Flame-synthesized LaCoO3-supported Pd. 2. Catalytic behavior in the reduction of NO by H2 under lean conditions

Article abstract of DOI:10.1016/j.jcat.2007.10.003

A 0.5 wt% Pd/LaCoO₃, prepared by flame-spray pyrolysis (FP), was tested as catalyst for the low-temperature selective reduction of NO by H₂ in the presence of excess O₂. In particular, the effect of the precalcination and prereduction temperature on catalytic activity was compared with that of a similar Pd/LaCoO₃ sample prepared by impregnation with a Pd solution of FP-prepared LaCoO₃. The FP-made catalyst allowed full NO conversion at 150 °C, with 78% selectivity to N₂, thus outperforming the catalytic behavior of the corresponding sample prepared by impregnation. The higher activity of the FP-made catalyst has been attributed to the formation of segregated Co metal particles, not present in the impregnated sample, formed during the precalcination at 800 °C, followed by reduction at 300 °C. Two reaction mechanisms can be deduced from the temperature-programmed experiments. The first of these, occurring at lower temperatures, indicates cooperation between the Pd and Co metal particles, with formation of active nitrates on cobalt, successively reduced by hydrogen spillover from Pd. The second, occurring at higher temperature, allows 50% conversion of NO, with >90% selectivity to N₂, and involves N adatoms formed by dissociative NO adsorption over Pd. Prereduction at 600 °C led to a slight increase in catalytic activity, due to the formation of a Pd{single bond}Co alloy, which is more stable on reoxidization compared with Pd alone. Moreover, the cooperative reaction mechanism seems to be favored by the proximity of Co and Pd in metal particles.

Full text of DOI:10.1016/j.jcat.2007.10.003

Journal of Catalysis 252 (2007) 137–147
www.elsevier.com/locate/jcat
Flame-synthesized LaCoO -supported Pd
3
2
. Catalytic behavior in the reduction of NO by H under lean conditions
2
a
b
b
a,∗
b
Gian Luca Chiarello , Davide Ferri , Jan-Dierk Grunwaldt , Lucio Forni , Alfons Baiker
a
Dipartimento di Chimica Fisica ed Elettrochimica, Università degli Studi di Milano, via C. Golgi 19, I-20133 Milano, Italy
b
Institute for Chemical and Bioengineering, Department of Chemistry and Applied Biosciences, ETH Zurich, Hönggerberg, HCI, 8093 Zurich, Switzerland
Received 18 July 2007; revised 25 September 2007; accepted 2 October 2007
Abstract
A 0.5 wt% Pd/LaCoO , prepared by flame-spray pyrolysis (FP), was tested as catalyst for the low-temperature selective reduction of NO by
3
H in the presence of excess O . In particular, the effect of the precalcination and prereduction temperature on catalytic activity was compared
2
2
with that of a similar Pd/LaCoO sample prepared by impregnation with a Pd solution of FP-prepared LaCoO . The FP-made catalyst allowed
3
3
◦
full NO conversion at 150 C, with 78% selectivity to N , thus outperforming the catalytic behavior of the corresponding sample prepared by
2
impregnation. The higher activity of the FP-made catalyst has been attributed to the formation of segregated Co metal particles, not present in
◦
◦
the impregnated sample, formed during the precalcination at 800 C, followed by reduction at 300 C. Two reaction mechanisms can be deduced
from the temperature-programmed experiments. The first of these, occurring at lower temperatures, indicates cooperation between the Pd and Co
metal particles, with formation of active nitrates on cobalt, successively reduced by hydrogen spillover from Pd. The second, occurring at higher
temperature, allows 50% conversion of NO, with >90% selectivity to N , and involves N adatoms formed by dissociative NO adsorption over Pd.
2
◦
Prereduction at 600 C led to a slight increase in catalytic activity, due to the formation of a Pd–Co alloy, which is more stable on reoxidization
compared with Pd alone. Moreover, the cooperative reaction mechanism seems to be favored by the proximity of Co and Pd in metal particles.
©
2007 Elsevier Inc. All rights reserved.
Keywords: Flame-spray pyrolysis; Lean conditions de-NOx; Pd/LaCoO perovskite; In situ fluorescence EXAFS
3
1
. Introduction
We considered the following reactions:
NO + 2H2 → N2 + 2H2O,
NO + H2 → N2O + H2O,
NO + O2 → 2NO2,
2
2
2
(1)
(2)
(3)
Hydrogen is an effective and clean alternative energy carrier.
However, the high temperature of hydrogen combustion can
lead to an even more abundant formation of NO than in nor-
x
mal fossil fuel combustion. Thus, the catalytic system needed
to remove NOx from the exhausts of lean-burn systems, using
H2 as a fuel, should be able to selectively reduce NOx in the
presence of excess oxygen, exploiting the traces of unburnt H2
present in the same exhaust. Moreover, H2 represents a valid
clean reducer alternative to NH3, which is widely accepted as
the reducing agent of choice for the selective catalytic reduction
and
2
H2 + O2 → 2H2O.
(4)
Reaction (1) is the desired one; the others are undesired side
reactions. In particular, reaction (4) competes with reaction (1),
decreasing NO conversion. The competition of reaction (1) with
reactions (2) and (3) decreases the selectivity to N2.
Over the last two decades, several catalyst formulations have
been proposed for selective NOx reduction by hydrogen (H2-
SCR) [2–10]. Among these, perovskite-like mixed oxides ap-
pear promising, due to their great versatility and redox proper-
ties. Doped perovskites [11,12] were found to be very active
(
SCR) of NO [1].
x
*
Corresponding author. Fax: +39 02 50314300.
E-mail address: lucio.forni@unimi.it (L. Forni).
0021-9517/$ – see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2007.10.003

138
G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
for the H2-SCR in the absence of O2. A new stimulus was
given by Costa et al. [13,14], who reported that Pt supported
by impregnation on a perovskite displayed higher NO conver-
sion and selectivity to N2 at low temperature in the NO/H2/O2
system compared with Pt/γ -Al2O3. Similarly, it has been re-
ported that the addition of water promoted reduction of NO over
Pd/LaCoO3 without affecting selectivity, whereas an inhibiting
effect was observed over Pd/Al2O3 [15]. In addition, Pd-doped
perovskites (e.g., LaFe0.95Pd0.05O3) seem to have a sort of self-
problem can be overcome by properly setting the FP operation
parameters, that is, by adjusting the organic solvent composi-
tion governing the flame temperature, taking into account the
possible noble metal cation mobility, leading to the surface seg-
regation of noble metal particles.
In an accompanying contribution [32], we reported the struc-
tural changes occurring in the FP-made catalyst on calcina-
tion and reduction, namely particle sintering, Pd and Co seg-
regation, formation of a Pd–Co alloy, and irreversible reduc-
tion of the support to La O + Co. The aim of the present
n+
regenerative character, in which Pd can move into (as Pd ) or
2
3
0
out of (as Pd ) the framework, depending on the redox fluctua-
work was to investigate the effect of these changes on the
tions of exhaust gases, thus preventing sintering of Pd particles
catalytic activity for H -SCR under lean-burn conditions in
the temperature range 50–450 C, by comparing a FP-prepared
2
◦
[
16–18].
Different mechanisms of reaction over noble metal-based
0.5 wt% Pd/LaCoO (PdLCO) with a 0.5 wt% Pd/LaCoO
3
3
catalysts have been proposed. Although all authors agree on
the dissociative adsorption of H2 on noble metals, different
interpretations regarding the interaction of NO and its succes-
sive reaction with H adatoms to give N2 (and N2O) and H2O
have been advanced. Burch et al. [19] proposed two routes of
N2 formation over Pt/SiO2, passing through chemisorbed N-
containing species on the surface of Pd. The latter can react ei-
ther with gaseous or physisorbed NO through an “impact” route
or with another identical N-containing species, each deliver-
ing one nitrogen atom to the final product. This mechanism has
been confirmed and implemented by Costa and Efstathiou [20],
who also compared it with the paths on Pt/La0.5Ce0.5MnO3,
thereby demonstrating the important role played by the support
(impPdLCO) prepared by the traditional Pd impregnation of a
FP-made LaCoO . In situ X-ray absorption near-edge structure
3
(XANES) and extended X-ray absorption fine structure (EX-
AFS) around the Pd K-edge were performed to investigate the
structural changes intervening in the catalyst during reaction
◦
between 50 and 500 C. Finally, transient experiments, namely
NO temperature-programmed desorption (NO-TPD) in flowing
He and temperature-programmed surface reaction (TPRS) in
flowing H /He or H /O /He, were performed to collect infor-
2
2
2
mation about the possible reaction mechanism.
2. Experimental
◦
chemical composition. It was found that at Tmax = 140 C, on
2.1. Catalyst preparation
Pt/SiO2 the active N species (particularly nitrosyls and uniden-
tate nitrates) was present exclusively on the noble metal particle
surface, whereas on Pt/La0.5Ce0.5MnO3, the two active inter-
mediate NOx species were nitrosyls and bidentate nitrates, both
present on the support surface. The reduction of these inter-
mediate species to N2 and N2O was due to hydrogen spillover
from metallic Pt on the support surface. The latter aspect was
also reported by Machida et al. over Pd/MnOx–CeO2 [6]. Ueda
et al. [4] observed two NO reduction conversion maxima at
The PdLCO catalyst was prepared by FP, as described in de-
tail in our accompanying study [32]. In brief, La(CH COO) ·
3
3
xH O (Aldrich, purity > 99.9%, 7.5 wt% H O, determined
2
2
by TGA), Co(CH COO) ·4H O (Merck, purum), and Pd(CH -
3
2
2
3
COO) (Fluka, purum) in the desired ratios and 0.15 M overall
2
concentration, referred to La or Co ions, were dissolved in a
mixture of propionic acid, n-propanol, and water (5:4:1 vol%).
The solution thus obtained was fed at a rate of 4 mL/min to
◦
◦
1
00 C and 300 C over Pd/TiO2, and suggested that at low
the nozzle of a vertical burner, along with 5 L/min of O . The
2
temperature, NO reacts directly with H2 over Pd, whereas at
high temperature, NO oxidizes to NO2, which is subsequently
reduced by H2. Finally, Qi et al. [9] observed the formation
spray was ignited and sustained by a premixed methane/oxygen
flame issuing from an annular ring surrounding the nozzle. The
pressure drop of oxygen across the nozzle was kept constant
at 1 bar, to ensure a sonic flow regime [24]. The produced
particles were collected on a glass fiber filter connected to a
vacuum pump. Two catalysts were prepared for comparison:
a FP-made LaCoO3 (LCO), synthesized using the same para-
meters of PdLCO and a 0.5 wt% Pd/LaCoO3 (impPdLCO),
prepared by the traditional impregnation method, by deposit-
ing Pd from a (NH4)2PdCl4 (Fluka, puriss) solution on LCO,
+
of an intermediate NH4 in the NO/H2/O2 reaction over Pd-
+
V2O5/TiO2–Al2O3, with NH4 considered responsible for the
higher activity of this catalyst with respect to a V2O5-free sam-
+
ple in which, in contrast, almost no NH4 formation occurs.
A plausible intermediate NO–NO dimeric species also has been
suggested [20–23], forming either onto the metal cation sites or
on oxygen vacancies.
◦
Flame pyrolysis (FP) is a low-cost, continuous single-step
method recently optimized by our group for perovskite prepa-
ration [24,25]. Whereas the FP preparation of noble metal on
single oxides has been demonstrated [26–28], some difficulties
have arisen in the direct synthesis of noble metals supported
on perovskite, due to the possible replacement of the noble
metal for the cation at the B site of the ABO3 perovskite frame-
work [16,29,30]. Indeed, FP is an effective technique for the
synthesis of doped perovskite-like mixed oxides [29,31]. This
preannealed at 800 C. The impregnated catalyst was then first
◦
◦
dried overnight at 90 C and then calcined at 500 C for 2 h in
flowing 10% O2/He (30 mL/min) gas mixture.
2.2. Catalytic activity
Catalytic activity tests for the selective reduction of NO by
H2 in the presence of excess O2 were carried out by a bench-
scale continuous reaction unit, including a U-shaped tubular

G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
139
quartz reactor (i.d. = 7 mm) heated by an electric furnace (Car-
bolite, model MTF 12/38A). The reactor temperature was mea-
sured by a thermocouple placed in a thin quartz tube beside the
catalyst bed. The catalyst (ca. 0.15 g, 0.20–0.30 mm particle
size, diluted 1:4 by weight with quartz powder of the same parti-
cle size) was loaded between two flocks of quartz wool. Before
in the flowing reaction mixture and repeating the temperature
ramp to measure the residual NO conversion. In addition, the
catalyst demonstrating the best performance was submitted to
three activation–deactivation cycles. The sample was reoxi-
2
dized by the reaction gas mixture (containing excess O ) when
◦
◦
ramping to 320 C, followed by reduction at 300 C in flowing
◦
each run, the catalyst was activated at 600 or 800 C for 2 h
10% H /He.
2
3
◦
in flowing 10% O2/He (20 cm /min), then cooled to 40 C in
the same gas flowing mixture. If not stated otherwise, a reduc-
tion in flowing 10% H2/He (20 mL/min) followed, while tem-
2
.3. Transient experiments
◦
◦
perature was increased to either 300 or 600 C at 10 C/min.
Temperature-programmed experiments were performed in
the same setup used for the catalytic activity tests. The fresh
catalyst bed (ca. 0.15 g, 0.20–0.30 mm particle size, diluted
The activity tests were carried out by feeding a mixture com-
posed of 0.25 vol% NO, 1 vol% H , 5 vol% O (He balance) at
2
2
1
00 mL/min overall flow rate, while increasing temperature up
1
:4 by weight with quartz powder of the same particle size)
◦
◦
to 450 C by 2 C/min.
3
was pretreated in 10% O2/He (20 cm /min) at either 600 or
8
H2/He (20 cm /min) at 300 C for 30 min, followed by reox-
idation at 160 C for 30 min in 20% O2/He. In this way, the
The reactor outlet was connected to an HP micro GC (model
M200H), equipped with PoraPLOT U and MS-5A PLOT
columns. The former was used for H2, O2, N2, and NO quan-
tification; the latter, for N2O. The composition of the gas phase
was monitored with an Omnistar Pfeiffer Vacuum GSD 30102
quadrupolar mass spectrometer (QMS) connected to the reactor
◦
00 C. The Pd-containing catalyst was prereduced in 10%
3
◦
◦
reduced support was reoxidized to LaCoO3, whereas the sur-
face Pd and Co particles retain their bulk metallic state [32].
This step was done to favor adsorption rather than reaction of
adsorbed NO with the reduced support. The outlet gas compo-
sition was monitored by QMS. The m/z mass fragments 2, 12,
◦
outlet through a stainless steel capillary tube kept at ca. 150 C.
The NO mol% conversion (XNO,T ), defined as the ratio of re-
acted over fed NO, and the percent selectivity to N2 (SN ,T ),
2
1
4, 18, 28, 30, 32, 44, and 46, corresponding to H2, C, N, H2O,
defined as the ratio of twice the moles of produced N2 over the
moles of reacted NO, at temperature T , were calculated by the
following equations:
N2, NO, O2, N2O, and NO2, were monitored. In particular, N2
and N2O were distinguished from CO and CO2 by the trend of
m/z = 12. Indeed, the fact that the latter signal remained un-
perturbed during the experiment indicates that no C-containing
species were formed or consumed.
ANO,0 − ANO,T
X
=
× 100
NO,T
ANO,0
and
The sequence of steps for the NO-TPD and TPSR experi-
ments is summarized in Table 1. In all cases, the overall gas flow
rate was 30 mL/min and the temperature ramp was 5 C/min.
2
αAN ,T
2
SN ,T =
× 100,
◦
2
CNO,0XNO,T
where ANO,0 is the NO chromatographic peak area, measured
on the feeding gas when bypassing the reactor; ANO,T and
2
.4. In situ fluorescence XANES and EXAFS during
NO/H2/O2 reaction
AN ,T are the NO and N2 chromatographic peak areas, respec-
2
tively, measured on the outlet gas at reaction temperature T .
CNO,0 is the NO concentration (= 0.25 vol% in this case) in the
feeding gas, and α is the slope of the titration curve of the chro-
XANES and EXAFS data were collected at the Norwegian–
Swiss beamline (SNBL, BM01B) of the European Synchrotron
Radiation Facility, ESRF, Grenoble, France. The spectra were
obtained using a specially designed in situ cell [33], loaded with
matographic peaks of N2. Thus, αAN ,T represents the vol%
2
concentration of N2 produced at temperature T .
0
5
.180 g of sample powder, pelletized, crushed, and sieved to
0–100 µm. For in situ oxidation, reduction, and reaction, 21%
The effect of pretreatment conditions on catalytic activity
was investigated on samples labeled (Pd)LCOc6(or 8)r3(or 6),
depending on the presence or absence of Pd, with “c” and “r”
representing “precalcined” and “prereduced,” respectively, and
O2/He, 5% H2/He, and 0.25% NO/1% H2/5% O2/He (ramp
◦
2
C/min), respectively, were fed to the reactor cell. The evolv-
◦
◦
◦
ing products during TPR/TPO were monitored by a Balzers
Thermostar QMS instrument.
3
(= 300 C), 6 (= 600 C), and 8 (= 800 C) indicating the
temperature of the treatment. For example, PdLCOc8r3 rep-
resents 0.5 wt% Pd/LaCoO calcined at 800 C, followed by
reduction at 300 C.
◦
The spectra around the Pd K-edge (24.350 keV) were
recorded in fluorescence mode using a 13-element Ge-solid
state detector (Canberra). For this purpose, the fluorescence sig-
nals from the 13 detector elements were preamplified and then
fed into the XIA digital pulse processing electronics (DXP 2X).
The Pd Kα-fluorescence regions were selected and calibrated
on the basis of a full spectrum for each channel, collected be-
fore the measured scans. To minimize the fluorescence from the
matrix, aluminum and chromium filters were added. The result-
ing SCA data were added up manually after each measurement,
3
◦
Catalyst stability versus time on stream was measured at the
temperature of maximum NO conversion (Tmax) for 100 h under
◦
flowing reaction mixture, increasing T from 50 C to Tmax at a
◦
rate of 2 C/min. GC analysis was done every hour for the first
1
6
2 h and then every 2 h from 12 to 26 h, every 4 h from 26 to
0 h, and every 10 h from 60 to 100 h.
Thermal resistance of the catalyst was measured by carrying
◦
out the reaction on samples prereduced at 300 C, then cooled

140
G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
Table 1
Sequence of steps of the different temperature-programmed experiments
Experiment
Sequence of steps
◦
◦
NO-TPD
TPSR-1
TPSR-2
TPSR-3
TPSR-4
2.5% NO/He (30 C; 30 min) → He (30 C; 30 min) → TPD in He
◦
◦
2.5% NO/He (30 C; 30 min) → He (30 C; 30 min) → TPSR in 4% H /10% O /He
2
2
◦
◦
◦
1% NO/20% O /He (150 C; 30 min) → cooling in the same flowing gas mixture to 30 C → He (30 C; 30 min) → TPSR in 4% H /10% O /He
2
2
2
◦
◦
◦
1% NO/20% O /He (150 C; 30 min) → cooling in the same flowing gas mixture to 30 C → He (30 C; 30 min) → TPSR in 10% H /He
2
2
◦
◦
On the sample after the TPSR2 → cooling in 4% H /10% O /He to 30 C → He (30 C; 30 min) → TPSR in 10% H /He
2
2
2
◦
Note. Common conditions: 0.150 g of catalyst; 30 mL/min total gas flow rate; 5 C/min temperature ramp.
and the spectra were background-corrected and normalized us-
ing WINXAS 3.1 software [34]. Data fitting was performed
3
in the R space on the Fourier-transformed k -weighted EX-
AFS functions (Fourier transform in the range of 3.5–14 Å,
2
0
S = 0.7). Theoretical scattering amplitudes and phase shifts
of the Pd–Pd shell and the Pd–O shell (only first shell) were
calculated using the FEFF code [35].
3
3
3
. Results
.1. Catalytic activity
.1.1. Effect of calcination temperature
The plots of NO conversion (XNO), selectivity to N (SN2 ),
2
and H2 conversion (XH ) versus temperature over Pd-modified
2
and -unmodified FP-LaCoO3, are reported in Fig. 1. LCOc6
showed only oxidative activity of NO to NO2 (reaction 3) in the
◦
◦
1
60–450 C range, with a maximum XNO of 86% at 310 C.
Preannealing at higher temperature (sample LCOc8) provoked
a partial catalyst deactivation, with a maximum X
at 380 C. However, a small reductive activity to N2 (reac-
of 53%
NO
◦
tion (1)) was detected in this range, with a maximum SN2 of
◦
8
% at 290 C (XNO = 15.4%). In contrast, the Pd-containing
◦
perovskite reduced at 300 C displayed two XNO maxima, one
◦
at 100–250 C, due to the reduction of NO to N2 and N2O (re-
actions (1) and (2)), and the second one at higher temperature,
due to the oxidation activity of the perovskite support. In par-
ticular, PdLCOc6r3 was less active for the reduction of NO to
◦
N (T
(
= 210 C, X = 67%, S = 62%) than PdLCOc8r3
Fig. 1. Effect of precalcination temperature on catalytic activity for the
NO/H2/O2 reaction of LCO (triangles) and PdLCO (diamonds). Calcination
2
max
NO
N2
◦
Tmax = 150 C, XNO = 100%, SN = 67%). Thus, a higher cal-
2
◦
◦
temperature: 600 C (empty dots) and 800 C (full dots). The Pd-containing
cination temperature had a beneficial effect on the reduction
activity but a detrimental effect on the oxidation activity. Both
◦
catalysts were reduced at 300 C before the catalytic test. Reaction conditions:
◦
0
.15 g of catalyst; 0.25% NO/1% H /5% O /He (100 mL/min); 2 C/min tem-
2
2
the PdLCOc6r3 and PdLCOc8r3 catalysts attained the highest
perature ramp.
◦
S
between 200 and 260 C, just before the start of oxidation
N2
activity to NO2, whereas H2 (XH ) and NO (XNO) conversions
2
◦
started at the same temperature. Except for LCOc8, in all other
a wider temperature range (140–180 C, SN = 77%). Fur-
2
◦
cases XH2 attained 100% at T > 350 C, and PdLCOc8r3 al-
thermore, in both cases, the FP-made PdLCO had superior
catalytic activity compared with the same catalyst prepared
by impregnation (impPdLCOc5r3). Indeed, the latter never
reached full NO conversion under the reaction conditions used
◦
lowed full conversion of H2 already at 150–210 C.
3
.1.2. Effect of reduction temperature
◦
Fig. 2 shows that the FP-made PdLCO, preannealed at
here (Tmax = 180 C, XNO = 78%, SN = 67%). In contrast,
2
◦
◦
8
00 C, showed a first XNO maximum (21%) at 250 C
the impregnated catalyst was the most active for H2 conversion
◦
(
SN = 86%). This activity was not observed for LCOc8 and is
(XH ) in the 200–450 C temperature range. Indeed, whereas
impPLCOc5r3 maintained a 100% XH , XH decreased to 90%
for PdLCOc8r3 and to 56% for PdLCOc8r6. Because XNO was
always lower with impPLCOc5r3 than with the prereduced FP-
made catalyst throughout the entire temperature range, we can
2
2
similar to that reported by Engelmann-Pirez et al. [15]. Reduc-
tion at 300 C in flowing 10% H /He led to improved NO reduc-
2
2
◦
2
tion according to reaction (1) (vide supra), whereas the catalyst
◦
reduced at 600 C (PdLCOc8r6) displayed X = 100% for
NO

G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
141
Fig. 3. Catalyst resistance vs time-on-stream at the temperature of max NO
◦
◦
conversion: 160 C (squares) for PdLCOc8r3; 180 C (triangles) for impPdL-
COc5r3. Full dots: NO% conversion, empty dots: selectivity to N .
2
Fig. 2. Effect of prereduction temperature on catalytic activity for the
◦
NO/H /O reaction of PdLCO (precalcined at 800 C). (×) PdLCOc8 (not pre-
2
2
◦
reduced); (Q) PdLCOc8r3 (reduced at 300 C); (P) PdLCOc8r6 (reduced at
◦ ◦
00 C); (!) impPdLCOc5r3 (perovskite support precalcined at 800 C before
6
Pd impregnation). Reaction conditions: 0.15 g of catalyst; 0.25% NO/1% H /
2
◦
5
% O /He (100 mL/min); 2 C/min temperature ramp.
2
Fig. 4. Catalytic activity of PdLCOc8r3 after three activation cycles in flowing
10% H /He and deactivation in reaction gas flowing mixture (NO/H /O /He)
up to 320 C. (") NO% conversion after the third reactivation cycle; (Q) resid-
conclude that with impPLCOc5r3, hydrogen reacted preferably
with O2 to form water, rather than with NO to form N2, thus
decreasing the catalytic activity for NO reduction.
2
2
2
◦
ual NO% conversion after the third thermal deactivation; (P) selectivity to N
2
after the third deactivation cycle.
3
.1.3. Time on stream and thermal resistance of catalyst
The stability of PdLCOc8r3 (Fig. 3) and PdLCOc8r6 (not
idation activity (reactions (3) and (4)) became predominant.
To check whether such a drop in catalytic activity was due
to a thermal phenomenon or to catalyst deactivation, a ther-
mal resistance test was performed on PdLCOc8r3 (Fig. 4), as
reported in Section 2. This test simulates a possible catalyst
shown) was tested under steady-state reaction conditions at
◦
1
60 C (Tmax). Both catalysts displayed full NO conversion af-
ter 100 h on stream. Furthermore, while remaining constant at
8% for PdLCOc8r6, SN2 increased from 68% to 78% for PdL-
7
◦
COc8r3 during the first 10 h and then attained the performance
of PdLCOc8r6. In addition, the conversions of H and O were
overheating from the Tmax of 160 C. The data show that the
2
2
catalyst underwent a partial deactivation, with the first maxi-
◦
stable at 100 and 10%, respectively. In contrast, in the case of
mum shifted to 240 C (X = 48%, S = 88%). A similar
NO
N2
impPdLCOc5r3 (Fig. 3), XNO decreased from 78 to 55% at
result was obtained when PdLCOc8r3 was overheated up to
500 C; however, the catalyst was fully reactivated in flowing
10% H /He at 300 C, even after several cycles of thermal de-
◦
◦
1
80 C (Tmax) during the first 10 h and then stabilized at this
◦
value for the remaining 90 h on stream. The trends for XH2 ,
XO , and SN were similar to those exhibited by PdLCOc8r3.
2
activation. Fig. 4 shows the activity of PdLCOc8r3 after three
cycles of reactivation and the residual NO conversion after the
2
2
Fig. 2 shows that either PdLCOc8r3 or PdLCOc8r6 exhib-
◦
◦
ited a saddle in the XNO in the 200–320 C range, where the
subsequent thermal deactivation at 320 C under a flowing re-
reduction activity (reactions (1) and (2)) dropped and the ox-
action mixture.

142
G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
◦
◦
◦
Fig. 5. TPD-NO spectra of NO preadsorbed at 30 C and species released in the 50–450 C range with heating rate = 5 C/min.
◦
◦
3
.2. TPD of NO
ima at 306 C and at 350 C, due to decomposition of surface
nitrites (M–O–N=O → M + NO + ½O2) and mono or biden-
Fig. 5 shows the thermal desorption patterns of some se-
tate nitrates (M–O–NO [or M–O–(NO)–O–M] → M + NO +
2
◦
◦
lected samples at 50–450 C after NO chemisorption at 30 C
followed by TPD in flowing He with a temperature ramp rate
O ) [38,39], formed by interaction with surface oxygen anions
2
(α-oxygen). Indeed, in the last window, O desorption also was
2
◦
of 5 C/min. For LCOc8, PdLCOc8, and impPdLCOc5, three
detected, along with NO. Only nitrite structures have been asso-
◦
3+
NO desorption windows can be recognized: at 50–100 C, due
to desorption of physisorbed NO; at 125–250 C, due to des-
ciated with La ions [38], whereas nitrosyl and nitrate species
◦
typically form on surface Co or Pd-containing species [36,37].
Some important differences arise from the comparison of the
thermal desorption patterns of the three samples. In particular,
orption of nitrosyls chemisorbed on surface metallic Co and Pd
◦
atoms or cations [36,37]; and at 225–425 C, with two max-

G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
143
the Pd-containing catalysts, PdLCOc8 and impPdLCOc5, ex-
◦
hibited a more intense NO desorption in the 125–225 C range
(
◦
Tmax = 185 C). The origin of this NO evolution can be reason-
ably attributed to nitrosyls adsorbed on segregated Co3O4 and
PdO. Furthermore, impPdLCOc5 displayed more pronounced
◦
NO + O2 evolution at 270 C, likely due to decomposition of
nitrates formed on the more abundant PdO surface particles
present on the impregnated catalyst.
With the Pd-containing catalysts, noticeable changes in the
◦
NO-TPD patterns occurred after reduction at 300 C in flow-
ing 10% H2/He, followed by bulk reoxidation of the support
◦
to the perovskite at 160 C in flowing 20% O2/He. Both PdL-
COc8r3 and impPdLCOc5r3 exhibited a single NO desorption
◦
◦
peak at 125–225 C (Tmax = 185 C for impPdLCOc5r3 and
◦
2
00 C for PdLCOc8r3), followed by a weak evolution of N2
◦
and N2O at around 225 C. The formation of N2 and N2O is
due to the coupling of two N adatoms, produced by dissociative
NO adsorption, and to the reaction of an N adatom with another
NO molecule, respectively. Similar patterns were reported by de
Wolf and Nieuwenhuys [36] for NO adsorbed at 27 C over a
Pd(111) single-crystal surface. The peak at 200 C can be at-
◦
◦
tributed to desorption of nitrosyls from Pd particles, although a
contribution of nitrosyls desorbed from the perovskite surface
cannot be excluded. Because neither NO nor O2 evolution was
detected in the temperature range of nitrite and nitrate decom-
◦
position, we can conclude that the reoxidation at 160 C was not
sufficient to restore the surface oxygen anion sites (α-oxygen)
of PdLCOc8, over which gaseous NO interacts to form the sur-
face nitrates and nitrites. However, the formation of other more
strongly adsorbed NO species, neither desorbing nor decom-
posing within the temperature range considered here, cannot be
excluded.
Moreover, the NO peak area, proportional to the amount of
desorbed NO, obtained with impPdLCOc5r3, was 10% smaller
than that obtained with PdLCOc8r3. This result could be cor-
related to a higher Pd dispersion in the FP-made catalyst com-
pared with the impregnated catalyst.
Finally, in the NO-TPD patterns of PdLCOc6r3, along with
◦
the desorption peak at 200 C, also observed on PdLCOc8r3,
◦
a second contribution appeared at 265 C. This second peak
might be attributed to NO adsorbed on the support surface. The
latter could be either lacking after calcination at 800 C, mainly
Fig. 6. QMS analysis of the species evolved during the TPSR-1 (a) and TPSR-2
(b) experiments over PdLCOc8r3 and TPSR2 (c) over impPdLCOc5r3.
◦
due to particle sintering, or desorbing at lower temperature, thus
overlapping to the NO desorbed from Pd. Moreover, the amount
of desorbed NO was ca. 4 times greater than that desorbed from
PdLCOc8r3, clearly due to the higher specific surface area of
shown in Fig. 6a. The main evolved species was again NO
◦
at around 200 C, whereas a weak evolution of N and N O
2
2
◦
◦
occurred at 100–200 C. Thus, NO adsorbed at 30 C on Pd
contributed only slightly to the reaction. Different and more
interesting results were achieved when NO was co-adsorbed
◦
2
the sample calcined at 600 C (20 vs 8 m /g).
.3. TPSR
In the previous section, we found that NO adsorbed at 30 C
◦
3
with O at 150 C (TPSR-2; Fig. 6b). Indeed, in this case, the
2
more intense signal was that of N2, evolved mainly in two dif-
◦
◦
◦
ferent zones: 100–180 C, with a maximum at 160 C, which
on surface Pd particles on PdLCOc8r3 and desorbed under He
flow at around 200 C. To establish the role of this NO species
also corresponds to the temperature of maximum X (Fig. 2),
NO
◦
◦
◦
and at 180–300 C, with a maximum at 210 C. In contrast,
N O showed a QMS signal intensity lower than that of N and
in the catalytic activity and thus in the reaction mechanism,
we tested its stability under a H2/O2/He flow while increas-
2
2
evolved only in the former temperature range. These results are
◦
◦
ing the temperature to 300 C (TPSR-1; Table 1). The QMS
in line with the experimental SN2 trend at 100–300 C observed
◦
signal trends, similar to those obtained on impPLCOc5r3, are
on PdLCOc8r3 (Fig. 2). Indeed, at 180–300 C, where no evo-

144
G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
lution of N2O was detected, SN2 increased to 90%. Finally,
Fig. 7b shows the results of a TPSR experiment in 10%
H2/He (TPSR-4) on PdLCOc8r3 after TPSR-2. In this case, the
reduction of the residual N-containing species was obtained.
But the latter do not participate in the reaction, remaining as
spectator species [19,20] accumulating onto the catalyst sur-
face, stabilized by the presence of oxygen. The only remaining
the signal corresponding to NO displayed a peak centered at
◦
2
00 C. Thus no relevant change in the NO desorption curve
was noticed with respect to the TPSR-1 experiment.
The TPSR-2 experiment conducted on impPdLCOc5r3 is re-
ported in Fig. 6c. The amount of evolved N was clearly lower
2
◦
contribution was that at 180 C. Note that no ammonia was de-
than that in PdLCOc8r3, and, in particular, the contribution at
◦
tected in this case. Finally, the uptake of H exhibited in these
1
60 C was completely absent. This result represents a signif-
2
last two experiments was due mainly to the reduction of the
catalyst itself [32], particularly of LaCoO3 to La3Co3O8 (steps
icant difference with respect to PdLCOc8r3 and suggests the
presence of a greater amount of active species, and thus active
sites, which would explain the better catalytic performance of
the FP-made catalyst.
A third TPSR experiment (TPSR-3) was conducted to inves-
tigate the stability of the N-containing species produced after
1
a and 1b in Fig. 7) and of La3Co3O8 to La2Co2O5 (steps 2a
and 3b). Furthermore, step 2b, which was absent during TPSR-
3
, was due to the reduction of cobalt oxide particles. Indeed,
◦
the reduction process at 300 C produced metallic Co particles,
stable in oxygen-rich atmosphere at 160 C [32]. At higher tem-
◦
◦
adsorption of NO + O at 150 C under a 10% H /He flow,
2
2
perature, they reoxidized, as occurred after TPSR-2.
that is, in an oxygen-free atmosphere. The results are reported
in Fig. 7a. The area under the N2 signal was three times larger
than that obtained in TPSR-2. Thus, in the absence of O2, more
N2 evolved from the catalyst. Moreover, no N2O was detected,
3
.4. In situ XANES and EXAFS during the NO/H /O reaction
2 2
In situ XANES measurements on PdLCOc8r3 during the
whereas the formation of ammonia cannot be excluded at 100–
◦
NO/H /O reaction (Fig. 8a) showed that at 160 C [i.e., the
◦
2
2
1
50 C. In fact, the m/z = 17 signal corresponds to the sum
temperature of maximum NO conversion (Fig. 2)], Pd was still
prevalent in the reduced state, whereas almost full oxidation
−
of both the parent NH3 and the OH fragment of water, which
amounts to the 23% of the main m/z = 18 signal. The weighted
subtraction of the 18- and 17-m/z signals should give, as a
residual, the contribution due to ammonia only; the result was a
◦
occurred at 500 C. In particular, the trend of the fraction of
2+
Pd at a given temperature (Fig. 8b), obtained by linear com-
bination analysis (LCA) of the XANES spectra recorded at that
temperature with those of PdLCOc8 (fully oxidized Pd) and of
PdLCOc8r3 (fully reduced Pd), revealed that the reoxidation
◦
peak centered at 120 C. Fig. 7a also shows the deconvolution
of the N2 signal, which led to four contributions at 120, 145,
◦
1
70, and 210 C.
◦
of Pd occurred in three steps at 25–500 C. Under the reaction
◦
mixture, Pd was stable up to 150 C, followed by mild oxidation
ca. 20%) at 160–200 C, due in part to surface reoxidation and
◦
(
reoxidation of the smallest Pd particles. No further oxidation
◦
occurred in the 200–300 C range, whereas rapid reoxidation
◦
was recorded from 300 C on. In contrast, PdLCOc8r6 dis-
0
played different behavior, because Pd was found to be more
stable under the reaction gas mixture. Indeed, LCA showed that
◦
reoxidation occurred only over 300 C. Moreover, in the 25–
2
◦
70 C range, LCA did not fit the experimental data when the
XANES spectrum of PdLCOc8r3 was taken as reference for
0
Pd , whereas the spectrum of PdLCOc8r6 did fit, due to the
presence of the Pd–Co alloy. The opposite occurred with LCA
◦
of the spectra recorded above 270 C, where fitting was pos-
sible using the spectrum of PdLCOc8r3; therefore, the Pd–Co
◦
alloy was stable up to 270 C.
The FT-EXAFS spectra of PdLCOc8r3 during reaction at
◦
1
60 and 500 C are presented in Fig. 9. In the former case,
although a weak contribution at 1.6 Å was observed due to
the oxygen first-neighbor shell, indicating a partial oxidation,
the main contribution was still at 2.5 Å, typical of Pd–Pd dis-
tance of metallic Pd (compare with FT-EXAFS of Pd foil in
Fig. 9). Hardly any contribution was found at the Pd–La dis-
tance (3.2 Å), indicating that no Pd incorporation in the per-
◦
ovskite framework occurred at 160 C. On the other hand, the
spectrum of PdLCOc8r3 (Fig. 9b) measured during reaction
◦
at 500 C demonstrated clear reoxidation of the sample with
Fig. 7. QMS analysis of the species evolved during the TPSR-3 (a) and TPSR-4
b) experiments according to Table 1.
an increase of the feature at 1.7 Å (Pd–O distance). Peak de-
convolution in the 2–4 Å range showed the presence of three
(

G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
145
Fig. 9. Radial distribution function (FT-EXAFS spectra at the Pd K-edge)
around Pd of some selected samples. PdLCOc8r3 during NO/H /O reaction at
2
2
◦
◦
1
(
60 C (a) and after 500 C (b). Interatomic distances: (×) Pd–O; (+) Pd–Co;
0
") Pd–Pd in Pd particles; (F) Pd–Pd in PdO particles; (e) Pd–La.
Pd oxidation state, and formation of the Pd–Co alloy [32]. The
highest catalytic activity in H2-SCR was attained after precalci-
◦
nation at 800 C, followed by reduction, over both PdLCOc8r3
and PdLCOc8r6. Furthermore, the FP-made catalyst proved
superior to the same material created by the conventional im-
pregnation method (impPdLCOc5r3), in terms of conversion
of NO (XNO; Fig. 2) and lifetime (Fig. 3). Momentarily ne-
glecting PdLCOc8r6, which has a completely different catalyst
formulation (Pd–Co alloy on CoOx–La2O3), what really differ-
entiates PdLCOc8r3 from PdLCOc6r3 and impPdLCOc5r3 is
the presence of cobalt metal nanoparticles at the catalyst sur-
face. Pd particle size seems not to be a crucial property in the
NO/H2/O2 reaction over this catalyst. Indeed, PdLCOc6r3 dis-
played lower catalytic performance, with the maximum XNO
lower and shifted to higher temperature, despite the fact that
it had Pd particles smaller than impPdLCOc5r3 but similar to
PdLCOc8r3 [32]. This might indicate that dissociative hydro-
gen adsorption over Pd is not the rate-determining step.
Fig. 8. (a) In situ fluorescence XANES at the Pd K-edge during NO/H /O re-
action over PdLCOc8r3 at different temperatures. (b) Fraction of Pd obtained
2
2
2+
by linear combination analysis of the in situ fluorescence XANES spectra dur-
◦
ing reduction of PdLCOc8 in 10% H /He up to 350 C (1); NO/H /O reaction
2
2
2
over PdLCOc8r3 (P) and over PdLCOc8r6 (×).
contributions centered at 2.5, 2.9, and 3.3 Å, which were as-
cribed to the Pd–Pd distance in metallic Pd, the Pd–Pd distance
in PdO, and the Pd–La distance of Pd placed at the B site of the
ABO3 perovskite framework, respectively. Therefore, after ex-
The TPSR-2 experiment (Fig. 6b) clearly showed that the
N-containing active species form on NO and O2 co-adsorption
◦
◦
at Tmax = 160 C, whereas the TPSR-1 experiment (Fig. 6a)
posure at 500 C under the reaction gas mixture, only a small
◦
showed that NO preadsorbed at 30 C on Pd reacted only
fraction of Pd was still reduced (ca. 10%, Fig. 8b), whereas the
remainder was oxidized, in part as surface PdO and in part as
slightly with H2 in the presence of O2. Moreover, although
the TPSR-1 results collected on PdLCOc8r3 and on impPdL-
COc5r3 were similar, involving the same NO adsorbed species
on Pd, the TPSR-2 data were considerably different (Fig. 6).
According to Costa and Efstathiou [20], two structurally dif-
ferent and adjacent active N-species lie on the support, namely
bidentate nitrates and nitrosyls, that can be efficiently reduced
to N2 and water by spillover of hydrogen dissociatively ad-
sorbed on Pd particles [6]. Because nitrates do not form on
2+
Pd dissolved into the perovskite. This demonstrates that Pd
can segregate on the surface during the reduction process and
move back into the perovskite bulk when exposed to oxygen-
rich atmosphere at relatively high temperature. Similar behavior
was reported for other Pd-doped perovskites [16].
4
. Discussion
3+
2+
The present results show how the temperature of precalci-
La [38], although NO/O2 co-adsorption over Co sites is
known to lead to the formation of surface nitrates [40], we
can conclude that one of the two active N-species (the biden-
tate nitrates) forms over Co sites. Therefore, we suggest that
the enhanced catalytic activity of PdLCOc8r3 compared with
nation and successive reduction decisively affect the catalytic
performance of the FP-made catalyst, as a consequence of
the different structural modifications during these processes,
namely sintering, Co O and Pd segregation, a change in the
3
4

146
G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
Fig. 10. Simplified scheme of the cooperative reaction mechanism occurring in
◦
the 125–175 C temperature range over PdLCOc8r3.
impPdLCOc5r3 is due to the surface Co metal particles act-
ing as more efficient NO/O2 co-adsorption sites compared with
the bare perovskite surface. This hypothesis is supported by the
TPSR-2 experiment (Fig. 6), which showed a greater amount of
active species present on PdLCOc8r3 than on impPdLCOc5r3,
where no surface Co particles were seen (vide supra). There-
◦
fore, the reaction occurring at 160 C on PdLCOc8r3 could
follow a cooperative mechanism, going through the oxidation
2+
to Co of Co metal particle surface by co-adsorption of NO
and O2 to give cobalt nitrates. The latter then readily react with
nitrosyls and subsequently reduce to N2, N2O, and water by
the hydrogen spillover from Pd, thus restoring the Co metal site
(
scheme of Fig. 10). Further work is now in progress, through
in situ DRIFT analysis, to better elucidate this aspect. Fig. 11
presents a STEM image of a possible active site on PdLCOc8r3,
showing the proximity of Co and Pd particles on LaCoO . As
3
mentioned in the accompanying paper [32], surface Co particles
have not been detected by STEM analysis on either impPdL-
COc5r3 or PdLCOc6r3.
When the temperature was increased to 300 C, the Co0
particles were irreversibly oxidized to Co3O4, thus suppress-
ing the mechanistic route proposed here. The oxidation of
◦
CoO to Co O also has been found to be detrimental for the
3
4
NO/C3H6/O2 reaction over CoOx/ZrO2 [37]. Moreover, in situ
XANES measurements demonstrated that ca. 20% of Pd reox-
idized at 160–300 C (Fig. 8b). These two aspects are in line
Fig. 11. HAADF-STEM images and EDX spectra of a possible reaction site
of PdLCOc8r3, showing the proximity of surface Pd (area 1) and Co (area 2)
particles on LaCoO3 (area 3). The Cu peak is caused by the sample-supporting
grid.
◦
with the partial catalyst deactivation observed after heating at
◦
3
00 C under the reaction gas mixture (thermal resistance tests
◦
of Fig. 4). Successive reduction at 300 C reactivated the cata-
lyst by restoring the Co and Pd metal particles. The residual NO
reduction activity at 180–280 C seems to be the same as that
intimate contact between Pd and Co atoms. As we report in
the accompanying paper [32], PdLCOc8r6 displayed the high-
est coordination number (11.8 ± 1, calculated by EXAFS data
fitting; Table 2), indicating that the Pd-containing particles were
remarkably large. But also in this case, this parameter seemed
to not affect the catalytic performance; on the other hand, larger
particles are less prone to reoxidation. Although the alloy was
◦
found over PdLCOc8 (Fig. 2) and can be reasonably attributed
to a different reaction mechanism. The latter implies the dis-
sociative NO adsorption over Pd particles, forming N adatoms,
which can either couple to form N or react with another NO
2
molecule to form N2O (vide supra). Oxygen adatoms are re-
moved by reaction with H adatoms to form water.
◦
◦
stable up to 300 C, XNO started to decrease already at 180 C,
◦
PdLCOc8r6 proved to be the most active catalyst, display-
ing full NO conversion in a wider temperature range (140–
whereas XH remained unchanged at 100% up to 250 C, and
2
XO increased slightly (from 10 to 12%). This suggests that
2
◦
◦
◦
1
80 C), with greater selectivity to N at 160 C than for the
over 180 C, the direct reaction of H2 with O2 to give H2O be-
2
other catalysts (77% vs 67%). The improved performance can
be attributed to formation of the Pd–Co alloy, for two rea-
sons: The alloy is more stable to reoxidation than separated
Pd and Co metal particles, as revealed by XANES analysis
came predominant, whereas the slight increase of XO was due
to oxidation of CoOx to Co3O4.
2
Finally, a significant result came from in situ EXAFS analy-
◦
sis on PdLCOc8r3 after reaction at 500 C (Fig. 9), demon-
(
(
Fig. 8a), and the proposed cooperative reaction mechanism
Fig. 10) can occur more efficiently on the alloy due to the
strating the reoxidation and thus the possible dissolution of Pd
within the perovskite framework, typical of the so called “intel-

G.L. Chiarello et al. / Journal of Catalysis 252 (2007) 137–147
147
Table 2
References
Fitting of the EXAFS data at the Pd K-edge
2
[1] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B Environ. 18 (1998)
1.
Entry Sample
Shell
CN
R
σ
ꢀE
Residual
Å) (Å ) (eV) error
%)
0
2
(
[
[
[
[
[
[
[
2] K. Yokota, M. Fukui, T. Tanaka, Appl. Surf. Sci. 121 (1997) 273.
3] B. Frank, G. Emig, A. Renken, Appl. Catal. B Environ. 19 (1998) 45.
4] A. Ueda, T. Nakao, M. Azuma, T. Kobayashi, Catal. Today 45 (1998) 135.
5] R. Burch, M.D. Coleman, Appl. Catal. B Environ. 23 (1999) 115.
6] M. Machida, D. Kurogi, T. Kijima, J. Phys. Chem. B 107 (2003) 196.
7] M. Machida, S. Ikeda, J. Catal. 227 (2004) 53.
(
1
2
3
PdLCOc8
PdLCOc8r3
PdLCOc8r3, during Pd–O
Pd–O
6.3 2.01 0.003 11.2
Pd–Pd 6.9 2.73 0.006
3.5 2.01 0.002
Pd–Pd 5.0 2.73 0.007
Pd–O 5.0 2.01 0.002
8.9
2.8
6.0
6.0
5.0
1.8
9.8
2.5
9.6
◦
reaction at 160 C
PdLCOc8r3, after
reaction at 500 C
4
5
8] T. Nanba, C. Kohno, S. Masukawa, J. Uchisawa, N. Nakayama, A. Obuchi,
Appl. Catal. B Environ. 46 (2003) 353.
◦
PdLCOc8r6
Pd–Pd 7.5 2.70 0.007
Pd–Co 4.3 2.70 0.007
2.3
6.3
3.7
3.7
[9] G.S. Qi, R.T. Yang, F.C. Rinaldi, J. Catal. 237 (2006) 381.
[
10] C.N. Costa, A.M. Efstathiou, Environ. Chem. Lett. 2 (2004) 55.
[11] A. Lindstedt, D. Stromberg, M.A. Milh, Appl. Catal. A Gen. 116 (1994)
09.
2
Note. CN: coordination number; R: interatomic distance; σ : Debye Waller
1
factor; ꢀE : energy shift.
0
[
[
[
12] D. Ferri, L. Forni, M.A.P. Dekkers, B.E. Nieuwenhuys, Appl. Catal. B
Environ. 16 (1998) 339.
13] C.N. Costa, V.N. Stathopoulos, V.C. Belessi, A.M. Efstathiou, J. Catal.
ligent” catalysts [16]. This is indeed a welcome phenomenon
for exhaust gas after-treatment catalysts, because it prevents
sintering of Pd particles. (Pd dissolves into the perovskite in-
stead of sintering.) Following a cyclic segregation as small sur-
face clusters during further reduction and redissolution during
reoxidation, this behavior confers a longer life to the catalyst
compared with the traditional formulation based on Pd sup-
ported on silica or alumina. It is worth keeping in mind that
197 (2001) 350.
14] C.N. Costa, P.G. Savva, C. Andronikou, P.S. Lambrou, K. Poly-
chronopoulou, V.C. Belessi, V.N. Stathopoulos, P.J. Pomonis, A.M. Efs-
tathiou, J. Catal. 209 (2002) 456.
[
15] M. Engelmann-Pirez, P. Granger, G. Leclercq, Catal. Today 107–108
(2005) 315.
[
16] Y. Nishihata, J. Mizuki, T. Akao, H. Tanaka, M. Uenishi, M. Kimura, T.
Okamoto, N. Hamada, Nature 418 (2002) 164.
[17] Y. Nishihata, J. Mizuki, H. Tanaka, M. Uenishi, M. Kimura, J. Phys.
Chem. Sol. 66 (2005) 274.
[18] M. Uenishi, M. Taniguchi, H. Tanaka, M. Kimura, Y. Nishihata, J. Mizuki,
T. Kobayashi, Appl. Catal. B Environ. 57 (2005) 267.
19] R. Burch, A.A. Shestov, J.A. Sullivan, J. Catal. 188 (1999) 69.
in the accompanying paper [32], we showed that after reduc-
tion at 300 C of PdLCOc8, the support reduced to La Co O ,
but fully reoxidized to LaCoO3 already at 150 C in an oxygen-
◦
2
2
5
◦
[
◦
containing atmosphere. Thus PdLCOc8r3 at Tmax = 160 C in
[20] C.N. Costa, A.M. Efstathiou, J. Phys. Chem. B 108 (2004) 2620.
21] A. Snis, I. Panas, Surf. Sci. 413 (1998) 477.
[
[
[
[
the O2-rich reaction mixture actually consists of Pd supported
on LaCoO3. Therefore, despite exhibiting a slightly lower low-
temperature activity than PdLCOc8r6, the catalyst prereduced
22] F. Acke, M. Skoglundh, J. Phys. Chem. B 103 (1999) 972.
23] F. Acke, I. Panas, J. Phys. Chem. B 103 (1999) 2195.
24] G.L. Chiarello, I. Rossetti, L. Forni, J. Catal. 236 (2005) 251.
[25] G.L. Chiarello, I. Rossetti, P. Lopinto, G. Migliavacca, L. Forni, Catal.
Today 117 (2006) 549.
◦
at 300 C (PdLCOc8r3) must be preferred for practical applica-
tions; indeed, it exhibited the aforementioned self-regeneration
behavior, permitting simple regeneration by reduction and thus
greater durability, as confirmed by our thermal stability test
[
[
[
26] R. Strobel, F. Krumeich, W.J. Stark, S.E. Pratsinis, A. Baiker, J. Catal. 222
(2004) 307.
27] R. Strobel, L. Madler, M. Piacentini, M. Maciejewski, A. Baiker, S.E.
Pratsinis, Chem. Mater. 18 (2006) 2532.
28] S. Hannemann, J.D. Grunwaldt, F. Krumeich, P. Kappen, A. Baiker, Appl.
Surf. Sci. 252 (2006) 7862.
(Fig. 4).
5
. Conclusion
[29] L. Fabbrini, A. Kryukov, S. Cappelli, G.L. Chiarello, I. Rossetti, C. Oliva,
L. Forni, J. Catal. 232 (2005) 247.
[
[
[
30] M. Uenishi, H. Tanaka, M. Taniguchi, I. Tan, Y. Sakamoto, S. Matsunaga,
K. Yokota, T. Kobayashi, Appl. Catal. A Gen. 296 (2005) 114.
31] C. Oliva, S. Cappelli, A. Kryukov, G.L. Chiarello, A.V. Vishniakov, L.
Forni, J. Mol. Catal. A 255 (2006) 36.
32] G.L. Chiarello, J.D. Grunwaldt, D. Ferri, F. Krumeich, C. Oliva, L. Forni,
A. Baiker, J. Catal. 252 (2007) 127.
In the low-temperature selective reduction of NO by H2
in the presence of excess O2, the FP-made Pd/LaCoO3 cata-
lyst exhibited greater activity and durability than a correspond-
ing sample prepared by the traditional impregnation method.
The temperature of precalcination and prereduction strongly
[33] J.D. Grunwaldt, M. Caravati, S. Hannemann, A. Baiker, Phys. Chem.
Chem. Phys. 6 (2004) 3037.
affected the catalytic activity of the FP-made Pd/LaCoO . In
3
[34] T.J. Ressler, Synchr. Rad. 5 (1998) 118.
particular, the best performance was obtained with samples pre-
[35] S.I. Zabinsky, J.J. Rehr, A. Ankudinov, R.C. Albers, M.J. Eller, Phys. Rev.
B 52 (1995) 2995.
◦
calcined at 800 C, followed by reduction. We suggest a co-
operative mechanism involving surface Co and Pd particles,
leading to the formation of nitrate species on Co, subsequently
reduced by hydrogen spillover from Pd. Although reduction at
[
36] C.A. de Wolf, B.E. Nieuwenhuys, Surf. Sci. 469 (2000) 196.
[37] D. Pietrogiacomi, M.C. Campa, S. Tuti, V. Indovina, Appl. Catal. B Envi-
ron. 41 (2003) 301.
[38] L.G. Tejuca, J.L.G. Fierro, Properties and Applications of Perovskite-Type
Oxides, Marcel Dekker, New York, 1993.
◦
◦
6
00 C gave better catalytic activity, reduction at 300 C seems
preferable, conferring a self-regeneration behavior to the cata-
lyst.
[39] J.M.D. Tascòn, L.G. Tejuca, C.H. Rochester, J. Catal. 95 (1985) 558.
[40] M. Kantcheva, A.S. Vakkasoglu, J. Catal. 223 (2004) 352.