Studies of the mechanism of ammonia oxidation into nitrous oxide over Mn-Bi-O/α-Al2O3 catalyst

Article abstract of DOI:10.1016/j.jcat.2003.09.029

A complex of kinetic and physicochemical methods: temperature-programmed surface reaction (TPSR), pulsing NH₃ or NH₃/ ¹⁶O₂ (¹⁸O₂) reaction mixture, and infrared and photoelectron spectroscopies, was used for characterization of the highly selective supported manganese-bismuth oxide catalysts and for the study of the mechanism of the ammonia oxidation. Ammonia oxidation was demonstrated to proceed via alternating reduction and reoxidation of the catalyst surface with participation of the lattice oxygen. NH₃ interacts with weakly bonded oxygen species through hydrogen atom abstraction to form adsorbed [N] species, which are localized on Mn²⁺ and Mn^δ+ (2<δ<3). Manganese ions with different oxidation degrees (Mn ³⁺ (Mn⁴⁺) and Mn^δ+) serve as active sites of the catalyst surface. The correlation between the selectivity toward N₂O and the portion of manganese in the Mn³⁺ (Mn ⁴⁺) state was established. Bismuth oxide plays an important role by increasing the quantity, mobility, and thermal stability of the subsurface oxygen. The reaction kinetic scheme is suggested based on the experimental results. Numerical simulation of TPSR data confirms the reliability of the proposed reaction mechanism.

Full text of DOI:10.1016/j.jcat.2003.09.029

Journal of Catalysis 222 (2004) 129–142
www.elsevier.com/locate/jcat
Studies of the mechanism of ammonia oxidation into nitrous oxide
over Mn–Bi–O/α-Al O catalyst
2
3
a,∗
a
b
a
a
a
E.M. Slavinskaya, S.A. Veniaminov, P. Notté, A.S. Ivanova, A.I. Boronin, Yu.A. Chesalov,
a
a
I.A. Polukhina, and A.S. Noskov
a
Boreskov Institute of Catalysis SB RAS, Pr. Akad. Lavrentieva, 5, 630090 Novosibirsk, Russia
b
Solutia Europe Louvain-la-Neuve, Belgium
Received 7 July 2003; revised 28 August 2003; accepted 1 September 2003
Abstract
A complex of kinetic and physicochemical methods: temperature-programmed surface reaction (TPSR), pulsing NH or NH / O ( O )
1
6
18
3
3
2
2
reaction mixture, and infrared and photoelectron spectroscopies, was used for characterization of the highly selective supported manganese–
bismuth oxide catalysts and for the study of the mechanism of the ammonia oxidation. Ammonia oxidation was demonstrated to proceed
via alternating reduction and reoxidation of the catalyst surface with participation of the lattice oxygen. NH interacts with weakly bonded
3
2
+
δ+
oxygen species through hydrogen atom abstraction to form adsorbed [N] species, which are localized on Mn
and Mn (2 < δ < 3).
3
+
4+
δ+
Manganese ions with different oxidation degrees (Mn (Mn ) and Mn ) serve as active sites of the catalyst surface. The correlation
3
+
4+
(Mn ) state was established. Bismuth oxide plays an
between the selectivity toward N O and the portion of manganese in the Mn
2
important role by increasing the quantity, mobility, and thermal stability of the subsurface oxygen. The reaction kinetic scheme is suggested
based on the experimental results. Numerical simulation of TPSR data confirms the reliability of the proposed reaction mechanism.

2003 Elsevier Inc. All rights reserved.
Keywords: Ammonia oxidation mechanism; Nitrous oxide; Supported manganese–bismuth oxide catalyst; TPSR; Pulse method
1
. Introduction
(2) NO, as the main intermediate, interacts with NHx to
form N2 and with Nads to form N2O [6,7].
(
3) Formation of a dimmer (hydrazine) from NH2 species
results in the formation of N2, and formation of nitroxyl
Numerous papers are devoted to the mechanism of low-
temperature ammonia oxidation into nitrous oxide and nitro-
gen on heterogeneous catalysts. Experimental studies were
carried out with metals [1–7] and oxides [8–14]. Ammonia
oxidation was shown to occur by alternating reduction and
oxidation of the catalyst, similar to the oxidation of hydro-
gen, CO, and general organic substances [1].
Numerous authors consider formation of NHx species
through oxydehydrogenation of ammonia as the first reac-
tion stage. As to the next stages, there are several opinions:
(
HNO) through interaction of NHads and Oads leads to
formation of N2O [10,12,13].
A common feature of these stages is that the oxygen–
catalyst bond is either formed or broken. The catalytic activ-
ity and selectivity are shown [8,15] to substantially depend
on the oxygen bond energy measured by the heat of reac-
tion: 0.5O2 + [ ] → [O]. The higher the heat of reaction,
the higher the selectivity to nitrogen. Nitrous oxide selec-
tivity decreases in the transition metals series as follows:
Pt > Pd > Ni > Fe > W > Ti.
Similarities between catalytic ammonia oxidation with
gaseous oxygen and solid oxide reduction with ammonia [8]
provide independent evidence for the participation of [O]
species and the determination of the role of their surface con-
centration in the reaction kinetics. These similarities are as
follows:
(
1) Successive abstraction of hydrogen atoms results in the
formation of Nads species. Interaction of Nads, Oads, and
two Nads produces NO and N2, respectively, [2,4,5].
*
Corresponding author.
E-mail addresses: lesla@catalysis.nsk.su (E.M. Slavinskaya),
ppnott@solutia.com (P. Notté).
0021-9517/$ – see front matter  2003 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2003.09.029

130
E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
(
(
(
(
1) products of the same composition are formed during the
catalytic reaction and the reduction;
2) longer reduction time results in an overall reduction of
the rate of ammonia transformation;
3) the rate of the catalytic reaction equals the reduction rate
at the same surface oxygen coverage;
4) selectivity to nitrogen increases while selectivity to ni-
trous oxide decreases when the reduction time is ex-
tended, i.e., when the surface oxygen coverage shrinks
and, correspondingly, the surface bond energy increases.
ble impregnation of the support: incipient wetness impreg-
nation of a weighed support sample with a solution of mixed
manganese (II) and bismuth (III) nitrates taken in a certain
◦
volume was followed by drying at 80–130 C. The dried
sample was impregnated once again with a solution of the
◦
said nitrates and dried at 80–130 C.
All the synthesized samples were calcined at 400, 550,
◦
and 750 C for 2 or 4 h. The salts were manganese nitrate,
Mn(NO3)2·6H2O, Aldrich index 28,864-0, and bismuth ni-
trate, Bi(NO3)3·5H2O, Aldrich index 38,307-4. Alumina in
α- and γ -modifications were used as the supports.
Hence, interaction of ammonia with the surface is accom-
panied by electron transfer from ammonia molecules to the
catalyst.
The mechanism of ammonia oxidation over complex ox-
ide catalysts is not clearly understood. However, some analo-
gies between the ammonia oxidation occurring on complex
oxide catalysts, on metals, and on simple oxides are pre-
sented in the literature. These are similar temperature prod-
uct formation relations [16] and the same product composi-
tions formed by reduction and by catalytic oxidation reaction
These manganese–bismuth oxide catalysts contain mul-
tiple phases [19]. Apart from the support, they comprise
individual manganese oxides (MnO2, Mn2O3), bismuth ox-
ide (Bi2O3), and their interaction products.
◦
Catalysts Mn–Bi–O/α-Al2O3 calcined at 400 C com-
prise MnO2, traces of undecomposed Bi(NO3)3, and α-
Al2O3. A higher temperature or longer time of the thermal
treatment favors the formation of phases of β-Mn2O3, α-
Bi2O3, and their interaction product, Bi2Mn4O10. The pro-
portion of the latter increases as the temperature is elevated
[
8,17]. The same correlations between oxygen bond energy
◦
to 750 C, MnO2 and Bi2O3 phases not being detected upon
and the catalyst activity and selectivity are valid both for
complex catalysts and for simple oxides. It should be noted
that earlier investigations dealt with low-selectivity catalysts
◦
treating at 750 C (Table 1).
Approximately the same phase composition is estab-
◦
lished for catalysts Mn–Bi–O/γ-Al2O3 calcined at 400 C.
(
less then 40%) with respect to nitrous oxide. Supported
Elevation of the calcination temperature or a longer calcina-
tion gives rise to formation of the following phases: Mn2O3,
α-, β-Bi2O3, Bi2Mn4O10; bismuth aluminate Bi2Al4O9 also
being formed. The proportion of the two latter phases in-
creases with the calcination temperature. The lattice constant
equal to 7.906 Å is characteristic of the support phase (γ -
Al2O3) that is lower than the constant of the pure oxide.
These data indicate formation of a γ -Al2O3-based solid so-
lution.
manganese–bismuth oxide systems have been proposed [18]
as selective catalysts for the oxidation of ammonia into ni-
trous oxide. The present study aims at understanding the
mechanism of ammonia oxidation over Mn–Bi oxides/α-
Al2O3 (noted as Mn–Bi–O/α-Al2O3) particularly in the 85–
9
0% selectivity range toward nitrous oxide.
2
. Experimental and methods
Based on flowthrough reactor data, Mn–Bi–O/α-Al2O3
is more selective toward nitrous oxide compared with Mn–
Bi–O/γ -Al2O3 with respective selectivities between 80 and
85% versus between 43 and 59%; such selectivity provided
by Mn–Bi–O/α-Al2O3 further increases at higher calcina-
tion temperatures.
2
.1. Catalysts/gases
Supported manganese–bismuth oxide catalysts were pre-
pared according to the procedure described elsewhere [19].
Mn–Bi–O-supported catalysts were synthesized using dou-
Table 1
Physicochemical properties Mn–Bi oxide catalysts
Catalyst
Calcination
Active component
S
Phase composition
specific
2
◦
temperature ( C) MnO ,
Bi O , Mn/Bi, Σ(Mn + Bi), (m /g)
2
2 3
2
(duration, h)
mass (%) mass (%) mass
(atom/nm )
Mn–Bi–O/α-Al O
400(2)
13.0
9.0
1.01
75.1
10
8.7 β- MnO , β-Mn O , α-Bi O , α-Al O
2 3
MnO , Bi(NO ) (traces), amorphous phase, α-Al O
2
3
2 3 3 2 3
550(2)
550(4)
750(2)
2
2
3
2
3
8.0 β-MnO , β-Mn O , α-Bi O , Bi Mn O , α-Al O
2
2
3
2
3
2
4
10
2 3
7.5 β-Mn O , Bi Mn O , α-Al O
2
3
2
4
10
2 3
Mn–Bi–O/γ -Al O
400(2)
21.1
12.0
1.23
5.8
100
101
103
β- MnO , Bi O , Bi(NO ) (traces), γ -Al O
2 2 3 3 3 2 3
2
3
550(2)
50(4)
β- MnO , β-, α-Bi O , γ -Al O
2 2 3 2 3
5
β-MnO , Mn O , β-, α-Bi O , Bi Mn O , Bi Al O ,
2
2
3
2
3
2
4
10
2
4 9
γ -Al O
2
3
750(2)
97
β-Mn O , Bi Mn O , Bi Al O , γ -Al O
2 3 2 4 10 2 4 9 2 3

E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
131
All gases used in this study are high purity (about
9.999%). Isotech provided 18O2 with an isotopic purity of
about 99%.
sorbed thereon. Spectra of the surface compounds were iden-
tified by subtracting the background spectrum of the catalyst
recorded before flowing ammonia from the acquired spectra.
9
2
.2. Methods
2 3 2 3
2.2.4. XPS studies of Mn–Bi–O/α-Al O , Mn–O/α-Al O ,
and Bi–O/α-Al2O3 samples
Catalytic properties were studied by two methods using
the RXM-100, catalyst testing/characterization equipment
ASDI, USA), installed at Solutia Inc (Pensacola, USA),
equipped with multigas pulsing and on-line mass spectrom-
etry. Mass axes were calibrated and corrected digital output
numbers used for each mass were followed up during the
experiments. Signals were deconvoluted using breakdown
patterns determined under the same conditions prior to the
experiments.
An X-ray photoelectron spectroscopy (XPS) technique
was used to study states of oxygen and the surface species
formed during the catalyst treatment with ammonia, to de-
termine surface concentrations and chemical states of these
compounds. XPS spectra were acquired in a VG ESCALAB
spectrometer, which was calibrated with respect to the bind-
(
ing energies of the reference levels Au4f7/2 (E = 84.0 eV)
b
and Cu2p3/2 (E = 932.7 eV) [20] using AlK radiation
b
α
(hν = 1486.6 eV). Catalyst samples were fixed on a holder
either by rubbing into a fine-mesh nickel grid or by sup-
porting from their alcohol suspension. The charging effect
during photoemission was taken into account using an in-
ternal reference (line Al2p of the α-alumina support with
E = 73.8 eV), and the C1s line (E (C1s) = 284.8 eV) of
1
6
18
2
.2.1. Pulsing of ammonia or ammonia/ O2 ( O2) on the
catalyst
Typically 0.1 to 0.4 g of fresh catalyst is placed in a
flowthrough quartz reactor connected to the RXM-100. The
b
b
◦
catalyst is heated to 350 C and pretreated with 11 pulses
the surface carbon contaminations [20,21]. The concentra-
tion of surface species was calculated based on the mea-
surements of the intensity (I) of certain XPS lines by the
following formula:
of 0.25 cc of oxygen spaced at 120 s each. Helium at 40
cc/min was used as a carrier gas resulting in an oxygen pulse
length of 0.375 s. Pulses of ammonia or reaction mixture
1
6
18
(
53% ammonia/47% O2 ( O2)) are also passed through
Ix/(ASF)x
the pretreated catalyst.
nx/n
Al
=
,
IAl/(ASF)Al
2
.2.2. Temperature-programmed surface reaction (TPSR)
TPSR of chemisorbed ammonia with catalyst oxygen was
where nx is the element concentration (at%) in the analy-
sis zone, (ASF)i are atomic sensitivity factors of elements.
According to the available data [21], the following atomic
sensitivity factors were used: Al2p = 0.193, N1s = 0.477,
O1s = 0.711, and Mn2p = 2.420.
conducted over Mn–Bi–O/α-Al2O3 calcined at 400, 550,
and 750 C and over Mn–Bi–O/γ -Al2O3 calcined at 400 C.
The catalysts were pretreated in helium at 450 C and evac-
uated at room temperature. Catalysts were pretreated with
ammonia at room temperature and further treated at room
temperature with helium to remove excess and labile ammo-
nia. Temperature-programmed reaction was conducted by
heating such catalyst from room temperature to 650 C at
a rate of 15 C/min.
◦
◦
◦
The package of proprietary CALC programs was used
for detailed spectra processing. The CALC programs al-
low carrying out the curve fitting analysis using Gaussian–
Lorenzian (GL) with a variable Gaussian proportion.
A sample of Mn–Bi–O/α-Al2O3 catalyst calcined at
◦
◦
◦
550 C was pretreated in helium at 473 K for an hour, treated
with ammonia at room temperature for another hour, and
blown by air for 30 min.
2
.2.3. In situ IRS studies
In situ IR studies were carried out in a Fourier IR spec-
trometer BOMEM MB-102 using a high-temperature flow
quartz IR cell reactor with CaF2 optic windows. A cata-
XPS spectra were recorded at different temperatures (300
to 769 K) for samples of Mn–Bi–O/α-Al2O3 catalyst cal-
◦
cined at 400, 550, and 750 C, samples of Mn–O/α-Al2O3
◦
lyst pellet (1 × 3 cm, 50 mg) was mounted in the cell (V =
and Bi–O/α-Al2O3 calcined at 400 C.
3
1
.5 cm ). Before each run the tablet was pretreated directly
◦
in the IR cell in flowing air at 250 C for an hour. The cell
was cooled to the required temperature and flowing ammo-
nia was substituted for air. Ammonia was passed through the
cell during the period of time needed to reach some station-
ary ammonia concentration in the gas phase inside the cell
and then stopped and the cell was sealed. In several minutes
the ammonia gas concentration decreased practically to be-
low the detectable level. The procedure was repeated again.
IR spectra were acquired, for 1.5 min (30 scans) each, in
the course of the experiment. The spectra looked like a su-
perposition of spectra of the catalyst and compounds ad-
3. Results
◦
3.1. Reduction of Mn–Bi–O/α-Al2O3—500 C catalyst with
ammonia
A reduced catalyst sample was prepared by exposing an
oxygen-treated catalyst to a series of fifty 0.25 cc ammonia
pulses.
Mass spectrometry data show that ammonia is oxidized
by catalyst oxygen to produce N2O, N2, and NO. Fig. 1

132
E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
Fig. 1. Rate constants of catalyst reduction with ammonia as a function of
the degree of surface reduction.
Fig. 3. Selectivity to reaction products (N O, N , and NO) during ammonia
oxidation vs the degree of surface reduction.
2
2
Fig. 2. Selectivity to reaction products (N O, N , and NO) during catalyst
2
2
reduction with ammonia vs the degree of surface reduction.
presents the rate constant of ammonia oxidation versus the
degree of reduction of the catalyst surface. The rate constant
of ammonia oxidation is calculated by the flow reactor equa-
tion k = −(ln(1 − x))/τ, where x is ammonia conversion,
Fig. 4. Selectivity of the oxidized and reduced catalysts to reaction products
N O, N ) versus the number of reaction mixture pulses.
(
2
2
−15
2
τ = 2.24 × 10
cm s/molecules is the contact time.
3.2. Interaction of the reaction mixture (NH3 + O2) with
The degree of catalyst reduction is calculated as the ra-
tio of the amount of oxygen eliminated from the catalyst to
the total amount of surface oxygen species from the catalyst
monolayer. It is assumed that the monolayer coverage cor-
oxidized and reduced catalysts
Pulses of the reaction mixture containing ammonia and
oxygen were fed to reduced and oxidized catalyst samples.
Figs. 3 and 4 demonstrate that the oxidized catalyst is
more selective to nitrous oxide and less selective to nitrogen
compared with the reduced catalyst. Variations in selectiv-
ity to N2O and N2 are observed with increasing time of
catalyst treatment with the reaction mixture (Fig. 4). Selec-
tivity to N2O decreases by using the oxidized catalyst and
increases with the reduced catalyst, while selectivity to nitro-
gen decreases with the reduced catalyst and increases with
the oxidized form.
The degree of surface reduction of a stationary operat-
ing catalyst can be determined from comparative rate con-
stants of the reduction and of the catalytic reaction over the
oxidized and reduced catalysts [22]. It is 24.6% for the ox-
idized catalyst (point 1 in Fig. 1) and 77% for the reduced
catalyst (point 2 in Fig. 1). The further extension of the treat-
ment time may lead to equal activities and selectivities with
3
2
responds to the amount of 0.238 cm O2/m of the catalyst
surface [22]. The catalyst activity is seen to decrease with an
increased degree of reduction. The amount of oxygen elimi-
nated from the catalyst during the whole reduction period is
3
3
.805 cm ; this is almost half of the total amount of oxygen
in the catalyst’s active components. This shows the rather
high mobility of bulk oxygen. Elimination of half of the
catalyst oxygen species can cause changes in the phase com-
position. XRD data [19] show that phases of Mn3O4, MnO,
and bismuth metal are present in the ammonia-reduced cata-
◦
lyst at 400 C.
Correspondingly, selectivity to the reaction products
changes with the increased catalyst surface reduction (Fig. 2):
the selectivity to nitrogen increases while the selectivity to
nitrous oxide decreases; selectivity to nitrogen oxide is un-
changed and remains low.

E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
133
(a)
Fig. 5. Total oxygen consumption for the formation of the reaction products,
oxygen consumption for formation of products from gas phase, and oxygen
consumption for formation of products from the catalyst vs the number of
reaction mixture pulses.
respect to N2O and N2 unless irreversible phase transforma-
tions occur during reduction.
Thus, the reaction of ammonia oxidation is probably pro-
ceeding through alternate reduction and reoxidation of the
catalyst surface.
In order to verify if the catalyst oxygen species are in-
volved in the reaction, the total oxygen consumption for the
formation of the reaction products was compared to the oxy-
gen consumption for the formation of the products from the
gas phase. In addition, the composition of the reaction prod-
(b)
Fig. 6. Amount of ammonia converted into N 16O, N 18O, N (a) and into
1
8
16
ucts was studied by substitution of O2 for O2 in the
reaction mixture containing ammonia and oxygen.
2
2
2
16
18
N
O, N O (b) vs the duration of the catalyst treatment with the reaction
18
mixture containing NH and O .
3
.3. Comparison of oxygen consumption for the formation
3
2
of the reaction products
3
.4. Reaction with labeled oxygen 18O2
Fig. 5 shows the dependence of the total oxygen con-
sumption for the formation of the reaction products (Σ[O])
and of the oxygen consumption for the formation of the
Fig. 6 shows the amount of ammonia converted to N216O,
N218O, N O, N O, and N versus the number of pulses
16
18
2
0
current
) (where q0 is
products from the gas phase (q −q
of the reaction mixture. Since the stationary selectivity to
nitrogen does not change, the amount of nitrogen formed
through ammonia oxidation by 16O2 and by O2 can be
separately calculated. Results show that the amount of prod-
[
O]
[O]
[O]
current
the oxygen amount in a pulse and q
is the oxygen
[O]
18
amount in gas phase) versus the number of pulses of the
reaction mixture. The difference between the total oxygen
amount consumed for the formation of the reaction prod-
ucts and the oxygen amount consumed for the formation of
the reaction products from the gas phase equals the amount
of oxygen consumed from the catalyst for the formation of
the products. The amount of oxygen eliminated from the
catalyst decreases with the increasing number of pulses. As
soon as the stationary state is reached, the amount of oxygen
consumed from the gas phase becomes equal to the amount
quantified in the oxidation products. The amount of oxygen
eliminated from the catalyst during the catalytic reaction can
be calculated based on these data. It is 21.5% of the mono-
layer. That agrees well with the above estimated degree of
surface reduction (24.6%) for the stationary operating cata-
lyst.
1
6
ucts formed by ammonia oxidation with O2 gradually de-
1
8
creases while it increases for O2. Data on the consumption
1
6
18
of O2 and O2 during the course of the reaction are shown
in Fig. 7. During the experiment, an equivalent amount of
3
16
2.47 cm of O2 is transferred from the catalyst in the re-
action products; this corresponds to 2.3 oxygen monolayers.
This corresponds to one-third of the total oxygen of the cata-
lyst (7.8 monolayers) as assessed by hydrogen reduction in
a temperature-programmed reduction [19]. These data sup-
port the assumption of a high mobility for the bulk oxygen
species.
Therefore, it is reasonable to conclude that catalyst oxy-
gen is involved in the reaction of ammonia oxidation through
alternate reduction and reoxidation of the catalyst surface.

134
E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
16
18
Fig. 7. Consumption of
O
and
O
for the formation of the reaction
2
2
products vs the duration of the catalyst treatment with the reaction mixture
Fig. 10. Temperature-programmed surface reaction of ammonia over
18
◦
containing NH and O .
Mn–Bi–O/α-Al O calcined at 750 C.
3
2
2 3
Fig. 8. Temperature-programmed surface reaction of ammonia over
◦
Mn–Bi–O/α-Al O calcined at 400 C.
2
3
Fig. 11. Temperature-programmed surface reaction of ammonia over
◦
Mn–Bi–O/γ -Al O calcined at 400 C.
2
3
Evolution of nitrous oxide from Mn–Bi–O/α-Al2O3 cal-
◦
cined at 400 C occurs over a broad range of temperatures
◦
with maximum at 176, 305, and 496 C, while nitrogen
evolution shows two sharp peaks with maximum at 300
◦
and 598 C. Nitrogen oxide evolved at a much higher tem-
perature compared with nitrous oxide with a maximum at
◦
◦
5
33 C. Catalyst calcination at 550 C leads to very similar
◦
results whereby calcination at 750 C results in considerable
changes. Indeed, the low-temperature maximum for nitro-
gen evolution is shifted toward higher temperatures (377 C)
but the high-temperature peak (at 598 C) disappears. Two
peaks appearing at 458 and 516 C characterize the NO evo-
◦
◦
◦
lution. Formation of nitrous oxide is proportionally reduced
and characterized by three peaks with maximum at 183, 233,
and 350 C.
Fig. 9. Temperature-programmed surface reaction of ammonia over
◦
◦
Mn–Bi–O/α-Al O calcined at 550 C.
2
3
Fig. 11 shows the TPSR observed for Mn–Bi–O/γ-Al2O3
◦
3
.5. Temperature-programmed surface reaction of
calcined at 400 C. Nitrous oxide appears as three poorly
◦
chemisorbed ammonia
separated peaks with maximum at 205, 318, and 605 C.
Three maximum at 187, 336, and 605 C are observed for
◦
Figs. 8–11 show that no ammonia is formed during the
temperature-programmed surface reactions.
nitrogen formation and two peaks with maximum at 505 and
605 C for nitrogen oxide.
◦

E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
135
Fig. 12. IR spectra of surface compounds acquired after adsorption of am-
Fig. 13. N1s spectrum acquired with the Mn–Bi–O/α-Al O catalyst (cal-
cined at 550 C) after treatment with ammonia followed by blowing with
oxygen. The spectra are recorded at 300 K.
2
3
◦
monia by the Mn–Bi–O/α-Al O catalyst calcined at 550 C. The spectra
◦
2
3
are recorded during desorption in helium at different temperatures.
3
.5.1. In situ IR studies
Fig. 12 shows spectra of surface species (SS) formed
E = 397.0 eV is within the region, which is characteristic
b
of metal nitrides or nitrogen adsorbed on the surface of met-
upon interaction of NH3 with the surface at room temper-
ature (20 C) at single and double feeding. In the IR spectra
als and oxides. The value E = 399.3 eV is in the end of
b
◦
the spectral N1s region, which is typical both for the nitro-
−
1
absorption bands at 1610, 1545, and 1411 cm are ob-
gen constituent of oxynitrides [24–26] and for the nitrogen
constituent of adsorbed NO, N–H, NHOH, or nitrosyl com-
plexes [20,21,23]. Therefore, XPS data alone are insufficient
for unambiguous conclusions about the nature of nitrogen
−1
served. A low intense band at 1610 cm relates to adsorbed
−1
molecular ammonia. The bands at 1545 and 1411 cm , in
accordance with available data [10], are assigned to NH2 and
NH adsorbed species, respectively. The repeated feeding of
ammonia results in an increase in the concentration of SS.
The SS concentrations do not practically change after
blowing with helium at room temperature. Elevation of the
temperature during blowing with helium results in a pro-
gressive decrease in the SS concentration; the bands of SS
with E = 399.3 eV. At the same time, the above discussed
b
IRS data indicate the absence of bands related to N–O and
N–H bonds in the samples treated with ammonia and blown
by air at room temperature or by helium at 393 K. Conse-
quently, it seems reasonable to assign the lines at E = 397.9
b
2
+
and 399.3 eV to nitrogen species localized on Mn and
Mn (2 < δ < 3) ions, respectively. It is shown in the pre-
◦
δ+
disappear from the IR spectra upon blowing at 120 C. Also
SS concentration decreases if blowing with air at room tem-
perature for 2 min occurs. Note that the reaction products or
NH3 are not detected in the gas phase while the SS concen-
trations decrease in the course of these experiments. Hence,
the decrease is accounted for not by desorption of the com-
pounds to the gas phase but by their further transformation
to form SS which are not detected in the measured frequency
vious paper [19] that species of manganese in different oxi-
dation states are formed upon treatment of the catalysts with
the reaction mixture or ammonia.
The surface nitrogen concentration expressed as the
N/Mn ratio is calculated from N1s and Mn2p intensity ra-
tio; it equals ca. 0.5. The study of the thermal stability of
these nitrogen species reveals that the line N1s decreases in
intensity on heating the sample immediately in the analyzer
chamber of an electron spectrometer that makes nitrogen un-
detectable in the XPS spectra at 700 K.
−1
range of the IR spectra (4000–1300 cm ).
3
.5.2. XPS studies of Mn–Bi–O/α-Al2O3, Mn–O/α-Al2O3,
and Bi–O/α-Al2O3
XPS studies were used to identify surface species formed
on adsorption of ammonia. The survey XPS spectra demon-
strate that a N1s line, which is not seen before treatment,
becomes observable after adsorption of ammonia and blow-
ing with air. The chemical state of the element is established
by the high-resolution inspection of the spectra N1s region
4. Discussion
4.1. Studies of surface intermediates
In situ IRS studies of the surface species formed through
interaction of ammonia with Mn–Bi–O/α-Al O —550 C
◦
(
see Fig. 13). Curve fitting shows that the resulting spec-
2
3
trum is composed of two components with Eb = 397.0 eV
and 399.3 eV. The analysis of the literature XPS data of
nitrogen-containingmaterials and nitrogen-containingmole-
cules on the oxide surfaces has been done [20,21,23]. Com-
parison with the experimental results shows that the peak at
3
reveal that NH abstracts hydrogen atom during the inter-
action to form NH and NH . Blowing with helium at a tem-
2
◦
perature elevation up to 120 C or blowing with air at room
temperature results in the disappearance of SS bands from
IR spectra. Evolution of either NH3 or the reaction products

136
E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
is not observed in this case but XPS data indicate that ni-
trogen remains on the surface. Therefore, it is reasonable to
conclude that the further abstracting of hydrogen atoms oc-
curs to produce nitrogen atoms by the parallel pathway:
NH2 → N,
NH → N.
XPS data are evidence of the localization of nitrogen atoms
on Mn2 and Mn (2 < δ < 3) ions.
+
δ+
4
.2. Studies of temperature-programmed surface reaction
of chemisorbed ammonia
As ammonia is not desorbed during the TPSR, the reac-
tion rate constant (Kr) is much higher than the desorption
rate constant (Kd) in the process:
Fig. 14. Temperature dependence of TPSR rate of formation of nitrogen in
linearized coordinates.
Kd
Kr
◦
NH3(g)
NH (a)
N2O(g) + N2(g) + NO(g).
One can see for Mn–Bi–O/α-Al O calcined at 400 C
3
2
3
that the amount of sites responsible for the formation of
nitrous oxide and nitrogen is approximately half of the to-
tal number of ammonia adsorption sites at the temperature
Therefore, the surface ammonia concentration during
heating is completely determined by the rate of the surface
reaction. Reaction activation energy (Er) can be calculated
by Eq. (1) with the assumption that A = 10
◦
range used for the catalytic reaction (315 to 350 C). Among
1
3
−1
s [27],
them, 70% are sites responsible for the formation of nitrous
oxide. The proportion of the sites responsible for the forma-
tion of nitrous oxide at low activation energy (27 kcal/mol)
is ca. 0.6% of the total number of ammonia adsorption sites.
Higher catalyst calcination temperatures result in a pro-
portional increase of these sites: 1.7% after calcination at
ꢀ
ꢁ
A
Er
2
p
Er = RT exp −
,
(1)
β
RTp
Tp is the peak maximum temperature; β is the heating rate
15 /min).
The quantity of the reaction products formed during the
TPSR is used to calculate the amount of oxygen removed
from the catalyst. It is 21.8% of monolayer.
The appearance of several peaks of the same reaction
product in the course of TPSR may indicate nonuniformity
of the active surface with respect to the given reaction, i.e.,
the occurrence of various types of active sites. This is valid
when there is no influence of the processes of diffusion, des-
orption, and reaction kinetics on the characteristics of the
spectrum.
◦
(
◦
◦
5
50 C and 2.6% after calcination at 750 C. Also calcina-
◦
tion at 750 C creates sites with higher activation energy (30
kcal/mol) at about 7.3%. In this case, increasing activation
energy for nitrogen formation is observed in agreement with
an increase in the selectivity to N2O.
◦
In the Mn–Bi–O/γ -Al2O3 calcined at 400 C, the pro-
portion of the sites responsible for the formation of nitrous
oxide is lower than that of the sites responsible for the for-
mation of nitrogen. At the same time, the proportion of sites
that produce nitrogen, with low metal–oxygen bond energy
activation energy of about 27.6 kcal/mol), is much higher
than the proportion of the sites responsible for the formation
of nitrous oxide.
The reaction order was determined for the formation
of nitrogen by TPSR. A linear correlation is obtained for
the nitrogen formation expressed as ln(W/N ) versus 1/T
From TPSR data, nitrous oxide starts releasing into the
gas phase at temperatures above 120 C. Therefore, one can
(
◦
conclude on the basis of IRS and XPS data that there are
only species of one type, namely [N], before the evolution
of gaseous products. Then, stages of NH2 and NH transfor-
mations into the N species are very fast, so the hydrogen
abstraction reactions do not determine the shape of TPSR
spectrum.
Formation of nitrous oxide during TPSR needs to break
the oxygen–catalyst bond. Hence, the existence of several
temperature peaks of nitrous oxide evolution may argue for
the presence of surface sites with different energies of the
oxygen–catalyst bonds.
2
(
Fig. 14), where W is the reaction rate and N the concentra-
tion of the adsorbate. This leads to a probable second order
for the nitrogen formation [28]. It is difficult to provide for
a correct determination of the orders for the formation of
nitrous oxide and nitrogen oxide because of the poorly re-
solved peaks. Nevertheless, by analogy with nitrogen, the
former reaction order is probably 2.
Table 2 presents the TPSR results and more particularly
the calculated activation energies, the amounts of products
released, and the fractional quantity of ammonia adsorption
sites. Poorly resolved peaks of nitrous oxide were geometri-
cally divided in order to evaluate the respective amounts of
N2O produced from the different sites.
4.3. Studies of catalyst reduction by ammonia
Studies of the catalyst reduction with ammonia reveal that
the reduction rate constant decreases with an increased de-
gree of surface reduction (Fig. 1), the reduction being higher

E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
137
Table 2
Comparative kinetic TPSR parameters of chemisorbed ammonia over various catalysts
◦
Catalyst/calcination
temperature ( C)
Peak
T ( C)
Er (kcal/mol)
Amount of product
2
Fractional quantity of NH
3
adsorption sites (%)
◦
(µmol/m
)
cat
◦
Mn–Bi–O/α-Al O —400 C
N
2
N O
2
300
176
34.6
26.9
34.9
46.8
49.2
53.3
0.148
0.007
0.290
0.063
0.893
0.173
13.1
0.6
25.8
5.6
39.7
15.3
2
3
3
4
05
96
NO
533
598
N
2
◦
Mn–Bi–O/α-Al O —550 C
N
N O
2
301
176
34.6
26.9
34.9
46.8
47.1
49.9
53.5
0.138
0.017
0.238
0.080
0.703
13.7
1.7
23.6
7.9
2
3
2
3
4
06
96
NO
500
44
34.8
5
N
N
601
0.187
18.5
2
2
◦
Mn–Bi–O/α-Al O —750 C
377
183
39.4
27.3
30.4
37.7
44.5
48.1
0.690
0.038
0.108
0.167
0.970
46.4
2.55
7.26
11.2
32.6
2
3
N O
2
2
3
33
50
NO
458
16
5
◦
Mn–Bi–O/γ -Al O —400 C
N
2
187
27.6
36.8
53.7
28.7
35.7
53.7
47.4
53.7
0.635
0.48
38.5
29.0
32.5
2
3
3
6
36
05
N O
2
205
3
6
18
05
NO
505
05
0.535
6
◦
α-Al2O3 catalyst calcined at 550 C. This is reflected by the
presence of surface sites of different natures: there are three
types of such sites responsible, respectively, for the forma-
tion of N2O, N2, and NO. In addition, each type may be
energetically inhomogeneous. For example, three types of
sites (activation energies equal to 27, 35, and 47 kcal/mol)
are responsible for the formation of nitrous oxide, two types
of sites (activation energies equal to 35 and 53.5 kcal/mol)
for the formation of nitrogen, and two types of sites (activa-
tion energies equal to 47 and 50 kcal/mol) for the formation
of nitrogen oxide. A significant reduction in the rate con-
stant observed during ammonia oxidation into nitrous oxide
is explained by the fact that N2O formation requires that a
larger number of oxygen–catalyst bonds are broken com-
pared with the formation of N2 [5]. Elimination of weakly
bonded oxygen results in a reduction of the overall rate con-
stant of ammonia consumption and in a significant reduction
in the rate constant for ammonia oxidation into N2O com-
pared to the rate constant of ammonia oxidation into N2.
As a consequence, the selectivity to N2 increases with the
oxygen–catalyst bond energy [5,29–32]. On the other hand,
the lower reduction rate constant may result from the lock-
ing of the active surface with strongly adsorbed species. The
related TPSR data show that approximately half of the sites
are occupied by strongly bonded species eliminated at con-
Fig. 15. Rate constant of ammonia oxidation into reaction products (N O,
2
N , and NO) versus the degree of reduction of the catalyst surface.
2
for the oxidation of ammonia into nitrous oxide compared
with oxidation to nitrogen (Fig. 15). These observations may
be explained by a low mobility of the bulk oxygen, by the
locking of the active surface with strongly bonded ammo-
nia species, or by a different oxygen state at the catalyst
surface. This study shows that the mobility of the bulk oxy-
gen is significant; therefore, the lower reduction rate con-
stant can be attributed to the latter two factors. The different
states of oxygen at the catalyst surface are indeed observed
from the TPSR data (Fig. 9) obtained with the Mn–Bi–O/

138
E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
siderably higher temperature than the normal reaction tem-
perature.
4
.4. Influence of calcination temperature on the proportion
of various sites for ammonia adsorption
Based on the TPSR data, we observe an increase in the se-
lectivity of Mn–Bi–O/α-Al2O3 to nitrous oxide with higher
calcination temperatures. This results from an increase in
the number of ammonia adsorption sites having a lower
manganese–oxide bond energy (low activation energy of
the reaction) responsible for the formation of nitrous ox-
ide. The oxygen–manganese bond energy is 62.2 kcal/gat
O in MnO2, 76.1 kcal/gat O in Mn2O3, and 82.8 kcal/gat
O in Mn3O4 [33]. Hence, an increase in the proportion of
the sites with lower bond energy means an increase in the
proportion of the oxidized manganese state. This conclu-
sion agrees with the data cited elsewhere [19] and indicates
(a)
that the Mn3 /Mn ratio increases with higher calcination
temperature. The selectivity to N2O varies with this ratio.
Hence, it is reasonable to conclude that the reduced man-
ganese species are responsible for the formation of nitrogen
and the oxidized species for the formation of nitrous oxide.
The selectivity of the catalyst on γ -alumina is low due to a
higher proportion of low bond energy sites involved in the
formation of nitrogen.
+
δ+
(b)
4
.5. The role of bismuth oxide in the manganese–bismuth
Fig. 16. (a) Relative intensities of O1s/Al2p lines in XPS spec-
tra of samples recorded at different temperatures: 1, α-Al O ; 2,
oxide catalyst
2
3
◦
Bi–O/α-Al O ; 3, Mn–O/α-Al O ; 4, Mn–Bi–O/α-Al O —750 C; 5,
2
3
2
3
2 3
◦
◦
Comparative data on activity and selectivity toward N2O
obtained with samples of Mn–O/α-Al2O3, Bi–O/α-Al2O3,
and Bi2Mn4O10/α-Al2O3 [19] demonstrate that the selectiv-
ity to N2O is around 75% on the Mn–O/α-Al2O3 catalyst,
the selectivity is equal to 63.6% on the Bi2Mn4O10/α-Al2O3
catalyst, while the Bi–O/α-Al2O3 is not active in ammo-
nia oxidation. Thus, the manganese oxide phase is active
in ammonia oxidation toward N2O. A comparison of the
selectivity toward N2O of Mn–O/α-Al2O3 and Mn–Bi–O/
α-Al2O3 catalysts shows that bismuth oxides play a key
role in the selectivity of manganese–bismuth oxide cata-
lysts. The selectivity toward nitrous oxide is higher over the
manganese–bismuth oxide catalyst than over the manganese
oxide catalyst.
Mn–Bi–O/α-Al2O3—550 C; 6, Mn–Bi–O/α-Al2O3—400 C. (b) Nor-
malized relative concentration of oxygen in the samples.
concentration of NHx species and, correspondingly, the re-
action rate (I) decrease abruptly to cause an increase in the
selectivity toward N2O.
Addition of bismuth oxide to manganese oxide leads to
improvement of the catalyst performance for three reasons.
First, bismuth oxide makes manganese oxide more dispersed
[
19] which results in an increase in the amount of active
oxygen. Second, bismuth oxide makes the active compo-
nent surface more defective and, as a result, oxygen more
labile. The third reason is the increased thermal stability of
oxygen. XPS studies of the oxygen behavior during heat-
ing samples in the analyzer chamber in situ regime support
this conclusion. XPS data are illustrated in Fig. 16. Fig. 16a
shows the O1s/Al2p intensity ratio for the samples of α-
Al2O3, Mn–Bi–O/α-Al2O3, Mn–O/α-Al2O3, and Bi–O/α-
Al2O3. These comparative values for an individual support
and catalyst characterize an increase in the oxygen loading
in the catalysts due to supporting of the active component.
It is seen in Fig. 16a that supporting the active component
results in an increase in the O1s/Al2p ratio. The ratio in-
creases considerably in the catalyst samples against that of
Nitrous oxide is known to form due to the presence of
active oxygen species in the catalyst [34], which can produce
NO from Nads and lattice oxygen (Olat) [35]. In accordance
with earlier paper [7], the selectivity toward N2 and N2O is
determined by the relative rates of reactions (I) and (II):
NHx + NO → N2 + OHx,
N + NO → N2O.
The reaction rates depend on the surface oxygen cover-
age. The more surface oxygen coverage, the faster the reac-
tions of NOads and Nads formation. As a result, the surface
(I)
(II)
the supported individual oxides Bi O (curve 2) and MnO
2
2
3
(curve 3). Since manganese and bismuth are loaded in ap-

E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
139
proximately equal amounts in all the samples, the sharp
increase of the O1s/Al2p ratio is accounted for by disper-
sion of manganese oxide under the action of bismuth oxide
supported on α-Al2O3. The surface sensitivity in XPS mea-
surements is determined by the mean free path of emitted
electrons λ (λ ∼ 20–30 Å); therefore, it is reasonable to
suppose that the observed process of dispersing leads to for-
mation of thin Mn–Bi–O film that coats the support particles.
Note that the highest O1s/Al2p ratio is observed with the
It is known that the reaction products are not formed in
the absence of oxygen in the gas phase [37] on the Mn–O/γ-
Al2O3. NO interacts with NH3 in the absence of oxygen to
form nitrogen. Addition of oxygen to the NO/NH mixture
3
results in the acceleration of N2 formation, no N2O being
formed. The catalyst pretreated with the NO/O mixture in-
2
teracts with ammonia to form N and a small amount of
2
N O. It is reasonable to assume based on these data that
2
the lattice oxygen is not labile and inactive in the Mn–O/γ-
Al O catalyst.
2 3
◦
catalyst calcined at 400 C. Elevation of the calcination tem-
perature results to some decrease in the ratio due to sintering
processes detected by XRD and TEM [19].
The presence of bismuth oxide in the catalyst supported
on γ -Al O causes an increase in the proportion of active
2
3
Fig. 16a also shows some decrease in the O1s/Al2p ra-
tio upon sample heating in vacuum that may happen due
to desorption of the active oxygen. To know in more de-
tail about the oxygen loss on heating, the curves in Fig. 16a
are normalized (Fig. 16b). The changes in the temperature-
dependent relative oxygen content in the active phase are
seen to follow different ways in the one-component samples
2 3
oxygen but to a lesser extent than in the α-Al O -based
catalyst. Hence, the presence of bismuth oxide makes the
γ -Al O -based catalyst active to formation of nitrous oxide,
2
3
but less active than the catalyst based on α-Al O .
2
3
4.6. Studies of behavior of the catalyst during treatment
with the reaction mixture
(
Mn–O/α-Al2O3, Bi–O/α-Al2O3) and in the catalysts under
study. For example, the relative oxygen loss is noticeably
lower in the catalysts than in the individual oxides Mn–O/α-
Al2O3 and Bi–O/α-Al2O3. A comparison reveals that the
presence of bismuth oxide makes the thermal stability of the
active oxygen much higher in the catalyst.
Hence, manganese oxide looses oxygen more rapidly, i.e.,
is easier reduced, in the absence of bismuth oxide. The same
conclusion follows from the TPR data [19] showing that the
temperature of reduction of MnO2 into Mn3O4 is lower for
The catalyst was tested with pulses of ammonia and
the ammonia/oxygen reaction mixture. The same products,
N O, N , and NO, were formed during both the reduction
2
2
and the catalytic reaction with the rate constants being equal
for the catalytic reaction and the reduction at the same de-
gree of surface reduction (Fig. 1). As stated above (Fig. 5),
comparison of the total oxygen consumption for the reaction
products with the consumption of the gas-phase oxygen for
the products demonstrates the participation of oxygen of the
catalyst in the reaction. Our results lead to the conclusion
that ammonia oxidation proceeds via alternate reduction and
reoxidation of the catalyst surface. When the reaction mix-
ture slightly affects the buried oxygen layers in the catalyst,
the rate of the reduction stage equals the rate of the catalytic
reaction. At higher reduction, the phase composition can be
changed and the oxygen supply from the reaction mixture is
not sufficient in restoring the initial activity (Fig. 1, point 2).
the Mn–O/α-Al2O3 catalyst (the peak maximum is equal to
◦
3
13 C) than for the Mn–Bi–O/α-Al2O3 catalyst (the peak
◦
maximum is equal to 348 C). One can conclude based on
these data that the reduction degree is considerably higher
in the Mn–O/α-Al2O3 catalyst under steady-state operation
than in the Mn–Bi–O/α-Al2O3 catalyst. As demonstrated
above, the selectivity toward N2 increases but the selectiv-
ity toward N2O decreases as the surface reduction degree
increases. Therefore, the manganese–bismuth oxide catalyst
under steady-state operation in ammonia oxidation is more
selective toward N2O than the manganese catalyst. Note that
the same was concluded for the Mn/La/SiO2 system [36].
The manganese–bismuth oxide catalyst supported on γ -
Al2O3 is considerably lower selective toward N2O than the
one supported on α-Al2O3. The observed distinctions may
be accounted for by the following reasons. As shown else-
where [19], in the catalyst based on γ -Al2O3 the active com-
ponents interact with the support to produce the Bi2Al4O9
phase and solid solutions of manganese ions in alumina that
is indicated by a lower lattice constant than that of pure alu-
4.7. The reaction mechanism
It is shown in our previous work [19] that an increase
in the Mn3 /Mn ratio correlates with the increase in the
+
δ+
selectivity toward N O that is supported again by the data re-
2
ported here. The studies on the catalyst reduction under the
action of ammonia pulses reveal that the selectivity toward
N increases but toward N O decreases with an increase of
2
2
the reduction degree. Experimental studies including feed-
ing of the reaction mixture in the pulse mode demonstrate
that the selectivity toward N O is much higher over the ox-
2
mina. Since rMn4+ = 0.52 < r 3+ = 0.57 Å, then alumina
idized catalyst than over the catalyst treated with ammonia.
Al
4
+
may be promoted by Mn ions. Moreover, the surface con-
centration of manganese and bismuth, in accordance with
XPS data, is lower by an order of magnitude than those in
the catalysts supported on α-Al2O3. As a result, the amount
of active oxygen responsible for formation of N2O is con-
siderably lower than that in the Mn–Bi–O/α-Al2O3 catalyst.
The selectivity into N2O decreases with an increase of num-
ber of pulses over oxidized catalyst and increases over the
reduced catalyst. The latter phenomenon indicates changes
in the oxidation state: The oxidized sample is reduced under
the action of the reaction mixture and the reduced sample is
reoxidized.

140
E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
The results obtained allow the reaction mechanism to be
presented as follows: ammonia is adsorbed on active sites
of the catalyst, which are manganese ions in different oxida-
tion states, i.e., Mn3 (Mn ) and Mn . While ammonia is
+
4+
δ+
adsorbed, the manganese sites are partially reduced to form
δ+
surface complexes [Mn –N], if ammonia is adsorbed by
3
+
2+
δ+
Mn and [Mn –N], if ammonia is adsorbed by Mn
.
That the complexes are formed is supported by IRS and XPS
studies; [NH2] and [NH] adsorbed complexes are identified
in the spectra, which are then transformed into [N] species
under the action of oxygen.
Interaction of the surface intermediates at a compara-
δ+
tively high oxidation state [Mn –N] leads to formation of
N2O. N2 is formed through the interaction of more reduced
intermediates. In addition, TPSR data show that elevation of
temperatures up to 200–250 C gives rise to decomposition
Fig. 17. Comparative experimental and calculated TPSR spectra of the
◦
Mn–Bi–O/α-Al O catalyst calcined at 550 C.
2
3
◦
of the surface complexes to release NO. It seems like NO is
formed, similar to N2O, on the oxidized manganese sites.
The reaction pathways corresponding to the suggested
mechanism are as follows:
low rate of the nitrogen atoms diffusion over the catalyst sur-
face causes a sharp decrease in the probability of the surface
species interactions as the coverage decreases. It is notewor-
thy that exactly the different dependencies of the stage rates
of N2O and N2 formation on coverage result in the shift of
the NO peak (the dependence of first order is in the case of
NO) in comparison to N2O peak position in the TPSR spec-
tra.
The presence of less pronounced peaks of N2O (peaks at
◦
◦
Tp = 176 and 496 C) and NO (peak at Tp = 544 C) evolu-
tion may be explained as follows.
The surface coverage by intermediates is no more than
.1; therefore, the proportion of intermediates occupying
Numerical modeling of TPSR was carried out in terms of
the suggested mechanism. The process of evolution of the
reaction products in the course of heating was described by
a set of differential equations:
0
neighboring sites is not high. If so, the occurrence of three
peaks of nitrous oxide evolution may be determined from
several reasons. First, the intermediates [N] can interact both
with the same intermediates localized on the neighboring
sites and with the labile [N] species diffusing over the sur-
face. The labile [N] species can be formed on the sites of all
types as the temperature increases. At the same time, upon
Nk
ꢂ
∂
∂
θk
T
−
Ei/(RT )
¯
θkfi(θ),
U
U
= −
kie
i=1
Nm
ꢂ
j=1
∂
Cm
−Ej /(RT )
¯
θj fj (θ),
= τb
kj e
δ+
temperature elevation the [Mn –N] intermediates can be
∂
T
reoxidized by oxygen supplied from the subsurface layers to
produce the [Mn3 –N] intermediate. Due to different Mn–O
+
x+
where θk (k = 1, 2) is the concentration of adsorbed [Mn
δ+
3+
bond energies, the interaction of [Mn –N] and [Mn –N]
intermediates may also be responsible for appearance of the
extra N2O peak in the TPSR spectrum. The presence of the
extra NO peak may be attributed to the formation of the
–
N] complexes, Cm is the concentration of the reaction prod-
ucts (NO, N2, N2O), ki(j) are constants of stage rates, Ei(j)
are activation energies, τ is contact time, b is the ratio of the
molecule concentrations in the reaction volume and on the
3
+
¯
[Mn –N] intermediate.
catalyst surface, and f (θ) is the coverage function of power
type.
◦
The second peak of nitrogen evolution (Tp = 601 C) is
The results obtained are illustrated in Fig. 17. One can
see that the peak of N2 evolution and main peaks of N2O
and NO evolution are described well enough. The calculated
activation energies are 35.8 kcal/mol for the stage of N2 for-
mation (III), and 14.3 kcal/mol for the stages of N2O and
NO formation. It is important to note that the rates of N2O
and N2 formation depend rather strongly on the coverage,
namely, of the fourth order with respect to the intermediate
species. That is most likely to result from strong bonds be-
tween the adsorbed complexes and the manganese sites. The
observed in the high-temperature range over Mn–Bi–O/α-
◦
Al2O3 catalysts calcined at 400 and 550 C but not over the
catalyst calcined at 750 C. The peak may be assigned to the
◦
adsorption of ammonia and the formation of a [N] species
adsorbed on bismuth oxide. This assumption agrees with the
◦
fact that the catalyst calcined at 750 C does not contain the
phase of bismuth oxide; for this reason the high-temperature
peak of nitrogen evolution does not exist.
The process of the catalyst reduction by ammonia also
can be considered in terms of the suggested scheme.

E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
141
δ+
The first ammonia pulse produces [Mn –N] due to the
presence of weakly bonded and labile oxygen. A part of
(4.2) NH3 interacts with weakly bonded oxygen species
through hydrogen atom abstraction, giving rise
3
+
δ+
2+
Mn
sites are reduced to Mn . Ammonia is adsorbed
by the Mn sites to form [Mn –N] species. [Mn –N]
and labile N species interact to form N2O, while [Mn
to key intermediate [N] localized on Mn and
δ+
2+
δ+
δ+
Mn (2 < δ < 3).
2
+
–
(4.3) The reaction mechanism is proposed. The main
features are as follow: ammonia is adsorbed
by manganese ions of different oxidation states
δ+
N] forms N2. Since [Mn –N] is formed over the oxidized
2
+
catalyst in a much larger amount than [Mn –N], then N2O
also is formed in a larger amount during the first pulse. More
than four oxygen monolayers are removed during reduction
by ammonia which makes it reasonable to assume that the
3
+
4+
δ+
Mn (Mn ) and Mn . The surface complexes
δ+
3+
[Mn –N] are formed on Mn sites, while the
2
+
δ+
[Mn –N] species are formed on Mn . The
reduced Mn2 sites are reoxidized by subsurface oxygen
+
δ+
2
[Mn –N] species are responsible for N O forma-
2
+
(
Oss) as the surface oxygen (Os) is consumed. The reoxi-
tion and [Mn –N] species are intermediates of
N formation. Formation of NO, similar to N O,
dation degree depends on the strength of oxygen/manganese
bonds and on the rate of oxygen diffusion to the surface.
The stronger the bonds, the lower the reoxidation rate. As a
2
2
is most likely to proceed on oxidized manganese
sites.
result, [Mn2 –N] species formed on Mn increase in num-
+
δ+
(4.4) The reaction kinetic scheme is proposed based on
the experimental results. Numerical simulation of
TPSR data confirms the reliability of the reaction
mechanism. Activation energies were estimated
for the stages of formation of the reaction prod-
ucts.
δ+
3+
ber but [Mn –N] formed on Mn decrease. Therefore,
2
+
the concentration of [Mn –N] increases with an increase
in the number of pulses and, correspondingly, the selectiv-
ity to nitrogen increases. This is just what is experimentally
observed.
Acknowledgment
5
. Conclusions
The authors are grateful to V.V. Mokrinskii for the prepa-
ration of the Mn–Bi–O catalysts.
The pulse kinetic method and methods of temperature-
programmed surface reaction, IRS, and XPS were used for
characterization of the highly selective supported manga-
nese–bismuth oxide catalyst. The studies revealed that:
References
3
+
(
1) Manganese ions at different oxidation states, viz Mn
[1] N.I. Il`chenko, G.J. Golodets, I.M. Avilova, Teoret. Exp. Khim. (The-
oret. Exp. Chem.) 11 (1975) 56 (in Russian).
4
+ δ+
(
Mn ) and Mn , behave as active sites of the catalyst.
[
[
2] G.S. Selwyn, G.T. Fujimoto, M.C. Lin, J. Phys. Chem. 86 (1982) 760.
3] V.V. Gorodetskii, M.Yu. Smirnov, A.R. Cholach, in: Proc. X Intern.
Congr. on Catalysis (Budapest, 1992), Vol. B, Akad. Kiado, Budapest,
1993, p. 1587.
Formation of the reaction products: N2O, N2, and NO,
was revealed. The reaction pathways are determined by
the energetically nonuniform state of the surface oxygen
involved in the manganese oxide. Relatively reduced
sites Mn bear the responsibility for the formation of
N2 but more oxidized Mn sites for the formation of
N2O.
[4] W.D. Mieher, W. Ho, Surf. Sci. 322 (1995) 151.
[5] B. Afsin, P.R. Davies, A. Pashusky, M.W. Robets, D. Vincent, Surf.
Sci. 284 (1993) 109.
δ+
3
+
[6] A.C.M. Van den Broek, J. Van Grondelle, R.A. Van Santen, J. Ca-
tal. 185 (1999) 297.
(
(
2) N2O selectivity correlates with a proportional increase
of active sites with low manganese–oxygen bond en-
ergy. A relative portion of these active sites is increased
with the catalyst calcination temperature.
3) Bismuth oxide plays an important role for the following
reasons: First, it disperses manganese oxide on the sup-
port surface and prevents the manganese oxide particles
from aggregation during calcination and the reaction
mixture action. Second, it increases the defect extent of
the supported active Mn–Bi–O phase, resulting in high
mobility of subsurface oxygen. Finally, it increases the
thermal stability of the active oxygen.
[7] B.G. Lu Gang, J. Anderson, R.A. Van Grondelle, J. van Santen, J. Ca-
tal. 199 (2001) 107.
[
[
8] N.I. Il`chenko, G.J. Golodets, J. Catal. 39 (1975) 57.
9] M. De Boer, H.M. Huisman, R.J.M. Mos, R.G. Leiveld, A.J. Van
Dillen, J.W. Geus, Catal. Today 17 (1993) 189.
[
10] G. Ramis, L. Yi, G. Busca, M. Turco, E. Kotur, R.J. Willey, J. Ca-
tal. 157 (1995) 523.
[11] G. Ramis, L. Yi, G. Busca, Catal. Today 28 (1996) 373.
[
[
[
[
12] G. Bagnasco, G. Peluso, G. Russo, M. Turco, G. Busca, G. Ramis, in:
rd World Congress on Oxidation Catalysis, 1997, p. 643.
13] J.M.G. Amores, V.S. Escribano, G. Ramis, G. Busca, Appl. Catal. B 13
1997) 45.
3
(
14] M. Trombetta, G. Ramis, G. Busca, B. Montanari, A. Vaccari, Lang-
muir 13 (1997) 4628.
15] G.J. Golodets, Yu.I. Pyatnitskii, Katal. Katal. (Catal. Catal.) 4 (1968)
(
4) Studies of the mechanism of ammonia oxidation showed
that:
25 (in Russian).
[
[
[
16] I. Zavadski, Discuss. Faraday Soc. 8 (1950) 140.
(
4.1) Ammonia oxidation proceeds via alternating re-
duction and reoxidation of the catalyst surface of
manganese oxide particles.
17] W. Krauss, A. Neubaus, Z. Phys. Chem. B 50 (1941) 323.
18] V.V. Mokrinskii, E.M. Slavinskaya, A.S. Noskov, I.A. Zolotarsky, PCT
Int. Appl. WO 9825698 (1998), priority: RU 96-96123343.

142
E.M. Slavinskaya et al. / Journal of Catalysis 222 (2004) 129–142
[
19] A.S. Ivanova, E.M. Slavinskaya, V.V. Mokrinskii, I.A. Polukhina, S.V.
Tsybulya, I.P. Prosvirin, V.I. Bukhtiyarov, V.A. Rogov, V.I. Zaikovskii,
A.S. Noskov, J. Catal. 221 (2004) 213.
[28] A.V. Sklyarov, Uspehi Khim. (Progr. Chem.) LV (1986) 450 (in
Russian).
[29] G.K. Boreskov, S.A. Veniaminov, N.N. Sazonova, Yu.D. Pankratiev,
A.N. Pitaeva, Kinet. Katal. (Kinet. Catal.) 16 (1975) 1442 (in Russian).
[20] D. Briggs, M.P. Seach, Practical Surface Analysis by Auger and X-Ray
Photoelectron Spectroscopy, Wiley, New York, 1983.
[
[
[
[
30] T.V. Andrushkevich, G.K. Boreskov, Yu.D. Pankratiev, G.Ya. Popova,
O.N. Kozmenko, A.K. Kuliev, React. Kinet. Catal. Lett. 7 (1977) 187.
31] G.Ya. Popova, T.V. Andrushkevich, G.A. Metalkova, React. Kinet.
Catal. Lett. 12 (1979) 469.
[
21] J.F. Moulder, W.F. Stickle, P.E. Sobol, et al. (Eds.), Handbook of
X-Ray Photoelectron Spectroscopy, Perkin-Elmer Corporation, Eden
Prairie, MN, 1992.
[
22] V.S. Muzykantov (Ed.), Mehanismy Geterogenno-Kataliticheskih
Reakcii Okisleniya (Mechanisms of Heterogeneous Catalytic Reac-
tions of Oxidation), Novosibirsk, 1993 (in Russian).
32] V.M. Bondareva, T.V. Andrushkevich, Yu.D. Pankratiev, V.M. Turkov,
React. Kinet. Catal. Lett. 32 (1986) 387.
33] M.Kh. Karapetiantc, M.L. Karapetiantc, Osnovnye Termodinamich-
eskie Konstanty Neorganicheskih i Organicheskih Vecshestv (The Ba-
sic Thermodynamic Constants of Inorganic and Organic Substances),
Chemistry, Moscow, 1968.
[
23] NIST X-Ray Photoelectron Spectroscopy Database. NIST Standard
Reference Database 20, Version 3.3 (Web version). Data compiled and
evaluated by C.D. Wagner, A.V. Naumkin, A. Kraut-Vass, J.W. Alli-
son, C.J. Powell, and John R. Rumble Jr.
[34] F. Kapteijn, L. Singoredjo, A. Andreini, J.A. Moulijn, Appl. Catal. B 3
[
[
[
[
24] B.M. Biwer, S.L. Bernasek, J. Electron Spectrosc. Relat. Phemon. 40
(1994) 173.
(1986) 339.
[35] M. De Boer, H.M. Huisman, R.J.M. Mos, R.G. Leiveld, A.J. Van
Dillen, J.W. Geus, Catal. Today 17 (1993) 189.
25] R. Marchand, D. Agliz, L. Boukbir, A. Quemerais, J. Non-Cryst.
Solids 103 (1988) 35.
26] C. Guimon, D. Gonbeau, G.P. Guillouzo, O. Dugne, A. Guette, R.
Naslain, M. Lahaye, Surf. Interface Anal. 16 (1990) 440.
27] P.A. Redhead, Trans. Faraday Soc. 57 (1961) 641.
[
36] R. Craciun, N. Dulamita, Catal. Lett. 46 (1997) 229.
37] F. Kapteijn, L. Singoredjo, M. van Driel, A. Andreini, J.A. Moulijn,
G. Ramis, G. Busca, J. Catal. 150 (1994) 105.
[