Active sites in HZSM-5 with low Fe content for the formation of surface oxygen by decomposing N2O: Is every deposited oxygen active?

Article abstract of DOI:10.1016/S0021-9517(03)00222-7

Surface-active centers were detected in HZSM-5 with a low content of iron activated by steaming and high-temperature calcination in inert atmospheres. Thee centers led to the formation of surface oxygen (O)_ad species from N₂O. The total amount of active centers was proportional to the Fe content in the zeolites. Only a part of (O)_ad deposited by decomposing N₂O was active in CO oxidation at 523 K and appeared as sharp O₂ peaks at 666 K during TPD measurements, which were performed after N₂O interaction or after CO interaction with adsorbed oxygen. A binuclear Fe center was suggested featuring a diamond core structure as an active center. The active O atoms were assigned to the paired terminal oxygen atoms each bonded to one Fe site in the diferric [Fe₂O₂H]⁺ cluster. Oxygen pretreatment at 823 K decreased somewhat the total amount of active sites but did not affect the active/inactive oxygen ratio. Zeolite presaturated by water vapor at 473-523 K generated (O)_ad species from N₂O completely inactive in the CO oxidation. The total amount of the deposited oxygen corresponded to half of the stoichiometric amount of the surface Fe atoms and suggested that water blocked a half of the binuclear [Fe₂O₂H]⁺center, the remaining acting as a single Fe site.

Full text of DOI:10.1016/S0021-9517(03)00222-7

Journal of Catalysis 219 (2003) 273–285
www.elsevier.com/locate/jcat
Active sites in HZSM-5 with low Fe content for the formation of surface
oxygen by decomposing N O: is every deposited oxygen active?
2
∗
Lioubov Kiwi-Minsker, Dmitri A. Bulushev, and Albert Renken
Laboratory of Chemical Reaction Engineering, Swiss Federal Institute of Technology, LGRC-EPFL, CH-1015 Lausanne, Switzerland
Received 11 December 2002; revised 14 March 2003; accepted 29 April 2003
Abstract
Surface-active centers were detected in HZSM-5 with a low content of iron (< 1000 ppm) activated by steaming and high-temperature
calcination in inert atmospheres (up to 1323 K). These centers lead to the formation of surface oxygen (O) species from N O and were
ad
2
characterized quantitatively by the transient response method. The total amount of active centers was proportional to the content of iron in the
zeolites. Only a part of (O) deposited by decomposing N O was active in CO oxidation at 523 K and appeared as sharp O peaks at 666 K
ad
2
2
during the TPD measurements. A binuclear Fe center is suggested featuring a “diamond core” structure, similar to that of the monooxygenase
enzyme, as an active center. The active O atoms were assigned to the paired terminal oxygen atoms each bonded to one Fe site (ferryl groups)
+
in the diferric [Fe O H] cluster. Oxygen pretreatment at 823 K decreases somewhat the total amount of active sites but does not affect
2
2
the ratio of active/inactive oxygen. Zeolite presaturated by water vapor at 473–523 K generates (O) species from N O completely inactive
ad
2
in the CO oxidation. Part of it appears as a broad peak at 940 K in the TPD profile. The total amount of the deposited oxygen corresponds
+
to half of the stoichiometric amount of the surface Fe atoms and suggests that water blocks a half of the binuclear [Fe O H] center, the
2
2
remaining acting as a single Fe site.
2003 Elsevier Inc. All rights reserved.

Keywords: Fe-containing HZSM-5; α-Oxygen; Ferryl groups in zeolite; [Fe O ] diamond core; Monooxygenase; N O decomposition; Oxo-bridged species;
2
2
2
Transient response method; Temperature-programmed desorption of oxygen
1
. Introduction
lytic function of isolated redox sites within zeolite cavities
(
or channels) may result in the material with specific redox
properties [2]. One fascinating example is the N2O decom-
position over Fe-containing ZSM-5. This process has been
An efficient method for controlling zeolite activity/selec-
tivity is modification of the zeolite by various metals. It is
well known that the introduction of trivalent heteroatoms,
like Al, Ga, B, or Fe into the crystalline framework of MFI-
silicalites via isomorphous substitution of Si atoms, results
in the formation of strong Brønsted sites affecting acid-
catalyzed reactions. Thermal and hydrothermal treatment of
H-zeolites induces the dehydroxylation and the migration of
heteroatoms into extraframework positions. As a result, the
Lewis acid sites appear and the zeolites possess both types
of acid sites [1].
suggested to proceed via Fe2 transition to Fe , form-
ing active oxygen species called by Panov and co-workers
atomic “α-oxygen” [3,4]. This oxygen can be transferred
to benzene or methane forming selective oxidation products
at temperatures as low as 243 K [4] similar to the enzy-
matic catalysis by methane monooxygenase (MMO) [5].
MMO is known to mediate the selective oxidation of hy-
drocarbons by molecular oxygen at ambient temperature and
pressure [6–8]. Significant efforts have been made to develop
an effective catalyst based on HZSM-5 zeolites for the direct
oxidation of benzene to phenol by nitrous oxide (N2O). Af-
ter catalyst optimization, selectivity to phenol close to 100%
was achieved with a conversion up to 40%. However, the
nature of active sites responsible for the α-oxygen species
formation (hereafter “α-sites”) remains unclear. Some au-
thors claim that Lewis acid sites of the extraframework
aluminum [9,10] or gallium [11] are responsible for the ben-
+
3+
The nature of trivalent heteroatom influences the prop-
erties of the resulting zeolite. Contrary to Al–silicalite,
Fe-containing zeolites show redox properties, since Fe can
2
+
3+
4+
easily change its oxidation state (Fe , Fe , or Fe ). Cata-
*
Corresponding author.
E-mail address: lioubov.kiwi-minsker@epfl.ch (L. Kiwi-Minsker).
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00222-7

274
L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
the zeolite surface:
N2O + ( ) = (O)ad + N2.
(1)
Until now the concentration of active sites (α-sites) has
been measured in a static vacuum setup from the amount
of the N2 released or the N2O consumed, since these two
amounts were found equal [3]. Herein we report the first
application of a transient response method to quantitatively
assess their concentration. Moreover, we aim to quantita-
tively clarify if every oxygen atom deposited on the surface
Scheme 1. Formation of α-sites and active α-oxygen [3].
zene to phenol transformation. Nevertheless, the fact that
commercial zeolites used in these studies contained some
amount of iron must be taken into account. Moreover, from
the recent research results obtained by various groups, the
active centers in benzene hydroxylation have been related to
the presence of Fe species [4,12–14].
(
O)ad from N2O has the same oxidative power.
2
. Experimental
It has been understood that α-sites are formed during
zeolite activation by steam treatment followed by high-
temperature calcination [3,13]. Breakage of the Fe–O–Si
bonds takes place and leads to the formation of well-
dispersed extraframework iron. Panov and co-workers [3]
It is known that catalytic properties of Fe-containing MFI
systems depend on the method of their preparation and acti-
vation. Two main preparation procedures have been reported
to achieve a high dispersion of iron species inside the zeolite
channels: (1) by Fe addition during hydrothermal synthe-
sis followed by NaZSM-5 conversion to the H form [3,16]
and (2) by the introduction of postsynthesis via ion ex-
change from an aqua solution or from the gas phase [17–19].
The first method gives isomorphous framework-substituted
FeZSM-5, which is less sensitive to activation rendering
stable ex-FeZSM-5 active in benzene hydroxylation. This
method, from our point of view, is more adequate for the
preparation of HZSM-5 with a low Fe content, since the par-
ent zeolite consists of homogeneously distributed Fe atoms
and was therefore employed in our study.
2
+
claim that active iron exists in the form of binuclear (Fe )α
complexes, where each (Fe2 )α site would be able to gener-
+
ate α-oxygen when reacting with N2O (Scheme 1).
The binuclear complex B consists of a µ-hydroxo group,
but its role in the formation of active α-oxygen has not been
investigated. Nevertheless, the importance of Brønsted acid
sites for the selective benzene transformation into phenol by
N2O has been reported in the literature [12,15]. Notté [12]
claimed an importance of Brønsted acid sites in addition to
the α-sites primarily linked to iron. The latter study reports
the relation among α-sites, Brønsted acid sites, and catalyst
performance in the phenol formation.
In our previous work [15] the detailed kinetics of benzene
transformation into phenol by N2O was studied. The kinetic
results obtained suggest benzene and nitrous oxide mole-
cules adsorbed on two distinct active sites and reacting by a
Langmuir–Hinshelwood mechanism. The first would be an
active α-site and the second was suggested to be a Brønsted
acid site responsible for the activation of benzene molecule.
The surface reaction proceeds when these two sites are in
close vicinity. The structure B (Scheme 1) could explain the
importance of the OH group for the catalysis during the ben-
zene transformation to phenol. Still the question remains: Is
this µ-proton important only for benzene activation during
its transformation to phenol, or is this structure important
for the formation of active oxygen (Oα) itself? Moreover, no
experimental evidence, with the exception of Panov’s group,
is available on the question: Is every oxygen atom deposited
on the surface from N2O, (O)ad, able to participate in oxida-
tive reactions and can be considered as an active α-oxygen?
The aim of our work is to contribute to the understanding
of the nature of iron sites in the steamed HZSM-5, with a
low content of isomorphous framework-substituted Fe and
to develop a method for the quantitative determination of
active sites in the zeolites under normal pressure and at the
reaction temperature. The number of active sites is defined
as the number of N2O molecules that can be decomposed on
2.1. Catalyst preparation
The initial parent HZSM-51000Fe with Si/Al ratio = 25
was kindly provided by Uetikon SA (Switzerland). The
initial parent HZSM-5200Fe and HZSM-570Fe zeolites with
Si/Al ratio 36–42 (Table 1) were synthesized as reported
elsewhere [20,21]. The synthesis solution was prepared by
adding sodium aluminate (52.5 wt% NaAlO , Riedel–de
Haën), sodium chloride (99.5 wt%, Fluka), and tetrapropyl-
2
ammonium hydroxide (20 wt% in water, Fluka), in deion-
ized water at room temperature. Then, the tetraethyl orthosil-
icate (TEOS, 98%, Merck–Suchardt) was introduced under
vigorous stirring. The molar ratio was as follows: TPA–
OH:Si(OEt) :NaCl:NaAlO :H O = 2.16:5.62:3.43:0.13:
4
2
2
1000. Homogenization of the mixture were performed for
2–3 h. The gel was afterward poured in a 250-ml teflon-lined
autoclave and the temperature was increased within 1 h up
to 444 K. The synthesis took place for 2 days under auto-
genously regulated pressure. The crystalline material was
filtered, washed with deionized water, and dried at 393 K.
After calcination at 873 K in air for 5 h, the zeolite was con-
verted into the H form using ammonia buffer and heated in
nitrogen at 823 K for 5 h.
The main characteristics of the parent zeolites are sum-
marized in Table 1. It is important to note that an iron source

L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
275
Table 1
Characteristics of parent HZSM-5 zeolites with low Fe content
Catalyst
Si/Al atomic
Fe content
ppm
Fe content
SSABET
(m /g)
Pore volume
(cm /g)
−18
2
3
ratio
(atoms ×10
/g)
HZSM-5
HZSM-5
HZSM-5
Uetikon SA
home-made
home-made
25
42
36
1000
200
70
10.7
2.14
0.75
335
370
340
0.32
0.27
0.21
1
2
7
000Fe
00Fe
0Fe
intentionally was not added to the synthesis solution in or-
der to obtain zeolite samples containing a “natural,” impurity
level of Fe. Therefore, the HZSM-5200Fe was prepared with
deionized water, and during the HZSM-570Fe synthesis only
bidistilled water was used.
lowing peaks was measured simultaneously: 4, 18, 28, 30,
32, 40, 44, and 46 m/e.
The amount of a catalyst placed into the reactor was 1.04–
1.09 g. The catalysts were pretreated in He (> 99.999%,
50 ml (STP)/min) at different temperatures 823–1323 K for
1 h and then cooled in He to the temperature of the transient
response experiment (473–653 K). The latter was performed
by a quick switch of the He flow to the flow of gaseous mix-
ture: 2 vol% N2O/2 vol% Ar/96 vol% He at a flow rate
20 ml (STP)/min. Argon served as an inert tracer.
For activation, all zeolites were pretreated by steam
(
310 mbar) in nitrogen at 823 K for 3 h. Before the mea-
surements, the samples were calcined in flow of He for 1 h
at 823 K. This pretreatment will be referred to from now on
as “standard.”
Effect of oxygen pretreatment was also studied after pre-
treatment of the catalyst at 1323 K in He and cooling to
2
.2. Catalyst characterization
8
23 K in He where a 10 vol% O2 mixture in He was in-
troduced for 1 h. Then the reactor was cooled in He to the
temperature of the transient response experiment.
Nitrous oxide reacts with the catalyst resulting in the for-
mation of gaseous nitrogen and oxygen bound by the catalyst
The analysis of chemical composition of the zeolites was
carried out using atomic absorption spectroscopy (AAS)
via a Shimadzu AA-6650 spectrometer with air–acetylene
flame. For these measurements the samples were dissolved
in hot aqua regia containing several drops of HF.
[
reaction (1)]. The amount of active sites was calculated by
integration of the nitrogen peak area. The amount of nitro-
gen formed at 523 and 603 K was the same, indicating a
complete occupation of active sites by oxygen. Formation of
molecular oxygen was observed only at higher temperatures.
To study an effect of water on the amount of active
sites, water was pulsed to the catalyst using a 0.5-µl sy-
ringe via a special injector after the pretreatment of the
catalyst (HZSM-5200Fe and HZSM-51000Fe) and cooling to
The specific surface areas (SSA) of the catalysts were
measured using N2 adsorption–desorption at 77 K via a
Sorptomatic 1990 instrument (Carlo Erba). Before the mea-
◦
surements, the catalysts were heated in vacuum at 250 C for
2
h. The SSA of the samples was calculated employing the
BET method while the Dollimore/Heal method was applied
for the calculation of pore volume.
X-ray diffraction (XRD) patterns of catalysts were ob-
tained on a Siemens D500 diffractometer with Cu-Kα mono-
chromatic radiation (λ = 1.5406 Å).
4
73–523 K. Saturation of the sample with water vapor was
controlled by the mass spectrometer. After a 15-min purge
by He the N2O interaction was performed for 5 min.
Reactivity of adsorbed oxygen was characterized by CO
interaction with the catalyst at 523 K. It was carried out as
follows: after a pretreatment followed by the interaction with
N2O to form (O)ad, the reactor was purged by He for 15 min
and the mixture of 3 vol% CO in He (20 ml (STP)/min) was
introduced for 6 min.
TPD measurements were performed in He (20 ml (STP)/
min) after N2O interaction or after CO interaction with ad-
sorbed oxygen. The gaseous composition was monitored in
the range of 523–1250 K at a heating rate 30 K/min.
Infrared diffuse-reflectance spectroscopy (DRIFTS) stud-
ies were performed via a FT-NIR spectrometer (Perkin-
Elmer 2000) with a TGS detector. The spectra of the samples
were taken in Ar at 823 K after the pretreatment under the
same conditions for 1 h in a high-temperature in situ cell
(
SpectraTech 0030-102). The spectra were obtained by aver-
−1
aging of 64 scans with a resolution of 4 cm with respect
to KBr powder.
2
.3. Transient response, reactivity, and
temperature-programmed desorption studies
The determination of the concentration of active sites, re-
activity, and temperature-programmed (TPD) studies were
performed in a Micromeritics AutoChem 2910 analyzer pro-
vided with a quartz plug-flow reactor. A ThermoStar 200
3. Results
3.1. Method of quantitative determination of active sites
based on the transient response
(
Pfeiffer Vacuum) mass spectrometer was used to control the
composition of the gas phase. It was calibrated using gaseous
mixtures of known compositions. The intensity of the fol-
The transient response method is widely used in catalytic
research. In the present study the gas phase was monitored

276
L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
in the reactor outlet after a step change of gas composition in
the reactor inlet. The gas composition at the outlet as a func-
tion of time after a step change in the reactor inlet from He to
a mixture of 2 vol% N2O, 2 vol% Ar, and 96 vol% He for the
HZSM-51000Fe sample is presented in Fig. 1. The nonideal
reactor behavior, as compared to the plug flow, is represented
by the concentration–time profile of an inert tracer (Ar). In
order to control if internal diffusion affects the responses,
the blank experiment was carried out using quartz beats in-
stead of the zeolite in the catalytic bed. The Ar responses
were similar (Fig. 2), indicating that diffusion limitations do
not play an important role.
with Ar. A certain delay (about 35 s) is observed for N2O.
It can be concluded that when all active sites are blocked,
the reaction stops. No molecular oxygen was detected at
the outlet. The concentration of sites (Cad) is estimated
from the amount of N2 released assuming that a single oxy-
gen atom is chemisorbed on each active site and after the
standard pretreatment the concentration found was Cad =
1
7
4.9 × 10 sites/g.
In order to control if the formation of (O)ad is reversible
after interaction with N2O, the reactor was purged by He
and the transient response experiment with the N2O mix-
ture was repeated (Fig. 1). No evolution of N2 was detected,
indicating that the formation of adsorbed oxygen was irre-
versible. At the same time, a delay in the N2O response was
again observed showing that N2O is reversibly adsorbed. In-
It is seen from Fig. 1 that besides N2O and Ar, also
N2 appears, indicating the stoichiometric reaction (1) with
concomitant formation of (O)ad. N2 appears simultaneously
Fig. 1. Transient response curves obtained after introduction of the 2 vol% N O, 2 vol% Ar mixture in He on the HZSM-5
2
catalyst at 523 K after
1000Fe
pretreatment in He at 823 K for 1 h.
Fig. 2. Transient response curves obtained after introduction of the 2 vol.% N O, 2 vol% Ar mixture in He on quartz beats (triangles), HZSM-5
, and on
1000Fe
2
the same sample after “non-aqueous back Na exchange” (only N response is shown) at 603 K after pretreatment in He at 823 K for 1 h.
2

L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
277
Fig. 3. DRIFT spectra of different HZSM-5
and HZSM-5 catalysts taken with respect to KBr at 823 K after pretreatment in Ar in the in situ
1000Fe
200Fe
DRIFTS cell for 1 h at the same temperature.
tegration of the area between two responses (Ar and N2O)
tional calcination of this catalyst at 1323 K in He results
in a considerable decrease of OH groups, but they are still
observed in the spectra.
allows estimation of the amount of CN O. It is worth not-
2
ing that the amount of reversibly adsorbed N2O is more than
1
8
one order of magnitude higher (CN O = 12.0 ×10 sites/g)
2
compared to the Cad. So, the Cad concentration was always
determined from the N2 released, but not from the amount
of N2O consumed. This is in variance with the static method
of Panov [3], which reported the equality of these amounts
and used both of these values for the determination of the
α-site concentration (corresponding in the present study to
Cad). This variance is attributed to the higher partial pres-
sure of N2O used in the present work (15 Torr instead of
3.3. Na-exchanged zeolites
The main difficulty in zeolite characterization used in this
study is due to the low content of iron, hindering quanti-
tative determination of their physico-chemical parameters.
Therefore, the activated zeolites underwent a chemical treat-
ment. It is known from the literature that sodium exchange
of dealuminated ZSM-5 is not an easy process [12] since
the exchange selectivity of HZSM-5 for various cations fol-
0
.2–0.4 Torr).
+
In order to verify the influence of the temperature on
lows inversely their ionic radii. This favors the H3O with
the concentration of active sites obtained (Cad), the tran-
sient response experiment was carried out at 603 K for
the same catalyst (HZSM-51000Fe) after the standard pre-
treatment (see Experimental). The results are presented in
Fig. 2. The N2 evolution is steeper, but the total amount
of sites determined by the N2-peak integration remains al-
an ionic radius of 1.5 Å versus sodium with an ionic radius of
0.95 Å. Even in the presence of large sodium excess under
reflux and repeated operation, a complete exchange could
not be observed. In order to prevent the active sites formed
during catalyst activation (steaming and calcination) against
+
the effect of water leading to the formation of H3O , the
1
7
most the same: Cad = 5.7 × 10 sites/g. At the same time,
Na exchange was performed from a nonaqueous solution of
sodium methanolate in dried tetrahydrofuran (THF) called
from now on “nonaqueous back Na exchange.” Under these
the amount of reversibly adsorbed N2O diminished about 4
1
8
times: CN O = 3.0 × 10 molecule/g, indicating a temper-
2
+
ature dependence of the adsorption equilibrium constant.
“dry” conditions the ion exchange of H (an ionic radius
+
of 0.35 Å) will be favored toward Na (an ionic radius of
3
.2. DRIFT spectra of the samples after different treatments
0.95 Å) and should proceed to a much higher extend, if com-
pared to the “aqueous conditions.”
Fig. 3 reports the spectra of the HZSM-5200Fe and HZSM-
1000Fe catalysts taken at 823 K after different treatments.
The AAS analysis showed that iron was not removed af-
ter nonaqueous back Na exchange and its content remained
1000 ppm for the sample HZSM51000Fe. The amount of Na
introduced in this sample corresponded to 7650 ppm. In or-
der to characterize Brønsted acidity, the DRIFT spectra were
taken for the Na exchanged and H form of the ZSM-51000Fe
after standard pretreatment. The spectra are presented on the
Fig. 3. It is seen that the amount of OH groups diminished,
5
The spectra consist of three main regions 3735, 3650–3680,
and 3590 corresponding to terminal Si–OH, OH groups at-
tached to extraframework cations, and bridged Si(OH)Al hy-
droxyls, respectively. It is seen from the spectra that zeolite
steaming increases the OH groups on the extraframework
−1
cations (3680 cm ) of the HZSM-5200Fe sample. Addi-

278
L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
especially in the region of more acidic hydroxyl groups
responses are shown in Fig. 4. The amount of (O)ad formed
after N2O contact with zeolites is proportional to the total
amount of iron (see inset in Fig. 4) and is independent on the
Si/Al ratio. This is in agreement with similar observations
for HZSM-5 zeolites with low Fe content reported [4,12].
−1
(
3650 and 3590 cm ), which correspond to OH groups
attached to extraframework cations and bridged Si(OH)Al
hydroxyls.
At the same time the nonaqueous back Na exchange leads
to a complete loss of the HZSM-51000Fe activity in the ben-
zene to phenol transformation [21]. Thus, at 623 K the ben-
zene conversion over HZSM-51000Fe (standard activation)
was 25% with a phenol selectivity of 98%, but no conver-
sion was observed over the same sample after nonaqueous
back Na exchange. These results are at variance with the re-
sults reported by Sobolev et al. [22] who did not observe
any change of activity after Na back exchange. We attribute
this difference to the conditions of Na exchange, which was
performed in a 1 M NaCl aqueous solution. Probably, un-
der these conditions the protons in the active sites were not
exchanged.
The transient response of N2 over the HZSM-51000Fe
sample after nonaqueous back Na exchange is presented on
Fig. 2. As it is seen, only small N2 evolution was observed,
indicating that almost no oxygen was deposited from N2O
during its contact with the Na exchanged zeolite. So, the
presence of Brønsted acidity in the zeolite seems to be im-
portant not only for the benzene reacting to form phenol by
activation of the benzene molecule, but for the (O)ad for-
mation as well. These results are in agreement with the ob-
servations by Notte [12], who reported a decrease of α-site
concentrations simultaneously with a decrease of Brønsted
acid sites.
3.5. Effect of calcination condition on the formation of
active sites
The catalytic activity of Fe-containing HZSM-5 in N2O
decomposition [16,18], genesis of α-oxygen [3], and ben-
zene hydroxylation by N2O [14] have been reported to be
strongly dependent on the pretreatment conditions (catalyst
activation), like temperature, time, and the gas atmosphere
used. The activity in benzene oxidation to phenol and the
concentration of (Oad) may vary by an order of magni-
tude [3], depending on the activation conditions, especially
for the catalysts prepared from the isomorphously substi-
tuted HZSM-5 zeolites with a low Fe content. The necessity
of mild steaming for these zeolites is well understood, but
the conditions needed for the calcination step are not. To
clear the latter points, after “standard steaming” (see Ex-
perimental), the calcination in He flow was performed for
1 h at different temperatures. After cooling the samples
to 523 K, the active site concentration was measured by
N2O decomposition under transient conditions. The results
for HZSM-5200Fe are presented in Fig. 5 showing Cad as
a function of the pretreatment temperature. Apparently, the
calcination temperature strongly affects the Cad value. The
temperature increase from 823 to 1023 K does not change
the amount of active sites, heating further to 1173 K dou-
bled their amount, and the calcination at 1323 K results in a
drastic concentration change of more than one order of mag-
3
.4. Relationship between (O)ad concentration and Fe
content
1
7
18
Transient response experiments were carried out for the
home-made catalysts after standard pretreatment and the N2
nitude (from 1.1 × 10 to 1.2 × 10 sites/g). This agrees
with the observations by Panov and co-workers who reported
Fig. 4. N response obtained after introduction of the 2 vol% N O, 2 vol% Ar mixture in He on the HZSM-5
, HZSM-5
, and HZSM-5
catalysts
2
2
1000Fe
200Fe
70Fe
at 603 K after pretreatment in He at 823 K for 1 h. Concentration of active sites as a function of the Fe content (inset).

L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
279
a sixfold increase in α-site concentrations after calcination
gen calcination (10 vol% O2 in He) at 823 K for 1 h. The
results are shown on Fig. 7, indicating that the (O)ad con-
in air at 1173 K [3]. If the calcination at 1323 K was per-
1
8
18
formed for 8 h, the Cad increased to 2.1 ± 0.1 × 10 sites/g
and reached the stoichiometric maximum in the case of the
sample HZSM-5200Fe with the total amount of Fe atoms of
centration decreased about 20% from 2.1 ± 0.1 × 10 to
1
8
1.75×10 atoms/g in a reproducible way. If the calcination
in He flow at 1323 K was repeated, all active sites were re-
generated. The effect of oxygen on the total amount of (O)ad
was also checked with HZSM-51000Fe calcined in He for 17
h. Similar to the HZSM-5200Fe catalyst, oxygen pretreatment
at 823 K led to a small decrease (∼ 25%) of the total amount
of active sites.
1
8
2
.14 × 10 atoms/g. The response of N2 at 523 K is shown
for the catalyst pretreated in He at 1323 K in Fig. 6. No
molecular oxygen was formed during this run; however, if
the temperature was raised to 653 K, N2O decomposed sto-
ichiometrically forming N2 and O2 with 100% conversion
(
see responses of O2 and N2 in Fig. 6).
According to Panov’s work, α-sites are the Fe(II)-con-
3.6. Effect of water on the formation of (O)ad
taining sites, which cannot be oxidized by molecular oxygen
at 823 K. In order to verify if the oxygen treatment can in-
fluence the Cad value, the transient response was performed
after pretreatment in He at 1323 K followed by an oxy-
The presence of Brønsted acidity, as noted before, seems
to be necessary for oxygen formation from N2O. From an-
other view, a strong decrease of the zeolite activity in the
Fig. 5. Effect of the time and temperature pretreatment in He after steaming on the concentration of the active sites formed in the HZSM-5
catalyst.
200Fe
Fig. 6. Transient response curves obtained after introduction of the 2 vol% N O, 2 vol% Ar mixture in He on HZSM-5
2
at 523 K and at 653 K after
200Fe
pretreatment in He at 1323 K.

280
L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
Fig. 7. Transient response curves obtained after introduction of the 2 vol% N O, 2 vol% Ar mixture in He on HZSM-5
at 523 K after pretreatment in He
200Fe
2
at 1323 K (1) and after the same pretreatment followed by either oxygen (10 vol% O in He) calcination for 1 h at 823 K (2) or presaturation by water vapor
2
at 523 K (3).
presence of water vapors has already been reported [17].
In order to understand this effect, we studied the influence
of water on the (O)ad concentration. After the active site
formation was completed, the sample HZSM-5200Fe was
cooled to 523 K and water was introduced into the reac-
tor by stepwise addition until saturation of the sites, which
may adsorb water irreversibly. The transient response ex-
periments are presented in Fig. 7. It is seen that the Cad
calculated from the N2 released decreases upon introduc-
tion of water. The (O)ad concentration of the sample satu-
rated by water was Cad = (1.0 ± 0.1) × 10 sites/g, cor-
responding to about half of the active sites measured over
dry zeolite. This result indicates that the chemisorption of
oxygen atoms from N2O is strongly affected by the pres-
ence of water. H2O and N2O molecules, both being highly
polar, may compete for adsorption sites. It is interesting
to note that only half of the sites are blocked by water
and another half is still reactive toward N2O decomposi-
tion.
3.7. Reactivity of (O)ad in CO oxidation
The main particularity of the oxygen species formed by
N O decomposition during its contact with (Fe)HZSM-5 is
2
a high reactivity in oxidation reactions [4]. In the present
study, the activity of (O)ad was characterized by the oxida-
tion of CO. The CO responses during the CO interaction
2
with the catalyst containing (O)ad deposited from N O are
presented in Fig. 8. The responses differ depending if the
2
sample was presaturated or not with water before oxygen de-
position from N O. Thus, if (O) is formed on dry zeolite,
1
8
2
ad
a quick formation of CO was seen. The integration of the
2
peak area gives the total amount of CO produced and char-
2
acterizes the zeolite reactivity. For the dry zeolite the total
amount of the (O)ad consumed (or CO formed) corresponds
2
to about 65% of the total amount of (O)ad chemisorbed from
N O (C ). This means that the pool of deposited oxygen
2
ad
(O)ad consists of two different subpools: one with the oxy-
gen capable of reacting with CO and the other subpool with
the (O)ad being not active. The same result was reported by
Sachtler and co-workers during the study of the isotopic ex-
change of oxygen on Fe/MFI [23]. The authors observed
that the number of exchangeable oxygen atoms was smaller
than the number of O atoms adsorbed from N O, and as-
signed this difference to the partial consumption of oxygen
for the reoxidation of prereduced iron species. In the present
study reductive pretreatment of zeolite in hydrogen was not
carried out. But the calcination at high temperature in inert
flow is also known to reduce strongly the Fe centers.
To see the influence of the temperature on the blocking
of the active sites by water, N2O was reacted with zeolite at
4
73 K after H2O was irreversibly adsorbed at the same tem-
perature. The total amount of (Oad) almost did not change, as
1
8
compared to 523 K, being equal to (0.9± 0.1)×10 sites/g.
In order to verify if this effect is valid for other catalysts,
the (O)ad concentration was measured on the sample HZSM-
2
5
1000Fe before and after saturation by water as described
1
8
above. The value Cad = 0.53 × 10 sites/g was obtained
1
8
on wet zeolite in comparison with Cad = 1.01 ×10 sites/g
measured on the dry sample pretreated in He at 1323 K for
The oxidation of zeolite by molecular oxygen (10 vol%
in He) at 823 K before interaction with N2O, diminished
the (O)ad concentration as noted above. Despite this pre-
treatment only ∼ 70% of deposited (O)ad is active in CO
1
h.
Fig. 7 shows a strong decrease in the amount of reversibly
adsorbed N2O after zeolite presaturation by H2O, suggest-
ing that the sites responsible for N2O adsorption were also
affected by H2O.
1
8
oxidation (C = 1.75 ×10 sites/g, (O) active in CO ox-
ad
ad
1
8
idation is only 1.27×10 sites/g) and this is an unexpected

L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
281
Fig. 8. CO response curves obtained during interaction of CO (3 vol% CO in He, 523 K) with oxygen deposited from N O on the HZSM-5 catalyst at
200Fe
2
2
5
(
23 K after pretreatment in He at 1323 K (1) and after the same pretreatment in He followed either by oxygen (10 vol% O in He) calcination for 1 h, 823 K
2
2), or presaturation by water vapor (3.1 × 1019 molecules/g) at 523 K (3).
Table 2
−18
Concentration of oxygen (in O atoms ×10
Conditions of treatment
/g) after 5 min interaction with N O at 523 K. The samples were activated in He at 1323 K for 8 h
2
Deposited from
N O, 523 K
2
Reacted with
CO at 523 K
Desorbed as O in TPD
2
(after He at 1323 K)
up to 1323 K
–
–
1
1
1.9 ± 0.1
2.0 ± 0.1
1.8 ± 0.1
1.8 ± 0.1
1.0 ± 0.1
Not performed
1.4 ± 0.1
1.2 ± 0.1 (sharp at 666 K)
0
0 vol% O , 823 K, 1 h
2
1.3 ± 0.1
0
0 vol% O , 823 K, 1 h
Not performed
Not performed
1.2 ± 0.1 (sharp at 666 K)
2
Presaturation with water
vapor at 523 K
0.8 ± 0.2 (broad at 940 K)
Presaturation with water
vapor at 523 K
1.1 ± 0.1
0
0.7 ± 0.2 (broad at 940 K)
result. This indicates that molecular oxygen oxidized only
a fraction of reduced iron sites and another fraction can be
oxidized only by N2O. The high-temperature calcination in
inert atmosphere (1323 K for 1 h) restored the initial value of
Cad and (O)ad concentration active in CO oxidation, show-
ing that molecular oxygen was able to affect reversibly active
sites.
experiments two types of (O)ad, active and inactive, were
detected upon pre-saturation by water with water, and there-
fore, the TPD measurements were performed over dry
(Fig. 9a) and wet, or presaturated by water before interaction
with N O (Fig. 9b) of the HZSM-5
2
200Fe
sample. The evolu-
tion of O from the dry sample appears as a sharp peak with
2
a maximum at 666 K. Integration of the peak area gives the
Presaturation of zeolites with water drastically affects
the surface reactivity. Not only the amount of O atoms
chemisorbed from N2O becomes lower (see Fig. 7 and Ta-
ble 2), but the oxygen species become inactive. Thus, after
zeolite presaturation by water only half of (O)ad was formed
total amount of (O)ad recombined, which was found equal
to the total amount of “active” (O)ad in CO oxidation. When
TPD was performed from the catalyst reacted with CO, no
O2 evolution was observed (curve after CO, Fig. 9a). This
indicates that (O)ad, which interacts with CO, corresponds
to the oxygen desorbed at 666 K. This is characteristic of
active (O)ad species.
The TPD profile of O2 evolution from the sample preox-
idized by molecular oxygen at 823 K is also presented in
Fig. 9a. The form of the sharp peak at 666 K remains, but
the total amount of the recombined oxygen did not change
noticeably, being in quantitative agreement with the amount
of (O)ad reacted with CO.
1
8
(
(
Cad = 1.1 × 10 sites/g), if compared to the dry zeolite
Fig. 7). Even more important is that this oxygen species was
not active for CO oxidation (Fig. 8, curve 3). A similar ef-
fect of water blocking the Fe sites for the isotopic exchange
was reported by Sachtler and co-workers [23].
3
.8. Temperature-programmed desorption of oxygen
TPD was applied to characterize the ability of (O)ad
species to recombine forming O2. During CO oxidation
The TPD profile of O2 evolution from the “water pre-
saturated” sample (Fig. 9b) is completely different if com-

282
L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
Fig. 9. TPD of oxygen after its deposition from N O (523 K) on the HZSM-5
catalyst: (a) pretreated at 1323 K in He (curve 1) and after the same
2
200Fe
pretreatment followed either by oxygen (10 vol% O in He) calcination for 1 h at 823 K (curve 2) or after CO interaction at 523 K (curve 3); (b) after
2
irreversible adsorption of water vapor at 523 K and after the same procedure followed by interaction with CO.
pared to the dry sample. Oxygen is desorbed showing a
broad peak with a maximum at 940 K, indicating that an-
other mechanism, like a surface spillover may be involved
in (O)ad recombination. If previous to the TPD experiment,
CO was contacted with the pre-saturated by water catalyst,
no CO2 was formed (Fig. 8). At the same time no signifi-
cant change in the TPD profile was observed (response after
CO, Fig. 9b). This confirms that (O)ad formed on a wet sur-
face is inactive. So, the TPD profile for the evolution of O2
can be considered as a characteristic (“fingerprint”) for ac-
tive/inactive species.
4. Discussion
Based on the measurements of active site concentrations
in HZSM-5 zeolites containing Fe in the range from 70 to
1
000 ppm, it was found that the presence of Fe is essential
for the formation of surface oxygen (O)ad species from N2O.
Even Fe concentrations of 200–500 ppm, which are common
in commercial or laboratory samples, make zeolite capable
of (O)ad chemisorptions from N2O and zeolites become ac-
tive in benzene to phenol oxidation by N2O. This conclusion
is in line with observations made by other groups [4,14].
It also follows from the present study that the presence of
Fe in HZSM-5 is required, but not sufficient by itself to allow
surface oxygen genesis and oxidative catalytic activity. The
isomorphously substituted zeolites with Fe in the framework
must be preactivated using mild steaming followed by high-
temperature calcination in an inert atmosphere (see Fig. 5).
It is worth noting that for the dry sample, the total amount
1
8
of desorbed oxygen species ((1.2 ± 0.1) × 10 atoms/g) is
close to the amount removed during CO interaction ((1.3 ±
1
8
0
(
.1) × 10 atoms/g) and corresponds to about 65% of the
O)ad chemisorbed from N2O (concentration of active sites,
Cad = (2.0 ± 0.1) × 10 atoms/g).
1
8

L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
283
An important conclusion is that not all oxygen species
formed on the surface from N2O are chemically active
for CO oxidation: the number of O atoms reacted with
CO forming CO2 is not always identical to the number
of (O)ad deposited from N2O after calcination in He at
1
323 K. The maximum number of active (O)ad species
at temperature 523 K was ∼ 65% from the total amount
of chemisorbed oxygen atoms. Moreover, the chemisorbed
(
O)ad atoms formed on the zeolite presaturated by water
were completely inactive in CO oxidation at 523 K. This
unexpected result is at variance with the observations by
Panov’s group [24], who reported that all O atoms adsorbed
(
which they call α-oxygen) and interacted with CO giving a
stoichiometric amount of CO2. The reason for this disagree-
ment is not clear, but probably is related to the difference in
the catalyst pretreatment. The oxidative treatment by mole-
cular oxygen at 823 K was observed to decrease the total
amount of (O)ad with a concomitant decrease in the concen-
tration of (O)ad active in the CO oxidation. By the Panov
group definition, the identity between the O species formed
on the surface from N2O (α-sites) and the O species active in
oxidative transformations is implicit. This assumption may
be misleading when trying to explain the catalyst reactivity,
since it may change drastically depending on the zeolite pre-
treatment.
In order to discuss the structure of iron-complex and ac-
tive oxygen formation, based on the results of this study
and taking into account the results reported on benzene
to phenol transformations [3,12,15,21], some additional as-
sumptions are necessary. The recent studies by the group
of Sachtler on N2O decomposition must be taken into con-
sideration, despite the fact that the zeolites were prepared
differently and contained a high amount of iron (Al/Fe = 1,
CFe = 4 wt%) [17,23,25,26].The active sites participating in
the aforementioned processes seem to be identical, and seem
to involve a binuclear Fe complex. Contrary to the present
study, which deals mainly with the sample having a low con-
tent of iron, most of the studies were performed on zeolites
with a high concentration of Fe. In these cases, besides an
active binuclear Fe complex, other forms of iron may be
present, as was confirmed experimentally. Combining these
results with recent theoretical studies [27–29], a tentative
structure for the active center and its formation can be pro-
posed.
+
2
Scheme 2. Structure of [Fe O H ] cluster formed during high-temperature
2
calcinations of ZSM-5
in inert flow and formation from N O of paired
(O) atoms active (III) in CO oxidation. Pathway of water interaction with
[Fe2O2H ] clusters and formation from N2O a single (O)ad inactive in CO
2
00Fe
2
ad
+
oxidation.
two lattice charges, we include in our scheme only the first
one. The cluster with structure I has only one Al atom in the
vicinity for the charge compensation. This, in turn, increases
the probability that such complexes exist, especially in the
zeolites with high Si/Al ratios. Calcination at high tempera-
ture in inert atmosphere leads to dehydroxylation (as shown
by DRIFTS experiments in this study) and concomitant au-
toreduction of binuclear center I, forming binuclear structure
II according to the Scheme 2.
The existence of a reduced binuclear iron complex with
bridging oxygen was previously postulated by Sachtler et
al. [17,23], and Panov et al. [3,32,33] based on studies by
IR, Mössbauer, ESR spectroscopies, and TPR. Literature re-
sults indicate that the Fe complexes should be located at
the ion-exchange positions of the zeolite, having an overall
monopositive charge. The structure of binuclear Fe com-
plex II requires an Al-centered tetrahedral configuration in
its vicinity for charge compensation. In Scheme 2 this is
Water vapor was reported to accelerate the hydrolysis
of Si–O–Fe bonds, leading to migration of Fe atoms from
the crystalline lattice into the micropore space of the zeo-
lite, thus facilitating Fe ex-framework formation. The ex-
framework Fe center consists of a single metal oxo-cations
+
+
(
[FeO] , or [Fe(OH) ] , as proposed by Bell and co-
2
workers [30] and Sachtler and co-workers [17]. The steam-
ing and calcination seem to induce surface migration form-
ing a binuclear Fe center I (see Scheme 2). The presence of
such binuclear Fe clusters in FeZSM-5 has been proven ex-
perimentally [19]. Although Prins and co-workers [19,31]
considered two types of Fe clusters compensating one or
−
represented as Z , indicating that the zeolite framework it-
self {(Si,Al)On} acts as a crystalline acidic anion [34,35].
−
The presence of proton in the diamond core equilibrates the
charge of the [Fe2O2]2 complex. In this form it can act as
+
+
extraframework monopositive [Fe2O2H] counterion.

284
L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
The structure of active binuclear Fe center II is rather dif-
ferent from that proposed by Sachtler and co-workers [17]
[
(Fe2O3 nanoparticles, single metal oxo-cations, framework
ions, etc.), as observed by other groups for this type of zeo-
lite, but these iron species were reported to be not catalyti-
cally active.
Temperature-programmed desorption results also support
the structure stated above: coupling of neighboring (O)ad
in structure III is more easy, if compared to the single
Fe–O–Fe]2 , which features both metal atoms bound at
+
ion-exchange positions. The latter structure is also different
compared to structure B proposed by Panov (see Scheme 1),
which consists of a µ-hydroxo group, but does not have
the second oxygen atom bound to both Fe atoms. We pro-
+
pose the active Fe center in the form of [Fe2O2H] clus-
(O)ad atoms recombination from the structure IIIH O, which
2
ter featuring a “diamond core” structure similar to that of
the monooxygenase enzyme [5,6,36,37]. From our point of
view, structure II follows from the reductive dehydroxyla-
tion of complex I, the existence of which was confirmed
by EXAFS experiments by Prins and co-workers [19]. No
other structures of Fe-containing species are considered in
this study due to the fact that for the sample HZSM-5200Fe
all Fe atoms (2.1×10 atoms/g) were observed to be active
toward (O)ad formation and water adsorption diminished this
amount by one-half. For sample HZSM-51000Fe, only part of
Fe atoms was active, but after saturation by water a half also
remained active toward (O)ad formation. We consider this re-
sult as an indirect sign of paired arrangement of Fe atoms in
the active center.
When structure II reacts with N2O, structure III is
formed with two terminal ferryl groups [Fe=O]. This is in
line with the prediction by DFT analysis for a terminal Fe–
O group distance of 1.61 Å, indicating strong double-bond
character [28]. We assume that CO interacts with the active
terminal oxygen atoms forming CO2.
should involve surface diffusion. Therefore, the O2 evolu-
tion from complex III appears as a sharp peak at 666 K in
the TPD profile, and from the complex IIIH O as a broad fea-
2
ture between 600 and 1250 K with the maximum at 940 K.
The structure of the binuclear Fe center proposed explains
the results obtained for the activity of (O)ad in CO oxidation.
As shown the Scheme 2, structure II can accept two O atoms
leading to the form III. We assume that CO interacts with
active terminal oxygen atoms of structure III forming CO2.
1
8
+
Only paired terminal O atoms adsorbed on the [Fe2O2H]
sites of the binuclear center seem to be active in CO interac-
tion, and the single Oad atoms generated on an Fe site in the
hydroxylated binuclear complex (IIIH O) are inactive. This
2
leads us to the conclusion about the importance in redox
processes of the paired arrangement of active terminal oxy-
gens bound to two Fe atoms in a diamond core geometry in
analogy to the binuclear structure of active Fe sites in MMO.
Another explanation of inactivity of single Oad-atoms in the
complex (IIIH O) is based on its instability due to easy elec-
2
2
+
4+
tron transfer from Fe to Fe resulting in the structure I.
This activation becomes impossible in the complex saturated
by water.
The effect of zeolite presaturation by irreversibly ad-
sorbed water is also shown in Scheme 2. The adsorption of
one water molecule proceeds on two neighboring sites (O–
2
+
Fe –) forming structure IIH O, and the binuclear Fe center
2
becomes a single Fe site. Therefore, the remaining part of the
binuclear diamond core cannot accept a second H2O mole-
cule. But it is able to adsorb N2O forming terminal oxygen
5. Conclusions
1. A method for quantitative determination of the active
Fe-containing sites responsible for the formation of
(O)ad species from N2O is developed based on the
transient response with a sensitivity level up to few
dozen ppm of Fe.
2. The Fe presence in HZSM-5 is required, but not suffi-
cient, for the surface-active site formation. The isomor-
phously substituted zeolites with a low content Fe in
the framework must be preactivated using mild steaming
followed by high-temperature calcinations in an inert at-
mosphere.
in the structure IIH O and releasing N2.
2
The assumption about H2O blocking half of the Fe atoms
may explain “a magic number” of 0.5 × Cad obtained af-
ter saturation by water in two of the samples studied. The
hydroxylated form of a binuclear Fe center acts as a single
Fe site. This conclusion is in agreement with the results of
DFT calculations [28], which showed that the hydroxylated
form of a binuclear Fe center consists of one Fe ion com-
pletely saturated by water in the first coordination sphere.
The result with H2O blocking one-half of the Fe atoms also
indirectly confirms the presence of proton in the binuclear
diamond core, since otherwise two molecules of H2O could
3. Based on the literature results and the findings of this
study a tentative structure of an active binuclear Fe cen-
ter featuring a diamond core geometry, similar to that
of the monooxygenase enzyme, is suggested. It includes
be accommodated during adsorption on a [Fe2O2]2 center.
It is important to note that the ability of zeolite to form
active sites in the form of a binuclear Fe center depends
on different factors like total Fe concentration, pretreatment
conditions, and probably, Si/Al ratio. Thus, in the sample
ZSM-5200Fe all the Fe atoms introduced in the zeolite were
involved in the active sites forming (O)ad, contrary to the
ZSM-51000Fe sample, in which only a part of the Fe atoms
was active. So, Fe atoms may be present in different forms
+
+
one H ion and has an overall monopositive charge
+
[Fe2O2H] .
4. Only a part of (O)ad adsorbed from N2O was active in
CO oxidation at 523 K. This active oxygen quantita-
tively corresponded to the O atoms, which appeared in
the TPD profile as a sharp peak at 666 K, indicating a
desorption mechanism via direct recombination. These

L. Kiwi-Minsker et al. / Journal of Catalysis 219 (2003) 273–285
285
active O atoms were assigned to the paired terminal oxy-
gens each bound to one Fe site in the diferric [Fe2O2H]
diamond core cluster.
[11] M. Hafele, A. Reitzmann, D. Roppelt, G. Emig, Appl. Catal. A 150
1997) 153.
+
(
[12] P.P. Notte, Top. Catal. 13 (2000) 387.
[13] A. Ribera, I. Arends, S. de Vries, J. Perez-Ramirez, R.A. Sheldon,
J. Catal. 195 (2000) 287.
5
. Zeolite presaturated by water vapor, generates (O)ad
species from N2O, which appeared in a TPD profile
as a broad peak with a maximum at 940 K, indicat-
ing a slow, probably diffusion-controlled recombination
mechanism. The total amount of the adsorbed oxygen
corresponded to half of the stoichiometric amount of
Fe atoms in the sample. This suggests that water blocks
one-half of the binuclear center, and the remaining part
acts as a single Fe site. The (O)ad species adsorbed on
single Fe sites were completely inactive in the CO oxi-
dation.
[14] P. Kubanek, B. Wichterlova, Z. Sobalik, J. Catal. 211 (2002) 109.
[15] B. Louis, L. Kiwi-Minsker, P. Reuse, A. Renken, Indust. Eng. Chem.
Res. 40 (2001) 1454.
[
16] J. Perez-Ramirez, F. Kapteijn, G. Mul, J.A. Moulijn, Appl. Catal. B 35
2002) 227.
17] E.M. El-Malki, R.A. van Santen, W.M.H. Sachtler, J. Catal. 196 (2000)
12.
(
[
2
[18] Q. Zhu, B.L. Mojet, R.A.J. Janssen, E.J.M. Hensen, J. van Grondelle,
P. Magusin, R.A. van Santen, Catal. Lett. 81 (2002) 205.
[
[
[
19] P. Marturano, L. Drozdova, A. Kogelbauer, R. Prins, J. Catal. 192
2000) 236.
20] B. Louis, C. Tezel, L. Kiwi-Minsker, A. Renken, Catal. Today 69
2001) 365.
21] Louis B. PhD thesis, EPFL, 2002.
(
(
Acknowledgments
[22] V.I. Sobolev, K.A. Dubkov, E.A. Paukshtis, L.V. Pirutko, A.S. Khari-
tonov, G.I. Panov, M.A. Rodkin, Appl. Catal. A 141 (1996) 185.
[
[
[
23] J.F. Jia, B. Wen, W.M.H. Sachtler, J. Catal. 210 (2002) 453.
24] K.A. Dubkov, E.A. Paukshtis, G.I. Panov, Kinet. Catal. 42 (2001) 205.
25] E.M. El-Malki, R.A. van Santen, W.M.H. Sachtler, Micropor. Meso-
por. Mater. 35–6 (2000) 235.
Authors highly appreciate the contributions of Dr. B. Louis,
E. Casali, and A. Udriot to this work. Financial support from
Swiss National Science Foundation is gratefully acknowl-
edged.
[26] Z.X. Gao, S. Qi, W.M.H. Sachtler, Appl. Catal. B 33 (2001) 9.
[27] A.L. Yakovlev, G.M. Zhidomirov, R.A. van Santen, J. Phys. Chem.
B 105 (2001) 12297.
[
[
28] A.L. Yakovlev, G.M. Zhidomirov, Catal. Lett. 63 (1999) 91.
29] G.M. Zhidomirov, A.L. Yakovlev, M.A. Milov, N.A. Kachurovskaya,
I.V. Yudanov, Catal. Today 51 (1999) 397.
References
[
[
[
1] A. Corma, Stud. Surf. Sci. Catal. 94 (1995) 736.
[30] J.A. Ryder, A.K. Chakraborty, A.T. Bell, J. Phys. Chem. B 106 (2002)
059.
2] D.E.D. Vos, B.F. Sels, P.A. Jacobs, Cattech. 6 (2002) 14.
3] K.A. Dubkov, N.S. Ovanesyan, A.A. Shteinman, E.V. Starokon, G.I.
Panov, J. Catal. 207 (2002) 341.
7
[31] P. Marturano, L. Drozdova, G.D. Pirngruber, A. Kogelbauer, R. Prins,
Phys. Chem. Chem. Phys. 3 (2001) 5585.
[4] G.I. Panov, A.K. Uriarte, M.A. Rodkin, V.I. Sobolev, Catal. Today 41
[32] G.I. Panov, V.I. Sobolev, K.A. Dubkov, V.N. Parmon, N.S. Ovanesyan,
A.E. Shilov, A.A. Shteinman, Reaction Kinet. Catal. Lett. 61 (1997)
(1998) 365.
[
[
5] A.A. Shteinman, Russ. Chem. Bull. 50 (2001) 1795.
6] L.J. Shu, J.C. Nesheim, K. Kauffmann, E. Munck, J.D. Lipscomb, L.
Que, Science 275 (1997) 515.
251.
[
[
[
33] G.I. Panov, Cattech. 4 (2000) 18.
34] A. Corma, Chem. Rev. 95 (1995) 559.
35] I. Balint, M.-A. Springel-Huet, K.-I. Aika, J. Fraissard, Phys. Chem.
Chem. Phys. 1 (1999) 3845.
[
[
[
7] L. Que, Pure Appl. Chem. 70 (1998) 947.
8] D.A. Kopp, S.J. Lippard, Curr. Opin. Chem. Biol. 6 (2002) 568.
9] J.L. Motz, H. Heinichen, W.F. Holderich, J. Mol. Catal. A 136 (1998)
[36] A.C. Rosenzweig, P. Nordlund, P.M. Takahara, C.A. Frederick,
S.J. Lippard, Chem. Biol. 2 (1995) 409.
175.
[
10] L.M. Kustov, A.L. Tarasov, V.I. Bogdan, A.A. Tyrlov, J.W. Fulmer,
[37] B.J. Brazeau, R.N. Austin, C. Tarr, J.T. Groves, J.D. Lipscomb, J. Am.
Chem. Soc. 123 (2001) 11831.
Catal. Today 61 (2000) 123.