Pd/TiO2_A and Pd/ZnO_A sam-
ples by CO-DRIFT spectra reported in Figs. 4(a) and 6. Namely these
two catalysts are the ones showing the higher selectivity.
The only exception to this general picture was represented by
Pd/CeO2_B catalyst, that keeps high selectivity to butene (always
higher than 90%) over all the temperature range investigated with
a butane selectivity never exceeding 10%. This peculiar behavior
can be ascribed to the unique morphology (i.e. exposed planes),
among the samples here reported, of palladium nanoparticles.
Also Pd/ZnO_A shows comparable performances, being more
active at low temperature but with a slightly lower selectivity to
butene (only 73% at 100 °C).
exceptions. Indeed Pd/CeO2_B sample shows a selectivity to butene
higher than 90% in a wide range of reaction temperatures that was
ascribable to the peculiar morphology of the Pd NPs.
Acknowledgements
Dr. B. Rizzo is gratefully acknowledged for part of the experi-
mental work. A.G. gratefully acknowledges support from Regione
Lombardia, Project ‘‘ACCORDO QUADRO Regione Lombardia e
CNR per l’attuazione di programmi di ricerca e sviluppo’’. A.N.
thanks the Italian Ministry of Education, University and Research
for financial support through the Project ‘‘ItalNanoNet’’ (Rete
Nazionale di Ricerca sulle Nanoscienze; Prot. No. RBPR05JH2P).
References
[1] J.P. Wilcoxon, B.L. Abrams, Chem. Soc. Rev. 35 (2006) 1162.
[2] F. Klasovsky, P. Claus, in: B. Corain, G. Schmid, N. Toshima (Eds.), Metal
Nanoclusters in Catalysis and Materials Science: The Issue of Size Control,
Elsevier, Boston, 2008, pp. 167–182.
[3] P. Serp, P. Kalck, R. Feurer, Materials. Chem. Rev. 102 (2002) 3085.
[4] C. Vahlas, B. Caussat, P. Serp, G.N. Angelopoulos, Mater. Sci. Eng., R 53 (2006) 1.
[5] S. Watano, H. Nakamura, K. Hamada, Y. Wakamatsu, Y. Tanabe, R.N. Dave, R.
Pfeffer, Powder Technol. 141 (2004) 172.
[6] A.C. Jones, H.C. Aspinall, P.R. Chalker, in: A.C. Jones, M.L. Hitchman (Eds.),
Chemical Vapour Deposition, RSC Publishing, 2009, pp. 357–412.
[7] V. Dal Santo, A. Gallo, V. De Grandi, R. Psaro, L. Sordelli, S. Recchia, J. Mater.
Chem. 19 (2009) 9030.
[8] V. Dal Santo, C. Mondelli, V. De Grandi, A. Gallo, S. Recchia, L. Sordelli, R. Psaro,
Appl. Catal. A: Gen. 346 (2008) 126.
[9] A. Beretta, A. Donazzi, G. Groppi, P. Forzatti, V. Dal Santo, L. Sordelli, V. De
Grandi, R. Psaro, Appl. Catal. B: Environ. 83 (2008) 96.
[10] J.P. Boitiaux, P. Cosyns, M. Derrien, G. Leger, Hydrocarbon Proc. 64 (1985) 51.
[11] J. Silvestre-Albero, G. Rupprechter, H.J. Freund, J. Catal. 240 (2006) 58.
[12] L. Piccolo, A. Valcarcel, M. Bausach, C. Thomazeau, D. Uzio, G. Berhault, PCCP 10
(36) (2008) 5504.
[13] V. Dal Santo, C. Dossi, A. Fusi, R. Psaro, C. Mondelli, S. Recchia, Talanta 66 (3)
(2005) 674.
[14] V. Dal Santo, L. Sordelli, C. Dossi, S. Recchia, E. Fonda, G. Vlaic, R. Psaro, J. Catal.
198 (2) (2001) 296.
Finally the increase of reaction temperature above 300 °C (data
not shown) leads to a decrease in butadiene conversion as well as
butene selectivity and can be originated from the sintering of Pd
nanoparticles, generating less selective low-index planes, as al-
ready reported for Pd catalyst reduced at high temperatures
[25,26].
[15] J. Gonzalez, J.C. Hernandez, M. Lopez-Haro, E. del Rıo, J.J. Delgado, A.B. Hungrıa,
S. Trasobares, S. Bernal, P.A. Midgley, J.J. Calvino, Angew. Chem., Int. Ed. 48
(2009) 5313.
[16] G. Agostini, R. Pellegrini, G. Leofanti, L. Bertinetti, S. Bertarione, E. Groppo, A.
Zecchina, C. Lamberti, J. Phys. Chem. C 113 (2009) 10485.
[17] F. Di Gregorio, L. Bisson, T. Armaroli, C. Verdon, L. Lemaitre, C. Thomazeau,
Appl. Catal. A: Gen. 352 (2009) 50.
[18] Y. Li, B.W.L. Jang, Appl. Catal. A: Gen. 392 (2011) 173.
[19] P. Hollins, Surf. Sci. Rep. 16 (1992) 51.
[20] C. Li, Y. Sakata, T. Arai, K. Domen, K. Maruya, T. Onishi, J. Chem. Soc., Faraday
Trans. 1 (85) (1989) 929.
[21] J.A. Wang, J.M. Dominguez, A. Montoya, S. Castillo, J. Navarrete, M. Moran-
Pineda, J. Reyes-Gasga, X. Bokhimi, Chem. Mater. 14 (2002) 4676.
[22] F.M. Hoffmann, Surf. Sci. Rep. 3 (1983) 107.
[23] A. Arteaga, F.M. Hoffmann, A.M. Bradshaw, Surf. Sci. 119 (1982) 79.
[24] L.L. Sheu, W.M.H. Sachtler, J. Mol. Catal. 81 (1993) 267.
[25] J. Goetz, M.A. Volpe, R. Touroude, J. Catal. 164 (1996) 369.
[26] A. Sarkany, Z. Zsoldos, B. Furlong, J.W. Hightower, L. Guczi, J. Catal. 141 (2)
(1993) 566.
4. Conclusions
Pd catalysts supported on different oxides, namely TiO2, CeO2
and ZnO, prepared by CVD of Pd(hfac)2 precursor show some valu-
able properties for heterogeneous catalysts like high dispersion of
metallic phase, easiness of preparation and generally high activity
in the desired reaction (1,3-butadiene hydrogenation). In addition
the proper choice of support and the tuning of its surface proper-
ties allow one to control the structural and surface properties of
the supported Pd NPs. Catalysts obtained from pretreated supports
show the presence of highly dispersed NPs; on the contrary, in non
pretreated supports larger NPs were present. Moreover, it was also
possible to tune the morphology of NPs (the exposed planes). The
catalytic behavior was found to depend mainly on these two
structural parameters. In particular A series samples showed high-
er selectivity to butene with respect to B series ones with some