The positive effect of NO on the N2O decomposition activity of Fe-ZSM-5: A combined kinetic and in situ IR spectroscopic study

Article abstract of DOI:10.1016/j.jcat.2005.11.012

NO-assisted N₂O decomposition over four different Fe-ZSM-5 samples prepared by wet ion exchange (WIE) or chemical vapor deposition (CVD) of FeCl₃ was investigated by steady-state kinetics, in situ infrared (IR) spectroscopy, and transient response methods. Despite their lower iron loading, the samples prepared by WIE had the highest activity for N₂O decomposition in the presence of NO. The in situ IR experiments showed that the most active sample was characterized by a high concentration of adsorbed NO, as well as adsorbed NO₂ under reaction conditions. Step response experiments proved that NO₂ is an intermediate of the catalytic cycle and functions as intermediate oxygen storage. IR and transient kinetic experiments showed that WIE catalysts behave qualitatively different in NO-assisted N₂O decomposition than CVD catalysts. These differences are discussed in terms of the different structure of the iron species in the two types of samples.

Full text of DOI:10.1016/j.jcat.2005.11.012

Journal of Catalysis 237 (2006) 237–247
www.elsevier.com/locate/jcat
The positive effect of NO on the N2O decomposition activity of Fe-ZSM-5:
A combined kinetic and in situ IR spectroscopic study
Gerhard D. Pirngruber a,∗, Johannis A.Z. Pieterse b
a
Institute for Chemical and Bioengineering, Swiss Federal Institute of Technology (ETH), Zürich, Wolfgang-Pauli Str. 10, CH-8093 Zürich, Switzerland
b
ECN Clean Fossil Fuels, P.O. Box 1, Westerduinweg 3, NL-1755 ZG Petten, The Netherlands
Received 19 September 2005; revised 11 November 2005; accepted 13 November 2005
Abstract
NO-assisted N O decomposition over four different Fe-ZSM-5 samples prepared by wet ion exchange (WIE) or chemical vapor deposition
2
(CVD) of FeCl was investigated by steady-state kinetics, in situ infrared (IR) spectroscopy, and transient response methods. Despite their lower
3
iron loading, the samples prepared by WIE had the highest activity for N O decomposition in the presence of NO. The in situ IR experiments
2
showed that the most active sample was characterized by a high concentration of adsorbed NO, as well as adsorbed NO under reaction conditions.
2
Step response experiments proved that NO is an intermediate of the catalytic cycle and functions as intermediate oxygen storage. IR and transient
2
kinetic experiments showed that WIE catalysts behave qualitatively different in NO-assisted N O decomposition than CVD catalysts. These
2
differences are discussed in terms of the different structure of the iron species in the two types of samples.
2005 Elsevier Inc. All rights reserved.
Keywords: Fe-ZSM-5; CVD; Wet ion exchange; N O decomposition; NO; In situ IR; Step response; Pulse response
2
1. Introduction
of NO increased N2O conversion more than stoichiometrically.
A redox mechanism with nitrites and nitrates as the redox
species was given, but no spectroscopic evidence was provided:
N2O emissions make a major contribution (6%) to the global
greenhouse effect [1]. A considerable fraction of the total N2O
emissions are caused by the chemical industry in adipic acid
and nitric acid plants. Abatement of these emissions by cat-
alytic decomposition of N2O is an important issue. Noble metal
catalysts have very high intrinsic activities for N2O decompo-
sition [2]. But in the exhaust gas stream of adipic and nitric
acid plants, N2O is present together with NO and other compo-
nents, such as O2, H2O, and so on. Most noble metal catalysts
are strongly deactivated by the presence of these other com-
ponents, particularly NO [3,4]. Iron zeolites show the opposite
behavior, with N2O decomposition activity strongly influenced
by the presence of NO in the feed. This was originally attributed
to a stoichiometric reaction of NO with N2O, yielding N2 and
NO2 [5], but later it was shown that NO has a truly catalytic
effect on the N2O decomposition activity [6]. Small amounts
−
N2O + ∗–NO3− → ∗–NO2− + N2 + O2,
(1)
(2)
−
N2O + ∗–NO2 → N2 + ∗–NO3
.
Mul et al. [7] observed that a small amount of NO led to an
increase in the conversion of N2O by significantly more than
one N2O per NO added (as would be the case for the stoichio-
metric reaction). The effect of NO was explained by a different
model [8]: NO that is adsorbed on the catalyst reacts with the
oxygen atom that N2O deposits on the catalyst surface and
forms adsorbed NO2. Thereby, the site for the activation N2O
is liberated, and a second oxygen atom from N2O is deposited
there. O2 desorption occurs by recombination of the second
deposited oxygen atom with an oxygen atom from NO2. The
following reaction equations show that the role of NO in the
cycle in purely catalytic; it functions as an intermediate storage
for deposited oxygen atoms:
∗ + N2O → ∗–O + N2,
(3)
(4)
*
Corresponding author. Fax: +41 1 632 1162.
E-mail address: pirngruber@chem.ethz.ch (G.D. Pirngruber).
∗–O + ∗–NO → ∗ + ∗–NO2,
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2005.11.012
2005 Elsevier Inc. All rights reserved.

238
G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
∗ + N2O → ∗ –O + N2,
∗–O + ∗–NO2 → ∗ + ∗–NO + O2,
(5)
Table 1
Elemental composition of the samples and fraction of the Brønsted OH groups
exchanged against Fe, as determined by IR spectroscopy
(6)
Sample
Fe (wt%)
Fe/Al
OH exchange
Sum: 2N2O → 2N2 + O2.
Fe-ZSM-5-24 WIE
Fe-ZSM-5-24 CVD
Fe-ZSM-5-12 WIE
Fe-ZSM-5-12 CVD
1.0
4.4
2.8
5.4
0.26
1.1
0.45
0.87
0.25
0.38
0.47
0.34
The beneficial role of NO2 in O2 formation is reminiscent of
mechanistic studies of NO decomposition over copper zeo-
lites [9,10]. In this system NO2 enhances the rate of O2 for-
mation by facilitating the recombination of two remote surface
oxygen atoms via transport in the gas phase, that is,
washing and another calcination at 773 K [13]. WIE was car-
ried out with FeSO4 ·7H2O at 353 K for 6 h in N2 atmosphere,
to avoid oxidation of Fe2+ to Fe3+. Exchange with Fe2+ ions
is preferred over exchange with Fe3+ ions, because Fe3+ ions
tend to precipitate as Fe2O3 clusters on the outer surface of the
zeolite. The exchange solution included 1 mmol of iron per 1 g
of NH4-ZSM-5, Si/Al 12 and 0.5 mmol of iron per 1 g of NH4-
ZSM-5, Si/Al 24. For filtration, 2 × 1000 mL of demiwater per
10 g of cake was poured on the cake while keeping the cake un-
der vacuum. The samples were subsequently dried in an oven
at 353 K, then calcined at 773 K for 5 h (at a heating rate of
1 K/min) in static air under shallow bed conditions. During
calcination, oxidation of the exchanged Fe2+ ions to Fe3+ oc-
curs. Ultraviolet–visible light (UV–vis) spectra of the calcined
iron zeolites were recorded in diffuse reflection on a Cary 400
UV–vis spectrometer equipped with a Praying Mantis sample
stage from Harrick. The Fe and Al content of the samples was
determined by atomic absorption spectroscopy after the zeolites
were dissolved in HF and HNO3 (see Table 1).
∗–O + NO → ∗ + NO2,
(4a)
(6a)
NO2 +∗–O → ∗ + NO + O2.
This sequence is identical to (4) and (6), but involving NO2 in
the gas phase instead of adsorbed NO2.
In contrast, conventional N2O decomposition occurs via the
recombination of two deposited oxygen atoms on the surface,
that is,
2∗ + 2N2O → 2∗–O + 2N2,
(7)
(8)
2∗–O → O2.
The enhancement of the N2O decomposition activity by NO de-
pends on the relative rate of the two catalytic cycles. Recently,
Pieterse et al. [11] studied N2O decomposition in the presence
of NO, oxygen, and water over Fe-ZSM-5 samples prepared
by wet ion exchange (WIE), chemical vapor deposition (CVD),
and steam-activated isomorphous substitution. These authors
concluded that WIE resulted in catalysts with the highest N2O
conversion, although containing much less iron than the CVD
samples studied. They proposed that the differences observed
among the Fe-ZSM-5 samples prepared by the various meth-
ods were related to the ability of the predominant iron species
to catalyze NO-assisted N2O decomposition.
The samples are coded as Fe-ZSM-5-x-CVD (WIE), where x
represents the Si/Al ratio of the parent and CVD or WIE spec-
ifies the preparation method.
2.2. Activity measurements
Catalytic tests were carried out in a computer-controlled
multiport flow setup. Quartz reactors with an internal diame-
ter of 5 mm were placed in a heated reactor vessel. The catalyst
sieve fraction (0.25–0.5 mm) was placed on a quartz grid. Space
velocities (GHSV) are reported at room temperature and at-
mospheric pressure. The feed consisted of 3000 ppm N2O and
0 or 800 ppm NO, the balance being N2. Gas purity was 99%
for N2O and 99.8% for NO. The quantitative analysis of N2O,
N2, and O2 was performed using a micro gas chromatograph
equipped with Poraplot and Molsieve columns and flame ion-
ization and thermal conductivity detectors. NOx was analyzed
with a NOx (chemoluminescence) analyzer. At the start of each
activity experiment, the reactor temperature was increased to
533 K at a rate of 3 K/min under N2 flow and flushed for 3 h.
Then the reaction gas mixture was fed to the catalyst. Precondi-
tioning was set for 20 min at each temperature. Data were col-
lected at ascending temperature using a ramp rate of 5 K/min,
to a maximum of 773 K.
To explore this possibility in more detail, we studied the
NO-assisted N2O decomposition over Fe-ZSM-5 samples, pre-
pared by CVD or WIE, from two different parent zeolites. The
beneficial effect of NO differed among the samples. By com-
bining characterization data, in situ infrared (IR) spectroscopy
and transient response measurements, we offer an explanation
for that phenomenon. Furthermore, we discuss the mechanism
of NO-assisted N2O decomposition in detail.
2. Experimental
2.1. Sample preparation
Two parent ZSM-5 samples were used, MFI-P46 (Süd-
chemie; Na-H-ZSM-5, Si/Al = 24, synthesized with template)
and SM27 (Alsi Penta/Südchemie; NH4-ZSM-5, Si/Al = 12,
template-free synthesis). MFI-P46 was transformed in the NH4
form by three-fold exchange with a 1 M NH4NO3 solution. The
NH4 form of the samples was used for the CVD of FeCl3 [12]
or for WIE with an aqueous solution of FeSO4. For CVD, the
samples were first treated in O2 at 773 K for 1 h; then FeCl3
was sublimed onto the zeolite at 593 K, followed by extensive
2.3. In situ IR spectroscopy
A total of 6 mg of sample was pressed into a self-supporting
pellet 6 mm in diameter, placed in a gold sample holder, and

G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
239
inserted into a cylindrical oven. The oven was screwed onto
a stainless steel cell block equipped with gas connections and
CaF2 windows to let the IR beam pass through the cell and the
sample. The gas steam flows along the axis of the cylindrical
oven through the sample pellet to the exit on the other side. Be-
cause the oven does not completely fill the cross-section of the
cell, about 10% of the gas bypasses the sample, with the rest
flowing through it. The cell represents a reasonable approxima-
tion of a plug flow reactor and, because of its small volume, is
suited for a rapid switch from one gas mixture to another.
The samples were heated in a stream of 5% O2 in He at
673 K, then kept for 30 min in a flow of pure He and cooled
to reaction temperature (usually 573 K was chosen as the ini-
tial temperature). The initial treatment in O2 serves to remove
hydrocarbon impurities, and the subsequent treatment in He al-
lows for partial autoreduction of Fe3+ to Fe2+. The reaction
was started by switching from He to a mixture of 3000 ppm
N2O and 800 ppm NO in He. NO was purified with a cold trap
at 173 K. The reaction was followed for 30 min by IR. The gas
phase was simultaneously analyzed by gas chromatography and
mass spectrometry; then the feed was switched back to He. The
temperature was raised to 673 K for 30 min, and the procedure
was repeated at the next temperature. The total gas flow through
the cell was kept constant at 20 mlNTP/min. For a 6-mg pellet,
this corresponds to GHSV = 120,000 h−1 (using a density of
0.5 g/ml).
Fig. 1. UV-vis spectra of Fe-ZSM-5-12/24 CVD and WIE.
3. Results
3.1. UV–vis
The UV–vis spectra of both WIE samples (Fig. 1) showed
only two intense ligand-to-metal charge transfer bands at
43,000 and 38,000 cm−1, as was expected for isolated iron
ions [14]. In the Fe-ZSM-5-24 CVD sample, a shoulder ap-
peared at 27,000 cm−1, indicating clustering of iron atoms [15].
Absorption at 20,000 cm−1 demonstrated that some Fe2O3
particles were also present. In the UV-vis spectrum of Fe-ZSM-
5-12 CVD, the band of Fe2O3 was more intense. The spectrum
of Fe-ZSM-5-12 CVD differed from the others in that it was
The IR spectra were recorded on a Biorad FTS 3000 MX
spectrometer equipped with a broad band MCT detector. The
spectral resolution was 4 cm−1. The catalyst pellet at the re-
action temperature in a flow of He was used as background.
The ZSM-5-12 samples were not very transparent, which had a
negative effect on the signal to noise ratio. The corresponding
spectra were Fourier filtered, to reduce the noise. The effluent
of the IR cell was monitored by a Omnistar mass spectrome-
ter. m/e values of 4, 28, 30, 32, 44, and 46 were recorded. All
signals were normalized to the signal of He (m/e = 4), and the
concentrations were calculated from a calibration matrix. The
low sensitivity of the mass spectrometer for NO2 did not allow
us to obtain reliable NO2 concentrations. The concentration of
NO2 can be inferred from the difference cN − 2cO . This was
dominated by two intense bands at 28,000 and 43,000 cm−1
.
The position of the bands suggests an assignment to Cl → Fe
charge transfer transitions [16]; that is, the chloride could not
be completely removed by the repeated washing and calcina-
tion of that sample. Note that the band at 28,000 cm−1 was
observed earlier on a Fe-ZSM-5-12 sample prepared from the
same parent zeolite using a FeCl2 precursor, but in our opinion
was erroneously assigned to binuclear iron species [17].
2
2
3.2. IR spectroscopy
verified by a few selected experiments with a NOx analyzer.
The IR spectra of the parent zeolites ZSM-5-12 and ZSM-
5-24 showed distinct differences (Fig. 2). ZSM-5-24 had two
intense bands in the region of the OH stretching vibrations,
at 3735 cm−1 (silanol groups) and at 3600 cm−1 (Brønsted
OH groups). A very weak band at 3660 cm−1 was attributed
to Al–OH groups on extra-framework species. In the case of
ZSM-5-12, the Brønsted OH band was more intense, due to the
higher Al concentration, and the silanol band was very weak.
The two bands at 3735 and 3715 cm−1 were assigned to exter-
nal and internal silanol groups [18]. After ion exchange, the
intensity of the Brønsted OH band decreased (see Table 1).
In the ion-exchanged samples, the fraction of exchanged OH
groups corresponds to the Fe/Al ratio; that is, one Fe atom
exchanged for one Brønsted site, as has been observed for
Cu2+ [19,20]. For reasons of charge balancing, the iron must
2.4. Transient response measurements
A 50-mg pelletized sample was placed in a 4-mm i.d. quartz
tube and treated in a He flow at 673 K for 1 h. The reac-
tor was then cooled to 648 K, and the feed was switched to
800 ppm NO in He. Pulses of 5000 ppm N2O were given into
the flow of NO at intervals of ∼100 s. Then, after letting the
system re-equilibrate for 10 min, a step to 3000 ppm N2O in He
was performed. The reaction was followed for ∼30 min, after
which the feed was stepped back to NO and finally once again
to N2O. The procedure was repeated at 673 K. The total flow
rate in the experiments was 50 ml NTP/min, corresponding to
GHSV = 30,000 h−1. The reaction products were analyzed by
mass spectrometry as described earlier.

240
G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
Table 2
Temperature at which the catalysts reach a N O conversion of 20%. Feed con-
2
−1
centration 3000 ppm N O, 0 or 800 ppm NO, GHSV = 60,000 h
2
with NO
without NO
ꢀT
Fe-ZSM-5-12 WIE
Fe-ZSM-5-24 WIE
Fe-ZSM-5-12 CVD
Fe-ZSM-5-24 CVD
331
350
360
367
432
451
422
447
−101
−101
−62
−80
Fig. 2. IR spectra of the dehydrated samples measured at 523 K. (a) ZSM-5-12
parent (solid), WIE (dash), CVD (dotted); (b) ZSM-5-24 parent (solid), WIE
(dash), CVD (dotted).
Fig. 4. NO concentration profile. Feed 3000 ppm N O, 800 ppm NO, balance
2
−1
N , GHSV = 60,000 h . ZSM-5-12 WIE (squares), ZSM-5-24 WIE (dia-
2
monds), ZSM-5-12 CVD (triangles), ZSM-5-24 CVD (circles).
tivity of the WIE samples was strongly boosted by NO, and
these samples became significantly more active than the CVD
samples, despite their lower iron loading. For a semiquantita-
tive description, we compared the temperatures at which the
catalysts reached a N2O conversion of 20% (Table 2). Adding
NO shifted the activity curve by −100 K for the WIE samples
and by −80 and −60 K for the CVD samples with Si/Al = 24
and 12, respectively.
The NO concentration profile shows how NO influences the
conversion of N2O (Fig. 4). When the temperature is raised,
the concentration of NO decreases and NO2 is produced (not
shown). NO2 is formed not through reaction of NO with O2,
originating from decomposition of N2O, but rather by direct
reaction with N2O [8], that is,
Fig. 3. N O decomposition activity of the samples. Feed 3000 ppm N O, 0 or
2
2
−1
800 ppm NO (full and open symbols), balance N , GHSV = 60,000 h
.
2
ZSM-5-12 WIE (squares), ZSM-5-24 WIE (diamonds), ZSM-5-12 CVD (tri-
angles), ZSM-5-24 CVD (circles).
be bound as [Fe(OH)2]+. If a dimer bridging two ion-exchange
sites is formed, then charge balancing can also be achieved by
[OH–Fe–O–Fe–OH]2+ or [OH–Fe-µ-(OH)2–Fe–OH]2+. The
vibration of the Fe–OH groups can be recognized as a broad
band at 3660 cm−1 in the spectra of both ion-exchanged sam-
ples (Fig. 2). In the CVD samples, on average three iron atoms
were bound to one Al site in the form of small iron oxide
clusters. Earlier studies reported exchange degrees correspond-
ing to roughly two iron atoms per one Al atom for the CVD
method [21,22].
NO + N2O → NO2 + N2.
(9)
Most of the N2O conversion in temperature range 550–650 K is
due to this stoichiometric reaction of NO with N2O (see Fig. 5).
Above ∼650 K, the effect of NO becomes mainly catalytic.
N2O decomposition strongly increases, and NO concentration
increases once again.
3.4. In situ IR spectroscopy
3.3. Comparison of catalytic activity
In situ IR spectroscopy was used to explain the different ac-
tivity of the four samples based on the nature and concentration
of adsorbed surface species. Thin catalyst pellets were exposed
to a flow of 3000 ppm N2O and 800 ppm NO in He. The space
velocity in these experiments was a factor of two lower than
in the activity measurements described above. The adsorbed
A comparison of the N2O decomposition activity of the sam-
ples reveals two trends (Fig. 3): Si/Al = 12 > Si/Al = 24 and
CVD > WIE. In essence, activity increased with the iron load-
ing of the samples. Adding NO changed the situation. The ac-

G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
241
Fig. 5. Conversion of N O vs. conversion of NO. Solid line corresponds to a
2
Fig. 7. IR spectra after 30 min in 3000 ppm N O + 800 ppm NO at 573 K.
2
−1
conversion of one N O per NO. Fe-ZSM-5-12 WIE, GHSV = 60,000 cm
.
2
In addition, new bands emerged in the high-frequency region
of the spectrum. The most intense band was a doublet at 3660
and 3630 cm−1 (Fig. 6). These bands were observed before and
during reaction of N2O with Fe-ZSM-5 and were assigned to a
Fe(OH)2 group [27]. They were generated during the oxidation
of Fe2+ to Fe3+ [28]. The source of the additional OH-groups
must be traces of water in the feed. The OH bands did not ap-
pear when only NO was flown over the catalyst.
Fig. 7 compares the spectra of the four investigated catalysts
in the region of the relevant NOx vibrations. The two WIE sam-
ples exhibited more intense nitrosyl bands than the two CVD
samples. The intensity of the nitro band was similar on all cat-
alysts. On Fe-ZSM-5-12 WIE, an additional broad absorption
was found at ∼1500 cm−1. The wavenumber suggests assign-
ment to a surface nitrito species, that is, a –O–N=O− group
bound to surface area via the oxygen atom [23]. The band did
not disappear when switching to He, supporting the assignment
to a strongly bound charged species. The CVD samples exhib-
ited a weak band at 1530 cm−1. This band had high thermal
stability and thus must be due to a charged species (nitrito or
nitrato), although it was not exactly assigned. Fe(OH)2 bands
at 3660 and 3630 cm−1 were found on both catalysts prepared
by WIE (not shown). On the CVD samples, only a broad band
due to hydrogen-bonded hydroxyls appeared, due to the higher
density of iron sites on the CVD materials.
A control experiment with the parent ZSM-5-12 sample
(600 ppm Fe) yielded entirely different spectra. No new OH
stretching vibrations were observed. A weak nitrosyl band was
formed due to adsorption of NO on the Fe impurities or on
Al3+ [29], but the intensity of the nitrosyl band was 10 times
weaker than on Fe-ZSM-5-12 WIE. Further, a multitude of
bands was formed between 1800 and 1300 cm−1, attributed
to surface nitro, nitrato, and nitrito groups. These bands were
highly stable on the catalyst surface and behaved like spectators
when switching from the reaction mixture to He and back. The
behavior of the parent was completely different in the CVD and
WIE samples. Activity tests in the plug flow reactor showed that
the activity of the parent sample was negligible compared with
that of Fe-ZSM-5. Therefore, we can state that (a) the catalysis
runs on the iron cations (clusters), which were introduced by
exchange, and that (b) the in situ IR measurements probe the
Fig. 6. Time evolution of IR spectra of adsorbed species on Fe-ZSM-5-24
WIE at 573 K during reaction with 3000 ppm N O and 800 ppm NO.
2
Fe-ZSM-5-24 spectra after 30 s, 1, 2, 4 and 19 min (each offset by 0.002).
−1
GHSV = 120,000 h
.
species were characterized in the temperature range of 573–
648 K, the range in which first the stoichiometric reaction and
then the catalytic reaction between NO and N2O set in.
Fig. 6 shows the time evolution of the spectra measured for
Fe-ZSM-5-24 WIE at 573 K. A nitrosyl band at 1878 cm−1 in-
creased rapidly, in parallel with the concentration of NO in the
gas phase. After 4 min, the nitrosyl band reached its maximum
intensity and then decreased slightly. A band at 1623 cm−1 in-
creased gradually with time. According to its wavenumber, this
band can be assigned to a nitrato (NO3−) or a weakly bound
nitro group (NO2) [23,24]. The band disappeared rapidly when
switching from the reaction mixture to He. Thus we assigned
the b−and to a weakly bound NO2 species [25] rather than to a
NO3 group, which would have greater thermal stability. The
appearance of the nitro band indicates that oxidation of NO to
NO2 occurred. As expected from the stoichiometry of reaction
(9), a higher concentration of N2 was observed in the gas phase
during the buildup of NO2 on the surface.
The doublets at 3498/3464, 2583/2547, and 2238/2208 cm−1
were due to gas-phase N2O. A low-frequency tail of the
2208 cm−1 band can be attributed to NO+ species, but its con-
centration was much lower than in a feed of NO + O2 [26].

242
G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
(a)
(b)
Fig. 8. IR spectra after 30 min in 3000 ppm N O + 800 ppm NO at (a) 623 K and (b) 648 K.
2
(a)
(b)
−1
Fig. 9. Pulses of 5000 ppm N O into a flow of 800 ppm NO at (a) 648 K and (b) 673 K. Fe-ZSM-5-12 WIE, GHSV = 30,000 h
.
2
adsorption/reaction on these iron sites, not on the zeolite ma-
trix.
the nitro band was observed, but a band at 1630 cm−1 remained
and decreased only very slowly with time. This finding demon-
strates that the nitro groups on Fe-ZSM-5-12 WIE are very
labile, whereas part of the surface nitro groups on the other cat-
alysts have rather high thermal stability.
At 648 K, the nitrosyl band further decreased in intensity
compared with at 623 K (see Fig. 8b). The intensity of the
nitro band increased for Fe-ZSM-5-12 WIE and slightly de-
creased for the other samples. The intensity of the NO2 band
at 648 K follows the order Fe-ZSM-5-12 WIE > Fe-ZSM-5-24
WIE > Fe-ZSM-5-24 CVD > Fe-ZSM-5-12 CVD. The same
order can be obtained at 623 K if only the labile nitro groups
(i.e., those that desorb within 1 min in He) are counted. Note
that the intensity of the nitrosyl band at lower temperatures
(573 K) follows the same order.
The adsorbed species on Fe-ZSM-5 reached steady state
much quicker at 623 K than at 573 K (not shown). The intensity
of the nitrosyl bands at 1880 cm−1 stabilized at a value approx-
imately a factor of five lower than at 573 K (see Fig. 8a). In
contrast, the intensity of the NO2 band increased for the two
WIE samples and remained roughly constant for Fe-ZSM-5-24
CVD. Due to the high noise level of the spectrum, Fe-ZSM-5-
12 CVD was not included in the comparison. In addition to the
nitrosyl and nitro bands, Fe-ZSM-5-12 WIE exhibited absorp-
tion at 1570 and at 1500 cm−1. In analogy to Cu-ZSM-5, the
former band can be assigned to chelating nitrate species (see
Scheme 1) [30,31], and the latter is due to nitrite species (vide
supra).
When switching from the reaction mixture back to He, the
nitrosyl band disappeared very quickly (in parallel with the
decreased NO concentration in the gas phase). In the case of
Fe-ZSM-5-12 WIE, the nitro band also disappeared within a
few minutes. For the other catalysts, a fast initial decrease of
The N2O conversions measured in the in situ IR experiments
were approximately a factor of 10 lower than those measured in
the plug flow reactor. At 573 K, hardly any measurable activity
was observed in the IR cell; at 648 K, conversions ranged from
2 to 7%.

G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
243
(a)
(b)
Fig. 10. (a) Step from 800 ppm NO to 3000 ppm N O in He at 648 K and step back to NO. (b) Zoom of the first 500 s after the step. Fe-ZSM-5-12 WIE,
2
−1
GHSV = 30,000 h
.
3.5. Pulse and step experiments
To obtain more insight into the reactivity of the surface
species, pulse and step experiments were performed in a tubular
quartz reactor at 648 and 673 K and in the IR reactor at 648 K.
The space velocity in the IR reactor was roughly a factor of four
lower than that in the plug flow reactor.
When N2O was pulsed into a flow of NO over the cata-
lyst Fe-ZSM-5-12 WIE, a peak of N2 was formed (see Fig. 9).
A peak of O2 was detected at 673 K, but not at 648 K. The
O2 peak strongly tailed with respect to the N2 peak. The same
phenomenon was observed in earlier TAP experiments [8]. In
the IR reactor, the nitrosyl band at 1880 cm−1 decreased dur-
ing the N2O pulse and regained its original intensity after the
pulse; that is, some NO was replaced by the incoming N2O. No
nitro bands were detected. It is interesting to note that the pulse
of N2O was strongly broadened and tailed, especially at 648 K.
This indicates that some N2O was molecularly adsorbed on the
iron sites. The effect was observed on both WIE catalysts, but
not on Fe-ZSM-5-24 CVD. Fe-ZSM-5-24 CVD also did not
produce any O2 during the N2O pulses, even at 673 K.
After the pulses of N2O, a step from NO to N2O was per-
formed (see Fig. 10). A peak of N2 was produced immediately
after the step. This peak of N2 was not related to NO2 for-
mation, but rather was due to the reoxidation of Fe2+ sites to
Fe3+ [32]. NO was desorbed from the catalyst (indicated by the
peak of NO in Fig. 10b), being replaced by N2O. After some de-
lay, the production of O2 began. After 40 s, N2 and O2 decayed
in parallel to the steady-state value. The steady-state activity
was due to NO-unassisted N2O decomposition.
Fig. 11. Step from 800 ppm NO to 3000 ppm N O in He at 648 K in the IR re-
2
actor. Intensity of the nitrosyl band (circles) and of the NO band (×5, squares)
2
on right axis. Gas phase concentration of N (solid), NO (dotted), and O (×2,
2
2
−1
dashed). Fe-ZSM-5-12 WIE, GHSV ∼120,000 h
.
Table 3
Quantification of the initial peak of N formed after a step from 800 ppm NO to
2
3000 ppm N O in He, at 648 or 673 K, as well as of the amount of O formed
2
2
−1
in the subsequent N O decomposition. GHSV = 30,000 h
2
a
a
N
peak
O
2
N
2
peak
O
2
2
(µmol)
(µmol) (mol/mol Fe) (mol/mol Fe)
648 K
Fe-ZSM-5-12-WIE 3.7
IR reactor
Fe-ZSM-5-24-WIE 1.9
Fe-ZSM-5-24-CVD 0.7
15
2.0
0.15
0.14
0.22
0.02
0.59
0.07
0.23
0
b
0
673 K
Fe-ZSM-5-12-WIE 1.21
Fe-ZSM-5-24-WIE 0.9
Fe-ZSM-5-24-CVD 0.5
21
4.7
0
0.05
0.11
0.01
0.85
0.54
0
The delayed onset of O2 formation, followed by the simulta-
neous decay of N2 and O2, were also observed in the IR reactor
(Fig. 11). Due to the higher space velocity, less O2 was formed.
The intensity of the surface nitrosyl band decreased in parallel
with the gas-phase concentration of NO. A desorption peak of
NO, as in the tubular quartz reactor, was not observed, again
due to the high space velocity. With a delay of ∼10 s (similar
to the delay of O2 formation), a nitro band was formed. The ni-
tro band increased in intensity as long as NO was still present
in the gas phase and then slowly decreased. Fig. 11 shows that
a
Steady state activity, due to NO-unassisted N O decomposition, subtracted.
2
b
−1
GHSV ∼ 120,000 h
.
the evolution of the NO2 band and the formation of O2 in the
gas phase were synchronized.
The behavior of the other catalysts was studied only in the
tubular quartz reactor. Fe-ZSM-5-24 WIE behaved qualitatively
like Fe-ZSM-5-12 WIE, but its activity for O2 formation af-
ter the step was lower (Table 3). With Fe-ZSM-5-24 CVD, no

244
G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
accelerated O2 formation was observed. The activity settled
immediately at its steady-state value, due to unassisted N2O de-
composition.
After the step back from N2O to NO, a small peak of O2
and an even smaller peak of N2 were observed. The O2 for-
mation decayed in less than 80 s. The amount of O2 formed
decreased from 0.015 mol/mol Fe in Fe-ZSM-5-12 WIE to
0.008 mol/mol Fe in Fe-ZSM-5-24 WIE to 0.001 mol/mol Fe
in Fe-ZSM-5-24 CVD. In the IR reactor, a nitro band increased
and decayed simultaneously with the O2 peak.
and 1575 cm−1 can also be formed from N2O [42,43]. In the
concentration and temperature range used in our study, how-
ever, N2O does not create any strongly IR absorbing surface
species [44]. Therefore, the source of the nitro/nitrate bands at
1625 and 1575 cm−1 in our experiments must be the oxida-
tion of NO by N2O. These bands are also formed on reaction of
NO+O2 on our catalysts, but the intensity and thermal stability
differ.
4.3. The role of the surface species in the reaction mechanism
Two different models have been proposed to explain the cat-
alytic effect of NO on N2O decomposition, one involving ni-
trito and nitrato species [6] and the other involving adsorbed
NO2, which functions as an intermediate oxygen storage [7,
8]. In our in situ IR experiments, bands of adsorbed NO2
(∼1625 cm−1), chelating nitrates (1575 cm−1), and weak bands
of nitrito species were observed. The band of molecularly ad-
sorbed NO2 was by far the most intense, and its intensity at high
temperatures correlated with the catalytic activity of the sam-
ples (Fig. 12). During the step experiments (i.e., during a step
from NO to N2O), the NO2 band appeared as the only new sur-
face species, and its intensity correlated with the O2 formation.
These three results prove that adsorbed NO2 is indeed part of
the catalytic cycle of NO-assisted N2O decomposition, as pro-
posed by Perez-Ramirez et al. [8]. We return later to the ques-
tion of whether the chelating nitrate (1575 cm−1) also plays a
role in the reaction mechanism.
4. Discussion
4.1. Structure of the iron sites in WIE and CVD catalysts
The UV–vis and IR spectra suggest that the WIE samples
contain isolated iron cations located at ion exchange positions.
However, recent EXAFS data indicate that iron dimers can be
formed at high ion exchange degrees [33], as has been reported
for Cu-ZSM-5 [34]. Because the Fe/Al ratio of our WIE sam-
ples is rather high (Fe/Al = 0.26 and 0.45), we suspect that a
mixture of isolated sites and dimers is present. Several stud-
ies have shown that the CVD samples contain small oligomeric
iron oxide clusters [35–37]. This is confirmed by the UV–vis
and IR data presented here.
4.2. The adsorbed species
With the help of our transient response experiments, we can
extract more details of the catalytic cycle described in Eqs. (3)–
(6). When N2O was pulsed into a flow of NO, unreacted N2O
was rather strongly retained on the WIE catalysts; that is, it was
molecularly adsorbed (Fig. 9). After a step of NO to N2O, N2O
partially replaced NO from the adsorption sites, leading to the
peak of NO observed in Fig. 10b. These two findings suggest
that N2O adsorbs molecularly on the WIE catalysts and that
N2O competes with NO for adsorption sites. After the step from
N2O to NO, NO2 species were formed only as long as N2O
was still present in the gas phase. We therefore propose that,
instead of adsorbed oxygen atoms, molecularly adsorbed N2O
is involved in the formation of NO2 on the WIE catalysts. We
can write the reaction mechanism as
The dominating surface species in our experiments are
mononitrosyl species (1880 cm−1) and adsorbed NO2 (1625
cm−1). Both bands were also observed in a similar study by
Mul et al. [7], although that study was performed at lower tem-
peratures. The stretching frequency of the nitrosyl band is only
slightly higher than the stretching frequency of NO in the gas
phase (1876 cm−1), indicating a very weak coordinative bond
of NO to iron. When NO adsorption is studied at room tempera-
ture, a multitude of bands, assigned to dinotrosyl and trinitrosyl
species [38] or to mononitrosyls adsorbed in different positions
in the ZSM-5 host [39], is observed. At the high temperatures
used in our in situ experiments, NO does not discriminate be-
tween these sites, and only a single band is observed. It is
generally agreed that NO adsorbs rather strongly on Fe2+ but
only weakly or not at all on Fe3+ [23,38]. Test experiments
confirmed that catalysts that were not allowed to autoreduce in
He before switching to the reaction mixture adsorbed less NO.
We therefore assign the band at ∼1880 cm−1 to mononitrosyl
species bound to Fe2+ [40].
The second prominent band in the spectra is the nitro band at
1625 cm−1. In the Fe-ZSM-5-12 WIE sample, the band exhibits
a shoulder at 1575 cm−1 at higher temperatures. A similar pair
of bands at 1625 and 1575 cm−1 is observed when NO + O2
is passed over a Fe-ZSM-5 catalyst [25,41]. When formed from
NO+O2, the band at 1625 cm−1 has rather high thermal stabil-
ity, and is probably due to a bridging nitrate species rather than
to an adsorbed nitro group. The two groups vibrate at very sim-
ilar wavelengths and can be distinguished based solely on their
thermal stability. At high concentrations (>1%), bands at 1625
Fe2+(NO) + N2O → Fe2+(NO)(N2O) → Fe2+(NO2) + N2,
(11)
Fe2+(NO2) + N2O → Fe2+(NO) + N2 + O2.
(12)
After the step from NO to N2O, N2O replaces NO from adsorp-
tion sites, as was also observed by Mul et al. [7]. Then NO2 is
formed by reaction of adsorbed NO with N2O [reaction (11)],
which is adsorbed on the same site or on a neighboring iron
site (if the iron sites are dimers). The concentration of adsorbed
NO2 reaches its maximum when the gas-phase NO concen-
tration decreases to zero. After that, the build-up of NO2 by
reaction (11) ceases, and only its consumption by reaction (12)
follows. Reaction (12) should restore adsorbed NO on the sur-
face. However, adsorbed NO was not observed 50 s after the
step to N2O (Fig. 11). The reason is that adsorbed NO, once

G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
245
regenerated by reaction (12), is rapidly reoxidized to NO2 by
reaction (11). In contrast, the consumption of NO2 by reaction
(12) is slow. This is in line with earlier work, which concluded
that O2 formation is the rate-limiting step of the reaction [7,8].
Instead of being oxidized to NO2 and refed to the catalytic cy-
cle, the adsorbed NO can also desorb into the gas phase. This
process terminates the catalytic cycle and leads to the slow
decay of the catalytic activity when NO is not present in the
feed. The competition between oxidation of NO to NO2, and its
desorption from the surface determines how much O2 can be
formed after the step from NO to N2O.
In contrast to the WIE samples, enhanced N2 formation after
the step from NO to N2O was not observed for the CVD cat-
alyst (see Table 3). The replacement of adsorbed NO by N2O,
resulting in a desorption peak of NO, also was not observed. We
presume that on the CVD catalyst, N2O adsorbs and dissociates
on sites remote from the NO adsorption sites. The deposited
oxygen atom has to migrate to the adsorbed NO to form NO2.
This process is slower than the direct reaction of adsorbed NO
with N2O on the WIE catalysts; hence significant amounts of
NO2 are not formed in the step experiment with the CVD cata-
lyst and enhanced N2 formation is not observed. Note, however,
that NO2 formation is not the rate-limiting step in steady state.
Therefore, the slower NO2 formation on the CVD catalysts
cannot explain their lower activity in NO-assisted N2O decom-
position.
Scheme 1. Structures of adsorbed NO and of chelating nitrate.
2
(see Scheme 1), which then decomposes to NO and O2. A sim-
ilar role of transporting oxygen atoms between distant sites to
facilitate O2 formation has been attributed to NO2 in the de-
composition of NO over Cu-ZSM-5 [9,10]. Note that the first
and last pathways can operate on isolated iron sites, whereas
the second pathway requires the presence of two iron sites for
the adsorption of oxygen and of NO/NO2, which are located
near each other, that is, a dimer or an even larger iron cluster.
Can we determine whether the contribution of adsorbed NO2
or that of gas-phase NO2 dominates on the iron zeolites? Perez-
Ramirez et al. [8] showed that when NO and N2O were alter-
nately pulsed over the catalyst at 773 K, NO2 appeared in the
gas phase only during the NO pulses, but O2 was formed after
the pulses of N2O, that is, when NO2 was not present in the gas
phase. This finding speaks in favor of adsorbed NO2. Our re-
sults show that the activity of the catalysts is correlated with the
concentration of adsorbed NO2 at 648 K. Keep in mind, how-
ever, that at 648 K the effect of NO/NO2 on N2O decomposition
is still mainly stoichiometric and not catalytic. The catalytic ef-
fect of NO becomes more dominating at higher temperatures.
Under these conditions, we can expect the adsorption of NO2
on the surface to be weak and the gas-phase mechanism to cer-
tainly gain in importance.
It is worthwhile to analyze O2 formation [reaction (12)],
which is rate-determining, in more detail. Different mechanistic
pathways can be considered.
Option 1:
Fe2+(NO2) + N2O → Fe2+(NO2)(N2O)
→ Fe2+(NO) + N2 + O2.
(12a)
(12b)
4.4. Comparison of CVD and WIE samples
Option 2:
N2O + Fe2+ → Fe–O + N2,
The IR measurements at 573 K correspond to a situation
in which conversion is close to zero. The concentration of ad-
sorbed NO under these conditions correlates with the activity of
the samples at higher temperatures (Fig. 12). NO adsorbs pref-
erentially on Fe2+. The concentration of Fe2+ on the samples
can be estimated from the initial peak of N2, which was always
observed in the step experiments. The peak of N2 is related to
reoxidation of Fe2+ to Fe3+, according to the equation
Fe–O + Fe2+(NO2) → Fe2+(NO) + O2.
Option 3:
N2O + Fe2+ → Fe–O + N2,
(12c)
Fe2+(NO2) → Fe2+ + NO2,
Fe–O + NO2 → Fe–O(NO2) → Fe2+(NO) + O2.
2Fe2+ + N2O → 2Fe3+ + O2− + N2.
(13)
The first option is a direct reaction of N2O with adsorbed
NO2, possibly through an intermediate in which both molecules
are adsorbed on the same iron site [45]. Alternatively, N2O may
first dissociate into N2 and an adsorbed oxygen atom. If the
oxygen atom is deposited in the vicinity of adsorbed NO2 or if
it has the potential to migrate, it can react with NO2 to give O2
and adsorbed NO [reaction (12b)]. This is in essence the mech-
anism proposed by Perez-Ramirez et al. [8]. A third possibility
is that NO2 desorbs into the gas phase and reacts with an oxy-
gen atom deposited from N2O. The band at 1575 cm−1 can be
tentatively ascribed to the intermediate formed between the de-
posited surface oxygen atom and a gas-phase NO2 molecule
We chose the values obtained in the steps from NO to N2O (see
Table 3) (which are more precise than those of in the in situ
IR experiments) and used the Fe2+/N2 stoichiometry of reac-
tion (13). The concentration of Fe2+ determined by this method
correlates with the concentration of adsorbed NO (Fig. 13). At
higher temperatures (i.e., with increasing N2O conversion), an
increasing fraction of adsorbed NO is converted into adsorbed
NO2. The more NO is adsorbed on the catalyst under nonre-
active conditions, the more can be converted to NO2 at higher
temperatures (Fig. 12) and thereby contribute to the increased
catalytic activity.

246
G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
the high ion-exchange capacity of the sample, which enables
stabilization of a high concentration of isolated iron ions or
dimers.
5. Conclusion
NO-assisted N2O decomposition was studied over Fe-ZSM-
5 samples prepared by two different methods (CVD and WIE)
from two different parent zeolites. Transient kinetic and IR ex-
periments provided direct proof that adsorbed NO2 is part of the
catalytic cycle of NO-assisted N2O decomposition. The mech-
anism of NO2 formation differs on WIE and CVD samples.
On WIE catalysts, NO2 is formed by the direct reaction of ad-
sorbed NO with N2O on an isolated iron site or dimer. On CVD
samples, containing oligomeric iron species, we presume that
NO and N2O adsorb on distant sites. NO2 formation is signifi-
cantly slower. But NO2 formation is not the rate-limiting step of
the reaction. O2 formation, although strongly accelerated com-
pared with unassisted N2O decomposition, is still the slowest
step of the reaction sequence. O2 is formed by the reaction of
N2O with adsorbed and/or gas-phase NO2. The involvement of
gas-phase NO2 is expected to lead to a strong increase in the
O2 formation rate in catalysts containing isolated iron sites, be-
cause it facilitates the recombination of oxygen atoms deposited
on distant sites. This effect partly explains why WIE samples
become more active in NO-assisted N2O decomposition than
the CVD catalysts even though the opposite sequence is found
in the decomposition of pure N2O. But the main reason for the
higher activity of the WIE samples is that WIE samples contain
a higher concentration of Fe2+ sites under reaction conditions.
NO preferentially adsorbs on Fe2+, and the concentration of
Fe2+ determines how many sites take part in the NO-assisted
N2O decomposition cycle.
−1
Fig. 12. Intensity of the nitrosyl band at 1880 cm at 573 K and of the nitro
−1
band at 1623 cm at 648 K vs. the N O conversion at 648 K (in the IR reactor,
2
−1
i.e., at GHSV ∼120,000 h ).
−1
Fig. 13. Normalized intensity of the NO band at 1880 cm obtained in the IR
experiments at 573 K (normalized by thickness of pellet) vs. concentration of
2+
Fe sites (values derived from Table 3, 648 K).
Acknowledgments
The addition of NO to the feed enhances the N2O decompo-
sition activity of the WIE catalysts more than that of the CVD
samples. Fig. 13 shows that the WIE samples have a signif-
icantly higher concentration of Fe2+ than the CVD samples,
despite the higher iron loading of the latter. WIE samples au-
toreduce more readily than catalysts prepared by CVD [46].
The higher concentration of Fe2+ leads to a higher concentra-
tion of adsorbed NO/NO2 on the WIE samples and explains
why the WIE samples are more active in NO-assisted N2O de-
composition than CVD materials. A second reason may be the
contribution of gas-phase NO2, which allows oxygen atoms that
were deposited on distant iron sites to recombine [see reac-
tion (12c)]. Therefore, gas-phase NO2 is expected to lead to a
stronger enhancement of the reaction rate on isolated iron sites
than on clustered iron sites, where O2 recombination is also
possible via migration over the cluster.
The authors thank Pijus K. Roy (ETH), Saskia Booneveld,
and Ruud van den Brink (ECN) for helping to characterize the
samples and engaging in helpful discussions.
References
[1] J.T. Houghton, IPCC 3rd assessment report, 2001, http://www.ipcc.ch/
pub/reports.htm.
[2] Y. Li, J.N. Armor, Appl. Catal. B 1 (1992) L21.
[3] G. Centi, A. Galli, B. Montanari, S. Perathoner, A. Vaccari, Catal. To-
day 35 (1997) 113.
[4] G. Centi, S. Perathoner, F. Vazzana, M. Marella, M. Tomaselli, M. Man-
tegazza, Adv. Environ. Res. 4 (2000) 325.
[5] F. Kapteijn, G. Marban, J. Rodriguez-Mirasol, J.A. Moulijn, J. Catal. 167
(1997) 256.
[6] C. Sang, C.R.F. Lund, Catal. Lett. 73 (2001) 73.
Along with comparing WIE and CVD, it is interesting to ex-
plore the role of the parent zeolite. Fe-ZSM-12-WIE was far
more active than the analogous sample prepared from ZSM-
5-24. This can be attributed to the higher iron loading of
the former sample (see Table 1). If activity is normalized by
the iron loading, then the two samples are similar. There-
fore, the high activity of Fe-ZSM-12-WIE is due mainly to
[7] G. Mul, J. Perez-Ramirez, F. Kapteijn, J.A. Moulijn, Catal. Lett. 77 (2001)
7.
[8] J. Perez-Ramirez, F. Kapteijn, G. Mul, J.A. Moulijn, J. Catal. 208 (2002)
211.
[9] B. Moden, P. Da Costa, B. Fonfe, D.K. Lee, E. Iglesia, J. Catal. 209 (2002)
75.
[10] B. Moden, P. Da Costa, D.K. Lee, E. Iglesia, J. Phys. Chem. B 106 (2002)
9633.

G.D. Pirngruber, J.A.Z. Pieterse / Journal of Catalysis 237 (2006) 237–247
247
[11] J.A.Z. Pieterse, S. Booneveld, R.W. van den Brink, Appl. Catal. B 51
(2004) 215.
[28] S. Kameoka, T. Nobukawa, S. Tanaka, S. Ito, K. Tomishige, K. Kunimori,
Phys. Chem. Chem. Phys. 5 (2003) 3328.
[12] H.Y. Chen, W.M.H. Sachtler, Catal. Today 42 (1998) 73.
[13] P.K. Roy, G.D. Pirngruber, J. Catal. 227 (2004) 164.
[14] G. Lehmann, Z. Phys. Chem. Neue Folge 72 (1970) 279.
[15] J. Perez-Ramirez, J.C. Groen, A. Brückner, M.S. Kumar, U. Bentrup, M.N.
Debbagh, L.A. Villaescusa, J. Catal. 232 (2005) 318.
[16] P. Marturano, L. Drozdova, G.D. Pirngruber, A. Kogelbauer, R. Prins,
Phys. Chem. Chem. Phys. 3 (2001) 5585.
[29] E.J.M. Hensen, Q. Zhu, R.A.J. Janssen, P.C.M.M. Magusin, P.J. Kooyman,
R.A. van Santen, J. Catal. 233 (2005) 123.
[30] M.V. Konduru, S.S.C. Chuang, J. Phys. Chem. B 103 (1999) 5802.
[31] M.V. Konduru, S.S.C. Chuang, J. Catal. 196 (2000) 271.
[32] G.D. Pirngruber, J. Catal. 219 (2003) 456.
[33] T. Nobukawa, M. Yoshida, K. Okumura, K. Tomishige, K. Kunimori,
J. Catal. 229 (2005) 374.
[17] R. Brosius, D. Habermacher, J.A. Martens, L. Vradman, M. Herskowitz,
L. Capek, Z. Sobalik, J. Dedecek, B. Wichterlova, V. Tokarova, O. Gon-
siorova, Top. Catal. 30/31 (2004) 333.
[18] A. Zecchina, S. Bordiga, G. Spoto, L. Marchese, G. Petrini, G. Leofanti,
M. Padovan, J. Phys. Chem. 96 (1992) 4991.
[34] P. Da Costa, B. Moden, G.D. Meitzner, D.K. Lee, E. Iglesia, Phys. Chem.
Chem. Phys. 4 (2002) 4590.
[35] A.A. Battiston, J.H. Bitter, F.M.F. de Groot, A.R. Overweg, O. Stephan,
J.A. van Bokhoven, P.J. Kooyman, C. van der Spek, G. Vanko, D.C. Ko-
ningsberger, J. Catal. 213 (2003) 251.
[19] J. Dedecek, B. Wichterlova, J. Phys. Chem. B 101 (1997) 10233.
[20] S.C. Larsen, A. Aylor, A.T. Bell, J.A. Reimer, J. Phys. Chem. 98 (1994)
11533.
[21] P. Marturano, L. Drozdova, A. Kogelbauer, R. Prins, J. Catal. 192 (2000)
236.
[36] E.J.M. Hensen, Q. Zhu, M. Hendrix, A.R. Overweg, P.J. Kooyman, M.V.
Sychev, R.A. van Santen, J. Catal. 221 (2004) 560.
[37] G.D. Pirngruber, P.K. Roy, N. Weiher, J. Phys. Chem. B 108 (2004) 13746.
[38] G. Berlier, G. Spoto, G. Ricchiardi, S. Bordiga, C. Lamberti, A. Zecchina,
J. Mol. Catal. A 182 (2002) 359.
[22] A.A. Battiston, J.H. Bitter, F.M.F. de Groot, A.R. Overweg, O. Stephan,
J.A. van Bokhoven, P.J. Kooyman, C. van der Spek, G. Vanko, D.C. Ko-
ningsberger, J. Catal. 213 (2003) 251.
[39] L.J. Lobree, I.C. Hwang, J.A. Reimer, A.T. Bell, J. Catal. 186 (1999) 242.
[40] G. Mul, J. Perez-Ramirez, F. Kapteijn, J.A. Moulijn, Catal. Lett. 80 (2002)
129.
[23] K.I. Hadjiivanov, Catal. Rev. Sci. Eng. 42 (2000) 71.
[24] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordination
Compounds, Part B: Applications in Coordination, Organometallic and
Bioinorganic Chemistry, 5th ed., John Wiley & Sons, New York, 1997.
[25] L.J. Lobree, I.C. Hwang, J.A. Reimer, A.T. Bell, Catal. Lett. 63 (1999)
233.
[26] K. Hadjiivanov, J. Saussey, J.C. Lavalley, Catal. Lett. 52 (1998) 103.
[27] B.R. Wood, J.A. Reimer, A.T. Bell, M.T. Janicke, K.C. Ott, J. Catal. 225
(2004) 300.
[41] H.Y. Chen, T. Voskoboinikov, W.M.H. Sachtler, J. Catal. 180 (1998) 171.
[42] E.M. El-Malki, R.A. van Santen, W.M.H. Sachtler, Microporous Meso-
porous Mater. 35–36 (2000) 235.
[43] E.M. El-Malki, R.A. van Santen, W.M.H. Sachtler, J. Catal. 196 (2000)
212.
[44] B.R. Wood, J.A. Reimer, A.T. Bell, J. Catal. 209 (2002) 151.
[45] C. Sang, B.H. Kim, C.R.F. Lund, J. Phys. Chem. B 109 (2005) 2295.
[46] G.D. Pirngruber, M. Luechinger, P.K. Roy, A. Cecchetto, P. Smirniotis,
J. Catal. 224 (2004) 429.