Selective Vapor-Phase Epoxidation of Propylene over Au/TiO2 Catalysts in the Presence of Oxygen and Hydrogen

Article abstract of DOI:10.1006/jcat.1998.2157

Gold has long been thought to be chemically inert, however, it has recently been proven that its catalytic performance is dramatically tunable by control of the particle size and by careful selection of the support metal oxide. A typical example is the selective oxidation of propylene in a gas containing oxygen and hydrogen. When gold is deposited on TiO₂ by a deposition-precipitation technique as hemispherical particles with diameters smaller than 4.0 nm it produces propylene oxide with selectivities higher than 90% and conversions of 1-2% at temperatures of 303-393 K. The oxidation of hydrogen to form water is depressed by propylene, whereas propylene oxidation is not only enhanced but also restricted to partial oxidation by hydrogen. The depression of hydrogen combustion by the presence of propylene and a new peak due to gold deposition in TPD spectra have indicated that propylene is adsorbed on the surfaces of both gold particles and the TiO₂ support. The reaction rate is almost independent on the concentration of propylene and increases linearly with increasing concentrations of O₂ and H₂. The above results suggest that propylene adsorbed on a gold surface may react with oxygen species formed at the perimeter interface between the gold particles and the TiO₂ support through the reaction of oxygen with hydrogen. The effect of gold loading is surprising in that the reaction product switches from propylene oxide to propane when gold loading is decreased to 0.1 wt%. Careful TEM observation indicates that gold particles larger than 2.0 nm in diameter produce propylene oxide, whereas smaller gold particles produce propane.

Full text of DOI:10.1006/jcat.1998.2157

JOURNAL OF CATALYSIS 178, 566–575 (1998)
ARTICLE NO. CA982157
Selective Vapor-Phase Epoxidation of Propylene over Au/TiO2
Catalysts in the Presence of Oxygen and Hydrogen
Toshio Hayashi,1 Koji Tanaka, and Masatake Haruta2
Osaka National Research Institute, AIST, Midorigaoka 1, Ikeda, 563-8577 Japan
Received September 23, 1997; revised April 30, 1998; accepted May 15, 1998
ical reactions to remain unsolved by catalysis (2). Despite
considerable effort, no economically viable route has yet
been found. Most patents, featuring the vapor-phase oxi-
dation of propylene to PO, show either low selectivity or
low productivity (3). Since direct oxidation is difficult to
achieve, a new process has been designed by Enichem to
produce PO by oxidation with H2O2 over a titanium sili-
cate catalyst (4, 5). For this process palladium supported on
titanium silicate can be used to replace H2O2 with a mix-
ture of molecular oxygen and hydrogen (6). The history of
the epoxidation processes shows a shift of the oxidant from
alkyl hydroperoxide with IV–VI elements toward hydro-
gen peroxide with Ti-silicate and, currently, toward hydro-
gen peroxide prepared in situ from dioxygen and hydrogen.
Recently, a few new approaches have been attempted. Gas-
phase partial oxidation of propylene in the copresence of
propane (7), electrochemical oxidation on a Pt-black anode
(8), and photochemical oxidation over SiO2 based catalysts
(9) can produce PO; however, selectivities are not suffi-
ciently high, 50% at the highest.
Although gold was and still is regarded by many as being
poorly active as a catalyst (10, 11), when gold is dispersed
as fine particles over select metal oxides it has been found
that the supported gold exhibits exceptionally high catalytic
activity (12, 13). Examples include CO oxidation (14–22),
hydrocarbon combustion (23, 24), hydrogenation of CO2
and CO (25, 26), water gas shift reaction (27, 28), and also
the reduction of NO to N2 (29). Quite often reactions occur
at low or subambient temperatures.
Gold has long been thought to be chemically inert, however, it has
recently been proven that its catalytic performance is dramatically
tunable by control of the particle size and by careful selection of the
support metal oxide. A typical example is the selective oxidation of
propylene in a gas containing oxygen and hydrogen. When gold is
deposited on TiO2 by a deposition–precipitation technique as hemi-
spherical particles with diameters smaller than 4.0 nm it produces
propylene oxide with selectivities higher than 90% and conversions
of 1–2% at temperatures of 303–393 K. The oxidation of hydrogen
to form water is depressed by propylene, whereas propylene oxi-
dation is not only enhanced but also restricted to partial oxidation
by hydrogen. The depression of hydrogen combustion by the pres-
ence of propylene and a new peak due to gold deposition in TPD
spectra have indicated that propylene is adsorbed on the surfaces
of both gold particles and the TiO2 support. The reaction rate is al-
most independent on the concentration of propylene and increases
linearly with increasing concentrations of O2 and H2. The above
results suggest that propylene adsorbed on a gold surface may react
with oxygen species formed at the perimeter interface between the
gold particles and the TiO2 support through the reaction of oxygen
with hydrogen. The effect of gold loading is surprising in that the
reaction product switches from propylene oxide to propane when
gold loading is decreased to 0.1 wt%. Careful TEM observation in-
dicates that gold particles larger than 2.0 nm in diameter produce
propylene oxide, whereas smaller gold particles produce propane.
c
ꢀ 1998 Academic Press
INTRODUCTION
Propylene oxide (hereafter denoted as PO) is one of the
most important chemical feedstocks for producing resins
such as polyurethane. Major conventional manufacturing
methods require two-stage processes, using chlorohydrin
or hydroperoxides as an oxidant (1). The direct synthesis of
PO, by the use of molecular oxygen, has long been desired
and has been nominated one of the most important chem-
Results reported in the past for the partial oxidation of
hydrocarbons over gold catalysts could not always be repro-
duced and may need for re-examination in view of chlorine
contamination, dispersion of gold, support effect, and so
forth. Gold catalysts investigated for C1 hydrocarbons are
Au/BaSO4 which can produce HCHO directly from CH4
at 923K with a yield of 0.4% (30) and Au gauze which
produces HCHO from CH3OH at 723 K with a selectiv-
ity of 100% (31). It was reported that Au/SiO2 catalyzed
propylene oxidation to form PO with a yield below 1% and
with 50% selectivity (32), whereas Cant and Hall reported
that acrolein was produced with a yield of 5% and with
1 Present address: Nippon Shokubai Co., Ltd., Nishi-Otabi 5-8, Suita,
564-0034 Japan.
2 To whom correspondence should be addressed. E-mail: haruta@onri.
go.jp.
566
0021-9517/98 $25.00
c
Copyright ꢀ 1998 by Academic Press
All rights of reproduction in any form reserved.

EPOXIDATION OF PROPYLENE
567
33% selectivity (33). Sinfelt described that the addition of precipitate, 41 m2/g), and ZrO2 (Nippon Aerosil Co. Ltd.,
gold to copper enhanced both activity and selectivity in the 40 m2/g) in a similar manner. Because Au(OH)3 could not
acrolein formation from proylene (34). Recently, Niiyama be deposited on SiO2, Au/SiO2 was prepared by the impreg-
and his coworkers have reported that the deposition of Au nation method.
onto Bi2Mo3O12 is effective to improve the yield of acrolein
Pd/TiO2 and Pt/TiO2 catalysts were prepared by the usual
from propylene maintaining its selectivity while Pd and Pt impregnation method by using (NH4)2PdCl4 and H2PtCl6,
accelerate complete oxidation to CO2 (35). Gold supported followed by drying, calcination in air at 673 K for 3 h, and
on Al2O3 and MgO was reported not to produce epoxide H2 treatment at 573 K for 2 h. These Pd and Pt metal par-
from ethylene and propylene because of two surface center ticles obtained were confirmed to be smaller than 2 nm in
adsorption per olefin molecule (36, 37).
diameter by TEM observation. In the case of Pd and Pt,
In continuation with our recent work on highly dispersed deposition–precipitation did not lead to any appreciable
gold catalysts, we present here the first evidence for the difference in particle size nor in any catalytic activity for
direct vapor-phase oxidation of propylene to PO in the CO oxidation (39). Cu/TiO2 and Ag/TiO2 catalysts were
presence of O2 and H2 using a catalyst comprising of gold also prepared by the impregnation method using nitrates as
deposited on TiO2:
the starting material, followed by drying, calcination, and
H2 treatment, in a similar manner.
==
–
CH3CH CH2 + O2 + H2 → CH3CH CH2 + H2O [1]
lO,
Catalyst Characterization
propylene
propylene oxide
The particle size of gold was determined by a TEM
(Hitachi H-9000NA). Except for Au loadings lower than
0.2 wt% , at least 200 particles were chosen to determine
the mean diameter by using a computerized image analyzer
(Nippon Avionics Co. Ltd., EXCEL). XPS measurements
were conducted by using a spectrometer (Shimadzu ESCA-
K1M) with a monochromatized Al focused X-ray beam and
an analyzer energy of 50 eV. TPD measurements were car-
ried out by using an apparatus equipped with TCD and FID
detector after treating the samples in an O2 or He stream
at 673 K and then cooling to 273 K.
EXPERIMENTAL
Catalyst Preparation
The TiO2-supported gold catalysts were prepared by
deposition–precipitation (DP) (38) of Au(OH)3 onto TiO2
(Nippon Aerosil, P-25, specific surface area; 50 m2/g) in an
aqueous solution of HAuCl4 at pH = 7. The precursors were
washed several times, dried, and calcined in air at 673 K for
3 h. By this method fine gold metal particles with diameters
smaller than 4 nm could be deposited with almost uniform
dispersion. They were mostly hemispherical in shape and
were attached to the TiO2 support at their flat planes (15).
Actual gold loadings of Au/TiO2 samples calcined were,
Catalytic Tests
Catalytic performance tests were made under atmo-
of course, smaller than the gold contents in the starting spheric pressure in a quartz reactor using a continuous-flow
solutions of HAuCl4, usually by 40–80% (38). The gold system at a temperature range of 273 K and 623 K. A mix-
loadings hereafter denoted are the gold contents in the fin- ture of propylene, oxygen, hydrogen, and argon as diluent
ished catalysts measured by X-ray fluorescence analysis and was passed through a fixed bed containing the catalyst sam-
are expressed as weight percentages to TiO2 or TiO2/SiO2 ples. The concentration of C3H6, O2, H2 was varied from 2 to
support.
20 vol% , 2 to 20 vol% , and 10 to 90 vol% , respectively. The
For comparison studies, the conventional impregnation feed gas compositions seem to be practical for an industrial
method was also used; however, it produced spherical Au process of the direct epoxidation of propylene; however,
particles simply loaded on the TiO2 support and with di- they need safety treatments because the explosion limit of
ameters larger than 10 nm. The Au/TiO2/SiO2 catalysts H2 is from 4.1 to 74.2 vol% in air and from 4.0 to 94.0 vol%
were also prepared by the same DP method by replacing in O2.
TiO2 with TiO2/SiO2 (TiO2 loading, 3 wt% to SiO2), which
Samples were sieved between 125 and 212ꢁm of par-
was prepared by impregnating the silica (Fuji Davison, ID ticle size for Au/TiO2 and between 355 and 1000ꢁm for
gel, specific surface area; 310 m2/g) with titanylacetylaceto- Au/TiO2/SiO2 and in most cases weighed to 0.5 g. The
nate, followed by drying and calcination in air at 773 K for temperature of the catalyst bed was measured by using a
2 h. The influence of metal oxide support was investigated thermocouple in a glass tube which was fixed just above the
by depositing gold on ꢀ -Al2O3 (JRC-ALO-4, a reference catalyst bed. The flow rate was 2000 ml/h unless otherwise
sample of the Catalysis Society of Japan, with a specific stated. All reaction products were identified by standard
surface area of 174 m2/g), Fe2O3 (prepared from hydrox- GC-Mass and NMR spectroscopic techniques and ana-
ide precipitate, 37 m2/g), Co3O4 (prepared from carbon- lyzed by two on-line gas chromatographs using four types
ate precipitate, 59 m2/g), ZnO (prepared from hydroxide of columns, Porapak Q (3m) for CO2, C3H6, and C3H8,

568
HAYASHI, TANAKA, AND HARUTA
molecular sieve 5A (3m) for CO, and activated carbon Ag supported on TiO2. The conversion tended to decrease
(3m) for H2 and O2, as packed columns with a thermal sharply soon after the reaction and then after 10 to 20 min
conductivity detector (TCD), and HR-20M (0.53 mm ꢁ it gradually decreased with time on stream. The deactiva-
50 m, Shinwa Chemical Industries, Co. Ltd.) capillary col- tion was caused mainly by oligomerization of PO and C3H6
umn for organic compounds with a flame ionization detec- and the catalysts used for reaction could be reactivated
tor (FID). The detection limit of the FID was 0.001 vol% by air oxidation at 525 K. Only Au/TiO2 prepared by the
for organic compounds.
deposition–precipitation method could selectively catalyze
the oxidation of C3H6 to PO. While C3H6 conversions are in
general not large and the consumption of H2 to form H2O
is much larger than that calculated from the stoichiometry
of Eq. [1], the selectivities to PO are very high and exceed
90% . The other products detected were acetone and car-
bon dioxide. An increase in gold loading from 0.98 wt% to
3.2 wt% leads to an appreciable increase in the space time
yield of PO but to a similar rate of PO formation per unit
weight of Au.
When the TiO2 support is replaced with TiO2 deposited
on SiO2, the reaction temperature is shifted toward higher
temperatures. The shift of reaction temperature can be cor-
related with a decrease in the catalytic activity for H2 oxi-
dation. It is also likely that the separation of active sites over
the SiO2 surface allows high temperature operation main-
taining a good selectivity to PO. When the reaction temper-
ature was raised to 373K with Au/TiO2 catalysts, it was only
the rate of H2 combustion that was appreciably increased.
The rate of PO formation was not increased but was de-
creased. It appears that there is an optimum temperature
region for the epoxidation of C3H6 with each gold cata-
RESULTS
Catalyst Characterization
Figures 1 and 2 show typical TEM photopgraphs of Au/
TiO2 (Au loading 0.98 wt% (a) and 0.02 wt% (b)) and the
distributions of the diameter of gold particles, respectively.
Small gold particles with diameters of 1–4 nm and of hemi-
spherical shape are homogeneously dispersed on TiO2 par-
ticles. The mean particle diameter of Au was 2.4 nm with
a standard deviation of 25% for 0.98 wt% Au loading and
1.7 nm with a standard deviation of 25% for 0.02 wt% load-
ing. The Au 4d and 4f spectra of XPS for 0.98 wt% Au/TiO2
showed no appreciable shift in binding energies from those
of bulk gold, indicating that the gold particles were metallic
in nature under high vacuum. Because of the very low load-
ing (0.02 wt% ), the XPS spectra for smaller gold particles
could not be obtained.
Catalytic Properties
Table 1 shows the conditions and products for the re- lyst. On the other hand, H2 combustion is monotonously
action of C3H6 with O2 and H2 over Au, Pd, Pt, Cu, and enhanced with an increase in reaction temperature. The
TABLE 1
Reaction of Propylene with Oxygen and Hydrogen over Au, Pd, Pt, Cu, and Ag Supported on TiO2 and over Au Supported on TiO2/SiO2
Selectivity,d %
PO formation rate
Metal
Reaction
PO
yield,f
%
Conversion, %
loading,a temperature, Reactionb
c
Catalyst
Au/TiO2
Au/TiO2
Au/TiO2
wt%
K
conditions C3H6
H2
POe Acetone C3H8 CO2
mmol/hr/g-cat mmol/hr/g-Au
0.98
1.0
3.2
323
353
303
323
393
298
298
393
393
A
A
B
B
C
A
A
A
A
1.1
0.2
0.3
0.6
2.5
57
12
<0.2
<0.1
3.2 >99
-
-
7
4
-
0.4
2
-
-
-
1.1
-
0.3
0.6
2.3
-
-
-
-
0.2
-
0.27
0.54
20.0
-
8.4
16.7
g
8.9
2.5
4.2
2.6
98
87
2.7
<0.1
-
<10
-
-
-
98
92
-
>70
-
-
7
93
96
93
-
Au/TiO2/SiO2
Pd/TiO2
Pt/TiO2
Cu/TiO2
Ag/TiO2
0.20
1.0
1.0
1.0
1.0
0.21
102.7
1
6
-
-
-
-
-
-
-
-
-
-
-
>99
-
-
-
a Actual metal loadings determined by X-ray fluorescence analysis.
b
A: Feed gas, C3H6/O2/H2/Ar = 10/10/10/70; flow rate, 2000ml/hr; catalyst, 0.5g; B: Feed gas, C3H6/O2/H2/Ar = 10/10/10/70; flow rate, 2000ml/hr;
catalyst, 0.1g; C: Feed gas, C3H6/O2/H2/Ar = 5/10/40/45: flow rate, 2000ml/hr; catalyst, 0.5g.
c Conversion of propylene = 100 (moles of all products detected by GC analysis)/(moles of propylene fed); moles of CO2 formed divided by three.
d Selectivity = 100 (moles of each product detected directly by GC analysis)/(moles of all products detected directly by GC analysis); moles of CO2
formed divided by three.
e PO = propylene oxide.
f PO Yield = 100 (moles of PO detected directly by GC analysis)/(moles of propylene fed).
g Prepared by impregnation method.

EPOXIDATION OF PROPYLENE
569
FIG. 1. TEM photographs for Au/TiO2 catalysts. Au loading: (a) 0.98 wt% , (b) 0.02 wt% .

570
HAYASHI, TANAKA, AND HARUTA
FIG. 2. Size distributions of gold particles for Au/TiO2 catalysts. Au loading: (a) 0.98 wt% , (b) 0.02 wt% .
feed gas composition used for the Au/TiO2/SiO2 catalyst idation of C3H6. The fact that H2 consumption at 350 K is
was also different from those of other catalysts: lower C3H6 only about 1.4 times as the stoichiometric amount required
concentration and larger H2 concentration, both of which to produce PO implies that the selective partial oxidation
were advantageous for improving C3H6 conversion (see may principally proceed via Eq. [1]. Dioxygen may be trans-
Fig. 4). The C3H6 conversion obtained was above 2% and formed by reduction with H2, to an active species which
the rate of PO formation per unit weight of Au was about has the ability to selectively oxidize C3H6 to PO. How-
five times as large as those of Au/TiO2 catalysts.
ever, as shown in Fig.3, the ratio of H2 consumption to oxy-
Neither Au nor TiO2 alone exhibited any notable cata- genate formation increases with reaction temperature, indi-
lytic activity and selectivity at a temperature range of cating that the combustion of H2 with O2 is more enhanced
303 K to 573 K, indicating that the combination of Au than the partial oxidation of C3H6. From the economical
with TiO2 was indispensable for the selective oxidation of point of view, it is requested to minimize the ratio down to
C3H6. However, it should be noted that PO did not form unity by retarding the consumption of H2 with O2 to form
at all over Au/TiO2 prepared by the conventional impreg- H2O.
nation method, which produced spherical Au particles sim-
Figure 4 shows the rate of PO formation as a function of
ply loaded on the TiO2 support and with diameters above partial pressure of feed gas, H2, O2, and C3H6, at the reac-
10 nm. The combination of gold with other metal oxides, tion temperature of 350 K and 393 K. The PO formation
such as Al2O3, SiO2, Fe2O3, Co3O4, ZnO, and ZrO2 did rate appreciably increases with increasing concentration of
not catalyze the partial oxidation of C3H6 and produced H2 and O2 and almost independent of the concentration of
H2O and CO2. Over Pd/TiO2 and Pt/TiO2, C3H6 is mostly C3H6 up to 10 vol% . The rate dependency on oxygen con-
transformed into C3H8, a hydrogenated product, with only centration is the first order and saturates at around 10 vol% .
trace amounts of acetone formed as an oxygenate. Even at The hydrogen-dependency does not level off and gives re-
a lower temperature, at 298 K, the conversions of C3H6 and action orders of 0.9–0.6. The above reaction orders imply
of H2 are much larger than those at 323 K over Au/TiO2. that C3H6 is strongly adsorbed on the catalyst surface nearly
Over Cu/TiO2 and Ag/TiO2, neither partial oxygenates of to saturation and that the active oxygen species are formed
C3H6 nor C3H8 were obtained and H2 was preferentially through the reaction of O2 with H2.
converted to H2O at temperatures above 393 K.
Figure 3 shows the rates of H2-O2, H2-O2-C3H6, and
C3H6-O2 reactions over Au(0.20 wt% )/TiO2/SiO2 as a func-
TPD of Propylene
tion of temperature. The oxidation of C3H6, in the absence
In TPD spectra, as shown in Fig. 5, a single broad des-
of H2, occurs only above 473 K and produces mainly CO2. orption peak is observed at around 313 K for TiO2. This
While the oxidation of C3H6 is enhanced by H2, the ox- peak can be assigned to C3H6 adsorbed on the surface of
idation of H2 to H2O is retarded by C3H6. This provides TiO2. Pretreatment in a stream containing H2O before the
optimal conditions for the selective low-temperature epox- introduction of C3H6 pulses reduces the intensity of the

EPOXIDATION OF PROPYLENE
571
face area of the TiO2 support in the Au (0.98 wt% )/TiO2
sample (40), the peak intensity ratio of 4 (Au/TiO2) to 3
(TiO2) suggests that the population density of C3H6 ad-
sorption is larger by more than 30 times on the Au sur-
face than on the TiO2 surface. The above results imply that
C3H6 can preferentially adsorb on the surface of Au parti-
cles under reaction conditions where H2O is continuously
formed.
Effect of Au Loading: Switching of Product Selectivity
The unique behavior of Au/TiO2 system is that the reac-
tion pathway switches from oxidation to hydrogenation at
a specific Au metal loading. Table 2 lists the yields of re-
action products over Au/TiO2 catalysts with different gold
loadings ranging from 0.02 wt% to 0.98 wt% . The catalysts
with Au loadings of 0.98 and 0.39 wt% catalyze oxidation to
produce mainly PO with a by-product of CO2. On the other
FIG. 3. Rate of oxygenate formation and of hydrogen consumption
in the oxidation of hydrogen and/or propylene over Au/TiO2/SiO2. Flow
rate, 2000ml/hr; catalyst, Au(0.20 wt% )/TiO2/SiO2 0.5g: 4, hydrogen con-
sumption in a feed gas of H2/O2/Ar = 40/10/50 (vol% ); ᭺, hydrogen
consumption and ᭹, PO (propylene oxide) formation in a feed gas of
H2/O2/C3H6/Ar = 40/10/2/48 (vol% ); ᮀ, CO2 formation and ᭿, the for-
mation of acetone and acrolein in a feed gas of C3H6/O2/Ar = 2/10/88
(vol% ).
desorption peak and brought about a new peak at around
423–573 K due to water desorption. For Au/TiO2 the C3H6
desorption was observed in two peaks at around 313 K and
around 323 K. The higher-temperature peak which does not
disappear after the moisture pretreatment can be ascribed
to the desorption of C3H6 from the surface of Au particles
or/and the perimeter interface on the TiO2 support. The
former possibility is assumed to be much larger, because
H2 oxidation which occurs on the surface of Au particles is
appreciably retarded in the copresence of C3H6. The whole
peak intensity for C3H6 desorption decreases by moisture
pretreatment (c) and (d) in the same ratio between TiO2
and Au/TiO2, 21% reduction from (b) to (c) and 51% re-
duction from (b) to (d), indicating that moisture affects the
adsorption of C3H6 on the TiO2 support and on the Au
particles similarly. Since the exposed surface area of the
FIG. 4. Rate dependency on reactant concentration over Au/TiO2/
SiO2: ᭺, H2/O2/C3H6 = 10-40/10/5 (393K); ᮀ, H2/O2/C3H6 = 10-90/5/5
(350K); 4, H2/O2/C3H6 = 40/2-20/5 (393K); ♦, H2/O2/C3H6 = 10/10/2-10
(350K); ᭹, H2/O2/C3H6 = 40/10/2-20 (350K); ᭡, H2/O2/ C3H6 = 40/10/2-10
(393K). Each gas was diluted with Ar up to 100 vol% in total. Flow rate,
gold particles occupies only less than 1% of the total sur- 2000ml/hr; catalyst, Au(0.20 wt% )/TiO2/SiO2 0.5g.

572
HAYASHI, TANAKA, AND HARUTA
Au particles was not so different for Au/TiO2 catalysts with
Au loadings below 0.20 wt% . On the other hand, the con-
trast of Au particles in TEM observation become weaker
with lower Au loadings. This implies that Au particles were
thinner or flatter and that we might have been unable to ob-
serve all the small Au particles in the samples with low Au
loadings. It isprobable that not onlyparticle size but also the
particle shape (thick or thin) influence the reaction of C3H6.
The flat shape of Au particles may account for the fact that
the formation rate of C3H8 normalized per gram Au appre-
ciably increases with a decrease in Au loading. However, it
can be primarily concluded that gold particles larger than
2 nm are responsible for the formation of PO, whereas those
below 2 nm catalyze the formation of C3H8. Gold particles
smaller than 2 nm may be no longer intrinsically gold but
may resemble Pt in catalytic nature which also selectively
produces C3H8 under the same reaction conditions. Since
Au clusters smaller than 1 nm could not be clearly observed
by TEM the possibility of the contribution of such clusters
to the catalytic reaction cannot be excluded.
DISCUSSION
The results obtained appear to show that over the
Au/TiO2-based catalysts molecular oxygen can be trans-
formed through the reaction with molecular hydrogen to
an active species, which is capable to selectively oxidize
C3H6 adsorbed on Au particles to PO. Although we have
attempted to detect oxygen species by ESR, no significant
information has yet been obtained. Many intermediates, for
example H2O2, are known to be involved during the reac-
tion between H2 and O2 (41). Naito and Tanimoto have
speculated that oxygen molecules on small Au particles
FIG. 5. TPD profile for Au/TiO2 (Au loading, 0.98 wt% ) and TiO2.
Pretreatment for Au/TiO2 and TiO2: (a) in O2 at 573K for 1 h, (b) in O2
at 573K for 1 h, followed by exposure to C3H6 pulses up to full uptake,
(c) in O2 at 573 K for 1 h, followed by exposure to He stream containing
H2O at 2.3 vol% for 10 min, then to C3H6 pulses up to full uptake, (d) in
O2 at 573 K for 1 h, followed by exposure to He stream containing H2O at
2.3 vol% for 20 min, then to C3H6 pulses up to full uptake. Heating rate,
5 K/min; rate of flow, 50 ml/min; sample weight, 200 mg.
hand, for Au loadings smaller than 0.10 wt% hydrogenation behave as a peroxo-like adsorbed species, which enhances
is catalyzed to produce C3H8 alone. Both PO and C3H8 were the dissociation of hydrogen molecules (42). Moreover, the
competitively produced for metal loadings of 0.20 wt% and products, epoxides from alkenes, are similar to those of the
0.10 wt% , and in the latter case C3H8 tended to exceed oxidations of hydrocarbons by H2O2 over titanium silicates,
PO. The reaction products for each catalyst never changed on which a peroxo or hydroperoxo species is regarded to
under the different reaction conditions investigated. These be responsible for the selective oxidation (4, 43). From this
results show that the reaction pathway switches over from we speculate that the active oxygen species may be a sim-
oxidation to hydrogenation over Au/TiO2 catalysts at a crit- ilar peroxo-like species, formed on a Ti cation site at the
ical Au loading of 0.20 wt% .
boundary between the Au particles and the TiO2 support.
Since the reaction pathway had been assumed to be cor- This assumption can also explain why Au/TiO2 prepared
related with the size of Au particles, we made careful TEM by impregnation does not lead to PO formation; the larger
observation for every catalyst. In the case of 0.98 wt% and spherical Au particles simply loaded on TiO2 have much
0.39 wt% Au samples, Au particles having hemispherical weaker interactions and a much shorter perimeter inter-
shape were homogeneously dispersed with diameters of face with the TiO2 on which the peroxo-like species may be
1–4 nm. Although the number of Au particles observed was produced.
not large for the catalysts with loadings below 0.10 wt% , a
Figure 6 shows a probable formation route for the active
major difference between Au loadings above 0.39 wt% and oxygen species on the Au/TiO2 catalyst. Although it is spec-
below 0.10 wt% is whether the majority of particles have ulative at the present stage of our research, at the border be-
diameters above or below 2 nm. The sample prepared with tween the gold particless and the TiO2 support there may be
3+
4+==
0
O Au ⇔ Ti -O-Au+.
...
a Au loading of 0.20 wt% appeared to have smaller particles an equilibrium state between Ti
than those with lower Au loadingsand the mean diameter of The presence of positively charged gold species has been

EPOXIDATION OF PROPYLENE
573
TABLE 2
Product Yield for the Reaction of Propylene with Oxygen and Hydrogen over Au/TiO2
with Different Gold Loadings and the Results of TEM Observation
Yield of products,b %
Propylene
TEM observaiton
Mean diameter, Most frequent
Au loadings,a
wt%
Total number of
Au particles
Standard
deviation, %
Propane
oxide
CO2
nm
diameter, nm
0.98
0.39
0.20
0.10
0.05
0.02
—
0.61
[0.109]c
(11.1)d
0.56
[0.100]
(25.6)
0.54
[0.096]
(48.0)
0.1
[0.018]
(18.0)
-
0.08
[0.014]
(1.4)
0.20
[0.036]
(9.2)
0.05
[0.009]
(4.5)
-
446
258
393
35
2.4
2.2-2.4
25
30
25
36
25
25
-
2.2
1.5
1.8
1.5
1.7
2.2-2.4
1.2-1.4
1.0-1.2
0.8-1.0
1.4-1.6
0.33
[0.059]
(29.5)
1.9
[0.346]
(346)
3.1
[0.554]
(1108)
2.5
-
-
79
-
80
[0.446]
(2230)
a Actual Au loadings determined by X-ray fluorescence analysis.
b Reaction conditions: feed gas, C3H6/O2/H2/Ar = 10/10/10/70; flow rate, 2000ml/hr; catalyst, 0.5 g; temperature, 353K.
c Formation rate (mmol/hr/g-cat) is given in parenthesis [ ].
d Formation rate (mmol/hr/g-Au) is given in parenthesis ( ).
indicated by a smaller carbonyl peak at a higher wavenum- substrate, hydrocarbon:
ber region (ꢂ CO) in FT-IR (44). Molecular oxygen is taken
up by a Ti3+ cation site and is activated, probably to a
negatively charged molecular oxygen species, which forms
S0 + O2 + 2H+ + 2eꢂ → S0O + H2O.
[2]
hydropeoxo- or peroxo-like species directly through reac- Dioxygen is transformed to an activated species by the
tion with hydrogen. The oxygen species adsorbed on Ti site interaction with two protons and electrons. One oxygen
can react with C3H6 mainly adsorbed on the surface of Au atom is donated to the organic substrate and the other to
particles to produce PO.
form water. The catalytic features of Au/TiO2 catalysts ob-
Monooxygenase enzymes can excellently catalyze the ox- served show reasonable similarity to the enzymes in terms
idation ofalkenesto epoxidesat ambient temperatureswith of the reductive activation of dioxygen with molecular hy-
high selectivities through the reductive activation of molec- drogen and, also, by the presence of multifunctional active
ular oxygen (45, 46), as expressed in Eq. [2], where So is a sites (46). The key role of Au appears to be in providing
FIG. 6. A probable formation route for the active oxygen species on Au/TiO2 catalyst.

574
HAYASHI, TANAKA, AND HARUTA
binding sites for C3H6 and to aid electron transfer from
H2 to O2.
REFERENCES
1. Ainsworth, S. J., C&EN, March 2, 9 (1992).
2. Roth, J. F., CHEMTECH, June, 357 (1991).
3. For example, U.S. Patent 5,241,088 (1993) and Europ. Patent
0640598A1 (1994).
Several attempts have been reported where the behav-
ior of monooxygenase enzymes have also been successfully
mimicked concerning the activation of molecular oxygen.
Pd-Fe zeolite (47) and Pd-titanosilicate (6, 48) can catalyze
the oxidation of both alkanes and alkenes to oxygenates
in the liquid phase, by using H2O2 generated in situ. Vapor
phase processes by use of oxygen and hydrogen have also
been reported for CH4 over FePO4 (49) and benzene over
Pd-Cu (50).
The occurrence of hydrogenation of C3H6 over Au/TiO2
with very low Au loadings appears consistent with the re-
sults reported by Naito et al. for Au/SiO2 catalyst (42),
where using very low Au loadings it was found that the
4. Notari, B., Catal. Today 18, 163 (1993).
5. E. Occhiello, Chem. Ind., October 6, 761 (1997).
6. Sato, A., Miyake, T., and Saito, T., Shokubai (Catalyst) 34(2), 132
(1992).
7. Hayashi, T., Han, L., Tsubota, S., and Haruta, M., Ind. Eng. Chem.
Res. 34, 2298 (1995).
8. Otsuka, K., Yamanaka, I., and Ushiyama, T., Shokubai (Catalyst)
37(6), 408 (1995).
9. Yoshida, H., Tanaka, T., Yamamoto, M., Funabiki, T., and Yoshida, S.,
J. Chem. Soc., Chem. Commun., 2125 (1996).
10. Schwank, J., Gold Bull. 16, 103 (1983).
11. Hammer, B., and Nørskov, J. K., Nature 376, 238 (1995).
hydrogenation of olefins was accelerated by the presence 12. Haruta, M., Catal. Today 36, 153 (1997).
13. Haruta, M., Catal. Surveys Jpn. 1, 61 (1997).
14. Haruta, M., Yamada, N., Kobayashi, T., and Iijima, S., J. Catal. 115,
of oxygen. Our work has further demonstrated that the re-
action of C3H6 switches between partial oxidation and hy-
drogenation, depending on the size of Au particles. Flat
Au particles with diameters smaller than 2 nm might be
301 (1989).
15. Haruta, M., Tsubota, S., Kobayashi, T., Kageyama, H., Genet, M. J.,
and Delmon, B., J. Catal. 144, 175 (1993).
oxidic in nature in the copresence of O2, namely, positively 16. Tanielyan, S. K., and Augustine, R. L., Appl. Catal. A: General 85, 73
(1992).
charged to a certain extent, whereasAu particleslarger than
2 nm might be still metallic. The positively charged state of
Au will have electronically some similarities to that of Pt
metal and can catalyze hydrogen dissociation and then hy-
17. Lin, S. D., Bollinger, M., and Vannice, M. A., Catal. Lett. 17, 245
(1993).
18. Baiker, A., Maciejewski, M., Tagliaferri, S., and Hug, P., J. Catal. 151,
407 (1995).
drogenation. It is probable that Au can remain oxidic only 19. Kang, Y., and Wan, B., Appl. Catal. A: General 128, 53 (1995).
20. Hoflund, G. B., Gardner, S. D., Schryer, D. R., Upchurch, B. T., and
Kielin, E. J., Appl. Catal. B: Environ. 6, 117 (1995).
21. Yuan, Y., Asakura, K., Wan, H., Tsai, K., and Iwasawa, Y., Catal. Lett.
in the presence of O2 which is electrophilic and therefore
O2 isnecessary for hydrogenation over Au/TiO2. Further in-
vestigation is now underway to understand the structure–
42, 15 (1996).
catalytic function relations and to describe the concerted
mechanism of metal and support.
22. Cant, N. W., and Ossipoff, N. J., Catal. Today 36, 125 (1997).
23. Haruta, M., and Ueda, A., in “Proc. Intern. Workshop on Catal.
Comb., Tokyo, 1994,” p. 2.
24. Haruta, M., Ueda, A., Tsubota, S., and Sanchez, R. M. T., Catal. Today
29, 443 (1996).
CONCLUSION
25. Sakurai, H., and Haruta, M., Appl. Catal. A: General 127, 93 (1995).
26. Sakurai, H., and Haruta, M., Catal. Today 29, 361 (1996).
27. Andreeva, D., Idakiev, V., Tabakova, T., and Andreev, A., J. Catal.
158, 354 (1996).
28. Sakurai, H., Ueda, A., Kobayashi, T., and Haruta, M., J. Chem. Soc.,
Chem. Commun., 271 (1997).
29. Ueda, A., Oshima, T., and Haruta, M., Appl. Catal. B: Environ. 12, 81
(1997).
30. Zavarykina, L. V., Yushkova, E. I., Kazakova, G. D., and Mardashev,
Yu. S., Russian J. Phys. Chem. 68, 322 (1994).
The selective partial oxidation of propylene to propy-
lene oxide has been achieved over a supported gold cata-
lyst in a vapor-phase containing H2 and O2. Combination
of Au and TiO2 is indispensable for the selective oxida-
tion. Over Au/TiO2, the presence of H2 not only enhances
the oxidation of C3H6 but also leads to the epoxidation to
produce propylene oxide at 303 K to 393 K with selectiv-
ities above 90% . It appears that the reaction pathway can
be tuned, by carefully controlling the size of the Au par-
ticles and switches from oxidation to hydrogenation at a
critical size of around 2 nm. Although no direct experi-
mental evidence was obtained, it can be suggested that the
active oxygen species are produced by the reductive activa-
tion of molecular oxygen with molecular hydrogen at the
perimeter boundary between the Au particles and the TiO2
support.
31. Hodges, C. N., and Roselaar, L. C., J. Appl. Chem. Biotechnol. 25, 609
(1975).
32. Blues, E. T., Bryce-Smith, D., Lawston, I. W., and Wall, G. D., J. Chem.
Soc., Chem. Commun., 513 (1974).
33. Cant, N. W., and Hall, W. K., J. Phys. Chem. 75, 2914 (1971).
34. Sinfelt, J. H., and Cusumano, J. A., in “Advanced Material in Cata-
lysis” (J. J. Burton and R. L. Garten, Eds.), p. 1. Academic Press, New
York, 1977.
35. Nakamura, R., Suzuki, S., Aida, R., and Niiyama, H., in “Proc. 24th
Meet. Chem. Engng. Soc. Japan. I, 1991,” p. 166.
36. Manara, G., and Parravano, G., J. Catal. 23, 379 (1971).
37. Manara, G., and Parravano, G., J. Catal. 32, 72 (1974).
38. Tsubota, S., Cunningham, D. A. H., Bando, Y., and Haruta, M., in
“Preparation ofCatalystsVI” (G. Poncelet et al., Eds.), p. 229. Elsevier,
Amsterdam, 1995.
ACKNOWLEDGMENT
We thank Mrs. T. Nakamura for performing the particle size analysis.

EPOXIDATION OF PROPYLENE
575
39. Bamwenda, G. R., Tsubota, S., Nakamura, T., and Haruta, M., Catal. 44. Boccuzzi, F., Chiorino, A., Tsubota, S., and Haruta, M., J. Phys. Chem.
Lett. 44, 83 (1997). 100, 3625 (1996).
40. Iizuka, Y., Fujiki, H., Yamauchi, N., Chijiiwa, T., Arai, S., Tsubota, S., 45. Green, J., and Dalton, H., J. Biol. Chem. 264, 17698 (1989).
and Haruta, M., Catal. Today 36, 115 (1997). 46. White, R. E., and Coon, M. J., Ann. Rev. Biochem. 49, 315 (1980).
41. Olivera, P. P., Patrito, E. M., and Sellers, H., Surf. Sci. 313, 25 47. Herron, N., and Tolman, C. A., J. Am. Chem. Soc. 109, 2837 (1987).
(1994).
48. Tastumi, T., Yuasa, K., and Tominaga, H., J. Chem. Soc., Chem. Com-
mun., 1446 (1996).
42. Naito, S., and Tanimoto, M., J. Chem. Soc., Chem. Commun., 832
(1988).
49. Wang, Y., and Otsuka, K., J. Catal. 155, 256 (1995).
43. Huybrechts, D. R. C., Bruycker, L. De., and Jacobs, P. A., Nature 345, 50. Kitano, T., Wani, T., Ohnishi, T., Jiang, L-F., Kuroda, Y., Kunai, A.,
240 (1990). and Sasaki, K., Catal. Lett. 11, 11 (1991).