Selective hydrogenation of α,β-unsaturated ketones to α,β-unsaturated alcohols on gold-supported catalysts

Article abstract of DOI:10.1016/j.jcat.2003.11.003

The liquid-phase reduction of α,β-unsaturated ketones [trans,4-phenyl,3-buten,2-one (benzalacetone C₆H₅CHCHCOCH₃), 4 -methyl,3-penten,2-one (CH₃) ₂CCHCOCH₃, and 3-penten,2-one CH₃CHCHCOCH ₃] to the corresponding α,β-unsaturated alcohols has been investigated on gold-supported catalysts. Au/Fe₂O₃ and Au/Al₂O₃ have been prepared by coprecipitation and deposition-precipitation. The catalytic behavior of a Au/Fe₂O ₃ "reference" catalyst supplied by the World Gold Council has been also investigated. In the hydrogenation of benzalacetone and 4-methyl,3-penten,2-one on the "homemade" Au/Fe₂O ₃ catalysts the unsaturated alcohol is the main reaction product. Chemoselectivity higher than 60% was achieved. On Au/Al₂O ₃, the selectivity is 10%. It is noteworthy that within the gold supported on iron oxide samples, the reference catalyst shows the lowest selectivity toward the formation of the unsaturated alcohol. In the hydrogenation of 3-penten,-2-one on the homemade Au/Fe₂O ₃, the saturated ketone is the main reaction product and the selectivity toward the formation of the unsaturated alcohol is 15% at conversion 90%. It is likely that the absence of bulky substituents on the conjugated CC double bond favors its adsorption on the catalytic sites, leading to the formation of the saturated carbonyl compounds as the main reaction product. A detailed characterization of the investigated catalysts by TEM and XRD is also reported.

Full text of DOI:10.1016/j.jcat.2003.11.003

Journal of Catalysis 222 (2004) 348–356
www.elsevier.com/locate/jcat
Selective hydrogenation of α, β-unsaturated ketones to α, β-unsaturated
alcohols on gold-supported catalysts
a
a
a
a
b
a,∗
C. Milone, R. Ingoglia, A. Pistone, G. Neri, F. Frusteri, and S. Galvagno
a
Dipartimento di Chimica Industriale e Ingegneria dei Materiali, Università di Messina, Salita Sperone 31, I-98166 Messina, Italy
b
Istituto CNR-TAE, Salita S. Lucia sopra Contesse 5, I-98126 Messina, Italy
Received 8 August 2003; revised 5 November 2003; accepted 5 November 2003
Abstract
The liquid-phase reduction of α,β-unsaturated ketones [trans,4-phenyl,3-buten,2-one (benzalacetone C H CH=CHCOCH ), 4-methyl,3-
6
5
3
penten,2-one (CH ) C=CHCOCH , and 3-penten,2-one CH CH=CHCOCH ] to the corresponding α,β-unsaturated alcohols has been
3
2
3
3
3
investigated on gold-supported catalysts. Au/Fe O and Au/Al O have been prepared by coprecipitation and deposition-precipitation. The
2
3
2 3
catalytic behavior of a Au/Fe O “reference” catalyst supplied by the World Gold Council has been also investigated. In the hydrogenation
2
3
of benzalacetone and 4-methyl,3-penten,2-one on the “homemade” Au/Fe O catalysts the unsaturated alcohol is the main reaction product.
2
3
Chemoselectivity higher than 60% was achieved. On Au/Al O , the selectivity is 10%. It is noteworthy that within the gold supported on
2
3
iron oxide samples, the reference catalyst shows the lowest selectivity toward the formation of the unsaturated alcohol. In the hydrogenation
of 3-penten,-2-one on the homemade Au/Fe O , the saturated ketone is the main reaction product and the selectivity toward the formation
2
3
of the unsaturated alcohol is 15% at conversion >90%. It is likely that the absence of bulky substituents on the conjugated C=C double
bond favors its adsorption on the catalytic sites, leading to the formation of the saturated carbonyl compounds as the main reaction product.
A detailed characterization of the investigated catalysts by TEM and XRD is also reported.

2003 Elsevier Inc. All rights reserved.
Keywords: Gold catalysts; Selective hydrogenation; α,β-Unsaturated ketones; α,β-Unsaturated alcohols
1
. Introduction
per (I) hydride complexes are able to reduce the conjugated
C=O double bond in the presence of molecular H2 [4]. On
heterogeneous catalysts the selective hydrogenation of the
conjugated C=O bond is generally obtained by using iso-
propanol as a hydrogen source [5–8].
Gargano et al. have studied the chemoselective reduction
of the trans,4-phenyl,3-buten,2-one (benzalacetone) and 5-
hexen,2-oneon metal oxides, via hydrogentransfer from iso-
propyl alcohol [5]. The allylic alcohols have been obtained
with remarkable selectivity and yield. During the reduc-
tion of 5-hexen,2-one the selectivity toward the formation
of the unsaturated alcohol was higher than 90% on CaO and
La2O3, whereas in the hydrogenation of benzalacetone the
chemoselectivity reached values higher than 80% on La2O3,
Y2O3, and Al2O3.
The selective reduction of α, β-unsaturated ketones to
α, β-unsaturated alcohols by molecular hydrogen has been
attempted on heterogeneous catalysts for some time with
very limited success [1,2].
A recent paper dealing with the selective hydrogenation
of the C=O bond of a highly functionalized cyclic α, β-
unsaturated ketone (ketoisophorone) reports selectivities to
the unsaturated alcohol higher than 90% [3]. However, such
appreciable results are obtained only in the hydrogenation of
ketoisophorone. In the reduction of other substrates, having
a sterically unhindered C=O group, the C=C double bond
is preferentially hydrogenated.
The selective reduction of the conjugated C=O bond in a
ketone can be easily achieved by using LiAlH4 as reductant.
Homogeneous catalysts such as phosphine-stabilized cop-
Coman et al. [6] report that the reduction of a sterically
hindered unsaturated ketone to a secondary allylic alcohol
occurs with high yields on PtSn catalysts, in the presence of
isopropanol as hydrogen donor. It has been suggested that
the high chemoselectivity is favored by the low hydrogen
*
Corresponding author.
E-mail address: galvagno@ingegneria.unime.it (S. Galvagno).
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2003.11.003

C. Milone et al. / Journal of Catalysis 222 (2004) 348–356
349
coverage obtained by using isopropanol instead of H2 and
by the presence of bulky substituents on the C=C double
bond which hinders its adsorption on the catalytic sites. The
importance of a steric hindrance has been confirmed by ex-
periments carried out with methyl vinyl ketone which, under
similar reaction conditions, was mainly reduced to the satu-
rated ketone [6]. It has also been proposed that the activation
the catalysts was determined by measuring the adsorbance
of gold solutions at λ = 400 nm [15]. The solutions were
prepared by dissolving the catalysts in HCl–HNO (3:1 by
3
volume), and then adding a solution of tin(II) chloride in hy-
drochloric acid.
The reference catalyst, 4.4 wt% Au supported on Fe O ,
2
3
prepared by coprecipitation, was supplied by the World Gold
Council.
n+
of the carbonyl group occurs on Sn sites, in analogy with
the widely accepted mechanism reported for the hydrogena-
tion of α, β-unsaturated aldehydes [1,2,9].
The Ru/Fe O sample (Ru nominal content = 2 wt%)
2
3
was prepared by deposition of the metal colloid on the pre-
formed support. The ruthenium nanoparticles were obtained
In the present work we report the results of a study on
the catalytic activity of gold-supported catalysts in the re-
duction of α, β-unsaturated ketones by dihydrogen. Prelimi-
nary results have already demonstrated that Au supported on
Fe2O3 is able to hydrogenate the trans,4-phenyl,3-buten,2-
one (benzalacetone) to the corresponding unsaturated alco-
hol with a selectivity higher than 60% at 90% of conver-
sion [10].
by reduction of RuCl in ethylene glycol following the pro-
3
cedure reported in Ref. [16]. Fe O was then added to the
2
3
colloidal solution under vigorous stirring to form a suspen-
sion. The catalysts was filtered, washed with water, and dried
at 393 K for 2 h.
2
.2. Sample characterization
It is known that gold catalysts are very selective in the
hydrogenation of α, β-unsaturated aldehydes to the corre-
sponding unsaturated alcohols in reactions occurring in ei-
ther the gas phase on the liquid phase [11–13].
Surface area measurements were made using the BET
nitrogen adsorption method in a conventional volumetric ap-
−4
paratus, after outgassing (10 mbar) the sample at 353 K
In this paper we will detail the study of the hydrogena-
tion of benzalacetone on gold catalysts and we will also
present the results obtained in the hydrogenation of other
α, β-unsaturated ketones. Moreover, the catalytic behavior
of “homemade” samples will be compared with a Au/Fe2O3
gold “reference” catalyst supplied by the World Gold Coun-
cil.
for 2 h.
XRD studies were carried out with an Ital-Structures dif-
fractometer using nickel-filtered CuKα radiation by mount-
ing the powder samples on Plexiglas holders. Diffraction
peaks were compared with those reported in the JCPDS Data
File.
TEM analysis was performed on a Philips instrument
Model CM12 operating at 120 kV and directly interfaced
with a computer for real-time image processing. The cat-
alyst specimens for electron microscopy were prepared by
gently grinding the powder samples in an agate mortar, sus-
pending, and sonicating them in isopropanol, and a drop of
the suspension was placed on a carbon film copper grid.
After evaporation of the solvent, the specimens were intro-
duced into the microscope column. In order to obtain metal
particles-size distribution, several hundred particles visible
on the micrographs were counted.
2
. Experimental
2
.1. Sample preparations
Au-supported catalysts were prepared by coprecipitation
and deposition-precipitation. The coprecipitation was car-
ried out using HAuCl4 (Fluka) and Fe(NO3)3·9H2O (Fluka)
as starting materials. An aqueous mixture of HAuCl4 and
Fe(NO3)3·9H2O was poured into an aqueous solution of
Na2CO3 1 M (pH 11.9) and kept at 353 K under vigorous
stirring. The solids were digested overnight at room tem-
perature and then washed with water until free of chloride
ions (AgNO3 test) and then dried under vacuum at 353 K for
The average value of the gold particle sizes was calcu-
lated by the following equation:
ꢀ
nidi
¯
d =
.
ni
1
day.
The deposition-precipitation of gold onto the supports
2.3. Catalytic activity
was carried out in agreement with Haruta’s procedure [14],
by adding the support to an aqueous solution of HAuCl4
previously adjusted at pH 7–8 with NaOH. The slurry was
maintained at 343 K, under vigorous stirring, for 2 h. Then
the samples were filtered, washed with deionized water un-
til elimination of chloride, and then dried under vacuum at
The catalytic experiments were carried out at atmospheric
pressure under H2 flow, at 333 K, in a 100-ml four-necked
batch reactor fitted with a reflux condenser, dropping fun-
nel, thermocouple, and magnetic stirrer head coupled with
a gas stirrer (Model MRK 1/20-BR purchased by Premex
Reactor). This gas stirrer was used to optimize the hydrogen
transfer from the gas to the liquid phase. The catalyst (weight
0.5–1 g; particle size = 160–200 mesh) was added to 25 ml
of solvent and treated at 343 K for 1 h under gaseous H2.
3
53 K for 1 day.
Al2O3 support was a commercial product (Alpha Aesar
purity 99.7%). Fe2O3 was prepared in agreement with the
procedure used for the coprecipitation. The gold content of

350
C. Milone et al. / Journal of Catalysis 222 (2004) 348–356
−
4
The substrate (6 × 10 mol) was injected through one
Catalysts obtained by deposition-precipitation of the gold
hydroxide onto preformed Al O and Fe O supports have a
arm of the flask. The reaction mixture was stirred at 700 rpm.
The trans,4-phenyl,3-buten,2-one (benzalacetone) was
supplied by ACROS, whereas the other reactants (4-methyl,
2
3
2
3
gold loading ranging between 2 and 6 wt% and BET surface
2
area of about 120 m /g. Calcination in air of the AF5dp cat-
3
-penten,2-one and 3-penten,2-one) were supplied by Fluka.
The progress of the reaction was followed by sampling
alyst at 673 K for 2 h leads to a strong decrease of the surface
area.
a sufficient number of microsamples and analyzing them
by means of GC-MS (Shimadzu GC-QP5000 Mod.). The
reaction products were separated on a CP-WAX capillary
column (1.20 µm, 60 m, 0.32 mm).
Catalytic activity was measured in terms of initial rates
and calculated from the slope of the curves conversion ver-
3
.2. Transmission electron microscopy measurements
Representative TEM microphotographs of the “as pre-
pared” coprecipitated and deposited-precipitated samples
are shown in Figs. 1a and b. Different particles sizes and
microstructures of the iron oxide support are obtained on the
samples prepared with these two methods. On the AF16.6,
prepared by coprecipitation, the support is in the form of
aggregates of round nanometer particles of about 3–5 nm
in diameter (Fig. 1a). No phase contrast characteristics of
an amorphous material can be observed. On the AF5dp cat-
alyst, the support exhibits a heterogeneous microstructure,
likely due to the presence of different (crystalline) phases
sus time (t) at t = 0. Selectivities were calculated by the
ꢀ
expression Si = Ci/ Cp where Ci is the concentration of
product i and Cp the total concentration of the products.
Preliminary tests carried out with different amounts of cata-
lysts (0.2–1.0 g) grain size (80–200 mesh) and stirring rate
(
500–1000 rpm) indicated that under the conditions used the
reaction is carried out in the absence of mass-transfer limi-
tation.
(
Fig. 1b). On both catalysts, very large and irregular gold
particles with low contrast have been observed, suggesting
the presence of gold in an oxidized state.
3
. Results and discussion
TEM observations show that the morphological charac-
teristics of the samples change after 1 h reduction in H2,
at 343 K in ethanol (Figs. 1c and d). In particular, small
gold particles were clearly visible on the reduced catalysts.
This may be due to the reduction of oxidized gold species
to metallic gold and/or to the growth of native gold sub-
nanometer particles favored by the H2 flowing at 343 K, in
ethanol. Moreover, on the coprecipitated catalysts the parti-
cles of support grew up to 7–10 nm in diameter (Fig. 1c).
The gold particle size distributions and the mean parti-
cle diameters of the reduced catalysts are reported in Fig. 2
and Table 1. On the coprecipitated catalyst AF3.1, the count
of gold particles was very difficult, likely because of the
low contrast with the support. Therefore, for this sample it
was not possible to draw a true reliable distribution. Despite
this limitation, TEM analysis clearly indicates that coprecip-
3
.1. Chemical composition and surface area
Table 1 lists the chemical and morphological properties
of the gold samples.
Catalysts prepared by coprecipitation have a gold loading
ranging between 3 and 17 wt% and show an increase of the
surface area with the gold content (entries 1–3). For sake of
completeness the surface area of the reference catalyst has
been measured under the same conditions as for the home-
made solids and it is reported in Table 1. The lower BET
surface area of the latter with respect to the analogous copre-
cipitated catalysts can be ascribed to calcination treatments
to which the sample has been subjected. It should be noted
that, unless otherwise specified, our samples were treated at
3
53 K.
Table 1
Chemical composition, surface area (SA), mean gold particle size (nm), and catalytic activities (V ) of gold-supported catalysts
i
a
Entry
Catalyst
Au
wt%)
Support
Preparation
method
Mean gold
particle size
d (nm)
SA
(m /g)
V
i
2
7
(
(10 )
c
1
2
3
4
5
6
7
AF3.1
AF5.3
AF16.6
AF5dp
AF5dp/c
3.1
5.3
16.6
5.4
5.4
4.4
Fe O
Coprecipitation
Coprecipitation
Coprecipitation
–
68
122
178
121
34
3.5 ± 0.2
3.0 ± 0.2
2.1 ± 0.1
27.9 ± 1.4
< 0.1
2
3
3
3
3
3
3
3
Fe O
6.3
6.6
3.2
39.4
2.8
4.3
2
Fe O
2
Fe O
Deposition-precipitation
Deposition-precipitation
Coprecipitation
2
b
Fe O
2
d
Au/Fe O
Fe O
39
124
1.1 ± 0.1
2
3
2
AAl2.3dp
2.3
Al O
2
Deposition-precipitation
1.0 ± 0.1
a
−1 −1
).
V = initial rate of hydrogenation of benzalacetone (moles converted g
s
i
Au
b
c
d
Calcined at 673 K for 2 h.
Not determined due to the poor contrast with the support background.
Gold “reference” catalyst, Type C, Lot No. Au/Fe O 02-3.
2
3

C. Milone et al. / Journal of Catalysis 222 (2004) 348–356
351
Fig. 1. TEM microphotographs of Au/Fe O catalysts: (a) AF16.6 “as prepared,” (b) AF5dp “as prepared,” (c) AF16.6 reduced, (d) AF5dp reduced.
2
3
itated samples AF5.3 and AF16.6 have a larger particles size
distribution than the AF5dp sample.
has been ascribed to the retarding of the crystal growth of the
iron phase by the presence of the gold ions. The diffraction
peaks relative to metallic gold have never been detected.
Fig. 3a shows the TEM microphotograph of the AF5dp
calcined at 673 K for 2 h. A remarkable increase of the
mean gold diameter was registered with respect to the un-
calcined sample. On the calcined catalyst the majority of
gold particles have a diameter between 10 and 80 nm. TEM
microphotographs reported in Figs. 3b and c refer to the
Au/Fe2O3 reference catalyst and the Au/Al2O3 sample, re-
spectively. On the reference catalyst it can be clearly seen the
plate-like morphology typical of well-crystallized hematite.
On both catalysts small, well isolated, round and with good
contrast gold crystallites having a narrow size distribution
After 1 h reduction in H , at 343 K, in ethanol, XRD pat-
2
terns of the coprecipitated catalysts show some significant
changes. As an example the XRD diffraction pattern of the
reduced AF16.6 sample (pattern d) is reported. Reflections
are attributed to magnetite (Fe O ) coming from the low-
3
4
temperature reduction of the amorphous iron(oxy-hydroxy)
species which is catalyzed by the presence of gold. The re-
duction at low temperature of the iron oxide support has been
previously evidenced by TPR experiments [15]. Moreover,
the main diffraction peak of metallic gold at 38.1 (2θ) is
noted.
(
Fig. 2) have been detected.
Fig. 4 (pattern e) shows the XRD diffraction pattern of the
reference catalyst. It shows a well-resolved reflection pattern
attributed to hematite. This sample, obtained according to
the coprecipitation method adopted by Haruta et al. [18], has
been calcined during its preparation at high temperature and
therefore the support is transformed into a crystalline phase.
X-ray diffraction patterns of the AF5dp and of the par-
ent support are reported in Fig. 5 (patterns a and b). On
the iron oxide the XRD analysis shows the presence of
different crystalline iron oxy-hydroxides phases. Hematite
(α-Fe2O3) is the prevailing crystallographic phase, but
goethite (α-FeOOH) and ferrihydrite have also been found.
Upon dispersing gold, through the deposition-precipitation
3
.3. X-ray diffraction measurements
Structural characteristics of the catalysts have been inves-
tigated by X-ray diffraction (XRD). Fig. 4 shows the XRD
of representative coprecipitated samples (patterns a–d). The
as prepared catalysts (patterns a–c), dried at 353 K, show an
almost amorphous feature regardless of the metal loading.
These results are in agreement with our previous findings on
similar catalysts [17] where it is reported that coprecipitated
Au/Fe2O3 catalysts, with a gold loading higher than 1 wt%,
show an amorphous feature, whereas at lower gold loading
the support is better crystallized as hematite. This behavior

352
C. Milone et al. / Journal of Catalysis 222 (2004) 348–356
Fig. 2. Particle-size distribution of gold-supported catalysts: (a) AF5.3 reduced, (b) AF16.6 reduced, (c) AF5dp reduced, (d) AF5dp/c calcined at 673 K for
h, (e) Au/Fe O reference, (f) AAl2.3dp reduced.
2
2
3
method, the ferrihydrite phase almost disappears. It is likely
that during the deposition of gold, ferrihydrite is transformed
into the more ordered hematite phase [19]. It is interesting
to note that after 1 h reduction in H2, at 343 K in ethanol,
the XRD pattern of the AF5dp catalyst (pattern c) is substan-
tially unmodified with respect to that of the as-prepared sam-
ple. This indicates the stability of the (crystalline) hematite
phase under reducing conditions at low temperatures.
Calcination of AF5dp at 673 K for 2 h leads to the growth
of the iron oxide particles, and consequently the intensity of
the reflection peaks increases (Fig. 5, pattern f). The main
crystallographic phase is hematite, whereas the diffraction
peaks of goethite disappear. Moreover, the diffraction peaks
of metallic gold are also observed.
3.4. Catalytic activity
Addition of H to the unsaturated ketones can be de-
2
scribed through the following simplified scheme where
ꢀ
ꢀ
R = H, R = C H for benzalacetone, R = H, R = CH for
6
5
3
ꢀ
3-penten,2-one, and R = CH , R = CH for 4-methyl,3-
3
3
penten,2-one.
Fig. 6 shows a typical course of the hydrogenationof ben-
zalacetone on gold preparations.
In Table 1 is reported the rate of hydrogenation of benza-
lacetone (C H CH=CHCOCH ) on gold preparations, ex-
6
5
3
pressed as initial rate of reaction per gram of metal.
It must be highlighted that the reduction of benzalace-
tone occurs by means of molecular H2 and any mechanism

C. Milone et al. / Journal of Catalysis 222 (2004) 348–356
353
Fig. 4. XRD spectra of catalysts prepared by coprecipitation: (a) AF3.1,
(
b) AF 5.3, (c) AF16.6, (d) AF16.6 reduced, (e) Au/Fe O reference.
2 3
(F) hematite; (1) metallic gold; (P) magnetite.
Fig. 5. XRD spectra of catalysts prepared by deposition-precipitation:
a) Fe O , (b) AF5dp, (c) AF5dp reduced, (d) AF5dp/c calcined at 673 K
(
2
3
for 2 h. (F) hematite; (1) metallic gold; (2) goethite; (") ferrihydrite.
Fig. 3. TEM microphotographs of gold-supported catalysts: (a) AF5dp/c,
calcined at 673 K for 2 h, (b) Au/Fe O reference, (c) AAl2.3dp reduced.
2
3
involving the hydrogen transfer from the alcohol to the sub-
strate was ruled out. This conclusion was drawn through a
series of experiments carried out in the absence of hydro-
gen. As reported in Section 2, catalysts were prereduced in
situ at 343 K for 1 h; the reactor was cooled down at the reac-
tion temperature (333 K). In order to eliminate any residual
H2, the reactor was flushed with N2. At this point the sub-
strate was introduced into the reaction vessel. Under these
conditions no conversion of benzalacetone was observed up
to 3 h of contact time, thus indicating that the reducing agent
is molecular H2. Moreover, when H2 was readmitted into the
reactor, the hydrogenation of benzalacetone occurred.
Fig. 6. Typical reaction course during the hydrogenation of benzalacetone
on the “homemade” Au/Fe O catalysts. (Reaction conditions: T = 333 K,
2
3
−4
PH = 1 atm, ethanol = 25 ml, benzalacetone = 6 × 10
mol, AF5dp
2
Catalytic activity data of the homemade Au/Fe2O3 cat-
alysts (Table 1) clearly show that the rate of reaction is
catalyst = 0.500 g.) Benzalacetone (F), saturated ketone (Q), unsaturated
alcohol (2), saturated alcohol (P).

354
C. Milone et al. / Journal of Catalysis 222 (2004) 348–356
strongly influenced by the preparation method. Catalyst pre-
pared by coprecipitation (entry 2) is much less active than a
analogous sample prepared by deposition-precipitation (en-
try 4). Within the same preparation method, instead, the
catalytic activity decreases with increasing gold content (en-
tries 1–3).
The higher catalytic activity of the sample prepared by
deposition-precipitation, AF5dp, with respect to the AF5.3
sample could be due to the higher dispersion of gold (Ta-
ble 1) and/or to a higher amount of gold exposed on the
external surface of the support. It is well known that in
solids prepared by coprecipitation and dried at 353 K, gold
is mainly buried into the iron oxy-hydroxide matrix [20],
whereas a higher amount of exposed gold is obtained
through the deposition-precipitation method [21].
Table 2
Catalytic activity and selectivity toward the formation of the unsaturated
alcohol in the hydrogenation of benzalacetone
Catalyst
Au
(wt%)
3.1
Conversion
(%)
Selectivity (%)
UA
SK
SA
AF3.1
AF5.3
AF16.6
AF5dp
23.9
95.1
21.6
42.4
1.4
35.8
1.5
28.4
11.2
35.8
18.6
77.9
77.3
100
51.8
57.0
57.3
56.2
64.2
67.5
61.3
58.5
20.4
19.8
–
5.8
41.6
6.9
42.3
7.4
21.2
2.9
23.0
1.7
5.3
16.6
5.4
9
5.1
20.4
0.4
9
17.3
91.2
17.6
a
Au/Fe O
2
4.4
3
84.0
1.9
–
3.5
AF5dp/c
AAl2.3dp
5.4
2.3
11.0
21.9
85.1
11.4
The Au/Fe2O3 reference catalyst shows a lower catalytic
activity with respect to the homemade iron oxide-supported
catalysts. This low activity is likely related to the calcination
procedure used in the preparation of the reference sample to
obtain metallic gold. Calcination of the solid is, in fact, detri-
mental for the catalytic activity. As it is shown in Table 1, the
hydrogenation rate of the AF5dp/c (calcined at 673 K) dra-
matically decreases with respect to that of the as-prepared
catalyst.
From the data reported in Table 1 no definitive conclusion
on the effect of gold particle size on the rate of reaction can
be drawn. On the three samples prepared by the same copre-
cipitation method, the decrease of activity can be simply due
to a decrease of the gold surface area.
SK, saturated ketone; UA, unsaturated alcohol; SA, saturated alcohol.
a
Gold reference catalyst, Type C, Lot No. Au/Fe O 02-3.
2
3
It can be noted that the selectivity to the unsaturated al-
cohol (UA) remains almost constant regardless of the level
of conversion whereas the selectivity to the saturated alco-
hol (SA) increases as that of the saturated ketone decreases
(
SK). This indicates that the formation of the saturated ke-
tone and of the unsaturated alcohol occurs through parallel
reactions, whereas the saturated alcohol is mainly obtained
by the further hydrogenation of the C=O bond of the satu-
rated ketone (Scheme 1).
The formation of the unsaturated alcohol has been ob-
served on all gold supported catalysts and the selectivity
measured at low and high conversion are reported in Table 2.
All the homemade Au/Fe2O3 preparations, but AF5dp/c,
show a remarkable selectivity toward the hydrogenation of
the C=O bond in the whole range of conversion investigated.
Scheme 1.

C. Milone et al. / Journal of Catalysis 222 (2004) 348–356
355
Table 3
AF5.3 catalysts, we could conclude that the dimensions of
the metal particles do not influence the selectivity. Indeed,
the selectivity of the two samples toward the hydrogena-
tion of the conjugated C=O bond is quite similar (Table 2),
notwithstanding the difference in the mean gold particles
size (Table 1). However a small increase in selectivity is
observed on the AF16.6 sample having larger gold parti-
cles (Table 1) and prepared similarly to the AF5.3 sample.
A more detailed investigation on the effect of the metal par-
ticle size is necessary to elucidate this effect on the selective
hydrogenation of unsaturated ketones.
Catalytic activity and selectivity toward the formation of unsaturated alco-
hol in the hydrogenation of α,β-unsaturated ketones
Substrate
Catalyst Conversion
%)
17.3
1.2
22.1
9.1
20.3
0.2
Selectivity (%)
(
SK
UA
SA
Benzalacetone
AF5dp
AF5dp
AF5dp
35.8 61.3
2.9
9
18.6 58.5 23.0
34.6 60.7 4.7
9.8 64.9 12.3
83.3 14.3 2.4
59.8 16.2 24.0
4
-methyl,3-penten,2-one
-penten,2-one
8
3
9
SK, saturated ketone; UA, unsaturated alcohol; SA, saturated alcohol.
It should also be noted that the chemical nature of the
support strongly influences the activation of the conjugated
C=O bond. Among the uncalcined samples, gold supported
on Al2O3 shows the lowest selectivity toward the formation
of the unsaturated alcohol, indicating that iron oxide pro-
motes the hydrogenation of the C=O bond with an increase
of the selectivity toward the formation of the unsaturated al-
cohol (Table 2).
Calcination of the AF5dp sample at 673 K decreases
strongly the catalytic activity. On this calcined sample the
only reaction product was the saturated ketone (Tables 1
and 2). A lower selectivity has also been measured on the
reference sample, likely due to the calcination treatment
used to obtain metallic gold. From the XRD characterization
it seems that the promotion effect is caused by poor crystal-
lized oxy-hydroxide phases. When all the iron is in the form
of well-crystallized hematite the catalytic activity and selec-
tivity drop drastically.
We have also studied the hydrogenation of 4-methyl,3-
penten,2-one [(CH3)2C=CHCOCH3] and of 3-penten,2-one
[
CH3CH=CHCOCH3]. The product distribution, at low and
high conversion, is reported in Table 3. It can be seen that
Au/Fe2O3 catalysts are able to hydrogenate selectively the
conjugated C=O bond on substrates having a sterically un-
hindered C=O bond with respect to benzalacetone. The re-
placement of the bulky phenyl group of benzalacetone with
two methyl groups on the C=C bond in the 4-methyl,3-
penten,2-one does not lead to substantial differences in the
product distribution. The α, β-unsaturated alcohol remains
the main reaction product with a selectivity higher than 60%
in the whole range of conversion investigated. In the hydro-
genation of the 3-penten,2-one, which has only one methyl
group as a substituent on the C=C bond, the selectivity to-
ward the formation of the unsaturated alcohol drops to al-
most 15%. In analogy with the behavior observed in the hy-
drogenation of the α, β-unsaturated aldehydes, the presence
of less bulky substituents on the conjugated C=C bond fa-
vors its adsorption on the catalytic sites leading to the prefer-
ential formation of the saturated carbonyl compound [1,22].
In the hydrogenation of benzalacetone on the homemade
Au/Fe2O3 catalysts prepared by coprecipitation (Table 2),
the selectivity toward the hydrogenation of the C=O group
increases slightly with gold loading. Au/Fe2O3 prepared
by deposition-precipitation shows a slight higher selectiv-
ity than analogous sample prepared by coprecipitation (Ta-
ble 2). Within the gold supported on iron oxide catalysts,
the reference sample shows the lowest selectivity toward the
formation of the unsaturated alcohol.
It should be highlighted that the presence of iron ox-
ide is not sufficient to enhance the hydrogenation rate of
the C=O with respect to the C=C bond in the hydrogena-
tion of the α, β-unsaturated ketones, but it is necessary to
have both gold and iron oxides. This indicates that the cat-
alytic behavior arises from a synergetic effect between the
two constituents. Indeed, when the hydrogenation of benza-
lacetone is carried out on Ru supported on Fe O , prepared
2
3
dispersing colloids of ruthenium on the preformed Fe O
2
3
(see Section 2), the main reaction product is the saturated
ketone and the selectivity toward the formation of the un-
saturated alcohol is less than 2%. For sake of completeness,
it must be noted that the Ru/Fe O has been reduced un-
2
3
der the same conditions as gold preparation (1 h in H2,
at 343 K, in ethanol). The XRD diffraction pattern of the
reduced Ru/Fe O is similar to that of the parent support
Claus et al. [11] in their investigation of the hydrogena-
tion of acrolein on Au-supported catalysts have reported that
the selectivity toward the formation of the unsaturated alco-
hols increases with increasing particle size in the range 1–
2
3
(Fig. 5a). Therefore in the Ru/Fe O the iron oxide is mainly
2
3
presents as poorly crystallized oxy-hydroxide phase, which
is suitable to promote the C=O hydrogenation on the gold
catalysts.
5
nm. They speculate that the adsorption of the C=O group
of the α, β-unsaturated aldehyde is favored by face atoms,
whereas sites with low coordination strongly favor the acti-
vation of the C=C bond. A similar correlation has also been
found by Hutchings and co-workers [12] in their work on the
hydrogenation of buten-2-al on Au/ZnO catalysts.
In order to gain more information on this point we
have also studied the hydrogenation of cinnamaldehyde
(trans,3-phenyl,2-propen,1-alC H CH=CHCHO) the α, β-
6
5
unsaturated aldehyde corresponding to benzalacetone. Ta-
ble 4 reports the product distribution obtained in the hydro-
genation of the cinnamaldehyde on the different catalysts.
Au/Fe2O3 are the most selective samples regardless of gold
In our case, it seems that the selectivity to the unsaturated
alcohol cannot be univocally correlated to the gold particle
sizes. If we compare the results obtained on the AF5dp and

356
C. Milone et al. / Journal of Catalysis 222 (2004) 348–356
Table 4
4. Conclusions
Selectivity toward the formation of α,β-unsaturated alcohol in the hydro-
genation of cinnamaldehyde
From the results reported in this paper, the following main
conclusions can be drawn: (a) gold is an intrinsically selec-
tive catalyst for the hydrogenation of the C=O bond; (b) iron
oxide favors the activation of the C=O bond; (c) the promot-
ing effect of iron oxide depends strongly on the substrate
and on the catalytically active metal. No clear effect of the
gold particle size on the catalytic activity and selectivity has
been observed. Calcination of the iron oxide decreases the
catalytic activity and selectivity. From the characterization
results it is not possible to identify definitely the nature of
the iron sites responsible for C=O activation.
Catalyst
Me
Conversion
(%)
Selectivity (%)
UA
(wt%)
SAL
SAC
AF5.3
AF5dp
5.3
5.4
4.4
2.3
2
50.2
56.3
48.2
50.1
50.3
51.2
13.6
3.3
25.3
33.9
8.1
83.8
86.9
73.2
60.3
74.9
10.2
2.6
9.8
1.6
5.8
17.0
9.1
a
Au/Fe O
2
3
AAl2.3dp
Ru/Fe O
2
3
3
Ru/Al O
2
80.7
2
Reaction conditions: T = 333 K, PH = 1 atm, ethanol = 25 ml, cinnam-
2
−4
aldehyde = 6 × 10 mol, catalyst weight = 0.500 g. SAL, saturated alde-
hyde; UA, unsaturated alcohol; SAC, saturated alcohol.
a
Gold “reference” catalyst, Type C, Lot No. Au/Fe O 02-3.
2
3
Acknowledgment
loading and preparation method, in agreement with our pre-
vious findings [13]. It is noteworthy that an enhancement
of the selectivity is also obtained when Ru is dispersed on
Fe2O3. This Ru sample is totally unselective in the hydro-
genation of benzalacetone.
This work was performed with the financial support
of Ministero dell’Istruzione, Università e Ricerca (MIUR)
through a COFIN 2001 grant.
To explain the promoting effect of iron it has been previ-
ously suggested that it acts as a Lewis acid which increases
the polarization of the C=O bond facilitating the attack
of the weak nucleophilic hydrogen adsorbed on the metal
sites [9,13].
The promoting effect of the Lewis acid, however, is
not always operating. In the hydrogenation of the α, β-
unsaturated aldehydes it is known that when palladium is
used as catalyst only the corresponding saturated aldehyde
is formed, regardless of the presence of a promoter. On the
other hand, when α, β-unsaturated ketones are hydrogenated
the promoter has no effect on the selectivity on all but the Au
metal catalysts.
The reason why palladium is so selective toward the for-
mation of the saturated aldehydes has been attributed to the
preferential adsorption of the conjugated system as a 1,4-
C=C–C=O species. The hydrogenation of such a species
would result in the formation of an unstable enol intermedi-
ate which will isomerize rapidly to the saturated aldehyde.
Therefore no effect of promoter on the selective hydrogena-
tion of the C=O bond can be observed [23]. On these bases
it could be suggested that, when a substituent is added to
the C=O group (ketone) the direct adsorption through the
carbonylic function (1,2-C=O adsorbed species) is hindered
and therefore the 1,4-C=C–C=O mode will be the preferen-
tial adsorption. As suggested for palladium, this species can
lead only to the saturated ketone and therefore no effect of
the promoter can be observed. Opposite to palladium, gold is
likely a metal on which the 1,4-C=C–C=O adsorption mode
is not favored and therefore the 1,2-C=O adsorbed species
is formed also in the case of ketones. The scarce abundance
of the 1,4-C=C–C=O species on gold catalysts is in agree-
ment with the higher intrinsic selectivity observed also in
the hydrogenation of unsaturated aldehydes in the absence
of promoter [11–13].
References
[
[
1] V. Ponec, Appl. Catal. A: Chem. 149 (1997) 27.
2] R.L. Augustine, Catal. Today 37 (1997) 419.
[3] M. von Arx, T. Mallat, A. Baiker, J. Mol. Catal. A: Chem. 148 (1999)
275.
[4] J.-X. Chen, J.F. Daeuble, D.M. Brestensky, J.M. Stryker, Tetrahe-
dron 56 (2000) 2153.
[5] M. Gargano, V. D’Orazio, N. Ravasio, M. Rossi, J. Mol. Catal. 58
(1990) L5.
[6] S.N. Coman, V.I. Parvulescu, M. De Bruyn, D.E. De Vos, P. Jacobs,
J. Catal. 206 (2002) 218.
[7] A.J. Blacker, S.M. Brown, C. Bubert, J.M.J. Williams, WO 0244111
(2002).
[
[
8] A.J. Blacker, B.V. Mellor, US patent 2002/0156282 (2002).
9] Z. Poltarzewki, S. Galvagno, R. PietroPaolo, P. Staiti, J. Catal. 102
(1986) 190.
[
[
[
[
10] C. Milone, R. Ingoglia, M.L. Tropeano, G. Neri, S. Galvagno, Chem.
Commun. (2003) 868.
11] P. Claus, A. Brückner, C. Mohr, H. Hofmeister, J. Am. Chem. Soc. 122
(2000) 11430.
12] J.E. Bailie, H.A. Abdullah, G.J. Hutchings, et al., Phys. Chem. Chem.
Phys. 3 (2001) 4113.
13] C. Milone, M.L. Tropeano, G. Gulino, G. Neri, R. Ingoglia, S. Gal-
vagno, Chem. Commun. (2002) 868.
[
[
14] M. Haruta, Catal. Technol. 6 (3) (2002) 102.
15] F.E. Beamish, J.C. van Loon, in: Analysis of Noble Metals, Academic
Press, New York, 1977.
[16] A. Miyazaki, I. Balint, K.-i. Aika, Y. Nakano, J. Catal. 204 (2001) 364.
[17] G. Neri, A.M. Visco, S. Galvagno, M. Panzalorto, Thermochim.
Acta 329 (1999) 39.
[18] M. Haruta, S. Tsubota, T. Kobayashi, H. Kageyama, M. Genet, B. Del-
mon, J. Catal. 144 (1993) 175.
[19] U. Schewertmann, R.M. Cornell, in: Iron Oxides in the Laboratory:
Preparation and Characterization, VCH, Weinheim, 1991.
[20] M. Haruta, H. Kageyama, N. Kamijo, T. Kobayashi, F. Delannay, Stud.
Surf. Sci. 44 (1988) 33.
[21] S. Scirè, S. Minicò, C. Crisafulli, C. Satriano, A. Pistone, Appl. Catal.
B 40 (2003) 43.
[22] P. Gallezot, D. Richard, Catal. Rev.-Sci. Eng. 40 (1&2) (1998) 81.
23] F. Delbecq, P. Sautet, J. Catal. 152 (1995) 217.
[