NO reduction with ethanol on Pd-Mo/Al2O3 catalysts

Article abstract of DOI:10.1016/S0021-9517(03)00272-0

The NO reduction with ethanol was studied on Pd/Al₂O ₃ and Pd-Mo/Al₂O₃ catalysts. The Pd/Al ₂O₃ catalyst was more active for NO conversion, although the selectivity for N₂ formation was the same on both catalysts. IR and TPD analysis of adsorbed ethanol showed formation of stable acetate species upon dehydrogenation of ethanol to acetaldehyde at reaction temperature. TPSR measurements suggested that NO preferentially adsorbs on Pd while ethanol selectively adsorbs on alumina and/or partially reduced molybdenum oxide (MoO_x). The catalytic results further suggested that the reaction mechanism was the same on both catalysts and that the reaction occurred between adsorbed NO and acetate species at the metal/oxide interface. A reaction mechanism was proposed where the limiting step was the dehydrogenation of ethanol to acetaldehyde.

Full text of DOI:10.1016/S0021-9517(03)00272-0

Journal of Catalysis 220 (2003) 358–371
www.elsevier.com/locate/jcat
NO reduction with ethanol on Pd–Mo/Al O catalysts
2
3
1
∗
L.F. de Mello, F.B. Noronha, and M. Schmal
NUCAT-PEQ-COPPE, Universidade Federal do Rio de Janeiro, Ilha do Fundão, CP 68502, CEP 21941, Rio de Janeiro, Brazil
Received 17 March 2003; revised 10 June 2003; accepted 12 June 2003
Abstract
The NO reduction with ethanol was studied on Pd/Al O and Pd–Mo/Al O catalysts. The Pd/Al O catalyst was more active for NO
2
3
2
3
2 3
conversion, although the selectivity for N formation was the same on both catalysts. IR and TPD analysis of adsorbed ethanol showed
2
formation of stable acetate species upon dehydrogenation of ethanol to acetaldehyde at reaction temperature. TPSR measurements suggested
that NO preferentially adsorbs on Pd while ethanol selectively adsorbs on alumina and/or partially reduced molybdenum oxide (MoOx). The
catalytic results further suggested that the reaction mechanism was the same on both catalysts and that the reaction occurred between adsorbed
NO and acetate species at the metal/oxide interface. A reaction mechanism was proposed where the limiting step was the dehydrogenation
of ethanol to acetaldehyde.

2003 Elsevier Inc. All rights reserved.
Keywords: Pd–Mo catalysts; TPD of ethanol; IR; NO + ethanol reaction
1
. Introduction
tion to the need for more fundamental research regarding
the adsorption properties and reaction mechanistic studies
in order to understand and develop adequate catalytic sys-
tems. In this sense, Cordi and Falconer [5] studied the ox-
idation of ethanol and acetaldehyde on alumina-supported
palladium catalysts using temperature-programmed desorp-
tion (TPD) and oxidation (TPO) while Idriss et al. [6] and
Yee et al. [7] studied the reactions of acetaldehyde [6] and
ethanol [7] on the surface of CeO2-supported palladium cat-
alysts also using temperature-programmed desorption and
infrared analyses. Ukisu et al. [8] and Miyadera et al. [9]
have performed infrared study of the reduction of NO with
alcohols on alumina-supported silver catalysts and they have
identified the presence of isocyanate species as an important
reaction intermediate. Toops et al. [10] studied the NO re-
duction with methanol over La2O3 and they have suggested a
mechanism where two types of adsorption sites for methanol
and NO were present.
Recently, the use of oxygenated organic compounds in
gasoline has been growing [1]. The addition of such com-
pounds reduces the emission of olefins, aromatics, complex
hidrocarbons, and SOx. The most commonly used additives
are MTBE (methyl tert-butyl ether) and alcohols (methanol
and ethanol). However, the use of ethanol, for instance,
may increase the direct emission of aldehydes (acetaldehyde
and formaldehyde) and also of unburned alcohol [2]. Apart
from their potential carcinogenic effects, these new emission
products may react in the atmosphere to form other toxic
compounds and also contribute to the formation of urban
smog [3]. Therefore, these facts must be taken into account
when projecting and testing catalytic systems with the pur-
pose of automotive pollution control.
Although much work has focused attention on the reduc-
tion of NO with several hydrocarbons on various catalysts
[
1,4], little has been done to determine the effect of the pres-
In previous works [11–13], the Pd–Mo system was shown
to be very active and selective for N2 formation during the
NO + CO reaction and both the surface characterization
and the reduction of NO with CO on Pd–Mo/Al2O3 cata-
lysts were thoroughly investigated. For instance, Schmal et
al. studied Pd–xMo/Al2O3 catalysts, where x indicates the
amount of Mo in each catalyst (2, 8, 14, and 20%). The au-
thors performed diffuse reflectance spectroscopy (DRS) and
X-ray diffraction (XRD) measurements on the oxidized sam-
ence of oxygenated organic compounds on the reduction
of NO in exhaust gases. Recent papers have called atten-
*
Corresponding author.
E-mail addresses: fabiobel@int.gov.br (F.B. Noronha),
schmal@peq.coppe.ufrj.br (M. Schmal).
1
Present address: Instituto Nacional de Tecnologia - INT, Av. Venezuela
8
2, CEP 20081-310, Rio de Janeiro, Brazil.
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0021-9517(03)00272-0

L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
359
ples and the XRD results clearly showed the presence of a
crystalline phase (bulk MoO3) for the samples containing 14
and 20% Mo, while the pattern for the 2% Mo-containing
catalyst was similar to that for pure alumina. With the aid
of DRS analyses, they showed that for the 8% Mo catalyst
a monolayer of octahedrally coordinated molybdenum was
the main species present. Schmal et al. [12] and Noronha
et al. [13] used temperature-programmed reduction (TPR)
experiments on the Pd–Mo catalysts and observed that the
Pd–Mo samples presented interesting differences when com-
pared to the Pd-only and Mo-only catalysts, clearly showing
an enhancement on the reduction of MoO3 (lower reduction
temperatures) when Pd was present. According to them, this
result indicated that Pd facilitates the reduction of molybde-
num oxide, thus implying that palladium is in close contact
and interacts with the Mo species.
When these catalysts were used for the NO + CO reac-
tion [12], it was found that the presence of Mo enhanced
catalytic activity and selectivity for N2 production. Schmal
et al. suggested a redox mechanism to explain the increase
in activity and selectivity for these catalytic systems and
the intimate contact between Pd particles and MoOx species
would favor this mechanism.
The present paper is a continuation of the same line of
work as the previous ones and the main objective here is to
study the NO reduction with ethanol on the Pd–Mo catalysts,
trying to understand the mechanistic differences between
this reaction and the NO + CO reaction, thus giving an inte-
grated view of the Pd–Mo system. For this purpose, the ad-
sorption properties of ethanol and also the reduction of NO
with ethanol on Pd–Mo/Al2O3 catalysts were investigated
using temperature-programmed desorption, temperature-
programmed surface reaction (TPSR), and infrared (IR)
analyses as well as catalytic activity measurements.
the catalysts were dried at 423 K, for 30 min. The sam-
ples were then reduced under pure hydrogen (30 cm /min)
3
at 773 K (5 K/min) and vacuum was made at the reduc-
tion temperature for 1 h followed by cooling to the ad-
sorption temperature. The irreversible adsorption was de-
termined at 343 K according to the method described by
Benson et al. [14].
2.2.2. Infrared spectroscopy of adsorbed ethanol
Ethanol adsorption was monitored using a Fourier trans-
form infrared spectrometer (Perkin Elmer 2000). Before the
analysis, the catalysts (25 mg pellets) were reduced with H2
at 773 K for 1 h. After evacuation at the reduction tempera-
ture for 1 h and cooling down to room temperature, ethanol
uptake was allowed until saturation at 298 K, followed by
FTIR measurements under vacuum at 298, 373, 423, 473,
523 and 573 K. For the 20% Mo-containing catalysts, no in-
frared measurements were made because there was no trans-
mittance through the reduced samples.
An additional infrared measurement was made for the
Pd8Mo sample. After the catalyst was reduced (same pre-
treatments as explained above), the pellet was exposed to a
mixture of 7.6 Torr NO/5.0 Torr ethanol initially at room
temperature and then raised up to 573 K and cooled to room
temperature again. IR measurement were taken before and
after heating. A flow of He was passed through the sample
and then another measurement was performed. Finally, vac-
uum was established and IR was taken again.
2
.2.3. Temperature-programmed desorption (TPD)
The temperature-programmed desorption experiments of
adsorbed ethanol were made in a similar manner as in a pre-
vious work [11]. Prior to TPD analyses, all samples were
3
purged under helium flow (50 cm /min) from room tem-
perature up to 823 K (10 K/min). The samples were then
cooled to room temperature and reduced under flowing H2
2
. Experimental
3
(
30 cm /min) up to 773 K (5 K/min), remaining at that tem-
perature for 2 h. Following reduction, the system was purged
with helium flow at the reduction temperature for 30 min and
cooled to room temperature. The adsorption of ethanol was
made at room temperature through pulses of an ethanol/He
mixture (until saturation), which was obtained by flowing
He through a saturator containing ethanol at 298 K. After
2
.1. Catalyst preparation
The 8%Mo and 20%Mo/Al2O3 catalyst were prepared
by wet impregnation of γ -Al2O3 (Engelhard; BET area =
2
2
4
16 m /g) with an aqueous solution of (NH4)6Mo7O24·
H2O. The samples were dried at 383 K for 22 h and cal-
adsorption, the catalyst was heated at a 20 K/min rate up to
cined under flowing air at 773 K for 2 h. Pd/Al2O3 and
Pd–Mo/Al2O3 samples were prepared by wet impregnation
of Al2O3 and Mo/Al2O3, respectively, with a 10% nitric
acid solution of Pd(NO3)2 (Aldrich). The samples were then
dried at 373 K and calcined under flowing air at 773 K for
3
8
23 K in flowing helium (50 cm /min). The products were
monitored using a quadrupole mass spectrometer (Balzers,
PRISMA).
The calibration of the mass spectrometer as well as the
correction procedure to determine the distribution of the des-
orbed products were described in detail elsewhere [11].
2
h. The composition of the catalysts are presented in Ta-
ble 1.
2
2
.2. Catalyst characterization
2.2.4. Temperature-programmed surface reaction
The temperature-programmed surface reaction was made
following the same procedure as described for TPD analysis.
.2.1. Hydrogen chemisorption
The H2 chemisorption measurements were made on an
3
All samples were first purged with helium flow (50 cm /min)
ASAP 2000C (Micromeritics) equipment. Prior to reduction,
from room temperature up to 823 K (10 K/min), cooled to

360
L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
room temperature, and reduced under flowing H2 (30 cm3/
min) up to 773 K (5 K/min), remaining at that temperature
for 2 h. After reduction, the system was purged with helium
at the reduction temperature for 30 min and cooled to room
temperature. Ethanol was adsorbed at room temperature un-
til saturation of the surface. After adsorption, the samples
of MoO3 upon exposure to H2 at high temperatures orig-
inates a suboxide that tends to migrate onto the Pd surface,
thus preventing its contact with H . Due to this effect, the Pd
2
dispersion was calculated only for the Pd/Al O and it was
2
3
found to be 23%. Such low Pd dispersion is usually achieved
when palladium nitrate is used as the precursor salt [13].
were purged with He and a flow of a 1%NO/He mixture
3
(
50 cm /min) was passed as the temperature was raised at
3.2. Infrared spectroscopy of adsorbed ethanol
a 20 K/min rate until 823 K. The reaction products were
monitored using a quadrupole mass spectrometer (Balzers,
PRISMA), as described before [11].
The IR analysis of ethanol adsorption on alumina (Fig. 1)
showed characteristic bands of ethoxy species at 1075, 1120,
−1
1
168, 1389, and 1447 cm . Similar bands were also ob-
2
.2.5. Catalytic activity
served by Greenler [17] after adsorption of ethanol on alu-
−1
The catalytic experiments were performed in a glass mi-
mina. The features present at 1585 and 1463 cm might be
related to the asymmetric and symmetric stretching of ac-
etate species, respectively. In fact, Greenler [17] observed
the appearance of such acetate species after heating the alu-
mina surface previously exposed to ethanol. Several reports
have also observed the appearance of the above-noted bands
related with acetate species upon adsorption of ethanol on
different surfaces [6,7,18]. However, in this work, an IR
spectrum for the alumina surface without adsorption of
ethanol was also taken (not shown) and very weak bands
croreactor at atmospheric pressure. Prior to the reaction,
the catalysts (ca. 140 mg) were pretreated in flowing he-
lium (50 ml/min) at 823 K for 30 min and then reduced
with pure H2 at 773 K for 1 h. The feed mixture for the
NO reduction with ethanol consisted of 0.2% ethanol/0.3%
NO in He (balance) at a flow rate of 250 ml/min (space
−1
velocity = 82,500 h ). For the reaction gas feed, a 0.97%
NO/He (AGA) gas mixture was used while the ethanol/He
mixture was obtained by flowing He through a saturator con-
taining 99.999% pure ethanol (Merck) at 278 K. The efflu-
ent was analyzed by gas chromatography (Chrompack with
TCD detector, Chromosorb 102 column, and cryogen).
−1
around 1479 and 1586 cm were also observed. Although
alumina was calcined in flowing air at 773 K for 2 h, it is
possible that carbonate species still remained at the surface
and originated the bands noted above. Therefore, it is possi-
−1
ble that the bands presented in Fig. 1 (1463 and 1585 cm
)
3
. Results
3
.1. Hydrogen chemisorption
H2 chemisorption results are presented in Table 1. The
Mo-only catalysts showed virtually no hydrogen chemisorp-
tion while for the Pd-containing catalysts there was a drastic
decrease in the amount of H2 chemisorbed as the Mo con-
tent increased. This result shows that the presence of MoOx
species reduces the amount of Pd adsorption sites, which
in turn may indicate that the molybdenum oxides might be
covering the Pd particles. The decoration of Pd particles by
the partially reduced molybdenum oxide species was also
observed in previous works [12,13] and reported in other
papers [15,16]. Konopny et al. [15] have studied the prepara-
tion and characterization of alumina-supported Pd–Mo cata-
lysts and observed a similar H2 chemisorption effect on the
bimetallic samples. According to the authors, the reduction
Table 1
Catalyst composition and H chemisorption
2
Catalyst
Pd
Mo
H chemisorption
2
(wt%)
(wt%)
(µmol H /gcat)
2
Pd/Al O
0.69
–
0.63
–
–
8.3
8.3
18.6
17.0
6.64
0.00
1.40
0.11
0.16
2
3
8
Mo/Al O
2
3
Pd8Mo/Al O
2
3
20Mo/Al O
2
3
Pd20Mo/Al O
0.60
2
3
Fig. 1. IR spectra of adsorbed ethanol on Al O .
2
3

L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
361
Fig. 3. IR spectra of adsorbed ethanol on 8Mo/Al O .
2
3
Fig. 2. IR spectra of adsorbed ethanol on Pd/Al O .
2
3
sorbed on the partially reduced molybdenum oxide (MoOx).
For the adsorption of ethanol on CeO2-supported Pd cata-
lysts, Yee et al. [7] observed infrared bands at 1037 and
are relative to characteristic alumina bands (due to carbonate
species) and/or to the presence of acetate species. After heat-
ing the sample, the band intensities related to ethoxy species
decreased, while those initially related to acetate species
were not modified.
−1
1
078 cm on reduced cerium oxide and they have also
attributed these bands to ethoxy species on the surface of
partially reduced cerium oxide.
The IR spectra of adsorbed ethanol on Pd/Al2O3 (Fig. 2)
showed, at room temperature, the same bands attributed to
ethoxy species on alumina. However, as the temperature
was raised, the band intensities of the ethoxy species (1075,
For the 8Mo catalyst, the bands related to acetate species
−1
on alumina (around 1474 and 1587 cm ) were very weak as
the temperature was raised. On the other hand, these bands
were very clear on the Pd8Mo catalyst (mainly at 573 K),
suggesting, once again, that the presence of Pd favors the
formation of such species. On both samples, however, weak
−1
1
120, 1168, 1388, and 1446 cm ) decreased whereas the
intensity of the bands related to acetate species (1466 and
−1
1
581 cm ) increased. Such increase in intensity may be
−1
associated with the formation of acetate species, suggest-
ing that the presence of Pd favors the appearance of these
species.
bands appeared around 1562–1566 and 1654–1662 cm
.
Although it is not yet clear, these bands are possibly associ-
ated with acetate species adsorbed on the surface of partially
reduced molybdenum oxide. Infrared analysis of adsorbed
acetaldehyde on 8Mo and Pd8Mo (not shown here) pre-
sented similar bands in this region after heating, suggesting
that acetaldehyde might acts as a precursor in the formation
of these acetate species and that the presence of MoOx also
contributes to acetate formation.
Figs. 3 and 4 show the IR spectra of ethanol adsorbed on
8
Mo and Pd8Mo catalysts. In both cases the bands related to
ethoxy species adsorbed on alumina were less intense when
compared with the spectra obtained for the support (Fig. 1)
and for the Pd/Al2O3 catalyst (Fig. 2). This is probably due
to the covering of the alumina surface by the MoO3. A simi-
lar effect has been observed in previous works [11,12] and it
has also been reported that a monolayer of MoO3 on alumina
is obtained for Mo loadings between 8 and 12% [12,19].
Fig. 5 shows the results for the IR measurements after
a sample of the Pd8Mo catalyst was exposed to a mixture
of NO/ethanol at 573 K. Several bands were observed be-
−1
−1
However, new bands around 1044 and 1093 cm were ob-
served and are possibly associated with ethoxy species ad-
tween 1200 and 1800 cm . Fig. 5a shows the spectrum for
this sample before being heated to 573 K, and bands around

362
L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
Fig. 6. TPD profile of ethanol adsorbed on Al O .
2
3
Fig. 4. IR spectra of adsorbed ethanol on Pd8Mo/Al O .
2
3
Fig. 7. TPD profile of ethanol adsorbed on Pd/Al O .
2
3
3
.3. Ethanol temperature-programmed desorption
The TPD profiles after adsorption of ethanol on alumina
and Pd/Al2O3 catalyst are presented in Figs. 6 and 7. Ethanol
and ethylene were the main desorption products for the sup-
port. Ethanol desorbed in two peaks (395 and 500 K) while
a great formation of ethylene was observed around 550 K.
The Pd/Al2O3 catalyst showed a decrease in ethylene pro-
duction when compared to alumina. Besides that, three dis-
tinct features were observed: first, the formation of CO, CH4,
and H2 (ethanol decomposition products) at 495 K, second,
the formation of acetaldehyde at 530 K due to ethanol dehy-
drogenation; and finally, a further desorption of CO, H2, and
CO2 was detected at higher temperatures (above 723 K).
Figs. 8 and 9 present the TPD profile after adsorption of
ethanol on 8Mo and Pd8Mo catalysts. These results show a
further decrease in ethylene formation (mainly on Pd8Mo).
For the 8Mo catalyst, there was desorption of ethanol at
Fig. 5. IR spectra after exposure of Pd8Mo to NO + ethanol mixture at:
(
a) room temperature, (b) room temperature after being heated to 573 K,
c) sample in b after He flow, and (d) sample in c after vacuum.
(
3
86 and 475 K, while for Pd8Mo, ethanol desorbed only
−
1
1
390, 1460, 1691, and 1800 cm besides a small shoulder
at 405 K. The simultaneous formation of CO, CH , and H
around 475 K (in a similar way as the Pd/Al O catalyst)
2 3
was only observed for the Pd8Mo sample. In addition, CO,
H , and CO were also detected at high temperatures, not
seen on 8Mo. Furthermore, both samples presented forma-
4
2
−1
at 1715 cm were observed, although the region between
500 and 1620 cm was not well resolved. The spectra
presented in Figs. 5b–d were very similar, showing bands
around 1268, 1462, 1577, 1690, and 1800 cm
−1
1
2
2
−1
.

L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
363
Fig. 8. TPD profile of ethanol adsorbed on 8Mo/Al O .
Fig. 11. TPD profile of ethanol adsorbed on Pd20Mo/Al O .
2 3
2
3
Fig. 9. TPD profile of ethanol adsorbed on Pd8Mo/Al O .
2
3
Fig. 12. NO + ethanol TPSR profile on 8Mo/Al O .
2
3
3
.4. Temperature-programmed surface reaction
During TPSR measurements, some products were identi-
fied by secondary fragment signal in the mass spectrometer
analysis. Since the signal m/e = 28 is attributed to both CO
and N2 (also a secondary fragment of CO2), the signal used
to distinguish one from the other, together with their main
signals, was the CO secondary mass fragmentation signal
m/e = 12. In the same way, the signal m/e = 44 may be
related to CO2 or N2O and the use of secondary fragments
(m/e = 28 and 12 for CO2 and m/e = 44 and 30) were also
needed.
Fig. 10. TPD profile of ethanol adsorbed on 20Mo/Al O .
2
3
tion of acetaldehyde in two peaks (455 and 532 K for 8Mo
and 479 and 522 K for Pd8Mo).
The TPD results after adsorption of ethanol on 20Mo and
Pd–20Mo catalysts were very similar (Figs. 10 and 11). Both
showed little formation of ethylene and the presence of two
peaks corresponding to acetaldehyde around 473 and 520 K.
Although not as intense as before, the desorption of CO,
CO2, and H2 at higher temperatures could also be observed.
TPSR results for the 8Mo catalyst are shown in Fig. 12.
The profile is very similar to that observed for the TPD of
adsorbed ethanol (Fig. 8) with ethylene formation around
540 K, two peaks of acetaldehyde at 465 and 535 K, and
ethanol desorption at 400 and 490 K. NO consumption was
observed only above 600 K in a wide temperature range

364
L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
tected around 455 and 545 K. NO consumption began at
slightly higher temperatures, showing a well-defined peak
at 690 K and two small shoulders around 640 and 612 K.
The NO consumption was followed by the simultaneous in-
crease in the intensity of signals m/e = 44 and 28. On the
other hand, the signal m/e = 12 (associated with CO and/or
CO2 formation) was seen only around 605 and 695 K.
3
.5. Catalytic activity
Table 2 displays NO and ethanol conversion and selec-
tivity values for each temperature on all samples. The 8Mo
and 20Mo catalysts were inactive for NO conversion, but ac-
tive for ethanol decomposition. Pd8Mo was more active for
NO conversion than Pd20Mo catalyst; however, in this tem-
perature range, the Pd/Al2O3 catalyst presented better NO
conversion than the others. This behavior is different from
the previous one for the NO + CO reaction [12], where the
Pd–Mo-containing catalysts were more active. Nevertheless,
at 593 K, the samples containing Pd showed similar NO
conversion. Considering H2 chemisorption results, it was
possible to calculate the turnover frequency (TOF) over the
exposed Pd sites for the Pd and Pd8Mo samples. The results
obtained are very interesting and may be seen in Table 2.
They show that the Pd–Mo catalyst presented higher values
when compared to the Pd-only sample.
Fig. 13. NO + ethanol TPSR profile on Pd/Al O .
2
3
N2 and N2O were the only nitrogen-containing products
observed for this reaction and regardless of the presence or
amount of Mo, the selectivity for N2 was approximately the
same for all active samples. Once again, this result is quite
different from previous NO + CO reaction [12], where the
addition of Mo to Pd remarkably improved N2 selectivity.
Acetaldehyde was not observed for the Pd and Pd8Mo cata-
lysts, while the other samples showed a high selectivity for
this product. Note that the Pd/Al2O3 and 8% Mo-loading
samples exhibited significant ethylene formation, while the
2
0% Mo-loading catalysts did not. Furthermore, no CO for-
mation was seen on the Mo-only catalysts.
Fig. 14. NO + ethanol TPSR profile on Pd8Mo/Al O .
2
3
4
. Discussion
and together with the increase in intensity of the signals
m/e = 28 and 44.
The results presented above are a continuation of a line
of work in which the alumina-supported Pd–Mo catalysts
are being studied. Previously published papers [11–13] have
thoroughly investigated the NO adsorption properties of
these catalytic systems, which we find important and rele-
vant for the discussion and conclusions drawn here. There-
fore, before the discussion on the results presented in this
work, we find it is suitable to describe some important points
of our previous works regarding NO adsorption.
Figs. 13 and 14 show the TPSR profiles for Pd and
Pd8Mo catalysts. On Pd/Al2O3, the results show the desorp-
tion of ethanol at 390 and 500 K and formation of ethylene
at 545 K. Unlike the TPD results for this catalyst (Fig. 7), the
simultaneous formation of CO, CH4, and H2 around 495 K
was not observed and virtually no acetaldehyde was formed.
A large consumption of NO was seen at 590 and 690 K, with
the simultaneous increase of signals m/e = 44, 28, and 12.
For the Pd8Mo, ethanol desorption was observed at 400
and 490 K, with formation of ethylene at 545 K but, again,
simultaneous CO, CH4, and H2 were not seen. Unlike the
Pd/Al2O3 catalyst, a small amount of acetaldehyde was de-
The previously published papers [11–13] studied the NO
adsorption/decomposition properties on the Pd–Mo/Al O
2
3
surface using infrared as well as temperature-programmed
desorption experiments. In a general sense, the TPD pro-
files for NO adsorption on Pd, Mo, and Pd–Mo catalysts

L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
365
Table 2
Conversion and selectivity values for the NO + ethanol reaction
Catalyst
Temperature
K)
523
NO conversion
TOF
N selectivity
2
Ethanol conversion
(%)
Selectivity of carbon species (%)
−
1 a
(
(%)
(s
)
(%)
CO
CO
Acetald.
Ethyl.
2
Pd
0
20
29
49
0
–
–
–
5
13
11
–
33
45
44
–
0
0
0
553
573
593
0.91
1.32
2.34
72
62
66
78
79
85
62
42
45
Pd8Mo
523
0
8
34.5
46
0
1.74
7.50
–
55
66
74
75
30
44
38
34.5
61
50
47
0
0
0
0
9
6
15
25
553
573
593
61
67
68
10.00
40.5
Pd20Mo
523
0
0
13
45
–
–
–
–
–
–
63
67
–
–
0
7
–
56
56
51
–
44
37
26
–
0
0
553
573
593
51
60
71
11
16
8Mo
523
0
0
0
0
–
–
–
–
–
–
–
–
60
69
71
78
0
0
0
0
37
47
33
42
50
29
52
32
13
23
15
25
553
573
593
20Mo
523
0
0
0
0
–
–
–
–
–
–
–
–
39
48
60
70
0
0
0
0
43
29
22
7
57
71
67
80
0
0
11
13
553
573
593
Acetald., acetaldehyde; ethyl., ethylene.
a
Mole of reacted NO/s (×1025).
were quite different, indicating that NO interacts differently
in each case. While on the Pd-only and 8Mo samples the
formation of N2 during TPD occurred only above 600 K
and Falconer [5], acetaldehyde or dehydrogenation products
were not observed during TPD of adsorbed ethanol on alu-
mina. The IR analyses of ethanol on alumina (Fig. 1) showed
a decrease in intensity of the bands related to ethoxy species
with increasing temperature. This suggests that part of the
ethoxy species are desorbed as ethanol (Fig. 6) and part de-
hydrated to ethylene.
(
broad peaks), on the Pd–8Mo sample there was a sharp,
intense peak of N2 at 520 K (and another peak at 773 K). Be-
sides lowering the temperature of N2 formation (first peak),
the presence of Pd and Mo on the Pd–Mo catalyst also in-
creased the selectivity for N2 during TPD, when compared
to the other samples. According to the authors’ discussion,
the high temperature N2 formation is attributed to inactive
adsorbed nitrogen species (Na species formed upon NO ad-
sorption) which are strongly bounded and inhibit further NO
adsorption and dissociation. The addition of molybdenum
oxide to the palladium catalyst decreased the fraction of the
inactive Na species, thus promoting the NO dissociation to
N2 at lower temperatures. The authors further suggested that
the increase in NO dissociation on the Pd–Mo catalyst could
be due to the modification of the electronic properties of pal-
ladium in the presence of MoOx species (they observed a
decrease in adsorption strength of CO on Pd on the Pd–Mo
catalyst). They concluded that the formation of new adsorp-
tion sites between Pd and MoOx could be responsible for
decreasing the stabilization of adsorbed nitrogen species on
the surface of palladium.
The Pd/Al2O3 catalyst significantly changed the product
distribution during TPD analysis (Fig. 7) showing simultane-
ous formation of CO, CH4, and H2 at 495 K due to ethanol
decomposition on Pd sites. However, the infrared analy-
ses of ethanol adsorption on the Pd/Al2O3 catalyst (Fig. 2)
was very similar (at low temperatures) to the Al2O3 sup-
port (Fig. 1), showing mainly the presence of ethoxy species
adsorbed on alumina. These results suggest that the ethoxy
species adsorbed on alumina migrate to the Pd particles and
decompose. A similar result was reported by Cordi and Fal-
coner [5] as they suggested that part of the ethanol adsorbed
on alumina diffuses to the Pd sites, where decomposition oc-
curs. In addition, according to them the α-carbon forms CO
and the β-carbon forms CH4 during ethanol decomposition.
Noteworthy is the formation of CO, CO2, and H2 at high
temperatures. Cordi and Falconer [5] also observed this dur-
ing TPD of adsorbed ethanol on Pd/Al2O3 catalysts and sug-
gested that it might be due to the decomposition of a more
stable carbon species formed during ethanol decomposition
at lower temperatures. In fact, the infrared measurements
(Fig. 2) indicated the formation of acetate species as the
temperature was raised from 523 to 573 K, which is the
same temperature range as acetaldehyde formation during
TPD (Fig. 6). It is possible that while ethanol (adsorbed
4
.1. Adsorption of ethanol
The high selectivity of alumina for the dehydration of
ethanol to ethylene during TPD analyses (Fig. 6) agrees well
with the work of McCabe and Mitchell [20], who attributed
this to the presence of strong acid sites. Contrary to Cordi

366
L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
as ethoxy species) is dehydrogenated on Pd sites to form
acetaldehyde, more stable acetate species are also formed.
These species remain adsorbed and are the precursors for
CO, CO2, and H2 production at high temperature. Cordi and
Falconer [5] further suggested that a partial oxidation prod-
uct from ethanol, such as acetaldehyde and/or acetic acid,
might be the precursor for the more stable carbon species.
However, no formation of acetic acid was observed in this
work, indicating that acetaldehyde is probably the precursor
for the acetate species.
For the catalysts containing Mo, there was a drastic re-
duction of ethylene formation, which is a reaction that is
mainly catalyzed by the acid sites on the support. However, a
large fraction of the alumina surface should be covered since
a MoO3 monolayer is usually formed between 8 and 12%
Mo loading [12,19]. This result is supported by the infrared
measurements of adsorbed ethanol on the 8% Mo-containing
catalysts (Figs. 3 and 4), since the bands related to the ethoxy
species adsorbed on alumina were less intense.
file for the 20Mo catalyst (Fig. 10). The high concentration
of Mo on these samples favors the presence of bulk MoO3,
as observed previously [12]. Therefore, this might indicate
that bulk MoO3 partially covers the Pd metallic sites, mak-
ing it appear as if the presence of Pd had no direct effect on
the properties of the Pd20Mo catalyst for the adsorption of
ethanol. In addition, the covering of Pd particles by MoOx
species is supported by H2 chemisorption results and it has
also been reported in the literature [15,16].
The infrared analysis of adsorbed ethanol on the 8Mo and
Pd8Mo catalysts showed bands at 1662–1654, 1562–1566,
−1
1586, and 1475 cm as the samples were heated. The first
group of bands was possibly associated with acetate species
on MoOx while the second group was attributed to acetate
species on alumina. When comparing the Pd-free samples
with the ones containing Pd, it becomes evident that the
presence of Pd favors the formation of acetate species, al-
though the presence of MoOx species also contributes to
this. Furthermore, the acetate species are formed at the same
temperature range as acetaldehyde is produced during TPD
analysis, thus suggesting that acetaldehyde might be the pre-
cursor of the formation of such species. Once again, it seems
that, since these species are more stable, they remain ad-
sorbed until they decompose at higher temperatures (above
723 K), originating CO, CO2, and H2 as seen on the TPD
profiles.
Another feature presented by the Mo-containing catalysts
was that acetaldehyde formation exhibited two peaks dur-
ing TPD (Figs. 8 and 9). The first one was only present
on the catalysts containing molybdenum oxide and might
be attributed to the oxidative dehydrogenation of ethanol
adsorbed on the partially reduced molybdenum oxide. In
fact, Iwasawa et al. [21] studied the reaction intermediates
during ethanol oxidation over silica-supported molybdenum
oxide catalysts using IR spectroscopy and in situ Raman
spectroscopy. The authors observed bands which were at-
tributed to the presence of adsorbed ethoxide structures. The
in situ Raman spectroscopy measurements at reaction con-
ditions identified two types of ethoxide species, which were
associated with Mo=O and Mo–O–Mo sites. A very simi-
lar result was obtained by Zhang et al. [22] while studying
ethanol oxidation over MoO3/Al2O3 catalysts. According
to both works, the ethoxide species bonded to a terminal
oxygen group produced acetaldehyde, while the ethoxide
species bonded to a bridging oxygen group produced ethy-
lene, which agrees well with the results observed here.
On the other hand, the second peak of acetaldehyde for-
mation (at a slightly higher temperature) was observed on
all samples, with the exception of the support alone. One
possible explanation for this would be the dehydrogenation
of the ethoxy species initially adsorbed on alumina, which
migrates to the active sites (MoOx, in the case of the Mo cat-
A final and important issue regarding the adsorption prop-
erties of ethanol on the catalysts studied here may be pointed
out by comparing the TPD profiles of the Pd, Mo, and
Pd–Mo catalysts. When looking at the Pd–Mo TPD results
and comparing these to their respective Pd-free or Mo-free
pair (for example, Pd8Mo compared to Pd/A2O3 and/or to
8Mo), there are clearly no new or distinct features arriv-
ing from a possible Pd–Mo interaction. All features ob-
served on the bimetallic samples are either attributable to
the presence of Pd, MoOx, or alumina. Although Pd is
in close contact and interacts with molybdenum suboxide
species as observed in previous studies by TPR and CO
infrared measurements [13], this apparently has no effect
on the adsorption/dissociation properties of ethanol. This
is in contrast to what was observed previously for the ad-
sorption/dissociation properties of NO [11,13], since the
presence of closely associated Pd and MoOx considerably
changed NO dissociation and product distribution during
TPD of adsorbed NO.
0
0
alysts; Pd in the case of the Pd catalysts; and MoOx or Pd ,
in the case of the Pd–Mo catalysts). The migration of such
species could be the limiting step of the second acetaldehyde
formation, thus occurring at higher temperature.
4.2. Ethanol + NO reaction
The NO + ethanol temperature-programmed surface re-
action for the 8Mo catalyst (Fig. 12) was very similar to
the ethanol TPD profile (Fig. 8), showing the same ethanol
desorption/decomposition pattern for temperatures below
600 K. Apparently, the presence of NO had no influence on
the adsorption properties of ethanol on the reduced molyb-
denum oxide. NO consumption was observed only above
600 K, together with the increase of intensity of signals
For the Pd8Mo catalyst, the simultaneous formation of
CO, CO2, and H2 at 475 K (Fig. 9) was, once again, due to
the decomposition of ethanol on Pd sites. However, despite
the presence of Pd, the Pd20Mo catalyst did not show high
selectivity for the decomposition of ethanol to CO, CH4, and
H2 at low temperatures (Fig. 11). In fact, the TPD profile
for the Pd20Mo catalyst was very similar to the TPD pro-

L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
367
m/e = 28 and 44 (very weak), indicating the formation of
both N2 (main product) and N2O. Since the intensity of the
signal m/e = 12 did not increase, no CO or CO2 was pro-
duced above 600 K, thus showing that NO decomposition
on the partially reduced molybdenum oxide surface was the
only event contributing to NO consumption. In other words,
the reaction between flowing NO and adsorbed ethanol or
acetate species did not take place. In a previous work [13],
TPD measurements of adsorbed NO on this catalyst showed
simultaneous formation of N2 and N2O only above 600 K,
which agrees with the results observed here. These observa-
tions suggest that neither NO or ethanol directly compete for
adsorption sites on the partially reduced molybdenum oxide
surface.
For the Pd8Mo catalyst, the TPSR results also indicated
that NO adsorption on the Pd particles hindered ethanol de-
composition since there were two ethanol desorption peaks
and the absence of the ethanol decomposition products (CO,
CH4, and H2). On the other hand, there were two distinct
peaks of acetaldehyde formation at the same temperature
as observed during TPD (Fig. 9), attributed to the oxidative
dehydrogenation of ethanol on the partially reduced molyb-
denum oxide. Once again, this shows that the presence of
NO does not modify the adsorption properties of ethanol on
MoOx.
NO consumption began only above 500 K, with the in-
crease of intensity of the signals m/e = 44, 28, and 12.
However, unlike the Pd/Al2O3 catalyst, one NO consump-
tion peak (690 K) and two shoulders (640 and 612 K) could
be identified, but only the peak at 690 K and the shoulder
at 612 K were followed by the increase of intensity of the
signal m/e = 12 (695 and 605 K), indicating that the re-
action between NO and adsorbed acetate species occurred
around 690 and 610 K. These are approximately the same
temperatures at which the reaction takes place during TPSR
on Pd/Al2O3, which indicates that the reaction proceeds in
the same way on both catalysts. That is, the reaction between
NO and adsorbed acetate species probably follows the same
pathway and apparently the presence of closely associated
Pd–MoOx does not originate new reaction sites for this re-
action. Furthermore, the TPSR results for the 8Mo sample
showed that the MoOx species alone do not promote the re-
action between adsorbed carbon species and NO. All of this
suggests that such reaction occurs at the Pd–alumina or Pd–
MoOx interface. The NO consumption peak at 640 K is most
possibly due to the dissociation of NO on MoOx with the
formation of both N2 (m/e = 28) and N2O (m/e = 44), and
it seems to have no influence on the NO + ethanol reaction.
The catalytic activity measurements (Table 2) showed
that Pd/Al2O3 was the most active catalyst for NO conver-
sion. This is in contrast with the previously studied NO+CO
reaction [12] on the same catalysts. For the NO + CO re-
action system, the addition of Mo to the Pd/Al2O3 catalyst
improved NO conversion and this was explained by a redox
mechanism in which NO adsorbed and dissociated on the
partially reduced molybdenum oxide surface and the oxy-
gen atom from NO was made available for the oxidation of
CO on Pd. Apparently, the NO + ethanol reaction on these
catalysts occurs through a different reaction pathway.
The TPSR measurements for the Pd-containing catalysts
(
Figs. 13 and 14) showed rather similar characteristics and
the discussion that will be carried on for the Pd/Al2O3 sam-
ple will later be referenced by the discussion on the Pd8Mo
sample.
When comparing the TPD profile of adsorbed ethanol
(
Fig. 7) with the TPSR results for the Pd/Al2O3 catalyst,
some important features may be pointed out. First, dur-
ing TPD, only one ethanol desorption peak was observed
(
(
around 395 K), whereas for the TPSR there were two peaks
at 390 and 500 K). The simultaneous formation of CO,
CO2, and H2 at 495 K seen during TPD was not observed
on the TPSR profile. As discussed above, such product for-
mation around 500 K (for TPD) was due to the decomposi-
tion of ethanol (originally adsorbed on alumina as an ethoxy
specie) on Pd particles. For some reason, during TPSR the
adsorbed ethanol is not allowed to decompose on the Pd sites
and instead it simply desorbs as unreacted ethanol (hence
the second ethanol peak at 500 K during TPSR) in exactly
the same way as was seen for the TPD profile on alumina
(
Fig. 6). It suggests that when NO is present, the NO mole-
cules readily adsorb on the Pd sites, removing and/or not
allowing ethanol adsorption nor reaction on the metallic par-
ticles. McCabe and Mitchell [20] studied the direct oxidation
of ethanol on alumina-supported Pd and Pt catalysts and
observed that the presence of NO inhibited the catalytic ac-
tivity. They attributed this to the fact that the NO molecules
adsorbed more strongly than ethanol on the metallic sites.
For the same reason as discussed above, almost no acetalde-
hyde was formed during TPSR on Pd/Al2O3.
Two well-defined NO consumption peaks were observed
with the simultaneous increase in the intensities of the sig-
nals m/e = 44, 28, and 12. The signals m/e = 28 may be at-
tributed to either N2, CO, or both while the signal m/e = 44
may be related to N2O, CO2, or both. However, the increase
in intensity of the signal m/e = 12 is indicative of the for-
mation of CO and/or CO2, revealing that NO reacts with
adsorbed carbon-containing species. On the other hand, the
IR results showed that at the temperature range in which
such reaction took place during TPSR, only acetate species
were present, suggesting that these are the carbon-containing
species that react with NO.
The NO conversion pattern for the reaction studied here
was as follows: Pd/Al2O3 > Pd8Mo > Pd20Mo > 8Mo and
20Mo (no activity). This suggests that as Mo is added, the
activity for NO reduction with ethanol diminishes. On the
other hand, it was seen [12] that as the amount of Mo in-
creases, a monolayer of molybdenum oxide is formed and
after that, bulk MoO3 is present. Through the ethanol TPD
results, it was seen that MoOx might cover part of the Pd
particles and this, of course, is more evident on the Pd20Mo
sample. In addition, hydrogen chemisorption results showed
less exposed Pd particles when Mo is present. In fact, the

368
L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
covering of Pd particles by MoOx suboxide species has been
reported in the literature, even at low Mo loadings [15]. Such
effect of the presence of Mo may be responsible for the ac-
tivity drop on the Pd–Mo catalysts. In other words, it seems
that for this reaction, it is very important that the metallic
sites are exposed. This was not true for the NO + CO re-
action, since the Pd20Mo catalyst was more active than the
Pd/Al2O3 catalyst [11]. Such great differences from one re-
action to another called our attention and in order to explain
the results obtained here, a comparison with the NO + CO
reaction studied before [11,12] will be made.
As noted above, for the NO + CO reaction the addition of
Mo to the Pd catalyst changed the prevailing reaction mech-
anism and a new redox mechanism was proposed where NO
adsorbed and reacted on the MoOx surface, while CO was
oxidized on the Pd particles. According to this mechanism,
NO conversion is enhanced by the presence of Pd–MoOx in
a bifunctional-type mechanism. On the other hand, for the
NO + ethanol reaction, it seems that the Pd particles have to
be exposed, which indicates that the main reaction step must
take place on the metallic sites or at the metal/oxide inter-
face, as suggested by TPSR results. In addition, the N2 selec-
tivity was approximately the same on all samples, indicating
that the reaction mechanism is probably the same, regard-
less of the presence of Mo (for the NO + CO reaction, N2
selectivity was higher on the Mo-containing catalysts). Nev-
ertheless, the fact that the selectivity was the same could also
mean that the reaction was at the equilibrium point. How-
ever, thermodynamical calculations were made for this sys-
tem and this possibility was discarded. Furthermore, TPSR
analysis also suggested that the reaction pathway was the
same on all Pd-containing catalysts.
As discussed above, the TPSR results showed that for this
reaction system the NO molecules adsorb preferentially on
the Pd sites, while ethanol adsorbs either on the partially re-
duced molybdenum oxide surface and/or the support. This is
in total contrast with what happened during the redox mech-
anism for the CO+NO reaction, where CO mainly adsorbed
on Pd and NO on the MoOx species, suggesting, once again,
that a similar redox mechanism for the NO + ethanol reac-
tion is unlikely and that the main reaction step occurs on the
Pd particles. Besides that, the TPSR analyses, together with
the IR and TPD results, also indicated that NO probably re-
acts with the acetate species (formed above 450 K) which
are adsorbed on the Al2O3 and/or the MoOx surface and that
such reaction takes place at the metal/oxide interface.
However, an interesting result was obtained when the
turnover frequency was calculated (Table 2). By assuming
that the NO conversion is dependent on the exposed Pd and
using the H2 chemisorption results as an estimate of exposed
Pd particles, it was possible to calculate the TOF over ex-
posed Pd sites for the Pd and Pd8Mo catalysts. These values
were much higher when Mo was present, indicating that al-
though MoOx in close contact with Pd hampers overall NO
conversion, it somehow enhanced the reaction rate at the
metal oxide interface.
Other important results that must be noted are the fact
that ethylene formation was seen on all samples and that the
selectivity for its formation was higher on the catalysts with
more exposed alumina surface. Indeed, the dehydration of
ethanol to ethylene is a characteristic reaction on alumina
(also observed during TPD experiments) and this is a side
reaction that occurs simultaneously with the NO + ethanol
reaction, thus contributing to ethanol conversion.
Acetaldehyde formation was observed on all samples ex-
cept for the Pd/Al2O3 and Pd8Mo, which were the most
active catalysts for NO conversion. According to the TPD
and IR results, it seems that acetaldehyde formation is an im-
portant step for originating more stable acetate species and
it is believed that these species react with adsorbed NO dur-
ing TPSR. Apparently, the formation of acetaldehyde is a
necessary step for reaction to take place. Since no acetalde-
hyde was detected for the more active catalysts, it seems that
acetaldehyde might be an intermediate product during the
reaction process and that its formation (dehydrogenation of
ethanol) should be the limiting step. In other words, as soon
as dehydrogenation of ethanol occurs, the acetaldehyde in-
termediate readily originates the acetate species, which im-
mediately react with the adsorbed NO, thus explaining the
absence of acetaldehyde on the most active catalysts.
Some works dealing with the oxidation of ethanol by
oxygen (much more potent oxidation agent than NO) have
shown that the activation of ethanol is the rate-determining
step. Zhang et al. [22] studied the oxidation of ethanol on
supported molybdenum oxide catalysts using in situ infrared
and Raman measurements. They observed that the acetalde-
hyde formation on such catalysts occurred in two steps: the
first would be the adsorption of ethanol with formation of
ethoxy species while the second, rate-limiting step, would
be the decompositionof these species to acetaldehyde. Nagal
and Gonzales [18] studied the oxidation of ethanol and ac-
etaldehyde on Pt/SiO2 catalysts using in situ infrared analy-
ses and according to their observations, they proposed a
reaction mechanism in which the dehydrogenation of an ad-
sorbed ethoxy species would be the limiting step. These
dehydrogenated species could than be oxidized to form in-
termediate acetate species that would react originating CO2.
According to the authors, the formation of acetate species
also occurs via acetaldehyde formation, as follows:
S
CH3CH2OH → CH3CH2O–S → H2 + CH3CHO
Ethanol
Ethoxy
Acetaldehyde
→
+ H–S.
Acetate species
Some works have investigated the NO reduction by al-
cohols [10,23–25]. Noto et al. [23] studied the reduction
of NOx by ethanol on silver-based catalysts and found that
NOx is reduced after ethanol is oxidized to acetaldehyde

L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
369
and, furthermore, when oxidized or partially oxidized prod-
ucts are formed by acetaldehyde oxidation, thus showing
that acetaldehyde formation is an important step in this re-
action. Toops et al. [10] studied the reduction of NO by
methanol over La2O3 and based on kinetic data, they sug-
gested a mechanism involving two types of adsorption sites,
as follows:
Reaction between adsorbed species then proceeds at the
metal/oxide interface:
8
NO–Pd + 2HNO–Pd + 2CH3COO–S
−[O]
→ 5N + 2CO + 4H O + 2S + 10Pd,
2
2
2
(5)
where [O] represents surface oxygen.
Some considerations must be made regarding other sur-
face reactions. As shown by the results presented before,
a side reaction may also occur during NO reduction by
ethanol, which is the dehydration of ethanol to ethylene on
the alumina surface, thus contributing to ethanol conversion:
6
2
2
[NO + S ꢀ NO–S],
[CH3OH + ∗ ꢀ CH3OH–∗],
[NO–S + CH3OH–∗ → HNO–S + CH3O–∗]
rate-determining step,
CH3CH2O–S + H–S → C2H4 + H2O + 2S.
(6)
4
NO–S + 2HNO–S + 2CH3O–∗ ꢀ 3N2 + 4H2O + 2CO2
Although reaction (5) shows total oxidation products, the
partial oxidation of adsorbed acetate species may also take
place and in fact, product distribution during catalytic mea-
surements did show the presence of both CO and N2O.
According to the reaction mechanism proposed above and
to the results presented in Table 2, it may be seen that since
no acetaldehyde was formed on the Pd and Pd8Mo catalysts,
it is suggested that reaction (3) must be the limiting step on
this mechanism. That is, as soon as acetaldehyde is formed,
it is converted into acetate species which then react with the
adsorbed NO species. On the other hand, there was acetalde-
hyde formation on the Pd20Mo and this may be explained by
the fact that for this sample, a large fraction of the Pd par-
ticles are covered by MoOx (due to the high concentration
of Mo), as discussed above, thus hampering steps 1 and 5
and allowing acetaldehyde to be formed. In other words, for
this sample, not all of the acetaldehyde formed during reac-
tion (3) is converted in reactions (4) and (5). Nevertheless, as
the temperature is raised and the NO conversion increases,
acetaldehyde selectivity considerably drops and this is better
observed when comparing the selectivity values of samples
Pd20Mo and 20Mo at 593 K.
+
6S + 2∗.
According to the authors, the rate-determining step involves
the reaction between adsorbed NO and adsorbed methanol,
originating methoxy species. Based on their observations,
they have discarded the possibility of methanol decompo-
sition to CO and H2 with these compounds then react-
ing with adsorbed NO. Although the rate-determining step
would involve the methanol molecule, they have not investi-
gated the possible adsorption intermediates involved during
adsorption of methanol, such as the formation of formate
species. Several works in the literature have reported formate
species formation upon methanol adsorption at high tem-
peratures [26–28] and this could possibly be an important
reaction intermediate involved in the rate-determining step.
In this work, we believe that the ethanol + NO reaction
occurs similarly to what was proposed by Toops et al. [10]
for the methanol + NO reaction, except that according to
the results obtained here and discussed above, we find there
is evidence to suggest that acetate species are an important
reaction intermediate and that its formation is linked to the
dehydrogenation of ethanol to acetaldehyde. We also believe
that there are two adsorption sites for NO (Pd particles) and
ethanol (Al2O3 and/or MoOx) and that the reaction between
the involved species occurs at the metal/oxide interface.
Based on what was discussed above, a possible reaction
mechanism was proposed for this system.
These observations suggest that there are possibly two
different processes affecting the overall reaction. The first
would be the need for exposed Pd particles and its contact
with MoOx species while the second would be the ability to
favor the dehydrogenation step [reaction (3)], which is rate
determining. This is better understood by comparing the NO
conversion data with TOF results. The turnover frequency
was calculated as the NO conversion rate over Pd sites and
by doing this we consider only the exposed Pd particles, thus
eliminating the effect of Pd decoration by MoOx. In this
case, TOF numbers indicate that as the amount of MoOx in-
creases, so does the reaction rate at the metal/oxide interface.
This could be explained by looking at the reaction mecha-
nism proposed above. According to the mechanism and to
the discussion presented before, acetaldehyde formation is a
critical step in the reaction pathway since it is a precursor in
originating acetate species. On the other hand, both TPD and
TPSR measurements showed that when MoOx was present,
acetaldehyde formation was favored when compared to the
Pd/Al2O3 sample, since two peaks of acetaldehyde were ob-
served on the Mo-containing catalysts while just one peak
First, NO adsorbs preferentially on Pd particles while
ethanol adsorbs dissociatively forming adsorbed ethoxy
species:
1
2
0[NO + Pd → NO–Pd],
(1)
(2)
[CH3CH2OH + 2S → CH3CH2O–S + H–S].
At the reaction temperature, these ethoxy species are
dehydrogenated, providing the overall dehydrogenation of
ethanol to acetaldehyde:
2
[CH3CH2O–S + H–S → CH3CHO + H2 + 2S].
Once again, at the reaction temperature, acetaldehyde is
readily activated originating intermediate acetate species:
(3)
[O]
2
[CH3CHO + S + NO–Pd → CH3COO–S + HNO–Pd]. (4)

370
L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
was seen on the Pd-only catalyst during TPD. In other words,
the MoOx species were active for the dehydration of ethanol
to acetaldehyde, while alumina was not. Furthermore, ac-
etaldehyde formation during TPD on the Mo-containing cat-
alysts began at lower temperatures than on Pd-only cata-
lysts. Since it was seen that Pd particles and MoOx are in
close contact [12,13], and that acetaldehyde formation [re-
action (3)] on MoOx is favored, it would be reasonable to
think that the reaction rate between NO and acetate species
at the Pd–MoOx interface would be favored over the Pd–
Al2O3 interface, thus explaining the higher TOF numbers
for the Mo-containing catalysts.
Nevertheless, when we look at the NO conversion results,
we clearly see that the presence of exposed Pd particles are
an important factor to be taken into account. For the Pd20Mo
catalyst, for example, the amount of exposed Pd particles is
very low (as seen by H2 chemisorption), while the amount
of MoOx species is very large, so even though the MoOx
favors step 3 on the reaction mechanism, the reaction rate
at the Pd–MoOx interface is not high enough to account for
acetaldehyde formation on MoOx (which is achieved only at
higher temperatures).
alysts was studied using IR analyses and only a very weak
−1
absorption band at 1775 cm (assigned to linearly adsorbed
NO on Pd) was observed and this was attributed to the poor
Pd dispersion presented by these catalysts. In fact, the Pd
dispersion for the Pd/Al2O3 catalyst was only 23% and in
addition to this, part of the Pd particles are covered by Mo
species on the Pd8Mo sample and therefore, it was virtually
impossible to identify the NO adsorption band on Pd in the
results presented in Fig. 5.
Nevertheless, the results obtained after the sample was
heated up to 573 K and cooled to room temperature (Figs. 5b
and 5c after He flow and Fig. 5d after vacuum) showed,
besides increased intensity of the bands related to NO ad-
sorbed on MoOx, the appearance of intense bands at 1577
−1
and 1462 cm . These new bands are at the same position
as the acetate bands observed after ethanol adsorption, ex-
cept that in this case, they were much more intense, thus
evidencing the participation of such adsorbed intermediates
in the reaction mechanism.
5. Conclusion
As an attempt to support the proposed mechanism, an ad-
ditional IR measurement was performed on a sample of the
Pd8Mo catalyst that was exposed to a mixture of NO/ethanol
at reaction temperature. The results are displayed in Fig. 5.
Although it is difficult to establish band identifications
since there may be overlapping of bands related to different
adsorbed species, some considerations may still be made.
Fig. 5a shows the IR result after the sample was exposed
to the NO/ethanol mixture, but prior to heating and only
The TPD and IR measurements showed that ethanol ad-
sorbs as ethoxy species and that as the surface is heated,
more stable acetate species are formed and decompose only
at higher temperatures. The results further suggest that ac-
etaldehyde (ethanol dehydrogenation product) is an impor-
tant precursor in the formation of such acetate species, which
were identified on both Pd/Al2O3 and Pd–Mo catalysts.
However, the presence of Pd seems to favor their formation.
The TPSR as well as the catalytic tests results showed
that partially reduced molybdenum oxide has no activity for
the reduction of NO with ethanol and that such reaction
occurs only in the presence of Pd. Furthermore, the TPSR
analyses together with the TPD and IR measurements in-
dicated that NO preferentially adsorbs on the Pd particles,
while ethanol selectively adsorbs on alumina and/or partially
reduced molybdenum oxide (MoOx). In addition, the com-
parison of the TPSR results with TPD and IR suggested that
NO reacts with adsorbed acetate species and that such reac-
tion would occur at the metal/oxide interface.
The catalytic tests also showed that the addition of molyb-
denum oxide to the palladium catalyst decreased the activity
for NO conversion, indicating that the main reaction step
should take place at the Pd sites. In other words, it is im-
portant, for this reaction, that Pd particles are exposed. Nev-
ertheless, the selectivity for N2 formation was similar on all
active samples, suggesting that the reaction mechanism is
the same, independently of the presence of Mo. This was in
striking contrast with the previously studied NO reduction
with CO on these catalysts. Furthermore, since no acetalde-
hyde formation was observed for the more active catalysts
and since acetaldehyde is an important intermediate in the
formation of acetate species, the dehydrogenation of ethanol
to acetaldehyde should be the rate-limiting reaction step.
−1
bands around 1800 and 1691 cm as well as the bands at
−1
1
390 and 1455 cm were observed. According to the litera-
ture [11–13,29], NO adsorption on partially reduced molyb-
denum oxide originates absorption bands around 1815 and
−1
1
710 cm (symmetric and antisymmetric stretching vibra-
tion modes, respectively) and therefore, the bands observed
−1
here at 1800 and 1691 cm are most probably due to the
adsorption of NO on MoOx species. The bands at 1390 and
−1
1
455 cm were also observed after adsorption of ethanol
on Pd8Mo (Fig. 3) and were associated with ethoxy species
−1
on alumina. The region between 1500 and 1620 cm was
not well resolved.
Several works [30–32] have also identified different NO
adsorption bands on Pd-based catalysts in the same re-
gion. Hoost et al. [30] studied the adsorption of NO on
reduced Pd–Mo catalysts and they observed bands which
were attributed to NO adsorbed on Pd(111) at 1740 and
− −1
1
1
550 cm and NO adsorbed on Pd(100) at 1670 cm . In
another work, Hoost et al. [31] investigated the adsorption
of NO on Pd/alumina catalysts and identified three absorp-
−1
tion regions as follows: 1753–1750 cm (linear), 1615–
− −1
1
1
599 cm (twofold bridged), and 1580–1572 cm (three-
fold bridged), where the linear form was favored when the
samples were reduced. However, in a previous work [13] the
adsorption of NO on the Pd/Al2O3 and Pd–Mo/Al2O3 cat-

L.F. de Mello et al. / Journal of Catalysis 220 (2003) 358–371
371
In accordance with this, a new reaction mechanism was
proposed where NO adsorbs on Pd particles and adsorption
of ethanol on the support and/or MoOx give rise to acetate
species which may decompose and/or react with the NO
molecules at the metal/oxide interface. Indeed, an IR mea-
surement after exposure of Pd8Mo to a NO/ethanol mixture
clearly revealed the presence of acetate species at reaction
conditions. On the other hand, TOF results over exposed
Pd particles showed that the presence of MoOx enhanced
the reaction rate at the metal/oxide interface, which was ex-
plained by the fact that such partially reduced molybdenum
oxide species are in intimate contact with Pd and they favor
the dehydrogenation of ethanol to acetaldehyde (probably
the rate-determining step), as observed by TPD and TPSR
analysis. Finally, we called attention to two important fac-
tors effecting the NO conversion. The first is the covering of
Pd particles by MoOx species while the second is the dehy-
drogenation of ethanol to acetaldehyde.
[7] A. Yee, J. Morrison, H. Idriss, J. Catal. 186 (1999) 279.
[8] Y. Ukisu, T. Miyadera, A. Abe, K. Yoshida, Catal. Lett. 37 (1996) 265.
9] T. Miyadera, Appl. Catal. B 2 (1993) 199.
10] T.J. Toops, A.B. Walters, M.A. Vannice, Catal. Lett. 64 (2000) 65.
11] M.A.S. Baldanza, L.F. de Mello, A. Vannice, F.B. Noronha, M.
Schmal, J. Catal. 192 (2000) 64.
12] M. Schmal, M.A.S. Baldanza, M.A. Vannice, J. Catal. 185 (1999) 138.
13] F.B. Noronha, M.A.S. Baldanza, M. Schmal, J. Catal. 188 (1999) 270.
14] J.E. Benson, H.S. Hwang, M. Boudart, J. Catal. 30 (1973) 46.
[15] L.W. Konopny, A. Juan, D.E. Damiani, Appl. Catal. B 15 (1998) 115.
[16] G.M. Tonetto, M.L. Ferreira, D.E. Damiani, J. Mol. Catal. A 193
[
[
[
[
[
[
(2003) 121.
[
[
17] R.G. Greenler, J. Chem. Soc. 82 (1962) 2488.
18] M. Nagal, R.D. Gonzalez, Ind. Eng. Chem. Prod. Res. Dev. 24 (1985)
525.
[
[
19] I. Halasz, A. Brenner, M. Shelef, Appl. Catal. B 2 (1993) 131.
20] R.W. McCabe, P.J. Mitchell, Ind. Eng. Chem. Prod. Catal. Dev. 23
(1984) 196.
[21] Y. Iwasawa, Y. Nakano, S. Ogasawara, J. Chem. Soc., Faraday
Trans. 74 (1978) 2986.
22] W. Zhang, A. Desikan, S.T. Oyama, J. Phys. Chem. 99 (1995) 14478.
23] T. Noto, T. Murayama, S. Tosaka, Y. Fujiwara, Soc. Automotive
Eng. SP-1626 (2001) 45–52.
[
[
Acknowledgments
[
24] T. Chafik, S. Kameoka, Y. Ukisu, T. Miyadera, J. Mol. Catal. A 136
(
1998) 203.
We thank CAPES for the scholarship (L.F.M.), and CNPq
for financial support.
[
25] S. Kameoka, Y. Ukisu, T. Miyadera, Chem. Phys. 2 (3) (2000) 367.
[26] G.C. Cabilla, A.L. Bonivardi, M.A. Baltanas, J. Catal. 201 (2001) 213.
[27] E. Finocchio, M. Daturi, C. Binet, J.C. Lavalley, G. Blanchard, Catal.
Today 52 (1999) 53.
[
[
28] N. Narishige, M. Niwa, Catal. Lett. 71 (2001) 63.
29] L. Portela, P. Grange, B. Delmon, Catal. Ver. Sci. Eng. 37 (4) (1995)
References
6
99.
[
[
[
[
[
[
1] J.N. Armor, Appl. Catal. B 1 (1992) 221.
[
30] T.E. Hoost, G.W. Graham, M. Shelef, O. Alexeev, B.C. Gates, Catal.
2] D.A. Guerrieri, P.J. Caffrey, V. Rao, SAE paper 950777, 85, 1995.
3] A.P. Altshuller, Atmos. Environ. 27A (1993) 21.
4] V.I. Pârvulescu, P. Grange, B. Delmon, Catal. Today 46 (1998) 233.
5] E.M. Cordi, J.L. Falconer, J. Catal. 162 (1996) 104.
6] H. Idriss, C. Diagne, J.P. Hindermann, A. Kiennemann, M.A. Barteau,
J. Catal. 155 (1995) 219.
Lett. 33 (1996) 57.
[31] T.E. Hoost, K. Otto, K.A. Laframboise, J. Catal. 155 (1995) 303.
[32] M. Valden, R.L. Keiski, N. Xiang, J. Pere, J. Aaltonen, M. Pessa, T.
Maunula, A. Savimäki, A. Lahti, M. Härkönen, J. Catal. 161 (1996)
614.