Effect of reductants in N2O reduction over Fe-MFI catalysts

Article abstract of DOI:10.1016/j.jcat.2004.11.009

We investigated the effect of reductants over ion-exchanged Fe-MFI catalysts (Fe-MFI) based on the catalytic performance in N₂O reduction in the presence and absence of an oxygen atmosphere. In the case of N₂O reduction with hydrocarbons (CH₄, C₂H ₆, and C₃H₆) in the presence of excess oxygen, the order of N₂O contribution was as follows: CH₄ > C₂H₆ > C₃H₆. This indicates that CH₄ is a more efficient reductant than C₂H₆ and C₃H₆. The TOFs of N₂O decomposition and the N₂O reduction by various reductants (H₂, CO, CH ₄) in the absence of oxygen increased with increasing Fe/Al ratio (Fe/Al≥0.15), wheras the TOFs were lower and constant in the range of Fe/Al≤0.10. Temperature-programmed reduction with hydrogen (H ₂-TPR) showed that the catalysts with a higher Fe/Al ratio were reduced more easily than those with a lower Fe/Al ratio. Temperature-programmed desorption of O₂ (O₂-TPD) showed that oxygen was desorbed at lower temperatures over the catalysts with a higher Fe/Al ratio. As the result of extended X-ray absorption fine structure (EXAFS) analysis, only mononuclear Fe species were observed over Fe(0.10)-MFI after treatment with N₂O or O₂. On the other hand, binuclear Fe species and mononuclear Fe species were observed over Fe(0.40)-MFI after treatment with N₂O or H₂. More reducible Fe species, which gave lower-temperature O₂ desorption, can be due to Fe binuclear species. Since the N₂O reduction with reductants proceeds via a redox mechanism, the reducible binuclear Fe species can exhibit higher activity. Furthermore, CH₄ can be oxidized by N₂O more easily than can H₂ and CO, although it is generally known that the reactivity of methane is very low.

Full text of DOI:10.1016/j.jcat.2004.11.009

Journal of Catalysis 229 (2005) 374–388
www.elsevier.com/locate/jcat
Effect of reductants in N2O reduction over Fe-MFI catalysts
Takeshi Nobukawa a, Masanori Yoshida a, Kazu Okumura b, Keiichi Tomishige a,∗,
Kimio Kunimori a,∗
a
Institute of Materials Science, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8573, Japan
b
Department of Materials Science, Faculty of Engineering, Tottori University, Koyama-cho Minami, Tottori 680-8552, Japan
Received 16 September 2004; revised 8 November 2004; accepted 9 November 2004
Available online 23 December 2004
Abstract
We investigated the effect of reductants over ion-exchanged Fe-MFI catalysts (Fe-MFI) based on the catalytic performance in N O reduc-
2
tion in the presence and absence of an oxygen atmosphere. In the case of N O reduction with hydrocarbons (CH , C H , and C H ) in the
2
4
2
6
3 6
presence of excess oxygen, the order of N O contribution was as follows: CH > C H > C H . This indicates that CH is a more efficient
2
4
2
6
3
6
4
reductant than C H and C H . The TOFs of N O decomposition and the N O reduction by various reductants (H , CO, CH ) in the absence
2
6
3
6
2
2
2
4
of oxygen increased with increasing Fe/Al ratio (Fe/Al ꢀ 0.15), wheras the TOFs were lower and constant in the range of Fe/Al ꢁ 0.10.
Temperature-programmed reduction with hydrogen (H -TPR) showed that the catalysts with a higher Fe/Al ratio were reduced more eas-
2
ily than those with a lower Fe/Al ratio. Temperature-programmed desorption of O (O -TPD) showed that oxygen was desorbed at lower
2
2
temperatures over the catalysts with a higher Fe/Al ratio. As the result of extended X-ray absorption fine structure (EXAFS) analysis, only
mononuclear Fe species were observed over Fe(0.10)-MFI after treatment with N O or O . On the other hand, binuclear Fe species and
2
2
mononuclear Fe species were observed over Fe(0.40)-MFI after treatment with N O or H . More reducible Fe species, which gave lower-
2
2
temperature O desorption, can be due to Fe binuclear species. Since the N O reduction with reductants proceeds via a redox mechanism,
2
2
the reducible binuclear Fe species can exhibit higher activity. Furthermore, CH can be oxidized by N O more easily than can H and CO,
4
2
2
although it is generally known that the reactivity of methane is very low.
2004 Elsevier Inc. All rights reserved.
Keywords: N O reduction; Fe-MFI; Fe binuclear; Turnover frequency; Reductant; Wet ion-exchange; EXAFS
2
1. Introduction
[11–20] and ammonia [21–23] have been proposed as an ef-
fective method of N2O abatement.
Recently, a number of researches have reported that Fe
binuclear species are active sites for NOx SCR and N2O
decomposition on the basis of characterization by means
of H2-TPR, CO-TPR, FTIR, EPR, EXAFS, and Möss-
bauer spectroscopies [24–32]. Koningsberger et al. [24–26]
have reported evidence for a binuclear iron-oxo complex
on the basis of EXAFS study of fully exchanged Fe-MFI
(Fe/Al = 1). They have also proposed the evolution and re-
activity of the binuclear iron-oxo species in thermal pretreat-
ment on fully exchanged Fe-MFI prepared by sublimation.
Prins et al. [27,28] have reported that the diiron structure re-
sembles the core unit in methane monooxygenase (MMO),
in terms of an Fe–Fe distance of approximately 0.3 nm.
Nitrous oxide (N2O) has been long considered a relatively
harmless gas and has suffered from a lack of interest from
environmental scientists and engineers. However, during the
last decade there has been a growing concern since N2O is
a harmful gas in our environment, contributing to the green-
house effect and ozone layer depletion. Therefore, catalytic
decomposition of N2O [1–10] and selective catalytic reduc-
tion (SCR) of N2O with reductants such as hydrocarbons
*
Corresponding authors. Fax: +81-29-855-7440.
E-mail addresses: tomi@tulip.sannet.ne.jp (K. Tomishige),
kunimori@ims.tsukuba.ac.jp (K. Kunimori).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.11.009
2004 Elsevier Inc. All rights reserved.

T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
375
Joyner and Stockenhuber [33] have shown that the nature
Table 1
Reaction conditions in N O reduction
a
2
of the Fe species in Fe-MFI depends markedly on the ion-
exchange method used for the preparation and on the type of
pretreatment. On the basis of EXAFS results, these authors
reported that in Fe/ZSM-5 zeolites prepared with different
ion-exchange methods, Fe was stabilized in different forms,
ranging from isolated metal ions to large oxide clusters, and
concluded that ion-oxo nanoclusters are most active in the
NOx SCR reaction [33]. On the other hand, mononuclear
iron-oxo species, which exist mainly on Fe-MFI with a low
exchange level, have been proposed as active species for the
reduction of N2O [13,22]. Delahay et al. [22] have reported
that the most active species in the reduction by NH3 on Fe-
BEA in an oxygen-rich atmosphere are in higher proportion
at low exchange levels. Furthermore, Segawa et al. [13] have
also reported that even low-exchange Fe-ZSM-5 catalysts
exhibited high activity when C3H6 was used as a reductant.
Recently we reported that the catalytic activity in N2O
reduction with CH4 on Fe-MFI was dependent on the ex-
change level of Fe ion [19]. From the results of activity test
and catalyst characterization, we have concluded that the ac-
tive sites are Fe species that are more reducible and give
a lower-temperature O2 desorption peak [19]. From these
comparisons, it is expected that the active structure of iron
species over zeolites can be influenced by the kinds of reduc-
tant. In this study, we investigated the relation between the
exchange level of iron and the activity of N2O reduction with
various reductants such as CH4, C2H6, C3H6, H2, and CO.
Furthermore, we characterized the structure of Fe species on
our Fe-MFI catalysts by means of temperature-programmed
desorption of O2 (O2-TPD), temperature-programmed re-
duction with H2 (H2-TPR), and extended X-ray absorption
fine structure (EXAFS). We discuss the nature of the active
sites of N2O reduction on the basis of the relation between
catalyst structure and performance.
Reductant
Gas composition
Catalyst
weight
(mg)
W/F
(g h/mol)
N O
2
(ppm)
Reductant
(ppm)
O
2
(%)
CH
950
1000
1000
500
250
250
10
10
20
50
100
100
0.41
0.41
0.41
4
C H
C H
3
2
6
6
CH
950
1000
1000
500
1000
1000
0
0
0
50
100
100
0.41
0.41
0.41
4
H
2
CO
None
a
950
0
0
50
0.41
All gases were balanced with He.
80% in the case of Fe/Al = 0.40. The catalyst is denoted
Fe(X)-MFI, where X stands for the molar ratio of Fe/Al.
2.2. Activity tests
Catalytic reduction of N2O with various reductants in the
presence of excess O2 or in the absence of O2 was carried out
in a fixed-bed flow reactor. The composition of the reactant
gases is listed in Table 1. All of these research-grade gases
were purchased from Takachiho Trading Co. Ltd., and they
were used without further purification. The catalyst weight
was 50 or 100 mg, the total pressure was 0.1 MPa, and
W/F (W (g)—catalyst weight, F (mol/h)—total flow rate)
was 0.41 g h/mol. The catalysts were pretreated at 773 K
with O2 for 1 h in the reactor. The products were monitored
with an on-line TCD gas chromatograph (Shimadzu GC-8A)
equipped with a Molecular sieve 5A column for N2 and O2
and a Porapak Q column for N2O; a FID gas chromatograph
(Shimadzu GC-14B) equipped with a Gaskuropak 54 col-
umn and a methanator for CO, CO2 and CH4; and another
gas chromatograph equipped with a VZ-10 column for C2H6
and C3H6. The sampling and analysis of effluent gas was
carried out for 1 h at each reaction temperature. The results
of the activity tests shown in the figures were obtained under
steady-state conditions.
2. Experimental
2.1. Catalyst preparation
2.3. Catalyst characterization
Fe-MFI catalysts were prepared by an ion-exchange
method with an aqueous solution of FeSO4 · 7H2O (Wako
Pure Chemical Industries Ltd.; 98%) for 20 h at 323 K under
a nitrogen atmosphere to avoid the precipitation of Fe(OH)3
[34]. Na-MFI (TOSOH Co.; SiO2/Al2O3 = 23.8) was used
as the catalyst support. The catalyst was separated from the
solution by filtration after an ion exchange procedure. And
it was washed thoroughly with distilled water and dried at
383 K overnight, followed by calcination in air at 773 K for
3 h. The loading amount of Fe on MFI was determined by
subtraction of the Fe amount in the solution after the sepa-
ration, analyzed by ICP analysis, from the total amount. The
exchange efficiency of FeSO4, the percentage of the iron
salt incorporated into the zeolite, was almost 100% in the
case of Fe/Al = 0.05–0.24. On the other hand, it was almost
Temperature-programmed reduction (TPR) with H2 was
performed in a fixed-bed flow reactor. The sample was pre-
treated in 100% O2 flow at 773 K for 1 h, and then it was
cooled to room temperature and exposed to helium flow to
purge the line. The TPR profile of each sample was recorded
from room temperature to 973 K under a flow of 5.0% H2
diluted with Ar. The flow rate of H2/Ar was 30 ml/min,
and the catalyst weight was 50 mg. The heating rate was
10 K/min, and the temperature was maintained kept con-
stant for 10 min after it reached 973 K. The consumption of
H2 was monitored continuously with a TCD gas chromato-
graph equipped with a Molecular Sieve 5A to remove H2O
from the effluent gas.

376
T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
Temperature-programmed desorption of O2 (O2-TPD)
search Organization in Tsukuba, Japan. The storage ring was
operated at 2.5 GeV with a ring current of 300–450 mA.
A single Si(111) crystal was used to obtain a monochro-
matic X-ray beam. The monochromator was detuned to 60%
of the maximum intensity to avoid higher harmonics in the
X-ray beam. Two ion chambers filled with N2 and 15% Ar
diluted with N2 were used as detectors of I0 and I, respec-
tively. EXAFS data were collected in transmission mode at
liquid nitrogen temperature. For EXAFS analysis, the oscil-
lation was first extracted from the EXAFS data by a spline
smoothing method [35]. The oscillation was normalized by
the edge height around 50 eV. The Fourier transformation
of the k3-weighted EXAFS oscillation from k space to r
space was performed over the range 23–120 nm−1 to obtain
a radial distribution function. The inversely Fourier filtered
data were analyzed by a common curve-fitting method [36,
37]. For the curve-fitting analysis, the empirical phase shift
and amplitude functions for Fe–O were extracted from the
data for ferric acetylacetonate. The phase shift and backscat-
tering amplitude of Fe–Si and Fe–Fe bonds were calcu-
lated with the software FEFF8.2 [38]. The program ATOMS
[39] was used to calculate coordination numbers and in-
teratomic distances from reported XRD data for reference
compounds. The theoretical Fe–Fe reference was calibrated
against EXAFS data obtained from Fe2O3 (hematite) at liq-
was carried out in a fixed-bed reactor equipped with a
quadrupole mass spectrometer (Balzers QMS 200 F). The
catalysts (30 mg) were pretreated with O2 flow (100% O2,
773 K, 1 h) or N2O flow (10% N2O/He, 773 K, 1 h). Af-
ter the pretreatment, they were cooled to room temperature.
Helium gas (flow rate 55 ml/min) was introduced into the
reactor, and the sample was heated at a rate of 10 K/min
from room temperature to 1273 K. The sample temperature
was kept constant for 30 min once it reached 1273 K. De-
sorbed O2 in He flow (flow rate, 55 ml/min) was analyzed
with a quadrupole mass spectrometer.
2.4. EXAFS measurement and analysis
We prepared the sample for EXAFS measurement by
pressing 60 mg of catalyst powder. The thickness of the sam-
ples was chosen to be 0.6–0.7 mm (10 mm φ) to give an edge
jump of 0.2–0.7. The sample was pretreated at 773 K with
11 kPa O2, H2, or N2O for 0.5 h in a closed circulating re-
actor. After these kinds of pretreatment, we transferred the
samples to the measurement cell without exposing the sam-
ple disk to air, using a glove box filled with nitrogen.
Fe K-edge XAFS was measured at the BL-12C station
of the Photon Factory at the High Energy Accelerator Re-
Fig. 1. Reaction temperature dependence of catalyst performance of Fe-MFI catalyst in N O reduction with C H under excess O atmosphere: (a) N O
2
3
6
2
2
conversion, (b) C H conversion, (c) CO selectivity (CO /(CO + CO )), and (d) C H conversion as function of N O conversion. ": Fe(0.40)-MFI,
3
6
2
2
2
3
6
2
Q: Fe(0.24)-MFI, 2: Fe(0.15)-MFI, a: Fe(0.10)-MFI, and F: Fe(0.05)-MFI. Gas composition: 1000 ppm N O, 250 ppm C H , 20% O (He balance). Open
2
3
6
2
marks represent the data at which carbon deposition was observed.

T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
377
uid nitrogen temperature, by fitting in R space. EXAFS data
were analyzed with the REX2000 program (Rigaku Co.;
Version 2.3.3).
under an air atmosphere. On the other hand, for the data in-
dicated by closed symbols in Fig. 1, the mass balance was
good. N2O conversion increased with the loading amount
of Fe. According to Fig. 1b, C3H6 conversion increased
with reaction temperature over Fe-MFI with low loading
of Fe (Fe/Al = 0.05 and 0.10). In contrast, C3H6 conver-
sion reached almost 100% in a rather low temperature range
over Fe-MFI (Fe/Al = 0.15 and 0.24). CO2 and CO were
observed as the carbon-containing products, and CO2 selec-
tivity is represented in Fig. 1c. Fig. 1d shows the relation
between N2O conversion and C3H6 conversion. It is clearly
shown that C3H6 conversion reached 100%, even for a low
level of N2O conversion. Furthermore, C3H6 conversion was
observed even when N2O conversion was zero, which means
that C3H6 reacted with oxygen, not with N2O.
Fig. 2 shows the temperature dependence of the catalyst
performance of Fe-MFI in N2O reduction with C2H6 under
an excess oxygen atmosphere. N2O and C2H6 conversions
increased with reaction temperature and the loading amount
of Fe (Figs. 2a and 2b). In the relation between N2O and the
reductant conversion, the behavior in N2O reduction with
C2H6 was much different from that in N2O reduction with
C3H6. As shown in Fig. 2d, C2H6 conversion increases pro-
portionally with N2O conversion in a N2O conversion range
of 0–50%. This means that C2H6 conversion is almost zero
3. Results and discussion
3.1. N2O reduction with hydrocarbons over Fe-MFI under
an excess oxygen atmosphere
Fig. 1 shows the temperature dependence of the catalyst
performance of Fe-MFI in N2O reduction with C3H6 under
an excess oxygen atmosphere. In this experiment, the activ-
ity test was carried out from high temperature (773 K) to
low temperature (at most 473 K) to avoid the influence of
carbon deposition, which was observed at lower tempera-
ture (open symbols in Fig. 1a). In this temperature region,
the color of the catalyst became black during the reaction,
and a disagreement in the carbon mass balance was ob-
served. The introduced carbon amount did not agree with
the carbon amount in the effluent gas at the lower tempera-
ture described by open marks. This behavior due to carbon
deposition was reported previously [40]. In addition, we con-
firmed the carbon deposition on the catalyst after reaction
by thermogravimetry analysis (TGA). In TGA, the weight
loss due to coke combustion was observed during heating
Fig. 2. Reaction temperature dependence of catalyst performance of Fe-MFI catalyst in N O reduction with C H under excess O atmosphere: (a) N O
2
2
6
2
2
conversion, (b) C H conversion, (c) CO selectivity (CO /(CO + CO )), and (d) C H conversion as function of N O conversion. ": Fe(0.40)-MFI,
2
6
2
2
2
2
6
2
2: Fe(0.15)-MFI, a: Fe(0.10)-MFI, and F: Fe(0.05)-MFI. Gas composition: 1000 ppm N O, 250 ppm C H , 10% O (He balance).
2
2
6
2

378
T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
Fig. 3. Reaction temperature dependence of catalyst performance of Fe-MFI catalyst in N O reduction with CH under excess O atmosphere: (a) N O conver-
2
4
2
2
sion, (b) CH conversion, (c) CO selectivity (CO /(CO+CO )), and (d) CH conversion as function of N O conversion. ": Fe(0.40)-MFI, Q: Fe(0.24)-MFI,
4
2
2
2
4
2
2: Fe(0.15)-MFI, a: Fe(0.10)-MFI, and F: Fe(0.05)-MFI. Gas composition: 950 ppm N O, 500 ppm CH , 10% O (He balance).
2
4
2
when N2O conversion is zero. This suggests that N2O is nec-
essary for the activation of C2H6. Similar behavior was also
observed on the Fe-BEA catalyst in our previous study [17].
In addition, the carbon amount in the effluent gas was bal-
anced with that introduced in the case of C2H6 in all temper-
ature ranges, unlike the case of C3H6.
than the stoichiometry of N2O reduction with hydrocarbons,
as listed in Table 1, N2O conversion can reach 100% before
the conversion of hydrocarbons when the reaction proceeds
stoichiometrically. However, this is not the case when hy-
drocarbons react with oxygen, and this tendency is related to
the reactivity between hydrocarbons and oxygen. Therefore,
for evaluation of the contribution of N2O in a total oxidiz-
ing agent, the results are shown in Fig. 4. In the case of N2O
reduction with CH4, the contribution of N2O can be calcu-
lated by (consumed N2O)/(4CO2 + 3CO) on the basis of
these equations:
Fig. 3 shows the temperature dependence of catalyst per-
formance in N2O reduction with CH4 under an excess oxy-
gen atmosphere. Conversions of N2O and CH4 increased
with Fe amount (Figs. 3a and 3b). It is known that CH4 is
less reactive than C2H6 and C3H6. However, N2O reduc-
tion with CH4 started at almost the same temperature as that
with C2H6 and C3H6. For example, the reduction of N2O
started at almost the same temperature (about 523 K) over
Fe(0.40)-MFI in the case of CH4 and C2H6. It is character-
istic that N2O conversion in Fig. 3a reached 100% at about
700 K over Fe(0.40)-MFI in the case of N2O reduction with
CH4. In contrast, in the case of N2O reduction with C2H6
and C3H6, N2O conversion did not reach about 100% be-
low ca. 760 K (Figs. 1a and 2a). In addition, it should be
pointed out that there was a plateau in CH4 conversion at
which N2O conversion reached 100% (Fig. 3b), especially
over Fe-MFI (Fe/Al = 0.40). This indicates that CH4 cannot
react directly with O2, even at 700 K. These results indicate
that CH4 is the most efficient reductant for the SCR of N2O.
Since the feeding ratio of hydrocarbons to N2O is higher
CH4 + 4[O] → CO2 + 2H2O,
CH4 + 3[O] → CO + 2H2O,
[O]: oxygen atom originating from N2O and O2.
The details of C2H6 and C3H6 are referred to in the cap-
tion of Fig. 4. On both Fe-MFI catalysts, the N2O contri-
bution to the N2O/CH4/O2 system is higher than that in the
case of the N2O/C2H6/O2 and N2O/C3H6/O2 systems. This
indicates that CH4 reacted with N2O more efficiently than
did C2H6 and C3H6. In other words, a higher concentration
of reductants must be supplied for higher N2O conversion
and removal in the case of C2H6 and C3H6 compared with
CH4. From these viewpoints, CH4 is a more suitable reduc-
tant when N2O is removed from containing air. In terms of
the practical aspect, the suitability of hydrocarbons as a N2O

T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
379
Fig. 4. N O contribution in total oxidizing agents during N O reduction with various hydrocarbons under excess oxygen atmosphere: (a) Fe(0.40)-MFI and
2
2
(b) Fe(0.10)-MFI. ": CH , Q: C H , and 2: C H . The reaction of CO and CO formation is assumed below, C H + (m + n/2)[O] → mCO + (n/2)H O,
m
n
4
2
6
3
6
2
2
C
H
+ (2m + n/2)[O] → mCO + (n/2)H O. On the basis of the assumptions, N O contribution is calculated as follows. N O contribution = (consumed
m
n
2
2
2
2
N O)/((1 + n/2m)CO + (2 + n/2m)CO ).
2
2
Fig. 5. Reaction temperature dependence of catalyst performance of Fe-MFI catalyst in N O decomposition in the absence of O : (a) N O conversion;
2
2
2
": Fe(0.40)-MFI, Q: Fe(0.24)-MFI, 2: Fe(0.15)-MFI, a: Fe(0.10)-MFI, and F: Fe(0.05)-MFI. (b) TOF as a function of molar ratio of Fe/Al; !: 673 K,
1: 723 K. Gas composition: 950 ppm N O (He balance).
2
reductant can be estimated as follows: CH4 > C2H6 > C3H6
on the basis of the order of N2O contribution. In addition,
N2O conversion was almost proportional to CH4 conversion,
as shown in Fig. 3d, and the line goes through the origin.
This also suggests that N2O is necessary for the activation
of CH4. On the other hand, C3H6 can be activated with-
out N2O on the basis of the result in Fig. 1d. In the three
reactions, Fe-MFI with higher Fe loading exhibited higher
activity. The relation between the activity and Fe loading is
discussed in detail in the next section. From these results,
we can conclude that CH4 is superior to C2H6 and C3H6 in
N2O reduction with hydrocarbons under an excess oxygen
atmosphere.
clude the reaction of the reductant with oxygen in order to
investigate the effect of a reductant in N2O reduction. Re-
action temperature dependence in N2O decomposition over
Fe-MFI catalysts is shown in Fig. 5. These results are also
based on the activity under steady-state conditions. There-
fore, the product N2/O2 ratio was 2:1, which means that the
reaction proceeded stoichiometrically. N2O conversion was
dependent on Fe loading. Furthermore, the starting tempera-
ture was also influenced by Fe loading. This behavior can be
related to the O2 desorption temperature in O2-TPD profiles.
The starting temperature of oxygen desorption in O2-TPD
profiles became lower over Fe-MFI with higher Fe loading.
As shown in Fig. 6, the starting temperature of O2 desorption
can be estimated to be 670 K and 790 K over Fe(0.40)-MFI
and Fe(0.10)-MFI, respectively. The temperature at which
N2O decomposition starts to proceed corresponds well with
the temperatures in O2-TPD of N2O-treated catalysts. This
suggests that the rate-limiting step of N2O decomposition
is oxygen desorption from the catalyst surface. This is also
supported by previous reports [10,41,42]. Fig. 5b shows
turnover frequency (TOF) of N2O decomposition as a func-
tion of the molar ratio of Fe/Al in Fe-MFI. It is found that
3.2. Reaction of N2O with various reductants over Fe-MFI
catalysts
In the previous section we described the activity test, car-
ried out in the presence of an excess oxygen atmosphere.
We did this because N2O should be removed from the gas
containing air in the practical process. In contrast, in this
section we describe the activity test as it is carried out in the
absence of oxygen. We do this because it is possible to ex-

380
T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
Fig. 6. Temperature-programmed desorption of oxygen after O and N O treatments normalized by Fe loading amount: (a) Fe(0.40)-MFI and (b) Fe(0.10)-MFI.
2
2
Reaction temperature: from room temperature to 1273 K, 10 K/min, 30 min hold. Gas composition: He, 55 ml/min. Catalyst weight: 30 mg. Pretreatment:
100% O flow, 773 K, 1 h, or 10% N O/He flow, 773 K, 1 h.
2
2
Fig. 7. Reaction temperature dependence of N O conversion of Fe-MFI catalyst in N O reduction with H and CO in the absence of O : (a) H , (b) CO,
2
2
2
2
2
and (c) TOF as a function of molar ratio of Fe/Al calculated at 573 K (1: CO) and 623 K (P: H ). ": Fe(0.40)-MFI, Q: Fe(0.24)-MFI, 2: Fe(0.15)-MFI,
2
a: Fe(0.10)-MFI, and F: Fe(0.05)-MFI. Gas composition: (a) 1000 ppm N O, 1000 ppm H (He balance), and (b) 1000 ppm N O, 1000 ppm CO (He
2
2
2
balance).
the TOF of N2O decomposition at 673 K was almost zero
in the range of Fe/Al ꢁ 0.10; however, the TOF drastically
increased with Fe loading (Fe/Al > 0.10). The profile of
O2-TPD over Fe(0.10)-MFI gave only one peak. In con-
trast, another sharp peak at a lower desorption temperature
(∼ 873 K) appeared over Fe(0.40)-MFI (Fig. 6a). This kind
of Fe species can contribute to the high catalytic activity over
Fe(0.40)-MFI. The behavior of TOF can be explained by the
Fe species that gave the lower-temperature peak. As shown
in Fig. 6a, the amount of O2 desorption after N2O treatment
was larger than that after O2 treatment, the origin of which
has been discussed extensively in our previous report [19].
Fig. 7(a) shows the temperature dependence of N2O con-
version in N2O reduction with H2. N2O conversion in the
N2O + H2 reaction was much higher than that in N2O
decomposition. Catalytic activity and the temperature at
which the N2O + H2 reaction started was also dependent
on the loading amount of Fe. Fig. 8 shows the profiles of
temperature-programmed reduction (TPR) of Fe-MFI af-
ter O2 treatment. Fe-MFI (Fe/Al = 0.05 and 0.1) started

T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
381
ilar to that in N2O + H2. Although the difference in starting
temperatures between H2-TPR and the N2O + H2 reaction
was clearly observed, it is thought to be explained by the
pressure difference. Fig. 7b shows the result of N2O reduc-
tion with CO. Although N2O conversion in the N2O + CO
reaction was higher than that of N2O+H2 reduction, the ten-
dency of the N2O + CO reaction was similar to that of the
N2O+H2 reaction. The relations between the TOF value and
the Fe/Al ratio are shown Fig. 7(c). The TOFs of N2O + H2
and the N2O+CO reaction increased with increasing Fe/Al,
which can be interpreted on the basis of TPR profiles. More
reducible Fe species can work as the active site at lower reac-
tion temperatures. The amount of more reducible Fe species
increased with increasing Fe loading, and this can explain
the relations between the TOFs and Fe/Al.
Fig. 9 shows the temperature dependence in N2O re-
duction with CH4 in the absence of oxygen. The carbon
amount in the effluent was balanced with that introduced in
this reaction. This means that the results correspond to the
steady-state activity and coke was not deposited. Although
the details are not shown here, a large amount of coke was
deposited in the N2O + C2H6 and N2O + C3H6 reactions
in the absence of excess oxygen. From this viewpoint, the
data in the case of C2H6 and C3H6 were not available un-
der this type of conditions. CH4 conversion had a plateau
Fig. 8. Temperature-programmed reduction in hydrogen over Fe-MFI
catalysts normalized by Fe loading amount: (a) Fe(0.05)-MFI, (b) Fe
(0.10)-MFI, (c) Fe(0.15)-MFI, (d) Fe(0.24)-MFI, and (e) Fe(0.40)-MFI.
Reaction temperature: from room temperature to 973 K, 10 K/min. Gas
composition: 5.0% H /Ar, 30 ml/min. Catalyst weight: 50 mg. Pretreat-
2
ment: 100% O flow, 773 K, 1 h.
2
to be reduced in H2 at about 600 K. In contrast, Fe-MFI
(Fe/Al = 0.15–0.40)started to be reduced at a lower temper-
ature than Fe-MFI (Fe/Al = 0.05 and 0.10). For example,
the reduction started at 450 K on Fe(0.40)-MFI. This sug-
gests that Fe-MFI catalysts with higher Fe loading contain
more reducible Fe species. The tendency in H2-TPR is sim-
Fig. 9. Reaction temperature dependence of catalyst performance of Fe-MFI catalyst in N O reduction with CH in the absence of O : (a) N O conver-
2
4
2
2
sion, (b) CH conversion, (c) CO selectivity, (d) TOF calculated at 573 K. ": Fe(0.40)-MFI, Q: Fe(0.24)-MFI, 2: Fe(0.15)-MFI, a: Fe(0.10)-MFI, and
4
2
F: Fe(0.05)-MFI. Gas composition: 950 ppm N O, 500 ppm CH (He balance).
2
4

382
T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
Fig. 10. Comparison of N O conversion in N O reduction with various reductants in the absence of O over Fe-MFI catalysts: (a) Fe(0.40)-MFI and
2
2
2
(b) Fe(0.05)-MFI. ": CH , Q: CO, 2: H . Reaction conditions were shown in the figure caption in Figs. 7 and 9.
4
2
since N2O conversion reached almost 100%. These pro-
files are also observed in N2O reduction with CH4 in the
presence of excess oxygen. The temperature at which the
N2O + CH4 reaction started was also dependent on Fe load-
ing, and the tendency is also observed in the N2O + H2 and
N2O + CO reactions. Fig. 9d shows the relation between
the TOF of the N2O + CH4 reaction and Fe/Al. The TOF
value jumped at Fe/Al = 0.15 and increased with increasing
Fe/Al. This TOF behavior is similar to that in the N2O + H2
and N2O + CO reactions. It is suggested that more reducible
Fe species can also contribute to the enhancement of TOF in
the case of the N2O + CH4 reaction.
tra. The Fourier transforms (FTs) of k3-weighted EXAFS os-
cillations for Fe(0.10)-MFI are shown in Fig. 11c, and their
fitting results are shown in Figs. 11d and 11e. The details of
the curve-fitting parameters are listed in Table 2. Two major
peaks were observed in the regions of 0.1–0.2 nm and 0.2–
0.3 nm in FT spectra. One peak with a shorter bond length
of 0.1–0.2 nm can be assigned to the Fe–O bond, which has
already been reported in EXAFS studies [25,27,43]. And
the other peak, with a longer bond length of 0.2–0.3 nm,
can be assigned to the Fe–Si bond, as shown later. In the
O2-TPD profile of Fe(0.10)-MFI, the amount of O2 desorp-
tion from O2- and N2O-treated samples is estimated to be
O2/Fe = 0.16 and 0.17, respectively. These values are less
than O2/Fe = 0.25, which is based on the equation
Fig. 10 shows the comparison of N2O conversion in vari-
ous reactions over Fe(0.40)-MFI and Fe(0.05)-MFI. The or-
der of N2O conversion over Fe(0.40)-MFI and Fe(0.05)-MFI
was CH4 > CO > H2 in all temperature ranges. It should be
noted that the N2O + CH4 reaction gave the highest N2O
conversion over all of the catalysts and in all temperature
ranges. Considering that the reactivity of methane is gener-
ally much lower than those of H2 and CO, it is very interest-
ing that methane has a higher reactivity than H2 and CO for
N2O reduction. According to the results of TOF in various
reactions, the structure of the active sites is almost the same
on Fe-MFI (Fe/Al = 0.05 and 0.10) catalysts, and other ac-
tive sites appear on Fe-MFI (Fe/Al = 0.15–0.40), which are
more reducible iron species.
4Fe3+ + 2O2− → 4Fe2+ + O2.
This suggests that a part of Fe can produce oxygen desorp-
tion, and the other cannot. Therefore, it is expected that there
are two kinds of Fe–O bond on the basis of the amount of O2
desorption. This suggestion led us to use two kinds of Fe–O
bond in the curve-fitting analysis (Table 2).
The Fe–O bond, with a shorter distance of 0.187 nm,
can be ascribed to O atoms of the OH species coordinated
with Fe ion species. In previous reports, Koningsberger et al.
[25,26] assigned an Fe–O distance of 0.188 nm to a terminal
OH group and an Fe–O distance of 0.193 nm to the bridg-
ing Fe–O–Fe oxygen atom. The Fe–O bond, with a longer
distance of 0.204 nm, is ascribed to O atoms in the zeolite
lattice, by which the Fe atoms are stabilized to the zeolite
lattice as Fe–O–Si and Fe–O–Al, and it is expected that the
oxygen species are difficult to desorb. This is supported by
previous reports [25,28].
From the fitting results of EXAFS data, we could fit our
spectra with the Fe–Si shell. Joyner et al. [33] have reported
that the distance of the Fe–Si shell is about 0.32 nm, which
is almost the same as our result. On the other hand, Choi
et al. [43] reported EXAFS analysis of Fe-ZSM-5 prepared
by solid ion exchange, and they assigned the peak between
0.25 and 0.30 nm to the Fe–Al bond. Generally speaking, it
is difficult to distinguish between Al and Si as a backscat-
tering atom because the difference in atomic number is very
3.3. Fe K-edge EXAFS analysis after different treatment
We carried out a structural analysis of the Fe(0.10)-MFI
and Fe(0.40)-MFI catalysts with some treatments on the
basis of the results from O2-TPD by means of EXAFS.
Figs. 11a and 11b show the Fe K-edge EXAFS oscillations
for Fe(0.10)-MFI measured after treatment with O2 or N2O.
It seems that the oscillation on Fe(0.10)-MFI treated with O2
is similar to that with N2O. As shown in Fig. 6b, only one O2
desorption peak was observed on Fe(0.10)-MFI, and the de-
sorption profile of the sample after O2 treatment was almost
the same as that after N2O treatment. This means that the ad-
ditional oxygen deposition by N2O treatment did not occur
on Fe(0.10)-MFI. This can explain the similarity of the spec-

T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
383
3
3
Fig. 11. (a, b): k -weighted Fe K-edge EXAFS for Fe(0.10)-MFI catalyst after treatment with (a) O or (b) N O. (c): Fourier transform of k -weighted Fe
2
2
−1
−1
K-edge EXAFS for Fe(0.10)-MFI catalyst after treatment with O (Fourier transform range: 25–115 nm ) or N O (Fourier transform range: 23–115 nm ).
2
2
(d, e): Fourier filtered EXAFS data (solid line) and calculated data (dotted line) of Fe(0.10)-MFI after treatment with (d) O or (e) N O. Fitting results are
2
2
listed in Table 2.
Table 2
Fe K-edge EXAFS fitting results for Fe(0.10)-MFI catalyst pretreated under different conditions
a
b
−1
c
−1
d
e
R
Shells
CN
R
(×10 nm)
σ
(×10 nm)
ꢀE (eV)
(%)
f
0
f
Fitting results of Fe(0.10)-MFI after O treatment
2
Fe–O
Fe–O
Fe–Si
1.7 0.1
1.7 0.1
1.9 0.2
1.87 0.01
2.04 0.01
3.17 0.01
0.041 0.012
0.045 0.015
0.058 0.018
−0.1 0.8
3.8 0.9
−5.6 1.0
0.69
1
2
g
Fitting results of Fe(0.10)-MFI after N O treatment
2
Fe–O
Fe–O
Fe–Si
1.6 0.1
2.0 0.1
1.9 0.2
1.88 0.01
2.04 0.01
3.19 0.01
0.064 0.011
0.070 0.011
0.066 0.018
−5.8 1.0
3.1 0.9
−4.7 1.1
1.09
1
2
a
Coordination number.
b
c
d
e
f
Distance.
Debye–Waller factor.
Difference in the origin of photoelectron energy between the reference and the sample.
Residual factor.
Fourier filtering range: 0.083–0.304 nm.
g
Fourier filtering range: 0.077–0.304 nm.

384
T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
3
3
Fig. 12. k -weighted Fe K-edge EXAFS for Fe(0.40)-MFI catalyst after treatments with (a) O , (b) N O, and (c) H . (d) Fourier transform of k -weighted
2
2
2
−1
Fe K-edge EXAFS for Fe(0.40)-MFI catalyst after treatments with O , N O, and H . Fourier transform range: 24–115 nm
.
2
2
2
small. However, this tendency can be explained by the local
structure of Fe on MFI. It has been proposed that an Fe ion
over the acid site is neighboring one Al and two or more Si
atoms through the lattice oxygen atoms. This suggests that
the Si atoms make a greater contribution than the Al atom.
The possibility of an Fe–Fe bond over Fe(0.10)-MFI cannot
be accepted, because this catalyst contains Fe ions almost
exclusively, with lower reducibility from H2-TPR and O2-
TPD results (Figs. 6b and 8). In addition, the Fe loading
of the Fe(0.10)-MFI catalyst is low, and because the cata-
lyst was prepared by severely controlled wet ion exchange,
it is expected that the Fe species over the Fe(0.10)-MFI cat-
alyst can be isolated. In fact, it is difficult to fit the FT peak
with one wave of Fe–Fe. From these comparisons, for the
Fe(0.10)-MFI after O2 treatment, three-shell fitting (Fe–O1,
Fe–O2, Fe–Si) gave the appropriate result. The comparison
of the results for the O2 and N2O treatments shows that the
coordination number and the distance were almost the same
values. This also shows that N2O treatment did not lead to a
structural change in the Fe species.
higher k region. In addition, it was also found that the oscil-
lations of Fe(0.40)-MFI are similar to that of Fe(0.10)-MFI
in the case of O2 treatment. The FT spectra of k3-weighted
EXAFS oscillations for Fe(0.40)-MFI with different treat-
ments are shown in Fig. 12d. Their fitting results are shown
in Figs. 13a–13c and listed in Table 3. Regarding Fe(0.40)-
MFI after O2 treatment, the curve fitting was carried out
on the basis of the results of Fe(0.10)-MFI because of the
similarities in EXAFS oscillations. Curve-fitting results for
these two samples were also similar. In addition, a different
point is the length of the short Fe–O1 bond. It seems that the
length of the Fe–O1 bond in Fe(0.40)-MFI is a little longer
than that in Fe(0.10)-MFI. Koningsberger et al. [26] have
reported that the Fe–O distance of Fe–O–Fe bridged oxy-
gen (0.193 nm) is longer than that of terminal OH species
(0.188 nm). In our case, although the difference in the bond
distance is not so large, we think that the contribution of the
bridged oxygen species can be added to the bond of Fe–O1
on Fe(0.40)-MFI. This interpretation can be supported by
the O2-TPD profiles (Fig. 6). In the case of Fe(0.40)-MFI
after O2 treatment, a desorption peak was observed at lower
temperatures, and this was not observed over Fe(0.10)-MFI.
This indicates that Fe(0.40)-MFI has oxygen species dif-
ferent from that of Fe(0.10)-MFI, which can be desorbed
Figs. 12a–12c show EXAFS oscillations for Fe(0.40)-
MFI after the treatments with O2, N2O, and H2. It is found
that the oscillation of Fe(0.40)-MFI after N2O treatment
is different from that after O2 treatment, especially in the

T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
385
Fig. 13. Fourier filtered EXAFS data (solid line) and calculated data (dotted line) of Fe(0.40)-MFI after treatments with (a) O , (b) N O, and (c) H . Fitting
2
2
2
results are listed in Table 3.
Table 3
Fe K-edge EXAFS fitting results for Fe(0.40)-MFI catalyst pretreated under different conditions
a
b
−1
c
−1
d
e
R
Shells
CN
R
(×10 nm)
σ
(×10 nm)
ꢀE (eV)
(%)
f
0
f
Fe(0.40)-MFI after O treatment
2
Fe–O
Fe–O
Fe–Si
1.4 0.1
2.5 0.2
1.5 0.2
1.90 0.01
2.05 0.01
3.19 0.01
0.063 0.013
0.078 0.010
0.060 0.023
−4.3 1.1
4.7 0.8
−5.3 1.4
0.54
1
2
g
Fe(0.40)-MFI after N O treatment
2
Fe–O
Fe–O
Fe–Si
Fe–Fe
2.7 0.1
2.3 0.1
1.5 0.2
0.6 0.1
1.89 0.01
2.05 0.01
3.19 0.01
2.95 0.01
0.077 0.007
0.066 0.010
0.057 0.022
0.059 0.021
5.5 0.7
0.6 0.8
0.44
1
2
−9.8 1.3
3.0 1.7
h
Fe(0.40)-MFI after H treatment
2
Fe–O
Fe–O
Fe–Si
Fe–Fe
0.9 0.1
2.3 0.2
0.9 0.2
0.4 0.1
1.89 0.01
2.06 0.01
3.22 0.02
2.91 0.02
0.077 0.023
0.074 0.010
0.064 0.039
0.060 0.031
4.6 2.1
0.6 0.8
0.90
1
2
−2.9 2.3
−13.7 2.6
a
Coordination number.
b
c
d
e
f
Distance.
Debye–Waller factor.
Difference in the origin of photoelectron energy between the reference and the sample.
Residual factor.
Fourier filtering range: 0.077–0.298 nm.
Fourier filtering range: 0.083–0.304 nm.
Fourier filtering range: 0.080–0.304 nm.
g
h
more easily. In the case of Fe(0.40)-MFI, N2O treatment can
make the Fe structure change, and this is not the case with
Fe(0.10)-MFI. The difference was clearly observed in the
FT spectra. N2O treatment decreased the peak intensity at
0.1–0.2 nm, and, in contrast, it increased the peak intensity
at 0.2–0.3 nm. N2O treatment increased the O2 desorption
amount from the TPD profiles; however, the peak intensity
due to the Fe–O bond decreased. According to our curve-
fitting analysis, the increase in the contribution of Fe–O1 can
explain this behavior. On the other hand, in the FT peak at
the larger region, we added the Fe–Fe bond, because of the
oscillation difference at the higher k region. Four-shell fitting

386
T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
of Fe(0.40)-MFI after N2O treatment requires 16 parame-
ters, and this cannot be allowed on the basis of the limitation
of the independent parameters (NI) [44]. Here, NI can be
calculated to be 15.3 from ꢀk = 91 nm−1, ꢀR = 0.23 nm.
For the profile of Fe(0.40)-MFI after O2 treatment, the best
fitting result was obtained by a three-shell fitting. However,
we could not obtain good fitting results for Fe(0.40)-MFI af-
ter N2O treatment by the three-shell fitting. As a result, the
curve fitting was carried out with two Fe–O, one Fe–Si, and
one Fe–Fe shells. The results are listed in Table 3. Consid-
ering also other characterization results from H2-TPR and
O2-TPD, it is strongly suggested that binuclear Fe species
are formed on Fe(0.40)-MFI.
an octahedral environment. These results also support that
Fe binuclear species are present on our Fe(0.40)-MFI cat-
alyst. However, we could not observe the Fe–Fe bond in
Fe(0.40)-MFI after O2 treatment. It seems that some of the
bridged oxygen on binuclear Fe species is not present on
Fe(0.40)-MFI after O2 treatment. In contrast, an Fe–Fe bond
is formed by N2O treatment, and this is due to bridging of
two Fe ions induced by deposited oxygen during the N2O
treatment. From the H2-TPR profile, it is found that all of
the iron species can be reduced from Fe3+ to Fe2+ with H2
reduction at 773 K. After H2 treatment (Fig. 13c), the co-
ordination number of Fe–O1 decreased because of oxygen
removal during H2 treatment (see Table 3). This behavior is
observed more significantly in Fe–O1. On the other hand,
the coordination number of the Fe–O2 shell, which can be
assigned to the bond between Fe and lattice oxygen, was al-
most constant. This can be inferred from the presence of the
oxygen species, which is hardly removed even after H2 treat-
ment.
From the results of EXAFS studies, many research
groups have reported various structures of Fe species pre-
pared by different methods [24–33]. Sachtler et al. [30–32]
reported that Fe/MFI prepared by a sublimation method has
a molar ratio of Fe to Al-centered tetrahedral of 1:1 and
that active species over MFI are oxygen-bridged binuclear
iron species. Koningsberger et al. [24–26] determined the
structure of the Fe binuclear complex with Fe–O–Fe bridges
on Fe/ZSM-5 catalyst, which was prepared with FeCl3 by
EXAFS. Prins et al. [27,28] reported that the diiron structure
resembles the core unit in methane monooxygenase (MMO).
In addition, Panov et al. [29] indicated that α-oxygen can
participate in the direct catalytic oxidation of benzene to
phenol by N2O over Fe-MFI, and the oxygen species are re-
lated to the presence of binuclear Fe complex. These reports
suggest that the Fe species with the coordination of 0.25–
0.30 nm are binuclear. On the other hand, Joyner et al. [33]
reported the presence of very small iron–oxygen clusters of
unusual structure such as Fe4O4, with a short iron–iron dis-
tance of ca. 0.25 nm. However, we could not observe the
short Fe–Fe bond in EXAFS spectra. This can be interpreted
that there are no iron-oxo nanoclusters such as Fe4O4 on
our Fe-MFI catalysts, and this is also supported by H2-TPR,
in which no H2 consumption due to Fe oxide was observed
(see Fig. 8). In the previous H2-TPR study, H2 consumption
with the peak at 850 K was observed on Fe2O3 [19]. Al-
though the presence of oligonuclear Fe species is possible
(binuclear, trinuclear, etc.), highly aggregated oligonuclear
Fe species can be ruled out, because the Fe–Fe coordina-
tion number is determined to be 0.6 after N2O treatment,
which is clearly less than unity (Table 3). This also indicates
that binuclear Fe species and isolated Fe ions are present.
This interpretation is also supported by O2-TPD and H2-
TPR: binuclear iron species are more reducible, and they
give lower-temperature O2 desorption. In addition, the co-
ordination number of the Fe–O1 shell was larger after N2O
treatment than that after O2 treatment. This behavior can be
explained by oxygen deposition on the Fe species by N2O
treatment (see Fig. 6).
3.4. Relation between catalyst structure and performance
in N2O reduction
From the results of EXAFS analysis combined with other
characterization, it is found that Fe(0.40)-MFI contains bin-
uclear and mononuclear Fe species, and Fe(0.10)-MFI con-
tains only a mononuclear species. In the case of N2O reduc-
tion with H2, CO, and CH4, TOF increased with increasing
Fe loading, and this can be due to the formation of binuclear
Fe species. From the comparison with H2-TPR and O2-TPD,
more reducible Fe species are also assigned to binuclear Fe
species.
The effect of reductants on N2O reduction activity over
Fe(0.40)-MFI is interesting in terms of the effectiveness of
methane. Generally speaking, the reactivity of methane is
very low compared with H2 and CO. However, methane can
play an important role for the reductant at almost the same
temperature range as H2 and CO. Pérez-Ramírez et al. [45]
proposed that the N2O + CO reaction proceeds by a differ-
ent mechanism on the different Fe ion species. Mononuclear
Fe3+ species participated in this reaction via coordinated
CO species on Fe3+ ions, and their oxidation state was not
changed during the reaction. The reaction over oligonuclear
FexOy clusters proceeds via a redox Fe3+/Fe2+ process
through the formation of O− species as a reaction intermedi-
ate. Here, the Fe3+ in oligonuclear FexOy clusters can easily
be reduced by CO. Therefore, they have concluded that ac-
tive sites of N2O+CO reaction are oligonuclear FexOy clus-
ters [45]. In our case, no formation of oligonuclear FexOy
clusters was observed on the basis of H2-TPR. The structure
sensitivity of N2O + CO in our case can be explained by the
reducibility of binuclear Fe species.
The sum of Fe–O coordination is 5.0, which indicates
Fe ion species are octahedrally coordinated with zeolite lat-
tice oxygen, bridged oxygen, and OH ligands. In a previ-
ous report [25] it is shown that Fe binuclear species have
We can also conclude that the N2O + H2 reaction is also
structure-sensitive and the behavior is similar to that of the
N2O + CO reaction. In the case of the N2O + H2 reac-
tion, the structure sensitivity can be related to the results

T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
387
from H2-TPR (Fig. 8), which indicate that the reducibility
Acknowledgments
of Fe species increased with increasing Fe loading and the
formation of Fe binuclear species. The structure sensitivity
can also be inferred from the more reducible Fe binuclear
species, which gives higher TOF.
A part of this research was supported by the Research In-
stitute of Innovative Technology for the Earth (RITE) and
a Center of Excellence Grant-in-aid for the 21st Century
COE Program for Frontiers from the Ministry of Education,
Culture, Sports, Science and Technology. This work was per-
formed under the approval of the Photon Factory Advisory
Committee (Proposal No. 2003G255). The authors are grate-
ful to the Chemical Analysis Center, University of Tsukuba,
for ICP analysis of Fe-MFI catalysts.
In this work, we have shown that CH4 is the most effec-
tive reductant in the presence and absence of excess oxygen.
From the result of in situ FTIR study [20], we observed
methoxy species formation during the N2O + CH4 reaction
on a Fe-BEA catalyst. From the result of pulse reactions, we
have proposed that formation of methoxy species promotes
the reduction of Fe ion species on the active sites, which are
suggested to be binuclear Fe species [19]. In addition, higher
TOF was observed in the N2O + CH4 reaction (Fig. 9d) over
Fe-MFI with higher Fe loading. These results indicate that
the structure sensitivity of the N2O + CH4 reaction can be
related to the formation of methoxy species from methane
activated with oxygen supplied by N2O dissociation over
binuclear Fe species [19].
References
[1] F. Kapteijn, J. Rodrigues-Mirasol, J.A. Moulijn, Appl. Catal. B 9
(1996) 25.
[2] M. Nakamura, H. Mitsuhashi, N. Takezawa, J. Catal. 138 (1992) 686.
[3] S. Tanaka, K. Yuzaki, S. Ito, S. Kameoka, K. Kunimori, J. Catal. 200
(2001) 203.
[4] X.F. Wang, H.C. Zeng, Appl. Catal. B 17 (1998) 89.
[5] G. Centi, L. Dall’Olio, S. Perathoner, J. Catal. 192 (2000) 224.
[6] Y. Li, J.N. Armor, Appl. Catal. B 1 (1992) L21.
[7] T. Turek, Appl. Catal. B 9 (1996) 201.
[8] J. Leglise, J.O. Petunchi, W.K. Hall, J. Catal. 86 (1984) 392.
[9] T. Nobukawa, S. Tanaka, S. Ito, K. Tomishige, S. Kameoka, K. Kuni-
mori, Catal. Lett. 83 (2002) 5.
4. Conclusions
[10] J. Pérez-Ramírez, F. Kapteijn, G. Mul, J.A. Moulijn, J. Catal. 208
(2002) 211.
[11] G. Centi, F. Vazzana, Catal. Today 53 (1999) 683.
[12] R.W. van den Brink, S. Booneveld, J.R. Pels, D.F. Bakker, M.J.F.M.
Verhaak, Appl. Catal. B 32 (2001) 73.
[13] K. Yamada, S. Kondo, K. Segawa, Mocropor. Mesopor. Mater. 35–36
(2000) 227.
[14] A. Satsuma, H. Maeshima, K. Watanabe, K. Suzuki, T. Hattori, Catal.
Today 63 (2000) 347.
1. In the selective catalytic reduction (SCR) of N2O with
hydrocarbons in the presence of excess O2, CH4 and
C2H6 were not activated by O2 but by N2O, although
C3H6 could react with oxygen. The order of N2O con-
tribution in the SCR under excess oxygen is CH4 >
C2H6 > C3H6. This means that CH4 is a more efficient
reductant for the SCR of N2O.
[15] S. Kameoka, T. Suzuki, K. Yuzaki, T. Takeda, S. Tanaka, S. Ito, T. Mi-
yadera, K. Kunimori, Chem. Commun. (2000) 745.
[16] S. Kameoka, K. Yuzaki, T. Takeda, S. Tanaka, S. Ito, T. Miyadera,
K. Kunimori, Phys. Chem. Chem. Phys. 3 (2001) 256.
[17] S. Kameoka, K. Kita, S. Tanaka, T. Nobukawa, S. Ito, K. Tomishige,
T. Miyadera, K. Kunimori, Catal. Lett. 79 (2002) 63.
[18] S. Kameoka, T. Nobukawa, S. Tanaka, S. Ito, K. Tomishige, K. Kuni-
mori, Phys. Chem. Chem. Phys. 5 (2003) 3328.
[19] M. Yoshida, T. Nobukawa, S. Ito, K. Tomishige, K. Kunimori, J. Ca-
tal. 223 (2004) 454.
[20] T. Nobukawa, M. Yoshida, S. Kameoka, S. Ito, K. Tomishige, K. Ku-
nimori, J. Phys. Chem. B 108 (2004) 4071.
[21] B. Coq, M. Mauvezin, G. Delahay, S. Kieger, J. Catal. 195 (2000) 298.
[22] G. Delahay, M. Mauvezin, B. Coq, S. Kieger, J. Catal. 202 (2001) 156.
[23] G. Delahay, M. Mauvezin, A. Guzmán-Vargas, B. Coq, Catal. Com-
mun. 3 (2002) 385.
[24] A.A. Battiston, J.H. Bitter, F.M.F. de Groot, A.R. Overweg, O. Ste-
phan, J.A. van Bokhoven, P.J. Kooyan, C. van der Spek, G. Vankó,
D.C. Koningsberger, J. Catal. 213 (2003) 251.
[25] A.A. Battiston, J.H. Bitter, W.M. Heijboer, F.M.F. de Groot, D.C. Kon-
ingsberger, J. Catal. 215 (2003) 279.
[26] A.A. Battiston, J.H. Bitter, D.C. Koningsberger, J. Catal. 218 (2003)
163.
2. The TOFs of N2O reduction by H2, CO, and CH4 and
N2O decomposition in the absence of oxygen increased
with increasing molar ratio of Fe/Al, when Fe/Al is
above 0.15, whereas TOFs were lower and constant in
the range of Fe/Al ꢁ 0.10.
3. From H2-TPR, more reducible Fe species were formed
over Fe-MFI with higher Fe loading. In O2-TPD,
Fe(0.40)-MFI had oxygen species that could be de-
sorbed at lower temperatures than Fe(0.10)-MFI.
4. From the result of EXAFS analysis, only mononuclear
Fe species were observed over Fe(0.10)-MFI after treat-
ment with O2 or N2O. On the other hand, binuclear Fe
species and mononuclear Fe species were observed over
Fe(0.40)-MFI after treatment with N2O or H2. More re-
ducible Fe species, which gave lower-temperature O2
desorption, can be due to Fe binuclear species over
Fe(0.40)-MFI.
5. Although the reactivity of methane is usually very low,
CH4 can be oxidized by N2O more easily than by H2
and CO. Formation of methoxy species on the Fe ion
species promotes the reduction of Fe ion sites even in
excess oxygen, and this can promote the redox cycle of
binuclear Fe species.
[27] P. Marturano, L. Drozdová, A. Kogelbauer, R. Prins, J. Catal. 192
(2000) 236.
[28] P. Marturano, L. Drozdová, G.D. Pirngruber, A. Kogelbauer, R. Prins,
Phys. Chem. Chem. Phys. 3 (2001) 5585.
[29] K.A. Dubkov, N.S. Ovanesyan, A.A. Shteinman, E.V. Starokon, G.I.
Panov, J. Catal. 207 (2002) 341.

388
T. Nobukawa et al. / Journal of Catalysis 229 (2005) 374–388
[30] H.-Y. Chen, W.M.H. Sachtler, Catal. Today 42 (1998) 73.
[31] H.-Y. Chen, El-M. El-Malki, X. Wang, R.A. van Santen, W.M.H.
Sachtler, J. Mol. Catal. A 162 (2000) 159.
[32] El-M. El-Malki, R.A. van Santen, W.M.H. Sachtler, J. Catal. 196
(2000) 212.
[33] R. Joyner, M. Stockenhuber, J. Phys. Chem. B 103 (1999) 5963.
[34] W.N. Delgass, R.L. Garten, M. Boudart, J. Chem. Phys. 50 (1969)
4603.
[38] L. Ankudinov, B. Ravel, J.J. Rehr, S.D. Conradson, Phys. Rev.
B (1998) 7565.
[39] B. Ravel, version 3.0beta8 4 March, The university of Washington,
2003.
[40] S. Kameoka, K. Yuzaki, T. Takeda, S. Tanaka, S. Ito, T. Miyadera,
K. Kunimori, Phys. Chem. Chem. Phys. 3 (2001) 256.
[41] D.A. Bulushev, L. Kiwi-Minsker, A. Renken, J. Catal. 222 (2004) 389.
[42] G.D. Pirngruber, J. Catal. 219 (2003) 456.
[35] J.W. Cook, D.E. Sayers, J. Appl. Phys. 52 (1981) 5024.
[36] K. Okumura, J. Amano, N. Yasunobu, M. Niwa, J. Phys. Chem. B 104
(2000) 1050.
[37] K. Okumura, S. Matsumoto, N. Nishiaki, M. Niwa, Appl. Catal. B 40
(2003) 151.
[43] S.H. Choi, B.R. Wood, J.A. Ryder, A.T. Bell, J. Phys. Chem. B 107
(2003) 11843.
[44] E.A. Stern, Phys. Rev. B 48 (1993) 9825.
[45] J. Pérez-Ramírez, M.S. Kumar, A. Brückner, J. Catal. 223 (2004) 13.