Sie sind auf Seite 1von 10

Applied Catalysis A: General 262 (2004) 167176

Methane and propane combustion over lanthanum transition-metal


perovskites: role of oxygen mobility
M. Alifanti b , J. Kirchnerova a, , B. Delmon b , D. Klvana a
b

a Department of Chemical Engineering, Ecole Polytechnique, P.O. Box 6079, Station Centre-ville, Montreal, Canada QC H3C 2A7
Unit de Catalyse et de Chimie des Matriaux Divises, Universit Catholique de Louvain, Place Croix du Sud 2/17, B-1348 Louvain la Neuve, Belgium

Received 17 February 2003; received in revised form 24 November 2003; accepted 25 November 2003

Abstract
Catalytic hydrocarbon combustion is a technologically important, but still relatively poorly understood reaction. To shed more light on
the role of various physical and chemical characteristics of the catalyst on its activity for hydrocarbon combustion, La1x Srx M1y My O3
perovskites (M and M represent transition metals) were used as a model system. Four representative compositions were prepared and fully
characterized by different methods and their activity was determined in methane and propane combustion. Oxygen desorption reflecting
oxygen mobility is mainly a function of composition, more or less independently from specific surface area (SSA). On the other hand, the
results confirm that SSA is the important factor determining the high activity. Yet, in these oxide type catalysts (perovskites) the loss of activity
due to lower SSA, resulting from aging at elevated temperatures, seems, to a large part, be compensated for by fast oxygen mobility assured
in oxygen nonstoichiometric compositions.
2003 Elsevier B.V. All rights reserved.
Keywords: Perovskite catalysts; Catalytic hydrocarbon combustion; Oxygen mobility in oxide catalysts; Oxygen desorption characteristic;
La1x Srx M1y My O3 perovskites; Methane catalytic combustion; Propane catalytic combustion

1. Introduction
Catalytic combustion of hydrocarbons, especially methane, the main component of natural gas, is recognized as
an important process allowing the reduction of noxious NOx
[17]. The key to this process is a highly efficient catalyst.
Many materials show an interesting activity, but very few
sustain operation at high temperatures. The sensitivity of the
combustion catalysts to temperature, or their thermal stability, is related to the phase stability of the active component
and the resistance to the loss of active sites by sintering.
While all materials lose their activity on exposure to high
operation temperature, the onset, rate and the degree of this
activity loss varies greatly [2,3,7]. For example, supported
noble metal-based catalysts (especially palladium oxide)
exhibit an outstanding activity for combustion initiation, but
are particularly sensitive to operation at elevated temperatures starting at about 1100 K [36]. Palladium oxide loses

Corresponding author. Tel.: +1-514-3404711; fax: +1-514-3404159.


E-mail address: jitka.kirchnerova@polymtl.ca (J. Kirchnerova).

0926-860X/$ see front matter 2003 Elsevier B.V. All rights reserved.
doi:10.1016/j.apcata.2003.11.024

its activity mainly due to its decomposition, while platinum


suffers from sintering and volatilization [2,3,5,6]. On the
other hand, a variety of transition metal based mixed oxides
of specific structures, such as perovskites may withstand
substantially elevated temperatures [2,3,7,8], but the activity
of these oxide catalysts is lower, as demonstrated by higher
light-off temperatures. Development of a very versatile material, possessing a sufficient activity to initiate the reaction
at low temperatures and being thermally stable, remains an
important challenge. Indeed, material combining the two
important characteristics, i.e. very low (<700 K) light-off
temperatures and exceptional thermal stability, may perhaps
never be found. Thus, some trade-off between lower initiation activity and higher thermal stability must be accepted.
A survey of numerous experimental data available in
the literature indicates that some structures, in particular
lanthanum transition metal based perovskites, exhibit substantially better thermal stability than the parent transition
metal oxide in addition to their high activity. Indeed, some
compositions may even compete in activity with supported
platinum [912]. These perovskites own their excellent oxidation activity to a unique crystal structure in which the

168

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176

larger, body centered cation (La3+ , Sr2+ ) is twelve fold


coordinated by oxygen, whereas the smaller (transition
metal) cation located at the corners of a cube is octahedrally
coordinated by six oxygen ions. This structure permits to
accommodate a wide variety of metal ions of different valences and has an unusual capacity to accept a number of
different types of defects [10]. Depending on composition,
this structure, unlike any other, gives rise to properties important in heterogeneous catalysis such as high electronic
and ionic conductivity and excellent capacity for reversible
oxygen sorption. Especially invaluable is the ability of the
structure to stabilize mixed-valence, as well as unusual
valence states of useful, catalytically active metals such as
cobalt, manganese, nickel, iron, or copper.
Catalytic oxidation of methane and other hydrocarbons
has been studied for decades [1,13,14], but the mechanism
of their combustion (total oxidation) is still relatively poorly
understood. While it is well accepted that the hydrocarbon
activation proceeds via a homolytic abstraction of hydrogen
by some surface bound oxygen species [7,1517], the exact
nature of this species, the sequence of consecutive steps
and the role of principal physico-chemical properties of the
catalyst remain rather vague. In particular, no consensus
about the role of oxygen mobility, often assumed as being
reflected by reversible O2 -TPD has been established yet,
although a number of studies relating the O2 -TPD data to
activities has been published [11,1820]. Perovskite oxides
are not only very good complete oxidation catalysts, but may
also serve as a model for mechanistic studies [10,12,20].
This contribution deals with the activity in methane and
propane combustion of four La1x Srx M1y My O3 perovskites in relation to their Specific Surface Area (SSA)
and oxygen desorption characteristics. All selected compositions, a priori expected to be good catalysts, comprise transition metal (M, M ) oxides that are well known
to confer a high oxidation activity through their easy
change between two oxidation states and a certain degree
of oxygen nonstoichiometry (Co, Ni, Cu). Two of the
compositions (La0.87 Sr0.13 Mn0.2 Ni0.8 O3 (LSMN) and
La0.8 Sr0.2 Cu0.15 Fe0.85 O3 (LSCuF)) previously studied in
nitric oxide decomposition [21] were prepared with higher
SSA. The other two perovskites (La0.66 Sr0.34 Ni0.2 Fe0.8 O3
(LSNF) and La0.66 Sr0.34 Co0.2 Fe0.8 O3 (LSCF)) are known
as excellent mixed electronic and ionic conductors [22] and
were considered as potentially more thermally stable. All
samples were characterized by XRD, XPS, O2 -TPD anal-

ysis and SSA measurements. The aim of the study was to


elucidate the role of different physical and chemical characteristics in the activity for hydrocarbon combustion. Comparison of activities of the mentioned catalysts in methane
versus propane combustion taking place at different temperatures serves very well this purpose.

2. Experimental
2.1. Catalyst preparation
Powders of the four La1x Sr x M1y My O3 perovskites included in this study (LSMN, LSCuF, LSNF and LSCF, compositions given in Table 1) were prepared by a new method,
which is based on reactive precursor slurry of metal hydroxides processed by spray-freezing/freeze-drying [23]. The
precursor slurry was obtained by suspending under vigorous
stirring lanthanum hydroxide paste in a solution of metal
nitrates. The fluid paste of lanthanum hydroxide was prepared by wetting pure lanthanum oxide with a solution of
strontium nitrate. All components were mixed in strictly stoichiometric ratios. The smallest amount of water necessary
to dissolve all nitrates and to give fluid, but sufficiently stable suspensions for further processing, was used. The suspension was quickly frozen by spraying or by direct pouring
in liquid nitrogen. The frozen material was then separated
from the liquid nitrogen and dried under vacuum on a commercial freeze-drier. The dry precursor mixture was crushed
when needed and calcined in air under controlled conditions,
typically 12 h at 863 K followed by 4 h at 923 K. To prepare LSCuF perovskite, iron hydroxide was first precipitated
from ferric nitrate solution by using ammonia solution and
washed to remove soluble salts. The washed precipitate was
then mixed with other components and the resulting slurry
processed as described previously [21]. This sample was additionally aged 5 h at 973 K. To prepare LSCF and LSNF,
fine powder of red iron oxide (SSA = 10 m2 /g) was used as
part (80%) of iron precursor, using ferric nitrate nanohydrate
for the remaining part. Due to the anticipated lower reactivity of the precursor mixture, the initial calcination regime
was followed by 16 h at 1135 K, to assure a complete formation of perovskite phase. Finally, these two samples were
aged 6 h at 1223 K. Alumina or magnesia crucibles were
used to hold the powders during calcination. Generally, at
least 20 g of powder was prepared in a single batch. Before

Table 1
Preparation conditions and SSA of the studied perovskites
Composition

I.D.

precursors

Calcination + Aging (K/h)

SSA (m2 /g)

La0.87 Sr0.13 Mn0.2 Ni0.8 O3


La0.8 Sr0.2 Cu0.15 Fe0.85 O3

LSMN
LSCuF

863/12 + 900 /6
863/12 + 900/6 + 973/4

12.7
10.1

La0.66 Sr0.34 Ni0.2 Fe0.8 O3


La0.66 Sr0.34 Co0.2 Fe0.8 O3

LSNF
LSCF

La2 O3 ; Sr, Mn, Ni nitrates solution


La2 O3 ; Sr, Cu, Fe nitrates solution
Fe(OH)3 preliminary precipitation
La2 O3 ; Sr, Ni nitrate solution, red-Fe2 O3
La2 O3 ; Sr, Co nitrate solution, red-Fe2 O3

873/16 + 923/6 + 1135/16 + 1223/6


873/16 + 923/6 + 1135/16 + 1223/6

3.7
3.4

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176

use, all samples were lightly milled to obtain a fine powder


with particle size smaller than 5 m.
2.2. Catalyst characterization
Formation of perovskite phase was confirmed by X-ray
powder diffractometry (XRD) using a Philips XPert diffractometer with Cu K radiation.
SSA of each sample was determined on a Micromeritics
Flow Sorb II 2300 apparatus by using a single-point BET
method with a 30% N2 /70% He gas mixture as the adsorbate.
XPS spectra were recorded at room temperature and under a vacuum of 107 Pa on a SSX-100 Model 206 Surface
Science Instrument spectrometer with monochromatized Al
K radiation (h = 1486.6 eV). The charge neutralization
was achieved using an electron flood-gun adjusted at 10 eV
and placing a Ni grid 3 mm above the sample. Atomic composition of the surface was calculated using the sensitivity
factors (Scofield) provided by the instrument software. For
deconvolution, peaks were considered to be Gaussian for
85% and Lorentzian for 15%. The baseline was considered
linear and tangent to the peak wings. The charge correction
was made considering that the C1s signal of contaminating
carbon (CC or CH bonds) was centered at 284.8 eV [24].
2.3. Thermo-programmed desorption of oxygen (O2 -TPD)
O2 -TPD measurements were performed at atmospheric
pressure in a conventional installation. 0.05 g of pelletized
catalyst powder was loaded in a commercial U-shape quartz
microreactor having a reaction zone of 1 cm diameter with
inserted frit to hold the catalyst and a thermocouple well
pointing to the center of the catalyst bed. The pretreatment
procedure comprised heating the catalyst from room temperature to 973 K in a flow of 5% O2 in helium at a rate of
10 K/min, maintaining 30 min at 923 K, followed by cooling to room temperature. The gas flow (45 ml/min) was then
switched to pure He and the system was thoroughly flushed
with helium for 2 h. For the measurements, the temperature
was raised to 973 K at 10 K/min while the oxygen evolution
was followed by a Hewlett-Packard G1800A gas chromatograph equipped with a quadrupole mass-spectrometer. Calibration was made by injecting different volumes of certified
5% oxygen in helium (Air Liquid). It was indirectly confirmed by measurement on the LaMnO3.123 perovskite and
compared well with other work [24,25].
2.4. Catalytic activity measurements
2.4.1. Methane combustion
Catalytic activity in the combustion of methane in air
was evaluated using 0.10 g pelletized catalyst broken to
40100 m particle size, loaded in a U-shape quartz microreactor having the same design as the one used for TPD
measurements [24]. The reactor was operated in a down-flow
mode at atmospheric pressure. Prior to each evaluation, the

169

catalyst was conditioned 2 h at 973 K under flow of air and


than cooled down to 573 K. For actual activity determination, 1 vol.% methane in air flowing at 75 ml/min (i.e. specific gas flowrate 45,000 ml/gcat h) was admitted and the reactor was heated up to 973 K at a rate of 2 K/min The inlet
and outlet gas compositions were followed using an on-line
Delsi 2000 GC, equipped with a Carbosphere packed column and TCD. Helium was used as a carrier gas at a flow
of 25 ml/min with the oven temperature set at 423 K. The
only reaction products were CO2 and H2 O.
2.4.2. Propane combustion
Activity in the catalytic combustion of propane in air
was determined at a steady state using an integral U-shape
plug-flow reactor consisting of a 30 cm long stainless steel
tube, 0.7 cm i.d., with 0.5 g catalyst powder dispersed in
7 ml of precalcined pumice (particle size 300500 m, SSA
1 m2 /g). The feed mixture of 0.5 or 1% propane in air
flowed across the catalytic bed at 400 or 200 ml/min, respectively. The temperature, increased in steps of 25 degrees,
was monitored by two K-type thermocouples touching the
front and the end of the catalytic bed. The effluents, after
being stripped of water by passing over a drying agent, were
analyzed by gas chromatography using a Porapak Q separation column to determine concentrations of carbon dioxide
and propane; molecular sieve 5 column was used to determine the presence of carbon monoxide, which, however,
was never detected. Carbon dioxide and water were the only
products. Typically it took about thirty minutes to reach a
steady temperature and stable conversion [26].

3. Results
3.1. Physical characteristics and the oxygen TPD
Based on powder XRD analysis run between 10 and
60 2, all four catalysts calcined at conditions indicated in
Table 1 were nearly pure perovskites [23]. Diffractograms
of LSCuF and LSCF showed only peaks of perovskite phase
[21,23]. Diffractograms of LSMN and LSNF showed, in addition to prominent peaks of perovskite phase, traces (less
than 5% of the peaks corresponding to perovskite phase)
of secondary intermediate phases. The SSA varied between
12.7 and 3.4 m2 /g, reflecting the conditions of the heat treatment. XPS analysis carried out on as-prepared samples indicated that only LSMN perovskite had the surface cation
population close to the anticipated bulk composition. The
surface of the three remaining samples was significantly enriched by A-site cations (Sr2+ and La3+ ) with respect to
B-site cations, and in particular by Sr2+ (Table 2). This enrichment, often reported in the literature [10,20,24], is especially important in LSNF and LSCF, the two compositions
with the high nominal strontium content. For the three iron
containing perovskites, the surface Sr2+ enrichment clearly
seems to correlate with the overall strontium concentration.

170

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176

Table 2
XPS surface composition of the studied La1x Sr x My M1y O3 perovskites
[La + Sr]/
[M + M ]

[La]/
[La + Sr]

M /
[M + M ]

Olat /Oads

La0.87 Sr0.13 Mn0.2


Ni0.8 O3
La0.8 Sr0.2 Cu0.15
Fe0.85 O3
La0.66 Sr0.34 Ni0.2
Fe0.8 O3
La0.66 Sr0.34 Co0.2
Fe0.8 O3

1.03

0.71

0.18

0.53

2.16

0.66

0.31

1.01

1.95

0.37

0.34

0.67

2.28

0.39

0.24

1.02

Relative intensity

Composition

LSCF
LSNF
LSMN
LSCuF

One more characteristic worth of noting concerns the ratio


of the minor versus the major B-cation (column 4, Table 2),
which in LSCuF and LSNF is significantly higher than expected. This feature may possibly reflect the relative difficulty of copper and nickel to form trivalent cations. It is
conceivable that formation of ABO3 perovskite phase proceeds via A2 BO4 phases, such as La2 CuO4 . Even though
XRD of LSCuF showed only peaks of an ABO3 perovskite
phase, the presence of La2 CuO4 , especially at the surface
cannot be excluded. Another interesting result of XPS is the
significantly higher population of adsorbed oxygen on the
surface of the two nickel containing compositions (LSMN
and LSNF), with respect to that of the other two samples.
This feature seems characteristic of trivalent nickel that exhibits rather weak bonds with oxygen, even when stabilized
in the perovskite structure. Oxygen desorption curves are
shown in Fig. 1. Note, however, that the traces do not completely represent the total (integrated) amount reported in
Table 3. Actually, once the maximum temperature of 973 K
was reached, the oxygen evolution was followed, at this temperature, until the evolution virtually ceased, typically for
additional 30 min. As often observed, the evolution takes
place in two clearly distinguishable peaks suggesting a presence of at least two energetically different oxygen species,
below and above 690 K, respectively. (In fact, the trace below 690 K for LSCF shows two distinguishable low tem-

300

400

500

600

700

800

900

1000

Temperature [ K ]
Fig. 1. Traces of TPD-O2 for the four studied perovskites.

perature peaks (one before the last line of Table 3)). The
temperature of the first peak of oxygen evolution increases
in the order of LSCuF < LSMN LSNF < LSCF, following the anticipated strength of oxygen bond in the less
stable oxide of the (trivalent) B-site metal. The integrated
amounts of oxygen corresponding to a given temperature
range (ambient to 690 K and 690973 K) are also given in
Table 3. These quantities agree well with those obtained
previously for similar nonstoichiometric perovskite compositions [19,27,28]. Nevertheless, for all samples, the amount
of desorbed oxygen, regardless of the temperature range, is
more than an order of magnitude higher than that typically
observed for single transition metal oxides [10,15,18], or for
stable, oxygen stoichiometric perovskites such as LaCoO3
[24]. For example, Iwamoto et al. [18], for temperatures up
to 833 K, obtained 8.4 and 6.4 mol/g in the case of Co3 O4
and MnO2 , respectively, and 1.08 and 0.99 mol/g for CuO
and Fe2 O3 . From LaCoO3 only 6.2 mol/g desorbed below
683 K, with the total amount up to 973 K of 18.2 mol/g
[24]. There is a general tendency to relate the desorbed oxy-

Table 3
Composition and oxygen desorption characteristics of perovskites oxides
Composition

La0.87 Sr0.13 Mn0.2 Ni0.8 O3


La0.8 Sr0.2 Cu0.15 Fe0.85 O3
La0.66 Sr0.34 Ni0.2 Fe0.8 O3
La0.66 Sr0.34 Co0.2 Fe0.8 O3
La0.66 Sr0.34 Ni0.3 Co0.7 O3 f
a
b
c
d
e
f

0.065
>0.100
0.17
0.17
0.17

M.W.b

238.1
233.7
225.9
225.9
228.3

SSA (m2 /g)

12.7
10.1
3.7
3.4
10.2

O2 -TPD (Tmax , Kc )

524
513
543
578/613
585f

mol O2 /g
<690 K

>690 K

Total observed

Total calculatedd

28
75
120
80
165f

107
195
240
240
145f

135
270
360
320
310f

136
214 (374)e
376
376
372f

Based on the strontium content.


Assuming that all oxygen vacancies are occupied, i.e. = 0.
Temperature of the first desorption peak.
Assuming that all oxygen occupying the vacancies is desorbed.
The value in the bracket includes the oxygen associated with copper ( = 0.1 + 0.075).
Data from reference [27]; sample equilibrated at 773 K, TPD at 20 K/min to only 930 K.

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176


100

8.0
LSCF

90

7.0

LSNF
LSCuF

80

ln k [ mol/g s bar ]

Methane conversion [ % ]

171

LSMN

70
60
50
40

6.0
5.0
4.0
3.0
2.0

LSCF
LSNF
LSCuF

30

1.0

LSMN
reg. <823K

20

0.0
0.0010

10

0.0011

0.0012

0.0013
-1

Temperature

0
550

650

750

850

950

0.0014

-1

[K ]

1050

Temperature [ K ]
Fig. 2. Conversion of 1% methane in air over four perovskites as a
function of temperature; 0.1 g catalyst, 75 ml/min, heating 2 /min. The
points correspond to the experimental results and the lines are calculated
according to the first-order rate model using the kinetic parameters derived
from the data of Fig. 3 and given in Table 4.

gen to surface adsorption and consequently to report them


per surface area [15,18], disregarding thereby the possible
contribution due to bulk oxygen evolution. However, the
data in Table 3, or similar ones available in the literature
[27,28], clearly indicate that for a number of materials SSA
seems to have little, or only a secondary effect on the total amount of evolved oxygen. Indeed, from the samples
used in this study with lower SSA (LSNF and LSCF), more
oxygen evolved than from the other two having SSA three
times higher. On the other hand, the total amount of desorbed oxygen appears approximately proportional to based
on the strontium content (last column of Table 3) as suggested by a good agreement between experimental and calculated values. Thus, the amount of oxygen evolved during
the TPD includes not only that adsorbed at the surface, but
also the relatively weakly bound, highly mobile bulk oxygen. In the present case (perovskites) this bulk oxygen is
directly related to the nominal (based on strontium substitution) number of oxygen vacancies that can be reversibly
occupied.

Fig. 3. Arrhenius plots for catalytic combustion of 1% methane in air


over four perovskites, based on data shown in Fig. 2. The plots include
data for up to 90% conversion.

3.2. Activity in methane combustion


The high activity in methane combustion of the four studied perovskites is illustrated in Figs. 2 and 3. Linear Arrhenius plots (Fig. 3) corresponding to a simple first-order rate
model applied by assuming an integral mode, were obtained
for the whole range of conversions up to 90%. The lines
traced on Fig. 2 were calculated by using kinetic constants
obtained by linear regression of Arrhenius plots (Fig. 3)
and summarized in Table 4. For comparison, Table 4 includes previously published activity data, determined under steady state conditions, for La0.66 Sr0.34 Ni0.3 Co0.7 O3
(LSNC) prepared by the same method as the LSMN sample
[23,29]. The simple first-order model gives clearly a very
good fit. Similarly good agreement was observed for a given
set of experimental data in the previous work concerning
other catalysts or perovskite compositions [7,19,24]. However, the data for LSMN give sign of some change of kinetics at around 823 K. While, for this catalyst, all data could
be satisfactorily fitted by a single line, a better fit is obtained
when separating the data in two sets covering temperatures
<823 and >823 K. Only the line for data below 823 K is
shown in Fig. 2. Activities of all present samples are apparently superior to those of some commercial supported

Table 4
Activity in methane combustion of different perovskites
Catalyst

SSA (m2 /g)

T50 (K)

Eapp (kJ/mol)

ln A (mol/g s bar)

k723 (mol/g s bar)b

k823 (mol/g s bar)b

LSMN
LSCuF
LSNF
LSCF
LSNCa

12.7
10.1
3.7
3.4
10.1

800
820
880
885
795

81
90
100
95
92

18.05
19.17
19.65
18.87
19.58

97
67
20
22
72

475
410
154
145
460

a
b

La0.66 Sr0.34 Ni0.3 Co0.7 O3 prepared similarly as LSMN and calcined 12 h at 863 K + 4 h at 900 K [23].
Pseudo first-order kinetic constant at 723 and 823 K, respectively.

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176

platinum catalysts [29], in agreement with the early study by


Arai et al. [9]. The two higher SSA perovskites, LSMN and
LSCuF, show activities comparable to that of LSNC [23] but
somewhat lower than La0.9 Ce0.1 CoO3 , having the same SSA
and LaMnO3 and La0.9 Ce0.1 MnO3 , with SSA of 16.5 and
32.2 m2 /g, respectively, the latter three perovskites prepared
by the citrate method and aged at 973 for a short period of
5 h [24,30]. Although at higher conversions (temperatures)
the activities of LSMN and LSCuF are nearly identical, the
nickelmanganese based perovskite shows measurably better activity below about 830 K. This is consistent with the
known high activity of manganese and nickel oxides and corresponding perovskites [11,30,31]. The earliest onset of catalytic combustion (with respect to temperature) over LSMN
possibly also relates to the presence of the most labile oxygen species evidenced by the earliest onset of oxygen evolution, in comparison with the other three perovskites. As
was expected, the two perovskites of similar composition
aged for a relatively long time at higher temperatures (16 h
at 1135 K and 6 h at 1223 K), namely, samples LSNF and
LSCF, had smaller SSA and their nearly identical activity
was correspondingly lower than that of LSMN and LSCuF.
However, with respect to the degree of aging, the apparent activity of LSNF and LSCF is actually very good and
compares well with apparent activities of some other similarly aged perovskites, or is even better. For example, the
activity at 723 and 823 K (k723 and k823 ) of La0.8 Sr0.2 FeO3
aged 22 h at 1173 K was 16 and 135 mol/g s bar, respectively [32], whereas that of La0.8 Sr0.2 MnO3 supported on
NiAl2 O4 calcined at 1223 K for an unspecified time and
having SSA = 22 m2 /g was 18 and 120 mol/g s bar, respectively [33]. Activity of LSNF and LSCF is certainly
higher than that of Co3 O4 prepared by the citrate method and
aged 5 h at 1083 K, for which the k823 was 126 mol/g s bar
[24]. It can also be noted that the apparent activity of the
LSCuF sample determined by the method described in our
work (dynamic heating) agrees very well with that determined under a steady state and under different conditions
[34].
3.3. Activity in propane combustion
In view of preceding studies indicating a strong dependence of the apparent activity on the flowrate [26,35],
catalytic propane combustion was evaluated at 400 and
200 ml/min with 0.5 and 1% propane in air, respectively.
This provided enough information allowing a plausible
speculation about the mechanism of propane catalytic combustion in spite of the relatively small number of experiments reported here. In addition, a sample of red-Fe2 O3
used in the preparation of LSNF and LSCF perovskites
and having an SSA equal to 10 m2 /g was included among
the catalysts for comparison. Although the activity of this
red-Fe2 O3 was appreciable, it was considerably lower than
that of LSMN and LSCuF. In fact, below 700 K red-Fe2 O3
was even less active than LSNF and LSCF. Furthermore,

100
LSMN

90

LSCuF
LSNF

80
Propane conversion [ % ]

172

LSCF
red-Fe2O3-cal

70
60
50
40
30
20
10
0
400

500

600

700

800

Temperature [ K ]

Fig. 4. Steady state conversion of 0.5% propane in air (400 ml/min)


over 0.5 g catalyst as a function of temperature. Points represent experimental data; dotted lines are calculated using kinetic parameters for the
pseudo-first-order model given in Table 5. Full line represents the calculated conversions for red-Fe2 O3 .

when calcined 12 h at 1113 K, this sample lost more than


90% of its initial activity, the SSA decreasing to 1.9 m2 /g.
This observation further demonstrates the superior thermal
stability of perovskites with respect to the simple corresponding transition metal oxides. Fig. 4 illustrates the
conversion data for 0.5% propane, 400 ml/min, and 0.5 g
catalyst, where, similarly as in the case of methane, the
lines represent conversions calculated by using kinetic parameters for a simple pseudo-first-order model given in
Table 5. Note that for clarity, for red-Fe2 O3 only calculated
conversions are shown in Fig. 4. Basically identical light-off
curves were obtained for 200 ml/min (1% propane), over
the same catalyst amount, which means a nearly two times
lower apparent (pseudo-first-order) activity, as demonstrated
by the data in Table 5. For the lower flowrate (and higher
propane concentration) the activity is represented in terms
of Arrhenius plots of the pseudo-first-order kinetic constant
in Fig. 5. The observed behavior of propane combustion, in
agreement with previous work [26,35], contrasts with that
of methane, for which the light-off curves typically shift
to lower temperatures as the flowrate decreases, regardless of concentration, and the apparent first-order kinetic
constants are only slightly dependent on the flowrate [29].
Such behavior seems consistent with the Mars van Krevelen
type mechanism involving two consecutive (controlling)
rate steps: that of propane oxidation resulting in transient
catalyst surface reduction, and that of catalyst (surface)
reoxidation. Indeed, assuming a Mars van Krevelen model
for propane combustion [26]:
r=

kp Pp ko Po0.5
5kp Pp + ko Po0.5

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176

173

Table 5
Apparent activity in propane catalytic combustion
SSA (m2 /g)

T50 (K)

Eapp (kJ/mol)

ln A (mol/g s bar)

k550 (mol/g s bar)b

k600 (mol/g s bar)b

0.5 g, 0.5% propane, 400 ml/min


LSMN
12.7
LSCuF
10.1
LSNF
3.7
LSCF
3.4
red-Fe2 O3
10
8.8
LSNCFa

609
643
685
685
693
637

101
98
93
91
119
88

25.91
24.28
22.39
22.02
26.71
22.67

46
17
8
8
2
31

288
104
45
44
19
156

0.5 g, 1% propane,
LSMN
LSCuF
LSNF
LSCF
red-Fe2 O3

609
643
685
685
693

91
97
90
88
121

23.05
23.53
21.19
20.87
26.41

24.3
9.5
4.8
4.9
0.9

127
56
25
25
8

Catalyst

200 ml/min
12.7
10.1
3.7
3.4
10

a La Sr Co
0.7 0.3
0.67 Ni0.29 Fe0.04 O3 , recently prepared for some other studies and aged 8 h at 973 K in addition to initial calcination 12 h at 863 K + 4 h
at 900 K.
b Pseudo first-order kinetic constant at 550 and 600 K, respectively.

the difference between the two apparent first-order constants


(Table 5) is qualitatively understandable on the grounds of
the following expressions:
k1 =

kp

(5kp Pp /ko Po0.5 ) + 1

k1 =

kp
(5 kp Pp /ko Po0.5 ) + 1

and

where k1 and k1 correspond to data based on 0.5 vol.%


propane (Pp = 0.005) and 1 vol.% (Pp = 0.01), respectively. In principle, the value of the k1 /k1 ratio should not
exceed that of Pp /Pp , i.e. 2. With the exception of red-Fe2 O3
this is satisfied. It is also worth noting that while the order
of activities of the individual catalysts is similar to that in
methane combustion and again LSNF and LSCF show the
same activity, the ratios of these activities are somewhat dif8
LSMN

LSCuF

ln k1 [ mol / g s bar ]

LSNF

LSCF
red-Fe2O3

5
4
3
2
1
0
0.0012

0.0014

0.0016
-1

Temperature

0.0018

0.002

-1

[K ]

Fig. 5. Arrhenius plots for catalytic combustion of 1% propane in air


(200 ml/min) over 0.5 g catalyst. Lines represent linear regressions.

ferent. In particular, the larger difference between the activity of LSMN and LSCuF possibly reflects the higher population of surface weakly bound oxygen species suggested
by XPS (Table 2).
The comparatively low activity of red-Fe2 O3 can undoubtedly be related to a rather small population of labile oxygen available at the surface as well as low oxygen mobility
[15,18], a situation dictated by the structure of this oxide
and the strength of the FeO bond in this oxide.

4. Discussion
When examining the reaction rate data for methane combustion over the four oxygen nonstoichiometric perovskites,
one is tempted to conclude that the single important factor
determining the activity is SSA. The effect of SSA on the
activity seems particularly prominent in the case of (oxygen) nonstoichiometric transition metal based perovskites
that, due to their composition, exhibit a good activity. Such a
conclusion would seem well illustrated by Fig. 6, where we
have plotted the apparent specific kinetic constants at 823 K
as a function of SSA, including data for three different samples of La0.66 Sr0.34 Co0.2 Fe0.8 O3 (LSCF-I) prepared from
different iron precursors and aged at different conditions
[36]. In fact, a similar trend is observed also in the case of
propane combustion, although not exactly to the same extent. Of course this emphasis on the importance of SSA,
supported by numerous other data, including those added
at the bottom of Table 4, prevails in the literature for many
years. Indeed, activity in hydrocarbon combustion has for
long been perceived by many authors as directly proportional
to SSA, as such has been expressed per area. However, on
closer analysis and especially from data of propane combustion, it becomes evident that the prominence of SSA should
be considered with some nuances. First of all, the data in
Fig. 6 demonstrate, in a similar way as in the earlier work

174

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176

by specific surface area. Nevertheless, it has to be accepted


that the extrinsic activity is also to some degree dependent
on the catalyst composition. The extent of contribution of
each of the two components may vary dramatically and will
greatly depend on selected preparation method and heat
treatment. Accepting this notion of intrinsic and extrinsic
activity provides an explanation for the failure of the direct
proportionality between activity (in methane combustion)
and SSA illustrated in Fig. 6 and previously observed or
discussed for other catalysts [23,39].
In view of available data and information, catalytic
methane combustion may be considered to proceed via the
following schematic steps, outlined in a previous work [7]:

600

k823 [mol/g s bar]

500
400
300
LSMN
LSCuF
LSNF
LSCF
regr.
LSCF-I
regr.I

200
100
0
0

10

12

14

S S A[m2/g]
Fig. 6. The apparent first-order kinetic constant for methane combustion over different perovskites at 823 K, as a function of SSA. Dotted
line represents regression for the four perovskites of this study (LSMN,
LSCuF, LSNF and LSCF), while full line is regression for the set of
LSCF-I perovskites. Note that LSCF and LSCF-I have the same nominal
composition.

[39], that activity (kinetic constant) is not directly proportional to SSA. While SSA clearly plays a vital role [7], other
catalyst composition related characteristics, particularly the
availability of surface oxygen and its overall mobility, may
become equally or even more important, as discussed for
the case of ceria substituted lanthanum cobaltates [24] and
manganites [31]. This is particularly true for materials aged
at higher temperatures (1100 K). Our O2 -TPD results also
strongly suggested that all oxygen available in catalysts can
play some role.
Catalytic combustion of methane and other hydrocarbons corresponds to a fairly complex mechanism, involving
several reaction steps, each apparently depending on specific parameters of operation and catalyst characteristic,
or their combination. Numerous experimental data available in the literature since several decades indicate that
only noble metals and transition metal oxide based catalysts show reasonable activity, which decreases in the order
PdO Pt > (Ag) > Co3 O4 > MnO2 > NiO CuO >
Cr2 O3 Fe2 O3 [13,14] for catalysts with reasonably large
(>10 m2 /g) SSA. This order in activity seems to correlate
with the strength of the metaloxygen bond. In fact, even
ceria (CeO2 ) when properly doped with, for example, CuO
[37], or in combination with MnOx [38] or Co3 O4 [24]
exhibits an excellent combustion activity.
The apparent activity of any given material depends
strongly not only on its composition and crystal structure
but also on the surface morphology and catalyst texture.
The overall activity may be perceived as due to a combination of two main components (factors): (a) intrinsic,
which is related to the composition and the corresponding
crystalline structure and (b) extrinsic, which is primarily
related to the texture and surface morphology, as reflected

1. H3 CH + [O] [Mn + Om ]
[H3 C ] + [OH] [Mn + Om ]
2. [H3 C ] + [Mn + Om ] [M(n 1)+ Om3 CO2 ]
+ [OH2 ]
3. H2 O and CO2 desorption (slow)
4. [M(n 1)+ Ox ] + O2 [O] [Mn + Om ]
In the first activation step, the initial CH bond is broken,
most likely homolytically. The initial, typically the slowest
or controlling, step is followed by presumably faster reaction. This reaction, step 2, is apparently of spillover type, or
at least involves a movement of the methyl (alkyl) radical
leading to the reaction with additional oxygen species favorably coordinated around a strong electron acceptor Mn + .
This step leads to surface bound carbon dioxide and water.
In many cases, desorption of these products, step 3, is comparatively slow, thereby inhibiting the overall rate of catalytic combustion [29]. Although at high enough temperatures the catalytic oxidation takes place even on oxides of
poor electron acceptors, for example MgO, the selectivity
to carbon dioxides, i.e. complete oxidation, tends to be low
[7]. Thus, the presence of [Mn + Om ] seems essential for fast
and complete oxidation with carbon dioxide and water as
the only products. Knowing that over transition metal based
perovskites the oxidation of methane and other hydrocarbons leads almost exclusively to carbon dioxide and water,
even in fuel rich mixtures [29], it is reasonable to consider
the octahedral coordination and (sufficiently) fast oxygen
mobility as the critical characteristics.
The overall kinetics will also depend on the availability
of the reactive oxygen species and its coordination at the
surface, regardless of its actual nature, as expressed by the
rate of step 4, r4 . It is easy to understand that r4 should be
related to the overall oxygen mobility. This is a function of
oxygen surface exchange and bulk oxygen ion conductivity
(diffusivity). The strength of the electron acceptor may be
related to the equilibrium governing the catalyst (surface)
reoxidation:
[M(n1) + Ox ] + O2 [O][Mn+ Om ]
On the other hand, the rate of electron transfer is likely
reflected by the electronic conductivity. Depending on the

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176

strength of the initial CH bond and consequently temperature dictating the rate of individual steps, the overall reaction
may be controlled either by the first step, or by the reoxidation step which involves oxygen dissociation. Typically, the
rate of reoxidation r4 is faster or comparable to the sum of
r1 and r2 .
It is obvious that the availability of appropriate oxygen
species and the rate of catalyst reoxidation are crucial in catalytic combustion. Although it is logical to relate the (surface) reoxidation to oxygen mobility, it is rather difficult to
assess the extent of its influence. Nevertheless, experience
has shown that above a certain temperature, in the case of
perovskites at about 650 K [20], when the oxygen mobility
becomes sufficiently fast (r4 r1 + r2 ), the rate of catalytic combustion is nearly oxygen independent. In fact, this
is typically observed in the case of methane. In contrast,
relatively facile catalytic oxidation of higher light hydrocarbons such as propane takes place at considerably lower
temperatures, at which the reoxidation (r4 ) is so slow that
it influences the combustion rate. Such a situation may be
accounted for by the Mars van Krevelen model [26] and is
apparently reflected also by data of this work (Table 5). In
view of these considerations, it is easy to understand that,
in this lower temperature region, the activity variation with
composition is typically more noticeable than at higher temperatures [9,19]. This indeed is also suggested by data in
Table 5. Propane combustion is clearly more dependent on
the characteristics related to nuances of composition, than
on SSA, in contrast to methane combustion. Inasmuch the
proposed mechanism is based on the analysis of available
data and appears highly plausible, more data might be necessary for its validation.
Overall, it is legitimate to conclude that the data of this
work serve as additional examples of the factors determining
the apparent catalyst activity in hydrocarbon combustion.
The composition of the perovskites studied in this work has
been selected in the hope to obtain highly active catalysts.
The results confirm this expectation. They also provide some
new information about the relative role of SSA and oxygen
mobility.

5. Summary and conclusions


A direct comparison of activities of selected representative
perovskites in methane and propane combustion provided a
good basis for the assessment of the relative importance of
the variety of intervening factors. In view of this work and
in view of the known characteristics of different methane
combustion catalysts it can be concluded that elevated activity (of La1x Srx M1y My O3 perovskites) in methane and
other hydrocarbon combustion stems from a combination
of electronic and structural characteristics of the material.
Among these, the magnitude of SSA definitely plays an important role. Very high oxygen mobility does not appear to
be comparatively as important in determining the activity at

175

intermediate temperatures, because in oxygen nonstoichiometric perovskites this mobility is always above the minimum value to be satisfied. Nevertheless, in aged catalysts, it
is apparently a high oxygen mobility that compensates for
the negative effect of lower SSA, i.e. number of active sites.
Thus, this characteristic is very important in aged, low SSA
catalysts and should be considered when designing catalysts
for high temperature catalytic combustion.
Acknowledgements
This work was in part financed by grants from Natural Sciences and Engineering Research Council of Canada.
The authors also thank to the European Community for an
additional financial support for the research under contract
ENV4-CT97-0599.
References
[1] R. Prasad, L.A. Kennedy, E. Ruckenstein, Catal. Rev.-Sci. Eng. 26 (1)
(1984) 1.
[2] M.F.M. Zwinkels, S.G. Jrs, P.G. Menon, Catal. Rev.-Sci. Eng.
35 (3) (1993) 319.
[3] R.A. Dalla Betta, Catal. Today 35 (1997) 129.
[4] R.E. Hayes, S.T. Kolaczkowski, Introduction to Catalytic Combustion, Gordon and Breach Science Publishers, 1998.
[5] J. Lee, D. Trimm, Fuel Process Technol. 42 (23) (1995) 339.
[6] P. Glin, M. Primet, Appl. Catal. B: Environ. 39 (2002) 1.
[7] J. Kirchnerova, D. Klvana, Catal. Lett. 67 (2000) 175.
[8] M. Machida, K. Eguchi, H. Arai, J. Catal. 123 (1990) 477.
[9] H. Arai, T. Yamada, K. Eguchi, T. Seiyama, Appl. Catal. 26 (1986)
265.
[10] L.G. Tejuca, J.L.G. Fierro, J.M.D. Tascn, Structure and Reactivity of
Perovskite-Type Oxides, in: D.D. Eley, H. Pines, P.B. Weisz (Eds.),
Advances in Catalysis, vol. 36, Academic Press, New York, 1989,
p. 237.
[11] J.G. McCarty, H. Wise, Catal. Today 8 (1990) 231.
[12] T. Seiyama, Catal. Rev.-Sci. Eng. 34 (4) (1992) 281.
[13] G.K. Boreskov, in: J.R. Anderson, M. Boudart (Eds.), Catalysis (Science and Technology), vol. 3, Springer-Verlag, Berlin, Heidelburg,
New York, 1982, p. 113.
[14] G.I. Golodets, Heterogeneous Catalytic Reactions Involving Molecular Oxygen, Studies in Surface Science and Catalysis, vol. 15, Elsevier, Amsterdam, 1983, pp. 437469.
[15] A. Bielanski, J. Haber, Oxygen in Catalysis, Marcel Dekker, New
York, 1991.
[16] E. Finocchio, G. Busca, V. Lorenzelli, R.J. Wiley, J. Chem. Soc.,
Faraday Trans. 90 (21) (1994) 3347.
[17] E. Finocchio, G. Gusca, V. Lorenzelli, R.J. Wiley, J. Catal. 151
(1995) 204.
[18] M. Iwamoto, Y. Yoda, N. Yamazoe, T. Seiyama, J. Phys. Chem. 82
(1978) 2564.
[19] P. Salomonsson, T. Griffin, B. Kasemo, Appl. Catal. A: Gen. 104
(1993) 175.
[20] M.A. Pea, J.L.G. Fierro, Chem. Rev. 101 (2001) 1981.
[21] C. Tofan, D. Klvana, J. Kirchnerova, Appl. Catal. A: Gen. 223 (12)
(2001) 275.
[22] K. Huang, H.Y. Lee, J.B. Goodenough, J. Electrochem. Soc. 145
(1998) 3220.
[23] J. Kirchnerova, D. Klvana, Solid State Ionics 123 (1999) 307.
[24] J. Kirchnerova, M. Alifanti, B. Delmon, Appl. Catal. A: Gen. 231
(2002) 65.

176

M. Alifanti et al. / Applied Catalysis A: General 262 (2004) 167176

[25] T. Nitadori, S. Kurihara, M. Misono, J. Catal. 98 (1986) 221.


[26] K.S. Song, D. Klvana, J. Kirchnerova, Appl. Catal. A: Gen. 213
(2001) 113.
[27] J.-P. Joly, D. Klvana, J. Kirchnerova, React. Kinet. Catal. Lett. 68
(1999) 249.
[28] H.M. Zhang, Y. Shimizu, Y. Teraoka, N. Miura, N. Yamazoe, J.
Catal. 121 (2) (1990) 432.
[29] D. Klvana, J. Kirchnerova, P. Gauthier, J. Delval, J. Chaouki, Can.
J. Chem. Eng. 75 (1997) 509.
[30] M. Alifanti, J. Kirchnerova, B. Delmon, Appl. Catal. A: Gen. 245 (2)
(2002) 231.
[31] M. Alifanti, R. Auer, J. Kirchnerova, F. Thyrion, P. Grange, B.
Delmon, Appl. Catal. B: Environ. 41 (2003) 71.

[32] C. Batiot-Dupeyrat, F. Martinez-Ortega, M. Ganne, J.M. Tatibout,


Appl. Catal. A: Gen. 206 (2001) 205.
[33] R. Burch, P.J.F. Harris, C. Pipe, Appl. Catal. A: Gen. 210 (2001) 63.
[34] C. Tofan, D. Klvana, J. Kirchnerova, Appl. Catal. B: Environ. 36 (4)
(2001) 311.
[35] D. Klvana, K.S. Song, J. Kirchnerova, Korean J. Chem. Eng. 19 (6)
(2002) 932.
[36] J. Kirchnerova, D. Klvana, Catal. Today 83 (14) (2003) 233.
[37] W. Liu, M. Flytzani-Stephanopoulos, J. Catal. 153 (1995) 304.
[38] D. Terribile, A. Trovarelli, C. de Leitenburg, A. Primavera, G. Dolcetti, Catal. Today 47 (1999) 133.
[39] J. Kirchnerova, J. Vaillancourt, D. Klvana, J. Chaouki, Catal. Lett.
21 (1993) 77.

Das könnte Ihnen auch gefallen