The kinetics of NOx decomposition and NO reduction by CH4 over La2O3 and Sr/La2O3

Article abstract of DOI:10.1006/jcat.1996.0071

The kinetics of four different reactions over two rare earth oxide catalysts have been determined at 923 K - N₂O decomposition, NO decomposition, NO reduction by CH₄ in the absence of O₂, and NO reduction by CH₄ with excess O₂ present. After examination of many different reaction mechanisms and the rate expressions derived from each of them, the sequence of elementary steps and the resulting rate equation which best describe the experimental results for each are reported here. The Langmuir-Hinshelwood-type expressions indicate that N₂O proceeds via unimolecular decomposition, whereas the rate determining step for NO decomposition involves a surface reaction between two adsorbed NO molecules. The reduction of NO by CH₄ in the absence of O₂ proceeds with methyl radical formation via hydride abstraction by NO implied as the slow step, and the rate determining step for NO reduction when both CH₄ and O₂ are present again appears to be methyl radical formation by reaction between adsorbed CH₄ and NO₂. The rate enhancements for NO reduction to N₂ and O₂ when CH₄, or CH₄ plus O₂, is utilized are attributed to the formation of methyl radicals and subsequent C-containing surface intermediates which rapidly interact with NO. The proposed chemistry is consistent with both that associated with methane oxidative coupling over rare earth oxides and that describing homogeneous free-radical reactions involving these intermediates. Consequently, a complete catalytic cycle of elementary steps is proposed for each reaction to demonstrate the possible presence of these various intermediates and to allow the testing of these models in the future.

Full text of DOI:10.1006/jcat.1996.0071

JOURNAL OF CATALYSIS 159, 119–126 (1996)
ARTICLE NO. 0071
The Kinetics of NOx Decomposition and NO Reduction
by CH4 over La2O3 and Sr/La2O3
M. Albert Vannice,1 Arden B. Walters, and X. Zhang
Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802-4400
Received August 18, 1995; revised November 13, 1995; accepted November 27, 1995
the SCR process utilizing ammonia. Correspondingly, nu-
merous studies of the roles of metals and oxygen have been
conducted; however, only a few measurements of kinetic
parameters have been reported (2–4) and only one detailed
sequence of reaction steps involving CH4 based on kinetic
behavior has been proposed (5). A family of rare earth ox-
ide (REO) catalysts has been prepared, characterized, and
tested (6–9), and the results have shown that good methane
oxidative coupling catalysts can also be good NO reduction
catalysts, even in the presence of excess O2. Among the
rare earth oxides tested, La2O3 and Sm2O3 with and with-
out Sr doping were the most active catalysts, although their
rates per gram were still 1–2 orders of magnitude lower
than that for Co/ZSM-5; however, when normalized to unit
surface area or represented as a turnover frequency based
on NO adsorption sites, the specific activities of these REO
catalysts were comparable to those reported for the zeo-
lite catalysts (7). Furthermore, unlike the zeolite catalysts,
the activities of these catalysts continuously increased with
reaction temperature with no bend-over, and high selectiv-
ities of methane utilization could be achieved over the best
catalysts.
This paper examines, in both the absence and presence
of O2, the kinetic behavior of NO reduction by CH4 as well
as NO decomposition over La2O3 and 4% Sr/La2O3, and
it also addresses N2O decomposition over these two cata-
lysts. Possible reaction mechanisms are discussed and for
each reaction the derived rate expression which best fits the
data is presented. The proposed sequences of elementary
steps strive for consistency with the chemistry associated
with methane oxidative coupling as well as that describing
homogeneous, free-radical reactions among the possible re-
action intermediates.
The kinetics of four different reactions over two rare earth oxide
catalysts have been determined at 923 K—N2O decomposition, NO
decomposition, NO reduction by CH4 in the absence of O2, and
NO reduction by CH4 with excess O2 present. After examination
of many different reaction mechanisms and the rate expressions
derived from each of them, the sequence of elementary steps and
the resulting rate equation which best describe the experimental re-
sults for each are reported here. The Langmuir–Hinshelwood-type
expressions indicate that N2O proceeds via unimolecular decom-
position, whereas the rate determining step for NO decomposition
involves a surface reaction between two adsorbed NO molecules.
The reduction of NO by CH4 in the absence of O2 proceeds with
methyl radical formation via hydride abstraction by NO implied as
the slow step, and the rate determining step for NO reduction when
both CH4 and O2 are present again appears to be methyl radical
formation by reaction between adsorbed CH4 and NO2. The rate
enhancements for NO reduction to N2 and O2 when CH4, or CH4
plus O2, is utilized are attributed to the formation of methyl radicals
and subsequent C-containing surface intermediates which rapidly
interact with NO. The proposed chemistry is consistent with both
that associated with methane oxidative coupling over rare earth ox-
ides and that describing homogeneous free-radical reactions involv-
ing these intermediates. Consequently, a complete catalytic cycle of
elementary steps is proposed for each reaction to demonstrate the
possible presence of these various intermediates and to allow the
c
testing of these models in the future. ꢀ 1996 Academic Press, Inc.
INTRODUCTION
Lean NOx reduction has attracted great attention in both
university and industry laboratories because of the need to
find catalysts which can work effectively under lean-burn
conditions to reduce NOx emissions. The most significant
progress in this area recently has been the discovery of zeo-
lite catalysts, such as Cu/ZSM-5 and Co/ZSM-5, which can
effectively convert NO into N2 under appropriate condi-
tions as discussed in the review by Armor (1–3). Of partic-
ular interest has been the use of CH4 to reduce NOx because
of the availability of natural gas and its potential to replace
EXPERIMENTAL
Kinetic studies of NO decomposition and NO reduction
by CH4 in both the absence and presence of O2 over La2O3
and Sr/La2O3 have been described in detail in two previ-
ous papers, and some of the kinetic data presented in this
paper were contained in those studies (6, 9). Briefly, 4%
1
To whom correspondence should be addressed.
119
0021-9517/96 $18.00
c
Copyright ꢀ 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.

120
VANNICE, WALTERS, AND ZHANG
RESULTS AND DISCUSSION
TABLE 1
Kinetic Behavior of N2O Decomposition over 4% Sr/La2O3
at 923 K
Tables 1–4 summarize the results of the kinetic studies
over La2O3 and Sr/La2O3. The reaction orders listed here
were determined by the best fit of the data to a power rate
law. All pressure units in the rate expressions are in Torr.
During the runs in Table 3, the CH4 pressure was held at
3.4 Torr while the NO pressure was varied, and the NO
pressure was held at 13.6 Torr while the CH4 pressure was
varied. In Table 4, the respective O2 and NO pressures were
7.56 and 13.7 Torr while the CH4 pressure was varied, the
respective O2 and CH4 pressures were 7.56 and 3.42 Torr
while the NO pressure was varied, and the respective CH4
and NO pressures were 3.40 for Sr/La2O3 (25.6 for La2O3)
and 13.6 Torr while the O2 pressure was varied.
Pessure (Torr)
a
App. Eact
Reaction
ordera
N2O
O2
(kcal/mole)
0.38–3.40
0.38–3.40
0
7.5
24
—
1.02
0.97
r (ꢀmole N2/sec ꢁ m2) = 0.081 PN O/(1 + 0.027 PN
)
O
2
b
2
a From Ref. (9).
b Pressure units are Torr.
Sr/La2O3, prepared by following the method employed for
Li-promoted MgO (10), and La2O3 were pretreated in situ
at973Kunder9.8%O2 inHeataflowrateof20cm3/minun-
til no CO2 could be detected. All kinetic studies were made
at 923 K and atmospheric pressure under steady-state reac-
tion conditions in a quartz microreactor system containing
ca. 100 mg of catalyst, as described previously (6, 10). The
ranges of partial pressure are summarized in Tables 1–4.
NO conversions were typically below 5% during NO de-
composition and reduction by CH4 and were usually less
than 10% during NO reduction by CH4 in excess O2. The
reactor effluent was analyzed with a Perkin-Elmer Sigma
2B gas chromatograph equipped with either a Carboxen1000
column (Supelco) or a 5A molecular sieve column and a
P-E Nelson 1020S integrator. All gases used were UHP
Grade (MG Ind.) except for the NO mixed in He, which
was 99.0+ %, with major mixture impurities of N2 (ꢂ600
ppm), N2O (ꢂ100 ppm), and CO2 (ca. 100 ppm). All rates
were based upon the formation of N2. Data-fitting was con-
ducted using a nonlinear regression technique contained
in a SAS computer package employing the Marquardt
method with a convergence criterion requiring a change
in the residual to decrease below 10ꢃ12. Three adjustable
parameters were used for all equations except that for NO
reduction by CH4 with O2 present, which contained four
parameters.
N2O Decomposition
This reaction is of interest not only because N2O is con-
sidered an environmental pollutant arising as a coproduct
from certain processes (11), but also because N2O has been
detected in the product stream during NO reduction by
CH4 over certain REO catalysts (6–9). Thus the possibil-
ity exists that it may be a reaction intermediate during this
sequence; alternatively, it may be formed as a product in a
parallel reaction during NO reduction. The analysis of the
effluent stream from this reaction was very easy and the
ratio of the N2 and O2 products was routinely near 2.
One of the simplest reaction sequences that can be pro-
posed is a Langmuir–Hinshelwood (L–H) model involving
unimolecular decomposition of adsorbed N2O as the rate
determining step (rds), i.e.,
KN2
O
*
2[N2O + ꢅ o N Oꢅ]
[1]
[2]
)
2
k
2[N2Oꢅ → N2 + Oꢅ] (rds)
1/KO2
*
2Oꢅ o O + 2ꢅ
[3]
[4]
)
2
2N2O → 2N2 + O2.
TABLE 2
Kinetic Behavior of NO Decomposition at 923 K
Partial pressure (Torr)
a
Eact
Reaction order
Catalyst
La2O3
(kcal/mole)
NO
O2
in NO
23a
—
2.70–13.5
2.73–13.7
0
7.53
1.22
1.46
r (ꢀmole N2/sec ꢁ m2) = 8.3 ꢄ 10ꢃ5 PN2O/(1 + 0.10 PNO + 0.26 P1/2
)
2 c
O
2
Sr/La2O3
28b
—
2.67–13.4
2.71–13.5
0
7.48
1.08b
1.11b
r (ꢀmole N2/sec ꢁ m2) = 2.4 ꢄ 10ꢃ4 PN2O/(1 + 0.077 PNO + 0.52 P1/2
)
2 c
O
2
a From Ref. (6).
b From Ref. (9).
c Pressure units are Torr.

NOx DECOMPOSITION AND NO REDUCTION BY CH4
121
TABLE 3
Kinetic Behavior of NO Reduction by CH4 at 923 K
Partial pressure (Torr)
Reaction order
NO
Eact
(kcal/mole)
Catalyst
La2O3
CH4
NO
CH4
24a
0.68–3.40
2.72–13.6
0.26a
0.98a
ꢃ3
2 c
r (ꢀmole N2/sec ꢁ m2) = 1.6 ꢄ 10
P
PCH /(1 + 0.0041 PNO + 0.30 PCH
)
4
NO
4
Sr/La2O3
26b
0.68–3.40
2.72ꢃ–313.6
0.19b
0.73b
2 c
r (ꢀmole N2/sec ꢁ m2) = 1.7 ꢄ 10 PNOPCH /(1 + 0.036 PNO + 0.49 PCH
)
4
4
a From Ref. (6).
b From Ref. (9).
c Pressure units are Torr.
Here ꢅ represents an empty active site while KN O and KO
O2 exchange occurs on La2O3 at these temperatures (12),
represent equilibrium adsorption constants. A site balance thus the model above is preferred.
2
2
fortheconcentrationofactivesites, L, combinedwithquasi-
The general L–H model for Eq. [4] has been proposed
equilibrated steps [1] and [3], gives the rate expression.
by others for N2O decomposition over Mn2O3 (13, 14)
and REOs including La2O3 (15, 16), whereas a redox se-
quence involving lattice oxygen has been proposed for Fe-
exchanged zeolites (17). At lower temperatures, Winter
found the best rate expression was r = kPN O/P1/2, which
ꢀꢁ
ꢂ
r = LkꢁN O = k0 KN O PN O 1 + KN O PN O + K1/2 PO1/2
.
2
2
2
2
2
O2
2
[5]
2
O2
Fitting this equation to the experimental data produced the
following equation with pressures in units of Torr.
implies that reversible oxygen chemisorption now becomes
significant and covers most of the sites, and his apparent Eact
value for La2O3 was 20 kcal/mole (15). For many different
oxides over a range of temperatures, a rate expression of
ꢀꢁ
ꢂ
r = 0.081PN O 1 + 0.027PN O
,
[6]
2
2
the form r = kPN O/(1 + K P1/2) has been reported (18),
2
again consistent with the abovOe2mechanism. In the present
study, O2 does not inhibit the reaction, in contrast to these
earlier results. This difference could be due simply to the
fact that a higher temperature (923 K) was employed in
this work, under which the rate of recombination and des-
orption of adsorbed atomic O would be enhanced. In fact,
when CH4 was co-fed along with N2O at 923 K, the de-
composition of N2O over Sr/La2O3 was not increased, but
actually was slightly decreased. These results suggest that
the removal of adsorbed O, which could be facilitated by
CH4 combustion, is not a slow step under our conditions.
whose correlation with the rate data is shown in Fig. 1. The
decomposition reaction was conducted in the presence and
absence of O2 in the feed and the lack of dependence on the
O2 pressure is clearly evident. This implies the surface con-
centration of chemisorbed O atoms is very low under these
conditions. The apparent activation energy of Eq. [5] be-
tween 773 and 973 K was 24 kcal/mole (9). It should be men-
tioned that other sequences can also provide satisfactory
rate expressions. For example, Eq. [5] is also obtained if
no rds is assumed, both steps [2] and [3] are irreversible,
and the surface coverage of O atoms is assumed to be small
compared to that of adsorbed N2O; however, rapid isotopic
TABLE 4
Kinetic Behavior of NO Reduction by CH4 in the Presence of O2 at 923 K
Partial pressure (Torr)
Reaction order
Eact
Catalyst
La2O3
(kcal/mole)
CH4
NO
O2
CH4
NO
O2
26a
0.69–3.42 2.74–13.7
0.38–22.8
0.62a
0.53a
0.50a
r (ꢀmole N2/sec ꢁ m2) = 6.1 ꢄ 10
P
PCH4P1/2/(1 + 0.081 PNO + 0.093 PCH4 + 0.035 P1/2
)
O
2
ꢃ3
2 c
NO
O
2
Sr/La2O3
26b
0.69–3.42 2.76–13.8
0.38–7.56
0.87b
0.55b
0.26b
r (ꢀmole N2/sec ꢁ m2) = 1.9 ꢄ 10
P
PCH4P1/2/(1 + 0.10 PNO + 0.068 PCH4 + 0.32 P1/2
)
O
2
ꢃ2
2 c
NO
O
2
a From Ref. (6).
b From Ref. (9).
c Pressure units are Torr.

122
VANNICE, WALTERS, AND ZHANG
FIG. 2. Rate dependence on NO pressure for NO decomposition over
FIG. 1. Rate dependence on N2O pressure for N2O decomposition
(a) La2O3 (open symbols) and (b) Sr/La2O3 (filled symbols): (ꢇ), (ꢆ), with-
out O2 in the feed; ( ), ( ), with 1% O2 in the feed; (——) fittings obtained
with Eq. [12] as represented in Table 2.
over Sr/La2O3: (ꢆ) without O2 in the feed; ( ) with 1% O2 in the feed;
(——) fitting obtained with Eq. [6].
Further support of this conclusion comes from the fact that
almost all the surface oxygen produced from NO adsorp-
tion and decomposition desorbed below 923 K, as revealed
by our recent TPD studies with Sr/La2O3 (19).
of two adsorbed NO species as the rds, as shown below:
KNO
*
2[NO + ꢅ o NOꢅ]
[7]
[8]
[9]
)
k
2NOꢅ → N2Oꢅ + Oꢅ (rds)
NO Decomposition
K
*
N Oꢅ o N + Oꢅ
)
2
2
Analyses of these effluent streams were more difficult be-
cause the O2 formed reacted with NO to form NO2 between
the reactor and the GC columns; consequently, little or no
molecular O2 was detected by gas chromatography. Under
differential reaction conditions with no O2 in the feed, O2
partial pressures were negligible and could be ignored in
the curve-fitting procedures; when 1% O2 was added to the
feed, the O2 partial pressure utilized was that represented
1/KO
*
o
)
2Oꢅ
2 O2 + 2ꢅ
[10]
[11]
2NO → N2 + O2.
If the surface coverage of N2O is assumed small due to
its much weaker adsorption and high reactivity, then the
following rate expression can be easily derived,
*
by the 2NO + O o 2NO equilibrium at 923 K. However,
)
2
2
ꢀꢁ
ꢂ
this did not alter the inlet pressure significantly.
2
r = LkꢁN2O = LkKN2 O PN2O 1 + KNO PNO + K1/2 PO1/2
,
O2
2
The reaction orders on NO were consistently higher than
first-order, whether O2 was co-fed or not, as listed in Table 2.
[12]
This strongly suggests a bimolecular surface reaction was where KNO and KO again represent equilibrium adsorp-
2
occurring. Because of the NO chemisorption we have mea- tion constants. After data fitting, the two quantitative rate
sured on these catalysts combined with the finding that it equations obtained are given in Table 2, and the correla-
decreases as temperature increases from 300 to 573 K (7, 9) tions with the data are shown in Fig. 2.
and that some NO desorbs molecularly above 400 K (19),
Experimentally measured reaction orders in NO for its
a quasi-equilibrated NO adsorption step was assumed and decomposition over oxides have fallen between 0 and 2, as
different bimolecular surface reactions were proposed as reviewed by Amirnazmi et al. (20), and two of these stud-
the rate determining step. These included adsorbed NO re- ies by Winter and Amirnazmi et al., in which a first-order
acting with another NO molecule, or an adsorbed N atom dependence was observed, proposed NO adsorption as the
or an adsorbed O atom, as well as the recombination and rds with O2 desorption being quasi-equilibrated (20, 21). A
desorption of 2N atoms or 2O atoms. Some of the derived similar reaction sequence has been proposed recently by
rate expressions could be rejected because they did not pre- Li and Hall for Cu-exchanged zeolites (22). Although the
dict the observed inhibitive effect of O2 when it was co-fed, adsorption sites on these REO surfaces have not been un-
as evidenced in Fig. 2. One model gave a clearly superior ambiguously identified, it is reasonable to associate them
fit—again it was a simple L–H sequence with the reaction with anion vacancies, which can be F-centers if they contain

NOx DECOMPOSITION AND NO REDUCTION BY CH4
123
a trapped electron, as originally proposed by Winter (15,
21), and more recently associated with the La2O3 surface
(12). Winter reported an activation energy of 16 kcal/mole
for NO decomposition over La2O3 (21), which is noticeably
lower than the value of 23 kcal/mole obtained in this study.
NO Reduction by CH4 (Absence of O2)
Analyses of the effluent stream for this reaction were not
complicated as no O2 or NO2 was detected; however, H2O
peaks could not be quantified and sometimes with the 5A
column neither could the CO2 peaks. The reaction orders
in CH4 were relatively low and positive, while those in NO
were near but less than one, as given in Table 3.
The use of CH4 as a reductant significantly enhances the
rate of N2 formation over REO catalysts (6–9), and this
can be seen by a comparison of Figs. 2 and 3. Seven differ-
ent reaction sequences were considered—these included
quasi-equilibrated dissociative and nondissociative NO ad-
sorption along with a rds invoking either CH3 formation via
hydride abstraction, a surface reaction between adsorbed
CH3 and adsorbed NO, or a surface reaction between ad-
sorbed CH2 and NO. These either gave derived rate expres-
sions with inappropriate reaction orders or they did not fit
the data well. Only one kinetic sequence provided a good
fit of the data—a L–H model assuming a surface reaction
between adsorbed NO and adsorbed CH4 to form a methyl
radical, with NO and CH4 adsorption in quasi-equilibrium,
as shown below:
FIG. 3. Rate dependence on reactant pressure for NO reduction by
CH4 over (a) La2O3 (open symbols) and (b) Sr/La2O3 (filled symbols):
(ꢇ), (ꢆ), CH4 with NO constant; ( ), ( ), NO with CH4 constant; (——)
fittings obtained with Eq. [24] as represented in Table 3. Experimental
points from references (6) and (9).
The curve-fitting procedure gave the two quantitative rate
equations for La2O3 and Sr/La2O3 listed in Table 2.
We have found no other kinetic model in the literature
describing NO reduction by CH4 in the absence of mole-
cular oxygen. Essentially, this sequence suggests that, in the
absence of O2, adsorbed NO acts as a hydride abstractor to
form methyl radicals, thus replacing chemisorbed oxygen
as proposed by Lunsford and co-workers for the methane
oxidative coupling reaction over La2O3 and Sr/La2O3 (23,
24). This same role for NO has been stated previously for
the gas-phase reaction to form CH3 ꢁ radicals (25). We have
admittedly speculated here on the surface chemistry fol-
lowing the rds that could constitute a complete catalytic cy-
cle; however, all the proposed elementary steps are based
on known homogeneous gas-phase reactions between free
radicals (26), and the homogeneous reaction between NO
and HCN is known (27). This sequence contains adsorbed
N2O as an intermediate, and its desorption, rather than de-
composition, could account for the small amounts of N2O
that are observed, sometimes up to 5–20% of the nitrogen
products (6–9). Clearly additional work is needed to ver-
ify this complicated sequence because other intermediates
and other steps can also be suggested, but this series of el-
ementary steps is proposed in a heuristic spirit to close the
catalytic cycle and to allow its testing; however, another
equilibrated sequence does not alter the kinetic rate equa-
tion.
KNO
*
4[NO + ꢅ o NOꢅ]
[13]
[14]
)
KCH
4
*
CH + ꢅ o CH ꢅ
)
4
4
k
NOꢅ + CH4ꢅ →HNOꢅ + CH3ꢅ (rds)
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
*
NOꢅ + HNOꢅ o N Oꢅ + OHꢅ
)
2
*
NOꢅ + CH ꢅ o HCNꢅ + H O + ꢅ
)
3
2
*
NOꢅ + HCNꢅ o HCOꢅ + N + ꢅ
)
2
*
HCOꢅ + Oꢅ o COꢅ + OHꢅ
)
*
COꢅ + Oꢅ o CO + 2ꢅ
)
2
*
N Oꢅ o N + Oꢅ
)
2
2
*
2OHꢅ o H O + Oꢅ + ꢅ
)
2
4NO + CH4 → 2N2 + CO2 + 2H2O.
[23]
If NOꢅ and CH4ꢅ are assumed to be the most abundant
surface intermediates, then a site balance gives L = [NOꢅ]
+ [CH4ꢅ] + [ꢅ] and the following rate expression can be
derived:
NO Reduction by CH4 (Presence of O2)
r = LkꢁNOꢁCH
Analyses of this product mixture were the most difficult
due to the reactivity of excess O2 and NO to form NO2 be-
[24] tween the reactor exit and the GC column. Although rates
4
ꢀꢁ
ꢂ
2
= LkKNO KCH PNO PCH 1 + KNO PNO + KCH PCH
.
4
4
4
4

124
VANNICE, WALTERS, AND ZHANG
FIG. 4. Rate dependence on reactant pressure for NO reduction by CH4 in the presence of O2 over (a) La2O3 (open symbols) and (b) Sr/La2O3
(filled symbols): (ꢇ), (ꢆ), CH4 with NO constant; ( ), ( ), NO with CH4 constant; (4), (N), O2 with both CH4 and NO constant; (——) fittings obtained
with Eq. [43] as represented in Table 4. Data points for O2 dependence in (a) from reference (6); data points for NO and CH4 dependencies from
reference (9).
of N2 formation could still be monitored, N2 balances could associated with CH4 activation during oxidative coupling
not be obtained because NO2 was not eluted quantitatively over REOs, and (iii) was also consistent with known gas-
from the gas chromatograph. NO conversions were typi- phase, free-radical species and reactions. At the tempera-
cally below 10%; however, due to the CH4 combustion re- tures used, we have assumed that heterogeneous reactions
action which can occur in parallel with NO reduction, CH4 predominate over any homogeneous reactions. More than
conversions could be high and they varied from 1 to 45%. a dozen sequences were examined involving, among others,
Consequently, the partial pressures could change signifi- dissociative and nondissociative NO adsorption as well as
cantly in some cases and, when necessary, average partial the presence and absence of NO2 species along with the
pressures in the reactor were calculated in the following following different rate determining steps: CH3 radical for-
manner:
mation by an adsorbed O atom or an NO2 species, reaction
between adsorbed CH3 and an NO or NO2 species, and re-
action between adsorbed CH2 and NO species. We have
found only one kinetic model which can satisfy these con-
straints and still describe the reaction well over both REO
catalysts. It invokes a surface reaction between adsorbed
CH4 and adsorbed NO2 to generate a methyl species, and
one possible complete catalytic cycle, again based on known
gas-phase species (26), is
ꢃ
ꢄ
XNO
PNO
=
=
1 ꢃ
PNꢇO
,
[25]
2
ꢃ
ꢄ
X
CH4
PCH
1 ꢃ
PCꢇH
,
,
[26]
[27]
4
4
2
PH O = XCH PCꢇH + PHꢇ O
2
4
4
2
and
ꢅ
ꢆ
1
4
1
2
PO = POꢇ ꢃ XCH PCꢇH
ꢃ
X
NO PNꢇO
ꢃ
PNO .
2
[28]
KNO
2
4
2
4
*
4[NO + ꢅ o NOꢅ]
[29]
[30]
[31]
[32]
)
KO2
Xi is the fractional conversion of i, Pꢇ is the inlet partial
pressure of i, Pi is the average partial pressure across the
catalystbed, andPNO isthepartialpressureofNO2 at923K
*
i
2[O + 2ꢅ o 2Oꢅ]
)
2
KCH
4
2
*
2[CH4 + ꢅ o CH ꢅ]
)
4
*
)
asdeterminedbytheequilibriumof2NO+O o 2NO . The
2
2
KNO
2
last correction was small.
*
3[NO ꢅ +Oꢅ o NO ꢅ +ꢅ]
)
2
With these REO oxide catalysts, the addition of O2 to the
feed significantly increased activity compared to CH4 only
(6–9), as can be seen by a comparison of Figs. 3 and 4. Con-
sequently, a reaction mechanism was sought that (i) could
explain the rate-enhancing role of excess O2 but still retain
chemical consistency with the three reaction sequences pro-
posed earlier, (ii) was consistent with the surface chemistry
k
2[CH4ꢅ + NO2ꢅ → CH3ꢅ + HNO2ꢅ] (rds) [33]
*
2HNO ꢅ o NOꢅ + NO ꢅ + H O
[34]
[35]
[36]
)
2
2
2
*
2[CH ꢅ + NOꢅ o HCNꢅ + H O + ꢅ]
)
3
2
*
2[HCNꢅ + Oꢅ o CNꢅ + OHꢅ]
)

NOx DECOMPOSITION AND NO REDUCTION BY CH4
125
*
2[CNꢅ + NO ꢅ o NCOꢅ + NOꢅ]
[37] the absence of O2 gives specific activities and activation en-
ergies that are very similar to those for NO reduction by
)
2
*
2[NCOꢅ + NOꢅ o N Oꢅ + COꢅ]
[38]
)
2
CH4 in the presence of O2 (6), thus suggesting that the for-
mer reaction proceeds with surface chemistry similar to the
*
2[COꢅ + Oꢅ o CO + 2ꢅ]
[39]
)
2
latter. Fourth, as in the NO + CH4 reaction, the genera-
*
2[N Oꢅ o N + Oꢅ]
[40]
)
2
2
tion of methyl radicals, which is known to occur in methane
oxidative coupling (23, 24), is again the rds; hydride ab-
straction by an adsorbed O atom can also occur, but reac-
*
2OHꢅ o Oꢅ + H O + ꢅ
)
[41]
2
tion orders in O2 as high as 0.5 could not be attained in
the derived rate equations if this is the only CH4 activation
4NO + 2CH4 + 2O2 → 2N2 + 4H2O + 2CO2.
[42]
The derivation of the rate expression is straightforward.
If the concentrations of NOꢅ, CH4ꢅ, and Oꢅ are assumed
to be predominant—NO2ꢅ is assumed to be small com-
pared to NOꢅ, in agreement with the gas-phase equilibrium
concentrations at this temperature—a site balance gives
L = [ꢅ] + [NOꢅ] + [CH4ꢅ] + [Oꢅ] and the rate expression
becomes
step. Finally, it should again be emphasized that different
sequences and intermediates can be proposed following the
rds to complete the cycle, but the derived reaction kinetics
are independent of this chemistry as the last nine steps can
be combined into a single, overall quasi-equilibrated step.
The only other in-depth kinetic study which has re-
ported partial pressure dependencies and proposed a com-
plete detailed catalytic cycle is that of Li et al. involving
Co/ferrierites operating at lower temperatures (4, 5). The
sequence represented by steps [29]–[41] is similar to that of
Lietal. inthatabsorbedNO2 isproposedtogeneratemethyl
radicals in a rds, but it differs in that the model of Li et al.
utilizes different chemical steps and does not explicitly pro-
pose the involvement of adsorbed oxygen or adsorbed CH4,
thus it has an invariant half-order dependence on O2 and
first-order dependence on CH4. Also, if the first two steps in
their sequence are multiplied by a stoichiometric number of
2 to balance all species, the overall stoichiometry of their re-
action is 3NO+1.5O2 +CH4 → N2 +CO2 +2H2O+NO2,
thus NO2 is generated as a product. The sequence repre-
sented by steps [29]–[41] is consistent with the study of Cant
andco-workerswhichindicatedthattherdsoverCo/ZSM-5
r = LkꢁCH ꢁNO
4
2
ꢀ
= LkKCH KNO KNO K1/2 PCH PNO P1/2
4
2
4
O2
O2
ꢁ
ꢂ
2
1 + KNO PNO + KCH PCH + KO1/2 PO1/2
.
[43]
4
4
2
2
Fitting this to the data gave the two rate equations for La2O3
and Sr/La2O3 provided in Table 4, and the correlations are
quite good, as shown in Fig. 4. The only dependence on
H2O and CO2 expected from this model would come if they
were included in the site balance because of competitive
adsorption; however, the data were successfully described
without their inclusion.
Several comments are appropriate at this point. First, all
the catalytic sequences examined which involved no NO2
species gave a stoichiometry of 6NO + 2CH4 + O2
→
–
involves the breaking of a C H bond (31).
3N2 + 2CO2 + 4H2O rather than that given by Eq. [42],
i.e., 2NO + CH4 + O2 → N2 + CO2 + 2H2O; however, a
catalytic cycle with the former stoichiometry can also be
obtained with a surface NO2 species by proposing different
intermediates. The sequence above is suggested primarily
SUMMARY
In the kinetic analyses of these four reactions, we have ex-
for heuristic proposes and is preferred because steps 35–41 amined many different reaction sequences: first, to find the
are known gas-phase free radical reactions (26, 27) and N2O ones mathematically capable of fitting the data; second, to
is an intermediate, thus explaining its gas-phase presence selectmodelsconsistentwithbothoxidativecouplingchem-
and providing consistency with the previous sequences. This istry and known homogeneous free-radical chemistry; and
stoichiometry also suggests that definitions of selectivity third, to retain consistency in the surface chemistry among
(CH4 combustion vs NO reduction) may have to be reex- the final four rate expressions best describing each of these
amined. Second, although this reaction clearly represents reactions. The results of this effort are offered here not in
complex chemistry, step [34] is known chemistry (28) and a dogmatic manner, but in the spirit of providing guidance
step [33] has been proposed recently by Li et al. (5). Hall toward a better understanding of these complicated reac-
and co-workers have also proposed that NO2 is an interme- tion systems. The rate expressions given here allow predic-
diate whose role in this reaction over Co/zeolite catalysts tions to be made and tests to be conducted to verify their
is to generate CH3 radicals (29), and Kikuchi and Yogo validity.
have proposed an initial reaction between NO2 and CH4 to
All the rate expressions derived here have been based
form a reactive intermediate on Ga/ZSM-5 and In/ZSM-5 on L–H models, i.e., a surface reaction represents the rate
(30). We have already shown that NO2 is very active, but determining step. The chemistry for unimolecular N2O
nonselective, for CH4 combustion under oxidative coupling decomposition is simple and in excellent agreement with
conditions over La2O3 (6). Third, NO2 reduction by CH4 in past studies. For NO decomposition at these temperatures,

126
VANNICE, WALTERS, AND ZHANG
ourresultsupportabimolecularreactionbetweenadsorbed
NO as the rds, which is consistent with some older studies,
but in others NO adsorption has also been proposed as the
rds on certain oxides. The chemistry associated with NO
reduction by CH4 in the presence or absence of molecular
O2 is very complex, but complete catalytic cycles can be
obtained based on free-radical species and reactions, and
we have taken the liberty to suggest one possible sequence
for each situation. However, the kinetic behavior is of pri-
mary interest, and the results here indicate that in the ab-
3. Li, Y., and Armor, J. N., Appl. Catal. B 2, 239 (1993).
4. Li, Y., and Armor, J. N., J Catal. 150, 376 (1994).
5. Li, Y., Slager, T. L., and Armor, J. N., J. Catal. 150, 388 (1994).
6. Zhang, X., Walters, A. B., and Vannice, M. A., Appl. Catal. B 4, 237
(1994).
7. Zhang, X., Walters, A. B., and Vannice, M. A., J. Catal. 155, 290
(1995).
8. Zhang, X., Walters, A. B., and Vannice, M. A., Catal. Today in press.
9. Zhang, X., Walters, A. B., and Vannice, M. A., Appl. Catal. B in press.
10. Zhang, X., Walters, A. B., and Vannice, M. A., J. Catal. 146, 568
(1994).
11. Li, Y., and Armor, J. N., Appl. Catal. B 1, L21 (1992).
sence of both CH4 and O2, activation of NO is the slow step. 12. Lacombe, S., Geantet, C., and Mirodatos, C., J. Catal. 151, 439
(1994).
The addition of only CH4 enhances the rate by being more
reactive with NO, forming a methyl radical intermediate,
and allowing very rapid subsequent reduction reactions to
occur between these C-containing intermediates and NO.
13. Rheaume, L., and Parravano, G., J. Phys. Chem. 63, 264 (1959).
14. Yamashita, T., and Vannice, M. A., submitted for publication.
15. Winter, E. R. S., J. Catal. 15, 144 (1969).
16. Read, J. F., J. Catal. 28, 428 (1973).
The presence of O2 with CH4 further increases the rate be- 17. Fu, C. M., Korchak, V. N., and Hall, W. K., J. Catal. 68, 166 (1981).
18. Roussel, P., and Teichner, S. J., Catal. Rev. 6, 133 (1972).
19. Zhang, X., and Vannice, M. A., unpublished data.
20. Amirnazmi, A., Benson, J. E., and Boudart, M., J. Catal. 30, 55
cause adsorbedNO2 ismorereactivewithCH4 than NOand
forms a methyl radical species more rapidly. Thus a consis-
tent trend in the rate-controlling surface chemistry appears
to exist; i.e., the rate of reaction between two adsorbed
(1973).
21. Winter, E. R. S., J. Catal. 22, 158 (1971).
NO molecules is slower than that between an absorbed NO 22. Li, Y., and Hall, W. K., J. Catal. 129, 202 (1991).
23. Lin, C.-H., Campbell, K. D., Wang, J.-X., and Lunsford, J. H., J. Phys.
Chem. 90, 534 (1986).
24. Xu, M., and Lunsford, J. H., Catal. Lett. 11, 295 (1991).
molecule and an adsorbed CH4 molecule while the reac-
tion between an adsorbed NO2 species and an adsorbed
CH4 molecule is more rapid than either of the previous
reactions.
25. Wojciechowski, B. W., and Laidler, K. J., Can. J. Chem. 38, 1027
(1968).
26. Miller, J. A., and Bowman, C. T., Prog. Energy Combust. Sci. 15, 287
ACKNOWLEDGMENTS
(1989).
27. Ismagilov, Z. R., and Kerzhentsev, M. A., Catal. Rev.-Sci. Eng. 32, 51
(1990).
28. Cotton, F. A., and Wilkinson, G., “Advanced Inorganic Chemistry.”
p. 349. Interscience Publishers, New York, 1966.
29. Lukyanov, D. B., Sill, G., D’Itri, J. L., and Hall, W. K., J. Catal. 153, 265
(1995).
The work was supported by the National Science Foundation under
Grant CTS-9211552. We also thank Professor W. N. Delgass for invaluable
comments related to the derivation of these kinetic rate expressions.
REFERENCES
30. Kikuchi, E., and Yogo, K., Catal. Today 22, 73 (1994).
31. Cowan, A. D., Du¨mpelmann, R., and Cant, N. W., J. Catal. 151, 356
(1995).
1. Armor, J. N., Appl. Catal. B 1, 221 (1992).
2. Li, Y., and Armor, J. N., Appl. Catal. B 1, L31 (1992).