Fe-based perovskites substituted by copper and palladium for NO + CO reaction

Article abstract of DOI:10.1016/j.jcat.2006.05.033

Nanoscale Fe-based perovskites with nominal formula LaFe_1-x(Cu,Pd)_xO₃ were generated by reactive grinding and characterized by N₂ adsorption, X-ray diffraction (XRD), temperature-programmed reduction by hydrogen (H₂-TPR), temperature-programmed desorption (TPD) of O₂, NO, and CO, temperature-programmed surface reduction (TPSR) of NO under CO/He flow, and activity test toward NO + CO reaction. The catalytic performance of LaFeO₃ can be considerably improved via 20% Cu substitution, leading to a 74% N₂ yield and 72% CO conversion at 350 °C, under an atmosphere of 3000 ppm NO and 3000 ppm CO in helium at a space velocity of 50,000 h^-1. This improvement was ascribed to the facility in generation of anion vacancies after Cu incorporation, which plays a crucial role on NO adsorption and dissociation. In addition, the enhanced reducibility of LaFe_0.8Cu_0.2O₃ after Cu substitution results in the promotion of CO oxidation and anion vacancy regeneration, providing another clue for this improvement. N₂O decomposition (31% N₂ yield at 500 °C) is much easier than NO decomposition (below 5% at T < 500 ° C) over LaFe_0.8Cu_0.2O₃. Conversion of both NO and N₂O is significantly improved by the presence of the reducing agent. A mechanism was proposed with dissociation of chemisorbed NO, forming N₂ and/or N₂O, and an oxidized perovskite surface, which was continuously reduced by CO with the generation of CO₂. Great performance at low temperature was found over LaFe_0.97Pd_0.03O₃ with a 96% NO conversion and 86% CO conversion at 300 °C, corresponding to the outstanding redox properties of this catalyst. O₂ has a strongly detrimental effect, leading to the consumption of the reducing agent by oxidation.

Full text of DOI:10.1016/j.jcat.2006.05.033

Journal of Catalysis 242 (2006) 241–253
www.elsevier.com/locate/jcat
Fe-based perovskites substituted by copper and palladium
for NO + CO reaction
a
b
a,∗
Runduo Zhang , Houshang Alamdari , Serge Kaliaguine
a
Department of Chemical Engineering, Laval University, Ste Foy (QC), G1K 7P4, Canada
b
Nanox Inc., 4975 rue Rideau, Local 100, Quebec, G2E 5H5, Canada
Received 10 March 2006; revised 25 May 2006; accepted 31 May 2006
Available online 17 July 2006
Abstract
Nanoscale Fe-based perovskites with nominal formula LaFe
(Cu,Pd)xO were generated by reactive grinding and characterized by N
2
1−x
3
adsorption, X-ray diffraction (XRD), temperature-programmed reduction by hydrogen (H -TPR), temperature-programmed desorption (TPD) of
2
O , NO, and CO, temperature-programmed surface reduction (TPSR) of NO under CO/He flow, and activity test toward NO + CO reaction. The
2
catalytic performance of LaFeO can be considerably improved via 20% Cu substitution, leading to a 74% N yield and 72% CO conversion at
3
2
◦
−1
3
50 C, under an atmosphere of 3000 ppm NO and 3000 ppm CO in helium at a space velocity of 50,000 h . This improvement was ascribed
to the facility in generation of anion vacancies after Cu incorporation, which plays a crucial role on NO adsorption and dissociation. In addition,
the enhanced reducibility of LaFe Cu O after Cu substitution results in the promotion of CO oxidation and anion vacancy regeneration,
0
.8 0.2 3
◦
providing another clue for this improvement. N O decomposition (31% N yield at 500 C) is much easier than NO decomposition (below 5% at
T < 500 C) over LaFe Cu O . Conversion of both NO and N O is significantly improved by the presence of the reducing agent. A mechanism
2
2
◦
0.8 0.2
3
2
was proposed with dissociation of chemisorbed NO, forming N and/or N O, and an oxidized perovskite surface, which was continuously reduced
2
2
by CO with the generation of CO . Great performance at low temperature was found over LaFe
CO conversion at 300 C, corresponding to the outstanding redox properties of this catalyst. O has a strongly detrimental effect, leading to the
Pd
O with a 96% NO conversion and 86%
2
0.97 0.03 3
◦
2
consumption of the reducing agent by oxidation.
©
2006 Published by Elsevier Inc.
Keywords: NO reduction; Carbon monoxide; Reactive grinding; High surface area; Iron; Perovskite
1
. Introduction
modified by partial substitution in A and/or B sites with metals
of different oxidation states, which plays an important role in
various redox reactions [1,2].
Perovskite-type oxides with the general formula ABO3,
As long ago as 1952, perovskites were being used as cat-
alytic materials for CO oxidation [3]. Twenty years later,
their potential application as catalysts for automobile ex-
haust purification was pointed out by Libby [4]. Much atten-
tion has been paid in the last three decades to perovskites
as candidates for NO reduction using CO as a reducing
gas to simultaneously control the emission of those two
pollutants from motor vehicles, with the studies focusing
mainly on the following compositions: La(Fe, Mn)O3 [5],
La1−xSrxFeO3 [6], La1−x−ySrxCeyFeO3 [7], La2CuO4 [8],
La(Cr, Mn, Co, Ni)O3 [8], and LaCoO3 [9]. Copper and iron
lanthanates were considered the most active among the lan-
thanum perovskites of the first transition element series in the
where A usually stands for a rare earth metal cation coor-
dinated by 12 oxygens and B usually designates a transition
metal cation surrounded by six oxygens in octahedral coordi-
nation, exhibit interesting physicochemical properties. A great
diversity of perovskites can be achieved by a careful adjust-
ment of their compositions with different metal ions in A or
B sites while satisfying the limit of tolerance factor t = (rA +
√
rO)/ 2(rB + rO), where rA, rB, and rO are the ionic radii for
A, B, and O. Moreover, their physicochemical properties can be
*
Corresponding author. Fax: +1 418 656 3810.
E-mail address: kaliagui@gch.ulaval.ca (S. Kaliaguine).
0021-9517/$ – see front matter © 2006 Published by Elsevier Inc.
doi:10.1016/j.jcat.2006.05.033

242
R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
CO + NO reaction according to the literature [10]. The activ-
ity of La2CuO4 for NO reduction with CO can be enhanced
after Fe substitution by means of textural promotion of the dis-
trogen adsorption using an automated gas sorption system
(NOVA 2000, Quantachrome). Before each measurement, the
sample (about 200 mg) was degassed at 300 C under vac-
uum until complete removal of humidity (approximately 6 h).
The specific surface area was obtained from the linear part
◦
persion of Cu2 ions, as reported by Peter et al. [10]. A high
reactivity of isolated copper in NO elimination was reported by
Iwamoto [11] and confirmed by our early work [12–15]. Thus,
good catalytic performance can be expected by means of evenly
distributing Cu into lanthanum ferrite via B-site substitution.
Noble metal palladium is less expensive than Pt and Rh, has
a good activity for the oxidation of hydrocarbons and carbon
monoxide, and shows an excellent potentiality for NO elim-
ination with no-rhodium catalysts [16–18]. It was found to
be a suitable procedure to stabilize palladium against reaction
with support and volatilization by incorporating Pd into a per-
ovskite lattice. Substituted Pd can also move in and out of the
perovskite framework, resulting in better resistance to sinter-
ing [17]. In addition, significantly enhanced activity of simple
perovskites for NO removal was achieved from addition of even
small amounts of palladium [16].
+
of the BET curve (P/P = 0.01–0.10). The average pore
0
diameter was calculated from the desorption branch of N
2
isotherms according to the Barrett–Joyner–Halenda (BJH) for-
mula.
2
.2.2. Elemental analysis
The chemical composition (Fe, Cu, and Pd) of the prepared
solids and the residual impurity (Zn) were analyzed by atomic
absorption spectroscopy (AAS) using a Perkin–Elmer 1100B
spectrometer. The amount of La in solid solutions was estab-
lished by inductively coupled plasma (ICP) spectrometry (Op-
tima 4300DV, Perkin–Elmer). All materials were dissolved in a
◦
mixture of 10% HCl solution and concentrated HF at 60 C for
2
4 h before elemental analysis.
The use of perovskites synthesized by the traditional ceramic
method was limited by their low BET surface areas, usually
2
.2.3. Crystal phase identification and crystallite size analysis
The crystal structure of the prepared samples was deter-
2
<
5 m /g [19]. A new method of perovskite preparation, called
reactive grinding, was developed by our group with the aim of
improving the specific surface area while getting highly dis-
persed metal ions into the solid phase and also satisfying the
requirements of several practical applications [20–23].
mined by XRD using a Siemens D5000 diffractometer us-
ing CuKα radiation (λ = 1.5406 Å) with step scans from 20
◦
◦
to 70 in a 2θ angle and 2.4 s for each 0.05 step. Parti-
cle sizes (D) were evaluated by means of the Scherrer equa-
The objective of this work was to investigate the influence
of substitution of trivalent Fe cation by divalent Cu or Pd cation
on the physicochemical properties of lanthanum ferrite and to
clarify the correlation between perovskite properties and their
catalytic behavior in the NO + CO reaction. An attempt was
also made to propose a mechanism for Fe-based perovskites.
This work is therefore a continuation of our recently reported
study of NO reduction by propene in the presence of oxygen
over LaFe1−x(Cu, Pd)xO3 catalysts [13].
tion after Warren’s correction for instrumental broadening,
ꢀ
2
2
D = 0.9λ/( w − w cosθ), where λ is the incident wave-
1
0
length, w1 is the full width at half-maximum (FWHM) of
◦
X-ray reflection at 2θ ≈ 32 , w0 is the instrumental broaden-
ing determined through the FWHM of the X-ray reflection at
◦
2
θ ≈ 28 of SiO2 with particles larger than 1 µm. Identifica-
tion of the crystal phases was achieved via the JCPDS data
bank.
2
. Experimental
2.2.4. Temperature-programmed analysis
Temperature-programmed reduction by hydrogen (H2-TPR);
temperature programmed desorption (TPD) of O2, NO, and
CO; and temperature-programmed surface reduction (TPSR)
of NO under CO/He flow were performed using a multifunc-
tional characterization system (RXM-100, ASD Inc.) equipped
with a quadrupole mass spectrometer (MS) (UTI 100) and a
thermal conductivity detector (TCD). Before H2-TPR, the sam-
ples (50 mg) were pretreated under 10% O2/He flow at a rate of
2
.1. Preparation of perovskites
A series of LaFe1 (Cu, Pd) O mixed oxides were pre-
−x
x
3
pared by reactive grinding. Powders of La2O3 (Alfa, 99.99%),
Fe2O3 (Baker & Adamson, 97.49%), PdO (Aldrich, 99.98%),
or CuO (Aldrich, 99%) were mixed in the desired proportions,
then introduced into a vial with three tempered steel balls. The
◦
◦
La2O3 powder was calcined at 600 C for 24 h to transform any
20 ml/min (STP) for 1 h at 500 C, cooled to room temperature
La(OH)3 to La2O3 before grinding. The high-energy grinding
was performed via a SPEX 800 shaker mill normally at a speed
of 1100 rpm. This grinding process was conducted in two steps
of 8 h for synthesis and 10 h for refining with ZnO as the grind-
ing additive.
under the same atmosphere and purged by 20 ml/min helium
for 40 min to remove the physisorbed O2, then switched to a
◦
5% H2/Ar stream with the temperature rising up to 800 C at
◦
a heating rate of 5 C/min. The water in effluent gas of TPR
process was condensed via a cold trap filled with a mixture
of dry ice and ethanol. H2 consumption was recorded con-
tinuously by TCD with 20 ml/min of 5% H2/Ar as reference
gas.
Before the TPD of O2, NO, and CO, 50 mg samples were
treated with the various atmospheres of 10% O2, 3000 ppm
NO, and 3000 ppm CO, respectively, with a total flow rate of
2
2
.2. Characterization
.2.1. BET surface area and pore size distribution
The specific surface area and pore size distribution of the
◦
samples calcined at 500 C for 5 h were determined by ni-

R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
243
Table 1
Properties of LaFe
◦
(Cu, Pd)xO mixed oxides after calcination at 500 C for 5 h
1−x
3
a
Sample
Chemical composition
Specific surface
Pore diameter
(nm)
Crystallite
size (nm)
Crystal phase
LaFeO /Fe O
LaFeO
LaFeO
Perovskite
symmetry
2
area (m /g)
b
LaFeO
3
La
La
La
Fe
O
Pd
Cu
30.5
48.1
41.6
16.2
11.8
14.6
18.7
18.9
16.8
Orthorhombic
Orthorhombic
Orthorhombic
0.99 1.0 3 ± δ
3
3
3
2
3
LaFe
Pd
O
3
Fe
O
O
0.97 0.03
3
0.96 0.97 0.03 3 ± δ
LaFe Cu
O
Fe
0.8 0.2
0.97 0.80 0.20 3 ± δ
a
Zn can be detected as contaminant with a weight percent less than 2% of the total weight of Fe-based samples.
Trace.
b
◦
2
0 ml/min at 500 C for 1 h and then cooled down to room tem-
3. Results
.1. Physicochemical properties
The chemical compositions of the prepared solids deter-
perature under the same atmosphere, subsequently, flushed by
He as mentioned in H2-TPR tests followed by heating the reac-
3
◦
◦
◦
tor up to 500 C (800 C for O2-TPD) at a speed of 10 C/min.
O2, NO, CO, and CO2 desorbed during TPD experiments were
monitored by on-line MS with mass numbers of 32, 30, 28,
and 44, respectively. For ascription of mass 44 to either N2O
or CO2, masses 30 and 12 were systematically monitored with
ratios 30/44 or 12/44, allowing the ascription of mass 44 with-
out ambiguity.
mined by AAS and ICP are close to the nominal values (Ta-
ble 1). The BET specific surface areas and pore diameters of
the materials prepared by reactive grinding are also reported in
2
Table 1, showing a specific surface area approximately 30 m /g
◦
for the LaFeO3 solid solution even after calcination at 500 C.
This value is enhanced after Pd or Cu substitution reaching val-
TPSR of NO under CO/He flow over LaFe0.8Cu0.2O3 was
performed with the same sample pretreatment as that for TPD
of NO experiments. However, the thermodesorption was ac-
complished in 1000 ppm CO/He instead of He. The signals of
desorbed N , O , NO, CO, and (CO + N O) were recorded
2
ues in excess of 40 m /g.
The crystal phase analysis was performed by XRD (not
shown), revealing a main orthorhombic LaFeO3 perovskite-
type structure (JCPDS card 74-2203) for all prepared Fe-
containing catalysts and small amounts of α-Fe2O3 (JCPDS
card 86-0550) for the unsubstituted LaFeO3 sample. No dif-
fraction lines corresponding to PdO (JCPDS card 85-0713) or
CuO (JCPDS card 80-1917) were observed, suggesting that Pd
2
2
2
2
with the mass numbers of 14, 32, 30, 28, and 44, respectively.
The gaseous responses obtained by MS or TCD were calibrated
using standard gas mixtures.
and Cu metals were fully incorporated into the LaFeO struc-
3
2
.3. Activity tests
ture or the formed particles of PdO and CuO were too tiny to be
detected by XRD. The crystallite sizes were calculated using
Scherrer’s equation as given in Table 1, with values <20 nm.
Based on the result that both pore diameters and crystallite
sizes are around 10–20 nm, the porous structure of the present
perovskites was believed to be realized via clustering of their
individual nanoscale primary particles, in agreement with the
literature [24].
The catalytic activity tests of about 100 mg LaFe1−x(Cu,
Pd)xO3 catalysts for CO + NO reaction were performed in
a fixed-bed reactor under the mixture of 3000 ppm NO and
3
000 ppm CO balanced by He at a total flow rate of 60 mL/min,
−
1
corresponding to a gas hourly space velocity of 50,000 h . In
addition, the atmospheres of 3000 ppm N2O + 3000 ppm CO,
3
000 ppm NO, 3000 ppm N2O, 3000 ppm CO, and 3000 ppm
CO + 1500 ppm O2 were also adopted for the reaction of
N2O + CO, NO or N2O decomposition, and CO oxidation
by perovskite or oxygen, respectively, over LaFe0.8Cu0.2O3 at
the same space velocity. The reactor was heated externally
via a tubular furnace regulated by a temperature controller
3
.2. Temperature-programmed reduction by hydrogen
(H -TPR)
2
2
H -TPR experiments for the prepared perovskites were per-
formed to investigate their reducibility; the results are shown in
(
Omega CN3240), via a thermocouple inserted in the catalyst
Fig. 1. A multiple-step reduction was observed over LaFeO ,
showing a small peak centered at 215 C, followed by a broader
3
◦
◦
◦
bed, over the range of 100–500 C in steps of 50 C. Reac-
tant and product content in the reactor effluent was recorded
only after steady state was achieved at each temperature step.
Nitrogen oxides (NO and NO2) were analyzed using a chemilu-
minescence NO/NO2/NOx analyzer (Model 200AH, API Inc.).
N2O and CO gases were monitored using a FTIR gas ana-
lyzer (FTLA 2000, ABB Inc.). N2 was analyzed with a gas
chromatograph (GC) (Hewlett Packard 5890) equipped with a
TCD and columns of molecular 13X (45–60 mesh, 2.4 m) com-
bined with Silicone OV-101 (100–120 mesh, 0.6 m) operating
peak and then a third intense one, fully overlapping, with max-
◦
ima at 440 and 573 C, respectively. After partial substitution
of 3% Fe in lanthanum ferrite by Pd, H consumption oc-
2
◦
curred at relatively low temperature (78 C), followed by suc-
◦
cessive consumption at 219, 328, and 465 C. The TPR profile
of LaFe0.8Cu0.2O exhibited one sharp reduction peak centered
3
◦
◦
at 241 C and another centered at 460 C with shoulders at 360
◦
and 550 C.
Small amounts of highly reducible Fe4 have been reported
in LaFeO [25–27], and its reduction to Fe occurred at T <
300 C in H -TPR experiments [26]. Therefore, the minor re-
duction peak at 215 C in the H2-TPR profile of LaFeO3 was
+
◦
3+
at an oven temperature of 30 C. Calibration was done with a
standard gas containing known concentrations of the compo-
nents.
3
◦
2
◦

244
R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
Fig. 1. H -TPR profiles of Fe-based perovskites [13].
2
Fig. 2. TPD of O profiles of Fe-based perovskites [13].
2
4
+
3+
attributed to the reduction of Fe to Fe . The quantitative
Table 2
Amount of O desorbed from Fe-based perovskites during O -TPD experi-
ments
3+
analysis of the successive Fe reduction peaks with maxima
2
2
◦
at 440 and 573 C correlated well with the H2 consumption
3
+
2+
necessary for the reduction of Fe to Fe . Thus, those two
Sample
Amount of oxygen desorbed
peaks were ascribed to Fe3 → Fe reduction occurring over
+
2+
α-O (µmol/g)
<700 C
β-O (µmol/g)
700–800 C
Total
(µmol/g)
2
2
◦
◦
surface and in bulk of parent LaFeO3, respectively. Finally, the
◦
minor reduction peak above 700 C was ascribed to the par-
LaFeO
LaFe
LaFe Cu O
0.8 0.2
38.2
62.8
196.5
76.2
173.1
202.2
114.4
235.9
398.7
3
2+
tial reduction of Fe to metallic iron. The low temperature
78 C) H2 consumption that occurs over LaFe0.97Pd0.03O3,
corresponds to the reduction of Pd to Pd according to the
literature [28]. This result reveals the excellent redox proper-
ties of Pd-substituted perovskite, possibly leading to a good
Pd
O
3
0
.97 0.03
3
◦
(
2
+
0
◦
desorbed from LaFeO3 at T < 700 C is designated as α-O2
and ascribed to oxygen species bound to the surface anion va-
4+
catalytic performance especially at low temperatures. Fe
→
◦
cancies of perovskite [22]. More O2 desorbed at T > 700 C
3+
◦
Fe reduction is present at 219 C, along with a downward
is referred to as β-O2 and attributed to oxygen species lib-
erated from the lattice [22]. Compared with LaFeO3, similar
α-O2 desorption but more intense β-O2 desorption was found
over LaFe0.97Pd0.03O3 during TPD of O2. Interestingly, a broad
3
+
2+
shift of Fe → Fe
reduction in the H2-TPR profile of
LaFe0.97Pd0.03O3 with respect to that of LaFeO3. This down-
ward shift likely corresponds to the effect of atomic hydrogen
generated on the Pd already formed at this temperature level.
The sharp peak at 241 C observed in the H2-TPR profile of Cu-
0
◦
plateau-like α-O2 desorption peak at 200–700 C together with
◦
◦
a sharp β-O2 peak maximum at 796 C was observed in the
substituted lanthanum ferrite was ascribed to the reduction of
O2-TPD profile for LaFe0.8Cu0.2O3 perovskite. This significant
enhancement of adsorbed α-O2 is likely related to the surface
oxygen vacancies generated on Cu substitution.
lattice Cu2 to Cu , whereas the shoulder at 550 C may be re-
+
+
◦
+
0
lated to the complete reduction from Cu to Cu . These values
mostly coincide with those observed in H2-TPR of Cu/MCM-
Fig. 3 shows the desorption signals of NO (m/e = 30) dur-
ing the NO-TPD process after the adsorption of 3000 ppm
NO/He for the three Fe-based perovskites. The amount of NO
desorbed from the perovskite was calculated after deconvolu-
4
[
1 [12], LaCo1−xCuxO3, and LaMn1−xCuxO3 (x = 0.1, 0.2)
4
+
14,15]. The Fe reduction peak possibly overlaps with the
2+
+
rising part of the Cu → Cu reduction peak, whereas the
3+
2+
Fe → Fe reduction peak shifts to lower temperatures in
◦
+
tion and is given in Table 3. A minor desorption peak at 84 C
Cu-substituted samples, suggesting that the lattice Cu ions
◦
and an intense one centered at 214 C with a slight shoulder
produced at this temperature interact with the iron ion, mak-
ing it more easily reduced.
at its falling part were observed in the NO desorption trace of
LaFeO3. Similar NO desorption behavior was found over Pd-
and Cu-substituted perovskite except for an enhanced intensity
3
.3. Temperature-programmed desorption of O , NO and CO
2
◦
of the second peak at approximately 220 C. Low O2 desorp-
(O2, NO and CO-TPD)
tion parallel with only the high-temperature NO shoulder was
observed during NO-TPD studies (not shown).
The mobility of oxygen for Fe-based perovskites was inves-
tigated by O2-TPD experiments, as illustrated in Fig. 2. The
amount of O2 released from perovskites was calculated af-
ter deconvolution of the O2 desorption curve via Lorentzian
peak fitting and reported in Table 2. The small amount of O2
The MS signals of CO (m/e = 28) and CO2 (m/e = 44) dur-
ing the TPD of CO experiments over lanthanum ferrites were
recorded. A quantitative analysis of the various carbonaceous
gases desorbed from perovskites is presented in Table 4. CO

R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
245
Fig. 3. TPD of NO profiles of Fe-based perovskites.
Fig. 4. MS signals during TPSR of NO under CO/He flow over LaFe0.8Cu0.2O3
perovskite, conditions: flow rate = 20 ml/min, 3000 ppm NO, 1000 ppm CO.
Table 3
Amount of NO desorbed from Fe-based perovskites during NO-TPD experi-
ments
oxygen signal detected by MS rapidly diminished over the
◦
range of 100–350 C simultaneously with consumption of CO,
◦
whereas recovery of this O2 signal occurred at T > 400 C. At
Physisorbed NO
Chemisorbed NO (µmol/g)
◦
(µmol/g)
temperatures below 200 C, the desorption peak of NO in Fig. 3
Nitrosyl
species
Nitrate
species
Total
(2.1 µmol/g) appeared again in Fig. 4 (6.1 µmol/g) with a slight
◦
◦
◦
increase in density and a small upward shift. However, the NO
thermodesorption peak centered at 225 C (36.3 µmol/g) and
the shoulder at its falling part (3.3 µmol/g) in Fig. 3 is to-
LaFeO
LaFe
84 C
4
214 C
283 C
1.8
22.0
27.1
39.6
3
◦
.5
20.2
◦
◦
◦
Pd
O
3
80 C
.1
222 C
317 C
0
.97 0.03
3
4
24.2
2.9
tally suppressed in Fig. 4, leaving only a NO desorption of
◦
◦
◦
LaFe Cu
O
86 C
.1
225 C
334 C
0.8 0.2
◦
0
.7 µmol/g at about 280 C. This result indicates that the corre-
2
36.3
3.3
sponding adsorbed NO species almost entirely reacted with CO.
◦
Some N2 desorption (4.9 µmol/g) with a maximum at 160 C
Table 4
was indeed observed but not in amounts comparable with the
content of desorbed NO reported for the same sample in Fig. 3.
The appearance of mass 44, which could in this case be ascribed
to both CO and N O, coincided with the disappearance of CO.
Amounts of CO and CO desorbed from Fe-based perovskites during CO-TPD
experiments
2
Sample
CO
CO
Total
carbonaceous
species
2
2
2
◦
◦
T ( C) Amount T ( C) Amount
(µmol/g)
(µmol/g)
3
.5. Activity tests
Amount (µmol/g)
LaFeO
>250
>250
>225
53.2
36.9
25.0
>350
>200
>220
43.9
102.5
136.9
97.1
139.4
161.9
A blank test was performed under 60 ml/min of 3000 ppm
CO and 3000 ppm NO balanced by helium in an empty reac-
3
LaFe
Pd
O
3
0.97 0.03
3
LaFe Cu
O
0.8 0.2
tor (not shown). No significant NO conversion (<2%) or CO
◦
conversion (<5%) was observed up to 500 C. Thus, the con-
and CO2 were desorbed from LaFeO3 at temperatures above
versions generated from homogeneous reaction in the absence
of catalyst are essentially negligible under usual reaction con-
ditions. The temperature dependence of NO conversion over
the three Fe-based perovskites is shown in Fig. 5a. Over un-
3
◦
2
50 and 350 C, respectively. The CO desorption is gradually
suppressed simultaneously with clearly enhanced CO2 desorp-
tion after Pd or Cu incorporation into the B sites of LaFeO3.
This result implies that CO transformation into adsorbed CO2
via oxidation over lanthanum ferrite is promoted by Pd or Cu
substitution. The total carbonaceous species (CO + CO2) des-
orbed from Fe-based perovskites follow the general trend of
LaFe0.8Cu0.2O3 > LaFe0.97Pd0.03O3 > LaFeO3.
substituted LaFeO , the NO conversion became detectable at
◦
250 C and increased progressively with temperature, reaching
◦
69% at 350 C. An obvious enhancement was observed over
lanthanum ferrite after Cu partial substitution, resulting in ini-
◦
◦
tiation at 200 C and then reaching a value of 84% at 350 C.
◦
◦
NO conversion was 30% at 200 C and up to 96% at 300 C
over LaFe0.97Pd0.03O . A similar order of LaFe0.97Pd0.03O >
3
.4. Temperature-programmed surface reduction (TPSR) of
3
3
NO under CO/He flow
LaFe0.8Cu0.2O3 > LaFeO3 for N2 yield over perovskites was
also observed (Fig. 5b). N2O was detected in the effluent at
◦
TPSR of the NO profile measured under a flowing 1000 ppm
CO/He mixture over LaFe0.8Cu0.2O3 is depicted in Fig. 4. The
low temperature, with a maximum yield of 13% at 300 C
◦
over LaFeO3, 16% at 300 C over LaFe0.8Cu0.2O3, and 41%

246
R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
(a)
(b)
(c)
(d)
−
1
Fig. 5. (a) NO conversion in CO + NO reaction over Fe-based perovskites, conditions: GHSV = 50,000 h , 3000 ppm NO, 3000 ppm CO. (b) N yield in
2
−
1
CO + NO reaction over Fe-based perovskites, conditions: GHSV = 50,000 h , 3000 ppm NO, 3000 ppm CO. (c) N O yield in CO + NO reaction over Fe-based
2
−
1
perovskites, conditions: GHSV = 50,000 h , 3000 ppm NO, 3000 ppm CO. (d) CO conversion in CO + NO reaction over Fe-based perovskites, conditions:
−
1
GHSV = 50,000 h , 3000 ppm NO, 3000 ppm CO.
◦
at 250 C over LaFe0.97Pd0.03O3, as depicted in Fig. 5c. The
likely related to the ease of its N–O bond cleavage. Never-
theless, it is noted that N2O decomposition cannot occur at
T < 350 C, indicating that sufficient energy is still necessary
conversion of CO described in Fig. 5d showed a progressive
increase up to 95% in the region of 200–500 C for LaFeO3.
◦
◦
The conversion level was improved after Cu substitution, re-
sulting in an increase from 32% to 42% at 300 C compared
for the dissociation of this N–O bond. The transformation of
nitrogen oxides (NO and N2O) into N2 was significantly im-
proved by CO as a reducing agent, resulting in N2 yields of
40% in NO + CO reaction and 70% in N2O + CO reaction at
◦
with unsubstituted LaFeO3. A remarkable CO conversion at
low temperature was found over LaFe0.97Pd0.03O3, achieving a
◦
◦
conversion of 62% at temperature as low as 250 C and reach-
ing 86% at 300 C.
300 C. The higher N2 yield in N2O reduction by CO is again
◦
attributed to the facility of N2O molecular activation compared
with NO.
N2 yields in the reaction of NO + CO, N2O + CO, NO
decomposition, and N2O decomposition for LaFe0.8Cu0.2O3
are depicted in Fig. 6a. The objective of these experiments is
a better understanding of NO and CO transformations in the
catalytic process. In the absence of reducing agent, little NO
CO conversions in the reaction of NO + CO, N2O + CO,
CO oxidation by perovskite (no oxygen), and CO + O2 over
LaFe0.8Cu0.2O3 are compared in Fig. 6b. CO oxidation can
◦
still occur with a maximum conversion of 40% at 450 C in
◦
decomposition (below 5%) was found at 350–500 C. N2O de-
the absence of O2, which is likely realized by consuming oxy-
gen species from the perovskite surface. With the diminution
of the remaining O2 species by previous CO oxidation and by
composition was much easier than NO decomposition starting
◦
◦
at 350 C and reached a N2O conversion of 31% at 500 C,

R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
247
that the N2 production obtained in the absence of O2 was ap-
proximately twice its value in the presence of 500 ppm O2. The
introduction of 5% O2 gas in feed led to a total inhibition of N2
yield (Fig. 7a). N2O concentration in effluent severely declined
with an increase in O2 feed concentration (Fig. 7b). NO2 pro-
duction was negligible during the reaction of 3000 ppm NO and
3
000 ppm CO in the absence of O2 and increased in its pres-
ence (Fig. 7c). CO conversion was indeed also promoted with
increasing O2 concentration in feed (Fig. 7d).
4
. Discussion
4
.1. O2-TPD study
TPD of O2 experiments were carried out to study the O2
species formed over Fe-based perovskites after O2 adsorption;
the results are reported in Fig. 2 and Table 2. Among three Fe-
containing samples, the lowest α- and β-O2 desorptions were
observed for LaFeO3, revealing its low coverage of molecular
oxygen and poor reducibility of lattice oxygen, in accordance
with previous reports [29,30]. β-O2 desorption of lanthanum
ferrite can be significantly enhanced via substitution of just
(a)
3
% Fe3 in the B sites by Pd , likely attributed to the im-
+ 2+
2+
proved mobility of lattice oxygen due to highly reducible Pd
,
as proven by H2-TPR analysis for LaFe0.97Pd0.03O3 (Fig. 1).
With the substitution of 20% Fe3 by Cu , significantly en-
hanced α-O2 was achieved, ascribed to the generation of a pos-
itive charge deficiency after incorporation of Cu into the lattice,
which is compensated for by oxygen vacancies. A similar en-
hancement in β-oxygen desorption by Cu substitution as by Pd
substitution was also found and attributed to the improvement
+
2+
of lattice oxygen mobility according to the H -TPR trace for
2
LaFe0.8Cu0.2O3 (Fig. 1).
Based on our previous description [22], the following
process was assumed to occur during calcination of LaFeO3
perovskite:
1) Transient generation of anion vacancy during calcination
of fresh sample
(
b)
Fig. 6. (a) N yield in various reactions over LaFe Cu O , conditions:
2
0.8 0.2 3
−
1
GHSV = 50,000 h , 3000 ppm NO, 3000 ppm N O, 3000 ppm CO.
2
(
b) CO conversion in various reactions over LaFe Cu O , conditions:
0.8 0.2 3
−
1
GHSV = 50,000 h , 3000 ppm NO, 3000 ppm N O, 3000 ppm CO,
2
−
OH
OH
|
O
1
500 ppm O .
2
|
|
−
H2O
3+
3+
2+
3+
.
Fe + Fe
−→ Fe
ꢀ
+ Fe
(1)
rapid O2 desorption at relatively higher temperature, a decline
in CO oxidation occurred. This rapid desorption of O2 species
at relatively higher temperature was indeed confirmed by O2-
TPD (Fig. 2) and TPSR (Fig. 4) studies. Nitrogen oxides (NO
and N O) promoted CO oxidation of up to 42% and 72% at
00 C, respectively. N2O was again more active than NO in
(anion vacancy)
2
) Instantaneous formation of α-O2 by adsorbing O2 at anion
vacancy
−
O
2
2
|
◦
3
Fe2+ꢀ + O2 −→ Fe
3+
.
(2)
donating its oxygen atom and oxidizing CO. Gaseous oxygen
with the same atomic O concentration as 3000 ppm NO or N O
The following two reactions are proposed for the α-O2 desorp-
2
tion process:
led to the highest CO oxidation, achieving a value of 85% even
at 300 C. The results, shown in Fig. 6b, suggest an order of
◦
−
−
2−
\
O
|
2
O
|
O
the efficiency in oxygen utilization for CO oxidation as ad-
/
−
sorbed oxygen molecule (O ) > oxygen atom (O) in NO or
3+
Fe + Fe −→ Fe3+ Fe + O2,
3+
3+
2
(3)
(4)
−
2−
N2O > lattice oxygen (O or O ).
−
The influence of O2 feed concentration on the catalytic per-
formance of LaFe0.8Cu0.2O3 is presented in Figs. 7a–7d to clar-
ify the role of gaseous O2 in NO + CO reaction. It is observed
O
2
|
Fe −→ Fe2+ꢀ + O2.
3
+

248
R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
(a)
(b)
(c)
(d)
−
1
Fig. 7. (a) Effect of O on N yield in CO + NO reaction over LaFe Cu
O
0.8 0.2 3
perovskite, conditions: GHSV = 50,000 h , 3000 ppm NO, 3000 ppm CO, 0,
2
2
−
1
5
00 ppm, 5% O . (b) Effect of O on N O yield in CO + NO reaction over LaFe Cu O perovskite, conditions: GHSV = 50,000 h , 3000 ppm NO, 3000 ppm
2
2
2
0.8 0.2 3
−
1
CO, 0, 500 ppm, 5% O . (c) Effect of O on NO yield in CO + NO reaction over LaFe Cu
O
perovskite, conditions: GHSV = 50,000 h , 3000 ppm NO,
2
2
2
0.8 0.2 3
−1
,
3
000 ppm CO, 0, 500 ppm, 5% O . (d) Effect of O on CO conversion in CO + NO reaction over LaFe Cu
O
perovskite, conditions: GHSV = 50,000 h
2
2
0.8 0.2 3
3000 ppm NO, 3000 ppm CO, 0, 500 ppm, 5% O .
2
In Eqs. (1)–(4) iron ions are indicated as examples. In the sub-
N-containing species formed during NO adsorption over Fe-
based perovskites were investigated by means of monitoring
the NO desorption traces (m/e = 30) in the effluents by MS
(Fig. 3). The amount of NO species desorbed was subsequently
quantified, as given in Table 3. A broad NO desorption peak
3+
2+
2+
stituted samples, pairs of one Fe and one Cu or Pd are
also present. Therefore, reactions (1) and (3) may occur on
these new ion pairs. Processes (2) and (4) also occur in the co-
ordination sphere of copper or palladium ions.
◦
◦
The oxygen desorbed above 700 C was assigned as the β-
occurred at 210–230 C with minor shoulders at the rising and
oxygen liberated from the lattice leaving bulk oxygen vacancies
and reduced cations. All of the steps for its desorption were
described in our previous study on similar compounds [13].
falling parts, respectively, during NO-TPD experiments. Centi
and Perathoner [33] concluded that the thermal stability of ad-
sorbed species increased with increasing the oxidation state of
nitrogen in N-containing adspecies according to transient cat-
alytic experiments conducted over Cu/ZSM-5. In our previous
NO+O2-TPD studies over Co-, Mn-, and Fe-based perovskites
[13–15], three NO peaks appearing at similar temperatures as
those in the present NO-TPD experiments, apart from minor
desorption at medium temperature and an intensive one at high
temperature, were also found and ascribed to the desorption
4
.2. NO-TPD study
The formation of (mono-, di-) nitrosyl and (monodentate,
bidentate, bridging) nitrate species was found over LaFeO [31],
3
Cu/ZSM-5 [32], and Cu/MCM-41 [12] after NO adsorption
as reported in the literature. In the present work, adsorbed

R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
249
of physisorbed NO, nitrosyl species, and nitrate species ac-
cording to their thermal stability. The NO peaks obtained at
low (80–90 C), medium (210–230 C), and high (280–340 C)
temperatures in present NO-TPD analyses were thus correlated
to physisorbed NO, nitrosyl, and nitrate species, respectively.
CO oxidation was proposed as the following step:
◦
◦
◦
(7)
O desorption parallel with the minor NO desorption occur-
ring at 283 C for LaFeO3 (not shown) gives another clue to the
2
Furthermore, the lower binding energy for this surface lattice
oxygen is favorable for CO oxidation [40]. The improved CO
oxidation after Pd and Cu substitution can be explained by the
enhanced mobility of the lattice oxygen, which was confirmed
◦
ascription of nitrate species in agreement with the report that
−
the desorption of NO3 species would appear as a NO desorp-
◦
tion peak at high temperature (T > 300 C) associated with O2
by H -TPR test in the present study (Fig. 1):
2
desorption [13–15,34,35]. It is noticeable that only a minor ni-
trate species was formed over perovskites after NO adsorption,
appearing as a shoulder rather than as a peak with respect to ni-
trosyl species, although its density was slightly enhanced on Cu
or Pd substitution.
(8)
Fig. 3 shows that the intensity of the broad NO peak at
In the presence of O2, this CO oxidation can be significantly
accelerated by α-O2 [40]:
◦
2
14 C for LaFeO was enhanced via Pd and Cu substitution.
3
Based on the fact that more oxygen vacancies can be gener-
ated after trivalent iron ions substituted by bivalent palladium
(9)
or copper ions (as confirmed by O -TPD experiments (Fig. 2)),
2
this intense NO desorption related to nitrosyl species was likely
formed by NO adsorption on anion vacancies, as verified by
Shin et al. via FTIR study [36]. The NO is chemisorbed as
A higher content of adsorbed carbonaceous species was found
over Cu- or Pd-substituted samples with respect to lanthanum
ferrite, likely related to the enhanced density of anion vacancies
−
a negatively charged form (NO ), which is thought to be the
first important step for NO + CO reaction [37], because back-
donation occurs through antibonding orbitals and thus deter-
mines a weakening of the N–O bond with a net electron transfer
from metal to NO. The formation of nitrosyl species is formu-
lated on the basis of the above discussion:
(
the adsorption sites for CO and O ) after substitution.
2
4
.4. TPSR of NO under CO/He flow
Fig. 4 shows that NO desorption related to nitrosyl species
◦
(
observed at 225 C in the NO-TPD profile of LaFe0.8Cu0.2O3
in Fig. 3) almost vanishes in the presence of CO. Only the peak
at low temperature, which was ascribed to physisorbed NO, is
still significant in the NO desorption trace. This result implies
that both nitrosyl and nitrate species are reactive toward CO
and fully consumed by this reducing agent, in agreement with
the previous conclusion that the chemisorption of NO seems
to play an important role in the NO + CO reaction catalyzed
by perovskites [41]. Nitrate species were proven to be highly
active toward propene over perovskites [13–15] and toward
(
5)
4
.3. CO-TPD study
Tascón et al. [38] pointed out an inhibiting effect of NO on
subsequent CO adsorption, which was greater than the inhibit-
ing effect of CO on NO adsorption over LaFeO . Therefore,
CO absorption appears to be weaker than NO absorption on
the same site of Fe-based perovskites. The CO adsorption thus
likely involves adsorption at anion vacancies with a coordina-
tive bond without electron transfer:
3
methane over Ag/Al O [34] and Ag-ZSM-5 [42], resulting in
2
3
good deNOx activity at high temperature. The reactivity of this
nitrate species toward CO was confirmed again in the present
TPSR study. Nevertheless, the nitrate species formed over per-
ovskites was minor (3.3 µmol/g) compared with the nitrosyl
species (36.3 µmol/g) according to the NO-TPD study. Fur-
◦
thermore, NO reduction was already remarkable at T < 300 C
(6)
(Fig. 5a), whereas nitrate species seemed to be stable over this
temperature range and corresponded only to high-temperature
deNO activity. Hence, nitrosyl species is believed to make the
most significant contribution to NO reduction by CO.
x
A decline in CO desorption and an opposite tendency in CO2
desorption from Pd- or Cu-substituted samples compared with
those from parent LaFeO3 (Table 4) were observed, suggesting
that CO oxidation occurred over Fe-based perovskites and was
further improved by divalent cation substitution. The oxidation
of CO has been suggested by Voorhoeve [39] to be a suprafacial
Simultaneously with the CO consumption and CO + N O
2
2
(m/e = 44) formation, the MS signal of O decreased pro-
2
◦
gressively in the range of 100–400 C due to rapid CO oxida-
tion. Nevertheless, recovery of the O signal was obtained at
2
◦
T > 400 C, likely related to rapid desorption of O species at
2
−
catalytic process involving a surface lattice oxygen (O ). Thus,
2 2
relatively high temperature. The O consumption by CO to CO

250
R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
oxidation was also slightly inhibited (see the m/e = 44 trace in
Fig. 4), possibly due to a serious surface coverage of carbonate
species, providing another explanation for these phenomena.
N2 was also detected during this experiment, again suggest-
ing that the reduction of nitrosyl species by CO is associated
with the NO catalytic elimination.
(Fig. 5c). Based on our previous studies and the present work,
a mechanism for NO + CO reaction, including the adsorption
and transformation of reactants over perovskite and redox cat-
alytic recycling during reaction, with the product components
depending on reaction temperature, was proposed that involves
the following steps:
At low temperature, adsorbed nitrosyl species combine to
4
.5. Activity tests and reaction mechanism
2 2
yield adsorbed N O species. This can decompose over the re-
duced perovskite, yielding N2O and N2 simultaneously with
oxidation of the perovskite surface. The latter was recognized as
a rate-determining step for NO reduction by CO at low tempera-
The direct decomposition of NO would represent the most
attractive solution in NO emission control, because no addi-
tional reactant was required for the reaction with potential prod-
ucts of only N2 and O2. In principle, simple NO decomposition
is a thermodynamically favored reaction at low temperature;
however, this reaction is very slow, due to the high dissociation
energy of NO (153.3 kcal/mol), and sufficiently efficient cata-
lysts have not yet been discovered [36]. Up to now, Cu-zeolites
ture. The formation of the dimeric NO species (N O ) involving
2
2
the N–N bond formation is followed by N–O bond cleavage
[47,48]. Two parallel reactions for chemisorbed NO dissocia-
tion thus occur over a reduced surface with N O and N as the
2
2
respective products:
+
have been considered the best catalysts, with isolated Cu ions
considered as active centers in NO decomposition [11]. Per-
ovskites exhibit better activity than most other metal oxides
for this reaction, but at 500 C they are still much less active
(10)
◦
than the best copper zeolites [43–45]. The fact that NO hardly
◦
decomposes over LaFe0.8Cu0.2O3 perovskite at <500 C was
confirmed again in this work (Fig. 6a).
(11)
N2O is not toxic, but it does play a role in ozone deple-
tion. Its decomposition over LaFe0.8Cu0.2O3 was examined,
and the results (shown in Fig. 6a) demonstrated better activ-
Reaction (11) involves the breakage of only one N–O bond,
making this reaction occur much more easily than reaction (10),
indicating that reaction (11) is dominating at low temperature.
The above two reactions can also proceed on copper and pal-
◦
ity in terms of N2 yield (31% at 500 C) than that achieved in
◦
NO decomposition (3% at 500 C). N2O is known to have a lin-
ear structure N–N–O, in which the N–O distance is increased
by about 0.03 Å relative to free NO with a bond order of 2.5,
resulting in an easy cleavage of its N–O bond and better dissoci-
ation behavior compared with NO. However, sufficient energy
is still necessary for N–O bond breakage, according to the re-
ladium cations. CO can be oxidized by Fe3
+
O
−
species, as
illustrated in Eq. (7), together with regeneration of anion va-
cancies on the surface for the continuous NO dissociation.
The reaction routes occurring at low temperature can thus be
described as two parallel reactions:
sult that N2O was merely decomposed at temperatures above
◦
3
50 C (Fig. 6a).
Route (a): 2NO + 2CO → N + 2CO2
(12)
2
With the assistance of CO, N2 yields for both NO and
N2O transformations were significantly improved, up to 74%
and
in NO + CO reaction and 89% in N2O + CO reaction even at
Route (b): 2NO + CO → N2O + CO2 (fast).
(13)
◦
3
50 C (Fig. 6a). Chien et al. [41] observed a higher NO adsorp-
tion rate for activated (reduced) than for unactivated LaCoO3
and La0.85Ba0.15CoO3. A reduced perovskite surface with large
amount of anion vacancies was reported to be crucial for the NO
dissociation according to FTIR and EPR spectroscopic analy-
ses [46]; thus, the role of CO seems to be to maintain a reduced
surface, which is necessary for successive NO dissociation. CO
oxidation can also be promoted by the O atom coming from NO
or N2O (Fig. 6b) via a catalytic process. The above results and
discussions indicate that these perovskites likely act as “oxygen
reservoirs,” transferring oxygen atoms from nitrogen oxides to
CO and thus achieving NO reduction and CO oxidation simul-
taneously.
Although N2O is detected in the effluent as a byproduct at low
temperature (see Fig. 5c), its further conversion as an inter-
mediate must not be excluded [29,46,49]. Pomonis proposed
a partially successive N2O decomposition [Route (c): 2N2O →
2N2 + O2], occurring at low temperature simultaneously with
reaction (13) over La1−xSrxFeO3 [5,6,50] and LaMnO3 [5].
However, N2O decomposition was found to be quite difficult at
◦
<350 C LaFe0.8Cu0.2O3 (see the line for N2O decomposition
in Fig. 6a) and is believed to occur only over a reduced per-
ovskite with the participation of CO to consume surface oxygen
species (Fig. 6a):
Route (d): N2O + CO → N2 + CO2.
(14)
The high reactivity of nitrosyl species toward CO over
LaFe0.8Cu0.2O3 was established via TPSR of NO under a
CO/He flow test. N2O formation for this catalyst was observed
during NO reduction by CO with a maximum yield of 16%
The ratio of NO conversion to CO conversion must be equal
to unity for Route (a) and to 2 for Route (b), whereas the mo-
lar ratio [CO2]/([N2] + [N2O]) in the products should be 2 for
Route (a) and 1 for Route (b) according to Eqs. (12) and (13).
◦
at 300 C and ultimately diminishing at higher temperature

R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
251
Fig. 8. Molar ratios of X /X
and [CO ]/([N ] + [N O]) in CO + NO re-
1
NO CO
2
2
2
Fig. 9. O effect on ratio of NO conversion to CO conversion in CO + NO
2
−
action over Fe-based perovskites, conditions: GHSV = 50,000 h , 3000 ppm
−1
,
reaction over LaFe Cu
O
perovskite, conditions: GHSV = 50,000 h
0
.8 0.2 3
NO, 3000 ppm CO.
3
000 ppm NO, 3000 ppm CO, 0, 500 ppm, 5% O .
2
Therefore, the plots of XNO/XCO as well as [CO2]/([N2] +
4
.6. The influence of O2 in feed on the catalytic behavior of
[
N2O]) versus temperature provide an easy way to probe the
LaFe0.8Cu0.2O3
reaction route involved. XNO/XCO values of approximately 1.5
◦
at low temperatures and 1 at T > 400 C for lanthanum fer-
A serious inhibiting effect of O2 was observed during the re-
action of 3000 ppm NO and 3000 ppm CO over LaFe0.8Cu0.2O3
rites were observed in the present study; moreover, the ra-
tios of [CO2]/([N2] + [N2O]) for these catalysts are close to
(Figs. 7a–7d). The N2 yield and N2O concentration were
◦
1
.3 at low temperatures and 2 at T > 400 C, as depicted in
severely depressed by 500 ppm O2. In the presence of 5% O2,
N2 yield was essentially inhibited with values below 6%. In
contrast, a remarkable improvement in CO conversion was ob-
Fig. 8.
The reaction between CO and N2O [Route (d)], as well as the
decomposition of N2O [Route (c)], were accelerated at rising
temperature, as illustrated in Fig. 6a. As a result, all N2O gen-
erated during NO dissociation could be rapidly consumed, and
the reaction occurring at high temperature can be simplified as
Route (a) with XNO/XCO = 1 and [CO2]/([N2] + [N2O]) = 2.
The above analysis may be interpreted as more evidence
supporting a mechanism dominated by Route (b) at low tem-
perature and then gradually progressing to Route (a) at elevated
temperature.
tained with the O in the feed. Significant NO generation was
2
2
observed at 5% O2. The corresponding ratio of XNO/XCO under
different O2 feed concentrations as a function of temperature is
plotted in Fig. 9, showing that XNO/XCO values diminished with
increasing O2 concentration and were much lower than 1 in the
◦
◦
presence of 5% O2 at 200 C. Above 400 C, a slight decline
in XNO/XCO was observed, possibly correlated with the inhibi-
tion of NO oxidation to NO2 by 5% O2 (see Fig. 7c) due to the
unfavorable thermal equilibrium for this reaction at high tem-
perature. The above results strongly imply that the reaction was
totally controlled by the unfavorable oxidation of the reducing
A significant improvement in N yield was achieved after the
2
incorporation of Cu into the B sites of Fe-based perovskite. Be-
sides the essential effect of copper ions in the transformation of
nitrogen oxides [33], Cu substitution promoted the generation
of anion vacancies, which facilitated the adsorption of reac-
tants (confirmed in NO-TPD (Fig. 3) and CO-TPD experiments
agent CO, as formulated in Eq. (9) with 5% O . At 500 ppm O ,
2
2
the XNO/XCO values increased progressively up to 0.59 as the
◦
reaction temperature increased from 200 to 500 C, revealing
a competitive reaction between NO reduction by CO with CO
oxidation by O2. It seems that the percentage of NO reduction
increased gradually at elevated temperature.
(Table 4)), as well as the dissociation of adsorbed NO. In addi-
tion, the regeneration of anion vacancies was also accelerated
after incorporation of Cu (see Eq. (8)) due to enhanced mobil-
ity of lattice oxygen. At low temperature, outstanding catalytic
5. Conclusion
performance was found over LaFe0.97Pd0.03O (Figs. 5a–5d),
3
yielding NO conversion of 96% and CO conversion of 86% at
Nanoscale LaFe1 (Cu, Pd) O perovskites were prepared
by reactive grinding with specific surface areas above 30 m /g,
−x
x
3
◦
2
temperature as low as 300 C. A large amount of N2O was de-
◦
◦
tected with a maximum yield of 41% at 250 C. Taking into
even after calcination at 500 C for 5 h, and crystal domain sizes
2+
account the high reducibility of LaFe0.94Pd0.06O3 found by
H2-TPR (Fig. 1), the great performance of the Pd-containing
sample for NO removal at low temperature was attributed to its
excellent redox properties, which can initiate the redox reaction
with a small energy barrier.
below 20 nm. More anion vacancies were generated after Cu
partial substitution of Fe3 ions in the B sites of the lattice due
to a positive charge deficiency. The nitrosyl species formed via
NO chemical adsorption on these anion vacancies was highly
reactive toward CO according to TPSR experiments. On the
+

252
R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
other hand, Cu substitution also enhanced the reducibility of
lanthanum ferrite and accelerated the regeneration of anion va-
cancies on the surface. Therefore, a significant improvement in
catalytic performance of NO + CO reaction was achieved over
LaFe0.8Cu0.2O3 compared with LaFeO3.
[5] A.E. Giannakas, A.K. Ladavos, P.J. Pomonis, Appl. Catal. B 49 (2004)
47.
6] A.A. Leontiou, A.K. Ladavos, P.J. Pomonis, Appl. Catal. A 241 (2003)
33.
7] V.C. Belessi, C.N. Costa, T.V. Bakas, T. Anastasiadou, P.J. Pomonis, A.M.
Efstathiou, Catal. Today 59 (2000) 347.
8] S.D. Peter, E. Garbowski, V. Perrichon, M. Primet, Catal. Lett. 70 (2000)
27.
9] F. Garin, L. Simonot, G. Maire, Appl. Catal. B 11 (1997) 181.
1
[
[
[
[
1
The reducibility of LaFeO3 was remarkably enhanced after
incorporation of Pd into the lattice, resulting in a reduction peak
◦
at a temperature as low as 78 C. This outstanding redox char-
[
[
[
10] S.D. Peter, E. Garbowski, V. Perrichon, B. Pommier, M. Primet, Appl.
acteristic of LaFe0.97Pd0.03O3 led to great performance for NO
reduction and CO oxidation at low temperatures.
Catal. A 205 (2001) 147.
11] M. Iwamoto, H. Yahiro, T. Kutsuno, S. Bunyu, S. Kagawa, Bull. Chem.
Soc. Jpn. 62 (1989) 583.
12] Y. Wan, J.X. Ma, Z. Wang, W. Zhou, S. Kaliaguine, J. Catal. 227 (2004)
242.
At low temperatures, the chemisorbed NO (nitrosyl species)
was dissociated over perovskites with the formation of both N2
and N2O, as well as an oxidized surface. This oxidized surface
subsequently can be reduced by CO with regeneration of anion
vacancies for the continuous reaction. N2O was further reduced
by CO, and this transformation became pronounced with in-
creasing temperature.
[13] R.D. Zhang, A. Villanueva, H. Alamdari, S. Kaliaguine, J. Catal. 237
2006) 368.
14] R.D. Zhang, A. Villanueva, H. Alamdari, S. Kaliaguine, Appl. Catal. B 64
2006) 220.
(
[
[
(
15] R.D. Zhang, A. Villanueva, H. Alamdari, S. Kaliaguine, Appl. Catal.
A 307 (2006) 85.
[16] N. Guilhaume, S.D. Peter, M. Primet, Appl. Catal. B 10 (1996) 325.
[17] Y. Nishihata, J. Mizuki, T. Akao, H. Tanaka, M. Uenishi, M. Kimura,
T. Okamoto, N. Hamada, Nature 418 (2002) 164.
NO decomposition was difficult over LaFe0.8Cu0.2O3 at
◦
temperatures below 500 C. In contrast, N2O could be decom-
◦
◦
posed at T > 350 C, achieving a N2 yield of 31% at 500 C
−1
and 50,000 h GHSV. Nevertheless, both NO and N2O trans-
formations were obviously promoted in the presence of CO.
O2 strongly inhibited N2 yield by suppressing the reducing
agent through unfavorable CO oxidation. As a result, the reac-
tion of NO and CO in the presence of 5% O2 was controlled by
CO oxidation.
[
[
[
18] Y.-H. Chin, W.E. Alvarez, D.E. Resasco, Catal. Today 62 (2000) 291.
19] M. Crespin, W.K. Hall, J. Catal. 69 (1981) 359.
20] S. Kaliaguine, A. Van Neste, Process for synthesizing perovskites using
high energy milling, U.S. patent 6017504, 2000.
[21] J. Shu, S. Kaliaguine, Appl. Catal. B 16 (1998) L303.
[22] S. Kaliaguine, A. van Neste, V. Szabo, J.E. Gallot, M. Bassir, R. Muzy-
chuk, Appl. Catal. A 209 (2001) 345.
NO reduction was previously studied over the same cata-
lysts but using propene as a reducing agent and in the presence
of oxygen [13], suggesting that NO reduction proceeded in-
[
[
[
23] V. Szabo, M. Bassir, A. van Neste, S. Kaliaguine, Appl. Catal. B 43 (2003)
81.
24] S. Royer, A. van Neste, R. Davidson, S. McIntyre, S. Kaliaguine, Ind. Eng.
Chem. Res. 43 (2004) 5670.
25] S. Varma, B.N. Wani, N.M. Gupta, Appl. Catal. A 241 (2003) 341.
−
volving NO species over the surface. However, the conclusion
3
−
obtained in the present work shows that nitrosyl species (NO )
is the main adsorbed reactant participating in the NO reduc-
tion by CO [Eqs. (10) and (11)]. The higher efficiency of CO
[26] P. Ciambelli, S. Cimino, L. Lisi, M. Faticanti, G. Minelli, I. Pettiti, P. Porta,
Appl. Catal. B 33 (2001) 193.
[27] P. Porta, S. Cimino, S. de Rossi, M. Faticanti, G. Minelli, I. Pettiti, Mater.
Chem. Phys. 71 (2001) 165.
(
[
compared with propene) in reducing the perovskite surface
Eqs. (7)–(9)] explains the difference between these two sys-
[28] W.J. Shen, M. Okumura, Y. Matsumura, M. Haruta, Appl. Catal. A 213
tems.
(2001) 225.
−
The surface concentration of O2 species in the presence of
propene and oxygen was higher (compared with CO alone),
thus resulting in a higher NO3 density over the surface. This
[29] L.G. Tejuca, J.L.G. Fierro, J.M.D. Tascón, Adv. Catal. 36 (1989) 237.
[30] T. Seiyama, N. Yamazoe, K. Eguchi, Ind. Eng. Chem. Prod. Res. Dev. 24
(1985) 19.
−
[
[
[
31] L.A. Isupova, A.A. Budneva, E.A. Paukshtis, V.A. Sadykov, J. Mol. Catal.
nitrate species was subsequently reduced by propene. The sur-
face density of nitrosyl species was lower (again compared with
CO in the gas phase), and nitrosyl species was also inactive
toward propene. The NO catalytic reduction mechanisms with
these two reducing agents were accordingly different.
A 158 (2000) 275.
32] E. Giamello, D. Murphy, G. Magnacca, C. Morterra, Y. Shioya, T. No-
mura, M. Anpo, J. Catal. 136 (1992) 510.
33] G. Centi, S. Perathoner, Appl. Catal. A 132 (1995) 179.
[34] X. She, M. Flytzani-Stephanopoulos, J. Catal. 237 (2006) 79.
[
35] B. Coq, D. Tachon, F. Figuéras, G. Mabilon, M. Prigent, Appl. Catal. B 6
1995) 271.
(
Acknowledgments
[
36] S. Shin, H. Arakawa, Y. Hatakeyama, K. Ogawa, K. Shimomura, Mater.
Res. Bull. 14 (1979) 633.
Financial support of NSERC through its industrial chair pro-
gram is gratefully acknowledged. The authors thank Nanox Inc.
for the preparation of the perovskite samples.
[37] V.A. Matyshak, O.V. Krylov, Catal. Today 25 (1995) 1.
[38] J.M.D. Tascón, L.G. Tejuca, C.H. Rochester, J. Catal. 95 (1985) 558.
[39] R.J.H. Voorhoeve, Advanced Materials in Catalysis, New York, 1977,
p. 129.
References
[40] B. Viswanathan, Catal. Rev.-Sci. Eng. 34 (1992) 337.
41] M.W. Chien, I.M. Pearson, K. Nobe, Ind. Eng. Chem. Prod. Res. Dev. 14
1975) 131.
[42] C. Shi, M. Cheng, Z. Qu, X. Yang, X. Bao, Appl. Catal. B 36 (2002) 173.
[43] Y. Teraoka, H. Fukuda, S. Kagawa, Chem. Lett. (1990) 1.
[44] H. Yasuda, N. Mizuno, M. Misono, J. Chem. Soc. Chem. Commun. (1990)
1094.
[
(
[
[
1] M.L. Rojas, J.L.G. Fierro, L.G. Tejuca, A.T. Bell, J. Catal. 124 (1990) 41.
2] L. Lisi, G. Bagnasco, P. Ciambelli, S. De Rossi, P. Porta, G. Russo,
M. Turco, J. Solid State Chem. 146 (1999) 176.
[3] G. Parravano, J. Chem. Phys. 20 (1952) 342.
[4] W.F. Libby, Science 171 (1971) 499.

R. Zhang et al. / Journal of Catalysis 242 (2006) 241–253
253
[
[
[
45] C. Tofan, D. Klvana, J. Kirchnerova, Appl. Catal. A 223 (2002)
[48] H.X. Dai, H. He, P.H. Li, L.Zh. Gao, C.-T. Au, Catal. Today 90 (2004)
231.
[49] P.J. Gellings, H.J.M. Bouwmeester, Catal. Today 12 (1992) 1.
[50] V.C. Belessi, P.N. Trikalitis, A.K. Ladavos, T.V. Bakas, P.J. Pomonis,
Appl. Catal. A 177 (1999) 53.
275.
46] L. Forni, C. Oliva, T. Barzetti, E. Selli, A.M. Ezerets, A.V. Vishniakov,
Appl. Catal. B 13 (1997) 35.
47] J. Valyon, W.K. Hall, J. Phys. Chem. 97 (1993) 1204.