Effect of CaO on the selectivity of N2O decomposition products: A combined experimental and DFT study

Article abstract of DOI:10.1016/j.susc.2016.04.004

The effect of CaO on N₂O decomposition and the selectivity of its decomposition products (NO and N₂) was investigated using a fixed-bed flow reactor with varying temperatures from 317°C to 947°C. The selectivity of NO from CaO-catalyzed N₂O decomposition is much lower than the N₂ selectivity with the N₂/NO products ratio greater than 12.1. Compared to N₂O homogeneous decomposition with the minimum N₂/NO products ratio of 6.2 at 718°C, CaO also decreases the NO selectivity from 718°C to 947°C. Density functional theory calculations provide possible N₂O decomposition routes on the CaO (1 0 0) surface considering both N₂ and NO as N₂O decomposition products. The N₂ formation route is more favorable than the NO formation route in terms of energy barrier and reaction energy, and NO formation on the CaO (1 0 0) surface is likely to proceed via N₂O + O_surf^2- → N₂ + O_{2, surf}^2- and N₂O + O_{2, surf}^2- → 2NO + O_surf^2-.

Full text of DOI:10.1016/j.susc.2016.04.004

Surface Science 651 (2016) 128–136
Contents lists available at ScienceDirect
Surface Science
journal homepage: www.elsevier.com/locate/susc
Effect of CaO on the selectivity of N O decomposition products: A
2
combined experimental and DFT study
a,b
b
b
b
b,
a,
Lingnan Wu , Xiaoying Hu , Wu Qin , Pan Gao , Changqing Dong ⁎, Yongping Yang ⁎
a
School of Energy, Power and Mechanical Engineering, North China Electric Power University, Beijing 102206, China
National Engineering Laboratory for Biomass Power Generation Equipment, North China Electric Power University, Beijing 102206, China
b
a r t i c l e i n f o
a b s t r a c t
Article history:
The effect of CaO on N O decomposition and the selectivity of its decomposition products (NO and N ) was inves-
2
2
Received 27 October 2015
Received in revised form 11 April 2016
Accepted 12 April 2016
tigated using a fixed-bed flow reactor with varying temperatures from 317 °C to 947 °C. The selectivity of NO
from CaO-catalyzed N O decomposition is much lower than the N selectivity with the N /NO products ratio
greater than 12.1. Compared to N O homogeneous decomposition with the minimum N /NO products ratio of
.2 at 718 °C, CaO also decreases the NO selectivity from 718 °C to 947 °C. Density functional theory calculations
2
2
2
2
2
Available online 16 April 2016
6
provide possible N
decomposition products. The N
2
O decomposition routes on the CaO (1 0 0) surface considering both N
2 2
and NO as N O
Keywords:
formation route is more favorable than the NO formation route in terms of
NO selectivity
Density functional theory
CaO (1 0 0)
2
energy barrier and reaction energy, and NO formation on the CaO (1 0 0) surface is likely to proceed via
2-
2-
2-
2-
N
2
O+Osurf→N
2
+O2,surf and N
2
O+O2,surf→2NO+Osurf.
2
N O decomposition
© 2016 Elsevier B.V. All rights reserved.
1
. Introduction
1
2
CaO þ SO2 þ O2→CaSO4 ΔH ¼ ꢀ481:4 kJ=mol
ð2Þ
Power generation from coal and biomass combustion is an impor-
tant source of nitrogen oxides (NO ) emissions. Compared to conven-
tional pulverized coal-fired power plants, circulating fluidized bed
CFB) boilers have lower furnace temperatures, which is beneficial for
reducing thermal NO formation. N O emission, however, soars to the
range of 20–360 mg/Nm [1,2], which is about ten times the amount
of conventional pulverized coal-fired power plants. N O is a powerful
greenhouse gas (310 times and 21 times greater than CO and CH , re-
x
The effect of limestone (or CaO) on the formation and decomposi-
tion of N O and NO has been extensively studied [6–17]. In general,
limestone addition can increase NO emissions [7,8,10–13], but experi-
mental results are controversial about the effect of CaO on N O emis-
sions depending on the experimental scale. For full-scale experiments,
Åmand et al. [7] found that limestone addition could reduce N O emis-
sion, while Bonn et al. [2] and Jensen et al. [10] found that there was lit-
tle influence, or the influence depended on the furnace temperature
[17]. For laboratory-scale experiments, it is widely agreed that CaO
2
(
x
2
2
3
2
2
2
4
spectively, in terms of global warming potential [3]) and it is speculated
to be the most important ozone-depleting substance in the 21st century
[
4]. Therefore, the effective control of both NO
equal importance for CFB boilers.
The in situ deSO process in the furnace of CFB boilers affects N
and NO emissions simultaneously. For atmospheric CFB boilers, the in
situ deSO process proceeds via Reactions (1) and (2) [5] with the lime-
stone calcined into calcium oxide (CaO), which acts as the SO sorbent
x
and N
2
O emissions is of
could reduce N
sor (mainly HCN) to generate N
position [15,16,18–22]. The attention of these studies has focused on the
catalytic ability of CaO to affect the formation and decomposition of
2
O emissions by decreasing the selectivity of N
2
O precur-
2
O decom-
2
O [9–12], or by catalyzing N
x
2
O
x
x
N
2
O, or the overall emissions of N
To the best of our knowledge, however, how CaO affects the selectivity
of N O decomposition products is not well studied.
2
O and NO in large-scale experiments.
x
in the furnace. CaO is therefore widely used to study the influence of
2
limestone on in-furnace processes of CFB boilers.
Besides experimental studies, theoretical investigations using densi-
ty functional theory (DFT) calculations have been performed to explain
CaCO3→CaO þ CO2 ΔH ¼ 182:1 kJ=mol
ð1Þ
the catalytic effect of CaO for N
0) surface was commonly used to study various CaO-involved pro-
cesses [24–26,30–32] as it is the most stable low-index surface of CaO
33]. Previous studies have reported possible N O adsorption geome-
tries on the CaO (1 0 0) surface [26,28,29], the O atom transfer from
O molecule to the surface [20,23,25,26,28], and the recovery of CaO
1 0 0) surface either by N O [23–25] or CO [27]. These studies outlined
the mechanism of N O decomposition on the CaO surface and
2
O decomposition [20,23–29]. The CaO (1
0
[
2
⁎
Corresponding authors.
N
(
2
E-mail addresses: wulingnan@126.com (L. Wu), huxy@ncepu.edu.cn (X. Hu),
qinwugx@126.com (W. Qin), gaopan@ncepu.edu.cn (P. Gao), cqdong1@163.com
C. Dong), yypncepu@163.com (Y. Yang).
2
(
2
http://dx.doi.org/10.1016/j.susc.2016.04.004
039-6028/© 2016 Elsevier B.V. All rights reserved.
0

L. Wu et al. / Surface Science 651 (2016) 128–136
129
considered N
fects NO formation from N
anism of NO formation during the N
important as both N O and NO emissions need to be strictly controlled.
To this end, we investigated the effect of CaO on NO formation and
the selectivity of N O decomposition products using a fixed-bed flow re-
actor. Possible NO formation mechanisms from N O decomposition on
2
and O
2
as N
2
O decomposition products, but how CaO af-
O decomposition is still unclear. The mech-
O decomposition process is also
selectivity of decomposition products with 14 mm inner diameter and
600 mm heating length (heated by an electric furnace). The furnace
temperature was monitored by a thermocouple mounted in the middle
of the electric furnace and controlled by a PID temperature controller.
The temperature of the reaction zone where the catalysts were placed
was measured before performing experiments, and its correlation
with the furnace temperature is shown in Fig. S2. A linear fitting be-
tween the furnace temperature and the reaction zone temperature
2
2
2
2
2
the CaO (1 0 0) surface were explored using DFT calculations. Homoge-
neous and CaO-catalyzed NO formation routes and their competition
2
was performed and the adjusted R value of 0.9887 was obtained. In
with the N
The results are beneficial for a better understanding of the effect of
CaO on N O decomposition and NO formation in CFB boilers, which
will contribute to proposing synergetic NO–N O–SO control strategies.
2
formation route from N
2
O decomposition was considered.
2
the following experiments, the effect of CaO on N O conversion, NO
yield and NO selectivity was investigated with furnace temperature
ranging from 300 °C to 950 °C with 100 °C temperature intervals. Corre-
sponding reaction temperatures are 317 °C, 418 °C, 520 °C, 620 °C,
2
2
x
718 °C, 812 °C, 902 °C, and 947 °C respectively.
2
. Methodology
Analytically pure CaO (98%) was used in all experiments. A sheet of
quartz wool was placed on the supporter for supporting CaO (for the
blank group, only a sheet of quartz wool was placed). The particle
diameter of CaO ranges from 0.18 to 0.25 mm. The CaO sample is char-
acterized using Brunner–Emmet–Teller (BET), X-ray diffraction (XRD),
scanning electron microscope (SEM), and X-ray fluorescence (XRF)
methods. Obtained BET specific surface area of the CaO sample is
2
.1. Experimental details
Fig. 1 shows the flowsheet of the laboratory-scale experimental ap-
paratus. Flow rates of N O and balance gas (argon) were controlled by
mass flowmeters and a total flow rate of 3000 mL/min at standard con-
2
2
ditions was selected. An on-line FTIR analyzer (MB3000, ABB) was used
2.761 m /g, and the average pore diameter is 10.58 nm. From the XRD
to analyze tail-gas composition. NO, N
with their IR spectrum shown in Fig. 2, and the concentrations of NO,
O and NO can be correlated with the intensity of the absorbance
spectrum. Measured values of standard NO, N O and NO gases at spe-
cific concentration are shown in Fig. S1. A good correlation can be ob-
tained when calibrating the measured concentrations of NO, N O and
NO against their known concentrations. In this way, the concentrations
of NO, N O and NO can be quantitatively obtained. However, N and O
cannot be detected as they are not IR active. To the best of our knowl-
edge, only N and NO have been discussed as possible N O homoge-
neous decomposition products [34]. Besides, NO is found to be
2
O and NO
2
are IR active species
result shown in Fig. S3, it can be seen that CaO is well crystallized, and
little amount of Ca(OH)
CaO and the H O in the air. The XRF result of the CaO sample is tabulated
in Table S1 in metal oxide form. The loss on ignition value is 3.86%,
which is speculated to be the contribution from H O captured by CaO
2
is detected as a result of the reaction between
N
2
2
2
2
2
2
2
in the air. The weight of CaO relative to the total weight detected by
XRF analysis is calculated to be 98.88%. The morphology of the CaO sam-
ple characterized by SEM method is shown in Fig. S4. From XRD result
2
2
2
2
2
2
(Fig. S3), only small amounts of Ca(OH) species are detected except
2
2
CaO. XRF result (Table S1) reports that the purity of CaO reaches as
high as 98.88%. It can also be seen from the XRF results that the impurity
of the CaO sample is composed of a mixture of 16 species rather than a
2
negligible (always lower than 5 ppm) from N
periments. Therefore, the concentration of N
2
O decomposition in ex-
is calculated according
2
single kind of species, of which Al
rities of the CaO sample with 0.55% and 0.38% weight ratio. Al
well-known inert material for N O decomposition and often used as cat-
alyst carrier [35]. MgO is more active than Al in decomposing N
but its activity is still lower than CaO [20]. Therefore, considering the
O
2 3
and MgO are the two main impu-
to the measured conversion of N
formation.
2
O and the concentration of NO
2 3
O
is a
2
The inlet N
2
O concentration was fixed at 2000 ppm. A quartz reactor
O decomposition and the
O
2 3
2
O
was used to study the effect of CaO on N
2
Fig. 1. Schematic of experimental apparatus.

130
L. Wu et al. / Surface Science 651 (2016) 128–136
2 2
Fig. 2. Absorbance spectrometry of (a) NO; (b) N O; (c) NO .
amount and the activity of detected impurities, the observed experi-
mental results of the decomposition and selectivity of N O are attribut-
ed to the effect of CaO rather than other species. To evaluate the effect of
CaO on N O conversion, NO yield and NO selectivity, the N O conversion
XN2O) is defined as:
periodic CaO surface slab model with a vacuum layer thickness of
92 Å. A N O molecule was placed 40 Å away from the CaO (1 0 0) surface,
and the electronic energies of the CaO (1 0 0) surface with N O mole-
cule, pure CaO (1 0 0) surface and N O molecule were calculated. Results
showed that the absolute value of BSSE error was only 0.002 eV, which
was very trivial compared to the energy profiles in this study.
CaO belongs to the cubic crystal system with Fm3m (225) space
group numbering. The lattice constant was 4.854 Å (4.883 Å [30] and
2
2
2
2
2
2
(
ꢀ
ꢁ
nN2O;out
nN2O;in
XN2O ¼ 1 ꢀ
ꢁ 100%
4
.840 Å [33]) after geometric optimization. The CaO (1 0 0) surface
where nN O,in is the inlet N
2
O concentration (ppm) of the reaction tube,
2
was used as it was the most stable low-index facet of CaO [33]. To estab-
lish the CaO (1 0 0) surface, a five-layer CaO (1 0 0) slab was cleaved
from a perfect CaO crystal with the top two layers fully relaxed and bot-
tom three layers frozen. A sheet of 12 Å vacuum layer was built above
the slab surface to eliminate the interference from imaging layers. The
surface was expanded with a 2 × 2 supercell containing 80 cations
and anions (40 Ca cations and 40 O anions) (Fig. 3). A 3 × 3 × 1
Monkhorst–Pack k-point sampling grid was used for Brillouin zone inte-
gration during geometric, energetic and transition state search calcula-
tions. Atomic radius orbital cut-off was selected as 4.8 Å. 0.005 Ha
Fermi smearing was selected for convergence acceleration. A combina-
tion of linear synchronous transit (LST) and quadratic synchronous
transit (QST) methods [40] were employed to search the transition
state of reactions. The transition state was then further optimized
with eigenvector following (EF) method to guarantee only one imagi-
nary frequency exists.
and nN2O,out is the outlet N
analyzer.
2
O concentration (ppm) measured by the FTIR
The NO yield (YNO) is defined as:
nNO;out
nN2O;in
YNO ¼
ꢁ 100%
where nNO,out is the outlet NO concentration (ppm) after the reaction
gas passes the catalyst layer.
The NO selectivity (SNO) is defined as:
nNO;out
nN2O;in ꢀ nN2O;out
SNO ¼
ꢁ 100%
2
.2. Computational details
DFT calculations were carried out using DMol3 code [36,37] with
To evaluate the interaction between the adsorbate and the CaO
(1 0 0) surface, the adsorption energy (Ead) is defined as:
generalized gradient approximation (GGA) and Perdew–Burke–
Ernzerhof (PBE) functional [38] for exchange and correlation potentials.
The atomic basis functions were expanded by double numerical basis
Ead¼ Esys ꢀ Eads ꢀ Esur
3
sets plus polarization function (DNP). DMol code with GGA and PBE
where Esys is the system total energy after adsorption, Eads is the energy
of adsorbate, and Esur is the energy of the surface before adsorption.
To consider the influence of temperature on the elementary reaction
steps, free Gibbs energy is calculated, which considers the zero point
functionals provided satisfactory calculation results with negligible
basis set superposition error [39], which was comparable with the
Gaussian 6–31G** basis set [25]. The BSSE error was tested using a

L. Wu et al. / Surface Science 651 (2016) 128–136
131
The Gibbs free reaction energy ΔG is the Gibbs free energy difference
between the product state (GFS) and the reactant state (GIS), which is:
ΔG=GFS-GIS
.
3. Results and discussion
3.1. Experimental results
2
The effect of CaO on N O conversion is illustrated in Fig. 4a. For the
blank group, N
2
O conversion is only 1.3% at 620 °C, and it increases
with the increase of reaction zone temperature, which is 5.0%, 30.8%,
0.4% and 94.5% at 718 °C, 812 °C, 902 °C and 947 °C. Although no cata-
lyst is placed in the reaction zone, the N O conversion even reaches
4.5%. at 947 °C, which agrees with the unstable thermodynamic prop-
8
2
9
2 2
erties of N O. The addition of CaO greatly enhances the N O conversion
for the temperature range from 620 °C to 947 °C. At 620 °C, XN2O reaches
5.0%, and it increases sharply to 99.5% at 718 °C. From 718 °C to 947 °C
the XN2O remains higher than 99.5%. Therefore, it is clear that CaO shows
strong catalytic ability for N O decomposition in comparison to N
Fig. 3. Structure of CaO (1 0 0) surface.
6
2
2
O
energy and the thermal contributions to the electronic energy from DFT
calculations obtained from vibrational analysis. The frequency analysis
is performed on the periodic surface slab. The translational and rotation-
al movements of surface adsorbed species are incorporated in vibration-
al analysis, and the vibrations of the atoms in the model are based on the
harmonic oscillator approximation. The harmonic oscillator approxima-
tion method, however, suffers from an underestimation of the entropies
of surface adsorbed species [41]. Based on vibrational analysis results,
enthalpy and entropy at different temperatures can be calculated, and
the Gibbs free energy can be obtained. The activation Gibbs free energy
homogeneous decomposition. This result is consistent with previous
experimental studies of Hansen et al. [15] using a laboratory-scale
fixed-bed quartz reactor in the temperature range from 800 °C to
950 °C, Snis et al. [20] using a fixed-bed quartz reactor in the tempera-
ture range from 500 °C to 950 °C, and Piskorz et al. [25] using a fixed-
bed microreactor from 400 °C to 650 °C.
The effect of CaO on NO yield from N
Fig. 4b. For the blank group and CaO-catalyzed N
group, NO is not formed from N O decomposition when the reaction
2
O decomposition is shown in
2
O decomposition
2
zone temperature is lower than 620 °C. When the temperature of the re-
action zone is higher than 620 °C, the NO yield begins to increase. There
is almost no difference for NO yield between the blank group and the
CaO group below 812 °C, but when temperature reaches 902 °C and
G
a
is the Gibbs free energy difference between the transition state (GTS
)
and the reactant state (GIS), which is:
9
47 °C, the NO yield of the blank group is 1.5% and 1.7% higher than
a
G =GTS-GIS.
the CaO group. Therefore, the addition of CaO could slightly reduce the
2
Fig. 4. The effect of CaO on (a) N O conversion; (b) NO yield and; and (c) NO selectivity.

132
L. Wu et al. / Surface Science 651 (2016) 128–136
2
NO yield from N O decomposition when temperature is between 812 °C
and 947 °C.
The generated O radical can also react with a second N
two NO molecules following Reaction (7):
2
O forming
The effect of CaO on NO selectivity during the N
process is shown in Fig. 4c. The NO selectivity below 620 °C is insignifi-
cant as NO will not be formed from N O decomposition when tempera-
ture is lower than 620 °C. When the reaction zone temperature reaches
18 °C, a significant increase of NO selectivity is seen for the blank
group. This increase for the blank group can be attributed to the low
conversion of N O and the increase of NO formation at 718 °C (around
4.0%). Its error range is around 4% due to the fact that the absolute con-
2
O decomposition
N O þ O → NO þ NO
2
ð7Þ
2
Reactions (5) to (7) consider the N–O bond cleavage during N
mogeneous decomposition. The possibility of N–N bond cleavage and its
2
O ho-
7
succeeding reaction with N
8) and (9):
2
O are also discussed according to Reactions
2
(
1
2
version of N O at 718 °C is so low that the small relative increase of the
NO product is prominently observable in the NO fraction. In comparison
to the catalyzed sample, this is correct, as the total conversion is much
N2O → NO þ N
ð8Þ
ð9Þ
N2O þ N → NO þ N2
2
larger at this temperature for the catalyzed process leading to N rather
than NO. The absolute conversion increases to 5% with about 1% NO,
leading to a selectivity of NO (out of all products) of about 20%. In
Fig. 4c, this is about 14% with a 5% error, and that matches well enough.
When reaction temperature further increases from 718 °C, a decreasing
trend for the blank group and an increasing trend for the CaO group is
seen, and the NO selectivity of the blank group is 7.2%, 3.4% and 2.2%
higher, respectively, than that of the CaO group. Therefore, the presence
Fig. 5 provides the energy profiles and the geometries of Reactions
(5) to (9). The homogeneous N–O bond cleavage of N O needs
2.926 eV to generate an O radical and a N molecule, while the genera-
tion of a N radical and NO is much more demanding with 5.768 eV need-
ed (Fig. 5a).
2
2
Then the generated O radical can attack both the O end and the N
of CaO is able to reduce the NO selectivity from N
18 °C to 947 °C. For the CaO catalyzed N O decomposition, the N
products ratio decreases with the increase of temperature. Its value is
09.7 at 718 °C, 19.9 at 812 °C, 14.8 at 902 °C and the N /NO products
ratio reaches the lowest value of 12.1 at 947 °C. For the homogeneous
O decomposition, the N /NO value reaches its lowest value of 6.2 at
2
O decomposition from
end of a second N
shown in Fig. 5b. The generation of N
barrier of 1.017 eV. By comparison, the reaction between O radical and
O only needs to overcome a 0.400 eV energy barrier and generates
an ONNO complex. The generated ONNO complex needs additional
0.476 eV to detach into two NO molecules, which is the rate determin-
ing step for NO formation. Another possibility of NO formation is
shown in Fig. 5c, which is caused by the reaction of N radical with a sec-
2
O molecule following Reactions (6) and (7) as
7
2
2
/NO
2
and O needs to pass the energy
2
1
2
N
2
N
2
2
7
9
18 °C, and it gradually increases to 8.6 at 812 °C, reaches a peak of
.3 at 902 °C and decreases to 9.2 at 947 °C.
Argon is used as the balance gas instead of N
between the diluent and generated O . The generated N
composition, however may also react with the generated O
the Zeldovich mechanism through N +O→NO+N and O
2
to avoid interference
from N O de-
following
+N→
ond N
2
O molecule, with the energy barrier of 1.305 eV. Comparing the
O decomposition is determined by the bond
2
2
2
above possible routes, N
cleavage step, and the cleavage of the N–O bond is more likely than
the N–N bond as a matter of lower energy barrier. For this section, the
2
2
2
2
NO+O [42]. This reaction is not considered in the following discussion
as thermal formation of NO is not significant until temperature exceeds
energy profile of N
parison with CaO-catalyzed N
2
higher quality calculations on N O homogeneous reaction with O ( P)
based on ab initio CASPT2//CASSCF methods can be referenced else-
where [34].
2
O homogeneous decomposition is provided for com-
2
O decomposition. Further details of
3
1
9
800 K (1527 °C) [43], which is beyond the temperature range (below
47 °C) of this study. Furthermore, when using a N and O mixture as
2
2
the inlet gas (with 2:1 M ratio) from room temperature to 947 °C, NO
was not detected. This is consistent with the Zeldovich mechanism
[
42] and rules out the NO formation route from homogeneous reaction
between N and O
To investigate the effect of CaO on NO formation during the N
3.3. Mechanism of N
(1 0 0) surface
2 2 2
O decomposition into N and O on the CaO
2
2
.
2
O cat-
alytic decomposition process, the mechanism of NO formation on the
CaO (1 0 0) surface is studied in the following sections based on density
functional theory calculations.
The mechanism of N
0 0) surface is investigated and summarized in Fig. 6 for comparison
with NO formation routes. N
CaO (1 0 0) surface initiates from the O atom transfer from a N
2 2 2
O decomposition into N and O on the CaO (1
2
O decomposition into N
2
and O
2
on the
O mol-
2
3
.2. Homogeneous decomposition of N
2
O
ecule to the surface O anion site, forming surface peroxy species with
a 0.989 eV energy barrier, which is much lower than that of homoge-
2
The mechanism of N O homogeneous decomposition is first investi-
2
neous N O decomposition (2.926 eV). The effect of exchange and corre-
gated for comparison purposes. Two kinds of N-containing products
lation functionals on the energy barrier of O atom transfer step is listed
in Table S2 in supplementary materials. It is found that GGA + PW91,
GGA + BP, GGA + BOP, GGA + VWNBP, and LDA + VWN give similar
results compared to the GGA + PBE functional, while GGA + BLYP
generated from N
2
O decomposition are considered, N
2
and NO, as
formulated in Reactions (3) and (4):
and LDA
+ PWC gives relatively lower energy barriers and
2
N2O → 2N2 þ O2
N2O → 2NO þ N2
ð3Þ
ð4Þ
GGA + HCTH gives a much higher energy barrier.
Surface recovery needs to be taken into account at high surface cov-
2
erage as the atomic O from N
face O anion. The surface adsorbed atomic O from N
can be removed following the Eley–Rideal (ER) mechanism by reacting
with a second N O (Fig. 6a) or the Langmuir–Hinshelwood (LH) mech-
anism by recombination with a neighboring adsorbed atomic O
Fig. 6b). The calculated energy barriers of the ER and LH mechanisms
2
O decomposition poisons the active sur-
2
O decomposition
2 2 2
The elementary reactions of N O decomposition into N and O
2
following Reaction (3) are:
(
are 1.200 eV and 1.400 eV respectively, and both of them are higher
than that of the O atom transfer process. Therefore, whether under
N2O → N2 þ O
ð5Þ
ð6Þ
high or low surface coverage conditions, the energy barrier of N
2
O de-
N2O þ O → N2 þ O2
composition on the CaO (1 0 0) surface is much lower than the

L. Wu et al. / Surface Science 651 (2016) 128–136
133
Fig. 5. Energy profile of N
O.
2 2
O homogeneous decomposition: (a) N–O and N–N bond cleavage; (b) reaction between O radical and a second N O; (c) reaction between N radical with a second
N
2
homogeneous case, and CaO is therefore able to catalyze N
sition, which explains the experimental observations.
2
O decompo-
bond cleavage of N
proceed.
2
O, this route is still very energy intensive in order to
Partial density of states (PODS) analysis is performed to investigate
the electronic properties of the adsorbed atomic O and N on the surface
3
.4. Mechanism of N
2
O decomposition into NO and N
2
on the CaO
(
1 0 0) surface
O anion site (Fig. 8). The 2s and 2p orbitals of the O atom of N
surface O anion are located at different energy levels, which is consis-
tent with the weak interaction between the surface and the N O mole-
cule before the O atom transfer process. When the O atom of N O is
2
O and the
In analogy with the NO homogeneous formation mechanism, two
NO formation routes from N O decomposition on the CaO (1 0 0) surface
2
2
2
are considered, depending on the order of bond cleavage (the N–O bond
cleavage first and N–N bond cleavage first). The CaO (1 0 0) surface
transferred to the surface O anion site, a good match between the elec-
tronic states of the adsorbed atomic O and the surface O anion is seen at
−21.1 eV, −19.2 eV and −14.1 eV (2s orbital) and −5.0 eV, −4.1 eV
and −0.2 eV (2p orbital). A covalent bond is formed between the
adsorbed atomic O and the surface O anion after the O atom transfer
could catalyze the N–O bond cleavage of N
2
O according to Reaction
(
10) (the O atom transfer step) or the N–N bond cleavage according
to Reaction (11):
process. Similarly, the electron states of the N atom of N
face O anion are located in different energy levels before N atom transfer
from the N O molecule to the surface O anion. When N O reaches the
chemisorption state, the electrons of the N atom overlaps the surface
O anion in a much more delocalized range from −4.8 eV to Fermi
level (2p orbital). With the cleavage of the N–N bond of the
2
O and the sur-
2
ꢀ
→N2 þ O2ꢀ
N2O þ O
ð10Þ
ð11Þ
surf
2;surf
2
2
N2O þ O →NO þ N ꢀ O2ꢀ
2ꢀ
surf
surf
where O 2s u- rf is the unsaturated five-fold O anion site on the CaO
2
chemisorbed N O, the electron states of the adsorbed N atom and the
O anion are still in good match but in a more localized area with the
peaks at −5.6 eV and Fermi level (2p orbital), indicating a covalent
bond is formed between the adsorbed N and the surface O anion.
The activity of the adsorbed atomic N and O on the CaO (1 0 0) sur-
face O anion site for NO generation is then investigated by studying its
(
1 0 0) surface.
The energy profiles of the N–O and N–N bond cleavage catalyzed by
the CaO (1 0 0) surface are shown in Fig. 7. The cleavage of the N–O bond
of N O is greatly facilitated by the energy barrier decrease from 2.926 eV
to 0.989 eV as mentioned previously. For the N–N bond cleavage on the
CaO (1 0 0) surface, an energy barrier of 1.412 eV is needed for the N
2
reaction with a second N
2
O molecule (Fig. 9). Reactions between the
2
O
2
N O molecule and the surface adsorbed atomic O and N proceeds fol-
molecule to chemisorb with a bent NNO structure on the surface O
anion site. The chemisorption process needs to adsorb 0.671 eV, and fur-
ther N–N bond cleavage needs 3.312 eV to generate a NO molecule and a
surface adsorbed atomic N. Although the CaO surface catalyzes the N–N
lowing Reactions (12) and (13):
N2O þ O ꢀ → 2NO þ O2ꢀ
2
ð12Þ
2;surf
surf

134
L. Wu et al. / Surface Science 651 (2016) 128–136
2
Fig. 8. PDOS analysis of the adsorbed atomic O and N before and after N O decomposition.
2
Fig. 6. N O decomposition routes on the CaO (1 0 0) surface: (a) following Eley–Rideal
mechanism; (b) following Langmuir–Hinshelwood mechanism.
the CaO surface and detach into two free NO molecules. The lower
route of Fig. 9 is the reaction between a second N O with surface
adsorbed atomic N, which needs to pass an energy barrier of 0.616 to
form a surface adsorbed NO molecule and a N molecule, with
.940 eV heat release. The adsorbed NO molecule needs 0.612 eV to de-
2
N2O þ N ꢀ O →NO þ N2 þ O2ꢀ
2ꢀ
ð13Þ
surf
surf
2
2
The energy profiles of Reactions (12) and (13) are illustrated in
Fig. 9, and the upper part shows the reaction route between surface
sorb from the CaO (1 0 0) surface. It is clear that the surface adsorbed N
is more active than the surface adsorbed atomic O, but the formation of
atomic N is much harder than that of atomic O with energy barriers of
3.983 eV and 0.989 eV respectively. Therefore, the NO formation from
2
adsorbed atomic O with a second N O forming two NO molecules. The
N
2
O molecule will first overcome an energy barrier of 1.043 eV to attack
the adsorbed atomic O forming an ONNO species with its N end
adsorbed on the surface, which further needs 1.645 eV to desorb from
N
2
O decomposition is more likely to proceed with an O atom transfer
O+O 2s urf →N
-
+O2,surf) followed by a subsequent
2-
process first (N
2
2
Fig. 9. NO formation from the reaction between surface adsorbed atomic O and N with a
second N O molecule (the surface adsorbed O is highlighted for clarity).
2
Fig. 7. Energy profiles of N–O and N–N bond breakage of N O on the CaO (1 0 0) surface.
2

L. Wu et al. / Surface Science 651 (2016) 128–136
135
reaction between the adsorbed atomic O and the N
2
O molecule forming
O+O 22 ,surf→2NO+Osurf), which needs to pass an
overall energy barrier of 1.918 eV.
surface adsorbed bent NNO species decreases with the increase of
temperature. The bent NNO species on the CaO (1 0 0) surface needs
formidable energy to break its N–N bond and forms NO and a surface
adsorbed atomic N, making this NO formation route less likely than
the other reaction route. Fig. 10c is the effect of temperature on the re-
action energy of the same reaction steps as in Fig. 10a. For O atom trans-
fer, ER, and LH surface recovery processes, the reaction energy is always
-
2-
two NO molecules (N
2
3
.5. Temperature dependence of elementary reaction steps
As the above analysis only considers the electronic energy of respec-
tive reaction processes, the influence of temperature on each respective
reaction step is investigated in this part to incorporate the thermal con-
tributions to the elementary reaction steps. Fig. 10a shows the influence
negative for the considered temperature range, indicating that N
decompose spontaneously on the CaO (1 0 0) surface. However, the re-
action energy of adsorbed atomic O reacting with a following N O into
ONNO is always positive, indicating that it cannot proceed spontaneous-
ly. The reaction energy of N O reacting with the CaO (1 0 0) surface into
the bent NNO species is also greater than zero for the calculated temper-
ature in Fig. 10d, and this process cannot proceed spontaneously for the
calculated temperature range.
2
O can
2
of temperature on the energy barrier of N
1 0 0) surface with initial N–O bond cleavage. The black line in Fig. 10a
depicts the temperature influence on O atom transfer from N O to the
2
O decomposition on the CaO
(
2
2
surface O anion. A decreasing trend is observed with increasing temper-
ature for the calculated temperature range. The red lines in Fig. 10a
show the energy barrier change of surface recovery processes (follow-
ing the ER and LH routes). Similarly, both of them show a decreasing
trend with the increase of temperature. O atom transfer and ER or LH
In terms of the above calculated activation Gibbs free energy and
Gibbs free reaction energy, N
face is more likely to generate N
and LH surface recovery process, while NO formation is comparatively
less likely to be generated from N O decomposition. This can be attrib-
uted to the higher energy barrier of NO formation routes compared to
formation routes, and the fact that NO formation routes cannot pro-
ceed spontaneously. Therefore, N has a higher selectivity than NO
when N O decomposes on the CaO (1 0 0) surface and the calculated re-
2
O decomposition on the CaO (1 0 0) sur-
2
and O following O atom transfer
2
surface recovery processes are succeeding reaction steps for N
composition into N and O . The energy barrier of the O atom transfer
step is lower than the ER and LH surface recovery, indicating that N
decomposition into N and O is determined by the surface recovery
2
O de-
2
2
2
2
O
N
2
2
2
process. The blue lines of Fig. 10a depict the influence of temperature
on NO formation routes from adsorbed atomic O on the surface O
2
2
anion reacting with a following N
2
O molecule. The energy barrier of
sults are consistent with our experimental observations. Meanwhile, for
the homogeneous case, calculated results show that the generated O
2
N O reacting with the adsorbed atomic O decreases with the increase
of temperature, which forms surface adsorbed ONNO species, and its
energy barrier is lower than the LH and ER surface recovery routes.
The energy need for cleaving the formed ONNO into two NO molecules
first decreases and then increases with the increase of temperature,
which is higher than the LH and ER surface recovery routes and deter-
mines the rate for NO formation. Another NO formation route from
radical is more likely to react with a secondary N
than forming N (which is offset by the O radical recombination reac-
tion). The calculated selectivity for N is higher than NO on the CaO (1
0 0) surface, so the selectivity of NO formation from N O decomposition
is higher when CaO is present compared to N O homogeneous decom-
position, which is in agreement with the experimental observations
that CaO decreases the selectivity of NO from N O decomposition com-
pared to the N O homogeneous decomposition.
2
O forming NO rather
2
2
2
2
N
2
O decomposition on the CaO (1 0 0) surface with initial N–N bond
2
cleavage is illustrated in Fig. 10b. The energy barrier for N O to form a
2
2
Fig. 10. Effect of temperature on the (a) energy barrier of elementary reaction steps with N–O bond cleavage at first; (b) energy barrier of elementary reaction steps with N–N bond
cleavage at first; (c) reaction energy of elementary reaction steps with N–O bond cleavage at first; (d) reaction energy of elementary reaction steps with N–O bond cleavage at first.

1
36
L. Wu et al. / Surface Science 651 (2016) 128–136
[
[
3] United States Environmental Protection Agency, Overview of GreenhouseGases, (in)
http://epa.gov/climatechange/ghgemissions/gases/n2o.html.
4] A.R. Ravishankara, J.S. Daniel, R.W. Portmann, Science 326 (2009) 123.
4
. Conclusions
Our experiments investigated the effect of CaO on N
2
O decomposi-
[5] E.J. Anthony, D.L. Granatstein, Prog. Energy Combust. Sci. 27 (2001) 215.
[
[
[
6] J.E. Johnsson, Fuel 73 (1994) 1398.
7] L.-E. Åmand, B. Leckner, K. Dam-Johansen, Fuel 72 (1993) 557.
8] K. Dam-Johansen, L.-E. Åmand, B. Leckner, Fuel 72 (1993) 565.
tion and the selectivity of its decomposition products. Results show
that CaO can catalyze N O decomposition compared to the homoge-
neous case, and N has a higher selectivity than NO. Meanwhile, CaO de-
creases the NO selectivity compared to the homogeneous case. The N
NO products ratio of N O homogeneous decomposition reaches the
minimum of 6.2 at 718 °C, while the N /NO products ratio of CaO cata-
lyzed N O decomposition reaches its minimum of 12.1 at 947 °C. Densi-
ty functional calculations were used to investigate the N and NO
formation routes from N O decomposition on the CaO (1 0 0) surface.
Two NO formation routes were proposed and compared to N and O
formation routes from N O decomposition both homogenously and het-
erogeneously on the CaO (1 0 0) surface. N formation from N O decom-
2
2
[9] A.N. Hayhurst, A.D. Lawrence, Combust. Flame 105 (1996) 511.
[10] A. Jensen, J.E. Johnsson, K. Dam-Johansen, Symp. Combust. 26 (1996) 3335.
11] W. Lin, J.E. Johnsson, K. Dam-Johansen, C.M. van den Bleek, Fuel 73 (1994) 1202.
12] T. Shimizu, Y. Tachiyama, D. Fujita, K. Kumazawa, O. Wakayama, K. Ishizu, S.
Kobayashi, S. Shikada, M. Inagaki, Energy Fuel 6 (1992) 753.
[13] T. Shimizu, Y. Tachiyama, A. Kuroda, M. Inagaki, Fuel 71 (1992) 841.
14] L.A.C. Tarelho, M.A.A. Matos, F.J.M.A. Pereira, Fuel 85 (2006) 967.
15] P.F.B. Hansen, K. Dam-Johansen, J.E. Johnsson, T. Hulgaard, Chem. Eng. Sci. 47 (1992)
419.
[16] B. Shen, T. Mi, D. Liu, B. Feng, Q. Yao, F. Winter, Fuel Process. Technol. 84 (2003) 13.
2
/
[
[
2
2
2
[
[
2
2
2
2
2
[
[
17] A. Lyngfelt, B. Leckner, Fuel 72 (1993) 1553.
18] V. Barišić, A.K. Neyestanaki, F. Klingstedt, P. Kilpinen, K. Eränen, M. Hupa, Energy
Fuel 18 (2004) 1909.
2
2
2
position is more likely than the NO formation route as a result of the
relatively lower energy barrier and lack of spontaneous NO formation
route. NO formation on the CaO (1 0 0) surface is more likely to proceed
[19] V. Barišić, F. Klingstedt, A. Naydenov, P. Stefanov, P. Kilpinen, M. Hupa, Catal. Today
100 (2005) 337.
[
[
20] A. Snis, H. Miettinen, J. Phys. Chem. B 102 (1998) 2555.
21] X. Hou, H. Zhang, S. Yang, J. Lu, G. Yue, Chem. Eng. J. 140 (2008) 43.
2
-
2-
2-
2-
via N
2
O+Osurf→N
2
+O2,surf and N
2
O+O2,surf→2NO+Osurf
.
[22] D. Shun, H.S. Chang, Y.S. Park, D.H. Bae, G.T. Jin, Korean J. Chem. Eng. 18 (2001) 630.
[
[
[
23] E.J. Karlsen, M.A. Nygren, L.G.M. Pettersson, J. Phys. Chem. A 106 (2002) 7868.
24] A. Snis, I. Panas, J. Chem. Phys. 103 (1995) 7626.
25] W. Piskorz, F. Zasada, P. Stelmachowski, A. Kotarba, Z. Sojka, J. Phys. Chem. C 117
Acknowledgments
(2013) 18488.
The authors are grateful for the support of the National Natural Sci-
ence Foundation of China (Grant No. 51206045), the Major State Basic
Research Development Program of China (973 Program) (Grant No.
[26] L.N. Kantorovich, M.J. Gillan, Surf. Sci. 376 (1997) 169.
[
[
[
27] A. Snis, I. Panas, D. Strömberg, Surf. Sci. 310 (1994) L579.
28] A. Snis, D. Strömberg, I. Panas, Surf. Sci. 292 (1993) 317.
29] L. Wu, W. Qin, X. Hu, C. Dong, Y. Yang, Appl. Surf. Sci. 333 (2015) 194.
2
015CB251501), 111 Project (Grant No. B12034), and the Fundamental
[30] S. Zhang, X. Hu, Q. Lu, J. Zhang, Energy Fuel 25 (2011) 2932.
[
[
[
31] L. Wang, Appl. Surf. Sci. 257 (2011) 5499.
32] D. Strömberg, Surf. Sci. 275 (1992) 473.
33] P. Broqvist, H. Grönbeck, I. Panas, Surf. Sci. 554 (2004) 262.
Research Funds for the Central Universities (Grant No. 2015ZZD02).
Special thanks to Aric Rousso from Princeton University for grammatical
revisions to this paper.
[34] M. González, R. Valero, R. Sayos, J. Chem. Phys. 115 (2001) 2540.
[
35] K. Doi, Y.Y. Wu, R. Takeda, A. Matsunami, N. Arai, T. Tagawa, S. Goto, Appl. Catal. B
Environ. 35 (2001) 43.
Appendix A. Supplementary data
[
[
[
[
[
36] B. Delley, J. Chem. Phys. 92 (1990) 508.
37] B. Delley, J. Chem. Phys. 113 (2000) 7756.
38] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865.
39] Y. Inada, H. Orita, J. Comput. Chem. 29 (2008) 225.
40] N. Govind, M. Petersen, G. Fitzgerald, D. King-Smith, J. Andzelm, Comput. Mater. Sci.
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.susc.2016.04.004.
28 (2003) 250.
[
[
[
41] C.T. Campbell, J.R.V. Sellers, J. Am. Chem. Soc. 134 (2012) 18109.
42] Y.B. Zeldovich, Acta Physicochim. URSS 21 (1946) 577.
43] P. Glarborg, A.D. Jensen, J.E. Johnsson, Prog. Energy Combust. Sci. 29 (2003) 89.
References
[
[
1] M.A. Wójtowicz, J.R. Pels, J.A. Moulijn, Fuel Process. Technol. 34 (1993) 1.
2] B. Bonn, G. Pelz, H. Baumann, Fuel 74 (1995) 165.