Supported CuO/Ce1-x Zr x O2 catalysts for the preferential oxidation of CO in H2-rich gases

Article abstract of DOI:10.1134/S0023158406050156

The catalytic properties of CuO supported on ceria or ceria-zirconia mixed oxides have been investigated in the preferential oxidation of CO in H ₂-rich gases. CuO/CeO₂ shows very high activity towards the oxidation of CO with a light-off temperature of about 70°C. This catalyst is very selective for the oxidation of CO rather than of H₂ in the low temperature region (70-120°C), while at higher temperatures, the oxidation of hydrogen begins, causing of a maximum of CO conversion to arise with increasing temperature.

Full text of DOI:10.1134/S0023158406050156

ISSN 0023-1584, Kinetics and Catalysis, 2006, Vol. 47, No. 5, pp. 756–764. © MAIK “Nauka /Interperiodica” (Russia), 2006.
Published in Russian in Kinetika i Kataliz, 2006, Vol. 47, No. 5, pp. 779–787.
Supported CuO/Ce1 – xZrxO2 Catalysts for the Preferential
Oxidation of CO in H2-Rich Gases1
T. Caputoa, R. Pironeb, and G. Russoa
a Dipartimento di ingegneria chimica, Universitá degli studi di Napoli “Federico II”; P.le Tecchio, 80–80125 Napoli, Italy
b Istituto di Ricerche sulla Combustione, CNR; P.le Tecchio, 80–80125 Napoli, Italy
Received June 6, 2005
Abstract—The catalytic properties of CuO supported on ceria or ceria-zirconia mixed oxides have been inves-
tigated in the preferential oxidation of CO in H2-rich gases. CuO/CeO2 shows very high activity towards the
oxidation of CO with a light-off temperature of about 70°C. This catalyst is very selective for the oxidation of
CO rather than of H2 in the low temperature region (70–120°C), while at higher temperatures, the oxidation of
hydrogen begins, causing of a maximum of CO conversion to arise with increasing temperature.
DOI: 10.1134/S0023158406050156
1
Increasing the contact time in the catalytic tests does technologies appear quite far from large-scale applica-
not have any effect on the selectivity of the process, yet tion for both technical and economic reasons. The only
it increases the conversion of CO. This result, together way to sustain the great demand for energy needed by
with catalytic tests of CO and ç2 oxidation fed one our industrialized society remains the conversion of
reactant at a time, allows us to assume that the very high fossil fuels into H2 and the direct conversion of it into
selectivity of these catalysts is a unique function of electricity via processes that would be globally more
temperature and probably stems from the different acti-
vation energy of the two competitive reactions. The
presence of ZrO2 in the support has been revealed as not
beneficial for catalyst performances, while an optimal
content of CuO supported on pure ceria at about 5 wt %
has been found. Finally, good resistance to the presence
of reformate species (carbon dioxide and water) in con-
centrations close to those of practical interest has been
verified.
exothermic and efficient than traditional ones.
Processes of hydrogen production have been devel-
oped over a long time and have been consolidated in the
chemical industry to produce valuable chemicals in sta-
tionary applications (primarily ammonia and methanol
production) and we should turn to these for automotive
applications. The autothermal reforming of light hydro-
carbons, followed by processes of high-temperature and
low-temperature water gas shift can be easily recon-
verted in car engines, but it produces a gaseous stream
typically containing 40–70 vol % H2, 15–20 vol % CO2,
5–10 vol % H2O, and 0.5–1 vol % CO (nitrogen as bal-
ance). Even this relatively low CO concentration repre-
sents a poison for polymer electrolyte membrane fuel
cells (PEMFCs), which are the most promising fuel
cells for vehicle applications due to their low operation
temperature (80°C) and high resistance towards refor-
mate species. The standard CO tolerance limit for
PEMFCs is about 10 ppm, although new Pt–Ru anode
catalysts have increased this limit up to 100 ppm [1].
For nonstationary applications, the simplest method to
decrease CO concentration appears the preferential oxi-
dation of carbon monoxide (CO-PROX). The process
would be effective provided a proper catalyst is
designed. Ideal features of the catalytic system needed
for the CO-PROX process can be summarized as fol-
lows: high activity towards CO oxidation in the range
fixed by the fuel processor (80–200°C), high selectivity
towards the oxidation of CO rather than H2 to limit fuel
consumption, high resistance towards CO2 and H2O in
the gas mixture, and inactivity towards the reverse
INTRODUCTION
The great energy demand required by industrial
development brings about serious economical and envi-
ronmental problems. On the one hand, these are strong
driving forces to find novel processes of energy produc-
tion able to exploit alternative raw materials. On the
other hand, use of traditionally employed fossil fuels in
energy-production processes should be rationalized in
order to achieve a higher efficiency for such processes.
For this purpose, hydrogen-fuelled fuel cells are
becoming a reliable option to replace internal combus-
tion engines in vehicle applications because of their
higher potential efficiency in energy production and
lack of environmental impact. Unfortunately, H2 is not
quite a raw material and must be produced from other
H-sources. One attractive route could be the production
of H2 from water by means of renewable energy forms
(such as solar or wind sources), but up to now these
1
This article was submitted by the authors in English.
756

SUPPORTED CuO/Ce1 – xZrxO2 CATALYSTS
757
water gas-shift reaction that could limit the maximum activity of this catalyst certainly correlates to the high
CO conversion.
oxygen storage capacity of cerium oxide, that is, pro-
portional to the concentration of defects in the oxide
structure. Defect concentrations can be increased by
doping ceria with other metals or compounds. In partic-
ular using zirconium oxide as promoter, it is possible to
increase not only the thermal stability, but also redox
and oxygen supplier properties of ceria [15]. Moreover,
the state of copper when supported on ceria has been
quite well investigated [16–17], but the kind of species
active and selective for CO PROX remains doubtful; as
well, the reasons for its high selectivity are still not
completely understood.
Ratnasamy et al. [18] studied the catalytic activity of
CuO samples supported over high-surface area CeO2,
CeO2–ZrO2, and ZrO2 samples synthesized by copre-
cipitation and evidenced a significant effect of the sup-
port. However, the CO oxidation activity/selectivity
increases in the order CuO–ZrO2 < CuO–CeO2-ZrO2 <
CuO–CeO2; i.e. doping the support with zirconia has no
beneficial effects on PROX performance. Moreover,
they exhibit an optimal CuO content existing at around
5 wt %.
In this work we propose screening different catalytic
systems constituted by copper oxide supported on com-
mercial mixed Ce–Zr oxides with different contents of
copper and zirconium. One aim of the study is to estab-
lish the optimal catalyst composition with a compari-
son of the activity in the CO PROX reaction under
experimental conditions as closest as possible to those
of the practical application and an attempt to correlate
such experimental results to the effect of the different
active species that are generated with different catalyst
compositions. Another important goal of this study is to
elucidate the basic reasons for the high selectivity of
such catalysts by means of catalytic and TPR tests. The
effect of the presence of water vapor and CO2 under
large amounts of hydrogen has been characterized,
while comparison of CO PROX tests with conventional
CO and H2 oxidation tests carried out separately, along
with a study of the reducibility of active sites with
either CO or H2, has allowed to explain the different
reaction kinetics between the desired CO oxidation and
the undesired oxidation of H2.
The most investigated catalysts are those of the plat-
inum group, with particular attention paid to Pt- and
Au-based catalysts [2–7]. Platinum-based catalysts
work in a high temperature range (150–200°C) and
exhibit an optimal resistance to the presence of refor-
mate species in the gas mixture and good selectivity
towards CO oxidation (~40%) even at a low O2/CO
ratio [3], due to the properties of strongly adsorbing
carbon monoxide at low temperatures. Gold-based cat-
alysts applied in CO-PROX process are much more
active and selective than the Pt ones showing very low
operation temperatures (80°C) and an intrinsic selectiv-
ity towards CO oxidation rather than H2 oxidation [8].
However, the activity of gold-based catalyst is signifi-
cantly depressed by the presence of high concentration
of CO2 and H2O in the gas mixture.
More recently, Avgouropoulos [9] tested the
CuO/CeO2 catalyst for the CO-PROX reaction, obtain-
ing values of activity and selectivity higher than on Pt-
based catalysts and comparable to gold catalysts. The
catalytic activity of such materials appears potentially
less sensitive to the presence of CO2 and H2O than a
gold-based catalyst but more than a Pt-based catalyst
[10]. However, the significantly lower cost of copper
makes this system very interesting and promising for
possible application.
Unfortunately, CuO/CeO2, analogously to noble
metal-based catalysts, is very selective towards the oxi-
dation of CO only up to a certain temperature, above
which H2 oxidation become more relevant and the con-
version of CO paradoxically decreases. A maximum in
CO conversion with temperature hence arises, whose
value can be only increased by a more active catalyst if
the contact time is taken around values significant for
practical application. The reasons for the presence of
such a maximum is still debated [9].
To improve the activity of this promising catalyst in
the presence of reformate species, an attempt has been
made to change the preparation method using coprecip-
itation [9, 11], urea-nitrate combustion [12], and sol-gel
[10]. Even if the literature results often refer to different
operating conditions, it seem possible to conclude that
the best results are obtained with those preparation
methods that allow increasing dispersion of copper
oxide on the support [11].
EXPERIMENTAL
Supported Cu-based catalysts were prepared by wet
impregnation using copper acetate as a precursor. Com-
mercial CeO2 and CeO2/ZrO2 (with two different
ceria/zirconia ratios (CZ): 85/15 and 60/40 w/w) pow-
ders from GRACE were used as supports. The support
was suspended in an aqueous solution of copper salt
and mixed in vacuum in a rotating evaporator at
100 rpm, 80 mbar, and 50°C. Subsequently, the sam-
ples were dried overnight at 110°C and calcined for 2 h
at 450°C. The amount of CuO loaded on CeO2 was var-
The increased activity of copper-based catalysts
when deposited on CeO2 certainly correlates to the
interaction between the active phase and support. Such
a catalytic system has been quite largely studied, prima-
rily due to the interest in its potential use in the three-
way catalyst [13]. However, there is a quite evident lack
of knowledge about the properties that make it seem
like a very promising catalyst for the preferential oxida-
tion of CO.
For the CO-PROX reaction, a Mars–Van Krevelen ied (2 to 8 wt %), while the effect of the support was
redox mechanism [14] has been proposed. Thus, the studied on samples at identical CuO contents (5 wt %).
KINETICS AND CATALYSIS Vol. 47 No. 5 2006

758
CAPUTO et al.
Table 1. Code, content of CuO and ZrO2, crystalline mean
size, and BET surface area for all catalysts and supports in-
vestigated
Table 1 lists the results of the analyses of copper
content, XRD, and surface area on all the samples pre-
pared (labeled with specific codes that will be used in
what follows). The copper content was evaluated by
spectrophotometric analysis.
wt %
Mean
Catalyst code
CeO2
BET, m2/g
size, nm
CuO ZrO2
In agreement with the results of Fernandez-Garzia
et al. [16] and Trovarelli et al. [15], the average crystal-
line size estimated by XRD spectra tends to diminish
when ceria is mixed with zirconia, since they form a
solid solution with higher defect concentrations. The
CeO2 and 5 CuO/CeO2 samples exhibit a mean size of
crystallines of about 11 nm; CZ (60/40) and 5 CuO/CZ
(60/40), of about 5 nm. The distortion of the fluorite-
type structure of ceria induced by the introduction of
Zr4+ results in a cell contraction since the ionic radius
of Zr4+ (84 pm) is lower than that of Ce4+ (97 pm).
Moreover, the presence of CuO dispersed in the support
does not significantly alter the crystalline average size;
the small difference listed in Table 1 is attributable to
experimental noise.
0
0
15
40
0
11.2
6.43
5.4
56
–
CZ (85/15)
0
CZ (60/40)
0
85
55
50
48
–
2 CuO/CeO2
5 CuO/CeO2
8 CuO/CeO2
5 CuO/CZ (85/15)
5 CuO/CZ (60/40)
2
12.09
11.6
12.5
6.8
5.4
7.8
4.8
4.7
0
0
15
40
5.6
79
The catalysts were characterized by means of a Carlo
Erba Sorptomatic 5 (BET analysis) and a Philips PW
1710 diffractometer (XRD analysis).
The activity tests were performed in a lab-scale
apparatus where a mixture containing 0.5–1 vol % of
In the X-rays of 8 CuO/CeO2, the signal of bulk
CuO was clear, suggesting that the segregation of a
fraction of copper occurs, which can be explained tak-
ing into account that the theoretical monolayer cover-
age corresponds to about 5 wt %.
CO, at λ = 2PO /PCO = 2, together with 50 vol % H2,
2
0−15 vol % CO2 and 0–1.4 vol % H2O (N2 as balance),
was fed to the catalytic reactor. The apparatus was
equipped with five Brooks 5850 mass flow controllers
for all components except water, which was instead fed
by saturating the reactive mixture in a thermostatic bath
at room temperature. A hydrogen CLAIND generator
was used to feed H2 by distilled water electrolysis.
Heated lines were used from the saturator to the reactor,
which was placed in an electrical oven. The reactor for
catalytic activity measurements consisted of a fixed bed
of 250 mg of catalyst, to which a flow rate of 30 l/h was
fed. The gas mixture was continuously analyzed by two
different ND-IR detectors for the measurements of CO
and CO2 concentrations, by a TCD for H2, and a para-
magnetic detector for O2 (EMERSON NGA2000 on-
line analyzer), after water removal by a chemical trap
(CaCl2).
Hocevar et al. [19], coupling XRD and XPS analy-
ses, showed a reduction of the unit cell parameter with
increasing Cu content in Cu/ceria catalysts prepared by
coprecipitation and a corresponding increase in the
concentration of both Ce3+ and Cu+ species. This result
was interpreted with the formation of a CuO–CeO2
solid solution, which should be possible only between
Cu+ and Ce4+ ions, which exhibit similar values of ionic
radius and consequent reduction of Ce4+ ions to Ce3+ to
compensate the charge defect.
Nevertheless, phase identification for Cu/ceria cata-
lysts prepared by coprecipitation is still debated. Actu-
ally, the hypothesis by Hocevar et al. [19] of a solid
solution is contrasted with an alternative one, according
to which a mixed phase Cu/Ce with a perovskite-type
structure is formed [16, 17].
Results of our analyses allow no appreciation of sig-
nificant variations in the unit cell parameter. The
absence of such a phenomenon, if not attributable to the
insufficient sensitivity of the apparatus, could be attrib-
uted to the fact that in our samples prepared by impreg-
nation, the amount of the eventual mixed phases or
solid solutions should not be relevant.
The following conditions were used in TPR analy-
sis. Heating rate: 10 K/min; Catalyst load: 300 mg;
Flow rate: 14 l/h; Feed gas: 1 vol % CO or 2 vol % H2
and balance N2.
RESULTS AND DISCUSSION
Figure 1 reports the results of activity tests carried out
The physical properties of all the samples prepared over a 5 CuO/CeO2 catalyst sample for the preferential
were investigated by BET and XRD analyses. From oxidation of CO in a mixture containing 0.5 vol % CO,
BET data, it resulted that the samples supported on 0.5 vol % O2, 50 vol % H2 in N2. The catalyst exhibits
ceria-zirconia exhibit higher values of surface area, high oxidation capacity at a temperature as low as 70°C
while XRD spectra showed that copper oxide homoge- (Fig. 1a) and high selectivity towards CO oxidation
neously deposed on all the investigated supports and (Fig. 1d). One-hundred-percent CO selectivity is visi-
was well dispersed in the porous structure, not evidenc- ble until 100°C, in accordance with conversion of CO of
ing the characteristic signals of CuOx phases up to the 40% (Fig. 1a); this highest value of selectivity was
highest copper concentration (8 wt % CuO).
detected by measuring stoichiometric O2 consumption
KINETICS AND CATALYSIS Vol. 47 No. 5 2006

SUPPORTED CuO/Ce1 – xZrxO2 CATALYSTS
759
Xëé, Sëé, %
Xé , %
XH , %
2
2
100
(a)
100
80
60
40
20
(b)
SCO
2.0
80
60
40
20
1.5
1.0
0.5
XCO
100
50
150
200
50
100
150
200
í, °C
Fig. 1. (a) CO conversion and O selectivity to CO as functions of temperature on 5 wt % CuO/CeO . (b) O and H conversion as
2
2
2
2
3
functions of temperature on 5 wt % CuO/CeO . Test conditions: [CO] = 0.5 vol %; λ = 2; [H ] = 50 vol % and W/F = 0.03 (g s)/cm .
2
2
for the conversion of carbon monoxide alone. At higher the reverse water gas shift is probably much larger than
temperatures, the oxidation of hydrogen also begins,
yielding a higher O2 conversion; the CO conversion
increases with increased temperature, but only up to
150°C, where a maximum is observed. Indeed, at
higher temperature, carbon monoxide conversion starts
to decrease (whereas the consumption of O2 is quite
complete).
that of the oxidation of CO, since under the same con-
tact time involved in the CO PROX tests, the conversion
of CO2 to CO is negligible, although the thermody-
namic limit is not very strict under the operating condi-
tions of interest. Actually, the conversion of CO2
becomes measurable only at temperatures higher than
200°C, i.e., outside the range of interest for the process
of CO removal in H2 streams to fuel cells. This means
The presence of a maximum for CO conversion with
temperature is frequently observed in the CO PROX in principle that the catalyst is not prevented to promote
tests [9]; its origin could be due either to the thermody-
namic limit imposed on CO conversion by the reverse
water gas shift reaction or to a different dependence on
temperature of the kinetics of the two reactions of
hydrogen and CO oxidation (different activation ener-
gies).
However, the evidence that O2 is completely con-
sumed when the conversion of CO is maximum sug-
gests that the oxidation of hydrogen, whose kinetics is
negligible at the lowest temperatures investigated,
becomes faster and not negligible at the highest ones.
Moreover, some experimental tests were performed in
order to verify the capability of CuO/CeO2 to catalyze
the reaction of reverse water gas shift (WGS) reaction.
the complete oxidation of CO to CO2, up to reduce its
concentration to few ppm and that the volcano-shaped
curve of CO conversion as function of temperature
must be explained by other phenomena than the pro-
ceeding of the reverse WGS reaction.
In order to investigate this phenomenon, we have
also carried out parallel tests of CO and H2 oxidation on
the same catalyst and in the same experimental condi-
tions of CO PROX tests, but by feeding one reactant at
Outlet CO concentration, %
1.0
In order to understand whether the reverse WGS
reaction is catalyzed by the catalyst in the temperature
range of interest (50–170°C), experiments were per-
formed feeding only hydrogen (50 vol %) and CO2 to
the reactor and measuring the amounts of CO eventu-
ally produced. Figure 2 shows the results of two differ-
ent series of such tests. On the one hand, the activity of
catalyst has been tested under conditions as close as
possible to those of PROX experiments: 1 vol % CO2
and 50 vol % H2 (diluted by nitrogen). On the other
hand, the eventual production of CO has been verified,
also under more favorable conditions for the reverse
WGS reaction, with a larger concentration of CO2
employed (20%).
0.8
1
2
0.6
0.4
0.2
0
3
4
50
100
150
200
250
300
350
í, °C
Fig. 2. CO outlet concentration as a function of temperature
on 5 wt % CuO/CeO in the reverse WGS reaction. Tests
conditions: [CO ] = 1 vol % (1) or 20 vol % (2) [H ] =
2
2
2
3
50 vol %; balance N . W/F = 0.03 (g s)/cm . Equilibrium
conditions represented by the dashed line for [CO ] = 1 vol %
(3) and continuous line for [CO ] = 20 vol % (4).
In Fig. 2 the limit values of CO concentration
expected by thermodynamics have been plotted.
Results of such experiments show that the time scale of
2
2
2
KINETICS AND CATALYSIS Vol. 47 No. 5 2006

760
CAPUTO et al.
XCO, %
XH , %
XO , %
2
2
100
80
60
40
20
100
80
(a)
(b)
(c)
2.0
1.5
1.0
0.5
H2 ox
H2 + O2
CO + O2 + H2
CO-PROX
CO ox
60
40
20
CO + O2
CO + O2 + H2
50
100
150
200 50
100
T, °C
150
200 50
100
150
200
Fig. 3. (a) CO conversion as functions of temperature on 5 wt % CuO/CeO for the CO oxidation ( ) and CO-PROX ( ) reactions.
2
(b) H conversion as functions of temperature on 5 wt % CuO/CeO for H oxidation ( ) and CO-PROX ( ) reactions. (c) Oxygen
2
2
2
conversion as a function of temperature in the CO oxidation ( ), H oxidation ( ), and CO-PROX ( ) reactions. Tests conditions:
2
CO oxidation CO/O /N = 0.5/0.5/99 (vol %); H oxidation H /O /N = 50/0.5/49.5 (vol %); CO-PROX CO/H /O /N =
2
2
2
2
2
2
2
2
2
3
0.5/50/0.5/49 (vol %). W/F = 0.03 (g s)/cm .
a time. Two different gas mixtures were used as the reac- dation of CO starts at lower temperatures and exhibits
tor feed: one containing 0.5 vol % CO and 0.5 vol % O2 a 100% selectivity until the hydrogen oxidation reac-
and the other containing 50 vol % H2, 0.5 vol % O2 in tion starts (in the range from 70 to 100°C). On the other
N2. Figure 3 shows that the light-off temperature for the hand, when the hydrogen oxidation starts, it subtracts
oxidation of CO is significantly lower than for the oxi- oxygen from the reaction environment, thus limiting
dation of H2: the difference can be quantified in about the CO conversion, which can decrease even with
40°C, thus showing that the CuO/CeO2 catalyst exhib- increasing temperature.
its an intrinsically higher capability in converting CO
If all this is true, the selectivity of the process should
(light-off temperature at <70°C, Fig. 3a) rather than H2
basically depend on the reaction temperature and very
(light-off at about 100°C, Fig. 3b), even if the concen-
poorly or not at all on the CO/H2 ratio in the reaction
tration of CO is significantly lower than that of H2.
mixture, unless a very large difference exists between
Moreover, it seems that the activation energy for the
the kinetic laws of the two competitive reactions. For
oxidation of CO is lowered by the catalyst to values
example, in the case of Pt catalyst for CO PROX such a
much lower than for the oxidation of hydrogen and,
consideration does not apply, since the selectivity
consequently, with increasing temperature, the differ-
towards the preferential oxidation of CO rather than of
ence between the two reaction rates tends to diminish,
H2 lies in the strong adsorption of carbon monoxide on
since the derivative of H2 oxidation rate is higher (being
the catalyst which in the range of optimal temperatures
proportional to the activation energy) up to make the
of the process largely cover the catalyst surface and pre-
two reactants competing each other for the reaction
vents hydrogen to reach the active centers [3, 6]. Thus,
with oxygen. In this way, the conversion of CO reaches
selectivity is strongly dependent on the CO/H2 ratio and
a maximum, because it is limited by the consumption of
decreases when the concentration of CO becomes too
O2 due to H2 oxidation.
low. In order to test such behavior, we carried out CO-
Moreover, by comparing the conversion plots of
both CO and H2 in the two cases of simple feeding of
one reactant at time or in the CO-PROX tests (namely,
when both reactant are fed to the reactor), it is evident
that the oxidation of the two compounds are very
weakly influenced by the simultaneous presence of
both. In particular, the conversion of hydrogen appears
quite unaffected by the presence of CO, even though
carbon monoxide reacts with oxygen in a faster way
and reduces O2 concentration in the reactor. The con-
version of CO is instead weakly lowered by the pres-
ence of H2, but a difference of about 30°C in the light-
off temperatures between the two reactions is still evi-
dent.
PROX tests at constant temperature and variable con-
tact times.
In such experiments, it is possible to measure the
selectivity of the process while changing the CO/H2
ratio without varying the temperature.
The results reported in Fig. 4 show that at low tem-
perature (82°C), where selectivity is still 100%
(namely, H2 is not converted at all), the conversion of
CO increases with increasing contact time, leaving the
selectivity of the process at a constant level (namely, H2
remains unconverted also when the H2/CO ratio
becomes much larger). Moreover, a similar behavior is
also observed at 122°C, where selectivity is lower
Finally, when CO and H2 are both present in the (about 80%) and conversion much higher (close to
reaction mixture, it is possible to confirm that the oxi- 100%): in such conditions, selectivity still remains con-
KINETICS AND CATALYSIS Vol. 47 No. 5 2006

SUPPORTED CuO/Ce1 – xZrxO2 CATALYSTS
761
stant with increasing contact time, while the conversion
of CO increases.
Xëé, Sëé, %
CO emission, ppm
600
For this purpose, we have reported the amount of
unconverted carbon monoxide, since the conversion is
too high and variations are not very easily appreciable.
In these latter experimental conditions (relatively
high temperature and very high CO conversion) similar
results were obtained by Ratnasamy et al. [18], who
confirmed that the selectivity is constant with changing
contact time.
Anyway, the main requirement for the CO-PROX
step in the fuel processor is to decrease the CO concen-
tration in the gas mixture to the PEMFC tolerance limit
(~10 ppm) sacrificing as little fuel as possible (H2). The
main goals of the copper–ceria catalyst in the CO-
PROX reaction compared to gold- and platinum-based
ones are the highest oxygen selectivity, also at high CO
conversion, and the possibility of achieving a deep
abatement level by only increasing the contact time,
since the selectivity only depends on temperature.
The catalytic activity of copper–ceria systems has
been investigated by changing the support: the same
amount of CuO (5 wt %) has been impregnated on two
different CeO2-ZrO2 supports (ceria/zirconia = 85/15;
60/40 wt %). The comparison among the activity of the
different samples is reported in Table 2 in terms of
light-off temperatures (T0) and of 50% CO conversion
(T50). It is evident that the catalyst supported on ceria
exhibits better catalytic performance than those sup-
ported on ceria–zirconia, in agreement with Ratnasamy
[18], who compared the activity of two CuO supported
catalysts, one on pure ceria, the one on 50/50 wt %
ceria–zirconia. In particular, the higher the zirconium
content in the support the higher the T50. Actually, these
differences cannot be explained by the higher surface
area of the sample supported on pure ceria. On the con-
trary, as reported in Table 1 and already discussed
above, the mixed ceria–zirconia oxides, being charac-
terized by lower values of the crystalline mean size and
a more defective structure, typically show higher values
of surface area than pure ceria. Moreover, Table 2 also
reports the value of the selectivity towards the oxida-
tion of CO into CO2 in the CO-PROX tests, as mea-
sured at 50% CO conversion. Results show that at equal
CO conversion, the selectivity decreases with increas-
ing Zr content as well, thus suggesting that the use of
zirconium as a promoter of the OSC properties of the
support does not have a favorable effect in this process.
400
100
80
60
40
20
0
SCO
T = 122°C
T = 82°C
3
2
1
200
~
~
S
CO = 100%
XCO
0.02
0.04
0.06
0.08
0.10
0.12
0.14
W/F, (g s)/Òm3
Fig. 4. CO conversion (1) at 82°C, unconverted CO (2) and
O selectivity to CO (3) at 122°C as functions of contact
time. Catalyst: 5 CuO/CeO ; Tests conditions: [CO] =
2
2
2
0.5 vol %; λ = 2; [H ] = 50 vol % in N .
2
2
involve study of the effect of copper concentration,
since CuO is the active component of the system. By
comparison of the activity of three different copper-
based catalysts characterized by different amounts of
CuO (2, 5, 8 wt %), the effect of copper concentration
on CO-PROX activity has been investigated on ceria
supported samples. Results of the activity tests are
reported in Fig. 5.
Among the catalysts tested, 5 CuO/CeO2 exhibits
the best performance for the CO-PROX reaction in
terms of activity and selectivity towards the CO oxida-
tion. As the copper concentration is increased from 2 to
5 wt %, the activity and the selectivity increase, while
another increase in copper concentration from 5 to
8 wt % lowers both the activity and the selectivity of
the catalyst.
It is known that only the copper species well-dis-
persed on the support are active towards CO oxidation
in CuO/CeO2 catalysts, while bulk CuO cannot adsorb
CO and contributes only a little to the activity [20]. For
a copper concentration lower than 5 wt % in catalysts
supported on CeO2 and prepared by wet impregnation,
no visible CuO crystal phases could be observed [20].
From our XRD analysis, the characteristic signal of
CuO is evidenced only in 8 wt % CuO/CeO2 catalyst,
Table 2. Light-off temperature (T0), T50, and CO selectivity for
5 wt % CuO/CZ (100/0; 85/15; 60/40) catalysts. Feed composi-
tion: CO/O2/H2/N2 = 1/1/50/48 (vol %); W/F = 0.03 (g s)/cm3
Opportune tests of CO and H2 oxidation as reported
in Fig. 3 have been also performed for catalyst samples
supported on mixed ceria–zirconia (not reported here).
They demonstrated that the increased surface area (as
well as the increased reducibility of the active phase)
does not significantly enhance the activity towards CO,
while producing an increment in the H2 oxidation rate,
consequently lowering the overall process selectivity.
SC50O , %
CATALYST CODE T0, °C
T50, °C
5 CuO/CeO2
50
50
60
60
50
109
116
165
146
110
100
100
70
5 CuO/CZ (85/15)
5 CuO/CZ (60/40)
2 CuO/CeO2
100
100
In addition, the search for an optimal catalyst com-
position for the CO-PROX reaction must necessarily
8 CuO/CeO2
KINETICS AND CATALYSIS Vol. 47 No. 5 2006

762
CAPUTO et al.
X
ëé, %
Xé , %
S
ëé, %
2
2
3
100
80
60
40
20
0
(a)
(b)
(c)
2
3
2 3
1
1
1
50
100
150
200 50
í, °C
100
150
200 0
í, °C
20
40
60
80 100
ëé, %
X
Fig. 5. Activity test on 2 wt % CuO/CeO (1); 5 wt % CuO/CeO (2); 8 wt % CuO/CeO (3). (a) CO conversion versus temperature,
2
2
2
(b) O conversion versus temperature; (c) O selectivity to CO versus CO conversion. Reaction condition: [CO] = 1 vol %; λ = 2;
2
2
2
3
[H ] = 50 vol % at W/F = 0.03 (g s)/cm .
2
Xëé, %
Sëé, %
100
80
60
40
20
(a)
(b)
1
2
3
4
50
100
150
200 0
í, °C
20
40
60
80
100
ëé, %
X
Fig. 6. (a) CO conversion as a function of the temperature and (b) oxygen selectivity as a function of the CO conversion for
5 wt %CuO/CeO catalysts. Feed composition: (1) CO/O /H /N = 0.5/0.5/50/49 vol %; (2) CO/O /CO /H /N = 0.5/0.5/15/50/34 vol %;
2
2
2
2
2
2
2
2
3
(3) CO/O /H O/H /N = 0.5/0.5/1.4/50/47.6 vol %; (4) CO/O /CO /H O/H /N = 0.5/0.5/15/1.4/50/33.6 vol %. W/F = 0.03 (g s)/cm .
2
2
2
2
2
2
2
2
2
meaning that in all other cases the copper species H2O. Results of such tests are reported in Fig. 6, where
should be well integrated in the support structure. Thus, it shows the effect of the presence of 15 vol % CO2
an increase in catalytic activity in CO-PROX reaction and/or 1.4 vol % H2O in the standard feed containing
when copper concentration increases from 2 to 5 wt % CO (0.5 vol %), O2 (0.5 vol %) and H2 (50 vol %) (N2
could correlate to the increased number of CuO parti- is used as diluting compound). In the presence of CO2,
cles dispersed in the support, while a further increase in the catalyst activity is depressed, as can be seen by the
copper content (from 5 to 8 wt %) could induce the for- CO conversion curves shifting to a higher temperature
mation of inactive copper species that, on the contrary, than in the absence of carbon dioxide. The selectivity of
could decrease the selectivity of CO oxidation in favor O2 to CO appears changed by the presence of CO2, even
of the conversion of H2. Moreover, we have also seen apparently increasing, as shown by comparison of the
that the active phase is more dispersed and reducible values at equal temperatures.
when supported on CeO2 than on CeO2–ZrO2 supports,
However, if a more correct comparison in terms of
thus explaining the superior activity/selectivity of the
O2 to CO at equal CO conversion is carried out
CuO/CeO2 sample.
(Fig. 6b), the presence of carbon dioxide proves not
In addition, as we pointed out in the introduction very relevant in changing the intrinsic selectivity of the
section, since the actual effectiveness of a catalytic sys- process, meaning that both reactions, the oxidations of
tem is closer to CO-PROX reaction conditions, activity CO and H2, are depressed by the presence of CO2, and
tests must be carried out in the presence of CO2 and in a similar way.
KINETICS AND CATALYSIS Vol. 47 No. 5 2006

SUPPORTED CuO/Ce1 – xZrxO2 CATALYSTS
763
H or CO consumption
However, according to the literature [13, 21, 22],
ceria shows two reduction peaks, the first at a tempera-
ture of about 500°C and corresponding to the reduction
of surface ceria, and the second at about 800°C and
related to the reduction of bulk ceria. Both peaks appear
hence in a range of temperatures not investigated in the
present tests, but it is also known that doping the ceria
structure with transition metal oxides enhances the
reducibility of surface cerium. Such a “strong metal-
support interaction” is responsible for the higher con-
sumption of CO and H2 in the corresponding TPR with
respect to the expected ideal value if only copper is
involved in the reduction process. Moreover, TPR tests
also confirm an intrinsic higher tendency of this cata-
lyst to be reduced by CO rather than by H2. Not only is
the overall CO consumption higher, but also the tem-
peratures at which reduction with carbon monoxide
occurs, are lower. Even at ambient temperature, the cat-
alyst is reduced by CO, as already demonstrated by
Martinez-Arias et al. [17]. Probably, the exact 0.5 ratio
measured for CO consumed per Cu atom in the ambient
temperature peak would suggest that all of the copper is
present in the Cu+ state in the catalyst, also under reac-
tion conditions, notwithstanding the presence of a small
amount of oxygen.
2
H2/Cu = 1.47
CO/Cu = 0.5
2
CO/Cu = 1.5
1
0 3 6 50 100 150 200 250 300 350 20 40
Time, min í, °C Time, min
Fig. 7. TPR of CO (1) and H (2). The plot represents the
2
consumption of the reductant as a function of time (in the
isothermal part at 30°C and at 400°C), or as a function of
temperature (when the sample is heated).
Likewise, the addition of 1.4 vol % of H2O
(CO/O2/H2/H2O/N2 = 0.5/0.5/50/1.4/47.6) also pro-
duces a decrease in catalyst activity, but without rele-
vant variations in the dependence of O2 to CO selectiv-
ity on CO conversion and finally, by the simultaneous
addition of CO2 and H2O (CO/O2/H2/H2O/CO2/N2 =
0.5/0.5/50/1.4/15/33.6), the catalyst activity is still
more depressed, being the two effects addictive. Both
effects can be explained in similar way by assuming
that CO2 and H2O are adsorbed on the same active site
of the reaction. The competitive adsorption between
reactants and these two additional compounds also
exists when they are not present in the feed, as they are
always formed as reactions products. Carbon dioxide,
as has been demonstrated in individual TPD tests (not
reported here), can be still very strongly adsorbed at
temperatures relatively higher than 200°C, while water
is known to strongly interact with copper.
CONCLUSIONS
CuO/CeO2 catalysts shows very high activity
towards the oxidation of CO in a hydrogen-rich envi-
ronment in a range of relatively low temperatures
(70−170°C). The oxidation of CO rather than of H2 is
strongly favored in the low temperature region
(70−120°C), while at higher temperatures, the oxida-
tion of hydrogen starts to take place and assumes a
more relevant role, decreasing the overall selectivity of
the process and causing a maximum of CO conversion
plot with increasing temperature.
The very high selectivity of these catalysts, which is
specific feature of such systems in comparison with
noble metals, is consequently a unique function of tem-
perature, probably due to the different activation energy
of the two competitive reactions, and does not depend
on the CO conversion and hence on the CO/H2 ratio in
the reacting mixture. Actually, if the contact time is
increased, it is potentially possible to achieve a com-
plete abatement of CO without a decrease in the selec-
tivity. The presence of ZrO2 in the support is not bene-
ficial for catalyst performance, while an optimal con-
tent of CuO supported on pure ceria at about 5 wt % has
been found.
To investigate the selectivity shown by copper-
based catalysts towards CO oxidation rather than H2,
TPR analysis has been performed using H2 or CO as a
reducing agent (Fig. 7).
TPR of CO exhibits four reduction peaks, the first of
which at ambient temperature, prior to imposing the
heating ramp of the sample. The total amount of CO
consumed in the experiment is in a CO/Cu ratio of
about 2, suggesting that not only copper but also ceria
is reduced in the range of temperatures investigated, as
generally proposed for the strong metal-support inter-
action between reducible transition metals and the sup-
porting ceria [13, 21]. On the other hand, the hydrogen
TPR shows three peaks in the range 120–200°C, the
first two at the same temperature and in the same
amount with respect to the TPR of CO. The total
amount of H2 consumed in the experiment is in a ratio
H2/Cu of about 1.5, meaning that ceria is reduced in H2
TPR as well.
Finally, good resistance to the presence of reformate
species (carbon dioxide and water) in concentrations
close to those of practical interest has been verified.
REFERENCES
1. Trimm, D.L. and Onsan, Z.I., Catal. Rev., 2000, vol. 43,
nos. 1–2, p. 31.
KINETICS AND CATALYSIS Vol. 47 No. 5 2006

764
CAPUTO et al.
2. Oh, S.H. and Sinkevitch, M.R., J. Catal., 1993, vol. 142, 13. Liu, W. and Stephanopoulos, M., J. Catal., 1995,
p. 254.
vol. 153, p. 304.
14. Sedmak, G., Hocevar, S., and Levec, J., J. Catal., 2003,
3. Kahlich, M.J., Gasteiger, H.A., and Behm, R., J. Catal.,
vol. 213, p. 135.
1997, vol. 171, p. 93.
15. Trovarelli, A., Catalysis by Ceria and Related Materials,
Catalytic Science Series, London: Imperial College,
2002, vol. 2.
16. Fernandez-Garzia, M., Rebollo, E.G., Ruiz, A.G.,
Conesa, J.C., and Soria, J., J. Catal., 1997, vol. 172,
p. 146.
4. Iragashi, H., Uchida, H., Suzuki, M., Sasaki, Y., and
Watambe, M., Appl. Catal., A, 1997, vol. 159, p. 159.
5. Haruta, M., Tsubota, S., Kobajashi, T., Kageyama, H.,
and Genet, M., J. Catal., 1993, vol. 144, p. 175.
6. Schubert, M., Kahlich, M.J., Gasteiger, H., and Behm, J.,
J. Power Sources, 1999, vol. 84, p. 175.
17. Martinez-Arias, A., Fernandez-Garcia, M., Soria, G.J.,
7. Torres and Sanchez, J. Catal., 1997, vol. 168, p. 125.
and Conesa, J.C., J. Catal., 1999, vol. 182, p. 367.
8. Kahlich, M.J., Gasteiger, H.A., and Behm, R., J. Catal.,
18. Ratnasamy, P., Srinivas, D., Satyanarayana, C.V.V.,
Manikandan, P., Senthil Kumaran, R.S., Sachin, M., and
Vasudev, N.Sh., J. Catal., 2004, vol. 221, p. 455.
1999, vol. 182, p. 430.
9. Avgouropoulos, G., Ioannides, T., Matralis, H., Batista, J.,
and Hocevar, S., Catal. Lett., 2001, vol. 73, p. 33.
19. Hocevar, S., Batista, J., and Levec, J., J. Catal., 1999,
vol. 184, p. 39.
10. Avgouropoulos, G., Ioannides, T., Papadopoilou, Ch.,
Batista, J., Hocevar, S., and Matralis, H.K., Catal. Today,
2002, vol. 75, p. 157.
20. Luo, M.F.,Yuan, X.X., and Zheng, X.M., Appl. Catal., A,
1997, vol. 162, p. 121.
11. Kim, D.H. and Cha, J.E., Catal. Lett., 2003, vol. 86,
21. Zimmer, P., Tschope, A., and Birringer, R., J. Catal.,
p. 107.
2002, vol. 205, p. 339.
12. Avgouropoulos, G. and Ioannides, T., Appl. Catal., A, 22. Boaro, M., Vicario, M., de Leitenburg, C., Dolcetti, G.,
2003, vol. 244, no. 1, p. 155.
and Trovarelli, A., Catal. Today, 2003, vol. 77, p. 407.
KINETICS AND CATALYSIS Vol. 47 No. 5 2006