Selective catalytic reduction of NOx by methane over Co-H-MFI and Co-H-FER zeolite catalysts: Characterisation and catalytic activity

Article abstract of DOI:10.1016/S0021-9517(02)00153-7

Co-H-MFI and Co-H-FER were investigated as catalysts for NO reduction with CH₄, under diluted conditions. Only the features due to zeolitic frameworks were observed. After outgassing, Co ions were mainly present as Co²⁺. There were r

Full text of DOI:10.1016/S0021-9517(02)00153-7

Journal of Catalysis 214 (2003) 179–190
www.elsevier.com/locate/jcat
Selective catalytic reduction of NO by methane over Co-H-MFI and
x
Co-H-FER zeolite catalysts: characterisation and catalytic activity
a,b
a,b
a,c
a,c
a,c
a,c,∗
d
C. Resini, T. Montanari, L. Nappi, G. Bagnasco, M. Turco, G. Busca,
F. Bregani,
d
d
M. Notaro, and G. Rocchini
a
Consorzio INSTM, via Benedetto Varchi, 59, 50132 Florence, Italy
b
Dipartimento di Ingegneria Chimica e di Processo, Università degli Studi di Genoa, Genova, Italy
c
Dipartimento di Ingegneria Chimica, Università “Federico II” di Napoli, Naples, Italy
d
BU Processi per la Generazione, CESI S.p.A., via Reggio Emilia, 39, 20090 Segrate (MI), Italy
Received 11 July 2002; revised 11 November 2002; accepted 19 November 2002
Abstract
Co-H-MFI and Co-H-FER were studied as catalysts for NO reduction with methane, under diluted conditions. Catalyst bulk and surface
structure was characterised by chemical analysis, XRD, TG-DTA, SEM micrographs, IR studies, UV–vis spectroscopy, ammonia TPD,
and TPR. Only the features due to zeolitic frameworks were observed. The UV–vis spectra showed that, after outgassing, Co ions are mainly
2
+
present as Co . FT-IR detected residual bridging hydroxyl and silanol groups in spite of the apparent over-exchange, deduced from the
chemical analysis. Co ions gave rise to Lewis acid sites of medium strength, evidenced through in situ analysis of the adsorption of a basic
probe molecule. NH -TPD measurements fully confirmed these results. TPR measurements detected easily reducible and oxidizable Co
3
2
+
species, possibly related to nanosized Co oxide phases, together with hardly reducible Co ions. Catalytic tests highlighted the Co-H-FER
and Co-H-MFI activity, both in NO-selective catalytic reduction by methane and in NO oxidation to NO , the latter being the predominant
2
reaction at low temperatures. Some mechanistic features of the CH -SCR reaction are discussed.
4

2003 Elsevier Science (USA). All rights reserved.
Keywords: CH -SCR; NOx reduction; Co-exchanged zeolites; Co-H-FER; Co-H-MFI; Co-H-ZSM5; Acidity characterisation; IR spectroscopy; Nitrile
4
adsorption; UV–vis spectroscopy
1
. Introduction
A valid alternative process for denitrification of exhaust
gases from power plants (as NO decomposition is not yet
applicable) implies the use of methane as the reducing agent.
The likely overall reaction for the CH4-SCR technique
is [3]
The removal of nitrogen oxides from flue gases of thermal
power plants is currently carried out with the so-called NH3-
SCR technology, i.e., the selective catalytic reduction of
NO by ammonia [1,2] over V2O5–WO3/TiO2 or V2O5–
MoO3/TiO2 catalysts through the overall reaction
CH4 + 2NO + O2 → N2 + CO2 + 2H2O.
(2)
The possibility of reducing NO with hydrocarbons such
as olefins and higher alkanes was first proposed in 1990 by
Held et al. [4] and by Iwamoto et al. [5] and Cu-MFI cat-
alysts were found to be particularly active. However, these
catalysts are poorly effective with methane as reductant.
Methane as a reductant is, however, the preferred choice for
NOx removal from flue gases because it is already present in
methane-fueled power plants. Co-containing zeolites, such
as Co-MFI and Co-FER, were found by Armor [6] to be
4
NH3 + 4NO + O2 → 4N2 + 6H2O.
The need to store ammonia is one of the main drawbacks
(1)
of this technology when applied to power plants. Addition-
ally, the possible formation of ammonium sulphates due to
reaction of SO2, oxygen, water, and ammonia and the dan-
ger of ammonia leakage may also limit the conditions for
application of such process.
*
particularly active in CH -SCR in the presence of oxygen.
Corresponding author.
4
E-mail address: Guido.Busca@unige.it (G. Busca).
These catalysts work in the temperature range from 573
0021-9517/03/$ – see front matter  2003 Elsevier Science (USA). All rights reserved.
doi:10.1016/S0021-9517(02)00153-7

180
C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
to 773 K and at space velocities that allow application in
high-dust configurations.
2.2. Catalyst characterisation
2
+
Co ions exchanged in a zeolitic matrix are considered
to be the active sites [7,8]. The reaction mechanism for these
catalysts was also investigated [9–11]. The activity of the
catalysts is significantly inhibited by steam [12] but this
effect can be limited, for instance, by mixing the zeolite
catalysts with metal oxides such as manganese oxide [13].
The main drawback of these catalysts is their hydrother-
mal instability due to dealumination and loss of the active
phase [14]. It has been reported that deactivation by dealumi-
nation is slower for small than for large crystal size Co-MFI
samples [15]. Moreover, these catalysts are SOx sensitive,
although that effect also can be limited by modifying the ze-
olite composition [16].
Chemical analysis of the catalysts was performed by
atomic absorption spectroscopy.
Surface area and pore size distribution measurements
were performedby N2 adsorption at 77 K using the Dubinin–
Radushkevich method for the calculation of the surface area
and the Horvath method for the pore size distribution.
XRD analysis was performed with a Philips PW1710/
1729 diffractometer (Cu-Kα radiation). The cell parameters
were calculated using the ASPO program. The crystal size
was calculated using the Sherrer formula and considering the
most intense XRD peaks. TG-DTA analyses were effected
with a Setaram T92 analyser.
To improve such catalysts’ stability, addition of other
components to Co-zeolites has been attempted [17,18]. In
spite of many efforts, the durability problems of transition-
metal-containing zeolites have not yet been resolved. On the
other hand, several more questions have been opened about
these catalysts. The reason these materials are more active
than mixed oxides containing the same metal oxides is not
yet fully understood.
Scanning electron microscope (SEM) observations were
made using a model Cambridge Leo microscope equipped
with a model Link Isis microanalysis system of the disper-
sion energy type. All images were obtained with secondary
electrons.
Diffuse reflectance spectra (DR-UV–vis–NIR) of pure
self-supported sample powder disks were recorded by a
−1
Jasco V-570 apparatus in the range 50,000–4000 cm
Recently, it has been proposed that cooperation between
transition metal cations and Brønsted acid sites can occur
upon CH4-SCR. According to Yan et al. [19], Co cations act
in the oxidation of NO to NO2 while Brønsted acid sites act
in the true reduction step by methane. In contrast, Kauchy
et al. [8] suggest that Brønsted sites enhance the activity in
oxidising NO to NO2.
at room temperature (rt) in air, and after dehydration at
773 K for 5 h by outgassing through a conventional gas
manipulation/outgassing ramp connected to a quartz cell.
The IR spectra were recorded on a Nicolet Protégé 460
Fourier transform instrument. The surface characterisation
was performed using pressed disks of pure NH4-zeolite and
Co-H-zeolite powders (15 mg, 2 cm diameter), activated by
outgassing at 773 K in the IR cell. A conventional manip-
ulation/outgassing ramp connected to the IR cell was used.
Spectra were collected after the activation procedure and af-
ter adsorption of acetonitrile (AN), supplied by Aldrich and
distilled under vacuum prior to use. The adsorption pro-
cedure involves contact of the activated sample disk with
vapours at rt and at a pressure of 15 Torr, and outgassing
in steps from rt to higher temperatures.
In the present work Co-exchanged MFI and ferrierite
+
(
FER) zeolites obtained from NH4 -zeolites were studied
for CH4-SCR. The study was aimed to obtain information
on the nature of active sites and to investigate the role of
protonic sites. To gain this purpose several characterisation
techniques were employed that can give complementary
results, such as FT-IR, UV–vis diffuse reflectance, NH3-TPD
and TPR.
Temperature programmed desorption of ammonia (NH3-
TPD) was performed in a flow laboratory plant equipped
with a thermal conductivity detector (TCD). The sample,
treated in He flow at 773 K, was saturated with a 1%
NH3/He mixture at rt and purged with He at rt; then
NH3-TPD was effected by heating in He flow at a rate of
2
. Experiments and methods
2
.1. Catalyst preparation
−1
1
0 K min . A KOH trap was placed between the sample
2
−1
NH4-MFI (SiO2/Al2O3 = 50, S = 425 m g ) and
and the TCD in order to avoid interference by H2O.
Temperature programmed reduction (TPR) tests were
performed by a Micromeritics TPD/TPR 2900 instrument.
2
−1
NH4-FER (SiO2/Al2O3 = 55, S = 480 m g ) supplied
by Zeolyst were used as starting materials. NH4-zeolite
powders were contacted with a 0.02 M (CH3COO)2Co ·
−
1
The sample was heated at a rate of 10 K min up to 1073 K
−1
4
H2O aqueous solution (10 g powder for 1 L solution) under
under a gas flow (22 mL min ) of 2% H2 in Ar. In such
a way the catalyst samples were reduced and the amount
of consumed H2 was measured by a TCD detector. The
H2O produced in the reduction was removed by a cold trap
(melting isopropanol) placed before the TCD.
stirring at a constant temperature of 353 K for 24 h. The
resulting mixture was filtered and washed three times with
double-distilled water. After centrifugation, powders were
dried at 353 K for 10 h and calcined at 823 K for 4 h.

C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
181
2
.3. Catalytic activity measurements
Catalytic tests were performed in a flow laboratory
plant equipped with a fixed bed reactor operating at at-
mospheric pressure. The operation conditions were: reacting
mixture composition, NO = 1500 ppm, CH4 = 1500 ppm,
O2 = 25,000 ppm, He balance; T = 523–773 K; GHSV =
−1
3
1
0,000–60,000 h . Grain dimensions of the catalysts were
50–212 µm. The analyses of NO, NO2, and N2O was per-
formed by an ABB URAS 14 continuous spectrophotometer
analyser equipped with a NO2 converter. CH4, N2, CO,
and CO2 concentrations were measured by an HP 5890 gas
chromatograph with a TCD detector and a 5A-Porapak Q
double-packed molecular sieve column.
3
. Results
3
.1. Structural and morphological catalyst
characterisation
The SEM micrographs of the exchanged samples are
reported in Fig. 1. The crystal shape is the same as for
the original ammonium zeolites, showing that for the MFI
sample the particles are nearly spherical with an apparent
diameter of 100–250 nm. For FER the particles are clearly
sheet-like, nearly 50 nm thick. The XRD patterns of both
unexchanged and exchanged samples correspond to those of
the related pure zeolite, without any other detectable phase.
The crystal sizes, calculated by the Sherrer method on the
most intense peaks, are nearly 50 nm for FER and 25 nm
for MFI.
Fig. 1. SEM micrographs of Co-exchanged zeolites.
FT-IR skeletal spectra do not provide any evidence for
a significant perturbation of the zeolite framework spectra
due to the ion-exchange process. On the other hand, both
Co-zeolites show, in the far-IR region, an additional weak
band at 250 cm which could be assigned to the in-cavity
cobalt–oxygen stretching, as previously reported for other
metal-exchanged zeolites [20].
it can be observed that the Co-exchange procedure and fol-
lowing calcination at 823 K do not modify the surface areas
of the parent zeolites.
−1
3
.2. UV–vis spectroscopic characterisation
The DR-UV–vis spectra of both ammonium and Co-
The TG-DTA analyses of the parent ammonium zeolites
and of Co-exchanged samples agree with those expected for
these samples, as regards water release and ammonium de-
composition. In all cases, water and ammonia desorption is
complete around 773 K. For Co-MFI a further small weight
loss, possibly associated with decomposition of Co oxide
species, occurs around 973 K. Above that temperature slow
weight losses are observed for all the samples, possibly due
to dehydration of structural OHs, with corresponding ze-
olitic structures decomposition.
zeolites, recorded under ambient conditions, are reported
in Fig. 2. The UV–vis spectrum of the most stable cobalt
2
−1
oxide, Co O (supplied by Carlo Erba, S = 15 m g ), is
3
4
n+
reported too. Absorptions due to Co can be clearly seen
in the spectra of cobalt-containing zeolites as a very strong
−1
broad band centred near 26,000 cm and a shoulder near
−
1
19,000 cm . In the case of Co-containing MFI significant
Table 1
Textural properties and analytical data of Co-zeolites
The results of chemical analysis and textural character-
isation of the catalysts are reported in Table 1. It can be
observed that the Si/Al ratios experimentally evaluated are
very close to the theoretical ones. The cobalt content of both
samples is higher than that corresponding to complete ion
exchange capacity evaluated from the Si/Al ratio. Moreover
Si/Al Si/Al Co wt% Co/Al
S
VMP
−1
DP
Å
2
−1
3
exp. theor.
exp.
exp.
m g
cm g
Co-H-FER 28.51 27.5
Co-H-MFI 28.31 25.0
2.21
1.71
0.73
0.55
451
466
0.160
0.159
5.3
5.5
S = specific surface area; VMP = micropore volume; DP = average pore
diameter.

182
C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
Fig. 3. UV–vis spectra of Co-MFI (a) and Co-FER (b) after outgassing
at 773 K for 5 h.
of FER. Co2 in this site is responsible for a single band
+
−1
at 14,800–15,000 cm . A second family of absorption
components is related to Co sitting in a deformed six-
Fig. 2. UV–vis spectra of H-zeolites (···) and Co-exchanged zeolites (—)
at rt in air: (a) MFI, (b) FER. The spectrum of Co O is reported too.
2+
3
4
ring made of six framework oxygen atoms allocated at the
intersection of the straight and the sinusoidal channels of
MFI and in the eight-ring channel of FER. This species is
considered to be responsible for four absorption components
absorption is also evident in the region between 20,000 and
−1
−1
1
0,000 cm with a further maximum near 14,000 cm .
The spectrum of Co-H-FER does not give any evidence
for the presence of massive Co3O4. On the contrary, the
spectrum of Co-H-MFI in these conditions is similar to that
of Co3O4. In Fig. 3 the spectra of cobalt-exchanged FER
and MFI recorded after evacuation at 773 K for 5 h are
reported. The very strong and broad absorption observed in
the ambient conditions spectra does not appear any longer,
showing that activation under vacuum causes a change in the
−1
in the range 15,900–21,000 cm with the main maximum
−1
near 17,000 cm . According to these authors, this kind
of site is that predominately occupied at all Co loadings.
The last site type is a boat-shaped site that can be found
in the wall of the sinusoidal channel of MFI and in the
eight-ring channel of FER. The exact geometry and the
number of coordinated oxygen atoms are still unknown for
the investigated materials; it is occupied only at high Co
loadings and gives rise to two absorption components in the
n+
oxidation and/or coordination state of Co ions. The very
−1
−1
high-energy region (above 40,000 cm ) is unavailable in
the spectrum because our vacuum cell causes strong noise
there. Weak absorptions are found which can be interpreted
as a triplet near 19,000 cm , near 17,000 cm and near
5,000 cm . No other absorption occurs in the range
range 20,000–22,000 cm .
To check the picture given by Kauchky et al. [21] and by
Dedechek et al. [22], we further investigated, under the same
conditions, the UV spectroscopy of other Co-containing
samples, such as Co–silica–aluminas. We find on these sam-
ples spectra quite similar to those on Co-MFI and Co-FER,
−1
−1
−1
1
4
−1
0,000–22,000 cm . The spectrum we observe is close to
−1
those reported and deeply discussed by Kauchy et al. for
Co-FER [21] and by Dedechek et al. for Co-MFI [22]. These
with absorption again in the region 22,000–14,000cm . So
we can suppose that Co species located at the external zeolite
surface (which likely exist, as deduced from IR spectra, see
below, and are likely similar to those of Co–silica–alumina)
cannot be distinguished from the internal Co species. The
presence of these species has not been considered in pre-
vious studies [21,22]. In agreement with Dedechek et al.
authors assigned the spectra to Co2 in exchanged positions
+
2
+
in the zeolite cavities. Three families of Co ions in the
zeolite cavities would exist, according to these authors. In
one of the sites Co2 is coordinated to four oxygen atoms
arranged in a rectangle which is situated in the wall of the
straight channel of MFI and in that of the main channel
+
2
+
absorption bands characteristic neither of hydrated Co

C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
183
We can note that, for bulk absorptions such as the skeletal
−
1
vibrations in the region 1200–400 cm and the overtones
−1
in the region 2200–1500 cm , band intensities are just the
same for Co-H-FER and H-FER and for Co-H-MFI and
H-MFI. Thus, the intensity decrease of the OH bands is
reasonably due to the disappearance of such bands due to
the exchange of the protons for Co ions. Although a certain
error (but probably small) can occur in this evaluation, we
can conclude that for both Co-H-MFI and Co-H-FER, the
exchange of protons is evidently only very partial, although
more pronounced for MFI than for FER.
These data indicate that, in spite of the significant
amount of cobalt loaded onto the zeolites and the absence
of detectable Co-containing species other than those that
exchanged for the protons, actually only a relatively small
part of the bridging OHs underwent cation exchange. In
particular, the large decrease of the band of the terminal
silanols indicates that a significant part of cobalt ions
actually exchanged for the OHs located at the external
zeolite structure and are consequently thought to remain
outside the channels.
To test whether the partial exchange with cobalt modified
the acidity of the residual OHs and if the same Co sites
are actually accessible, we investigated the adsorption of
acetonitrile (AN) onto the catalyst surface (Figs. 5–7).
As previously reported [27], the adsorption of AN on
H-FER and H-MFI surfaces perturbs the OH groups. The
Fig. 4. FT-IR spectra in OHs stretching region of H-zeolites (···) and
Co-exchanged zeolites (—) after outgassing at 773 K for 1 h: (a) MFI,
(b) FER.
nor of [Co2 (OH) ] species have been detected after out-
+
− +
gassing.
3
.3. IR studies
The IR spectra of the OH stretching bands of hydroxyl
groups of activated Co-zeolites are compared to those
of the corresponding H-zeolites in Fig. 4. As discussed
elsewhere [23] the spectrum of H-FER shows a very sharp
−1
band at 3747 cm
which is due to the terminal silanol
groups located on the external surface and weakly acidic;
−1
a broader band at 3600 cm
is due to the bridging
Si–OH–Al groups that are exclusively on the inner surface
and possess a strong Brønsted acidity. The Co-FER sample
shows the same bands but with a loss of intensity. The
absolute intensity of the band of the external silanol groups
is near 70% for Co-H-FER with respect to the H-FER band,
whereas the band of the internal bridging OHs is still 88%
with respect to that of the unexchanged zeolite.
The spectra of Co-H-MFI and H-MFI samples show
−
1
similar bands at 3747 and 3615 cm , but with an additional
−1
weak band at 3660–3670 cm that can be assigned to the
OHs of extraframework species. Besides, H-MFI sample
presents a component in the region 3730 cm
actually unusually strong [24–26]. We can calculate that
7% of the band due to terminal silanols and 50% of the band
−1
that is
5
Fig. 5. FT-IR subtraction spectra (ABC pattern) of H-zeolites (—) and
Co-exchanged zeolites (—) after AN adsorption and evacuation at rt; the
spectra of activated samples have been subtracted. (a) FER, (b) MFI.
due to bridging OHs are still residual in the MFI exchanged
sample.

184
C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
Fig. 6. FT-IR subtraction spectra of MFI samples (OH stretching region,
left) after AN adsorption and evacuation, and of AN (–CN stretching region,
right) adsorbed on catalyst surfaces. The spectra of the activated samples
have been subtracted. (a) H-MFI, (b) Co-H-MFI; from the top: liquid AN
Fig. 7. FT-IR subtraction spectra of FER samples (OH stretching region,
left) after AN adsorption and evacuation, and of AN (–CN stretching region,
right) adsorbed on catalyst surfaces. The spectra of the activated samples
have been subtracted. (a) H-FER, (b) Co-H-FER; from the top: liquid AN
(— —), evacuation at rt (···), at 373 K (—), and at 473 K (—), for 10 min.
(— —), evacuation at rt (···), at 373 K (—), and at 473 K (—), for 10 min.
perturbation of the bridging hydroxyl groups, in particular,
gives rise to the so called ABC pattern (Fig. 5): two intense
and broad bands at around 2800 and 2400 cm , a less
gion 2600–2000 cm 1 (right side): the spectra of activated
materials have been subtracted. Correspondingly, the sub-
traction spectra in the OH stretching region are shown (range
−
−1
−
1
intense one at around 1600 cm , and a window with a
−1
−1
4000–3300 cm , left side). For each zeolite the spectra col-
lected after the evacuation steps at rt, 373 K, and 473 K are
compared for exchanged and not-exchanged samples. Liquid
minimum at around 2600 cm , corresponding to ∼2 δOH.
This so-called ABC pattern has been assigned to very strong
quasi-symmetrical hydrogen bonding, where the proton is
partially but not completely transferred to the base. It arises
from the Fermi resonances between νOH and the overtones
of δOH and γ OH when the hydroxyl groups strongly
interact with bases, as discussed elsewhere [28]. In Fig. 5
the results of subtraction of the spectra of activated zeolites
from those obtained after adsorption of AN and evacuation at
rt for 10 min, in order to eliminate the excess vapour phase,
are shown. The ABC profiles of the Co–H-zeolites can be
compared to those of H-zeolites. The addition of Co seems
−1
acetonitrile shows a strong doublet at 2294 and 2254 cm
,
the latter band being definitely stronger than the former.
They are due to Fermi resonance between the –CN stretching
and a δCH3 +νC–C combination. This characteristic pattern
changes when the –CN group adsorbs onto an acidic surface:
the bigger the electron-withdrawal effect of either Brønsted
or Lewis sites with which the N lone pair interacts the larger
the shift upwards of the components and the inversion of
their relative intensity, as previously reported [29].
In the case of AN adsorbed onto H-FER, a quite complex
group of bands can be seen (Fig. 6a). For H-FER the doublet
at 2307 and 2270 cm
the bridging OHs and responsible for the ABC pattern.
The further doublet at 2334 and 2307 cm
−1
to determine a small (5–10 cm ) but significant shift down
of the overall ABC spectrum. This should correspond to an
increased average acidity of the residual hydroxyl groups,
−1
is due to AN interacting with
due either to a direct effect of Co2 species on the acidity of
the residual OHs or to an easier exchange of the most weakly
acidic OHs. In any cases, the patterns of the exchanged
samples are much less intense as a result of the partial
disappearance of the bridging hydroxyl groups.
+
−1
has been
assigned to the AN interacting with extraframework or
external Al Lewis sites, which are very strong acids. For
H-MFI (Fig. 7a), the spectrum is more complex: here the
interactions with Lewis sites are evidenced by a doublet
In Figs. 6 and 7, the spectra of adsorbed species arising
from AN adsorption onto the catalysts are shown in the re-
−
1
at 2337 and 2299 cm and, again, the main doublet is due

C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
185
−
1
to H-bonded species (2309 and 2270 cm ). Actually, an
−1
additional component is observed at 2286 cm
.
The spectra of the Co-exchanged samples appear signif-
icantly different from those of the H-precursors and similar
to each other (Figs. 6b and 7b). Both of them show a sin-
gle doublet exactly at the same wavenumbers, 2319 and
−1
2
290 cm , very strong in both cases, that resists outgassing
at 473 K, characterised by a comparable intensity of the two
components. This doublet is certainly due to AN interacting
with the Co sites typical of the Co-exchanged zeolites. In
principle, we cannot distinguish whether this cobalt actually
is exchanging for the bridging OHs in the inner surface of
the cavities or the external silanols. On the other hand both
are likely similar, because on both the internal surface and
the external one Co ions are nearly naked, oxide anions be-
ing substantially absent. The shape of these spectra seems
to point out that Co species behave as medium-strong Lewis
acidic sites.
Fig. 9. NH -TPD curves of (a) Co O , (b) H-FER activated at 773 K,
(c) Co-H-FER.
3
3 4
2
−1
plied by Carlo Erba, 15 m g ). The spectrum of H-MFI
shows two partially overlapping peaks at low temperature
(
(
379, 416 K) and a well-resolved peak at higher temperature
632 K). Similar TPD curves were reported for the H-MFI
To have even more information, we also investigated
the adsorption of 2,2-dimethylpropionitrile (pivalonitrile) as
a basic probe. In the case of Co-containing samples we found
an adsorbed species characterized by a CN stretching band
sample with comparable Si/Al ratio [30–33]. The low-
temperature signals are supposed to be due to NH3 hydrogen
bonded to weak acid sites, probably silanol groups. The
−1
+
shifted to very high frequencies, up to above 2300 cm
,
high-temperature signal is due to decomposition of NH4
+
which is not found on protonic zeolites. Taking into account
that this molecule does not enter the cavities of MFI and
FER zeolites [29] it is evident that new Lewis sites, i.e.,
Co cations, are also located at the external surfaces of
Co-containing zeolites, in agreement with the data arising
from the OH stretching bands.
ions formed by interaction of NH3 with zeolitic H sites,
as it is the only peak observed during deammoniation of
+
NH4 –MFI. The Co-H-MFI sample also shows two TPD
peaks at low-temperature followed by a high-temperature
signal, similar to the parent H-MFI. However, the over-
all intensity is greatly increased (see also Table 2), and a
broad unresolved additional component, likely centered at
about 573 K, absent in the spectrum of the parent materi-
als, appears. This shows the presence of Co species acting
as medium-strong Lewis acid sites, in agreement with IR
studies. It can be noted that the high-temperature signal, rep-
resentative of protonated ammonia, is also present; by taking
into account that this peak appears on the tail of a large sig-
nal, we can say that it is of lower intensity in comparison
with the spectrum of the parent H-MFI. Therefore, we can
3
.4. NH3-TPD measurements
The surface acidity properties of the samples have also
been investigated by the NH3-TPD technique. The results of
TPD measurements are reported in Figs. 8 and 9 and in Ta-
ble 2. Only total amounts of desorbed NH3 are calculated
and reported in Table 2 because in many cases the over-
lapping of the peaks in TPD spectra makes their resolution
arbitrary. In Fig. 8 the NH3-TPD spectrum of Co-H-MFI is
reported together with those of the parent H-MFI (NH4-MFI
activated at 823 K) and the reference Co3O4 oxide (sup-
+
deduce that a significant amount of H sites are still present
in the Co-zeolite. This agrees with the uncompleted Co ex-
change, evaluated by FT-IR measurements, notwithstanding
that the Co amount is close to the theoretical exchange ca-
+
pacity. The peak due to NH4 decomposition, in the case of
Co-H-MFI, is shifted a little bit upwards, in agreement with
a slightly increased acidity of the residual acid sites, as de-
duced from IR spectra, too.
Table 2
NH TPD results
3
Total desorbed NH
Peak temperature
(K)
3
−
1
(mmol g
)
H-MFI
Co-H-MFI
H-FER
0.28
1.00
0.59
0.72
0.09
379, 416, 632
406, 445, 654
467, 674
467, 713
497
Co-H-FER
Fig. 8. NH -TPD curves of (a) Co O , (b) H-MFI activated at 773 K,
3
3 4
Co O
3
4
(c) Co-H-MFI.

186
C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
The NH3-TPD spectrum of Co-H-FER is reported in
Fig. 9, together with the spectra of the parent H-FER and of
the reference Co3O4. Similarly to H-MFI and in agreement
with literature data [34], the spectrum of H-FER shows a
low-temperature signal due to hydrogen-bridge-bondedNH3
and a high-temperature signal due to protonated NH3. The
spectrum of Co-exchanged zeolite shows a signal at 467 K,
with a shoulder at about 383 K, and a well-resolved peak
at 713 K. The high-temperature signal appears less broad
and more regular and shifted to higher temperature in
comparison to H-FER. The amount of ammonia adsorbed
onto Co-H-FER is higher than that onto H-FER, suggesting
the formation of new NH3 adsorbing sites due to Co ions.
The high-temperature peak is indicative, as observed for
+
Co-H-MFI, of the presence of a large amount of residual H ,
suggesting that only a partial Co exchange took place. The
shift of the maximum to higher temperature suggests an
increase of the strength of protonic acid sites, as already
observed in FT-IR spectra.
The amount of adsorbed ammonia in Co-H-zeolites
(
Table 2) indicates that two ammonia molecules can adsorb
at each Co ion, and this amount adds to ammonia adsorbed
on residual protons.
Fig. 10. H -TPR curves of (A) Co O pretreated in air, (B) Co-H-MFI, and
2
3 4
(C), Co-H-FER. (a) pretreatment in air, (b) pretreatment in He.
3
.5. TPR analysis
a peak at 820 K is evidenced. In contrast, the most intense
peak at 1023 K is not influenced by the atmosphere of
thermal treatment. This last signal cannot be attributed to
Co oxide, since it is completely reduced at about 773 K
(Fig. 10A). As suggested by Wang et al. [35], this signal
The reducibility of Co sites in Co-exchanged zeolites
has been studied by H2-TPR. These measurements have
been carried out on samples pretreated at 773 K either
in air or in He flow. TPR measurements have also been
performed on the parent zeolites and on the reference
Co3O4, for comparison. The TPR profiles are compared
in Figs. 10A–C, the amounts of consumed H2, obtained
from integration of TPR signals, are reported in Table 3.
As expected, the original zeolites give no reduction signals.
As shown in Fig. 10A, Co3O4 gives two TPR peaks at 596
and 714 K. The hydrogen consumption corresponding to
the two reduction peaks (Table 3) is very close to the
can be attributed to Co2 exchanging protons, which should
be more resistant to reduction, because it is stabilised by
the negatively charged zeolite framework. The intensity of
the TPR peaks appearing at lower temperatures is affected
by the treatment atmosphere (i.e., helium or air). This
suggests that they are representative of reduction of cobalt
oxide species that can easily change the oxidation state.
The correspondence of the temperatures of such peaks to
those of reduction of the Co3O4 phase seems to confirm
this hypothesis. We can suppose that small amounts of
easily reducible cobalt oxide phase, perhaps in nanosized
particles, escaping the above characterisation techniques
(FT-IR, UV–vis, XRD), have formed. An additional signal
+
3
+
2+
expected values for the two-step reduction: Co → Co
and Co2 → Co . The spectrum of Co-H-MFI pretreated in
air (Fig. 10B) shows three peaks at 559, 723, and 1023 K.
For the sample pretreated in He we can notice that the first
peak disappears while the second decreases in intensity and
+
0
Table 3
TPR results: peak temperatures, consumed H amounts, and molar ratios H /Co
2
2
Peak 1
nH2 c
Peak 2
nH2 c
Peak 3
nH2 c
Total
Catalyst
T (K)
H /Co
2
T (K)
H /Co
2
T (K)
H /Co
2
nH2 c
H /Co
2
a
b
a
Co-H-MFI
Co-H-MFI
Co-H-FER
559
–
594
612
596
0.03
–
0.06
0.04
4
0.10
–
0.15
0.10
0.32
723
728
–
–
714
0.05
0.05
–
–
12.5
0.17
0.17
–
–
1.00
1023
1032
1014
1017
–
0.23
0.20
0.30
0.32
–
0.80
0.69
0.80
0.85
–
0.31
0.25
0.36
0.36
16.5
1.0
0.8
0.9
0.9
1.32
b
Co-H-FER
a
Co O
3
4
a
Pretreated in air.
Pretreated in He.
Consumed hydrogen, mmol g
b
c
−1
.

C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
187
is observed at somewhat lower temperature (503–563 K).
Similar behaviour was reported by Jong and Chen [36] for
Co-MFI prepared by impregnation. Taking into account the
results obtained by Wang et al. [35], this signal can be
attributed to the presence of polynuclear Co oxo-ions that are
0
reduced to Co nanoclusters by H2. Such multinuclear ions
can play an important role in adsorption and activation of
NOx species [35]. However, the same result can arise from a
fraction of exchanged Co ions oxidized to the trivalent state.
The amount of consumed hydrogen per cobalt ion ranges
from 0.8 to 1 (Table 3), thus being lower than the expected
one for reduction of total cobalt, which includes exchanged
2
+
Co
and cobalt oxide species. This could be due to a
defective estimate of the high-temperature peak, since the
Fig. 11. NO and CH conversion and selectivity to N on Co-H-MFI and
reduction of exchanged Co2+ occurs at a slow rate and
4
2
−1
Co-H-FER. GHSV = 60,000 h
.
probably is not complete under isothermal conditions within
the time of the tests.
greatly exceeding values. This clearly suggests the occur-
rence of a side reaction also at high temperature, mainly
methane combustion, as previously reported [6]. The occur-
rence of side reactions consuming CH4 hardly explains the
presence of maximum of NO conversion because methane
is never limiting even at high temperature. The maximum
in NO conversion is related to the occurrence of NO oxida-
tion to NO2, which predominantly occurs at low temperature
but approaches equilibrium conversion at high temperature
The H2-TPR profiles of Co-H-FER (Fig. 10C) are quite
similar to those of Co-H-MFI. They show peaks at 594 K
that can be attributed to easily reducible cobalt species, and
the predominant one at 1014 K due to hardly reducible
species. The intensity of low-temperature peaks is affected
by the pretreatment atmosphere, while the high-temperature
one is insensitive to pretreatment conditions. The dominant
2
+
peak can be attributed again to reduction of exchanged Co
ions. The signals at 594 K is likely related to the presence
of polynuclear Co oxo-ions, as observed for the Co-H-MFI
sample. The somewhat higher peak temperature seems to
indicate a lower reducibility of these species in comparison
with Co-H-MFI. Moreover the signal in the range 673–
(
see below), and gives rise to a decrease of NO conversion
with temperature due to thermodynamics. For Co-H-FER
the maximum of NO conversion is shifted to higher tem-
perature probably because this catalyst is less active than
Co-H-MFI also in NO oxidation catalysis, and thus shows
the increase of NO conversion to NO2 at higher temperature.
The catalytic behaviour of H-zeolites was also investi-
gated. It can be observed that the not-exchanged zeolites
show low activity, with maxima of NO conversion of about
7
23 K is very weak, suggesting the absence of a nanosized
Co oxide phase, if any, different from the Co-H-MFI sample.
The value of total H2 consumption is lower than expected for
reduction of all Co, probably due to very slow reduction of
exchanged Co2
.6. Catalytic activity tests
The results of catalytic activity tests are reported in
+
.
−
1
1
0% at 650–700 K (GHSV = 60,000 h ) and decreasing
3
values at higher temperatures. The samples are active mainly
for NO oxidation to NO2, which is practically the only reac-
tion product at T > 650 K, and to a negligible extent towards
NO reduction, in agreement with results reported elsewhere
[33,37]. Similar behaviour was observed for H-FER.
Figs. 11–14. In all tests the main products were CO2, NO2,
and N2 with trace amounts of CO. In no conditions was
N2O detected. In preliminary tests it was ascertained that no
homogeneous NO oxidation to NO2 occurred.
The NO and CH4 conversions for Co-H-MFI and Co-H-
FER are reported in Fig. 11 together with selectivity to N2.
Co-H-MFI appears more active than Co-H-FER in the whole
temperature range. Both catalysts show a maximum in NO
conversion, shifted to higher temperature for Co-H-FER. Se-
lectivities to N2 are increasing with temperature and exceed
The influence of GHSV on the most active catalyst, that
is Co-H-MFI was investigated (Fig. 12). As expected, the
increase of GHSV leads to a decrease of NO conversion,
which reaches a maximum value of about 35% at 673 K.
However, NO was converted mainly to NO2, which was
the main product at low temperature. As shown in Fig. 13,
with increasing temperature, conversion to NO2 decreases
while conversion to N2 increases and overcomes that to
NO2 at 773 K. These data clearly indicate a high activity
of Co-H-MFI towards NO oxidation at low temperature.
In order to obtain more information about the catalyst ac-
tivity towards this reaction, tests were carried out feeding
NO and O2 in the same concentrations as during CH4-SCR
tests, but in the absence of methane. The results are re-
ported in Fig. 14, which also reports NO conversion to
5
0% only at high temperatures. This is due to the predom-
inant formation of NO2 at low temperature in agreement
with previous data for Co-MFI catalysts [8,11]. By contrast,
CH4 conversion is always increasing with temperature, thus
leading to a varying CH4/NO reaction ratio: at tempera-
tures lower than the maximum, this is close to or slightly
higher than 1:2, while at higher temperatures it reaches

188
C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
Fig. 12. NO conversion and selectivity to N on Co-H-MFI at differ-
2
ent GHSV.
−1
Fig. 14. NO oxidation to NO on Co-H-MFI. GHSV = 30,000 h
.
2
NO2 in equilibrium conditions. It can be observed that
Co-H-MFI shows significant activity and that conversion to
NO2 reaches equilibrium values starting from 650 K. At
T ꢀ 650 K conversion to NO2 is practically the same as
that observed in CH4-SCR tests, while it is higher at higher
temperatures. This suggests that under CH4-SCR conditions
NO2 formation (prevailing at low temperature) can compete
with the main reaction, giving rise to N2; NO reduction by
methane occurs only at higher temperatures where oxidation
to NO2 is thermodynamically limited. Similar results were
reported by Cant and Liu for Co-MFI [11].
3
4
. A reactive intermediate containing C, H, N, and O
nitromethane- or nitrosomethane-like) is formed from
the methyl group and an adsorbed NOy species; this
intermediate reacts with gaseous NO or NO2, leading
to the formation of the N2 molecule.
(
. Methane activation is the rate-determining step.
Otherwise, several different details are proposed by the
different authors. In particular, disagreement can be found
concerning the role of NO2 either as a gaseous intermediate
or only in the adsorbed state.
Different points of view are also reported in recent litera-
ture on the nature of active sites. Kauchy et al. [8] reported
4
. Discussion
that exchanged Co sites (characterised as Co2 [21]) exhibit
+
The picture concerning the reaction mechanism upon
activity in CH -SCR while protons or oxide-like Co species
4
CH4-SCR is quite complex, and different authors came only
to partial agreement. In particular, taking into account the
papers published by Armor and co-workers [6,12], Cant
and Liu [11], Kauchy et al. [8], Wang et al. [38], and
Ivanova et al. [39], the following points about the reaction
mechanism appear largely shared:
contribute to CH -SCR by enhancement of the oxidation of
4
NO to NO . Wang et al. [38] point to the redox properties of
2
catalysts and suppose that the activity of Co-MFI catalysts
is due to the redox couple Co2 /Co . Zerovalent Co species
can be formed by reduction of Co polynuclear oxo-cations
that appear easily reducible. According to Yan et al. [19] Co
cations act in the oxidation of NO to NO2 while Brønsted
acid sites act in the true reduction step by methane.
+
0
1
2
. NO in the presence of O2 is oxidized to an adsorbed
Ng aOs eyo us sp eN c Oi e2s. (y = 2, 3) and can also give rise to
. Methane is activated by adsorbed NOy species via
abstraction of a hydrogen atom.
Our own experimental data show that under CH -SCR
4
conditions the following overall reactions can actually oc-
cur:
CH4 + 2NO + O2 → N2 + CO2 + 2H2O,
(2)
(3)
(4)
1
NO + O2 = NO2,
2
CH4 + 2O2 → CO2 + 2H2O.
At low temperatures reaction (3) is the main one over
Co-H-zeolites. At high temperatures, when NO oxidation (3)
is thermodynamically limited, reactions (2) and (4) become
the prevailing ones. This suggests that NO2 could not be
a necessary intermediate in the CH4-SCR reaction starting
from NO; otherwise, its formation should be inhibited to
have a good catalyst. The data we present here suggest that
the mechanism is rake-type and that an intermediate NOy
(
like that previously observed, e.g., by Ivanova et al. [38]
Fig. 13. Conversion of NO and conversion to N and NO on Co-H-MFI.
2
2
for Co-zeolites, identified as a nitrate species) is likely

C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
189
formed from NO and oxygen and can desorb as NO2
at low temperature when NO2 is a stable compound) or
as NO or react with methane, giving rise finally to N2.
This agrees, among other things, with the effect of space
velocities towards selectivity to nitrogen upon CH4-SCR
zeolites could play a nonnegligible role in catalysis. Such
a contribution, generally disregarded in the literature, is now
under investigation.
(
(
(
Fig. 12) and with the effect of methane on NO oxidation
Fig. 14). Interestingly, Co-H-FER, which is less active in
5. Conclusions
converting NO to NO2 at low temperature than Co-H-MFI,
gives, however, higher selectivities to N2.
Partially exchanged Co-H-FER and Co-H-MFI have been
prepared and their properties characterised using different
physicochemical techniques. Their catalytic activity towards
CH4-SCR and NO oxidation was studied under diluted
conditions.
Exchanged Co cations are certainly active in all three
reactions (2)–(4). They have been characterized to be in
the bivalent state in vacuum by UV–vis spectroscopy, in
agreement with previous data [21], and this also agrees with
their evident medium-strong Lewis acidity, found here both
by IR of adsorbed acetonitrile and by TPD of adsorbed
ammonia. TPR data obtained in the present work show
The main conclusions of this study can be summarized as
follows:
– Partial Co exchange occurs with strong acid sites in
the inner zeolitic cavities and also with weak silanol
acidic groups on the external crystals surface; Co ions
at the external zeolite surface are detected by IR, and
could have a role in catalysis and in the spectroscopic
behaviour in the UV–vis region.
that Co2 ions exchanged in MFI and FER zeolites are
reduced by hydrogen at about 1000 K; thus it is unlikely that
such species can be reduced under the reaction conditions.
On the other hand, far more easily reducible species are
detected from TPR low-temperature signals. These species
appear to be also easily oxidizable, since the TPR signals are
clearly influenced by the atmosphere of thermal treatment.
+
– Both Co-H-FER and Co-H-MFI retain large amounts of
residual protonic sites whose acid strength seems to be
a little increased by the Co exchange; residual silanol
groups are also found at the external zeolite surface.
– Cobalt ions act as medium-strength Lewis acid sites;
2
+
ions, either those
This suggests that a fraction of Co
exchanging protons or a fraction which are clustered in
nanosized oxide particles, are oxidized to Co3 when treated
in air at 773 K: the subsequent reduction of Co to Co in
TPR measurement contributes to the low-temperature peaks.
+
3
+
2+
2+
TPR measurements give evidence that exchanged Co
ions are hardly reducible. However, a fraction of Co
oxide species (either exchanged cations or polynuclear
oxo-species) are easily reducible and oxidizable.
– Co-H-FER and Co-H-MFI zeolites are active both in
NO selective catalytic reduction by methane and in the
NO oxidation to NO2. H-zeolites are weakly active,
mainly for NO oxidation.
Thus, our data suggest that a Co2 /Co redox couple can
be active for CH4-SCR as well as for NO oxidation to NO2.
This couple is likely active also for methane and other alkane
oxidation, as discussed previously for catalytic combustion
over cobalt oxides [40,41].
+
3+
Protonic zeolites are weakly active in NO oxidation to
NO2 and, even much less, for CH4-SCR. On the other hand,
the very acidic OHs of zeolites are well known to be active
in adsorbing and perturbing hydrocarbon molecules [42]
and nitrogen oxides as well, and can also act as methane
combustion catalysts [43]. In particular, it is evident that
hydrocarbons whose size allows penetration into the cavities
tend to concentrate there, as reported elsewhere [42]. Our
data indicate that the average Brønsted acidity of the residual
bridging OHs is even a little increased with respect to
nonexchanged protonic zeolites. It seems quite reasonable
that the ability of these sites to adsorb both reactants can
favor in more than one way the CH4-SCR reaction. So the
reported cooperating effect of Co and proton sites can be due
to the enhanced rate of formation of the NOy intermediate
or to the enhanced rate of reaction of this intermediate with
methane, whose concentration in the cavities is enhanced
due to its interaction with the protons. In our opinion, this
explains the cooperating effect of protons better than their
weak activity in NO oxidation, although both things can
occur to some extent.
– NO oxidation to NO2 competes with and prevails over
CH4-SCR when it is thermodynamicallyfavoured, in the
sense that NO2 is formed instead of N2. CH4-SCR is
predominant when NO oxidation to NO2 is no more a
favoured reaction.
– It is proposed that NO reduction occurs via a redox
mechanism involving essentially a Co2 /Co redox
couple and an NOy adsorbed intermediate (possibly a
nitrate species). This species can either give rise to
NO2 gas or react with methane.
+
3+
– According to previous studies, it is proposed that pro-
tonic sites cooperate in CH4-SCR mainly by adsorbing
either NO (so favouring the formation of the interme-
diate) or methane (so favouring its reaction with the
intermediate) or both.
Acknowledgments
The present work has been supported by the MICA
(Italian Ministry of Industry, Trade and Handicraft) in the
frame of Energy Research Program for the Italian Electrical
Moreover, it must be considered that the sites, cobalt
cations and protons, present on the external surface of

190
C. Resini et al. / Journal of Catalysis 214 (2003) 179–190
System (MICA Decree of January 26, 2000), Project on
Interaction between the electrical system and environment
SOSTE).”
[21] D. Kauchy, J. Dedechek, B. Wichterlova, Micropor. Mesopor. Ma-
ter. 31 (1999) 75.
“
(
[22] J. Dedechek, D. Kauchy, B. Wichterlova, Micropor. Mesopor. Ma-
ter. 35–36 (2000) 483.
[23] M. Trombetta, G. Busca, M. Lenarda, L. Storaro, M. Pavan, Appl.
Catal. A 182 (1999) 225.
References
[24] T. Armaroli, M. Trombetta, A. Gutièrrez Alejandre, J. Ramirez Solis,
G. Busca, Phys. Chem. Chem. Phys. 2 (2000) 3341.
[
25] R.M. Dessau, K.D. Schmitt, G.T. Kerr, G.L. Woolery, L.B. Alemany,
J. Catal. 104 (1987) 484.
[
[
[
[
1] H. Bosch, F.J. Janssen, Catal. Today 2 (1988) 369.
2] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B 18 (1998) 1.
3] Y. Li, J.N. Armor, Appl. Catal. B 1 (1992) L31.
4] W. Held, A. Konig, T. Richter, L. Puppe, Soc. Aut. Eng., 1990, Paper
9
[26] A. Zecchina, S. Bordiga, G. Spoto, L. Marchese, G. Petrini, G. Leo-
fanti, M. Padovan, J. Phys. Chem. 88 (1992) 2959.
[
[
27] M. Trombetta, G. Busca, J. Catal. 187 (1999) 521.
28] A.G. Pelmenschikov, G.H.M.C. van Wolput, J. Jänchen, R.A. van
Santen, J. Phys. Chem. 99 (1995) 3612;
00496.
5] M. Iwamoto, H. Yahiro, Y. Shundo, Y. Yu-u, N. Mizuno, Shokubai 32
1990) 430.
[
(
A.G. Pelmenschikov, R.A. van Santen, J. Phys. Chem. 97 (1993)
[6] J.N. Armor, Catal. Today 26 (1995) 147.
10678.
[7] V. Indovina, M.C. Campa, S. De Rossi, G. Ferrarsi, Appl. Catal. B 8
[
[
[
[
[
29] G. Busca, Phys. Chem. Chem. Phys. 1 (1999) 723.
30] H. Ichibashi, M. Kitamura, Catal. Today 73 (2002) 23.
31] R.Q. Long, R.T. Yang, J. Catal. 198 (2001) 20.
(
1996) 315.
8] D. Kauchy, A. Vondrovà, J. Dedechek, B. Wichterlova, J. Catal. 194
2000) 318.
9] A.J. Desai, V.I. Kovalchuk, E.A. Lombardo, J.L. D’Itri, J. Catal. 184
1999) 396.
[
[
(
32] F. Lonyi, J. Valyon, Micropor. Mesopor. Mater. 47 (2001) 293.
33] F.G. Requejo, J.M. Ramallo-Lopez, E.J. Lede, E.E. Miró, L.B.
Pierella, O.A. Annunziata, Catal. Today 54 (1999) 553.
34] P. Cañizares, A. Carrero, P. Sànchez, Appl. Catal. A 190 (2000) 93.
35] X. Wang, H. Chen, W.M.H. Sachtler, Appl. Catal. B 29 (2001) 47.
36] S. Cheng, S.-J. Jong, Appl. Catal. A 126 (1995) 51.
37] E. Kikuchi, M. Ogura, N. Aratani, Y. Sugiura, S. Hiromoto, K. Yogo,
Catal. Today 27 (1996) 35.
(
[
10] S.A. Beloshapkin, E.A. Paushktis, V.A. Sadykov, J. Mol. Catal. A
[
[
[
[
Chem. 158 (2000) 355.
[
[
[
[
[
11] N.W. Cant, I.O.Y. Liu, Catal. Today 63 (2000) 133.
12] Y. Li, P.J. Battavio, J.N. Armor, J. Catal. 142 (1993) 561.
13] M. Misono, CATTECH 2 (1998) 183.
14] M. Shelef, Chem. Rev. 95 (1995) 209.
15] P. Praserthdam, N. Mongkolsiri, P. Kanchanawanichkun, Catal. Com-
mun. (2002), in press, available on the Internet.
[
38] X. Wang, H.Y. Chen, W.M.H. Sachtler, J. Catal. 197 (2001) 281.
[39] E. Ivanova, K. Hadjiivanov, D. Klissurski, M. Bevilacqua, T. Armaroli,
G. Busca, Micropor. Mesopor. Mater. 46 (2001) 299.
[
[
[
16] Z. Li, M. Flytzani-Stephanopoulos, Appl. Catal. B 22 (1999) 35.
17] L. Gutierrez, A. Boix, J. Petunchi, Catal. Today 54 (1999) 451.
18] L. Ren, T. Zhang, D. Liang, C. Xu, J. Tang, L. Lin, Appl. Catal. B 35
[40] E. Finocchio, G. Busca, V. Lorenzelli, V. Sanchez Escribano, J. Chem.
Soc. Faraday Trans. 92 (1996) 1587.
[41] G. Busca, V. Lorenzelli, G. Ramis, V. Sanchez Escribano, in: L. Guczi,
F. Solymosi, P. Tetenyi (Eds.), New Frontiers in Catalysis, Elsevier,
Amsterdam, 1993, pp. 2661–2664.
[42] T. Armaroli, M. Trombetta, A. Gutièrrez Alejandre, J. Ramirez Solis,
G. Busca, Phys. Chem. Chem. Phys. 2 (2000) 3341.
[43] J.R. Anderson, P. Tsai, Appl. Catal. 19 (1985) 141.
(
2002) 317.
19] J.-Y. Yan, H.H. Kung, W.M.H. Sachtler, M.C. Kung, J. Catal. 175
1998) 294.
20] W.M. Butler, C.L. Angell, W. McAllister, W.M. Risen, J. Phys. Chem.
1 (1977) 2061.
[
[
(
8