Kinetics of the CO + N2O reaction over noble metals: II. Rh/Al2O3 and Pt-Rh/Al2O3

Article abstract of DOI:10.1016/j.jcat.2004.01.017

The subsequent reduction of N₂O by CO is an important subreaction in the overall CO+NO reaction. This study reports a kinetic investigation of the single CO+N₂O reaction on Rh/Al ₂O₃, and on Pt-Rh/Al₂O₃, between 264 and 310°C with inlet partial pressures in the range 3.4-13. 8×10^-3 and 1.5-5.9×10^-3 atm respectively for CO and N₂O. Our mechanism proposal differs from those proposed earlier in the literature mainly by the dissociation step of chemisorbed N ₂O species involving a nearest neighbor vacant site. The difference in the selectivity behavior toward the transformation of NO into N₂O on Pt and Rh has been related to competitive adsorptions between NO and N ₂O. A comparison of the values of the equilibrium constant for the adsorption of N₂O with those earlier calculated for NO from the kinetic study of the CO+NO reaction on similar catalysts [J. Catal. 175 (1998) 194] shows that, on Pt, this competition is in favor of N₂O, which can readsorb and dissociate at low temperatures and low NO conversions. In contrast, the strongest NO adsorption on Rh inhibits the subsequent CO+N ₂O reaction. In the second part of this study, the selected mechanism has been tentatively validated in a wider range of temperatures and conversions. The adsorbate coverage dependencies of the strength of the metal-adsorbate bond has been accounted for to model temperature-programmed conversion curves.

Full text of DOI:10.1016/j.jcat.2004.01.017

Journal of Catalysis 223 (2004) 142–151
www.elsevier.com/locate/jcat
Kinetics of the CO + N O reaction over noble metals
2
II. Rh/Al O and Pt–Rh/Al O
2
3
2 3
∗
P. Granger, P. Malfoy, and G. Leclercq
Université des Sciences et Technologies de Lille, Laboratoire de Catalyse de Lille, UMR CNRS No. 8010, Bât. C3, 59655 Villeneuve d’Ascq cedex, France
Received 17 July 2003; revised 24 November 2003; accepted 13 January 2004
Abstract
The subsequent reduction of N O by CO is an important subreaction in the overall CO + NO reaction. This study reports a kinetic
2
◦
investigation of the single CO + N O reaction on Rh/Al O , and on Pt–Rh/Al O , between 264 and 310 C with inlet partial pressures in
2
2
3
2 3
−3
−3
the range 3.4–13.8 × 10 and 1.5–5.9 × 10 atm respectively for CO and N O. Our mechanism proposal differs from those proposed
2
earlier in the literature mainly by the dissociation step of chemisorbed N O species involving a nearest neighbor vacant site. The difference
2
in the selectivity behavior toward the transformation of NO into N O on Pt and Rh has been related to competitive adsorptions between NO
2
and N O. A comparison of the values of the equilibrium constant for the adsorption of N O with those earlier calculated for NO from the
2
2
kinetic study of the CO + NO reaction on similar catalysts [J. Catal. 175 (1998) 194] shows that, on Pt, this competition is in favor of N O,
2
which can readsorb and dissociate at low temperatures and low NO conversions. In contrast, the strongest NO adsorption on Rh inhibits the
subsequent CO + N O reaction. In the second part of this study, the selected mechanism has been tentatively validated in a wider range of
2
temperatures and conversions. The adsorbate coverage dependencies of the strength of the metal–adsorbate bond has been accounted for to
model temperature-programmed conversion curves.

2004 Elsevier Inc. All rights reserved.
Keywords: Three-way catalyst; Kinetics of the CO + N O reaction; Pt- and Rh-based catalysts; Pt–Rh/Al O ; Reaction mechanism; Nitrous oxide;
2
2 3
N O selectivity
2
1
. Introduction
Typically, the overall transformation of NO over auto-
subsequent reduction are of great interest. Among the differ-
ent strategies already suggested for improving the selectivity
toward the transformation of NO into N O, changes in the
2
motive exhaust three-way catalysts obeys competitive and
successive reactions involving the intermediate formation of
N2O:
adsorption properties of noble metals (Pt and Rh) by vari-
ous additives, such as ceria or lanthana, could be attractive
[1–3]. Unfortunately, such modifications are mainly benefi-
cial to the activity, whereas no significant improvement on
the selectivity toward the production of N2 is observed at
least at low temperatures and conversions [4–8]. Hence, it
seems obvious that a better understanding of the specific role
of active phases, and of reaction mechanisms involved over
automotive three-way catalysts (TWC), should bring crucial
information for further practical applications.
Cho and co-workers [9–11] were one of the first to state
the real importance of the subsequent reduction of N2O dur-
ing the overall transformation of NO. These authors sug-
gested that the rate of N2O conversion, formed during the
partial reduction of NO, is sharply enhanced above the light-
off temperature because of a lessening of the CO-inhibiting
2
CO + 2NO → N2 + 2CO2,
(1)
(2)
(3)
CO + 2NO → N2O + CO2,
CO + N2O → N2 + CO2.
Nitrous oxide (N2O) is an undesirable side product be-
cause of its greenhouse gas behavior. Currently, noble metal-
based catalysts are efficient in selectively converting NO into
N2 under three-way conditions, except during the cold start,
N2O being the main N-containing product on Pt- and Rh-
based catalysts. Further technological developments either
to minimize the initial formation of N2O or to enhance its
*
effect due to repulsive interactions between Nads and COads
.
Corresponding author.
E-mail address: pascal.granger@univ-lille1.fr (P. Granger).
Based on this suggestion, competitive adsorptions between
0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2004.01.017

P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
143
the reactants and the products on three-way catalysts could
affect the selectivity behavior of noble metals. On this sub-
ject, comparative kinetic studies of the CO + NO and CO +
N2O reactions should bring relevant arguments for further
discussions.
The kinetics of the overall CO + NO [12,13] and of the
single CO + N2O [14] reactions have been previously in-
vestigated in our laboratory mainly over noble metals. Re-
garding the kinetics of the CO + N2O reaction on Pt/Al2O3,
a mechanism scheme has been selected involving a neigh-
bor vacant site for the dissociation step of chemisorbed N2O
molecules (step (6)),
This paper deals with the kinetics of the CO + N2O re-
action on Rh/Al2O3 and on a bimetallic Pt–Rh/Al2O3 cat-
alyst. The calculation of these kinetic and thermodynamic
parameters and then their comparison with those previously
estimated from the kinetic studies of the CO + NO reaction
[12,13] on similar catalysts give arguments for explaining
the changes in the selectivity behavior of Pt and Rh toward
the production of N O. In the second part of this study,
2
experimental and predicted rates have been compared in a
wider range of temperatures and conversions.
2
. Experimental
CO + ∗ ⇐⇒ COads,
(4)
(5)
(6)
(7)
N2O + ∗ ⇐⇒ N2Oads,
2.1. Catalysts
N2Oads + ∗ → N2 + Oads + ∗ (or N2ads + Oads),
The catalyst preparation and characterization were de-
scribed earlier [13]. Supported noble metal catalysts were
prepared using a conventional coimpregnation of hexachlo-
roplatinic acid and rhodium trichloride solutions with ad-
justed concentrations to obtain 1 wt% Pt and 0.2 wt% Rh.
The impregnated solids were dried and calcined in air at
450 C and then reduced in hydrogen at 450 C for 2 h.
According to the preparation procedure, the formation of
bimetallic Pt–Rh particles is favored on Pt–Rh/Al2O3, rather
than the segregation of Pt and Rh on alumina [18]. Addi-
tional infrared spectroscopic measurements of chemisorbed
CO species on the metals evidenced the presence of sur-
face Pt-enriched particles [19]. The residual chlorine content
varied in the range 0.2–1.2 wt%. The metal dispersion, cal-
culated from H2 chemisorption at room temperature, using a
pulse technique, assuming a H/Ms stoichiometry of 1 (with
Ms = Pts and/or Rhs), were 0.93 and 0.64 respectively on
Rh/Al2O3 and on Pt–Rh/Al2O3. The catalysts were in pow-
der form with an average grain diameter of 80 µm.
COads + Oads → CO2 + 2∗,
where ∗ stands for a vacant adsorption site.
The rate Eq. (8) has been derived assuming competitive
adsorption of the reactants at equilibrium, step (6) as rate
determining, and CO and N2O predominantly adsorbed on
noble metals:
◦
◦
k6λN OPN O
2
2
rN O =
.
(8)
2
2
(
1 + λN OPN O + λCOPCO)
2
2
This rate equation differs from that established earlier by
Belton and Schmieg [15] and Mac Cabe and Wong [16], re-
spectively, on Rh/Al2O3 and Rh (111), who found more reli-
able correlations between experimental and calculated rates
with
k10λN OPN O
2
2
rN O =
,
(9)
2
1
+ λN OPN O + λCOPCO
2
2
that takes into account no additional adsorption site for the
dissociation of chemisorbed N2O molecules according to
N2Oads → N2 + Oads.
(10)
2.2. Catalytic measurements
Cho and co-workers [9,10] also argued on the basis of
Eq. (9) for explaining the rate enhancement of the subse-
quent reduction of N2O on Rh above the light-off tempera-
ture.
The dissociative adsorption of N2O can also be consid-
ered on Rh according to [17]
Details on the experimental setup were reported else-
where [12]. Kinetic measurements were performed in a re-
cycle fixed-bed flow reactor running under differential con-
ditions, with a high recycle ratio of about 180. Preliminary
steady-state experiments showed that the rate remained un-
changed by modifying the catalyst loading, with a constant
space velocity, indicating that external mass transfers were
minimized. A rough estimation of the effectiveness factor by
assuming a Knudsen regime, with an efficient coefficient of
N2O + ∗ → N2 + Oads.
(11)
In that case, the following rate Eq. (12), derived from a
mechanism involving steps (4), (7), and (11), should be ac-
counted for:
−6
2
−1
about 10 m s , led to values close to 1 in our experimen-
tal conditions suggesting that internal diffusion phenomena,
should not operate significantly [20]. Finally, the estimates
k11PN O
2
rN O =
.
(12)
2
1
+ λCOPCO
−4
of the Prater number in the range 2–3.4×10 , correspond-
k6, k10, and k11, in Eqs. (8), (9), and (12), are respectively
ing to a temperature deviation between the core and the sur-
◦
the rate constants of N2O dissociation related to steps (6),
face of the grain less than 0.2 C, suggested the achievement
(
10), and (11), λi and Pi being respectively the equilibrium
of a kinetic regime during rate measurements [21].
Ca dOs o, rNp 2t iOo n) .constant and the partial pressure of reactant i (i =
Typically, the catalyst loadings were 0.2 g mixed with
0.8 g of α-Al2O3. The total flow rate adjusted to 10 L h
−1

144
P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
corresponded to a space velocity of 25,000 h 1. Reactants
and products were analyzed with a HP5890 Series II chro-
matograph fitted with a thermal conductivity detector. Prior
to their quantification, N2O, CO2, CO, and N2 were sepa-
rated on a CTR1 column from Alltech. The specific rate was
calculated according to
−
Diτi
m
ꢀ
−1ꢁ
molh 1
−
g
,
(13)
ri =
where τi is the conversion, Di the molar flow rate for the
compound i, and m the mass of catalyst. The intrinsic rate,
expressed per surface atom, was estimated based on the
number of surface atoms calculated from H2 chemisorption
measurements.
3
3
3
. Results
.1. Previous temperature-programmed experiments
Fig. 1. Temperature-programmed conversion curves for the single
CO + N O reaction under stoichiometric conditions with inlet partial pres-
.1.1. The single CO + N2O reaction
2
−3
−1
.
sures of CO and N O of 5 × 10 atm and a space velocity of 25,000 h
2
Temperature-programmed experiments were performed
under stoichiometric conditions with partial pressures of
−
3
N2O and CO equal to 5 × 10 atm. The conversion pro-
files in Fig. 1 obtained on Rh/Al2O3 and Pt–Rh/Al2O3 are
compared with that previously obtained on Pt/Al2O3 [14].
The overall catalytic activity can be qualitatively compared
on Rh/Al2O3 and Pt–Rh/Al2O3 (Figs. 2b and 2c). It is also
noticeable that the sharp decrease in SN O, above the NO
2
light-off temperature, when the subsequent CO + N2O re-
action becomes predominant, occurs at much higher conver-
sion on Rh/Al2O3 and Pt–Rh/Al2O3 than on Pt/Al2O3. The
difference is clear when comparing the conversion of NO
from the light-off temperatures of respectively 295, 315, and
25 C on 0.2 wt% Rh/Al2O3, 1 wt% Pt–0.2 wt% Rh/Al2O3,
◦
3
and 1 wt% Pt/Al2O3, corresponding to 50% N2O con-
version. This comparison underlines the higher activity of
Rh/Al2O3 compared to Pt/Al2O3 which could be explained
mainly by the higher intrinsic activity of Rh, the number of
Pt surface atoms being higher than that of Rh over compara-
ble weight sample, even though Pt dispersion is lower than
that of Rh. Clearly, the intermediate conversions obtained
on Pt–Rh/Al2O3 between those of Rh/Al2O3 and Pt/Al2O3
corroborate previous observations showing that Pt and Rh
preserve their individual adsorption properties in surface Pt-
enriched bimetallic particles [13].
where SN O starts its sharp decrease: 40% for Pt/Al2O3, 80%
2
for Pt–Rh/Al2O3, and 100% for Rh/Al2O3. Additional com-
parisons in Fig. 3, by plotting the conversion of NO into N2O
vs the overall conversion of NO can be obtained. A maxi-
mum in the production of N2O, associated with the forma-
tion and the successive transformation of N2O, shifting to
higher NO conversion following the sequence Pt/Al2O3 <
Pt–Rh/Al2O3 < Rh/Al2O3, is observed. Of course the in-
fluence of temperature should be taken into account for
explaining such a tendency. However, the temperature effect
on the selectivity changes is usually much less pronounced
than on the activity. Based on these considerations, it seems
obvious that the subsequent reduction of N2O, during the
overall CO + NO reaction, occurs more readily on Pt/Al2O3
than on Rh/Al2O3, Pt–Rh/Al2O3 exhibiting an intermediate
selectivity behavior.
3
.1.2. The overall CO + NO reaction
Temperature-programmed reduction of NO by CO has
been investigated on the same Rh/Al2O3 and Pt–Rh/Al2O3
catalysts under stoichiometric conditions with PNO and PCO
−3
of 5 × 10 atm. The conversion and selectivity profiles
in Fig. 2 are compared with those previously obtained on
3
.2. Steady-state kinetic measurements
Pt/Al2O3 [14]. NO light-off temperatures of respectively
◦
2
62, 286, and 409 C on Rh/Al2O3, Pt–Rh/Al2O3, and
Pt/Al2O3 lead to the same reactivity sequence previously es-
tablished for the single CO + N2O reaction. Consequently,
Rh/Al2O3 still exhibits the highest overall activity. Now
regarding the selectivity behavior, it is obvious that the se-
lectivity profile of Pt/Al2O3 differs from those obtained on
Rh/Al2O3 and Pt–Rh/Al2O3. As observed in Fig. 2a, at low
N3 .22 O. 1 .r e Ta ec mt i op ne rature dependency of the rate of the CO +
Steady-state rate measurements were performed under
◦
stoichiometric conditions between 264 and 310 C with in-
−3
let partial pressures of CO and N2O of 6 × 10 atm. The
values of the apparent activation energy calculated from the
slopes of the Arrhenius plots in Fig. 4 are respectively equal
conversions (below the light-off temperature), SN O gradu-
2
−1
ally decreases on Pt/Al2O3 while it remains quasi-constant
to 137 and 133.6 kJ mol on Rh/Al2O3 and Pt–Rh/Al2O3.

P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
145
(a)
Fig. 3. Plots of the conversion of NO into N O as a function of the overall
2
−3
NO conversion during the CO+NO reactions with P
= PCO = 5×10
NO
atm on Pt/Al O , Rh/Al O , and Pt–Rh/Al O .
2
3
2
3
2 3
(b)
Fig. 4. Arrhenius plots for the CO + N O reaction on Rh/Al O (2) and
2
2 3
Pt–Rh/Al O (1) under stoichiometric conditions with inlet partial pres-
2
3
−3
sures of CO and N O of 6 × 10 atm. Predicted curves calculated from
2
Eq. (16) (—).
3
.2.2. Partial pressure dependencies of the rate of N2O
transformation
The influence of the reactant partial pressures (PCO and
◦
PN O) on the rate was investigated at 300 C in the range
3
2
−3
−3
.4–13.8×10 and 1.5–5.9×10 atm for CO and N2O,
respectively. Figs. 5 and 6 illustrate the influence of the par-
(
c)
tial pressure of N2O and CO on the rate of the CO + N2O
reaction, at constant PCO and PN O, respectively. The val-
Fig. 2. Comparative temperature-programmed experiments of the
2
ues for the apparent average kinetic orders with respect to
CO + N O and CO + NO reactions under stoichiometric conditions: (!)
2
N O selectivity in the CO + NO reaction; (1) conversion of N O (in the
the partial pressure of N O are 0.75 ± 0.11 and 0.71 ±
2
2
2
single CO + N O reaction); (2) conversion of NO on Pt/Al O (a), on
2
2
3
0.11 on Rh/Al O and on Pt–Rh/Al O , while the nega-
2
3
2
3
Rh/Al O (b), on Pt–Rh/Al O (c).
2
3
2 3
tive values obtained for the CO order, respectively equal to

146
P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
CO + N2O reaction, particularly on Rh. Previous investiga-
tions on the kinetics of the CO + NO reaction highlighted
the beneficial effect of rhodium, dissociating NO more eas-
ily than platinum [22,23], and also the structure sensitivity of
such a reaction [23–25]. In the case of the single CO + N2O
reaction, comparisons in Fig. 1 also indicate that rhodium is
more active than Pt in the transformation of N2O, in accor-
dance with previous results obtained for the transformation
of N2O the presence and/or in the absence of CO [26].
Currently, there is no general agreement on the mecha-
nism of the CO + N2O reaction, the involvement of a disso-
ciative adsorption, or a twostep mechanism involving mole-
cular adsorption and dissociation steps, being questionable
on Rh. In this latter case, the involvement of a nearest neigh-
bor site could also be considered. In a previous investigation
performed in our laboratory, it was found that the mecha-
nism, noted in the Introduction, is the most representative on
Pt/Al2O3 for depicting the single CO + N2O reaction. Alter-
nately, previous workers [10,11,15,16]disagree with step (6)
on Rh, and rather propose step (9) (N2Oads → N2 + Oads).
Fig. 5. N O partial pressure dependency of the rate of the CO + N O
2
2
◦
reaction at 300 C on Rh/Al O (2) and on Pt–Rh/Al O (1). Inlet
2
3
2 3
= 6 × 10 3 atm.
−
Consequently molecular or dissociative adsorption for N O
P
2
CO
could be equally considered on Rh, as well as the require-
ment or not of an additional nearest neighbor vacant site for
the dissociation of chemisorbed N2O molecules.
The clarification of the mechanism of the CO + N2O is
of interest for further explaining the formation and the sub-
sequent transformation of N2O during the reduction of NO
by CO. As illustrated in Figs. 2 and 3, it seems obvious that
the subsequent CO + N2O reaction in the presence of NO,
occurs more readily on Pt/Al2O3 than on Rh/Al2O3 and Pt–
Rh/Al2O3 despite the fact that Rh/Al2O3 is more active in
converting N2O than Pt/Al2O3 in the absence of NO (see
Fig. 1). Such a discrepancy could be equally explained by
differences in the ability of Pt and Rh to dissociate N2O, or
by different properties toward the adsorptions of CO, NO,
and N2O, when they compete in accordance with previous
authors [10,11,27].
4
.1. On supported Rh catalysts
In this section, kinetic data obtained on Rh/Al2O3 have
Fig. 6. CO partial pressure dependency of the rate of the CO + N O
been interpreted on the basis of previous findings mainly
obtained on rhodium-based catalysts [15,16], and those ob-
tained more recently in our laboratory on Pt/Al O [14].
2
◦
reaction at 300 C on Rh/Al O (2) and on Pt–Rh/Al O (1). Inlet
2
3
2 3
−3
P
= 6 × 10 atm.
N O
2
2
3
First, we have attempted to check for Rh/Al2O3 the valid-
ity of the mechanism earlier established for Pt/Al O .
−
0.54± 0.08 and −0.58± 0.09, reflect the well-known CO-
2
3
inhibiting effect on the reaction rate. These reaction orders
are in qualitative agreement with those reported in the liter-
ature [15,16].
No relevant conclusion can be drawn from the compa-
rable CO order values of respectively −0.51 and −0.54
on Pt/Al O and Rh/Al O . They underline the usual CO-
2
3
2
3
inhibiting effect on the rate of N2O transformation that can
be modeled either by Eqs. (8), (9), or (12). On the other
4
. Discussion
hand, the examination of the apparent N O order value on
2
Rh/Al2O3, invalidates Eq. (12) which only assumes a value
Currently, there is a general agreement on the mechanism
equal to one. Consequently, the occurrence of a dissociative
adsorption of N2O, via step (11), seems to be unlikely on
Rh/Al2O3. Let us now examine a two-step mechanism with a
scheme for displaying the CO + NO reaction over noble
metals, whereas some divergences arise in the case of the

P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
147
Table 1
◦
Comparison of kinetic and thermodynamic parameters for the CO + N O reaction calculated on Rh/Al O at 300 C
Reaction
2
2 3
a
b
ꢀ
k
α
β
i
λ
CO
λ
Residual sum
of square
i
N O
2
n
−
1
−1
2
−1 −1
(10 mol h at )
s
step
i = CO
i = N O
i = CO
i = N O
(atm
)
(atm
)
2
2
(6)
34.15
31.0
0.78
0.82
70 ± 25
53 ± 16
42 ± 13
88.9 ± 26.7
5
5
4 ± 16
261.5 ± 78.5
c
c
c
−8
0 ± 8
36 ± 6
34 ± 5.1
7.86 × 10
(10)
75.4
57.7
0.51
0.66
268 ± 80
350 ± 105
3
1
35 ± 100
438 ± 131
c
c
41.8 ± 6.3 c
−8
94 ± 29
113 ± 17
10.5 × 10
ꢂ
ꢂ
a
Slopes of the linear plots
P
P
/r
,
P
/r ,P
/r
vs P
and P
.
CO
N O CO CO
N O CO N O CO
N O
2
2
2
2
ꢂ
ꢂ
b
c
Intercepts of the linear plots
Adjusted values from the statistical method.
P
P
/r
,
P
/r ,P
/r
vs P
2
and P
.
CO
N O CO CO
2
N O CO N O CO
N O
2
2
dissociation of molecular chemisorbed N2O species involv-
ing or not an additional site. In that case, the corresponding
rate Eqs. (8) and (9) can be linearized leading respectively to
ꢃ
investigations found that the orientation of adsorbed N2O
molecules over metallic sites could have an influence on the
further elementary steps [28–30]. If N2O chemisorbs via the
terminal N atom, its subsequent dissociation is likely to ne-
cessitate a neighbor free adsorption site. On the contrary, if
N2O is coordinated via the oxygen atom, then the dissocia-
tion could take place without an additional site. As a matter
of fact theoretical calculations performed in our laboratory
over monometallic and bimetallic Pt and Rh clusters [31],
based on the density functional theory, showed that N2O ad-
sorbs via the terminal N atom, whereas the configuration
PN O
1 + λN OPN O + λCOPCO
2
2
2
=
ꢂ
,
(14)
(15)
rN O
6 N2O
k λ
2
PN O
1 + λN OPN O + λCOPCO
2
2
2
=
.
rN O
k10λN O
2
2
ꢂ
In both cases, linear plots PN O/rN O and PN O/rN O
2
2
2
2
versus PN O or PCO are obtained. A mechanism has been
2
with N O bound to the metal via the oxygen is thermody-
2
tentatively selected by comparing the parameters kn and λi
estimated from the intercepts βi and the slopes αi of the
straight lines. Unfortunately the calculations, summarized in
Table 1, lead to positive values in both cases and thus can-
not discriminate between steps (6) and (10). On the other
hand, the comparison of the values of λCO obtained in this
study, with those previously estimated from the kinetic in-
vestigation of the CO + NO reaction on the same catalyst
namically unstable over noble metals. Hence, both theoret-
ical and kinetic results obtained in our laboratory speak in
favor of molecular N2O adsorption, and subsequent dissoci-
ation of N2O involving active sites. Such a result is in line
with conclusions of previous authors who suggested that the
dissociation of N2O on Ru (001) necessitates at least four
bare Ru atoms [27].
ꢀ
The comparison of the optimized values for λi and k on
n
[
13], could be relevant since comparable values should be
Pt/Al2O3 and Rh/Al2O3, in Table 2, shows slight changes in
obtained. According to the margin of error, the estimates
λN O while λCO does not significantly vary probably within
2
of λCO, using Eq. (8), are in good agreement with the op-
the margin of error. As far as the intrinsic rate of N2O disso-
−1
ꢀ
timized value of λCO = 71 atm earlier obtained for the
CO+NO reaction [13], whereas the numerical solutions ob-
tained from Eq. (9) diverge. Such a discrimination can also
be achieved using the solver setup on Excel 5 from Microsoft
by minimizing the square differences between experimental
and calculated rates using Eqs. (8) and (9). The optimization
routine, described elsewhere [13], leads to a better adjust-
ment from Eq. (8) with a lower residual sum of squares. In
addition, a better agreement between the set of values ob-
tained from the optimization and the graphical procedure is
obtained using Eq. (8). Consequently, the results obtained
from both methods seem to be in better agreement with the
fact that the mechanism earlier described on Pt/Al2O3, in-
volving a nearest neighbor vacant site for the dissociation
of chemisorbed N2O molecules (step (6)), is still the most
representative on Rh/Al2O3. Such a proposal seems consis-
tent with previous surface science studies which showed that
the adsorption and the subsequent dissociation of N2O on
Rh (111) necessitate geometrical requirements [28]. Earlier
ciation (k ) is concerned, only slight variations are observ-
6
−1
−1
able (3400 on Rh/Al2O3 against 1000 molec h at on
s
Pt/Al2O3). Nevertheless, they could reflect changes in the re-
activity of chemisorbed N2O molecules on Rh and Pt. Such
a statement is supported by theoretical calculations [31],
which show that modifications in the geometrical configu-
rations of chemisorbed N2O species on Pt and Rh. N2O,
coordinated to Pt, are upright, while bent N2O molecules,
bonded with two adjacent Rh atoms, are stabilized on Rh
clusters. Such changes in configurations lead to a weaken-
ing of the terminal N–O bond being more accentuated on
Rh than on Pt. Hence, these results corroborate our kinetic
data showing that the dissociation of the N–O bond in N O
2
molecules is easier on Rh/Al O than on Pt/Al O .
2
3
2
3
The comparison between λ calculated for the single
i
CO + N O reaction with those previously estimated for the
2
CO + NO reaction on Pt/Al O and Rh/Al O is interest-
2
3
2
3
ing because it could explain changes in selectivity for the
CO + NO reaction (Figs. 2 and 3). In the overall CO + NO

148
P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
Table 2
◦
Comparison of the kinetic and thermodynamic data obtained at 300 C for the CO + NO and CO + N O reactions
Catalyst
2
ꢀ a
k
n
Reaction
P
P
or P
λ
or λ
λ
CO
3
NO
N O
NO
N O
CO
−1
2
2
)
−
−3
−1
2
−1 −1
(10 mol h at )
s
(
10 atm)
(10 atm)
(atm
(atm
)
Pt/Al O
2
CO + NO
5–9
3.6–13.8
1.5–5.6
1.8–7.4
11 ± 2
121 ± 18
4.0 ± 0.6
3
CO + N O
90 ± 13
78 ± 12
10.0 ± 1.5
2
Rh/Al O
CO + NO
3–13.6
3.6–13.8
0.5–3.2
1.8–7.4
472 ± 71
71 ± 11
50 ± 8
123.1 ± 18.5
2
3
CO + N O
36 ± 6
34.0 ± 5.1
2
Pt–Rh/Al O
2
CO + NO
3–8
3.8–10.4
1.5–5.6
1.5–5.6
505 ± 76
66 ± 10
122 ± 18
2.8 ± 0.4
4.2 ± 0.6
3
CO + N O
91 ± 14
2
a
Intrinsic rate constant for the dissociation step of N O (N O + ∗ → N + O + ∗) and of NO (NOads + ∗ → Nads + Oads) in the mechanism of the
2
2
ads
2
ads
CO + N O and CO + NO reactions.
2
reaction, the initial selectivity toward the transformation of
NO into N2O depends on the relative rates of reactions (1)
tained on Pt/Al2O3 than on Rh/Al2O3. Such a comparison
corroborates previous results indicating that Pt–Rh/Al2O3 is
a surface Pt-enriched catalyst.
and (2), while at higher conversions, SN O is also affected
2
by the subsequent reduction of N2O via reaction (3). The
differences in the selectivity behavior of Pt and Rh are very
Now regarding the selectivity toward the transforma-
tion of N2O, Fig. 3 shows that the selectivity behavior of
Pt–Rh/Al2O3 is closer to that of Rh/Al2O3 than that of
Pt/Al2O3. As illustrated in Fig. 3, the subsequent CO + N2O
reaction becomes significant above 80% NO conversion on
Pt–Rh/Al2O3 and Rh/Al2O3, whereas it occurs at much
lower NO conversion on Pt/Al2O3. According to previous
comparison on Rh, such kinetic features on Pt–Rh/Al2O3
can easily be explained by the difference in the equilibrium
adsorption constant of NO and N2O (see Table 2). This com-
parison indicates that competitive adsorption is still in favor
of NO on Pt–Rh/Al2O3. Consequently, N2O can readsorb
and then decompose when NO is quasi-completely depleted
from the surface near 100% NO conversion.
well explained by comparing the values obtained for λN O
2
and λNO, λN O being approximately one order of magni-
2
tude higher than λNO on Pt/Al2O3; consequently, this can
explain the better reactivity of N2O, since its surface cover-
age is higher than that of NO. Such a difference in adsorp-
tion behavior could also satisfactorily explain the changes of
SN O in the CO + NO reaction on Rh/Al2O3, N2O, initially
2
formed on Rh/Al2O3, which can hardly adsorb because of
its much lower λN O compared with λNO. The reverse ten-
2
dency is observed on Pt, because λNO is significantly much
lower than λN O. In such a case, N2O can be adsorbed even
2
in the presence of NO and further dissociate, explaining the
In our case, the previous explanations suggested earlier
by Cho do not seem to be adequate since only slight varia-
tions in λCO are observed, probably varying within the mar-
gin of error. According to Cho’s statements, the adjusted val-
ues of λCO should be lower in the case of the CO + NO, ac-
counting for repulsive interactions between Nads and COads,
than those calculated from the single CO+N2O, the dissoci-
ation of the N–N bond being not thermodynamically favored
in N2O molecules. In fact, the results in Table 2 show the re-
verse trend, the lowest values being systematically obtained
from the CO + N2O reaction.
faster decrease of SN O on Pt than on Rh.
2
4
.2. On supported bimetallic Pt–Rh catalysts
With regard to Pt–Rh/Al2O3, two borderline cases can
be envisaged because of competitive adsorptions of the re-
actants or not. Such a suggestion is supported by previ-
ous kinetic and spectroscopic investigations of the CO +
NO reaction respectively on Pt–Rh/Al2O3 [12], and on Pt–
Rh/SiO2 [32]. In the former case, noncompetitive adsorp-
tions of NO and CO were concluded based on the compari-
son of λNO and λCO on Pt/Al2O3 and Rh/Al2O3, λNO being
substantially higher on Rh than on Pt, while the reverse trend
was observed on Pt/Al2O3 for λCO. The values reported in
4
.3. Simulation of temperature-programmed experiments
under stoichiometric conditions
Table 2 show only slight differences on the estimates of λN O
2
In a previous paper an attempt for modeling the CO +
N2O on Pt/Al2O3 under three-way conditions, near 100%
N2O conversion, is reported [14]. Temperature-programmed
and λCO on Pt/Al2O3 and Rh/Al2O3. Accordingly, compet-
itive adsorptions of N2O and CO seem more probable on
Pt–Rh/Al2O3, Eq. (8) being still valid for depicting the re-
actant partial pressure dependencies of the rate of N2O re-
(
TP) conversion curves can be predicted under differential
conditions using
◦
ꢀ
duction on Pt–Rh/Al2O3 at 300 C. The parameters k and
6
DN OτN O
λi have been calculated at that temperature using procedures
similar to those applied for monometallic catalysts. The ad-
2
2
rN O =
(16)
2
m
0
justment routine leads to average values for λN O between
k λ
6
(1 − τ
N O
2
)P
N2O N O
2
2
N2O
=
,
those calculated on Pt/Al2O3 and Rh/Al2O3 (see Table 2).
However, it is worthwhile to note that the optimized values
0
0
CO
2
(
1 + λN O(1 − τN O)P
+ λCO(1 − τCO)P
)
2
2
ꢀ
0
i
for k and λN O on Pt–Rh/Al2O3 are closer to those ob-
P is the inlet partial pressure of N2O.
6
2

P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
149
Table 3
Comparison of the adjusted kinetic and thermodynamic parameters for the CO + N O reaction on monometallic and bimetallic Pt- and Rh-based catalysts
2
b
c
Catalyst
Preexponential factor
Adsorption enthalpy
Activation energy
9
−1
−2
−1
−2
−1
−1
A (10 s
)
f
(10 atm
)
f
(10 atm
)
ꢀH
ꢀH
E (kJ mol
112 ± 17
)
CO
N O
2
ads,CO
ads,N O
2
a
Pt/Al O
2
4.1 ± 0.6
650 ± 98
3.9 ± 0.6
12.0 ± 1.8
1.4 ± 0.2
10.0 ± 1.5
5.2 ± 0.8
64 ± 10
−28.6 ± 4.3
−39.1 ± 6.0
−32.5 ± 4.9
−34.3 ± 5.2
−19.3 ± 2.9
−33.3 ± 5.0
3
Rh/Al O
130.3 ± 19.5
117.7 ± 17.7
2
3
Pt–Rh/Al O
2
6.3 ± 1.0
3
a
See Ref. [14].
b
c
−1
In kJ mol
.
Related to step (6).
sorption properties of noble metals during the temperature-
programmed CO + N2O reaction. Typically, deactivation
phenomena could alter differently the surface properties of
noble metals with the buildup of oxygen species yielding
electrophilic sites [33]. In the particular case of Rh-based
catalysts, the well-known oxidative-disruption/reductive ag-
glomeration could alter the adsorption properties of Rh par-
ticles [34]. Alternately, the accumulation of adsorbates can
also change the adsorption properties of noble metals. Such a
statement is supported by previous CO chemisorption mea-
surements performed on Pt–Rh/Al2O3 [35], which showed
a weakening of the metal–CO bond with an increase in CO
coverage. Such an observation is in line with previous sur-
face science studies explaining such an effect by a lessening
∗
of the electron back-donation into the π antibonding mole-
cular orbital of CO, or lateral repulsive interactions between
adsorbed CO molecules mainly at high CO coverage [36,37].
Such a surface coverage dependency of CO adsorption on
Pt–Rh/Al2O3 has been modeled using Freundlich’s assump-
tions [35], ꢀHads,CO being a linear function of ln nCO
on Pt–Rh/Al2O3 (nCO stands for the amount of adsorbed
CO molecules at a given PCO). According to these pre-
vious observations, semilogarithmic relationship have also
been considered for modeling changes in ꢀHads,i during
temperature-programmed experiments of the CO + NO and
of the CO + N2O reactions respectively on Pt–Rh/Al2O3
Fig. 7. Correlation between experimental and calculated rates on Rh/Al O
2
3
(
2) and on Pt–Rh/Al O (!).
2 3
Typically, the simulation of temperature-programmed
conversion curves consists in quantifying the temperature
dependency of the rate constant k6, related to step (6), and
of the equilibrium constants for CO and N2O adsorptions ac-
cording to Eqs. (17) and (18). The parameters A and fi , the
preexponential factors of respectively k6 and λi , E the acti-
vation energy related to step (6), and ꢀHads,i the adsorption
enthalpy of the reactant i have been optimized from steady-
state rate measurements performed at low conversion in
Figs. 4–6. The adjusted values for these different parameters
are collected in Table 3. Fig. 7 shows a good fit between ex-
perimental and calculated rates from Eqs. (8), (17), and (18).
[
35] and Pt/Al2O3 [12]. The influence of the partial pressures
of CO and N2O (i.e., the surface coverage) on ꢀHads,i can be
modeled by empirical equations which account for mutual
interactions involving only CO, and only N2O molecules, or
either N2O or CO molecules. The extent of such interactions
depends on the N2O and CO coverages and therefore on the
partial pressure ratio PN O/PCO:
2
ꢄ
k6 = Aexp −
ꢄ
ꢅ
ꢀ
ꢀ
Hads,N O = aN O + b lnPN O + c ln PCO,
(19)
(20)
2
2
2
E
and
(17)
(18)
ꢀ
ꢀ
Hads,CO = aCO + b lnPN O + c lnPCO.
RT
2
ꢅ
ꢄ
ꢅ
ꢀ
Hads,i
ꢀSads,i
Under stoichiometric conditions, PN2O = P = P , then
CO
i
λi = fi exp −
with fi = exp
.
RT
R
Eqs. (19) and (20) can be simplified, leading to
However, divergences were found between experimental and
calculated TP conversion curves using the optimized values
of the adsorption enthalpies for CO and N2O adsorption in
Table 3. Such discrepancies have not been related to signifi-
cant heat and mass-transfer restrictions during temperature-
programmed experiments, but mainly to changes in ad-
ꢀ
ꢀ
Hads,N O = aN O + bN O lnPi,
(21)
(22)
2
2
2
Hads,CO = aCO + bCO lnPi.
The parameters E, ai, and bi and the preexponential
factor A and fi have been adjusted on Rh/Al2O3 and Pt–
Rh/Al2O3 by comparing experimental and predicted TP con-

150
P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
Table 4
Influence of partial pressures of the reactants on the adsorption enthalpies of CO and NO over supported Pt- and Rh-based catalysts
9
−1
−1
a
−2
−1
−1
−1
bi (kJ mol )
Catalyst
A (10 s
)
E (kJ mol
)
Adsorbate
f
(10 atm
)
ai (kJ mol
)
i
b
Pt/Al O
2
N O
2
CO
6.4 ± 1.0
−12.2 ± 1.8
3.9 ± 0.6
3
18.0 ± 2.7
−2.8 ± 0.4
5.0 ± 0.8
Rh/Al O
550 ± 82
2.0± 0.3
131.8 ± 19.8
110.1 ± 16.5
N O
2
CO
7.0 ± 1.0
−10.1 ± 1.5
−16.7 ± 2.5
4.0 ± 0.6
1.7 ± 0.3
2
3
5.0 ± 0.8
Pt–Rh/Al O
2
N O
2
CO
11.2 ± 1.7
93 ± 14
−7.3 ± 1.1
4.0 ± 0.6
3
−13.6 ± 2.1
1.3 ± 0.2
a
/RT ] (atm-1) with f = exp[ꢀSads,i/R].
λ = f exp[−ꢀH
i
i
ads,i
i
b
See Ref. [14]
cules on nearest neighbor vacant sites according to previ-
ous findings on Pt/Al2O3. The estimation of the equilibrium
constants of N2O, CO and of the rate constant of N2O dis-
sociation on Pt and Rh and their comparison suggests that
CO and N2O compete for adsorption on Pt–Rh/Al2O3 and
that N2O dissociates more readily on Rh/Al2O3. Additional
comparisons with the equilibrium constants for CO and NO
adsorption previously calculated from the kinetics of the
CO + NO reaction give arguments for explaining changes
in selectivity toward the formation of N2O particularly dur-
ing the cold start. It appears that competitive adsorptions
between NO and N2O govern the selectivity behavior of
Pt and Rh. Strongly chemisorbed NO species on Rh would
prevent the readsorption of N2O, initially formed, and its
subsequent dissociation. On the other hand, such a competi-
tion is in favor of N2O on Pt. Such a difference in adsorption
properties correctly explains the most interesting selectivity
toward the production of N2 on Pt/Al2O3. Correlations be-
tween predicted and experimental temperature-programmed
conversion curves can be achieved by taking into account the
surface coverage dependency of ꢀHads,i.
Fig. 8. Comparison between experimental and predicted N O con-
2
version curves vs temperature from temperature-programmed experi-
ments performed under stoichiometric conditions on Rh/Al O (2) and
2
3
Pt–Rh/Al O (1); (—) predicted curves.
2
3
versions using Eq. (16). As observed in Fig. 8, the results in
Table 4 lead to a good correlation between experimental and
calculated conversions, except at quasi-complete conversion
when Pi tends toward 0. Under such conditions, the numeri-
cal solutions from Eqs. (21) and (22) diverge, both equations
being invalid when Pi = 0. As seen in Table 4, the optimized
References
[1] J. Cunningham, J.N. Hickey, R. Cataluna, J.C. Conesa, J. Soria, A.
Martinez-Arias, Stud. Surf. Sci. Catal. 101 (1996) 681.
[2] J.H. Holles, M.A. Switzer, R.J. Davis, J. Catal. 190 (2000) 247.
[3] N.T. Pande, A.T. Bell, J. Catal. 98 (1986) 7.
[4] P. Granger, J.-F. Lamonier, N. Sergent, A. Aboukais, L. Leclercq, G.
Leclercq, Top. Catal. 16/17 (2001) 89.
values for aCO and aN O are still in agreement with previ-
2
ous trends, observed in Table 3, demonstrating that N2O still
adsorbs more strongly on Pt/Al2O3 than on Rh/Al2O3 near
saturation. Finally, both results show that a rate expression
derived from a selected mechanism, accounting for adsor-
bate coverage dependencies of the adsorption enthalpies, can
be profitably used for modeling light-off curves on three-
way catalysts.
[
5] P. Fornasiero, N. Hickey, J. Kaspar, T. Montini, M. Graziani, J. Ca-
tal. 189 (2000) 339.
[6] R. Di Monte, P. Fornasiero, J. Kaspar, P. Rumori, G. Gubitosa, M.
Graziani, Appl. Catal. B 24 (2000) 157.
[
[
7] D. Ciuparu, A. Bensalem, L. Pfefferle, Appl. Catal. B 26 (2000) 241.
8] P. Granger, L. Delannoy, J.-J. Lecomte, C. Dathy, H. Praliaud, L.
Leclercq, G. Leclercq, J. Catal. 207 (2002) 202.
[9] B.K. Cho, B.H. Shanks, J.E. Bailey, J. Catal. 115 (1989) 486.
[
[
[
10] B.K. Cho, J. Catal. 138 (1992) 255.
11] B.K. Cho, J. Catal. 148 (1994) 697.
12] P. Granger, J.-J. Lecomte, C. Dathy, L. Leclercq, G. Mabilon, M. Pri-
gent, G. Leclercq, J. Catal. 173 (1998) 304.
5
. Conclusion
[
13] P. Granger, J.-J. Lecomte, C. Dathy, L. Leclercq, G. Leclercq,
J. Catal. 175 (1998) 194.
The kinetics of the CO + N2O reaction has been in-
vestigated on Rh/Al2O3 and Pt–Rh/Al2O3. The mechanism
selected involves a dissociation of chemisorbed N2O mole-
[14] P. Granger, P. Malfoy, P. Esteves, L. Leclercq, G. Leclercq, J. Ca-
tal. 187 (1999) 321.

P. Granger et al. / Journal of Catalysis 223 (2004) 142–151
151
[
[
[
[
15] D.N. Belton, S.J. Schmieg, J. Catal. 138 (1992) 70.
16] R.W. Mac Cabe, C. Wong, J. Catal. 121 (1990) 422.
17] R.R. Sadhankar, J. Ye, D.T. Lynch, J. Catal. 146 (1994) 511.
18] S. Kacimi, D. Duprez, in: A. Crucq (Ed.), in: Catalysis and Automotive
Pollution Control II, vol. 71, Elsevier, Amsterdam, 1991, p. 581.
19] P. Granger, L. Delannoy, L. Leclercq, G. Leclercq, J. Catal. 177 (1998)
[25] D. Loffreda, D. Simon, P. Sautet, J. Catal. 213 (2003) 211.
[26] P. Malfoy, P. Granger, J.-F. Lamonier, L. Leclercq, G. Leclercq, Phys.
Chem. 41 (1997) 109.
[27] H.H. Huang, C.S. Seet, Z. Zou, G.Q. Xu, Surf. Sci. 356 (1996) 181.
[28] Y. Li, M. Bowker, Surf. Sci. 348 (1996) 67.
[29] V.P. Zhdanov, J. Catal. 162 (1996) 47.
[30] A.A. Diamantis, G.J. Sparrow, J. Colloid Interface Sci. 47 (1974) 455.
[31] P. Granger, P. Malfoy, P. Esteves, J.-F. Paul, L. Leclercq, G. Leclercq,
Europacat IV, Book of Abstracts, Rimini, 1999.
[32] R.F. van Slooten, B.E. Nieuwenhuys, J. Catal. 100 (1986) 360.
[33] Y.J. Mergler, B.E. Nieuwenhuys, J. Catal. 161 (1996) 292.
[34] F. Solymosi, M. Pasztor, J. Phys. Chem. 89 (1985) 4789.
[35] P. Granger, J.-J. Lecomte, L. Leclercq, G. Leclercq, Appl. Catal. A 208
(2001) 369.
[
[
[
147.
20] P. Trambouze, H. Van Landeghem, J.-P. Wauquier, Les réacteurs chim-
iques, conception, calcul, mise en oeuvre, Technip, Paris, 1984.
21] J.-F. Lepage, J. Cosyns, P. Courty, E. Freund, J.-P. Franck, Y. Jacquin,
B. Juquin, C. Marcilly, G. Martino, J. Miquel, R. Montarnal, A. Sugier,
H. Van Landeghem, Catalyse de contact, Technip, Paris, 1978.
[22] R.J. Gorte, L.D. Schmidt, J.L. Gland, Surf. Sci. 109 (1981) 367.
[23] D.T. Wickham, B.A. Banse, B.E. Koel, Surf. Sci. 243 (1991) 83.
[24] S.W. Jorgensen, N.D.S. Canning, R.J. Madix, Surf. Sci. 179 (1987)
[
[
36] E.G. Seebauer, A.C.F. Kong, L.D. Schmidt, Surf. Sci. 176 (1986) 134.
37] G. Ertl, M. Neuman, K.M. Streit, Surf. Sci. 64 (1977) 393.
322.