NO decomposition on PdMo/γ-Al2O3 catalysts

Article abstract of DOI:10.1016/S1381-1169(03)00122-5

The NO decomposition reaction on Pd/γ-Al₂O₃, Mo/γ-Al₂O₃, and PdMo/γ-Al₂O₃ catalysts was studied. The binary catalyst had different behavior for the NO decomposition, being the principal differences a longer steady state activity and an increase of oxygen evolution at elevated temperatures. Pd physicochemical properties were altered by an effective interaction with molybdenum. A decoration effect of Mo over Pd particle was observed. This interaction was responsible for the observed modification in NO activity and selectivity. MoO₃ reduced at higher temperatures was more active. PdMo was more active than Pd and MoO₃ reduced at low temperature and that the activity of PdMo was not the sum of those of Pd and MoO₃ reduced at low temperature.

Full text of DOI:10.1016/S1381-1169(03)00122-5

Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
NO decomposition on PdMo/␥-Al2O3 catalysts
A.M. Eberhardta, E.V. Benvenuttib, C.C. Morob, G.M. Tonettoa,∗, D.E. Damiania,1
a
PLAPIQUI (UNS-CONICET), Camino La Carrindanga km 7, CC 717, 8000 Bah´ıa Blanca, Argentina
b
Instituto de Qu´ımica, Universidad Federal de Rio Grande do Sul—UFRGS, CP 15003, 91501-970 Porto Alegre, RS, Brazil
Received 1 October 2002; received in revised form 11 February 2003; accepted 16 February 2003
Abstract
Studies of the NO decomposition reaction on Pd/␥-Al2O3, Mo/␥-Al2O3 and PdMo/␥-Al2O3 catalysts were made. The
catalytic tests revealed that the binary catalyst has different behavior for the NO decomposition, being the principal differences
a longer steady state activity and an increase of oxygen evolution at elevated temperatures. Characterization by TPR, hydrogen
chemisorption and FT–IR of CO indicated that palladium physicochemical properties are altered by an effective interaction
with molybdenum. The results would suggest a decoration effect of Mo over Pd particle. This interaction appears to be
responsible for the observed modification in NO activity and selectivity.
© 2003 Elsevier Science B.V. All rights reserved.
Keywords: NO decomposition; Pd catalyst; Molybdenum
1. Introduction
of nitrogen oxides in the combustion of typical fuels.
The first is the oxidation of atmospheric N2 at the
high temperatures of combustion. The other one is
the oxidation of nitrogen-containing compounds in the
fuel.
At present, one of the significant problems in air
pollution is the removal of nitrogen oxides. Nitrogen
oxides contribute to smog, can affect human health,
and are harmful to plants and animals. They con-
tribute to acidic deposition. Furthermore, NOx reacts
with certain volatile organic compounds (VOCs) in the
presence of sunlight to form photochemical oxidants,
including ozone.
The reduction of NO in the presence of CO and/or
H2 on supported noble metal catalysts is intensively
investigated. There are many publications about sup-
ported Pd catalysts associated with the removal of NO
in automotive exhausts [1,2]. Ghandi et al. [3] were
among the first in the use of molybdenum in auto-
motive three-way catalysts. They found that improved
net NOx activity, with minimum NH3 formation, is
accomplished by the incorporation of molybdenum
oxide in Pt and Pt-Rh catalysts. Halasz et al. stud-
ied the catalytic reduction of nitric oxide by CO, H2
and CO + H2 on PdMo/Al2O3 [4–6]. They reported
that at higher temperatures molybdenum can improve
the performance of the catalyst, especially at slightly
oxidizing conditions. Schmal et al. [7] confirmed the
high selectivity for N2 in the CO + NO reaction on
The term “NOx” can refer to all of the oxides of
nitrogen, but in air pollution working NOx generally
refers only to NO and NO2. There are two sources
∗
Corresponding author. Present address: University of Western
Ontario, Faculty of Engineering, 1151 Richmond Street, London,
Ont., Canada N6A 5B9. Tel.: +1-519-661211;
fax: +1-519-6613498.
E-mail addresses: tonetto@surf.ssw.uwo.ca (G.M. Tonetto),
ddamiani@plapiqui.edu.ar (D.E. Damiani).
1
Co-corresponding author. Tel.: +54-291-4861700;
fax: +54-291-4861600.
1381-1169/03/$ – see front matter © 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S1381-1169(03)00122-5

248
A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
PdMo/Al2O3. Based on TPD analysis of NO and CO
adsorption, a redox mechanism for NO reduction to
N2 involving partially reduced molybdenum oxide was
proposed. The same group [8] studied the effect of
MoO3 on Pd/Al2O3 catalysts for the reduction of NO
by methane and demonstrated that the selectivity to-
wards N2 was enhanced due to the promoting effect
of NO decomposition on the MoOx surface.
In view of that, the objective of this work is to
examine the effects of molybdenum on the activity
of Pd catalysts for NO decomposition. Experimental
results about NO decomposition reaction are presented
and discussed.
In all cases, prior to impregnation the support was
dried under N2 at 150 ◦C for 2 h. The palladium and
molybdenum contents were determined by atomic ab-
sorption spectroscopy.
2.2. H2 chemisorption
Hydrogen chemisorption experiments were con-
ducted in a conventional glass volumetric apparatus
at 25 ◦C. Catalysts were reduced in flowing H2 at
500 ◦C for 1 h. The samples were then outgassed
overnight before cooling to 25 ◦C and measuring the
first isotherm between 10 and 500 Torr. This con-
sumption is related to the total uptake of hydrogen.
A second isotherm was then obtained after pumping
for 30 min, in order to measure the amount of ab-
sorbed and weakly held hydrogen. The corresponding
uptake values were determined by extrapolating the
linear portion of the isotherms to zero pressure. The
amount of irreversibly adsorbed gas was obtained by
subtracting these values.
2. Experimental
2.1. Catalysts preparation
Pd/Al2O3 sample was prepared by wet impregna-
tion of the support with a solution of Pd(C5H7O2)2
(Alfa) in toluene (Merck). The support material was
␥-Al2O3 (Condea, Puralox, 148 m2 g−1). After im-
pregnation, the catalyst was dried in Ar at 150 ◦C for
2 h and then calcined in air (chromatographic grade)
at 500 ◦C during 2 h. Mo/Al2O3 catalyst was prepared
in similar way using a solution of MoO2(C5H7O2)2
(Alfa) in toluene.
The bimetallic samples were prepared by wet im-
pregnation utilizing the same precursors. The metals
were co-impregnated or impregnated in sequence. Af-
ter impregnation the sample was dried and calcined
similarly to the monometallic catalysts. Table 1 sum-
marizes catalyst preparation data.
The fraction of exposed palladium was calculated
assuming that one hydrogen atom is adsorbed per sur-
face palladium atom.
2.3. O2 sorption experiments
Oxygen sorption experiments were carried out at
400 ◦C in the same volumetric apparatus used for H2
chemisorption measurements. Previously, the samples
were pre-reduced in H2 flow at 500 ◦C for 2 h and
evacuated at this temperature overnight. An ultimate
vacuum near 10−5 Torr was obtained. The O/Me ratio
(Me: Pd, Mo or Pd + Mo) ratios were determined by
Table 1
Characterization of the catalysts
a
Catalyst
Metal loading (wt.%)
Pd
Preparation
Pdexp (%)
Mo
Pd/Al2O3
Mo/Al2O3
PdMo(c)/Al2O3
PdMo/Al2O3
MoPd/Al2O3
0.82
–
0.45
1.00
0.95
–
w.i.b
w.i.
55
–
36
45
39
0.71
0.74
0.87
0.76
w.i.–co-impregnation
w.i.–sequencec
w.i.–sequenced
Metallic content and hydrogen uptakes as exposed metal percentage.
a Fraction of exposed palladium.
b w.i.: Wet impregnation.
c Pd impregnated on Mo/Al2O3.
d Mo impregnated on Pd/Al2O3.

A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
249
extrapolating the sorption isotherm to zero pressure to
correct for physisorption on the support. In order to
test whether the oxygen taken up forms a stable oxide,
a second isotherm was recorded after an evacuation
of 30 min at 500 ◦C. High purity oxygen was used
(99.99% from AGA).
mocouple placed in the furnace close the catalyst bed.
The reaction mixture contained 1100 ppm of NO in
He. The reaction products were analyzed in a Varian
3700 gas chromatograph operating in the TCD mode.
Two Porapak Q columns operated isothermally at 40
and −20 ◦C with He as carrier gas were used for sep-
aration. Chromatographic peaks were integrated by
means of a Hewlett-Packard 3390 A integrator.
2.4. TPR experiments
The catalytic activity measurements were done
on catalyst samples that were previously reduced at
300 ◦C in the presence of a 5% H2 in Ar mixture.
The tests were done in two ways. (1) The catalyst
sample was heated to 400 ◦C in He flow. Once the
final temperature was reached the catalyst was con-
tacted with the reaction mixture during 2 h. Samples
of reaction products were taken regularly. (2) The
catalyst sample was heated in the reaction mixture
from 20 to 500 ◦C. Samples of the reaction products
were taken at different reaction temperatures. Cata-
lyst weight was adjusted in order to keep constant
the weight of palladium in the different experiments.
The reaction mixture flow rate was maintained at
TPR experiments were carried out in an appara-
tus similar to that described by Robertson et al. [9]
to which some changes were introduced [10]. Stan-
dard methods of gas purification were used. The ref-
erence and the sample were held in an oven that could
be heated at linear rate of 10 ◦C min−1 between 0
and 500 ◦C. The temperature of the catalyst bed was
measured with a 1/16 in. od stainless steel sheathed
Chromel–Alumel thermocouple placed in contact with
the top of the catalyst bed. The flow rate of reducing
gas (a 5%H2–95%Ar mixture) was 20 cm3 min−1
.
Fresh samples of catalysts (40 mg) were oxidized in
flowing air chromatographic grade at 500 ◦C for 1 h,
purged and cooled in Ar to −50 ◦C, followed by TPR.
35 cm3 min−1
.
2.5. IR studies
3. Results and discussion
Five square centimeter disks of catalysts, weighting
50 mg, were made by pressing the catalyst powder.
They were heated to 300 ◦C in vacuum (10−5 Torr)
during 1 h. Then they were exposed to 200 Torr of
oxygen at 200 ◦C during 1 h. The oxygen was pumped
off to a final pressure of 10−5 Torr during 1 h at 300 ◦C.
Following this, the material was exposed to 300 Torr
of hydrogen at 200 ◦C for 1 h, and finally evacuated
for an additional hour at 300 ◦C. The treatments were
carried out using a quartz IR cell [11]. The infrared
spectra were recorded on a FT–IR 8300 Shimadzu
spectrometer with 4 cm−1 resolution and 100 scans.
The spectra were recorded at room temperature. The
CO pressure in the analysis camera was 25 Torr.
3.1. Catalysts preparation and H2 chemisorption
Table 1 presents palladium and molybdenum
contents and summarizes the hydrogen chemisorp-
tion measurement results. Exception made of the
PdMo(c)/Al2O3 the remaining samples resulted
with about the same Pd and Mo loading. The table
shows a decrease in the H2 chemisorption capacity
of palladium due to molybdenum addition. Similar
results were reported by Konopny et al. [12] for
PdMo-alumina catalysts prepared using different Pd
precursors. Also, it has been observed in our labo-
ratory for RuMo samples [13,14] and by others in
RhMo catalysts [15]. High temperature oxidation fa-
vors the formation of MoO3 that after exposure to
pure H2 at 300 ◦C reduces to suboxides that could
migrate onto the palladium surface impeding its con-
tact with H2. It is well known that low molybdenum
loading on alumina lead to quite stable molybdenum
species. However, it has been reported evidences
of MoO3 formation on bimetallic RuMo catalysts
2.6. Catalytic activity measurements
The catalysts were tested in a glass-made packed
bed reactor (6 mm od) placed in an electrically heated
oven. A temperature controller was used to control
the reaction temperature (Omega CN3800), which was
measured by means of a shielded J-type Omega ther-

250
A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
despite of their low molybdenum content [14]. Sur-
face octahedral molybdate and tetrahedral species
have been detected, even with molybdate loading
down to 0.1 wt.% on ␥-alumina. Xiong et al. [16]
reported that the octahedral molybdate is formed on
␥-alumina at all molybdate loading.
Noronha et al. [17] found that molybdenum addi-
tion induces modifications in the Pd metallic phase. In
Pd-MoO3/Al2O3 catalyst they observed: (i) a decrease
in the amount of adsorbed hydrogen; (ii) an increase
in the COlineal/CObridge ratio; and (iii) the disappear-
ance of the NO adsorption band on palladium. These
results suggest the blockage of palladium surface by
the molybdenum oxide species.
unity that was attributed to the presence of small quan-
tities of PdO2. This explanation can not account for
the high O/Pd ratio obtained for the monometallic Pd
sample since PdO2 can not be formed under the exper-
imental conditions prevailing during the experiments
reported in Table 2. The formation of Pd2O3 is also
discarded since it decomposes at 0 ◦C.
Hoost and Otto [19] showed evidence of oxygen
uptake higher than unity for PdO during temperature
programmed oxidation (TPO) of Pd/␥Al2O3 catalysts
having low Pd loading. The authors suggested a higher
oxidation state of Pd or the presence of highly dis-
persed oxygen-rich species following similar results
by Lesage-Rosemberg et al. [20].
The precursor and the impregnation sequence also
influence the exposed metallic fraction. When molyb-
denum was incorporated by impregnation on the
Pd/Al2O3 catalyst, the exposed palladium fraction
was smaller than on catalysts prepared using an in-
verse order of impregnation. That is to say, when a
Mo/Al2O3 sample was impregnated with a Pd(Acac)2
solution. The lowest value in the exposed metallic
fraction was presented in the PdMo(c)/Al2O3 catalyst,
in which the metallic precursors were co-impregnated
on the support.
The amount of oxygen consumed in the oxidation
of Mo is obtained as the difference between the to-
tal oxygen uptake by the sample and the amount of
oxygen adsorbed on Pd [21]. The following balances
apply:
H2 balance
PdO + H2 → Pd0 + H2O
(1)
(2)
Pd0 + H2 → Pd-H
1
2
3.2. O2 sorption
H/Pd (titration with H2) − Pd(s)/Pd (= H/Pd)
= [H2 used to reduce PdO]·
(3)
The oxygen consumption is reported related to the
moles of Pd present in the sample. Previous experi-
ments carried on alumina showed that physisorption
onto the support is negligible. Therefore, the observed
oxygen uptake is due to the metals. Data is given in
Table 2.
The O/Pd ratio obtained for the monometallic pal-
ladium sample is higher than expected. Leon y Leon
and Vannice [18] also found O/Pd ratios higher than
O2 balance
Pd0 + O2 → PdO
(4)
(5)
1
2
1
2
MoOx−1 + O2 → MoOx
According to Eq. (1), there are two hydrogen atoms
per oxygen atom, therefore, to obtain the amount of
oxygen used to oxidize Pd, Eq. (3) must be divided
by 2. Therefore,
Table 2
Characterization of the catalysts
b
[H/Pd (titration with H2) − H/Pd]
Catalyst
O–Pda (Treduction = 500 ◦C)
NMo
O/Pd⊕
=
·
(6)
2
Pd/Al2O3
Mo/Al2O3
PdMo(c)/Al2O3
PdMo/Al2O3
1.47
0.23
2.42
2.75
–
5.5
5.0
3.3
Considering Pd and Mo loading and their molecular
weight the oxygen uptake by Mo is:
Pd(%) PMMo[O/Pd (from Table 2)
−O/Pd⊕]
O2 chemisorption.
a O2 chemisorption on catalysts pre-reduced at 500 ◦C.
b Mo oxidation state (in samples reduced at 500 ◦C).
O/Mo =
·
(7)
Mo(%) PMPd

A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
251
Then the oxidation state of Mo is:
catalyst evidences very low hydrogen consumption
just before 500 ◦C. It is important to note that no
change in the sample color was observed (Mo/Al2O3
is white after calcination and this color remains after
the TPR experiment).
NMo = 6 − 2 (O/Mo) ·
(8)
According to the data shown in Table 2 the presence
of Pd increases Mo reducibility. De Vries et al. [22]
reported that the reduction of MoO3-PtO2/␥-Al2O3
with H2 leads to a mixture of Mo(VI), Mo(V) and
Mo(IV) species. They found a relationship between
the extent of Mo(VI) reduction and the amount of
PtO2 in the sample. Similar results were given for the
Pd-Mo system [17,23].
A characteristic negative peak attributed to palla-
dium hydride decomposition is usually observed in the
TPR profile of Pd catalysts. This peak correlates well
with the average Pd particle size since the hydride for-
mation is a bulk phenomenon. In the present case, the
negative peak is hardly visible. The relative high dis-
persion of the monometallic sample may inhibit the
formation of the hydride. The presence of broad hydro-
gen consumption peak in the 80–200 ◦C temperature
range could prevent the observation of the negative
peak in the bimetallics. This broad peak would be re-
lated to highly dispersed palladium oxide, which is
reduced at temperatures higher than 100 ◦C [24] or
palladium–molybdenum species more strongly linked
3.3. TPR experiments
TPR profiles of catalysts are given in Fig. 1, where
the hydrogen consumption is plotted against tempera-
ture. The positive peaks between 10 and 40 ◦C in the
TPR profiles are assigned to the reduction of palladium
oxide. On the other hand, the TPR profile of Mo/Al2O3
Fig. 1. TPR profiles. Oxidation temperature = 500 ◦C.

252
A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
to the support than the PdO species, which would ex-
plain the reduction at higher temperature. Another pos-
sible explanation would be the bronze formation, due
to the hydrogen spillover from the noble metal toward
molybdenum trioxide. The process can be represented
as follows:
According to Tessier et al. [26], the absorption
band at 2150 cm−1 can be assigned to Pd2+–C O.
≡
They found ionic palladium in catalysts prepared
using acetilacetonate precursors, after reduction at
300 ◦C. These authors consider that the high exposed
metallic fraction (65%) favors a transfer of electrons
from palladium to the support. This type of species
was not detected at higher reduction temperatures.
Evacuation of the phase gas during 30 min at room
temperature induces a strong decrease in the inten-
sity of the band corresponding to lineal CO. This phe-
nomenon was also observed by Pitchon et al. [29] and
it was pointed out as an indication that lineal species
are adsorbed on low coordination sites without any
interaction with bridged species. The intensity of the
assigned signal to CO multiply connected remains
practically unaffected, although a shift to smaller wave
numbers was observed. This result is coincident with
previous works. Eischens and Pliskin [30] observed
that the frequencies of the bands move toward higher
wave numbers upon increasing the surface coverage
with CO. This was assigned to the stronger competi-
tion among adsorbed CO molecules for electrons of
superficial metallic atoms.
H2 + MoO3 → HxMoO3
where the stable stoichiometry of bronze is H1.6MoO3
[25]. This H2 consumption, which is not present in
the Pd/Al2O3 sample, demonstrates the interaction be-
tween palladium and molybdenum.
3.4. IR studies
As it is presented in Fig. 2, the IR spectrum of
CO adsorbed on Pd/Al2O3 catalyst showed two main
bands of absorption at 2094 and 1975 cm−1, a shoulder
weakly insinuated at 1940 cm−1 and a small band at
2150 cm−1. The first two bands are similar to those re-
ported by numerous authors for CO adsorption on sup-
ported palladium catalysts [26–28]. They are assigned
to lineal and multiply connected CO, respectively. The
bands of low frequency are generally attributed to ad-
sorbed species in bridge form on the (1 0 0) and (1 1 1)
planes of palladium [27,28]. The shoulder present at
low frequency has been attributed to multiply coordi-
nated species [29].
Fig. 3 shows the FT–IR spectrum of CO ad-
sorbed on PdMo/Al2O3 sample. The same absorption
bands as in Pd/Al2O3 catalyst were found. The most
Fig. 2. FT–IR spectra of CO absorbed on Pd/Al2O3 sample
(treatment = 300 Torr H2 at 200 ◦C for 1 h). References: (a) cata-
lyst; (b) 25 Torr of CO; (c) 30 min vacuum.
Fig. 3. FT–IR spectra of CO absorbed on PdMo/Al2O3 sample
(treatment = 300 Torr H2 at 200 ◦C for 1 h). References: (a) cata-
lyst; (b) 25 Torr of CO; (c) 30 min vacuum.

A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
253
evident difference among spectra is the change in the
COlineal/CObridged relationship.
According to Peri [33], characteristic absorption
bands for CO on molybdenum are seen around 2025
Determining the concentration of palladium species
by integration of signals intensities is not possible be-
cause the absorption coefficients “ε” are not known.
However, for samples containing the same species
(showing bands at identical positions) it is reasonable
to admit that ε-values do not vary significantly from
one sample to the other. In this way, the relationship of
relative abundance between lineal and multiply con-
nected CO could be estimated integrating the intensi-
ties of signals at 2094, 1975 and 1940 cm−1 expressed
in the following way:
and 2045 cm−1 for Mo0 and at 2100 cm−1 for Moδ+
.
We did not find any signal for CO adsorbed on Mo
neither on Mo/Al2O3 nor on PdMo/Al2O3 (see Fig. 3).
This would be due to the absence of partially reduced
species. It has not been observed CO adsorption on
calcined Mo/alumina catalysts [33].
For Pd/Al2O3 and PdMo/Al2O3 spectra, the in-
crease in intensity of bands at 1440 and around
1650 cm−1 is assigned to carboxylate formation over
alumina [34].
3.5. Catalytic activity measurements
I2094 cm−1
τ =
·
)
Fig. 4 shows the NO conversion as a function of time
at 400 ◦C after reduction treatments at 300 ◦C. Since
the same palladium mass was used in reaction, rela-
tive performances can be discussed. The Pd contain-
ing samples presented a similar qualitative behavior:
the initial NO decomposition at 400 ◦C on the reduced
surface was almost complete and then it diminished
gradually. The only observed products were N2 and
N2O. It would indicate that the decrease of the activ-
ity in the time was originated by the oxygen retention
on the metallic surface. In all the cases, the nitrogen
balance allowed to discard the possible formation of
NO2.
Fig. 5 presents the activity corresponding to
Mo/Al2O3 sample. The Mo catalyst reduced at low
temperature was weakly active for the NO decom-
position. The activity fluctuated among 2–5% during
the reaction time. The total content of molybdenum
in the catalyst was similar to the amount of molybde-
num in the PdMo/Al2O3 sample used in the catalytic
tests. According to TPR results (Fig. 1), molybdenum
is present as MoO3 over the support after reduction
at 300 ◦C.
It has been reported that alumina shows some activ-
ity for NO decomposition [35]. Fig. 5 superimposes
NO conversion obtained on alumina (after being ex-
posed to hydrogen at 300 ◦C) with the results obtained
on the Mo/Al2O3 sample. This result indicates that
the catalytic activity showed by Mo/Al2O3 reduced at
300 ◦C could de due to the contribution of the sup-
port. Therefore, this would indicate that the promotion
effect that molybdenum exerts on the NO decomposi-
tion reaction is due to its interaction with palladium.
(I1975 cm−1 + I1940 cm−1
For Pd/Al2O3 and PdMo/Al2O3 samples, τ was 0.5
and 0.7, respectively. These results suggest that molyb-
denum addition generates a decrease of multiply con-
nected species on palladium. This observation could
be explained assuming that Pd is partially covered by
molybdenum species, and this result is in agreement
with those of H2 chemisorption. Thus, when the ex-
posed Pd surface decreases, the adsorption of linearly
adsorbed CO is also favored.
Molybdenum addition also originates that the as-
signed signal for lineal CO on palladium moves to-
ward lower wave numbers. Balakrishnan et al. [31]
reported the same effect in PtAu/SiO2 catalyst versus
the one for the monometallic sample. They attributed
this to a decrease of dipole–dipole interactions. Volpe
reported a comparable shift on the IR spectra of CO
adsorbed on Pd-Pb/alumina [32]. This author postu-
lates that although a larger number of molecules lin-
early adsorbed is found in the bimetallic system, these
molecules are not placed in adjacent positions and the
interaction between dipoles is expected to be smaller
due to dilution that originates the Pb addition. How-
ever, it is also pointed out that Pb addition, forming a
superficial alloy, could produce an electronic change
in palladium. If the electronic density is increased in
=
this metal, it could diminish the strength of C O bond
because of a larger backbonding of electrons from Pd
to C. The latter would be the cause of the signal shift
to smaller frequencies in bimetallic catalyst. This ar-
gument has been used in many occasions to postulate
electronic effects in bimetallic systems.

254
A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
Fig. 4. NO conversion as a function of reaction time at 400 ◦C. SV = 4 × 104 mol feed/(h mol Pd).
The same figure reveals an increase in the activity of
Mo/Al2O3 sample when both, the reduction temper-
ature and reduction time increase (500 ◦C for 2 h). A
severe reduction treatment would originate a higher
number of oxygen vacancies. These are postulated in
[36,37] as the possible active site for NO decomposi-
tion on metallic oxides. These oxygen vacancies would
be originated during a reduction treatment not achiev-
able in the TPR study.
In general, the bimetallic catalysts presented a bet-
ter behavior that the palladium sample. The data pre-
sented in Fig. 4 can be analyzed in terms of the activity
presented by the catalysts after a reaction time of 2 h,
named from now as “residual conversion” (Xresidual).
Fig. 5. NO conversion over alumina and Mo/Al2O3 sample (after different reduction treatments) at 400 ◦C. SV = 4 × 104 mol feed/(h mol
Pd).

A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
255
Fig. 6. Residual conversion after 2 h in reaction at 400 ◦C. SV = 4 × 104 mol feed/(h mol Pd).
Fig. 6 presents the residual conversion of all the stud-
ied samples. The addition of molybdenum originates
an increase of Xresidual in bimetallic catalysts with re-
gard to Pd/Al2O3 catalyst.
temperature while the residual activity remained the
same for the three catalysts. Increasing the reduction
temperature results in a more severe Pd coverage by
Mo species. This was observed also for RuMo cata-
lysts [13,14]. The Mo suboxides migrate towards pal-
ladium particles during the reduction treatment, di-
minishing the Pd exposed area. This decoration effect
originates a decrease in tx = 100% value, since molyb-
denum is much less active than Pd for the decomposi-
tion of NO. For this reason, tx = 100% diminishes when
the reduction temperature increases.
It is more difficult to understand why Xresidual value
remains constant in spite of the different reduction
treatment since the higher the reduction temperature,
the higher the number of co-ordinative unsaturated
sites (CUS) presents on MoOx species. The fact that
equal Xresidual values are obtained leads to conclude
that this parameter, although depending on Mo, it is
not a direct function of the CUS number in MoOx.
Fig. 7 can be interpreted taking into account the
ideas represented in Scheme 1(A). Upon increasing the
reduction temperature, the number of MoOx species
increases. The Pd decoration process modifies the ex-
tension of the Pd–MoOx interface. The more severe
the decoration process, the shorter the Pd–MoOx in-
terface. Therefore, the number of active sites in the in-
terface drops during the process of Pd masking. At the
same time, those in the MoOx region increase. In this
Comparing the activity of Pd/Al2O3, Mo/Al2O3 and
different Pd-Mo systems in terms of conversion, it is
found that those corresponding to the bimetallics are
not the sum of the activities of the monometallics. It
should be considered that in the bimetallic catalysts
there are different potential active sites for the studied
reaction. We speculate that those sites could be: Pd,
MoO3, Al2O3, MoOx and the Pd–MoOx interface. The
catalytic activity of Pd, MoO3 and Al2O3 is known. It
was shown in Figs. 4 and 5. In the bimetallic catalysts,
the interaction between the metals is important. The
reducibility of Mo increases in these catalytic systems,
therefore, there will be more MoOx species in Pd-Mo
samples than in Mo/Al2O3 catalyst.
To further elucidate the relationship between the
activity and a Pd–Mo interaction, the reaction was
carried out over PdMo/Al2O3 catalyst, using fresh
samples reduced at different temperatures: 300, 400,
and 500 ◦C. During the studies, a mass of catalyst
smaller than the one employed during the first reaction
test, was used. The results are presented in Fig. 7. It
can be observed that the time during which the catalyst
exhibited very high conversion (from now on named
t
x = 100%) diminished with the increase of reduction

256
A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
Fig. 7. NO conversion over PdMo/Al2O3 catalyst (after different reduction treatments over fresh samples) at 400 ◦C. SV = 4 × 104 mol
feed/(h mol Pd).
Scheme 1. Possible sites on bimetallic catalyst.
way, the different variables under consideration could
be compensated and it would explain experimental re-
sults. In Scheme 2, the relationship between different
parameters is visualized: residual activity depends on
Pd, MoOx and Pd–MoOx. The total number of sites
is approximately constant, since the increase of active
sites on MoOx is compensated by the loss of sites on
the Pd surface and the Pd–MoOx interface.
The following reduction–reaction sequence was car-
ried out on a given sample in order to study a possible
reactivation of the catalysts by means of a reduction.
• Reduction at 300 ◦C in Ar/H2: 95/5 mixture during
Scheme 2. Relationship between residual activity with Pd, MoOx
and Pd–MoOx sites.
30 min, purge in He.

A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
257
Fig. 8. NO conversion over PdMo/Al2O3 catalyst (after different reduction treatments over the same sample) at 400 ◦C. SV = 4 × 104 mol
feed/(h mol Pd).
• Reaction at 400 ◦C during 120 min, purge in He.
• Reduction at 500 ◦C in Ar/H2: 95/5 mixture during
30 min, purge in He.
The experimental results point out a degradation of
the active sites during the reaction. The initial activity
in each cycle is similar to the final activity for the
previous cycle, indicating a “memory” of the catalyst
that is not possible to modify with severe reduction or
oxidation treatments.
The information obtained from the monometallic
catalysts suggests that this behavior could be related
essentially to palladium. On the other hand, the resid-
ual activity of PdMo/Al2O3 catalysts after the sec-
ond reduction at 300 ◦C is smaller than the one of
Mo/Al2O3. This would indicate a Pd–Mo interaction.
Therefore, the aging or deterioration of the active site
would also include Mo species.
Recently, it was reported [38] that palladium is
re-dispersed in the presence of NOx. The formation
of (PdO)NO volatile species is suspected. To analyze
this last hypothesis, X-ray fluorescence analysis of the
fresh and aged samples were done. The equipment
used was a X-ray fluorescence spectrometer, PHILIPS
PW 1400 with chromium anode, operated at 60 kV and
40 mA. The results indicated that some loss of molyb-
denum (9.8%) took place after 6 h in reaction. This
loss is associated to the relative high partial pressure
of molybdenum oxides. There was not any palladium
loss.
• Reaction at 400 ◦C during 120 min, purge in He.
• Oxidation at 500 ◦C in air during 90 min, purge in
He.
• Reduction at 300 ◦C in Ar/H2: 95/5 mixture during
30 min, purge in He.
• Reaction at 400 ◦C during 120 min.
Fig. 8 shows the NO conversion on the PdMo/Al2O3
catalyst. A gradual deactivation of the catalyst is ob-
served after the sequence of different treatments. It
is found that the residual activity diminishes with the
progressive cycles of reduction–reaction. It is also no-
torious that initial activity in each cycle is similar to
the final activity of the previous cycle.
Therefore, the reduction of an aged catalyst does
not allow reactivating the catalyst. An oxidation at
500 ◦C followed by a mild reduction does not turn to
the original. This treatment is expected to produce the
formation of MoO3 leading to a Mo free Pd surface
almost identical to the original one.
Similar results, not shown here, were obtained for
Pd/Al2O3. In the case of the Mo/Al2O3 sample, its ac-
tivity increased when the reduction treatment tempera-
ture was increased (Fig. 5). As already mentioned, this
would be a consequence of an increase of the number
of CUS on the molybdenum surface.
Using transmission electron microscopy (JEOL
100CX), the statistical distribution of particle size
was obtained for the same sample. As it is observed

258
A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
Fig. 9. Particle size histograms obtained from TEM micrographs of fresh and aged PdMo/Al2O3 catalyst exposed to NO decomposition.
in the histograms presented in Fig. 9, the average
particle diameters were 3 and 6.7 nm for fresh and
aged PdMo/Al2O3 catalyst, respectively. Also, after
reaction, the particles lost their rounded shape.
Conrad et al. [39] found that the interaction of NO
with Pd(1 1 1) at 727 ◦C causes the formation of PdO
by means of the NO → N2 +PdO reaction. They con-
sider that the oxidation of the surface originates the
formation of PdO microcrystals whose (1 0 0) planes
are parallel to the Pd(1 1 1) surface. The oxide phase
persists even after vacuum treatments at high temper-
atures. This is not expected for PdO bulk according to
its thermodynamic properties. After elimination of the
upper layers, the LEED pattern turns to be identical
to the one of Pd(1 1 1). But the photoemission spec-
trum showed evidences of existence of considerable
quantities of oxygen inside the metal.
Figs. 10 and 11 show the product distribution as
a function of the reaction temperature. For both the
catalysts, the maximum formation of N2O occurs at
350 ◦C and it concludes at 400 ◦C, but there is a higher
formation of N2O in the monometallic sample. Our
results also show a much higher selectivity to N2 in
the presence of molybdenum. This agrees with data
Fig. 10. Temperature dependence of decomposition of NO over Pd/Al2O3 catalyst. SV = 4 × 104 mol feed/(h mol Pd).

A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
259
Fig. 11. Temperature dependence of decomposition of NO over PdMo/Al2O3 catalyst. SV = 4 × 104 mol feed/(h mol Pd).
reported by da Silva and Machado [40]. They pub-
lished the selectivity values calculated from tempera-
ture programmed desorption (TPD) profiles of NO on
Pt-yMo/␥-Al2O3 catalysts that show that N2 selectiv-
ity increases with molybdenum content in the catalyst.
The oxygen evolution is different in the studied
samples. Over Pd/Al2O3, oxygen production begins at
350 ◦C, it shows a maximum value at 400 ◦C. On the
other hand, over PdMo/Al2O3 sample, the oxygen for-
mation starts when the reaction temperature reaches
300 ◦C and it continues increasing with the tempera-
ture. This result could indicate that molybdenum ad-
dition to the palladium catalyst favors the release of
oxygen from the surface. N2 formation at 500 ◦C on
Pd/Al2O3 sample is not enough to close the nitrogen
balance. Iwamoto et al. [41] consider that the loss of
nitrogen and oxygen in the balance can be attributed
to the formation of NO2.
soning are increased. The active species seems to be
a complex of cobalt and silver oxides. The enhance-
ment observed with silver may be due to its low affin-
ity for oxygen. El Hamadaoui et al. [2] indicated that
chromium promotes the dissociation of NO compared
to pure palladium. Under static conditions, the behav-
ior of this catalyst is similar to that of Rh. It was sug-
gested that this behavior was due to the influence of
this element on the electronic properties of palladium.
Another explanation for the nature of the interac-
tions between Pd and Mo could be considered. The
activity of the catalytic system could be related to the
formation of oxygen vacancies lying on the thin ox-
ide overlayer that covers the metal. The mechanism
of formation is associated to a junction effect between
the metal and the oxide as proposed by Frost [43] us-
ing a detailed analysis of the electronic properties of
the surface.
On metal oxide compounds, it is generally accepted
that active sites are adjacent anion vacancies [36]. IR
studies show [40] that NO adsorbed preferentially on
MoOx species in Pt-yMo/␥-Al2O3 catalyst. Schmal
et al. [7] suggested that in CO + NO reaction on
PdMo/␥-Al2O3 catalysts, NO adsorption and dissoci-
ation occur on species Mo4+. It has also been reported
that a promoter can reduce oxygen inhibition. Hamada
et al. [42] have shown that by adding Ag to cobalt
oxide, both the activity and resistance to oxygen poi-
We have been working with M-Mo systems for sev-
eral years in different catalytic systems. It is clear to
us that Mo tends to cover nearby metallic particles
of most of Group VIII metals. There is an impor-
tant difference in the heat of vaporization of Mo and
Group VIII metals (for example: the heat of vaporiza-
tion is 33 kcal/(g mol) and 148 kcal/(g mol) for MoO3
and Ru, respectively). Besides the vapor pressure of
MoO3 at temperatures close to 400 ◦C is 1 mmHg ap-
proximately, while for bulk Pd a similar vapor pressure

260
A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
is reached at substantially higher temperatures. There
are also very important differences in free surface en-
ergy, being considerably lower the free surface energy
of MoO3 that is mobile at the temperatures used in the
pre-treatments of our paper.
In sum, we considered exhaustively the activity of
the studied catalytic system in terms of the forma-
tion of oxygen vacancies. Although it looks plausi-
ble, based on our previous experience on Mo-modified
metals used in a family of reactions and in the IR and
TPR experiments of the present paper we believe that
our proposal of active sites formed in the Pd–MoOx
interface offers a reasonable explanation of the ob-
served phenomena.
Although the catalytic decomposition of NO to O2
and N2 over noble metals or oxide catalysts is too slow
to be considered on a practical scale, this reaction is
significant for its activity in other important pollution
control reactions [46].
In all the M-Mo systems that we studied in the last
12 years we always found a decrease in hydrogen or
CO chemisorption on the metal M, with M being Pd,
Ru or Rh, when Mo was added to the catalyst regard-
less the procedure used in the preparation. This (to-
gether with other experiments) indicated some sort of
surface coverage by Mo of the M surface. We never
found, however, that the presence of the overlayer in-
creased the activity in the case of the syngas reac-
tion to produce oxygenates. We found a severe loss of
activity, although accompanied by an increase of the
selectivity to methanol and ethanol [13,14] that theo-
retical studies suggested was associated to changes in
the adsorption modes of CO induced by the presence
of MoOx moieties on the surface of Ru [44]. How-
ever, using PdMo catalysts we found that the bimetal-
lic sample was more active than monometallic Pd for
the combustion of methane [12]. Although in this case,
the increase of activity (or its absence) for methane
combustion depended on the Pd precursors used in the
catalyst preparation.
Therefore, it seems that the same combination of
metal and oxides do not result in similar effect in
all cases. It is evident that the explanation proposed
by Frost fits better in those cases, where vacancies
are postulated to be part of the reaction mechanism.
In the case of the paper under discussion, it is clear
that MoO3 reduced at higher temperatures is more ac-
tive. It is also clear that PdMo is more active than
Pd and MoO3 reduced at low temperature and that
the activity of PdMo is not the sum of those of Pd
and MoO3 reduced at low temperature. However, the
PdMo could indeed be the sum of the activities of
Pd and MoO3 on top of Pd behaving like MoO3 re-
duced at high temperature as a consequence of the
presence of vacancies generated by a process like
Frost’s. It is not clear why the TPR or the IR ex-
periments do not detect the formation of reduced Mo
oxides in PdMo. Although we found that totally cov-
ered Pd by MoOx moieties allowed the formation of
Pd hydride indicating that the blockage of the sur-
face was not complete and that reduction was possible
[45].
4. Conclusions
The catalytic tests revealed that the binary catalysts
has different behavior for the NO decomposition re-
action, being the principal differences a better perfor-
mance in the reaction time and an increase of oxygen
evolution an elevated temperatures for PdMo/alumina
catalyst when comparing with its molybdenum free
counterpart. Characterization by TPR, hydrogen
chemisorption and FT–IR of CO indicated that pal-
ladium physicochemical properties are altered by an
effective interaction with molybdenum. The results
would suggest a decoration effect of Mo over Pd par-
ticle. This interaction appears to be responsible for
the observed modification in NO activity.
Acknowledgements
We thank the Universidad Nacional del Sur
(UNS) and the Consejo Nacional de Investigaciones
´
Cientıficas y Técnicas (CONICET) for their financial
support.
References
[1] M. Valden, J. Aaltonen, E. Kausisto, M. Perra, C. Barnes,
Surf. Sci. 307 (1994) 193.
[2] A. El Hamadaoui, G. Bergeret, J. Massardier, M. Primet, A.
Renouprez, J. Catal. 148 (1994) 47.
[3] H. Gandhi, H. Yao, H. Stepien, Catalysis Under Transient
Conditions, in: A. Bell, L. Hedegies (Eds.), ACS Symp. Ser.

A.M. Eberhardt et al. / Journal of Molecular Catalysis A: Chemical 201 (2003) 247–261
261
No. 178, Am. Chem. Soc., Washington, DC, 1982,
p. 143.
[23] S. Kawi, O. Alexeev, M. Shelef, B. Gates, J. Phys. Chem.
99 (1995) 6926.
[4] I. Halaz, A. Bremmer, M. Shelef, Appl. Catal. B 2 (1993)
131.
[5] I. Halaz, A. Bremmer, M. Shelef, Catal. Lett. 16 (1992)
311.
[6] I. Halaz, A. Bremmer, M. Shelef, Catal. Lett. 18 (1993)
289.
[7] M. Schmal, M. Baldanza, M. Vannice, J. Catal. 185 (1999)
138.
[24] H. Lieske, J. Volter, J. Phys. Chem. 89 (10) (1985) 1842.
[25] P. Sermon, G. Bond, J. Chem. Soc., Faraday Trans. 1 (72)
(1976) 730.
[26] D. Tessier, A. Rakai, F. Bozon-Verduraz, J. Chem. Soc.,
Faraday Trans. 88 (5) (1992) 741.
[27] F. Hoffmann, Surf. Sci. Rep. 3 (1983) 107.
[28] A. Ortega, F. Hoffmann, A. Bradsshaw, Surf. Sci. 119 (1982)
79.
[8] L. de Mello, M. Baldanza, F. Noronha, M. Schmal, Studies
in Surface Science and Catalysis, vol. 130, in: A. Corma, F.
Melo, S. Mendioroz, J. Fierro (Eds.), Elsevier, Amsterdam,
2000, pp. 647.
[9] S. Robertson, B. Mc Nicol, J. de Baas, S. Kloet, J. Jenkins,
J. Catal. 37 (1975) 424.
[29] V. Pitchon, M. Primet, H. Praliaud, Appl. Catal. 62 (1990)
317.
[30] R. Eischens, W. Pliskin, Adv. Catal. 10 (1958) 1.
[31] K. Balakrishnan, A. Sachev, J. Schwank, J. Catal. 121 (1990)
441.
[32] M. Volpe, Thesis, UNS, 1991.
[10] D. Damiani, E. Pérz Millán, A. Rouco, J. Catal. 101 (1986)
162.
[11] J. Foschiera, T. Pizzolato, E. Benvenutti, J. Braz. Chem. Soc.
12 (2001) 159.
[12] L. Konopny, A. Juan, D. Damiani, Appl. Catal. B 15 (1998)
115.
[13] A. Juan, D. Damiani, J. Catal. 137 (1992) 77.
[14] A. Juan, D. Damiani, Catal. Today 15 (1992) 169.
[15] B. Kip, E. Hermans, J. van Wolput, N. Hermans, J. van
Grondelle, R. Prins, Appl. Catal. 35 (1987) 109.
[16] G. Xiong, C. Li, Z. Feng, P. Ying, Q. Xin, J. Liu, J. Catal.
186 (1999) 234.
[17] F. Noronha, M. Baldanza, M. Schmal, J. Catal. 188 (1999)
270.
[18] C. Leon y Leon, M. Vannice, Appl. Catal. 69 (1991) 569.
[19] T. Hoost, K. Otto, Appl. Catal. A 92 (1992) 39.
[20] E. Lesage-Rosemberg, G. Vlaic, H. Dexpert, P. Legarde, E.
Freund, Appl. Catal. 22 (1986) 211.
[33] J. Peri, J. Phys. Chem. 86 (9) (1982) 1615.
[34] M. Pereira da Silva, R. Mello Viera, M. Schmal, Appl. Catal.
190 (2000) 177.
[35] A. Amirnazmi, J. Benson, M. Boudart, J. Catal. 30 (1973) 55.
[36] E. Winter, J. Catal. 22 (1971) 158.
[37] T. Yamashita, A. Vannice, J. Catal. 163 (1996) 158.
[38] B. Adelman, W. Sachtler, Appl. Catal. B 14 (1997) 1.
[39] H. Conrad, G. Ertl, J. Küppers, E. Latta, Surf. Sci. 65 (1977)
245.
[40] M. da Silva, B. Machado, in: Proceedings of the Acta Symp.
Iberoam. Catal., Brazil, 2000, p. 641.
[41] M. Iwamoto, H. Yahiro, Y. Mine, S. Kagawa, Chem. Lett.
(1989) 213.
[42] H. Hamada, Y. Kintaichi, M. Sasaki, T. Ito, Chem. Lett.
(1990) 1069.
[43] C. Frost, Nature 334 (1988) 577.
[44] A. Juan, D.E. Damiani, N.J. Castellani, Catal. Lett. 19 (1993)
43.
[21] S. Kawi, O. Alexeev, M. Shelef, B. Gates, J. Phys. Chem.
99 (1995) 6926.
[22] J. De Vries, H. Yao, R. Baird, H. Gandhi, J. Catal. 84 (1983)
8.
[45] A. Juan, C.C. Moro, D.E. Damiani, Colloids Surf. Part A:
Physicochem. Eng. Aspects 122 (1997) 257.
[46] R. Burch, P. Millington, A. Walker, Appl. Catal. B 4 (1994)
65.