Synergistic catalysis of carbon monoxide oxidation over copper oxide supported on samaria-doped ceria

Article abstract of DOI:10.1006/jcat.2002.3580

CO oxidation over copper oxide catalysts supported on fluorite-type oxygen ion-conducting oxides was studied. The catalytic activities depended on the propensity of copper oxide species toward valence variations and, thus, the facility of withdrawal of their surface lattice oxygen. The reduction of copper oxide occurred more readily when CO concentration was increased in the reducing gas. Copper oxides supported on fluorite-type oxygen-ion-conducting oxides, or specifically samaria-doped ceria (SDC), exhibited significant light-off behavior. These catalysts had a negligible induction period and appreciable activity at low temperatures. These were attributed to the surface oxygen vacancies of SDC facilitating the withdrawal by CO of the surface lattice oxygen of CuO at the interface between the copper oxide species and the oxygen vacancies of SDC, lowering the reduction temperature of copper oxides. Over these CuO/SDC catalysts, the active sites composed of metastable copper species and two types of oxygen vacancies showed the Langmuir-Hinshelwood-type mechanism although the redox-cycle mechanism is still operative.

Full text of DOI:10.1006/jcat.2002.3580

Journal of Catalysis 208, 370–380 (2002)
doi:10.1006/jcat.2002.3580
Synergistic Catalysis of Carbon Monoxide Oxidation over Copper Oxide
Supported on Samaria-Doped Ceria
∗
,1
Jenshi B. Wang, De-Hao Tsai,† and Ta-Jen Huang†
∗
Department of Chemical Engineering, I-Shou University, Kaohsiung, Taiwan 840, Republic of China; and †Department of Chemical Engineering,
National Tsing Hua University, Hsinchu, Taiwan 300, Republic of China
Received October 24, 2001; revised February 6, 2002; accepted February 19, 2002
In addition, since the concepts of strong metal–support
interaction (3, 4) and catalysis at the metal–support inter-
face (5, 6) were introduced, supports have been studied to
explore the advantages of the interaction in order to seek
insight into developing new catalysts. Fluorite-type oxygen-
ion-conducting oxides such as ceria, yttria-stabilized zirco-
nia (YSZ), gadolinia-doped ceria, and samaria-doped ceria
(SDC), when used as supports or catalysts, significantly en-
hance the catalytic activity of Pt (6), Rh (7), and CuO (8, 9)
for CO oxidation and NO reduction. The improvements in
catalytic activity by these oxides have been attributed to
the oxygen storage/transport characteristics of the support
(6, 7) and to the generation of very active centers existing
at the interface between metal and support (8–11). Many
studies have shown that the oxygen vacancies of metal ox-
ides such as V2O5 (12) and CuO (13) play an important role
in enhancing the catalytic activities of CO oxidation. A re-
cent article by Dow and Huang (8) reported the effect of
oxygen vacancies in YSZ on the activity enhancement and
stability properties of copper catalysts for carbon monoxide
oxidation.
Copper oxide supported on samaria-doped ceria (SDC) is cho-
sen as a model catalyst for carbon monoxide oxidation to illustrate
interfacial metal oxide–support interaction and synergism. The ox-
idation catalysis by CuO/SDC was studied to further elucidate the
effect of oxygen vacancies on activity enhancement and light-off
behavior. It was found that an approximate proportionality exists
between the reducibility and the activity of catalysts. It is concluded
that reducibility and susceptibility of change of oxidation states of
copper oxides is key to catalytic activities, and the nonstoichiomet-
ric metastable copper oxide species formed during reduction are
very active because of their superior capability to transport surface
lattice oxygen. A preliminary mechanism involving surface lattice
oxygen of CuO and oxygen vacancy participation is proposed to
fully explicate the synergistic process leading to light-off. The cause
of light-off is attributed to the formation of co-shared oxygen ions
coupled with the creation of high-turnover-frequency active sites
which are composed of metastable copper oxide species and oxygen
vacancies of two types, Vo,CuO and Vo,SDC. Over these CuO/SDC
catalysts, the active sites exhibit the Langmuir–Hinshelwood-type
mechanism characteristic of precious metals although the redox-
cycle mechanism is still operative.
ꢀc 2002 Elsevier Science (USA)
Key Words: copper oxide; samaria-doped ceria; CO oxidation;
Ceria has long been used as an effective component in
automotive exhaust control using three-way catalysts be-
cause of its ability to store oxygen and improve dispersion
of noble metals (14, 15). Doped ceria has been the subject
of great interest because the addition of dopants greatly in-
creases concentrations of oxygen vacancies (16), improves
thermal stability of ceria (17), or enhances catalytic activity
for CO oxidation and NO reduction (7). Rare earth ox-
ides with trivalent cations are the best possible dopants
for modifying structural and chemical properties of ceria
because rare earth elements can easily replace cerium on
its regular site (18). Copper oxide supported on doped ce-
ria appears to provide an enhanced and unique catalytic
property, that is, metal oxide–support synergism, to CO
oxidation.
synergism; surface lattice oxygen.
1. INTRODUCTION
In exhaust emission control, to meet increasingly strin-
gent environmental regulations in an economical way the
complete oxidation of carbon monoxide is of prime impor-
tance. Precious metals have long been used as the most
efficient oxidation catalysts with high activity and stability
to control exhaust gas emissions. Due to the cost and lim-
ited availability of precious metals, considerable attention
has been paid to transition metals and their oxides in these
uses. Among base metals, copper has been explored as a
possible substitute for precious metals because of its high
activity toward CO oxidation (1, 2).
Countless works have been devoted to the study of car-
bon monoxide oxidation over copper catalysts because of
its significance in understanding a number of major indus-
trial applications, in addition to exhaust emission control.
It has been generally agreed that CO oxidation over copper
1 To whom correspondence should be addressed. Fax: 886-3-571-5408.
E-mail: tjhuang@che.nthu.edu.tw.
370
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

SYNERGISTIC CATALYSIS OF CO OXIDATION OVER CuO/SDC
371
oxide catalyst involves surface oxygen/oxygen vacancy par- vacuum at 110◦C for 4 h, calcined in air at 300◦C for 2 h and
ticipation (19–21). However, to date, there is a dearth of at 700◦C for 4 h, and then immediately quenched in water.
information on the interaction of fluorite-type oxygen-ion-
The copper catalysts were prepared by impregnating
conducting oxide supports with base metal catalytic agents SDC support with an appropriate amount of aqueous solu-
for CO oxidation. As a result, even though significant en- tion of copper nitrate trihydrate (SHOWA, Japan). After
hancement in CO oxidation activity over copper supported excess water was evaporated at 80◦C, the catalysts were
on doped oxide has been recognized, its mechanism re- dried under vacuum at 80◦C for 12 h, and then calcined
mains poorly understood. Liu and co-workers (22) arrived in air at 260◦C for 1.5 h and at 500◦C for 3.5 h. The cal-
at the conclusion that copper in Cu-promoted ceria cata- cination of the supports and catalysts was conducted by
lysts for CO oxidation existed in the form of clusters which passing air at the rate of 1 L/min and by raising the temper-
were stabilized by “strong interaction” with the cerium ox- ature from room temperature at a rate of 10◦C/min . Pow-
ide matrix. In work by Dow and Huang (8), the catalytic dered samples of Cu (99% purity, Aldrich), Cu2O (99.5%,
activity enhancement of CO oxidation was attributed to SHOWA), and CuO (98.5%, SHOWA) were ground to be
the surface oxygen vacancies of YSZ and to the synergism used in temperature-programmed reduction studies.
brought about by the interfacial metal oxide–support in-
teraction (IMOSI). A very recent study by Wang et al. (9)
ascribed the synergistic effect enhancing CO oxidation ac-
tivity to the formation of interfacial active centers between
copper oxide clusters and SDC support but gave no detail
about the involvement of surface lattice oxygen of copper
oxide and did not elaborate on mechanistic schemes at all.
The purpose of this work is to focus on CO oxidation over
copper oxide catalysts supported on fluorite-type oxygen-
ion-conducting oxides to further elucidate the effect of
oxygen vacancies on activity enhancement and light-off
behavior. A preliminary mechanism involving surface lat-
tice oxygen of CuO and oxygen vacancy participation is
proposedtofullyexplicatethesynergisticprocessleadingto
light-off. CuO/SDC was employed as a model catalyst since
the samaria–ceria system reportedly displayed superior
ionic conductivity (23), and CuO/SDC was demonstrated
to exhibit excellent activity for CO oxidation (9). Activ-
ityevaluationscoupledwithCO-temperature-programmed
reduction were carried out to study the oxidation proper-
ties of the model catalyst. We hope this preliminary study
will aid the development of a better understanding of the
synergism between the copper oxide species and the oxide
support brought about by IMOSI.
2.2. Definition of Catalyst Symbols
1. The number in front of Cu is the copper loading in
weight percent, which is calculated with respect to the
weight of the SDC support.
2. The number before SDC is the samaria content in
mole percent, which is evaluated based on moles (samarium
plus cerium) of SDC.
2.3. Activity Measurement
The time-on-stream activity tests of the catalysts were
conducted for 360 min under atmospheric pressure in a
continuous-flow reactor charged with 0.3 g of catalyst. The
reactor was an 8-mm-ID Pyrex U-tube imbedded in a silica–
oil circulating tank (Mandarin Scientific Co.). Two K-type
thermocouples were employed. One was placed in the
silica–oil bath to monitor and control the bath tempera-
ture, while the other was inserted into the catalyst bed to
measure the reaction temperature in the bed.
Tworeactionatmosphereswereadoptedforthetest, agas
mixture composed of 1.80% CO plus 2.95% O2 in argon for
oxygen rich, and 1.85% CO, 0.56% O2, and the balance ar-
gon for oxygen lean environments, both at a total flow rate
of 300 ml/min. The flowing of the reactants, CO (99.9%
purity, Air Products), O2 (99.995%), and Ar (99.9995%),
was regulated by mass flow controllers (Hastings,
Model HFC-202). The reactor outflow was analyzed on-
line by a CO-NDIR (Beckman 880), a gas chromatograph
(Shimadzu GC-8A) equipped with a thermal conductivity
detector, and an oxygen analyzer (Beckman 755A). Both
the signals of CO-NDIR and the temperatures of the cata-
lyst bed were transmitted to a Y-t recorder (Yokogawa,
LR-4110).
2. EXPERIMENTAL
2.1. Catalyst Preparation and Characterization
SDC was prepared from reagent-grade (99% purity,
Strem Chemical Co.) metal nitrates Sm(NO3)3 · 6H2O,
Ce(NO3)3 · 6H2O by a coprecipitation method. Appropri-
ate amounts of samarium nitrate and cerium nitrate were
dissolved in deionized water to make 0.08 M solutions. Hy-
drolysis of the metal salts to hydroxides was obtained by
adding NH4OH solution and in the meantime stirring to
keep the pH of the solution greater than 9. A distinct deep
purple color of precipitate/gel was formed when NH4OH
2.4. CO-Temperature-Programmed Reduction (CO-TPR)
The CO-TPR was carried out by using CO of different
was added into the nitrate solution. Vacuum filtration was percentages in argon as a reducing gas to obtain reduction
employed to isolate the gel, which was then washed twice by data more related to the catalytic oxidation of CO. The TPR
water and ethanol. After washing, the gel was dried under reactor was made up of an 8-mm-ID quartz U-tube with

372
WANG, TSAI, AND HUANG
catalyst samples of varying weights mounted on loosely
packed quartz wool. The flow rate of the reducing gas
was kept at 44 ml/min by a mass flow controller. The tem-
perature of the reactor was increased from room temper-
ature to 550◦C at a rate of 10◦C/min by a temperature-
programmable controller (Eurotherm, 815P). The rate
of CO consumption was analyzed by a CO-NDIR and
recorded by an online personal computer. The peak areas
of TPR were separated and integrated using software de-
veloped by SISC, Taiwan.
3. RESULTS
3.1. CO-TPR of Copper(I) and Copper(II) Oxides
A γ -reduction peak appeared in the CO-TPR patterns
of CuO as shown in Fig. 1. The appearance of the reduc-
tion peak indicated that metallic copper had been formed
directly from carbon monoxide reduction of copper(II) ox-
ide (24, 25). The peak temperature and area under the peak
increased with the increase in copper(II) oxide loadings,
which seemed to have negligible effect on the temperature
at which reduction began.
Figure 2 depicts the TPR profiles of copper(II) oxide over
three reduction atmospheres. With the increase in CO con-
centration, the reduction peak became sharper and moved
toward lower temperatures, while the low-temperature-
reduction area (the portion to the left of the dashed line
in the figure) increased. This suggests that the reducibility
of copper(II) oxide in the early phase of reduction is ap-
proximately proportional to the concentration of the reduc-
ing gas.
FIG. 2. CO-TPR patterns of CuO. Operating conditions: 10◦C/min;
total flow rate, 44 ml/min; sample weight, 10.1 mg. Reduction atmosphere:
(a) 0.67% CO in Ar, (b) 1.13% CO in Ar, and (c) 1.70% CO in Ar.
In order to have more understanding of the activity of the
oxidation states of copper, the CO-TPR of copper(I) oxide
was carried out as shown in Fig. 3. Comparison of profile
a with profiles b and c, and relating profile b to profile e,
shows that the effects of copper(I) oxide loading and CO
concentration were analogous to those of copper(II) ox-
ide, as shown in Figs. 1 and 2, respectively. However, more
than one reduction peak was observed in the TPR patterns
of Cu2O, which implied that other reduction states existed
FIG. 3. CO-TPR patterns. Operating conditions: 10◦C/min; total flow
rate, 44 ml/min. (a) Cu2O, 4.6 mg, 1.13% CO in Ar; (b) Cu2O, 14 mg, 1.13%
CO in Ar; (c) Cu2O, 27 mg, 1.13% CO in Ar; (d) CuO, 7.5 mg, 1.70% CO
in Ar; (e) Cu2O, 14 mg, 1.70% CO in Ar.
FIG. 1. CO-TPR patterns of CuO. Operating conditions: 10◦C/min,
1.70% CO in Ar with total flow rate of 44 ml/min. Sample weight: (a) 4.6,
(b) 10.1, and (c) 15 mg.

SYNERGISTIC CATALYSIS OF CO OXIDATION OVER CuO/SDC
373
TABLE 1
The Integrated Values of CO-TPR Peaks of CuO/SDC Catalystsa
during the course of reduction (24). Since cuprous oxide
when exposed to air is susceptible to being oxidized, it is
presumed that the second reduction peak, γ2, was due to
the air exposure of the oxide during sample weighing and
handling.
Absolute amount of α
Fraction of
Catalyst
peak (mol CO)
α peakb (%)
A comparison of profiles Fig. 1a and Fig. 3a shows that the
γ1 reduction peak of Cu2O occurred at a much lower tem- 5Cu/10SDC
perature than that of CuO. This indicates that Cu2O is more
2.5Cu/10SDC
5.95E-06
8.66E-06
1.10E-05
15.11
11.00
9.27
7.5Cu/10SDC
easily reduced than CuO. However, CuO displayed higher
peak area, which is natural because, on the basis of the same
weight, CuOpossessesmorelatticeoxygenthanCu2Odoes.
If we consider Figs. 3d and 3e, which have the same number
of moles, or equivalently the same number of lattice oxy-
gens, it appears that these two oxides had no appreciable
difference in total peak areas, whereas the peak tempera-
tures of Cu2O were lower than that of CuO. Accordingly, it
becomes apparent that Cu2O starts to be reduced at much
lower temperatures than CuO, and thus Cu2O is considered
to exhibit a higher reducibility than CuO.
a Operating conditions of CO-TPR: 10◦C/min; 1.13% CO in Ar with
total flow rate of 44 ml/min; sample weight, 0.1 g.
b Fractionofα peak = (α peak area)/(total peak area of CuO) × 100%.
tion of α, β, and γ peaks was obviously due to the reduction
of supported CuO. The area of β + γ peaks increased with
increasing copper loading, which seems to have little effect
on the inception temperature of α-peak reduction. This is
expected since the formation of the α peak at low temper-
ature was attributed to the CO consumption of interfacial
oxygen ions of the copper oxide species interacting with the
surface oxygen vacancies of ceria in SDC, i.e., the effect of
IMOSI (9). The amount of these copper oxide species is
measured by the α-peak area. Thus, for quantitative anal-
ysis, all of the reduction peaks were integrated to evaluate
the absolute and fractional α-peak areas, which are listed
in Table 1. The absolute α-peak area increased with the in-
creaseinCuOloading. Theincreasein α-peakareais clearly
a result of the increasing copper oxide species interacting
with the surface oxygen vacancies of ceria in SDC.
3.2. CO-TPR and Activity Tests of CuO/SDC Catalysts
For the CuO/SDC catalysts studied in this work, es-
pecially the high-copper-content xCu/10SDC samples, the
presence of bulk CuO phase can be verified by the XRD
spectra of 5Cu/10SDC/γ -Al2O3 and 7Cu/10SDC/γ -Al2O3
samples reported earlier (9, 26), in which the diffraction
peaks characteristic of copper oxide could be easily discrim-
inated. As displayed in Fig. 4, three peaks, α, β, and γ , could
be distinguished in the CO-TPR profiles of xCu/10SDC
catalysts. It is clearly observable that in pure SDC support
noreductionpeakappearedbelow500◦C, sothattheforma-
Figure 5 shows the activities of CO oxidation catalyzed
by CuO/SDC with different copper loadings under oxygen-
rich conditions. It is interesting to observe that the CO ox-
idation activity was enhanced with the increase in copper
FIG. 4. CO-TPR patterns of CuO/SDC. Operating conditions:
10◦C/min; 1.13% CO in Ar with total flow rate of 44 ml/min; sample
weight, 0.1 g. Support and catalysts: (a) SDC support, (b) 2.5Cu/10SDC,
(c) 5Cu/10SDC, and (d) 7.5Cu/10SDC.
FIG. 5. Activity behavior of CuO/SDC under oxygen-rich condi-
tions. Sample weight, 0.3 g; reaction temperature, 175◦C. Catalysts:
(a) 2.5Cu/10SDC, ꢀ; (b) 5Cu/10SDC, ꢁ; and (c) 7.5Cu/10SDC, ꢂ.

374
WANG, TSAI, AND HUANG
tance at the inception stage of CO catalytic oxidation, and
on the reduction behavior of CuO/SDC catalysts are also
provided.
4.1. Light-Off Behavior of CuO/SDC Catalysts
As shown in Figs. 5 and 6 regarding the activity of
CuO/SDC catalysts, a correlation exists between the activ-
ity and the reducibility of catalysts, and a light-off behav-
ior occurs in the presence of rich oxygen. Figure 7 reveals
that the CuO/SDC-catalyzed CO oxidation behavior over
time-on-stream is quite different from that catalyzed by
CuO. Over CuO/SDC, in contrast to that catalyzed by CuO,
no induction period existed and the CO concentration
FIG. 6. Activity behavior of CuO/SDC under oxygen-lean condi-
tion. Sample weight, 0.3 g; reaction temperature, 175◦C. Catalysts:
(a) 2.5Cu/10SDC, ꢀ; (b) 5Cu/10SDC, ꢁ; and (c) 7.5Cu/10SDC, ꢂ.
loading, and a sharp rise in activity, right after the reaction
was started, occurred for every catalyst exhibiting the light-
off behavior characteristic of precious metal catalysts. Cata-
lyzed by CuO/SDC in an oxygen-lean environment, the
conversion of CO increased drastically at the outset of the
reaction and then declined to a steady one, as depicted in
Fig. 6. It is also seen that the catalytic activity of CO con-
version increased with an increase in copper loadings.
From Figs. 5 and 6, it is obvious that the order of activ-
ity among the catalysts was independent of the prevailing
oxygen concentration in the reaction. In addition, the or-
ders of activity coincide well with the sequence of α-peak
areas shown in the TPR profiles of Fig. 4. This again asserts
that the IMOSI resulting from CuO/SDC interaction is the
driving force leading to activity enhancement.
4. DISCUSSION
Synergistic effects between metal/metal oxide and sup-
port leading to catalytic activity enhancement of CO oxida-
tion have been studied extensively. The activity enhance-
ment was ascribed to the formation of active sites, at which
the turnover frequency is very high, between metal/metal
oxide particles and the support (6, 8, 27, 28). The interfacial
active sites and CO activity enhancement which result
from synergism and interaction between copper oxide and
SDC support have recently been investigated, but with
no elaboration on catalytic mechanisms (9). In this work,
using CuO/SDC as a model system, we elucidate a catalytic
mechanism involving surface lattice oxygen of CuO and
participationofoxygenvacancyleadingtoactivityenhance-
ment and light-off behavior. Further discussions on the
FIG. 7. Variations of catalytic behavior exhibited by (a) CuO sup-
ported on an oxygen-ion-conducting material, SDC; (b) pure CuO.
(A) Oxygen rich; (B) Oxygen lean. All samples have the same loading
reduction of copper oxide species, which is of prime impor- of copper(II) oxide.

SYNERGISTIC CATALYSIS OF CO OXIDATION OVER CuO/SDC
375
[2]
dropped sharply soon after the reaction was started. These oxide, follows afterward:
striking differences in activity behavior can be attributed to
Cu2+ . . . O2− . . . Cu0 → Cu+ . . . O2− . . . Cu+.
the effect of IMOSI induced by the oxygen-ion-conducting
support, SDC (3). In general, CO oxidation over base–
metal oxide catalysts follows the Mars–van Krevelen re-
dox mechanism (1, 29–32). Jernigan and Somorjai (30)
indicated that the redox mechanism for CO oxidation
over CuO catalyst cycles between CuO and Cu2O where
the rate-limiting step is the reduction of copper(II) oxide
by CO.
As the reduction takes place, the CuO clusters become
metastable because the loss of lattice oxygen results in un-
balancedtotalvalencesintheclusters. Severinoetal. (25)in-
dicated that the dominant species present in CO-pretreated
catalysts was Cu0, which was then transformed to a mixture
of Cu0 and Cu+ species when exposed to CO + O2 under
reaction conditions. Other authors (37–39) have also pro-
posed the coexistence of Cu0 and Cu+ over supported CuO
catalyst. Thus, an extended agreement seems to exist in the
literature that both Cu0 and Cu+ are essential for CO oxi-
dation on copper-based catalysts (40–42). Once the copper
is reduced to Cu0 in the prereduction step, the reduced su-
rface is partially oxidized during the course of reaction, and
formation of Cu+ to some extent will occur. Liu et al. (22)
further identified by XPS the formation of Cu+ species in
their study of the interaction of fluorite oxides with copper
catalysts for CO oxidation. Thus, the proceeding of reaction
[2] can be readily concluded.
Oxygen-ion-conducting material such as ceria was used
successfully to improve the catalytic activity of precious
metal catalysts, inhibited by strong adsorption of CO at
low temperatures, by providing interfacial lattice oxygen.
An interfacial oxygen vacancy would thus be created after
the interfacial lattice oxygen of ceria had been removed by
the interfacial metal-adsorbed CO. Over CuO/YSZ, Dow
and co-workers (8, 33) proposed a mechanism of CO ox-
idation that involves the nested oxygen ion reacting with
CO to produce CO2, thus forming an interfacial active cen-
ter. In a recent investigation the oxygen vacancies of SDC
were reported to play a vital role in CO oxidation activity
enhancement by CuO/SDC due to the synergistic effect of
copper oxide and the oxygen vacancies of SDC by means
of formation of interfacial active centers (9). Details of this
mechanism involving redox cycle and active center partici-
pation leading to activity enhancement and light-off behav-
ior are proposed as follows and are shown in Fig. 8.
Step a: The CuO/SDC catalyst at the outset of reaction.
Steps b–c: The nested oxygen ion of CuO neighboring
with the oxygen vacancy of SDC (Vo,SDC) is attacked by
CO, so that CuO is reduced to a transient Cu0. An oxygen
vacancy of CuO (Vo,CuO) is thus formed at this step:
Since having copper partly in the +1 state is essential for
the catalytic activity, it is conceivable that the activity will be
enhanced appreciably when Cu+ is formed in the CuO clus-
ter. Sadykov and Tikhov (43) pointed out the existence of
an oxygen defect phase which occurs as a metastable state
and influences catalytic activity. Nagase et al. (44) have pro-
posed that the occurrence of a metastable cluster, including
an oxygen vacancy, is the active site in the reaction. Pierron
et al. (36) reported that CO reduction of cupric oxide
resulted in the generation of reduced copper phases and
the formation of lattice heterogeneities. And Dekker
and coworkers (45) have shown that over copper-based
catalysts possible redox couples are Cu+–Cu2+, Cu0–Cu2+
,
Cu2+ . . . O2− . . . Cu2+ . . . O2− . . . Vo,SDC + CO
and Cu+–Cu0. In sum, owing to electron transfer the non-
stoichiometric clusters, composed of ion pairs such as Cu0–
Cu2+, Cu+–Cu2+, Cu+–Cu+, and Cu+–Cu0, are very active
because of their superior redox capability. Therefore, reac-
tion [2] represents the electron transfer in the metastable
cluster.
→ Cu2+ . . . O2− . . . Cu0 + CO2 + Vo,SDC + Vo,CuO
.
[1]
Steps a–c have been proposed and confirmed by Dow
and Huang (8) and Wang et al. (9) for supports of YSZ and
SDC, respectively. Jernigan and Somorjai (30) have shown
similarly that CO2 production must result from the surface
reduction of copper(II) oxide by CO. The CO-TPR result in
Fig. 4 shows that two types of reduction behavior exist and,
according to Dow et al. (33) and Wang et al. (9), the α-type
reductionthateasilyoccursatlowtemperaturesisthemajor
behavior of reaction [1]. This agrees with studies (25, 34–36)
showing that the reduction of CuO may proceed directly to
Cu at temperatures below 200◦C without undergoing two
steps, i.e., Cu2+ → Cu+ → Cu0. As a result, reaction [1]
proposed in the mechanism is reasonably established.
Steps c–d: The transient Cu0 will be affected by the neigh-
boring oxygen ions, and formation of Cu+ . . . O2− . . . Cu+,
due to electron transfer in the metastable cluster of copper
Step d: CO prefers to be adsorbed to the metastable cop-
per cluster, while O2 is drawn to the oxygen vacancies that
would be a good adsorption site for oxygen.
Step e: The dissociative adsorption of O2 occurs on the
oxygen vacancies of SDC and CuO, the latter having just
been formed by CuO reduction at steps b–c. CO adsorbs
on the metastable copper cluster:
2Vo,CuO + O2 → 2Vo,CuO . . . O
2Vo,SDC + O2 → 2Vo,SDC . . . O
[3]
[4]
Cu+ . . . O2− . . . Cu+ + CO → Cu+ . . . O2− . . . Cu+ . . . CO.
[5]

376
WANG, TSAI, AND HUANG
FIG. 8. The mechanism of CO oxidation over copper oxide supported on an oxygen-ion-conducting material, SDC. Steps a–k: described in text.
Zhang et al. (46) pointed out that the oxygen vacancies metal oxide play an important role as dissociation cen-
of oxygen-ion-conducting solids make suitable sites avail- ters for O2. O2 dissociation only occurs when electrons are
able for oxygen adsorption. These authors further con- donated from the surface to the O2 molecule, leading to
cluded that the common features of oxygen chemisorp- a metal cation and oxygen anion pair. Because of these
tion on the oxygen-ion-conducting materials might be (i) studies, coupled with other investigations (6, 35, 40), re-
the high capacity for oxygen adsorption, and (ii) the ten- actions [3] and [4] are readily verified with the existence
dency to form atomic-type oxygen species on the sur- of oxygen vacancies. Considering the chemisorption char-
face. Dow and Huang (8) indicated that oxygen vacancies acter of CO, owing to the lone pair of electrons on car-
acted as good defect sites for the adsorption of atomic- bon, CO prefers to adsorb on the nonstoichiometric cop-
type oxygen species. Moreover, Mergler et al. (47) sug- per cluster (48). Accordingly, the copper cluster, which can
gested that for CO oxidation, oxygen vacancies on the supply electron holes, will give a higher activity for the

SYNERGISTIC CATALYSIS OF CO OXIDATION OVER CuO/SDC
377
donor-type reaction of CO, such as the one shown in
reaction [5].
In the CO-TPR of CuO/SDC, the (β + γ )-type reduction
of the CuO cluster is looked upon as the major character-
Steps e–f: The “co-shared” oxygen ions are formed be- istic of the light-off behavior represented by reaction [10].
tween the oxygen vacancies and the adsorbed CO and then As elaborated in reaction [2] above, reaction [10] also in-
reacted with the adsorbed CO to form CO2 which then des- volves electron transfers in CuO clusters. Note that reac-
orbs:
tion [10] is similar to reactions [1]–[2] without the partici-
pation of the oxygen vacancies of SDC. Thus, it may occur
over unsupported copper catalysts. It is apparent that the
key points for the occurrence of light-off in the CO oxida-
tion over the CuO/SDC catalyst is the formation of oxygen
vacancies of CuO and then fast turnover of the adsorbed
reactants on the active sites associated with these oxygen
vacancies.
Cu+ . . . O2− . . . Cu+ . . . CO + Vo,CuO . . . O
→ Cu+ . . . O2− . . . Cu+ . . . CO . . . O . . . Vo,CuO [6]
Cu+ . . . O2− . . . Cu+ . . . CO + Vo,SDC . . . O
→ Cu+ . . . O2− . . . Cu+ . . . CO . . . O . . . Vo,SDC [7]
Cu+ . . . O2− . . . Cu+ . . . CO . . . O . . . Vo,CuO
Step k: Substantial “co-shared” oxygen ions are being
formed and then converted to CO2, liberating far more ac-
tive sites to sustain the reaction.
→ Cu+ . . . O2− . . . Cu+ + CO2 + Vo,CuO
[8]
Cu+ . . . O2− . . . Cu+ . . . CO . . . O . . . Vo,SDC
The repetition between steps j and k sustains the reaction
at high conversions after light-off takes place. As shown
in Fig. 5, the fact that the activities of CuO/SDC catalysts
jumped drastically soon after the reaction was started and
then was maintained at a roughly steady value provides
clues for the assumption of repetition between j and k. Due
to this repetition, the turnover frequency increases consid-
erably so that the activities and conversions can be main-
tained at elevated levels.
Based on the above arguments, it is suggested here that
for CO oxidation catalyzed by copper oxide supported on
oxygen-ion-conducting materials, once ignited by the sur-
face oxygen vacancies of the latter, the activity enhance-
ment and light-off behavior can be attributed to the oxygen
vacancies of CuO and the (β + γ )-type CuO clusters. The
active sites which are composed of metastable copper clus-
ters and two types of oxygen vacancies (Vo,CuO, Vo,SDC) ex-
hibit the L–H-type mechanism of precious metals, although
the redox-cycle mechanism is still operative in the reaction
over the CuO/SDC catalysts since CuO/SDC still displays
CO oxidation activity in the absence of O2, as evidenced
by the results of Fig. 4. Over CuO/YSZ, Dow and Huang
(8) also observed that CO oxidation by surface lattice oxy-
gen could not be neglected, especially for high copper
loadings.
→ Cu+ . . . O2− . . . Cu+ + CO2 + Vo,SDC
.
[9]
As
a description comparable to the Langmuir–
Hinshelwood (L–H)-type mechanism of precious metal
catalysts, the above behavior is stated as being that CO ad-
sorbs on the active site of the CuO/SDC surface and reacts
with the neighboring O(ad) at the interface between CuO
and SDC. Many studies have verified that these interfacial
reactions have an appreciable contribution for CO oxida-
tion over Pt and Pd catalysts (6, 27, 28) and over CuO/YSZ
catalysts (8), leading to a higher initial activity.
Steps f–g: Additional molecules of CuO neighboring with
oxygen vacancy of CuO (Vo,CuO) are reduced by CO, form-
ing more metastable clusters and oxygen vacancies.
Steps g–h: More reactant molecules participate and react.
Thereactionsequenceof[6]–[9]isconsideredtobedueto
the formation of coshared oxygen ions, igniting neighboring
sites to create additional oxygen vacancies consecutively.
The reaction spreads out from the interface between CuO
and SDC to the surface of the (β + γ )-type CuO cluster,
defined as those copper species with (β + γ )-type reduction
in the foregoing CO-TPR results, bringing about further
formation of active sites.
Steps a–h can be viewed as the induction mechanism
for the CuO/SDC catalyst, for which, after the reduction
is triggered by surface defects, the induction proceeds very
rapidly, thereby drastically reducing its induction period, as
depicted in Figs. 5, 6, and 7. The repetition between steps
g and h persists until the reaction condition reaches that
suitable for the occurrence of light-off. This repetition is
regarded as the incubation of light-off, the mechanism of
which is exemplified by steps i–k.
4.2. Reduction of Copper Oxides
For CO oxidation over copper oxide catalysts, it is gener-
ally agreed that the reaction obeys a redox cycle mechanism
(32),
CO + OSL → CO2 + ꢃs,
O2 + ꢃs → 2OSL
[11]
[12]
Steps i–j: The number of reduced CuO and oxygen va-
cancies of CuO multiplies, creating active sites in large
quantities:
,
where OSL is a surface lattice oxygen and ꢃ is a surface
s
oxygen vacancy on the metal oxide surface. Usually, re-
oxidation of the catalytic surface, reaction [12], is rather
fast and thus oxygen withdrawal from the metal oxide,
xCu2+ . . . O2− . . . Cu2+ . . . O2− + xCO
→ xCu+ . . . O2− . . . Cu+ + xCO2 + xVo,CuO
.
[10]

378
WANG, TSAI, AND HUANG
reaction [11], appears to be the rate-determining step (29–
31). The reaction rate will be enhanced when more surface
lattice oxygen is withdrawn from the metal oxide.
From the viewpoint of catalytic oxidation reaction, if the
reaction is of a redox mechanism, weaker bond strength
of oxygen ions usually leads to more significant activity en-
hancement, which has been verified for many reactions (32,
49, 50). These findings suggest that the weakly bound oxy-
gen ions could be removed more easily. For the same rea-
son, if weakly bound oxygen ions exist on the surface of
the metal oxide, the reduction of the metal oxide will read-
ily take place at this position. From the CO-TPR shown in
Fig. 3, the fact that the reduction of Cu2O occurred in the
low-temperature regime well before the incipience of CuO
reduction implies that Cu2O is more easily reduced and its
surface lattice oxygen is more liable to come loose.
Nagase et al. (44) proposed that variations in copper va-
lence during CO oxidation over CuO and Cu2O cycled be-
tween II and 0 for CuO, while for Cu2O the valence changed
from I to II first and then cycled between II and I. This result
coupled with the discussion above suggests that the rates of
CO oxidation over different copper oxide species are dif-
ferent. On the basis of the same weight, CuO possesses
more lattice oxygen, or equivalently more surface lattice
oxygen, than Cu2O. Under oxygen-rich conditions, there is
no change in the oxidation state of CuO, but Cu2O and Cu
are oxidized partially to increase their surface lattice oxy-
gen. On the other hand, under oxygen-lean conditions, CuO
and Cu2O are liable to be reduced to lower oxidation states,
resulting in the loss of surface lattice oxygen. With this in
mind, one can elucidate the activity scenario in terms of
the role that variations of oxidation states of copper would
play during CO oxidation. The valence variation of copper
is accompanied by a change in the number of surface lat-
tice oxygens. Thus, the oxidation activity of copper oxide
species depends on their ability to transport surface lattice
oxygen.
FIG. 9. Reduction scheme of CuO.
lar arguments have been given in the literature. Boon and
co-workers (13) have confirmed that surface oxygen vacan-
4.3. Reduction Behavior of CuO/SDC Catalysts
The reduction of a metal oxide by CO is, in a general cies of copper oxide are the active sites for the TPR of
sense, a heterogeneous catalytic reaction involving CO and the supported copper oxides. Lo Jacono et al. (35) pointed
oxygen ions of the metal oxide. Based on the concept of out that a good adsorption site for the reduction of the
a heterogeneous catalytic reaction, if a very active site ex- alumina-supported copper oxide involved special config-
ists on the surface of the metal oxide, the reduction will urations, such as coordinatively unsaturated surface Al3+
take place easily and occur at lower temperatures. In other ions, close to edges or other defects. As triggers, i.e., oxy-
words, the reduction of the metal oxide needs some starting gen vacancies, are formed, the valences of copper around
points, and the “trigger” of the reduction reaction may be a the triggers are changing, so that the copper oxide cluster
defect such as a surface oxygen vacancy or a weakly bound becomes unstable. The metastable cluster, made up of ion
surface oxygen ion.
pairs such as Cu0–Cu2+, Cu+–Cu2+, and Cu+–Cu0 (45), is
There is no trigger present in CuO at the outset of the susceptible to being reduced or oxidized to other oxidation
reaction, for which the rate-determining step is to form a states. The metastable cluster of CuO is very active because
trigger such as oxygen vacancies on the CuO surface. It ofitssuperiorabilitytotransportsurfacelatticeoxygen, and
is presumed that the reduction of CuO is initiated at the the activity of catalysts will be significantly enhanced when
surface of the copper cluster, as illustrated in Fig. 9. Simi- nonstoichiometric copper oxides are formed.

SYNERGISTIC CATALYSIS OF CO OXIDATION OVER CuO/SDC
379
For CuO/SDC catalysts, the occurrence of β and γ peaks the course of CO oxidation because of its superior ability
has been ascribed to the reduction of highly dispersed to transport surface lattice oxygen.
copper oxide species and bulklike CuO, respectively, and
2. The reduction of copper oxides has been shown to
is considered an intrinsic behavior of copper oxide (33). be affected to some extent by the prevailing reduction at-
However, as shown in Fig. 4, these reduction-peak tem- mosphere. The reduction of copper oxide takes place more
peratures are much lower than those of unsupported CuO, readily when CO concentration is increased in the reducing
such as those shown in Fig. 2. This behavior is associated gas. However, for CuO/SDC catalysts, there is no significant
with the formation of the α peak. Unlike inert supports, difference in the kickoff temperature of α-peak reduction
SDC provides active sites good for oxygen adsorption, so since the formation of α peak is attributed to the effect
that the surface lattice oxygen of CuO neighboring the sup- of surface oxygen vacancies of the oxygen-ion-conducting
port is liable to be seized and carried away from the bulk support. As long as it attains a certain incipient tempera-
CuO by the reducing gas (8, 51). In this regard, Wang and ture, the oxygen vacancies provided by SDC are able to trig-
co-workers (9) have reported that the downward shift in ger the reduction of neighboring CuO species. An IMOSI
reduction peak temperatures is because of improved dis- mechanism is involved at the inception of the reduction,
persion of copper oxide through interaction with the oxy- followed by the induced successive reduction of the bulk
gen vacancies of SDC. Differently from the cases of β and copper oxides.
γ peaks, the α-peak temperatures are almost not affected
3. The conditions for the occurrence of light-off are pri-
by reduction atmospheres and copper loadings, as revealed marily that (i) the reaction temperature must be elevated
by Fig. 4. Dow and co-workers (8, 33) have indicated that enough and (ii) the reactant molecules participating in the
the formation of the α peak is not due to the isolated cop- reaction must be sufficient in number. Both of these con-
per ions but to the effect of surface oxygen vacancies of an ditions are associated with the turnover frequency (TOF)
oxygen-ion-conducting support. Based on the above discus- of the reaction, which increases considerably during light-
sions, we further elucidate here that the α-peak formation off. The cause of light-off was attributed to the formation
is very likely due to the release and carrying away by CO of of coshared oxygen ions coupled with the creation of high
the surface lattice oxygen of CuO at the interface between TOF active sites which are composed of metastable cop-
copper oxide cluster and oxygen vacancies of SDC, such as per species and two types of oxygen vacancies, i.e., Vo,CuO
that illustrated in steps b–c of the reaction mechanism.
and Vo,SDC. At high TOF, activities and conversions can be
The results in Table 1 reveal that the fraction of the α maintained at elevated levels.
peak decreases with increased copper loading. The fraction
4. Copper oxides supported on fluorite-type oxygen-ion-
of the α peak can be viewed as the number of α-type lattice conducting oxides, or specifically SDC, exhibit significant
oxygens needed to ignite those of other types, i.e., the (β + light-off behavior. A negligible induction period and ap-
γ )-type lattice oxygen. In a low-temperature regime, the preciable activity at low temperatures are characteristic of
rate of reduction is controlled by the number and fraction these catalysts. These are attributed to the surface oxygen
of α-type lattice oxygen. After having been triggered by the vacancies of SDC facilitating the withdrawal by CO of the
α-type reduction reaction, the (β + γ )-type lattice oxygen surface lattice oxygen of CuO at the interface between the
then plays a major role in determining reduction rates, as copper oxide species and the oxygen vacancies of SDC,
illustrated by step f of the mechanism proposed earlier.
lowering the reduction temperature of copper oxides. Once
ignited by the surface oxygen vacancy of SDC, the activity
enhancement and light-off behavior can be attributed to the
oxygen vacancies of CuO and (β + γ )-type CuO clusters.
5. CONCLUSION
CO oxidation catalyzed by copper oxides supported on Over these CuO/SDC catalysts, the active sites composed of
samaria-doped ceria (SDC) has been studied. On the basis metastable copper species and two types of oxygen vacan-
of the results in this work, a preliminary mechanism involv- cies exhibit the Langmuir–Hinshelwood-type mechanism
ing surface lattice oxygen of CuO and oxygen vacancy par- although the redox-cycle mechanism is still operative.
ticipation, which leads to activity enhancement and light-off
behavior, has been proposed and the following conclusions
can be drawn.
REFERENCES
1. Kummer, J. T., Prog. Energy Combust. Sci. 6, 177 (1980).
2. Huang, T. J., and Yu, T.-C., Appl. Catal. 71, 275 (1991).
3. Tauster, S. J., Fung, S. C., and Garten, R. L., J. Am. Chem. Soc. 100,
170 (1978).
4. Stevenson, S. A., Dumesic, J. A., Baker, R. T. K., and Ruckenstein,
E., Eds., “Metal–Support Interaction in Catalysis, Sintering, and Re-
dispersion.” Van Nostrand–Reinhold, New York, 1987.
5. Burch, R., and Flambard, A. R., J. Catal. 78, 389 (1982).
6. Metcalfe, I. S., and Sundaresan, S., AIChE J. 34, 195 (1988).
1. Reducibility and susceptibility of change of oxida-
tion states of copper oxides is key to the catalytic activi-
ties of CO oxidation. The catalytic activities depend on the
propensity of copper oxide species toward valence varia-
tions and, hence, the facility of withdrawal of their surface
lattice oxygen. The nonstoichiometric metastable copper
oxide species formed during reduction are very active in

380
WANG, TSAI, AND HUANG
29. Bielan˜ski, A., and Haber, J., in “Oxygen in Catalysis,” p. 244. Dekker,
7. Cho, B. K., J. Catal. 131, 74 (1991).
8. Dow, W. P., and Huang, T. J., J. Catal. 160, 171 (1996).
9. Wang, J. B., Shih, W.-H., and Huang, T. J., Appl. Catal. A 203, 191
(2000).
10. Silver, R. G., Hou, C. J., and Ekerdt, J. G., J. Catal. 118, 400 (1989).
11. Dilara, P. A., and Vohs, J. M., J. Phys. Chem. 97, 12919 (1993).
12. Roozeboom, F., Jos van Dillen, A., Geus, J. W., and Gellings, P. J., Ind.
Eng. Chem. Prod. Res. Dev. 20, 304 (1981).
New York, 1991.
30. Jernigan, G. G., and Somorjai, G. A., J. Catal. 147, 567 (1994).
31. Golodets, G. I., in “Heterogeneous Catalytic Reactions Involving
Molecular Oxygen,” p. 280. Elsevier, Amsterdam, 1983.
32. Boreskov, G. K., in “Catalysis: Science and Technology” (J. R.
Anderson and M. Boudart, Eds.), Vol. 3, p. 40. Springer-Verlag,
Berlin/Heidelberg, 1982.
13. Boon, A. Q. M., van Looij, F., and Geus, J. W., J. Mol. Catal. 75, 277
(1992).
33. Dow, W. P., Wang, Y. P., and Huang, T. J., J. Catal. 160, 155 (1996).
34. Dow, W. P., and Huang, T. J., J. Catal. 147, 322 (1994).
35. Lo Jacono, M., Cimino, A., and Inversi, M., J. Catal. 76, 320
(1982).
36. Pierron, E. D., Rashkin, J. A., and Roth, J. F., J. Catal. 9, 38 (1967).
37. Choi, K. I., and Vannice, M. A., J. Catal. 131, 22 (1991).
38. Panayotov, D., and Mehandjiev, D., Stud. Surf. Sci. Catal. 75, 1449
(1983).
14. Yao, H. C., and Yu Yao, Y. F., J. Catal. 86, 254 (1984).
15. Summers, J. C., and Ausen, S. A., J. Catal. 58, 131 (1979).
16. Choudhary, C. B., Maiti, H. S., and Subbarao, E. C., in “Solid Elec-
trolytes and Their Applications” (E. C. Subbarao, Ed.), p. 47. Plenum,
New York, 1980.
17. Kubsh, J. E., Rieck, J. S., and Spencer, N. D., in “Catalysis and Au-
tomotive Pollution Control II” (A. Crucq, Ed.), p. 125. Elsevier,
Amsterdam, 1991.
39. Sakurai, K., Okamoto, Y., Imanaka, T., and Teranishi, S., Bull. Chem.
Soc. Jpn. 49, 1732 (1976).
18. Etsell, T. H., and Flengas, S. N., Chem. Rev. 70, 339 (1970).
19. Maitra, A. M., Appl. Catal. A 104, 11 (1993).
20. Spivey, J. J., Ind. Eng. Chem. Res. 26, 2165 (1987).
21. Satterfield, C. N., “Heterogeneous Catalysis in Industrial Practice,”
2nd ed. McGraw–Hill, New York, 1991.
22. Liu, W., Sarofim, A. F., and Flytzani-Stephanopoulos, M., Chem. Eng.
Sci. 49, 4871 (1994).
23. Yahiro, H., Eguchi, Y., Eguchi, K., and Arai, H., J. Appl. Electrochem.
18, 527 (1988).
24. Jones, A., and McNicol, B., “Temperature-Programmed Reduction for
Solid Materials Characterization.” Dekker, New York, 1986.
25. Severino, F., Brito, J. L., Laine, J., Fierro, J. L. G., and Lopez Agudo,
A., J. Catal. 177, 82 (1998).
26. Shih, W.-H., M.S. thesis. Dept. Chem. Eng., National Tsing Hua Univ.,
Taiwan, R.O.C., 1995 (in Chinese).
27. Jin, T., Okuhara, T., Mains, G. J., and White, J. M., J. Phys. Chem. 91,
3310 (1987).
28. Serre, C., Garin, F., Belot, G., and Marie, G., J. Catal. 141, 9 (1993).
40. Liu, W., and Flytzani-Stephanopoulos, M., Chem. Eng. J. 64, 283
(1996).
41. Liu, W., and Flytzani-Stephanopoulos, M., J. Catal. 153, 304 (1995).
42. Ska˚rman, B., Reine Wallenberg, L., Larsson, P.-O., Andersson, A.,
Bovin, J.-O., Jacobsen, S. N., and Helmersson, U., J. Catal. 181, 6 (1999).
43. Sadykov, V. A., and Tikhov, S. F., J. Catal. 165, 279 (1997).
44. Nagase, K., Zheng, Y., Kodama, Y., and Kakuta, J., J. Catal. 187, 123
(1999).
45. Dekker, N. J. J., Hoorn, A. J. J., Stegenga, S., Kapteijn, F., and Moulijn,
J. A., AIChE J. 38, 385 (1992).
46. Zhang, Z., Verykios, X. E., and Baerns, M., Catal. Rev.–Sci. Eng. 36,
507 (1994).
47. Mergler, Y. J., van Aalst, A., van Delft, J., and Nieuwenhuys, B. E.,
Appl. Catal. B 10, 245 (1996).
48. Voge, H. H., and Atkins, L. T., J. Catal. 1, 171 (1962).
49. Klissurski, D., and Rives, V., Appl. Catal. A 109, 1 (1994).
50. Deo, G., and Wachs, I. E., J. Catal. 146, 323 (1994).
51. Liu, W., and Flytzani-Stephanopoulos, M., J. Catal. 153, 317 (1995).