NOx storage on BaO(100) surface from first principles: A two channel scenario

Article abstract of DOI:10.1021/jp0126457

Lean-burn engines, where the engine is operated at oxygen excess, yields a more efficient combustion and reduces the fuel consumption in comparison to stoichiometric combustion . A drawback of lean-burn engine is that NO_x cannot be effectively removed with a catalytic converter. A possible solution to this problem comprises the NO_x storage concept, where a storage function in the catalytic converter is combined with mixed operating conditions of the engine. Spin polarized GGA density functional theory was used to study NO₂ adsorption at a BaO(100) surface. A periodic supercell procedure was utilized and two redox reaction channels were mapped out, involving two chemisorbed NO₂ molecules per supercell. The reaction paths were initiated by NO₂ adsorption in the form of a nitrite over a Ba² site, which generated an electron hole among the surrounding surface oxygen atoms. A reaction path branching occurred as the second NO₂ either acted as surface oxidant, forming a surface nitrile-peroxide pair by releasing NO(g), or bound to an O^-_surf site to form a formal surface nitrate. A redox reaction involving surface nitrite-nitrate interconversion was elucidated. The calculated results were used to interpret experimental observations of surface nitrites, peroxides, NO(g0 desorption, and surface Ba(NO₃)₂ formation. The understandings were elucidated in the context of the NO_x storage concept of lean-burn catalysis.

Full text of DOI:10.1021/jp0126457

J. Phys. Chem. B 2002, 106, 137-145
137
NOx Storage on BaO(100) Surface from First Principles: a Two Channel Scenario
Peter Broqvist,†,‡ Itai Panas,*,§ Erik Fridell,†,‡ and Hans Persson†,‡
Department of Applied Physics, Competence Centre for Catalysis, and Department of Inorganic
EnVironmental Chemistry, Chalmers UniVersity of Technology, SE-412 96 Go¨teborg, Sweden
ReceiVed: July 11, 2001; In Final Form: October 23, 2001
NO2 adsorption at a BaO(100) surface is investigated by means of spin polarized GGA density functional
theory. A periodic supercell procedure is employed, and two redox reaction channels are mapped out, involving
two chemisorbed NO2 molecules per supercell. The chemisorption is studied in two subsequent steps. The
reaction paths are initiated by NO2 adsorption in the form of a nitrite over a Ba2+ site. This generates an
electron hole among the surrounding surface oxygen atoms. A reaction path branching occurs as the second
NO2 either (a) acts as surface oxidant, forming a surface nitrite-peroxide pair by releasing NO(g), or (b)
binds to an O- site to form a formal surface nitrate. A redox reaction involving surface nitrite-nitrate
surf
interconversion is also addressed. The computed results are employed to interpret experimental observations
of surface nitrites, peroxides, NO(g) desorption, and surface Ba(NO3)2 formation. The understandings are
discussed in the context of the NOx storage concept of lean-burn catalysis.
1. Introduction
conditions.7-9 This initial oxidation step, employing a Pt catalyst,
limits the storage rate for the temperature range below where
The demand for lower fuel consumption in gasoline engines
has led to the introduction of lean-burn engines where the engine
is operated at oxygen excess.1 This concept yields a more
efficient combustion and reduces the fuel consumption in
comparison to “stoichiometric combustion”. Increased combus-
tion efficiency implies reduced CO2 emissions in conjunction
with effectively eliminating CO and hydrocarbon emissions. A
drawback of the lean-burn engine is that NOx cannot be
efficiently removed with a catalytic converter. One possible
solution to this problem comprises the NOx storage concept,
where a storage function in the catalytic converter is combined
with mixed operation conditions of the engine.2 During lean
periods NOx is trapped by the storage material, while during
short fuel-rich intervals it is regenerated by NOx release and
subsequent reduction with hydrocarbons H2 and CO to form
N2, H2O, and CO2. Detailed mechanistic knowledge of the
interaction of NOx with the storage component is of vital
importance for optimizing this concept. Key parameters to
investigate include, e.g., temperature stability, NOx storage
capability, regeneration times, and hindering of sulfur deactiva-
tion.
Several investigations regarding NOx storage on alumina-
supported barium oxide are found in the literature.2-9 Experi-
ments with simplified gaseous compositions and model catalysts
have been employed in order to gain knowledge of various
aspects of the real system.6,7 Thus, influence of, e.g., carbonate
and hydroxyl formation has been removed in order to obtain
knowledge of the interplay between NOx and the storage
compound alone. Indeed, competition between CO2 and NO2
for the same adsorption sites has previously been demonstrated.9
Also relevant is the finding that, in case of NO, the NOx storage
capacity is crucially dependent on an intermediate NO oxidation
step to NO2 before the storage can take place under lean
the storage capacity displays its maximum, i.e., T < 380 °C.
At T > 450 °C, the stored compounds are found to decompose
rapidly.7,8
When a storage catalyst is exposed to NO2, both trapping of
NO2 and release of NO can be observed. This tells of a redox
process involving oxidation of either the oxide surface or an
adsorbed species upon NO2 reduction. Fridell et al.8 discuss this
NO release in view of two different possible processes, either
surface nitrites are converted to nitrates, or, formation of surface
peroxides result from oxidation of the oxide surface, predomi-
nantly BaO, by NO2. FTIR studies of the interaction of NO2
with BaO/Al2O3 samples show formation of surface nitrites and
nitrates. Experiments at 100 °C indicate that the formation of
nitrites precedes the formation of nitrates. Formation of bulk
nitrates is not observed on the relevant time scales for NOx
storage. Thus, a plausible mechanism for NOx storage becomes
one where the oxidation of NO to NO2 over Pt is followed by
formation of surface nitrites on the BaO surface. These are
further oxidized to nitrates by NO2 with desorbing NO as a
result. However, a mechanism involving BaO2 cannot be ruled
out. Kinetic modeling of NOx storage on Pt/Al2O3 and Pt/BaO/
Al2O3 has been undertaken based on these two mechanisms.7
The present work aims at complementing the available
experimental work with a theoretical study. In this paper we
have chosen to focus on the interaction between NOx and barium
oxide. NOx in the form of NO2 is employed, as experiments
conclude that the oxidation of NO to NO2 precedes the actual
storage of NOx. The storage compound, barium oxide, is in the
form of a clean BaO(100) surface, cleaved from the geometry
optimized BaO bulk structure. The (100) surface is generally
considered to be the most stable one for BaO(s). It consists of
five coordinated Ba and O atoms in the top layer of atoms. The
complexity of surface properties due to steps, oxygen vacancies,
etc. is not considered in this paper. Enhancement effects of
surface defects such as kinks and steps have been demonstrated
on CaO surfaces10 where it may be of great importance for NOx
* Corresponding author. E-mail: itai@inoc.chalmers.se
† Department of Applied Physics.
‡ Competence Centre for Catalysis.
§ Department of Inorganic Environmental Chemistry.
10.1021/jp0126457 CCC: $22.00 © 2002 American Chemical Society
Published on Web 11/30/2001

138 J. Phys. Chem. B, Vol. 106, No. 1, 2002
Broqvist et al.
adsorption. Below, we demonstrate that the BaO (100) surface
is highly reactive toward NO2 adsorption, and activation
becomes unnecessary. Similar ideas have been proposed by
Pettersson et al.11
The interaction between NO2 and BaO(100) is studied using
periodic supercell density functional theory (DFT) within the
gradient corrected local density approximation (GGA). Pseudo
potentials are employed to describe the inert core electrons.12
Key steps in the NO2 adsorption on a BaO(100) surface are
generated from first-principles, and an NO2 storage scenario is
formulated. Possible connections to the proposed storage
mechanism by Fridell et al.8 are highlighted.
2. Method
2.1. Theoretical Method and Computational Details.
-
Structures and binding energies were produced for NO2, NO2
,
NO3-, and NO bound at the barium surface site (Ba2+surf) and
oxygen surface site (O2-surf) of BaO(100). All calculations were
performed using the commercial version of CASTEP12 as
implemented in Cerius2.13 A large body of existing work
indicates that CASTEP is indeed useful for predicting structural
properties and energy changes associated with chemical trans-
formations involving oxides (see, e.g., refs 14-16).
CASTEP employs a plane wave expansion of the wave
functions associated with the valence electrons. This basis set
is determined by selecting a kinetic energy cutoff, Ecut. The
tightly bound core electrons are represented by nonlocal ultrasoft
pseudopotentials as proposed by Vanderbildt.17 Reciprocal space
integration over the Brillouin zone is approximated through a
careful sampling at a finite number of k points using the
Monkhorst-Pack scheme.18 The exchange-correlation contribu-
tion to the total electronic energy is treated in the spin polarized
generalized gradient approximation (GGA) extension to the local
density approximation (LDA) by Perdew et al.,19 as implemented
in the CASTEP software package (GGS-PBE).12 In all calcula-
tions, the kinetic energy cutoff and the density of the Monkhorst-
Pack k-point mesh were chosen high enough so as to ensure
convergence of the computed structures and energetics. The Ecut
Figure 1. Position of chemisorbed NO2 with N centered over a surface
oxygen site. (a) Top view and (b) side view with bond length [Å] and
angle [degrees] indicated.
was set to 300 eV and the k-point spacing was set to 0.05 Å-1
.
The methodological accuracy in adsorption energies and mo-
lecular bond distances were converged to within ( 0.05 eV,
and ( 0.01 Å respectively. This was deemed sufficient for the
present investigation.
To model the BaO(100) surface, a two-layer slab was cleaved
from the geometry optimized bulk structure and modeled by a
periodic supercell. A x2 × x2 supercell was employed in
conjunction with a 9 Å vacuum thickness. Tests on convergence
of adsorption energies were performed by using a three-layer
slab and by varying the vacuum thickness in the interval 9-15
Å. Insignificant differences in binding energies were obtained.
Following usual convention, the adsorption energies Eads were
calculated according to the following expression:
In case of chemisorption involving a surface oxygen, the position
of the latter was included in the geometry optimization. Earlier
studies addressing surfaces of the alkaline earth metal oxides
MgO and CaO have shown the surface not to change signifi-
cantly upon chemisorption when allowing the initial bulk
parameters to relax.10,20 Such secondary relaxation may be larger
in case of the BaO substrate due to the weaker restoring
Madelung potential. Such effects will be quantified in a future
study.
2.2. Experimental Method. NO2 adsorption and NO/NO2
desorption from a BaO/Al2O3 monolith catalyst were monitored
by TPD experiments in a flow reactor system. The monolith
catalyst was prepared by wet deposition methods as described
in ref 5, and the resulting washcoat consisted of 20 wt % BaO
and 80 wt % γ-Al2O3. In situ detection of NO and NO2 in the
flow reactor was made with a chemiluminescence detector as
described in ref 21. TPD measurements were performed by first
conditioning the sample at 600 °C in Ar and then cooling it to
the adsorption temperature in Ar for 1 min. 500 ppm NO2 in
Ar was subsequently introduced for 10 min, and after another
min in Ar, a temperature ramp from the adsorption temperature
up to 600 °C with 20 °C/min in Ar was commenced.
Eads ) Eslab + molecule - (Eslab + Emolecule
)
(1)
where Emolecule is the energy of the isolated adsorbate in its
equilibrium configuration, Eslab is the total energy of the isolated
slab, and E(molecule+slab) is the total energy of the adsorbate +
slab supersystem. Negative adsorption energies denote an
exothermic adsorption process.
The purpose of this study is to produce information regarding
structures and relative stabilities for the different NO2 derivatives
on the surface. Therefore, all calculations were performed with
a fixed surface, while the sorbate geometries were optimized.

NOx Storage on BaO(100) Surface
J. Phys. Chem. B, Vol. 106, No. 1, 2002 139
Figure 2. Position of chemisorbed nitrite over a surface barium site.
(a) Top view and (b) side view with bond lengths [Å] and angle
[degrees] indicated.
3. Results and Discussion
Electronic structure calculations and geometry optimizations
were performed in order to determine the relative stabilities of
the most relevant species formed when a clean BaO(100) surface
is exposed to NO2(g). The scenario is unravelled as follows:
in section 3.1 initial adsorption of NO2 on a surface oxygen
site is considered; section 3.2 contrasts the former by the initial
adsorption of NO2 on a surface barium site; section 3.3 addresses
chemisorption of two NO2 molecules as nitrite-nitrate pairs,
where the nitrite adsorbs on a barium site and the nitrate involves
a surface oxygen. Section 3.3 addresses also subsequent redox
reactions among the adsorbates to form nitrate-nitrite pairs and
the stability of an all-nitrate surface phase at 25% coverage
resulting from NO2(g) reduction to NO(g). Section 3.4 describes
a competing reaction channel to that described in section 3.3
as it involves the formation of transient nitrite-peroxynitrite
pairs, followed by disproportionation of the Osurf-O-NO
species into NO(g) and a surface peroxide.
Figure 3. (a) Spin density plot for NO2 adsorbed over a barium site.
The spin is seen to delocalize over the adsorbate and the surrounding
surface oxygen ions. (b) Spin density plot for NO3 adsorbed over a
surface barium site. The spin density is seen to delocalize over the
surface oxygens alone. (c) Charge density plot emphasising the
electrostatic nature of the bond between the NO2(ads) and the Ba2+
site, i.e., nonoverlapping adsorbate/surface charge distributions.
-
The theoretical findings are discussed in detail below and
are used to interpret experiments.
3.1. Initial Adsorption of NO2 at an Osurf Site. When
molecular NO2 approaches a clean BaO(100) surface at low

140 J. Phys. Chem. B, Vol. 106, No. 1, 2002
Broqvist et al.
Figure 4. Optimized structure of the surface nitrite-nitrate pair on
BaO(100). (a) Top view and (b) side view with bond lengths [Å]
indicated.
Figure 5. Optimized structure of the surface nitrate-nitrite pair on
BaO(100). The two NO2 molecules in Figure 4 have undergone a
disproportionation (redox) reaction. (a) Top view and (b) side view
with bond lengths [Å] indicated.
coverage, electrostatic interactions guide the positive nitrogen
of NO2 toward a surface oxygen site. An initial Osurf-NO2 bond
distance of 2.39 Å is obtained for the adsorbate/substrate
compound and it is depicted in Figure 1. This suggests a
physisorbed species, but binding energies in the range of 0.8 (
0.05 eV in conjunction with the structure of the adsorbate
surface nitrite, i.e., (i) an electron transfer from substrate to
-
adsorbate occurs, (ii) the structure approaches that of NO2
,
and (iii) a 0.7 eV adsorption energy is obtained. Although
apparently trivial, this surface species is decisive to the
mechanism for further NO2 storage. Given a priori that NO2(g)
has an unpaired electron, it implies that the BaO surface
becomes partially oxidized when forming a nitrite at the Ba
site. This feature can be demonstrated by performing spin-
polarized calculations and plotting the spin density, see Figure
3. It is clearly seen how this density is delocalized over both
adsorbate and adjacent substrate oxygen ions. Thus, while the
NO2 adsorption energy could be interpreted to result from
electrostatic interactions between an ionic surface site and a
neutral but polar molecule, the spin density distribution and
indeed also the NO2(ads) structure (RN-O ) 1.26 Å, V(O-N-
O) ) 118°) both clearly infer electron transfer. Thus the bonding
must be described as a chemisorption rather than a physisorption
event.
-
approaching that of NO2 imply the adsorbate to be chemi-
sorbed. Indeed, the experimental 1.24 Å N-O bond length in
the nitrite ion and the 115.4° O-N-O bond angle22 compare
reasonably with the 1.25 Å and 121° obtained here for the
surface species. The latter result is also reflected in the spin
density map, which has the spin to a great extent delocalized
over the adsorption site, i.e., Osurf, indicating a partial electron
transfer from the oxygen site to the adsorbed NO2 molecule.
The fact that NO2 is found to chemisorb weakly with a long
Osurf-N bond distance is chemically intuitive, since it amounts
to the formation of an “NO32-” surface species, which has the
unpaired electron in the antibonding Osurf-N orbital. The
alternative would be the formation of an NO3- surface species,
but the extra electron would then have to reduce Ba2+, which
is not observed in the calculations.
-
The degree to which the Ba2+surf-NO2 interaction is
noncovalent can be appreciated from the fact that binding
energies in the range 0.5-0.7 eV are obtained for the four most
stable chemisorption arrangements, comprising the monodentate
structures Ba-NO2 and Ba-O-NO, in conjunction with the
bidentate structures Ba-O2N and Ba-ON-O. Still, the latter
3.2. Initial Adsorption of NO2 on a Basurf Site. Crucial to
the NO2 loading is the adsorption configuration whereby a single
NO2 is found to bind at a Ba2+surf site, see Figure 2. Similar to
the description in section 3.1, also here NO2 chemisorbs as a

NOx Storage on BaO(100) Surface
J. Phys. Chem. B, Vol. 106, No. 1, 2002 141
Figure 6. Optimized structure of the surface nitrate-nitrate pair on
BaO(100). (a) Top view and (b) side view with bond lengths [Å]
indicated.
is found to display the global minimum on the potential energy
surface. In this structure, the molecule binds with the molecular
plane orthogonal to the surface plane. The atomic arrangement
is such that one oxygen atom points away from the surface,
while the molecule binds to the surface with the doubly occupied
highest occupied molecular orbital of local π symmetry with
respect to the remaining NO bond. The low electron density
between N and Ba (see Figure 3c) supports the understanding
of an ionic bond.
Figure 7. Optimized structure of a single surface nitrate ion over a
Ba2+-site on BaO(100). (a) Top view and (b) side view with bond
lengths [Å] indicated.
the initial nitrite formation channel, described in section 3.2, is
associated with the formation of a delocalized unpaired electron
-
on the surface oxygens in the vicinity of the Ba2+surf-NO2
site. This result suggests that initial formation of surface nitrites
over barium sites activates new chemisorption sites comprising
The net stability of the surface nitrite species can be
understood to result from three contributions, including (a) the
significant electron affinity of NO2 (2.27 eV),23 (b) the resulting
the adjacent O- sites. Thus, a key intermediate step in the
surf
NO2 loading on BaO(s) involves the pairwise formation of
Ba2+ to NO2 electrostatic interaction, and (c) the loss of
nitrate analogues [Osurf-NO2]- and surface Ba nitrites.
-
surf
lattice energy as an O2- is effectively oxidized to O-. To the
To test this possible understanding, a second NO2 molecule
was introduced into the supercell and placed on a surface oxygen
adjacent to the Ba2+surf-NO2- site. Geometry optimization was
performed and an additional 1.0 eV adsorption energy is
obtained, which is slightly greater than what was found for a
single NO2 at a barium site. This result supports possible
formation of transient nitrite-nitrate pairs at intermediate initial
nitrite coverages. The resulting bonding configuration for the
nitrite-nitrate pair is depicted in Figure 4, where the proximity
between adjacent nitrite and nitrate ions can be appreciated. The
distance between O on the nitrate and N on nitrite is ap-
proximately 3 Å, which limits the NO2 loading at the BaO-
extent that the Madelung potential is decisive for the stability
of the formal doubly charged oxygen anions in the alkaline earth
oxides, the lattice constant must be deemed critical for the initial
formation of surface nitrites.
3.3. Further NO2 Storage and Surface Nitrite-Nitrate
Redox Reaction. The increasingly undressed columbic interac-
tions among the adsorbates with increasing nitrite coverage
would be the main mechanism for controlling the NO2 storage
along the initial adsorption channels outlined above. It is found
that a channel for additional loading comes into play already
before this limit is reached. Indeed, it was stressed above, that

142 J. Phys. Chem. B, Vol. 106, No. 1, 2002
Broqvist et al.
Figure 9. Optimized structure of the surface nitrite species and
peroxide on BaO(100). (a) Top view and (b) side view with bond
lengths [Å] and angle [degrees] indicated.
Figure 8. Structure of the surface nitrite-peroxynitrite pair transient,
which may lead to surface peroxide formation and NO(g) release. (a)
Top view and (b) side view with bond lengths [Å] indicated.
electron affinity (EA) of NO3 than that of NO2, i.e., 3.91 eV24
vs 2.27 eV,23 respectively. An efficient illustration as to the
significance of this EA difference is found in the spin density
distribution for the Ba2+site-NO3- surface compound (see Figure
3b), where the spin density is seen to delocalize exclusively on
the surface oxygens. This is in contrast to the nitrite formation
at the barium site, where significant spin density is found to
remain on the adsorbate (see Figure 3a).
3.4. Peroxide Formation on BaO(100). A surface reaction
competing with the formation of surface nitrite-nitrate pairs
comprises the nitrite-peroxynitrite channel. When an oxygen
site has been activated by nitrite formation at an adjacent barium
ion, a second NO2 can form an Osurf-O-NO transient (see
Figure 8), which subsequently dissociates into an Osurf-O
species and NO(g) (see Figure 9). Loss in stability of ∼0.4 eV
is associated with this reaction. Analysis of the electronic spin
density reveals that a surface peroxide is formed as no spin
density is found on this species. Rather, the spin is located on
the nitrite and the remaining Osurf sites. This NO(g) release is
thus understood to reactivate the surface, although some Osurf
sites are replaced by surface peroxides. The fact that the unpaired
electron is shared among fewer Osurf sites results in a more NO2-
(100) surface. In principle, it would be possible for the nitrite
and nitrate to switch sites, and one possible rearrangement
comprises nitrate-to-nitrite oxygen transfer. Indeed, an additional
0.3 eV stability gain is observed upon such a rearrangement
(see Figure 5). The resulting nitrite ion formed by this redox
reaction can be understood to consist of an NO molecule bound
to a surface O- site. This species in the (nitrate on Ba)-(nitrite)
pair displays slightly greater stability toward NO desorption than
does NO2 in the (nitrite on Ba)-(nitrate) pair (1.1 eV vs 1.0
eV for the latter). Introduction of a third NO2 molecule allows
oxidation of the NO at the O-surf site, whereby a nitrate-nitrate
surface pair is formed and an NO(g) molecule is released. This
reaction comes out marginally endothermic by 0.2 eV. The
resulting structure is depicted in Figure 6.
To complete the picture, we describe the situation resulting
from a nitrate sitting at the Ba2+ site, depicted in Figure 7,
surf
when first removing NO/NO2 from the O-surf site. As expected,
the Ba2+surf nitrate is found to be more stable toward desorption
of an NO3(g) species than is Ba2+ nitrite toward desorption
surf
of NO2, i.e., 0.9 eV vs 0.7 eV. This is reflected in the greater

NOx Storage on BaO(100) Surface
J. Phys. Chem. B, Vol. 106, No. 1, 2002 143
Figure 10. Diagram summarizing the relative stabilities of the species considered in the present study. Transient species 1 and 2 correspond to
initial adsorptions of single NO2 molecules. Two channels for further NO2 storage, indicated A and B, refer to intermediate peroxide formation and
direct pairwise NO2 loading, respectively. Adsorption energies are given in eV; triplet excited state in brackets, denoted *.
TABLE 1: Results from Geometry Optimization Calculationsa
reaction
NO2(g) + BaO f Ba2+-[Osurf-(NO2)]2-
cf. Figure 10
∆E [eV]
0f1
0f2
-0.8
-0.7
0.4
-0.6
-1.0
-0.3
0.8
-0.1
1.1
0.2
-0.9
NO2(g) + BaO f Ba2+(NO2-)-O-
surf
NO2(g) + Ba2+(NO2-)-O-surf f Ba2+(NO2)-[Osurf-O]2- + NO(g)
2f4A
4Af5
2f3B
3Bf4B
3Bf5
3Bf6
4Bf5
4Bf6
5f6
Ba2+(NO2)-[Osurf-O]2- f Ba2+(NO3- )-Osurf
-
NO2(g) + Ba2+(NO2-)-O-surf f Ba2+(NO2- )-[Osurf-(NO2)]-
Ba2+(NO2- )-[Osurf-(NO2)]- f Ba2+(NO3- )-[Osurf-(NO)]-
Ba2+(NO3- )-[Osurf-(NO2)]- f Ba2+(NO3-)-O-surf + NO2(g)
NO2(g) + Ba2+(NO2-)-[Osurf-(NO2)]- f Ba2+(NO3- )-[Osurf-(NO2)]- + NO(g)
Ba2+(NO3- )-[Osurf-(NO)]- f Ba2+(NO3-)-O-surf + NO(g)
NO2(g) + Ba2+(NO3- )-[Osurf-(NO)]- f Ba2+(NO3- )-[Osurf-(NO2)]- + NO(g)
NO2(g) + Ba2+(NO3-)-O-surf f Ba2+(NO3- )-[Osurf-(NO2)]-
a The corresponding reaction scheme is depicted in Figure 10. ∆E is calculated according the expression: ∆E ) Eads(products) - Eads(reactants).
Negative ∆E corresponds to an exothermic reaction.
like species at the Ba site. This is seen in the slight shortening
of the bond length (0.01 Å) and the slightly increased bond
angle (119.5°) for this species.
The detailed energy balance in the peroxide formation channel
is emphasized. Such a bond displays a typical stability of 2.5
eV.25 The simultaneous N-O bond dissociation in the NO2
molecule, whereby NO(g) is formed, results in an accidental
approximate cancellation of energy contributions.
studies,6,8 an initial NO(g) release is observed during this
process. Three contributions to the experimentally observed
initial NO(g) desorption during NO2 loading (see Figure 11a)
were discussed above. These comprise NO2 reduction by surface
peroxide formation (see transient from 3A to 4A in Figure 10),
heterogeneous Ba2+-NO2 oxidation (see transient from 3B
-
to 6 in Figure 10), and heterogeneous oxidation of [Osurf-NO]-
(see transient from 4B to 6 in Figure 10). Both the latter
reactions along the reaction path B produce the formal surface
Ba(NO3)2 compound.
Finally, the oxidative power of the surface peroxy species
toward oxidation of the Ba2+ associated nitrite ion was tested.
The results give that a Ba2+ associated nitrate is the energetically
more favorable surface species by 0.2 eV. This property can be
taken to infer that spurious peroxides at the BaO surface do
not inhibit NOx loading at lean combustion conditions, since as
Temperature programmed desorption (TPD) measurements
of NOx (see Figure 11b) are determined to come from two
contributions. For a heating rate of 20 °C/min from 300 to 600
°C, NO2 is seen to desorb with a peak at 470 °C (see Figure
11b) in conjunction with an NO desorption peak at 530 °C.
This ordering is understood from the computed coverage
dependence of initial NO2 adsorption found in the range 0.7-
1.0 eV (see transient species 2 and 3B in Figure 10). When
considering the NO release, two sources contribute from
transient species 4B and one from the surface Ba(NO3)2 end
product. Direct NO release from transient species 4B produces
transient species 5 with an energy cost of 1.1 eV, while the
long as NO2 ions are present at the Ba2+ sites, these will act
-
to reduce the surface.
3.5. Results and Interpretation of Experiment. The theo-
retical results, presented in sections 3.1-3.4, are summarized
in Figure 10 and in Table 1. These findings are used to explain
the experimental results, depicted in Figure 11.
The NO(g) and NO2(g) concentrations during NO2 storage
at 400 °C are depicted in Figure 11a. As learned from other

144 J. Phys. Chem. B, Vol. 106, No. 1, 2002
Broqvist et al.
sponding adsorption energies range from 0.5 to 0.8 eV. The
adsorbate becomes more nitrite like with increasing adsorption
energy in the case of adsorption over a Ba2+ site. The initial
adsorption step involves the binding of one NO2 molecule over
either a surface O2- or a surface Ba2+ site. The former is
2-
formally an NO3 ion but is better understood as a [Osurf
-
NO2]2- species with the extra unpaired electron mainly localized
in an Osurf-NO2 antibonding orbital. This configuration displays
a binding energy of 0.8 eV, see Figure 1 and transient species
1 in Figure 10. The competing Ba2+ associated nitrite species,
see Figure 2 and transient species 2 in Figure 10, opens the
route toward surface Ba(NO3)2 formation. This is so because it
oxidizes the substrate, thus generating a delocalized electron
vacancy among the adjacent surface oxygen. The unpaired Osurf
electron, which results from NO2 binding at the Ba2+ site,
provides bonding for a second NO2 molecule, which can react
with the surface oxygen in two ways, depicted channel A and
B in Figure 10. Either a metastable peroxynitrite species
corresponding to channel A is formed, i.e., [Osurf-O-NO], or
a straightforward formation of surface nitrate species follows
(see channel B), i.e., [Osurf-NO2]-. The latter binds by 1.0 eV
to the surface, while the former constitutes an endothermic
transient species (see 3A in Figure 10). Interestingly, though,
the peroxynitrite is unstable toward surface peroxide formation
in conjunction with the evolution of NO(g). This surface
peroxide, in turn, is seen to be unstable toward decomposition
by oxidation of the nitrite located over the barium site, forming
a nitrate (see transient from 4A to 5 in Figure 10). This reac-
tion comes out exothermic by 0.6 eV, which implies that
inhibition of the NOx loading capacity of BaO(s) due to spur-
ious surface peroxides is deemed unlikely. Peroxide decomposi-
tion by oxidizing a Ba2+ nitrite in the presence of a
[Osurf -NO2]- surface nitrate (see Figure 4) leads to the all-
nitrate end product (see surface species 6 in Figure 10 and Figure
6).
If the second NO2 molecule approaches O-
with the
surf
nitrogen end, then reaction channel B follows (see transient
species 3B in Figure 10) and a stable nitrite-nitrate pair is
formed. Redox reactions among the nitrite-nitrate pairs lead
to the 0.3 eV more stable transient species 4B. Two additional
paths to nitrite oxidation at the surface Ba2+ site come about
from (i) the above surface peroxide acting as oxidant or (ii) an
Figure 11. (a) Concentration of NO and NO2 during exposure of 500
ppm NO2 to a BaO/Al2O3 sample at 400 °C. Note the appearance of
an NO peak at t ) 100 s and its exponential decay with time. (b)
Concentration of NO and NO2, detected during temperature pro-
grammed desorption experiment after exposure of 500 ppm NO2 to a
BaO/Al2O3 at 300 °C. Note the NO2 peak at 1250 s (460 °C) followed
by an NO peak at 1580 s (540 °C).
additional NO2 molecule acting as oxidant toward the Ba2+
-
surf
-
NO2 species. The latter corresponds to the transient from 3B
to 6 (see Figure 10).
subsequent indirect dissociation/decomposition of NO3 costs 0.9
eV. Similarly, NO2 release from Ba(NO3)2 costs 0.9 eV. This
implies a 0.7-1.0 eV stability range for the NO2(g) producing
surface species. The corresponding range for the NO(g) produc-
ing surface species is 0.9-1.1 eV. This can be taken to explain
the observed desorption ordering in Figure 11b.
Having said this, a contributing factor to the appearance of
the NO curve in Figure 11b originates to some extent from
desorbing NO2, which is converted to NO at, e.g., reactor walls,
since NO is thermodynamically favored in the experimental
temperature range. Support for the fact that at least part of the
NO(g) observed originates from direct desorption is found from
experiment on an Al2O3 system under similar conditions to those
employed here.26
Some straightforward observations regarding the initial NO-
(g) release observed experimentally during NO2 storage (see
Figure 11a) are highlighted. There are two different sources for
NO(g) evolution along the route to the all-nitrates end product.
These involve either an NO2(g) oxidant toward surface nitrite
oxidation or the channel for surface peroxide formation. Naively,
the probability for choosing the peroxide forming channel A is
twice that for choosing channel B. This observation exaggerates
the importance of this reaction channel due to the short lifetime
of the [Osurf -O-NO]- species, because it has an unreacted
desorbing NO2 molecule as a likely product. Still, this source
of NO(g) is expected to dominate at low temperatures and
sufficiently high coverages, as longer residence times and
availability of O-surf sites are necessary prerequisites. The sole
reducing agents at high coverages comprise the surface nitrites
[Osurf -NO]-, which are oxidized by NO2(g) to form surface
Ba(NO3)2 under NO(g) release.
4. Summary and Conclusions
The NOx storage process on BaO(s) is initiated by the
adsorption of one NO2 on a clean BaO(100) surface, as shown
in Figure 10. There are two different adsorption sites and several
different adsorption configurations for NO2(ads). The corre-
In conclusion, a first-principles semiquantitative mechanistic
investigation regarding the NOx loading property of BaO(s) has

NOx Storage on BaO(100) Surface
J. Phys. Chem. B, Vol. 106, No. 1, 2002 145
(6) Fridell, E.; Persson, H.; Olsson, L.; Westerberg, B.; Amberntsson,
A.; Skoglundh, M. Top. Catal. 2001, 16/17, 133.
(7) Olsson, L.; Persson, H.; Fridell, E.; Skoglundh, M.; Andersson, B.
J. Phys. Chem. B 2001, 105, 6895.
(8) Fridell, E.; Persson, H.; Westerberg, B.; Olsson, L.; Skoglundh,
M. Catal. Lett. 2000, 66, 71-74.
(9) Rodrigues, F.; Juste, L.; Potvin, C.; Tempe`, J. F.; Blanchard, G.;
Dje´ga-Mariadassou, G. Catal. Lett. 2001, 72(1-2), 59-64.
(10) Snis, A.; Panas, I. Surf. Sci. 1998, 413, 477.
(11) Nygren, M. A.; Pettersson, L. G. M. in manuscript.
(12) Payne, M. C.; Allan, D. C.; Arias, T. A.; Johannopoulus, J. D. ReV.
Mod. Phys. 1992, 64, 1045.
been undertaken and used to interpret experimental TPD data.
Signatures of the proposed mechanism include (a) initial
pairwise NO2 chemisorption at Ba2+ and Osurf sites to form
surface nitrites and nitrates, respectively, (b) a surface redox
step to form nitrate-nitrite pairs, and (c) subsequent heteroge-
neous oxidation of the remaining nitrite to form an all-nitrate
final product.
Acknowledgment. This work has been performed within
the Competence Centre for Catalysis, which is financially
supported by the Swedish National Energy Administration and
the member companies: AB Volvo, Johnson Matthey CSD,
Saab Automobile AB, Perstorp AB, MTC AB, Eka Chemicals,
and Swedish Space Corporation. Support from the Swedish
Research Council (P.B. and I.P.) is gratefully acknowledged.
Prof. Bengt Andersson is gratefully acknowledged for fruitful
discussions.
(13) Available from Accelrys Inc., formerly Molecular Simulation Inc.,
version 4.2, 1 December 1999.
(14) Sorescu, D. C.; Yates, J. T. J. Phys. Chem. B 1998, 102, 4556.
(15) Refson, K.; Wogelius, R. A.; Fraser, D. G.; Payne, M. C.; Lee, M.
H.; Milman, V. Phys. ReV. B 1995, 52, 10823.
(16) Rodriguez, J. A.; Hanson, J. C.; Chaturvedi, S.; Maiti, A.; and Brito,
J. L.; J. Chem. Phys. 2000, 112, 935.
(17) Vanderbildt, D. Phys. ReV. B 1990, 41(11) 7892.
(18) Monkhorst, H. J.; and Pack, J. D. Phys. ReV. B 1976, 13, 5188.
(19) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77,
3865.
References and Notes
(20) Schneider, W. F.; Li, J.; Hass, K. C. J. Phys. Chem. B 2001, 105,
6972.
(1) Shelef, M. Chem. ReV. 1995, 95, 209.
(21) Skoglundh, M.; Johansson, H.; Lo¨wendahl, L.; Jansson, L.; Dahl,
L.; Hirschbauer, B. Appl. Catal. 1996, B7, 299.
(22) Greenwood, N. N.; Earnshaw, A. Chemistry of the elements;
Pergamon Press: New York, 1984
(23) Ervin K. M.; Ho, J.; Lineberger W. C. J. Phys. Chem. 1988, 92,
5405.
(24) Davidson, J. A.; Fehsenfeld, F. C.; Howard, C. J. Int. J. Chem.
Kinet. 1977, 56, 17.
(25) Snis, A.; Panas, I. J. Chem. Phys. 1995, 103, 7626.
(26) Fridell, E., unpublished results.
(2) Takahashi, N.; Shinjoh, H.; Iijima, T.; Suzuki, T.; Yamazaki, K.;
Yokota, K.; Suzuki, N.; Miyoshi, N.; Matsumoto, S.; Tanizawa, T.; Tanaka,
T.; Tateishi, S.; Kasahara, K. Catal. Today 1996, 27, 63.
(3) Bo¨gner, W.; Kra¨mer, M.; Krutzsch, B.; Pischinger, S.; Voigtla¨nder,
D.; Wenninger, G.; Wirbeleit, F.; Brogan, M. S.; Brisley, R. J.; Webster,
D. E. Appl. Catal. B 1995, 7, 153.
(4) Fridell, E.; Skoglundh, M.; Johansson, S.; Westerberg, B.; To¨rn-
crona, A.; Smedler, G. Stud. Surf. Sci. Catal. 1998, 116, 537.
(5) Fridell, E.; Skoglundh, M.; Westerberg, B.; Johansson, S.; Smedler,
G. J. Catal. 1999, 183, 196.