On the catalytic activity of Co3O4 in low-temperature CO oxidation

Article abstract of DOI:10.1016/S0021-9517(02)93738-3

Emissions of CO and VOC in ventilation air from chemical industries may be reduced using oxidation catalysts. CO oxidation over Co₃O₄ was studied with flow reactor experiments, and in-situ spectroscopic and structural methods. The rate of deactivation increased with increasing CO or CO₂ gas-phase concentration but decreased with increased O₂ concentration or increased temperature. Regeneration of the catalyst in 10% O₂/Ar was more efficient than regeneration in Ar alone. The presence of surface carbonates, carbonyl, and oxygen species was observed. The active site in the mechanism, where CO can adsorb and react at low temperature, was an octahedrally coordinated surface Co³⁺ ion. The decrease in activity was due to a surface reconstruction of the cobalt oxide making the cobalt ions inactive for CO adsorption. The deactivated catalyst could be regenerated by oxidizing at 250°C, thus, creating the active Co₃O₄ surface state.

Full text of DOI:10.1016/S0021-9517(02)93738-3

Journal of Catalysis 211, 387–397 (2002)
doi:10.1006/jcat.2002.3738
On the Catalytic Activity of Co3O4 in Low-Temperature CO Oxidation
∗,1
¨
Jonas Jansson, Anders E. C. Palmqvist,† Erik Fridell,† Magnus Skoglundh,† Lars Osterlund,†
Peter Thorma¨hlen,† and Vratislav Langer‡
∗
Department of Chemical Reaction Engineering, Chalmers University of Technology, SE-412 96 Go¨teborg, Sweden; †Competence Centre
for Catalysis, SE-412 96 Go¨teborg, Sweden; and ‡Department of Environmental Inorganic Chemistry, SE-412 96 Go¨teborg, Sweden
Received March 13, 2002; revised June 20, 2002; accepted June 27, 2002
peratures, which in the future will call for a need to use
more efficient low-temperature active catalysts (8). Low-
temperatureactiveoxidationcatalystsalsofindapplications
in recombining dissociated CO2 in CO2 lasers used in or-
biting applications (9, 10), removal of odor and CO from
air at room temperature (11, 12), and in selective CO gas
sensors (13, 14).
Oxidation of CO over Co3O4 at ambient temperature was stud-
ied with flow reactor experiments, and in-situ spectroscopic and
structural methods. The catalyst deactivates during the reaction.
The rate of deactivation increased with increasing CO or CO2 gas-
phase concentration but decreased with increased O2 concentra-
tion or increased temperature. Regeneration of the catalyst in 10%
O2/Ar was more efficient than regeneration in Ar alone. The pres-
ence of carbonates and surface carbon on the deactivated catalyst
was concluded from TPO experiments. None of these species could,
however, be correlated with the deactivation of the catalyst. In-situ
FTIR showed the presence of surface carbonates, carbonyl, and
oxygen species. The change in structure and oxidation state of the
catalyst was studied by in-situ XRD, in-situ XANES, XPS, and flow
reactor experiments. One possible explanation for the deactivation
of the catalyst is a surface reconstruction hindering the redox cycle
c
Cobalt oxide shows very high CO oxidation activity in
CO/O2 mixtures even at ambient temperature (15) and this
reaction has been studied by several groups. Yu Yao (16)
found that both pure Co3O4 and Co3O4 supported on alu-
mina had high activity for CO oxidation and obtained ki-
netic parameters for the reaction. Different CO oxidation
activities for Co3O4 in terms of light-off temperature, T50,
have been reported. Haruta et al. (12) found a T50 of about
120◦C for pure Co3O4. Cunningham et al. (17) found a T50
of −54◦C for pure Co3O4 and Thorma¨hlen et al. (15) re-
ported a T50 of −63◦C for Co3O4 supported on Al2O3. It
was only the preoxidised cobalt oxide catalyst that showed
activity at these low temperatures. The activity over a pre-
reduced catalyst or a cobalt oxide without oxidative pre-
treatment is lower, and for such catalysts T50 has been re-
ported to be about 160◦C (15, 18, 19, 20). Upon adding Pt
or Pd to a Co3O4/Al2O3 catalyst, Meng et al. (20) found
that the temperature required for full conversion in CO
oxidation could be decreased by 60 K. Mergler et al. (21–
23) found that a Pt/CoOx /SiO2 catalyst had an improved
CO oxidation activity compared to CoOx /SiO2 or Pt/SiO2.
However, Thorma¨hlenetal. (15)showedthatatlowtemper-
atures Pt does not promote the CO oxidation over CoOx /
Al2O3.
of the reaction.
ꢀ 2002 Elsevier Science (USA)
Key Words: cobalt oxide; CO oxidation; in-situ FTIR; in-situ
XRD; in-situ XANES; XPS, low-temperature activity; catalyst de-
activation.
INTRODUCTION
Low-temperature active oxidation catalysts are impor-
tant in many applications. During the cold start of a car
CO and hydrocarbons are emitted untreated into the at-
mosphere. About 80–90% of all emissions from a car are
released during the cold start (1–3). Although the length
of the cold-start period may be shortened by an improved
heat-up of the catalyst (4, 5), the use of low-temperature
active oxidation catalysts may further reduce the amount
of these emissions. Emissions of CO and VOC in ventila-
tion air from chemical industries, restaurants, and printeries
may be reduced by the use of oxidation catalysts. However,
the temperature of the exhaust gases are often too low to
start the catalytic reaction, in which case low-temperature
active catalysts would be preferred (6, 7). In diesel engines
the improved fuel economy leads to lower exhaust gas tem-
The low-temperature activity of Co3O4 is inhibited by
the presence of, e.g., water (16, 17), hydrocarbons, and NO
(16). Also in the absence of these inhibitors a slow decrease
in activity is seen during steady-state CO oxidation at low
temperature. Cunningham et al. (17) reported a decrease
in CO oxidation activity at −54◦C after about 100 min of
reaction over Co3O4. The deactivation was suggested to be
due to accumulation of carbonates on the catalyst surface.
Our group has previously seen that a gradual deactivation
1 To whom correspondence should be addressed. Fax: +46 31 772 30 35.
E-mail: Jonas.Jansson@volvo.com.
387
0021-9517/02 $35.00
c
ꢀ 2002 Elsevier Science (USA)
All rights reserved.

388
JANSSON ET AL.
takes place during the reaction (24); however, the deac- used as balance. The recovery, R, during the regeneration
tivation could not be correlated with the accumulation of is defined as
surface carbonates (25). The objective of this study is to fur-
ther investigate the mechanisms behind the deactivation of
Co3O4 during CO oxidation at ambient temperature and
the regeneration of the catalyst.
R = (x1 − xd)/(x0 − xd)
x1 = CO conversion (%) at 22◦C after regeneration
xd = CO conversion (%) at 22◦C after 70 min
of deactivation
x0 = CO conversion (%) at 22◦C over the
EXPERIMENTAL
preoxidised catalyst (not deactivated).
Materials
The influence of prolonged deactivation was tested by de-
activating the Co3O4 sample in 1% CO + 0.6% O2 at 22◦C
during 70 min and 14 h, respectively. After the deactivation
time 10% O2 was flowed through the reactor, the temper-
ature was raised linearly (10 K/min), and the CO2 concen-
tration was recorded with a mass spectrometer. In this ex-
periment 100.8 mg of catalyst was used, the flow rate was
20 ml (NTP)/min, and Ar was used as balance.
The Co3O4 catalyst used in all studies was 99.9985% pure
Co3O4, Puratronic, from Johnson Matthey with a BET area
of 4.53 m2/g.
Flow Reactor Studies
Flow reactor studies were performed in a fixed bed re-
actor connected to a quadrupole mass spectrometer. The
reactor setup has been described in detail elsewhere (24).
The influence of the stoichiometric ratio, S = 2 · [O2]/
[CO], CO concentration, CO2 concentration, and temper-
ature, T, on the CO oxidation over Co3O4 was investigated
in a full 24 factorial experiment. The levels of the factors
are presented in Table 1.
In all reactor experiments the catalyst was first oxidised
in 10% O2 at 550◦C for 20 min and cooled in 10% O2 to
the reaction temperature (30◦C or 80◦C according to the
experimental plan). Using mass flow controllers CO, O2,
and CO2 were then introduced to the reactor during 70 min
in the amounts given in Table 1, while the CO conversion
continuously was recorded. In these experiments 69.9 mg
of catalyst was used, the flow rate was 20 ml (NTP)/min,
and Ar was used as balance.
Regeneration of the deactivated Co3O4 catalyst was per-
formed by first deactivating the preoxidised catalyst in 1%
CO + 0.6% O2 during 70 min at 22◦C and then treating the
catalyst for 5 min in either 100% Ar or 10% O2 at differ-
ent temperatures (150, 200, 250, 300, 350, 400, 450, 500, or
550◦C). After this treatment (regeneration) the activity for
CO oxidation was tested at 22◦C in a flow of 1% CO + 0.6%
O2. The activity was compared with the activity for the pre-
oxidised catalyst. In these experiments 100.8 mg of catalyst
was used, the flow rate was 20 ml (NTP)/min, and Ar was
In-Situ FTIR
The in-situ FTIR spectra were recorded by diffuse re-
flectance using a BioRad FTS 6000. The catalyst pow-
der was placed in a small (6.2-mm diameter × 3.2-mm
height)in-situflowreactor(HarrickPrayingMantisDRIFT
cell). The flows of Ar (99.9997% pure), CO, and O2
(99.998% pure) were controlled by separate mass flow con-
trollers (Bronkhorst High-Tech). The total flow rate was
400 ml (NTP)/min and Ar was used as balance. When the
spectra were acquired, 40 scans were averaged to minimise
the signal noise. Spectra were recorded in the wavenumber
range of 950–8000 cm−1 with a resolution of 1 cm−1
.
Reference spectra of the catalyst were taken in 100%
Ar flow at 20◦C after preoxidation in 10% O2 at 550◦C for
90 min and after prereduction in 5% H2 at 550◦C for 30 min.
The effect of simultaneous CO + O2 exposure was studied
by introducing 0.3% CO + 0.3% O2 to preoxidised Co3O4
at 20◦C. Subsequent FTIR spectra were recorded as a func-
tion of time during 85 min (steady-state CO oxidation). The
areas of the peaks at 988, 1005, 1044, 1223, 1248, 1270, 1290,
1301, 1321, 1422, and 2164 cm−1 were calculated by fitting
a partly linear baseline and integrating the spectrum using
the method of gaussian deconvolution. In the case of the
peak at 2164 cm−1, the contribution from gas-phase CO was
identified using the CO rotational lines and then subtracted
from the spectra before the area was calculated.
TABLE 1
Factorial Design for Flow Reactor Study
of the CO Oxidation over Co3O4
In-Situ XRD
In-situ XRD spectra were obtained with a Siemens
Factor
−
+
D5000 X-ray powder diffractometer using Cu Kα radiation
❛
S
1.2
0.5
0
3
1
1
(λ = 1.54056 A). The Co3O4 powder sample was placed in
a small corundum (α-Al2O3) crucible mounted on a plat-
inum heating plate and positioned inside the in-situ flow
reactor cell (AP Paar HTK-10). Reduction of the Co3O4
CO concentration (%)
CO2 concentration (%)
Temperature (◦C)
30
80

ON THE CATALYTIC ACTIVITY OF Co3O4
389
catalyst was studied by flowing 5% H2 over the fresh sam- XPS Analysis
ple and increasing the temperature stepwise from 25 to 150,
XPS analysis was performed using a Perkin Elmer PHI
5000 C system equipped with a pretreatment cell allow-
ing the samples to be exposed to various gas mixtures at
1 atm and transferred between the pretreatment cell and
the UHV chamber without being exposed to ambient air.
The analysis was performed on a pressed disk of Co3O4
after different pretreatments. The sample was initially ox-
idised in 19% O2 at 550◦C for 20 min, cooled in oxygen,
transferred to the UHV chamber and analysed. After this
the sample was returned to the pretreatment cell and ex-
posed to either a mixture of 1% O2 and 0.6% CO or 4%
CO at 25◦C for 70 min and then analysed again. Finally the
sample was reduced in 8% H2 at 550 ◦C for 20 min and anal-
ysed again. The total flow was 100 ml/min. The XPS spectra
were recorded using nonmonochromatic Mg Kα radiation.
The energy scale was internally calibrated by setting the
C1s peak at 284.3 eV.
250, 350, 450, and 550◦C. XRD patterns were recorded at
each temperature while keeping the temperature constant.
Oxidation of the catalyst was studied by first reducing the
Co3O4 catalyst in 5% H2 at 550◦C, cooling it to 25◦C in H2,
and recording the XRD pattern. The gas was then switched
to 5% O2 and the temperature was increased stepwise from
25 to 150, 250, 350, 450, and 550◦C. XRD patterns were
recorded at each temperature while keeping the temper-
ature constant. CO oxidation was studied by first preox-
idising the Co3O4 catalyst in 5% O2 at 550◦C and then
cooling the catalyst to 25◦C and recording the XRD pat-
tern. The gas flow was then switched to 2% CO + 4.9%
O2. XRD patterns were then recorded after 30 min at each
of the temperatures 25, 100, and 170◦C. In all experiments
the total gas flow rate was 500 ml/min and He was used as
balance.
RESULTS
In-Situ XANES Analysis
Dispersive Co K-edge XANES measurements were per-
formed in transmission mode at beamline ID24 at ESRF in
Grenoble, France, on pressed pellets of powder samples of
Co3O4. The pellets were mounted in an in-situ temperature-
controlled flow-through catalytic reactor cell, which was fed
with a reactant gas mixture obtained via mass flow con-
trollers. The total flow was 100 ml (NTP)/min. The gas mix-
ture at the reactor outlet was continuously analysed with
a Balzers Prisma mass spectrometer. The beamline uses a
Si [111] polychromator crystal, operating in Bragg mode,
for selection of the desired range of X-ray wavelengths,
and a 1152 × 1242 pixel CCD solid-state detector for spec-
tral analysis. Harmonic rejection was achieved through the
use of two additional mirrors in the beamline, and the es-
Influence of O2, CO, CO2, and Temperature
The factor experiments for the Co3O4 showed that the
activity increased and the rate of deactivation was less se-
vere with a higher oxygen concentration. For example, the
conversion after 70 min of reaction at 30◦C and with a CO
concentration of 0.5% increased from 79 to 94% by increas-
ing the oxygen concentration from 0.3% (S = 1.2) to 0.75%
(S = 3). A higher CO concentration (with constant stoichio-
metric ratio) only influenced the initial activity slightly but
the rate of deactivation increased. These results agree with
previous findings (24), although in the previous study the
CO concentration was kept constant at 2% while in the
present work the effect of different CO concentrations was
studied. Carbon dioxide present in the reactant gas mixture
also increased the rate of deactivation. When the tempera-
ture was increased the catalyst showed an increased activity
and a slower rate of deactivation. The deactivation of the
catalyst could be almost completely avoided by increasing
the temperature from 30 to 80◦C. The initial conversion of
CO to CO2 and the conversion after 70 min of reaction for
all the levels of the parameters are given in Table 2.
timated energy resolution was ꢀE/E ≈ 1.2 − 1.5 × 10−4
.
The data analysis was performed using the program Delia
in the XAID toolbox of the XOP 2.0 software package (26).
The energy scale of the CCD detector was calibrated be-
fore and after each measurement series using a Co metal
foil and fitting the measured data points to those of a Co foil
spectrum taken from the database DABAX in the XOP 2.0
package. The average energy and current in the storage ring
during the entire series of measurements was 6 GeV and
180 mA, respectively. XANES spectra were recorded dur-
ing H2-TPR in 10% H2 after preoxidation, O2-temperature
Regeneration
The result of the regeneration experiments is presented
programmed oxidation (TPO) in 10% O2 after prereduc- in Fig. 1. The regeneration does not occur at low tempera-
tion, temperature programmed CO oxidation in 1% CO + ture (during the reaction) but the catalyst must be heated
1% O2 after either preoxidation or prereduction, and dur- to 250◦C or above. Regeneration in 10% O2 was more ef-
ing exposure to 1% O2 + 1% CO + 10% CO2 at 50◦C for ficient than regeneration in 100% Ar. More than 60% re-
25 min. Preoxidation was performed at 460◦C in 10% O2 covery of the initial catalytic activity was obtained after re-
and prereduction at 460◦C in 10% H2. The temperature generating in 10% O2 at 250◦C while regenerating in 100%
ramp speed was 20 K/min in all experiments and He was Ar less than 30% recovery was obtained even at 350◦C
used as balance.
(see Fig. 1).

390
JANSSON ET AL.
TABLE 2
a) O2 turned on
b) O2 turned on
-4
x 10
CO Oxidation over Preoxidised (10% O2 at 550◦C for 20 min)
Cobalt Oxide. Effect of Stoichiometric Ratio (S = 2[O2]/[CO]), CO
Concentration, CO2 Concentration, and Reaction Temperature at
Steady-State Reaction Conditions
7
a) CO2
b) CO2
6
5
4
3
2
1
0
600
500
400
300
200
100
0
Initial
Conversion after
70 min (%)
S
CO (%) CO2 (%) T (◦C) conversion (%)
1.2
3
1.2
3
1.2
3
1.2
3
1.2
3
1.2
3
1.2
3
1.2
3
0.5
0.5
1
0
0
0
0
1
1
1
1
0
0
0
0
1
1
1
1
30
30
30
30
30
30
30
30
80
80
80
80
80
80
80
80
95
98
93
98
92
99
89
98
99
99
99
100
99
100
99
79
94
60
82
59
74
44
72
97
99
96
98
93
97
92
98
b) CO
1
a) CO
0.5
0.5
1
-10
0
10
20
30
40
50
60
70
1
0.5
0.5
1
Time (minutes)
FIG. 2. CO and CO2 evolution during TPO after deactivation (1%
CO + 0.6% O2) at 22◦C during (a) 70 min (solid lines) and (b) 14 h
(dashed lines).
1
0.5
0.5
1
1
100
case (b) the catalyst had completely lost its activity and no
CO2 was produced. Since no CO2 had been let into the MS
chamber for a long time in the latter case the CO2 signal had
diminished to a low value. The time to reach this low value
was longer than 70 min, which was why the CO2 signal did
not reach zero at the end of the TPO experiment in Fig. 2.
The area below the graph corresponds to 5.5 respectively
3.9 µmol CO2 for the sample deactivated during 70 min re-
spectively 14 h. The weight of the sample was 100.8 mg. The
total number of surface cobalt atoms can thus be estimated
from the BET area (4.53 m2/g) to 6.8 µmol surface atoms
by assuming a number of 9 × 1018 cobalt atoms/m2 (27) for
the Co3O4 surface. This means that carbonates and Csurf
probably cover a large part of the catalyst surface but not
the entire surface.
Deactivation during Prolonged Exposure Time
After 14 h of exposure to the reactant gas mixture the
catalyst had completely lost its activity. Figure 2 shows the
TPO diagrams of the Co3O4 after exposure to the 1% CO +
0.6% O2 during 70 min and 14 h, respectively. In both cases
a large CO2 peak is seen at about 120◦C and a smaller
peak at about 350◦C. Previous analysis of TPO profiles
has identified the 120◦C peak as carbonates desorbing from
the catalyst and the high-temperature peak as oxidation of
some sort of surface carbon species (24). The catalyst deac-
tivated for 14 h also showed desorption of CO that started at
room temperature. In case (a) in Fig. 2 the TPO started with
a high CO2 concentration since the catalyst still had a high
activity and hence a large amount of CO2 was produced. In
In-Situ FTIR
The in-situ FTIR spectrum of preoxidised Co3O4 is
shown in Fig. 3, spectrum a, where, in order to present
the cobalt–oxide specific bands, the prereduced catalyst was
used as background when the absorbance was calculated.
The preoxidised Co3O4 showed strong absorption bands at
988, 1005, 1044 cm−1, and a broad band at 1290 cm−1. The
latter consisted of several bands: at 1248, 1270, 1290, and
1
10% O
2
0.8
0.6
0.4
0.2
0
Ar
1301 cm−1. There were also bands at 1146 and 1180 cm−1
.
Davydov assigned bands in this region to metal–oxygen
stretchings of coordinatively unsaturated metal ions and
surface oxygen atoms (28). These bands are not present
in the reduced catalyst but appear upon preoxidation.
The bands at 1248, 1270, 1290, and 1301 cm−1 also ap-
pear upon preoxidation. However, these frequencies are
too high to belong to metal–oxygen stretchings. Carbon-
ates have stretching frequencies in this region, but in this
-0.2
150 200 250 300 350 400 450 500 550
Temperature (°C)
FIG. 1. Effect of regenerating the Co3O4 catalyst in 10% O2 (squares)
or 100% Ar (crosses) at different temperatures after the catalyst had been
deactivated for 70 min in 1% CO + 0.6% O2.

ON THE CATALYTIC ACTIVITY OF Co3O4
391
carbonate species (25, 28). We have previously shown that
these peaks are doublets consisting of two peaks at 1219 and
a
1229 cm−1, respectively, two peaks at 1417 and 1432 cm−1
.
0.3
0.25
0.2
These bands originated from two different surface carbon-
ate species: one that occurred upon CO2 adsorption and
the other after CO exposure. During the reaction between
CO and O2 both of these are present since CO2 is formed
in the reaction. The band at 1591 cm−1 is also characteristic
of surface carbonates.
Preoxidised Co O
in 0.3% CO + 0.3% O
3
4
2
0.15
0.1
The evolution of the bands at 1223, 1248, 1270, 1301,
1422, 1591, and 2164 cm−1 during the reaction is shown in
Fig. 3. All bands except the band at 1248 cm−1 first quite
rapidly reached a maximum and then decreased. The de-
crease in the 1301 and 2164 cm−1 bands was faster than the
decrease in the 1223, 1422, and 1270 cm−1 bands. The band
at 2164 cm−1 is associated with CO adsorbed to a cobalt ion.
Careful inspection of this band (not shown) and the fact that
it is quite broad indicates that it consists of two peaks lo-
cated approximately at 2155 and 2178 cm−1. The frequency
of the CO stretch depends on the valency of the cobalt ion.
TheCOstretchisreportedat2023–2025cm−1 onCo0, 2070–
2110 cm−1 on Co+, 2120–2170 cm−1 on Co2+, and 2178–
2180 cm−1 on Co3+ (30–32). The CO stretching frequency
can also vary with the exposed cobalt oxide facet, probably
because of the different coordination of the cobalt atoms
on the different facets affects the orbitals involved in the
adsorption of CO to cobalt and thus the CO stretching fre-
quency. For Co2+, Scho¨nnenbeck et al. (33) reported a CO
stretching frequency of about 2146 cm−1 when adsorbed
to CoO(100) and 2170 cm−1 when adsorbed to CoO(111).
The band at 2164 cm−1 is thus probably a combination of
CO adsorbed on Co3+ and Co2+ sites or adsorbed on cobalt
ions with different coordination.
0.05
0
Preoxidised Co O in Ar
3
4
2200
2000
1800
1600
1400
1200
1000
Wavenumber (cm-1)
b
-1
1248 cm
1
-1
-1
-1
1270 cm
1223 cm
1422 cm
0.8
-1
1591 cm
0.6
0.4
0.2
0
-1
1301 cm
-1
2164 cm
0
10
20
30
40
50
60
Time (minutes)
FIG. 3. (a) In-situ FTIR spectra of preoxidised Co3O4 in Ar at 20◦C
(bottom spectrum) and in 0.3% CO + 0.3% O2 (top spectrum). (b) The
evolution of the area of the bands at 1223, 1248, 1270, 1301, 1422, 1591,
and 2164 cm−1 during the exposure to 0.3% CO + 0.3% O2 as a function
of time.
The split of the 2164 cm−1 band in the two bands at
2155 and 2178 cm−1 was even more pronounced when CO
was adsorbed at 20◦C without the presence of oxygen (not
experiment neither CO nor CO2 was introduced to the shown in the figure). In this case the creation of oxygen
catalyst. Another candidate is an oxide species O2x−
.
vacancies via the formation of CO2(g) is more pronounced
Charged O2x− species with x between 0 and 1 have vibration than during the reaction between CO and O2. As the
frequencies between 1550 and 1130 cm−1 (28). Al-Mashta creation of oxygen vacancies changes the coordination
et al. (29) studied O2 adsorption on preoxidised α-Fe2O3 number of the surface cobalt ions the two bands at 2155
with IR spectroscopy. They found absorption bands at 1270, and 2178 cm−1 probably arise from CO adsorbed on
1300, 1325, and 1350 cm−1, which were assigned to a per- surface cobalt ions with different coordination. The lower
turbed O−2 species (intermediate between O2 and O−2 ), and frequency species could also arise from CO adsorbed on
bands at 930, 990, 1010, and 1090 cm−1, which were assigned a cobalt ion that has been reduced to Co2+ in the catalytic
to perturbed O22− (intermediate between O2− and O22−) or redox cycle. The decrease in the 2164 cm−1 band means that
==
to Fe O surface bonds.
the number of cobalt ion sites available for CO adsorption
Figure 3, spectrum b, shows the change in the spectrum decreases during the reaction. One possible reason for this
when CO and O2 were introduced to the sample. Here the could be that the surface reconstructs in a way that cobalt
absorbance was calculated using the spectrum of the preox- ions available for CO adsorption become less abundant on
idised catalyst as background. The same peaks as those in the surface. Another possibility is that carbonates or OH
Fig. 3, spectrum a, were present and in addition new bands groups block these sites. Traces of OH stretchings could
at 1223, 1321, 1422, 1591, and 2164 cm−1 appeared. The be detected in the 3300–3700 cm−1 region of the FTIR
bands at 1223 and 1422 cm−1 can be assigned to surface spectra. However, these bands were so weak that it was not

392
JANSSON ET AL.
possible to determine whether they increased or decreased
during the reaction.
Co3O4
CoO
Co
α-Al2O3 (crucible)
Pt (heating plate)
Reference
The 1248 cm−1 peak increased first rapidly then more
slowly during the 85 min of the reaction. It is tempting to
assign this band to a surface carbonate accumulating on
the surface during the reaction explaining the deactivation
of the catalyst. However, this band was also seen on the
preoxidised sample when neither CO nor CO2 were in-
troduced so that no carbonates could have been formed.
The preoxidation temperature (550◦C for 90 min) should
also be sufficient to remove possible contamination of car-
bonates or carbon containing compounds from the surface.
Al-Mashta et al. (29) assigned this band to a perturbed su-
peroxide species. The bands at 1270, 1290, and 1301 cm−1
are in the same region as the 1248 cm−1 band and occurred
all upon oxidation of the catalyst without any introduction
of CO or CO2. The bands at 1270 and 1301 cm−1 first rapidly
increased and then decreased while the band at 1290 cm−1
increased during the first 5 min but then remained constant.
Following Al-Mashta et al. (29) we assign the bands at 1248,
1270, 1290, and 1301 cm−1 as originating from the same O2x−
species adsorbed onsurface sitesofdifferentstructures. The
increase in the 1248 cm−1 band and the decrease in the 1270
and 1301 cm−1 bands could then be explained as a redistri-
bution of the type of surface sites where O2x− is adsorbed,
the redistribution being due to the surface reconstruction
during the reaction.
at 25°C
550°C
450°C
350°C
250°C
150°C
25°C
30
35
40
45
50
55
60
65
70
2θ
FIG. 4. Reduction of fresh Co3O4 in 5% H2 at different temperatures.
seemed to be formed as an intermediate. The diffrac-
tograms were collected by scanning from small θ to large
θ. This explains why only the CoO peak at 36.3◦ is clearly
seen, whereas the peak at 42.6◦ is almost absent, and the
peak at 61.7◦ is not present. Collecting a diffractogram took
about 30 min and by the time the angle 42.6◦ was recorded
the CoO phase had almost disappeared.
The in-situ XRD measurements are in agreement with
previous TPR and TPO data of cobalt oxide catalysts (24,
35–37). In TPR a main reduction peak was seen at 370◦C
(24), which agrees with the transition from Co3O4 to metal-
lic Co seen at 350◦C in the XRD pattern (Fig. 4). XRD also
reveals that the reduction takes place via CoO since this
phase is present as an intermediate. During the oxidation a
transition from Co to Co3O4 was seen at 350◦C (Fig. 5). This
The bands at 988, 1005, 1044, 1290, and 1321 cm−1 all in-
creased during the first 5 min but then remained constant
during the reaction time. The increase of the 988, 1005, and
1044 cm−1 peaks means that the amount of surface oxygen
atoms bonded to coordinatively unsaturated cobalt ions in-
creased during this time. This might be due to the creation
of oxygen vacancies during the reaction. Some of the sur-
face oxygen atoms reacts with CO or CO2 to form CO2 or
surface carbonates thus creating oxygen vacancies and low-
ering the coordination number of the surface cobalt ions.
The frequencies of the CoO vibration of the remaining oxy-
gen atoms then shifted from the fundamental frequencies
(at 571 and 664 cm−1 (34)) to the 1000 cm−1 region leading
to an increase in these peaks.
Co3O4
CoO
Co
α-Al2O3 (crucible)
Pt (heating plate)
Reference
at 25°C
550°C
450°C
350°C
In-Situ XRD
According to the XRD pattern in Fig. 4 the fresh cobalt
oxide sample consisted of pure Co3O4. When the sample
was heated in 5% H2 the Co3O4 phase remained at 150 and
250◦C. At 350◦C, however, the peaks for the Co3O4 phase
decreased and the largest Co peak appeared. Also present
at 350◦C was one of the peaks for the CoO phase (at 2θ =
36.3◦). At 450◦C the Co3O4 and CoO peaks disappeared
completelyandonlytheCophasewasvisible. At550◦Conly
Co was present, and even its weaker peak (at 2θ = 51.1◦)
was more pronounced than at 450◦C.
250°C
150°C
25°C
H2 at 25°C
30
35
40
45
50
55
60
65
70
2θ
FIG. 5. Oxidation of reduced Co3O4 in 5% O2 at different temper-
atures. The waved line indicates that the Pt reflection at 2θ = 46.3◦ has
been truncated in the figure.
Upon treatment in 5% H2, Co3O4 was reduced to Co
at a temperature around 350◦C. At this temperature CoO

ON THE CATALYTIC ACTIVITY OF Co3O4
393
2θ = 42.2◦. But after this the Co peaks at 44.0◦ and 51.2◦
could be seen so the sample was not yet completely oxi-
dised at this stage. At higher angles still, Co3O4 could be
seen at 2θ = 65.1◦ and CoO at 61.7◦. At 450◦C the Co phase
had completely disappeared, and the Co3O4 peaks had in-
creased substantially. Some CoO was still present (small
peaks at 36.3◦, 42.2◦, and 61.7◦). At 550◦C only Co3O4 was
detected XRD.
For all diffractograms collected at elevated temperature
the XRD peaks were shifted to lower 2θ values compared
to the references which refer to 25◦C. This was caused by
the thermal expansion of the crystal lattice.
Co3O4
CoO
Co
α-Al2O3 (crucible)
Pt (heating plate)
Reference
at 25°C
170°C
100°C
25°C
During the CO oxidation reaction only peaks from the
Co3O4 phase could be seen at 25◦C (Fig. 6). There was no
visible difference between the diffractogram of the preoxi-
dised sample in O2 at 25◦C and that of the sample in the re-
actant gas mixture. Even upon increasing the temperature
to 100 and 170◦C no visible change in the diffractograms
occurred, and the only phase present was Co3O4.
From previous isotope labeling experiments (24) the
amount of surface Co atoms could be estimated to 3.8%
of the total amount of cobalt atoms. Of these surface atoms
only a fraction are deactivated (since the catalyst is still
quite active). It is thus not probable that this change could
be observed with the XRD equipment used, since it mea-
sures both surface and bulk Co atoms, and does not detect
crystalline phases present in a concentration below 0.25%
by volume.
O2 at 25°C
30
35
40
45
50
55
60
65
70
2θ
FIG. 6. In-situ XRD during reaction between 2% CO + 4.9% O2 over
preoxidised Co3O4. The waved lines indicate that the α-Al2O3 reflections
at 2θ = 43.4◦ and 68.2◦, and the Pt reflection at 2θ = 46.3◦ have been
truncated in the figure.
agrees fairly well with previous TPO data where a main ox-
idation peak was present at 245◦C with a shoulder at about
350◦C.
Exposing the reduced sample for a flow of 5% O2 at 25◦C
does not affect the bulk structure of the sample according
to the XRD patterns shown in Fig. 5. The sample remained
as Co also at 150 and 250◦C. At 350◦C a decrease in the Co
peaks (2θ of 44.0◦ and 51.2◦) was observed, accompanied
by the appearance of Co3O4 peaks (2θ of 36.7◦ and 65.1◦)
and CoO peaks (2θ of 36.3◦ and 42.2◦). This temperature
In-Situ XANES
The result of in-situ XANES analysis during H2-TPR
region for formation of Co3O4 from the lower oxide agrees of a preoxidised Co3O4 sample is presented in Fig. 7a.
well with the temperature region where the catalyst is re- The sample remained in an oxidised form up to ca. 425◦C,
generated (see Fig. 1). Interestingly Co, CoO, and Co3O4 and the transition to metallic cobalt started at about
seem to coexist at this temperature. Given that lower θ were 460◦C. After the sample was kept at 460◦C for 5 min
recorded first; CoO (2θ = 36.3◦) and Co3O4 (2θ = 36.7◦) the sample was completely reduced to metallic cobalt.
were detected first, followed by the larger CoO peak at The result of in-situ XANES analysis during O2-TPO of
2.50
a)
b)
T=460 C, 5 min
T=460 C
T=396 C
T=234 C
T=112 C
T=41 C
T=166 C
T=303 C
T=423 C
2.00
1.50
1.00
0.50
0.00
T=460 C
T=41 C
T=460 C, 3 min
T=460 C, 5 min
T=460 C, 9 min
T=41 C
T=460 C, 9 min
T=460 C, 5 min
T=460 C, 3 min
T=112 C
T=234 C
T=396 C
T=460 C
T=423 C
T=303 C
T=166 C
T=41 C
T=460 C
T=460 C, 5 min
7680
7700
7720
7740
7760
7780 7680
7700
7720
7740
7760
7780
Photon Energy (eV)
Photon Energy (eV)
FIG. 7. Co K-edge XANES analysis during (a) H2-TPR in 10% H2 of preoxidised Co3O4 and (b) O2-TPO in 10% O2 of prereduced Co3O4.

394
JANSSON ET AL.
2.00
1.50
1.00
0.50
0.00
T=460 C, 6 min
a)
b) T=460 C
T=40 C
T=213 C
T=373 C
T=460 C, 5 min
T=375 C
T=134 C
T=50 C
T=42 C, in CO + O2
T=42 C, in 10% H2
T=42 C, in 10% H2
T=42 C, in CO + O2
T=50 C
T=134 C
T=375 C
T=460 C
T=460 C, 6 min
7680
7700
7720
7740
7760
7780 7680
7700
7720
7740
7760
7780
Photon Energy (eV)
Photon Energy (eV)
FIG. 8. Co K-edge XANES analysis during CO oxidation in 1% CO + 1% O2 over (a) preoxidised Co3O4 and (b) prereduced Co3O4.
a prereduced Co3O4 sample is shown in Fig. 7b. After oxidation state of Co3O4 during CO oxidation according to
prereduction the sample was in the form of metallic cobalt. XANES, and the catalyst stayed as Co3O4.
A slight oxidation of the sample was seen throughout the
XPS Analysis
temperature range 41–396◦C, but the sample was only
completely oxidised at 460◦C. The reoxidation seemed to
The resulting Co2p spectra from the XPS analysis are
go faster than the reduction.
given in Fig. 9. The prereduced sample (a) shows a spectrum
that closely resembles what one would expect for pure CoO
(38). The spectra for the preoxidised sample (b) and for the
sample deactivated in 1% O2 and 0.6% CO (c), and 4%
CO (not shown) at room temperature for 70 min, are very
similar. They resemble reference spectra for pure Co3O4
(39). Also the O1s spectra are very similar for the three
latter pretreatments. No significant accumulation of carbon
could be observed after the deactivation pretreatment. Oku
and Sato (40) studied cobalt oxide with in-situ XPS and
could detect a reversible transition from CoO to Co3O4 at
347◦C during oxidising conditions. This temperature agrees
with the temperature where the oxidation of CoO to Co3O4
can be seen with XRD in Fig. 5. Bridge and Lambert (41)
observed with LEED that Co3O4 grown on Co(0001) was
reduced to CoO when exposed to CO and heated to 200◦C,
and that the CoO could then be oxidised back to Co3O4
by exposure to O2 and annealing at 327◦C. However, in
our experiments performed at 25◦C, no change in oxidation
state was observed during the CO oxidation. Not even when
the catalyst was severely deactivated by exposing it to 4%
CO for 70 min at 25◦C could a change in oxidation state of
cobalt be detected.
Figure 8a shows the result of CO oxidation over the pre-
oxidised Co3O4. During the reaction the catalyst stayed as
Co3O4 at all temperatures. The preoxidised sample did not
show any sign of reduction, which was expected since at
460◦C the rate of reoxidation to Co3O4 is high. The result
of CO oxidation over the prereduced Co3O4 is shown in
Fig. 8b. The catalyst was initially in the form of metallic
cobalt but was gradually oxidised during the reaction. The
oxidation started already below 50◦C and increased as the
temperature was raised. The oxidation was not completed
at 460◦C but continued as the sample was held there for
a period of 6 min. Introduction of CO2 did not affect the
450x103
Co2p
Adjusted at C1s (284.3 eV)
400
350
300
a
prereduced
DISCUSSION
b preoxidised
250
deactivated
in CO + O
770
2
c
The oxidation of CO over Co3O4 has been proposed
to follow a redox cycle where gas-phase CO adsorbs on a
cobalt site, and the adsorbed CO reacts with a lattice oxygen
atom forming CO2(g) and an oxygen vacancy, thus reducing
the oxidation state of the cobalt site (24, 25). Reoxidation of
the cobalt site then occurs with gas-phase oxygen (24, 25).
810
800
790
780
Binding energy (eV)
FIG. 9. Co2p XPS spectra (a) after prereduction in H2 at 550◦C, (b)
of the preoxidised Co3O4, and (c) after 70 min reaction in 1% CO + 0.6%
O2 at 25◦C.

ON THE CATALYTIC ACTIVITY OF Co3O4
395
In Table 2 it can be seen that the CO conversion over Co3O4 carbon could be observed with XPS after the deactivation
was initially high (89–100%) but the catalyst became deac- pretreatment.
tivated with time. The rate of deactivation increased when
Traces of surface OH groups were detected with in-situ
increasing the gas-phase CO or CO2 concentration, while FTIR. This is because traces of water are always present in
increasing the stoichiometric ratio (oxygen concentration) the gas bottles even when using gases of high purity. It is
or increasing the temperature decreased the rate of deacti- known that small amounts of water deactivate the cobalt
vation(Table2). Severalpossiblemechanismscouldexplain oxide catalyst (17) and this deactivation probably occurs
the observed deactivation of the Co3O4 catalyst during low- by H2Oads or OH groups blocking the active cobalt ions.
temperature CO oxidation. Previously, we have proposed However, this hypothesis cannot explain why the catalyst
the deactivation to occur by irreversibly reducing the oxida- was regenerated at a lower temperature when regenerating
tion state of the surface cobalt ions so that they could not be in 10% O2 compared to Ar, nor why the deactivation is
reoxidised at the low reaction temperature (25). Another faster in the presence of a higher CO concentration and
possibility is that the active sites are blocked by carbonyl, lower at higher stoichiometric ratios.
carbonate, coke, or H2O/OH species. A third deactivation
The second mechanism for the deactivation of Co3O4
mechanism could be that the creation of oxygen vacancies during CO oxidation could be a change in oxidation state
when reacting with adsorbed CO leads to a reconstruction of the cobalt in the oxide surface. The observed increased
of the cobalt oxide surface making the cobalt sites unavail- deactivation rate of Co3O4 with increasing CO concentra-
able for CO adsorption.
tion and exposure time agrees with the deactivation model
Several hypotheses can thus be put forward to explain proposed in (25). Increasing the CO concentration would
the deactivation of the cobalt oxide catalyst during CO ox- increase the rate of the irreversible reduction of the ac-
idation at ambient temperature. It is often believed that tive cobalt sites leading to a higher rate of deactivation. In
deactivation occurs by blocking the active sites with sur- this oxygen-vacancy-deactivation mechanism (or slow irre-
face carbonates. In this work, however, it was seen that versible reduction mechanism) two types of oxygen vacan-
a temperature of about 400◦C was needed to obtain 80% cies or reduced cobalt sites exist. The first site is an oxygen
recovery of the activity when regenerating in Ar. This is vacancy created when CO reacts with a lattice oxygen atom
much higher than what could be expected for desorbing and desorbs as CO2. This oxygen vacancy is readily reoxi-
surface carbonates, which is about 100◦C (25). With FTIR, dised and the reoxidation proceeds at 20◦C. The second site
the bands originating from surface carbonates at 1223 and is a more reduced site (with lower or partially lower oxida-
1422 cm−1 were seen to decrease during the reaction, op- tion state), which is created when the first type of oxygen
posite to what would be the case if surface carbonates built vacancy is further reduced by CO. This second type of site
up on the catalyst and thus deactivated it. A third piece of is not as readily reoxidised as the first type and a higher
evidence opposing the carbonate explanation is the TPO temperature (about 250◦C) is needed in order to reoxidise
after 14 h of deactivation, where the carbonate peak was it. This site is thus referred to as an “irreversibly reduced
not larger (but smaller) than the carbonate peak from the site.” At 80◦C the rate of reoxidation of the first type of site
catalyst deactivated during only 70 min.
is higher leading to a lower surface concentration of these
The blocking of active sites by adsorbed CO is not a prob- sites. This means that the rate of formation of the second
able candidate for the deactivation since the bands for CO typeofsitefromthefirsttypeofsitewillbeloweratthistem-
adsorbed on cobalt sites were seen to decrease with FTIR. perature explaining the lower rate of deactivation at 80◦C.
If there were a blocking of the sites by CO one would have However, if sites of the second type are formed during the
expected these bands to increase or at least remain con- reaction at 80◦C they will not be reoxidised but the tem-
stant during the reaction. Adsorbed CO would also prob- perature has to be increased to 250◦C in order to reoxidise
ably have desorbed in the regeneration experiment when these sites. Cobalt Co3+ ions are active for CO oxidation but
regenerating in 100% Ar already at about 100◦C (15).
during the reaction these are reduced to a lower oxidation
It is also possible that carbonaceous species, Csurf, might state resulting in a slow decrease in the number of these
be formed during the reaction, deactivating the catalyst. ions. The reduced cobalt ions cannot be reoxidised to the
However, regeneration in oxygen at 250◦C was sufficient active state at ambient temperature and the activity is lost.
to regain more than 60% of the activity while the oxida- A similar type of deactivation mechanism has been pro-
tion temperature for Csurf is about 450◦C (24). If Csurf was posedforthereactionbetweenCOandNOoverCuO/ZrO2
the major deactivating species a larger high-temperature at 250◦C (42). In that catalytic cycle Cu2+ was reduced to
TPO peak would be expected in the TPO of the catalyst Cu+ and then reoxidised to Cu2+. Deactivation occurred
deactivated during 14 h compared to the sample deacti- when Cu+ was reduced by CO to Cu0. Deactivation by re-
vated during 70 min. However this was not seen, and the duction of Cu2+ has also been observed for CuO/CeO2,
high-temperature peak is of the same size independent of which is active for CO oxidation at 70◦C (43). However,
deactivation time. Further, no significant accumulation of as far as we know, experimental observations supporting

396
JANSSON ET AL.
this deactivation mechanism have not been reported for reaction a surface reconstruction occurs where active octa-
room temperature CO oxidation over Co3O4. The deacti- hedrally coordinated Co3+ ions are transformed to tetra-
vation mechanism explains why regeneration in O2 is more hedrally coordinated cobalt ions, the catalyst would lose
efficient than regeneration in Ar alone. During regenera- activity, such as has been observed experimentally. Since
tion at 250–350◦C in O2 the deactivated (reduced) cobalt no change in oxidation state of the cobalt ions occurs, this
ions are reoxidised to the active state, which could also be process would not be possible to detect with XPS.
observed with XRD and XANES. The regeneration tem-
perature agrees with the temperature region for oxidation
of cobalt oxide previously seen with TPO (24). The positive
CONCLUSIONS
effect of regenerating in 100% Ar seen at a temperature of
Co3O4 deactivated slowly during steady-state CO ox-
approximately 350◦C is probably explained by bulk oxy-
idation at room temperature. The rate of deactiva-
gen in Co3O4 becoming mobile at this temperature. Bulk
tion increased with increasing gas-phase CO or CO2
oxygen can thus diffuse to the surface and fill the oxygen
concentration but the deactivation could be made less se-
vacancies creating the active Co3O4 phase. This mechanism
vere by increasing the gas-phase O2 concentration or in-
is also consistent with previous studies which have shown
creasing the reaction temperature. Neither the formation
that a prereduced catalyst is less active than a preoxidised
of surface carbonates, surface carbonyls, nor Csurf could ex-
catalyst (15): prereduction of the catalyst creates more in-
plain the slow deactivation of the cobalt oxide. Further,
active reduced cobalt ions, i.e., having the same effect as
an irreversible reduction mechanism is not supported by
deactivating the catalyst. However, this mechanism is not
XPS data. Instead we suggest that the decrease in activity is
consistent with the XPS measurements where no reduction
due to a surface reconstruction of the cobalt oxide making
could be seen during the reaction, not even after exposure
the cobalt ions inactive for CO adsorption. The deactivated
catalyst could be regenerated by oxidising it at 250◦C thus
re-creating the active Co3O4 surface state.
to 4% CO. One explanation could be that the number of
active sites is small and thus the change is too small to be
observed with XPS.
A third explanation for the deactivation of Co3O4 during
CO oxidation could be that the coordination of the surface
cobalt ions is changed but there is no change in oxidation
state. The valency and coordination of the surface cobalt
ions have been subject to investigation by several authors.
Busca et al. (32) found that Co3O4 after calcination exposed
only Co3+ sites. These sites were readily reduced by CO at
room temperature, producing Co2+. Shirai et al. (44) found
that Co3O4 grown on α-Al2O3 (0001) after oxidation at
600◦C exposed coordinatively unsaturated oxygen atoms
bridging Co3+ ions. Omata et al. (45) studied the CO and
H2 oxidation over Co3O4. By successively substituting the
Co2+ and Co3+ ions in Co3O4 with less active Zn2+ and
Al3+ ions respectively they concluded that octahedrally co-
ordinated Co3+ ions in the surface are the active site and
that tetrahedrally coordinated Co2+ is less active. Using a
similar ion substitution technique Jacobs et al. (46) drew
the conclusion that octahedrally coordinated Co3+ ions in
the surface are active whereas no Co2+ ions existed at the
surface. We thus suggest that the active site in our mecha-
nism, where CO can adsorb and react at low temperature,
is an octahedrally coordinated surface Co3+ ion. Yao and
Shelef (47) reported that CO chemisorbed on octahedrally
coordinated cobalt sites independent of the oxidation state
ACKNOWLEDGMENTS
The Competence Centre for Catalysis is financially supported by the
Swedish National Energy Administration and member companies: AB
Volvo, Johnson Matthey-CSD, Saab Automobile AB, Perstorp AB, MTC
AB, Akzo Nobel, and the Swedish Space Corporation. The European
Synchrotron Radiation Facility is thanked for granting the XANES beam
time, and Dr. Sakura Pascarelli at beamline ID24 is gratefully thanked for
her assistance.
REFERENCES
1. Shelef, M., and McCabe, R. W., Catal. Today 62, 35 (2000).
2. Lenaers, G., Sci. Total Environ. 189/190, 139 (1996).
3. Rijkeboer, R. C., Catal. Today 11, 141 (1991).
4. Lafyatis, D. S., Ansell, G. P., Bennett, S. C., Frost, J. C., Millington,
P. J., Rajaram, R. R., Walker, A. P., and Ballinger, T. H., Appl. Catal.
B 18, 123 (1998).
5. Umehara, K., Tateishi, T., Nishimura, H., and Misumi, M., JSAE Rev.
18, 67 (1997).
6. Chi-Sheng Wu, J., Lin, Z.-A., Tsai, F.-M., and Pan, J.-W., Catal. Today
63, 419 (2000).
7. Drewsen, A., Licentiate thesis, Chalmers University of Technology,
Gothenburg, Sweden, 1999.
8. Erkfeldt, S., Jobson, E., and Larsson, M., Top. Catal. 16/17, 127 (2001).
9. Gardner, S. D., Hoflund, G. B., Upchurch, B. T., Schryer, D. S.,
Kielin, E. J., and Schryer, J., J. Catal. 129, 114 (1991).
(Co2+ and Co3+), while tetrahedrally coordinated cobalt 10. Sato, H., and Tsuchida, E., Trans. IEICE E73(9), 1525 (1990).
sites were inactive for CO adsorption. In Co3O4, the Co3+
11. Watanabe, N., Yamashita, H., Miyadera, H., and Tominaga, S., Appl.
Catal. B 8, 405 (1996).
12. Haruta, M., Tsubota, S., Kobayashi, T., Kageyama, H., Genet, M. J.,
and Delmon, B., J. Catal. 144, 175 (1993).
13. Yamaura, H., Moriya, K., Miura, N., and Yamazoe, N., Sens. and
ions are originally in the octahedral positions and Co2+ in
tetrahedral positions. As discussed above, it has been shown
that octahedrally coordinated Co3+ is the ion, which is ex-
posed at the surface of Co3O4. If during the CO oxidation
Actuators, B 65, 39 (2000).

ON THE CATALYTIC ACTIVITY OF Co3O4
397
14. Funazaki, N., Hemmi, A., Ito, S., Asano, Y., Yamashita, S.,
Kobayashi, T., and Haruta, M., Sens. Actuators B 13–14, 536 (1993).
15. Thorma¨hlen, P., Skoglundh, M., Fridell, E., andAndersson, B., J. Catal.
188, 300 (1999).
31. Mergler, Y. J., van Aalst, A., van Delft, J., and Nieuwenhuys, B. E.,
J. Catal. 161, 310 (1996).
32. Busca, G., Guidetti, R., and Lorenzelli, V., J. Chem. Soc., Faraday
Trans. 86(6), 989 (1990).
16. Yu Yao, Y.-F., J. Catal. 33, 108 (1974).
17. Cunningham, D. A. H., Kobayashi, T., Kamijo, N., and Haruta, M.,
Catal. Lett. 25, 257 (1994).
33. Scho¨nnenbeck, M., Cappus, D., Klinkmann, J., Freund, H.-J.,
Petterson, L. G. M., and Bagus, P. S., Surf. Sci. 347, 337
(1996).
18. Simonot, L., Garin, F., and Maire, G., Appl. Catal. B 11, 167 (1997).
19. Simonot, L., Garin, F., and Maire, G., Stud. Surf. Sci. Catal. 96, 203
(1995).
20. Meng, M., Lin, P., and Fu, Y., Catal. Lett. 48, 213 (1997).
21. Mergler, Y. J., Hoebink, J., and Nieuwenhuys, B. E., J. Catal. 167, 305
(1997).
34. Christoskova, St. G., Stoyanova, M., Georgieva, M., and
Mehandjiev, D., Mater. Chem. Phys. 60(1), 39 (1999).
35. Arnoldy, P., and Moulijn, J. A., J. Catal. 93, 38 (1985).
36. van’t Blik, H. F. J., and Prins, R., J. Catal. 97, 188 (1986).
37. Wang, W.-J., and Chen, Y.-W., Appl. Catal. 77, 223 (1991).
38. Shen, Z. X., Allen, J. W., Lindberg, P. A. P., Dessau, D. S., Wells, B. O.,
Borg, A., Ellis, W., Kang, J. S., Oh, S. J., Lindau, I., and Spicer, W. E.,
Phys. Rev. B 42, 1817 (1990).
22. Mergler, Y. J., van Aalst, A., van Delft, J., and Nieuwenhuys, B. E.,
Stud. Surf. Sci. Catal. 96, 163 (1995).
23. Mergler, Y. J., van Aalst, A., van Delft, J., and Nieuwenhuys, B. E.,
Appl. Catal. B 10, 245 (1996).
24. Jansson, J., J. Catal. 194, 55 (2000).
39. Moulder, J. F., Stickle, W. F., Sobol, P. E., and Bomben, K. D.,
“Handbook of X-Ray Photoelectron Spectroscopy,” Perkin Elmer
Corporation Wellesley, MA, 1992.
25. Jansson, J., Skoglundh, M., Fridell, E., and Thorma¨hlen, P., Top. Catal.
16/17, 385 (2001).
26. Sanchez del Rio, M., and Dejus, R. J., in “XOP: Recent Develop-
ments,” Vol. 2448, p. 340. SPIE—Int. Soc. Opt. Eng., Bellingham, WA,
1998.
27. Takita, Y., Tashiro, T., Saito, Y., and Hori, F., J. Catal. 97, 25 (1986).
28. Davydov, A. A., in “Infrared Spectroscopy of Adsorbed Species on
the Surface of Transition Metal Oxides” (C. H. Rochester, Ed.),
p. 6–62. Wiley, New York, 1990.
29. Al-Mashta, F., Sheppard, N., Lorenzelli, V., and Busca, G., J. Chem.
Soc., Faraday Trans. 1 78(3), 979 (1982).
40. Oku, M., and Sato, Y., Appl. Surf. Sci. 55, 37 (1992).
41. Bridge, M. E., and Lambert, R. M., Surf. Sci. 82, 413 (1979).
42. Okamoto, Y., Kubota, T., Gotoh, H., Ohto, Y., Aritani, H., Tanaka, T.,
and Yoshida, S., Faraday Trans. 94, 3743 (1998).
43. Harrison, P. G., Ball, I. K., Azelee, W., Daniell, W., and Goldfarb, D.,
Chem. Mater. 12, 3715 (2000).
44. Shirai, M., Inoue, T., Onishi, H., Asakura, K., and Iwasawa, Y., J. Catal.
145, 159 (1994).
45. Omata, K., Takada, T., Kasahara, S., and Yamada, M., Appl. Catal. A
146, 255 (1996).
46. Jacobs, J.-P., Maltha, A., Reintjes, J. G. H., Drimal, J., Ponec, V., and
Brongersma, H. H., J. Catal. 147, 294 (1994).
47. Yao, H. C., and Shelef, M., J. Phys. Chem. 78(24), 2490 (1974).
30. Todorova, S., Zhelyazkov, V., and Kadinov, G., React. Kinet. Catal.
Lett. 57(1), 105 (1996).