Hydrogenation of crotonaldehyde on different Au/CeO2 catalysts

Article abstract of DOI:10.1016/j.jcat.2007.11.018

The gas-phase hydrogenation of crotonaldehyde was carried out at 120, 150, and 180 °C over Au catalysts supported on ceria with low and medium surface areas (150 and 80 m² g^-1, respectively). An initial deactivation period was observed, followed by a steady-state regime. Ethanol was the main product in the deactivation period, whereas crotyl alcohol, butanal, butanol, and condensation products were produced under steady-state conditions. The activity and selectivity to crotyl alcohol (in the 20-32% range) were lower than those for the high-surface area ceria catalysts studied previously [B. Campo, M. Volpe, S. Ivanova, R. Touroude, J. Catal. 242 (2006) 162]. Samples were characterized by TPR, XPS, TEM, and XRD. The analysis of catalytic and characterization results indicates that gold particles supported on low- and medium-surface area ceria were relatively large, and the promotional effect of Ce³⁺ species was not achieved for the corresponding catalysts. Moreover, under reductive conditions, sintering still increased gold particle size.

Full text of DOI:10.1016/j.jcat.2007.11.018

Journal of Catalysis 254 (2008) 71–78
www.elsevier.com/locate/jcat
Hydrogenation of crotonaldehyde on different Au/CeO2 catalysts
Betiana Campo a, Corinne Petit b, María A. Volpe a,∗
a
PLAPIQUI, Camino Carrindanga km 7, 8000 Bahía Blanca, Argentina
b
LMSPC, UMR 7515 du CNRS, ECPM, ULP, 25, rue Becquerel, 67087 Strasbourg Cedex 2, France
Received 9 October 2007; revised 28 November 2007; accepted 28 November 2007
Available online 16 January 2008
Abstract
◦
The gas-phase hydrogenation of crotonaldehyde was carried out at 120, 150, and 180 C over Au catalysts supported on ceria with low and
2
−1
medium surface areas (150 and 80 m g , respectively). An initial deactivation period was observed, followed by a steady-state regime. Ethanol
was the main product in the deactivation period, whereas crotyl alcohol, butanal, butanol, and condensation products were produced under steady-
state conditions. The activity and selectivity to crotyl alcohol (in the 20–32% range) were lower than those for the high-surface area ceria catalysts
studied previously [B. Campo, M. Volpe, S. Ivanova, R. Touroude, J. Catal. 242 (2006) 162]. Samples were characterized by TPR, XPS, TEM,
and XRD. The analysis of catalytic and characterization results indicates that gold particles supported on low- and medium-surface area ceria
3+
were relatively large, and the promotional effect of Ce species was not achieved for the corresponding catalysts. Moreover, under reductive
conditions, sintering still increased gold particle size.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Gold; Ceria; Crotonaldehyde hydrogenation; Selective hydrogenation
1. Introduction
genation of the conjugated C=O bond. This property was ob-
served for catalysts with gold particle size <10 nm, however.
In addition, the support exerts a certain influence on the
selectivity of gold. Milone et al. [2,3] suggested that an elec-
tron transfer from iron oxide support to gold particles creates
electron-enriched gold particles on which the C=O bond is ac-
tivated. Regarding crotonaldehyde hydrogenation, Bailie and
Hutchings [5], working with Au/ZnO, Au/ZrO2, and Au/SiO2,
found that the selectivity to crotyl alcohol depends on the nature
of the support. For the same reaction, Okumura et al. [11] re-
ported selectivity to crotyl alcohol ranging from 10 to 25% for
Au/TiO2, Au/Al2O3, and Au/SiO2 catalysts. Because the gold
particle size of these catalysts was in a similar range (3–5 nm),
these authors concluded that the desired selectivity would be
sensitive to the selection of the support. Schimpf et al. studied
gold supported on SiO2, TiO2, and ZrO2 [8] for the hydrogena-
tion of acrolein and crotonaldehyde and reported a significant
variation in selectivity (between 23 and 43%) for the different
catalysts and they concluded that only the edges of crystallites
are active sites for the preferred C=O hydrogenation. Because
the support determines the morphology of the particles, its na-
ture strongly influences the selectivity.
The hydrogenation of a carbonyl bond in a molecule that
also contains a C=C double bond (as in α, β-unsaturated com-
pounds) is extremely difficult. Thermodynamics favors the hy-
drogenation of the C=C bond over the C=O bond by about
35 kJ/mol [1]; moreover, for kinetic reasons, the reactivity of
the C=C bond is greater than that of the C=O bond. Con-
ventional hydrogenation catalysts based on Pt, Ru, Ni, and Pd
produce mainly saturated aldehyde. The development of new
catalysts able to selectively hydrogenate the carbonyl group
is important both from a fundamental standpoint and for the
eventual synthesis of speciality chemicals. In this context, gold
catalysts have been studied.
Several research groups have reported that gold-supported
catalysts show a remarkable selectivity toward the hydrogena-
tion of the conjugated C=O bond in the hydrogenation of α, β-
unsaturated compounds [2–10]. These results demonstrate that
a peculiarity of gold is its intrinsic selectivity toward hydro-
*
Corresponding author.
E-mail address: mvolpe@plapiqui.edu.ar (M.A. Volpe).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.11.018

72
Journal of Catalysis 254 (2008) 71–78
In a previous study, some of us [12] studied the hydrogena-
Catalyst reducibility was studied by temperature-program-
med reduction (TPR) in a conventional flow system. Samples
were previously calcined at 300 ◦C and then purged with Ar
at the same temperature. Then the catalyst was cooled in Ar
and switched to an H2/Ar (5%) mixture. The temperature was
increased linearly up to 900 ◦C, with hydrogen consumption
monitored by a thermal conductivity detector.
tion of crotonaldehyde over gold supported on high-surface area
ceria (240 m2 g−1), Au/HSA–CeO2. In that work, high selectiv-
ity toward the formation of crotyl alcohol was found. The gold
particles were smaller than 4 nm, and the high selectivity value
was considered an intrinsic property of gold nanoparticles. In
addition, the redox properties of ceria promoted hydrogenation
of the C=O bond. This effect was enhanced under certain pre-
treatment conditions, due to increased selectivity. However, the
ceria acid–base characteristics gave rise to the formation of side
products ethanol and condensation products.
Another important disadvantage of Au/HAS–CeO2 is asso-
ciated with the thermal instability of the support. Any treatment
of this system at temperatures above 120 ◦C turns the catalyst
inactive. In this context, it seems logical to study gold supported
on ceria with surface area <240 m2 g−1, over which side prod-
uct formation would be diminished and thermal stability would
increase.
X-ray diffraction (XRD) was used to estimate the gold par-
ticle size from Scherrer’s analysis. XRD experiments were car-
ried out on a D5000 Siemens diffractometer using CuKα radi-
ation (λ = 0.15406 nm). The scattering intensities were mea-
sured over an angular range of 15◦ < 2θ < 65◦ for all the sam-
ples with a step size (2θ) of 0.02◦ and a step time of 2 s.
X-ray photoelectron spectroscopy (XPS) measurements
were recorded on a VG-ESCA III spectrometer with MgKα
X-rays as photon source (1253.6 eV). Spectra were obtained
for the C 1s, O 1s, Au 4f, and Ce 3d regions. For each case, the
binding energy (BE) and the area of the corresponding peaks
were measured. Due to the charging effect, the XPS peaks
were found to shift toward higher BE. For this reason, C 1s
BE (284.6 eV) was taken as an internal reference to correct
peak position. For Au, 20 scans were accumulated for increas-
ing the signal-to-noise ratio, whereas for Ce and C, 5 scans
were accumulated. The Au/Ce atomic ratio was estimated by
relating Au 4f (both peaks) and Ce 3d (all of the transitions
in the corresponding region) areas after Shirley background
subtraction and correction by mean scape depth and Scofield
cross-section [14].
In the present work, we studied gold supported on ceria with
two different surface areas, 150 and 80 m2 g−1, to further in-
vestigate the influence of the properties of ceria on the catalytic
properties of gold for C=O activation.
2. Experimental
2.1. Catalyst preparation
Ceria with different surface areas were used as supports for
gold, with 80 and 150 m2 g−1, respectively. The former was ob-
tained from the calcination of a ceria (Rhône Poulenc, Acalys
HSA) with 240 m2 g−1 surface area and 0.2 ml g−1 pore vol-
ume at 800 ◦C for 8 h. The latter was a commercial ceria from
Rhône Poulenc, with 150 m2 g−1 surface area and 0.4 ml g−1
pore volume.
The catalysts were prepared by the direct anionic-exchange
method [13]. The support was dried in air at 500 ◦C and then
mixed at 70 ◦C with an aqueous solution of HAuCl4·xH2O
(Alfa Aesar) under vigorous stirring. The concentration of the
gold precursor solution was fixed to obtain 2 wt% loaded cat-
alysts. The suspension was washed with ammonia solution
(4 mol L−1) to eliminate Cl− and then filtered and washed
with water to eliminate ammonium species. The catalysts were
then dried at 100 ◦C for 12 h and calcined under air at 300 ◦C
(1 ◦C/min) for 4 h before being stored in a sealed vessel. The
two catalysts are designated Au/CeO2150 and Au/CeO280, re-
spectively.
2.3. Catalytic test
The hydrogenation of crotonaldehyde was performed in a
glass flow microreactor operating at atmospheric pressure, as
described elsewhere [12]. Samples were submitted to two dif-
ferent in situ pretreatments: (a) reduction at 120, 300, and
500 ◦C for 1 h in H2 and (b) calcination at 300 ◦C for 1 h
with 25 cm3 min−1 of air, cooling to room temperature, and re-
duction for 1 h at 120 ◦C. Between treatments, the reactor was
exhaustively evacuated and purged with Ar.
The crotonaldehyde (Fluka puriss, 100–200 µl) was intro-
duced in a trap, set before the reactor tube, and maintained
at 0 ◦C to achieve constant crotonaldehyde partial pressure of
1.6 kPa. Two catharometers, inserted in the line at the input and
output of the reactor, respectively, measure the reactant pressure
throughout the experiment. When crotonaldehyde was injected,
the reactor was kept closed, to avoid air contamination. Ap-
proximately 20 min later, the reactor was opened, after which
the output signal immediately dropped to zero value. If no ad-
sorption occurs on the catalyst, the output signal should recover
to its initial level in about 30 s, the time it takes for gas to reach
the catharometer.
2.2. Catalyst characterization
The concentrations of gold and chlorine were determined
using atomic absorption spectroscopy (AAS) at CNRS, Vernai-
son, France. BET surface areas of the supports and the catalysts
were measured by N2 physisorption at −196 ◦C with an au-
tomatic NOVA 1200e analyzer. Before each measurement, the
samples were degassed for 1 h in helium at 100 ◦C. The cat-
alysts were characterized by transmission electron microscopy
(TEM) on a Jeol 100 CX2 (Tokyo, Japan) apparatus.
The reaction products as well as the unreacted crotonalde-
hyde were analyzed at 10-min intervals online by a gas–liquid
chromatograph equipped with a flame ionization detector (FID)
and a 30-m-long, 0.5461-mm-diameter DB-Wax column (J&W
Scientific). The condensation products were analyzed in a
Chrompack CP-Wax-58CB column connected to a FISONS

Journal of Catalysis 254 (2008) 71–78
73
2
−1
◦
Fig. 1. (a) TPR on CeO (240, 150, and 80 m g ), (b) Au/CeO 80 and Au/CeO 150 nonreduced and pre-reduced at 500 C/1 h.
2
2
2
mass spectrometer, and the hydrocarbons using a chromato-
graph equipped with an FID and a CP-SIL5CB column. Further
details of the analysis are available elsewhere [12].
The activity and selectivity toward the different products
were measured, along with time on stream (TOS). The activity
is reported in µmol converted per gram of catalyst per second
(µmol s−1 g−1). The selectivity was calculated as a ratio be-
tween a determined product and the sum of all products formed.
Fig. 1b. A strong diminished intensity of the low-temperature
peak compared with the unreduced sample can be observed. Be-
cause the textural properties of the support were not modified
by the reduction pretreatment (as measured by BET determi-
nation), the diminution of consumed hydrogen was related to
depletion of the concentration gold–ceria interphase sites due to
gold particle sintering. The same trends were observed for the
Au/CeO2150 catalyst. This result suggests that gold crystallites
undergo a sintering process after reduction at high temperature.
We attempted to determine the mean particle size for the
Au/CeO2 samples, because this feature plays an important role
in determining the catalytic pattern in the hydrogenation of α,
β-unsaturated aldehydes [5,10]. First, we conducted a TEM
study for the Au/CeO280. Although the contrast between the
support and the gold particles was rather low, the observation of
some crystals on a micrograph allowed us to estimate a metal
crystallite size of 9 nm. Moreover, to study the dependence of
gold particle size on the reduction pretreatment temperature, we
compared TEM results corresponding to the Au/CeO280 sam-
ple for the sample reduced at 120 and 500 ◦C. The particle sizes
ranged from 9 to 15 nm with increasing reduction temperature.
For the Au/CeO2150 sample, TEM images showed a very
small contrast between gold and ceria, as was reported pre-
viously for the high-surface area catalyst, Au/CeO2240 [12].
Haruta et al. failed to observe Au nanoparticles supported on
CeO2 by TEM [17]; however, in this case the support area (not
reported by the authors) likely was relatively low, considering
that the study was carried out over a model catalyst.
Taking into account that no clear morphologic characteri-
zation was obtained for the Au/CeO2150 catalyst by TEM, we
performed an XRD analysis of the samples thus formed. Table 1
reports the results corresponding to gold particle size obtained
from Scherrer’s analysis of the XRD main gold diffraction peak
(111 plane). It can be seen that gold crystals were relatively
large for both low- and medium-surface area catalysts (11 and
10 nm, respectively).
We performed an XPS study to investigate the possible sin-
tering of gold particles submitted to reduction treatment at high
temperature. The 4f XPS peaks of Au on the Au/CeO280 and
Au/CeO2150 catalysts are shown in Fig. 2. From the corrected
relation of the area of these peaks and those corresponding to
Ce 3d, we calculated the Au/Ce atomic ratio; these ratios are
reported in Table 1. A decrease in these values with increasing
reduction temperature can be seen, demonstrating unambigu-
3. Results
3.1. Characterization
The gold content was approximately the same in the Au/
CeO2150 and Au/CeO280 catalysts (1.52 and 1.53 wt%, re-
spectively) despite the difference in the specific surface area of
the support. Almost all of the chlorine species were eliminated
from the gold precursor, because the amount of Cl detected in
the solids was <200 ppm.
TPR profiles corresponding to the CeO280 and CeO2150
support (surface areas of 80 and 150 m2 g−1, respectively)
are shown in Fig. 1a. In both cases, two broad peaks of hy-
drogen consumption at 430 and 850 ◦C were observed. The
lower of these is attributed to the surface reduction of O2−
and O− anions attached to Ce4+ [15] and also to the reduc-
tion of surface Ce4+ to Ce3+. The second peak (at 850 ◦C) is
associated with reduction of the bulk oxygen and formation of
lower-oxidation state cerium cations. The consumption associ-
ated with the CeO280 sample exhibited a smaller surface than
that associated with the CeO2150 support, due to the latter’s
higher specific area. For the sake of comparison, Fig. 1a also
shows the profile corresponding to ceria with 240 m2 g−1
.
The TPR profiles of the Au/CeO280 and Au/CeO2150 cat-
alysts demonstrated the two characteristic peaks of ceria. For
both samples, the first consumption peak was shifted to a lower
temperature than in the bare support. This result was previously
reported in previous work [12] and by other groups [16]. The
low reduction peak (with maximum at 137 ◦C for Au/CeO280
and 106 ◦C for Au/CeO2150) was larger for the catalyst with
the higher surface area, as expected.
In other experiments, the Au/CeO2150 and the Au/CeO280
samples were submitted to reduction at 500 ◦C before the usual
TPR pretreatment. The profile corresponding the Au/CeO2150
catalyst previously reduced at high temperature is shown in

74
Journal of Catalysis 254 (2008) 71–78
Table 1
◦
Physicochemical properties of Au/CeO catalysts reduced at 120 and at 500 C
2
c
Sample
Gold content Particle size
BET surface area
(m g
Au/Ce
2
−1
)
(wt%)
(nm)
a
b
Au/CeO 80
2
Red. 120 C
1.52
9 –11
80
0.32
0.26
0.20
0.18
◦
a
Au/CeO 80
15
80
2
◦
Red. 500 C
b
Au/CeO 150 1.53
10
150
150
2
◦
Red. 120 C
Au/CeO 150
2
◦
Red. 500 C
a
Particle size obtained from TEM.
Particle size obtained from Scherrer’s analysis of XRD data.
Atomic ratio as measured by XPS.
b
c
ously that sintering of gold particles with temperature occurred
for both samples, in agreement with the TPR and TEM results
for samples prereduced at high temperatures.
Fig. 2. XPS analysis: Au 4f in Au/CeO 150 and Au/CeO 80.
2
2
Although the BE of gold peaks can be measured from the
data given in Fig. 2, the analysis of the chemical state of gold
in the present samples is beyond the scope of the present con-
tribution.
of the Au/CeO2150 sample, all these parameters in the steady
state phase was lower for the former than for the latter cat-
alysts. Thus, Au/CeO2150 is considered to be more selective
than Au/CeO280.
3.2. Catalytic results
3.2.3. Crotonaldehyde hydrogenation on Au/CeO280 and
AuCeO2150 at 120 ◦C, pretreatment b
3.2.1. Crotonaldehyde hydrogenation on bare ceria
No significant activity was observed on the bare supports.
However, for both low- and medium-surface area ceria, an ex-
tended adsorption of the crotonaldehyde was detected. This
period was shorter for CeO280 (5 min) than for CeO2150
(10 min).
Based on previous results [12] indicating a strong bene-
ficial effect of calcination pretreatment on the yield toward
crotyl alcohol of Au/CeO2240, we submitted the Au/CeO280
and Au/CeO2150 samples to pretreatment b. The initial yields
to ethanol are strongly influenced by the pretreatment; ethanol
production in the samples submitted to pretreament a was
higher than that in the samples submitted to pretreatment b. The
initial yields to ethanol (during the deactivation period) were 70
3.2.2. Crotonaldehyde hydrogenation on Au/CeO280 and
Au/CeO2150 at 120 ◦C, pretreatment a: reduction for 1 h at
120 ◦C under H2
Figs. 3a and 3b report the dependence of conversion and of
the selectivity to different products on TOS for the Au/CeO280
and Au/CeO2150 catalysts respectively. Both samples demon-
strate a deactivation stage during which the product distri-
butions strongly varied with TOS. In the first of these, the
main product in both cases was ethanol (EOL), although in
−1
−1
and 17 × 10−8 mol s
g
(see Table 2) for the medium- and
cat
the low-surface area catalysts, respectively.
Regarding the steady-state regime, Fig. 4 compare the yields
to the different products for the Au/CeO280 catalyst corre-
sponding to pretreatments a and b. The activity corresponding
to pretreatment b was higher than that for pretreatment a. The
yield to CP was negligible for both pretreatments. The same
catalytic trends were observed for Au/CeO2150. For this cata-
lyst, the production of CP f−or1 pretreatment b was null, whereas
a yield of 0.2 × 10−8 mol s gc−a1t was measured over the same
sample submitted to pretreatment a.
On a steady-state regime, slightly higher crotyl alcohol pro-
duction was detected after calcinations pretreatment for the
Au/CeO2150 catalyst, whereas no enhancement of the produc-
tion of the desired product was detected for the low-surface area
sample. Table 1 reports the selectivity to crotyl alcohol, as well
as the crotyl alcohol-to-butanal ratio, indicating less improve-
ment than was found for the Au/CeO2240 sample [12].
the case of Au/CeO2150, its production was much higher than
for Au/CeO280 (85 and 44 × 10−8 mol s
g
−1, respectively).
−1
cat
Crotyl alcohol (UOL), butanal (BAL), and butanol (BOL) were
also detected, whereas C4 hydrocarbons were produced in mi-
nor amounts.
Afterward, a quasi-steady state was reached. Ethanol almost
disappeared at the start of this period, at the same time that con-
densation products (CPs) appeared. The main CP was 2-, 4-,
6-octatrienal, as identified by mass spectrometry. This fact sug-
gests that both CP and UOL were formed over the same sites.
For both samples, in the steady-state phase, crotyl alcohol
deceased with TOS while butanal increased, becoming the main
product. This indicates that the samples turn into rather uns-
elective catalysts under reaction conditions. The selectivity to
UOL and the yield to this product and the crotyl alcohol-to-
butanal ratio are reported in Table 1. Though catalytic trends
corresponding to the Au/CeO280 catalysts were similar to those
3.2.4. Crotonaldehyde hydrogenation on Au/CeO280 and
Au/CeO2: Influence of the temperature of the reduction
pretreatment
The temperature of the reduction pretreatment was increased
in an attempt to create more Ce3+ species or induce an in-

Journal of Catalysis 254 (2008) 71–78
75
◦
◦
Fig. 3. Crotonaldehyde hydrogenation on (a) Au/CeO 150 and (b) Au/CeO 80. Samples reduced 1 h at 120 C, reaction temperature 120 C. Dependence of activity
2
2
and selectivity to different products on TOS.
◦
◦
Fig. 4. Crotonaldehyde hydrogenation on (a) Au/CeO 150 and (b) Au/CeO 80. Samples reduced at 120 C, reaction temperature 120 C. Yield to the different
2
2
products corresponding to a and b pre-treatments at steady state regime.
Table 2
a
Hydrogenation of crotonaldehyde
b
Sample
Initial total yield
a pre-treatment
0.17
Initial yield
to ethanol
c
b pre-treatment
Au/CeO 80
2
0.45
6.1
0.3
4.8
0.2
◦
120 C
Au/CeO 80
0.11
0.81
0.08
2
◦
500 C
Au/CeO 150
0.73
2
◦
120 C
Au/CeO 150
2
◦
500 C
a
◦
Samples reduced at 120 C in hydrogen.
−1
cat
−1
cat
b
−7
−1
−1
10 mol s
g
measured at 10 min TOS. Deactivation period.
c
−8
10 mol s
g
, b pre-treatment.
Fig. 5. Crotonaldehyde hydrogenation on Au/CeO 150. Sample pre-reduced at
2
◦
120, 300 and 500 C. Yields to different products under steady state regime.
teraction between the gold and the support. The deactivation
period observed previously was not detected for the catalysts
prereduced at 500 ◦C; that is, the conversion was nearly con-
stant throughout the TOS. Moreover, the activity attained under
the steady-state regime was strongly dependent on the tem-
perature of the reduction pretreatment; the higher the temper-
ature of the reduction pretreatment, the lower the activity of
the sample. Fig. 5 shows the yields to the different products
(butanal, butanol, crotyl alcohol, and condensation products)
corresponding to the Au/CeO280 and Au/CeO2150 catalysts
under the steady-state regime for the different temperatures of
reduction pretreatment. As shown, the dependence product dis-
tribution was not greatly influenced by the prereduction temper-
ature.
3.2.5. Crotonaldehyde hydrogenation: Effect of the reaction
temperature (samples reduced at 120 ◦C, pretreatment a)
Fig. 6 reports the yields to each product in the steady-state
phase over Au/CeO2150 for reaction temperatures of 120, 150,
and 180 ◦C. The increase in reaction temperature caused an in-

76
Journal of Catalysis 254 (2008) 71–78
Fig. 7. Model of gold particles in Au/CeO 80, Au/CeO 150, and Au/CeO 240.
2
2
2
Fig. 6. Crotonaldehyde hydrogenation on Au/CeO 150 and Au/CeO 80. Sam-
ples reduced at 120, reaction temperatures 120, 150, and 180 C. Yields to the
different product under steady state regime.
2
2
◦
cles to sintering. Nanogold structures supported on high-surface
area ceria particles had a high resistance to sintering, with the
opposite occurring for Au/CeO280 and Au/CeO2150, as the
TPR and XPS results clearly showed. This would account for
the fact that after these samples were reduced at high tem-
perature, the gold particle size increased from 9 to 15 nm
for Au/CeO280. The entire scenario is depicted in the scheme
shown in Fig. 7.
crease in the yields to crotyl alcohol and butanal but a decrease
in condentation products. The formation of butanal was favored
over that of crotyl alcohol, due to the differing activation ener-
gies for the hydrogenation of the olefinic and carbonyl bonds
(30.8 and 42.3 kJ/mol−1, respectively).
4. Discussion
Regarding the hydrogenation of crotonaldehyde over Au/
CeO280 and Au/CeO2150 performed at 120 ◦C, the catalysts
demonstrated strong deactivation with TOS. No C3 hydrocar-
bons were detected over all of the TOS studied; thus, deac-
tivation due to decarbonylation reaction producing strong CO
adsorption of metal surface is disregarded. It can be speculated
that catalyst decay resulted from the formation of coke deposits
produced from dienic polymers originating from adsorbed cro-
tonaldehyde.
Here we compare the results of the present work on gold sup-
ported on ceria with low and medium surface areas (80 and 150
m2 g−1, respectively), with previously reported results for gold
supported on high-surface area ceria, Au/CeO2240 [12] and dis-
cuss the catalytic and characterization results. Regardless of the
specific surface area of the support (80, 150, or 240 m2/g),
the loading of gold fixed by the solid remains almost the same
(1.5–1.8 wt%). To explain this fact, it could be argued that dur-
ing the impregnation step, the acid pH of the precursor solution
modifies the support surface in all the cases, creating a similar
concentration of acid Lewis sites (cationic vacancies) responsi-
ble for the anchoring the metal.
The main product of the deactivation period was ethanol,
formed following a reaction postulated in a previous study [12],
CH3–CH=CH–CHO → CH3–CHOH–CH2–CHO
→ 2CH3–CH2OH.
(1)
The first step, hydration of the C=C bond, occurs on ceria
sites, consuming surface OH groups. In the second step, hy-
drogenolysis of the C(2)–C(3) bond, hydrogen atoms coming
from dissociated H2 on the metallic gold particles participate.
Considering that H2 chemisorption occurs on surface atoms
of low coordination number [22], reaction (1) would occur on
highly dispersed gold. Both active sites are necessary for the
formation of ethanol and OH from the support and metallic
gold particles. This explains why the yield of ethanol depends
on the specific surface area of the support, as well as on the
gold particle size. The yield for the medium-surface area cata-
lyst (Au/CeO2150) was higher that that of the low-surface area
catalyst (Au/CeO280), due to the former’s larger specific sur-
face area and smaller particle size. Au/CeO2240, with a high
concentration of both active sites due to its high surface and the
relatively small particle size, exhibited high initial activity for
ethanol formation.
Based on TPR results, we determined that the concentra-
tion of surface oxygen anions and reducible ceria species were
(as expected) lower on low-surface area ceria than on medium-
surface area ceria. Because these species are stabilized on nu-
cleated oxygen vacancies on the ceria surface [18,19], TPR
can account for a dependence of the vacancy concentration on
the surface area of ceria. Wahlström et al. [20] postulated that
oxygen vacancies are active nucleation sites for Au clusters
on rutile surface, whereas Lai and co-workers [21], based on
FTIR of adsorbed CO results, indicated that low-coordination
ceria sites in association with oxygen vacancy sites are re-
sponsible for the strong binding of small gold particles. One
could argue that high-surface area ceria can stabilize gold at
high dispersion. This would be in line with the fact that gold
particles are relatively large on low- and medium-surface area
ceria (as measured by TEM and XRD analysis) but are <4 nm
for high-surface area ceria (as deduced from XRD analysis in
Ref. [12]). Furthermore, the different surface area of the sup-
ports also would play a role in the resistance of gold parti-
It was previously postulated [12] that the formation of
ethanol results in iginates Lewis acid sites. In a second step,

Journal of Catalysis 254 (2008) 71–78
77
these Lewis center are the active sites for the formation of
condensation products (see reaction (3) in Ref. [12]). For this
reason, the higher the production of ethanol in the initial state,
the higher the production of CP under the steady-state regime.
In line with this, the yield to CP was higher−fo1r Au/CeO2150
increasing prereduction temperature would be related to an en-
largement of crystallites after high-temperature reduction. This
particle size modification was clearly demonstrated by the XPS
and TPR experiments. A decreased specific surface area was
ruled out based on BET analysis of the samples reduced at high
temperature.
(initial yield to ethanol, 85 × 10−8 mol s gc−a1t ) than for
In steady-state conditions at 120 ◦C, Au/CeO280 and Au/
CeO2150 have lower activity than Au/CeO2240. The differ-
ence in the activity between high- and low-surface area samples
can be ascribed to a size effect; the gold particles on the low-
surface area ceria catalyst were larger than those supported on
the high-surface area ceria catalyst, but are rather inactive to-
ward hydrogenation reactions.
Au/CeO280 (initial yield to ethanol, 44 × 10−8 mol s
g
−1).
−1
cat
In previous work [12], it was concluded that pretreatment b
leads to a diminished number of OH groups on the ceria sur-
face and concomitantly a diminished yield to ethanol. In the
samples studied in the present work, the initial yield of ethanol
was lower for pretreatment a than for pretreatment b.
For previously studied Au/CeO2240, the decreased forma-
tion of condensation products led to an increase in the yield
of crotyl alcohol in the steady-state regime. The improved se-
lectivity was due to an increase in the availability of the sites
responsible for formation of condensation products: surface
oxygen vacancies, V0O. In turn, the formation of Ce3+ centers
also increased, following:
5. Conclusion
The supports used in this work (with surface areas of 150
and 80 m2 g−1) should not be considered convenient materi-
als for gold, because the catalysts present relatively low activity
and selectivity to crotyl alcohol in the hydrogenation of croton-
aldehyde at atmospheric pressure. This occurs because (i) the
concentration of surface sites of the support responsible for a
promotional effect on gold catalytic properties are too low, and
thus no increase of gold selectivity can be achieved due to the
support, and (ii) gold particles are relatively large and quite
unstable under reaction conditions. Comparing the present re-
sults with those corresponding to Au supported on high-surface
area ceria (240 m2 g−1) indicates that the latter support leads
to highly selective catalyst in which ceria redox and acid-base
properties are engaged in a promotional effect.
V0O + Ce4+ ꢀ Ce3+ + e−.
(2)
The formation of crotyl alcohol was higher when the equilib-
rium (2) was displaced to the right, due to a promotional effect
of Ce3+ cations. In this way, the low- and medium-surface area
ceria, showing meager production of the secondary products,
would give rise to a highly selective catalyst. However, the ex-
pected increase in selectivity was not observed for the low-
and medium-surface area catalysts; the selectivity to crotyl al-
cohol was approximately 10% for Au/CeO280 versus 63% for
Au/CeO2240 (steady-state values, for samples submitted to pre-
treatment a). It could be speculated that the concentration of
V0O sites of the low-surface area support was not sufficient to
produce Ce3+ sites following reaction (2). Concomitantly, no
promotional effect of the support occurred in the low-surface
area ceria.
Acknowledgments
The authors thank Svetlana Ivanova and Raymonde Touro-
ude for fruitful discussions.
The selectivities to crotyl alcohol for the Au/CeO280 cata-
lyst under different reaction and pretreatment conditions were
similar to the values reported by other groups for Au supported
on “inert” supports. For example, Haruta et al. [11] found a
selectivity to crotyl alcohol of 10% for an alumina-supported
catalyst. The selectivity corresponding to the Au/CeO2150
sample was higher (30–33%) than that corresponding to the
Au/CeO280 catalyst. The difference in selectivity to the desired
product between these samples is related to the fact that the
V0O concentration for the CeO2150 was higher than that of the
CeO280.
References
[1] M.A. Vannice, B. Sen, J. Catal. 115 (1989) 65.
[2] C. Milone, R. Ingoglia, A. Pistone, G. Neri, F. Frusteri, S.J. Galvagno,
J. Catal. 222 (2004) 348.
[3] C. Milone, R. Ingoglia, L. Schipilliti, C. Crisafulli, G. Neri, S. Galvagno,
J. Catal. 236 (2005) 80.
[4] J.E. Bailie, G.J. Hutchings, Catal. Commun. 2 (2001) 29.
[5] J.E. Bailie, G.J. Hutchings, Chem. Commun. (1999) 2151.
[6] M. Shibata, N. Kawata, T. Masumuto, H. Kimura, J. Chem. Soc. Chem.
Commun. (1988) 154.
[7] C. Mohr, P. Claus, Sci. Prog. 84 (4) (2001) 311.
[8] S. Schimpf, M. Lucas, C. Mohr, U. Rodemerck, A. Brückner, J. Radnik,
H. Hofmeister, P. Claus, Catal. Today 72 (2002) 63.
[9] P. Claus, Appl. Catal. A. Gen. 291 (2005) 222.
At this point, the different catalytic results have been dis-
cussed based mainly on the different surface areas of the sup-
ports. However, the role of particle size should be consid-
ered for the present catalysts with a relatively large particle
size (9–11 nm). Bailie and Hutchings [5], studying Au/ZnO,
Au/ZrO2, and Au/SiO2, suggested that high activity and selec-
tivity of Au catalysts are associated with small particles. An
effect of particle size on selectivity toward hydrogenation of
the C=O bond also would explain the difference between the
samples that we studied and the high-surface area catalyst. Fur-
thermore, the diminished selectivity toward crotyl alcohol with
[10] R. Zanella, C. Louis, S. Giorgio, R. Touroude, J. Catal. 223 (2004) 328.
[11] M. Okumura, T. Akita, M. Haruta, Catal. Today 2688 (2002) 1.
[12] B. Campo, M. Volpe, S. Ivanova, R. Touroude, J. Catal. 242 (2006) 162.
[13] S. Ivanova, V. Pitchon, Y. Zimmermann, C. Petit, App. Catal. A Gen. 298
(2006) 57.
[14] J.H. Scofield, J. Electron. Spectrosc. Relat. Phenom. 8 (1976) 29.
[15] S. Scirè, S. Minicò, C. Crisafulli, C. Satriano, A. Pistone, Appl. Catal. B
Environ. 40 (2003) 43.
[16] D. Andreeva, V. Idakiev, T. Tabakova, L. Ilieva, P. Falaras, A. Bourlinos,
A. Travlos, Catal. Today 72 (2002) 51.

78
Journal of Catalysis 254 (2008) 71–78
[17] T. Akita, M. Okumura, K. Tanaka, M. Kohyama, M. Haruta, Catal. Today
117 (2006) 62.
[18] Q. Fu, H. Saltsburg, M. Flytzani-Stephanopoulos, Science 301 (2003) 935.
[19] Q. Fu, W. Deng, H. Saltsburg, M. Flytzani-Stephanopoulos, Appl. Catal.
B Environ. 56 (2005) 57.
[20] E. Wahlström, N. Lopez, R. Schaub, P. Thostrup, A. Rønnau, C. Africh,
E. Lagsgaard, J. Nørskov, F. Besenbaches, Phys. Rev. Lett. 90 (2003)
026101.
[21] S.-Y. Lai, Y. Qiu, S. Wang, J. Catal. 237 (2006) 303.
[22] L. Stobinsky, L. Zommer, R. Dus, Appl. Surf. Sci. 141 (1999) 319.