TiO2-supported metal oxide catalysts for low-temperature selective catalytic reduction of NO with NH3: I. Evaluation and characterization of first row transition metals

Article abstract of DOI:10.1016/j.jcat.2003.09.003

Low-temperature SCR of NO with NH₃ in the presence of excess oxygen on the oxides of V, Cr, Mn, Fe, Co, Ni, and Cu supported on anatase TiO₂ has been studied. Among the catalysts tested, Mn/TiO ₂ supported on Hombikat TiO₂ provided the best performance with 100% N₂ selectivity and complete NO conversion at temperatures as low as 393 K under numerous conditions. The catalytic performance for various transition metal oxides supported on TiO₂ decreased in the following order: Mn > Cu ≥ Cr ? Co > Fe ? V Ni. For Mn-based catalysts the activity increases with an increase in Mn loading and the reaction temperature. TiO₂ alone did not give any NO conversion at ≤573 K, and calcination at low temperature (≤673 K) is preferable. XRD coupled with XPS confirmed the presence of MnO₂ as a major phase (peak at 642.2 eV) with Mn₂O₃, and partially undecomposed Mn-nitrate as the minor phases for supported manganese catalysts. It is proposed that MnO₂ contributes to the high activity of Mn/TiO₂. XPS results also confirmed a higher concentration of active metal oxides on the surface of Mn/TiO₂ compared to the other catalysts. The NH₃ FT-IR study showed the presence of Lewis acid sites for the most active catalysts, while the peak corresponding to Broensted acid sites was weak or absent. This strongly suggests that Broensted acid sites are not necessary for the reaction to occur at low temperatures. The H₂ TPR study indicated the difficulty of reducing Mn oxide when the metal loading is low and/or the catalysts are calcined at temperatures higher than 773 K. It is concluded that lower catalyst calcination temperatures, Lewis acidity, the redox properties of metal oxides and their higher surface concentration are important for very high SCR activity at low temperatures. Mn/TiO₂ provided the best performance at 50,000 h ^-1 when the catalysts were tested in the presence of 11 vol% H ₂O. Under these conditions, the catalytic activity of the transition metal oxides decreases in the following order: Mn > V ? Co > Cu > Cr > Fe ? Ni.

Full text of DOI:10.1016/j.jcat.2003.09.003

Journal of Catalysis 221 (2004) 421–431
www.elsevier.com/locate/jcat
TiO -supported metal oxide catalysts for low-temperature selective
2
catalytic reduction of NO with NH3
I. Evaluation and characterization of first row transition metals
1
∗
Donovan A. Peña, Balu S. Uphade, and Panagiotis G. Smirniotis
Chemical and Materials Engineering Department, University of Cincinnati, Cincinnati, OH 45221-0012, USA
Received 29 April 2003; revised 28 July 2003; accepted 3 September 2003
Abstract
Low-temperature SCR of NO with NH in the presence of excess oxygen on the oxides of V, Cr, Mn, Fe, Co, Ni, and Cu supported
3
on anatase TiO has been studied. Among the catalysts tested, Mn/TiO supported on Hombikat TiO provided the best performance with
2
2
2
1
00% N selectivity and complete NO conversion at temperatures as low as 393 K under numerous conditions. The catalytic performance for
2
various transition metal oxides supported on TiO decreased in the following order: Mn > Cu ꢀ Cr ꢀ Co > Fe ꢀ V ≫ Ni. For Mn-based
2
catalysts the activity increases with an increase in Mn loading and the reaction temperature. TiO alone did not give any NO conversion at
2
ꢁ
573 K, and calcination at low temperature (ꢁ 673 K) is preferable. XRD coupled with XPS confirmed the presence of MnO as a major
2
phase (peak at 642.2 eV) with Mn O , and partially undecomposed Mn-nitrate as the minor phases for supported manganese catalysts. It is
2
3
proposed that MnO contributes to the high activity of Mn/TiO . XPS results also confirmed a higher concentration of active metal oxides
2
2
on the surface of Mn/TiO compared to the other catalysts. The NH FT-IR study showed the presence of Lewis acid sites for the most active
2
3
catalysts, while the peak corresponding to Brönsted acid sites was weak or absent. This strongly suggests that Brönsted acid sites are not
necessary for the reaction to occur at low temperatures. The H TPR study indicated the difficulty of reducing Mn oxide when the metal
2
loading is low and/or the catalysts are calcined at temperatures higher than 773 K. It is concluded that lower catalyst calcination temperatures,
Lewis acidity, the redox properties of metal oxides and their higher surface concentration are important for very high SCR activity at low
−1
temperatures. Mn/TiO provided the best performance at 50,000 h when the catalysts were tested in the presence of 11 vol% H O. Under
2
2
these conditions, the catalytic activity of the transition metal oxides decreases in the following order: Mn > V ꢀ Co > Cu > Cr > Fe ꢀ Ni.

2003 Elsevier Inc. All rights reserved.
Keywords: Low-temperature SCR; Ammonia; Nitric oxide reduction; Titania; Transition metal oxides
1
. Introduction
occurring is
NO + 4NH3 + O2 → 4N2 + 6H2O.
The well-known industrial catalyst for this process is
4
(1)
Nitrogen oxides (NOx) emitted from automobiles and
stationary sources such as oil- and coal-fired power plants,
waste incinerators, industrial ovens, and chemical processes
based on V O /TiO (anatase) promoted with either WO
3
2
5
2
(
i.e., nitric acid production plants) are major causes for acid
or MoO [4–9]. The required operating temperature for the
3
rain, smog, and ozone depletion [1–3]. Selective catalytic re-
duction (SCR) of NOx to N2 using NH3 as a reductant is now
considered to be the most effective process for the treatment
of stack gases from stationary sources. The general reaction
above industrial catalyst is typically 573–673 K. This makes
it necessary to locate the SCR unit upstream of the de-
sulfurizer and/or particulate control device in order to avoid
reheating the flue gas. Despite the low sensitivity of this ca-
talyst to SO2 poisoning, deactivation does occur. The life of
the catalyst is shortened because of high concentrations of
SO2 and ash in the flue gas. This can be avoided by locating
the SCR reactor downstream in the tail-end configuration,
where the flue gas is relatively clean after passing through
*
Corresponding author.
E-mail address: Panagiotis.Smirniotis@UC.EDU (P.G. Smirniotis).
Current address: Chemical Engineering Division, National Chemical
1
Laboratory, Pune 411 008, India.
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2003.09.003

422
D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
the desulfurizer and particulate removal device. This strat-
egy would be especially advantageous for high-sulfur coals,
as the removal of SO2 before NOx removal would minimize
both the catalyst deactivation and the detrimental oxidation
of SO2 to SO3. However, the flue gas temperature at this
location drops below 433 K, which makes it necessary to
reheat the flue gas back to the SCR operating temperature.
This makes the overall operation very expensive. One solu-
tion to this problem is the development of an inexpensive
low-temperature SCR system that operates at ꢁ 423 K. This
would eliminate the need to reheat the stack gas and increase
the operational lifetime of the catalyst relative to existing
configurations, which results in a lower investment and also
a thermally more efficient process.
2. Experimental
2.1. Catalyst preparation
The TiO2 used in this study was Hombikat UV 100 from
Sachtleben Chemie. As determined by N2 adsorption, it
2
had a specific surface area of 309 m /g, a pore volume of
3
0
.37 cm /g, and a pore diameter of 4.5 nm. The transition
metal oxides were deposited using aqueous solutions of their
nitrates by the solution impregnation method as previously
reported [25,26]. In one of the catalyst preparation experi-
ments, manganese acetate was also used as the precursor for
comparison. In a typical synthesis, 50 ml of deionized water
was added to a 100-ml beaker containing 1.0 g of support.
The mixture was heated to 343 K under continuous stirring.
A measured quantity of nitrate or acetate precursor was then
added, and the mixture was evaporated to dryness. The paste
obtained was further dried overnight at 383 K, ground, and
sieved (80–120 mesh). The calcination of the catalysts was
2
Supported transition metal oxides, such as MnOx/Al2O3
[
10–12], CuO/Al2O3 [13], CuO/TiO2 [14], Fe2O3/TiO2 [15],
and V2O5/activated carbon [16–18], are shown to be active
in medium temperature SCR of NO with NH3 in the pres-
ence of excess oxygen. Recent efforts have been made to
develop robust catalysts for low-temperature SCR of NO
using NH3 [19,20]. Industrial researchers have particularly
shown an interest in this area. Taking into account the var-
ious advantages of low-temperature operation, a process
based on carbon as a catalyst has been developed (Bergbau–
Farschuang and Mitsui) for simultaneous removal of H2S,
SO2, and NOx from industrial process gases and off gases.
The reaction products of H2S and SO2 are adsorbed on the
carbon catalyst whereas NOx is reduced by NH3 accord-
ing to reaction (1) at ꢀ 363 K. However, carbon has the
danger of getting consumed slowly [reaction (3)], thereby
decreasing steadily the activity and available capacity of car-
bon during operation [21].
performed at 673 K for 2 h in a flow of O (4.0% in helium).
2
.2. Characterization of supports and catalysts
Prior to characterization, the samples were calcined at
73 K for 2 h in O2 (4.18% in He) at a total flow rate of
0 ml/min. The catalysts were stored in a vacuum oven
6
2
maintained at 328 K and 30 inches Hg of vacuum.
X-ray diffraction was used to identify the anatase and
rutile phases of titania and the crystalline phases of any tran-
sition metal oxides deposited on titania. XRD studies were
performed on a Siemens D500 diffractometer with a Cu-Kα
radiation source (wavelength 1.5406 Å). Aluminum holders
were used to support the samples in XRD measurements.
Anatase and rutile phases of titania were determined from
2
SO2 + O2 + 2H2O → 2(H2SO4)ads,
(2)
(3)
◦
the strongest peaks corresponding to anatase [2θ = 25.3 for
◦
H2SO4 + C → 2SO2 + 2H2O + CO2.
the (101) reflection of anatase] and rutile [2θ = 27.5 for
the (110) reflection of rutile]. A combination of Raman and
XRD results was used to determine the crystalline phases of
titania. Self-supported pellets (100–150 mg) were used for
the Raman study, which was collected using an Ar laser at
Shell [22] and Rhône-Poulenc [23] also have developed V-
based supported catalysts that are claimed to operate in the
temperature range of 313–523 K. The good low-temperature
performance of Shell’s industrial processes is attributed to
a better reactor design (lateral flow) and an improved cat-
alyst preparation method using promoters such as Fe, Mo,
and Co.
This paper describes work performed in our labora-
tory [24,25] in an effort to find active catalysts for the SCR
of NO at low temperatures (ꢁ 423 K). Catalysts were pre-
pared from various transition metal oxides (V, Cr, Mn, Fe,
Co, Ni, and Cu) deposited by the solution impregnation
method using nitrate or acetate precursors on titania. It is
shown here that MnOx/TiO2 catalysts can operate effec-
tively below 423 K with both high NO conversion and N2
selectivity. In Part II of this work, in situ FT-IR studies have
been used extensively to develop a low-temperature SCR re-
action mechanism.
2
mW under ambient conditions.
The specific surface areas were measured by nitrogen ad-
sorption at 77 K by the BET method using a Micromeritics
Gemini 2360. Pore-size measurements were performed us-
ing a Micromeritics ASAP 2010. Catalysts were pretreated
in situ at 523 K under vacuum, and the analysis of the sup-
ports was done at 77 K under vacuum.
X-ray photoelectron spectroscopy (XPS) was used to an-
alyze the atomic surface concentration on each catalyst. The
spectra were recorded on a Perkin-Elmer Model 5300 X-
ray photoelectron spectrometer using Mg-K (1253.6 eV)
as a radiation source at 300 W. The spectra were recorded in
the fixed analyzer transmission mode with pass energies of
89.45 and 35.75 eV for recording survey and high-resolution
spectra, respectively. The powdered catalysts were mounted
α

D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
423
onto the sample holder and degassed overnight at room tem-
heating rate of 5 K/min and finally keeping the temperature
constant for 1 h at 1173 K to ensure complete metal oxide re-
duction. The hydrogen consumed in the TPR was measured
by a TCD.
−7
perature at a pressure on the order of 10 Torr. Binding
energies (BE) were measured for C 1s, O 1s, Ti 2p, V 2p,
Mn 2p, Cr 2p, Ni 2p, and Cu 2p. Since the tape used to
mount the sample was a source of carbon contamination, the
surface composition of carbon is not included in the tables.
Recorded Auger spectra for Mn, Cr, and Ni were very weak.
Sample charging effects were eliminated by correcting the
observed spectra with the Ti 2p3/2 binding energy value of
2.3. Catalytic experiments
The SCR of NO at atmospheric pressure was carried out
in a fixed-bed quartz reactor (i.d. 6 mm) containing calcined
catalyst (80–120 mesh). Oxygen (Wright Bros., 4.18% in
He), ammonia (Matheson, 3.89% in He), and nitric oxide
(Air Products, 2.0% in He) were used as received. The re-
action temperature was measured by a type K thermocouple
inserted directly into the catalyst bed. Experiments were per-
formed under two different reaction conditions. While the
concentration of O2 remained constant (2 vol%), the reactant
concentrations and GHSVs varied between the two cases. In
4
58.5 eV. An estimated error of 0.1 eV can be considered
for all the measurements. The spectra were smoothed, and a
nonlinear background was subtracted. Peaks were deconvo-
luted, and the fitting was performed using a convolution of
Gaussian and Lorenzian curves. Quantification was accom-
plished by determining the elemental peak areas following
a Shirley background subtraction. Quantitative analysis was
carried out using the sensitive factors supplied with the in-
strument.
FT-IR and NH3-TPD were used to characterize and quan-
tify the acidity of the catalysts, respectively. The FT-IR ex-
periments were performed with a Bio-Rad spectrophotome-
ter (FTS-040). Circular self-supporting thin wafers (8 mm
diameter) prepared using 6–8 mg catalyst were used for the
study. The wafers were placed in a high-temperature cell
with AgBr windows and purged with prepurified grade he-
lium (30 ml/min) at 673 K for 2 h to remove any impurities
before cooling down to 323 K. NH3 (4.4 vol% in He) was in-
troduced to the cell and continued flowing (30 ml/min) for
a total of 1 h at 323 K to ensure complete saturation of the
sample. Helium (30 ml/min) was then replaced by NH3 and
the temperature was increased to 373 K. Physisorbed ammo-
nia was removed by flushing the wafer with He for at least
−1
the first case at 8000 h , the inlet concentrations of NO
and NH3 were each 2000 ppm. The reactants and products
were analyzed on-line using a Quadrapole mass spectrom-
−1
eter (MKS PPT-RGA). In the second case at 50,000 h
,
the inlet concentrations of NO and NH3 were each set at
400 ppm. The reactants and products were analyzed using
a GC (GOW-MAC) with Porapak Q and Carboxen 1000
columns and a NOx analyzer (Eco Physics CLD 70S). The
definitions used for NO conversion and nitrogen selectivity
are as follows:
NO conversion = 100 × (NOin − NOout)/NOin,
(4)
(5)
ꢀ
ꢁ
N2 selectivity = 100 × N2/(N2 + N2O) .
4
h. Subsequently, the FT-IR spectra were recorded by des-
3. Results and discussion
orbing NH3 at 373 K. All the spectra were recorded at 373 K
and normalized to avoid inconsistencies in the results.
TPD of NH3 was used to measure the total acidity of
the catalysts. The experiments were performed on a custom-
made TCD setup using 50 mg of catalyst. Prior to the ex-
periments the catalysts were pretreated at 773 K for 1 h
in a highly pure He (30 ml/min) stream. The furnace tem-
perature was lowered to 373 K, and the samples were then
treated with anhydrous NH3 (4.05% in He) at a flow rate of
3.1. Surface area and pore-size distribution
The BET surface area values of the various metal oxides
deposited on TiO2 are compared in Table 1. The results show
the expected drop in surface area upon impregnation of the
transition metal oxides on TiO2. The drop was most severe
when vanadia was deposited. It is reported in the literature
that V2O5/TiO2 (anatase) is quite unstable and that vanadia
facilitates transformation of the metastable anatase phase of
titania into the thermodynamically more stable rutile phase
at any temperature and pressure [27]. Vanadia also favors the
sintering of anatase along with the corresponding decrease
in surface area [28]. Therefore, to retard the phase transi-
tion and the loss of surface area, WO3 or MoO3 is added
to commercial V2O5/TiO2 (anatase) catalysts. The drop in
surface area was not very severe when metal oxides of Mn,
Cr, and Cu were deposited. This suggests that impregnation
with Mn, Cr, and Cu probably inhibits the sintering and/or
phase transformation of the support and also enhances dis-
persion of transition metal oxides on the supports. In fact,
the effect of Mn loading on TiO2 suggests that overall sur-
face area improves with an increase in metal loading on the
3
0 ml/min for 1 h. Physisorbed NH3 was removed by flush-
ing the catalyst with He at 373 K for 3–5 h before starting
the TPD experiments. Experimental runs were recorded by
heating the sample in He (30 ml/min) from 373 to 773 K at
a heating rate of 5 K/min and finally keeping the tempera-
ture constant for 1 h at 773 K to ensure complete ammonia
desorption.
The H2-TPR experiments were performed on a custom-
made setup using 100 mg of calcined catalyst. The cata-
lysts were pretreated at 673 K for 1 h in a highly pure He
(
25 ml/min) stream. The furnace temperature was lowered
to 373 K, and then the feed containing 6.0 vol% H2 in He
was fed at a flow rate of 25 ml/min. H2-TPR runs were per-
formed by heating the sample from 373 to 1173 K at a linear

424
D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
Table 1
Properties of catalysts used in the present study
a
Catalyst
Loading
SSA
(m /g)
XRD phases
observed
Total acidity
(µmol/g)
2
(wt%)
TiO
n/a
20
20
5
10
12.5
12.5
15
20
25
309
51
Anatase
None
n.d.
9.8
2
b
V
Cr
c
143
n.d.
n.d.
173
165
n.d.
204
n.d.
169
160
163
140
Cr O
2
55.8
n.d.
n.d.
n.d.
n.d.
n.d.
23.2
n.d.
33.6
31.6
42.7
12.5
3
c
Mn
Mn
Mn
Mn
Mn
Mn
Mn
n.d.
n.d.
n.d.
n.d.
n.d.
c
c
d
c
c
c
Mn O
2
3
n.d.
Fig. 1. XRD spectra of 20 wt% transition metals supported on Hombikat
c
Fe
20
20
20
20
Fe O
2
3
TiO (A, anatase; ✩, NiO; P, MnO ; e, Co O ).
2
2
3 4
c
Co
Co O
3
4
c
Ni
NiO
CuO
c
Cu
n.d., not determined; n/a, not applicable.
a
b
c
d
◦
Calcined at 400 C for 2 h in O + He flow.
2
Ammonium metavanadate + oxalic acid.
Nitrate precursor.
Acetate precursor.
support. In conclusion, the results in Table 1 clearly indicate
that with the exception of V2O5/TiO2 catalyst, the surface
area of other transition metals supported on TiO2 catalysts is
higher.
3
3
.2. X-ray diffraction and Raman spectroscopy
.2.1. TiO2
The phase compositions of titania determined from XRD
and Raman spectroscopy results are described below. Only
anatase is observed from the XRD results. As reported in
the literature [29], five peaks from Raman spectroscopy near
−1
6
38, 515, 396, 196, and 144 cm that correspond to anatase
were observed (not shown). The broadening of the peak
−1
at 144 cm in Hombikat TiO2 also indicates the possible
presence of the brookite phase of TiO2, as its two most in-
−
1
tense peaks (127 and 150 cm ) should be in the vicinity of
Fig. 2. Raman spectra of 20 wt% Mn/TiO , Mn O , and Mn O collected
at ambient conditions at 2 mW with an Ar laser.
2
3
4
2 3
−1
1
3
2
44 cm
.
.2.2. Transition metal oxides
The XRD patterns of the catalysts calcined at 673 K for
h are shown in Fig. 1, and the observed crystalline XRD
be detected for Mn in deconvoluted XPS spectra (MnO and
2
Mn O ), indicating the presence of these mixed metal oxide
2
3
phases over the TiO supports. It is well known that Mn O
3
2
2
phases for these catalysts are also tabulated in Table 1. None
of the XRD spectra presented here give intense or sharp
peaks for transition metal oxides despite having relatively
high metal concentrations (20.0 wt%) on the titania sup-
port. This may be due to the high dispersion and/or low
crystalline nature (amorphous) of the metal oxides on the
is more amorphous than MnO [30,31], indicating the diffi-
culty of its detection by XRD. The Raman spectra presented
in Fig. 2 could not help identify the different phases of man-
2
ganese oxides, as the spectrum for Mn/TiO is dominated
2
by the titania peaks discussed above. However, the Raman
spectra for the catalyst Mn/TiO as shown in Fig. 2 show
2
◦
support. MnO2 at 2θ values of 28.9, 37.7, and 57.1 with
the broadening of the peaks throughout the scanned region,
−1
relative intensities of 100, 55, and 55%, respectively (JCPDS
File No. 24-0735), was observed as the only phase for sup-
ported manganese oxide catalysts. The crystalline phase of
Mn2O3 (Fig. 1) could not be observed; however, it will be
shown below that at least two different oxidation states could
325–900 cm . Kapteijn et al. [32] have reported the pos-
sible presence of manganese oxide phases in their Raman
−1
spectra in the region of 635–650 cm , indicating that the
Raman spectra for Mn/TiO may have the presence of Mn-
2
oxide phases. Moreover, Bernard et al. [33] have reported

D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
425
that, although part of MnO2 transforms into haussmanite
cating a substantial amount of Mn on the surface. However,
when the same catalyst was calcined at the higher temper-
ature of 873 K, the Mn/Ti ratio severely decreased to 0.5.
This may be due to the diffusion of the metal oxide, MnOx,
into the TiO2 matrix and/or due to its sintering leading to
the decreased catalytic activity as discussed in the follow-
ing section. Additionally, the surface concentration of other
transition metals was poor. The lowest ratio of 0.1 was ob-
served for Ni. For Ni/TiO2, NiO seems to be mostly buried
under titania and may not be available on the surface for the
catalytic reaction.
−1
(
Mn3O4, 651 cm ) due to the pressure used during the
pellets formation of MnO2 powder, the spectrum of MnO2
shows three main peaks in its Raman spectra, viz. 650,
−1
5
76, and 523 cm . Also, the Raman spectra for Mn2O3 is
also reported [33] to show peaks at 581, 509 (broad), and
−1
6
30 cm (shoulder). Hence, taking the above points into
account and our own quite complex and broad Raman spec-
tra for Mn/TiO2 (Fig. 2), we may conclude that the presence
of various phases for Mn oxide (viz. MnO2, Mn2O3, and
Mn3O4) are likely to be present.
Fig. 1 also shows the presence of Co3O4 (JCPDS File
No. 43-1003) and NiO (JCPDS File No. 44-1159) as weak
crystalline phases for cobalt and nickel oxides supported
on TiO2, respectively. No XRD phases were observed for
TiO2-supported Cu- and Cr-oxide catalysts. Additionally,
crystalline phases could not be observed in the XRD spec-
tra of V/TiO2 and Fe/TiO2. This could also be attributed to
the their amorphous nature. This was most apparent in the
case of V/TiO2, which did not produce a single peak in the
XRD spectrum (not shown). The surprising disappearance
of all XRD peaks, including titania, in this catalyst indicates
that the support could be completely covered with vanadia
due to its very high loading, 20 wt%, further indicating the
highly amorphous nature of vanadia. It is reported in the lit-
erature [34,35] that vanadia is normally well dispersed on
titania and silica supports. As a result, no corresponding
XRD peak could be detected at 27.2 wt% vanadium load-
ing.
The XPS spectra for O 1s is given in Fig. 3 and shows
that the deconvoluted O 1s spectra gave two distinct peaks.
The lower BE peak is due to oxygen from TiO2, whereas
the higher BE peak is due to the respective transition metal
oxides. From Fig. 3 it is clear that the surface oxygen con-
centration in Mn/TiO2 is quite high as compared to that for
Cr/TiO2 and Ni/TiO2 catalysts. The Ti 2p3/2 (Fig. 4) spectra
3
.3. X-ray photoelectron spectroscopy
The surface atomic concentrations for oxygen, titanium,
and transition metals and the atomic ratios for V/Ti, Cr/Ti,
and Ni/Ti are presented in Table 2. The atomic concentra-
tions were based upon the binding energies observed for O
1
s, Mn 2p, Cr 2p, Ni 2p, and V 2p in the titania-supported
transition metal oxide catalysts. All the values are referenced
to the Ti 2p3/2 BE value of 458.5 eV. This peak was cho-
sen for its clarity, intensity, lack of contaminant interference,
and the absence of other oxidation states. It is remarkable
to note the very high Mn/Ti atomic ratio of 5.9 that was
observed for 20 wt% Mn/TiO2 when calcined at 673 K, indi-
Table 2
Atomic surface concentrations of the catalysts as determined by XPS
b
Catalyst
Calcination
Atomic concentrations (%)
a
composition
temperature (K) Oxygen Titanium
TM
TM/Ti
V/TiO
Cr/TiO
Mn/TiO
Mn/TiO
Ni/TiO
673
673
673
873
673
58.1
53.5
52.3
51.6
53.2
14.0
5.4
3.5
13.2
12.6
5.5
9.2
20.8
6.6
0.4
1.7
5.9
0.5
0.1
2
2
2
2
1.4
2
a
2
0 wt% metal loading.
b
Mg-Kα radiation at 300 W, 89.45- and 35.75-eV pass energies; the
remaining mass for each catalyst consisted of carbon.
Fig. 3. O 1s XPS spectra of Cr/TiO , Ni/TiO , and Mn/TiO (20 wt% metal
2 2 2
loading).

426
D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
Fig. 4. Ti 2p XPS spectra of Ni/TiO , Cr/TiO , and Mn/TiO (20 wt%
2
2
2
metal loading).
Fig. 6. FT-IR spectra of adsorbed species after contact with ammonia for
various transition metal oxides supported on TiO after outgassing at 373 K
2
(B, Brönsted acid site; L, Lewis acid site).
peak disappeared for the catalysts calcined at 873 K. The
confirmation of MnO2 and Mn2O3 from our deconvoluted
XPS spectra may be used as evidence to conclude that sup-
ported metal oxides active in low-temperature SCR of NOx
are present in at least two different oxidation states, possibly
suggesting a redox mechanism for the low-temperature SCR
reaction.
Fig. 5. Deconvoluted XPS results of Mn 2p for 20 wt% Mn/TiO .
2
comparing Ni, Cr, and Mn oxides supported on TiO2 cata-
lysts are shown in Fig. 4. It is evident from Fig. 4 that the
concentration of Ti 2p3/2 and thus that of TiO2 catalyst is
significantly less compared to Cr/TiO2 and Ni/TiO2 cata-
lysts. This is probably due to the higher surface coverage by
Mn oxide and/or its higher dispersion on the TiO2 support
as reported earlier by Kapteijn et al. [32] for Mn/Al2O3 ca-
talysts. This clearly indicates that the concentration of Mn
on the surface of Mn/TiO2 relative to the support is sig-
nificantly higher than the other metal oxides. The Mn 2p
spectra for 20 wt% Mn/TiO2 were deconvoluted to obtain
detailed information on the presence of possible metal ox-
ide phases and their relative intensities. The spectrum shows
3
.4. Acidity characterization of the catalysts
3.4.1. NH FT-IR
3
The spectra of ammonia adsorbed on V, Cr, Mn, and Cu
oxides supported on TiO after outgassing at 373 K are
2
compiled in Fig. 6. All the spectra were normalized and
recorded under identical operating conditions. It is reported
in the literature that anatase TiO adsorbs ammonia in coor-
2
dinated form over Lewis acid sites [36,37]. This indicates
(
(
Fig. 5) the presence of three peaks corresponding to Mn2O3
641.2 ± 0.2 eV), MnO2 (642.2 ± 0.4 eV) [32], and Mn-
that anatase TiO is purely Lewis acidic and any evolve-
ment of Brönsted acid sites should be related to the metal
2
nitrate at 644.2 ± 0.4 eV. The Mn-nitrate peak is expected,
as the catalysts were calcined at a relatively low tempera-
ture of 673 K. It should be noted that no reference in the
literature was found to verify our claim that the peak near
oxides deposited on TiO supports. Fig. 6 shows the pres-
2
−
1
ence of four prominent bands near 1175 cm (Lewis acid
−1
−1
sites), 1242 cm (Lewis acid sites), 1424 cm (Brönsted
−
1
acid sites), and 1604 cm (Lewis acid sites). Based upon
the number and size of the peaks observed over Cr/TiO2,
this catalyst clearly possesses the greatest amount of sur-
6
44 eV could be due to Mn-nitrate. Therefore, in order
to confirm the presence of partially undecomposed nitrate
species in the catalysts calcined at lower temperatures, an
uncalcined catalyst was analyzed using XPS. The prominent
peak at 644.5 eV belonging to Mn-nitrate was visibly seen,
confirming our claim of the presence of nitrate species for
our catalysts calcined at lower temperatures. Moreover, this
face acidity. V/TiO only possessed significant amounts of
2
Brönsted acidity, while both Mn/TiO and Cu/TiO only had
2
2
significant amounts of Lewis acid sites. Kapteijn et al. [38]
also reported in the case of alumina-supported manganese
oxide catalysts that the addition of Mn to alumina does not

D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
427
induce any Brönsted acidity. The mix of Lewis and Brön-
sted acid sites along with recording the spectra under actual
reaction conditions served to help identify the surface char-
acteristics necessary for operation at low temperatures. As it
will be shown in the discussion of catalytic activity, Brön-
sted acid sites do not appear to be necessary to carry out
the reaction. An exhaustive study describing the adsorption–
desorption characteristics of the gases, viz. NH3, NO, or
O2 or the mixture of the two or more gases, especially on
Mn/TiO2 catalyst by in situ FT-IR has been thoroughly in-
vestigated in the subsequent paper [39].
was observed in our NH3 FT-IR experiments as well (not
shown).
3.5. Catalytic activity
3.5.1. Results at low GHSV (8000 h 1
−
)
The performance of various catalysts for low-temperature
SCR at reaction temperatures of 373 and 393 K operat-
ing under identical experimental conditions is presented in
Fig. 8. The data show that catalysts with Cr, Mn, and Cu are
highly active at low temperature, and quantitative NO con-
version is achieved at 393 K with 100% N2 selectivity for
2
0 wt% Mn/Hombikat TiO2 catalyst. N2O and NO2 were
3
.4.2. NH3-TPD
not detected with our mass spectrometer. It should be noted
here that a much higher NO concentration (2000 ppm) was
used for this set of our experiments than that found in the
flue gas of industrial power plants at the tail end. As we
have demonstrated earlier [24,25] and as we show below
The amount and strength of the acid sites in the sam-
ples were probed by NH3-TPD. Fig. 7 shows the ammo-
nia desorption patterns for various transition metal oxides
supported on TiO2. The total acidity of various catalysts is
summarized in Table 1. Except for V and Cu, the rest of the
metal oxides supported on TiO2 distinctly show the presence
of broadly distributed strong acid sites that continue desorb-
ing above 523 K. When TPD curves for Mn and V supported
on TiO2 are compared, the overall area of ammonia desorp-
tion curve for the latter catalyst is much smaller than that of
the former one. It should be remembered that not a single
XRD peak was detected for V/TiO2, most probably due to
the amorphous nature of vanadium. This suggests that am-
monia that could have been adsorbed on titania acid sites
was absent in this catalyst resulting in an overall decrease in
acidity and corresponding loss in catalytic activity. It is of-
ten reported in the literature [36,40] that ammonia adsorbed
on Brönsted acid sites desorbs easier than Lewis-coordinated
ammonia during temperature-programmed desorption. This
−1
(
case of 50,000 h ) in this paper, the Mn-based catalysts
are equally effective for space velocities and operating con-
ditions found in industry. The other important factor could
be the possible redox behavior of our catalysts, which was
presented earlier with the XPS results. Very recently Yang
and others [41,42] have also reported active catalysts, viz.
Fe–Mn [41] and MnOx–CeO2 [42], for the low tempera-
ture (373–453 K) SCR of NO with NH3 at relatively high
−1
space velocities of 15,000 and 42,000 h , respectively, but
using relatively lower concentrations of NO (1000 ppm) in
the feed. V/TiO2, which is the catalyst of choice in commer-
cial medium temperature DeNOx systems, was found to be
ineffective in our study. The poor catalytic activity for the
Ni-based catalyst suggests that it may be in its reduced form
even under oxidizing conditions and/or is predisposed to
monovalent oxidation behavior. Moreover, the surface con-
centration of nickel on TiO2 (Table 2) is also very small
(
Ni/Ti = 0.1) compared to the other metals. This could also
be the reason why this particular catalyst is not active in our
system. It is believed that the high surface concentration of
active metal oxides on the surface, the ability of the metal
oxide to operate in the redox mode, and Lewis acidity are
the main contributors to excellent performance at these con-
ditions.
Fig. 8. Performance of 20 wt% transition metal/TiO catalysts at 373 and
2
393 K (GHSV = 8000 h−1, NO = NH3 = 2000 ppm, 2.0 vol% O , He
Fig. 7. NH -TPD spectra of various transition metals TiO (20 wt% metal
loading).
3
2
2
balance).

428
D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
Fig. 9. Comparison of manganese precursors (5.0 wt% Mn) supported
Fig. 10. Catalytic performance of Mn (nitrate)/TiO as a function of metal
2
−1
on TiO at different temperatures (GHSV = 8000 h , NO = NH3
=
−1
2
loading at 373 and 393 K (GHSV = 8000 h , NO = NH3 = 2000 ppm,
2
000 ppm, 2.0 vol% O , He balance).
2
2.0 vol% O , He balance).
2
After finding MnOx/TiO2 to be the most active cata-
lyst, the role of the Mn precursor was examined. As shown
in Fig. 9, a catalyst prepared from Mn-nitrate gives better
performance than one prepared from the acetate precursor
under these conditions. The higher surface area of the ca-
talyst prepared from the nitrate precursor and the formation
of MnO2 as the major and active phase could be the rea-
sons. In contrast to our results, Kapteijn et al. [43] have
reported that manganese acetate results in better dispersion
of manganese oxide on alumina and higher catalytic activity
than manganese nitrate at 373–453 K. That group concluded
from H2 TPR studies that acetate yields Mn2O3 while ni-
trate produces mainly MnO2. At the same time, the authors
claimed that MnO2 exhibits higher NO reduction activity
than Mn2O3 per unit surface area. This explanation clearly
indicates that MnO2 should be more active than Mn2O3
in SCR reactions based upon our XPS results (Fig. 5). We
could clearly observe the presence of MnO2 in our catalysts
along with better catalytic performance than the previously
reported catalysts. It should be noted that higher loadings of
Mn on the titania support gave better results, and this will be
discussed below. Surprisingly, V/TiO2 was found to be to-
tally inactive for SCR reaction at low temperatures. In our
study the actual surface area for the TiO2 support decreased
by about 5 times after vanadium deposition. Other metal ox-
ides do not lead to such a drastic decrease in surface area
Fig. 11. Catalytic performance of Mn (nitrate)/TiO as a function of calci-
2
−1
nation temperature at 373 and 393 K (GHSV = 8000 h , NO = NH3
=
2000 ppm, 2.0 vol% O , He balance, 20 wt% Mn).
2
form well. Only catalysts with higher Mn loadings which
are calcined at lower temperatures provide excellent cata-
lytic performance (Fig. 11). H2 TPR results for the catalysts
are shown in Fig. 12. The results show that higher reduction
temperatures are needed for catalysts synthesized with lower
Mn loadings or calcined at higher temperatures. For these
catalysts, there is likely to be a stronger metal–support inter-
action especially when the catalysts are calcined at higher
temperatures (> 673 K), thereby making it more difficult
to carry out the reduction. A similar trend was observed
by Richter et al. [44]. Increasing Mn loading decreases the
(
Table 1).
Mn loadings ranging from 5 to 25 wt% were studied at
reaction temperatures of 373 and 393 K. The results are
presented in Fig. 10. NO conversion increases with an in-
crease in metal loading, and an optimum loading is observed
near 20 wt% Mn. At 393 K quantitative NO conversion is
achieved at 17.5 wt% Mn loading. The selectivity for N2
was always 100% irrespective of catalytic activity. Catalysts
having lower Mn loadings, along with those having higher
loadings but calcined at higher temperatures, do not per-
temperature required for reduction of MnO , provided that
x
the catalysts are calcined at or below 673 K. In summary,
the catalysts with higher metal loadings and lower calci-
nation temperature provided the best catalytic performance,
and the lower reduction temperatures observed in those cases

D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
429
Fig. 12. H TPR of MnOx /TiO catalysts (with different Mn loadings and
2
2
prepared using Mn-nitrate precursor).
suggest the presence of MnO2 (major) and Mn2O3 (minor)
phases [10,43].
It is reported [45] that crystallites of V2O5 are formed
at the expense of polymeric vanadyl species if the vanadia
loading exceeds a theoretical monolayer. Crystalline V2O5
species are detrimental to both activity and N2 selectiv-
ity. Those in situ Raman spectroscopic studies revealed that
NH3 is chemisorbed on Ti4 centers associated with the
support and that the coordination of NH3 takes place only
with the isolated monomeric vanadyl species. We confirmed
from our XRD studies that high loading vanadia/titania ca-
talysts do not yield any XRD peaks, indicating that titania
is completely covered with homogeneously dispersed V2O5
+
species. Hence, there is no chance for NH3 to be adsorbed
on crucial Ti4+ sites of the support, and that could lead to
Fig. 13. Catalytic performance of transition metals supported on TiO as
2
−1
a function of temperature (20 wt% metal loading, GHSV = 50,000 h
,
its subsequent dissociation on monomeric vanadyl species.
The NH3-TPD results also indicated that the V/TiO2 cata-
lyst adsorbs very small quantities of NH3, further indicating
its relatively low acidity (Fig. 7). It can be concluded that
the higher amount of Brönsted acidity, possible coverage of
the titania support by amorphous V oxide, very low overall
acidity, and the low surface area of the actual catalyst rela-
tive to the support could be the main reasons for its dismal
catalytic performance at the lower temperatures. In addition
to this, the catalyst may not have the required activation en-
ergy for this reaction at these temperatures. In contrast with
the reported literature for medium-temperature SCR opera-
tions [46–50] where it is reported that Brönsted acidity is of
utmost importance to obtain better catalytic activity, we have
shown here that Lewis acidity is important in our system and
that in our best catalyst, MnOx/TiO2, the most active phase
found from XRD and XPS observations was MnO2.
NO = NH = 400 ppm, 2.0 vol% O , He balance).
3
2
talysts previously discussed were tested at 50,000 h 1. The
NO conversion and N2 selectivity results are presented in
Fig. 13. While it is expected that the catalytic activity would
−
−1
not be as high as that reported at 8000 h , MnOx/TiO2 pro-
vides respectable catalytic activity relative to the other tran-
sition metals tested at temperatures below 423 K. Cu/TiO2
also provides high catalytic activity under these conditions
and peaks in activity from 423 to 473 K. Mn is clearly most
useful at temperatures below 423 K, as its N2 selectivity
quickly declines below 90% as the temperature increases. It
appears that Cu/TiO2 outperforms Mn/TiO2 at higher space
velocities when both the activity and the selectivity data
are considered; however, SCR units in power plants must
have N2 selectivities above 95% to avoid releasing signif-
icant amounts of N2O into the atmosphere. Additionally,
industrial SCR units will be exposed to significant amounts
of water from combustion. Consequently, the catalytic per-
formance of all the supported transition metal oxides tested
−
3
.5.2. Results at higher GHSV of 50,000 h 1
Since SCR catalysts used in power plants will typically
−1
be used at GHSV values above 30,000 h , all of the ca-

430
D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
SCR of NO with NH3. The H2 TPR study over MnOx/TiO2
suggested that a higher reduction temperature was required
when either the Mn loading was low (ꢁ 12.5 wt%) or the
catalysts were calcined at higher temperatures (ꢀ 773 K).
In MnOx/TiO2 catalysts, XPS and XRD studies showed the
presence of MnO2 (major) and Mn2O3 (minor) phases. Since
MnO2 is present in higher concentrations, it is suggested to
be the most active phase for low-temperature SCR of NO
−1
with NH3. Catalytic experiments performed at 50,000 h
with 11 vol% water demonstrated that MnOx/TiO2 was the
most promising catalyst for low temperature SCR applica-
tions. Evidently, the presence of water increased N2 selectiv-
ity by preferentially adsorbing on those active sites respon-
sible for the formation of N2O.
Acknowledgments
Fig. 14. Catalytic performance of transition metals supported on TiO at
2
48 K (20 wt% metal loading, GHSV = 50,000 h 1, NO = NH3
−
The authors are grateful to the Ohio Coal Development
Office (OCDO), Columbus, Ohio, USA, for financial sup-
port and for allowing us to publish our findings. The authors
thank Dr. Ettireddy Padmanabha Reddy for thoughtful dis-
cussions.
4
4
=
00 ppm, 2.0 vol% O , 11 vol% H O, He balance).
2
2
in the presence of 11 vol% water is presented in Fig. 14 at
48 K and 400 ppm NO in the feed. As it can clearly be seen,
4
Mn/TiO2 provides the best overall performance when water
is present. The presence of water drastically suppresses the
formation of N2O, allowing the use of Mn/TiO2 at slightly
higher temperatures. This suggests that water preferentially
adsorbs on the active sites responsible for the N2O forma-
tion, and this phenomenon is beneficial for the application of
Mn/TiO2 catalysts in power plants. Under these conditions,
the catalytic activity of the transition metal oxides decreases
in the following order: Mn > V ꢀ Co > Cu > Cr > Fe ꢀ
Ni.
References
[
[
1] H. Bosch, F. Janssen, Catal. Today 2 (1988) 369.
2] J.G. Henry, G.W. Heinke, Environmental Science and Engineering,
Prentice-Hall, Englewood Cliffs, NJ, 1989.
[3] F. Luck, J. Roiron, Catal. Today 4 (1989) 205.
[4] F. Janssen, F. van den Kerkhof, H. Bosch, J.R.H. Ross, J. Phys. Chem.
1 (1987) 5921.
9
[
[
[
[
5] J.W. Byrne, J.M. Chen, B.K. Speronello, Catal. Today 13 (1992) 33.
6] G. Ramis, G. Busca, F. Bregani, Catal. Lett. 139 (1993) 353.
7] S.C. Wood, Chem. Eng. Progr. 90 (1994) 32.
8] L.J. Alemany, F. Berti, G. Busca, G. Ramis, D. Robba, G.P. Toledo,
M. Trombetta, Appl. Catal. B 248 (1996) 299.
4
. Conclusions
[9] J.A. Dumesic, N.-Y. Topsoe, H. Topsoe, T. Slabiak, J. Catal. 163
(
1996) 409.
10] L. Singoredjo, R. Korver, F. Kapteijn, J.A. Moulijn, Appl. Catal. B 1
1992) 297.
11] W.S. Kijlstra, D.S. Brands, H.I. Smit, E.K. Poels, A. Bliek, J. Ca-
tal. 171 (1997) 219.
12] W.S. Kijlstra, E.K. Poels, A. Bliek, B.M. Weckhuysen, R.A. Schoon-
heydt, J. Phys. Chem. B 101 (1997) 309.
13] G. Centi, S. Perathoner, D. Biglino, E. Giamello, J. Catal. 151 (1995)
Most of the catalysts studied, which included Mn, Fe, Co,
[
[
[
[
[
[
[
[
Ni, and Cu supported on titania, possessed mainly Lewis
acidity and had relatively weak Brönsted acid sites. Only
Cr/TiO2 and V/TiO2 possessed a significantly high amount
of Brönsted acidity, and that was coupled with a substan-
tial amount of Lewis acidity for Cr/TiO2. Comparing the
low-temperature (373 K) SCR activity of our catalysts, it
was concluded that catalysts having Lewis acidity are ac-
tive and that Brönsted acidity may not be contributing to
catalytic performance, as reported earlier by Busca and co-
workers [36]. The deposition of active metal oxide(s), in
addition to the catalyst preparation method, is paramount to
excellent catalytic activity at temperatures as low as 393 K.
In the case of V/TiO2, it was concluded that high levels of
Brönsted acidity, complete coverage of titania by vanadium
oxide, inability of the catalyst to activate at low tempera-
tures, very low overall acidity, and the low surface area of the
actual catalyst as compared to the support were the main rea-
sons for its dismal catalytic performance in low-temperature
(
75.
14] G. Ramis, L. Yi, G. Busca, M. Turco, E. Kotur, R.J. Willey, J. Ca-
tal. 157 (1995) 523.
15] A. Kato, S. Matsuda, F. Nakajima, M. Imanari, Y. Watanabe, J. Phys.
Chem. 85 (1981) 1710.
16] S. Kasaoka, E. Sasaoka, H. Iwasaki, Bull. Chem. Soc. Jpn. 62 (1989)
1226.
17] J. Pasel, P. Käßner, B. Montanari, M. Gazzano, A. Vaccari, W. Ma-
kowski, T. Lojewski, R. Dziembaj, H. Papp, Appl. Catal. B 18 (1998)
1
99.
18] M. Yoshikawa, A. Yasutake, L. Mochida, Appl. Catal. A 173 (1998)
39.
[
2
[
[
19] Z. Zhu, Z. Liu, S. Liu, H. Niu, Appl. Catal. B 23 (1999) L229.
20] Z. Zhu, Z. Liu, H. Niu, S. Liu, J. Catal. 187 (1999) 245.
[21] E. Richter, Catal. Today 7 (1990) 93.

D.A. Peña et al. / Journal of Catalysis 221 (2004) 421–431
431
[
22] C.J.G. Van der Grift, A.F. Woldhuis, O.L. Maaskant, Catal. Today 27
[35] H. Miyata, K. Fujii, T. Ono, J. Chem. Soc., Faraday Trans. I 84 (1988)
3121.
(1996) 23.
[
[
23] V. Visciglio, G. Blanchard, R. Surantyn, Catal. Lett. 40 (1996) 39.
24] P.G. Smirniotis, D.A. Peña, B.S. Uphade, US patent applied, filed
[36] J.M.G. Amores, V.S. Escribano, G. Ramis, G. Busca, Appl. Catal. B 13
(1997) 45.
2
000.
[37] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B 18 (1998) 1.
[38] F. Kapteijn, L. Singoredjo, M. van Driel, A. Andreini, J.A. Moulijn,
G. Busca, G. Ramis, J. Catal. 150 (1994) 105.
[39] D.A. Peña, E.P. Reddy, B.S. Uphade, P.G. Smirniotis, in preparation.
[40] L. Lietti, G. Ramis, F. Berti, G. Toledo, D. Robba, G. Busca, P. For-
zatti, Catal. Today 42 (1998) 101.
[41] R.Q. Long, R.T. Yang, R. Chang, Chem. Commun. (2002) 452.
[42] G. Qi, R.T. Yang, Chem. Commun. (2003) 848.
[43] F. Kapteijn, L. Singoredjo, A. Andreini, J.A. Moulijn, Appl. Catal. A 3
(1994) 173.
[44] M. Richter, A. Trunschke, U. Bentrup, K.-W. Brzezinka, E. Schreier,
M. Schneider, M.-M. Pohl, R. Fricke, J. Catal. 206 (2002) 98.
[45] G.T. Went, L.-J. Leu, R.R. Rosin, A.T. Bell, J. Catal. 134 (1992) 492.
[46] N.-Y. Topsøe, J. Catal. 128 (1991) 499.
[
[
[
[
[
25] P.G. Smirniotis, D.A. Peña, B.S. Uphade, Angew. Chem. Int. Ed.
Engl. 40 (2001) 2479.
26] N.V. Economidis, D.A. Peña, P.G. Smirniotis, Appl. Catal. B 23 (1999)
1
23.
27] I. Barin, Thermochemical Data of Pure Substances, Chemie, Berlin,
989.
1
28] J.L. Alemany, L. Lietti, N. Ferlazzo, P. Forzatti, G. Busca, G. Ramis,
E. Giamello, F. Bregani, J. Catal. 155 (1995) 117.
29] G. Busca, G. Ramis, J.M.G. Amores, V.S. Escribano, P. Piaggio,
J. Chem. Soc., Faraday Trans. 90 (1994) 3181.
30] E.R. Stobbe, B.A. de Boer, J.W. Geus, Catal. Today 47 (1999) 161.
31] J. Datka, A.M. Turek, J.M. Jehng, I.E. Wachs, J. Catal. 135 (1992)
[
[
186.
[
[
[
32] F. Kapteijn, A.D. van Langeveld, J.A. Moulijn, A. Andreini, M.A. Vu-
urman, A.M. Turek, J.-M. Jehng, I.E. Wachs, J. Catal. 150 (1994) 94.
33] M.-C. Bernard, A.H.-L. Goff, B.V. Thi, S.C. de Torresi, J. Electro-
chem. Soc. 140 (1993) 3065.
[47] N.-Y. Topsøe, Science 265 (1994) 1217.
[48] N.-Y. Topsøe, H. Topsøe, J.A. Dumesic, J. Catal. 151 (1995) 226.
[49] H. Schneider, S. Tschudin, M. Schneider, A. Wokaun, A. Baiker, J. Ca-
tal. 147 (1994) 5.
[50] N.-Y. Topsøe, J.A. Dumesic, H. Topsøe, J. Catal. 151 (1995) 241.
34] M.A. Reiche, Ortelli, A. Baiker, Appl. Catal. B 23 (1999) 187.