Chromia on anatase: A catalyst for selective hydrocarbon combustion in the presence of CO

Article abstract of DOI:10.1002/cctc.201402769

The catalytic activity and selectivity for propane combustion in the presence of CO have been studied with anatase, rutile, eskolaite, chromia-impregnated anatase, and titania-impregnated chromia and its physical mixtures. It was found that the selectivity and activity depend on the TiO₂ modification. Cr-doped anatase converts 80% propane with 100% selectivity, whereas chromia on rutile converts CO preferentially to propane. Cr₁₀@HOM 500/4 (HOM = Hombikat) and the physical mixture Cr₁₀ + HOM 500/4 are promising new catalysts for selective propane oxidation. High-resolution TEM and powder XRD analyses have shown that a solid solution of chromia in anatase is the active phase of these selective oxidation catalysts. Diffuse reflectance UV/Vis spectroscopy demonstrated the effects of Cr-doping of TiO₂, and Rietveld refinement analysis implied a solid solution of Cr³⁺ ions in anatase. Evidence for a Marsvan Krevelen mechanism was obtained. Catalyst deactivation was reduced by increasing the oxygen content in the feed. The catalyst is reactivated by treatment with oxygen.

Full text of DOI:10.1002/cctc.201402769

DOI: 10.1002/cctc.201402769
Full Papers
Chromia on Anatase: A Catalyst for Selective Hydrocarbon
Combustion in the Presence of CO
[
a]
Simon Swislocki, Klaus Stçwe, and Wilhelm F. Maier*
The catalytic activity and selectivity for propane combustion in
the presence of CO have been studied with anatase, rutile, es-
kolaite, chromia-impregnated anatase, and titania-impregnated
chromia and its physical mixtures. It was found that the selec-
pane oxidation. High-resolution TEM and powder XRD analyses
have shown that a solid solution of chromia in anatase is the
active phase of these selective oxidation catalysts. Diffuse re-
flectance UV/Vis spectroscopy demonstrated the effects of Cr-
tivity and activity depend on the TiO modification. Cr-doped
doping of TiO , and Rietveld refinement analysis implied
2
2
3
+
anatase converts 80% propane with 100% selectivity, whereas
chromia on rutile converts CO preferentially to propane.
Cr @HOM 500/4 (HOM=Hombikat) and the physical mixture
a solid solution of Cr ions in anatase. Evidence for a Mars–
van Krevelen mechanism was obtained. Catalyst deactivation
was reduced by increasing the oxygen content in the feed.
The catalyst is reactivated by treatment with oxygen.
10
Cr +HOM 500/4 are promising new catalysts for selective pro-
10
Introduction
The reduction of pollution from combustion engines is based
on modern catalysts that have been developed during the last
search for HC-selective mixed-oxide catalysts. Using propane
as a HC model led to the discovery of TiCrCeOx catalysts,
which provided a propane conversion of 58% and a CO con-
version of only 25% at 3508C in a propane/CO mixture with
30 years. The three-way catalytic converter (TWC) is highly
active for the oxidation of carbon monoxide (CO) and hydro-
[11]
carbons (HC) and the reduction of NO . This converter shows
large CO excess. Further studies led to a HC-selective TiCrOx
x
[12]
good stability, but the main drawback is its noble-metal con-
catalyst.
During a detailed mechanistic study Ce-doped
[1]
tent and its restriction to gasoline four-stroke engines. Little
attention has been paid to another large class of combustion
engines, the two-stroke engines found in scooters, small motor
bikes, construction equipment, and mobile household ma-
chines, such as lawn mowers, chain saws, leaf blowers, hedge
trimmers, portable generators, irrigation pumps, and many
others. Owing to the oil content of the gasoline–oil mixture as
fuel, the exhaust gases of two-stroke engines are dominated
by hydrocarbons, their fragments, and CO. Emission standards
for two-stroke scooters and motorcycles, stages Euro 2 to
TiCrOx was obtained, this species converted approximately
90% propane at 3758C with 100% selectivity in the presence
of a five-fold excess of CO. Mechanistic studies performed by
using diffuse-reflectance infrared Fourier-transform spectrosco-
py (DRIFTS) confirmed that propane forms combustible adsor-
bates (carboxylates) on the catalyst surface in the presence, as
well as absence, of molecular oxygen, whereas CO forms only
formates with surface OH groups in the absence of oxygen.
These formates are not oxidized, but are reversibly decom-
posed to CO and OH. The sol–gel-derived catalysts showed
good propane activity and excellent selectivity, but their ex-
À1
Euro 5, show an acceptance of 1000 mgkm CO, whereas the
À1
[13]
HC content ranges from 1200 to 100 mgkm depending on
pensive and time-consuming syntheses are a disadvantage.
[
2]
the stage. Apparently, the problem is not so much the CO
emission, but the HC fragments that are responsible for odor
and toxicity. Noble-metal catalysts are known to preferentially
This disadvantage prompted us to investigate catalysts synthe-
sized by completely different synthetic routes, such as wet im-
pregnation and solid-state reactions. Therefore, the objective
of this work was the simplification of the synthesis, the im-
[
3–6]
convert CO rather than HCs
and are, therefore, not suitable
for the cleaning of the two-stroke engine exhaust. Several
mixed-metal oxides have been reported as good alternatives
provement of the sol–gel-derived TiCr(Ce)O system discovered
x
[13]
previously, as well as the elucidation of the effect of the de-
fined phases, anatase (TiO ), rutile (TiO ), and eskolaite (Cr O ),
[
7–10]
to noble-metal catalysts for the oxidation of HCs.
There-
2
2
2
3
fore, it is attractive to develop catalysts that preferentially con-
vert HCs in the presence of CO. Based on sol–gel methods, we
have used high-throughput technologies in a highly diverse
on catalytic activity and selectivity. For anatase we used the
commercially available TiO , Hombikat. The catalysts are abbre-
2
viated according to preparation procedure; for impregnated
materials “@” is used (Cr @HOM 500/4 stands for 10 at.% chro-
10
[
a] S. Swislocki, Prof. Dr. K. Stçwe, Prof. Dr. W. F. Maier
Technische Chemie, Saarland University
Campus C4 2,D-66123 Saarbruecken (Germany)
E-mail: w.f.maier@mx.uni-saarland.de
mia impregnated on Hombikat, calcined for 4 h at 5008C), “+”
represents a physical mixture ground in a mortar (e.g. Cr +
1
0
HOM 500/4), and “Ti62.5Cr37.5O ” represents a sol–gel-derived ref-
x
erence material.
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cctc.201402769.
ChemCatChem 2015, 7, 261 – 270
261
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
of the study by mass-transport limitation under the reaction
conditions with this new catalyst system.
Results
Catalytic-activity measurements have been performed in a gas-
Mass-transport limitation, already excluded for the sol–gel
[13]
[14]
phase flow reactor as described in a previous publication.
materials,
was investigated with the most active catalyst,
Propane oxidation is highly exothermic, therefore, the catalyst
bed was always diluted with catalytically inactive sand (200 mg
sand, 20 mg catalyst, sieve fraction 100–200 mm). This ten-fold
dilution was necessary to allow reliable temperature control of
the catalyst bed. It was surprising to note, that the impregnat-
ed Hombikat showed activity and selectivity comparable to
the sol–gel reference materials from previous studies. For dif-
ferent Cr loadings (5, 10, 20, and 30 at.%), Cr @HOM 500/4
Cr @HOM 500/4. Although mass-transport limitation was pres-
10
À1
ent at flow rates below 30 mLmin , the conversion stabilized
À1
at flow rates of 50 mLmin and above. Because all other cata-
lysts studied here are less active than Cr @HOM 500/4, mass-
10
transport limitations can be excluded at our standard flow rate
À1
of 50 mLmin .
It was also of interest to investigate the catalytic activity and
selectivity of the pure oxide phases (see Table 1). For anatase,
commercially available Hombikat UV100 was used. Eskolaite
10
was the most active catalyst with a propane conversion to CO2
and H O of 80% at 3758C. CO conversion was not detected, in-
2
dicating a propane selectivity of 100% despite the five-fold
excess of CO in the feed gas. In fact, a slight CO formation of
approximately 2% was observed, indicating incomplete pro-
pane combustion. This observation was also supported by the
Table 1. Activity, selectivity, and specific surface area of the pure metal
oxides and the catalysts.
Conversion
Propane
Specific
substoichiometric conversion of O related to propane (ꢀ5%).
at 3758C [%]
selectivity
at 3758C [%]
surface area
2
2
À1
The temperature-dependent competitive conversion of pro-
pane and CO is shown in Figure 1. When determining the con-
Propane
CO
[m g
]
Cr O 500/4
2
3
15
0
4
0
79
–
50
11
312
8
TiO
2
2
(Hombikat UV100)
RUT 800/10
TiO
1
1
Cr10@HOM 500/4
Cr10 +HOM RT
Cr10 +HOM 500/4
Cr10@RUT 500/4
Cr10 +RUT 500/4
80
2
35
14
5
0
2
0
25
2
100
50
100
36
84
278
78
–
–
[
[
a]
a]
71
(
Cr
8
Ce
2
)@HOM 500/4
78
60
2
0
98
100
110
261
Ti62.Cr37.5
O
x
[
a] The specific surface area was too low to be determined.
(
Cr O ) was synthesized by calcination of Cr(NO ) ·9H O at
2 3 3 3 2
5
008C for 4 h according to the synthesis of Cr @HOM 500/4.
10
Therefore, we were interested to discover whether rutile, the
thermodynamically stable TiO modification, acts differently to
2
anatase. Rutile was synthesized by annealing the commercial
anatase and rutile mixture P25 at 8008C for 10 h. Diffraction
patterns showed no reflections corresponding to anatase, but
did show sharp reflections corresponding to rutile (see
3
+
Figure 9). The rutile powder was impregnated with Cr solu-
tion and calcined as before. Although anatase was catalytically
inactive, eskolaite showed a propane conversion of 15%. Rutile
showed very poor conversions of propane and CO and no se-
lectivity was observed (see Table 1). Physical mixtures were
prepared by grinding the two powders in pentane with
a mortar and pestle, followed by drying. Ball milling was also
attempted, but the resulting mixture was catalytically inactive,
therefore, the simple mortar approach was used. Although
Cr +HOM RT (RT stands for room temperature) showed only
Figure 1. Temperature-dependent conversion of propane and CO with
Cr10@HOM 500/4. No CO is converted. The arrows indicate decreasing and
increasing the temperature during the measurement.
versions of propane and CO separately (separated feed gases),
a CO conversion of 16% at 3758C was observed, whereas pro-
pane showed a conversion of 86%. This result shows that the
two gases compete in the oxidation process, but in a mixture,
propane combustion is favored. These results are consistent
with our mechanistic investigations for selectivity with the sol–
10
poor conversion and no selectivity, Cr +HOM 500/4 provided
10
[
13]
gel catalysts.
DRIFTS spectra were also recorded for
a 35% propane conversion with 100% selectivity (see Table 1).
The temperature-dependent catalytic activity of the annealed
physical mixture of Cr +HOM 500/4 is shown in Figure 2. The
Cr @HOM 500/4, showing the same bands as that of the sol–
10
gel catalyst (not shown here). This result suggests that the
same combustion mechanism is applicable to both systems.
Before continuing it was important to rule out any corruption
10
data for the various catalytic systems studied is summarized in
Table 1, which provides conversions as well as surface areas
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
262
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. Temperature-dependent conversion of propane and CO on
Cr10 +HOM 500/4. No CO is converted. The arrows indicate decreasing and
increasing the temperature during the measurement.
Figure 3. Temperature-dependent conversion of propane and CO on
Cr10@RUT 500/4. Both propane and CO are converted, with a selectivity of
64% for CO.
and propane selectivities. Propane selectivity is defined as
apply to sol–gel-derived catalysts for which small amounts of
[13]
SProp =XProp/(XProp +X ). Note that we did not define turnover
cerium led to more active catalysts.
CO
frequencies as usual for catalytic reactions because the
number of active centers on such bulk materials could not be
determined.
Another important factor for catalytic applications is long-
term stability. As shown by the black squares in Figure 4,
Cr @HOM 500/4 showed a rapid loss of activity during the first
10
The poor behavior of rutile was confirmed with the impreg-
nation of rutile with chromia. Cr @RUT 500/4 provided
24 h, after which its activity leveled out, after 3 days, to ap-
proximately 40% (the same observation was made with sol–
gel-derived Ce Ti Cr O , not shown here). Supposedly, the
10
a higher CO conversion than propane conversion resulting in
a propane selectivity of only 36% (see Table 1). The tempera-
ture-dependent catalytic activity of Cr @RUT 500/4 is shown in
2,5 50 47.5
x
active phase has to be formed at the interphase of the two
metal oxides owing to the annealing process. Therefore, the
10
Figure 3, documenting the reverse selectivity of this catalyst.
Apparently, only catalysts based on anatase (Hombikat) and
annealed with chromia (Cr O ) display the desired catalytic ac-
2
3
tivity and selectivity.
Impregnation of the chromia with titania (Ti @Cr O 500/4)
10
2
3
resulted in a catalyst less active and less selective than the
chromia alone (XProp =8%, XCO =3%), supporting the hypothe-
sis that chromia acts as a catalyst, whereas titania acts as a sup-
port.
The lower catalytic activity of Cr +HOM RT relative to the
10
pure Cr O 500/4 is attributed to the reduced amount of chro-
2
3
mia in the physical mixture because for all experiments 20 mg
of catalyst was used.
Cerium can play an important role in this catalytic reaction,
as shown in our earlier studies with the sol–gel-based cata-
[
13]
lysts, therefore, cerium was added to the chromia-impregna-
tion solution. A cerium-containing catalyst, (Cr Ce )@HOM 500/
8
2
4, was prepared by impregnation. This catalyst converted 78%
propane and approximately 2% CO at 3758C. On comparison
with Cr @HOM 500/4 there was no improvement in catalytic
10
activity, but a slight loss in selectivity (see Table 1). Therefore,
the addition of cerium to the catalyst did not improve the cat-
alytic activity of the catalysts prepared by impregnation and
was, therefore, abandoned in this study. This result does not
Figure 4. Long-term stability tests for Cr10@HOM 500/4. The temperature
was held at 3758C. The black squares show the deactivation over 72 h,
whereas the red squares show cycles of 24 h with reconditioning in between
by pure molecular oxygen for 1 h.
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
263
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
catalyst was annealed at 5008C for a further 100 h. Counter-dif-
4
+
3+
fusion processes of Ti and Cr in the lattice should be pro-
moted under these conditions, which may be the key to the
activity and selectivity of the catalyst. Activity tests of these
materials (denoted as 500/100) showed that there was no im-
provement in activity and the selectivity remained the same.
Cr @HOM 500/100 converted 70% propane at 3758C (not
10
shown here), which is a slight decrease in activity. The physical
mixture Cr +HOM 500/100 also showed no catalytic improve-
10
ment compared with that Cr +HOM 500/4. We concluded
10
that annealing at 5008C for 4 h is sufficient to obtain these
active and selective catalysts.
X-ray fluorescence analysis (XRF) before and after the stabili-
ty test showed no leaching of the metals over three days
under the reaction conditions [Ti/Cr=86:14 fresh and Ti/Cr=
84:16 after 72 h (at.%)], also no phase transformation could be
identified in the powder diffraction pattern (not shown). In our
previous study, evidence for a Mars–van Krevelen mechanism
[
13]
was obtained. Therefore, this rapid drop of catalytic activity
could result from depletion of lattice oxygen owing to the oxi-
dation process. If this is the case, additional oxygen should
result in restoration of catalytic activity, and higher oxygen
content in the feed gas should cause a reduction in the activity
drop. Cr @HOM 500/4 was tested again and pure oxygen was
Figure 5. Long-term stability tests of Cr10@HOM 500/4 with two different
oxygen concentrations at 3758C. Black squares: stoichiometric propane to
O
2
ratio (10.0 vol.% O
2
); Red squares: Doubled oxygen concentration
(20.0 vol.% O2).
dilution of the catalyst in the catalyst bed, as described in the
Experimental Section, and the small amount of catalyst used in
the experiments. The BET measurements provided 84 and
10
introduced at 3758C for one hour after every 24 h of reaction.
This procedure increased the propane conversion after 24 h
from 48 to 69%, showing that the catalyst could be recondi-
tioned by molecular oxygen. The catalyst was deactivated
again under the reaction conditions, but could be recondi-
tioned with molecular oxygen to increase the propane conver-
sion from 44 to 64% (see Figure 4, red squares). The sol–gel-
derived Ti Cr Ce Ox catalyst from our previous work
2
À1
2
À1
87 m g
for the fresh catalyst, 75 m g after 72 h and
2
À1
81 m g after 24 h followed by reaction with oxygen for 1 h.
Although the observed trend follows the loss of activity report-
ed above qualitatively, its absolute value cannot explain the
strong loss observed, especially considering the reproducibility
of such BET measurements. Therefore, sintering was excluded
as the main reason for deactivation. Another reason for the ex-
clusion of sintering as the main reason for deactivation is that
deactivation would not be reversible and would not be affect-
ed by oxygen concentration. The main sintering seems to take
place during the calcination at 5008C, as shown in Table 1, for
example, the surface area of fresh Cr +HOM 500/4 decreased
50
47.5
2.5
showed the same reconditioning effect (not shown here). If
the depletion of lattice oxygen is the main reason for deactiva-
tion under the reaction conditions, then an increase in the
concentration of oxygen should result in a reduction of the de-
activation owing to the higher equilibrium concentration of
oxygen.
10
2
À1
The effect of doubling the oxygen concentration in the feed
gas was investigated. Long-term stability tests with a stoichio-
metric propane to oxygen ratio (1:5, as used for all measure-
ments) and a propane to oxygen ratio of 1:10, whereby the CO
content was not considered are shown in Figure 5. Doubling
the oxygen concentration increased the propane conversion at
the beginning of the measurement from 78 to 97% (see
Figure 5). Surprisingly, it did not cause a higher CO conversion,
the selectivity for propane combustion remained at 100%,
which may be taken as evidence against hot-spot formation
from 278 to 78 m g . Nevertheless, the decrease in initial ac-
tivity over time, shown in Figure 4 and 5, that is not recovered
by oxygen treatment, may be attributed to some sintering.
However, this decrease could also result from the oxygen treat-
ment being too short or from coking. To investigate a potential
contribution to the deactivation from coking, a thermogravi-
metric analysis (TGA) experiment was carried out under air. The
results are shown in Figure 6. Fresh Cr @HOM 500/4 showed
10
a small weight loss of 2% below 1508C, attributed to the loss
of water, and another small weight loss above 5508C, attribut-
ed to carbon-containing surface contamination. The used cata-
lyst (2 days under the reaction conditions) showed the same
weight losses and an additional weight increase of <1%
above 3008C, which we can only attribute to the filling of
oxygen vacancies in the catalyst lattice. Therefore, TGA sup-
ports the Mars–van Krevelen mechanism and provides evi-
dence against the importance of catalyst coking during this re-
action. This is also supported by the light-green color of the
used catalyst, another indication against serious coking.
(which should lead to loss of selectivity). The deactivation was
slower and leveled out at 56% propane conversion after 72 h.
The results strongly support the Mars–van Krevelen mechanism
proposed earlier. Although evidence against sintering of the
catalysts as the main reason for deactivation has been present-
ed above, we have measured the BET surface of a fresh Cr +
10
HOM 500/4 catalyst before and after a 72 h oxidation experi-
ment and after a 24 h oxidation followed by 1 h reactivation
with oxygen. The experiment was difficult because of the high
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
264
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tration, indicating that most Cr ions must have diffused into
the anatase lattice of all anatase particles. During the HRTEM
study only one small Cr O particle could be found with a com-
2
3
position of 34 atom% Cr and 58 atom% O. Despite all efforts,
no evidence for surface decoration of the anatase particles
with chromia could be obtained, confirming that most of the
Cr species diffused into the anatase particles and the remain-
ing Cr species are only detectable as Cr O3 particles. The
2
HRTEM of Cr @RUT 500/4 showed a particle size of approxi-
10
mately 150 nm. All selected-area EDX analysis obtained from
the bulk, as well as from the edges, showed different ratios of
Cr/Ti, varying from 0 to 50 (see Figure S2 in the Supporting In-
formation). Clearly, Cr is only found on the outer part of parti-
cles, whereas the particles themselves consisted of pure TiO2.
This finding indicates that Cr does not diffuse into the rutile,
but remains as separate phase. It is also interesting to note,
that the particle size of the rutile must result from sintering at
8008C because P25 is known to have a particle size of approxi-
mately 20 nm.
Figure 6. Thermogravimetric analysis under air of fresh and used
Cr10@HOM 500/4 (used for 2 days under reaction conditions).
To learn more about the difference between Cr @HOM 500/
10
4
and Cr @RUT 500/4, high-resolution transmission electron
10
microscopy (HRTEM) studies were performed. Cr @HOM 500/4
The powder X-ray diffraction pattern (PXRD) of
Cr @HOM 500/4 (see Figure 8) showed two phases: Anatase
10
showed the expected coagulation of nanoparticles (10–20 nm
diameter) displayed in Figure 7a. Figure 7b (amplification of
the bottom cluster in Figure 7a) documents the crystalline
nature of these anatase particles. Micro-EDX (ꢀ10 nm areas,
EDX=energy-dispersive X-ray), performed on 15 selected areas
of nanoparticles (centers as well as edges) throughout the
sample provided a rather constant atomic ratio of Cr/Ti of
10
(TiO ) and eskolaite (Cr O ), confirming the pure anatase nature
2
2
3
of Hombikat UV100. The powder diffraction patterns of all
three Hombikat-based catalysts are summarized in Figure 8.
ꢀ
0.1 with a variation of 10%. Further details on HRTEM with
EDX analysis are shown in Figure S1 in the Supporting Informa-
tion. This analysis shows that all particles and areas sampled
are composed of Cr and Ti in approximately identical concen-
Figure 8. Powder X-ray diffraction pattern of selected Cr–HOM catalysts
(
TiO
2
1
space group: I4 /amd; inorganic crystal structure database (ICSD):
202242; Cr O
2 3
space group R 3¯ c; ICSD: 26791).
The rather broad and intense reflections correspond to anatase
confirming the nanoparticlular nature seen in HRTEM), where-
(
as the low-intensity reflections with small FWHM (full width at
half maximum) correspond to eskolaite, indicative of larger
particles. No other phases were observed. Cr +HOM RT
10
showed the broadest anatase reflections, indicative of the
nanocrystalline titania. The specific surface area of these cata-
lysts and the pure oxides were investigated by single-point
BET measurements and are summarized in Table 1. Hombika-
2
À1
t UV100 showed a specific surface area of 312 m g , whereas
2
À1
the physical mixture Cr +HOM RT revealed 278 m g . An-
10
nealing Hombikat UV100 and Cr +HOM RT at 5008C for 4 h
10
Figure 7. a,b) HRTEM micrographs of Cr10@HOM 500/4; b) is an amplification
of the lower part of (a) showing the crystalline nature of this anatase cata-
lyst.
resulted in a decrease of specific surface area to 92 and
2
À1
78 m g , respectively. After wet impregnation and calcination,
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
265
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
analysis of Cr @RUT 500/4 showed almost identical lattice pa-
10
rameters (a 4.59502; c 2.95903) when compared with pure
rutile (a 4.59426; c 295915), indicating that in this case very
little, if any, Cr is incorporated into the rutile lattice. This find-
ing was also confirmed by the phase composition, which was
3
+
9
2.3% rutile and 7.7% chromia. This facile migration of Cr
ions into the anatase lattice is not unreasonable because the
3
+
4+
ionic radius of Cr (75.5 pm) is very similar to that of Ti
[16]
(
74.5 pm).
To learn more about the role of the Cr ions in the active
phase, the catalyst powder samples were investigated by dif-
fuse reflectance UV/Vis spectroscopy (DR-UV/Vis). It was of in-
terest to clarify whether there is a difference between the inac-
tive physical mixtures and the active catalysts. Indeed, a distinct
difference in the reflectance spectra was observed (see
Figure 10). Hombikat UV100 started to absorb light at 400 nm
Figure 9. Powder X-ray diffraction pattern of rutile and Cr10@RUT 500/4 syn-
thesized by wet impregnation and calcined at 5008C (TiO space group: P4
mnm; ICSD: 82085; Cr
space group: R 3¯ c; ICSD: 26791).
2
2
/
2
O
3
the Cr @HOM 500/4 catalyst showed a specific surface area of
10
2
À1
only 84 m g . In contrast to the sol–gel-derived catalysts of
[
13]
our previous work, there is no correlation of surface area
with catalytic activity and selectivity. The PXRD shown at the
bottom of Figure 8 confirms the presence of eskolaite and ana-
tase in the physical mixture Cr +HOM RT, which showed no
10
catalytic activity (see Table 1). However, after calcination at
008C the propane conversion at 3758C reached 35%, where-
5
as no CO was converted (see Figure 2). Although less active
compared to the impregnated Hombikat, this facile solid-state
reaction of TiO and Cr O with cation counter diffusion is the
2
2
3
most simple and low-cost way to obtain an active and selec-
tive catalyst for this reaction. Note that no CO conversion was
detected despite the high CO excess in the feed gas. Small,
but sharp, reflections of eskolaite indicate relatively large resid-
ual chromia crystals in Cr @HOM 500/4 and Cr +HOM 500/4
Figure 10. DR-UV/Vis spectra of Hombikat UV100, Cr10 +HOM RT,
Cr10 +HOM 500/4, Cr10@HOM 500/4, and Cr O 500/4.
2 3
10
10
relative to those in Cr +HOM RT (Figure 8). This finding indi-
10
cates that because of the high-temperature treatment most of
the original Cr has disappeared (dissolved in the anatase lat-
tice).
and full absorbance was reached at 325 nm, which is typical
[15]
for anatase.
The physical mixture Cr +HOM RT showed
10
a similar behavior with a lower reflectance of approximately
45% above 400 nm. The two small reflectance minima at 470
and 600 nm indicate octahedrally coordinated Cr, more pre-
The PXRD pattern of impregnated rutile Cr @RUT 500/4
10
showed sharp reflections of rutile and two broad reflections of
eskolaite at 25 and 338 with low intensity (see Figure 9). These
reflections correspond to the two most intense reflections of
eskolaite; the lower intensity reflections of eskolaite were not
observed. In Cr @HOM 500/4, the Cr species is largely incorpo-
[17,18]
cisely Cr O , coexisting with TiO .
The band-gap region at
2
3
2
approximately 375 nm was unaffected by the presence of Cr.
In contrast to this, Cr +HOM 500/4 showed a drastic change
10
of the reflectance spectrum compared to the physical mixture.
We observed additional absorption in the range of 360–
600 nm as the most significant difference attributed to the an-
nealing at 5008C. Consequently, it is not surprising that the im-
10
rated in the anatase phase as shown by HRTEM and Rietveld
refinement, but about 1.5% remains as Cr O , according to
2
3
HRTEM large crystals, which may be responsible for the chro-
mia reflections in the PXRD (Figure 8). In Cr @RUT 500/4, the
pregnated catalyst Cr @HOM 500/4 showed a similar absorp-
10
10
Cr species does not dissolve in the rutile lattice; rather it “dec-
orates” the rutile crystals and as such does not provide chro-
mia reflections.
tion spectrum with an even broader reflectance ascent in the
range 360–620 nm. In the literature for similar (Cr,Ti)O systems
2
[17,18]
the spectral changes are attributed to Cr doping.
Cr O 500/4 absorbed nearly all of the light over the whole
Rietveld analysis of Cr @HOM 500/4 showed a change of
10
2
3
the anatase lattice parameters (a 3.7913; c 9.4958) compared
to pure HOM (a 3.7996; c 9.5105), indicative of incorporation
of Cr into the anatase lattice (solid solution). Phase composi-
tion of the sample was found to be 98.5% anatase and 1.5%
chromia, indicating that up to 8.5 at.% of the 10 at.% Cr ions
are incorporated in the anatase lattice. In contrast, Rietveld
measurement range. Again, two minima at 470 and 600 nm, as
already mentioned above, could be observed.
Cr @RUT 500/4, which showed the reverse catalytic selectiv-
10
ity, was investigated in the same way by DR-UV/Vis (see
Figure 11). The spectral changes of the impregnated catalyst
are very similar to those of Cr @HOM 500/4, the main differ-
10
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
266
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
active than and as propane selective as the original sol–gel
catalyst Ti62.5Cr37.5O , whereas Cr @RUT 500/4 was CO selective
x
10
and less active. Therefore, anatase is not only a support for the
chromia, but dramatically improves the combustion selectivity,
resulting in a highly selective catalyst for hydrocarbon combus-
3
+
tion. Rutile, which does not dissolve the Cr ions, has the op-
posite effect as a support for chromia, it improves the CO com-
bustion and reduces the relative hydrocarbon combustion. To
the best of our knowledge, this is the first time that the titania
modifications have had such a drastic effect on catalytic selec-
tivity. Attempts to reverse the catalyst texture were unsuccess-
ful. Titania (10 mol%) on chromia reduced the catalytic activity
and selectivity of the pure eskolaite, confirming the role of tita-
nia as support because on the catalyst surface it has a detri-
mental effect. The results also suggest that anatase has to be
doped with chromia to observe the desired catalytic per-
formance. Further support for this hypothesis was obtained by
physically mixing chromia with anatase (Cr +HOM RT), which
2 3
Figure 11. DR-UV/Vis spectra of rutile, Cr10@RUT 500/4, and Cr O 500/4.
ence is that the additional band is not as broad (400–550 nm)
as in the sample mentioned first. This indicates Cr doping into
the rutile surface in a similar manner to the case of Cr-impreg-
nated anatase. As with anatase, pure rutile is inactive for pro-
pane as well as for CO oxidation. Therefore, Cr doping must
also be the key to the activity and selectivity for the rutile-
based catalyst. Furthermore, we observed the two minima at
10
resulted in an unselective material of very low catalytic activity.
When the mixture was calcined at 5008C, the resulting material
showed a good activity (XPropane =35%) and 100% selectivity
for propane combustion in the presence of five-fold CO excess.
We assume that chromia and anatase form a solid solution
comparable to the impregnation of anatase with chromia. The
surface areas of the two catalysts are comparable, therefore,
470 and 600 nm, which are also apparent in the spectra of
Cr O 500/4. This indicates the presence of eskolaite (corun-
2
3
dum-type Cr O ), which could hardly be observed in the dif-
the lower activity of Cr +HOM 500/4 relative to that of
2
3
10
fraction patterns of Cr @RUT 500/4 (see Figure 9), therefore,
Cr @HOM 500/4 can be attributed to incomplete incorpora-
10
10
supporting the surface-decoration argument outlined above.
tion of chromia into the anatase lattice by the annealing pro-
cess relative to impregnation. The lack of catalytic selectivity of
Cr @RUT 500/4 is attributed to the lack of rutile to form
10
Discussion
a solid solution with chromia because it changes the propane
selectivity of pure chromia from 79 to 36%, thus improving
the undesired CO oxidation. Ruiz et al. have reported that
mixed oxides of chromia–titania with 5–30% Cr upon anneal-
ing only show the anatase phase, whereas traces of eskolite
formation can only be seen in the 30% Cr material. The dis-
solved Cr is not visible in the PXRD as was the case in our
study. Upon further heating to 9008C not only a phase change
to rutile was observed, but also segregation of additional chro-
mia appeared, indicating that the chromia dissolves in anatase
Sol–gel-derived TiCr(Ce) oxides have been developed as effec-
tive catalysts for selective propane combustion in the presence
[
11–13]
of CO.
Alternative catalysts have been prepared by the
more simple impregnation of Hombikat UV100 (anatase) with
3
+
Cr ion-containing solution followed by calcination at 5008C
for 4 h. Chromia (10 mol%) was found to provide the most
active impregnated catalysts, this finding is in strong contrast
to the sol–gel materials, for which a chromia content of nearly
[13]
40 mol% provided the most active material (Ti62.5Cr37.5Ox).
[19]
Considering the potential toxicity of chromium oxides, the sig-
nificantly lower Cr content of the impregnated catalysts may
be an advantage for potential applications. Calcination at
much better than it does in rutile. This observation is in per-
fect agreement with our data.
The most active catalyst Cr @HOM 500/4 for propane com-
10
5
008C provided the best catalysts, therefore, the defined
bustion in the presence of CO (XPropane =80%) shows 100%
propane selectivity. Calcination for 100 h at 5008C did not im-
prove the catalytic activity—propane conversion dropped
slightly to 70% and selectivity remained stable at 100%, indi-
cating the high structural stability of this new catalyst system.
The catalytic performance of Cr @HOM 500/4 also indicates
phases formed at this temperature [anatase (TiO ) and esko-
2
laite (Cr O )] may be connected to the unusual catalytic behav-
2
3
ior observed. The catalytic performance of the pure phases at
758C showed that anatase and rutile are inactive, whereas es-
3
kolaite shows a low activity for propane oxidation with a selec-
tivity of 79% relative to simultaneous CO oxidation (see
Table 1). This finding points to chromia as the catalytically
active phase, whereas anatase acts as a suitable support mate-
10
that TiO and Cr O do not have to be mixed from the begin-
2
2
3
ning of the synthesis on an atomic scale, as was originally con-
cluded from the sol–gel materials. HRTEM, as well as Rietveld
3
+
3+
rial, which dissolves the Cr ions in its lattice, as shown by
the TEM data and the Rietveld analyses. This dissolution of
analysis, indicate that the solid solution of Cr ions in anatase
is the active site of the catalyst. This analysis also explains the
much lower chromia content relative to the sol–gel catalysts,
in which a large portion of the chromia is hidden in the bulk
material, where it may not contribute to the catalysis. The sol–
3
+
Cr into the anatase lattice represents the active and selective
phase of Cr @HOM 500/4. When the two TiO modifications
10
2
were impregnated with chromia Cr @HOM 500/4 was more
10
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
267
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
gel procedure used is well known to provide true mixed
which implies that the lattice oxygen in the solid is replenish-
ed, supporting the above hypothesis. When the reaction was
continued, the catalyst deactivated again and reconditioning
with molecular oxygen again increased the catalytic activity
(see Figure 4). An increase of the oxygen content in the feed
should result in a higher equilibrium conversion with the risk
that CO oxidation may be promoted owing to excess oxygen.
Doubling the oxygen concentration in the gas feed resulted in
a higher propane conversion and a smaller loss in activity after
72 h with an equilibrium conversion of 56% (see Figure 5) and,
surprisingly, no additional CO was oxidized, so the propane
combustion selectivity was maintained. Owing to experimental
limitations and the explosion risk, higher oxygen concentra-
tions could not be studied, although they may be promising.
In this case microreactors should be used. We have shown that
most of the activity loss during the experiments can be attrib-
uted to loss of lattice oxygen, which is not filled fast enough
by the oxygen concentrations used in the feed. The fact that
the original catalytic activity was not recovered fully could be
due to irreversible sintering, it could also be due to incomplete
reoxidation or to some coking. Evidence against coking was
provided by the TGA experiment, which also supported the
Mars–van Krevelen mechanism. Because the particle size of
P25 changed from 20 to approximately 250 nm upon sintering
at 8008C, sintering remains the most likely option to explain
the unrecovered activity loss.
[20]
oxides, which may be best described as solid solutions. DR-
UV/Vis investigations, which mainly sample the surface layers,
revealed that Cr is incorporated into the TiO surface. This Cr-
2
doped TiO phase on the surface is the key to the observed
2
catalytic activity and selectivity. To determine the oxidation
state of Cr in the TiO lattice, X-ray photoelectron spectroscopy
2
(
XPS) was considered, but was not used owing to its known
3
+
4+
problems with differentiation between Cr
and Cr
ions
3
+
4+ [21]
(
overlap of Cr2p for Cr and Cr ). Oxygen XPS was consid-
ered to be too complex because Cr oxides and Ti have several
O1s signals. The dissolution of Cr in the anatase lattice is
3
+
strong evidence for the exclusive presence of Cr ions owing
4
+
2+
to ionic radius. Whether Cr or Cr ions are present or are
formed during the catalytic reactions cannot be clarified with
our methods and would require additional investigations.
More information on the detailed mechanism would require
better knowledge about the surface mobility of oxygen during
the catalysis. Isotopic exchange studies with oxygen are prom-
ising techniques as pioneered by Descorme and Duprez on
[
22]
mixed oxides. Unfortunately, to date, no studies have been
reported on the Cr–Ti mixed-oxide couples.
During the sol–gel-study, addition of Ce(NO ) solution was
3
3
[
13]
found to significantly improve the catalytic activity. It is well
3
+
4+
known that the Ce is converted to Ce (CeO ) during calci-
2
nation. The role of ceria was further investigated by impreg-
3
+
3+
nating Hombikat UV100 with a Ce and Cr solution. Al-
though there was no improvement in propane combustion ac-
tivity, the catalyst converted some CO, resulting in a drop in
selectivity. The lack of Ce to improve the catalytic performance
of the impregnated catalyst is attributed to its exclusive nature
to modify the surface. We believe that in the sol–gel synthesis
the Ce ions are incorporated into the mixed-oxide lattice (Cr–
Ti), where they increase the surface area and may support the
lattice oxygen contribution to the combustion.
More evidence for the origin of the selective propane com-
bustion activity was obtained with DR-UV/Vis spectroscopy.
The physical mixture of Hombikat UV100 and Cr O showed
2
3
that both phases coexisted alongside each other. The annealed
mixture (Cr +HOM 500/4) and the impregnated material
10
(Cr @HOM 500/4) showed an additional band ranging from
10
360 to 600 nm and 360 to 620 nm in the DR-UV/Vis spectra, re-
spectively. This is taken as evidence that Cr is incorporated
[17,18]
into the TiO lattice.
The wide range of this band could be
2
Testing the long-term stability of Cr @HOM 500/4 provided
an indication for gradual doping. Because this band was ac-
companied by increased propane combustion activity, this is
10
further insights into the oxidation mechanism. The doped cata-
lysts showed almost the same activity loss as the sol–gel-de-
rived catalysts. Activity decreased from 80 to 40% within 72 h.
Neither XRF nor PXRD showed any changes in the solid after
this period, which could indicate leaching of the metal ions or
phase changes. In our mechanistic study we observed, by
DRIFTS analysis, that propane formed carboxylates with the lat-
further evidence for the Cr-doped TiO phase as the key to the
2
selective propane oxidation. The impregnated rutile showed
a similar DR-UV/Vis spectrum, indicating that the additional
band may be associated with the catalytic activity, but does
not explain the catalytic selectivity. Our hypothesis on the role
of anatase as a support is strengthened by the poor activity of
the Ti-impregnated Cr O . The DR-UV/Vis spectra and the
[
13]
tice oxygen of the catalysts, supporting a Mars–van Krevelen
mechanism. When propane is oxidized by the solid, the solid
itself is reduced and is afterwards oxidized by the molecular
oxygen of the gas feed. Apparently, the oxidation of propane
and consequently the reduction of the solid are faster than its
2
3
powder diffraction pattern showed that there are two separate
phases of TiO and Cr O and in addition Cr-doped TiO is also
2
2
3
2
present. These measurements give further insights into the
origin of the active phase, but the roles of the two phases and
the doped phase in the selective oxidation of propane need to
be investigated in more detail by other techniques.
reoxidation by O . At the point where propane activity remains
2
constant (X=40%, see Figure 4), the reduction and the oxida-
tion of the solid are equilibrated. Clearly, with a gas-feed com-
position of 2.0 vol.% propane, 10.9 vol.% CO, and 10.0 vol.%
Conclusion
O the reaction was operated under reductive conditions (stoi-
2
chiometric combustion of CO and propane would require
Doping of anatase with chromia (Cr @HOM 500/4) provided
10
1
5.5 vol.%), which may restrain oxidation processes. After 24 h
a propane-selective combustion catalyst in the presence of
a five-fold excess of CO (100% selectivity). The same catalyst,
of reaction pure molecular oxygen was introduced, at 3758C
for 1 h, and most of the initial catalytic activity was restored,
when prepared with the TiO modification rutile, was CO selec-
2
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
268
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
tive and much less active for propane combustion. HRTEM
studies have shown that the chromia does not decorate the
anatase nanoparticles, but dissolves completely, whereas chro-
mia on rutile does decorate the crystal surface and shows
a lower tendency to dissolve in this lattice. These findings
were supported by Rietveld refinements. Surface-sensitive DR-
UV/Vis spectroscopy provided evidence that Cr is incorporated
into the surface of the anatase and to a lesser extent also into
the surface of the rutile phase. Rapid initial deactivation could
be assigned to loss of lattice oxygen, supporting a Mars–van K-
revelen mechanism. Catalyst activity could be restored by
purging with pure oxygen. Deactivation of the catalysts was
attributed to depletion of lattice oxygen by the reductive gas
feed, underlining the proposed Mars–van Krevelen mechanism.
TGA with the deactivated catalyst showed a slight weight in-
crease at 3008C, which is related to filling of oxygen vacancies,
thus further supporting the Mars–van Krevelen mechanism. In-
creasing the oxygen concentration in the feed resulted in im-
provements of initial and long-time conversion without caus-
ing undesired CO oxidation. The most active and selective cat-
alyst, Cr @HOM 500/4, was obtained by impregnation of Hom-
All the other catalysts were prepared in a similar way. For Ce im-
pregnation of Hombikat a Ce(NO ) ·6H O solution was used.
3 3
2
Activity measurement
The activity-measurement setup consisted of a glass-tube reactor
with an internal diameter of 6.1 mm. 20 mg of catalyst (100–
2
00 mm) diluted with 200 mg SiO (100–200 mm) was added to the
2
glass tube and pretreated in synthetic air at 3758C for 90 min. The
feed gas (C H /CO/CO /O /N =2.0/10.9/15.8/10.0/61.3 vol.%) and
3
8
2
2
2
the synthetic air (O /N =20/80 vol.%) had a total gas flow of
2
2
À1
5
0 mLmin . Both, the reactor containing the catalyst and the reac-
tor that acts as bypass (filled with SiO ) were encased in the same
2
brass heating unit. The feed gas composition was analyzed flowing
through the bypass by a micro gas chromatograph (GC) (Varian
CP-4900) before the first (3758C) and after the last run (2758C).
Measurements were taken at 275, 300, 325, 350, and 3758C, prod-
uct gas composition was monitored by GC. Between each sample
run, there was a hold time of 15 min. Investigations without pro-
pane or CO in the gas feed were carried out by replacing one of
the gases by N2.
Mass-transport limitation was investigated at 3758C with the same
setup as that described above. The gas flow rate as well as the cat-
alyst mass and SiO2 mass were changed by the same factor to
obtain a constant residence time. For each measuring point, fresh
catalyst was filled into the reactor and pretreated at 3758C for
10
3
+
bikat with Cr solution, which is a significant improvement of
the synthesis compared with our earlier sol–gel synthesis. Fur-
ther simplification of the synthesis was achieved by physically
mixing the two oxide phases, which after calcination were also
catalytically active and propane selective (Cr +HOM 500/4).
9
0 min in synthetic air.
10
Long-term stability tests were also carried out in the same arrange-
ment, holding the temperature at 3758C for the chosen period.
Catalyst deactivation that could not be restored could be at-
tributed to further sintering. Coking was excluded by TGA ex-
periments.
For the reconditioning, pure molecular oxygen was introduced at
À1
3
758C for 1 h at a gas flow rate of 50 mLmin
.
Chromia-doped anatase is a new heterogeneous catalyst
with unusual catalytic selectivities for hydrocarbon oxidation. It
may also have yet unrecognized properties as a selective oxi-
dation catalyst for a variety of reactions not studied yet.
Characterization
To specify the surface area of the catalysts, nitrogen physisorption
measurements were carried out on a single-point BET Fisons In-
struments Sorpty 1750. The adsorption isotherms were recorded at
À1968C with ground materials. The samples were outgassed at
Experimental Section
2
008C for 2 h.
Catalysts are identified by their relative molar composition given
as subscript after the element symbol. “@” is used for wet impreg-
nation, “+” for a physical mixture, “HOM” for Hombikat UV100 (a
Powder X-ray diffraction patterns were obtained using a PANalytical
X’Pert PRO system with
Ni-filtered Cu-Ka1 and Cu-Ka2 radiation (l=1.54060 ꢁ and l=
commercial nanocrystalline and pure anatase TiO from Sachtleben
2
1
.54443 ꢁ, respectively). The instrument has a Bragg–Bretano ge-
Chemie), “P25” for TiO P25 from Degussa, and “RUT” for rutile. Cal-
2
ometry and a PIXcel detector. The shown diffraction patterns are
background corrected with X’Pert Highscore Plus software.
cination is characterized by final temperature (8C) and duration (h).
Diffuse-reflectance UV/Vis (DR-UV/Vis) measurements were ob-
tained with a PerkinElmer UV/Vis-NIR spectrometer Lambda 19
equipped with a Labsphere RSA-PE-19 Ulbricht sphere. All samples
were ground and measured in a sample holder covered by quartz
glass.
Synthesis
The Cr @HOM 500/4 catalyst was prepared by wet impregnation
1
0
and final calcination at 5008C for 4 h. To obtain 20 mmol catalyst,
Cr(NO ) ·9H O (800 mg, Sigma Aldrich) was dissolved in deionized
3
3
2
H O (5 mL). Hombikat UV100 (1.438 mg) was added under stirring,
followed by treatment in an ultrasonic bath for 2 h. The suspension
X-ray fluorescence (XRF) analyses were performed on an EDAX
Eagle II m-Probe instrument with a rhodium anode as a radiation
source and evaluated with the program Vision 32 (Version 3.35).
The excitation voltage was 40 kV. Each powder sample was mea-
sured at three different spots for 60 s and the results were aver-
aged.
2
was stirred again for 15 min, dried at 1108C for 12 h and calcined
À1
at 5008C for 4 h. The heating ramp was 108Ch .
Cr +RUT 500/4 was prepared by solid-state reaction. First, P25
1
0
(
TiO2 from Degussa) was annealed at 8008C for 10 h to obtain
pure rutile, Cr(NO ) ·9H O (Sigma Aldrich) was annealed at 5008C
For high-resolution transmission electron microscopy (HRTEM) a Hi-
tachi HF-2000 instrument was used. Energy-dispersive X-ray analy-
sis (EDX) was performed by using a Noran EDX system with
a liquid nitrogen cooled Si(Li) detector. For sample preparation the
3
3
2
for 4 h to obtain Cr O . TiO (72.6 mg) was ground with Cr O3
2
3
2
2
(
7.27 mg) in n-pentane by using a mortar and pestle. The powder
was dried at 1108C for 12 h and calcined at 5008C for 4 h.
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
269
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
catalyst powders were deposited on Holey carbon grids (Cu,
00mesh, 3 mm diameter).
[6] B. S. Caglayan, ꢄ. Soykal, A. E. Aksoylu, Appl. Catal. B 2011, 106, 540–
549.
4
[
7] K. Alexopoulos, M. Anilkumar, M. F. Reyniers, H. Poelman, S. Cristol, V.
Balcaen, P. M. Heynderickx, D. Poelman, G. B. Marin, Appl. Catal. B 2010,
97, 381–388.
Thermographic analysis was performed on a Mettler Toledo TGA/
e
e
DSC 1 Star system (Software Star version 12.00a, 70 mL corundum
À1
À1
crucible, 30–9008C with 28Cmin , synthetic air 30 mLmin ).
[8] D. Shee, A. Sayari, Appl. Catal. A 2010, 389, 155–164.
9] G. Busca, E. Finocchio, V. Lorenzelli, G. Ramis, M. Baldi, Catal. Today
999, 49, 453–465.
[
1
[
[
10] J. E. Tasca, C. E. Quincoces, A. Lavat, A. M. Alvarez, M. G. Gonzꢅlez,
Ceram. Int. 2011, 37, 803–812.
11] P. Rajagopalan, K. Stçwe, W. F. Maier, Catal. Commun. 2010, 11, 1041–
Acknowledgement
We thank the Max-Planck Institut fꢀr Kohlenforschung in Mꢀl-
heim and Bernd Spliethoff for the HRTEM investigations and E.
Dietzsch, Chemnitz University, for BET-measurements.
1
044.
[12] P. Rajagopalan, K. Stçwe, W. F. Maier, Top. Catal. 2010, 53, 19–27.
[13] S. Swislocki, K. Stçwe, W. F. Maier, J. Catal. 2014, 316, 219–230.
[14] P. Rajagopalan, PhD Thesis, Saarland University, Saarbrꢂcken 2010.
[15] D. Reyes-Coronado, Nanotechnology 2008, 19, 145605.
Keywords: chromium
hydrocarbons heterogeneous catalysis
combustion · carbon monoxide · titania
·
exhaust
gas
treatment
·
[16] P. D. Shannon, Acta Crystallogr. Sect. A 1976, 32, 751–767.
[
[
17] C. C. Tsai, H. Teng, Appl. Surf. Sci. 2008, 254, 4912–4918.
18] N. Serpone, D. Lawless, J. Disdier, J. M. Herrmann, Langmuir 1994, 10,
·
·
propane
643–652.
[
19] A. M. Ruiz, G. Skai, A. Cornet, K. Shimanoe, J. R. Morante, N. Yamazoe,
Sens. Actuators B 2003, 93, 509–518.
[
[
1]
H. S. Gandhi, G. W. Graham, R. W. McCabe, J. Catal. 2003, 216, 433–442.
[
[
20] G. Frenzer, W. F. Maier, Annu. Rev. Mater. Res. 2006, 36, 281–331.
21] B. M. Weckhuysen, I. E. Wachs, R. A. Schoonheydt, Chem. Rev. 1996, 96,
2] Achtundzwanzigste Verordnung zur Durchfꢂhrung des Bundes-Immis-
sionsschutzgesetzes (2012).
3327–3350.
[
[
3] A. B. Mhadeshwar, D. G. Vlachos, Combust. Flame 2005, 142, 289–298.
4] C. Daniel, M. O. Clartꢃ, S. P. Teh, O. Thinon, H. Provendier, A. C. Van Veen,
B. J. Beccard, Y. Schuurman, C. Mirodatos, J. Catal. 2010, 272, 55–64.
5] V. P. Zhdanov, B. Kasemo, Appl. Surf. Sci. 1994, 74, 147–164.
[
22] C. Descorme, D. Duprez, Appl. Catal. A 2000, 202, 231–241.
[
Received: September 25, 2014
ChemCatChem 2015, 7, 261 – 270
www.chemcatchem.org
270
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim