Experimental studies and redox-type monte carlo simulations for the reduction reaction of NO by CO over supported copper

Article abstract of DOI:10.1246/bcsj.20120020

Kinetics experiments for the CONO reaction over copper oxide supported on zirconia are interpreted adequately by theoretical activity decay models. A redox-type reaction mechanism is proposed for the process. A kinetic Monte Carlo simulation algorithm is developed for the case of a redox mechanism. Its results give a reasonable interpretation of the order of the experimental reaction.

Full text of DOI:10.1246/bcsj.20120020

8
84
Bull. Chem. Soc. Jpn. Vol. 85, No. 8, 884­891 (2012)
© 2012 The Chemical Society of Japan
Experimental Studies and Redox-Type Monte Carlo Simulations for
the Reduction Reaction of NO by CO over Supported Copper
Joaquín Cortés,* Eliana Valencia, Carlos González, and Paulo Araya
Facultad de Ciencias, Físicas y Matemáticas, Universidad de Chile, P. O. Box 2777, Santiago, Chile
Received January 14, 2012; E-mail: jcortesgarrido@ing.uchile.cl
Kinetics experiments for the CO­NO reaction over copper oxide supported on zirconia are interpreted adequately by
theoretical activity decay models. A redox-type reaction mechanism is proposed for the process. A kinetic Monte Carlo
simulation algorithm is developed for the case of a redox mechanism. Its results give a reasonable interpretation of the
order of the experimental reaction.
1
2
Nitrogen oxides NO , i.e., NO and N O, are well known
BZ mechanism by Brosilow and Ziff or Yaldran and Khan,
x
2
1
3
14
15
to be responsible for detrimental effects to the environment
such as the greenhouse effect, acid rain, and ozone formation.
Catalytic NOx abatement reactions are considered very im-
portant for the effluent gases from diesel and lean-burn engines.
Such is the case of those that occur in the catalytic converters
Hecker and Bell, Oh et al. and later those of Cho should
be mentioned. Later, as a result of experimental work with
rhodium, Permana et al. and Peden et al. proposed a mech-
anism that has been largely accepted in current literature,
while Chuang and Tan take into account the existence of
the positively or negatively charged NO species Rh­NO and
1
6
16
1
7
+
used to control NO emission from mobile sources, such as
x
¹
automotive exhaust gases, like the reduction of NO by CO
Rh­NO on the surface to explain the behavior of the CO­NO
(
CO­NO reaction) on supported noble metals, that has been
reaction on supported Rh catalysts. Later, our group proposed a
1
8
studied extensively over the last 30 years, as has been reported
in the reviews by Taylor1 and Shelef and Graham. This
reaction has also been one of the classical prototype surface
reactions which under flow conditions are good examples of
nonequilibrium systems that show interesting behaviors such as
dissipative structures, oscillations, kinetics phase transitions,
and so forth, as has been very well reviewed by Evans,
Zhdanov, and Albano.5 The relation between both aspects
have been of great interest to our research group for several
years in our experimental work as well as in the lines in
new mechanism for the same system that took into account
2
19
the last experiments made by Zaera et al.
Most of the mechanisms proposed for the CO­NO reaction
on noble metals assume that the process occurs by an LH-type
mechanism. However, according to published studies, in the
case of supported copper oxide catalysts the reaction occurs
3
11
through a redox or Mars­van Krevelen mechanism. This kind
4
of mechanism, which we will discuss below, will be used in
the simulations and analysis of this paper based on schemes
proposed previously in the literature.
6
7
which we use Monte Carlo (MC) simulations and theoretical
developments.
As is well known, three oxidation states are found on the
8
2+
+
0
copper oxide surface, CuO (Cu ), Cu O (Cu ), and Cu (Cu ),
2
Especially for economic reasons, over the last few years
there has been increasing interest in the literature to study
supported copper catalysts that have shown activity in the
which change thermodynamically between each other as a
function of temperature and the partial pressure of oxygen.
Temperature-programmed reduction (TPR) studies of copper
9
9
2+
CO­NO reaction and in another important reaction in con-
supported on zirconia (Cu/ZrO ) showed that Cu species are
2
1
0
0
+
verters, namely the oxidation of CO (CO­O reaction). These
reduced stepwise to Cu via Cu species in a stream of CO
2
2+
+
+
0
new systems show an important difference in the micro-
scopic behavior of the reaction with respect to noble metal
catalysts. While most of the latter seem to take place through a
Langmuir­Hinshelwood (LH) type mechanism, copper cata-
lysts act by a redox or Mars­van Krevelen11 type mechanism.
This paper reports kinetics experiments with the CO­NO
(Cu to Cu at 100 °C and Cu to Cu at 170­180 °C). On the
other hand, the reduced Cu/ZrO2 surface is restored by NO
treatment (1% in He), a process that takes place completely at
a temperature above 250 °C. The above explains the redox
mechanism of the CO­NO reaction on Cu/ZrO suggested by
2
9
Okamoto, who used XAFS techniques showing the reducing
reaction over copper catalysts supported on ZrO , and a Monte
Carlo simulation algorithm is developed for a redox mech-
anism, which in this case will be used to interpret the system.
action of CO and the oxidizing action of NO, recovering the
surface during the process of the reaction. The products of the
2
CO­NO reaction are CO over the whole temperature range,
2
with N2 produced especially at high temperatures and N2O at
low temperatures, as will be seen later.
Reaction Mechanism
The study of the CO­NO reaction mechanism has a long
history that does not lack interesting controversies. Among the
most representative work in this relation, the old papers on the
The redox mechanism mentioned above, which we will
use to interpret the experimental information, is shown in
Scheme 1, valid at low temperatures at which N production is
2
Published on the web August 8, 2012; doi:10.1246/bcsj.20120020

J. Cortés et al.
Bull. Chem. Soc. Jpn. Vol. 85, No. 8 (2012)
885
2
+
2+
+
+
1
2
3
4
) CO(g) + Cu −O−Cu → CO
2
(g) + Cu −S + Cu
tographs equipped with HWD detectors. The chromatographs
had a HAYASEP D (2 m © 1/8 inch) column to analyze
CO and N O. The conversion of NO and CO was calculated
+
+
) NO(g) + Cu −S → Cu −NOadsorbed
2
2
from the C and N mass balance, considering that the only
nitrogen-containing products are N and N O, and the only
+
+
2+
2+
) NO(g) + Cu −NOadsorbed−Cu → N O(g) + Cu −O−Cu
2
2
2
+
+
carbon-containing product is CO , according to the following
2
) Cu −NOadsorbed + Cu −NOadsorbed
reaction pathways:
2
+
2+
→
N
2
O(g) + Cu −O−Cu + S
CO þ 2NO ¼ CO2 þ N2O
ð1Þ
ð2Þ
S: oxygen vacancy
1
CO þ NO ¼ CO2 þ N2
Scheme 1.
2
Therefore, the N2 concentration was estimated from the
equation
considered negligible. The first step shows the reducing action
of CO(g) present in the gas phase, where the CO (g) product is
2
1
obtained in an Eley Rideal (ER) form, which is desorbed
reducing the surface of the catalyst and producing an oxygen
vacancy, S. The recovery of the original surface occurs by the
oxidizing action of NO(g) in the gas phase, which is first
½N2ꢀ ¼ ð½CO2ꢀ ꢁ ½N2OꢀÞ
ð3Þ
2
where [CO ] and [N O] are the CO and N O concentrations,
respectively, in the reactor effluent.
2
2
2
2
+
adsorbed on the Cu active sites as shown in step (2). This is
9
Kinetic Monte Carlo Simulations
in agreement with the literature and with confirmations in our
laboratory using FTIR techniques. It is assumed that the oxida-
tion of the surface occurs through an ER type form according
to the scheme of stage (3) between a molecule of gaseous NO
and an adsorbed one, or through an LH type mechanism
between two molecules of adsorbed NO according to step (4).
The kinetic Monte Carlo (kMC) algorithm developed in this
paper is similar to others used previously by our group for
various reactions, such as the oxidation of CO or the reduction
2
0
21
of NO by CO, and to that reported recently by our group
for the CH ­O reaction. For those reactions, however, as is
4
2
Both steps produce N O(g) which is desorbed as a product,
often the case, a Langmuir­Hinshelwood (LH) type mechanism
was assumed. In this case we used the redox mechanism of
2
recovering the original surface. The necessary neighborhoods
between the reacting species in the corresponding cases are
taken into account in the Monte Carlo simulations as will be
explained later.
Scheme 1 for the CO­NO reaction over Cu/ZrO . As far as we
2
are aware, in the literature there seem to be no Monte Carlo
algorithms assuming a redox mechanism for the reaction.
The simulation algorithm begins by choosing an event of the
mechanism (adsorption or one of the reactions) according to the
The mechanism of Scheme 1 should become complicated at
higher temperatures at which N production is significant in
2
relation to that of N O. In this case superficial NO dissociation
steps could be included, for example, with the corresponding
probability p of the event defined by
2
i
ꢀ
X
pi ¼ ki
ki
ð4Þ
production of ¢ N (2N
in previous studies.
¼ N (g)) as we have considered
2
2
adsorbed
7
,8
where k corresponds to the rate constant of step i of the
i
mechanism. In the case of the adsorption steps, ki was
calculated from the collision of gas A molecules with a solid
surface (effusion) expression:
Experimental
A 2% Cu/ZrO2 catalyst was prepared by impregnating
zirconia (ZrO ) with an appropriate amount of aqueous solution
2
kiðadsÞ ¼ SA·ð2³MARTÞꢁ1=2PA
ð5Þ
of copper nitrate trihydrate (Merck, p.a.). ZrO was obtained by
2
direct calcination of a commercial hydrated zirconium oxide
where M is the molar mass of gas A, S is the corresponding
A A
(
Zr(OH) , MEI Chemical Corporation, FZ0922) at 500 °C for
sticking coefficient (S = 1 in this case), P is the partial
A A
4
three hours. The impregnated support was then dried in an
oven at 105 °C for 12 h.
pressure, and coefficient · is the area occupied by 1 mol of
superficial metal atoms and A can be CO or NO.
To determine the catalytic activity, 0.1 g of catalyst was
loaded into a 50 cm long and 1 cm diameter tubular reactor. The
The Monte Carlo simulations allowed a set of kinetic
parameters to be selected according to criteria based on obser-
vations of the behavior of the mechanism and on the magnitude
of the experimental data found for the system.
Since step (2) corresponds to the adsorption of NO, we
calculated its kinetic constant from the collisions expression
assuming that it has no activation energy. Step (1), on the other
hand, can be thought of as the combination of the adsorption
of CO and a kinetic process that produces CO2 which is
desorbed to the gas phase. The kinetic constant for this stage
3
¹1
catalyst was calcined in situ for 1 h at 500 °C in a 10 cm min
stream of pure O , cooled to 300 °C, and reduced in a flow of
2
3
¹1
3
0 cm min of a 5% H /Ar stream for 1 h. After that, the
2
feed was switched to pure He and maintained at 300 °C for 1 h.
The reactor temperature was then decreased to room temper-
3
¹1
ature and the reactants were allowed to flow (90 cm min )
at a CO and NO concentration of 14 torr, the balance He.
The temperature was then increased using an RKC model
¹1
REX-P100 programmer at a rate of 2 °C min until the desired
can be conceived as given by the Arrhenius expression k =
i
¹
1
value was reached. The space velocity (GHSV) was 35000 h .
The reactor inlet and outlet streams were analyzed by gas
chromatography using two Perkin-Elmer Autosystem chroma-
A exp(¹E /RT), with a frequency factor A = A equal to that
i
ai
i
1
of the collisions of CO with an additional activation energy
Ea1. A similar assumption was made for step (3). In step (4) we

8
86
Bull. Chem. Soc. Jpn. Vol. 85, No. 8 (2012)
Exp/MC for the (CO­NO)/Cu Reaction
have assumed an arbitrary value for the frequency factors,
Now, if the chosen event is reaction (3), an S site is first
chosen randomly. If that site does not contain adsorbed NO the
event ends, otherwise the search for two Cu continues among
7
1
© 10 , similar to the order of magnitude of the collisions
+
of the first two steps. This means that the activation energy
obtained is relative to a conventional choice of the frequency
factor.
the four copper species surrounding the NO. If there are more
+
that two Cu , two are chosen among the existing ones, and if
In a first analysis simulations were made eliminating one of
the two last steps of the kinetic mechanism. It was seen that
when step (3) did not exist and only step (4) was operational,
orders close to zero were obtained for NO and CO over a wide
range of parameters. In the opposite situation, when step (3)
was operational and step (4) was eliminated from the simu-
lation, an order close to one was obtained for NO, and it was
not possible to get magnitudes close to those obtained in the
experiment, which were in the order of 0.6. This led us to
conclude that the mechanism required both steps to interpret
experimental behavior. This analysis also made it possible
to restrict to a very small range the allowed activation energies
of steps (3) and (4) to remain in the range of the experimental
activities.
there are less than two the event ends without change. If the
event was successful, one molecule of N2O gas leaves the
surface, the site that contained an NO is occupied by an O
+
2+
atom, and the two Cu become Cu
.
If the chosen event is reaction (4), two NO adsorbed on
nearest neighbor S sites are required, and we proceed as
follows: a superficial site is first chosen randomly. If this site
corresponds to NO , a neighboring S site is chosen. If this site
(a)
does not contain an adsorbed NO, the event ends, otherwise
+
the search for Cu continues in the six copper sites neighboring
+
both. If two or more Cu are found the event is successful,
and sometimes it is necessary to choose randomly two of those
2+
sites, which change their charge to Cu . Otherwise the event
ends without change. If the event was successful, an N O gas
2
A complementary analysis showed that the activities and
orders obtained with the proposed mechanism depend very
molecule is produced and an O particle remains in the first
chosen S site. Computing time was measured in Monte Carlo
Steps (MCS), defined as a number of attempts equal to the
number of sites in the substrate. Times on the order of 250 to
600 million MCS were used in the simulations.
little on the activation energy of the CO production reaction
2
(step 1), provided the activation energy of step 1 is less than
that of step (3). Otherwise an order close to one is obtained for
CO and close to zero for NO, contradicting the experiment. It is
therefore concluded that experimental production is controlled
by the magnitudes of the activation energies of the last two
steps, in particular step (3).
Results and Discussion
Kinetics experiments were carried out with the CO­NO
reaction over copper oxide supported on zirconia, observing
first a general vision of its behavior with temperature. Then the
time evolution of activity toward the steady state was deter-
mined, allowing the system to be analyzed in the light of
deactivation models. Finally, determination of the reaction
order allowed the experiment to be compared with the mech-
anism of the reaction using simulation techniques of the Monte
Carlo type.
Based on the above considerations and assumptions, a set of
kinetic parameters was sought that would interpret in the best
possible way the orders and experimental data at 403 K, getting
¹
1
the following relative activation energies: E = 5 kcal mol ,
a1
¹
1
¹1
Ea2 = 0, Ea3 = 16 kcal mol , and E = 17 kcal mol . The
a4
reaction orders obtained in the MC simulation are shown in the
graphs of Figure 5, at 403 K, a temperature at which N pro-
2
duction is small compared to that of N O, a condition that is in
agreement with the model used in the MC.
Behavior of Activity with Temperature. Figure 1 shows
the behavior of the system’s activity with temperature through
the conversion of CO into CO and of NO into N and N O,
2
The substrate used in the simulations is a surface made of
copper sites located in an L © L square lattice (L = 30) with
periodic boundary conditions, and sites S located in the center
of the squares formed by four neighboring copper sites. It is
assumed that copper can only change its charge (in this case
2
2
9
2
similar to what was reported by Okamoto et al. No other
100
2
+
+
80
Cu and Cu ) and that the S sites can contain oxygen in the
O
2
¹
oxidation state or an NO molecule, or they can be vacant.
The MC algorithm begins with the selection of the event.
60
If it corresponds to the adsorption of NO, an S site is chosen
+
randomly and if it is empty, a Cu is sought in the four
+
surrounding sites. A neighboring Cu site is required for the
40
adsorption to take place, so if one of them is found a molecular
+
NO particle is adsorbed on the S site. If no Cu is found, the
event ends without change.
20
If the chosen event is reaction (1), a superficial site S is first
chosen randomly. If this site does not contain O, the event
ends, but if there is O the search for two Cu2 continues among
the four copper species that surround the oxygen. If there are
0
+
0
100
200
300
400
500
Reaction Temperature/°C
2+
less than two Cu the event ends, and if there are three or more
Figure 1. Catalytic activities of 2 wt % Cu/ZrO2 for CO­
NO reaction as a function of the reaction temperature. CO
conversion ( ), NO conversion to N ( ), NO conversion
2+
Cu , two are chosen among the existing ones, which become
+
Cu , and one molecule of CO gas leaves the surface, leaving
2
2
an empty site where the oxygen was.
to N O ( ), total NO conversion ( ).
2

J. Cortés et al.
Bull. Chem. Soc. Jpn. Vol. 85, No. 8 (2012)
887
-
-
-
-
-
5
6
7
8
9
Table 1. Fitting Parameters of the Corresponding Models and
Some Experimental Data of Figure 3
Equation
T
K
º
n
n
8
A ¼ 1 ꢁ K1t
(6)
(7)
403 0.00094 (K1)
463 0.001367 (K1) 0.4833
0.1247
Linear (Ref. 22)
1
A ¼ 1
403 0.001773 (K2) 0.0392
þ K2t
4
63 0.00434 (K2)
0.1082
Hyperbolic (Ref. 23)
4
03 0.0009 (K3)
0.87 (K4)
0.0287
0.1020
>
<
A ¼ K4 expðꢁK3tÞ (8)
Exponential (Ref. 24)
>
:
463 0.00158 (K3)
.77 (K4)
0
.0024
0.0025
0.0026
0.0027
0.0028
0
1
/T
8
><
Figure 2. Arrhenius like expression in the final steady state
PCO = 7.6 torr, PNO = 3.8 torr.
403 0.023 (K5)
0.56 (K6)
463 0.08547 (K5)
0.0022
1
A ¼ 1
(
(9)
K6
þ K5t
Ref. 25)
>
:
0.004149
1
1
1
4
2
0
0
.47 (K6)
One of the most interesting problems in catalysis, which is
also important because of its clear practical applications, is the
loss of catalytic activity that occurs in some systems when the
reaction takes place. The complexity of the phenomenon, due
to its multiple possible causes, makes it difficult to establish
models for its interpretation, explaining why it has not been
given sufficient importance in recent literature. In spite of this,
a number of models have been proposed, some of them with
a rather empirical character and others corresponding to an
assumed deactivation mechanism.
8
6
4
2
0
It is interesting to contrast these results with deactivation
models. Table 1 shows some cases of models for the activity
decay that we have considered as the most representative in
the literature and that we have used to interpret our experi-
mental data, such as eq 6, whose linear shape is the simplest
0
200
400
600
800
1000
1200
t/min
Figure 3. Experimental decay of normalized production
versus time at various temperatures ( ) 403, ( ) 463, and
(
2
2
) 523 K at PCO = 15.2 torr and PNO = 7.6 torr. The lines
possible; a hyperbolic law represented by eq 7 that has
been proposed in case the deactivation is due to aging by
have been drawn to guide the eyes.
2
3
sintering of the catalytic substrate; eq 8, which corresponds to
an exponential law proposed in some cases of poisoning by
products were found as a result of the reaction. An increasing
conversion of CO over the whole temperature range was
observed. Conversion of NO, on the other hand, shows that N2
increases constantly after about 100 °C, with lower conversion
into N O at an intermediate temperature range. At temperatures
higher than about 200 °C N O production decreases strongly
2
4
molecules that are irreversibly chemisorbed on the surface;
and eq 9, proposed in some examples of coking or dirtying
2
5
of the surface. The experimental activities RCO ðexpÞ of
2
the system for initial time (t = 0) and for any time t are
obtained in the experimental case from the conversion values
of CO (XCO) that are shown in Figure 3, using the expression
RCO = FCOXCO/n , where n is the number of moles of
2
2
until it disappears at approximately 300 °C, and above that
the only reaction products are CO and N . This results in the
2
2
2
Cu
Cu
maximum observed for the conversion of NO into N O.
active sites on the surface and FCO is the flow rate of CO in
moles per second. This relation is valid for XCO < 10%, which
is the case of our experimental results. These values allow
2
Figure 2 presents the thermal behavior of the reaction by
means of an Arrhenius graph in the indicated zone, which
shows an apparent activation energy for the system equal to
the determination of the decay of CO production expressed
2
¹
1
1
1 kcal mol .
through the normalized variable at the initial time defined by
the relation ACO ðexpÞ = RCO ðt ¼ tÞ=RCO ðt ¼ 0Þ.
Catalyst Deactivation Models.
Figure 3 describes the
2
2
2
temporal evolution of the system’s activity through the
conversion of CO versus reaction time at constant temperature
and pressure of the gas phase. These results show a deacti-
vation phenomenon observed at three temperatures. Deactiva-
tion is practically not observed at 250 °C.
The ACO ðexpÞ values obtained in the experiment were later
2
adjusted by the theoretical deactivation models that express
ACO ðteÞ as a function of time and of the parameters of
2
each model as shown in Table 1. The fit was obtained on
the computer according to the least-squares criterion, which

8
88
Bull. Chem. Soc. Jpn. Vol. 85, No. 8 (2012)
Exp/MC for the (CO­NO)/Cu Reaction
(a)
(d)
-
6
-6
6.3
6.6
m= - 0.11
n= 0.48
-
-
6.5
-
-
-7
7.5
-8
-6.9
-4.5
0
1
2
3
1
1.5
2
2.5
3
ln PCO
ln PNO
(b)
(e)
n= 0.58
m= 0.13
-
-
-
5.1
5.4
5.7
-
-
-
4.9
5.3
5.7
-6
1
1.5
2
2.5
3
1.5
1.9
2.3
2.7
ln PCO
ln PNO
(c)
(f)
0
.5
.4
.3
.2
m= 0.17
n= 0.25
-
-
-
-
-
5.7
5.9
6.1
6.3
6.5
0
0
0
0.1
0
1
1.5
2
2.5
3
0
1
2
3
ln PCO
ln PNO
Figure 4. Experimental reaction order of (a) CO, 403 K, PNO = 1.9 torr (b) CO, 523 K PNO = 7.6 torr (c) CO, 403 K, PNO = 15.2 torr
d) NO, 403 K, PCO = 15.2 torr (e) NO, 523 K, PCO = 15.2 torr (f) NO, 403 K, PCO = 3.8 torr.
(
allows getting the optimum parameters of the model as those
have good arguments to accept that situation. However, eq 9
can be associated in a wide sense with the alterations of the
active sites due to the mechanism of the reaction, such as the
variation of the chemical state of superficial Cu as will be
commented in the following section.
that correspond to the minimum value of the function º ¼
P
2
ðACO ðteÞ ꢁ ACO ðexpÞ Þ where the sum is made over all
i
2
i
2
i
the experimental points (i), comparing the values of ACO ðexpÞ
2
i
with those corresponding to each theoretical model ACO ðteÞ .
2
i
9
Table 1 shows, for two temperatures, the optimum param-
eters of the corresponding equation and the value of the
quadratic function º for each of the minima. It is seen that in
the case of our experimental data the best fit with the model is
given by eq 9. This equation, on the other hand, has been used
to interpret deactivation results due to coking of the catalytic
Okamoto et al. carried out experiments with the CO­NO
reaction over Cu/ZrO2 and got a curve similar to that of
Figure 1, which shows the relation of activity with temperature,
also showing a decay of the activity at low temperatures like
that obtained by us in Figure 3. They also related these results
with the catalytic hysteresis (XNO vs. T) that they found in their
experiments, which they associated with the gradual deactiva-
tion of the active species during the course of the reaction.
Although this agrees with the alteration of the superficial active
species that we commented in the previous paragraph, they
also relate it with a sintering effect that they found in their
2
6
surface. This phenomenon has been reported by Rainer et al.
in the case of the CO­NO reaction over small supported Pd
particles, commenting that a possible cause of the deactivation
was the blocking of active sites with carbon atoms produced by
the dissociation of CO on the surface. In our case we do not

J. Cortés et al.
Bull. Chem. Soc. Jpn. Vol. 85, No. 8 (2012)
889
(
a)
(c)
n = 0.65
-4
-
-
-
-
4
5
6
7
m = -0.003
-5
-6
-7
0
.5
1
1.5
2
2.5
3
3.5
1
1.5
2
2.5
3
3.5
ln PCO
ln PNO
(
b)
(d)
n = 0.63
-
-
-
-
4
5
6
7
-4
-5
-6
-7
m = -0.06
1
1.5
2
2.5
3
3.5
1
1.5
2
2.5
3
3.5
ln PCO
ln PNO
Figure 5. Monte Carlo reaction order of (a) CO, PNO = 1.9 torr (b) CO, PNO = 15.2 torr (c) NO, PCO = 3.8 torr (d) NO, PCO =
15.2 torr.
experiments. This phenomenon, on the other hand, has been
excluded by Rainer et al.,26 who comment that “the catalyst has
been exposed to much harsher temperature conditions during
initial reduction than it experiences in the CO­NO run.” Since
this has also been the case in our experiments, we have no
arguments to consider the effects of sintering. Neither do we
believe that the fits of Table 1 are sufficient proof to exclude
sintering, such as from the values of º of eq 7 in relation to
those of eq 9 if we consider that both equations were fitted with
different numbers of parameters.
example in the case of the CO­NO reaction, by an empirical
expression between the activity of CO2 and the pressure of
CO and NO:
m
n
RCO ðTONÞ ¼ kPCO PNO
(10)
2
The order m for CO or n for NO can be obtained, both in the
experiment and in the Monte Carlo simulations, from the slope
of the graph of the logarithm of RCO2 versus the logarithm
corresponding to each of the pressures while keeping the other
constant.
Another possible explanation of the phenomenon, given
Figure 4 shows the experimental results of these graphs for
the various situations indicated in each of the cases. In general
terms, straight lines with good correlation are found, showing
that the system’s experimental behavior fits reasonably well the
approximation that is assumed by eq 10. Figure 5 shows, on
the other hand, the same type of graphs obtained in the Monte
Carlo simulations assuming that the reaction behaves accord-
ing to the redox mechanism suggested in Scheme 1 and the
activity remains within the observed range in the experimental
2
6
by Rainer et al., is the presence of inactive nitrogen atoms
adsorbed on the surface, like those proposed by Oh and
1
4
Eickel to explain the behavior of the system in the case of
rhodium. However, associating this case with a given decay
equation is not a simple matter.
Decreasing activity was seen with time in the MC simu-
lations measured in Monte Carlo Steps (MCS) that is not
simple to associate with the deactivation observed in the
experiment, since in this case the decay includes the natural
stochastic evolution characteristic of MC. This, however, is an
interesting aspect that we are currently exploring. In this paper
we have used the steady state MC values corresponding to long
times in which the activity remains constant to compare them
with the theoretical mechanism. In our case this corresponded
to values on the order of 600 million MCS.
information around 403 K, when the production of N is small
2
compared to that of N O.
2
Table 2 gives a summary of the orders obtained from the
experiment and from the Monte Carlo simulations, together
with the only values we have found in the literature for the
same system.
Figure 6 shows in general that production decreases with the
concentration of CO, yCO, in the gas phase. On the other hand,
it is seen that the superficial configuration remains approx-
imately constant over the whole range of CO concentrations,
Reaction Order and Kinetics Mechanism. A practical
way of expressing the relation between the activity of a cataly-
tic system and the pressure of the reactants in the gas phase is to
define the order of the reaction for a given temperature, for
2
+
with slight variations such as, for example, an increase of Cu

8
90
Bull. Chem. Soc. Jpn. Vol. 85, No. 8 (2012)
Exp/MC for the (CO­NO)/Cu Reaction
Table 2. Reaction Order for CO­NO Reaction over Cu/
anism for the reaction. The variation of the activity of CO with
2
m
n
ZrO2 RCO ðTONÞ ¼ kPCO PNO : (a) Experimental 2 wt %
temperature shows an increasing evolution of CO and N with
2
2
2
Cu/ZrO2 (Figure 4); (b) Experimental 1 wt % Cu/ZrO2
temperature, while N O production presents a maximum.
2
(
fresh) (Ref. 9); (c) The Same as (b) Cu/ZrO2 (Deacti-
The experiments were carried out in zones in which N2
production was small, in order to contrast their results with a
simplified mechanism that assumes that CO and N O are the
vated); (d) Monte Carlo Simulation (Figure 5)
2
2
T/°C
m
n
only products of the reaction.
PCO > PNO
PNO > PCO
130
¹0.11
0.13
¹0.003
0.17
¹0.06
0
¹0.1
0.48
0.58
0.63
0.25
0.65
0.85
1.0
(a)
(a)
(d)
(a)
(d)
(b)ref(9)
(c)ref(9)
The experiments showed a time decay of the activity at
moderately low temperatures that was interpreted by models
which show that the decay is not caused by structural changes
of the substrate, but by alterations of the number of active
sites due to the mechanism of the reaction, for example by the
variation of the chemical state of superficial Cu.
2
1
50
30
130
130
250
250
A Monte Carlo type simulation algorithm was developed for
a redox type reaction mechanism. Its results interpret reason-
ably well the order of the reaction.
1
0.01
The authors acknowledge the financial support of this work
by FONDECYT under Project No. 1070351.
0
.8
.6
.4
.2
0
0.008
0.006
0
0
0
References
θi
RCO2
1
2
K. C. Taylor, Catal. Rev., Sci. Eng. 1993, 35, 457.
M. Shelef, G. W. Graham, Catal. Rev., Sci. Eng. 1994, 36,
0.004
0.002
0
4
33.
3
J. W. Evans, Langmuir 1991, 7, 2514.
4
V. P. Zhdanov, B. Kasemo, Surf. Sci. Rep. 1994, 20, 113;
V. P. Zhdanov, Surf. Sci. Rep. 1997, 29, 31.
E. V. Albano, Heterog. Chem. Rev. 1996, 3, 389; E. V.
5
Albano, in Computational Methods in Surface and Colloid Science
in Surfactant Science Series, ed. by M. Borowko, CRC Press,
0.0
0.2
0.4
0.6
0.8
1.0
yco
2
000, Vol. 89, p. 387.
P. Araya, F. Gracia, J. Cortés, E. E. Wolf, Appl. Catal., B
2002, 38, 77.
Figure 6. Production of CO2 and the surface concentration
ªi of species i versus concentration yCO (phase diagram)
obtained from Monte Carlo simulations for PCO + PNO =
6
7
J. Cortés, H. Puschmann, E. Valencia, J. Chem. Phys. 1997,
2
+
30 torr ( ) ªO, ( ) ªNO, ( ) ªCu , ( ) ªCu
þ
, RCO (
).
106, 1467; J. Cortés, E. Valencia, Phys. Rev. E: Stat., Nonlinear,
Soft Matter Phys. 2003, 68, 016111; J. Cortés, E. Valencia, J. Phys.
Chem. B 2004, 108, 2979.
2
+
corresponding to a small decrease of Cu . Over the whole
range of yCO, on the other hand, the concentration of vacant
sites is extremely small, indicating that only a small fraction of
the surface participates in the CO­NO redox reaction and the
surface always remains reduced. It should be recalled, however,
that the mechanism assumes that there is only production of
N O as well as of CO . The experiments were performed in the
8
J. Cortés, H. Puschmann, E. Valencia, J. Chem. Phys. 1998,
109, 6086; J. Cortés, E. Valencia, H. Puschmann, Phys. Chem.
Chem. Phys. 1999, 1, 1577; J. Cortés, A. Narváez, H. Puschmann,
E. Valencia, Chem. Phys. 2003, 288, 77.
9
Y. Okamoto, H. Gotoh, H. Aritani, T. Tanaka, S. Yoshida,
J. Chem. Soc., Faraday Trans. 1997, 93, 3879; Y. Okamoto, T.
Kubota, H. Gotoh, Y. Ohto, H. Aritani, T. Tanaka, S. Yoshida,
J. Chem. Soc., Faraday Trans. 1998, 94, 3743.
2
2
zones in which they can be compared with the assumptions of
1
0
G. G. Jernigan, G. A. Somorjai, J. Catal. 1994, 147, 567;
K. Nagase, Y. Zheng, Y. Kodama, J. Kakuta, J. Catal. 1999, 187,
23; W.-P. Dow, T.-J. Huang, J. Catal. 1996, 160, 171; W.-P. Dow,
the Monte Carlo simulations, which requires N production to
2
be low with respect to N2O.
1
Table 2 shows an order for CO close to zero in all cases,
both in our experimental results as well as in those from the
literature. These values are also well interpreted by the results
of our Monte Carlo simulations. In the case of the order for
NO, on the other hand, in general positive values less than 1.0
are found in the experiments as well as in the simulations.
Y.-P. Wang, T.-J. Huang, J. Catal. 1996, 160, 155; L. Liu, J. Cai, L.
Qi, Q. Yu, K. Sun, B. Liu, F. Gao, L. Dong, Y. Chen, J. Mol. Catal.
A: Chem. 2010, 327, 1.
11
P. Mars, D. W. van Krevelen, Chem. Eng. Sci. 1954, 3, 41.
B. J. Brosilow, R. M. Ziff, J. Catal. 1992, 136, 275; K.
1
2
Yaldram, M. A. Khan, J. Catal. 1991, 131, 369.
1
1
3
4
W. C. Hecker, A. T. Bell, J. Catal. 1983, 84, 200.
S. H. Oh, G. B. Fisher, J. E. Carpenter, D. W. Goodman,
Conclusion
Kinetics experiments were run for the CO­NO reaction over
copper oxide supported on zirconia, which were interpreted
reasonably well by means of activity decay models and Monte
Carlo simulations, in the latter case assuming a redox mech-
J. Catal. 1986, 100, 360; S. H. Oh, C. C. Eickel, J. Catal. 1991,
128, 526.
15 B. K. Cho, J. Catal. 1992, 138, 255; B. K. Cho, J. Catal.
1994, 148, 697.

J. Cortés et al.
H. Permana, K. Y. Simon Ng, C. H. F. Peden, S. J.
Bull. Chem. Soc. Jpn. Vol. 85, No. 8 (2012)
891
1
6
114, 11441.
Schmieg, D. N. Belton, J. Phys. Chem. 1995, 99, 16344; H.
Permana, K. Y. Simon Ng, C. H. F. Peden, S. J. Schmieg, D. K.
Lambert, D. N. Belton, J. Catal. 1996, 164, 194; C. H. F. Peden,
D. N. Belton, S. J. Schmieg, J. Catal. 1995, 155, 204.
22 D. D. Eley, E. K. Rideal, Proc. R. Soc. London, Ser. A 1941,
178, 429.
23 A. F. Ogunye, W. H. Ray, Ind. Eng. Chem. Process Des.
Dev. 1971, 10, 410.
17
18
19
S. S. C. Chuang, C.-D. Tan, J. Catal. 1998, 173, 95.
J. Cortés, E. Valencia, J. Phys. Chem. B 2006, 110, 7887.
F. Zaera, C. S. Gopinath, J. Chem. Phys. 1999, 111, 8088; F.
24 R. N. Pease, L. Stewart, J. Am. Chem. Soc. 1925, 47, 1235;
A. F. Ogunye, W. H. Ray, Ind. Eng. Chem. Process Des. Dev.
1970, 9, 619.
Zaera, C. S. Gopinath, Chem. Phys. Lett. 2000, 332, 209; F. Zaera,
C. S. Gopinath, J. Chem. Phys. 2002, 116, 1128.
25 A. Voorhies, Jr., Ind. Eng. Chem. 1945, 37, 318; C. D.
Prater, R. M. Lago, Adv. Catal. 1956, 8, 293.
26 D. R. Rainer, M. Koranne, S. M. Vesecky, D. W. Goodman,
J. Phys. Chem. B 1997, 101, 10769.
20
1
J. Cortés, E. Valencia, Chem. Phys. Lett. 2008, 463, 251.
J. Cortés, E. Valencia, P. Araya, J. Phys. Chem. C 2010,
2