Applied Catalysis A: General 288 (2005) 220–231
www.elsevier.com/locate/apcata
Mg–Fe–Al mixed oxides with mesoporous properties prepared
from hydrotalcite as precursors: Catalytic behavior
in ethylbenzene dehydrogenation
Yoshihiko Ohishi a, Tomonori Kawabata a,
Tetsuya Shishido b, Ken Takaki a, Qinghong Zhang c,
Ye Wang c, Kiyoshi Nomura d, Katsuomi Takehira a,*
a
Department of Chemistry and Chemical Engineering, Graduate School of Engineering, Hiroshima University,
Kagamiyama 1-4-1, Higashi-Hiroshima 739-8527, Japan
b
Department of Chemistry, Tokyo Gakugei University, Nukui-kita 4-1-1, Koganei, Tokyo 184-8501, Japan
c
State Key Laboratory for Physical Chemistry of Solid Surfaces, Department of Chemistry, Xiamen University, Xiamen 361005, China
d
School of Engineering, University of Tokyo, Hongo 7-3-1, Bunkyoku, Tokyo 113-8656, Japan
Received 18 January 2005; received in revised form 15 April 2005; accepted 27 April 2005
Available online 6 June 2005
Abstract
Supported iron catalyst was prepared from Mg–Fe–Al hydrotalcite-like compounds as precursors and successfully applied for the
ethylbenzene dehydrogenation to styrene. After the calcination, the iron-substituted hydrotalcite-like samples were converted to mixed oxides
with a high surface area as well as a mesoporous character; the XRD analysis indicates the formation of periclase Mg(Fe, Al)O as a main
phase. Catalytic tests of the Mg2FexAl1x catalysts showed that the styrene conversion increased with increasing the iron content up to
x = 0.75 and then decreased, while the selectivity was the highest at x = 0.25. The optimum temperature for the reaction was 550 8C, which
was lower than that used in the commercial process. No favorable effect of the addition of either CO2 or O2 in the reaction medium was
observed in this reaction. Actually, the pre-treatment with H2 resulted in an increase in the activity at the beginning of the reaction as well as a
stable activity during the reaction. The ethylbenzene conversion of 60% and the styrene selectivity of 95% were kept for 3 h over
Mg3Fe0.5Al0.5 catalyst at 550 8C. After the reaction for 3 h, the iron species on the catalyst was partially reduced to the valence state between
Fe2+ and Fe3+. The high catalytic performance can probably be attributed to the formation of partially reduced iron oxides on the surface of
catalyst and to the high surface area along with the porous structure, which originated from the Mg–Fe–Al hydrotalcite structure in the
precursors.
# 2005 Elsevier B.V. All rights reserved.
Keywords: Mg–Fe–Al hydrotalcite-like materials; Ethylbenzene dehydrogenation; The reduced iron oxide; Porous catalyst
1. Introduction
Styrene, an important basic chemical as a raw material for
polymers, is produced commercially by the dehydrogenation of ethylbenzene using a typical Fe-K-Cr oxide-based
catalyst in the presence of a large quantity of steam at high
temperatures of 600–700 8C, just below the temperature
* Corresponding author. Tel.: +81 824 24 7744; fax: +81 824 24 7744.
E-mail address: [email protected] (K. Takehira).
0926-860X/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2005.04.033
where the thermal cracking becomes significant [1,2]. The
main precursors of the technical catalyst are hematite
(Fe2O3) and the promoter potassium carbonate (K2CO3);
these are mixed and calcined. Small amounts of other metal
oxides like Cr2O3 are added as structural promoters to
stabilize the catalyst morphology and to prevent sintering.
Promotion of iron oxide with potassium enhances the
reactivity of iron oxide by an order of magnitude and reduces
the formation of carbonaceous surface deposits or coke that
deactivates the catalysts [1,3,4]. A study on the catalyst
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
structure and composition under reaction conditions by
Muhler et al. [5,6] gave evidence for the existence of KFeO2
which, in line with a proposal of Hirano [4], was interpreted
as essential for high activity. KFeO2 was identified as the
active phase also by Lundin et al. [7] and Coulter et al. [8].
Kinetic studies on model catalysts showed similar apparent
activation energies for K-promoted and unpromoted iron
oxide catalysts, suggesting that potassium only increases the
number of active sites, but does not change their nature [8].
Potassium stabilizes the catalyst against reduction [9] and
supports the removal of coke. K2CO3 is believed to represent
the active center for carbon gasification [4,10,11], whereas
benzene and toluene as the main side products may be
formed at different sites [4,8,12]. Experiments on unpromoted Fe2O3 gave clear evidence that defective surfaces are
more active than well-ordered ones [13]. Ranke and coworkers [14] reported that both unpromoted Fe2O3 and Kpromoted model catalysts show a similar high starting
activity, while that of Fe3O4 is clearly lower. Catalyst
deactivation by coking or oxide reduction was prevented by
the addition of potassium to the catalyst, and by the addition
of water or a small amount of oxygen to the feed.
Although the commercial catalysts are very active and
selective, they have some disadvantages: (i) The active
oxidation state is unstable; hematite (a-Fe2O3) is preferred
for styrene production, but tends to lower oxides and even to
elemental iron, both of which catalyze coke formation and
dealkylation [15,16]. (ii) The catalysts have low surface
area. (iii) They are deactivated with time, being susceptible
to poisoning by halides and residual organic chloride
impurities [1]. The most serious deactivation is caused by
the loss of potassium promoter, which migrates in two
directions as the catalyst ages. Potassium chlorides are found
down-stream in the water layer of the condensed product as
well as in catalyst pellets. In fact, a major migration of
potassium occurs within the catalysts pellets, because the
center of the pellets operates at a slightly lower temperature
than the periphery due to the endothermicity of the reaction.
In addition, potassium migrates also to the cooler part of the
reactor (exit) [1,15,16]. Also, the toxicity of the chromium
compounds causes damage to humans and to the environment [1,16]. Therefore, the search for new catalyst systems
which have high surface areas and can stabilize the active
state of iron, in the absence of potassium, is much needed.
Aluminum was proved to be an excellent promoter,
preventing sintering in iron-oxide catalysts [17,18]. Hirano
[19] studied the effect of adding a series of alkaline earth
oxides such as MgO, CaO, SrO, and BaO to a K-promoted
iron oxide catalyst, among which especially MgO had good
characteristics. Stobbe et al. [20] have developed methods to
support promoted iron oxide onto an MgO support material.
A supported system can prevent the physical degradation of
the catalyst particles.
The starting materials can also play a significant role in
the catalytic performance [21]. There has been for the past
years an increased interest in using hydrotalcite-like
221
compounds as precursors for mixed oxide catalysts in
various reactions as reviewed by Cavani et al. [22]. Indeed,
Clause et al. [23] reported that the catalysts obtained by
thermal decomposition of Ni–Al hydrotalcite precursors
have very interesting properties, such as high thermal
stability and good metal dispersion. The latter behavior has
been attributed by Kruissink et al. [24] to the particular
structure of the hydrotalcite through a homogeneous
distribution of the metallic cations in the brucite-type
sheets. The authors have reported that Mg–Ni–Al hydrotalcite-like compounds were effective as precursors of the
catalyst for steam and autothermal reforming of CH4
[25,26]. A new catalyst for the process of ethylbenzene
dehydrogenation to styrene under a CO2 or O2 atmosphere
was obtained using vanadium-substituted Mg–Al hydrotalcite-like compounds as precursors [27,28]. Fe–Mg mixed
oxides with high surface areas around 200 m2 g1 and with
mesopores around 15 nm in diameter were prepared from
hydrotalcite-like compounds and tested as the catalysts in
the Fisher–Tropsch reaction [29]. Moreover, surface acid/
base properties and the reducibility of the iron of the
catalysts were studied by microcalorimetric method and
Mössbauer spectroscopy [30]. Fe–Mg or Fe–Mg–Al mixed
oxides were prepared from hydrotalcite-like precursors and
were successfully tested as the catalysts for the reduction of
aromatic nitro compounds [31–33]. As for the dehydrogenation of ethylbenzene, Fe–Mg–Al [34], Fe–Mg–Zn–Al [35],
and Fe–Zn–Al [36] mixed oxides were tested under a CO2
atmosphere. The activity losses over the Fe-Mg-Al catalysts
with time-on-stream were completely restored by oxygen
pulses [34]. The Fe-Mg-Zn-Al catalysts afforded the highest
ehylbenzene conversion of 53.8% and a styrene selectivity
of 96.7% at 500 8C [35]. The Fe-Zn-Al catalysts were
effective and gave an areal rate of 4.15 mmol min1 m2
none the less the surface area was as low as 22.1 m2 g1
cat [36].
In the present work, a series of Fe–Mg–Al mixed oxides
were prepared from the corresponding hydrotalcite-like
precursors and were tested for the dehydrogenation of
ethylbenzene. Catalytic behaviors of Fe–Mg–Al mixed
oxides are completely different from those reported in the
preceding papers [34–36]; no favorable effect of either the
additions of CO2 and O2 or the oxidation pre-treatment of
the catalyst on the activity was observed, whereas the
reduction pre-treatment afforded the catalyst with somewhat
higher activity as well as higher stability. The results of
detailed studies on the catalyst structure and the activities
will be reported in this paper.
2. Experimental
2.1. Catalyst preparation
The Mg–Fe and Mg–Fe–Al hydrotalcite-like precursors
were synthesized using a co-precipitation of nitrates of each
metal component, following the method by Miyata and
222
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
Okada [37] after minor modifications. An aqueous solution
containing the nitrates of Mg2+, Fe3+ and Al3+ (300 ml) was
added slowly with vigorous stirring into an aqueous solution
of sodium carbonate (300 ml). By adjusting the pH of the
solution at 10 with an aqueous solution of sodium hydroxide,
we caused a heavy slurry to precipitate. After the solution
was aged at 60 8C for 24 h, the precipitates were washed
with de-ionized water (1000 ml), dried in air at 110 8C for
24 h, and calcined at 550 8C for 12 h in a muffle furnace in a
static air atmosphere. As a control, imp-Fe/MgO and imp-Fe/
g-Al2O3 catalysts were prepared by incipient wetness
method using magnesia (UBE) and alumina (JRC-ALO8)
as the carrier.
2.2. Characterizations of catalysts
Inductively coupled plasma (ICP) optical emission
spectroscopy was used for the determination of the metal
content in each sample synthesized above. The measurements
were performed with a Perkin-Elmer OPTIMA 3000; each
sample was completely dissolved in a mixture of HF and
HNO3 before the measurements. Powder X-ray diffraction
was recorded on a Rigaku powder diffraction unit, RINT
2250VHF, with mono-chromatized Cu Ka radiation
(l = 0.154 nm) at 40 kV and 300 mA. The diffraction pattern
was identified by comparing with those included in the JCPDS
database (Joint Committee of Powder Diffraction Standards).
Diffuse reflectance UV–vis spectroscopic measurements
were recorded on a JASCO UV/VIS/NIR (V-570) spectrophotometer. The spectra were collected in the range of 200–
700 nm referenced to BaSO4. The sample was loaded in an in
situ cell and was treated in a pure N2 gas flow at 200 8C for
dehydration before the measurement. X-ray absorption
spectroscopic measurements were performed at room
temperature in a transmission mode at the EXAFS facilities
installed at the BL01B1 line of Spring-8 JASRI, Harima,
Japan, using a Si(1 1 1) monochrometer. The data were
collected in a quick-XAFS mode. Data reductions were
performed with the FACOM M1800 computer system of the
Data Processing Center of Kyoto University [38]. The sample
was mixed with boron nitride as a binder and then pressed into
a disk (10 mm in diameter). Energy was calibrated with Cu Kedge absorption (8981.0 eV); the energy step for measurement in the XANES region was 0.3 V. The adsorption was
normalized to 1.0 at an energy position of 30 eV higher than
the adsorption edge. Transmission Mössbauer spectra of
pelletized powder samples were recorded at room temperature, using a constant acceleration mode (Topologic System
Co.) of a radiation source with about 40 MBq 57Co(Cr) and a
YAP scintillation counter. Doppler velocity was calibrated
with reference to a-Fe. Thermogravimetric-differential
thermal analysis (TG-DTA) of the catalyst was performed
under an inert atmosphere of N2 (20 ml min1) with
Shimadzu TGA-50 and DTA-50 analyzers using 50 mg of
sample at a rate of 10 8C min1. Temperature programmed
reduction (TPR) of the catalyst was performed at a heating
rate of 10 8C min1 from ambient temperature to 1100 8C
using a mixture of 5 vol.% H2/Ar at a rate of 100 ml min1 as
reducing gas, after passing through a 13 molecular sieve trap
to remove water. We used a U-shaped quartz tube reactor, the
inner diameter of which was 6 mm; it was equipped with a
TCD for monitoring the H2 consumption. Prior to the TPR
measurement, 50 mg of sample was calcined at 550 8C for 1 h
in Ar gas flow at a rate of 20 ml min1. N2 adsorption (77 K)
study was used to examine both BET surface area and the
porous property of the Mg–Fe–Al mixed oxide. The
measurement was carried out on a Bell Japan Belsorp
18SP instrument (volumetric); all samples were pre-treated in
vacuum at 200 8C for 12 h before the measurements. The
pore-size distribution was evaluated from the adsorption
isotherm by the Dollimore and Heal (DH) method [39].
2.3. Catalytic reaction
Dehydrogenation of ethylbenzene was conducted using a
fixed-bed flow reactor at atmospheric pressure. A quartz
glass tube with an inner diameter of 8 mm was used as a
reactor. In the dehydrogenation reactions, typically 0.15 g of
catalyst, which had been pelletized and crushed to the
particles 250–417 mm in diameter, was loaded into the
reactor. The material was treated in a gas flow containing N2/
O2 (10/10 ml min1) at 550 8C for 1 h and then purged with
N2. The reaction was started by introducing a gas mixture of
ethylbenzene, N2 and CO2, O2 or Ar to the reactor.
Ethylbenzene (0.6 ml min1; ca. 1.5 mmol h1) was fed by
bubbling a gas mixture of N2 (10 ml min1) and CO2, O2 or
Ar (30 ml min1) through liquid ethylbenzene held at
28.3 8C in a thermostat. The reaction products (styrene,
toluene, and benzene) and ethylbenzene were liquefied in a
cold trap of n-heptane and acetone at 0 8C, and analyzed
with an FID gas chromatograph using a packed ID-BPX 5
column and cyclohexanone as an internal standard. Analyses
of gaseous products (CO, CO2, and H2) were performed with
a TCD gas chromatograph using a packed Molecular Sieve5A column and Porapak Q. All the lines and valves between
the cold trap and the reactor were heated to 150 8C to prevent
any condensation of ethylbenzene or of the dehydrogenation
products. The presence of other hydrocarbons and of the
oxygenated products was checked by GC with FID using a
packed FFAP column, but none were detected. All data of
the reactions were collected after the reaction for 1 h and
used for the calculation of the reaction results unless
specifically mentioned, since the activity of the catalyst
sometimes gradually decreased during the reaction.
3. Results and discussion
3.1. Structure and properties of supported Fe catalysts
Metal compositions, surface areas and Fe contents of the
supported Fe catalysts are shown in Table 1. Metal
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
Table 1
Surface areas of supported Fe catalysts
Catalyst
Preparation Fe wt.%b Surface area Color
c
methoda
(m2 g1
cat )
Mg3Fe0.5Al0.5
imp-Fe/MgO
imp-Fe/g-Al2O3
Mg2Fe
Mg2Fe0.75Al0.25
Mg2Fe0.5Al0.5
Mg2Fe0.25Al0.75
Mg2Al
cp
imp
imp
cp
cp
cp
cp
cp
a
b
c
21.4
21.4
21.4
49.8
39.1
27.3
14.4
0
169.0
67.9
137.1
115.5
142.2
170.3
174.1
202.9
Off-white
Off-white
Reddish brown
Brown
Pale brown
Pale brown
Off-white
White
cp: Co-precipitation method; imp: impregnation method.
Measured by ICP method.
Measured by BET method.
compositions shown as suffixes in each catalyst are calculated
based on the amounts of metal nitrates used in the preparation.
Higher surface area was observed for the catalysts prepared
from hydrotalcite-like compounds by co-precipitation compared with those prepared by impregnation.
XRD patterns of Mg3Fe0.5Al0.5 hydrotalcite (HT)
obtained by co-precipitation showed typical reflection lines
of layered double hydroxides. After the calcination of
Mg3Fe0.5Al0.5 HT at 550 8C in N2 atmosphere (Fig. 1a), the
lines of periclase Mg(Fe, Al)O, i.e., periclase MgO phase
containing both Fe3+ and Al3+ as solid solutions, appeared
together with the weak line of Mg(Fe, Al)2O4 spinel. imp-Fe/
MgO also showed mainly the lines of periclase MgO
Fig. 1. XRD patterns of supported Fe catalysts effects of pre-treatment and
the reaction: (a) Mg3Fe0.5Al0.5 (after N2 treatment at 550 8C); (b)
Mg3Fe0.5Al0.5 (after O2 treatment at 550 8C); (c) Mg3Fe0.5Al0.5 (after H2
treatment at 550 8C); (d) Mg3Fe0.5Al0.5 (after H2 treatment at 1000 8C); (e)
Mg3Fe0.5Al0.5 (a, after the reaction); (f) Mg3Fe0.5Al0.5 (d, after the reaction).
(&) Periclase Mg(Fe, Al)O; (&) Mg(Fe, Al)2O4 spinel; (+) Fe metal; (*)
MgAl2O4; (*) Fe3C; (), graphite.
223
together with a weak line of MgFe2O4 spinel, suggesting that
a reaction between iron and magnesia occurred during
impregnation, though slowly. imp-Fe/g-Al2O3 showed
reflection lines of a-Fe2O3 (hematite) and g-Al2O3 alone,
and no other line was observed, indicating that no substantial
reaction between iron and alumina occurred.
To understand the decomposition procedure of the
Mg3Fe0.5Al0.5 HT, we analyzed the TG-DTA profile in
detail. As reported in references [22,29], two weight loss
stages were observed on the TG curve; these correspond to
the two endothermic peaks around 177 and 366 8C on the
DTA profile, demonstrating that the decomposition proceeded in two steps. When 10 mg of the sample was used,
the total weight loss was found to be 4.25 mg; 1.53 mg was
lost in the first step, while 2.72 mg was lost in the second
step. The sample loses water molecules contained in the
interlayer at the first step and hydroxide and carbonate
groups at the second step. Suppose that the initial sample has
the composition of Mg6Fe1Al1(OH)16CO3xH2O and that
the final mixed oxide has the composition of 6MgO 1/
2Fe2O3 1/2Al2O3, the formula weight of the initial sample
can be calculated to be 648.1 and the number of water
molecules contained in the interlayer was found to be 4.9,
which is approximately equal to 5. Thus, the formula of the
Mg3Fe0.5Al0.5-HT may be Mg6Fe1Al1(OH)16CO35H2O,
which seems to be reasonable since the number of water
molecule is just between Mg6Fe2(OH)16CO36H2O [29] and
Mg6Al2(OH)16CO34H2O [40].
The periclase Mg(Fe, Al)O phase was mainly detected
together with a small amount of spinel Mg(Fe, Al)2O4, in the
XRD patterns of Mg–Fe–Al samples after the calcination of
the precursors at 550 8C in both N2 and O2 atmospheres
(Fig. 1a and b). These results demonstrate that the mixed
oxide samples may mainly possess a brucite-like structure
with Fe3+ randomly distributed in the octahedral sites owing
to the ionic radius of Fe3+ (0.49 Å) being smaller than that of
Mg2+ (0.72 Å) [41]. We have reported that periclase
Mg(Ni2+, Al3+)O was formed by the calcination of Mg–
Ni–Al HT [25,26], where the ionic radii of Ni2+ and Al3+
were 0.69 and 0.39 Å, respectively [41]. It is possible to
consider the formation of Mg(Fe2+)O solid solutions, since
the ionic radius of Fe2+ is 0.78 Å [41]. Actually the
Mg(Fe2+)O solid solutions were formed by the reduction of
MgFe2O4 spinel during the preparation of Mg–Fe mixed
oxide from the HT precursors [42,43]. The formation of
Mg(Fe3+)O solid solutions was also reported in Mg–Fe
mixed oxide prepared from the HT precursors, by Shen and
co-workers [30] and Figueras and co-workers [33]. Moreover, the presence of Fe3+ in periclase Mg(Fe3+, Al3+)O can
be also supported by both Mössbauer spectra and X-ray
absorption spectra in the present work (vide infra).
3.2. Electronic state of iron
The color of the supported Fe catalysts is shown in
Table 1 since it is a simple indication of whether bulk iron
224
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
oxide exists [44]. All as-synthesized Mg–Fe–Al HT samples
exhibited a white color, suggesting that no bulk iron oxide
existed and that all iron cations were incorporated inside the
framework of hydrotalcite after the co-precipitation. After
the calcination at 550 8C, both Mg3Fe0.5Al0.5 and
Mg2Fe0.25Al0.75 became off-white in color, suggesting the
presence of some extraframework iron. On the other hand,
the color from pale brown to brown of Mg2Fe0.75Al0.25,
Mg2Fe0.5Al0.5, and Mg2Fe with increasing color density
with increase in the Fe content suggests that these samples
contain aggregated iron oxide clusters; none the less, XRD
did not detect any iron oxide. For the two supported Fe
catalysts prepared by impregnation, imp-Fe/MgO was offwhite in color, suggesting the well-dispersed iron formation
on this catalyst, whereas imp-Fe/g-Al2O3 exhibited reddish
brown color probably due to the formation of bulk iron oxide
as detected by XRD.
As shown in the Mössbauer spectra of Mg3Fe0.5Al0.5
before and after the reaction and in the spectra of imp-Fe/
MgO (Fig. 2a–c), these samples displayed a doublet with an
isomer shift (IS) of around 0.3 mm s1 and with a
quadrupole splitting (QS) of around 0.7 mm s1; such
values are typical of superparamagnetic Fe3+ species. The
spectrum of Mg3Fe0.5Al0.5 sample can be fitted with two
doublets, whose calculated parameters are reported in
Table 2. One of the two components is attributed to the Fe3+
in Mg(Fe3+, Al)O as solid solutions, as seen in the XRD.
Sometimes small ferric oxide particles or a-Fe2O3 were
observed in Mg–Fe [33] or Mg–Fe–Al [30] mixed oxides;
both showed the Fe3+ signal as a doublet in Mössbauer
spectra. Even though the off-white color suggests the
possibility of iron oxide (Table 1), XRD exhibited no clear
evidence of such species and, moreover, a-Fe2O3 usually
displays a sextet signal (vide infra). It is therefore most
likely that the other doublet component observed in the
Mössbauer spectra can be attributed to the Fe3+ in Mg(Fe3+,
Al)2O4 spinel. The XRD patterns showed the presence of
periclase Mg(Fe3+, Al)O as a main phase, while Mg(Fe3+,
Fig. 2. Mössbauer spectra of supported Fe catalysts: (a) Mg3Fe0.5Al0.5
(before reaction); (b) Mg3Fe0.5Al0.5 (after reaction); (c) imp-Fe/MgO; (d)
imp-Fe/g-Al2O3.
Al)2O4 spinel was observed as a minor phase or as an
amorphous phase (Fig. 1). It should be noted that the spinel
species was not strongly detected by XRD at the calcination
temperature of 550 8C. Moreover, the concentration of Fe3+
will probably be higher in Mg(Fe3+, Al)2O4 spinel than in
periclase Mg(Fe3+, Al)O, since Fe3+ possesses a regular site
in the former crystal structure but not in the latter. No clear
correlation was thus observed between the abundance of
Fe3+, shown as the value of relative contribution (Table 2),
and the analytical results obtained by XRD. After the
reaction, the Mg3Fe0.5Al0.5 sample exhibited a much broader
peak width and significantly larger QS in the Mössbauer
spectrum. This is probably due to the fact that a part of Fe3+
was reduced to Fe2+, as reported by Shen and co-workers
[30] and Figueras and co-workers [33]. The Mössbauer
spectrum of imp-Fe/MgO showed two doublets attributed to
Table 2
Mössbauer parameters of supported iron oxide catalysts
Catalyst
Component
Site
3+
ISa
QSb
LWc
Bhfd
Rel. contr. (%)e
Phases by XRD
Mg3Fe0.5Al0.5
Doublet
Doublet
Fe (1)
Fe3+(2)
0.42
0.20
0.68
0.73
0.52
0.52
65.0
35.0
Mg(Fe, Al)2O4
Mg(Fe, Al)O
Mg3Fe0.5Al0.5 (after reaction)
Doublet
Doublet
Fe3+(1)
Fe3+(2)
0.44
0.27
0.74
0.79
0.59
0.59
48.2
51.8
Mg(Fe, Al)2O4
Mg(Fe, Al)O
imp-Fe/MgO
Doublet
Doublet
Fe3+(3)
Fe3+(4)
0.37
0.15
0.67
0.77
0.51
0.51
64.0
36.0
MgFe2O4
Mg(Fe)O
imp-Fe/g-Al2O3
Doublet
Doublet
Sextet
Fe3+(5)
Fe3+(6)
Fe3+(7)
0.51
0.26
0.36
0.88
0.89
0.22
0.67
0.67
0.50
18.9
63.7
17.4
Superpara Fe2O3
Tetrahedral Fe-oxide
a-Fe2O3
a
b
c
d
e
Isomer shift (mm s1).
Quadrupole splitting (mm s1).
Line width (mm s1).
Inner magnetic field (T).
Relative contribution.
50.1
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
both periclase Mg(Fe)O and MgFe2O4 spinel (Fig. 2c and
Table 2). It is supposed that the Fe3+ in Mg(Fe, Al)2O4 spinel
is more quickly reduced than that in periclase Mg(Fe, Al)O
during the reaction, since the Fe3+ ions in periclase Mg(Fe,
Al)O are much more highly diluted and probably better
stabilized. Actually, the reflection line of Mg(Fe, Al)2O4
spinel disappeared after the reaction (Fig. 1). It is thus
concluded that Fe3+(1) site is located in Mg(Fe, Al)2O4
spinel while Fe3+(2) site is in periclase Mg(Fe, Al)O
(Table 2). imp-Fe/g-Al2O3 showed the Mössbauer spectra
composed of two doublets and one sextet (Fig. 2d and
Table 2), while a-Fe2O3 (hematite) and g-Al2O3 alone were
observed in the XRD. The sextet of the site Fe3+(7) with IS
of 0.36 mm s1 is assigned to the Fe3+ in a-Fe2O3, as
reported in the study on Mg–Fe–Al mixed oxides; the sextet
has IS of 0.32 mm s1, characteristic of a-Fe2O3 [43]. One
of the other doublets, Fe3+(5), is assigned to the Fe3+ in
superpara Fe2O3 of fine particles, since the inner magnetic
field of a-Fe2O3 was 50.1 T, a little smaller than the normal
values of 51–52 T. Another doublet, Fe3+(6), is assigned to
the Fe3+ in tetrahedral Fe oxide from the value of IS of
0.26 mm s1, since the IS of Fe3+O4 tetrahedra is generally
observed at 0.2–0.32 mm s1 while that of Fe3+O6 octahedra
is generally observed at 0.35–0.55 mm s1.
3.3. Coordination sphere of iron
Fig. 3 shows the diffuse reflectance UV–vis spectra. aFe2O3 exhibited a peak at ca. 540 nm (Fig. 3a) assigned to
Fe2O3 particles. Centi and Vazzana [45] observed absorption
bands at 270 and 330 nm in the UV–vis spectra of Fe/ZSM-5
Fig. 3. Uv–vis spectra of supported Fe catalysts: (a) Fe2O3; (b)
Mg3Fe0.5Al0.5-HT; (c) Mg3Fe0.5Al0.5; (d) imp-Fe/MgO; (e) imp-Fe/gAl2O3.
225
Fig. 4. Fe K-edge XANES spectra of supported Fe catalysts and materials
as control: (a) Fe2O3; (b) Mg3Fe0.5Al0.5-HT; (c) Mg3Fe0.5Al0.5; (d) imp-Fe/
MgO; (e) imp-Fe/g-Al2O3.
samples prepared by CVD and assigned these bands to
isolated Fe3+ species interacting strongly with the support.
Mg3Fe0.5Al0.5 HT as prepared exhibited a peak at 270 nm
(Fig. 3b). This should probably be assigned to Fe3+ which is
isolated and octahedrally coordinated in brucite layered
structure [29]. Both Mg3Fe0.5Al0.5 and imp-Fe/MgO showed
a broad peak around 330 nm together with a weak shoulder
around 480 nm. On imp-Fe/g-Al2O3, the latter shoulder
shifted toward much longer wavelength and appeared at
540 nm. The peak around 330 nm is assigned to isolated
Fe3+ in either periclase Mg(Fe, Al)O or Mg(Fe, Al)2O4
spinel in Mg3Fe0.5Al0.5 mixed oxide. The isolated Fe3+ is
possibly formed also in imp-Fe/MgO, since neither XRD
reflection of Fe2O3 nor brown color was observed and,
moreover, no signal of Fe2O3 was detected in Mössbauer
spectra for this sample. The latter peaks around 480 nm for
Mg3Fe0.5Al0.5 and imp-Fe/MgO can probably be assigned to
the aggregated Fe3+ oxide clusters.
Fig. 4 shows the Fe K-edge XANES spectra of the
supported Fe catalysts, along with hematite (a-Fe2O3) as
control. The XANES spectra exhibit a pre-edge peak at
approximately 7107 eV, which is 10 eV below the midpoint
of the absorption band and is commonly assigned to iron in
tetrahedral coordination [46,47]. Iron atoms are principally
in octahedral coordination in both Mg3Fe0.5Al0.5 HT and aFe2O3 and therefore showed the smallest pre-edge peak. If
one judges the relative intensities of the pre-edge peak for
Mg3Fe0.5Al0.5, imp-Fe/MgO, and imp-Fe/g-Al2O3 against
those for Mg3Fe0.5Al0.5 HT and a-Fe2O3, one will see that
the amount of iron in tetrahedral coordination must be rather
small and that a major part of iron exists in octahedral
coordination also in the former three catalysts. Fourier
transforms of the k3-weighted Fe K-edge EXAFS spectra of
the supported Fe catalysts along with hematite as control are
shown in Fig. 5. Mg3Fe0.5Al0.5 HT exhibited an intense peak
226
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
Fig. 5. Fourier transforms of the k3-weighted Fe K-edge EXAFS spectra of
supported Fe catalysts and materials as control: (a) Fe2O3; (b) Mg3Fe0.5Al0.5HT; (c) Mg3Fe0.5Al0.5; (d) imp-Fe/MgO; (e) imp-Fe/g-Al2O3.
of Fe–O at 1.6 Å (non-phase shift corrected) together with a
weak peak at 2.6 Å (non-phase shift corrected) while aFe2O3 showed two strong peaks at 2.6 and 3.2 Å (non-phase
shift corrected) in addition to that of the Fe–O bond. These
two peaks could be assigned the Fe–O–Fe linkages through
edge-shared and corner-shared FeO6 octahedra [48].
Mg3Fe0.5Al0.5 exhibited two peaks of Fe–O and Fe–O–Fe
even though the intensities were significantly weakened
compared with those of Mg3Fe0.5Al0.5 HT, suggesting that
the basic coordination structure around Fe was not changed
during the calcination. imp-Fe/MgO also showed almost
identical spectra to those of Mg3Fe0.5Al0.5 with much lower
intensities. On the other hand, imp-Fe/g-Al2O3 exhibited a
peak at 3.2 Å (non-phase shift corrected) in addition to those
observed for Mg3Fe0.5Al0.5, among which the peaks
assignable to the Fe–O–Fe linkage were enhanced. The
latter result well coincided with the fact that the aggregated
iron oxide clusters exist together with the isolated iron
species on imp-Fe/g-Al2O3 observed by both XRD and
Mössbauer spectroscopies.
3.4. Activities of Mg–Fe–Al mixed oxides
The effect of Fe–Al composition in Mg2Fe1xAlx mixed
oxide in the dehydrogenation of ethylbenzene is shown in
Fig. 6. Effect of Fe–Al composition on the dehydrogenation of ethylbenzene over Mg2FexAl1x: (*) ethylbenzene conversion; (*) styrene selectivity; (&) benzene selectivity; (~) toluene selectivity. Catalyst, 0.15 g; N2,
10 ml min1; ethylbenzene, 1.5 mmol h1; reaction temperature, 550 8C;
reaction time, 1 h.
Fig. 6. In the absence of Fe, i.e., on Mg2Al mixed oxide, the
conversion of ethylbenzene was extremely low, and benzene
was produced as a by-product, suggesting that a thermal
b-fission reaction took place. Upon introducing Fe
component into the Mg2Al mixed oxide, we found that
the conversion of ethylbenzene increased and we observed
selective formation of styrene. Increase in Fe content
resulted in an increase in the conversion of ethylbenzene, but
no substantial change was observed in the product
selectivities. The activity of Mg3Fe0.5Al0.5 mixed oxide is
shown in Table 3 together with those of imp-Fe/MgO and
imp-Fe/g-Al2O3. The selectivity to hydrogen was calculated
based on the stoichiometry of the dehydrogenation of
ethylbenzene to styrene (1). The activity of Mg3Fe0.5Al0.5
Ph-CH2 CH3 ! Ph-CH¼CH2 þ H2
(1)
mixed oxide was the highest among the catalysts tested
in the present work and was also higher than that obtained
on Cr-MCM-41 in the presence of CO2 [49]. The selectivity
to hydrogen was high on Mg3Fe0.5Al0.5 mixed oxide and
low on both imp-Fe/MgO and imp-Fe/g-Al2O3, none the
less the selectivity to styrene was not affected on either
catalyst.
Gao and co-workers [35] reported that, in the dehydrogenation of ethylbenzene over Mg–Fe–Zn–Al mixed
oxide prepared from the corresponding HT, the catalytic
activity was enhanced in the presence of CO2 and afforded
Table 3
Ethylbenzene dehydrogenation over supported Fe catalystsa
Catalyst
Mg3Fe0.5Al0.5
imp-Fe/g-Al2O3
imp-Fe/MgO
a
b
Conversion (%) of ethylbenzene
58.5
34.0
36.9
Selectivity (%)
Styrene
Benzene
Toluene
Hydrogenb
91.2
92.8
92.8
5.0
4.6
4.8
3.7
2.6
2.4
86.4
70.0
69.2
Reaction conditions: catalyst weight, 150 mg; reaction temperature, 550 8C; reaction time, 1 h; N2/Ar, 10/30 ml min1; ethylbenzene, 1.5 mmol h1.
Calculated based on the stoichiometry of ethylbenzene dehydrogenation to styrene.
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
Fig. 7. Effect of partial pressure of O2 on the dehydrogenation of ethylbenzene over Mg3Fe0.5Al0.5: (*) ethylbenzene conversion; (*) styrene
selectivity; (&) benzene selectivity; (~) toluene selectivity; (&) CO2
selectivity; (~) CO selectivity. Catalyst, 0.15 g; pre-treated in N2
(20 ml min1) at 550 8C for 1 h; reaction in N2/Ar (10/30 ml min1) and
ethylbenezene (1.5 mmol h1) at 550 8C; reaction time, 1 h.
the highest ethylbenzene conversion of 53.8% and a styrene
selectivity of 96.7% at 500 8C. Moreover, Mg–V–Al
mixed oxide prepared from the HT precursors showed high
activity in the dehydrogenation of ethylbenzene in the
presence of CO2; the XRD analysis indicates the formation
of Mg3V2O7 and Mg2Al2O4 in bulk, while the XPS results
point to the presence of a mixture of V5+ and V4+ ions on
the surface [27]. This Mg–V–Al mixed oxide showed
high activity also in the presence of O2; the highest
ethylbenzene conversion of 38% was obtained with a
styrene selectivity of 98% at 450 8C [28]. In the present
work, addition of neither CO2 nor O2 showed any
favorable effect in the dehydrogenation of ethylbenzene
on Mg3Fe0.5Al0.5 mixed oxide. The effect of the addition
of oxygen in the dehydrogenation of ethylbenzene on
Mg3Fe0.5Al0.5 mixed oxide is shown in Fig. 7. With
increasing partial pressure of O2, the ethylbenzene
conversion first decreased and then inversely increased
accompanied by an increase in CO2 selectivity. Styrene
formation was significantly suppressed in the presence of
O2, indicating that Mg3Fe0.5Al0.5 mixed oxide catalyzed
the combustion reaction and was not effective for the
oxydehydrogenation of ethylbenzene. The addition of
CO2 simply resulted in a decrease in the ethylbenzene
conversion; none the less, the selectivities to styrene,
toluene, and benzene were not affected. Moreover, during
the reaction for 3 h, the ethylbenzene conversion gradually
decreased in the presence of CO2 while no substantial
decline was observed in the absence of CO2. These results
suggest that CO2 does not work as the oxidant on
Mg3Fe0.5Al0.5 mixed oxide as observed on Mg–V–Al
mixed oxide [28] or on Cr-MCM-41 [49].
The effect of the reaction temperature on the dehydrogenation of ethylbenzene on Mg3Fe0.5Al0.5 mixed oxide is
shown in Fig. 8. When the reaction temperature was
increased from 500 to 575 8C, the ethylbenzene conversion
increased, while the styrene selectivity slowly decreased.
227
Fig. 8. Effect of reaction temperature on the dehydrogenation of ethylbenzene over Mg3Fe0.5Al0.5: (*) ethylbenzene conversion; (*) styrene selectivity; (&) benzene selectivity; (~) toluene selectivity. Catalyst, 0.15 g;
pre-treated in N2 (20 ml min1) at 550 8C for 1 h; reaction in N2/Ar (10/
30 ml min1) and ethylbenezene (1.5 mmol h1); reaction time, 1 h.
Further increase in the reaction temperature to 600 8C
resulted in a decrease in the ethylbenzene conversion
together with a drastic decrease in the styrene selectivity.
Also at 600 8C, benzene formation was significantly
enhanced, suggesting an occurrence of thermal cracking
reactions of ethylbenzene. This was supported by the fact
that significant coke formation was observed on the catalyst
after the reaction.
3.5. Effect of pre-treatment on catalytic behavior of
Mg3Fe0.5Al0.5
Temperature programmed reduction (TPR) of
Mg3Fe0.5Al0.5 mixed oxide showed two reduction peaks
at 433 and 907 8C (Fig. 9). Shen and co-workers [43]
reported that the TPR of Mg2Fe1Al1 mixed oxide showed
two peaks at 506 and 904 8C. At the first TPR peak (506 8C),
hematite (a-Fe2O3) was reduced to Fe2+ and Fe0 ((2) and
(3)) while Mg(Fe, Al)2O4
a-Fe2 O3 þ MgO þ H2 ! Mg1x FeII O þ H2 O
(2)
a-Fe2 O3 þ H2 ! a-Fe þ H2 O
(3)
Fig. 9. Temperature programmed reduction of Mg3Fe0.5Al0.5. Heating rate,
10 8C min1; in a mixture of 5 vol.% H2/Ar at a rate of 100 ml min1.
228
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
was transformed into Mg1xFexO (4). The second TPR peak
(904 8C) corresponded to
MgðFeIII ; AlÞ2 O4 þ H2 ! Mg1x FeIIx O þ H2 O þ MgðAlÞO
(4)
the reduction of Mg1xFexO4 to metallic Fe0 (5). The
reductions of Fe3+ to Fe2+ and Fe0
Mg1x FeIIx O þ H2 ! a-Fe þ H2 O þ MgO
(5)
were confirmed by both XRD and Mössbauer spectroscopy
[42]. In the present work, the temperatures of both TPR
peaks varied depending on the ratio of Mg/Fe/Al; the first
peak significantly moved between 456 and 552 8C. Similar
results were obtained for Mg–Fe mixed oxides prepared
from HT precursors [29]. It is therefore possible that Fe3+ in
Mg3Fe0.5Al0.5 mixed oxide was reduced to Fe2+ or Fe0 by H2
formed during the dehydrogenation reaction at 550 8C.
Effects of the conditions of pre-treatment of Mg3Fe0.5Al0.5
mixed oxide on the activity in the dehydrogenation reaction
are shown in Table 4. The pre-treatment in H2 atmosphere was
carried out under the same conditions as that of TPR to exactly
follow the effect of the reduction of Fe3+ to Fe2+ or Fe0.
Sometimes the reduction treatment with H2 brought overreduction to the catalyst, resulting in the formation of metallic
Fe. As a result, coking became significant on the metallic Fe
and the reaction was finally suppressed (vide infra). In the
XRD of the samples, reflection lines of periclase Mg(Fe, Al)O
were observed in all samples treated at 550 8C (Fig. 1a–c). A
weak reflection of Mg(Fe, Al)2O4 spinel was observed for the
samples treated in both N2 and O2 atmospheres (Fig. 1a and b),
whereas no such reflection was observed for the sample
treated in H2 atmosphere (Fig. 1c). Even with such changes in
the XRD, no significant effect of pre-treatments in N2, H2, and
O2 atmosphere at 550 8C was observed in the reaction results.
The pre-treatments in both N2 and H2 atmosphere at 550 8C
resulted in a similar high activity, whereas that in O2
atmosphere at 550 8C caused a little decrease in the activity.
These results suggest that the surface of Mg3Fe0.5Al0.5 mixed
oxide was almost same after 1 h of the dehydrogenation
reaction, i.e., the surface was reduced by H2 produced during
the reaction. This was actually seen in the facts that the
reflection line of Mg(Fe, Al)2O4 spinel disappeared (Fig. 1e)
and also that Mg3Fe0.5Al0.5 sample exhibited a much broader
peak width and significantly larger QS in the Mössbauer
spectra after the dehydrogenation reaction (Table 2).
After the pre-treatment in H2 atmosphere at 1000 8C, no
reaction could be continued since the reactor was quickly
plugged by significant coke formation (Table 4). In the XRD
of the sample after the reduction (Fig. 1d), the lines of
Mg(Al)O were sharpened while the lines of Fe metal
appeared, indicating that periclase Mg(Fe, Al)O was
reductively decomposed to Mg(Al)O and Fe metal. When
the reduced sample was used in the dehydrogenation
reaction, the lines of MgO were further sharpened and, at the
same time, many reflection lines of Fe3C carbide appeared
together with those of graphite and MgAl2O4 spinel in the
XRD (Fig. 1f). These results indicate that periclase Mg(Al)O
was decomposed into MgO and MgAl2O4 spinel and,
moreover, that Fe metal was converted into Fe3C carbide by
coke deposited on it during the reaction.
3.6. Possible active phase on Mg3Fe0.5Al0.5
The time course of the dehydrogenation reaction is shown
in Fig. 10. After the pre-treatment of Mg3Fe0.5Al0.5 mixed
oxide in N2 atmosphere (Fig. 10, full line), a short induction
period of 20 min was observed, i.e., the conversion of
ethylbenzene started at a low value and increased during
20 min to reach a steady state value. The selectivity to
styrene gradually increased while that to benzene decreased
during the reaction for 3 h. When the mixed oxide was pretreated in H2 atmosphere, the induction period almost
disappeared and both conversion and selectivity showed
high values, close to those of the steady state, even at the
beginning of the reaction (Fig. 10, dotted line). After the
reaction reached steady state, the reaction profiles were
almost the same as those obtained after the pre-treatment in
N2 atmosphere. It was thus clearly confirmed that the
reduction pre-treatment brought a favorable effect to the
catalytic activity of Mg3Fe0.5Al0.5 catalyst.
Many papers suggest that the active phase in iron oxide
catalyst is Fe3+. Ranke co-workers [14] reported that
unpromoted Fe2O3 was much more active than Fe3O4 and
that KFeO2 is the active species for K-promoted iron oxide
catalyst in the dehydrogenation of ethylbenzene. KFeO2 as
Table 4
Effect of pre-treatment of Mg3Fe0.5Al0.5 mixed oxide catalyst
Condition of pre-treatment
N2b
O2c
H2d
H2e
a
b
c
d
e
Conversion (%) of ethylbenzene
60.0
58.5
59.1
–
Selectivity (%)
Styrene
Benzene
Toluene
Hydrogena
91.7
91.2
93.2
–
4.7
5.0
4.0
–
3.6
3.7
2.8
–
92.0
86.4
92.0
–
Calculated based on the stoichiometry of ethylbenzene dehydrogenation to styrene.
With N2 (20 ml min1) at 550 8C for 1 h.
With O2/N2 (10/10 ml min1) at 550 8C for 1 h.
With H2/Ar (5/95 ml min1) from room temperature to 550 8C at a rate of 10 8C min1.
With H2/Ar (5/95 ml min1) from room temperature to 1000 8C at a rate of 10 8C min1.
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
Fig. 10. Time course of the dehydrogenation of ethylbenzene over
Mg3Fe0.5Al0.5: (*) ethylbenzene conversion; (*) styrene selectivity; (&)
benzene selectivity; (~) toluene selectivity. Solid line, pre-treated in N2
(20 ml min1) at 550 8C for 1 h; dotted line, pre-treated in H2/Ar (5/
95 ml min1) from room temperature to 550 8C at 10 8C min1. Catalyst,
0.15 g; reactionin N2/Ar (10/30 ml min1) and ethylbenezene (1.5 mmol h1)
at 550 8C.
the active phase was also identified using SEM and Auger
microscopy by Lundin et al. [7]. Moreover, a study on Kpromoted iron oxide catalyst films prepared on Ru(0 0 0 1)
as a model catalyst reported that the KFeO2 shell around a
K2Fe22O34 core was identified for the active phase [50]. In
the ethylbenzene dehydrogenation over K-promoted iron
oxide catalyst, the fully oxidized iron phases containing Fe3+
ions are responsible for high catalytic activity; their partial
reductions by generated H2 caused a deactivation, and the
treatment with steam leads to a partial re-oxidation, resulting
in the re-activation of the catalyst [51]. It was also reported
that the activity losses over the Fe-Mg-Al catalysts with
time-on-stream were completely restored by oxygen pulses
[34].
Fe K-edge XANES spectra of Mg3Fe0.5Al0.5 after the
reaction, along with that before the reaction and that of
Fe(II)O as control, are shown in Fig. 11. The iron species
was slightly reduced after the reaction; the charge of iron is
considered to be between Fe2+ and Fe3+ judging from the
location of the absorbance curve of Mg3Fe0.5Al0.5 after the
reaction in the spectra. In the present work, it must be
emphasized that the reduced iron species is active as well as
stable in the ethylbenzene dehydrogenation (Figs. 10 and
11). It is noteworthy that H2 is formed during the
dehydrogenation and reduces the surface of catalyst [52].
Reports also suggest the possibility of Fe2+ as the active
species. Lee [53] observed that a-Fe2O3 (hematite) as a main
component in the catalyst is reduced to Fe3O4 (magnetite)
during the reaction and that the latter is more selective.
Courty and Le Page [54] mentioned that the catalyst is more
stable after a reduction of Fe2O3 to Fe3O4 after the first 200 h
operation. Yang et al. [55], using Mössbauer spectroscopy
on K-promoted iron oxide catalyst, suggested that the active
site contained no K and was promoted by a rapid electron
exchange between Fe2+ and Fe3+ catalyzed by K. Use of
unpromoted Fe2O3 in the ethylbenzene dehydrogenation
229
Fig. 11. Fe K-edge XANES spectra of Mg3Fe0.5Al0.5 after the reaction: (*)
Mg3Fe0.5Al0.5 after the reaction for 1 h; (~) Mg3Fe0.5Al0.5 before the
reaction; (&) Fe(II)O.
gave clear evidence of the fact that defective surfaces are
more active than well-ordered ones [13]. Also, the in situ
study of MCM-41-supported iron oxide by XANES and
EXAFS showed that a 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 8C
[56]. Such effects of reduced iron species or defect structure
due to the partial reduction on the catalyst surface are likely
in the present work.
3.7. Possible roles of Mg and Al
The addition of Mg as well as Al will be also important in
the high activity of Mg–Fe–Al mixed oxides. A high
stability was observed by the addition of Mg on K-promoted
iron oxide catalyst; it was attributed to the fact that MgO
prevented the pore destruction of the catalyst [19]. By the
process of supporting promoted iron oxide catalyst on MgO,
coke formation was limited because the catalyst was
subjected to a suitable pre-treatment [20]. Aluminum was
proved to be an excellent promoter, preventing sintering in
iron-oxide catalysts [17,18]. The supported system can
prevent the physical degradation of the catalyst particles.
In the present work, the surface area of Mg3Fe0.5Al0.5
was 169.0 m2 g1
cat before the reaction, and it decreased
to 152.4 and 147.1 m2 g1
cat after the reaction for 1 and 3 h,
respectively, suggesting that the surface area was somewhat
stabilized during the reaction.
The pore-size of Mg3Fe0.5Al0.5 showed a wide distribution from 2 nm to above 25 nm as well as an increasing
distribution with an increase in radius (Rp). The lowest limit
of pore radius detectable in the present method is 2 nm, and
therefore no information concerning micropores smaller
than 2 nm could be obtained. However, no significant
distribution less than 5 nm was observed on Mg3Fe0.5Al0.5.
230
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
Mg3Fe0.5Al0.5 calcined at 850 8C showed a sharp distribution at less than 5 nm, together with a broad distribution
around 20 nm in radius. When the calcination temperature
was increased, the pores with large size diminished and the
pore-size distribution tended to converge to the small values.
Mg–Al(3/1) hydrotalcite showed a peak around 2–3 nm
together with a wide distribution up to 20 nm in radius. The
former pore-sizes are probably related to the layered
structure, while the latter is due to a ‘‘card house’’ structure
consisting of many small plates. We calculated the basal
interlayer spacing from the strong symmetric (0 0 3)
reflection (2u = 13.48) of Mg–Al(3/1) hydrotalcite. If the
thickness of the brucite-like layer is assumed to be 4.8 Å
[57], the interlayer distance corresponds to 2.9 Å. The high
surface area of Mg3Fe0.5Al0.5 is probably due to the presence
of meso- and macro-pores in the catalyst. The pore-size
distribution of Mg3Fe0.5Al0.5 showed no substantial change
from before to after the reaction.
The amounts of coke deposited on the catalyst were 15.7
and 22.5 mg g1
cat after 1 h of reaction in CO2 and Ar,
respectively; these amounts were far larger than the values,
i.e., 1.83 and 0.60 mg g1
cat , observed after 5 h of reaction
over Cr-MCM-41-DHT catalysts in CO2 and He, respectively [58]. These values suggest that the coking was not
main reason of the catalyst deactivation, since no deactivation was observed over Mg3Fe0.5Al0.5 catalyst while a clear
deactivation took place over Cr-MCM-41-DHT catalysts
[58], and, moreover, the deactivation of Mg3Fe0.5Al0.5
catalyst occurred in CO2 but was not observed in Ar
atmosphere. The porous structure with the high surface area
was sufficiently preserved and no remarkable deactivation
was observed on Mg3Fe0.5Al0.5 during the reaction for 3 h.
These stabilizing effects on the catalytic activity are
probably due to the fact that catalyst was prepared via
solid solutions, i.e., periclase Mg(Fe, Al)O, starting from
Mg–Fe–Al hydrotalcite as the precursors.
4. Conclusion
Mg-Fe-Al mixed oxide catalysts prepared from the
corresponding layered double hydroxides as the precursors
were successfully applied for the ethylbenzene dehydrogenation to styrene. By calcining the precursors, we
obtained the mixed oxides with a high surface area and a
mesoporous character, where periclase Mg(Fe, Al)O was
mainly detected by the XRD analysis. During the reaction,
the iron species on the catalyst surface was partially reduced
to the valence state between Fe2+ and Fe3+, and the partially
reduced iron species were effective for the dehydrogenation.
Neither CO2 nor O2 showed any favorable effect as the
additive in the reaction medium in the dehydrogenation
reaction. The ethylbenzene conversion of 60% and the
styrene selectivity of 95% were stably observed over
Mg3Fe0.5Al0.5 catalyst at the optimum temperature of
550 8C, which was lower than that employed in commercial
processes. The high catalytic performance is due to the
formation of partially reduced iron oxides on the surface of
catalyst, together with the high surface area along with the
porosity. All these characters are owing to the fact that
catalysts were prepared via periclase Mg(Al, Fe)O as solid
solutions intermediate starting from Mg–Fe–Al hydrotalcite
as the precursors.
References
[1] E.H. Lee, Catal. Rev. Eng. Sci. 8 (1973) 285.
[2] K. Kochloefl, in: G. Ertl, H. Knözinger, J. Weitkamp (Eds.),
Handbook of Heterogeneous Catalysis, vol. 5, VCH, Weinheim,
1998, p. 2151.
[3] T. Hirano, Appl. Catal. 26 (1986) 65.
[4] T. Hirano, Appl. Catal. 28 (1986) 119.
[5] M. Muhler, J. Schütze, M. Wesemann, T. Rayment, A. Dent, R.
Schlögl, G. Ertl, J. Catal. 126 (1990) 339.
[6] M. Muhler, R. Schlögl, G. Ertl, J. Catal. 138 (1992) 413.
[7] J. Lundin, L. Holmlid, P.G. Menon, L. Nyberg, Ind. Eng. Chem. Res.
32 (1993) 2500.
[8] K. Coulter, D.W. Goodman, R.G. Moore, Catal. Lett. 31 (1995) 1.
[9] S.C. Ndlela, B.H. Shanks, Ind. Eng. Chem. Res. 42 (2003) 2112.
[10] W.D. Mross, Catal. Rev.-Sci. Eng. 24 (1983) 591.
[11] R.L. Hirsch, J.E. Gallagher, R.R. Lessard, R.D. Wesselhoft, Science
215 (1982) 121.
[12] W.P. Addiego, C.A. Estrada, D.W. Goodman, M.P. Rosynek, R.G.
Windham, J. Catal. 146 (1994) 407.
[13] W. Weiss, D. Zscherpel, R. Schlögl, Catal. Lett. 52 (1998) 215.
[14] O. Shekhah, W. Ranke, R. Schlögl, J. Catal. 225 (2004) 56.
[15] B.D. Herzog, H.F. Raso, Ind. Eng. Chem. Prod. Res. Dev. 23 (1984)
187.
[16] A.B. Styles, in: B.E. Leach (Ed.), Applied Industrial Catalysis, Academic Press, New York, 1983, p. 137.
[17] G.C. Araújo, M.C. Rangel, Catal. Today 62 (2000) 201.
[18] H. Topsoe, J.A. Dumesic, M. Boudart, J. Catal. 28 (1978) 477.
[19] T. Hirano, Bull. Chem. Soc. Jpn. 59 (1986) 2672.
[20] D.E. Stobbe, F.R. van Buren, A.W. Stobbe-Kreemers, A.J. van Dillen,
J.W. Geus, J. Chem. Soc., Faraday Trans. 87 (1991) 1631.
[21] A.C. Oliveira, A. Valentini, P.S.S. Nobre, M.C. Rangel, React. Kinet.
Catal. Lett. 75 (2002) 135.
[22] F. Cavani, F. Trifiro, A. Vaccari, Catal. Today 11 (1991) 173.
[23] O. Clause, M. Gazzano, F. Trifiro, A. Vaccari, L. Zatorski, Appl. Catal.
73 (1991) 217.
[24] E.C. Kruissink, L.L. Van Reijen, J.R.H. Ross, J. Chem. Soc., Faraday
Trans. 177 (1991) 649.
[25] K. Takehira, T. Shishido, P. Wang, T. Kosaka, K. Takaki, Phys. Chem.
Chem. Phys. 5 (2003) 3801.
[26] K. Takehira, T. Shishido, P. Wang, T. Kosaka, K. Takaki, J. Catal. 221
(2004) 43.
[27] G. Carja, R. Nakamura, T. Aida, H. Niiyama, J. Catal. 218 (2003)
104.
[28] F. Malherbe, C. Forano, B. Sharma, M.P. Atkins, J.P. Besse, Appl. Clay
Sci. 13 (1998) 381.
[29] J. Shen, B. Guang, M. Tu, Y. Chen, Catal. Today 30 (1996) 77.
[30] M. Tu, J. Shen, Y. Chen, J. Solid State Chem. 128 (1997) 73.
[31] P.S. Kumbhar, J. Sanchez-Valente, F. Figueras, Tetrahedron Lett. 39
(1998) 2573.
[32] S.M. Auer, J.-D. Grunwaldt, R.A. Köppel, A. Baiker, J. Mol. Catal. A
139 (1999) 305.
[33] P.S. Kumbhar, J. Sanchez-Valente, J.M.M. Millet, F. Figueras, J. Catal.
191 (2000) 467.
[34] P. Kuśtrowski, A. Rafalska-Łasocha, D. Majda, D. Tomaszewska, R.
Dziembaj, Solid State Ionics 141–142 (2001) 237.
Y. Ohishi et al. / Applied Catalysis A: General 288 (2005) 220–231
[35] X. Ye, N. Ma, W. Hua, Y. Yue, C. Miao, Z. Xie, Z. Gao, J. Mol. Catal.
A 217 (2004) 103.
[36] N. Mimura, I. Takahara, M. Saito, Y. Sasaki, K. Murata, Catal. Lett. 78
(2002) 125.
[37] M. Miyata, A. Okada, Clays Clay Miner. 25 (1977) 14.
[38] K. Tanaka, H. Yamashita, R. Tsuchitani, T. Funabiki, T. Yoshida, J.
Chem. Soc., Faraday Trans. 84 (1988) 2987.
[39] D. Dollimore, G.R. Heal, J. Appl. Chem. 14 (1964) 109.
[40] A. Drits, T.N. Sokolova, G.V. Sokolova, V.I. Cherkashin, Clays Clay
Miner. 35 (1987) 401.
[41] R.D. Shannon, Acta Crystallogr. A 32 (1976) 751.
[42] L.A. Boot, A.J. van Dillen, J.W. Geus, A.M. van der Kraan, A.A. vand
der Horst, F.R. van Buren, Appl. Catal. A 145 (1996) 389.
[43] X. Ge, M. Li, J. Shen, J. Solid State Chem. 161 (2001) 38.
[44] P. Ratnasami, R. Kumar, Catal. Today 9 (1991) 329.
[45] G. Centi, F. Vazzana, Catal. Today 53 (1999) 683.
[46] D.W. Lewis, G. Sanker, C.R.A. Catlow, S.W. Carr, J.M. Thomas, Nucl.
Instrum. Methods Phys. Res. B 97 (1995) 44.
231
[47] S.A. Axon, K.K. Fox, S.W. Carr, J. Klinowski, Chem. Phys. Lett. 189
(1992) 1.
[48] Y. Ma, S.L. Suib, Chem. Mater. 11 (1999) 3545.
[49] Y. Ohishi, T. Kawabata, T. Shishido, K. Takaki, Q. Zhang, Y. Wang, K.
Takehira, J. Mol. Catal. A 230 (2005) 49.
[50] G. Ketteler, W. Ranke, R. Schlögl, J. Catal. 212 (2002) 104.
[51] X.M. Zhu, M. Shön, U. Bartmann, A.C. van Veen, M. Muhler, Appl.
Catal. A 266 (2004) 99.
[52] G.R. Meima, P.G. Menon, Appl. Catal. A 212 (2001) 239.
[53] E.H. Lee, Catal. Rev. 8 (1973) 285.
[54] Ph. Courty, J.F. Le Page, Stud. Surf. Sci. Catal. 3 (1979) 293.
[55] X. Yang, S. Weng, K. Jiang, L. Mao, Y. Euong, K. Jing, Hyperfine
Interact. 69 (1991) 863.
[56] S.-T. Wong, J.-F. Lee, S. Cheng, C.-Y. Mou, Appl. Catal. A 198 (2000)
115.
[57] M.A. Drezdon, Inorg. Chem. 37 (1988) 4628.
[58] K. Takehira, Y. Ohishi, T. Shishido, T. Kawabata, K. Takaki, Q. Zhang,
Y. Wang, J. Catal. 224 (2004) 404.