In-situ study of MCM-41-supported iron oxide catalysts by XANES and EXAFS

Abstract

Our study focuses on the structural evolution of MCM-41-supported iron oxide under the reducing environment of catalyst pretreatment and ethylbenzene dehydrogenation reaction. Powder X-ray diffraction (XRD) analysis showed that the iron oxide is well-dispersed on the surface of the support with no detectable diffraction peaks from iron oxide. X-ray absorption near edge structure (XANES) study indicates that iron oxide is being reduced during catalyst pretreatment under flowing helium from α-Fe2O3 at room temperature to Fe3O4 at 425°C. At 500°C, the oxide species is reduced even further. Curve-fitting analysis of the extended X-ray absorption fine structure (EXAFS) radial distribution function (RDF) profile of the catalyst pretreated at 500°C can be done with a basic tetragonal γ-Fe2O3 spinel structure. However, the cationic vacancies of the spinel on the octahedral position are almost filled with iron cations, indicating that the structure of the iron oxide species is approaching that of a ccp FeO. Stabilization of the FeO-like structure formed at 500°C is probably through iron oxide-support interactions, especially via condensation of the oxide terminal hydroxyl groups with the silanols of MCM-41. This distorted form of iron oxide species is metastable and contains labile surface oxide anions, which are probably responsible for the high initial catalytic activity during ethylbenzene dehydrogenation reaction at 500°C. In the presence of the reactant, however, the iron oxide is further reduced and metallic iron is formed during the reaction. The formation of metallic iron is probably through fragmentation of FeO particles, as shown by catalysis and EXAFS results. The reduction process contributes mainly to the deactivation of the catalyst.

Full text

Page 1

Applied Catalysis A: General 198 (2000) 115–126
In-situ study of MCM-41-supported iron oxide catalysts by
XANES and EXAFS
She-Tin Wonga, Jyh-Fu Leeb, Soofin Chenga, Chung-Yuan Moua,∗
aDepartment of Chemistry, National Taiwan University, 1, Roosevelt Road Section 4, Taipei, Taiwan
bSynchrotron Radiation Research Center, Hsinchu, Taiwan
Received 24 August 1999; received in revised form 24 November 1999; accepted 24 November 1999
Abstract
Our study focuses on the structural evolution of MCM-41-supported iron oxide under the reducing environment of catalyst
pretreatmentandethylbenzenedehydrogenationreaction.PowderX-raydiffraction(XRD)analysisshowedthattheironoxide
is well-dispersed on the surface of the support with no detectable diffraction peaks from iron oxide. X-ray absorption near
edge structure (XANES) study indicates that iron oxide is being reduced during catalyst pretreatment under flowing helium
from ?-Fe2O3at room temperature to Fe3O4at 425◦C. At 500◦C, the oxide species is reduced even further. Curve-fitting
analysis of the extended X-ray absorption fine structure (EXAFS) radial distribution function (RDF) profile of the catalyst
pretreated at 500◦C can be done with a basic tetragonal ?-Fe2O3spinel structure. However, the cationic vacancies of the
spinel on the octahedral position are almost filled with iron cations, indicating that the structure of the iron oxide species is
approaching that of a ccp FeO. Stabilization of the FeO-like structure formed at 500◦C is probably through iron oxide-support
interactions, especially via condensation of the oxide terminal hydroxyl groups with the silanols of MCM-41. This distorted
form of iron oxide species is metastable and contains labile surface oxide anions, which are probably responsible for the high
initial catalytic activity during ethylbenzene dehydrogenation reaction at 500◦C. In the presence of the reactant, however,
the iron oxide is further reduced and metallic iron is formed during the reaction. The formation of metallic iron is probably
through fragmentation of FeO particles, as shown by catalysis and EXAFS results. The reduction process contributes mainly
to the deactivation of the catalyst. ©2000 Elsevier Science B.V. All rights reserved.
Keywords: MCM-41; Iron oxide; XANES; EXAFS; Ethylbenzene dehydrogenation
1. Introduction
In heterogeneous catalysis, the knowledge of cat-
alyst structure is essential in order to understand the
chemistry actually occurring on the surface of a cat-
alyst. This makes structure characterization crucial
throughout the life cycle of the catalyst, from the
∗Corresponding author. Fax: +886-2-2366-0954.
E-mail address: cymou@ms.cc.ntu.edu.tw (C.-Y. Mou)
preparation step to the use in catalytic reaction. How-
ever, the majority of these studies were carried out in
an ex-situ manner where the conditions of the catalyst
are far from those generally present in a catalytic re-
actor. Besides, the phase composition and the texture
of the catalyst may change under the reaction con-
ditions, especially when the reaction temperature is
very high. Some studies have shown that static sur-
face structures and reaction mechanisms in vacuo are
generally different from dynamic ones under reaction
0926-860X/00/$ – see front matter ©2000 Elsevier Science B.V. All rights reserved.
PII: S0926-860X(99)00516-5

Page 2

116
S.-T. Wong et al./Applied Catalysis A: General 198 (2000) 115–126
conditions [1–5]. Thus, in-situ study has become in-
creasingly important, since the real chemistry of the
surface can only be understood by approaching the
real reaction condition as much as possible.
In the last two decades, due to the significant im-
provementofinstrumentationandcomputerpower,the
easier access to synchrotron radiation sources and the
development of powerful software for structural mod-
eling, it is possible to carry out structural analysis on
well-dispersed catalysts using X-ray absorption tech-
niques such as X-ray absorption near edge structure
(XANES) and extended X-ray absorption fine struc-
ture (EXAFS) [6,7]. This evolution is of paramount
importance for a supported metal or metal oxide cat-
alyst, since the active phase of an efficient catalyst
mostly exists in the form of highly dispersed state on
high surface area supports. In this regard, the main ad-
vantage of X-ray absorption spectroscopy over other
techniques is that it provides direct information on
charge density and the local environment of a specific
absorbing atom without the requirement of long-range
ordered structure and UHV conditions.
In this paper, we apply the XANES and EXAFS
techniques to an in-situ study of ethylbenzene dehy-
drogenation on iron oxide catalyst. EXAFS techniques
have been used to study the catalyst system in ethyl-
benzene dehydrogenation reaction but not in an in-situ
manner [8]. Moreover, the catalyst system studied was
too complicated.
The nature of the reaction centers for ethylben-
zene dehydrogenation is still a matter of controversy.
Numerous models have been put forward, which is
reasonable considering the vast number of catalyst
systems and experimental conditions studied. Among
these models, two are of particular interest to us. The
first one involves KFeO2as the active phase as pro-
posed by Hirano [9]. However, we believe that the
involvement of crystalline KFeO2phase in the reac-
tion process is not very likely since the direct syn-
thesis of KFeO2requires a much higher temperature
(1000◦C) than the usual reaction temperature of ethyl-
benzene dehydrogenation (550–600◦C). Another in-
teresting model is suggested by Koppe et al. [10],
involving potassium-doped ?-Fe2O3as the active in-
gredient. Here, the incorporation of potassium ions in
the ?-Fe2O3lattice sites leads to a defective ?-Fe2O3
structure. In general, all the models proposed in the
study of ethylbenzene dehydrogenation point toward
the importance of a K–O–Fe3+entity as the active
center of the catalyst.
As in the case of other related studies, the primary
goal of this study is aimed at unveiling the problem of
the active center(s) involved in the ethylbenzene de-
hydrogenation reaction. However, our basic approach
is to simplify the reaction system as much as pos-
sible, i.e. using an unpromoted catalyst and without
the cofeeding steam in the reactant. The basis of our
consideration is that the promoter such as the potas-
sium species used in the industrial catalysts only mod-
ifies but does not alter the nature of the active cen-
ter, e.g. the electronic density of iron center may have
changed, giving rise to an improved catalytic activ-
ity and styrene selectivity. In fact, literature studies
have shown that the active centers are the same for the
promoted and unpromoted catalysts [11,12]. But the
steam in the reaction stream only functions as a dilu-
ent and catalyst regenerating agent [13]. The absence
of steam will thus allow us to follow the catalyst de-
activation process more easily.
We are also interested in the study of supported
catalyst systems. It is known that MCM-41 has high
surface area and porosity [14], so it is possible to use
it as a support for preparing highly dispersed metal
or metal oxide catalysts. A direct strategy in catalyst
preparation is to use the mesoporous silica as a host
for reactive chemical species through the deposition or
impregnation of various organometallic precursors. In
addition, the surface of MCM-41 contains an abundant
supply of silanol groups; therefore, a wide variety of
stable and surface-anchored species can be made suc-
cessfully by this method via the surface reaction with
the silanol groups on the wall of the MCM-41 chan-
nels. Some of the MCM-41 supported species have
been claimed to display high catalytic activities [15].
On the other hand, there is very limited work in elu-
cidating the structures of these MCM-41 supported
species, particularly so in an attempt to correlate the
structures with their catalytic activities.
In this paper, we show that XANES and EX-
AFS analysis of the MCM-41 supported iron oxide
catalysts can provide structural information on the
surface-attached Fe–O species. From the proposed
structural model, active site(s) for ethylbenzene dehy-
drogenation reaction is proposed. Special interest has
been focused on correlating the ethylbenzene dehy-
drogenation activity and the mobility of iron cations

Page 3

S.-T. Wong et al./Applied Catalysis A: General 198 (2000) 115–126
117
in iron oxide, particularly so in the case of Fe3O4
spinel. A detailed discussion of the catalyst system
and catalytic performance is the subject of another
publication, but information related to this study is
included in Section 3 [16].
2. Experimental
2.1. Catalyst preparation and characterization
The method used for the synthesis of MCM-41
was adapted from our published literature [17].
MCM-41-supported catalysts were prepared by im-
pregnating the MCM-41 support with an aqueous
solution of ammonium iron(III) oxalate trihydrate,
(NH4)3Fe(C2O4)3·3H2O, or ammonium heptamolyb-
date, (NH4)6Mo7O24·4H2O. After being dried at
100◦C, the samples were calcined at 500◦C for 6h.
All the catalysts used in this study have a metal
loading of 6wt.%.
The catalyst and support were characterized with
various techniques such as powder X-ray diffraction
(XRD), surface area and pore size measurements, and
we found that the catalyst retained much of the peri-
odic nano-structure of the pristine MCM-41.
2.2. Catalytic reaction
Reactions were carried out in a continuous flow
micro-reactor system at atmospheric pressure. The
reactant, ethylbenzene, was injected continuously
(2.72mlh−1) into the nitrogen carrier gas stream
(effluent flow rate=30mlmin−1) and the reaction
product (gas and liquid) was analyzed off-lined by a
Shimadzu GC-7A gas chromatograph. Liquid product
was collected by a condenser (10◦C) positioned at the
outlet of the reactor and the components were sep-
arated with a packed column (5% SP-1200+1.75%
Bentone 34 on 100/120 Suplecoport, 6ft). Catalyst
regeneration was done at 500◦C under an air flow
of about 75mlmin−1for 1 day (iron sample) or 6h
(molybdenum sample). The amount of carbon left
on the catalyst surface after the reaction was deter-
mined with a Perkin–Elmer CHN-2400 elemental
analyzer.
Total conversion of ethylbenzene is defined as the
percentage of ethylbenzene converted to hydrocarbon
products. The products are light hydrocarbons, ben-
zene, toluene, styrene and polyaromatic compounds
such as dimers and trimers of styrene. Styrene selec-
tivity is defined as the percentage of styrene in the
observed product. Both the conversion and selectivity
data were based on the amount of carbon.
2.3. XANES and EXAFS studies
2.3.1. X-ray absorption measurements
X-ray absorption measurements were carried out
with synchrotron radiation using the EXAFS facilities
installed at the Synchrotron Radiation Research Cen-
ter, Hsinchu, Taiwan. The storage ring was operated
at about 1.5GeV with about 200mA ring current. For
the in-situ experiment, finely powdered sample was
pressed into pellet (≤1mm thick) and sat within a
sample holder. The temperature of the pellet can be
varied by changing the temperature of a home-made
in-situ cell. The carrier gas used is helium, which
was bubbled through an ethylbenzene reservoir at a
flow rate of about 30mlmin−1. The data for ex-situ
experiment was recorded at room temperature with
the sample uniformly spread onto a strip of adhesive
tape, which was then folded into two layers to obtain
an optimum absorption jump (?µt≈1) enough to be
free from thickness and pinhole effects [18]. All the
data were recorded in transmission mode using an Si
(111) channel-cut monochromator. Intensities of the
incident and transmitted beams were measured in ion-
ization chambers filled with a gas mixture of higher
density.
2.3.2. Data analyses
All the experimental spectra were analyzed with a
standard procedure similar to those outlined in [18].
The threshold energy (E0) of ?-Fe2O3determined at
half-height of the absorption edge jump (1s→4s tran-
sition) was 7123.2eV, which is in close agreement
with 7126.4eV reported by Sankar et al. [19]. The re-
sulting EXAFS spectra were k3-weighted in order to
compensate for the attenuation of EXAFS amplitude
at high k and then Fourier transformed in the range of
about2.55Å−1≤k≤11.5Å−1withaHanningapodiza-
tion function of dk=2Å−1.
To avoid the unnecessary computations in the
course of EXAFS fitting of the experimental spectrum

Page 4

118
S.-T. Wong et al./Applied Catalysis A: General 198 (2000) 115–126
(using feffit program version 2.32), the coordination
number (N) was fixed to the crystallographic value
of known compounds and the amplitude reduction
factor (S2
0) was set equal to 0.9 for all compounds
[18]. We also varied the mean square displacement
of path distance or Debye–Waller Factor (σ2
the threshold energy difference (?E0) for all paths
in order to optimize the fit of the theoretical data to
the experimental data. However, the magnitude of σ2
was constrained to reasonable values by referring to
the published results on σ2versus temperature rela-
tionship by Yokoyama et al. [20]. The hypothetical
structure used in the fitting process was obtained
from the published results of Shmakov et al. [21].
In this case, the atomic coordinates of a basic cubic
spinel-like unit cell of ?-Fe2O3 were fitted with a
tetragonal space group (Hermann–Maguin Notation:
P 41212) rather than the reported cubic space group
(P 4332). Nevertheless, a tetragonal basic unit cell
of ?-Fe2O3has been reported by Greaves [22] with
lattice constants a=8.3396Å and c=8.3221Å.
i) and
i
3. Results
3.1. Catalytic system
MCM-41 was chosen as the catalyst support be-
cause of its high surface area (1100m2g−1), which
enables a better dispersion of catalytic active materi-
als. In our MCM-41-supported iron oxide catalyst, for
example, iron oxide is highly dispersed on MCM-41
surfaces since no XRD peaks other than the pris-
tine MCM-41 are observed. The persistence of the
MCM-41 diffraction peaks indicates that the struc-
ture of MCM-41 remained intact during the catalyst
preparation process. The catalyst still retained much
of the surface area (774m2g−1) of MCM-41, but the
porosity of the catalyst is significantly lower (from
0.93 to 0.53mlg−1). Since the catalyst loading is only
6wt.%, the decrease in porosity should be due to the
partial collapse of the uni-dimensional channels of
MCM-41.
We have previously studied a series of MCM-41-
supported molybdenum oxide and iron oxide catalysts
in ethylbenzene dehydrogenation reaction. Depending
on the method of catalyst preparation, large surface
area MCM-41 supported catalysts can have 1.5–2.4
Fig. 1. Catalytic performance of MCM-41-supported iron oxide
catalysts at 500◦C.
times increases in steady state activity over amorphous
silica supported ones (surface area=377m2g−1).
We also found that molybdenum-based catalyst is
about 3.3 times higher in steady state activity than
iron-based catalyst. On the other hand, the selectivity
of styrene, which is the major product, is comparable
(93% versus 87%). However, due to the limitation
of our synchrotron source energy, we are not able
to carry out X-ray absorption study on molybdenum
samples, hence, only the results for iron catalysts are
presented.
Fig. 1 shows the catalytic performance of both fresh
and regenerated MCM-41-supported iron oxide cat-
alysts in the ethylbenzene dehydrogenation reaction.
The activity of the catalyst is high initially, but decays
at increasing time on stream to a steady state value. It
is possible that different active centers are involved in
the initial and steady stages of reaction. We also found
that the activity profile of the fresh catalyst fluctuates
during the initial period of reaction. This is not an ex-
perimental artifact, since a similar but opposite trend
is also observed for styrene selectivity. A possible
explanation is the fragmentation of large iron oxide
particles into smaller micro-particles via a reductive

Page 5

S.-T. Wong et al./Applied Catalysis A: General 198 (2000) 115–126
119
process. The participation of fresh active centers on
the newly exposed surface has an added effect on
the reaction activity. On the other hand, the styrene
selectivity varies inversely with activity since a more
active surface tends to catalyze more secondary re-
actions (refer to the first selectivity data). Secondary
reactions include styrene or ethylbenzene cracking
(giving benzene, toluene and light hydrocarbon) and
styrene oligomerization. The styrene selectivity fi-
nally stabilized at ≈87% after the fragmentation
process was completed. This behavior diminished
in the regenerated catalyst due to improved disper-
sion of iron oxide after a reduction–oxidation cycle.
Now, the particle size of iron oxide must be small
and the activity and styrene selectivity will decrease
monotonously with time on stream due to catalyst
deactivation.
Largerfluctuationsinactivityarealsoobservedover
freshMCM-41-supportedmolybdenumoxidecatalyst,
but at a lower reaction temperature of 400◦C. Here, the
styrene selectivity was even lower initially (≈23%),
indicating the abundance of cracking and oligomer-
ization products, but finally stabilized at ≈93% within
3h of reaction. When the reaction was carried out at
500◦C, the reductive disintegration process was too
fast to be observed.
The reliability of the X-ray in-situ cell in ethylben-
zene dehydrogenation reaction was also checked. A
much higher conversion of ethylbenzene (up to about
10-fold) than the catalytic micro-reactor was obtained
especiallyduringtheinitialstagesofreactionwhilethe
styrene selectivity was comparable. It should be em-
phasized that the design of the catalytic micro-reactor
is completely different from that of the in-situ cell so
that direct quantitative comparison of results should
be avoided. Only qualitative comparison should be
envisaged.
In this paper, we try to provide results from X-ray
absorption studies related to the two phenomena ob-
served in our catalytic dehydrogenation reaction: (i)
the nature of active sites involved in the initial and
steady states of reaction; (ii) reductive disintegration
of iron oxide during the process of reaction. Since
both of them are continuous processes, and in order
to avoid contamination, we carried out our study in an
in-situ manner. Besides, X-ray absorption techniques
can also provide us direct information on the structural
evolution of iron oxide at various temperatures.
Fig. 2. Fe K-edge XANES spectra of MCM-41-supported iron
oxide catalyst pretreated at different temperatures.
3.2. Carbon analysis
The carbon content of the catalyst sample after
the in-situ X-ray absorption study was also analyzed.
The amount of carbon left on the surface is less than
1.6wt.% which corresponds to nearly 1.2 atoms of
carbon per atom of iron. A comparable value was also
obtained in the actual catalytic run.
3.3. XANES study
In Fig. 2, we present a typical iron K-edge XANES
spectra of MCM-41-supported iron oxide catalyst pre-
treated at various temperatures. For the purpose of
comparison,thesespectrawererecordedafterthesam-
ple was maintained at similar time intervals (<0.5h)
at each temperature. As the pretreatment temperature
of the catalyst rises, the absorption edge shifts pro-
gressively towards lower absorption energy. With ref-
erence to the standard compounds, we found that the
iron oxide in the as-synthesized catalyst (at room tem-
perature) has a nature similar to that of ?-Fe2O3,
while at 425◦C, Fe3O4 is a more likely candidate.
The absorption spectrum and its first derivative of our