Perovskite-type oxide catalysts for low temperature, anaerobic catalytic partial oxidation of methane to syngas

Article abstract of DOI:10.1016/j.jcat.2010.07.004

The cyclic reaction between CH₄ and O₂ over perovskite-type La_0.8Sr_0.2Fe_0.8Co _0.2O_3-δ, La_0.75Sr _0.25Fe_0.6Co_0.15Al_0.25O _3-δ and La_0.8Sr_0.2Fe _0.8Co_0.2O_3-δ/γ-AlO(OH), impregnated with 0.5 wt% Rh or Pt, was studied at 873 K. Synchrotron X-ray and neutron diffraction patterns of La_0.75Sr_0.25Fe _0.6Co_0.15Al_0.25O_{3- δ} proved a rhombohedral structure with Al distributed over the octahedral B-site. The oxygen non-stoichiometry (δ) was determined by thermogravimetric analysis at 958 K for pO₂ > 10^-22 atm. High selectivity to the partial oxidation products CO and H₂ was observed when 3 - δ was lower than 2.76 and 2.78 for the mentioned oxides, with and without Al, respectively. The role of Rh relates solely to the activation of CH₄. A stable high selectivity throughout the pulse sequence was achieved when utilizing partially reduced materials. In situ synchrotron XRD proved that the oxides retained the perovskite structure during the reduction/oxidation cycle, and no phase decomposition occurred. The strong correlation between results from catalytic transient tests and thermogravimetric analysis suggests that the CO selectivity is ruled by the redox potential of the reducible oxide.

Full text of DOI:10.1016/j.jcat.2010.07.004

Journal of Catalysis 275 (2010) 25–33
Contents lists available at ScienceDirect
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
Perovskite-type oxide catalysts for low temperature, anaerobic catalytic partial
oxidation of methane to syngas
b
Federica Mudu a, Bjørnar Arstad b,c, , Egil Bakken , Helmer Fjellvåg a,c, Unni Olsbye a,c,
*
**
a Centre of Material Science and Nanotechnology, Department of Chemistry, University of Oslo, P.O. Box 1126 Blindern, N-0318 Oslo, Norway
b SINTEF Materials and Chemistry, Forskningsveien 1, 0314 Blindern, Oslo, Norway
c inGAP Centre of Research-Based Innovation, Department of Chemistry, University of Oslo, P.O. Box 1126 Blindern, N-0318 Oslo, Norway
a r t i c l e i n f o
a b s t r a c t
Article history:
The cyclic reaction between CH4 and O2 over perovskite-type La0.8Sr0.2Fe0.8Co0.2O3ꢀd, La0.75Sr0.25Fe0.6
-
Received 15 March 2010
Revised 25 June 2010
Accepted 4 July 2010
Available online 21 August 2010
Co0.15Al0.25O3ꢀd and La0.8Sr0.2Fe0.8Co0.2O3ꢀd -AlO(OH), impregnated with 0.5 wt% Rh or Pt, was studied
/
c
at 873 K. Synchrotron X-ray and neutron diffraction patterns of La0.75Sr0.25Fe0.6Co0.15Al0.25O3ꢀd proved a
rhombohedral structure with Al distributed over the octahedral B-site. The oxygen non-stoichiometry
(d) was determined by thermogravimetric analysis at 958 K for pO2 > 10ꢀ22 atm. High selectivity to the
partial oxidation products CO and H2 was observed when 3 ꢀ d was lower than 2.76 and 2.78 for the men-
tioned oxides, with and without Al, respectively. The role of Rh relates solely to the activation of CH4. A
stable high selectivity throughout the pulse sequence was achieved when utilizing partially reduced
materials. In situ synchrotron XRD proved that the oxides retained the perovskite structure during the
reduction/oxidation cycle, and no phase decomposition occurred. The strong correlation between results
from catalytic transient tests and thermogravimetric analysis suggests that the CO selectivity is ruled by
the redox potential of the reducible oxide.
Keywords:
Perovskite oxide
Synthesis gas
Methane partial oxidation
Reducible oxide
Thermogravimetric analysis
Rhodium
Ó 2010 Elsevier Inc. All rights reserved.
1. Introduction
The product compositions are determined by thermodynamics.
Both CH4 conversion and CO and H2 selectivities are favored by
The worldwide demand for natural gas (NG) is expected to con-
tinue its long-term upwards trend in the next decades [1,2] with a
slight increase in the share of unconventional gas [2]. Offshore, re-
mote NG represents a particular exploitation challenge, since float-
ing units for conversion of NG to easily transportable liquid
products (methanol or synthetic crude oil) need to be light-weight
and free of explosion risks associated with e.g. air distillation.
Conventional technologies for conversion of NG to methanol or
synthetic crude proceed via synthesis gas (syngas; CO + H2) pro-
duction over transition-metal (Ni in particular)-based catalysts
[3]. In order to obtain the appropriate H2:CO ratio in the syngas,
either steam reforming, followed by the water gas shift reaction
(reactions (1) and (2), respectively) or autothermal reforming (3),
may be used. The latter is a complex process comprising combus-
tion of part of the feed (4) followed by reactions (1) and (2) [4].
high temperatures and low pressures. Common outlet tempera-
tures in industrial processes are above 1000 K [3].
For offshore purposes, a small light-weight unit for syngas pro-
duction is desirable. Steam reforming units are limited by heat
transfer and typically consist of 5–600 reactor tubes, each with a
diameter of 70–130 mm [3], and do not meet relevant weight
requirements. On the other side, autothermal reforming uses a
much more compact reactor, but the process involves the use of
pure O2, commonly produced via cryogenic distillation and associ-
ated with non-zero explosion risks [5]. Such high temperature pro-
cesses are furthermore demanding with respect to materials
choice. For these reasons, a low-risk, low- or medium-temperature
process for selective oxidation of CH4 to syngas would represent an
attractive alternative.
One possible approach would be a selective oxidation process,
in which a reducible metal oxide is used as O2 source (5) and sub-
sequently reoxidized in an air separation unit (6).
CH4 þ H2O ! CO þ 3H2
CO þ H2O $ H2 þ CO2
CH4 þ 0:5O2 ! CO þ 2H2
CH4 þ 2O2 ! CO2 þ 2H2O
ð1Þ
ð2Þ
ð3Þ
ð4Þ
yCH4 þ MOx ¼ yCO þ 2yH2 þ MOxꢀy
MOxꢀy þ 1=2yO2 ¼ MOx
ð5Þ
ð6Þ
* Corresponding author.
Such split oxidation process has been widely studied for several
partial oxidation reactions [6,7], and was evaluated already in
1949 by Lewis et al. for conversion of CH4 to syngas using CuO as
oxygen source in a fluidized bed reactor at around 1200 K [8]. They
** Corresponding author. Address: inGAP Centre of Research-Based Innovation,
Department of Chemistry, University of Oslo, P.O. Box 1126 Blindern, N-0318 Oslo,
Norway.
E-mail address: unni.olsbye@kjemi.uio.no (U. Olsbye).
0021-9517/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2010.07.004

26
F. Mudu et al. / Journal of Catalysis 275 (2010) 25–33
suggested a two-step reaction; a first step where copper oxide was
rapidly reduced and part of CH4 totally oxidized, and a second
slower step where the CH4 was reformed with the formed carbon
dioxide and steam to syngas.
1000 min. The 1273 K treatment was carried out twice with crush-
ing between each treatment. Rh was deposited using the incipient-
wetness technique to a nominal amount of 0.5 wt% (elemental Rh).
Rh(NO3)3ꢁ2H2O from Alfa-Aesar was used. After impregnation, the
Rh/LSFC, Rh/LSFCA and Rh/LSFCB samples were heated at a rate of
5 K/min to 1073 K and held for 600 min. Additional Rh/LSFCA were
synthesized with 0.1 wt% and 1.0 wt% Rh loading.
More recently, Otsuka et al. investigated ceria, CeO2, as oxidant
in CH4 partial oxidation to syngas [9]. Ceria is one of the best
known oxygen storage materials and is e.g. used in three way cat-
alysts [10,11]. Several recent works describe ceria as oxidizer for
the conversion of CH4 to syngas [12–16]. Ceria is often combined
with platinum to increase the activity at low temperatures and/
or chemically modified with e.g. zirconium, samarium or bismuth
to tune oxygen diffusivity and capacity [17]. An important issue is
syngas selectivity. Otsuka observed high syngas selectivities over
Pt/CeO2 and suggested that CO and H2 were formed directly from
CH4 [9,12], but did not discuss total oxidation products. Later, Fathi
et al. studied Pt and Rh supported on CeO2 and observed that syn-
gas selectivity depended on the degree of reduction of ceria [14].
Pantu et al. made the same observation for Pt and Ru supported
on Ce1ꢀxZrxO2 and suggested that the selectivity to CO and H2 is
determined by the relative rates of CH4 and CO oxidation [15].
Perovskite-type oxides (ABO3) represent an alternative class of
reducible oxides with potential as partial oxidation catalysts [18].
Previously, such studies of reactions (5) and (6) have focused on
high temperature applications (above 1073 K) due to low activity
[19–23]. Dai et al. studied LaFeO3 and suggested that full combus-
tion products are formed from chemisorbed oxygen at particle sur-
faces, while selective oxidation products are formed from oxygen
diffusing from the bulk [24]. Recently, Kharton et al. reported a cor-
relation between the degree of reduction and the activity and
selectivity for syngas formation over La0.3Sr0.7Fe0.8M0.2O3ꢀd
(M = Ga, Al) and SrFe0.7Al0.3O3ꢀd [25].
In the present study, Rh-promoted (La, Sr)(Fe, Co, Al)O3ꢀd per-
ovskites were studied as CH4 partial oxidation catalysts at medium
temperatures (873 K). The purpose of the study was to elucidate
the reason for the reported oxide non-stoichiometry–syngas selec-
tivity correlation and further to elaborate the phase stability and
oxygen capacity of a novel perovskite composition for the title
reaction.
The materials were characterized in detail by in situ and ex situ
powder X-ray diffraction (XRD), neutron diffraction, thermogravi-
metric methods and cyclic fixed bed reactor tests. The results pro-
vide novel insight into materials – selectivity correlation, providing
a basis for future design of oxidizers for selective oxidations.
2.2. Characterization
BET surfaces of the samples were estimated using single point
isotherm measurements on a Quantachrome Monosorb instrument
at 77 K (liquid N2).
Powder X-ray diffraction patterns were recorded with a position
sensitive detector using a Siemens Bruker D5000 diffractometer
and Cu Ka1 radiation. Room temperature Neutron Powder Diffrac-
tion (NPD) data were collected with PUS (Powder Universal Spec-
trometer) two-axis diffractometer at the JEEP-II reactor at Kjeller,
Norway. Data were collected over a 2
H range of 10–130° using neu-
trons of wavelength 1.5561 Å and rebinned into steps of 0.05°.
Variable temperature synchrotron radiation X-ray diffraction
(SRXRD) patterns were collected on a MAR345 image plate system
at the Swiss-Norwegian Beam Line (SNBL) at ESRF, Grenoble,
France, at a wavelength of 0.694 Å. The detector resolution at the
working distance used corresponds to an effective step size of
0.01783° in 2H. The analyzed 2H range was 1–40°. With an expo-
sure time of 20 s and read out time of 87 s, the time resolution be-
tween consecutive scans is 107 s. The sample (0.1–0.2 mm
diameter particles) was loaded between two quartz wool pads in
a 0.7 mm diameter quartz capillary. Both ends of the capillary were
mounted to a flow cell by means of Swagelok valves. The inlet was
connected to a gas feed system and the outlet to a Mass Spectrom-
eter (Pfeiffer Omnistar) for continuous effluent analysis. The gas
feed system was equipped with Bronkhorst mass flow controllers
and remotely controlled. An air heater gun with remote tempera-
ture control was used for heating. Temperature calibration was
carried out by separate measurements of the thermal expansion
of silver. During the in situ experiments, Rh/LSFC and Rh/LSFCB
were heated to 1000 K in an O2:He = 10:90 gas flow. Once the tem-
perature was stable, the gas flow was switched to CH4:He = 10:90
for 15 min, thereafter switched back to O2 for 15 min, before a sec-
ond CH4 and O2 sequence, and finally cooled to 800 K and sub-
jected to two reduction–oxidation cycles.
Rietveld refinements of the NPD data and batch refinements of
the SRXRD patterns were performed using the GSAS/EXPGUI
[26,27] suite of programs. Peak shape was modeled using modified
Gaussian (NPD) and pseudo-Voigt (XRD) profile functions.
2. Experimental
2.1. Synthesis
2.3. Thermogravimetric analysis (TG)
Perovskite oxides with nominal formula La0.8Sr0.2Fe0.8Co0.2O3ꢀd
(LSFC) and La0.75Sr0.25Fe0.6Co0.15Al0.25O3ꢀd (LSFCA) were prepared
by the citric acid method. Sr(NO3)2 (Fluka, P99.0), Co(II)acetate
tetrahydrate (Fluka, P99.0), La2O3 (Sigma–Aldrich, 99.99), Al
(NO3)3ꢁ9H2O (Merck, P99.0) and Fe(NO3)3ꢁ9H2O (Merck, P99.0)
were used as reagents. La2O3 was dissolved in nitric acid, and the
other reactants were dissolved in water before mixing. A surplus
of citric acid monohydrate (C6H8O7ꢁH2O, 100%, VWR) was then
added. The mixture was heated to 430 K and held at this tempera-
ture until dry. For one sample, boehmite (AlOOH Vista Chemical
Company) was added to the viscous LSFC mixture before drying,
resulting in a composite material (65 wt% La0.75Sr0.25Fe(Co)0.75A-
TGA experiments were performed to investigate the equilib-
rium oxygen non-stoichiometry (3 ꢀ d) of Rh/LSFCA and Rh/LSFC
as a function of O2 partial pressure (pO2) at 958 K. Samples of (typ-
ically) 100 mg were measured by a SETARAM Setsys Evolution TGA
with an in-house gas mixing system giving 100 ml minꢀ1 reaction
gas. The gas mixing system allowed successive dilution of the reac-
tion gas to obtain a wide range of O2 partial pressures. CO2 and H2
with different dilutions were used to obtain the lower pO2 reaction
gas mixtures, and pO2 values were measured with a pO2 sensor
built in house. The non-stoichiometry versus pO2 isotherms were
measured for O2 partial pressures in the range 10ꢀ22–2 ꢂ 10ꢀ1 atm.
The absolute stoichiometry (3 ꢀ d) of the samples in air at room
temperature was determined by cerimetric titration. The 3 ꢀ d var-
iation from room temperature to 958 K was measured thermo-
gravimetrically, thus providing the absolute 3 ꢀ d value at 958 K.
XPS measurements (not reported here) showed that Rh was
l0.25O3, 20 wt% LaFe(Co)Al11O19 and 15 wt% Fe(Co)Al2O4) with cat-
ion ratio La:Sr:Fe:Co:Al = 4:1:4:1:12 (denoted LSFCB, where B
stands for Boehmite). The dried gels were crushed and heated to
720 K at a rate of 5 K/min, held for 1000 min, and thereafter the
temperature was raised to 1273 K (5 K/min) and held for

F. Mudu et al. / Journal of Catalysis 275 (2010) 25–33
27
oxidized after pretreatment in air at 973 K and reduced after pre-
3. Results and discussion
treatment in 10% CH4 at the same temperature. Therefore, the oxy-
gen loss due to Rh2O3 reduction to metallic Rh was subtracted
before the 3 ꢀ d was calculated, in the thermogravimetric mea-
surements, as well as in the catalytic measurements (see below).
3.1. Structural properties and phase analysis
Powder X-ray diffractograms (XRD) of the studied oxides are
shown in Fig. 1. The observed peaks and their splitting clearly
showed that LSFC and LSFCA took rhombohedral symmetry, in
good accordance with the findings by Riza et al. for
La0.8Sr0.2Fe0.8Co0.2O3ꢀd. However, they reported that even small
compositional variations could change the symmetry to ortho-
rhombic [28]. Earlier, Tai et al. [29] had reported orthorhombic
symmetry for the mentioned stoichiometry. Unit cell dimensions
for the rhombohedral LSFC sample as derived from Rietveld refine-
ments are listed in Table 1.
Analysis of synchrotron powder X-ray diffraction data for the
AlOOH-containing LSFCB revealed the presence of several crystal-
line phases (Fig. 2). Two aspects were of interest; minor Al-substi-
tutions to modified redox features of the oxide, and large additions
to provide an alumina-rich support for the redox-active perovskite.
For the Al-rich LSFCB sample, the latter multiphase situation was
expected. The perovskite-type main phase was rhombohedral (like
LSFC), but with lower unit cell volume and a lower degree of distor-
tion. The second main phase was lanthanum iron hexaaluminate
(LaFeAl11O19) with smaller amounts of FeAl2O4 in addition. These
Al-rich oxides originated from the boehmite added in the synthesis.
LaFeAl11O19 is interesting in itself for catalytic applications [30–32].
It has a stable specific surface area even after prolonged treatments
2.4. Catalytic tests in pulse setup
Catalytic testing was performed in a tubular quartz reactor (i.d.
8 mm) at 873 K and 1 atm pressure. The catalyst (0.1 g, 0.2–0.5 mm
diameter) was diluted by quartz (0.4 g, 0.2–0.5 mm). An automated
setup equipped with 4-way and 6-way VICI VALCO valves was
used. Feed gases (AGA, 99.996%) were used without further purifi-
cation, implying an O2 content not higher than 1 ꢂ 10ꢀ5 atm. The
4-way valve was used to switch between 10% CH4/He and 10%
O2/He. The 6-way valve was equipped with a loop (0.25 ml) and
used to send discrete pulses of gas, with a continuous stream of
Ar (30 ml/min), to the reactor. In one experiment, O2 pulses were
sent to the reactor every 2 min, while the catalyst was heated to
873 K. The sample was equilibrated in Ar flow for 10 min at
873 K and thereafter subjected to a series of 50 CH4 pulses (10%
CH4/He) followed by a series of 100 O2 pulses (10% O2/He). The
same test was performed on Rh/LSFC and Rh/LSFCA at 958 K. How-
ever, in this case, the sample was equilibrated in Ar flow for 24 h
before proceeding with CH4 pulses.
In a second set of experiments, the cycling properties of the
materials were studied. The catalyst was heated to 873 K as de-
scribed above, then 50 pulses of CH4 were let into the reactor fol-
lowed by 14 pulses of O2. Subsequently, 25 pulses of CH4 followed
by 14 pulses of O2 were sent to the reactor. This reduction–oxida-
tion cycle was repeated 13 times.
Effluent gases were monitored by an online mass spectrometer
(Pfeiffer Omnistar). Calibration values for quantification of CH4,
CO2, CO and H2 were found by sending known pulses of these gases
and integrating the area under the peaks of m/z 15, 44, 28 and 2,
respectively. The contribution of CO2 to m/z 28 was accounted
for before quantification of CO. Carbon balance was calculated for
each pulse as the sum of unconverted CH4, CO2 and CO. H2O was
quantified as the difference of the converted CH4 and the produced
H2.
Table 1
Surface area and refined unit cell values.
Sample
BET Surface area (m2/g)
a
c
Space group
LSFC
LSFCB
1.5
17.9
3.5
1.8
19
5.502
5.474
5.472
13.347
13.328
13.323
R-3C
R-3C
R-3C
LSFCAa
Rh/LSFC
Rh/LSFCB
Rh/LSFCA
4.3
a
Refined values from NPD. La, Sr in 6a 0,0,1/4; Fe, Co in 6b 0,0,0 and O in 18e
x,0,1/4, x = 0.5387 (2).
a
b
Rh/LSFCA
Rh/LSFCB
Rh/LSFC
20
40
60
80
56 57 58 59 60 61
2θ (deg)
2θ (deg)
Fig. 1. (a) Powder X-ray diffraction patterns for Rh/LSFC, Rh/LSFCB and Rh/LSFCA. Inset (b) emphasizes similarity in pattern between Rh/LSFCB and Rh/LSFCA when compared
to Rh/LSFC. In (b) merging of the three reflections (3 0 0), (2 1 4 and (0 1 8) centered at 58° in Rh/LSFCB and Rh/LSFCA is evident, reflecting a lower degree of rhombohedral
distortion.

28
F. Mudu et al. / Journal of Catalysis 275 (2010) 25–33
LSFCB
a
b
c
0
10
15
20
2θ (deg)
25
30
35
20
40
60
2θ (deg)
80
100
120
Fig. 2. Observed, calculated and difference powder diffraction pattern for LSFCB.
The bars indicate reflections from the different phases present in the material,
perovskite oxide (a), lanthanum iron hexaaluminate LaFe1Al11O19 (b), and spinel
FeAl2O4 (c).
Fig. 3. Observed, calculated, and difference powder neutron diffraction pattern for
LSFCA. The bars indicate positions for perovskite oxide reflections.
regation during cyclic operation at 800 and 1000 K. As expected,
expansion of the unit cell occured upon reduction, this is due to
the fact that ionic radii of Fe and Co will increase upon reduction.
There was a concomitant symmetry change from rhombohedral
to cubic (see Fig. S3) [34]. The pseudocubic unit cell (cube root of
lattice volume per ABO3) dimensions are shown as function of tem-
perature and gas atmosphere in Fig. 4. In most cases, the change in
unit cell size and symmetry was completed within the time frame
needed to record one diffractogram (107 s). The unit cell volume
change during cycling between O2 and CH4 atmosphere was a func-
tion of chemical composition. The ꢄ0.4% expansion during reduc-
tion for Al-containing LSFCA was just around half of that of the
parent LSFC (ꢄ0.8%) material. It is not likely that this difference
can be ascribed to the degree of reduction since the oxygen con-
sumption data in Table 2 and Fig. 5 suggest that the oxygen dona-
tion capacities differ by less than 20%. We rather suggest that
structural aspects are in play; Al does not readily take tetrahedral
coordination in perovskite-type oxides and may possibly stabilize
e.g. square pyramidal coordination in the reduced state. These pro-
vide in turn a more efficient packing than tetrahedral building
units.
at high temperature [30] and can incorporate transition elements
like Cr, Mn, Fe, Co and Ni. Due to this flexibility, the present LaFe-
Al11O19 may take a slightly deviating composition since Sr may en-
ter the La site, while Co could take the Fe site and Fe can easily enter
the Al site. Indeed, the unit cell size measured in this work was
slightly bigger than the reference LaFeAl11O19 (ICSD 203003), this
deviation might originate from a partial substitution of Al with Fe
(see ICSD 35174). This would in turn modify the stoichiometry of
the perovskite phase. This could be evidenced through unit cell
changes. Such were indeed observed when comparing LSFC and
LSFCB (Table 1). On the other hand, the dominating effect currently
appeared to be aluminum incorporated into the B-site. According to
Rietveld refinement of LSFCB (for fit see Fig. 2), some 25% of the B-
site atoms were aluminum; i.e. La0.75Sr0.25Fe0.60Co0.15Al0.25O3ꢀd
(note that since Fe and Co have too similar scattering factors to be
distinguished by XRD analysis, their relative amount was fixed to
the nominal 0.8:0.2).
With respect to the catalytic tests reported below, it should be
noted that the third target material, LSFCA, was aimed to corre-
spond completely to the pure perovskite phase identified in the
LSFCB composite. This was confirmed by analysis of neutron pow-
der diffraction data of LSFCA (Fig. 3) and the derived unit cell data
in Table 1. The neutron data strongly supported the Al-concentra-
tion at the B-site as determined from SRXRD data for LSFCB. Fur-
thermore, the strong scattering from oxygen atoms allowed a
reasonable determination of the oxygen stoichiometry. The Riet-
veld refinement of O-occupancy factors gave close to fully occupied
O-sites (no = 0.99(1)), suggesting d ꢃ 0 at room temperature for
LSFCA as confirmed by cerimetric titration. The rhombohedral dis-
3.3. Thermogravimetric analysis
At equilibrium, the oxygen non-stoichiometry for the perov-
skite oxide depends on temperature and O2 partial pressure.
.
.
.
.
.
.
.
.
tortion of LSFCA was less than for LSFC, i.e. the rhombohedral
a an-
gle is lowered from 60.43° to 60.22°. As a consequence, the
obviously split reflections for LSFC merged into a single peak for
LSFCA and LSFCB in Fig. 1 (note that peak broadening owing to
strain and particle size also partly blurred the splitting). This vari-
ation in distortion was in agreement with what would be predicted
from tolerance t-factor correlations using ionic radii provided by
Shannon [33]; i.e. t = 0.978 for LSFCA whereas 0.963 for LSFC.
3.2. In situ synchrotron radiation X-ray diffraction
The 0.5 wt% Rh impregnated samples of LSFC and LSFCB were
analyzed with variable temperature synchrotron radiation X-ray
diffraction (SRXRD). The results are reported in Fig. 4 and in sup-
porting information, Figs. S1 and S2. The study (Fig. 4) indicates that
both the Rh/LSFC and Rh/LSFCA perovskites were stable toward seg-
Fig. 4. Unit cell size (here reported as pseudocubic for sake of comparison)
variation as a function of temperature and gas atmosphere (CH4 intervals shown in
grey) for Rh/LSFC and Rh/LSFCA. The unit cell side was determined by the Rietveld
refinement of each diffractogram collected with Synchrotron X-ray diffraction.

F. Mudu et al. / Journal of Catalysis 275 (2010) 25–33
29
Table 2
Pulse test results. Accumulated carbon deposition and oxygen consumption during 50 methane pulses.
Sample
CH4 conversion (%)
CO selectivity (%)
C deposited (
l
mol/gꢀ1
)
O used (
l
mol/gꢀ1
)
CO2
(l
mol/gꢀ1
)
CO (l
mol/gꢀ1
)
H2 (l )
mol/gꢀ1
LSFC
LSFCB
LSFCA
Rh/LSFC
Rh/LSFCB
Rh/LSFCA
Rh/LSFCAa
24
16
28
75
99
95
94
2
4
nd
nd
nd
nd
10
4
46
25
48
109
81
105
35
11
6
<1
<1
<1
17
37
25
21
<1
<1
<1
34
81
60
43
<1
43
73
58
85
12
23
14
20
3
nd
a
Only partially reoxidized, cumulative results after 25 methane pulses.
on SrFe(Al)O3ꢀd was observed by Patrakeev et al. [36] and was re-
lated to tetrahedral Al3+ cations that act as traps for oxygen
vacancies.
3.4. Catalytic tests; CO selectivity and cycling
The LSFC, LSFCA and LSFCB materials, with and without impreg-
nated Rh, were tested as catalysts for CH4 partial oxidation in a
transient setup at 873 K. Results from the catalytic tests are shown
in Fig. 6 and Table 2. The average carbon mass balance (Cin/Cout) for
all reported tests was 1.00 0.02.
CH4 conversion versus CH4 pulse number is plotted in Fig. 6a.
Rh/LSFCB showed the highest initial activity; close to 100%, while
Rh/LSFCA and Rh/LSFC had 81% and 64% initial conversion, respec-
tively. For the non-promoted catalysts, LSFCA gave the highest ini-
tial conversion; 70%, while LSFCB and LSFC showed 55% and 41%,
respectively. Blank tests of CH4:O2 = 1:1 mixtures in an empty
reactor gave less than 5% CH4 conversion at the same temperature
and feed rate.
Fig. 5. Oxygen non-stoichiometry (3-d) as a function of oxygen partial pressure as
measured by thermogravimetric analysis for Rh/LSFCA and Rh/LSFC.
It is well established that metal elements facilitate CH4 activa-
tion [5]. Notably, the LSFC-based perovskite oxides were them-
selves active for CH4 oxidation at the studied temperature of
873 K, even in the absence of a metal promoter. For the Rh-con-
taining oxides, the CH4 conversion increased during the first CH4
pulses. This reflects probably a progressing Rh reduction in corre-
spondence with the literature data for Rh/Al2O3 catalysts [37]. Sub-
sequently, the activity decreased. For the pure oxide catalysts, the
activity decreased already from the first pulse and onwards.
The CO selectivity (and H2, cfr Fig. 7) was initially close to zero
(Fig. 6b) over all materials, but increased with time on stream. For
the Rh-catalysts, the CO selectivity approached 100% after about 30
pulses. Fewer pulses were required for Rh/LSFCB to reach high
selectivity than for Rh/LSFCA and Rh/LSFC. The selectivity increase
was concomitant with a decrease in CH4 conversion. For the pure
oxide catalysts, only LSFCB showed increasing CO selectivity with
increasing pulse number, approaching 24% after 50 pulses.
Average CH4 conversion and H2 and CO selectivity, and accumu-
lated carbon deposition and O2 consumption during 50 CH4 pulses,
are listed in Table 2. It is interesting to note that carbon deposition
(as monitored by COx evolution during subsequent O2 pulses) was
observed for Rh/LSFCA and Rh/LSFCB, i.e. the materials that gave
the highest CH4 conversion and H2 and CO selectivities. For these
two materials, the carbon balance gradually increased during the
last 20 CH4 pulses, summing up to 12% selectivity to carbon depo-
sition during these pulses over Rh/LSFCA. It is worth noting that
the LSFC sample with Al on the B-site showed higher CO yields
throughout the CH4 pulses compared to the parent material
(Fig. 6 and Table 2).
Thermogravimetry was used to measure the equilibrium oxygen
non-stoichiometry of Rh/LSFC and Rh/LSFCA, by fixing the pO2 of
the gas mixture flowing over the oxide and keeping the tempera-
ture constant. pO2 was varied in the range 10ꢀ22–2 ꢂ 10ꢀ1 atm.
Equilibrium was insured by monitoring the weight loss and allow-
ing the samples to reach constant weight at each pO2, in some
cases this required several hours.
The equilibrium non-stoichiometry (d) as a function of O2 par-
tial pressure at 958 K for Rh/LSFC and Rh/LSFCA is reported in
Fig. 5. These data show the behavior of the pure perovskite oxides,
i.e. the contribution from Rh2O3 reduction was taken into consider-
ation (see experimental section). The oxygen stoichiometry at
room temperature was measured by cerimetric titration. For both
samples, a value of 3.0 was found. The initial starting 3 ꢀ d value
at 958 K was determined as described in Section 2.4. The initial
3 ꢀ d is 3 for Rh/LSFC and 2.975 for Rh/LSFCA. This corresponds
to an average valence for the B-site of 3.2 for both samples. In
the case of Rh/LSFC, the 3 ꢀ d was quite independent of pO2 in
the range pO2 > 5.0 ꢂ 10ꢀ5 atm, while it decreased continuously
in the same range for Rh/LSFCA. At pO2 ꢃ 1 ꢂ 10ꢀ12 atm, the
3 ꢀ d reached a value of 2.88 for Rh/LSFC and 2.86 for Rh/LSFCA,
which corresponds to an average valence for Fe and Co of 2.93
for Rh/LSFCA and 2.96 for Rh/LSFC. Over the pO2 range
1.0 ꢂ 10ꢀ10–1.0 ꢂ 10ꢀ20 atm 3 ꢀ d was almost independent from
the pO2. These results are in good agreement with the literature
where a similar plateau was reported in the case of La1ꢀxSrxFeO3ꢀd
for a 3 ꢀ d value corresponding to an Fe valence of 3 [35]. The two
materials appeared to differ considerably at high pO2. Rh/LSFC had
a rather constant oxygen non-stoichiometry in the pO2 range 100–
10ꢀ5 atm, while Rh/LSFCA was continuously reduced in the same
interval. Hence, the introduction of Al into the B-site facilitates
reduction already in that pO2 range. A similar effect of Al doping
Earlier studies indicated a correlation between the CO selectiv-
ity and the degree of reduction for ceria-based materials [14,15]. In
order to elucidate whether such a correlation would exist for the
studied reducible Rh-promoted perovskites, the CO and H2 selec-
tivity was plotted as a function of oxygen consumption during 50

30
F. Mudu et al. / Journal of Catalysis 275 (2010) 25–33
100
80
60
40
20
0
100
80
60
40
20
Rh/LSFCA
Rh/LSFCB
Rh/LSFC
LSFCA
LSFCB
LSFC
a
b
0
0
10
20
30
40
50
0
10
20
30
40
50
Pulse of CH4
Pulse of CH4
Fig. 6. Methane conversion (a) and CO selectivity (b) as function of number of methane pulses in transient catalytic tests.
was subjected to 25 CH4 pulses followed by another 14 O2 pulses.
This group of sequences was repeated 13 times after the initial
(50 + 14) pulse sequence. The first point in Fig. 8 has a low average
selectivity and represents the data shown in Figs. 6 and 7. The sec-
ond group of CH4 pulse sequences gave a very high average CO
selectivity (98%). Thereafter, a drop in selectivity was observed be-
fore stabilizing around the fourth group of sequences to a stable CO
selectivity of 85%. The high CO selectivity obtained during the sec-
ond CH4 pulse sequence clearly supports the hypothesis that selec-
tive CH4 partial oxidation to synthesis gas can be achieved over a
partially oxidized catalyst. The decrease in average CO selectivity
during the third and fourth CH4 pulse sequences and the lower va-
lue observed during the rest of the experiment, compared to the
second sequence, were due to an imbalance between the con-
sumed and supplied oxygen.
Fig. 7. CO selectivity (full symbols) and H2 selectivity (empty symbols) as
function of consumed oxygen as registered during transient catalytic tests for Rh/
LSFCA, Rh/LSFCB and Rh/LSFC.
a
3.4.1. Rhodium loading
The role of Rh loading was investigated for LSFCA impregnated
with 0.1, 0.5 and 1 wt% Rh. The results from catalytic pulse exper-
iments are summarized in Fig. 9 and Table 3. Fig. 9a shows the CH4
conversion versus the number of CH4 pulses. A significantly lower
conversion for each CH4 pulse, starting at 56%, was observed for the
catalyst with the lowest Rh loading. During the first 6 CH4 pulses, a
decrease in conversion was observed followed by a plateau with
CH4 pulses in Fig. 7. The oxygen consumption was calculated on
the basis of oxygen present in the products (CO, CO2 and H2O) after
each CH4 pulse. The data show clearly that the first oxygen deliv-
ered by the solid oxygen source yielded total oxidation products,
whereas the subsequent oxygen release gradually led to increasing
CO (and H2) selectivity. The critical oxygen consumption (‘‘thresh-
old value”) at which a switch in product selectivity was observed
differs between the investigated materials. For the non-promoted
materials, the oxygen consumption during the 50 CH4 pulses were
below the threshold values registered for the corresponding pro-
moted materials (Fig. 7 and Table 2). This empirically explains
the low CO and H2 selectivity.
The CO and H2 selectivity of the Rh/LSFCB was higher at lower O
consumed per g of catalyst, compared to Rh/LSFCA. The reasons for
this observation was the lower fraction of the perovskite oxide
phase present per gram of the catalysts for LSFCB (65 wt%).
Fig. 7 suggests that a high CO and H2 selectivity was achieved
below a certain oxygen content of the reducible perovskite. If ruled
solely by this parameter, one would expect that a high CO and H2
selectivity should remain also subsequent to partial re-oxidization
as long as the oxygen content remains within the selective range
(i.e. below threshold). This was in turn verified by re-oxidizing
100
90
80
70
60
50
100
90
80
70
60
50
0
2
4
6
8
10
12
14
Group of Pulses
Fig. 8. CH4 conversion and CO selectivity as a function of number of groups of
pulses. Each group from nb 2–14 includes 25 methane pulses followed by 14
oxygen pulses. The first group includes 50 methane pulses followed by 14 oxygen
pulses.
Rh/LSFCA with 14 O2 pulses (14 lmoles O2, which corresponds
approximately to the amount of CO formed during the preceding
50 CH4 pulses, see Table 2, line 7) at 873 K. Thereafter, the catalyst

F. Mudu et al. / Journal of Catalysis 275 (2010) 25–33
31
100
80
60
40
20
0
100
80
60
40
0.1% Rh
0.5% Rh
1% Rh
20
0
a
b
0
10
20
30
40
50
0
20
40
60
80
100
Pulse of CH4
μmol of O consumed
Fig. 9. LSFCA loaded with 0.1, 0.5 and 1% Rh. Methane conversion as a function of methane pulses (a) and CO selectivity as a function of consumed oxygen from the perovskite
(b).
Table 3
Pulse test results for different Rh loading on LSFCA.
Rh (wt%)
CH4 conversion
CO selectivity (%)
C deposited
O useda
CO2
CO
H2
0.1
0.5
1
57
94
96
41
58
52
nd
4
8
78(77)
105(99)
109(97)
17
20
21
12
25
23
24
60
62
a
Numbers in brackets do not take into account the oxygen used to the Rh2O3 reduction.
approximately. 43% CH4 conversion. A much higher conversion,
with just a small and smooth reduction after some 20 pulses,
was observed for the catalysts with 0.5 and 1.0 wt% Rh loadings.
CO selectivity versus oxygen consumption is shown for the
three catalysts in Fig. 9b. The three catalysts reached 100% CO
selectivity at similar oxygen consumption values. As shown in Ta-
ble 3, the amount of coke on the catalyst surface after reaction in-
creased with increasing Rh loading.
100
80
60
40
20
0
873 K
923 K
973 K
873 K with Carbon
A key objective in the previous literature reports has been to de-
velop a selective catalyst for CO and H2 formation, according to a
sequential reaction scheme [38,39]:
\O"
\O"
CH4 ! CO;H2 ! CO2; H2O
ð7Þ
ka
kb
However, there is mounting evidence that the second reaction is
reversible, and rapid compared to first step for a large range of sup-
ported metal catalysts [40,41].
-30
-25
-20
-15
-10
-5
0
log pO2
Reducible metal oxides, such as perovskites and ceria-based
materials, may possibly contribute to a selective direct reaction
pathway for partial CH4 oxidation to synthesis gas [15,20,25]. It
has been suggested that there exist two types of oxygen species
in reducible oxides: one very reactive giving full oxidation and an-
other less reactive giving predominantly partial oxidation
[24,42,43]. The two oxygen species could either be ascribed to
framework (‘‘lattice”) versus loosely bound surface oxygen, or to
two types of framework oxygen. Presently, the amount of surface
oxygen (estimated by monolayer coverage; for specific surface area
as reported in Table 1) was initially some 5 lmoles per 100 mg
material. This amount was just a fraction of the oxygen consumed
before reaching the threshold value and shift in selectivity toward
CO, occurring around 35–70 lmoles of oxygen (Figs. 7, 9 and 10).
This strongly suggests that loosely bound oxygen at the catalyst
surface had no major influence on the selectivity at high pulse
numbers. Thus, the framework oxygen appears to play a key role.
This is supported by the experiment reported in Fig. 8, where it
was shown that as long as the total oxygen content is kept below
the ‘‘threshold value,” high selectivities to CO were retained during
Fig. 10. Calculated CO selectivity as function of oxygen partial pressure assuming
equilibrium between CO, CO2 and O2 (and C in one case) based on standard G of
G reported
D
formation reported in the NIST database. In the case of C, the standard
for amorphous carbon was used.
D
the full CH4 pulse cycle. The observed CO selectivity does hence
correlate with the degree of reduction of the perovskite oxide
(Fig. 7). Notably, neither the nature nor the amount of noble metal
at the surface influenced the CO selectivity (Fig. 9 and Fig. S5).
The oxygen non-stoichiometry of the (perovskite) oxide de-
pends at equilibrium on temperature and O2 partial pressure
[44]. At equilibrium, the partial pressure of O2 over the oxide is di-
rectly related to its degree of reduction. This suggests that syngas
selectivity over some reducible oxides might be controlled by ther-
modynamics just as observed in conventional catalytic partial oxi-
dation (see below). Calculated equilibrium selectivities to CO and
CO2 versus O2 partial pressure (at a constant carbon amount) at
873–973 K are shown in Fig. 10. CO selectivities above 90% are

32
F. Mudu et al. / Journal of Catalysis 275 (2010) 25–33
predicted at O2 partial pressure below 10ꢀ22 at 973 K and below
10ꢀ26 atm at 873 K. When including the reaction of carbon forma-
tion from CO in the calculations (considering equilibrium in both
est. For 0.5 wt% Rh/LSFCA at 958 K, this window is between
3 ꢀ d = 2.71 and 2.78.
the reactions 1/2O2 + Ca = CO,
D
G873 = ꢀ50.42 kJ/mol and 1/
4. Conclusion
2O2 + CO = CO2,
D
G873 = ꢀ49.36 kJ/mol), using thermodynamic val-
ues for amorphous carbon at 873 K, carbon formation becomes fa-
vored over CO formation at O2 partial pressures below 10ꢀ24 atm
with maximum CO selectivity 40%. It should be noted that carbon
formed by CH4 decomposition may differ significantly from the
amorphous carbon reported in thermodynamic tables. E.g. Rost-
rup-Nielsen recently reported that for noble metal-based catalysts
coke-free operation of CH4 reforming is achieved inside the C:H:O
regime predicted to give amorphous carbon [45]. In the current
study, carbon deposition was observed only during the 20 last
pulses of a CH4 pulse sequence (see comments to Fig. 6 and Table 2
above). We therefore assume that carbon formation was insignifi-
cant during the first 30 CH4 pulses.
Catalytic tests and thermogravimetric analysis have been com-
bined to investigate the thermodynamic dependency of the CO
selectivity over La0.8Sr0.2Fe0.8Co0.2O3ꢀd and La0.75Sr0.25Fe0.6Co0.15
-
Al0.25O3ꢀd impregnated with Rh during cyclic anaerobic CH4 partial
oxidation at 873 K. High CO yields and selectivity were obtained in
an O2 activity range below 10ꢀ22 atm. The perovskite-type oxygen
reservoir remained stable throughout several reduction/oxidation
cycles as seen by synchrotron in situ X-ray diffraction analysis.
The role of Rh related solely to the activation of CH4, without influ-
ence on selectivity. Chemical modification via Al-substitution af-
fected the redox properties of the material and reduced the unit
cell volume expansion during reduction (and cycling).
Fig. 5 shows the relation between equilibrium O2 partial pres-
sure over LSFC and LSFCA perovskites versus degrees of reduction,
obtained by thermogravimetry. These O2 partial pressure relations
and the thermodynamic data in Fig. 10 were next used to calculate
equilibrium CO selectivity versus degree of reduction of the oxide
catalysts. A direct comparison between thus predicted CO selectiv-
ities (based on TG data) and measured CO selectivities (from cata-
lytic tests) versus oxygen non-stoichiometry d for Rh/LSFC and Rh/
LSFCA is shown in Fig. 11a and b. The initial oxygen content 3 ꢀ d
for the materials in the catalytic tests was set to the equilibrium
value at pO2 = 10ꢀ5 atm, i.e. the highest pO2 of the carrier gas.
Fig. 11 shows a strong correlation between the data deduced from
thermodynamic studies and from transient catalytic tests. This
strongly suggests that the selectivity of the catalytic partial oxida-
tion of CH4 is determined by the very redox properties of the per-
ovskites and hence by thermodynamics. The observation that the
oxygen consumption before reaching high selectivity differs
among the oxides (Figs. 7 and 11) is explained by the lower oxygen
content in Rh/LSFCA at the start of the catalytic test (10ꢀ5 atm
pO2).
Acknowledgments
This publication is part of the Remote Gas project (168223/S30),
performed under the strategic program Petromaks of The Research
Council of Norway. The authors acknowledge the additional part-
ners; Statoil, UOP, Bayerngas Norge, Aker Solutions, DNV for
support.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.jcat.2010.07.004.
References
[1] BP, Statistical review of world energy. <http://www.bp.com/productlanding.
do?categoryId=6929&contentId=7044622>.
[2] I.E. Agency, World energy outlook fact sheet. <http://www.worldenergyoutlook.
org/>.
[3] J.M. Moulijn, M. Makkee, A. Van Diepen, Chemical Process Technology, Wiley,
2001.
[4] B.C. Enger, R. Lodeng, A. Holmen, Appl. Catal. A: Gen. 346 (2008) 1–27.
[5] J.-R. Rostrup-Nielsen, in: G. Ertl, H. Knözinger, F. Schüth, J. Weitkamp (Eds.),
Handbook of Heterogeneous Catalysis, Wiley, 2008, pp. 2882–2904.
[6] C.O. Bennett, in: W. Haag, B. Gates, H. Knoezinger (Eds.), Advances in Catalysis,
Vol. 44, Academic Press, 2000.
The present study suggests that an optimal, redox-active carrier
material for anaerobic, catalytic partial oxidation of CH4 to synthe-
sis gas should contain a large reservoir of oxygen which is only re-
leased at low O2 partial pressures, in the thermodynamic window
between selective CO and C formation for the temperature of inter-
[7] P. Silveston, R.R. Hudgins, A. Renken, Catal. Today, 1995, pp. 91–112.
[8] W.K. Lewis, E.R. Gilliland, W.A. Reed, Ind. Eng. Chem. 41 (1949) 1227–1237.
[9] K. Otsuka, T. Ushiyama, I. Yamanaka, Chem. Lett. (1993) 1517–1520.
[10] M. Funabiki, T. Yamada, K. Kayano, Catal. Today 10 (1991) 33–43.
[11] S. Kacimi, J. Barbier, R. Taha, D. Duprez, Catal. Lett. 22 (1993) 343–350.
[12] K. Otsuka, Y. Wang, E. Sunada, I. Yamanaka, J. Catal. 175 (1998) 152–160.
[13] K. Otsuka, Y. Wang, M. Nakamura, Appl. Catal. A: Gen. 183 (1999) 317–324.
[14] M. Fathi, E. Bjorgum, T. Viig, O.A. Rokstad, Catal. Today 63 (2000) 489–497.
[15] P. Pantu, K. Kim, G.R. Gavalas, Appl. Catal. A: Gen. 193 (2000) 203–214.
[16] V.A. Sadykov, T.G. Kuznetsova, G.M. Alikina, Y.V. Frolova, A.I. Lukashevich, Y.V.
Potapova, V.S. Muzykantov, V.A. Rogov, V.V. Kriventsov, D.I. Kochubei, E.M.
Moroz, D.I. Zyuzin, V.I. Zaikovskii, V.N. Kolomiichuk, E.A. Paukshtis, E.B.
Burgina, V.V. Zyryanov, N.F. Uvarov, S. Neophytides, E. Kemnitz, Catal. Today
93–95 (2004) 45–53.
[17] V.A. Sadykov, T.G. Kuznetsova, Y.V. Frolova-Borchert, G.M. Alikina, A.I.
Lukashevich, V.A. Rogov, V.S. Muzykantov, L.G. Pinaeva, E.M. Sadovskaya,
Y.A. Ivanova, E.A. Paukshtis, N.V. Mezentseva, L.C. Batuev, V.N. Parmon, S.
Neophytides, E. Kemnitz, K. Scheurell, C. Mirodatos, A.C. van Veen, Catal. Today
117 (2006) 475–483.
[18] T.V. Choudhary, S. Banerjee, V.R. Choudhary, Appl. Catal. A: Gen. 234 (2002) 1–23.
[19] Y. Zeng, S. Tamhankar, N. Ramprasad, F. Fitch, D. Acharya, R. Wolf, Chem. Eng.
Sci. 58 (2003) 577–582.
[20] X.P. Dai, Q. Wu, R.J. Li, C.C. Yu, Z.P. Hao, J. Phys. Chem. B 110 (2006) 25856–
25862.
[21] X. Dai, C. Yu, R. Li, Q. Wu, Z. Hao, J. Rare Earths 26 (2008) 76–80.
[22] X. Dai, C. Yu, Q. Wu, J. Nat. Gas Chem. 17 (2008) 415–418.
[23] J.N. Kuhn, U.S. Ozkan, Catal. Lett. 121 (2008) 179–188.
[24] X.P. Dai, R.J. Li, C.C. Yu, Z.P. Hao, J. Phys. Chem. B 110 (2006) 22525–22531.
[25] V.V. Kharton, M.V. Patrakeev, J.C. Waerenborgh, V.A. Sobyanin, S.A.
Veniaminov, A.A. Yaremchenko, P. Gaczynski, V.D. Belyaev, G.L. Semin, J.R.
Frade, Solid State Sci. 7 (2005) 1344–1352.
.
.
.
.
.
.
.
.
.
.
.
.
Fig. 11. CO selectivity as a function of oxygen non-stoichiometry as measured by
thermogravimetric analysis and as measured by transient catalytic tests for Rh/
LSFCA and Rh/LSFC.

F. Mudu et al. / Journal of Catalysis 275 (2010) 25–33
33
[26] B.H. Toby, J. Appl. Crystallogr. 34 (2001) 210–213.
[27] A.C. Larson, R.B. Von Creele, Los Alamos National Laboratory Report LAUR,
2000, pp. 86–748.
[36] M.V. Patrakeev, V.V. Kharton, Y.A. Bakhteeva, A.L. Shaula, I.A. Leonidov, V.L.
Kozhevnikov, E.N. Naumovich, A.A. Yaremchenko, F.M.B. Marques, Solid State
Sci. 8 (2006) 476–487.
[28] F. Riza, C. Ftikos, F. Tietz, W. Fischer, J. Eur. Ceram. Soc. 21 (2001) 1769–1773.
[29] L.W. Tai, M.M. Nasrallah, H.U. Anderson, D.M. Sparlin, S.R. Sehlin, Solid State
Ionics 76 (1995) 259–271.
[30] M. Machida, K. Eguchi, H. Arai, J. Catal. 103 (1987) 385–393.
[31] M. Machida, K. Eguchi, H. Arai, J. Catal. 120 (1989) 377–386.
[32] M. Machida, K. Eguchi, H. Arai, J. Catal. 123 (1990) 477–485.
[33] R.D. Shannon, Acta Crystallogr. Sect. A 32 (1976) 751–767.
[34] J.E. Readman, A. Olafsen, Y. Larring, R. Blom, J. Mater. Chem. 15 (2005) 1931–
1937.
[37] J.C. Slaa, R.J. Berger, G.B. Marin, Catal. Lett. 43 (1997) 63–70.
[38] D.A. Hickman, L.D. Schmidt, Science 259 (1993) 343–346.
[39] Y. Boucouvalas, Z. Zhang, X. Verykios, Catal. Lett. 40 (1996) 189–195.
[40] J. Wei, E. Iglesia, J. Phys. Chem. B 108 (2004) 7253–7262.
[41] J. Wei, E. Iglesia, J. Phys. Chem. B 108 (2004) 4094–4103.
[42] R. Li, C. Yu, S. Shen, J. Nat. Gas Chem. 11 (2002) 137–144.
[43] R. Li, C. Yu, G. Zhu, S. Shen, Petrol. Sci. 2 (2005) 19–23.
[44] S. Stolen, E. Bakken, C.E. Mohn, Phys. Chem. Chem. Phys. 8 (2006) 429–447.
[45] J.R. Rostrup-Nielsen, J. Sehested, J.K. Norskov, Adv. Catal. 47 (2002) 65–139.
[35] J. Mizusaki, M. Yoshihiro, S. Yamauchi, K. Fueki, J. Solid State Chem. 58 (1985)
257–266.