Promoting effects of Na and Fe impurities on the catalytic activity of CaO in the reduction of NO by CO and H2

Article abstract of DOI:10.1021/jp980944z

The heterogeneous reduction of NO by H₂ and CO over different CaO materials is investigated. The dependence of the specific NO reduction rate on the impurity content is demonstrated for both reducing species. The roles of two specific impurities, i.e., Na and Fe, as well as their combined effect are investigated. The apparent activation energies for the NO + CO and NO + H₂ reactions are determined for three different calcium oxides. Values between 26 and 28 kcal/mol are obtained. The influence of impurity content is found in the preexponential factor of the Arrhenius equation. A reaction mechanism based on a rate-determining surface-oxygen-abstraction step is suggested. This mechanistic understanding is explored to compare the activities of other alkaline-earth oxides. Particularly, a linear correlation between the apparent activation energy and the lattice parameter is observed.

Full text of DOI:10.1021/jp980944z

J. Phys. Chem. B 1998, 102, 5127-5134
5127
Promoting Effects of Na and Fe Impurities on the Catalytic Activity of CaO in the
Reduction of NO by CO and H2
Filip Acke* and Itai Panas
Department of Inorganic Chemistry, Chalmers UniVersity of Technology and Go¨teborg UniVersity,
S-412 96 Go¨teborg, Sweden
ReceiVed: January 29, 1998; In Final Form: April 16, 1998
The heterogeneous reduction of NO by H2 and CO over different CaO materials is investigated. The
dependence of the specific NO reduction rate on the impurity content is demonstrated for both reducing
species. The roles of two specific impurities, i.e., Na and Fe, as well as their combined effect are investigated.
The apparent activation energies for the NO + CO and NO + H2 reactions are determined for three different
calcium oxides. Values between 26 and 28 kcal/mol are obtained. The influence of impurity content is
found in the preexponential factor of the Arrhenius equation. A reaction mechanism based on a rate-determining
surface-oxygen-abstraction step is suggested. This mechanistic understanding is explored to compare the
activities of other alkaline-earth oxides. Particularly, a linear correlation between the apparent activation
energy and the lattice parameter is observed.
I. Introduction
et al.13 for Li-promoted MgO substrates. An initial increase of
the specific activity was observed with Li doping. The
Emissions of nitrogen oxides are a threat to the environment.
The use of catalytic materials to reduce NOx is one of the
techniques that is available to efficiently control these emissions.
Noble metals are an example of thoroughly investigated
materials that show a high activity toward NOx reduction. To
regenerate the active surface, the presence of reducing agents,
e.g., CO, H2, and hydrocarbons, has been shown to be
necessary.1 Model catalysts in the nano- and microscale2 and
single crystals3,4 are used in the search for a mechanistic
understanding of the reduction of NO. Transition-metal oxides
are another class of materials that have been shown to be
catalytically active for these reactions.5 Perovskites were
especially believed to be promising representatives,6,7 for which
lattice oxygens were suggested to participate in the reaction.8
The participation of lattice oxygens has also been discussed
for the oxidative coupling of methane over alkali-metal-doped
alkaline-earth oxides.9,10 In a previous work, this was shown
to even hold for the NO reduction over CaO.11 The activity of
CaO might be surprising because the solid has a filled valence
band and a bulk band gap of about 7 eV.12 Two models were
suggested to explain the involvement of promoters on the
participation of lattice oxygens in the reaction mechanism, i.e.,
a semiconductor model10 and a model based on F-type centers.9
In case of the former, alkali-metal ions are introduced into an
alkaline-earth oxide to provide electron acceptors in the band
gap, while in the latter they are introduced to stabilize F centers.
A mechanism was suggested based on an oxygen-abstraction
step and a reoxidation of the surface.9,10 The activation energy
for the oxygen-abstraction step using CO as the reducing agent
is 25 kcal/mol. The reoxidatation of the surface, using NO as
the oxidant, displays a 10 kcal/mol upper limit for the apparent
activation energy.11
importance of the impurity level as well as the chemical
surrounding of the impurity has been discussed for iron in
quartz.14 It was shown that Fe as an impurity is not catalytically
active when iron is incorporated in a well-defined crystal
structure (illite). To be active, the electronic structure had to
be affected or crystal defects produced.
In this article, a comparison of the catalytic activities will be
presented for different CaO materials toward the reduction of
NO with CO and H2. It will be shown that the different
impurities do not significantly affect the apparent activation
energy for the NO reduction with H2 but that the importance of
the impurities has to be found in the preexponential factor of
the Arrhenius equation, i.e., the amount of defects produced. It
will be shown that the results for H2 as a reducing agent agree
with those of CO after correction for CO2 poisoning. The effect
of Na and Fe impurities on the reduction of NO with H2 will
be carefully investigated by the preparation of CaO powders
containing equimolecular amounts of Na, Fe, and Na + Fe.
II. Experimental Section
II.1. Preparation of the CaO Powders. The tested powders
are listed in Table 1. They can be divided into three groups:
pure, as received, and promoted CaO powders. The purification
step, for the pure materials, is based on the low solubility of
CaCO3 in water. A CaO powder was first dissolved in milli-Q
water, and the dissolved Ca2+ ions were subsequently precipi-
tated as a carbonate by bubbling through CO2. This solution is
then filtered, and the precipitate is heat-treated at 950 °C for
2.5 h in an oxygen flow, resulting in the purified CaO. Other
carbonates such as Na2CO3 are much more soluble and will
stay in solution.
A dependence on the amount of dopant of the catalytic
activity for NO reduction with methane was shown by Zhang
Three different types of promoted CaO materials were
prepared by doping with Na, Fe, and Na + Fe. The promoters
were added as sodium carbonate and/or iron acetate to a calcium
acetate solution. By evaporation of the water, a promoted
* Corresponding author. Telephone: +46 (0)31 772 28 86, Fax: +46
(0) 31 772 28 53. E-mail: filip@inoc.chalmers.se.
S1089-5647(98)00944-4 CCC: $15.00 © 1998 American Chemical Society
Published on Web 06/06/1998

5128 J. Phys. Chem. B, Vol. 102, No. 26, 1998
Acke and Panas
TABLE 1: (a) Iron and Sodium Contents, and Specific Surface Areas of the Investigated CaO Materials and (B) Added
Amounts of Iron and Sodium (Both as wt % and mol %), as Well as Specific Surface Areas of the CaO Materials Prepared for
the Detailed Analysis of the Na, Fe, and Na + Fe Contents on the Catalytic Activities of CaO
Section A
AAS, wt %
material
Fe
Na
BET, (m2/g)
CaO/Fisher Scientific
CaO/Ph DAB pieces
CaO/Ph DAB pieces
CaO from Ca(OH)2/Merck
CaO from Ca(OH)2/Merck
CaO from Ca-acetate Fisher Scientific
Na doped CaO
0.052
0.029
0.004
0.042
0.004
0.003
0.003
0.003
0.003
0.008
0.010
0.011
0.004
0.014
0.003
0.005
0.200
0.508
0.470
0.268
2.3
1.6
9.2
as received
purified
as received
purified
4.9
14.4
11.6
0.8
0.6
0.4
label 1
label 2
label 3
label 4
0.6
Section B
added amount, wt %
Fe Na
added amount, mol %
material
Na 0.001
Na 0.004
Na 0.007
Fe 0.001
Fe 0.004
Fe 0.007
Fe
Na
BET, m2/g
0
0
0
0.12
0.37
0.74
0.12
0.37
0.74
0.05
0.15
0.30
0
0
0
0
1.24 × 10-3
3.72 × 10-3
7.44 × 10-3
0
7.3
3.7
2.7
8.1
7.4
7.2
4.6
3.1
1.7
1.24 × 10-3
3.72 × 10-3
7.44 × 10-3
1.24 × 10-3
3.72 × 10-3
7.44 × 10-3
0
0
0
0
Na + Fe 0.001
Na + Fe 0.004
Na + Fe 0.008
0.05
0.15
0.30
1.24 × 10-3
3.72 × 10-3
7.44 × 10-3
calcium acetate was obtained. Calcination of the acetate was
done in an oxygen flow at 950 °C for 2.5 h. To make clear
that the change in catalytic activity was due to the addition of
dopants and not to the treatment, a blank was prepared. The
Na-promoted materials can be divided in two groups, i.e., the
ones labeled 1-4 (Table 1A) and the ones labeled 0.001, 0.004,
and 0.007 (Table 1B). The latter are part of a group of materials
that were prepared to clarify the possible synergestic effects
between Na and Fe impurities. These materials contained
equimolecular amounts of Na, Fe, and Na + Fe.
II.2. Specific Surface Area. The specific surface areas of
the investigated materials were determined by nitrogen adsorp-
tion-desorption with a Digisorb 2600 instrument (Micromet-
rics). The BET surface area determinations were based on five
experiments at relative pressures of nitrogen in the range of
0.05-0.21. The used cross-sectional area of the nitrogen
adsorbate was 0.162 nm2.
II.3. Experimental Setups and Testing Procedures. The
first set of experiments investigate the catalytic actiVity of the
CaO materials listed in Table 1A toward the reduction of NO
with CO and H2. The experiments were performed in a fixed-
bed reactor. One-half gram of catalytic material was mixed
with 5 g of quartz sand (quartz fine granular, pro analysi from
Merck). The sample was placed on a sintered quartz filter in a
quartz reactor with a 28-mm inner diameter and a length of
1000 mm. The gas flow, measured at ambient temperature and
pressure, was 580 mL/min. This results in residence times for
the different tested materials of between 50 and 63 ms at
800 °C. Gas mixtures of 1000 ppm NO and 1000 ppm CO or
H2 in Ar were used for all samples. These gas mixtures simplify
the solution of the mass balance significantly. The temperature
was measured with a Pt-13% Rh/Pt thermocouple, shielded
with a quartz tube and situated about 0.5 cm above the bed.
The temperature was kept between 799 and 801 °C by using a
Eurotherm thermoregulator. A quadropole mass spectrometer,
Type QMG 421(Balzers), was employed to detect the decrease
in NO concentration between a gas flow bypassing the reactor
and the gas flow passing through the reactor. No mass-transport
problems and no catalytic activity other than the one due to the
added material were observed.
Measurements of the kinetic parameters were done for three
of the powders listed in Table 1A in a fixed-bed reactor
connected to a photoacoustic Fourier transform infrared (FTIR)
gas analyzer from Bru¨el and Kjaer (one of each group, i.e., a
pure, an as-received, and a Na-doped). One-half gram of the
tested catalytic materials was mixed with different amounts of
quartz sand (4, 6, and 10 g), changing the residence times in
the range of 33-176 ms. The mixtures were placed on a
sintered quartz filter. A quartz reactor with an inner diameter
of 22 mm and a length of 500 mm was used for the kinetic
study. The reactor had an asymmetric construction to avoid
heating of the upper metal fitting and the vacuum-tight Viton
O-ring. The temperature was measured with a nickel-chrome
thermocouple positioned 0.5 cm below the sintered quartz filter
and 3 mm below the outlet of the reactor. The positioning of
the thermocouple in this way avoids any interference on the
results. The bed materials were exposed to different gas
mixtures: 862 ppm NO/918 ppm CO or H2, 844 ppm NO/424
ppm CO or H2, and 635 ppm NO/890 ppm CO or H2. The
first gas mixture was tested at two different flows, 693.4 and
433.5 mL/min, while the other two mixtures were only tested
using a flow of 525 mL/min. All flows refer to room
temperature and atmospheric pressure. The results were checked
for influences of mass transport and interference of metal parts
in the reactor.
The latter experimental setup was also used to investigate
the possible synergestic effects between Na and Fe impurities
on the catalytic activity of doped CaO materials. A reaction
mixture of 1000 ppm H2 and 700 ppm NO was used to
determine the NO reaction rate for the materials listed in Table
1B, using a flow of 450 mL/min at 800 °C. H2 was used as
the reducing agent in order to avoid CO2 poisoning (see below).
The bed material consisted of a mixture of quartz sand and the
respective CaO powders, i.e., 4.000 g of SiO2 mixed with 0.125

Promoting Effects of Na and Fe Impurities
J. Phys. Chem. B, Vol. 102, No. 26, 1998 5129
surface area in the range of the pure materials, i.e., 11.6 m2/g,
is obtained. In Figure 1b, it is shown that Fe impurities have
a marginal effect on the specific surface area, while Na
impurities affect it strongly. The trend for the latter is in
agreement with the trend displayed in Figure 1a. Addition of
equimolecular amounts of Fe and Na results in a pronounced
decrease when a comparison is made based on the Na content
alone.
III.2. Catalytic Activity versus the Iron and Sodium
Contents. The decrease in specific surface area with increasing
Na + Fe content has important consequences when evaluating
the catalytic activities of these materials. To be able to evaluate
the impurity effect on the catalytic activity, the reaction rate
has to be formulated as a function of the surface area of the
respective materials. This specific reaction rate is calculated
from the material balance, assuming the order of reaction to be
one for both NO and CO15,16 and using the equimolecular
consumption of NO and CO. The following relations are found:
2
r ) (F/S)(1/yNO - 1/yNOf)yNO
(1)
r ) - (F/(SR))(ln(yNO/yNOf) -
ln((R + yNO)/(R + yNOf)))yNOyCO (2)
where F stands for the molecular flow, y for the molar ratio, r
for the specific reaction rate, S for the surface area, R for the
difference in CO and NO feed, i.e., yCOf - yNOf, and superscript
f for the feed conditions. The former relation (1) gives the
specific reaction rate for equimolecular feed mixtures, while
the latter (2) is applicable for all other gas mixtures.
In Figure 2a, a dependency is observed of the specific reaction
rate on the impurity content (Na + Fe) of the materials listed
in Table 1A. This is seen for both CO and H2 as reducing
agents. It is shown that an impurity content increase correlates
with an increased specific reaction rate for low impurity levels.
For high impurity contents, such as for the materials promoted
by sodium, the specific reaction rate is not that well-defined.
This is mainly due to the uncertainty in the determination of
the surface areas for these materials. Although both H2 and
CO behave similarly as reducing agents, some differences
between them are observed for the Na-doped materials (Figure
2b). This will be shown to be due to CO2 poisoning.
A relatively big spread in the specific reduction rates can be
observed for the as-received materials. This is suggested to be
due to the presence of other, unquantified impurities. To
quantify the effect of Fe, new materials were prepared with
controlled amounts of Na, Fe, and Na + Fe (Table 1B). In
Figure 2c, the specific reaction rate for the NO reduction with
H2 at 800 °C is shown. An increasing specific reaction rate
with increasing impurity content can be observed for all three
groups of materials. Fe is seen to enhance the catalytic activity
of CaO to a greater extent than Na when equimolecular amounts
are compared. In the presence of low amounts of both Na and
Fe, suppression of the catalytic activity is observed. An
increased amount of impurities results in a decreased suppression
as can be seen by the decreasing difference between the specific
reaction rate of the (Na + Fe)-doped CaO and the specific
reaction rates for the sum of the samples independently doped
with Na and Fe.
Figure 1. BET specific surface area (m2/g) as a function of impurity
content for the investigated CaO materials listed in section A and in
section B of Table 1.
g of CaO doped with Na and Na + Fe and 2.000 g of SiO2
mixed with 0.063 g of Fe-doped CaO. Finally, the FTIR gas
analyzer was replaced by a quadropole mass spectrometer
(Balzers, QMG 421) in order to measure the oxygen abstraction
by CO. More details on this procedure and the experimental
set up for surface oxygen abstraction can be found in previous
work.11
The gases used were 5000 ppm NO, CO, CO2, and H2 in Ar.
The gases were mixed with pure Ar, and the flows were
controlled using mass flow controllers (Brooks Type 5850E).
III. Results
III.1. Characterization of the Tested Materials. In Table
1, the Fe and Na contents as well as the specific surface areas
are listed for the investigated CaO powders. The Fe and Na
contents in Table 1A were determined by atomic absorption
spectroscopy, while in Table 1B the added concentrations are
given. It is observed in Table 1A that the purification step,
which was intended to lower the sodium content, also lowers
the Fe content. The specific surface area of all investigated
materials varies between 0.4 and 14.4 m2/g. In Figure 1, the
specific surface area of the CaO powders is shown as a function
of Na + Fe content. The results displayed in Figure 1a show
a decrease in specific surface area with an increased Fe + Na
content. It could be argued that this difference is not due to a
difference in impurity content, but that different treatments of
the respective powders influence the specific surface areas. This
assumption can be excluded based on the results for the CaO
material produced from calcium acetate, blank sample, which
was treated in the same way as the Na-doped materials. A BET
III.3. Measured Activation Energies for Heterogeneous
NO Reduction. The apparent activation energies for the NO
reduction were determined with both CO and H2 as reducing
agents. Three of the previously described powders, listed in
Table 1A, were investigated, i.e., one pure (calcined calcium

5130 J. Phys. Chem. B, Vol. 102, No. 26, 1998
Acke and Panas
Figure 3. Absolute values of the ln((R - XNO/R(1 - XNO)) versus
f
(R - 1)CNO τ for the NO + H2 reaction over a CaO Fisher Scientific
surface at 600, 700, and 800 °C.
reactor model, results in
ln((R - XNO)/R(1 - XNO)) ) k(R - 1)CNO
(4)
where R is the ratio of the feed concentration of CO or H2 to
that of NO, XNO is the fractional conversion of NO as defined
f
earlier, CNO is the feed concentration of NO, and τ is the
residence time, defined as the ratio between the void volume
of the bed and the volume flow rate.
When the absolute value of ln((R - XNO)/R(1 - XNO)) is
f
plotted versus (R - 1)CNO τ, the rate constant can be evaluated
from the slope at a given temperature. This is shown in Figure
3 for the NO + H2 reaction over the Fisher Scientific CaO
surface at 600, 700, and 800 °C. A good correlation with the
linear behavior is observed. This is consistent with the
assumption concerning the reaction orders in NO and reducing
agent. Similar curves were obtained for all materials at the
investigated temperatures and for both reactions. Knowing the
rate constants at the different temperatures, the Arrhenius
equation
k ) A exp(-Ea/RT)
(5)
gives the apparent activation energies (Ea) and the preexponen-
tial factors (A) of the investigated reactions as slopes and Y-axis
intercepts in ln(k) versus 1/T plots (Figure 4).
Figure 2. (a) Specific reaction rate for the reduction of NO with CO
or H2 as a function of Na + Fe content for the materials listed in Table
1A. (b) Difference in the specific reaction rate for CO and H2 as
reducing agents as a function of Na + Fe content for the materials
listed in Table 1A. (c) Specific reaction rate for the reduction of NO
with H2 as a function of Na + Fe content for the materials listed in
Table 1B.
The measured apparent activation energies and the corre-
sponding preexponential factors are given in Table 2. No
significant differences in activation energies between the
different materials are observed for the NO + H2 reaction.
Values between 26.1 and 26.6 kcal/mol are obtained. As
opposed to this, significant differences are noticed using CO
as the reducing agent. The spread in the apparent activation
energies is much greater as values between 27.5 and 32.6
kcal/mol are obtained. This difference cannot be accounted for
by experimental uncertainties. It was observed that an increase
in Ea correlates with an increase in the preexponential factor.
Such an effect has been reported in the literature for oxidative
coupling of methane over similar materials,17,18 where it was
assigned to CO2 poisoning. Korf et al. showed that CO2 blocks
the adsorption of the reactants, thereby lowering the activity
and affecting the activation energy.19 Also, apparent activation
energies are calculated from rate constants at different temper-
atures, assuming identical surfaces. The results of the present
study imply that, for the NO + CO reaction, the surfaces at the
different temperatures are not identical. A decrease in temper-
ature leads to an increase in CO2 adsorption and thus an
increased blocking of the active sites.
acetate), one as-received (CaO, Fisher Scientific), and one Na-
doped (label 1).
When investigating the mass balance, it was observed that
NO and CO were consumed equimolecularly and that the CO2
production corresponded with the amounts of CO that had
reacted. Based on these results, the following overall reaction
is suggested for CO as the reducing species:
NO + CO w CO2 + 1/2N2
This discussion can be repeated for H2 as the reducing agent
producing H2O, since only negligible amounts of NH3 were
detected (<5%).
The reaction rates were evaluated, assuming the reaction to
be first order in both NO and the respective reducing agent.15,16
These assumptions lead to the following rate equation for the
NO + CO reaction:
In Figure 5, the measured apparent activation energy for the
NO + CO reaction is plotted versus the amounts of CO2 per
squared meter of surface in the outlet, taken at 800 °C.
-d[NO]/dt ) k[NO][CO]
(3)
Solving this differential equation, by applying a plug flow

Promoting Effects of Na and Fe Impurities
J. Phys. Chem. B, Vol. 102, No. 26, 1998 5131
Figure 6. Specific preexponential factor (A: preexponential factor per
surface area) versus the Na + Fe content of the investigated materials.
The corrected activation energy of 27.6 kcal/mol for the
reduction of NO with CO is comparable with the 26.3
kcal/mol value for the NO + H2 reaction where no CO2
formation appears. The preexponential factors for the NO +
CO reaction over all three materials can now be recalculated
by constructing an Arrhenius plot based on the 800 °C results
and the calculated apparent activation energy of 27.6 kcal/mol
(excluding CO2 poisoning). The corrected values are 3.8 ×
103, 1.5 × 104, and 8.6 × 103 for CaO pure, CaO Fisher
Scientific, and Na/CaO, respectively. These are in qualitative
agreement with the preexponential factors for the NO + H2
reaction shown in Table 2. The corrected values are systemati-
cally slightly higher than those for the NO + H2 reaction and
could be interpreted to still indicate some CO2 poisoning at 800
°C. To make comparisons to the true preexponential factors,
they have to be divided by the surface area. The preexponential
factors related to the surface area are plotted versus the Na +
Fe content in Figure 6 for each investigated material. The
similarity between this figure and Figure 2a should be noted.
In fact, this could be expected, since all three materials have
similar activation energies, and therefore, the difference in NO
reduction efficiency should be found in the preexponential
factor.
Figure 4. Arrhenius plots for the NO + H2 (a) and NO + CO (b)
reactions catalyzed by Na-doped CaO, pure CaO, and CaO Fisher
Scientific.
III.4. Surface Oxygen Abstraction and Reoxidation. The
surface oxygen abstraction of the materials listed in Table 1B
is examined at 800 °C using CO as a probe molecule. The
amount of CO2 produced under CO exposure is a measure for
the oxygen atoms that can be abstracted from the CaO surface,
i.e., each CO2 molecule stands for one abstracted oxygen.
Typical CO2 formation curves are shown in Figure 7 for the
CaO powders promoted with Fe. Similar curves are obtained
for the other tested materials, i.e., Na- and (Na + Fe)-doped
CaO. A transient CO2 production is observed of which the peak
height and the peak width are correlated to the impurity content;
i.e., an increase in impurity concentration results in an increased
amount of formed CO2. The transient production is followed
by a slowly decaying background due to a continuous reduction
of the CaO matrix. In Figure 8, the integrated amounts of
produced CO2 under the first 700 s, i.e., the amounts of
abstracted oxygens under this period, are shown as a function
of the Na + Fe content. The amount of abstracted surface
Figure 5. Relation between apparent activation energies for the NO
+ CO reaction and the specific CO2 production for the different
investigated materials. Included is also a modeled point obtained by
adding 500 ppm CO2 to the reaction mixture for the CaO (Fisher
Scientific) powder.
Extrapolating the data to a CO2 partial pressure of 0, results in
an estimate of the true apparent activation energy for the NO
+ CO reaction of 27.6 kcal/mol. Support for such a treatment
despite a CO2 gradient along the reactor is found in Figure 5,
where the apparent activation energy was determined when 500
ppm CO2 was added to the feed reaction mixture for the Fisher
Scientific CaO powder. A shift in the specific CO2 production
was modeled and seen to be consistent with the observed change
in apparent activation energies for the different powders.
TABLE 2: Apparent Activation Energies for the NO + CO and NO + H2 Reactions over the Different CaO Surfaces
NO + CO
NO + H2
material
temp, °C
Ea
A
corr
Ea
A
corr
0.5 g CaO Fisher
0.5 g CaO pure
0.5 g Na/CaO
corrected
550-850
650-800
650-800
30.4
27.5
32.6
27.6
5.5 × 104
3.7 × 103
8.6 × 104
0.99
0.99
1
26.6
26.1
26.3
26.3
1.1 × 104
3.4 × 103
7.9 × 103
0.99
1
1

5132 J. Phys. Chem. B, Vol. 102, No. 26, 1998
Acke and Panas
Figure 9. Specific reaction rate as a function of the amount of
abstracted oxygens for the materials listed in Table 1B.
Figure 7. Transient CO2 production for the different Fe-doped CaO
materials under CO exposure at 800 °C.
Figure 10. Ratio between the N2 formed during reoxidation by NO
and the CO2 formed during CO exposure at 800 °C of the materials
listed in Table 1B.
Figure 8. Amounts of abstracted oxygens during the first 700 s of
CO exposure as a function of the Na + Fe content of the materials
listed in Table 1B.
III.5. Measured Activation Energies for Heterogeneous
NO Reduction over MgO, SrO, and BaO. The results so far
indicate that the apparent activation energies for the two
investigated reactions
oxygens is observed to increase with impurity content for each
group of materials. Comparing the oxygen abstraction for CaO
powders promoted with Fe or Na, Fe impurities are seen to be
more effective than equimolecular amounts of Na impurities.
NO + CO S 1/2N2 + CO2
NO + H2 S 1/2N2 + H2O
Addition of both Na and Fe leads to a suppression of the
oxygen abstraction. This is observed when the amount of
abstracted oxygens for the (Na + Fe)-promoted materials is
compared with the sum of the amounts for the separately
promoted materials. This suppression is most pronounced for
low impurity concentrations. Similar effects were observed for
the specific reaction rate of the NO reduction with H2 at
800 °C. And indeed, combination of these two results, i.e., the
amount of abstracted oxygens versus specific reaction rate, for
all investigated materials listed in Table 1B results in a linear
relation between the two as is shown in Figure 9. It is observed
that an increase in the amount of abstracted oxygens results in
an increased specific reaction rate. Extrapolation of this result
to zero oxygen abstraction results in a very low specific reaction
rate.
are independent of the impurity content and of the reducing
gas used, H2 or CO, if correction is made for the influence of
CO2 poisoning for the latter. This independence can be used
to determine the apparent activation energies of the other
alkaline-earth oxides, i.e., MgO, SrO, and BaO. The MgO was
provided by Fluka and had a maximum Na and Fe content of
0.2 and 0.005 wt %. The SrO (99.5% pure) and BaO (97%
pure) were provided by Johnson Matthey. The trace metal
content of the latter was specified to be less than 1%. The bed
materials were pretreated in Ar at 900 °C for MgO and SrO
and 1100 °C for BaO until neither H2O nor CO2 desorption
was observed. The obtained apparent activation energies for
the reduction of NO by H2 are given in Table 3 and are seen to
vary from 19.5 kcal/mol for MgO to 33.5 kcal/mol for BaO.
The temperature intervals for determination of the activation
energies (Table 3) were in all cases selected above the
decomposition temperature of the respective hydroxides. In
Figure 11, these values are plotted versus the respective lattice
parameters and a correlation is observed.
After exposure to CO, the materials were reoxidized by NO
at 800 °C. During the NO exposure, a transient N2 formation
was observed. The amounts of N2 formed were found to be
about half the amounts of abstracted oxygens. This was
expected from previous experiments11 and is shown again in
Figure 10.

Promoting Effects of Na and Fe Impurities
J. Phys. Chem. B, Vol. 102, No. 26, 1998 5133
+ Fe concentrations exceeding about 0.3 wt % and is in line
with the discussion put forward by Schoderbo¨ck and Lahaye
for Fe impurities in quartz sand.14
The role of O- species in the catalytic activity of Li-doped
MgO and Na-doped CaO substrates toward the oxidative
coupling of methane has been discussed extensively. These
materials display high selectivity toward this reaction.25 Al-
though the reaction mechanism is not fully understood, a redox
mechanism involving O- species introduced by substitution of
Mg or Ca by Li or Na has been suggested. The surface O-
species can abstract a hydrogen atom and produce an alkyl
radical from methane. The subsequent dehydroxylation of the
surface leads to an inactive surface for further alkyl formation.
The surface is regenerated by exposure to oxygen. This reaction
mechanism was proposed by Lin et al.,10 and it illustrates clearly
the role of the O- species in the catalytic activity of these types
of materials. The electric conductivity of pure MgO, known
to be an insulator, is enhanced by a factor of 104 by promotion
with lithium ions.26 Lin et al. referred to this result when
suggesting that defects in the bulk could also contribute to the
catalytic activity by a hole transport mechanism exploiting O-
species in the bulk:
Figure 11. Apparent activation energy for the NO + H2 reaction over
alkaline-earth oxides as a function of the lattice parameter of these
oxides.
TABLE 3: Apparent Activation Energies for the NO + H2
Reaction over Different Alkaline-Earth Metal Oxides
material
temp, (°C)
Ea, kcal/mol
MgO
CaO
SrO
650-850
650-800
750-850
900-1000
19.5
26.3
29.1
33.5
Na+O- + O2-s S Na+O2- + O-
BaO
s
IV. Discussion
Examination of the same reaction over ultrathin films of MgO
and Li-doped MgO and characterization of the respective films
by high-resolution electron energy loss spectroscopy lead Wu
et al. to a different conclusion.9 These authors suggested that
[Li+O-] defects promote the production of F centers in the
vicinity of the surface.
The presented results suggest that the rate-determining step
for the examined reactions (shown above) is the same. Tem-
perature-programmed reduction experiments have shown that
the activation energy for the surface-oxygen-abstraction step
with CO is about 25 kcal/mol,11 while the apparent activation
energy for the breaking of an NO bond is lower than 10 kcal/
mol.11 The former value, 25 kcal/mol, is similar to the apparent
activation energies measured here. Based on these results, it is
suggested that the rate-determining step can be ascribed to the
picking out of surface oxygens, leaving the electrons of the
removed oxygen ion in the substrate.
It is not clear how transition-metal ions are incorporated in
the structure, whether in solid solution or in a second phase, or
even in which oxidation state they appear. However, the
stability of a certain oxidation state will depend on, e.g.,
temperature, gas atmosphere, and other impurities present.
Promotion of an oxide matrix with transition-metal ions has
been addressed by Cimino.27 It was concluded that when the
guest ion has the same valence and is of similar size as the
host ion and the matrix offers sites of symmetry, compatible
with the electronic structure of the guest ion, then a true solid
solution can be formed. However, when the guest ion has a
different charge than the host ion, a solid solution will be
obtained only if its extra charge is compensated, e.g., Cr3+ by
Li+ in MgO. This is in agreement with electron paramagnetic
resonance (EPR) experiments for unstable [Li+O-] defects in
doped MgO.28 In Fe-containing samples, it was observed that
unstable [Li+O-] defects, created by γ irradiation at 77 K,
disappear at temperatures as low as 200 K, while simultaneously
the Fe3+ concentration increased. For the materials investigated
here, this understanding would imply that the highest oxidation
state for the respective transition-metal impurity is the most
likely, probably compensated by Na+ ions in its vicinity.
However, it has been shown by several authors that the addition
of a transition metal to alkali-metal-promoted alkaline-earth
oxide materials can give rise to significant improvements in
catalytic behavior toward the oxidative coupling of methane.29-32
Promotion of the CaO by transition-metal addition was observed
in the present study for the reduction of NO with H2. However,
simultaneous addition of equimolecular amounts of Na and Fe
was found to suppress the specific catalytic activity when
comparing to the sum of the reaction rates obtained for the
impurities added separately (Figure 2c). This is consistent with
the EPR results as discussed in ref 28, as an electrostatically
balanced situation based on Na+ and Fe3+ ions is understood
In Figures 2a and 6, the influence of the trace metal impurities
on the catalytic activity of CaO is shown for the materials listed
in Table 1A. Although these materials were only analyzed for
the sodium and iron contents, it is believed that the results will
not change if other impurities present in the materials would
be included in the analysis. In fact, a further improved
correlation between impurity content and specific reaction rate
for the NO reduction is expected. The trace metal impurities
can be separated into two main groups: metals having only
one oxidation state and metals appearing in several oxidation
states. The alkali metals and the alkaline-earth metals belong
to the former group, while the transition metals belong to the
latter. In the first group, a distinction should be made between
the alkali metals, with oxidation state +I, and the alkaline-earth
metals, oxidation state +II. Substituting a Ca2+ ion with a Na+
leads to an electrostatically unbalanced situation in the CaO
matrix. This can be neutralized by the formation of O- species
and/or oxygen vacancies in the crystal structure. The formation
of O- species has been shown by electron paramagnetic
resonance for a number of alkaline-earth systems doped with
alkali metals: e.g., Li-doped MgO and Li- and Na-doped
CaO.9,10,20-24 The amount of alkali-metal ions that can be kept
in solid solution in an alkaline-earth metal oxide matrix is
limited. From the moment the solid solubility is exceeded, a
Na-rich second phase will form. The sodium ions in this second
phase will not lead to the formation of O- species and will
have different catalytic properties toward the investigated
reactions. This is suggested to be observed in Figure 2 for Na

5134 J. Phys. Chem. B, Vol. 102, No. 26, 1998
Acke and Panas
to form. In such a scenario, the addition of Na does not lead
to the formation of O- species, and consequently, only Fe
impurities can participate in the redox process. The results
displayed in Figure 2c show further that addition of Fe is more
effective than addition of equimolecular amounts of Na. A
possible explanation would be the formation of oxygen vacan-
cies which compete with O- production in the case of Na
addition to maintain an electrostatically balanced situation.
Oxygen vacancies are not expected to form in the bulk for only
Fe addition. The role of Fe impurities on the catalytic activity
of CaO has not been addressed in the literature. A mechanism
based on a reduction and reoxidation of the surface is demon-
strated by an oxygen-abstraction step under CO exposure (Figure
7) and a subsequent N2 formation under NO exposure (Figure
11).
While details can be discussed, the main result of this work
is to demonstrate the extent to which impurity content correlates
with the catalytic activity (Figure 2a and -c) and with the amount
of abstracted oxygens (Figure 8). Combining these observations
demonstrates that a decrease in the amount of abstracted oxygen
from the substrate corresponds with a decreased specific reaction
rate (Figure 9). This demonstrates a definite relation between
the impurity content, the amount of abstracted oxygens, and
the specific reaction rates.
have previously been suggested to play a vital role in the surface
reaction mechanism. The precise function, whether localized
electron sinks or contributing to an O- impurity band in a
semiconductor model, could not be resolved. The role of Fe,
shown to be more effective than equimolecular amounts of Na,
is not fully understood. However, a redox mechanism based
on Fe2+/Fe3+ cannot be excluded.
Acknowledgment. We are in debt to the Swedish National
Board for Industrial and Technical Development for financial
support (NUTEK).
References and Notes
(1) Zhdanov, V. P.; Kasemo, B. Surf. Sci. Rep. 1997, 29, 31.
(2) Wong, K.; Johansson, S.; Kasemo, B. Faraday Discuss. 1996, 105.
(3) Goodman, D. W.; Peden, C. H. F. J. Phys. Chem. 1986, 92, 4839.
(4) Berlowitz, P. J.; Peden, C. H. F.; Goodman, D. W. J. Phys. Chem.,
1988, 90, 5213.
(5) Hamada, H. Catal. Today 1994, 22, 21.
(6) Chu, W. F.; Rohr, F. J. Solid State Ionics 1988, 28-30, 1540.
(7) Voorhoeve, F. J. H.; Remeika, J. P.; Freeland, P. E.; Matthias, B.
T. Science 1972, 177, 353.
(8) Voorhoeve, F. J. H.; Remeika, J. P.; Trimble, L. E. Ann. N. Y.
Acad. Sci. 1976, 272, 3.
(9) Wu, M.-C.; Truong, C. M.; Coulter, K.; Goodman, D. W. J. Catal.
1993, 140, 344.
(10) Lin, C.-H.; Wang, J.-X.; Lunsford, J. H. J. Catal. 1988, 111, 302.
(11) Acke, F.; Panas, I.; Stro¨mberg, D. J. Phys. Chem. B 1997, 101,
6484.
(12) Henrich, V. E. Rep. Prog. Phys. 1985, 48, 1481.
(13) Zhang, X.; Walters, A. B.; Vannice, M. A. J. Catal. 1994, 146,
568.
(14) Schoderbo¨ck, P.; Lahaye, J. Appl. Surf. Sci. 1996, 93, 109.
(15) Tsujumura, M.; Furusawa, T.; Kunii, D. J. Chem. Eng. Jpn 1983,
16, 132.
Similar understandings as produced for CaO are suggested
to also hold for the MgO, SrO, and BaO substrates. The linear
dependence between the apparent activation energy of the NO
reduction on the lattice parameter for these materials can be
interpreted to support both the semiconductor model and the
model based on stabilization of F centers.
(16) Tsujumura, M.; Furusawa, T.; Kunii, D. J. Chem. Eng. Jpn, 1983,
16, 524.
(17) Coulter, K.; Goodman, D. W. Catal. Lett., 1993, 20, 169.
(18) Xu, M.; Shi, C.; Yang, X.; Rosynek, M. P.; Lunsford, J. H. J. Phys.
Chem. 1992, 96, 6395.
(19) Korf, S. J.; Roos, J. A.; de Bruijn, N. A.; Van Ommen, J. G.; Ross,
J. R. H. J. Chem. Soc., Chem. Commun. 1987, 1433.
(20) Tohver, H. T.; Henderson, B.; Chen, Y.; Abraham, M. M. Phys.
ReV. B 1972, 5, 3276.
(21) Chen, Y.; Tohver, H. T.; Narayan, J.; Abraham, M. M. Phys. ReV.
B 1977, 16, 5535.
(22) Olson, D. N.; Orera, V. M.; Chen, Y.; Abraham, M. M. Phys. ReV.
B 1980, 21, 1258.
(23) Wang, J.-X.; Lunsford, J. H. J. Phys. Chem. 1986, 90, 5883.
(24) Lin, C.-H.; Ito, T.; Wang, J.-X.; Lunsford, J. H. J. Am. Chem. Soc.
1987, 109, 4808.
V. Conclusion
Correlation between the impurity content and the specific
reaction rate of the different types of CaO materials toward the
NO reduction with CO and H2 was demonstrated. The apparent
activation energy for the overall NO + H2 reaction was
determined to about 26.5 kcal/mol. A similar value, 27.5 kcal/
mol, was found for the NO + CO reaction after correction for
CO2 poisoning. Both reactions have previously been suggested
to comprise (i) a surface-oxygen-abstraction step and (ii) a
reoxidation step. The former one has been pointed out as the
rate-determining step.
(25) Ito, T.; Lunsford, J. H. Nature, 1985, 314, 721.
(26) Chen, Y.; Kernohan, R. H.; Boldu´, J. L.; Abraham, M. M. Solid
State Commun. 1980, 33, 441.
(27) Cimino, A. Chim Ind 1974, 56, 27.
(28) Boldu´, J. L.; Abraham, M. M.; Chen, Y. Phys. ReV. B, 1979, 19,
4421.
(29) Larkins, F. P.; Nordin, M. R. In Methane ConVersion ; Bibby, D.
M., Chamg, C. D., Howe, R. F., Yurchak, S., Eds.; Elsevier: Amsterdam,
1988; p 409.
(30) Bi, Y.; Zhen, K.; Jiang, Y.; Teng, C.; Yang, X. Appli.Catal. 1988,
39, 185.
(31) Bartsch, S.; Falkowski, J.; Hofmann, H. Catal. Today 1989, 4, 421.
(32) Korf, S. J.; McNamara, D.; van Ommen, J. G.; Ross, J. R. H. Appl.
Catal. 1989, 56, 119.
The presence of impurities in the CaO crystal structure was
shown to have no significant influence on the apparent activation
energy. The effect of increasing the impurity level results in a
proportional increase in the number of active sites as was
observed by an increase in the specific preexponential factor,
specific reaction rate, and amount of abstracted oxygens. The
other alkaline-earth oxides were also shown to be catalytically
active for the NO reduction, and the apparent activation energy
for the NO + H2 reaction was shown to correlate linearly with
the lattice parameter. Our results emphasize the importance of
O- sites, created by the Na+ impurities. Such electron sinks