Kinetic Analysis of the Decomposition of Nitrous Oxide over ZSM-5 Catalysts

Article abstract of DOI:10.1006/jcat.1997.1581

A detailed comparative kinetic analysis has been made of the N₂O decomposition over Co-, Fe-, and Cu-ZSM-5. The effect of partial pressure of N₂O, O₂, CO, and NO, the space time, and temperature have been investigated. The decomposition is first order in N₂O over Fe- and Co-ZSM-5 and has a slightly lower order over the Cu sample. Inhibition of oxygen observed for Cu-ZSM-5 is absent for Co and for Fe at the lower temperatures. N₂O destruction is enhanced by CO for all catalysts and by NO for Fe-ZSM-5 only. In the presence of NO, NO₂ is produced over Fe- and Co-ZSM-5. Over Cu-ZSM-5 the enhancement by CO passes through a maximum as a function of CO pressure due to the strong adsorption at reduced sites. A detailed kinetic model that accounts quantitatively for the observed dependencies and which deviates from the classical model foroxidic systems is advanced. Estimates of the maximum turnover rates for the various model steps range from 10^-4 to 1 s^-1.

Full text of DOI:10.1006/jcat.1997.1581

JOURNAL OF CATALYSIS 167, 256–265 (1997)
ARTICLE NO. CA971581
Kinetic Analysis of the Decomposition of Nitrous Oxide
over ZSM-5 Catalysts
Freek Kapteijn,1 Gregorio Marba´n, Jose´ Rodriguez-Mirasol,† and Jacob A. Moulijn
Industrial Catalysis, Department of Chemical Engineering, Delft University of Technology, Julianalaan 136, 2628 BL Delft, The Netherlands; Instituto
Nacional del Carbon, Oviedo, Spain; and †Department of Chemical Engineering, University of Malaga, Spain
Received September 20, 1996; revised December 16, 1996; accepted December 16, 1996
with Fe-zeolites (5, 7, 8, 10–13), but also Co and Cu sys-
tems exhibit high activities (6, 10, 11). ZSM-5 catalysts are
quite active (3) and especially Co-ZSM-5 is most attractive
in view of its thermostability (14), while Fe- and Cu-ZSM-5
are shown to be deactivated under hydrothermal conditions
(15). Detailed kinetic studies, needed for practical applica-
tion, have hardly been reported (3, 16), and even less is
known of the the influence of other components that may
be present, like O2, CO (16), H2O, NO, and SO2 (11). For
Fe-zeolites mainly a first order in N2O and a zero order
in O2 is reported (7, 8, 12), although also a positive influ-
ence of O2 has been found (13). Mechanistic studies mainly
concern Fe systems (7, 8, 10, 12).
Classically the reaction over oxidic catalysts is described
by adsorption followed by an oxidation of active sites,
and a subsequent removal of the deposited oxygen by
recombination (Eqs. [2]–[4]). The adsorption (Eq. [2]),
and desorption (Eq. [4]), are generally assumed to be in
quasi-equilibrium under decomposition conditions:
Adetailed comparativekineticanalysishasbeen madeoftheN2O
decomposition over Co-, Fe-, and Cu-ZSM-5. The effect of partial
pressure of N2O, O2, CO, and NO, the space time, and temperature
have been investigated. The decomposition is first order in N2O
over Fe- and Co-ZSM-5 and has a slightly lower order over the
Cu sample. Inhibition of oxygen observed for Cu-ZSM-5 is absent
for Co and for Fe at the lower temperatures. N2O destruction is
enhanced by CO for all catalysts and by NO for Fe-ZSM-5 only. In
the presence of NO, NO2 is produced over Fe- and Co-ZSM-5. Over
Cu-ZSM-5 the enhancement by CO passes through a maximum as
a function of CO pressure due to the strong adsorption at reduced
sites. A detailed kinetic model that accounts quantitatively for the
observed dependencies and which deviates from the classical model
for oxidic systems is advanced. Estimates of the maximum turnover
1
rates for the various model steps range from 10 4 to 1 s
.
1997
c
Academic Press
1. INTRODUCTION
Nitrous oxide has received increasing attention the past
decade, due to the growing awareness of its impact on the
environment, as it has been identified as an ozone deple-
tion agent and as a Greenhouse gas (1). Identified major
sources include adipic acid production, nitric acid, and fer-
tilizer plants, fossil fuel and biomass combustion, and de-
NOx treatment techniques, like three-way catalysis and se-
lective catalytic reduction (2, 3).
For the abatement of N2O emissions one can note much
interest in the development of catalysts that decompose
nitrous oxide into its elements at rates and conditions that
are compatible with the production sources (3, 4):
KN2
O
N2O+
↔
N2O
[2]
[3]
[4]
k
N2O → N2 + O
2 O ↔O2 + 2 .
KO2
For steady-state conditions and assuming a constant
number of active sites (17) this yields the rate expression
[5], where NT represents the active site concentration
(mol/gcat) and rN O the conversion rate of N2O (mol/s
2
gcat). Depending on the values of the parameters in this
expression it can be reduced to simpler forms (3). For weak
N2O adsorption and fast surface oxidation [2] and [3] are
often combined to one step:
2 N2O → 2 N2 + O2 (1r H (298) = 163 kJ/mol). [1]
Catalysts include oxides, mixed oxides (perovskites), and
zeolites (3). The latter, transition metal ion-exchanged ze-
olites, have been shown to exhibit high activities for the de-
composition reaction (5–11). Most published studies deal
kNT K
p
N2O N2O
p
rN O
=
.
[5]
2
1 + KN O pN O
+
KO pO
2 2
2
2
1
Hall et al. (7, 8) found for Fe-Y and Fe-Mor an absence
To whom correspondence should be addressed. Fax: +31 15 278 4452.
E-mail: f.kapteijn@stm.tudelft.nl.
of oxygen inhibition and a pure first order behavior for
256
0021-9517/97 $25.00
c
Copyright
1997 by Academic Press
All rights of reproduction in any form reserved.

DECOMPOSITION OF N2O OVER ZSM-5
257
nitrous oxide, similar to that of Fe-ZSM-5 (12). This can be
Recently, we compared qualitatively the behavior of
accounted for only if N2O adsorption is rate limiting, but three zeolitic catalysts in the N2O decomposition, viz.
then the catalytic sites must be in a reduced state, while it Co-, Cu-, and Fe-ZSM-5, and found striking differences
could be observed that the catalyst contained a lot of oxy- (11) with respect to activity, inhibition, and pressure de-
gen. Therefore they proposed [7] as the irreversible oxygen pendency. Here, we focus on the quantitative interpretation
removal reaction, in combination with the oxidation step of of the data to give a detailed microkinetic model for these
Eq. [6], resultingin Eq. [8]for the totalN2O conversion rate: catalysts.
k1
2. EXPERIMENTAL
N2O + → N2 + O
[6]
[7]
2.1. Catalysts
k2
N2O + O → N2 + O2
+
Cu-, Fe-, and Co-exchanged ZSM-5 zeolites have been
used as catalysts for the N2O decomposition. ZSM-5 ze-
olite (SiO2/Al2O3 = 37.2) in the sodium form (ZEOCAT
PZ-2/40 Na; Chemie Uetikon) was ion-exchanged, under
vigorous stirring, using aqueous solutions (pH between 5.5
and 6) of Cu(II) acetate (4.0 mM at 293 K), Fe(II) sulfate
(3.7 mM at 343 K), or Co(II) acetate (39.7 mM at 323 K).
The zeolites were then filtered and washed thoroughly
with deionized water at room temperature before drying at
383 K overnight. The metal content was determined by
ICP-AES and AAS. The exchange levels were calculated
on the basis of the amount Na+ disappeared and the amount
2k1k2 NT
k1 + k2
2k1 NT
rN O
=
pN O
=
pN O
.
[8]
2
2
2
k1/k2 + 1
The ratio k1/k2 in Eq. [8] equals [O ]/[ ] and so determines
the state of the active sites. For k1/k2 1 the difficult step
is Eq. [7] and the sites are oxidized, while for k1/k2 1 the
reverse holds.
The deposited oxygen, often denoted as extralattice oxy-
gen (ELO), has special properties. It catalyzes the O2/18O2
oxygen exchange reaction over Fe-ZSM-5 at room temper-
ature (12).
From N218O studies over Fe-Mor it appeared that this
oxygen was exchanged with lattice oxygen and with N2O
(Eq. [9]) (10). The latter can be interpreted as a nonpro-
ductive step compared to that of Eq. [7], and corroborates
the possibility of this latter step:
of transition metal TM introduced as Cu2+, Fe3+, or Co2+
The data are given in Table 1.
.
2.2. Experimental Setup and Procedures
The experimental setup for N2O decomposition con-
sisted of a gas mixing section, a reactor and a gas analysis
section. A quartz fixed bed reactor of 5 mm i.d. was used,
containing 20–100 mg of catalyst (106–212 m) diluted with
180 mg of SiC (106–212 m), to assure plug flow, and oper-
ated at a total pressure of 2.5 bar.
The SiC diluent did not contribute to the N2O decompo-
sition at the reaction temperatures applied. Prior to each
run, the catalyst was subjected to heating in He at 30 K/min
to 923 K and maintaining this temperature for 1 h. Subse-
quently, the temperature was decreased to the desired value
and the feed mixture was passed over the bed. Generally, 40
to 50 min after a change of conditions the conversion levels
were constant and considered as the steady-state values. At
least five analyses were averaged for a data point.
N218O + O ↔ N2O + 18O .
[9]
For each ELO two Fe ions are involved (18, 19), but with
N2O about 1 oxygen is deposited per 4–5 Fe ions (12). On
the other hand, more than eight times the maximum ELO
capacity could be exchanged with lattice oxygen (10), indi-
cating that this deposited oxygen readily loses its identity.
The isolated nature of the ions in the zeolite framework
resembles the dilute solid solution oxide systems, exten-
sively studied two decades ago for the N2O decomposition
(20, 21). The mechanistic proposals made for these systems
(20) may apply to zeolites, too (3). In essence, the oxy-
gen moves from the transition metal (TM) ion to a matrix
oxygen, becoming a peroxy-oxygen, and either reacts with
another such oxygen to O2, or with a newly incoming oxy-
gen on the TM ion. For zeolites Hall et al. called the TM
ion a “porthole.” The identity of the oxygen may be lost
by exchange with the framework oxygen. If the peroxy-
oxygen is not mobile the reaction is limited to the direct
vicinity of the TM ion, which is suggested by the value of
four to five exchangeable lattice oxygens per Fe ion. A simi-
lar exchange of lattice oxygen with N18O was observed over
Cu-ZSM-5 (10, 22). Recently, isotopic exchange reactions
were reported between adsorbed 15NO2 and NO over Co-
and Cu-ZSM-5, which indicate a rapid exchange of the ni-
trogen (23).
The product gases were continuously analyzed for
NO and NO2 using a chemiluminescence analyzer, and
TABLE 1
Catalysts Used
Metal loading
(wt% )
Metal exchange
level (% )
Sample
Cu-ZSM-5
Co-ZSM-5
Fe-ZSM-5
4.0
1.6
1.3
130
73
98

258
KAPTEIJN ET AL.
discontinuously for N2O, N2, CO, CO2, and O2 by GC
equipped with a thermal conductivity detector and an elec-
tron capture detector, specifically for the N2O analysis, us-
ing a Poraplot Q column and a molsieve 5A column for
separation.
Conditions. The influence of temperature, partial N2O
and O2 pressure, space time W/FN O , and gases like CO and
2
NO on the decomposition of N2O over the catalysts were
studied. The temperatures varied between 625 and 873 K.
The inlet partial N2O pressure ranged from 0.5 to 2 mbar,
the O2 pressure from 0 to 100 mbar, and the space time from
1.5 105 to 11.0 105 g s/mol. The total gas flow rates were
between 1 and 5 ml (STP)/s. NO or CO was added in molar
ratios of 0–2 with N2O (kept at 1 mbar). The low partial
pressures were obtained by using gas mixtures of 5% in
He and further dilution with helium by means of mass flow
controllers.
FIG. 1. Conversion as a function of temperature at 1 mbar N2O and
space time 1.44 105 g s/mol for Co-, Cu-, and Fe-ZSM-5.
Parameter estimation. Integral reactor behavior was
used for the interpretation of the experimental data. To
this purpose the reactor continuity Eq. [10] was integrated
numerically using the Bulirsch–Stoer method (24) to calcu-
The presence of O2 hardly affects the reaction over Fe-
(at lower temperatures) and Co-ZSM-5, but it inhibits for
the Cu system and for Fe at higher temperatures (Fig. 3).
The apparent Ea for Cu increased by nearly 40 kJ/mol.
Addition of CO enhances the N2O conversion, by about a
factor of two for Co and tremendously for Fe (Fig. 4). In the
latter case N2O is being converted at temperatures where
in absence of CO hardly any conversion is being observed.
For Cu a maximum in the N2O conversion appears as a
function of the CO/N2O ratio in the feed. This maximum
shifts to higher values with increasing temperature (Fig. 5).
The apparent activation energy for Co is hardly altered;
for Fe it decreased nearly 100 kJ/mol, while for Cu it is
increased by 50 kJ/mol.
late the exit conversion of nitrous oxide. In Eq. [10] rN O
2
represents the expression for the total N2O conversion rate,
as will be derived under Discussion:
dx
N2O
= rN O
.
[10]
2
W
d
FN02
O
The apparent rate parameters were estimated by nonlin-
ear least-squares methods (Simplex (25) and Levenberg-
Marquardt (26, 27)), minimizing the sum of squares of
the residual (=observed–calculated) N2O conversion. The
temperature dependency of the rate parameters was ex-
pressed in the Arrhenius form. The confidence limits of the
parameter estimates were calculated from their covariance
matrix at the 95% confidence level (28, 29). Transport lim-
itations could be neglected (30).
The product distribution over Fe-ZSM-5 (Fig. 6a) clearly
shows the 1 : 1 stoichiometry for the reaction between
N2O and CO (Eq. [11]). Over Cu-ZSM-5 at low CO
3. RESULTS
An impression of the activity of the different catalysts is
given in Fig. 1. The activity order Cu > Co > Fe corresponds
with literature (6). The N2O pressure dependency for Co-
ZSM-5isgiven in Fig. 2. Due to the integralreactor behavior
the relation between conversion and partial pressure shows
a curvature, but the reaction order equals 1 for Co and Fe
below 733 K, while lower values are found for Cu and for
Fe at higher temperatures. The fitting results of apparent
activation energies for the different experiments, assuming
a first order behavior, are given in Table 2. Included in this
table are also the apparent reaction orders for a combined
fit of the whole data set for an assumed nth order behavior.
Full results are available through the authors on request.
FIG. 2. N2O conversion as a function of N2O pressure for Co-ZSM-5
at various temperatures and a space time of 1.52 105 g s/mol. Drawn
lines are the model fits.

DECOMPOSITION OF N2O OVER ZSM-5
259
TABLE 2
Apparent Activation Energies (kJ/mol) and Reaction Ordersa
All data
O2/N2O CO/N2O NO/N2O
Only N2O
104
136 38
168 22
= 30
= 2
= 1.5
Ea
Order n
Co
Cu
Fe
7
110
175
187
122
179
78
134
138
—
106 15 1.00 0.07
138 17 0.88 0.11
182 31 0.79 0.15
a
95% Confidence limits; 1st order assumed, except for the combined
data set.
concentrations still some oxygen is observed (Fig. 6b), so
both reactions of Eqs. [1] and [11] can occur simultaneously,
depending on the conditions:
FIG. 4. The effect of CO on the N2O conversion at 1 mbar N2O and
space time W/FN = 1.52 105 g s/mol. Drawn curves are model fits.
O
2
N2O + CO → N2 + CO2.
[11]
Also addition of NO enhances the N2O conversion over
Fe-ZSM-5 tremendously and does not affect that for the
other two catalysts (Fig. 7). It is, however, noted that NO
is converted to NO2 not only over Fe-ZSM-5, but also over
Co-ZSM-5. The product composition for these two cata-
lysts is given in Figs. 8a and 8b. Again it is noted that over
Fe-ZSM-5 the conversion of N2O occurs at temperatures
where the decomposition does not noticeably take place
without NO. NO is converted to NO2 according to Eq. [12],
while still O2 formation takes place:
which can be rationalized by the two-step kinetic model
given by Eqs. [6] and [7]. Their different behavior becomes
especially apparent in their sensitivity to the presence of
oxygen, of a reducing agent like CO, and even of NO. The
enhancement by reducing agents can be expected only if
they remove the oxygen deposited by the N2O faster than
occurs in the pure decomposition reaction. This competi-
tion must become apparent in the kinetic modeling dis-
cussed below. The results of this modeling are given in
Table 3 (parameter estimates) and Figs. 9a–9c (predicted
versus observed N2O conversions for all experiments). As-
suming that all TM ions in the zeolite samples are active
centers, so their concentration represents NT, then from
the rate parameters in Table 3 turnover rates of the various
reaction steps in the kinetic models can be estimated for
the hypothetical case that all sites participate in that step.
These values are given in Table 4 for 1 mbar reactant pres-
sures, where applicable, at the temperatures indicated. The
N2O + NO → N2 + NO2
[12]
4. KINETIC MODELING AND DISCUSSION
The Cu-, Co-, and Fe-ZSM-5 catalysts are all active sys-
tems for the decomposition of N2O, but their behavior dif-
ferswith respect to conditionsand gasatmospheres. Theyall
seem to obey a (nearly) first order dependency toward pN O
,
2
FIG. 5. Effect of CO on the N2O conversion over Cu-ZSM-5 at 1 mbar
N2O and space time W/FN = 1.52 105 g s/mol. Drawn curvesare model
FIG. 3. The effect of oxygen on the N2O conversion at 1 mbar N2O
and different temperatures. Space time W/FN = 2.87 105 g s/mol.
O
2
fits.
O
2

260
KAPTEIJN ET AL.
FIG. 6. Product composition for the CO/N2O feed mixture at conditions of Figs. 4 over Fe-ZSM-5 (a) and Cu-ZSM-5 (b).
lowest values for a catalyst approach the turnover frequen- like CO would not have any effect. Apparently it does have
cies of the overall reaction for only N2O. The other values an influence, an argument in favor of the two-step model,
represent in principle the maximum turnover frequency for formed by Eqs. [6] and [7].
that step if all sites would participate in that step. In real-
The apparent rate constant contains the concentration of
ity the surface concentrations during steady-state operation active sites, NT (mol/g), and the two rate constants k1 and
will adjust in such a way that all steps turn over at rates in k2 of Eq. [8]. Assuming that all Co ions are active sites this
the catalytic cycle compatible with the overal reaction.
results in turnover frequencies of 10 3–10 2 s 1 (Table 4).
A further determination of the individual values of the two
rate constant was not possible on the basis of the data with
N2O alone. The experiment with CO addition, carried out
at 693 K and two space times, gave additional information.
Co-ZSM-5
The experimental data of Co-ZSM-5 could be excellently
described by the first order rate expression [8], with the
parameter values given in Table 3. The observed conversion
data are well predicted by this model (Fig. 9a). A first order
rate expression could also be derived on the basis of the
classical model if N2O adsorption were the rate determining
process. This would imply, however, that the surface oxygen
occupancy is nearly zero and addition of a reducing agent
TABLE 3
Estimated Model Parameters and Their 95% Confidence Intervalsa
Co-ZSM-5
All data
Data at 693 K
(includ. CO)
ln(k0NT) = 11.0 1.5
2k1NT = (0.17 0.03) 10
k1/k2 = 4.5 1.7
Ea = 104
7
2
1
1
mol s 1 bar gcat
—
—
k4/k2 > 40
Cu-ZSM-5
All CO data
ln(k02NT) = 17.7 0.2
Ea2 = 136b
Ea4 = 0b
ln(k04NT) = 3.96 0.35
ln(K05 k04/k01) = 19.0 3.0 1H5 + Ea4 Ea1
=
170 16
3
1
1
1
All data at 673 K 2k2NT = (2.04 0.08) 10
k4NT = 0.0217 0.0038
K5k4/k1 = 1935 420
k2/k3 = 413 39
mol s 1 bar gcat
mol s 1 bar gcat
bar
1
1
1
bar
1
k2/k1K3 = 363 71
bar
Fe-ZSM-5
N2O data vs T
CO data
(623–713 K )
NO data at 673 K k7NT = 0.098
k7/k8 = 14
ln(k02NT) = 20.4 3.5
ln(k04NT) = 6.55 1.10
ln(k04/k01) = 1.16 0.08
Ea2 = 168 22
Ea4 = 64
6
Ea4 Ea1
=
1.8 2.2
mol s 1 bar gcat
1
1
—
a
1
FIG. 7. Effect of NO on the N2O conversion at 1 mbar N2O and space
time W/FN = 1.52 105 g s/mol. Included is the NO conversion (dashed
Activation energies Eai in kJ/mol, k0iNT in mol s 1 bar gcat 1, K5 in
bar 1, K3 in bar.
b Fixed parameter value.
O
2
lines and open symbols).

DECOMPOSITION OF N2O OVER ZSM-5
261
FIG. 8. Product composition for a NO/N2O feed mixture at the conditions of Fig. 7 over Fe-ZSM-5 as a function of pNO (a) and over Co-ZSM-5
as a function of temperature at 1 mbar NO (b).
FIG. 9. Calculated model N2O conversion as a function of the observed N2O conversion for the data sets of Co-ZSM-5 (a), Fe-ZSM-5 (b), and
Cu-ZSM-5 (c).

262
KAPTEIJN ET AL.
order behavior and oxygen inhibition appears. The latter is
TABLE 4
in contrast with the results over oxidic catalysts, where inhi-
bition decreases with temperature, for the obvious thermo-
dynamic reason that adsorption is an exothermal process
and less favored at higher temperatures. This is a support
for the two-step model (Eqs. [6] and [7]) at lower temper-
atures. The increased sensitivity toward molecular oxygen
is ascribed to the fact that at these temperature levels this
catalyst starts to exhibit exchange reactions of molecular
oxygen (12), indicating a dissociation of O2. This implies an
additional pathway, the reverse of Eq. [4], through which
oxygen can be deposited on the active sites and which com-
peteswith the oxidation byN2O. Hence, a graduallyincreas-
ing inhibiting effect may be expected. The kinetic results
above 730 K contained too little information to take this
effect into account. Only the data below 730 K are used in
the present kinetic analysis.
Maximum Turnover Frequencies, TOF(max) (s 1), Estimated
for the Various Reaction Steps at 1 mbar Reactant Pressure
(See Text)
1
TOF(max) (s
)
From rate Co-ZSM-5 Fe-ZSM-5 Cu-ZSM-5
Reaction step
constant
(693 K)
(673 K)
(673 K)
3
3
2
4
Reaction with N2O,
Eq. [6]
Reaction O with N2O,
Eqs. [7], [21]
Desorption of O2,
Eq. [22]
O-removal by CO,
Eq. [13]
k1NT
6.1 10
1.7 10
3
3
2
k2NT
k3NT
k4NT
k7NT
1.4 10
2.8 10
1.6 10
3.9 10
3.4 10
2
> 0.25
3.2 10
0.4
Reaction O with NO,
Eq. [17]
The data for N2O alone can be considered as repre-
sentative for the activation energy of the pure reaction
(168 kJ/mol). The tremendous increase of the N2O conver-
sion in the presence ofCO and NO indicatesthat the oxygen
removal is the difficult step in the reaction and k1/k2 1,
so in principle the rate of Eq. [7] is being measured in pure
N2O, as given by Eq. [15]. TOFs reported in literature range
from 10 3 to 10 2 s 1 at 723 K (12), corresponding well with
our data:
Apparently CO removes oxygen from the oxidized centers
in competition with N2O (Eq. [7]):
k4
CO + O → CO2 + .
[13]
The resulting rate expression for the N2O removal is now
given by
(
)
rN O = 2k2 NT pN O
.
[15]
2
2
2 + (k4/k2) (pCO /pN O
)
2
rN O = k1 NT pN O
.
2
2
For the results with CO the same kinetic approach holds as
for Co-ZSM-5. Since now k2 k1, k4 the simplified expres-
sion that applies here is given by Eq. [16]. Estimated val-
ues for these parameters (623–713 K) are given in Table 3.
Figure 9b shows that all N2O conversions are well predicted
bythe modeling. The rate constantsk4 and k1 are ofthe same
magnitude and have about equal activation energies. The
low value of the former (64 kJ/mol) explains the decrease
in temperature dependency as indicated in Table 2 by the
apparent activation energies. The results (Tables 3 and 4)
indicate that the rate constant k2 is two orders of magnitude
smaller than k4 and k1.
1 + k1/k2 + (k4/k2) (pCO /pN O
)
2
[14]
For vanishing CO concentrations this reduces to Eq. [8],
so the pCO/pN O terms in Eq. [14] account for the effect
2
of CO in the N2O destruction. Expression [14] contains
three parameters that may be determined, k1NT, and the
ratios k1/k2 and k4/k2. The parameter estimation yielded
clear values for the first two, while for the latter ratio a
lower limit was estimated. These results indicate that the
rate constant of the oxidation step (Eq. [6]) is about 4.5
times larger than that of the removal by N2O (Eq. [7]),
whereas that for the removal by CO is about 10 times larger
than for the oxidation step. For the reaction in only N2O this
means that about 80% of the active sites are in the oxidized
state during decomposition, since the ratio k1/k2 represents
the ratio between oxidized and empty sites, [O ]/[ ]. The
temperature dependency of the two rate constants did not
differ that much to be able to determine their individual
activation energies.
k4 NT pCO
rN O
=
.
[16]
2
1 + (k4/k1) (pCO /pN O
)
2
–
Petunchiand Hallreported data on the CO N2O reaction
over Fe-Mor and Fe-Y (16). They did not follow the kinetic
approach presented here, but mentioned that the rate is
dependent on the CO/N2O ratio for Fe-Mor, with an acti-
vation energy of 76 kJ/mol, in agreement with our results
(Table 2). The TOF for this catalyst at 673 K, 2.4 s 1, is about
10 times higher than in our study. Fe-Y, being 500 times less
Fe-ZSM-5
The behavior of this catalyst is peculiar. At lower tem- active, exhibited a first order behavior in N2O, which can
peratures (<730 K) it exhibits first order behavior and be interpreted as k1 k4 and the rate of the oxidation step
no inhibition by oxygen, in agreement with earlier results of Eq. [6] is observed. These data indicate that for Fe the
(7, 8, 12). At higher temperaturesthe rate deviatesfrom first activity order for the zeolite matrices is Mor > ZSM-5 > Y.

DECOMPOSITION OF N2O OVER ZSM-5
263
This is not a general trend, because each TM ion has its own essential remains to be answered, although it is a tempting
optimal zeolite matrix (3). explanation for the inhibition. Our Cu samples do have
For the reaction with NO a similar approach can be some overexchange (Table 1).
followed as for CO, although it may be doubted whether
Assumingthat for Cu the “classical” kineticmodelshould
the model should be modified or not. Both NO and NO2 apply (Eqs. [2]–[4]), the rate parameter values indicate that
may adsorb at the active sites, as has been demonstrated the catalyst is not in a highly oxidized state at lower partial
by in situ DRIFTS experiments (11). Reduced Fe sites do pressures and conversion levels. DRIFTS characterization
not exist under these conditions as they are directly oxi- data, however, show that the catalyst is in a highly oxidized
dized (11). The product NO2 may be assumed to adsorb state in the presence of N2O (11). This suggests that des-
more strongly on the sites and it could be envisaged that orption is a difficult step in the process, supported by the
desorption of NO2 is enabled by the direct oxidation by enhancement effect of CO.
N2O, as has been observed for the desorption of CO2 from
An alternative kinetic model is proposed, formed by
oxidic catalysts effected by O2 in the CO oxidation (31). Eqs. [6], [21], and [22], in analogy to and as an extension of
The model becomes then the set of Eqs. [17] and [18]. Re- the two-step model for Co and Fe. The last step is reversible,
cently, an isotopic exchange study has been conducted be- but not assumed to be in quasi-equilibrium. The last step
tween 15NO2 and NO over Co-ZSM-5 and Cu-ZSM-5. Over can be proposed on the basis of thermal desorption exper-
Co-ZSM-5, which behaves similarly as Fe-ZSM-5 in the iments of oxygen from different Cu-zeolites (22, 33, 37). In
presence of NO, a rapid exchange between gaseous NO and the NO decomposition it turned out that oxygen started to
adsorbed NO2 was observed (23), which indicates an impor- desorb only above 573 K (38), while in TPD experiments
tant role for adsorbed NO2, like in the reactions of Eqs. [17] the peak maximum for O2 desorption occurred around 673
and [18]:
K. This indicates that oxygen desorption is a difficult step in
the process. The reversibility of this step is suggested by the
observed O2 inhibition here and for the NO decomposition
(34, 37):
k7
NO + O → NO2
[17]
[18]
k8
N2O + NO2 → N2 + NO2 + O
k1
N2O + → N2 + O
[6]
[21]
[22]
k7 NT pNO
rN O
=
.
[19]
2
k2
1 + (k7/k8) (pNO /pN O
)
2
N2O + O → N2 + O2
k3
The data do not allow an accurate estimation of the pa-
rameters in Eq. [19]; their values are highly correlated, but
as an indication those for 673 K are included in Table 3, im-
O2 → O2 + .
k
3
A suggestion for molecular O2 adsorption follows from
oxygen exchange experiments over Cu-ZSM-5. Cu exhib-
ited most pronounced a single step double exchange reac-
tion, i.e. 18O2 was exchanged directly for O2, without for-
mation of singly labeled molecules, at the temperatures of
interest here (22, 39). The model rate expression is given
by Eq. [23] and includes the simplification that the concen-
tration of empty sites is negligible compared to that of *O
and *O2
1
plying an even higher TOF(max) than with CO, 0.1–1 s
(Table 4). Thermodynamically the reaction of Eq. [12] al-
lows complete conversion to NO2 (3). This in contrast with
the reaction between NO and O2 which is equilibrium lim-
ited. The observed NO2 levels are much above those ac-
cording to this latter reaction, which evidences the reaction
path of Eq. [12]. The observed formation of O2 (Fig. 8) is
expected on thermodynamic grounds. At these tempera-
tures N2O itself still does not yield O2, and it is ascribed to
the interaction of NO2 with an oxidized site, schematically
represented by Eq. [20]:
2k2 NT p
N2O
rN O
=
.
[23]
2
1 + (k2/k3)pN O + (k2/k1 K3)pO
2
2
k9
NO2 + O → NO + O2 + .
[20]
Comparison of this rate expression with Eq. [5] indicates
the similar form, but the interpretation of the parameters is
different. The new model allows that the parameters in the
denominator may increase with increasing temperature, in
contrast to pure adsorption constants that thermodynami-
cally only are allowed to decrease.
The experimental data did not allow a statistical distinc-
tion between a molecular or a dissociative oxygen adsorp-
tion model. The latter was found for NO decomposition (34,
37). Table 3 indicates values for the parameters in Eq. [23]
Cu-ZSM-5
The observed N2O conversion data, apparent reaction
order <1 and oxygen inhibition, can be well described by a
rate expression of the form of Eq. [5], with either molecular
(32, 33) or dissociative adsorption of oxygen. The latter
is suggested by the NO dissociation properties especially
for overexchanged Cu samples (e.g., Refs. 34, 35). Whether
the presence of pairs or clusters of copper ions (32, 36) is

264
KAPTEIJN ET AL.
for a temperature of 673 K at which most of the experi- dependency of the last term in the denominator of Eq. [26]
ments had been conducted. These include the partial pres- corresponds to 1H5 + Ea4 Ea1 168 kJ/mol. Since the
sure variation of N2O and O2 and the addition of CO. TOFs adsorption enthalpy 1H5 is negative, with a value of 40 to
=
reported in literature for the O2 desorption (37, 39) corre-
60 kJ/mol estimated from infrared studies (11), and Ea4
is negligible, the value Ea1 for the first reaction step will
1
spond to the values given in Table 4, 10 3–10 2 s
.
For the enhancement by CO the model was extended amount to about 110–130 kJ/mol. It indicates that this is
with Eq. [13], the removal of oxygen, and Eq. [24], which the easier step in the decomposition.
represents the adsorption of CO at reduced sites and as-
Over Cu-ZSM-5 steady-state NO2 formation in the pres-
sumed to be in quasi-equilibrium. DRIFTS experiments ence of NO is not observed. Speculation offers two expla-
have demonstrated the strong reversible adsorption of CO nations; either it is not formed or it decomposes back to NO.
at Cu1+ sites at the applied conditions (11):
IR measurements show by the presence of a 2134 cm 1 band
that NO2 can be formed on Cu-ZSM-5 (34, 40), but from
TPD experiments it appears that it decomposes rapidly in
the range of 600–700 K (38). On the basis of isotopic ex-
change measurements nitrate structures are even proposed
(23). So, in our case this could mean that NO2 reacts with an
oxidized site to NO and O2 much more efficiently than over
Fe-ZSM-5 or Co-ZSM-5, acting as an oxygen carrier, and
hence competes with N2O. Apparently, this does, however,
result neither in an acceleration of the N2O decomposi-
tion nor in a significantlychangingtemperature dependency
(Table 2).
K5
CO +
↔
CO.
[24]
The complete rate expression for this model is given by
rN O
=
2
2k2 NT p
+ k4 NT pCO
N2O
.
pCO
k4
k2
k3
k2
k2
k1
1 + pN O
+
pO + (1 + K5 pCO
)
+
2
2
k1 K3
k1 pN2
O
[25]
From the CO experiments it became clear, like in the case
for Fe-ZSM-5, that k2 k1 and that 1 K5 pCO, leaving a
simplified expression for parameter estimation (Table 3):
Evaluating the results presented above, a detailed kinetic
picture of the decomposition of N2O over the studied cata-
lysts has been obtained. In the steady state the active sites
in Fe- and Cu-ZSM-5 are nearly fully oxidized, while for Co
80% of the sites are oxidized. From the literature it is well
established that the former catalysts operate in an oxidation
reduction cycle, Fe2+/Fe3+ and Cu+/Cu2+ (18, 19, 36). Co2+
in zeolites is hardly oxidized or reduced, but ESR studies on
–
rN O
=
2
2k2 NT p
+k4 NT pCO
N2O
.
1+(k2/k3)pN O +(k2/k1 K3)pO +(k4 K5/k1) p2 /pN O
2
2
2
CO
[26]
dilute solid solutions of Co in MgO indicate that Co3+
O
formation is possible, rapidly followed by a migration of the
deposited oxygen to lattice oxygen and reduction back to
Co2+ (41). A second deposited oxygen could then directly
form molecular oxygen. The involvement of lattice oxygen
as ELO carriers is receiving increasing suppport from stud-
ies with labeled NO and N2O (10, 22). Our kinetic analysis,
however, cannot yield insight in the detailed mechanism,
but is in good agreement with these ideas.
All the reported apparent activation energy values are
compatible with literature values for Fe-zeolites (5, 7,
12, 13) or dilute solid solutions of Co in MgO (42). The ki-
netic (and IR (11)) results with NO indicate that, like CO,
it can remove the oxygen from the surface of the Co and Fe
catalyst, too, thereby forming NO2. The NO does adsorb
on the Co catalyst, evidenced by IR, yielding a slight in-
crease in apparent activation energy. So, although at 723 K
no enhancement is observed, some is expected at higher
temperatures. Kinetically, the blocking of sites by the ob-
served NO adsorption by DRIFTS (11) on the Fe catalyst
is negligible compared to the enhancement achieved by the
oxygen removal effect.
The squared partialCO pressure term in the denominator
clearly accounts for the maximum in the N2O conversion
as a function of the CO partial pressure (Figs. 4 and 5). It
indicates the need to first remove an adsorbed CO from
a site before it can be oxidized by N2O and reduced by a
second CO in a consecutive reaction.
–
For the temperature dependent behavior of the N2O CO
reaction only the last term in the denominator could be es-
timated, due to the limited data with N2O and O2 pressure
variation at these temperatures. This model can describe all
conversion data well as is apparent from Fig. 9c. Also the
–
shift in the maximum with temperature for the N2O CO
experiments is well described (Fig. 5). The values for the
CO data set are included in Table 3, whereby the activation
energy Ea2 (no CO) was fixed at the value obtained for pure
N2O, and that of Ea4 at zero, since its value statistically did
not deviate from zero. This low value can be rationalized
by envisaging that a CO molecule first adsorbs at an oxi-
dized site before it reacts. The negative adsorption enthalpy
can compensate a low activation energy for the subsequent
step. Even an overcompensation is known in literature giv-
ing rise to negative apparent activation energies for the
cracking of n-alkanes over ZSM-5 (17). The temperature
The kinetic model proposed for Cu-ZSM-5 should be
further evaluated by using lower exchange levels and

DECOMPOSITION OF N2O OVER ZSM-5
265
exploring the adsorption and desorption of molecular oxy- 11. Kapteijn, F., Mul, G., Marban, G., Rodriguez-Mirasol, J., and Moulijn,
J. A., in “11th International Congress on Catalysis—40th Anniver-
sary” (J. W. Hightower and W. N. Delgass, Eds.), Studies in Surface
Science and Catalysis, Vol. 101, p. 641. Elsevier, Amsterdam, 1996.
gen with these samples in more detail. For the sake of com-
pletenessit isemphasized at thispoint that the compatibility
of the sets of elementary reaction steps with the observed
12. Panov, G. I., Sobolev, V. I., and Kharitonov, A. S., J. Mol. Catal. 61, 85
kinetics does not prove but only substantiates the proposed
kinetic models.
The observed NO2 formation offers the potential use of
these catalysts in nitric acid plants off-gas treatment, where
about equal amounts of NO and N2O are present. The pro-
duced NO2 can be reused in the nitric acid process (3). Fur-
(1990).
13. Chang, Y.-F., McCarty, J. G., and Zhang, Y. L., Catal. Lett. 34, 163
(1995).
14. Armor, J. N., and Farris, T. S., Appl. Catal. B Env. 4, L11 (1994).
15. Kharas, K. C. C., Robota, H. J., and Datye, A., in “Environmental
Catalysis” (J. N. Armor, Ed.), ACS Symposium Series, Vol. 552, p. 39.
ACS, Washington, DC, 1994.
thermore, it is obvious that application of these catalysts 16. Petunchi, J. O., and Hall, W. K., J. Catal. 78, 327 (1982).
17. Kapteijn, F., Moulijn, J. A., and Santen, R. A. v., in “Catalysis: An
strongly depends on the composition of the gas that has to
be treated and additional kinetic data on the inhibiting and
deactivating effects of, e.g., H2O and SO2 are required (see,
e.g., Refs. (43–46).
Integrated Approach to Homogeneous, Heterogeneousand Industrial
Catalysis” (J. A. Moulijn, P. W. N. M. v. Leeuwen, and R. A. v. Santen,
Eds.), Studies in Surface Science and Catalysis, Vol. 79, p. 69. Elsevier,
Amsterdam, 1993.
18. Garten, R. L., Delgass, W. N., and Boudart, M., J. Catal. 18, 90
(1970).
19. Dalla Betta, R. A., Garten, R. L., and Boudart, M., J. Catal. 41, 40
(1976).
20. Stone, F. S., J. Solid State Chem. 12, 271 (1975).
21. Cimino, A., La Chimica e l’Industria 56, 27 (1974).
22. Valyon, J., and Hall, W. K., J. Catal. 143, 520 (1993).
CONCLUSIONS
A detailed microkinetic model has been developed that
quantitatively predicts N2O decomposition over Co-, Fe-,
and Cu-ZSM-5 catalysts as a function of partial pressures of
N2O and O2, space time, and temperature and in the pres- 23. Beutel, T., Adelman, B. J., and Sachtler, W. M. H., Appl. Catal. B Env.
9, L1 (1996).
ence of CO and NO. Over Co- and Fe-ZSM-5 the reaction
is first order in N2O pressure and is not inhibited by O2,
while Cu-ZSM-5 suffers from O2 inhibition. In the kinetic
model basically N2O oxidizes an active site and removes it
24. Bulirsch, R., and Stoer, J., Num. Math. 8, 1 (1966).
25. Nelder, J. A., and Mead, R., Comput. J. 7, 308 (1965).
26. Levenberg, K., Q. Appl. Math. 2, 164 (1944).
27. Marquardt, D., SIAM J. Appl. Math. 11, 431 (1963).
in a second step, thereby forming O2. This second step is 28. Froment, G. F., and Hosten, L., in “Catalysis; Science and Technology”
(J. R. Anderson, and M. Boudart, Eds.), Vol. 2, p. 97. Springer, Berlin,
1981.
29. Kapteijn, F., and Moulijn, J. A., in “Handbook of Heterogeneous
the difficult one in all cases and addition of CO or NO en-
hances the conversion. CO is effective for all catalysts, but
it inhibits the reaction over Cu-ZSM-5 at concentrations in
excess of N2O due to strong adsorption. NO also enhances
Catalysis” (G. Ertl, H. Kno¨zinger, and J. Weitkamp, Eds.), Vol. A.
VCH, Weinheim, 1996.
the N2O conversion rate over Fe-ZSM-5 and forms NO2, 30. Kapteijn, F., and Moulijn, J. A., in “Handbook of Heterogeneous
Catalysis” (G. Ertl, H. Kno¨zinger, and J. Weitkamp, Eds.), Vol. A.
VCH, Weinheim, 1996.
31. Dekker, F. H. M., Dekker, M. C., Bliek, A., Kapteijn, F., and Moulijn,
like it does over Co-ZSM-5, but here without a net effect
on the N2O conversion rate.
J. A., Catal. Today 20, 409 (1994).
32. Jacobs, P. A., and Beyer, H. K., J. Phys. Chem. 83, 1174 (1979).
ACKNOWLEDGMENT
33. Iwamoto, M., Maruyama, K., Yamazoe, N., and Seiyama, T., J. Phys.
Chem. 81, 662 (1977).
34. Valyon, J., and Hall, W. K., J. Phys. Chem. 97, 1204 (1993).
These investigations have been supported by the European Union un-
der Contract JOU2-CT92-0229 and JR-M by a Human Capital and Mo-
bility grant. Dr. E. Ito is gratefully acknowledged for the catalyst samples.
35. Iwamoto, M., Yahiro, H., Mizuno, N., Zhang, W.-X., Mine, Y.,
Furukawa, H., and Kagawa, S., J. Phys. Chem. 96, 9360 (1992).
REFERENCES
36. Lei, G. D., Adelman, B. J., Sarkany, J., and Sachtler, W. M. H., Appl.
Catal. B Env. 5, 245 (1995).
1. Crutzen, P. L., J. Geophys. Res. 76, 7311 (1971).
37. Li, Y., and Hall, W. K., J. Catal. 129, 202 (1991).
38. Li, Y., and Armor, J. N., Appl. Catal. 76, L1 (1991).
39. Chang, Y.-F., Somorjai, G. A., and Heinemann, H., J. Catal. 154, 24
(1995).
2. Soete, G. G. d., Rev. Inst. Franc. Petr. 48, 413 (1993).
3. Kapteijn, F., Rodriguez-Mirasol, J., and Moulijn, J. A., Appl. Catal. B
Env. 9, 25 (1996).
4. Centi, G., D’Angelo, S., and Perathoner, S., “Book of Abstracts,”
p. 177. Europacat II, Maastricht, The Netherlands, 1995.
5. Sobolev, V. I., Panov, G. I., Kharitonov, A. S., Romannikov, V. N.,
Volodin, A. M., and Ione, K. G., J. Catal. 139, 435 (1993).
6. Li, Y., and Armor, J. N., Appl. Catal. B Env. 1, L21 (1992).
7. Leglise, J., Petunchi, J. O., and Hall, W. K., J. Catal. 86, 392 (1984).
8. Fu, C. M., Korchak, V. N., and Hall, W. K., J. Catal. 68, 166 (1981).
9. Chang, Y.-F., McCarty, J. G., Wachsman, E. D., and Wong, V. L., Appl.
Catal. B Env. 4, 283 (1994).
40. Hoost, T. E., Laframboise, K. A., and Otto, K., Catal. Lett. 33, 105
(1995).
41. Indovina, V., Cordischi, D., Occhiuzzi, M., and Arieti, A., J. Chem.
Soc. Faraday Trans. 1 75, 2177 (1979).
42. Cimino, A., and Pepe, F., J. Catal. 25, 362 (1972).
43. Dann, T. W., Schulz, K. H., Mann, M., and Collings, M., Appl. Catal.
B Env. 6, 1 (1995).
44. Li, Y., and Armor, J. N., Appl. Catal. B Env. 5, L257 (1995).
45. Mabilon, G., and Durand, D., Catal. Today 17, 285 (1993).
46. Li, Y., Battavio, P. J., and Armor, J. N., J. Catal. 142, 561 (1993).
10. Valyon, J., Millman, W. S., and Hall, W. K., Catal. Lett. 24, 215 (1994).