An in-depth study of supported in2O3 catalysts for the selective catalytic reduction of NOx: The influence of the oxide support

Article abstract of DOI:10.1021/jp0532824

The influence of the oxide support (i.e., A1₂O₃, Nb₂O₃, SiO₂, and TiO₂,) on the surface properties, reduction and oxidation properties, acid-base properties, and catalytic activity of supported indium oxide catalysts has been investigated by temperature-programmed reduction/oxidation, thermogravimetry coupled to differential scanning calorimetry, ammonia and sulfur dioxide adsorption calorimetry, and reduction of NO_x by ethene in highly oxygen-rich atmosphere. Two series of In₂O₃-containing catalysts at low (≈3 wt %) and at theoretical geometric monolayer (from 20 to 40 wt %) In₂O₃ content were prepared and their properties were compared with unsupported 10203 material. Supports able to disperse the In₂O₃ aggregates with high In stabilization gave rise to active catalytic systems. Among the studied oxide supports, Al ₂O₃ and, to a lower extent, TiO₂ were found to be the best supports for obtaining active de-NO_x catalysts.

Full text of DOI:10.1021/jp0532824

240
J. Phys. Chem. B 2006, 110, 240-249
An In-depth Study of Supported In2O3 Catalysts for the Selective Catalytic Reduction of
NOx: The Influence of the Oxide Support
A. Gervasini,† J. A. Perdigon-Melon,‡,§ C. Guimon,| and A. Auroux‡,*
Dipartimento di Chimica Fisica ed Elettrochimica, UniVersita` degli Studi di Milano, Via Camillo Golgi n. 19,
20133 Milano, Italy, Institut de Recherches sur la Catalyse, CNRS, 2, aV. A. Einstein, 69626, Villeurbanne,
France, Instituto de Quimica Fisica “Rocasolano”, CSIC, c/Serrano 119, 28006 Madrid, Spain, and
Laboratoire de Physico-Chimie Mole´culaire, 2 aVenue du Pre´sident Angot, 64053 Pau Cedex 09, France
ReceiVed: June 18, 2005; In Final Form: September 26, 2005
The influence of the oxide support (i.e., Al2O3, Nb2O5, SiO2, and TiO2,) on the surface properties, reduction
and oxidation properties, acid-base properties, and catalytic activity of supported indium oxide catalysts has
been investigated by temperature-programmed reduction/oxidation, thermogravimetry coupled to differential
scanning calorimetry, ammonia and sulfur dioxide adsorption calorimetry, and reduction of NOx by ethene in
highly oxygen-rich atmosphere. Two series of In2O3-containing catalysts at low (≈3 wt %) and at theoretical
geometric monolayer (from 20 to 40 wt %) In2O3 content were prepared and their properties were compared
with unsupported In2O3 material. Supports able to disperse the In2O3 aggregates with high In stabilization
gave rise to active catalytic systems. Among the studied oxide supports, Al2O3 and, to a lower extent, TiO2
were found to be the best supports for obtaining active de-NOx catalysts.
Introduction
acidic sites of the alumina support has been found to be involved
in de-NOx reactivity. In this way, it was recognized that not
only do the typical properties of supported metal phases affect
the catalytic activity (dispersion, distribution, surface structure,
etc.), but also the nature of the support dramatically influences
the activity and selectivity of the In-supported phase. The poor
performances of silica support for numerous transition metal
based catalysts toward SCR has been noted.6,11 On the other
hand, the acidic Brønsted or Lewis oxides (e.g., Al2O3, TiO2,
SiO2-Al2O3) are suitable supports for the SCR process.16-18
The high performance of In-containing catalysts in/on zeolitic
and oxide materials has been recently demonstrated for reactions
of environmental interest, in particular the selective catalytic
reduction of nitrogen oxides by hydrocarbons (HC-SCR
process).1-9 The main problem with the HC-SCR process
remains the identification of a selective catalyst that is able to
drive the reduction of NOx by the hydrocarbon, acting as a
reducing species, in a highly oxidizing atmosphere. In fact, on
the most active known de-NOx catalysts, nonselective combus-
tion by oxygen of the hydrocarbon begins to dominate the NOx
reduction above 773 K, and the NOx conversion concomitantly
begins to fall. In addition, the presence of water among the
gaseous combustion effluents containing NOx and vulnerability
to deactivation by sulfur oxide prevents zeolite-based materials
from being effectively used in this process as viable catalysts.
Among the numerous types of nonzeolitic catalytic systems
investigated for NOx abatement under net oxidizing conditions,
In2O3 2-5,10,11 and Ga2O312-14 supported on alumina (γ-Al2O3)
have demonstrated very promising performance. In particular,
these metal oxide phases are selective in HC-SCR, being able
to predominantly utilize the hydrocarbon for the NOx reduction
while limiting hydrocarbon reactivity with oxygen (combustion).
In the first part of this series,15 a detailed study of a supported
In2O3 on γ-Al2O3 system at different In loading (2 < wt % <
22) has been presented. Surface and bulk properties and catalytic
activity toward the selective reduction of NOx were investigated,
paying particular attention to the Brønsted and/or Lewis acidity
of the support and metal centers and to the redox character of
the metal active species. A balanced presence of In centers and
6
In particular, Maunula et al. reported on a sol-gel derived
alumina on which In2O3 and Ga2O3, outperform other existing
alumina supports.
In this study, we wanted to deepen the understanding of the
role of the support on the physicochemical properties and
catalytic performances of the In2O3-supported phase by studying
In-based materials on several different supports. We selected
as supports oxides with well differentiated acid properties, both
in terms of the nature of their surface sites and their acid
strength, including TiO2, Nb2O5, and SiO2, in addition to Al2O3.
For each support we also prepared low and high In-loading
samples to check the influence of In content. As it is not simple
to predict the acid character of a surface when depositing an
essentially basic oxide, such as In2O3 on an acidic or neutral
oxide support, the acidity of the prepared surfaces was carefully
studied. The redox properties of the catalysts and the In-
dispersion were determined by performing a series of experi-
ments using thermal techniques including temperature pro-
grammed reduction and oxidation (TPR-TPO), differential
scanning calorimetry (DSC), and thermogravimetric analysis
(TGA). To complete the study, the catalytic performances in
the selective reduction of NOx by ethene in an oxygen rich
atmosphere were studied and the results interpreted in light of
the different surface properties of the catalysts.
* Corresponding author. E-mail: aline.auroux@catalyse.cnrs.fr
† Universita` degli Studi di Milano.
‡ Institut de Recherches sur la Catalyse.
§ Instituto de Quimica Fisica “Rocasolano”.
| Laboratoire de Physico-Chimie Mole´culaire.
10.1021/jp0532824 CCC: $33.50 © 2006 American Chemical Society
Published on Web 12/08/2005

Oxide Supported In2O3 Catalysts
J. Phys. Chem. B, Vol. 110, No. 1, 2006 241
TABLE 1: Physicochemical Properties of the Samples
Experimental Section
In/M (molar ratio)
(M)Si,Al, Ti)
Catalyst Preparation. The samples were prepared by incipi-
ent wetness impregnation using the appropriate amounts of In-
(NO3)3‚5H2O from Aldrich (99.9%) in order to obtain two
different indium oxide loadings for each support, one series of
samples with loadings around 3wt % (hereafter named In(3)/Si
(Al, Ti, Nb) or low loading samples) and the second one with
an indium oxide loading close to the theoretical geometric
Sext
loading wt %
In2O3
sample
m2g-1
AES-ICP
XPS
In2O3
SiO2
21
100
208
215
146
112
113
82
92
101
78
In(3)/Si
In(hl)/Si
Al2O3
In(3)/Al
In(hl)/Al
TiO2
In(3)/Ti
In(hl)/Ti
Nb2O5
In(3)/Nb
In(hl)/Nb
3.1
37.2
0.01
0.26
0.02
19
monolayer (hereafter named In(hl)/Si (Al, Ti, Nb) or high
loading samples). The used supports were γ-Al2O3 and SiO2
from Degussa, TiO2 DT-51 (anatase) from Rhone-Poulenc, and
Nb2O5 (obtained by dehydration for 8 h at 523 K in O2 flow of
commercial niobium pentoxide hydrate, niobia HY-340, from
Companhia Brasileira de Metalurgia e Minerac¸a˜o Brasil). After
impregnation, the samples were dried at 393 K overnight and
calcined at 773 K in oxygen flow for 12 h. Bulk In2O3 was
prepared in the same way, by calcination of indium nitrate at
773 K after drying at 393 K overnight.
Catalyst Characterization. The concentrations of the sup-
ported indium oxides were determined by AES-ICP in a
Spectroflame-ICP instrument. Surface areas were determined
by the BET method from the adsorption of nitrogen at 77 K.
The crystallographic structure was examined by X-ray diffrac-
tion in a Bruker (Siemens) D5005 apparatus (Cu KR radiation,
0.154 nm). The oxidation states of indium were determined by
XPS performed at room temperature with an SSI 301 spec-
trometer.
3.2
26.8
0.01
0.13
0.02
0.05
3.3
20.4
0.02
0.15
0.06
0.31
108
88
71
3.2
18.3
0.03
0.27
0.39
associated with the process. In contrast to the TPR-TPO
experiments, the samples were held in a quartz crucible, which
does not allow the reactant gas to pass through the sample. As
in TPR-TPO, successive experiments of reduction and oxidation
have been carried out. The temperature was increased at 5 K
min-1 up to 923 K, keeping this temperature for 30 min, after
which the samples were cooled under helium atmosphere. In
reduction experiments a total flow of 30 cm3 min-1 with a
mixture of 68% hydrogen in helium was used. For oxidation
experiments the gas flow was 15 cm3 min-1 and the composition
was 46% O2/He.
The acidity and basicity of the supported indium samples, as
well as those of the bare support samples, were determined by
adsorption microcalorimetry of NH3 and SO2 respectively.20 The
samples were pretreated at 673 K in vacuum overnight before
the measurements, which were carried out isothermally at 353
K in a heat flow calorimeter (Setaram C80) coupled with a
standard volumetric apparatus. Successive doses of gas were
sent to the sample until a final pressure of 0.5 Torr was obtained.
The sample was then evacuated for 30 min at the same
temperature in order to remove the amount physically adsorbed,
and a second adsorption was performed. The quantity adsorbed
at 0.2 Torr in the first adsorption will be called VT. The
difference between the amounts adsorbed in the first and second
adsorptions at 0.2 Torr is the irreversibly chemisorbed amount
(Virr).
de-NOx Catalytic Activity. The catalytic tests were carried
out following the experimental procedure detailed in part I of
this series.15 The experiments were performed at a fixed time
of 1.6 g‚s‚mmol-1 and variable temperature (523-823 K) in a
quartz downflow tubular microreactor working close to the
atmospheric pressure. The feed mixture was composed of 2500
ppm of NO and 500 ppm of NO2, 3000 ppm of C2H4 and 40,000
ppm of O2, with He as balance gas. An FTIR spectrometer (Bio-
Rad with DTGS detector) was used as an analytical device. NOx
(NO+NO2) and C2H4 conversions as well as COx (CO+CO2)
yields were calculated directly from the experimental absorbance
values. The N2 yield was calculated by subtracting the amount
of N-containing species flowed out from the reactor from the
sum of all the N-containing species fed into the reactor.
The redox properties of the supported indium samples have
been studied by TPR-TPO experiments. The sample was held
on the frit of a U-shape quartz reactor, allowing the reductant
or oxidizing gas stream to pass through the sample. In TPR
experiments the temperature was increased at 5 K min-1 from
room temperature up to 1113 K, under a reducing atmosphere
of 5% H2 /Ar (flux 20 cm3 min-1). The temperature was kept
at 1113 K for 1 h, followed by cooling under argon atmosphere.
After TPR, TPO experiments were carried out using the same
experimental conditions, but since volatilization of In2O3 takes
place at 1123 K,21 the temperature was increased only up to
1073 K. The oxidizing atmosphere was 1% O2 /He. The
hydrogen or oxygen consumption was determined by means of
a TCD (Delsi Instruments DN11). Prior to TPR experiments,
the samples were reoxidized under oxygen at 773 K during 3 h
and cooled under argon atmosphere.
Results and Discussion
Physico-Chemical Characterizations. The surface areas of
the supports and supported indium oxide samples are sum-
marized in Table 1, as well as the indium oxide loadings
determined by AES-ICP. For the low loading samples (≈3 wt
% In2O3), no important variations are observed in the surface
area for In(3)/Si, In(3)/Al, and In(3)/Ti samples, except for the
In(3)/Nb sample which shows a decrease of 18%. In all cases
a marked decrease in surface area takes place for the high
loading samples reaching 30, 26, 15 and 34% for In2O3 on silica,
alumina, titania, and niobia respectively.
The XRD spectra of the high loading samples as synthesized
and bulk indium oxide are represented in Figure 1. Alumina,
silica, and niobia are amorphous, while titania presents peaks
corresponding to anatase. Bulk indium oxide presents a well-
crystallized phase. The peak intensities and their 2θ angles have
been identified as characteristic of the cubic structure of In2O3
(JCPDS card 6-0416). In the low loading samples, no diffraction
lines other than those of the supports are observed (spectra not
shown), while in the high loading supported samples diffraction
patterns characteristic of well developed crystalline phases of
For the high loading samples, the redox properties have been
also studied using a differential scanning calorimeter (DSC)
coupled to a microbalance (TGD-DSC 111 from Setaram). This
apparatus makes it possible not only to calculate the extent of
reduction or oxidation from the weight loss or gain of the
sample, but also to simultaneously measure the evolved heat

242 J. Phys. Chem. B, Vol. 110, No. 1, 2006
Gervasini et al.
Figure 2 represents the adsorption heats of ammonia and
sulfur dioxide on bulk indium oxide (Figure 2a) and supported
indium oxides versus coverage (Figure 2b-e). Table 2 sum-
marizes the total adsorbed amount (VT) and the irreversible
amount (Virr) at 0.2 Torr (see Experimental Section) concerning
bulk indium oxide and supported indium oxides. The heats of
ammonia adsorption decreased continuously with coverage,
while the SO2 adsorption heat remained constant to an adsorbed
amount of 26 µmol g-1. The fact that the adsorption heats and
irreversible volume are clearly higher for SO2 than for ammonia
adsorption indicates that indium oxide can be considered basic
rather than acidic.23
Figure 2b represents the adsorption heats of ammonia and
sulfur dioxide on both indium oxide samples supported on silica.
The bare support adsorbs neither ammonia nor SO2, indicating
that the silica surface is mainly inert. The ammonia adsorption
heats over both samples drop continuously with coverage. The
adsorption of SO2 gives rise to higher adsorption heats, and for
the In(hl)/Si sample an initial zone with constant adsorption
heats is observed. This feature is also observed for adsorption
on bulk indium oxide. Considering that there is no adsorption
on SiO2, we can conclude that the deposited indium oxide is
responsible for the adsorption. Nevertheless, the poor dispersion,
due to the inertia of the silica surface, makes the adsorbed
quantities very small, even for high indium oxide loading.
Figure 1. XRD spectra of as-synthesized supported indium oxide
samples.
indium oxide are seen. As in the case of bulk indium oxide,
these diffraction lines are characteristic of the cubic phase of
indium oxide. The intensities of the diffraction peaks and their
positions for the different supported indium oxide samples are
very similar to bulk indium oxide, indicating that the support
does not have a big influence on the crystallization of indium
oxide, and in particular does not favor the crystallization in a
special plane. The widths at half-height (fwhm) of the diffraction
peaks are related to the size of the crystal phases; with respect
to this criterion, the crystallinity varies in the order In2O3 >
In(hl)/Si > In(hl)/Al > In(hl)/Nb > In(hl)/Ti.
XPS gives information on the oxidation state and concentra-
tion of indium on the surface of the sample. For all the samples
used in the present work, only one oxidation state was found,
with In 3d5/2 binding energy between 444.5 and 444.8 eV. These
values are characteristic of In2O3.22
The dispersions of the supported phases in the high loading
samples can be compared by considering the molar ratios
between indium and the metal (M) cation of the support
determined by AES-ICP (bulk technique) and by XPS (surface
technique). Both values are summarized in Table 1. The sample
In(hl)/Si presents the lowest surface In/M ratio, despite having
the highest indium oxide loading; this is indicative of a poor
dispersion of indium oxide phase. In(hl)/Nb has a similar In/M
ratio determined by AES-ICP, but the In/Nb ratio found by
XPS is about 20 times larger, suggesting that most of the indium
is located at the surface and that the dispersion is higher. For
the In(hl)/Ti sample, the In/M ratio determined by AES-ICP
is lower than for the In(hl)/Nb sample, but the ratio determined
by XPS is of the same order in both cases, indicating that the
dispersion achieved on titania is bigger than on niobia. The In-
(hl)/Al sample presents an In/M ratio determined by AES-ICP
similar to In(hl)/Ti, but the ratio found by XPS is clearly lower,
although better than for the silica sample. This order in the
dispersions estimated by XPS (In(hl)/Ti > In(hl)/Nb > In(hl)/
Al > In(hl)/Si) is in good agreement with the crystal sizes found
by XRD. This behavior of dispersion can be expected from
considerations of surface reactivity: SiO2 is well-known to have
an inert surface, so even if the silica support has the highest
surface area of all the supports used, there are comparatively
few anchoring sites. Al2O3, TiO2, and Nb2O5 present more active
surfaces (see also below), which allow better dispersions to be
obtained.
The adsorption heats of NH3 and SO2 on the γ-Al2O3, In-
(3)/Al and In(hl)/Al samples are represented in Figure 2c. The
heats of ammonia adsorption present an interesting behavior.
Except for the initial part (up to 50 µmol‚g-1) where the heats
of ammonia adsorption on the three samples are identical, the
heats of ammonia adsorption on the supported indium oxide
samples are lower than on the bare support, with the lowest
heats corresponding to the highest amount of indium oxide. The
irreversibly adsorbed amounts (see Table 2) decrease in the order
Al2O3 > In(3)/Al > In(hl)/Al. Both results point out that the
deposition of indium oxide decreases the acidity of the sample.
The heats of SO2 adsorption on γ-Al2O3 and the indium samples
supported on alumina present a surprising behavior: while bare
alumina and In(hl)/Al are identical in the entire range of
adsorbed amounts and give rise to very similar irreversibly
adsorbed amounts (Table 2), the In(3)/Al sample presents lower
adsorption heats and a lower irreversibly adsorbed amount.
Contrary to expectations, the deposition of indium oxide
increases neither the heats nor the adsorption capacity of the
bare support.
Adsorption heats of ammonia and SO2 on TiO2 and TiO2-
supported indium samples are represented in Figure 2d. Am-
monia adsorption presents the same trend already observed for
alumina. The bare support and indium-loaded samples have the
same initial adsorption heats (up to 80 µmol‚g-1); afterward
the adsorption heats are higher on the bare support, and the
adsorbed amounts decrease with increasing indium oxide
content. The same trend is observed in the irreversibly adsorbed
amount (see Table 2), so it can be concluded that the acidity of
titania is lowered by indium oxide deposition. Contrary to
alumina, the basicity of bare titania is very low, with continu-
ously decreasing heats of SO2 adsorption. The In(hl)/Ti sample
presents initial adsorption heats that are markedly higher than
for the bare and low indium loading samples. The irreversibly
adsorbed amount of SO2 increases with the amount of supported
indium oxide (see Virr in Table 2), indicating that the basicity
of titania has been increased by the deposition of indium oxide.
Acidity and Basicity. The surface acidity and basicity are
among the most important features of a solid surface, and
influence in a broad range of different reactions. We have
studied both properties for bulk indium oxide, the supported
indium oxide samples, as well as the bare supports, to determine
the influence of the support on the deposited indium oxide.
Figure 2e represents the adsorption heats of ammonia and
SO2 on niobia and indium-supported on niobia samples. Except

Oxide Supported In2O3 Catalysts
J. Phys. Chem. B, Vol. 110, No. 1, 2006 243
Figure 2. Adsorption heats at 353 K of NH3 and SO2 on: (a) indium oxide and indium oxide supported on: (b) silica, (c) γ-Al2O3, (d) TiO2, (e)
Nb2O5.
TABLE 2: Irreversibly Adsorbed and Total Adsorbed
Amounts of NH3 and SO2 at 353K under an Equilibrium
Pressure of 0.2 Torr
irreversibly adsorbed amount corresponds to the lowest indium
loading, following the same trend as for the other supports. The
SO2 adsorption heats and the irreversible amounts increase with
the deposited amount of indium oxide. Hence, the deposition
of indium oxide decreases the acidity and increases the basicity
of niobium oxide.
NH3
SO2
VT
Virr
VT
Virr
sample
µmol g-1
µmol g-1
µmol g-1
µmol g-1
In2O3
SiO2
66
58
58
21
23
23
47
9
9
40
5
5
Redox Properties. The redox properties of the supported
indium oxide samples have been studied by means of TPR-
TPO experiments. The bare supports also have been tested. The
reduction profiles of Al2O3 and SiO2 do not show any hydrogen
consumption, while TiO2 presents a single peak with a maximum
at 823 K and a consumption of 0.6 mmol H2 g-1, and Nb2O5
also presents hydrogen consumption at temperatures around
1073 K. Figure 3a represents the TPR profiles of the low loading
samples. The In(3)/Si and In(3)/Al samples present broad peaks
with onset around 480 K. Table 3 summarizes the total hydrogen
consumptions per mol of indium oxide. The complete reduction
of indium oxide requires three molecules of hydrogen per
molecule of indium oxide. In both cases the reduction is almost
100% of the total deposited amount. In the other two low loading
In(3)/Si
In(hl)/Si
Al2O3
In(3)/Al
In(hl)/Al
TiO2
In(3)/Ti
In(hl)/Ti
Nb2O5
In(3)/Nb
In(hl)/Nb
77
30
47
40
269
259
206
403
348
270
220
237
152
161
139
125
265
215
159
123
130
84
217
192
199
159
165
160
189
163
173
91
102
128
21
26
5
10
in the initial zone where adsorption heats are greater when
indium oxide is deposited, the heats of ammonia adsorption are
not affected by the presence of indium oxide. The highest

244 J. Phys. Chem. B, Vol. 110, No. 1, 2006
Gervasini et al.
Figure 4. XRD spectra of supported In2O3 samples after TPR
experiments.
very similar to the peaks found in the other samples. The third
peak has been shifted to a lower temperature and its relative
intensity is 54%, clearly lower than for the previously mentioned
samples where it accounted for about 90% of the total H2
consumption. The peak with a maximum at 650 K is not
observed for the other samples. As remarked above, some titania
can be reduced under these conditions, but since the maximum
of the reduction peak for the bare support is at 823 K, titania
reduction should be masked in the last peak, so it can be
supposed that the second peak at 650 K is also due to supported
indium oxide. However, a peak at the same temperature was
observed for the low loading sample, so another explanation
could be that the presence of indium or indium oxide affects
the reduction temperature of titania. The amount of hydrogen
consumed is higher than expected, probably due to titania
reduction. Finally, for the In(hl)/Nb sample we found a profile
similar to that of the In(hl)/Ti sample. The two first peaks (548,
701 K) are displaced to slightly higher temperatures, and the
last one is cropped when it starts to appear. The last peak, with
an onset at 950 K and not totally developed, could be due to
niobia reduction. The hydrogen amount consumed (considering
only the first three peaks) is clearly lower than the expected
one: only 53% of the indium oxide has been reduced.
Figure 3. TPR experiments on supported indium oxide samples: (a)
low loading, (b) high loading.
TABLE 3: Hydrogen and Oxygen Consumptions in TPR/
TPO Experiments on Supported Indium Oxide Samples
TPR (1)
TPO
TPR (2)
sample mol H2/mol In2O3 mol O2/mol In2O3 mol H2/mol In2O3
In2O3
2.8
3.0
2.7
3.0
2.6
8.8
3.7
4.4
1.6
In(3)/Si
In(hl)/Si
In(3)/Al
In(hl)/Al
In(3)/Ti
In(hl)/Ti
In(3)/Nb
In(hl)/Nb
1.4
0.9
1.2
2.9
2.5
- -
- -
samples the consumed amounts are much larger than the
expected ones. For In(3)/Nb, the small difference can be due to
reduction of niobia, but in the case of In(3)/Ti the main peak
occurs at a lower temperature than titania reduction. This point
is discussed below.
On the high loading samples, the only indium phase detectable
by XRD after TPR is metallic indium (see Figure 4). Since the
reduction peaks at low temperature do not appear in the TPR
profile of bulk indium oxide, it is difficult to attribute these
peaks to different oxidation steps in the reduction from In3+ to
In0. The reduction of bulk indium oxide to metallic indium
seems to occur in one step,24 so the behavior observed on the
supported samples is generally attributed to the reduction of
indium oxide particles of different sizes.2 This feature is in
agreement with the present results. The dispersion of indium
oxide in In(hl)/Si is poor, and due to the high indium content,
indium oxide must be forming big aggregates, diminishing the
effect of the support and presenting a behavior very similar to
bulk oxide. The dispersion of indium oxide in the In(hl)/Al
sample is higher, presenting particles of smaller size that are
reduced at lower temperature. For In(hl)/Ti this effect is more
marked. This sample presents the highest indium oxide disper-
sion and the highest amount reduced at low temperature. The
same behavior is observed for the In(hl)/Nb sample, with two
peaks at low temperatures, slightly higher than for the titania-
based sample due to the lower dispersion. However, for this
sample there are two differences that deserve a more detailed
explanation: the reduced amount of indium oxide is only 53%,
and the third peak is abruptly cropped at 861 K. It has already
been reported that in this sample two effects take place at
temperatures around 873 K, namely the crystallization of the
The profiles of hydrogen consumption in the TPR experiments
for the high loading samples as well as bulk indium oxide are
represented in Figure 3b. For bulk indium oxide two peaks are
observed, the first one with no well-defined maximum and onset
at 448 K, and the second one more intense (96% of the total
consumed hydrogen) with an onset at 693 K and maximum at
1028 K. This second peak is almost but not completely
developed at the end of the temperature ramp. In this case the
total consumption is 2.8 mol of H2 per mol of In2O3, which
means that 93% of the indium is reduced. The In(hl)/Si sample
presents a profile very similar to bulk indium oxide. The
maximum of the second and main peak is at 1030 K, and as in
the case of bulk indium oxide, 93% of the supported indium
has been reduced. A similar trend is observed for the In(hl)/Al
sample, except that after the first peak a constant consumption
of hydrogen (not seen in the other samples) is observed up to
the beginning of the main peak. The temperature of the
maximum of this second peak is shifted to a lower temperature
(986 K) than for the In2O3 and In(hl)/Si samples. On the In-
(hl)/Ti sample three peaks are detected, with maxima at 543,
650, and 909 K. The positions of the first and third peaks are

Oxide Supported In2O3 Catalysts
J. Phys. Chem. B, Vol. 110, No. 1, 2006 245
Figure 5. TPO experiments on high loading supported indium oxide
samples.
Figure 7. Second TPR experiments after first TPR-TPO cycle for
samples In(hl)/Ti and In(hl)/Al.
Figure 6. XRD spectra of the In(hl)/Al sample after TPR experiment
(In TPR) and after TPO up to 503 K (In2O3 TPO 503).
niobia support with an important loss of surface area (which
can occlude the indium oxide) and the reaction of the supported
indium oxide with niobia to form InNbO4.25 Both processes are
detected in the XRD spectra measured after the TPR experiments
(Figure 4), where well crystallized niobium oxide in hexagonal
phase and InNbO4 are both detected, explaining the sharp
decrease in the amount of indium oxide available to be reduced.
The TPO experiments carried out after TPR evidenced
remarkable differences among the different samples (Figure 5).
For bulk indium, no consumption of oxygen was detected. On
In(hl)/Si, only one peak is observed, with a maximum at 612
K, and the amount of oxygen consumed is very close to the
expected amount (1.5 mol oxygen per mol of generated indium
oxide). The In/Al sample presents two different peaks. The
oxidation begins at lower temperatures than for In(hl)/Si, but
the second and more important peak has its maximum at the
same temperature as for the In(hl)/Si sample. The XRD spectrum
obtained on a sample where TPO was stopped at 503 K (Figure
6) shows the peaks corresponding to indium oxide and metallic
indium, suggesting that, as in the case of reduction, the oxidation
peak at low temperature is probably due to the smallest indium
crystallites and not to intermediate oxidized species. The In-
(hl)/Ti sample presents only one peak, larger and with an onset
and a maximum at higher temperatures than the two other
samples. This difference can hardly be attributed to differences
in the size of indium particles, because according to XRD, the
particle sizes must be similar; so in some way titania stabilizes
metallic indium. The oxidation process is more affected by the
support than the reduction, where the main differences are
related to particle size. The temperature of the maximum in
TPO experiments is lower than in TPR experiments, even if
after TPR experiments indium particles have been synthesized.
For In(hl)/Al and In(hl)/Ti, a second TPR experiment was
carried out after TPO; the profile of this second TPR is
represented in Figure 7. Comparing this profile with the TPR
carried out on the fresh sample (Figure 3b), a huge change is
Figure 8. Heat flow profiles of reduction of In2O3 and supported
indium oxide samples.
observed, especially for the In(hl)/Ti sample. The reduction peak
that took place at low temperature in the first TPR run has
disappeared in the second run, where reduction occurs only at
high temperature. For the In(hl)/Ti sample the reduction peak
temperature changes from 910 to 979 K, with a profile quite
similar to the TPR of bulk indium or In/Si. The XRD pattern
after the TPR experiments (Figure 4) shows very narrow peaks,
typical of metallic indium, indicating that after reduction at high-
temperature crystalline indium has been formed, as the behavior
of the supported indium oxide becomes closer to bulk indium
oxide. At the end of these experiments, a decrease in surface
area was observed, confirming the loss of dispersion. As in the
TPR on the fresh sample, the consumption of hydrogen is very
close to the expected 3 mol of H2 per mol of In2O3, indicating
that the process of reduction/oxidation of indium oxide is
reversible.
In addition to TPR-TPO, experiments of oxidation and
reduction have been carried out on the high loading samples
by TG-DSC. The heat flow profiles for reduction experiments
are presented for all samples in Figure 8. Though these results
could not be compared quantitatively with TPR experiments
because of the different experimental conditions (reactant
mixture composition, reactant gas flow, etc.), the results are
qualitatively the same. As in TPR, In(hl)/Ti is the most easily
reduced sample, followed by In(hl)/Al, In(hl)/Si, and bulk
indium oxide. In all cases the extent of reduction estimated from
the mass loss (0.17 g per gram of In2O3 reduced) is very close
to the expected extent of reduction (see Table 4), suggesting
that the total reduction of indium oxide takes place. In some
cases the mass loss is larger than expected, probably due to
water loss. Even if the main loss of water occurs up to 413 K,
the most strongly adsorbed water could be retained up to 700

246 J. Phys. Chem. B, Vol. 110, No. 1, 2006
Gervasini et al.
TABLE 4: Mass Variation and Evolved Heat in the
Reduction/Oxidation Process of Supported Indium Oxide
Samples
oxygen content atmosphere is dramatically dependent on the
support. Figures 10 and 11 show the results obtained for the
reaction performed in the 473-773 K temperature range on the
In2O3 phases dispersed over alumina, titania, niobia, and silica
supports. The results for the unsupported In2O3 bulk oxide are
also presented in Figures 10 and 11 for comparison.
reduction
heat
massa % kJ (mol In2O3)-1 massa % kJ (mol In)-1
oxidation
heat
sample
For each catalytic system, both the NO and NO2 conversions
(Figure 10a-e) and the C2H4 conversions to CO and CO2
(Figure 11a-e) are shown. As a general trend, the reactivity of
NO2 is higher than that of NO. The values of the NO2 conversion
attained 70% over In2O3 supported on alumina, about 40-50%
over In2O3 supported on titania or silica, and about 30-40%
over bulk In2O and In2O3 supported on niobia. The curves of
NO2 conversion as a function of temperature display a typical
trend. At low temperatures (below 573 K), there is a rapid
increase of NO2 conversion with temperature. As it is not
possible to observe a parallel high ethene conversion, the very
high NO2 consumptions observed could be ascribed to nitrite
and/or nitrate formation on the catalyst surfaces. The observed
behavior is in agreement with some recent literature findings.26
For higher temperatures, two different behaviors of the NO2
conversion-temperature curves could be observed. The In2O3
phases on alumina, titania, and silica display a more or less
regular increasing trend of NO2 conversion with temperature,
attaining a maximum value in the cases of In/Al and In/Ti but
not in the case of In/Si. This second increase in NO2 conversion
with temperature could be associated with NO2 reduction to
N2, as in parallel a high C2H4 conversion to CO2 was observed
(Figure 11a-e). Meanwhile, on the niobia-supported system,
no increasing trend of NO2 conversion could be observed in
the entire temperature range investigated.
In2O3
109
107
107
91
171.4
95.6
108.9
64.9
In(hl)/Si
In(hl)/Al
In(hl)/Ti
110
102
94
-284.9
-204.8
-39.5
a Ratio between the experimental mass loss/gain during reduction/
oxidation and the corresponding to theoretical mass variation (%).
Figure 9. Heat flow profiles of oxidation of supported indium samples.
K. The measured heats are in the order In2O3 > In(hl)/Al >
In(hl)/Si > In(hl)/Ti. It is useful to remember that the measured
heat includes not only the reduction heat but also the melting
of metallic indium. The melting point of metallic indium is 426.6
K,21 which is lower than the reduction onset in all cases, so
reduction and fusion must take place at the same time. This
fact is well observed in bulk indium oxide, for which the cooling
ramp features a very narrow exothermic peak without associated
mass variation and centered at 427.2 K, indicating the solidifica-
tion of metallic indium. For the other samples this peak does
not occur (Figures not shown).
No oxidation is observed in the case of reduced bulk indium.
For the three other samples the oxidation gives rise to an
exothermic peak (Figure 9). The same behavior is observed as
in the TPO experiments: In(hl)/Al presents two oxidation peaks
and the lowest onset temperature of the three samples, while
In(hl)/Si and In(hl)/Ti present only one oxidation peak, at a
higher temperature for In(hl)/Ti. Contrary to reduction experi-
ments, the melting point of indium is seen for all three samples,
always at the same temperature (430.9 K), indicating the absence
of influence of the support on the melting point, probably due
to the sintering of metallic indium particles. It is interesting to
point out that in the case of In(hl)/Al reoxidation takes places
before and during the fusion of metallic indium, while for In/
Si and especially In/Ti the metallic indium melts before
oxidation.
The curves of NO conversion vs temperature follow a
different trend from those of NO2 conversion, in particular in
the low-temperature range (below 573 K) where no increase of
the NO conversion could be observed. In the 573-823 K
temperature interval, a noticeable increase in the NO conversion
was observed for the catalysts on the alumina, titania, and silica
supports, but not on the niobia support, in agreement with the
behavior of the NO2 conversions. In this case too, the NO
conversion could be associated with N2 formation due to the
high C2H4 consumption observed.
C2H4 was selectively oxidized by NO and NO2 to CO2. Only
very limited amounts of CO were detected over the various
catalytic systems (Figure 11a-e). The CO yields decreased even
further at temperatures higher than 773 K. The highest amounts
of CO formation were observed over In2O3 supported on niobia
and silica. On these systems the CO2/CO ratio in the 773-823
K interval was around 5-20, while on the more active catalysts
(with alumina and titania supports) the CO2/CO ratio was as
high as 15-40. In any case, quantitative C2H4 conversion to
CO2 was never observed. The highest CO2 yields were observed
over In/Al, In/Ti, and In/Si (about 70%). In/Nb was poorly
active, both for reducing NOx and for oxidizing C2H4. Bulk
In2O3 had a marked oxidizing character, as demonstrated by
the high amount of CO2 formed.
The measured heats in the reduction process follow the order
In2O3 . In(hl)/Al = In(hl)/Si > In(hl)/Ti, while in the oxidation
process the order found is In(hl)/Si > In(hl)/Al . In(hl)/Ti. It
is interesting to note that the measured oxidation heat for the
In(hl)/Ti sample is much lower than for the two other samples,
even though the weight increase is very similar to that expected
for the total oxidation. This fact and the higher oxidation
temperature suggest that the oxidation of metallic indium on
this sample occurs in a different manner from the two other
samples.
The most selective catalysts were the In/Al and In/Ti systems.
They could be distinguished from the others by a higher ratio
between the amount of NOx reduction to N2 and the amount of
C2H4 oxidation to CO2 (compare Figures 10 and 11). This so-
called reaction selectiVity is due to the coexistence of two main
parallel reactions in the HC-SCR process.27-28 The two oxidant
species O2 and NOx are present simultaneously in the feed
mixture, and they compete for the oxidation of ethene. Because
the amount of O2 is much higher than that of NOx in the feed,
de-NOx Catalytic Activity. The activity of the indium-
containing catalysts in the reduction of NOx by C2H4 in high

Oxide Supported In2O3 Catalysts
J. Phys. Chem. B, Vol. 110, No. 1, 2006 247
Figure 10. Conversion of NOx as function of reaction temperature for (a) bulk indium oxide and the high (filled symbols) and low (open symbols)
loading indium oxide catalysts supported on: (b) silica, (c) alumina, (d) titania, (e) niobia.
the combustion of the hydrocarbon naturally occurs with a
higher rate than the parallel reaction between C2H4 and NOx.
Selective catalytic systems are able to favor the involvement
of the hydrocarbon in the NOx reduction reaction, limiting the
hydrocarbon combustion. Unlike the supported catalysts, bulk
In2O3 showed very poor selectivity.
when In2O3 was well dispersed on the surface of the support,
high activity could be observed, as in the case of alumina and
titania.
Figure 12 gives a comparative view of the collected results,
in terms of N2 yield over the various studied catalysts. The better
activity of the alumina and titania supports clearly emerges. No
remarkable differences could be detected between the bulk In2O3
and the In2O3 systems supported on niobia and silica. It can be
inferred that on these supports In2O3 has not been well dispersed;
this can be due to the surface properties of these oxides. Silica
is very poorly acidic and niobia is an acidic oxide possessing
poor amphoteric properties. The balanced presence of basic and
acidic sites on the support surfaces could play a key role in
stabilizing the dispersed In2O3 particles, limiting their aggrega-
tion into large indium islands, which are not as active in
reducing NOx.
It is interesting to observe the influence of the In loading on
the activity and selectivity of the catalysts for the different
supports. As regards C2H4 combustion, all the catalytic systems
followed a general trend: the high In loading systems were
much more active than those with low In loadings. However,
no general common behavior could be observed for the NOx
conversion. The high loading In2O3 catalysts supported on
alumina and silica, In(hl)/Al and In(hl)/Si, had a higher activity
for NOx reduction than the corresponding low In2O3 loading
systems. The In2O3 systems supported on titania and niobia had
an opposite behavior. The observed behavior is difficult to
explain; it could be related to differences in the In2O3 dispersion
on the surface of the different supports. Small amounts of very
poorly dispersed In2O3 gave rise to a low activity, as evidenced
by the niobia- and silica-supported systems. On the contrary,
Because of the incomplete coverage of the supports, both the
metal oxide and support sites might participate to the complex
reaction mechanism of selective reduction of NOx. Following
the reaction pathway recently proposed for Ga2O3/Al2O3 and
In2O3/Al2O3 systems by Haneda et al.,3 the surface species of

248 J. Phys. Chem. B, Vol. 110, No. 1, 2006
Gervasini et al.
Figure 11. Conversion of C2H4 to COx as function of reaction temperature for (a) bulk indium oxide and the high (filled symbols) and low (open
symbols) loading indium oxide catalysts supported on: (b) silica, (c) alumina, (d) titania, (e) niobia.
Figure 12. Comparison of N2 yields obtained for the low and high In loading catalysts over the different supports and bulk indium oxide at 823
K.
the support can be primarily involved in nitrate formation, which
can explain the low NO2 (and NO) conversion on silica and
niobia supports. The surface nitrate species can react with the
hydrocarbon, leading to thermally unstable organic nitro and

Oxide Supported In2O3 Catalysts
J. Phys. Chem. B, Vol. 110, No. 1, 2006 249
(5) Serban, M.; Halasz, I.; Datta, R. Catal. Lett. 1999, 63, 217.
(6) Maunula, T.; Kintaichi, Y.; Inaba, M.; Haneda, M.; Sato, K.;
Hamada, H. Appl. Catal. B 1998, 15, 291.
nitrile compounds that can be oxidized by O2 or reduced by
NOx, leading to the final N2 product. On the other hand, nitrate
species formed on dispersed In2O3 species can act as strong
oxidant for the hydrocarbon, realizing a direct red-ox reaction
justifying that the poorly dispersed In2O3 particles are less active
than the highly dispersed ones.
(7) Requejo, F. G.; Ramallo-Lo´pez, J. M.; Lede, E. J.; Miro´, E. E.;
Pierella, L. B.; Annunziata, O. A. Catal. Today 1999, 54, 553.
(8) Schmidt, C.; Sowade, T.; Schu¨tze, F.-W.; Richter, M.; Berndt, H.;
Gru¨nert, W. Stud. Surf. Sci. Catal. 2001, 135, 4973.
(9) Schu¨tze, F.-W.; Berndt, H.; Richter, M.; Lucke, B.; Schmidt, C.;
Sowade, T.; Gru¨nert, W. Stud. Surf. Sci. Catal. 2001, 135, 1517.
(10) Miyadera, T.; Yoshida, K. Chem. Lett. 1993, 1483.
(11) Hamada, H.; Kintaichi, Y.; Sasaki, M.; Ito, T.; Tabata, M. Appl.
Catal. 1991, 75, L1.
(12) Tabata, T.; Hamada, H.; Suganuma, F.; Yoshinari, T.; Tsuchida,
H.; Kintaichi, Y.; Sasaki, M.; Ito, T. Catal. Lett. 1994, 25, 55.
(13) Hamada, H.; Kintaichi, Y.; Yoshinari, T.; Tabata, M.; Sasaki, M.;
Ito, T. Catal. Today 1993, 17, 112.
Conclusions
The selective reduction of NOx by hydrocarbons (HC-SCR)
was studied over four different catalytic systems prepared by
dispersing various amounts of In2O3 over different supports.
Indium oxide can be considered as a basic solid. As a general
rule, the global acidity was decreased and the basicity increased
upon deposition of indium oxide on the support.
(14) Subramanian, S.; Kudla, R. J.; Chun, W.; Chattha, M. S. Ind. Eng.
Chem. Res. 1993, 32, 1805.
Indium oxide is reversibly reduced and oxidized; the tem-
perature of reduction is mostly related to the particle size,
whereas oxidation is more influenced by the support. The
oxidation of indium oxide take places at a lower temperature
than the reduction. The In(hl)/Ti sample presents the highest
oxidation temperature, indicating that titania stabilizes metallic
indium.
Alumina and titania were found to be the best supports in
order to prepare active and selective de-NOx catalysts. The better
stabilization of In2O3 by titania, detected by XRD and TPR/
TPO profiles, could explain the higher activity in the reaction
of NOx reduction by hydrocarbons observed for the low loading
sample In(3)/Ti. More generally, supports that are able to
disperse the In2O3 aggregates with high In stabilization could
give rise to active catalytic systems.
(15) Perdigon-Melon, J. A.; Gervasini, A.; Auroux, A. J. Catal. 2005,
234, 421.
(16) Haneda, M.; Kintaichi, Y.; Inaba, M.; Hamada, H. Appl. Surf. Sci.
1997, 121/122, 391.
(17) Kintaichi, Y.; Haneda, M.; Inaba, M.; Hamada, H. Catal. Lett. 1997,
48, 121.
(18) Ren, L.; Zhang, T.; Tang, J.; Zhao, J.; Li, N.; Lin, L. Appl. Catal.
B 2003, 41, 129.
(19) Innes, W. B. In Catalysis, Vol. I, Fundamental principles (Part I);
Emmett, P. H. Ed.; Reinhold Publishing Corporation: New York, 1954;
Chapter 6, p 258.
(20) Auroux, A. Top. Catal. 1997, 4, 71. Auroux, A. Top. Catal. 2002,
19, 205.
(21) Handbook of Chemistry and Physics, 70th ed; Weast, R. C.; Lide,
D. R.; Astle, M. J.; Beyer W. H., Eds.; CRC Press: Boca Raton, 1989-
1990.
(22) Poznyak, S. K.; Golubev, A. N.; Kulak, A. I. Surf. Sci. 2000, 545-
456, 396.
(23) Choudhary, V. R.; Jana, S. K.; Kiran, B. P. J. Catal. 2000, 192,
257.
(24) Miha´lyi, R. M.; Beyer, H. K.; Mavrodinova, V.; Minchev, Ch.;
Neinska, Y. Micropor. Mesopor. Mater. 1998, 24, 143.
(25) Petre, A. L.; Perdigo´n-Melo´n, J. A.; Gervasini, A.; Auroux, A.
Catal. Today 2003, 78, 377.
(26) Haneda, M.; Kintaichi, Y.; Bion, N.; Hamada, H. Appl. Catal. B
2003, 42, 57.
(27) Gervasini, A.; Carniti, P.; Ragaini, V. Appl. Catal. B 1999, 22,
201.
References and Notes
(1) Berndt, H.; Schu¨tze, F.-W.; Richter, M.; Sowade, T.; Gru¨nert, W.
Appl. Catal. B 2003, 40, 51.
(2) Park, P. W.; Ragle, C. S.; Boyer, C. L.; Balmer, M. L.; Engelhard,
M.; McCready, D. J. Catal. 2002, 210, 97.
(3) Haneda, M.; Joubert, E.; Me´ne´zo, J. C.; Duprez, D.; Barbier, J.;
Bion, N.; Daturi, M.; Saussey, J.; Lavalley, J. C.; Hamada, H. J. Mol. Catal.
A 2001, 175, 179.
(4) Sulikowski, B.; Kubacka, A.; Wloch, E.; Schay, Z.; Corte´s
Corbera´n, V.; Valenzuela, R. X. Stud. Surf. Sci. Catal. 2000, 130, 1889.
(28) Carniti, P.; Gervasini, A.; Modica, V. H.; Ravasio, N. Appl. Catal.
B 2000, 28, 175.