Surface properties of Sr- and Co-doped LaFeO3

Article abstract of DOI:10.1016/j.jcat.2007.10.005

The surface properties of Fe-based perovskite-type oxides with the formula La_0.6Sr_0.4Co_yFe_1-yO_3-δ for y = 0.1, 0.2, and 0.3 were investigated. Using methanol as a probe molecule, the amounts of basic and Lewis surface sites were determined. The types of sites were confirmed using DRIFTS. The surface basicity, as well as the reducibility, oxygen storage capacity, transition metal surface concentration, and methanol oxidation activity, were found to progress through an extrema for the intermediate Co level; that is, the behavior and performance of the samples are nonlinear with respect to increasing Co content. An electronic structural transition is proposed as the explanation for the observed trends.

Full text of DOI:10.1016/j.jcat.2007.10.005

Journal of Catalysis 253 (2008) 200–211
www.elsevier.com/locate/jcat
Surface properties of Sr- and Co-doped LaFeO3
∗
John N. Kuhn, Umit S. Ozkan
The Ohio State University, Department of Chemical and Biomolecular Engineering, Columbus, OH 43210, USA
Received 11 June 2007; revised 2 October 2007; accepted 4 October 2007
Available online 14 November 2007
Abstract
The surface properties of Fe-based perovskite-type oxides with the formula La Sr CoyFe
tigated. Using methanol as a probe molecule, the amounts of basic and Lewis surface sites were determined. The types of sites were confirmed
using DRIFTS. The surface basicity, as well as the reducibility, oxygen storage capacity, transition metal surface concentration, and methanol
oxidation activity, were found to progress through an extrema for the intermediate Co level; that is, the behavior and performance of the samples
are nonlinear with respect to increasing Co content. An electronic structural transition is proposed as the explanation for the observed trends.
O
for y = 0.1, 0.2, and 0.3 were inves-
0.6 0.4
1−y 3−δ
©
2007 Elsevier Inc. All rights reserved.
Keywords: Perovskite-type; Solid oxide fuel cell cathode; Surface characterization; Methanol oxidation; Probe molecule; Structure–property relation
1
. Introduction
All four major SOFC components (cathode, electrolyte, an-
ode, and interconnects) are commonly made from perovskite-
type oxides [8]. This fact attests to the wide variety of potential
behavior of perovskite-type oxides, because the components
require very different properties. For example, the electrolyte
must be an electronic insulator, and the interconnects must be
catalytically inactive, whereas the electrodes must be electronic
conductors and have catalytic activity. In particular, perovskite-
type oxides are especially important as electrode materials,
because cathode alternatives are limited to cermets containing
precious metals, such as Pt, and conventional anode materials
Perovskite-type (ABO3) oxide catalysts have been widely
researched due to their unique properties which have led to
applications in a number of reactions [1–3]. These applica-
tions include various total and partial oxidation reactions (e.g.,
CO, NH3, and organics), hydrogenation, dehydrogenation, hy-
drogenolysis, photocatalysis, and environmental applications
(e.g., SO2 reduction, de-NOx). For methane oxidation, sev-
eral formulations displayed performance similar to that of Pt
on Al2O3 [2]. In addition to these purely catalytic reactions,
perovskite-type materials have been used in several applica-
tions in which both their unique surface and bulk properties
are exploited. Oxygen separation membranes are a prime ex-
ample [4,5]. Because only oxygen (as oxide ions) is trans-
ported across the dense membranes, high-purity separations of
air can be achieved. These devices can be combined with cat-
alytic applications, and oxide ions can be used as the oxidant
source for ethane dehydrogenation and oxidative coupling of
methane [6,7]. The present work focuses on formulations of in-
terest for solid oxide fuel cell (SOFC) applications.
(
Ni-YSZ) deactivate due to sulfur impurities and carbon-based
fuels. Whereas Co-based materials were the first proposed
perovskite-type oxides as SOFC cathodes, Mn-based materials
quickly became popular because of their increased stability un-
der harsh conditions [8]. However, because there was a desire
to lower the SOFC operation temperature (mainly achieved by
decreasing the thickness of the electrolyte), more active materi-
◦
als were needed at intermediate temperatures (500 to 800 C).
At this range of operating temperatures, materials based on Fe
and Co and mixtures thereof are expected to provide improved
performance at lower temperatures through increased ionic con-
duction abilities.
As the ionic conductivity increases, the active reaction area
for oxygen reduction expands. When this expansion occurs, the
role of interfacial oxygen reduction kinetics becomes increas-
*
Corresponding author. Fax: +1 614 292 3769.
E-mail address: ozkan.1@osu.edu (U.S. Ozkan).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.10.005

J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
201
ingly important. As demonstrated through the work of several
researchers, surface processes that are chemical in nature, such
as surface oxygen exchange and oxygen surface diffusion, in-
fluence the performance of mixed conducting perovskite-type
materials [4,9–13]. For SOFC cathodes, Adler has proposed
that the performance of mixed conductors is co-limited by ionic
diffusion and the activation of molecular gas-phase oxygen.
However, despite their newfound importance, few studies have
focused on the characterization of the surface of these materi-
als.
The bulk oxygen pathway makes an exclusive study of the
surface difficult. For example, oxygen temperature-program-
med desorption is truly a bulk technique, because oxygen
evolves from the anionic matrix to form oxygen vacancies.
During heterogeneously catalyzed reactions, the mechanisms
can be intrafacial (involving mobile lattice oxygen) rather than
suprafacial (involving adsorbed oxygen). At temperatures high
enough for bulk oxygen to become mobile, methane oxidation
is an example of an intrafacial reaction [2].
A common surface characterization technique is the use of
probe molecules. Whereas simple molecules such as H2, CO,
CO2, and H2O are most common, hydrocarbons also have been
used to gain insight into oxidation behavior and evaluate surface
acid sites of perovskite-type oxides [2,3,14]. There has been
considerable success in using oxygenated hydrocarbons (e.g.,
alcohols) as probe molecules for supported and bulk oxide ma-
terials. Methanol was shown as a useful compound for studying
these surfaces. Product formation from methanol surface reac-
tivity was related to the nature of the active sites [15]. Acidic
surface sites led to dimethyl ether, whereas redox sites caused
formaldehyde production. Basic sites, such as those examined
for the perovskite-type oxides in the present study, resulted in
the formation of carbon dioxide. Moreover, the ability of a sur-
face to dissociate methanol yields information on the strength
of the sites [16,17]. However, such surface studies have not
been performed on perovskite-type oxides with stabilities rel-
evant for SOFC cathodes.
cations, catalytic properties (e.g., oxidation activity [23] and
oxygen storage capacity [24]) tend to increase with Co content
at the expense of stability [19] when Fe and Co are mixed on the
transition metal site. However, evidence (maximum methane
oxidation activity occurring for La0.8Sr0.2Fe0.6Co0.4O3−δ com-
pared with pure Co and Fe materials [20]) exists for complex
behavior when these transition metals are used simultaneously.
The present work examines the surface properties of La0.6-
Sr0.4CoyFe1−yO3−δ for y = 0.1, 0.2, and 0.3. In an extension
of the work on simple metal oxides, the surfaces of these mixed
perovskite-type oxides were characterized using methanol as a
probe molecule. The site density was measured using dynamic
chemisorption, whereas the nature of the sites was determined
using the products from temperature-programmed desorption
(TPD) and in situ diffuse reflectance Fourier transform infrared
spectroscopy (DRIFTS). Supporting surface information was
obtained by X-ray photoelectron spectroscopy (XPS). The re-
sults were then used to explain the product distribution and ac-
tivity differences for the different Co loadings during methanol
oxidation.
In a previous study [25], we have reported the bulk char-
acterization of the same materials. According to that work, all
samples, due to the doping of Sr, have rhombohedral symmetry
and are stoichiometric with respect of oxygen under ambient
conditions. At elevated temperatures, the transition to cubic
symmetry occurred, and the transition temperature dropped as
the Co content increased, as was expected due to size argu-
ments. For the reaction results discussed in this article, however,
all three samples are expected to be of rhombohedral symme-
try, because the lowest expected transition temperature is near
◦
500 C. The transition temperature is important because it may
lead to significant differences in phase stability under reducing
conditions [26,27]. Another important finding from our previ-
ous study [25] was that, in contradiction to what is generally
perceived for formulations containing Fe and Co, oxygen va-
cancy formation was not proportional to Co content. This trend
◦
occurred under air up to 700 C and for initial reductions un-
Fe-based materials have been the least widely studied of the
perovskite-type oxides, because they generally have lower ac-
tivity than their Co counterparts. For example, studies focusing
on the oxidation of propane [18], methanol [18], CO [2], propy-
lene [2], and isobutene [2] have demonstrated that LaCoO3 is
more active than LaFeO3. No studies have been performed on
Fe-rich LaFeO3 formulations doped with both Sr and Co. Fe-
based formulations are desirable because of improved system
compatibility, both in terms of lower thermal expansion coef-
ficients (TECs) and lower chemical reactivity with the most
common electrolytes (e.g., yttria-stabilized zirconia or YSZ),
compared with Co-rich formulations. These materials also may
catalyze oxidation reactions; recent work has shown that Fe
doping provides structural stability in Co-based materials [19].
Moreover, the addition of Sr into La(Co,Fe)O3−δ may lead
to an activity increase, as was reported for butane oxidation
over (La,Sr)CoO3−δ [20]. On the other hand, an optimal Sr
loading was observed for Fe-based materials used for ethanol
and propane oxidation [21,22] and Co-based materials used
for propane oxidation [18]. When La is doped with divalent
der more reducing environments (e.g., He and 10% H2/N2). In
addition, measured by in situ XRD under reducing conditions,
◦
the perovskite phase was the only one detected up to 800 C
in the presence of 5% H2/N2. This result indicates that the
perovskite-type phase was the only one present under the re-
action conditions used in this study.
2. Experimental
The La1−xSrxCoyFe1−yO3−δ samples were prepared by a
conventional solid-state route. The preparation and the BET
surface measurements were described previously [25]. The
mass specific surface areas of all samples were between 1.5 and
2
2.0 m /g.
2.1. Dynamic adsorption and temperature-programmed
desorption of methanol
Dynamic methanol adsorption and sequential temperature-
programmed desorption experiments were performed using a

202
J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
Thermo-Finnigan Trace Ultra differential scanning quadrupole
DSQ) gas chromatograph/mass spectrometer (GC/MS). Sam-
diation under high vacuum. Samples were ground into carbon
tape for analysis. The Mg anode, operated at 13 kV and 10 mA,
was selected as the X-ray source, to minimize interference
caused by Co and Fe Auger lines [28]. Deconvolution of the
Fe 2p region has been reported to be more difficult when an Al
source is used [19]. With a slot aperture setting, the spectrome-
ter was used in spectrum analyzer mode and hybrid lens mode.
The charge neutralizer was set at a current of 2.1 A, a bias of
1.3 V, and a charge of 2.2 V during analysis. A survey scan
(
ples (∼150 mg) were loaded into a U-tube shaped quartz reactor
between plugs of quartz wool. The samples were exposed to a
◦ ◦
3
0 mL/min flow of 20% O2/He at 850 C (10 C/min) with
an isothermal hold for 20 min. The samples were cooled to
◦
5
0 C under the same flow. At this temperature, the flow was
switched to He at 30 mL/min, and the tubing was purged for
◦
1
h. The adsorption was performed at 50 C to limit physical
adsorption and eliminate any combustion. A second He flow
also 30 mL/min) was saturated with methanol in a homemade
(
1 sweep/100 ms dwell) was acquired between 1200 and 0 eV.
Concurrent sweeps for the La 3d (8/200), Sr 3d (8/150), Co 2p
8/800), Fe 2p (8/400), O 1s (8/75), and C 1s (8/150) regions
(
bubbler at room temperature. The effluent from the bubbler was
(
◦
fed to a 50-µL loop on a 6-port valve (Valco) at 100 C for the
were obtained. The C 1s peak at 284.5 eV was used for bind-
ing energy correction. The deconvolution was performed with
Gaussian curves using the XPS Peak 4.1 program. Elemental
surface composition was calculated using transmission values
and relative sensitivity factors specific for the instrument and
the Mg source.
pulsed adsorption. All lines after the bubbler were heated to ap-
◦
proximately 100 C to reduce the methanol adsorption on the
walls of the tubing.
The chemisorbed amount was determined by averaging the
integrated 15 (CH3), 29 (COH), and 31 (CH3O) m/z fragments.
The 32 (CH3OH) m/z signal was not used, because it occasion-
ally displayed spikes caused by the valve switches. The amount
of methanol was calculated using the ideal gas and Antoine’s
equations. The adsorbed amount was calculated assuming that
the full pulses correlated with the predicted amount of methanol
contained in the loop. The adsorbed amount was converted to
site density using the specific surface area assuming 1 site per
adsorbed methanol molecule.
2
.4. Methanol oxidation activity
The catalytic activity of the samples was measured by
methanol oxidation reactions. The reactions were performed
with an Autochem II 2920 and a Cirrus residual gas-analyzing
mass spectrometer (RGA-MS). Samples were loaded on an
17
equal surface site basis (2.0 × 10 sites) into U-tube reactors
Once the signals for methanol related ions reached a steady
integrated value, the reactor was purged for 1 h in He. A linear
with the samples sitting on top of quartz wool. Samples were
◦
◦
◦
exposed to 10% O2/He at 850 C for 20 min before the re-
temperature program (50–900 C at 10 C/min) was enacted,
and the desorption products were monitored. The amount of
molecularly adsorbed methanol was determined by evaluating
the integrated area for methanol fragments using the full pulses
during the adsorption as a response factor. Peak integration was
performed with the Grams AI software package.
action. After the pretreatment, the sample was exposed to the
reaction mixture, which consisted of 10% O2/He at 23 mL/min
and He at 27 mL/min bubbled through the vapor generator. The
generator was held at a constant temperature of 33 C to pro-
vide excess methanol compared with oxygen. Using Antoine’s
equation, the methanol-to-oxygen ratio was found to be near 2.
◦
◦
2
.2. DRIFTS
All lines beyond the generator were held at 110 C. The reac-
tion was performed using a quasi-steady-state approach. The
In situ DRIFTS experiments were performed with a Ther-
◦
reaction temperature was ramped linearly at 5 C/min in 50 or
moelectron Nicolet 6700 FTIR spectrometer furnished with
a liquid nitrogen-cooled MCT detector and an environmental
◦
1
00 C intervals, then held constant for 30 min. The RGA-MS
was used to monitor the reactor effluent. The instrument was
used in scanning ion mode with the electron multiplier detector.
The signals were corrected using the ionization probability and
fragmentation patterns for all ions present. The conversions,
yields, and selectivities were calculated using the following ex-
pressions:
−
1
chamber. Spectra were obtained at a resolution of 4 cm using
−
1
5
00 scans between 650 and 4000 cm . Backgrounds were col-
lected in 25 mL/min of He after a 5-min hold after temperature
increases. After the final background was collected, the flow
was switched to air (25 mL/min) for 30 min to restore the oxy-
◦
gen content of the sample. The sample was then cooled to 50 C
before the flow was switched back to the He stream, and the
chamber was purged for 15 min. The sample was then exposed
to a steady stream of methanol (bubbler at room temperature)
for 1 h. After a 10-min purge, spectra acquisition began. The
mol of CH3OH converted
%
CH3OH conversion =
× 100,
mol of CH3OH fed
mol of O2 converted
% O conversion =
2
× 100,
◦
total mol of O2 fed
temperature was raised in 25 or 50 C increments, after which
the temperature was held constant for 5 min before the scan was
enacted.
mol of H2 produced
%
%
%
H2 yield =
2
× 100,
× (mol of CH3OH fed)
2
× (mol of H2 produced)
2
.3. X-ray photoelectron spectroscopy
H2 selectivity =
× 100,
mol of H in all products
X-ray photoelectron spectra (XPS) were collected on a
Kratos Ultra Axis Spectrometer using MgKα (1253.6 eV) ra-
mol of CO2 produced
mol of CH3OH fed
CO2 yield =
× 100,

J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
203
mol of CO2 produced
× 100.
mol of C in all products
After the surface was saturated with methanol as indicated
by constant pulse areas, desorption profiles were obtained by
TPD. Typical results are shown in Fig. 2 for La0.6Sr0.4Co0.2-
%
CO2 selectivity =
. Results
.1. Use of methanol as a probe molecule
An example of the chemisorption process is shown in Fig. 1a
3
◦
Fe0.8O3−δ. At low temperatures (80–150 C), molecularly ad-
sorbed methanol desorbed from the sample. These species have
been commonly misidentified as physically adsorbed methanol,
whereas they are actually intact Lewis-bound methanol adducts
3
[
16,17,29,30]. We present more detail on this topic when
discussing the DRIFTS results. At intermediate temperatures
for La0.6Sr0.4Co0.3Fe0.7O3−δ. As demonstrated by methanol
◦
(
4
150–300 C), chemisorbed methanol desorbed as CO2 (m/z =
(
(
m/z = 32 (CH3OH)) and its fragments (m/z = 15 (CH3), 29
4) and H2O (m/z = 18). These products formed after a sur-
COH), and 31 (CH3O)), slightly less than 2 complete pulses
face reaction of the chemisorbed methanol with available oxy-
gen from the sample’s lattice. No peaks were detected for
formaldehyde (m/z = 30), a product of redox sites. More-
over, there was no evidence of acidic sites linked to dimethyl
were adsorbed. Moreover, no peaks were detected for CO2
(
m/z = 44) besides a small amount of noise associated with
the valve switch. Adsorption profiles for the main methanol
fragment (m/z = 31 (CH O)) are shown in Fig. 1b for the
3
ether (m/z = 46). The peaks for m/z = 45 and 46 were ac-
different formulations. La0.6Sr0.4Co0.1Fe0.9O3−δ demonstrated
a greater adsorption capacity than the samples with larger
amounts of Co. The adsorption amount was converted into sur-
face site density as shown in the last column of Table 1 us-
ing the mass-specific surface areas. Generally, these values are
low compared with traditional metal oxides [15]. The surface
density measurements were reproduced for different batches
counted for by natural isotope levels of 13C (∼1%) and
18
O
◦
(
∼0.2%). Although high-temperature peaks (>300 C) were
seen for m/z = 32, these were not due to intact methanol de-
Table 1
2
Adsorption data (sites/nm ) from adsorbed methanol on La Sr Coy-
0.6 0.4
Fe
O
1
−y 3−δ
of La0.6Sr0.4Co Fe0.8O
0.2
within 0.1 site/nm of the results reported in Table 1.
, and the values were found to be
3−δ
y
Lewis-bound
Basic
Total
2
0.1
0.2
0.3
0.64
0.66
0.66
1.49
0.37
0.53
2.13
1.03
1.19
(a)
(a)
(b)
(b)
◦
Fig. 1. Pulsed methanol chemisorption at 50 C for La Sr CoyFe
(
O
.
Fig. 2. Desorption of chemisorbed methanol from La Sr Co Fe
as (a) intact methanol and (b) carbon dioxide and water after a surface reaction.
Profiles are offset for clarity.
O
0.6 0.4
1−y 3−δ
0.6 0.4 0.2 0.8 3−δ
a) Monitoring of several ions for y = 0.3. (b) m/z = 31 for different formula-
tions. Pulses indicated by arrows. Profiles are offset for clarity.

204
J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
sorption. At these temperatures, lattice oxygen evolved from
the bulk of the material to create oxygen vacancies [25].
Comparing the m/z = 32 profile obtained when no methanol
was adsorbed in our previous work with the present pro-
file shows that the relative amounts of oxygen desorbed at
low (α-oxygen) to high (β-oxygen) temperatures were differ-
ent. Here α-oxygen refers to the oxygen related to adsorbed
or near-surface oxygen. High-temperature oxygen desorption
or β-oxygen is associated with the transition metals on the
B-site. Although more β-oxygen desorbed when no methanol
was present, α-oxygen became the primary desorbing species
after methanol desorption. The results suggested that the lattice
oxygen replenished the near-surface oxygen that was consumed
to oxidize CH OH to CO and H O, and that oxygen vacancies
site density. All three samples demonstrated smaller desorption
peaks at higher temperatures than the major peak. These addi-
tional peaks show that some sites have stronger interactions and
indicate at least some surface heterogeneity.
As alluded to earlier, DRIFTS studies confirmed the na-
ture of surface species. DRIFT spectra for the desorption of
methanol are shown in Fig. 4. In the high wavenumber region,
−
1
−1
CH (∼2900 cm ) and OH (∼3700 cm ) vibrations indi-
3
cated the presence of adsorbed methanol at low temperatures.
Similarly, adsorbed methanol species was signified by the CO
−
1
vibrations (∼1050 cm ) in the low wavenumber region. As
the temperature was raised, the bands corresponding to sur-
face species decreased and the CH and OH vibrations were
3
◦
−1
3
2
2
no longer present by 150 C. Moreover, CO (∼2350 cm
2
)
◦
were more likely to form in the bulk than at or near the surface
after oxidation of the methoxy surface species.
appeared in the spectrum obtained at 125 C and existed during
the rest of the experiment. The wide temperature range of CO2
reinforced the idea of surface heterogeneity. For solid solutions
of perovskite-type materials, a heterogeneous surface was eas-
ily realized through Sr-doping, which created a mix of trivalent
and tetravalent B-site cations. Because the methanol desorbed
molecularly as methanol and as CO2 after surface reaction of a
methoxy species with oxygen from the sample, it was not sur-
prising that the CH3 vibrations were nearly extinct before the
CO evolved. A more in-depth examination of the CH vibra-
Fig. 3 compares the desorption profiles as a function of
Co content. As we discuss in the DRIFTS results section, this
species that desorbed at lower temperatures was a Lewis-bound
species. With the Lewis-bound sites quantified, it was possi-
ble to back-calculate the basic site density. The nature of the
sites is broken down in Table 1. Whereas the peak temperature
remained constant for the three samples, the intensity of the
CO desorption profiles followed the same trend as the basic
2
2
3
(a)
(a)
(b)
(
b)
Fig. 4. Methanol TPD from La Sr Co Fe
wavenumber region and (b) low wavenumber region. Profiles are offset for clar-
ity.
O
in DRIFTS. (a) High
0
.6 0.4 0.2 0.8 3−δ
Fig. 3. Desorption of chemisorbed methanol for La Sr CoyFe
a) Intact methanol desorption. (b) CO formation after surface reaction.
O
.
0.6 0.4
1−y 3−δ
(
2

J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
205
Scheme 1. Types of surface species formed during methanol adsorption.
was used as the probe molecule [2]. In this review by Pena
◦
Fig. 5. CH vibrations for adsorbed methanol species at 50 C following dif-
3
◦
and Fierro, desorption below 152 C was linked to monoden-
◦
ferent treatments at 550 C on La Sr Co Fe
O
in DRIFTS. Spectra
0
.6 0.4 0.2 0.8 3−δ
◦
tate carbonates, and desorption at 267–647 C was speculated
are offset for clarity.
to occur due to decomposition of a combination of bidentate
and bridged carbonates. Generally, these desorption peaks fit
with the temperature ranges given in Fig. 3b. Moreover, the
bidentate and bridge carbonates were linked to the extent of the
oxide reduction in the samples [2]. For this reason, the study
was reexamined after exposure to an inert environment at el-
Table 2
DRIFTS band assignments for Fig. 5
−
1
Band (cm
)
Vibration
Asymmetric CH stretch
3007
2951
2922
2866
2845
2827
3
Symmetric CH (LB) stretch
3
◦
Symmetric CH stretch
3
evated temperatures (550 C), as shown in Fig. 5, to look for
Asymmetric CH bend
3
effects caused by oxygen vacancies. La0.6Sr0.4Co0.2Fe0.8
O
3−δ
Symmetric CH (LB) bend
3
has been shown to be nearly stoichiometric with respect to oxy-
gen content at this temperature in air, whereas a significant
amount of oxygen vacancies form in He [25]. Whereas the bulk
and surface defects may not be exclusively linked (e.g., surface
vacancies may exist even when the bulk is fully oxidized or vice
versa), the amount of surface vacancies should increase as oxy-
gen vacancies form in the bulk.
Symmetric CH bend
3
tions yielded important information on the types of adsorbed
methanol species.
The CH3 vibrations are shown in much greater clarity in
Fig. 5. Results are shown after the adsorption of methanol
when an oxidative (air) and an inert (He) environment (at
Based on the two surface structures, one would intuitively
expect a more oxidized surface to lead to the dissociated sur-
face methoxy species, because oxygen is available to extract
hydrogen from methanol. For similar reasoning, formation of
the Lewis-bound methanol species would be favored on a less-
oxidized surface. Looking at the bands for CH3 stretching,
the intensities of the bands for Lewis-bound species remained
relatively constant compared with the bands for the methoxy
species as the environment before methanol adsorption was
changed. The lack of differences is explained by the replen-
ishment of the surface and near-surface oxygen with the lattice
oxygen, which was in agreement with the TPD data and SrO
suboxide present at the surface (discussed in Section 3.2). Be-
cause no bands for carbonate species were detected (due to dark
samples with low surface areas), direct assertions cannot be
reached for the intermediates between the methoxy species and
CO2. Instead, it can only be inferred from the TPD that the ma-
jor species involve a monodentate carbonate, whereas smaller
amounts of more stable bidentate or bridged carbonates exist.
◦
5
50 C in both cases) were used before adsorption. Band
assignments for the CH3 region are presented in Table 2.
The band designations have been compiled from literature on
methanol–oxide surface interactions [29,30]. The presence of
both surface methoxy and molecularly adsorbed or Lewis-
bound (LB) species has been identified. The Lewis-bound
methanol species are signified by the relatively sharp bands
at 2951 and 2845 cm . Generally, the undissociated Lewis-
bound methanol species are found at slightly lower frequencies
than the surface methoxy species [16]. Moreover, undissoci-
ated adsorbed methanol species have a sharp CO vibration
−
1
−1
near 1030 cm [16]. Fig. 4 shows that the sharp CO vibra-
tion existed at 1037 cm . The bands at 2827, 2866, 2922, and
−
1
−
1
3
007 cm were associated with various stretches and bends
(
first overtones) of surface methoxy species. With these des-
ignations, likely surface structures for the different types of
adsorbed methanol are formulated in Scheme 1. Reviews by
Lavalley and Busca were instrumental in constructing the sur-
face structures [16,17].
Besides methanol adsorbing as two different types of sur-
face species, the methoxy species led to CO2 evolution over a
wide range of temperatures. Similar types of desorption behav-
ior has been reported for perovskite-type materials when CO2
3.2. Surface characterization using XPS
Examples of the peak deconvolution and spectra compared
as a function of Co content are shown in Figs. 6–8, with the

206
J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
(a)
(b)
(c)
(d)
Fig. 6. X-ray photoelectron spectra for A-site constituents of La Sr CoyFe
O
. (a) Deconvolution for La 3d when y = 0.2. (b) Comparison of La 3d region
0.6 0.4
1−y 3−δ
for different B-site compositions. (c) Deconvolution for Sr 3d when y = 0.2. (d) Comparison of Sr 3d region for different B-site compositions. Spectra are offset for
clarity.
Table 3
837 eV that could be caused by hydroxyl formation [28,31].
Whereas the positions did not change with Co content, the sur-
face composition did. Sr existed in two divalent oxide forms
with 3d5/2 binding energies at 131.5 and 133.2 eV. These val-
ues match those reported for La0.4Sr0.6Co0.8Fe0.2O3−δ, and
the two peaks may arise from Sr in the perovskite phase and
a suboxide phase (although the presence of surface carbon-
ates cannot be excluded) [31]. The Sr-based suboxide phase,
which may be less likely to form oxygen vacancies than
the perovskite-type phase, may explain why the surface ap-
peared less likely to form oxygen vacancies than the bulk.
The Co 2p3/2 peaks were slightly above 780 eV. Values near
XPS peak parameters extracted from Figs. 6–8
Region
Position (eV)
Surface composition (%)
y = 0.1 y = 0.2 y = 0.3 y = 0.1 y = 0.2 y = 0.3
La 3d
Total
833.1
36.7
834.3
837.7
833.2
837.0
3.6
3.8
7.3
1.2
5.2
6.4
2.0
2.3
4.3
5/2
8
Sr 3d
132.5
131.8
131.6
133.1
13.4
7.8
5.7
8.5
7.4
5/2
133.4
Total
13.4
0.5
13.5
15.9
Co 2p
780.1
710.0
529.4
780.2
709.9
780.1
709.7
1.3
5.8
1.1
5.3
3/2
Fe 2p
6.4
3/2
∼
779.5 eV have been reported for trivalent species in Co2O3
O 1s
530.5
533.2
528.2
531.0
43.0
29.3
72.3
29.7
43.4
73.1
24.3
49.1
73.4
and LaCoO [1,28]. A small shift [31] toward higher binding
3
531.9
energy may indicate the presence of tetravalent species that
Total
formed through Sr doping. A value of 780.2 eV was reported
for La0.4Sr0.6Co0.8Fe0.2O3−δ [31]. As demonstrated by the lack
of the satellite peaks near 787.5 eV, divalent Co, an indicator of
impurities at the surface, was not present [1,19]. The Fe 2p3/2
peaks were near 710 eV. Although none of the usual Fe species
are in this region, the location fits well with the values given for
perovskite-type oxides with only La on the A site [19,28,31].
The O 1s region exhibited the most interesting surface behavior,
with large shifts detected as a function of Co content. The low
A-site cations in Fig. 6, the B-site cations in Fig. 7, and oxygen
in Fig. 8. The spectra for each sample were deconvoluted as
shown in the example. The peak positions and the elemental
surface compositions extracted from the spectra deconvolution
are given in Table 3.
La existed in two forms: a trivalent oxide with a 3d5/2
binding energy near 833.5 eV and a less-oxidized form near

J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
207
(a)
(b)
(c)
(d)
Fig. 7. X-ray photoelectron spectra for B-site constituents of La Sr CoyFe
O
. (a) Deconvolution for Co 2p when y = 0.2. (b) Comparison of Co 2p
0.6 0.4
1−y 3−δ
region for different B-site compositions. (c) Deconvolution for Fe 2p when y = 0.2. (d) Comparison of Fe 2p region for different B-site compositions. Spectra are
offset for clarity.
(a)
(b)
Fig. 8. O 1s region in X-ray photoelectron spectra for La Sr CoyFe
different B-site compositions. Spectra are offset for clarity.
O
. (a) Deconvolution for O 1s when y = 0.2. (b) Comparison of O 1s region for
0.6 0.4
1−y 3−δ
binding energy peak correlated to anion network of lattice ox-
ide ions in the perovskite structure [31,32]. The shifts displayed
the same trend as the bulk reducibility determined by TPR [25].
The high binding energy peak is related to chemisorbed oxygen
species.
present findings are in agreement with these conclusions. The
results given in Table 3 demonstrate relatively good agreement
with the composition expected from the perovskite phase (i.e.,
20% A site, 20% B site, and 60% anions), but also demonstrate
deviations from the bulk as O became enriched at the surface
in comparison to the transition metals. Excess oxygen at the
surface may be correlated with the fact that the surface is less
The surface composition of perovskite-type materials have
been known to deviate from the bulk composition [3], and our

208
J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
Partial methanol oxidation:
Scheme 2), it is noteworthy that the catalysts showed a slight
preference for forming CO2 compared with H2O at complete
conversion.
Because single carbon (C1) chemistry can be significantly
more complex [33] than discussed in Scheme 2, we briefly
discuss the apparently nonoperative reactions. Although the
studies were performed on Co- or Fe-based materials, other
perovskite-type oxides also have been used for converting
methanol to methyl formate and formaldehyde, with the pos-
sibility of dimethyl ether and methane impurity formation [34–
1
CH3OH + O2 → CO2 + 2H2,
2
CH3OH + O2 → CO2 + H2 + H2O,
1
2
CH3OH + O2 → CO + H2 + H2O.
Complete methanol oxidation:
3
2
CH3OH + O2 → CO2 + 2H2O.
Methanol decomposition:
CH3OH → CO + 2H2.
CO oxidation:
3
7]. Methane formed through the dry reforming of methyl for-
2
CO + O2 → 2CO2.
H2 oxidation:
H2 + O2 → 2H2O.
mate [34]. Perovskite-type oxides also exhibited activity for
hydrogenation, hydrogenolysis, and oxidative coupling reac-
tions [2,33]. Consistent with the TPD studies, neither dimethyl
ether nor formaldehyde formed under the conditions used in
this study. On the other hand, small amounts of methyl formate
2
Water–gas shift:
CO + H2O → CO2 + H2.
Methanol steam reforming:
CH3OH + H2O → CO2 + 3H2.
◦
and methane formed at 450 and 500 C, respectively. Thus, it
is possible that similar behavior occurred as has been reported
for Cu perovskite-type oxides [34]. Because basic metal oxides
have been linked to the direct dehydrogenation of methanol to
make methyl formate [34], its formation was not unexpected.
However, the presence of these species was small, and the sam-
ples behaved primarily as total oxidation catalysts.
Scheme 2. Possible reactions for methanol oxidation.
likely to contain anion vacancies (from differences in O-TPD
with and without methanol adsorption). Moreover, Sr was en-
riched at the surface relative to La, and a surface enriched in
the dopant may be responsible for the differences in oxygen va-
cancy preference between the surface and the bulk.
In the absence of a complete reaction network analysis,
several test reactions were monitored through temperature-
◦
◦
programmed reactions (100 mg, 10 C/min from 50 to 700 C,
and each reactant at 5% with He balance) for La0.6Sr0.4Co0.2-
Fe0.8O3−δ. This sample exhibited good activity for CO oxi-
dation and the water–gas shift reaction, but poor activity for
methane-steam reforming. Activity for CO oxidation and the
3
.3. Catalytic activity
Methanol oxidation was selected as a model reaction (im-
◦
portant reactions shown in Scheme 2) to link the reaction per-
formance to measured properties. All three samples showed
the same general trends as discussed here unless noted oth-
erwise. Appreciable reactant conversions were first observed
water–gas shift reactions began at 215 and 450 C and maxi-
mum conversion was obtained at 450 and 550 C, respectively.
◦
The sample exhibited no activity for methane-steam reforming
◦
below 700 C. It is also worth noting that exposure to 50 ppm
◦
at 250 C. At low methanol conversions (and, consequently,
◦
H2S/N2 at 700 C for 5 h had a minimal impact on the results;
low temperatures), the complete oxidation reaction occurred, as
demonstrated by high CO2 and low H2 selectivities. H2 and CO
for example, it shifted the CO oxidation profiles toward higher
◦
temperatures by only roughly 50 C and had no effect on the
◦
◦
were not detected until 350 C. The results at 300–400 C are
discussed in more detail later, because the primary differences
between samples were observed in this temperature range. The
methanol conversion was further increased by raising the tem-
water–gas shift results. Moreover, the samples demonstrated no
signs of coking when TPO experiments were conducted after
◦
steady reaction at 700 C for 4 h.
The reaction results for methanol oxidation are shown in Ta-
◦
perature, and the primary products changed. At 500 C, the CO2
yield progressed through a maximum. Also at this temperature,
H2 became the predominant product. At all temperatures, oxy-
◦
ble 4 as a function of temperature (300, 350, and 400 C) and
Co content. At these temperatures, the intermediate Co con-
tent led to the highest methanol and oxygen conversions. In
addition, the high-Co loading sample had better activity than
the low-Co loading sample. Thus, these results were consistent
with property changes that do not scale proportionally to the Co
content.
◦
gen was converted at higher rates than methanol. At 500 C,
the maximum in the CO2 yield aligned with the oxygen con-
version, reaching nearly 100%. Only ∼75% of the methanol
was converted. Thus, the oxidized products were favored when
the methanol conversion lagged behind the oxygen conversion.
Moreover, any oxygen escaping from the lattice in conjunction
with the formation of oxygen vacancies also would shift the
4
. Discussion
◦
reaction toward the oxidized productions. At 600 and 700 C,
H2 was even more favored over H2O, whereas CO and CO2
formed almost evenly. There was no longer a preference for the
formation of oxidized products, because the methanol and oxy-
gen conversions were 100%. Because the product distribution
from the partial oxidation can be different (as demonstrated in
We discuss the differences in the surface properties, with an
emphasis on trends with surface basicity and oxygen storage
capacity and influence of a potential electronic structure tran-
sition that occurred with Co content. We compare the catalytic
properties for the oxidation of volatile organic compounds with

J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
209
Table 4
Reaction comparison for methanol oxidation
Table 5
a
Surface area-normalized reaction rates and turnover frequencies for methanol
oxidation
Catalyst
Conversion (%)
1−y 3−δ CH OH
Yield (%)
CO
Selectivity (%)
H CO
2
La Sr CoyFe
O
Catalyst
Methanol rate
(mL methanol/(m min))
Methanol TOF
0.6 0.4
O
H
3
2
2
2
2
2
−1
La Sr CoyFe
O
(s
)
0.6 0.4
1−y 3−δ
◦
T = 300 C
◦
T = 300 C
y = 0.1
y = 0.2
y = 0.3
6
23
7
15
36
16
0
0
0
7
12
5
6
0
0
100
100
100
y = 0.1
y = 0.2
y = 0.3
0.4
6.6
2.4
0.07
2.56
0.81
◦
T = 350 C
◦
T = 350 C
y = 0.1
y = 0.2
y = 0.3
12
44
28
31
80
56
1
6
2
17
39
29
10
30
11
100
95
97
y = 0.1
y = 0.2
y = 0.3
0.7
12.7
9.6
0.14
4.90
3.23
◦
T = 400 C
◦
T = 400 C
y = 0.1
y = 0.2
y = 0.3
34
46
44
68
86
83
6
10
10
37
45
47
25
39
34
92
92
91
y = 0.1
y = 0.2
y = 0.3
2.1
13.3
15.2
0.39
5.13
5.07
a
Reaction conditions: CH OH/O /He = 10.5/4.9/84.6, feed flow rate =
3
2
50 mL/min, and equal surface sites.
tion and thermal stability. Second, the samples were prepared
through a solid-state route, as opposed to alternative methods
that can yield higher surface areas. In addition to its simple na-
ture, this preparation method was used, because low surface
area applications are of primary interest, deactivation through
sintering is minimized, and the lower cost of precursors makes
this route more economically feasible for commercialization.
Nonetheless, the preparation of perovskite-type oxides with
high surface areas while also maintaining good stability is im-
portant and currently under study [3,19,40].
other perovskite-type oxides used in similar reactions. We also
address differences in preparation techniques.
4
.1. Catalytic applications of doped LaFeO3 materials
We compared the Sr- and Co-doped samples with available
results for similar reactions as reported in the literature over
perovskite-type oxides and found that the samples studied in
this work were generally less active. Compared with LaNiO3,
◦
Despite the worse performance compared with previously
reported studies, the samples still may have catalytic applica-
tions due to their high thermal stability. As discussed previ-
ously [25], La0.6Sr0.4Co0.2Fe0.8O3−δ was stable under reducing
methanol conversions lagged behind by roughly 100 C, but the
differences more than likely can be attributed to the low flow
rate (4 mL/min) used in the study (insufficient information
given to compare the surface area-normalized reaction rate or
TOF) [36]. Higher temperatures were needed to achieve simi-
lar methanol oxidation activity compared with La0.6Sr0.4MnO3,
◦
conditions up to 800 C. Such stability is much higher than that
for many perovskite-type oxides and allows the use of these
samples in applications in which reducing conditions prevail
and the sample’s robust nature is exploited. For example, these
materials could be used as part of bifunctional catalysts for
autothermal reforming, where hydrogen is produced and high
local temperatures may exist due to the exothermic nature of
the oxidation reaction. Moreover, the high oxygen mobility may
allow for low carbon deposition rates for both catalysts, pro-
vided that good mixing allows for oxygen spillover from the
perovskite-type oxide. These materials also have been consid-
ered as SOFC anode materials [41,42]. In these studies, the
materials were found to have similar CH4 and CO oxidation
activity compared with the conventional SOFC anode material,
Ni-YSZ.
◦
which gave 50% methanol conversion at 164 C, the temper-
ature at which methanol conversion became significant in the
present study and a similar result as that for precious metal cat-
alysts supported on Al2O3 [38]. La0.6Sr0.4Co0.2Fe0.8O3−δ was
also less active than LaMnO3 for CO oxidation, because 100%
◦
◦
conversion was achieved near 150 C [39] or about 100 C be-
fore La0.6Sr0.4Co0.2Fe0.8O3−δ achieved significant conversion
levels. Surface area-normalized methanol oxidation rates have
been reported for formulations of LnTmO3 where Ln = La, Sm,
◦
and Gd and Tm = Co, Mn, and Fe [18] at 200 C and the rates
determined for La0.6Sr0.4Co0.2Fe0.8O3−δ in the present study,
possibly caused by Sr doping, are two orders of magnitude
greater. At this temperature, 4% of the methanol was converted,
2
which led to a rate of 1.1 mL methanol/(m min). For future
4
.2. Trends with oxygen storage capacity and surface basicity
comparisons, the reaction results have been converted in Ta-
ble 5 into surface area-normalized reaction rates and turnover
frequencies using the total surface site density.
Several possible reasons are proposed to explain why the Co-
and Fe-based materials in this study have lower catalytic activ-
ity than those referenced earlier in this section. First, Fe-based
perovskite-type oxides have been shown to have lower oxida-
tion activity than Co- or Mn-based materials [2]. This trade-off
is necessary to balance catalytic activity with anionic conduc-
For the three samples of the form La0.6Sr0.4CoyFe1−yO3−δ
studied in this work, the methanol and oxygen reactivity in the
◦
temperature range of 300–400 C decreased in the following
order: y = 0.2 > y = 0.3 > y = 0.1. Because this trend is non-
linear with respect to Co content, it is desirable to correlate the
interesting behavior with bulk and surface properties. Oxidation
activity for methane [43,44] and propane [22] has been linked

210
J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
to oxygen storage capacity. However, the surface composition
and basicity also may be influential in controlling activity.
As shown in our earlier study [25], the anion lattice is able
to store several different types of oxygen. The oxygen desorbed
at lower temperatures or α-oxygen is the oxygen that is related
to adsorbed or near-surface oxygen. The amount of α-oxygen
was inversely related to the methanol oxidation activity and de-
creased in the following order: y = 0.3 > y = 0.1 > y = 0.2.
High-temperature oxygen desorption or β-oxygen is associ-
ated with the transition metals on the B site. The amount of
β-oxygen desorption was found to be in agreement with the
trend in reaction activity, thus confirming the association be-
tween oxygen storage capacity and oxidation activity. Such
an explanation indicates an interfacial mechanism in which
the high oxygen mobility supplies available oxygen to oxidize
methanol rather than less mobile adsorbed oxygen species; that
is, oxygen is reduced and placed into the lattice at a faster rate
than it can directly react with surface methoxy species.
That the Lewis sites were relatively constant was not sur-
prising, because the samples contained the same amount of Sr.
The addition of Sr (+2) into the La (+3) matrix is known to
cause charge imbalances (either oxidation of the B-site tran-
sition metal or oxygen vacancies) compared with a pure La
material. These samples have been determined to maintain sto-
ichiometry with respect to oxygen under ambient conditions
despite the change in Co content [25].
Because methanol is a slightly acidic molecule, a certain
degree of basic strength is needed from the oxide surface to
activate methanol (i.e., proton extraction) and form a surface
methoxy species. Because the materials have nearly the same
total site density (i.e., the composition variations change the
unit cell by only a small amount), the greater amount of ba-
sic site density implies that more sites are above the minimum
basicity threshold for methanol activation. Thus, the basic site
density can be used as a benchmark for basic strength of the
surface and is inversely related to methanol and oxygen con-
versions. Because highly basic sites should be better at proton
extraction from methanol, the fact that the results did not fol-
low this trend points to a mechanism in which oxygen mobility
is more important than methanol activation.
The trends in the density and nature of surface sites have
been correlated to surface concentrations measured by XPS.
The amount of basic sites was determined to be a strong func-
tion of Co content, indicating that B-site composition exerted
a greater influence on the surface properties. Co at the surface
mirrored the reaction activity and was inversely related to the
surface basicity. Thus can be explained by the fact that Co is
generally more difficult to oxidize than Fe. Thus, the increased
surface Co may be linked to more surface oxygen vacancies un-
der reaction conditions that allow for more rapid incorporation
of oxygen into the lattice.
ferred from the site density. Surface diffusion rates are difficult
to measure, and thus a gap exists in the literature on this topic.
As the surface basicity increases, the surface oxygen species
are perhaps more stable and less likely to move freely, whereas
weakly basic sites allow for high surface diffusion rates.
4.3. Influence of electronic structure transition
As mentioned previously [25], an electronic structure tran-
sition is expected to occur in the range of Co levels studied
in this work. Pure Fe perovskite-type materials are localized
electron conductors, whereas pure Co material exhibit delo-
calized electronic conduction [12]. The delocalized electronic
structure has been shown to dominate at Fe levels <60% of the
B site [45], so the transition occurs between the formulations of
La0.6Sr0.4Co0.4Fe0.6O3−δ and La0.6Sr0.4FeO3−δ. It should be
noted that the compositions used in this study fall within this
range. The fact that the properties evaluated in the present work
(e.g., methanol oxidation activity, reducibility, surface basic-
ity, oxygen storage capacity) did not exhibit trends proportional
to the compositional changes (i.e., a maximum was observed)
suggests that changes in the electronic structure might play an
important role.
Previous studies on Co- and Fe-based perovskite-type ox-
ides have proposed Co charge disproportionation, ionic fac-
tors, and a preferential charge compensation of Fe over Co
[46,47]. Moreover, the nonlinear behavior of Co-rich formu-
lations (LaCo1−yFeyO3−δ where y = 0.1) has been explained
by the electronic structure complexity caused by low Fe con-
tent [19]. Merino et al. explained that compensation occurred
with the Co3 (high-spin state)–Co (low-spin state) pair
+
III
4+
at low Fe content, whereas Fe compensation dominated at
higher Fe content. The electronic structures of (La,Sr)CoO3−δ
and (La,Sr)FeO3−δ have been thoroughly studied by experi-
mental and theoretical means [48–58]. As reported by those
studies, a transition from Co (low-spin state) to Co (high-
spin state) was thermally induced for pure La-based materials,
and more complicated behavior was seen when Sr doping was
involved. Moreover, these studies observed that charge dispro-
III
3+
portionation between Fe3 and Fe can occur under certain
compositions and conditions. Because no studies have been
performed at La:Sr = 3:2 and an A-site charge imbalance has
been shown to be instrumental to an electronic structure change,
pinpointing the exact cause of the observed behavior is diffi-
cult [46,47]. But because the nonlinear behavior occurred at
both ends of the composition range (LaCo0.9Fe0.1O3−δ [19]
and La0.6Sr0.4Co0.1Fe0.9O3−δ [25] and the present study), it is
plausible that electronic structure complexity arose due to the
unusual charge balance behavior caused by slight B-site com-
positional changes.
+
5+
For ideal perovskite-type materials, the distance between B-
site cations is approximately 4 Å regardless of composition and
environment. Using this value, a site existed every 3–5 unit
cells. Consequently, there is the possibility of further modify-
ing the surface so that the active sites approach one per unit
cell. Moreover, the relative surface diffusion mobility can be in-
5. Conclusion
The surface properties of La0.6Sr0.4CoyFe1−yO3−δ with y =
0.1, 0.2, and 0.3 were examined using methanol as a probe
molecule, by XPS, and by methanol oxidation activity. The
findings demonstrate that methanol is an appropriate probe

J.N. Kuhn, U.S. Ozkan / Journal of Catalysis 253 (2008) 200–211
211
molecule for measuring surface site type and density in mixed
perovskite-type oxides, expanding the use of this approach from
supported and bulk metal oxides to more complex oxide struc-
tures. The different nature of these sites is likely to have im-
portant implications, leading to different activities for different
reactions.
The nonlinear trend in catalytic activity with respect to Co
content was found to agree with several measured parame-
ters, such as strength of basic surface sites, reducibility, oxygen
storage capacity, and B-site composition at the surface. These
trends were linked to a possible electronic structure transition,
which is supported by data in the literature. These findings may
have important implications in pinpointing this transition in the
composition range and establishing linkages to the chemical
properties. Although the activity measured in this study was
lower than reported for other perovskite-type oxides for sim-
ilar reactions, these materials may have catalytic applications
where their robustness is used.
[19] N.A. Merino, B.P. Barbero, P. Ruiz, L.E. Cadus, J. Catal. 240 (2006) 245.
[20] T. Seiyama, N. Yamazoe, K. Eguchi, Ind. Eng. Chem. Prod. Res. Dev. 24
(1985) 19.
[21] T. Shimizu, Appl. Catal. 28 (1986) 81.
[22] T. Nitadori, M. Misono, J. Catal. 93 (1985) 459.
[23] N.A. Merino, B.P. Barbero, C. Cellier, J.A. Gamboa, L.E. Cadus, Catal.
Lett. 113 (2007) 130.
[24] Y. Teraoka, H.-M. Zhang, N. Yamazoe, Chem. Lett. (1985) 1367.
[25] J.N. Kuhn, U.S. Ozkan, Catal. Lett. (2007), submitted for publication.
[26] L. Bedel, A.-C. Roger, C. Estournes, A. Kiennemann, Catal. Today 85
(2003) 207.
[27] L. Bedel, A.-C. Roger, J.-L. Rehspringer, Y. Zimmerman, A. Kiennemann,
J. Catal. 235 (2005) 279.
[28] J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of X-Ray
Photoelectron Spectroscopy, second ed., Perkin–Elmer, Eden Prairie, MN,
1992.
[29] L.J. Burcham, L.E. Briand, I.E. Wachs, Langmuir 17 (2001) 6164.
[30] L.J. Burcham, L.E. Briand, I.E. Wachs, Langmuir 17 (2001) 6175.
[31] M. Machkova, N. Brashkova, P. Ivanov, J.B. Carda, V. Kozhukharov,
Appl. Surf. Sci. 119 (1997) 127.
[
[
[
32] N. Yamazoe, Y. Teraoka, T. Seiyama, Chem. Lett. (1981) 1767.
33] J.L.G. Fierro, Catal. Lett. 22 (1993) 67.
34] I. Rodriguez-Ramos, A. Guerrero-Ruiz, M.L. Rojas, J.L.G. Fierro, Appl.
Catal. 68 (1991) 217.
Acknowledgments
[
[
[
35] H.R. Aghabozorg, B.H. Sakakini, A.J. Roberts, J.C. Vickerman, W.R.
Flavell, Catal. Lett. 39 (1996) 97.
Financial support was provided by the Ohio Coal Develop-
ment Office and the Ohio Department of Development through
the Wright Center of Innovation. Support for acquisition of the
XPS system was provided by the National Science Foundation
under NSF-DMR grant 0114098.
36] J. Choisnet, N. Abadzhieva, P. Stefanov, D. Klissurski, J.M. Bassat,
V. Rives, L. Minchev, J. Chem. Soc. Faraday Trans. 90 (1994) 1987.
37] K.S. De, M.R. Balasubramanian, J. Catal. 81 (1983) 482.
[38] W. Wang, H. Zhang, G. Lin, Z. Xiong, Appl. Catal. B Environ. 24 (2000)
19.
2
[
[
39] K.S. Song, H.X. Cui, S.D. Kim, S.K. Kang, Catal. Today 47 (1999) 155.
40] S. Kaliaguine, A. Van Neste, V. Szabo, J.E. Gallot, M. Bassir, R. Muzy-
chuk, Appl. Catal. A Gen. 209 (2001) 345.
Supporting material
[41] S.P. Scott, D. Mantzavinos, A. Hartley, M. Sahibzada, I.S. Metcalfe, Solid
State Ionics 152–153 (2002) 777.
The online version of this article contains additional support-
ing material.
[
42] A. Hartley, M. Sahibzada, M. Weston, I.S. Metcalfe, D. Mantzavinos,
Catal. Today 55 (2000) 197.
Please visit DOI: 10.1016/j.jcat.2007.10.005.
[
43] D. Ferri, L. Forni, Appl. Catal. A Gen. 16 (1998) 119.
44] S. Royer, H. Alamdari, D. Duprez, S. Kaliaguine, Appl. Catal. B Envi-
ron. 58 (2005) 272.
[
References
[
[
45] M.H.R. Lankhorst, J.E. ten Elshof, J. Solid State Chem. 130 (1997) 302.
46] L.W. Tai, M.M. Nasrallah, H.U. Anderson, D.M. Sparlin, S.R. Sehlin,
Solid State Ionics 76 (1995) 259.
[
[
[
1] L.G. Tejuca, J.L.G. Fierro, J.M. Tascon, Adv. Catal. 36 (1989) 237.
2] M.A. Pena, J.L.G. Fierro, Chem. Rev. 101 (2001) 1981.
3] L.G. Tejuca, J.L.G. Fierro (Eds.), Properties and Applications of Perovsk-
ite-Type Oxides, Marcel Dekker, New York, 1993.
[47] L.W. Tai, M.M. Nasrallah, H.U. Anderson, D.M. Sparlin, S.R. Sehlin,
Solid State Ionics 76 (1995) 273.
[
4] H.J.M. Bouwmeester, H. Kruidhof, A.J. Burggraaf, Solid State Ionics 72
[48] P.M. Raccah, J.B. Goodenough, J. Appl. Phys. 39 (1968) 1209.
[49] P.M. Raccah, J.B. Goodenough, Phys. Rev. 155 (1967) 932.
[50] S. Mathi Jaya, R. Jagadish, R.S. Rao, R. Asokamani, Phys. Rev. B 43
(1991) 13274.
(1994) 185.
[
5] J.F. Vente, S. McIntosh, W.G. Haije, H.J.M. Bouwmeester, J. Electrochem.
Soc. 10 (2006) 581.
[6] Y. Zeng, Y.S. Lin, S.L. Swartz, J. Membrane Sci. 150 (1998) 87.
[7] F.T. Akin, Y.S. Lin, J. Membrane Sci. 209 (2002) 457.
[8] R.M. Ormerod, Chem. Soc. Rev. 32 (2003) 17.
[9] B.C.H. Steele, Solid State Ionics 94 (1997) 239.
[51] V.G. Bhide, D.S. Rajoria, C.N.R. Rao, G.R. Rao, V.G. Jadhao, Phys.
Rev. B 12 (1975) 2832.
[52] P. Ganguly, P.S.A. Kumar, P.N. Santhosh, I.S. Mulla, J. Phys. Condens.
Matter 6 (1994) 533.
[
[
[
[
10] S.B. Adler, Solid State Ionics 111 (1998) 125.
11] S.B. Adler, Solid State Ionics 135 (2000) 603.
12] S.B. Adler, Chem. Rev. 104 (2004) 4791.
13] S.B. Adler, J.A. Lane, B.C.H. Steele, J. Electrochem. Soc. 143 (1996)
[53] M.A. Senaris-Rodriguez, J.B. Goodenough, J. Solid State Chem. 118
(1995) 323.
[54] H. Takahashi, F. Munakata, M. Yamanaka, Phys. Rev. B 57 (1998) 15211.
[55] J. Mizusaki, J. Tabuchi, T. Matsuura, H. Yamauchi, K. Fueki, J. Elec-
trochem. Soc. 136 (1989) 2082.
3554.
[
[
[
[
[
14] V.C. Corberan, L.G. Tejuca, A.T. Bell, J. Mater. Sci. 24 (1989) 4437.
15] M.B. Badlani, I.E. Wachs, Catal. Lett. 75 (2001) 137.
16] G. Busca, Catal. Today 27 (1996) 457.
17] J.C. Lavalley, Catal. Today 27 (1996) 377.
18] T. Nitadori, T. Ichiki, M. Misono, Bull. Chem. Soc. Jpn. 61 (1988) 621.
[56] A. Chainani, M. Mathew, D.D. Sarma, Phys. Rev. B 48 (1993) 14818.
[57] T. Ishikawa, S.K. Park, T. Katsufuji, T. Arima, Y. Tokura, Phys. Rev. B 58
(1998) R13326.
[58] J. Mizusaki, T. Sasamoto, W.R. Cannon, H.K. Bowen, J. Am. Ceram. Soc.
66 (1983) 247.