Active site and reaction mechanism for the epoxidation of propylene by oxygen over CuOx/SiO2 catalysts with and without Cs + modification

Article abstract of DOI:10.1016/j.jcat.2012.11.032

Among alkali metal ions, Cs⁺ was found to be the most efficient promoter of the CuO_x/SiO₂ catalyst for the epoxidation of propylene to propylene oxide (PO) by O₂. Stronger interactions between Cs⁺ and CuO_x nanoparticles were proposed to favor the selective formation of PO. Kinetic studies indicated that PO was formed as a primary product in parallel with acrolein over the CuO_x/SiO ₂ catalyst, but PO readily underwent isomerization to allyl alcohol, followed by oxidation to acrolein, without Cs⁺ modification. The modification by Cs⁺ inhibited the isomerization of PO because of the weakened acidity, contributing to the increase in PO selectivity. The inhibition of the reactivity of the lattice oxygen in the presence of Cs⁺ also promoted PO selectivity by suppressing the allylic oxidation route. In situ X-ray diffraction, CO-adsorbed Fourier-transform infrared, and pulse-reaction studies suggest that Cu^I sites generated during the reaction account for the epoxidation of propylene by O₂.

Full text of DOI:10.1016/j.jcat.2012.11.032

Journal of Catalysis 299 (2013) 53–66
Contents lists available at SciVerse ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Active site and reaction mechanism for the epoxidation of propylene by oxygen over
CuOx/SiO2 catalysts with and without Cs+ modification
⇑
Jieli He, Qingge Zhai, Qinghong Zhang, Weiping Deng, Ye Wang
State Key Laboratory of Physical Chemistry of Solid Surfaces, National Engineering Laboratory for Green Chemical Productions of Alcohols, Ethers and Esters, College of Chemistry
and Chemical Engineering, Xiamen University, Xiamen 361005, China
a r t i c l e i n f o
a b s t r a c t
Among alkali metal ions, Cs+ was found to be the most efficient promoter of the CuOx/SiO2 catalyst for the
epoxidation of propylene to propylene oxide (PO) by O2. Stronger interactions between Cs+ and CuOx
nanoparticles were proposed to favor the selective formation of PO. Kinetic studies indicated that PO
was formed as a primary product in parallel with acrolein over the CuOx/SiO2 catalyst, but PO readily
underwent isomerization to allyl alcohol, followed by oxidation to acrolein, without Cs+ modification.
The modification by Cs+ inhibited the isomerization of PO because of the weakened acidity, contributing
to the increase in PO selectivity. The inhibition of the reactivity of the lattice oxygen in the presence of Cs+
also promoted PO selectivity by suppressing the allylic oxidation route. In situ X-ray diffraction,
CO-adsorbed Fourier-transform infrared, and pulse-reaction studies suggest that CuI sites generated
during the reaction account for the epoxidation of propylene by O2.
Article history:
Received 15 September 2012
Revised 21 November 2012
Accepted 22 November 2012
Available online 11 January 2013
Keywords:
Propylene
Molecular oxygen
Epoxidation
Allylic oxidation
Selectivity control
Copper oxide catalyst
Cesium
Ó 2012 Elsevier Inc. All rights reserved.
CuI active site
1. Introduction
be obtained [3,19,20]. It is believed that the main difficulty arises
from the higher reactivity of the allylic C–H bonds in C3H6. The
Propylene oxide (PO) is an important bulk chemical for the pro-
duction of polyether polyols and propylene glycol. Currently, PO is
produced by the chlorohydrin and organic hydroperoxide pro-
cesses in industry, but these processes produce large amounts of
by-products together with PO and are not green chemical pro-
cesses. Many studies have been devoted to the catalytic epoxida-
tion of propylene using a proper oxidant [1–6]. PO can be formed
with good selectivities using H2O2 as an oxidant in the liquid phase
[7], and the production of PO using H2O2 catalyzed by TS-1 has
been commercialized by Dow–BASF [8]. H2–O2 gas mixture
[6,9–11] and N2O [12–18] are also effective for the epoxidation
of propylene in the presence of a proper catalyst. However, only
limited success has been achieved for the epoxidation of propylene
by O2, which is the most desirable route and has been viewed as
the ‘‘Holy Grail’’ for PO production [2].
allylic oxidation would proceed more readily than the epoxidation
in the case of C3H6 oxidation by O2 over most heterogeneous cata-
lysts including Ag, leading to lower selectivity to PO [6]. Previous
studies suggested that larger Ag particles favored selectivity to
PO [3,19,20]. However, a recent communication claimed that Ag
nanoparticles ꢀ3.5 nm in size aggregated from Ag3 clusters depos-
ited on ultrathin Al2O3 could afford high PO selectivities at lower
C3H6 conversions [21].
Besides the Ag-based catalysts, Au- and Cu-based catalysts have
also been exploited for the epoxidation of C3H6 by O2 [6]. Recently, a
few research groups found that the supported Au nanoclusters
could catalyze the epoxidation of C3H6 by O2 in the presence of
H2O [6,22–24]. For example, the Au/TS-1 with a mean size of Au
nanoclusters of 1.8 nm exhibited a PO selectivity of 52% at C3H6 con-
version of 0.88% at 473 K [24]. It is proposed that smaller Au nanocl-
usters could activate O2 and H2O to form hydroperoxy species
(ꢁOOH), which may account for the epoxidation of C3H6 [22–24].
Surface chemistry studies over Ag and Cu single-crystal surfaces
have demonstrated that the Cu surface preadsorbed with oxygen
species is more selective than the Ag surface for the epoxidation
of alkenes with allylic C–H bonds [25,26]. It is found that only iso-
lated oxygen adatoms over Cu surfaces can lead to epoxidation
while islands of ‘‘oxidic’’ oxygen cannot [25,26]. A theoretical study
also reveals that Ag(111) favors allylic hydrogen stripping from
On the other hand, the Ag-catalyzed epoxidation of ethylene by
O2 has been commercialized for several decades [1–3]. Many stud-
ies have attempted to apply Ag-based catalysts to the epoxidation
of C3H6 by O2, but the selectivity of PO was very low under the
reaction conditions where reasonably high C3H6 conversions could
⇑
Corresponding author. Fax: +86 592 2183047.
E-mail address: wangye@xmu.edu.cn (Y. Wang).
0021-9517/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2012.11.032

54
J. He et al. / Journal of Catalysis 299 (2013) 53–66
C3H6, while Cu(111) favors the formation of propylene metallacy-
cle, an intermediate for PO formation, owing to the lower basicity
of adsorbed oxygen species on Cu(111) [27]. However, Monnier
and Hartley [28] once pointed out that Cu might be difficult to
use as a true epoxidation catalyst under reaction conditions be-
cause metallic Cu may be readily oxidized into CuO or Cu2O by
O2. Several studies have claimed the epoxidation of C3H6 by O2
over metallic Cu (Cu0)-based catalysts [29,30]. High PO selectivities
(40–50%) can be obtained only at very low C3H6 conversions
(<0.3%), and the increase in C3H6 conversion to >1% with increasing
reaction temperature or O2 partial pressure led to a decrease in PO
selectivity to <5% over these catalysts, possibly due to the transfor-
mation of Cu0 to oxidized copper species (CuI or CuII) [29,30]. A re-
cent study showed that the addition of a small amount of Cu to Ag/
BaCO3 to form Cu–Ag bimetallic catalysts significantly promoted
the epoxidation of C3H6 by O2 [31].
On the other hand, recently, a few copper oxide (CuOx)-based
catalysts have been reported to show interesting catalytic behavior
for the epoxidation of C3H6 by O2. We found that CuOx/SBA-15 and
CuOx/SiO2 modified with K+, which started from CuII and did not
undergo any reductive pretreatments, could catalyze the epoxida-
tion of C3H6 by O2 [32–34]. The epoxidation occurred efficiently
even under O2-rich conditions over our catalysts, and the PO for-
mation rate was one order of magnitude higher than those ob-
tained over the previously reported Cu0-based catalysts [29,30].
The superior performance of these supported CuOx catalysts was
further demonstrated by other groups [35,36]. The modification
of CuOx with VOx or RuOx could enhance its activity and selectivity
for PO formation [37–40].
However, it is noteworthy that supported CuOx is also known as
a representative catalyst for the allylic oxidation of C3H6 to acrolein
by O2 [41–44]. It is generally believed that the allylic oxidation
competes with the epoxidation. There is little knowledge of selec-
tivity control in the oxidation of C3H6 by O2. Therefore, it is of great
significance to gain insights into the factors controlling the reac-
tion route for the formation of acrolein or PO over the CuOx-based
catalysts.
The present work is a continuation of our previous studies on
C3H6 oxidation by O2 over CuOx-based catalysts. In our previous
studies [32–34], we did not pay much attention to reaction net-
works over supported CuOx catalysts. The first aim of the present
paper is to clarify the reaction routes for the oxidation of C3H6 over
CuOx/SiO2 catalysts in the absence and in the presence of an alkali
metal ion modifier (particularly Cs+) through kinetic studies. The
roles of the alkali metal ion in affecting the reaction route will be
discussed. Furthermore, although we have proposed that CuI is
the active site for the epoxidation of C3H6 by O2 [34], further evi-
dence is required. This paper will report our recent studies on
the chemical states of copper under working conditions through
in situ X-ray diffraction (XRD) and CO-adsorbed Fourier-transform
infrared (FT-IR) studies. The roles of oxygen species in the forma-
tion of acrolein or PO will also be discussed through pulse-reaction
studies.
and then, the mixture was kept at 343 K under vigorous stirring.
A blue gel was finally obtained. After aging at room temperature
for 24 h and further at 373 K for 48 h, the gel was calcined at
823 K in air for 6 h to obtain the catalyst.
In this paper, besides the catalyst starting from CuII, catalysts
containing Cu0 and CuI as the starting copper species were also
prepared, to gain insights into the changes of Cu0 and CuI under
our reaction conditions and the possible active sites. The catalyst
containing Cu0 was produced by treating the catalyst prepared
above in a H2–He gas flow [P(H2) = 10.1 kPa] at 773 K for 30 min,
while that containing CuI was obtained by treating this reduced
catalyst in a N2O–He gas flow [P(N2O) = 20.3 K] at 573 K for
30 min. Generally, a lower temperature is employed in N2O oxida-
tion for the generation of CuI on the surfaces of Cu0 particles while
the bulk is kept as Cu0 [45]. We found that a higher temperature
(573 K) used for N2O oxidation could transform the whole Cu0 par-
ticles in the reduced catalyst to Cu2O particles. We confirmed that
no CuII or CuO was generated during the oxidation of the reduced
catalyst by N2O under our conditions.
2.2. Catalytic reaction
Catalytic reactions were carried out on a fixed-bed reactor oper-
ated under atmospheric pressure. The catalyst was pretreated in
the quartz reactor with a gas flow containing He (40 mL minꢂ1
)
and O2 (10 mL minꢂ1) at 823 K for 0.5 h, followed by purging with
He (60 mL minꢂ1) at the same temperature for another 0.5 h. After
the catalyst was cooled down to the reaction temperature (473–
573 K), the reactant gas mixture of C3H6 and O2 was introduced into
the reactor to start the reaction. In some cases, He was also used as a
balance gas in the reactant mixture to regulate the partial pressures
of C3H6 and O2. The products were analyzed by two online gas chro-
matographs equipped with three columns. The separation of PO,
acrolein, allyl alcohol, propanal, acetone, and acetaldehyde was
performed by a capillary column (FFAP, 50 m ꢃ 0.53 mm ꢃ 1.0
lm)
equipped with a flame ionization detector. The separation and
detection of other components, such as O2, C3H6, CO, and CO2, were
performed by Porapak Q and Molecular Sieve 5A columns and ther-
mal conductivity detectors. All the lines and valves between the
exit of the reactor and the gas chromatographs were heated to
393 K to prevent condensation of organic products.
Pulse reactions were performed to gain information about the
reactivity of the lattice oxygen and the initial catalytic behavior
of the catalysts pretreated under different conditions. For the pulse
reaction, He, typically with a flow rate of 80 mL minꢂ1, was used as
the carrier gas, and the pulse volume was typically fixed at 0.63 mL
(STP). Other pretreatment and reaction conditions were the same
as those used for the flow reaction.
2.3. Catalyst characterization
XRD measurements were carried out on a Panalytical X’pert Pro
Super X-ray diffractometer with Cu K
a radiation (40 kV and
30 mA). For in situ XRD measurements, the sample was loaded into
an XRK-900 cell, which was directly attached to the X-ray diffrac-
tometer. N2 sorption at 77 K was performed with a Micromeritics
TriStar 3000 surface area and porosimetry analyzer. The sample
was pretreated at 573 K in vacuum for 3 h before N2 adsorption.
The surface area was calculated using the BET method. Transmis-
sion electron microscopy (TEM) was performed on a Philips Analyt-
ical FEI Tecnai 30 electron microscope operated at an acceleration
voltage of 300 kV or on a JEM-2100 electron microscope operated
at an acceleration voltage of 200 kV. Samples for TEM observation
were suspended in ethanol and dispersed ultrasonically. Drops of
suspensions were applied on a copper grid coated with carbon.
2. Experimental
2.1. Catalyst preparation
The CuOx/SiO2 and the alkali-metal-ion-modified CuOx/SiO2 cat-
alysts were prepared by a sol–gel method. Briefly, an aqueous solu-
tion of alkali metal ion precursor (typically alkali metal carbonate)
was first added slowly into a mixed aqueous solution of Cu(NO3)2
and ethylene glycol (20 mL), to obtain a homogeneous solution
with a total volume of 60 mL. Subsequently, tetraethyl orthosili-
cate (TEOS, 10 g) was added dropwise into the mixed solution,

J. He et al. / Journal of Catalysis 299 (2013) 53–66
55
H2 temperature-programmed reduction (H2 TPR) and NH3 tem-
perature-programmed desorption (NH3 TPD) were carried out
using a Micromeritics AutoChem II 2920 instrument. Typically,
the sample (0.1 g) was pretreated in a quartz reactor with a gas
flow containing O2 and He at 823 K for 1 h, followed by purging
with high-purity He. For H2 TPR measurements, after the sample
was cooled down to 303 K, a H2–Ar (5 vol.% H2) mixture was intro-
duced into the reactor and the temperature was raised to 1173 K at
a rate of 10 K minꢂ1. The consumption of H2 was monitored by a
thermal conductivity detector. For NH3 TPD measurements, the
adsorption of NH3 was performed at 393 K in an NH3–He mixture
(10 vol.% NH3) for 1 h, and the remaining or weakly adsorbed NH3
was purged with high-purity He. TPD was performed in He flow by
raising the temperature to 1000 K at a rate of 10 K minꢂ1. The des-
orbed NH3 was detected with a mass spectrometer (ThermoStar
GSD 301 T2) by monitoring the signal with m/e = 16.
Table 2 shows the effect of the Cs/Cu molar ratio on catalytic
performances of the Cs+-modified CuOx/SiO2 catalysts in the epox-
idation of C3H6 by O2 at 523 K. It should be mentioned that no con-
version of C3H6 has been observed over Cs+/SiO2 without loading
CuOx, confirming that CuOx accounts for the oxidation of C3H6 by
O2. Table 2 clarified that the addition of Cs+ to the CuOx/SiO2 cata-
lyst, even with a low content (Cs/Cu = 0.1), caused a decrease in the
conversion of C3H6. However, as the ratio of Cs/Cu rose from 0.1 to
0.4, the conversion of C3H6 did not change significantly. At the
same time, the selectivity to PO increased gradually from 15% to
34%. A further increase in the Cs/Cu ratio decreased both C3H6 con-
version and PO selectivity. Thus, the Cs+–5 wt% CuOx/SiO2 catalyst
with a Cs/Cu ratio of 0.4 afforded the highest PO yield. Accompa-
nied by an increase in the selectivity to PO, the selectivity to acro-
lein declined.
In situ FT-IR spectroscopy studies of CO adsorption were carried
out with a Nicolet FTIR 380 or Nicolet 6700 instrument equipped
with an MCT detector. Transmission FT-IR spectra were recorded
with a resolution of 4 cmꢂ1. The sample was pressed into a self-
supporting wafer and was placed in an in situ IR cell, where it could
be treated directly under different conditions.
3.2. Kinetic and reaction route studies for 5 wt% CuOx/SiO2 catalysts
with and without Cs+ modification
Fig. 1 shows the dependence of catalytic performance on reaction
temperature under O2-rich conditions, that is, P(O2)/P(C3H6) = 39/1
[P(O2) = 98.8 kPa, P(C3H6) = 2.53 kPa], for the 5 wt% CuOx/SiO2 and
the Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalysts. The conversion of
C3H6 over both catalysts increased almost exponentially with reac-
tion temperature. The apparent activation energies of C3H6 conver-
sions calculated using the Arrhenius plots were 64 and 56 kJ molꢂ1
over the CuOx/SiO2 and the Cs+–CuOx/SiO2 catalysts, respectively.
Regarding the product selectivity, acrolein was the main product
at low temperatures under O2-rich conditions over the 5 wt% CuOx/
SiO2 catalyst without Cs+ modification (Fig. 1A). Besides acrolein and
COx, PO and allyl alcohol were also formed with lower selectivities.
The increase in temperature decreased the selectivities to acrolein,
PO, and allyl alcohol and increased that to COx. After the modifica-
tion by Cs+, PO was formed as a major product instead of acrolein
(Fig. 1B). PO selectivity of 53% could be obtained at 473 K (C3H6 con-
version, 1.7%). Although the selectivity became significantly lower,
acrolein was also formed after the modification by Cs+. However,
no allyl alcohol was observed even at lower temperatures. The in-
crease in temperature decreased the selectivity to PO, but a PO selec-
tivity of ꢀ23% could still be sustained at a C3H6 conversion of 17% at
573 K, providing a PO yield of 3.9%.
We have also investigated the effect of temperature on catalytic
behavior over the 5 wt% CuOx/SiO2 and the Cs+–5 wt% CuOx/SiO2
(Cs/Cu = 0.4) catalysts under conditions of P(O2)/P(C3H6) = 1
[P(O2) = P(C3H6) = 50.7 kPa]. Under such conditions, the conversion
of C3H6 over both catalysts was lower (Fig. 2). From the Arrhenius
plots derived from the C3H6 conversion rates versus reaction tem-
perature, the apparent activation energies for the catalysts with
and without Cs+ modification were evaluated to be 63 and
62 kJ molꢂ1, similar to those under the O2-rich conditions. How-
ever, the selectivity patterns became somewhat different at
3. Results
3.1. Catalytic behavior of Cs+-modified CuOx/SiO2 catalysts
Table 1 shows the effect of the modification of the 5 wt% CuOx/
SiO2 catalyst by various alkali metal ions on the catalytic perfor-
mance in epoxidation of C3H6 by O2. The CuOx/SiO2 without mod-
ification exhibited a high C3H6 conversion (15%) but a low PO
selectivity (1.8%). The main partial oxidation product was acrolein
(selectivity 49%) over this catalyst. This is consistent with the con-
sensus that the supported CuOx catalyzes the allylic oxidation of
C3H6 [41–44]. The modification of CuOx/SiO2 by an alkali metal
ion decreased the conversion of C3H6 and the selectivity to acro-
lein. At the same time, the selectivity to PO increased significantly,
and the yield of PO also increased remarkably despite the decrease
in C3H6 conversion. With the alkali metal ion changed from Li+ to
Cs+ in the modified catalysts, although there was no definite trend
in the change of C3H6 conversion, the selectivity to PO increased in
the order Li+ < Na+ ꢄ K+ < Rb+ < Cs+. The selectivity to PO over the
Cs+ꢂCuOx/SiO2 catalyst reached 34% at a C3H6 conversion of 7.5%.
A PO yield of 2.6% was attained, which was ꢀ10 times higher than
that over the 5 wt% CuOx/SiO2 catalyst. Although the selectivity to
COx was increased by modification with alkali metal ions, the Cs+-
ꢂCuOx/SiO2 catalyst exhibited the lowest selectivity to COx among
all the alkali-metal-ion-modified catalysts displayed in Table 1.
This catalyst exhibited a better PO formation activity than the
K+–CuOx/SBA-15 catalyst reported in our previous studies, which
provided PO selectivity of ꢀ20% at a C3H6 conversion of ꢀ5% [32].
Table 1
Catalytic performance of 5 wt% CuOx/SiO2 catalysts modified by different alkali metal ions.a
Catalystb
Conversion (%)
Selectivity (%)
PO
PO yield (%)
Acrolein
Othersc
COx
CuOx/SiO2
15
1.8
12
22
20
25
49
4.6
0.3
0.5
0.4
0.3
0.8
42
84
70
75
68
57
0.27
0.70
1.8
1.4
1.4
Li+–CuOx/SiO2
Na+–CuOx/SiO2
K+–CuOx/SiO2
Rb+–CuOx/SiO2
Cs+–CuOx/SiO2
5.8
8.0
6.8
5.4
7.5
3.8
6.7
4.5
7.3
8.5
34
2.6
a
b
c
Reaction conditions: W = 0.2 g, T = 523 K, F = 60 mL minꢂ1, P(C3H6) = 2.53 kPa, P(O2) = 98.8 kPa.
The molar ratio of alkali metal ion to Cu is 0.4.
Other oxygenates include allyl alcohol, propanal, acetone, and acetaldehyde.

56
J. He et al. / Journal of Catalysis 299 (2013) 53–66
Table 2
Effect of the molar ratio of Cs/Cu on catalytic performance of Cs+–5 wt% CuOx/SiO2 catalysts.a
Cs/Cu (Molar ratio)
Conversion (%)
Selectivity (%)
PO
PO yield (%)
Acrolein
Othersb
COx
0
15
1.8
15
21
34
28
49
35
27
8.5
7.1
4.6
1.6
0.7
0.8
1.6
42
48
51
57
64
0.27
1.1
1.5
2.6
1.4
0.1
0.2
0.4
0.6
7.1
7.3
7.5
5.1
a
Reaction conditions: W = 0.2 g, T = 523 K, F = 60 mL minꢂ1, P(C3H6) = 2.53 kPa, P(O2) = 98.8 kPa.
Other oxygenates include allyl alcohol, propanal, acetone, and acetaldehyde.
b
Fig. 1. Dependence of catalytic performance on reaction temperature: (A) 5 wt%
CuOx/SiO2; (B) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4). Reaction conditions: W = 0.20 g,
F = 60 mL minꢂ1, P(C3H6) = 2.53 kPa, P(O2) = 98.8 kPa.
Fig. 2. Dependence of catalytic performance on reaction temperature: (A) 5 wt%
CuOx/SiO2; (B) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4). Reaction conditions: W = 0.20 g,
F = 60 mL minꢂ1, P(C3H6) = P(O2) = 50.7 kPa.
the Cs+–5 wt% CuOx/SiO2 catalyst, and a PO selectivity of 63% was
P(O2)/P(C3H6) = 1. For the 5 wt% CuOx/SiO2 catalyst, although acro-
lein was still a main product, allyl alcohol, and PO were also
formed with considerable selectivity (>20% and >10%, respectively)
at low temperatures (Fig. 2A). The increase in the temperature de-
creased the selectivity to allyl alcohol and PO and increased that to
acrolein and COx. On the other hand, over the Cs+-modified CuOx/
SiO2 catalyst at P(O2)/P(C3H6) = 1, almost no allyl alcohol was ob-
served (Fig. 2B). However, the selectivity to acrolein also became
considerable (25–33%), although PO was still the main product.
To gain information about the reaction routes over the 5 wt%
CuOx/SiO2 and the Cs+ꢂ5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalysts,
we have investigated the dependence of catalytic performance on
the pseudo contact time, which is expressed as the ratio of catalyst
weight to total flow rate (W/F). Fig. 3 shows the results at 523 K un-
der O2-rich conditions [P(O2)/P(C3H6) = 39/1]. The conversion of
C3H6 increased almost proportionally to the contact time over both
catalysts. We evaluated the rates of C3H6 conversion, r(C3H6), using
the slopes of these straight lines; r(C3H6) were 2.9 and
1.6 mmol hꢂ1 gꢂ1 at 523 K over the catalysts with and without
Cs+ modification, respectively. Thus, the modification by Cs+ de-
creased the C3H6 conversion rate by a factor of 0.55. The main
products were acrolein and COx over the 5 wt% CuOx/SiO2, and
the increase in the contact time decreased the selectivity to
acrolein and increased that to COx. PO was a main product over
attained at a C3H6 conversion of 1.1% at
a shorter contact
time (Fig. 3B). Acrolein could still be obtained with a selectivity
of >10% at shorter contact times. The increase in the contact time
decreased the selectivity to PO and acrolein and increased that to
COx.
Fig. 4 shows the dependence of the catalytic performance of the
5 wt% CuOx/SiO2 and the Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) cata-
lysts on the contact time at 523 K at P(O2)/P(C3H6) = 1/1. From
the linear relationship between C3H6 conversion and contact time,
the rates of C3H6 conversion were calculated to be 5.3 and
5.1 mmol gꢂ1 hꢂ1 for the 5 wt% CuOx/SiO2 and the Cs+–5 wt% CuOx/
SiO2, respectively. Thus, although C3H6 conversion was lower at
P(O2)/P(C3H6) = 1/1, the rates of C3H6 conversion were higher.
Moreover, the difference in the rates of C3H6 conversion between
catalysts with and without Cs modification was smaller. At shorter
contact times, PO and allyl alcohol also formed with considerable
selectivities, besides acrolein, over the 5 wt% CuOx/SiO2 catalyst.
The increase in the contact time decreased the selectivity to PO
and increased that to allyl alcohol and acrolein slightly. The selec-
tivity to COx also increased with the contact time. On the other
hand, over the Cs+–5 wt% CuOx/SiO2 catalyst, almost no allyl alco-
hol was formed over the whole range of contact times. PO and
acrolein were both formed, and the selectivity to PO was signifi-

J. He et al. / Journal of Catalysis 299 (2013) 53–66
57
Fig. 3. Dependence of catalytic performance on contact time: (A) 5 wt% CuOx/SiO2;
(B) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4). Reaction conditions: W = 0.025–0.30 g,
T = 523 K, F = 60 mL minꢂ1, P(C3H6) = 2.53 kPa, P(O2) = 98.8 kPa.
Fig. 5. Effect of partial pressure of O2 on catalytic performance: (A) 5 wt% CuOx/
SiO2; (B) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4). Reaction conditions: W = 0.20 g,
T = 523 K, F = 60 mL minꢂ1, P(C3H6) = 50.7 kPa.
reaction orders of 0.51 and 0.54 with respect to O2 over the cata-
lysts without and with Cs+ modification, respectively. For the cata-
lyst without Cs+ modification, acrolein, PO, and allyl alcohol were
all formed, as well as COx and small amounts of other products,
including propionaldehyde and acetaldehyde. With increasing
P(O2), the selectivity to PO and ally alcohol decreased and that to
acrolein increased. The selectivity to COx changed only slightly at
the same time. After modification with Cs+, the selectivity to PO be-
came higher, while that to acrolein became lower. Allyl alcohol was
formed with a lower selectivity only at lower P(O2). The increase in
P(O2) decreased the selectivity to allyl alcohol and increased that to
acrolein. The selectivity to PO and COx did not change significantly
with changing P(O2).
The effects of P(C3H6) on the rates of C3H6 conversion and the
product selectivity over the 5 wt% CuOx/SiO2 and the Cs+–5 wt%
CuOx/SiO2 (Cs/Cu = 0.4) catalysts at a fixed P(O2) (50.7 kPa) are dis-
played in Fig. 6. From the plots of the logarithm of r(C3H6) against
the logarithm of P(C3H6), we obtained reaction orders of 0.58 and
0.55 with respect to C3H6 over the catalysts without and with Cs+
modification, respectively. Thus, we could express the rate equa-
tions for these two catalysts as follows:
0:51
for 5 wt% CuOx=SiO2 : rðC3H6Þ ¼ k1 PðC3H6Þ0:58PðO2Þ
;
ð1Þ
for Csþ ꢂ 5 wt% CuOx=SiO2 : rðC3H6Þ ¼ k2PðC3H6Þ0:55PðO2Þ0:54; ð2Þ
Fig. 4. Dependence of catalytic performance on contact time: (A) 5 wt% CuOx/SiO2;
(B) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4). Reaction conditions: W = 0.025–0.30 g,
T = 523 K, F = 60 mL minꢂ1, P(C3H6) = P(O2) = 50.7 kPa.
where k1 and k2 were the apparent rate constants for the two cata-
lysts. At a fixed P(O2), an increase in P(C3H6) increased the selectiv-
ity to PO and acrolein and decreased that to COx over the 5 wt%
CuOx/SiO2 catalyst. The selectivity to allyl alcohol did not vary sig-
nificantly at the same time. Over the Cs+–5 wt% CuOx/SiO2 catalyst,
the increase in P(C3H6) also increased the selectivity to PO and acro-
lein and decreased that to COx, but almost no allyl alcohol was
formed in the whole range of P(C3H6) in this case.
cantly higher than that of acrolein. The increase in the contact time
decreased the selectivity to PO and increased that to COx.
Fig. 5 shows the effects of partial pressure of O2, that is, P(O2),
on catalytic performances over 5 wt% CuOx/SiO2 and Cs+–5 wt%
CuOx/SiO2 (Cs/Cu = 0.4) catalysts at a fixed partial pressure of
C3H6 [P(C3H6) = 50.7 kPa]. Over both catalysts, the rate of C3H6 con-
version increased with increasing P(O2). From the plots of the log-
arithm of r(C3H6) against the logarithm of P(O2), we obtained
To further clarify the reaction routes, we have performed the
conversion of PO over the 5 wt% CuOx/SiO2 and the Cs+–5 wt%
CuOx/SiO2 (Cs/Cu = 0.4) catalysts. Table 3 shows that PO could be

58
J. He et al. / Journal of Catalysis 299 (2013) 53–66
alcohol and acrolein were still two main products, the selectivity
to COx increased in most cases over the Cs+–5 wt% CuOx/SiO2 (Cs/
Cu = 0.4) catalyst.
3.3. Characterizations of fresh CuOx/SiO2 and alkali-metal-ion-
modified CuOx/SiO2 catalysts
The XRD patterns for the 5 wt% CuOx/SiO2 samples modified by
different alkali metal ions are displayed in Fig. 7A. A broad peak at
2h of 22°–23°, assignable to the amorphous silica, was observed for
all the samples. The 5 wt% CuOx/SiO2 exhibited two distinct diffrac-
tion peaks at 2h of 35.4° and 38.6°, and these peaks could be as-
cribed to the (111) and (111) reflections of monoclinic CuO.
Although these diffraction peaks still remained after modification
with alkali metal ions, the peaks became significantly weakened
and broadened, suggesting that the crystallite size of CuO became
smaller, possibly due to strong interactions between CuO and the
alkali metal ions. We have estimated the sizes of CuO crystallites
in these samples using the Scherrer equation, and the results are
summarized in Table 4. The size of CuO crystallites decreased dra-
matically from 29 to 7.8 nm after modification by Cs+. These obser-
vations indicate an enhancement in the dispersion of the CuO
phase after modification with alkali metal ions, although the BET
surface areas for the modified samples become lower (Table 4).
The XRD patterns for the Cs+–5 wt% CuOx/SiO2 catalysts with
different Cs/Cu ratios (Fig. 7B) showed that the diffraction peaks
ascribed to CuO crystallites became weakened and broadened with
an increase in the ratio of Cs/Cu. At a Cs/Cu ratio to 0.6, the diffrac-
tion peaks of CuO crystallites almost disappeared, and only a broad
peak at 2h of ꢀ23°, belonging to amorphous SiO2, was observed.
The crystallite sizes of CuO calculated by the Scherrer equation
are also listed in Table 4. The increase in the ratio of Cs/Cu de-
creased the size of CuO crystallites.
Fig. 8 displays the typical TEM micrographs for the 5 wt% CuOx/
SiO2 catalysts modified by different alkali metal ions. CuO nanopar-
ticles were found to be dispersed homogeneously on SiO2 over
each catalyst. The mean sizes of CuO nanoparticles in these sam-
ples have been evaluated by counting 100–200 particles and are
listed in Table 4. The mean size of CuO particles was 18 nm over
the 5 wt% CuOx/SiO2 catalyst, and modification with an alkali metal
ion significantly decreased the CuO particle size. With the alkali
metal ion changed from Li+ to Cs+, the mean size of the CuO nano-
particles decreased in the order Li+ > Na+ > K+ > Rb+ > Cs+. For the
Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalyst, the mean CuO particle
size was 5.8 nm. This tendency is in agreement with that observed
Fig. 6. Effect of partial pressure of C3H6 on catalytic performance: (A) 5 wt% CuOx/
SiO2; (B) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4). Reaction conditions: W = 0.2 g,
T = 523 K, F = 60 mL minꢂ1, P(O2) = 50.7 kPa.
converted considerably over the 5 wt% CuOx/SiO2 catalyst. An in-
crease in P(O2) at 523 K or an increase in temperature at P(O2) of
84 kPa increased the conversion of PO significantly over this cata-
lyst. Propanal was formed as a main product, together with allyl
alcohol and acrolein, in the absence of O2. The former two products
should be formed by the isomerization of PO [46], while acrolein
may be formed by the oxidation of allyl alcohol by lattice oxygen
on the catalyst surface. In the presence of O2, allyl alcohol and acro-
lein became two predominant products over the 5 wt% CuOx/SiO2
catalyst. An increase in temperature decreased the selectivity to al-
lyl alcohol and increased that to acrolein, confirming that acrolein
was formed by the oxidation of allyl alcohol. On the other hand,
over the Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalyst, the conversion
of PO became significantly lower. This clearly indicates that the
modification by Cs+ inhibits the conversion of PO. Although allyl
Table 3
Conversion of PO over 5 wt% CuOx/SiO2 and Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalysts.a
Catalyst
P(O2) (kPa)
T (K)
Conv. (%)
Selectivity (%)b
Propanal
Allyl alcohol
Acrolein
COx
CuOx/SiO2
0
8.4
34
84
84
84
84
523
523
523
523
473
498
548
9.7
24
31
48
12
25
80
67
0.5
0.7
0
1.6
0.4
0
19
15
10
11
42
30
4.5
12
59
70
75
47
57
73
0
14
16
10
0
5.5
20
Cs+–CuOx/SiO2
0
8.4
34
84
84
84
84
523
523
523
523
473
498
548
0.8
2.9
3.3
3.7
0.9
2.0
6.1
10
0
0
0
0
54
11
9.0
11
69
27
9
25
31
30
29
25
30
25
0
50
57
58
0
0
0
40
65
a
Reaction conditions: W = 0.05 g, F = 60 mL minꢂ1, P(PO) = 0.084 kPa, P(Ar) = 8.4 kPa. He was used as a balance gas.
Other products include acetaldehyde and acetone.
b

J. He et al. / Journal of Catalysis 299 (2013) 53–66
59
there are no acid sites on SiO2 prepared in our work. After the load-
ing of CuOx, a broad NH3 desorption peak appeared at ꢀ520 K, indi-
cating the generation of acid sites. Obviously, this desorption peak
arises from the NH3 molecules coordinated to surface Cu species.
According to the electronic acid–base theory proposed by Lewis
[47], a metal ion, which can accept an electron pair from another
molecular entity, is defined as a Lewis acid. In our system, we spec-
ulate that CuII on CuO nanoparticles may function as a Lewis acid.
Such a Lewis acid may form an adduct with PO by sharing the elec-
tron pair furnished by the oxygen in the epoxy cycle, causing the
isomerization of PO by breaking the C–O bond and opening the cy-
cle [46]. After modification by an alkali metal ion, the peak of NH3
desorption shifted to a lower temperature, and the peak tempera-
ture decreased in the order Li+ (ꢀ510 K) > Na+ (ꢀ508 K) > K+
(ꢀ505 K) > Rb+ (ꢀ500 K) > Cs+ (ꢀ497 K). For the Cs+–5 wt% CuOx/
SiO2 series of catalysts with different Cs/Cu ratios, the peak tem-
perature decreased as the Cs/Cu ratio increased from 0.1 to 0.4
(Fig. 10B). These observations suggest that the Lewis acidity as-
cribed to CuO nanoparticles is decreased after modification by an
alkali metal ion, and Cs+ is the most efficient modifier to eliminate
the acidity.
3.4. Characterizations of CuOx/SiO2 and Cs+-modified CuOx/SiO2
catalysts under working conditions
To gain information on the crystalline phase in the working cat-
alyst, we performed in situ XRD measurements for the 5 wt% CuOx/
SiO2 and the Cs+ꢂ5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalysts under
reaction conditions. Fig. 11 shows the evolution of XRD patterns
with reaction time for both catalysts in a gas flow containing
C3H6 and O2 [P(C3H6) = 50.7 kPa, P(O2) = 50.7 kPa] after the pre-
treatment with O2-containing He [P(O2) = 8.4 kPa] at 823 K, which
is similar to that used for catalytic reactions. The temperature was
cooled down to 523 K after the pretreatment, and time zero de-
noted the point just before the introduction of the reactant gas.
Two diffraction peaks at 2h of 35.5° and 38.6° belonging to CuO
were observed for both catalysts. After the introduction of the reac-
tant gas of C3H6 and O2 at 523 K, new diffraction peaks at 2h of
36.4° and 42.3°, assignable to the (111) and (200) reflections of
cubic Cu2O, appeared for the 5 wt% CuOx/SiO2 catalyst after
10 min of reaction (Fig. 11A). These two diffraction peaks could
be observed during the reaction at 523 K for 240 min. For the
Cs+–5 wt% CuOx/SiO2 catalyst, the diffraction peak at 2h of 36.4°
belonging to Cu2O could also be observed clearly after the intro-
duction of the reactant gas, although the weak peak of (200)
reflection of Cu2O at 2h of 42.3° was hard to discern. These obser-
vations demonstrate that part of the CuO has been transformed
into Cu2O in both catalysts during the reaction in the in situ XRD
cell. It should be noted that the intensities of diffraction peaks of
CuO for these two catalysts did not change significantly with the
formation of Cu2O during the reaction. We speculate that this
may be because the sensitivity of CuO is lower than that of Cu2O
in XRD and only a small fraction of CuO has been transformed into
Cu2O under our reaction conditions. Moreover, the in situ reaction
might change the smoothness or roughness of the surface of the
sample pellet pressed in the sample holder, increasing the diffi-
culty in directly comparing the absolute intensities of XRD peaks.
We also investigated the evolution of XRD patterns for the cat-
alysts containing Cu0 and CuI as the starting copper species, which
were prepared by treating the calcined catalyst by H2 reduction or
by H2 reduction followed by N2O oxidation, respectively (see Sec-
tion 2.1). For the Cs+–5 wt% CuOx/SiO2 catalyst treated by H2 reduc-
tion, two diffraction peaks at 2h of 43.0° and 50.2° were observed
(Fig. 12A, 0 min), which could be attributed to the (111) and
(200) reflections of cubic Cu0. After the introduction of C3H6 and
O2 at 523 K, these two peaks disappeared rapidly, and another
Fig. 7. XRD patterns: (A) 5 wt% CuOx/SiO2 modified with different alkali metal ions
(the molar ratio of alkali metal ion to Cu is 0.4); (B) Cs+–5 wt% CuOx/SiO2 with
different Cs/Cu molar ratios.
Table 4
Some physical properties of alkali-metal-ion-modified CuOx/SiO2 samples.
Samplea
Ratio of
alkali metal
ion to Cu
BET surface
area
Crystallite
size of
Mean size
of CuO
(m2 gꢂ1
)
CuO (nm)b
(nm)c
CuOx/SiO2
–
693
323
191
198
230
390
401
329
214
29
15
12
8.9
8.5
7.8
24
20
–
18
8.9
7.6
6.9
6.3
5.8
–
Li+–CuOx/SiO2
Na+–CuOx/SiO2
K+–CuOx/SiO2
Rb+–CuOx/SiO2
Cs+–CuOx/SiO2
Cs+–CuOx/SiO2
Cs+–CuOx/SiO2
Cs+–CuOx/SiO2
0.4
0.4
0.4
0.4
0.4
0.1
0.2
0.6
–
–
a
b
c
CuO loading in each sample is 5.0 wt%.
Evaluated from XRD patterns using the Scherrer equation.
Evaluated from TEM images.
from the XRD measurements, indicating the strongest interactions
between CuO nanoparticles and Cs+.
The H2-TPR profiles for the 5 wt% CuOx/SiO2 catalysts in the ab-
sence and in the presence of alkali-metal-ion modifiers are shown
in Fig. 9. The main reduction peak for the 5 wt% CuOx/SiO2 catalyst
was observed at 573 K, and this peak was attributable to the reduc-
tion of CuO to Cu0. The modification by an alkali metal ion shifted
this reduction peak to a higher temperature (Fig. 9A), and the peak
temperature increased in the order Li+ (595 K) < Na+ (597 K) ꢄ K+
(597 K) ꢄ Rb+ (597 K) < Cs+ (602 K). The H2 TPR profiles for the
Cs+–5 wt% CuOx/SiO2 catalysts with different ratios of Cs/Cu
(Fig. 9B) showed that the content of Cs+ did not significantly affect
the peak temperature for the reduction of CuO. These observations
further confirm the interaction between alkali metal ions and CuO,
which retards the reduction of CuO.
Fig. 10 shows the NH3 TPD profiles for 5 wt% CuOx/SiO2 catalysts
modified by different alkali metal ions. For pure silica, no desorp-
tion of NH3 molecules was observed (Fig. 10A), suggesting that

60
J. He et al. / Journal of Catalysis 299 (2013) 53–66
Fig. 8. TEM micrographs: (A) 5 wt% CuOx/SiO2, (B) Li+–5 wt% CuOx/SiO2 (Li/Cu = 0.4), (C) Na+–5 wt% CuOx/SiO2 (Na/Cu = 0.4), (D) K+–5 wt% CuOx/SiO2 (K/Cu = 0.4), (E) Rb+–
5 wt% CuOx/SiO2 (Rb/Cu = 0.4), (F) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4).
two diffraction peaks at 2h of 36.4° and 42.3° ascribed to Cu2O
were observed. These two peaks became weaker after 60 min of
reaction, and at the same time, a shoulder peak at 35.4°, which
could be ascribed to CuO, appeared. This shoulder peak of CuO be-
came stronger with further prolongation of the reaction time.
These observations clarify that Cu0 is unstable under the present
reaction conditions and can be readily transformed into Cu2O, part
of which is further transformed into CuO after a longer reaction
time.
0 min). This confirms that Cu2O can be formed as the sole copper
phase by this pretreatment procedure. After the introduction of
C3H6 and O2 at 523 K, the two diffraction peaks of Cu2O did not
change significantly in the initial 30 min, but their intensities de-
creased gradually with further prolonging the reaction time.
Simultaneously, the diffraction peak at 2h of 35.4° belonging to
CuO appeared after 60 min and its intensity increased gradually
with reaction time.
The results described demonstrate that the catalyst containing
exclusively CuO, Cu2O, or Cu0 as the starting copper phase will
be transformed into a mixture of CuO and Cu2O under our reaction
conditions. However, the fraction of Cu2O in the mixture depends
For the Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalyst treated by H2
reduction followed by N2O oxidation, only the diffraction peaks at
2h of 36.4° and 42.3° attributed to Cu2O were observed (Fig. 12B,

J. He et al. / Journal of Catalysis 299 (2013) 53–66
61
Fig. 11. In situ XRD patterns for the catalyst, which has been pretreated with O2-
containing He [P(O2) = 8.4 kPa] at 823 K, under reaction conditions for different
times: (A) 5 wt% CuOx/SiO2; (B) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4). Reaction
conditions: T = 523 K, P(C3H6) = P(O2) = 50.7 kPa, F = 60 mL minꢂ1
.
Fig. 9. H2 TPR profiles: (A) 5 wt% CuOx/SiO2 modified with different alkali metal
ions (the molar ratio of alkali metal ion to Cu is 0.4); (B) Cs+–5 wt% CuOx/SiO2 with
different Cs/Cu molar ratios.
Fig. 12. In situ XRD patterns for the Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalyst
under reaction conditions for different times. Reaction conditions: T = 523 K,
P(C3H6) = P(O2) = 50.7 kPa, F = 60 mL minꢂ1. The catalyst has been pretreated with
different procedures: (A) pretreated with H2-containing He [P(H2) = 10.1 kPa] at
773 K; (B) pretreated with H2-containing He [P(H2) = 10.1 kPa] at 773 K, followed by
N2O [P(N2O = 20.3 kPa] oxidation at 573 K.
the catalysts starting from Cu0 and Cu2O, after 300 min of reaction
were quite similar to each other.
We further performed FT-IR studies using CO as a probing mol-
ecule for the 5 wt% CuOx/SiO2 and the Cs+–5 wt% CuOx/SiO2 (Cs/
Cu = 0.4) catalysts after different pretreatments and subsequent
reactions. It is known that CO can be adsorbed strongly onto CuI
sites, whereas the adsorption of CO onto CuII or Cu0 sites is weak
at room temperature [48,49]. Thus, the CO-adsorbed FT-IR studies
can provide insights into the chemical states of copper on catalyst
surfaces during the reaction. For the 5 wt% CuOx/SiO2 catalyst after
the pretreatment with O2 [P(O2) = 8.4 kPa] at 823 K, a weak IR band
at 2129 cmꢂ1 was observed (Fig. 13A, curve a), while almost no IR
band could be observed for the Cs+–5 wt% CuOx/SiO2 catalyst after
the same oxidative pretreatment (Fig. 13B, curve a). These observa-
tions are in agreement with the consensus that the adsorption of
CO on CuII is very weak. CO adsorbed onto the 5 wt% CuOx/SiO2
and the Cs+–5 wt% CuOx/SiO2 catalysts after H2 reduction showed
Fig. 10. NH3 TPD profiles: (A) 5 wt% CuOx/SiO2 modified with different alkali metal
ions (the molar ratio of alkali metal ion to Cu is 0.4); (B) Cs+–5 wt% CuOx/SiO2 with
different Cs/Cu molar ratios.
on the starting copper phase. The comparison of the intensities of
diffraction peaks ascribed to Cu2O (2h = 36.4°) and CuO (2h = 35.4°)
after a longer time (300 min) of reaction suggests that the catalyst
starting from the CuO phase (Fig. 11B) possesses a lower fraction of
Cu2O than the catalyst starting from the Cu0 and Cu2O phases
(Fig. 12). The fractions of Cu2O in the latter two catalysts, that is,
IR bands at 2120 and 2101 cmꢂ1
, respectively (curve b of

62
J. He et al. / Journal of Catalysis 299 (2013) 53–66
Fig. 13A and B), and both peaks could be ascribed to CO adsorbed
onto metallic Cu0 [49]. The pretreatments of the 5 wt% CuOx/SiO2
and the Cs+–5 wt% CuOx/SiO2 catalysts with H2 followed by N2O
oxidation provided strong IR bands at 2129 and 2123 cmꢂ1, respec-
tively (curve c of Fig. 13A and B). These bands were attributable to
CO strongly adsorbed on CuI sites [48–51]. After the reaction in a
C3H6 and O2 mixture [P(C3H6) = 50.7 kPa, P(O2) = 50.7 kPa] at
523 K, intense IR bands of CO adsorption at 2129 cmꢂ1 (for 5 wt%
CuOx/SiO2) and 2123 cmꢂ1 (Cs+–5 wt% CuOx/SiO2) were observed
for the catalysts with all three different pretreatments (curves d,
e and f of Fig. 13A and B). These observations suggest the presence
of CuI sites on catalyst surfaces after the reactions, irrespective of
the pretreatment procedures. However, the intensity of the IR band
of CO adsorption after the reaction over the 5 wt% CuOx/SiO2 and
the Cs+–5 wt% CuOx/SiO2 catalysts was dependent on the pretreat-
ment procedure, and pretreatment with O2 led to a relatively lower
the Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) samples pretreated with
O2-containing He at 823 K, followed by purging with He at the
same temperature and cooling down to 523 K in He. Table 5 shows
that the reaction of the C3H6 pulse with the 5 wt% CuOx/SiO2 at
523 K produces acrolein as the predominant product (selectiv-
ity P 95%), together with a small amount of COx. The amount of
C3H6 converted or the conversion of C3H6 decreased with an
increasing number of successively introduced C3H6 pulses, and al-
most no products could be detected after the reactions with ꢀ62
C3H6 pulses. By assuming that the lattice oxygen species associated
with CuII were consumed by C3H6 to form acrolein, CO and CO2, we
estimated that the reduction degree of CuII to CuI was ꢀ4.0% for the
5 wt% CuOx/SiO2. Similarly, the reaction of C3H6 pulses with the
Cs+–5 wt% CuOx/SiO2 also provided acrolein as the main product
(selectivity P 75%), together with COx. However, the reactivity of
the lattice oxygen in the Cs+-modified sample decreased drasti-
cally, becoming ꢀone order of magnitude lower than that in the
sample without Cs+ modification (Table 5). The estimated reduc-
tion degree decreased to ꢀ0.34% for the sample modified with
Cs+. These results clearly demonstrate that the lattice oxygen spe-
cies is responsible for the allylic oxidation of C3H6 to acrolein over
our catalysts, and the presence of Cs+ modifier remarkably sup-
presses the reactivity of the lattice oxygen. It is worth mentioning
that no PO could be detected in these experiments, suggesting that
the lattice oxygen does not participate in the epoxidation route.
The pulse-reaction technique may also provide information on
the catalytic behavior of the catalyst instantly after pretreatment,
which may contain a peculiar active species unstable under reac-
tion conditions [53]. To gain information on the catalytic behavior
of the Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4) catalyst containing Cu0,
CuI, or CuII as the starting copper species, we have performed reac-
tions of (C3H6 + O2) pulses over this catalyst after different pre-
treatments. The results are displayed in Fig. 14. For the catalyst
pretreated with O2-containing He gas flow, which contains CuII
as the starting copper species, the reaction of the first (C3H6 + O2)
pulse only gave a C3H6 conversion of 0.1% and a PO selectivity of
7%. This suggests that CuII is less active and less selective for the
epoxidation of C3H6. Fig. 14 shows that the increase in pulse num-
bers significantly increases both the conversion of C3H6 and the
selectivity to PO. After ꢀ250 successive (C3H6 + O2) pulses, C3H6
conversion and PO selectivity rose to ꢀ0.40% and ꢀ40%, respec-
tively, and they did not undergo significant changes with further
increases in pulse numbers. This observation implies the genera-
tion of active sites during the successive (C3H6 + O2) pulse reac-
tions. From the in situ characterization results described
previously, it is easy to deduce a conclusion that the CuI sites gen-
erated gradually on the catalyst surface during the successive
(C3H6 + O2) pulse reactions are responsible for the epoxidation of
C3H6 by O2.
intensity of IR band at 2129 or 2123 cmꢂ1
.
The results obtained from in situ XRD and CO-adsorbed FT-IR
studies are in harmony with each other and have provided useful
information on the chemical state of copper species during the
reaction. In short, it becomes clear that CuI is generated during
the reactions over both the 5 wt% CuOx/SiO2 and the Cs+–5 wt%
CuOx/SiO2 catalysts, which have been pretreated with O2-contain-
ing He gas (similarly to the pretreatment used for catalytic reac-
tions) and contain CuII as the starting copper species. We have
further demonstrated that Cu0 species over the 5 wt% CuOx/SiO2
and the Cs+–5 wt% CuOx/SiO2 catalysts pretreated with H2 would
be transformed rapidly into CuI and then into a mixture of CuII
and CuI after exposure to reactants. The catalysts containing CuI
as the sole starting copper species would also undergo changes
during the reaction, and part of CuI would be transformed into CuII.
3.5. Pulse-reaction studies and catalyst stability
The lattice oxygen has been proposed to account for the allylic
oxidation of propylene to acrolein over metal oxide catalysts
including CuOx-based catalysts [42,44,52]. To uncover the role of
the lattice oxygen in the conversion of C3H6 in our case, we have
performed reactions of C3H6 pulses with the 5 wt% CuOx/SiO2 and
On the other hand, when the Cs+ꢂ5 wt% CuOx/SiO2 catalyst had
been pretreated with H2 to give Cu0 or with H2 followed by N2O to
provide CuI as the starting copper species, C3H6 conversion was
higher (1.25% and 0.97%) during the reaction of the first
(C3H6 + O2) pulse (Fig. 14A). This indicates that Cu0 and CuI are
both active for the oxidation of C3H6 by O2. However, the selectiv-
ities to PO were low (7.6% and 12%) in the reactions of the first
(C3H6 + O2) pulse in these cases (Fig. 14B). The selectivity to PO in-
creased with an increase in the successive pulse number. At the
same time, the conversion of C3H6 decreased gradually. The stea-
dy-state PO formation activities achieved after ꢀ600 successive
(C3H6 + O2) pulses were similar for these two pretreatment proce-
dures. In steady states, PO selectivities in these two cases (ꢀ20%)
were lower than that in the case of pretreatment with O2 (ꢀ40%),
while C3H6 conversions were slightly higher. This suggests that a
lower concentration of CuI on catalyst surfaces favors the PO selec-
tivity. This will be discussed later.
Fig. 13. FT-IR spectra of adsorbed CO over (A) 5 wt% CuOx/SiO2 and (B) Cs+–5 wt%
CuOx/SiO2 (Cs/Cu = 0.4) after different pretreatments and subsequent catalytic
reactions. (a) Pretreatment with O2-contaning He [P(O2) = 8.4 kPa] at 823 K; (b)
pretreatment with H2-contaning He [P(H2) = 16.9 kPa] at 773 K; (c) pretreatment
with H2-contaning He [P(H2) = 16.9 kPa] at 773 K, followed by oxidation with N2O
[P(N2O) = 16.9 kPa] at 573 K; (d) reaction for 1 h after (a); (e) reaction for 1 h after
(b); (f) reaction for 1 h after (c). Reaction conditions: T = 523 K, P(C3H6) = -
P(O2) = 50.7 kPa, F = 60 mL minꢂ1
.

J. He et al. / Journal of Catalysis 299 (2013) 53–66
63
Table 5
Reaction of C3H6 pulses with 5 wt% CuOx/SiO2 and Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4).a
Sample
Pulse number
C3H6 converted (lmol)
C3H6 conv. (%)
Selectivity (%)
PO
Acrolein
COx
CuOx/SiO2
1
2
3
4
5
6
0.16
0.62
0.22
0.16
0.13
0.11
0.10
0
0
0
0
0
0
97
96
95
97
97
97
2.9
3.8
4.6
3.2
3.0
3.3
0.056
0.040
0.033
0.028
0.027
Cs+–CuOx/SiO2
1
2
3
4
5
6
0.016
0.011
0.0064
0.0056
0.0038
0.0032
0.061
0.043
0.025
0.021
0.015
0.012
0
0
0
0
0
0
88
86
80
75
79
83
12
14
20
25
21
17
a
Reaction conditions: W = 0.15 g, T = 523 K, amount of C3H6 pulse = 0.63 mL (STP) (ꢀ28
l .
mol), F(He carrier) = 80 mL minꢂ1
4. Discussion
4.1. Reaction pathways over CuOx/SiO2 and Cs+–CuOx/SiO2 catalysts
The analyses of the changes of product selectivities with vary-
ing kinetic parameters, particularly the contact time (expressed
as W/F), may provide useful information on the reaction pathways.
The primary products could be deduced from the product selectiv-
ities at very short contact times, that is, the initial reaction stage.
Over the 5 wt% CuOx/SiO2 catalyst, acrolein was a main product
(selectivity ꢀ50%) at shorter contact times (Figs. 3A and 4A). The
formation of PO could also be observed at shorter contact times,
and the selectivity to PO approached ꢀ20% as the contact time ap-
proached 0 (Figs. 3A and 4A). The increase in the contact time de-
creased the selectivity to PO and increased that to COx. The
selectivity to acrolein decreased slightly with increasing the con-
Fig. 14. Dependence of catalytic performance on the number of successive
(C3H6 + O2) pulses over the Cs+–5 wt% CuOx–SiO2 (Cs/Cu = 0.4) catalyst after
different pretreatments. (A) C3H6 conversion; (B) PO selectivity. (a) Pretreatment
with O2-containing He [P(O2) = 8.4 kPa] at 823 K; (b) pretreatment with H2-
containing He [P(H2) = 16.9 kPa] at 773 K; (c) pretreatment with H2-containing He
[P(H2) = 16.9 kPa] at 773 K, followed by oxidation with N2O [P(N2O) = 16.9 kPa] at
573 K. Reaction conditions: W = 0.10 g, T = 523 K, P(C3H6) = P(O2) = 50.7 kPa, pulse
volume = 0.63 mL (STP), F(He carrier) = 80 mL minꢂ1
.
We have further performed flow-mode reactions for a longer
time for the Cs+ꢂ5 wt% CuOx/SiO2 catalysts starting from both CuII
and CuI. The results are shown in Fig. 15. The reaction conditions
used here are the same as those employed for the pulse reactions
in Fig. 14, except for the reaction mode. Fig. 15A shows that, for
the catalyst after O2 treatment, which contains CuII as the starting
copper species, C3H6 conversion slightly decreases, while PO selec-
tivity slightly increases when the time on stream is prolonged to
4 h. After the treatment of the used catalyst with an O2–He flow,
the catalytic performance could be recovered. From Fig. 15B, we
clarified that, for the catalyst treated by H2 reduction followed by
N2O oxidation, which contained CuI as the starting copper species,
neither C3H6 conversion nor PO selectivity changed significantly
with time on stream.
Fig. 15. Changes of catalytic performance of the Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4)
catalyst after different pretreatments with time on stream: (A) after O2 pretreat-
ment in O2-He [P(O2) = 8.4 kPa] at 823 K; (B) after H2 pretreatment in H2–He
[P(H2) = 16.9 kPa] at 773 K followed by oxidation with N2O in N2O–He [P(N2-
O) = 16.9 kPa] at 573 K. Reaction conditions: W = 0.20 g, T = 523 K, F = 60 mL minꢂ1
P(C3H6) = P(O2) = 50.7 kPa.
,

64
J. He et al. / Journal of Catalysis 299 (2013) 53–66
tact time under O2-rich conditions (Fig. 3A), but it rather increased
slightly with the contact time under the condition P(O2)/
P(C3H6) = 1 (Fig. 4A). Allyl alcohol was also formed in significant
amounts at P(O2)/P(C3H6) = 1, and the selectivity to allyl alcohol in-
creased slightly with the contact time (Fig. 4A). These results allow
us to speculate that the primary products over the 5 wt% CuOx/SiO2
catalyst are acrolein and PO as well as COx (Fig. 16A). Allyl alcohol
must be an isomerization product (a secondary product) from PO,
and the consecutive oxidation of allyl alcohol further provides
acrolein. COx may further be formed via the consecutive oxidation
of PO and acrolein. In other words, there are two parallel routes for
the partial oxidation of C3H6 over the 5 wt% CuOx/SiO2 catalyst.
One is the allylic oxidation route to form acrolein, and the other
is the epoxidation route to form PO. PO could be converted further
to acrolein via allyl alcohol. Actually, the facile transformation of
PO to allyl alcohol and acrolein over the 5 wt% CuOx/SiO2 catalyst
has been confirmed by experimental results using PO as a reactant
(Table 3). Allyl alcohol could be obtained with higher selectivity at
lower temperatures in the conversion of both PO and C3H6 (Table 3
and Fig. 2A) over the 5 wt% CuOx/SiO2 catalyst. At lower tempera-
tures or shorter contact times, the sum of the selectivities to PO
and allyl alcohol can reach 30–40% (Figs. 2A and 4A), indicating
that the epoxidation route over the 5 wt% CuOx/SiO2 catalyst is
not a minor route, as expected. The lower selectivity to PO over this
catalyst should be due to the higher reactivity of PO over this cat-
alyst (Table 3). This is a significant result because supported CuOx
catalysts are well known for the allylic oxidation of C3H6 to acro-
lein [41–44] but not for the epoxidation reaction. Vaughan et al.
once reported that Cu/SiO2 could catalyze the epoxidation of
C3H6 by O2 with a selectivity of ꢀ53% at a low C3H6 conversion
(ꢀ0.25%), but they emphasized that the active phase was Cu0
[30]. Here, we have clearly demonstrated that CuOx/SiO2 catalyzes
the epoxidation of C3H6 to PO as well as the allylic oxidation to
acrolein, and part of the acrolein in the final products stems from
the epoxidation route via PO and allyl alcohol.
The reaction pathways over the Cs+–5 wt% CuOx/SiO2 catalyst
are summarized in Fig. 16B. PO was formed as a main product
(selectivity 50–65%) at short contact times (Figs. 3B and 4B). Acro-
lein was also formed with relatively lower selectivities (20–30%) in
the initial stage over this catalyst. The increase in the contact time
decreased the selectivities to PO and acrolein (particularly the for-
mer) and increased that to COx. No significant formation of allyl
alcohol could be observed over this catalyst even at lower temper-
atures and P(O2)/P(C3H6) = 1 (Fig. 2B). The selectivity to other oxy-
genates such as propanal and acetone also became lower over the
Cs+-modified catalyst (Tables 1 and 2). This suggests that the mod-
ification by Cs+ inhibits the isomerization of PO to allyl alcohol and
other oxygenates. The experiments using PO as a reactant con-
firmed the significantly lower activity of the Cs+–5 wt% CuOx/SiO2
catalyst in the conversion of PO (Table 3). Our results further sug-
gest that, as compared to the catalyst without Cs+ modification, the
Cs+–5 wt% CuOx/SiO2 catalyst possesses a higher initial selectivity
for the epoxidation route (Fig. 16).
4.2. Role of Cs+
Our catalytic studies have demonstrated that the main partial
oxidation product over 5 wt% CuOx/SiO2 is acrolein, and the modi-
fication of 5 wt% CuOx/SiO2 by an alkali metal ion significantly en-
hances the selectivity to PO. Among the alkali metal ions, Cs+ is the
most efficient promoter for enhancing the selectivity to PO (Table
1). Kinetic analyses reveal that both acrolein and PO are formed as
primary products via the allylic oxidation and epoxidation routes
over 5 wt% CuOx/SiO2, but PO would be further isomerized to allyl
alcohol and subsequently oxidized to acrolein over this catalyst.
The presence of Cs+ inhibits the consecutive conversion of PO. This
Fig. 16. Reaction routes for the oxidation of C3H6 by O2: (A) 5 wt% CuOx/SiO2 and (B) Cs+–5 wt% CuOx/SiO2 (Cs/Cu = 0.4). The number in parentheses below a primary product
is the estimated fraction of this product in all the primary products.

J. He et al. / Journal of Catalysis 299 (2013) 53–66
65
should be one of the main reasons for the increase in the selectivity
to PO after modification by Cs+. We have further clarified that the
presence of Cs+ also increases the selectivity to PO and decreases
that to acrolein in the initial reaction stage, although both are pri-
mary products irrespective of the presence of Cs+ (Fig. 16).
Characterizations of catalysts using XRD and TEM techniques
suggest that there exist strong interactions between CuOx species
and Cs+ or other alkali metal ions. The presence of an alkali metal
ion significantly decreased the size of CuOx nanoparticles. Such
interactions between CuOx and Cs+ were stronger than those be-
tween CuOx and other alkali metal ions because the size of CuOx
was the smallest over the Cs+–CuOx/SiO2 catalyst (Table 4). We fur-
ther clarified that the strong interactions between CuOx and alkali
metal ions affected the acidity and the redox properties of the
modified catalysts. NH3 TPD studies suggested that CuOx particles
possessed acidity, and modification by an alkali metal ion de-
creased the acidity of CuOx particles. Cs+ was the most effective
modifier to weaken the acidity of CuOx, since the peak temperature
for NH3 desorption was lowest over the Cs+–CuOx/SiO2 catalyst
(Fig. 10A). H2 TPR results indicated that the reducibility of CuII in
the catalyst was lowered due to the presence of an alkali metal
ion. The peak temperature observed from H2 TPR was the highest
over the Cs+-modified catalyst.
We propose that the decrease in acidity due to the presence of
Cs+ or another alkali metal ion mainly contributes to the inhibition
of the consecutive conversion of PO. It is known that Lewis acids
can catalyze the isomerization of PO to allyl alcohol, which can
be readily oxidized to acrolein in the presence of O2 [46]. On the
other hand, the decrease in the reducibility of CuOx after modifica-
tion by Cs+ or another alkali metal ion indicates a decrease in the
reactivity of the lattice oxygen in the CuOx species. The lower reac-
tivity of the lattice oxygen in the Cs+-modified catalyst has also
been confirmed by the pulse-reaction studies (Table 5). Our
pulse-reaction studies further clarify that the lattice oxygen is
responsible for the allylic oxidation of C3H6 to acrolein. This is in
agreement with the consensus that the lattice oxygen is a nucleo-
philic oxygen species, which preferentially attacks the allylic H
atoms in C3H6 molecule and results in allylic oxidation. Thus, the
decrease in the reactivity of lattice oxygen can contribute to the
decrease in the selectivity of acrolein in the primary products after
the modification by Cs+.
Fig. 17. Proposed reaction mechanism for the formations of primary products over
the CuOx/SiO2 catalysts with and without Cs+ modification. CuI generated over the
CuOx/SiO2 catalyst with and without Cs+ modification during the reaction accounts
for the formation of propylene oxide, while the lattice oxygen is responsible for
acrolein formation.
successive (C3H6 + O2) pulses significantly increases the conversion
and PO selectivity (Fig. 14). This confirms that CuII is not active for
the epoxidation of C3H6 by O2. Combining these results and analy-
ses, we conclude that CuI works as the active site for the epoxida-
tion of C3H6 by O2 over our present CuOx-based catalysts.
The catalysts, starting from both CuI and Cu0, would be trans-
formed into those containing a mixture of CuI and CuII under our
reaction conditions (Figs. 12 and 13). In these cases, the ratios of
CuI to CuII were higher than that over the catalyst starting from CuII
(Section 3.4). Our (C3H6 + O2) pulse-reaction and flow-reaction
studies both revealed that the PO selectivity was lower over the cat-
alyst starting from CuI or Cu0, although the C3H6 conversion was
higher (Figs. 14 and 15). These observations allow us to assume that
a lower concentration of CuI on the catalyst surface favors PO selec-
tivity. Generally, for a selective oxidation reaction, the spatial isola-
tion of active oxygen species is a key, as proposed by Grasselli [54].
Similarly, we speculate that a larger concentration of adjacent oxy-
gen species formed over the catalyst with a higher concentration of
CuI is unfavorable to the selective formation of PO.
Based on the results and discussion described above, we pro-
pose a reaction mechanism in Fig. 17 for the formation of acrolein
and PO, two main primary products, over the CuOx/SiO2 catalysts
with and without Cs+ modification. The lattice oxygen is responsi-
ble for the allylic oxidation of C3H6, producing acrolein, and the re-
duced CuI sites can be reoxidized to CuII by O2. Although several
early studies claimed that the lattice oxygen associated with CuI
in Cu2O was more selective for the formation of acrolein [42,44],
our pulse-reaction results suggested that the lattice oxygen in
CuO nanoparticles exhibited high selectivity for the conversion of
C3H6 to acrolein (Table 5). Besides the allylic oxidation route, we
have clearly demonstrated that there exists an epoxidation route
over the CuOx/SiO2 catalysts with and without Cs+ modification.
The CuI, with a lower concentration generated during the reaction,
functions as the active site for the epoxidation of C3H6 to PO. It is
likely that the CuI site activates molecular oxygen to form an active
oxygen species (CumꢂOꢁ), accounting for the epoxidation of C3H6
to PO by O2. PO would undergo isomerization to allyl alcohol and
further oxidation to acrolein over the catalyst without Cs+ modifi-
cation. The modification by Cs+ cannot only inhibit the consecutive
conversion of PO but also decrease the reactivity of the lattice oxy-
gen, contributing to high PO selectivity. Future studies using isoto-
pic oxygen (18O2) are needed to provide further evidence for the
roles of lattice oxygen and adsorbed oxygen species derived from
O2 in the formation of acrolein and PO.
4.3. Active sites and possible reaction mechanism
As described above, it is clear that the lattice oxygen species in
the present CuOx-based catalysts is responsible for the allylic oxi-
dation of C3H6 to acrolein. It is generally accepted that epoxidation
requires an electrophilic oxygen species, which is typically a kind
of adsorbed oxygen species such as Oꢂ2 ; O22ꢂ; Oꢂ, or ꢁOOH formed
by the activation of O2 or another oxidant on the active site [1,3–
6]. Here, we will first discuss the active site for the activation of
O2 over our catalysts.
For the 5 wt% CuOx/SiO2 and the Cs+–5 wt% CuOx/SiO2 catalysts
pretreated by O2-containing He, which contain CuII as the starting
copper species, our in situ XRD and CO-adsorbed FT-IR studies have
clarified that CuI is generated during the reaction. Moreover, it has
been demonstrated that the Cu0 species over the same catalysts
pretreated by H2 reduction is rapidly transformed into CuI and then
into a mixture of CuI and CuII under reaction conditions. The insta-
bility of Cu0 species under reaction conditions can exclude the pos-
sibility of Cu0 as the active site for the present catalytic system. On
the other hand, it is also difficult to assume that CuII can work for
the activation of O2. Actually, our pulse-reaction studies showed
that the reaction of the first (C3H6 + O2) pulse over the CuII-contain-
ing Cs+–5 wt% CuOx/SiO2 catalyst provided very low C3H6 conver-
sion and low PO selectivity, and the increase in the number of
5. Conclusions
The modification of the 5 wt% CuOx/SiO2 catalyst, which mainly
catalyzed the formation of acrolein, by an alkali metal ion in-

66
J. He et al. / Journal of Catalysis 299 (2013) 53–66
[12] V. Duma, D. Hönicke, J. Catal. 191 (2000) 93–104.
[13] E. Ananieva, A. Reitzmann, Chem. Eng. Sci. 59 (2004) 5509–5517.
[14] X. Wang, Q. Zhang, Q. Guo, Y. Lou, L. Yang, Y. Wang, Chem. Commun. (2004)
1396–1397.
[15] X. Wang, Q. Zhang, S. Yang, Y. Wang, J. Phys. Chem. B 109 (2005) 23500–23508.
[16] Q. Zhang, Q. Guo, X. Wang, T. Shishido, Y. Wang, J. Catal. 239 (2006) 105–113.
[17] B. Moens, H.D. Winne, S. Corthals, H. Poelman, R. De Gryse, V. Meynen, P. Cool,
B.F. Sels, P.A. Jacobs, J. Catal. 247 (2007) 86–100.
[18] S. Yang, W. Zhu, Q. Zhang, Y. Wang, J. Catal. 254 (2008) 251–262.
[19] J. Lu, J.J. Bravo-Suárez, A. Takahashi, M. Haruta, S.T. Oyama, J. Catal. 232 (2005)
85–95.
[20] J. Lu, J.J. Bravo-Suárez, M. Haruta, S.T. Oyama, Appl. Catal. A 302 (2006) 283–
295.
[21] Y. Lei, F. Mehmood, S. Lee, J. Greeley, B. Lee, S. Seifert, R.E. Winsns, J.W. Elam,
R.J. Meyer, P.C. Redfern, D. Teschner, R. Schlögl, M.J. Pellin, L.A. Curtiss, S. Vajda,
Science 328 (2010) 224–228.
[22] M. Ojeda, E. Iglesia, Chem. Commun. (2009) 352–354.
[23] S. Lee, L.M. Molina, M.J. López, J.A. Alonso, B. Hammer, B. Lee, S. Seifert, R.E.
Winans, J.W. Elam, M.J. Pellin, S. Vajda, Angew. Chem. Int. Ed. 48 (2009) 1467–
1471.
[24] J. Huang, T. Akita, J. Fays, T. Fujitani, T. Takei, M. Haruta, Angew. Chem. Int. Ed.
48 (2009) 7862–7866.
[25] R.M. Lambert, F.J. Williams, R.L. Cropley, A. Palermo, J. Mol. Catal. A 228 (2005)
27–33.
[26] R.L. Cropley, F.J. Williams, A.J. Urquhart, O.P.H. Vaughan, M.S. Tikhov, R.M.
Lambert, J. Am. Chem. Soc. 127 (2005) 6069–6076.
[27] D. Torres, N. Lopez, F. Illas, R.M. Lambert, Angew. Chem. Int. Ed. 46 (2007)
2055–2058.
[28] J.R. Monnier, G.W. Hartley, J. Catal. 203 (2001) 253–256.
[29] J. Lu, M. Luo, H. Lei, X. Bao, C. Li, J. Catal. 211 (2002) 552–555.
[30] O.P.H. Vaughan, G. Kyriakou, N. Macleod, M. Tikhov, R.M. Lambert, J. Catal. 236
(2005) 401–404.
[31] X. Zhang, Q. Zhang, Y. Guo, W. Zhan, Y. Guo, Y. Wang, G. Lu, J. Mol. Catal. A 357
(2012) 106–111.
[32] H. Chu, L. Yang, Q. Zhang, Y. Wang, J. Catal. 241 (2006) 225–228.
[33] Y. Wang, H. Chu, W. Zhu, Q. Zhang, Catal. Today 131 (2008) 496–504.
[34] W. Zhu, Q. Zhang, Y. Wang, J. Phys. Chem. C 112 (2008) 7731–7734.
[35] W. Su, S. Wang, P. Ying, Z. Feng, C. Li, J. Catal. 268 (2009) 165–174.
[36] I. Onal, D. Düzenli, A. Seubsai, M. Kahn, E. Seker, S. Senkan, Top. Catal. 53
(2010) 92–99.
[37] L. Yang, J. He, Q. Zhang, Y. Wang, J. Catal. 276 (2010) 76–84.
[38] S. Seubsai, M. Kahn, S. Senkan, ChemCatChem 3 (2011) 174–179.
[39] A. Seubsai, S. Senkan, ChemCatChem 3 (2011) 1751–1754.
[40] W. Long, Q. Zhai, J. He, Q. Zhang, W. Deng, Y. Wang, ChemPlusChem 77 (2012)
27–30.
[41] H.H. Voge, C.R. Adams, Adv. Catal. 17 (1967) 151–221.
[42] B.J. Wood, H. Wise, R.S. Yolles, J. Catal. 15 (1969) 355–362.
[43] T. Inui, T. Ueda, M. Suehiro, J. Catal. 65 (1980) 166–173.
[44] J.B. Reitz, E.D. Solomon, J. Am. Chem. Soc. 120 (1998) 11467–11478.
[45] S. Sato, R. Takahashi, T. Sodesawa, K. Yuma, Y. Obata, J. Catal. 196 (2000) 195–
199.
creased the selectivity and yield to PO in the oxidation of C3H6 by
O2, although the conversion of C3H6 decreased to some extent. The
Cs+–5 wt% CuOx/SiO2 catalyst exhibited the highest PO selectivity
and PO yield among a series of alkali-metal-ion-modified CuOx/
SiO2 catalysts. Our characterizations showed that the modification
by an alkali metal ion decreased the size of CuOx nanoparticles,
suggesting the existence of strong interactions between CuOx and
alkali metal ions. Such interactions were strongest between CuOx
and Cs+. The presence of Cs+ or another alkali metal ion decreased
the acidity arising from CuOx species and suppressed the reducibil-
ity of the catalyst. The kinetic studies clarified that acrolein and PO
were formed in parallel as primary products via allylic oxidation
and epoxidation routes over the catalysts with and without Cs+
modification. PO underwent isomerization into allyl alcohol and
further oxidation into acrolein over the 5 wt% CuOx/SiO2 catalyst
without Cs+ modification. The modification by Cs+ inhibited the
consecutive conversion of PO by decreasing the acidity. We con-
firmed via pulse-reaction studies that the lattice oxygen species
was responsible for the allylic oxidation of C3H6 to acrolein. The
lower reactivity of the lattice oxygen of the Cs+-modified catalyst
further contributed to the increase in PO selectivity. Through
in situ XRD, CO-adsorbed FT-IR, and pulse-reaction studies, we fur-
ther clarified that CuI generated during the reaction worked for the
epoxidation of C3H6 by O2. It was proposed that CuI activated O2,
forming the active oxygen species for the conversion of C3H6 to
PO. However, too high a concentration of CuI sites is unfavorable
to the PO selectivity.
Acknowledgments
This work was supported by the National Basic Program of Chi-
na (No. 2010CB732303), the Natural Science Foundation of China
(Nos. 21173174, 21033006, 21161130522, and 20923004), the
Program for Changjiang Scholars and Innovative Research Team
in University (No. IRT1036), and a Key Scientific Project of Fujian
Province (2009HZ0002-1).
References
[1] J.R. Monnier, Appl. Catal. A 221 (2001) 73–91.
[46] J.M. Coxon, R.G.A.R. Maclagan, A. Rauk, A.J. Thrope, D. Whalen, J. Am. Chem.
Soc. 119 (1997) 4712–4718.
[47] G.N. Lewis, Valency and Structure of Atoms and Molecules, Wiley, New York,
1923.
[2] M. McCoy, Chem. Eng. News 79 (43) (2001) 19–20.
[3] T.A. Nijhuis, M. Makkee, J.A. Moulijn, B.M. Weckhuysen, Ind. Eng. Chem. Res. 45
(2006) 3447–3459.
[4] Y. Wang, W. Zhu, Q. Zhang, Chin. J. Catal. 29 (2008) 857–865.
[5] Q. Zhang, W. Deng, Y. Wang, Chem. Commun. 47 (2011) 9275–9292.
[6] J. Huang, M. Haruta, Res. Chem. Intermed. 38 (2012) 1–24.
[7] M.G. Clerici, G. Bellussi, U. Romano, J. Catal. 129 (1991) 159–167.
[8] A. Tullo, Chem. Eng. News 82 (36) (2004) 15.
[9] T. Hayashi, K. Tanaka, M. Haruta, J. Catal. 178 (1998) 566–575.
[10] A.K. Sinha, S. Seelan, S. Tsubota, M. Haruta, Angew. Chem. Int. Ed. 43 (2004)
1546–1548.
[48] K. Hadjiivanov, H. Knözinger, Phys. Chem. Chem. Phys. 3 (2001) 1132–1137.
[49] K.I. Hadjiivanov, G.N. Vayssilov, Adv. Catal. 47 (2002) 307–511.
[50] D. Scarano, S. Bordiga, C. Lamberti, G. Spoto, G. Ricchiardi, A. Zecchina, A.C.
Otero, Surf. Sci. 411 (1998) 272–285.
[51] A. Frache, M. Cadoni, C. Bisio, L. Marchese, Langmuir 18 (2002) 6875–6880.
[52] J. Haber, W. Turek, J. Catal. 190 (2000) 320–326.
[53] Y. Li, D. An, Q. Zhang, Y. Wang, J. Phys. Chem. 112 (2008) 13700–13708.
[54] R.K. Grasselli, Top. Catal. 21 (2002) 79–88.
[11] B. Chowdhury, J.J. Bravo-Sárez, M. Daté, S. Tsubota, M. Haruta, Angew. Chem.
Int. Ed. 45 (2006) 412–415.