Rhodium-Catalyzed Methanation and Methane Steam Reforming Reactions on Rhodium–Perovskite Systems: Metal–Support Interaction

Article abstract of DOI:10.1002/cctc.201600262

Metal–support interaction in rhodium–perovskite systems was studied using LSF (La_0.6Sr_0.4FeO_3?δ) and STF (SrTi_0.7Fe_0.3O_3?δ) supports to disentangle different manifestations of strong or reactive metal–support interaction. Electron microscopy and catalytic characterization in methane steam reforming/CO₂ methanation reveal that reduction in hydrogen at 673 K and 873 K causes different extents of Fe exsolution. Depending on the perovskite reducibility, Fe–Rh alloy particles are observed. No signs of strong metal–support interaction (i.e., encapsulation of metal particles) by reduced oxide species were observed. As re-oxidation in oxygen at 873 K did not fully restore the initial structures, the interaction between Rh and the perovskites manifests itself in irreversible alloy formation. Catalytic effects are the suppression of methane reactivity with increasing prereduction temperature. The results show the limits of the strong metal–support interaction concept in complex metal–oxide systems.

Full text of DOI:10.1002/cctc.201600262

DOI: 10.1002/cctc.201600262
Full Papers
Rhodium-Catalyzed Methanation and Methane Steam
Reforming Reactions on Rhodium–Perovskite Systems:
Metal–Support Interaction
[a]
[a]
[a]
[b]
[b]
Ramona Thalinger, Thomas Gçtsch, Chen Zhuo, Walid Hetaba, Wolfgang Wallisch,
[b]
[c]
[a]
[a]
Michael Stçger-Pollach, Daniela Schmidmair, Bernhard Klçtzer, and Simon Penner*
Metal–support interaction in rhodium–perovskite systems was
studied using LSF (La Sr FeO ) and STF (SrTi Fe O ) sup-
of metal particles) by reduced oxide species were observed. As
re-oxidation in oxygen at 873 K did not fully restore the initial
structures, the interaction between Rh and the perovskites
manifests itself in irreversible alloy formation. Catalytic effects
are the suppression of methane reactivity with increasing pre-
reduction temperature. The results show the limits of the
strong metal–support interaction concept in complex metal–
oxide systems.
0
.6 0.4
3Àd
0.7
0.3 3Àd
ports to disentangle different manifestations of strong or reac-
tive metal–support interaction. Electron microscopy and cata-
lytic characterization in methane steam reforming/CO metha-
2
nation reveal that reduction in hydrogen at 673 K and 873 K
causes different extents of Fe exsolution. Depending on the
perovskite reducibility, Fe–Rh alloy particles are observed. No
signs of strong metal–support interaction (i.e., encapsulation
Introduction
Metal–support interaction has always been a key criterion for
the understanding and tailoring of catalytically active and se-
“strong” metal–support interaction has been almost exclusively
focused on the interaction of metals with binary, that is,
“simple”, oxides. Structural manifestations in this special case
are usually the formation of substoichiometric oxide phases. If
the mobility of these phases is high enough, they form a thin
covering layer around the metal particles, thus decreasing the
amount of catalytically active metal surface. Diffusion of more
strongly reduced or metallic species into the metal lattice
might then eventually lead to the above-discussed case of in-
termixing and formation of intermetallic compound/alloy
[
1–4]
lective species.
Initially neglected in metal–oxide systems,
viewing the oxide basically as a mediator to disperse the cata-
lytically active metal, increased interest in this effect was first
shown in the early 1980s with the classification of metal–sup-
[5]
port interaction effects and especially with the introduction
[
4,6]
of the term “strong metal–support interaction” (SMSI).
The
latter, initially coined by Tauster et al. and strictly reserved for
[
2,4,6]
metals on reducible oxides (such as TiO2 or CeO2),
was
[
11]
shown to induce drastic structural and electronic changes to
the catalytic entity upon high-temperature reduction, with
phases.
Recent focus in catalysis is increasingly put on the use of
more complex catalytic systems, which offer a higher possibili-
[
6–10]
equally notable changes in the catalytic properties.
Its
most extreme form, for which the new term “reactive metal–
ty of individually tailoring their catalytic properties by dedicat-
[11]
[12–16]
support interaction” (RMSI) has been introduced recently, is
usually connected to the structurally irreversible formation of
intermetallic compounds or alloy phases upon treatment in hy-
drogen at elevated temperatures. So far, the concept of
ed pretreatments.
Such systems include mixed oxide sys-
[
14,15]
[16]
tems
or (doped) ternary oxide compounds.
Especially
the latter offer a tremendous variability in structures and
[
12–16]
chemical composition.
One class of materials for which
this advantage is particularly obvious are perovskite systems.
Their use as catalysts in a variety of reactions, including hydro-
genation, reforming or water-gas shift reactions, has been re-
[
a] R. Thalinger, T. Gçtsch, C. Zhuo, Prof. Dr. B. Klçtzer, Dr. S. Penner
Institute of Physical Chemistry
University of Innsbruck
Innrain 80–82, A-6020 Innsbruck (Austria)
E-mail: simon.penner@uibk.ac.at
ported, but unfortunately does not come without strings at-
[12,13,17]
tached.
This is mainly connected to their structural and
electronic complexity, which is multiplied if the simple perov-
skite systems are converted to metal–oxide systems for further
[
b] W. Hetaba, W. Wallisch, Dr. M. Stçger-Pollach
University Service Center for Transmission Electron Microscopy (USTEM)
Vienna University of Technology
[
18,19]
improvement of their catalytic properties.
The so-created
Wiedner Hauptstraße 8–10, A-1040 Vienna (Austria)
metal–oxide interface is then particularly variable and thus, no-
toriously difficult to control. One of the most crucial problems,
significantly affecting their catalytic use, is thereby related to
the pronounced metal (or oxide) exsolution phenomena as
well as, consequently, alloy formation occurring during reduc-
[
c] D. Schmidmair
Institute of Mineralogy and Petrography
University of Innsbruck
Innrain 52d, A-6020 Innsbruck (Austria)
Supporting information for this article can be found under http://
dx.doi.org/10.1002/cctc.201600262.
[
20–23]
tive pretreatments.
This is a direct consequence of the re-
ChemCatChem 2016, 8, 2057 – 2067
2057
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ducibility of perovskite systems and obscures the effects of
classic” metal–support interaction. Exsolution does not occur
intermetallic compounds, but a number of distinct Fe–Rh alloy
[
26]
“
phases with varying composition. It is particularly interesting
whether such Fe–Rh alloy phases reproducibly arise through
perovskite reduction and Fe exsolution, which potentially give
rise to outstanding catalytic properties.
on simpler binary oxides, but is only observed on more com-
plex oxide systems with easily reducible components. Eventu-
ally, disentangling these effects is a major effort, as the two
structural effects are obviously prone to overlap and occur si-
multaneously at potentially comparable temperatures. To the
best of our knowledge, no studies on this particular subject
have been provided, despite its catalytic importance.
Regarding the catalytic investigations, focus is put on the
methane steam reforming and CO methanation reactions, as
2
only the combination of a catalytically active metal with these
perovskites (LSF or STF) shows considerable activity and selec-
[
12,13,17,18]
To close this knowledge gap, we provide a thorough elec-
tron microscopy and catalytic study on the archetypical metal–
tivity in the chosen reactions.
The methanation reac-
tion from CO (or CO ) and H to CH or its reverse reaction, the
2
2
4
perovskite systems Rh–La Sr FeO (lanthanum strontium
3Àd
methane steam reforming reaction, respectively, has been
0
.6 0.4
[
27–39]
ferrite, LSF) and Rh–SrTi Fe O (strontium titanium ferrite,
studied over a number of perovskite materials.
The CO2
0
.7
0.3 3Àd
STF). The choice of these particular systems is fuelled by the
outstanding catalytic properties of Rh and the two perovskite
systems, which in fact serve as dedicated model systems for an
easily reducible oxide (LSF) and a more stable oxide with
rather suppressed reducibility (STF). This will allow for direct
transfer of the ideas of (S)MSI already known for decades to
more complex and SOFC-anode-relevant metal–oxide systems.
In particular, combinations of a noble metal and a mixed
ionic–electronic conductor (MIEC) are very interesting candi-
dates for such applications as they provide strongly improved
methane reforming and electro-oxidation properties. Thus, the
so-introduced combined oxide anion and electronic conduc-
tion might give rise to applications as “direct methane SOFC
methanation reaction in particular has attracted ongoing inter-
est regarding energy-efficient storage solutions for renewable
[
40]
electricity. Thanks to the number of active catalysts (includ-
ing transition metals such as Rh, Ru, Co, or Ni), the available lit-
erature on the methanation reaction is considera-
[
27–29,31,36–38,40–44]
ble.
This also holds for the methane steam re-
forming reaction, which is also a well-established method for
[
30]
efficient hydrogen production. The reactivity of both perov-
skite systems is also compared to a Rh/Al O reference catalyst,
2
3
in which metal–support interaction effects are strongly sup-
pressed.
anodes”. However, such noble-metal–MIEC combinations, espe- Results and Discussion
cially if a noble-metal-doped perovskite is used, inevitably
Methane reforming reactivity
suffer from the strongly reducing conditions of SOFC usage,
and, in turn, are effectively prone to SMSI- or RMSI-related de-
activation effects. Thus, the outlined studies not only offer
a novel approach to the identification of special anode deacti-
vation mechanisms, but the presented studies also allow to
identify the experimental conditions to separate the effects of
Methane reforming on Rh–STF and Rh–LSF is shown in
Figure 1. The signals are presented as turnover frequency per
Rh site and second. Generally, the reactivity is strongly depen-
dent on the reducibility and the adsorption behavior of the
[
17]
pure perovskite under question.
Rh–LSF shows, besides
“
classic” metal–support interaction and effects arising from the
methane reforming reactivity, hydrogen desorption starting
below 573 K as a side effect of the reductive pretreatment
(673 K, H 1 bar 1 h). To confirm this phenomenon and quanti-
structural peculiarities of the chosen perovskite systems, which
is (Fe) metal exsolution. This will directly lead to conclusions
about the similarities and differences of the more complex
oxide structures compared to the simpler ones with respect to
metal–support interaction. Studies on Fe-containing perov-
2
,
fy this effect, a blank test (without any reactive gas) was con-
ducted after prereduction at the same conditions and the H2
yield obtained thereafter by desorption subtracted from the
one of the reaction profile. With respect to the onset tempera-
skites are especially worthwhile, because in particular Fe O4
3
used as a catalyst support for Pt particles is well-known to be
prone to enter a state of (S)MSI upon reduction in hydro-
ture of the reaction, an offset regarding CH consumption is
4
clearly visible. This is attributed to the ability of LSF to quench
oxygen vacancies resulting from the reductive pretreatment
with water. Another blank test with water was performed to
corroborate this thesis. Starting at approximately 600 K, a H2
signal could be detected (Supporting Information, Figure S1).
The following temperature-dependent reaction steps could
hence be identified, with increasing reaction temperature:
1) H2 desorption attributed to prereduction treatment (>
500 K) and water quenching of the oxygen vacancies with H2
as byproduct (>540 K), 2) Methane reactivity (>650 K), and
[
24,25]
gen.
To revisit the appearance and formation of alloy
phases and/or intermetallic compounds in catalytic pretreat-
ments or measurements, we note that this represents a quite
general phenomenon, but despite the catalytic importance, re-
ports of Rh-containing alloys or even intermetallic compounds
are relatively scarce with respect to other noble metals. To
[11a]
[11b,c]
[11d,e]
date, only on four oxides, namely CeO ,
SiO2,
intermetallic Rh–X (X=Si, Ce, V, Ti) compounds
or alloys have been observed through reactive metal–support
V O ,
2
2 3
[
11f,g]
and TiO2,
[
11]
interaction. As such, the high reducibility of iron oxides and
Fe-containing perovskites, the latter through Fe exsolution,
might give rise to the formation of distinct Fe–Rh alloy phases
via reductive activation. Having said that, the current Fe–Rh
binary alloy phase diagram does not report well-defined Fe–Rh
3) CO/CO desorption (>800 K).
2
Methane consumption therefore starts at approximately
650 K and accelerates up to 850 K. Methane conversions of
50% are obtained after a 20 min isothermal period at 873 K
(not shown in Figure 1). To explain the clear temperature offset
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2058
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
CO becomes the major product of methane steam reforming,
which is in line with the equilibrium CO yield at higher temper-
atures.
A generally similar behavior can be observed on the refer-
ence catalyst, with the only difference of a 120 K lower onset
temperature (Figure S1C). This offset is caused by metal–per-
ovskite interaction, and is mainly attributed to the electronic
and structural modifications of the Rh particles caused by the
different extent of alloy formation with Fe (see below).
CO methanation reaction
2
CO methanation as the reversal of methane steam reforming
2
was also studied as a function of prereduction temperature. In
Figure 2 the normalized reaction rates on Rh–LSF after pre-
treatment in H at 673 K are compared with those after pre-
2
treatment at 873 K. The reaction profile after reduction at
6
73 K (Figure 2A) is characterized by an inverse water gas shift
reactivity up to approximately 550 K, followed by an increase
of the methane reaction rate up to a maximum rate of 0.1”
À3 À1
À1
1
0
s site at approximately 700 K. The reverse reaction,
methane reforming to CO and CO , sets in above 800 K, and
2
Figure 1. Normalized methane reforming reaction rates on A) Rh–LSF and
B) Rh–STF after preoxidation and prereduction at 673 K. Low-temperature
hydrogen desorption as a side effect of the reductive pretreatment only ap-
both signals as well as that of hydrogen rise accordingly (see
also Figure S2).
pears on Rh–LSF and is subtracted from the H
2
signal. Reaction mixture: par-
After reduction at higher temperatures (Figure 2B), the cata-
lyst is completely deactivated for the methanation reaction
and could not be re-activated by another cycle of oxidation
and reduction, even at 673 K. The inverse water gas shift reac-
tial pressures for H
2
O and CH
4
each 20 mbar, Ar and He 10 mbar added to
À1
1
bar total pressure. Heating rate: 5 Kmin
.
between carbon dioxide/carbon monoxide production and
methane consumption, we might assume that also carbon
(
arising from methane decomposition) is adsorbed and
stored” at the surface, before it is finally released as CO and
“
2
CO above 800 K. As the methane rate again increases between
00 and 850 K, this also indicates an immediate reaction of
8
methane to CO and CO without induction period. The corre-
2
sponding pressure versus temperature plots are shown in Fig-
ure S1.
Hence, on prereduced Rh–LSF, a series of “non-catalytic”
conversions of the reactants alongside pronounced H2 and
carbon storage capacity of the material, is observed. This sets
Rh–LSF apart from Rh–STF, which exhibits a defined reforming
onset temperature and also stoichiometric conversion of the
reactants in the entire temperature region above 700 K. (Fig-
ure 1B) Moreover, Rh–STF does not feature any ability to
quench H -induced vacancies (i.e., no hydrogen desorption
2
prior to reaction is observed). Therefore, the onset temperature
of hydrogen evolution is significantly higher at 680 K and H is
2
exclusively formed from the reforming process. Compared to
Rh–LSF, the normalized reaction rate at comparable reaction
temperatures is much higher. Again, during the isothermal
period at 873 K (20 min), methane conversion reaches approxi-
mately 60%. The reaction profile shows no indication for su-
perimposed noncatalytic processes (hydrogen desorption or
water quenching reactions) as on Rh–LSF. On the contrary, an
abrupt boost of the normalized reaction rates is observed,
with a peak at 750 K. Above 770 K, the selectivity changes and
Figure 2. Normalized CO
after different prereduction temperatures of A) 673 K, B) 873 K (1 h, 1 bar
). Reaction mixture: partial pressures for H 54 mbar and CO 17 mbar for
the prereduction at 673 K, H 60 mbar and CO 17 mbar for the prereduction
2
methanation reaction rates measured on Rh–LSF
H
2
2
2
2
2
at 873 K; Ar and He 10 mbar added to 1 bar total pressure in both mixtures.
À1
Heating rate: 5 Kmin
.
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2059
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tion (i-WGSR), however, appears to be unaffected by the
changed reaction conditions, except its nonstoichiometry
above 700 K. The increased hydrogen consumption originates
possibly from structural changes of the catalyst. The progres-
sive deactivation for methane chemistry with increasing reduc-
tion temperature points to a general absence of free Rh metal
area and increasing Fe–Rh alloy formation (see electron micros-
copy results below). This also explains the generally higher
unalloyed Rh particles on Al O as a direct comparison to the
2
3
Rh–perovskite systems. The same experiments performed on
Rh–STF (Figure 3B) reveal a comparable picture to the Rh/
Al O3 case, but it clearly differs from the Rh–LSF case (Fig-
2
ure 2A). After low-temperature prereduction at 673 K, the reac-
tion profile is qualitatively similar to that of Rh/Al O , but the
2
3
respective temperature for the maximum methanation rate is
shifted to approximately 660 K. The maximum methane steam
reforming rate on this catalyst is then shifted beyond the maxi-
mum temperature of the experiment. By applying harsher re-
ductive pretreatments (873 K) the catalyst is even more deacti-
vated with a lower maximum methane formation rate at ap-
proximately 700 K. (Figure 4A 1st run). A small amount of CO
À3 À1
À1
maximum of the CO formation rate (ꢀ0.3”10
s site at
ꢀ
620 K). In this respect, the i-WGSR proceeds on the perov-
skite support and is potentially also aided by the increasing
iron content of the Rh particles, because iron compounds are
[45–47]
well known to catalyze the water gas shift reaction.
Thus,
obtaining information on the chemical and electronic state of
Fe within the perovskite after reduction is imperative. X-ray
photoelectron spectroscopy results (XPS, see also Figure 11)
demonstrate that the Rh–LSF sample reduced at 873 K consists
II
III
of three iron species, namely, Fe , Fe , and a low amount (1–
0
3
at.%) of segregated metallic Fe . The oxidized LSF perovskite,
II
III
in contrast, reveals a mixture of Fe and Fe , as in magnetite,
which is a commonly used catalyst for the water gas shift reac-
[45–47]
tion in the temperature range of 573–723 K.
The influence of the Rh–perovskite interface is best revealed
by comparative studies on the Rh/Al O reference catalyst (Fig-
2
3
ure 3A). The maximum methane formation rate is observed at
approximately 520 K and CO2 methanation proceeds up to
6
20 K. Above that temperature, the normalized methane rate
reverses and methane is effectively consumed to produce CO,
CO , and H . Thus, the maximum methane reforming rate is
2
2
observed at approximately 780 K. Note that this sample pro-
vides access to the methane steam reforming capabilities of
2
Figure 4. Normalized CO methanation reaction rates after reduction at
873 K. Two runs were performed using the same catalyst with A) 1st and
B) 2nd run. Reaction mixture: H
2
80 mbar and CO
2
20 mbar, Ar and He
À1
10 mbar added to 1 bar total pressure. Heating rate: 5 Kmin
.
from the inverse water-gas shift reaction is also observed. After
another cycle of reduction at 873 K (Figure 4B 2nd run), the
maximum methane rate is even lower, but still observed at the
same temperature. In contrast, the CO rate is significantly
higher than in the first run at 873 K, which points to increased
inverse water gas shift activity. As on Rh–LSF, the reaction pro-
files can be assigned to different extents of Fe–Rh alloy forma-
tion. In fact, while such alloy formation is of course absent for
the reference catalyst, the overall deactivation of Rh–STF is
best seen in Figure S5 highlighting the Arrhenius analysis and
points to alloy-dependent CO methanation reactivity. The cat-
2
alytic behavior appears to be again mostly steered by the per-
ovskite. Compared to the reference catalyst Rh–Al O , the
2
3
onset temperature is almost 150 K higher and also the maxi-
mum conversion is found to be lower (Rh–STF 55%, Rh–Al O3
2
Figure 3. Comparison of the normalized CO
A) reference catalyst Rh–Al (after reduction at 873 K) and B) Rh–STF (after
reduction at 673 K). Reaction mixture: H
2
methanation reactivities on
8
5% conversion, see Figure S3).
2
O
3
2
80 mbar and CO
2
20 mbar, Ar and
À1
He (10 mbar) added to 1 bar total pressure. Heating rate: 5 Kmin
.
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2060
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Catalytic behavior
more often the reaction was repeated, the higher the activa-
tion energy accordingly becomes (see Table 1).
To elucidate the influence of the perovskite and metal–support
interactions on catalytic properties, key reaction parameters of
Rh–STF/LSF were compared to those of the Rh/Al O reference
In general, the structure–activity correlations in Rh–perov-
skites are comparable to the values for the corresponding Ni–
[
18]
perovskite systems. The activation energy for Rh–STF is ap-
2
3
À1
catalyst, and activation energies and turnover frequencies
TOF) in the CO methanation reaction were calculated. For
proximately 10 kJmol higher than for Ni–STF, most likely be-
(
cause of the more pronounced alloy formation already at
lower prereduction temperatures. Nevertheless, “alloyed” Ni–
LSF and Rh–LSF are in good comparison with an activation
2
TOF determination, the partial pressure signals of methane
after reduction at 673 K were at first differentiated by time. For
total TOF calculation, each surface Rh atom was considered to
be an active site. Hence, the so-calculated TOF values repre-
sent lower limits of specific reactivity of Rh centers. Based on
the average particle size after reduction (STF 6.9 nm, LSF
À1
energy of approximately 145 kJmol . For both systems, pro-
nounced deactivation originating from the alloying of the
metal particle with iron exsolved from the support can be thus
deduced. The TOF values are relatively similar to those for the
Ni–perovskite catalysts and also in good agreement with, for
example, Rh/TiO2 samples treated under comparable condi-
6
.7 nm; see the electron microscopy images in Figure S8) and
the knowledge of the Rh loading (10 mol%), one can calculate
the average volume of these particles (approximated as half-
spheres as deduced from the TEM images) and with that, the
amount of accessible Rh surface sites. To check the validity of
À3 À1
[18,50]
tions (ꢀ1–5”10
s
at 473 K).
In addition, the effects of the particle nature of the Rh in the
two Rh–perovskite system and the reference Rh–Al O on their
2
3
this estimation, H adsorption measurements were performed
catalytic behavior have to be discussed. As derived from the
TEM analysis (see Figure S7 and Figure 6 and 7), the initial Rh
particles sizes in the Rh–perovskite systems and the reference
are comparable at approximately 2–4 nm. After reduction at
673 K, Rh particles on STF and LSF show considerable sintering
2
at ambient temperature (Figure S4). Using a Langmuir fit
model, the estimated values from the TEM measurements
could be directly confirmed. The TOF values per active site
(
and time) were thus calculated and are shown in Table 1. The
(
particle sizes ꢀ7 nm), but Rh on Al O is known to remain
2
3
[51]
highly dispersed.
Therefore, pure dispersion differences
Table 1. Summary of the maximum TOF values, particle sizes, catalyst dis-
cannot explain the observed catalytic changes. This is especial-
ly true for reduction at higher temperatures (873 K), at which
considerable additional sintering is absent and the changes in
catalytic behavior are tremendous (Figure 2 and 4).
a 2
persion, and activation energies (E ) for the CO methanation reaction on
Rh–STF, Rh–LSF, and Rh–Al
2
3
O .
Rh–Al
2
O
3
Rh–STF
(H673)
Rh–LSF
(H673)
(H873)
1
8
18
18
Sites (H
TOF (max) [s
2
adsorption)
6”10
2.2”10
1.02”10
(665 K)
129
3.4”10
1.09”10
(712 K)
144
Structural characterization
À1
À3
À3
À4
]
1.60”10
534 K)
(
To focus on the structural stability of the Rh–perovskite cata-
lyst entity with respect to metal–support interaction, XRD
measurements were conducted after different pretreatments
and after catalytic runs. An overview is given in Figure 5. After
the impregnation process and the reductive treatments the
À1
E
a
[kJmol
]
101
2.3
40
Particle size (TEM) [nm]
Dispersion [%]
6.9
14
6.7
14
Rh–STF
H673
129
H873 I
137
H873 II
146
À1
E
a
[kJmol
]
[
17]
structures of both STF and LSF appear to be unchanged. In
both cases, however, no sign of Rh or Rh O can be detected
2
3
maximum TOF was reached for the reference catalyst already
at 540 K, whereas the maximum TOF for Rh–STF is observed at
in the as-grown state (red). This is assigned to a particle size
effect. Using TEM overview images it can be shown that the
average polycrystalline Rh particle size of the as-grown state is
approximately 2 nm, which is too small to be clearly detectable
by XRD (see Figure S7). After reduction of Rh–STF at 873 K
(Figure 5, blue), Rh reflections (fcc, pattern 00-005-0685, PDF
1
25 K higher temperatures. For Rh–LSF, the TOF is approxi-
mately one magnitude lower at comparable temperatures than
for the other catalysts. To estimate the activation energies for
this reaction, the initial reaction rates were fitted by the Arrhe-
nius plot for all three catalysts. For better comparability, the
[
52,53]
4+ ICDD database
) arise. It is well known, that Rh shows
8
À1
preexponential factor was held constant at a value of 10 s
see Figure S5 and S6). The apparent activation energy of the
strong sintering behavior at elevated temperatures and espe-
[
54]
(
cially under reducing conditions. This is clearly visible in the
TEM images, confirming the presence of larger Rh crystallites.
After the first cycle of the catalytic measurement, the Rh peaks
are still present in the diffractogram. To test whether the deac-
tivation of the catalysts upon high temperature reduction is re-
À1
reference catalyst (ꢀ100 kJmol ) is in good comparison with
literature data and almost 30 kJmol lower than that of Rh–
[
48]
À1
À1
À1
STF (ꢀ130 kJmol ), which is approximately 15 kJmol lower
À1
than the value of Rh–LSF (ꢀ145 kJmol ). The reason for this
difference can be directly tracked down to the lower methana-
tion reactivity of exsolved iron and subsequently, the Fe–Rh
versible, a reoxidation treatment (1 bar O , 873 K, 1 h) was per-
2
formed. After this procedure, the diffractogram shows (besides
some remaining metallic Rh), peaks that are assigned to Rh O
[
49]
alloy, as compared to pristine Rh. The comparison of Rh–STF
samples after different prereduction treatments confirms this
assumption. The higher the reduction temperature and the
2
3
[
55]
(trigonal, pattern 00-041-0541, PDF 4+ ICDD database ). STF
and LSF structures remain unaltered, except for a change of
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2061
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 5. Summary of XRD measurements on Rh–LSF (left) and Rh–STF
(
(
middle and right). The as-prepared states, compared to after reduction
blue) and after catalysis (green), as well as the reoxidized sample (black),
are shown. The peak shift is denoted by the dashed lines and the arrows.
Figure 6. A) TEM bright field image of Rh–STF and B) high-angle annular
dark-field (HAADF) image of Rh–LSF, of the samples after preparation, with
according EDX measurements. EDX spectra in (A) were measured at the
color-coded spots, those in Panel B measured along the denoted line profile,
in which the color code highlights the location of the measured spectra. Fe
L (*), Cu L (+) peaks are additionally indicated.
the lattice parameter resulting from the formation and
quenching of oxygen vacancies. The associated peak shifts
were detected on the pure perovskite and have been dis-
[
17]
cussed elsewhere. The studies have thus been extended to
those after different reduction temperatures and numbers of
catalytic cycles (see Figure S9). In short, no differences could
be found in the diffractograms neither for LSF nor for STF, al-
though after catalysis, different behaviors were pointed out
(
see Figure 2 and 3). Apparently, XRD is unselective to detect
the catalysis-steered chemical changes of the Rh particles.
For a more detailed characterization, a combination of TEM
and energy-dispersive X-ray spectroscopy (EDX) characteriza-
tion of the calcined states was performed (Figure 6). On both
perovskites, small (ꢀ2 nm diameter on average, see Figure S8)
Rh O particles were characterized. For Rh–STF (Figure 6A), dif-
2
3
ferent representative sites were measured, including a free-
standing rhodium oxide particle (dark blue circle), a particle on
perovskite (green circle), and on pure STF (orange circle). For
the free-standing particle, no alloy formation with iron was
measurable. On LSF (Figure 6B), an EDX line scan across
a Rh O particle attached to the LSF grain was performed. The
2
3
Fe signal strictly follows the other perovskite signals, indicating
that no iron contamination in the Rh particles is present.
After reduction at 673 K, the situation on Rh–LSF changes
considerably (Figure 7). A core–shell-like structure of the metal
particle is clearly visible in the high-resolution (HR) TEM image
(
Figure 7A2). For the particle core, the lattice spacing fits to
the theoretical value of Rh (111) (dtheo =0.2196 nm), whereas
the shell (marked by the arrow) appears to be amorphous. An
EDX line scan was performed across this particle (Figure 7A1
and A3). On the outer parts, only the iron signal is detected.
This indicates that during the reductive pretreatment, some
metallic iron is exsolved from the perovskite lattice and subse-
quently diffusing into the Rh particle. The shell is likely formed
by postoxidation to iron oxide at the surface, once the catalyst
was removed from the reactor and transferred in air to the
electron microscope. Note that despite the fact that the parti-
cles appear in a core–shell structure, no decoration effects
from the pre-reductive treatment are evident. The Fe–Rh
Figure 7. Structural states of A) Rh–LSF and B) Rh–STF after prereduction at
673 K. A1) EDX line scans along the line indicated in the HAADF image in
A3. A2) HRTEM image of the same particle with pronounced core–shell
structure with marked lattice fringes of Rh (green) and LSF (red). For Rh–STF
the signals of the Fe K, Rh L, Sr L and Ti L edges were measured along an
EDX line scan (inset in Panel B) over three Rh particles (denoted by black-
ened red arrows).
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2062
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
binary alloy phase diagram is well known. It reveals complete
solid solubility of Fe in Rh at temperatures up to 1900 K at Fe/
Rh stoichiometries of approximately 1:1. No Fe–Rh intermetal-
lic compounds are reported, but a number of distinct alloy
phases with pronounced crystal parameters. At high Rh con-
centrations, this comprises a g-(Fe,Rh) alloy phase and a CsCl-
type phase at a Fe:Rh ratio of approximately 0.4, containing
Higher reduction temperatures also had a detrimental
impact on the reactivity of Rh–STF, distinctly suppressing the
normalized TOF in CO methanation (see Figure 4). This can be
2
directly confirmed by the microscopic study. HRTEM analysis
and EELS measurements (Figure 9) strongly suggest that under
the chosen conditions, Fe becomes mobile also in STF, ex-
solved and incorporated in the Rh particle to subsequently
form a Fe–Rh alloy. In the high-resolution images, lattice fring-
es of Fe O can be identified ((110) and (104) with the theo-
[26,56]
approximately 55 at.% Rh.
This ratio was also confirmed
by EDX analysis, performed as outlined in the Experimental
Section. Assignment of this stoichiometry to a defined alloy
phase is difficult because the individual alloy particles fre-
quently appear chemically heterogeneous, see below. The
alloy formation during the pretreatment causes direct changes
in the electronic structure of Rh and detrimentally affects the
methane chemistry. The catalyst is thus partially deactivated. In
contrast, for Rh–STF, the pretreatment at 673 K does not
induce such strong structural alterations. In Figure 7B the EDX
line scan over three metal particles attached to the perovskite
is presented. Both the Sr (red) and Ti (black) signals show the
same trend. The Rh intensity is independent from the perov-
skite signals. The Fe signal (blue) mainly follows those of Sr/Ti,
but especially for the middle particle, it appears that the iron
signal increases parallel to that of Rh, indicating Fe–Rh alloy
formation.
2
3
[
57]
retical spacings of 0.252 and 0.271 nm ), likely induced by
contact to air.
If the reduction temperature is increased to 873 K, the effect
of iron dissolution in the Rh particle becomes much more pro-
nounced. In Figure 8, the quantitative amounts of Rh, Fe (rang-
ing from 60 at.% to 1 at.%, with 35 at.% in average), and La
within a free-standing particle are followed along an electron
energy loss spectrometry (EELS) line scan (black arrow). At low
distances the amounts of Fe and Rh in the particle are compa-
rable (i.e., in a 1:1 stoichiometry), but at higher distances they
vary considerably. The Fe concentration appears lower in the
core of the particle. This, nevertheless, gives a tentative explan-
ation as to why the methanation activity is completely sup-
pressed and the i-WGSR more pronounced on this catalyst
after high-temperature reduction. The LSF lattice, represented
by its (012) fringes in the HRTEM image (Figure 8), appears un-
changed.
Figure 9. HAADF image of Rh–STF after reduction at 873 K (bottom left) and
the EELS line scan along the white line (top left), HRTEM image of a single
alloyed Fe–Rh particle (top right) with the lattice fringes of Rh (red) and
Fe O (green, blue) as marked in the fast Fourier transfer image derived
2 3
from the entire HRTEM image.
As indicated above, after two catalytic cycles, the reactivity
of Rh–STF is almost totally suppressed. Accordingly, in
Figure 10, an EDX line scan across two metal particles very
clearly shows the strong iron contamination within the Rh par-
ticles. The signals of Rh (green) and Fe (blue) thus follow each
other almost in parallel (the spectrum taken at d=10 nm
shows no Ti from the perovskite, but the simultaneous pres-
ence of Rh and Fe).
In situ measurements in H /H O mixtures of different reduc-
2
2
tion strengths on pure LSF samples have indicated that the
[
23]
metallic iron segregation can be reversible. To exclude the
reversibility of metal–support interaction in our system, and to
study any possibility to re-activate the catalyst for the CO2
methanation reaction, the deactivated samples were treated in
O for one hour at 873 K. For both systems, the Fe–Rh alloy
2
formation was irreversible. The EDX spectra along the marked
line in the HAADF image for Rh–LSF (Figure 11 left) indicate the
presence of iron also for the measurement position on the par-
ticle where it does not overlap with the perovskite (first spec-
trum). For Rh–STF, Fe was detected by an EDX line scan, too
Figure 8. HRTEM image of Rh–LSF after reduction in hydrogen at 873 K. The
red arrows indicate (012) lattice fringes of the LSF structure. Quantification
of the EELS line scan (Fe K (blue), Rh L (green), and La L edges (red)) mea-
sured along the black arrow is shown as inset.
(
see Figure 10). The high-resolution image in addition shows
lattice fringes of both Fe O and Rh O (Figure 11 right). In sum-
2
3
2
3
mary, the changes observed at ambient conditions were hardly
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2063
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 10. HAADF image and EDX spectra of Rh–STF after two catalytic
cycles in CO methanation, as shown in Figure 4B. The EDX line scan is
2
shown as inset.
Figure 12. Fe2p3/2 XP spectra of the pre-oxidized Rh–LSF sample (similar to
“
as prepared”) and after subsequent reduction in hydrogen at 873 K. Com-
III
II
ponents of Fe and Fe were fitted after previously measuring the binding
energy using commercial Fe and FeO obtained via in situ decomposition
of Fe oxalate (cf. Experimental section).
O
3
reversed by the oxidative treatment at 873 K, which demon-
strates the very high stability of the alloyed particles.
2
II
to the measured Fe2p3/2 peaks. Whereas the fully oxidized
Oxidation state of iron in the reduced sample
III
state can be described perfectly well by using the Fe refer-
To collect more information about the oxidation state of iron
in LSF during reduction and to gain information about the
electronic modifications leading to the observed decreased
catalytic reactivity, XPS measurements were performed
ence, in the reduced case three different species could be
found. Besides a very small amount (1–3 at.%) of metallic iron
II
that is expected to arise from the alloyed particles, Fe was
found to be the major component at approximately 60 at%.
Correlating the iron oxidation state with d (La0.6Sr0.4FeO3Àd),
one can assume that the relative difference between the fully
oxidized state and the reduced one is approximately 0.3,
(
Figure 12). The sample after oxidation (similar to after calcina-
tion) was compared with the reduced state after the hydrogen
treatment at 873 K (solid black lines), and a qualitative differ-
ence between them can be observed. To interpret the oxida-
tion state of Fe, conclusive Fe O (haematite) and FeO (wues-
[
58,59]
which is well comparable to the literature values.
2
3
tite) reference samples were measured and subsequently fitted
Figure 11. HAADF image and EDX spectra along the indicated line for Rh–LSF (left) and HAADF and HRTEM images and EDX line scan along the red arrow for
Rh–STF (middle and right) after reoxidation at 873 K.
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2064
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
the latter was dissolved in water and added dropwise to STF,
which was suspended in water (ꢀ150 mL) under permanent stir-
ring. Water was removed cautiously by gentle heating and the re-
sulting powder was heated overnight at 873 K to decompose
Rh(NO ) . A 10 mol% loading of Rh on LSF was prepared by
Conclusions
We were able to clarify the structural and catalytic manifesta-
tions of metal–support interaction in metal–perovskite “model
systems” (Rh on strontium titanium ferrite and lanthanum
strontium ferrite). The term “model” in this respect refers to
the properties of easy- and hard-to-reduce perovskite systems.
No signs of classical, reversible strong metal–support interac-
tion effects such as the encapsulation of active Rh metal area
by reduced perovskite species (or oxide species caused by the
decomposition of the perovskite structure) were observed.
This is directly corroborated by the missing structural and cata-
lytic restorage capabilities of the Rh–perovskite systems upon
re-oxidation, being a direct prerequisite of a metal–support in-
teraction to be classified as “strong”. Rather, the metal–support
interaction is a “reactive” and irreversible one. As such, it is do-
minated by iron exsolution effects, leading to the more or less
pronounced formation of Fe–Rh alloy phases and subsequently
spoiling the methane chemistry of Rh, both in methane re-
3
3
a water-free synthesis routine starting from the organometallic pre-
cursor Rh (CO) (Sigma Aldrich, melting point: 423 K). The latter
4
12
was dissolved in acetone and slowly added dropwise to LSF sus-
pended in acetone (ꢀ200 mL) during permanent stirring. Subse-
quently, acetone was removed by gentle heating the suspension at
its boiling point (330 K) and the resulting sample heated stepwise
(
50 K steps, 1 h each) up to 673 K. The mean particle sizes for both
preparation routes were comparable both after preparation
(
ꢀ2 nm) and after reduction (that is, before running the catalytic
tests; ꢀ7 nm). The particle size distributions were based on TEM
analysis of 50–70 particles in each case and the associated accessi-
ble Rh surface sites additionally corroborated by chemisorption
measurements (see Figure S4).
Pretreatments and catalytic experiments
forming and CO methanation. In this respect, future directed
2
Before each measurement, the catalyst was pretreated in oxygen
at 673 K for 1 h to remove surface carbon. Subsequently, reductive
treatments in hydrogen (1 bar) at 673 K and 873 K, respectively,
were performed to reduce Rh O , thus activating the catalyst. Cata-
and systematic synthesis approaches will eventually reveal the
intrinsic catalytic properties and physicochemical properties of
several well-defined Fe–Rh alloy phases, also in their unsup-
ported isolated state. This potentially goes beyond the mere
2
3
lytic measurements were performed in an all-quartz batch reactor
suppression of Rh methane chemistry or CO hydrogenation
using catalyst material of ꢀ50 mg. Details of the setup are de-
2
[17]
activity by reactive metal–support interaction. Significant cata-
lytic contributions of the perovskite phases, either by its hydro-
gen chemistry or the inverse water gas shift reaction at high
reaction temperatures, are also observed. Finally, we anticipate
that the results can directly be transferred to similar metal-on-
complex-oxide systems, whose interaction behavior by the in-
herently more complex chemistry is most likely much more
variable than simpler supporting oxides. This subsequently
opens the door for more systematic studies of metal–support
interaction effects in this catalytically increasingly important
class of materials. Eventually, regarding the use of noble-metal
pronounced mixed ionic–electronic conducting perovskite
anode materials, exsolution phenomena of the most easily re-
ducible oxide constituents (such as Fe) under the strong reduc-
ing conditions of SOFC anode operation, may strongly coun-
teract the originally anticipated catalytic promotion of fuel re-
forming and electrooxidation.
scribed elsewhere.
The gas-phase composition was detected
online by using a quadrupole mass spectrometer, connected to
the reactor by a capillary. Two different reactions were tested cata-
lytically, methane steam reforming [Eq. (1)],
CH þ x H O ! CO þ ð2 þ xÞH
x ¼ f1,2g
ð1Þ
4
2
x
2
and CO methanation [Eq. (2)],
2
CO
þ 4 H
! CH
þ 2 H
O
ð2Þ
2
2
4
2
The reaction mixture consisted of CH4 (20 mbar) and H O
2
(
20 mbar) for Reaction (1), and CO (20 mbar, for Rh–LSF: 15 mbar)
2
and H (80 mbar, for Rh–LSF: 60 mbar) for Reaction (2). To account
2
for the gas withdrawal through the leak, Ar (10 mbar) was admit-
ted to all reaction mixtures. Subsequently, He was added to a total
pressure of 1 bar. Admission temperature into the reactor was
373 K. After an equilibration phase (40 min) the reactor was heated
À1
up to 873 K at a rate of 5 Kmin . In some cases, the reaction was
repeated on the same catalyst to monitor the structural stability of
the system. For analysis of the catalytic turnover, the relative inten-
sities of the mass spectrometer signals were converted into partial
pressures by an external calibration using gas mixtures of defined
Experimental Section
À1
partial pressures, and are given in mbarmin vs. time (see Fig-
Materials preparation
ure S1–S3). Subsequently, the pressure signals were differentiated
and the rate per active Rh site (TOF) calculated, with the number
of Rh sites calculated from H2 adsorption measurements (Fig-
ure S4). Accounts of mass and heat transport limitations in the
chosen reactor setup have been thoroughly assessed in previous
publications and did not play a substantial role under the chosen
The perovskites were synthesized either by the Pechini routine
[60]
[17]
(
LSF) or a dedicated solid-state reaction (STF). The latter in-
cludes mixing TiO , SrCO , and Fe O (all 99.95% pure, Sigma Al-
2
3
2
3
drich) and calcining them at 1270 K for 2 h, followed by additional
mixing in a mortar und subsequent calcination at 1520 K for anoth-
[17]
[17]
er 2 h. Details are given elsewhere. To activate the chosen perov-
skites catalytically for the methane steam reforming/methanation
reactions, a synthesis routine involving impregnation of the perov-
skites with Rh particles was subsequently applied. For STF, a stan-
dard wet impregnation synthesis was performed, using Rh(NO ) as
reaction conditions. This followed from comparing the collision
À2
number Z with the maximum CO TOF value per area in cm on
2
the most active catalyst discussed in this work. Pore diffusion was
also excluded due to the low surface areas of the catalysts under
2
À1
question (ꢀ0.4 m g ). In addition, heat transfer limitation played
3
3
precursor material. To prepare a nominal Rh loading of 10 mol%,
a minor role because the low reaction rates suppress local temper-
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2065
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
ature gradients and heat transfer via the gas phase is limited be-
cause of the admission of He to all reaction mixtures.
[2] G. H. Haller, D. E. Resasco, Adv. Catal. 1989, 36, 173–236.
[
[
3] J. Liu, ChemCatChem 2011, 3, 934–948.
4] S. Tauster, Acc. Chem. Res. 1987, 20, 389–394.
As a reference catalyst, commercial Rh-Al O (Heraeus, 2% Rh) was
[5] M. Boudart, Adv. Catal. 1969, 20, 153–165.
2
3
measured in the same setup under identical experimental condi-
tions (Figure S7).
[6] K. Foger in Catalysis: Science and Technology, Vol. 6 (Eds.: J. R. Anderson,
M. Boudart), Springer, Berlin, 1984, p. 227.
[7] S. Bernal, G. Blanco, J. M. Gatica, C. Larese, H. Vidal, J. Catal. 2001, 200,
11.
4
[
8] For strong metal–support interactions, see:R. Burch in Hydrogen effects
in catalysis (Eds.: Z. Paal, P. G. Menon), Marcel Dekker, New York, 1998,
p. 347.
Structural and spectroscopic characterization
Powder XRD measurements were performed on a Bruker AXS D8-
Advance powder diffractometer in Bragg–Brentano q–q geometry.
The diffractometer exhibited parallel beam optics and a Sol-XE
energy dispersive detector. Diffractograms were recorded in the
range between 2q=28 and 908 with a step size of 2q=0.028 and
a counting time of 3 s per step. Samples were prepared on back-
ground-free Si single-crystal sample holders.
[9] S. J. Tauster, S. C. Fung, R. L. Garten, J. Am. Chem. Soc. 1978, 100, 170–
175.
[
[
10] G. C. Bond, Stud. Surf. Sci. Catal. 1982, 11, 1–10.
11] S. Penner, M. Armbrüster, ChemCatChem 2015, 7, 374–392; a) S. Penner,
D. Wang, R. Podloucky, R. Schlçgl, K. Hayek, Phys. Chem. Chem. Phys.
2004, 6, 5244–5250; b) S. Labich, A. Kohl, E. Taglauer, H. Knçzinger, J.
Chem. Phys. 1998, 109, 2052–2055; c) F. Sadi, D. Duprez, F. Gerard, S.
Rossignol, A. Miloudi, Catal. Lett. 1997, 44, 221–228; d) S. Penner, B. Je-
newein, D. Wang, R. Schlçgl, K. Hayek, Appl. Catal. A 2006, 308, 31–42;
e) B. Jenewein, S. Penner, K. Hayek, Appl. Catal. A 2006, 308, 43–49;
f) D. E. Resasco, R. S. Weber, S. Sakellson, M. McMillan, G. L. Haller, J.
Phys. Chem. 1988, 92, 189–194; g) S. Sakellson, M. McMillan, G. L. Haller,
J. Phys. Chem. 1986, 90, 1733–1738.
Surface areas using the BET method were determined for both
2
À1
pure perovskite samples to be approximately 0.4 m g . BET sur-
face areas were measured with a Quantachrome Nova 2000 Sur-
face and Pore Size Analyser.
Electron microscopy experiments were performed by using an ana-
lytical transmission electron microscope (FEI TECNAI F20 S-TWIN)
operated at 200 kV acceleration voltage. HRTEM and HAADF
images were acquired, and EEL as well as EDX spectroscopies were
performed. The EEL spectra were collected by using a Gatan GIF
Tridiem energy filter with an energy dispersion of 0.5 eV and an ex-
posure time of 1–2 s. For chemical quantification using EEL spec-
trometry, Hartree–Slater cross-section models for the O K, Fe L, Rh
M, and La M edges were employed. EDX quantification of the Fe K,
Sr K, Rh K, and LaÀL peaks was performed by using the Cliff Lorim-
er model.
[
[
12] M. A. PeÇa, J. L. G. Fierro, Chem. Rev. 2001, 101, 1981–2017.
13] S. Royer, D. Duprez, F. Can, X. Courtois, C. Batiot-Dupeyrat, S. Lassiri, H.
Alamdari, Chem. Rev. 2014, 114, 10292–10368.
[
14] P. Hirunsit, K. Faungnawakij, J. Phys. Chem. C 2013, 117, 23757–23765.
[15] B. M. Reddy, A. Khan, Catal. Rev. Sci. Eng. 2005, 47, 257–296.
[16] S. Hernµndez-Morejudo, A. Masso, E. Garcia-Gonzalez, P. Concepcion,
J. M. Lopez-Nieto, Appl. Catal. A 2015, 504, 51–61.
[
[
[
17] R. Thalinger, A. K. Opitz, S. Kogler, M. Heggen, D. Stroppa, D. Schmid-
mair, R. Tappert, J. Fleig, B. Klçtzer, S. Penner, J. Phys. Chem. C 2015, 119,
2
–14.
18] R. Thalinger, M. Gocyla, M. Heggen, R. Dunin-Borkowski, M. Grünbacher,
M. Stçger-Pollach, D. Schmidmair, B. Klçtzer, S. Penner, J. Catal. 2016,
3
37, 26–35.
The Rh–LSF samples were further analyzed by XPS using a Thermo
Scientific MultiLab 2000 spectrometer (with a base pressure in the
19] H. Wang, Y. Fang, Y. Liu, X. Bai, J. Nat. Gas Chem. 2012, 21, 745–752.
À11
À10
[20] D. Neagu, G. Tsekouras, D. N. Miller, H. Menard, J. T. S. Irvine, Nat. Chem.
013, 5, 916–923.
21] Q. Qin, G. Wu, S. Chen, W. Doherty, K. Xie, Y. Wu, Electrochim. Acta 2014,
27, 215–227.
high 10 to the low 10 mbar range), equipped with an Alpha
10 hemispherical sector analyzer as well as a monochromated AlKa
2
1
[
[
X-ray source. For charge compensation, electrons with a kinetic
energy of 6 eV were accelerated towards the sample. Internal cali-
bration of the energy shift was performed relative to the C1s peak,
set to a fixed binding energy of 284.8 eV. Reference spectra of
Fe O and FeO were recorded to unequivocally assign the chemical
1
22] A. K. Opitz, A. Nenning, C. Rameshan, R. Rameshan, R. Blume, M. Hä-
vecker, A. Knop-Gericke, G. Rupprechter, J. Fleig, B. Klçtzer, Angew.
Chem. Int. Ed. 2015, 54, 2628–2632; Angew. Chem. 2015, 127, 2666–
2670.
23] R. Thalinger, M. Gocyla, M. Heggen, B. Klçtzer, S. Penner, J. Phys. Chem.
C 2015, 119, 22050–22056.
24] K. Zhang, S. Shaikhutdinov, H. J. Freund, ChemCatChem 2015, 7, 3725–
2
3
[
[
[
state of the Fe 2p region especially in the oxidized and the re-
2
+
duced state. To avoid oxidation or disproportionation of Fe
during transport in air, FeO was prepared in situ by controlled de-
composition of iron(II)oxalate (FeC O ) upon heating to tempera-
3730.
2
4
25] M. G. Willinger, W. Zhang, O. Bondarchuk, S. Shaikhutdinov, H. J. Freund,
tures above 573 K using the in situ heating stage of the spectrom-
R. Schlçgl, Angew. Chem. Int. Ed. 2014, 53, 5998–6001; Angew. Chem.
[
61]
eter.
2014, 126, 6108–6112.
[
[
26] L. J. Swartzendruber, Bull. Alloy Phas. Diagr. 1984, 5, 456–462.
27] J. L. G. Fierro, Catal. Rev. Sci. Eng. 1992, 34, 321–336.
Acknowledgements
[28] T. Seiyama, Catal. Rev. Sci. Eng. 1992, 34, 281–300.
[
[
29] K. Ichimura, Y. Inoue, I. Yasumori, Catal. Rev. Sci. Eng. 1992, 34, 301–320.
30] J. Villoria, N. Mota, S. A. Al-Sayari, M. Alvarez-Galvan, Micro- and Nano-
syst. 2012, 4, 231–252.
We thank the FWF (Austrian Science Foundation) for financial
support under the project F4503-N16. The work was performed
at the University of Innsbruck within the platform “Material- and
Nanoscience”. Samples were kindly provided by A. K. Opitz, Insti-
tute of Materials Chemistry, Technical University Vienna.
[31] M. A. Ulla, R. A. Migone, J. O. Petunchi, E. A. Lombardo, J. Catal. 1987,
105, 107–119.
[
[
32] S. Choi, S. Moon, Catal. Today 2009, 146, 148–153.
33] H. Provendier, C. Petit, A. Kiennemann, C. R. Acad. Sci. Paris Serie Iic
2001, 4, 57–66.
[
[
[
34] H. Provendier, C. Petit, C. Estournes, S. Libs, A. Kiennemann, Appl. Catal.
A 1999, 180, 163–171.
35] H. Provendier, C. Petit, C. Estournes, A. Kiennemann, C. R. Surf. Sci. Catal.
Keywords: electron microscopy · iron · perovskite phases ·
rhodium · supported catalysts
1998, 119, 741–749.
36] J. Gao, L. Jia, W. Fang, Q. Li, H. Song, J. Fuel Chem. Technol. 2009, 37,
[
1] R. W. Joyner, J. B. Pendry, D. K. Saldin, S. R. Tennison, Surf. Sci. 1984, 138,
4–94.
573–577.
8
[37] R. A. Stowe, W. W. Russell, J. Amer. Chem.Soc. 1954, 76, 319–323.
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2066
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[
[
[
38] V. R. Choudhary, B. S. Uphade, A. A. Behelkar, J. Catal. 1996, 163, 312–
18.
39] K. Urasaki, Y. Sekine, S. Kawabe, E. Kikuchi, M. Matsukata, Appl. Catal. A
005, 286, 23–29.
40] P. Sabatier, J. B. Senderens, J. Soc. Chem. Ind. London 1902, 21, 504–
06.
41] W. Wei, G. Jinlong, Front. Chem. Sci. Eng. 2011, 5, 2–10.
42] G. Alex Mills, F. W. Steffgen, Catal. Rev. Sci. Eng. 1974, 8, 159–210.
43] T. Schaaf, J. Grünig, M. Schuster, T. Rothenfluh, A. Orth, Energy Sustaina-
ble Soc. 2014, 4, 1–14.
[52] PDF-4+ Version 4.1403 2014, ICDD International Centre for Diffraction
Data, Newtown, Square, Pennsylvania.
[53] H. E. Swanson, E. Tatge, R. K. Fuyat, Natl. Bur. Stand. Circ. 1953, 539, 39.
[54] G. Rupprechter, G. Seeber, H. Goller, K. Hayek, J. Catal. 1999, 186, 201–
213.
[55] J. M. D. Coey, Acta Crystallogr. Sect. B 1970, 26, 1876–1877.
[56] L. J. Swartzendruber, J. Phase Equilib. 1984, 5, 456–462.
[57] R. L. Blake, R. E. Hessevic, T. Zoltai, L. W. Finger, Am. Mineral. 1966, 51,
123.
[58] M. Søgaard, P. V. Hendriksen, M. Mogensen, J. Solid State Chem. 2007,
180, 1489–1503.
3
2
5
[
[
[
[
[
[
[
44] D. J. Elliott, J. H. Lunsford, J. Catal. 1979, 57, 11–26.
45] C. Rhodes, G. J. Hutchings, A. M. Ward, Catal. Today 1995, 23, 43–58.
46] C. Ratnasamy, J. P. Wagner, Catal. Rev. 2009, 51, 325–440.
47] C. R. F. Lund, J. E. Kubsh, J. A. Dumesic, ACS Symp. Ser. 1985, 279, 313–
[59] M. Kuhn, S. Hashimoto, K. Sato, K. Yashiro, J. Mizusaki, Solid State Ionics
2011, 195, 7–15.
[60] M. P. Pechini, U.S. Pat. 3,330,679, 1967.
3
38.
[61] A. F. Holleman, E. Wiberg, Lehrbuch der anorganischen Chemie, Walter de
Gruyter.
[
[
48] J. Wei, E. Iglesia, J. Catal. 2004, 225, 116–127. (110 kJ Rh Al O ).
2 3
49] G. Jones, J. G. Jakobsen, S. S. Shim, J. Kleis, M. P. Andersson, J. Rossmeisl,
F. Abild-Pedersen, F. Bligaard, T. Helveg, S. Hinnemann, B. Rostrup-Niel-
sen, J. Catal. 2008, 259, 147–160.
[
[
50] J. C. Matsubu, V. N. Yang, P. Cristopher, J. Am. Chem. Soc. 2015, 137,
Received: March 4, 2016
3
076–3084.
51] C. Force, A. Ruiz Paniego, J. M. Guil, J. M. Gatica, C. López-Cartes, S.
Revised: April 11, 2016
Bernal, J. Sanz, Langmuir 2001, 17, 2720–2726.
Published online on May 20, 2016
ChemCatChem 2016, 8, 2057 – 2067
www.chemcatchem.org
2067
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim