Mechanism of low-temperature selective catalytic reduction of NO with NH3 over carbon-supported Mn3O4 Role of surface NH3 species: SCR mechanism

Article abstract of DOI:10.1016/j.jcat.2004.05.022

In this work a complete mechanism for describing the low-temperature (125°C) selective catalytic reduction of NO with NH₃ over carbon-supported Mn₃O₄ is discussed. This study sets out to explain for the first time certain specific interactions among NH ₃, NO, O₂, and a manganese-based catalyst. A set of SCR reactions was obtained through a detailed TPD analysis of the surface NH ₃ species and by taking into account the conclusions of a previous study on the role of NO species [Phys. Chem. Chem. Phys. 6 (2004) 453]. The SCR reactions proceed via an Eley-Rideal mechanism, in which NO₂, and to a lesser extent NO, reacts from the gas phase with surface-active NH₃ species. The overall reaction path involves the simultaneous occurrence of two different SCR mechanisms in which either aminooxy groups or ammonium ions react with NO/NO₂. These NH₃-based species are related to the local phases that coexist in Mn₃O₄: (a) SCR by aminooxy groups (steady-state mechanism). Aminooxy groups formed on the locally octahedral environment of Mn₃O₄ (Mn₂O ₃) react with gaseous NO₂. O₂ cannot dissociate on this phase in order to reoxidize the reduced catalyst and therefore the overall SCR process is 6NO+4NH₃→5N₂+6H₂O. (b) SCR by ammonium ions (pseudo-steady-state mechanism). This mechanism occurs on the locally tetrahedral environment of Mn₃O₄ (MnO) and initially accounts for ～60% of the total NO reduction. However, it is gradually deactivated by the nitrates formed on those same hydroxyl groups that are available for ammonium formation. The ammonium ions formed on the hydroxyl groups of this tetrahedral environment react with gas-phase NO₂. The overall SCR process is 4NO+4NH₃+O₂→4N ₂+6H₂O.

Full text of DOI:10.1016/j.jcat.2004.05.022

Journal of Catalysis 226 (2004) 138–155
www.elsevier.com/locate/jcat
Mechanism of low-temperature selective catalytic reduction of NO
with NH3 over carbon-supported Mn3O4
Role of surface NH3 species: SCR mechanism
Gregorio Marbán ∗, Teresa Valdés-Solís, Antonio B. Fuertes
Instituto Nacional del Carbón (CSIC), c/Francisco Pintado Fe, 26, 33011 Oviedo, Spain
Received 13 January 2004; revised 17 May 2004; accepted 19 May 2004
Abstract
In this work a complete mechanism for describing the low-temperature (125 C) selective catalytic reduction of NO with NH over
◦
3
carbon-supported Mn O is discussed. This study sets out to explain for the first time certain specific interactions among NH , NO, O ,
3
4
3
2
and a manganese-based catalyst. A set of SCR reactions was obtained through a detailed TPD analysis of the surface NH species and
3
by taking into account the conclusions of a previous study on the role of NO species [Phys. Chem. Chem. Phys. 6 (2004) 453]. The SCR
reactions proceed via an Eley–Rideal mechanism, in which NO , and to a lesser extent NO, reacts from the gas phase with surface-active
2
NH species. The overall reaction path involves the simultaneous occurrence of two different SCR mechanisms in which either aminooxy
3
groups or ammonium ions react with NO/NO . These NH -based species are related to the local phases that coexist in Mn O : (a) SCR
2
3
3 4
by aminooxy groups (steady-state mechanism). Aminooxy groups formed on the locally octahedral environment of Mn O (Mn O ) react
3
4
2 3
with gaseous NO . O cannot dissociate on this phase in order to reoxidize the reduced catalyst and therefore the overall SCR process
2
2
is 6NO + 4NH → 5N + 6H O. (b) SCR by ammonium ions (pseudo-steady-state mechanism). This mechanism occurs on the locally
3
2
2
tetrahedral environment of Mn O (MnO) and initially accounts for ∼ 60% of the total NO reduction. However, it is gradually deactivated
3
4
by the nitrates formed on those same hydroxyl groups that are available for ammonium formation. The ammonium ions formed on the
hydroxyl groups of this tetrahedral environment react with gas-phase NO . The overall SCR process is 4NO + 4NH + O → 4N + 6H O.
2
3
2
2
2
2004 Elsevier Inc. All rights reserved.
Keywords: Selective catalytic reduction; Mechanism; Manganese oxide; Nitric oxide; Ammonia; Vacancies; Aminooxy; Ammonium
1. Introduction
have been analyzed by several authors [5,9,10] the exact na-
ture of the reaction mechanism remains unclear due in part
to the strong NO adsorption that produces partial catalyst de-
activation [11]. As a result it is difficult to discern between
the species related to the SCR reaction and catalyst deacti-
vation.
In a previous work we carefully analyzed the NO-
related species existing on the catalyst surface under low-
temperature SCR conditions [11]. In the present study the
role of surface NH3 species is analyzed with the aim of de-
termining the mechanism of the low-temperature selective
catalytic reduction of NO with NH3 over carbon-supported
manganese oxide catalysts. This work also provides infor-
mation which might be of help in understanding the behavior
of manganese oxides in other catalytic processes of similar
complexity.
Manganese-based catalysts have attracted widespread in-
terest because their unique redox properties make them
useful for a variety of applications. Thus, in recent years,
manganese oxides have been proposed as catalysts for
several applications such as the oxy-dehydrogenation of
propane [1], the selective reduction of nitrobenzene to ni-
trosobenzene [2], CO oxidation [3], and the low-temperature
selective catalytic reduction (SCR) of NO with NH3 [4–7].
In low-temperature SCR, manganese oxide-based catalysts
have been proven to be among the most active catalysts [4,8].
Although low-temperature SCR manganese-based catalysts
*
Corresponding author.
E-mail address: greca@incar.csic.es (G. Marbán).
0021-9517/$ – see front matter
doi:10.1016/j.jcat.2004.05.022
2004 Elsevier Inc. All rights reserved.

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
139
2. Summary of previous findings
carbonaceous support. This reduces the oxidation state of the
catalyst, thus increasing the number of oxygen vacancies:
This section summarizes the previous results relating to
the catalytic phase and the formation of NO-related species
on the catalyst surface reported in [11] and which are nec-
essary for an understanding of the discussion on the SCR
mechanism proposed here. The catalyst chosen for this
study is Mn3O4 [11] supported on carbon–ceramic cellu-
lar monoliths [4] or activated carbon fibers [6,7]. More
specifically, the active phase of the catalyst is a nonsto-
ichiometric Mn3O4, locally composed of an octahedral
Mn2O3 phase and a tetrahedral MnO phase, which can be
represented by the following empirical formulae (super-
scripts T and O mean tetrahedral and octahedral, respec-
tively) [11]:
O
{O=Mn–O–Mn=O} + CO
O
→ {O=Mn–ꢀ–Mn=O} + CO2.
NO is also adsorbed to a very limited extent as linear
(monodentate) and bridged nitrites on the oxygen atoms of
the octahedral Mn2O3:
(5)
ꢀ
ꢁ
O
O
N···O
O
ꢀ
|
ꢀ
{O=Mn–ꢀ–Mn=O} + NO ↔
.
(6)
Mn···O···Mn
Nitrosyls may react with gas-phase oxygen to produce gas-
phase NO2:
O
1
2
{O=Mn–[NO]–Mn=O} + O2
T
T
Mn3O4+0.5x−y ⇒ {Mn=O}1−x{O=Mn–O–Mn=O}x/2
O
↔ {O=Mn–ꢀ–Mn=O} + NO2.
(7)
O
O
{O=Mn–O–Mn=O}1−y{O=Mn–ꢀ–Mn=O}y .
This reacts with the hydroxyl groups (oxygen excess) of
the tetrahedral MnO phase to produce nitrates which, under
SCR conditions, cause the progressive deactivation of this
catalyst phase:
This formula accounts for, in order of appearance, the
tetrahedral Mn2+–O bonds, the tetrahedral Mn3+–O bonds
(oxygen excess), the octahedral Mn3+–O, and the octahe-
dral Mn2+–O bonds (oxygen vacancies), respectively. Frac-
tions x and y have values between 0 and 1, and account
respectively for oxygen excess and vacancies in the cata-
lyst.
T
{(HO)2Mn–O–Mn(OH)2} + 6NO2
T
→ {(NO3)2Mn–O–Mn(NO3)2} + 2H2O + 2NO.
(8)
The nonstoichiometry of the tetroxide is a result of the
low temperature used to prepare the catalyst, which gives
3. Experimental
rise to the oxygen excess bonded to the tetrahedral Mn2+
.
The distorted structure associated with this oxygen excess
may be relaxed by the formation of hydroxyls from water
adsorption:
3.1. Catalysts
The preparation of the catalysts is described in detail else-
where [4,6,7,11]. The following catalysts were prepared by
equilibrium adsorption impregnation: (i) a carbon–ceramic
monolith-supported manganese oxide catalyst (CCCM-Mn;
Mn content on a carbon basis = 7.1 0.9 wt%), whose
activated support will be denoted as CCCM-40 (the numeri-
cal suffix in the reference indicates the carbon weight loss
after steam activation), and (ii) an activated carbon fiber
composite-supported manganese oxide catalyst (ACFC-Mn;
Mn content in carbon basis = 4.8 wt%) prepared by impreg-
nation of the steam-activated support (ACFC-20). For all
experiments performed with CCCM-40 or ACFC-20 (FTIR,
TPD, etc.) the samples were previously oxidized with nitric
acid and heat treated in N2 at 400 ◦C for 1 h [11]. The effec-
tive size of the supported carbon layer in CCCM was found
to vary in the range 10–20 µm.
T
1
T
2{Mn=O} + O2 ↔ {O=Mn–O–Mn=O} ,
(1)
2
T
{O=Mn–O–Mn=O} + 2H2O
T
↔ {(HO)2Mn–O–Mn(OH)2} .
(2)
The oxygen in the octahedral long Mn3+–O bonds is
easily removed by mild-temperature treatments (∼ 200 ◦C)
under inert atmosphere. At 125 ◦C gas-phase oxygen can-
not dissociate on the oxygen vacancies of the octahedral
environment, but can partly oxidize the tetrahedral MnO
phase [11].
The carbonaceous support hardly adsorbs NO at 125 ◦C.
NO adsorbed on Mn3O4 is associated with different catalyst
centers. NO is weakly adsorbed on octahedral oxygen vacan-
cies as nitrosyls that can dissociate into N2 when adsorbed
in adjacent positions:
3.2. Step-response and temperature-programmed
desorption (TPD) experiments
O
O
{O=Mn–ꢀ–Mn=O} + NO ↔ {O=Mn–[NO]–Mn=O} ,
(3)
O
O
2{O=Mn–[NO]–Mn=O} → 2{O=Mn–O–Mn=O} + N2.
Samples of CCCM-40 supports and CCCM-Mn cata-
lysts, crushed to a particle size below 1 mm, were tested
for adsorption/desorption/reactions under different gas at-
mospheres and temperature programs in a vertical quartz
reactor (0.65 cm i.d.) connected to an Omnistar model mass
(4)
This is the main mechanism of catalyst reoxidation at
125 ◦C. Gas-phase oxygen may indirectly lead to an increase
in the amount of surface nitrosyls by generating CO from the

140
G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
Table 1
Deconvolution parameters for the NO and NH evolution curves during the TPD stage averaged for all the experiments
3
Gas
Starting TPD
temperature
( C)
LT peak
MT peak
HT peak
T
Ea
n
T
Ea
n
T
Ea
n
HT
LT
LT
LT
MT
MT
MT
HT
HT
◦
◦
◦
◦
−1
−1
−1
( C)
177
( C)
–
( C)
231
(kJ mol
)
(kJ mol
–
)
(kJ mol )
NO
125
175
6
Fig. 3
in [11]
92 10
1.6
1.6
–
–
8
6
Fig. 3
in [11]
132
1.8 0.2
1.8 0.5
201 11
–
–
242
6
NH
125
175
204
255
5
8
79
90
5
4
2.0
2.0
262
322
4
4
67
60
1
1
1.5
1.5
370 12
403
60
63
1
1.6
1.6
3
4
spectrometer. A typical sample mass of 1 g and a gas flow
rate of 150 mL (STP) min−1 were used during the experi-
ments. By means of a combination of mass-flow controllers
it was possible to obtain different mixtures of NO, NH3,
and O2 in He. A typical experiment comprised four stages:
(1) degasification of the sample in He at 200 ◦C for 30 min,
(2) single or multiple step-response experiments (see be-
low), (3) isothermal desorption in He at the temperature
of the final step-response experiment, and (4) temperature-
programmed desorption in He (TPD stage) at 5 ◦C min−1 up
to 400 ◦C. Step-response experiments were performed at 125
and 175 ◦C, and the response of the system to a given com-
bination of different gases (700 ppm NO; 800 ppm NH3;
3 vol% O2; 3.3 vol% H2O and He) was analyzed. With this
experimental procedure the entire experiment could be la-
beled with reference only to the step-response stage, since
(medium-temperature) peak, and an HT (high-temperature)
peak (examples of these are not included for brevity). The
values of maximum peak temperature (Ti), activation en-
ergy (Ea ), and reaction order (ni) for these peaks averaged
i
for all the experiments performed are shown in Table 1.
The values of the peak areas obtained for the different ex-
periments are displayed in Table 2 for the NO curves and
in Table 3 for the NH3 curves. In these tables the para-
TPD
NO
meter Φ
(ΦNTPHD) means the total moles of desorbed
3
NO (NH3) per manganese moles in the sample during the
TPD stage, whereas ΦLT , ΦMT, and ΦHT represent the
moles of desorbed NO or NH3 per manganese moles as-
cribed to low-temperature (LT), medium-temperature (MT),
and high-temperature peak (HT), respectively. Additionally,
these tables include the values of NO/NH3 desorbed during
the isothermal desorption stage per manganese moles (Φiso
)
the other stages were always performed in the same man-
for all the experiments, and the values of N2 formed dur-
125◦C
ing the TPD stage per manganese moles (ΦNTPD) for selected
ner. Thus, an experiment referred to as NO + O2
>
60
2
125 ◦C
NH3
consisted of stages 1, 3, and 4 as described
experiments, values that have already been discussed in the
case of NO [11].
The data displayed in Tables 2 and 3 are the main ba-
sis for the subsequent discussions on NO- and NH3-based
species that coexist on the catalyst surface under different
conditions.
40
above, and a step-response sequence (stage 2) comprising a
first step under NO + O2 atmosphere (700 ppm NO, 3 vol%
O2 in He) at 125 ◦C for 60 min followed by a second step
under NH3 atmosphere (800 ppm NH3 in He) at 125 ◦C for
40 min.
4.2. NH3 adsorption/reaction modes: SCR mechanism
4. Results and discussion
4.2.1. Adsorption of NH3 by the carbonaceous support
It can be seen from Table 3 that the carbonaceous support
itself adsorbs a significant amount of NH3 (codes h0, i0, d0,
11, 12), although the quantity is lower than that adsorbed by
the catalyst. Generally speaking, a clean carbon surface is
considered to have a basic character, so that ammonia must
be adsorbed on acidic functional groups formed on the sup-
port surface [13,14]. It is well known that nitric oxidation of
a carbon surface at moderate temperatures increases surface
acidity mainly through the creation of carboxylic functional
groups, although other less acidic groups such as phenols
and lactones are also formed. From the values provided by
the literature [15] and taking into account the fact that the
heat treatment performed after the oxidation step of CCCM-
40 is strong enough to decompose single carboxylic groups
to CO2 and to dehydrate adjacent carboxylic and pheno-
lic groups to anhydrides or lactones [16], it is possible to
4.1. Procedure for the quantification of the adsorbed
species
Both the NO and the NH3 evolution curves during the
TPD step were subjected to a numerical fitting proce-
dure with the purpose of finding convolved Arrhenius-type
curves [12]. The procedure consisted in finding a minimum
number of Arrhenius-type curves whose convolution would
closely resemble the experimental curve, and at the same
time show features common to all the performed experi-
ments. For the NO curves these premises can be fulfilled by
taking only two desorption peaks, a low-temperature peak
(LT peak) and a high-temperature peak (HT peak) [11].
For the NH3 evolution curves during the TPD experi-
ments three characteristic peaks are needed for a satis-
factory agreement: an LT (low-temperature) peak, an MT

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
141
Table 2
iso
NO
TPD
NO
TPD
N
2
Moles of NO (isothermal desorption stage (Φ ) and TPD (Φ
experiments
)) and N (TPD (Φ
)) released per manganese moles in the catalysts during the
2
iso
TPD
NO
LT
NO
TPD
Code
Sample
Step-response sequence
Φ
Φ
Φ
Φ
N
2
NO
HT
NO
Φ
a
125 ◦
C
1
CCCM-40
CCCM-40
CCCM-40
NO
0.007
0.008
0.020
0.004
0.004
0.010
–
–
–
66
a
125 ◦
C
2
NO + O
2
60
125 ◦
a
C
3
NO + O + NH
2
3
59
125 ◦
C
a2
b1
b2
f
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
NO
0.023
0.064
0.017
0.021
0.017
0.032
0.026
0.010
0.002
0.017
0.000
0.002
0.002
0.000
0.001
0.091
0.191
0.047
0.000
0.075
0.004
0.004
93
125 ◦
C
C
NO + O
NO + O
–
2
2
15
125 ◦
0.010
137
125 ◦
C
C
NO + O + H O
–
–
–
–
2
2
115
81
175 ◦
C
8
NO
64
175 ◦
C
9
NO + O
2
131
175 ◦
10
NO + O + H O
2
2
125 ◦
125 ◦
C
C
d1
d2
d3
d4
d5
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
NO + O + NH
0.101
0.102
0.064
0.035
0.030
0.009
0.014
0.014
0.004
0.004
0.006
0.016
0.022
0.032
0.060
0.009
0.019
0.019
0.027
0.026
2
3
3
3
3
3
10
20
45
60
120
NO + O + NH
2
125 ◦C
NO + O + NH
2
125 ◦
125 ◦
C
C
NO + O + NH
2
NO + O + NH
2
125 ◦
C
125 ◦
125 ◦
C
e2
e3
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
O
> NO
0.046
0.048
0.018
0.030
0.054
0.030
0.046
0.009
0.015
0.000
0.014
0.020
0.008
0.020
0.000
0.000
0.197
0.002
0.204
0.006
0.004
–
2
2
64
45
57
68
175 ◦C
C
O
> NO
–
b
125 ◦
C
125 ◦
C
j
NO + O
> NH
125 ◦
C
0.080
0.006
0.019
0.007
–
2
C
90
3
91
125 ◦
125 ◦
c
NH
NH
> NO
3
97
94
C
125 ◦
C
C
k
> NO + O
3
90
2
100
125 ◦
C
125 ◦
g
NH + O
> NO
3
2
94
101
125 ◦
C
125 ◦
C
125 ◦
C
c
a1, el
NO
> He
> O
30
C
15
C
2
85
125 ◦
125 ◦
125 ◦
125 ◦
> He
> He
> NO
15
15
30
30
C
C
> O
2
See Table 1 for peak temperatures.
a
Values of Φ evaluated by assuming a hypothetical Mn load equal to that of the corresponding experiments with CCCM-Mn.
b
iso
Isothermal desorption after the NH adsorption stage (Φ
= 0.039 after the NO + O adsorption stage).
3
2
NO
iso
c
Isothermal desorption after the last NO adsorption stage (Φ = 0.014 after the first NO adsorption stage).
NO
estimate the approximate molar distribution of oxygen func-
tional groups: ∼ 2–6 mmol g−1 =∼ 65% carbonyls (basic
groups), ∼ 15% phenolics, ∼ 20% (anhydrides + lactones).
to be released in this stage, both for the catalysts and for
the supports. Robb et al. [19] also established by means
of combined NH3-STPD and FTIR+that the thermal stabil-
ity ranges of NH3 bonded as NH4 to Brønsted hydroxyl
groups on Si and Al atoms of zeolites can be defined by des-
orption peaks located at 350, 440, and 540 ◦C. By combining
the work of Chughtai et al. [17] with the works on zeo-
lites [18,19], we can tentatively assign the different NH3 ad-
sorption modes on the CCCM-40 support as follows: (i) NH3
adsorbed via hydrogen bonding to anhydrides and lactones
is released in the isothermal desorption stage; (ii) the LT
peak is originated by the liberation of reactively adsorbed
ammonia (possibly NH2 species produced by hydrogen ab-
straction of weakly adsorbed NH3), and (iii) the MT and HT
peaks (peak temperatures for experiments at 175 ◦C: 322,
and 403 ◦C, respectively) are formed by the desorption of
NH3 linked as ammonium ions to phenolic groups of vary-
ing acid strength.
The total amount of NH3 desorbed by CCCM-40 after
125◦C
NH3
(code h0 of Table 3) is ∼ 0.2 mmolNH gC−1
,
64
3
which means that, if the basic carbonyls are not adsorb-
ing NH3, a moderate proportion of available oxygen surface
groups are performing the adsorption (∼ 10–30%). Anhy-
drides and lactones can physisorb NH3 via weak hydrogen
bonding on the oxygen atoms of –O–C=O groups, which
may partially evolve to yield more strongly bonded amine-
type species [17]. On the other hand, phenolic groups ad-
sorb NH3 via acid–base interaction to produce ammonium
ions linked to oxygen atoms [17]. In the isothermal de-
sorption stage used in our experiments, after the NH3 step,
the samples were left under a He atmosphere for ∼ 30 min
at 125 ◦C (175 ◦C in certain cases) so that, in accordance
with the literature [18,19], most of the NH3 weakly adsorbed
on oxygen atoms via hydrogen bonding can be expected

142
G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
Table 3
iso
NH
TPD
3
TPD
Moles of NH (isothermal desorption stage (Φ
) and TPD (Φ
)) and N (TPD (Φ
N
)) released per manganese moles in the catalysts during the
2
3
2
NH
3
experiments
iso
TPD
3
TPD
Code
Sample
Step-response sequence
Φ
Φ
Φ
Φ
NH
3
NH
N
2
LT
HN
MT
3
HT
Φ
Φ
NH
NH
3
3
a
125 ◦
C
h0
CCCM-40
CCCM-40
CCCM-40
CCCM-40
CCCM-40
NH
0.079
0.055
0.064
0.032
0.078
0.031
0.028
0.035
0.018
0.035
0.014
0.016
0.016
0.012
0.023
0.015
0.015
0.016
0.008
0.016
–
–
–
–
–
3
64
a
125 ◦
C
i0
NH + O
3
2
60
a
125 ◦
C
d0
NO + O + NH
2
3
59
a
175 ◦
C
11
NH
3
60
a
175 ◦
C
12
NH + O
3
2
131
125 ◦
C
h1
i1
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
NH
0.156
0.136
0.011
0.171
0.071
0.003
0.102
0.088
0.047
0.077
0.076
0.034
0.042
0.045
0.029
0.009
0.035
0.025
0.023
0.032
0.018
0.021
0.053
0.058
–
–
–
–
–
–
3
112
125 ◦
C
NH + O
3
2
119
125 ◦
C
m
NH + O + H O
3
2
2
122
175 ◦
C
13
14
15
NH
3
83
175 ◦
C
NH + O
3
2
125
175 ◦
C
NH + O + H O
3
2
2
122
125 ◦
C
d1
d2
d3
d4
d5
16
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
NO + O + NH
0.082
0.092
0.092
0.117
0.131
0.113
0.051
0.069
0.094
0.097
0.110
0.040
0.018
0.024
0.037
0.023
0.017
0.022
0.012
0.017
0.018
0.019
0.017
0.029
0.009
0.019
0.019
0.027
0.026
–
2
3
3
3
3
3
3
10
125 ◦C
125 ◦C
NO + O + NH
2
20
NO + O + NH
2
45
125 ◦
125 ◦
175 ◦
C
C
C
NO + O + NH
2
60
NO + O + NH
2
120
175 ◦
C
NO + O + NH
> NO + NH
2
75
3
134
125 ◦C
125 ◦C
> NH
3 91
j
CCCM-Mn
CCCM-Mn
CCCM-Mn
CCCM-Mn
NO + O
0.178
0.118
0.000
0.003
0.032
0.002
0.014
0.013
0.017
0.080
0.006
0.019
0.007
2
90
125 ◦
125 ◦
C
125 ◦
C
b
b
b
c
k
g
NH
NH
> NO
0.187 + 0.006
0.158 + 0.004
0.115 + 0.005
3
97
90
94
C
125 ◦
125 ◦
C
C
> NO + O
3
2
100
101
125 ◦C
NH + O
> NO
94
0.000
0.013
3
2
See Table 1 for peak temperatures.
a
Values of Φ evaluated by assuming a hypothetical Mn load equal to that of the corresponding experiments with CCCM-Mn.
NH desorption during the NO/NO + O step + NH desorption during the isothermal desorption stage.
b
3
2
3
4.2.2. NH3 weakly adsorbed by the catalyst (Φiso
)
NH3
Fig. 1 shows the NH3-TPD curves for the experiments of
NH3 adsorption at 125 and 175 ◦C performed on both the
catalyst and the carbonaceous support where similar pro-
files for both materials are observed with almost identical
maximum reactivity temperatures. The TPD curve corrected
for the support ((h1)–(h0) in Fig. 1) yields a similar pro-
file. Furthermore, the NH3-TPD curves obtained for the sup-
port could be resolved by using the same peak distribution
and kinetic parameters as those determined for the catalyst
(Table 1). This suggests strongly that for both the support
and the catalyst similar functionalities are used to link the
NH3-based species, oxygen atoms being the only function-
alities common to both materials. However, several authors
[5,9,10] consider that NH3 is adsorbed predominantly on
oxygen vacancies of transition metal oxides, such as those
used for linking NO in the form of nitrosyls. In this case
competition between both gases might be expected to occur.
However, from the results shown in Table 2 the opposite ef-
fect can be observed; the experiments in which ammonia was
passed prior to or together with NO gave rise to the enhanced
desorption of NO from nitrosyls in the isothermal stage. In
conclusion, at 125 ◦C the oxygen vacancies of Mn3O4 ad-
Fig. 1. Rates of NH release on a carbon mass basis during the TPD stage
3
◦
125 ◦
C
of the experiments of NH adsorption at 125 C (NH
) and
3
3
60–110
◦
175 ◦
C
175 C (NH
) with CCCM-40 and CCCM-Mn. Open squares:
3
60–80
125 ◦
C
CCCM-Mn curve corrected for CCCM-40 (NH
ble 3).
) (codes as in Ta-
3

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
143
Fig. 2. Evolution with time of NH and H O gas concentration and NO desorption rate based on carbon content during the step-response stage of
3
2
125 ◦
C
125 ◦
C
NH
(I) and NH + O
(II) experiments for CCCM-40 and CCCM-Mn.
3
3
2
sorb acid NO molecules but do not seem to adsorb basic NH3
molecules. This can be expected for electron-donor defects,
examples of which are the oxygen vacancies of ZrO2−y or
Nb2O5−y [16], which preferentially adsorb acid-type mole-
cules. A final piece of evidence that shows the preferred
adsorption of ammonia on oxygen atoms is indirectly pro-
vided by comparing NO–H2O competition with NH3–H2O
competition on CCCM-Mn. As regards the former, water
completely impedes the formation of species ascribed to the
LT peak of the NO-TPD (code f in Table 2), due to the com-
petitive adsorption on the oxygen atoms [11]. Since H2O
also impedes the weak adsorption of NH3 (codes m and 15
in Table 3) it follows that this process must take place on the
oxygen atoms.
passed through the catalyst, a significant amount of water is
also released by the catalyst (Fig. 2). For long reaction times
(3000–4000 s), when NH3 is no longer being adsorbed, the
water is only released from the carbonaceous support. Once
the amount of water desorbed by the carbonaceous support
has been subtracted from that desorbed by the catalyst, the
resulting amount is the same in both the absence or the pres-
ence of oxygen. Consequently the similarity of the curve
125◦C
shapes and the amounts of desorbed water for the NH3
125◦C
and NH3 + O2
experiments suggests that the source of
125 ◦C
water is the same in both cases. For the NH3
experi-
ment only two possible sources are available: (i) the reaction
of surface oxygen with the hydrogen atoms of either the
adsorbed ammonia or surface hydroxyls, and (ii) the compe-
tition of ammonia with weakly bonded water on the oxygen
atoms of the manganese oxide. If it were the first route, the
elimination of surface oxygen in the form of water molecules
would be accompanied by the creation of oxygen vacancies
on the catalyst surface.
In a previous work [11] we established an indirect way of
assessing the amount of oxygen vacancies on the surface of
the catalyst by determining the amount of N2 formed in the
initial moments of the NO(+O2) step. Fig. 3 offers the N2
release rates during the NO(+O2) steps (in square brackets
in the legend) of different experiments. As can be observed,
a NH3 step before the introduction of NO produces a sig-
The results above lead to the conclusion that ammonia
is preferentially linked to the oxygen atoms of manganese
oxides. Certain works also propose oxygen–ammonia in-
termediates for the SCR reaction [21–25]. It can be as-
sumed therefore that NH3 which desorbs at 125 ◦C during
the isothermal desorption stage corresponds to weakly ad-
sorbed NH3 on the oxygen atoms of manganese oxide via
hydrogen bonding. In principle this bonding is performed
O
on the octahedral phase of the oxide, {O=Mn–O–Mn=O} ,
since MnO seems to be a poor NH3 adsorbent. In this re-
◦
spect, an NH3 + NO175 C experiment over pure MnO parti-
cles (not shown) yielded negligible NH3 adsorption during
the step-response stage. As pointed out previously, coad-
sorption of H2O produces a strong decrease in the amount
of weakly adsorbed NH3 (code m in Table 3), indicating
that water is able to establish stable hydrogen bonds with
the oxygen atoms of the manganese oxides. When NH3 is
nificant increase in the formation of N2, from ΦN = 0.002
2
to 0.108. Most of this nitrogen comes from the SCR reaction
of NO with preadsorbed ammonia, specifically with all the
ammonia species ascribed to the LT peak of the TPD curve
and some of the ammonia species ascribed to the MT and

144
G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
Fig. 3. Evolution of N concentration in the exit gases during the NO(+O ) step (in brackets) of experiments performed with CCCM-Mn. The inset is an
2
2
enlarged image of the dashed box section (codes as in Table 2).
HT peaks. Thus the amount of N2 formed during the passage
pretreatment. The following equilibrium equation is intro-
duced to illustrate these results:
of NO (Fig. 3; ΦN = 0.108) is very similar to the amount
2
125 ◦C


O
of NH3 desorbed at the LT peak of the NH3
ment (code h1 of Table 3; Φ
experi-
O






LT
NH3
= 0.102), a peak which is




H
H
◦ ◦
absent in the TPD curve of the NH3125 C > NO125 C exper-
iment. In this experiment, therefore, the oxygen vacancies
are only responsible for a minute contribution to the total
N2 released during the passage of NO. By comparing curves
c and a2 in Fig. 3 (inset) it can be concluded that only the
shaded areas are related to the nitrogen formed by a direct
NO reaction on the oxygen vacancies. The similarity of both
areas implies that no oxygen vacancies were created during
the NH3 step. If oxygen is introduced together with NH3
before the NO step, then a sharper N2 peak appears in the
initial moments of the NO step (curve g in Fig. 3). The N2
peak is not accompanied by a parallel water release (curve
g (H2O) in Fig. 3), which means that it is produced by a re-
duction of NO on the oxygen vacancies rather than an SCR
reaction with preadsorbed NH3. Nevertheless, this peak is
almost coincident with that produced during the NO + O2
experiment (curve b2 in Fig. 3). The oxygen vacancies are
therefore created by CO produced by oxygen oxidation of
the carbonaceous surface [11] during the NH3 + O2 step and
not by the action of NH3.
+ NH3




O
O
2.85 Å






Mn
O
Mn


O
NH










H
H
←→
+ H2O.
(9)
2.6-2.9 Å




O
O






Mn
O
Mn
In the above representation of the octahedral Mn2O3
phase, the Mn–O–Mn bonds represent the long and labile
Mn–O bonds (1/3 of this phase) which are a precursor of
the oxygen vacancies [11], whereas Mn=O represents the
short Mn–O bonds (2/3 of this phase). The distances be-
tween both kinds of oxygen atoms [26] are indicated in the
equation, the H–H distances of H2O and NH3 being repre-
sented on the same scale as the =O =O distance. NH3 is
assumed to be adsorbed on the short Mn=O bonds since
adsorption on the labile oxygen atoms would imply a sig-
nificant decrease in NH3 adsorption at 125 ◦C, when O2 is
present in the gas phase, as a consequence of the formation
of oxygen vacancies. This decrease was not observed in any
of the experiments (points h1 vs i1 in Fig. 4II).
In conclusion, the adsorption of NH3 does not produce
oxygen vacancies on the manganese oxide, and so the con-
comitant water release during the passage of NH3 (Fig. 2I)
must correspond to H2O weakly adsorbed on the oxygen
atoms of the manganese oxide that has survived the catalyst
As can be observed in Fig. 2I, in the absence of O2 and
NO, the release of water from CCCM-Mn starts at ∼ 200 s
(ammonia being first adsorbed on the water-free sites) and
finishes at ∼ 3000 s. This means that attaining equilibri-

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
145
iso
NH3
Fig. 4I plots the experimental values of Φ
vs the NH3
exposure time for all the experiments performed with NH3.
The plotted values (indicated in Table 3) correspond to the
desorption of NH3 under the gas atmosphere of the stage
that followed the passage of NH3 (He in d, h, i, j, and m).
As the amount of ammonia adsorbed by the support was
quite high, it was necessary to subtract its value from the
overall amount adsorbed by the catalyst in order to estimate
properly the amount of NH3 desorbed by the active phase.
These results are plotted in Fig. 4II, where the increasing
iso
NH3
trend of Φ
with NH3 exposure time can easily be ap-
preciated. The points that clearly make up this trend are d1
to d4 (NO + NH3 + O2 experiments), whose lower Φiso
NH3
values with respect to the other points may be explained
in two ways: (i) by the previously noted slowness of the
NH3 adsorption–water displacement process (9); (ii) by a
transient reaction of weakly adsorbed NH3 toward an active
species in the SCR process. Both explanations are probably
valid, since the water displacement process is no longer sig-
nificant at exposure times of over ∼ 30 min (Fig. 2II) and the
weakly adsorbed NH3 has not yet reached its attainable max-
imum under SCR conditions (Fig. 4II). Nevertheless, the fact
that for long exposure times the amount of weakly adsorbed
NH3 in the absence of water is approximately constant re-
gardless of the composition of the atmosphere (corrected
iso
Fig. 4. Moles of desorbed NH per manganese moles versus NH expo-
Φ
= 0.075 0.009) supports the assumption that weakly
3
3
NH3
sure time for: (I) isothermal desorption after the NH (+O (+H O)) stage,
3
2
2
adsorbed NH3 on water-free sites or via equilibrium (9) is
not the active ammonia species in an SCR reaction.
On the other hand, weakly adsorbed NH3 does appear
(II) isothermal desorption corrected for desorption by the support. The
◦
legends indicate the corresponding step-response sequences at 125 C per-
formed over CCCM-Mn (codes as in Table 3).
to react partly with gas-phase oxygen, as can be deduced
iso
NH3
from the decrease in Φ
when oxygen is introduced to-
gether with NH3, especially at 175 ◦C (codes 13 vs 14 in
Table 3). Fig. 2II shows that during the passage of NH3 +O2
at 125 ◦C, NH3 is converted into NO at the same conversion
degree by both the catalyst and the carbonaceous support.
This means that oxidation of NH3 to NO is only catalyzed
by the support. The gas-phase oxidation of NH3 can be dis-
regarded since NO is not released from t = 0 (Fig. 2II). In
any case, the carbon-catalyzed oxidation of NH3 to NO only
um (9) is a slow process. After an NH3 or NH3 + O2 step
the amounts of desorbed NH3 in He (Φiso ; codes h1 and
NH3
i1 in Table 3) are similar to those determined in NO or in
NO + O2 (Φiso ; codes c, k, and g in Table 3). This sug-
NH3
gests that weakly adsorbed NH3 does not directly react with
NO or NO + O2, as do the ammonia species ascribed to the
different TPD peaks.
Table 4
Total moles of NO, N , and N O per carbon moles released during the different NH (+O (+H O)) stages after 60 min reaction
2
2
3
2
2
Sample
T
Stage
Gas/C = mol gas/mol C
◦
( C)
a
a
a
NO/C (%NH )
N /C (%NH )
N O/C (%NH )
2 3
3
2
3
CCCM-40
125
NH
NH + O
NH
NH + O
0
0
0
0
0
3
−3
1.3 × 10 (3.1%)
0
1.9 × 10 (4.4%)
0
3
2
2
2
175
125
0
0
3
−3
−4
−4
6.6 × 10 (3.5%)
0
3.3 × 10 (0.6%)
0
3
CCCM-Mn
NH
3
−3
−4
−4
NH + O
1.3 × 10 (4.2%)
8.2 × 10 (4.3%)
2.1 × 10 (0.7%)
3
−4
NH + O + H O
3.6 × 10 (1.3%)
0
0
3
2
2
175
NH
0
0
0
3
−4
−3
−4
NH + O
9.3 × 10 (2.9%)
1.7 × 10 (10.1%)
6.9 × 10 (5.5%)
3
2
−4
NH + O + H O
2.9 × 10 (1.2%)
0
0
3
2
2
a
Percentage of inlet NH consumed to give the corresponding gas at t = 60 min (pseudo-steady state).
3

146
G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
produces a diminution of ∼ 1–4 vol% in the inlet NH3 con-
centration under all the conditions studied. This is shown in
Table 4, where the amounts of N2, N2O, and NO formed in
1 h of the NH3(+O2(+H2O)) stages at 125 and 175 ◦C for
both the catalyst and the support are presented. For the sake
of comparison, the steady-state conversion of NH3 to N2 at
125 ◦C under SCR conditions was in the range of 50–60%.
The low NH3 reaction to NO cannot possibly explain the
high decrease in weakly adsorbed NH3 at 175 ◦C when oxy-
gen is present. This can only be explained by the reaction
of adsorbed NH3 with O2 to give N2 and, to a lesser extent,
N2O, which is only (125 ◦C) or mainly (175 ◦C) catalyzed
by the active phase, as can be inferred from the values indi-
cated in Table 4. The SCR reaction of NO that evolves from
the carbonaceous surface with the adsorbed NH3-derived
species can be rejected as a potential source of the observed
N2 (at least to any significant extent) since this gas was de-
tected from the very beginning of the stage and NO starts to
be released to the gas phase at t ≈ 550 s (Fig. 2II). In the next
section it will be proved that the adsorbed NO does not react
with NH3 to produce N2 (LH vs ER mechanism section). The
following reactions of direct oxidation are therefore consid-
ered:
tetrahedral phase of the catalyst (8). This mechanism would
explain the small though conspicuous HT peak found in the
◦
125◦C
NO-TPD of the NH3 +O2
94 > NO125 C101 experiment
(code g in Table 2).
4.2.3. LH vs ER mechanism: NO–NH3 competition
As noted in [11], there is still some controversy as to
the state of NO when it reacts with NH3 at low tempera-
tures. There is a broad consensus regarding the fact that NH3
always reacts from an adsorbed state, but results from vari-
ous sources suggest that the NO state (gaseous or adsorbed)
when reacting with adsorbed NH3 depends on the catalytic
system [5,9,10,28,29]. At this stage of the discussion we are
in position to determine for CCCM-Mn whether NO reacts
from the gas phase (ER mechanism), from an adsorbed state
(LH or Freundlich mechanism), or by a combined ER-LH
mechanism. An assessment can be reached by analyzing the
results of a specific step-response experiment. Fig. 5I dis-
plays the evolution of the N2 and NO concentrations during
125◦C
125◦C
the step-response stages of NO + O2
90 > NH3
.
91
In the first stage N2 is produced by NO reduction on the oxy-
gen vacancies of the catalyst. When the passage of NO + O2
is stopped and NH3 is introduced, the production of N2
ceases, even though the catalyst surface is covered with ni-
trosyls, nitrites, and nitrates. This is strong evidence that NO
must be in the gas phase to react with adsorbed NH3. Further
support for this hypothesis is provided by the TPD results of


O
NH










H
H
3
+ O2
2
2




O
O






Mn
O
Mn
ꢀ
ꢁ
O
O
→ 2
O + N2 + 3H2O,
(10)
Mn
NH
O
Mn


O










H
H
2
+ 2O2


O
O








Mn
O
Mn
ꢀ
ꢁ
O
O
→ 2
O + N2O + 3H2O.
(11)
Mn
O
Mn
According to Table 4, reaction (10) can convert around 10%
of the inlet NH3 into N2 at 175 ◦C (SCO reaction [27]),
the temperature at which reaction (11) also starts to be-
come noticeable (∼ 5% conversion of NH3). The fact that
the presence of water inhibits both reactions (Table 4) is yet
further proof that reactions (10) and (11) are produced from
weakly adsorbed NH3, displaced from the surface by water
(codes m and 15 in Table 3), and not from the more stable
ammonia-derived species desorbed during the TPD stage. It
is not surprising that NO is not produced by NH3 oxidation
on the active phase, since this reaction is known to occur
at higher temperatures than those needed for reactions (10)
and (11) [27]. Finally, the NO that evolves from the carbona-
ceous surface after NH3 oxidation may be partly adsorbed
as nitrosyls on the surface of the active phase (3), then con-
verted to NO2 (7), and eventually stored as nitrates in the
125 ◦C
125 ◦C
Fig. 5. Results of the NO + O
> NH
experiment:
91
2
90
3
(I) Evolution of N and NO during the step-response stage. (II) Evolution
of NH , NO, and N during the TPD stage (codes as in Table 2).
2
3
2

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
147
the same experiment (Fig. 5II). As already pointed out, N2
production begins only when NO starts to be desorbed into
the gas phase. Moreover, the maximum desorption rates for
both gases are coincident. Therefore the reaction between
NO (or NO2) and adsorbed NH3 follows an Eley–Rideal
mechanism on the CCCM-Mn catalyst.
It is somewhat surprising that the NO released during the
NH3 stage (Fig. 5I) does not produce N2 via further reac-
tion with adsorbed NH3, as does the NO released during
the TPD. This is because the nitrogen oxides released dur-
ing the TPD are mainly NO2 (from nitrate decomposition)
whereas the nitrogen oxides observed during the passage of
NH3 correspond to the NO produced by the desorption of
bridged nitrites. As noted before, gaseous NO reacts with
preadsorbed NH3 to give N2. However, at 125 ◦C this reac-
tion is very slow, as can be deduced from the slow decay
of N2 during the passage of NO (curve g in Fig. 3). It can
be assumed therefore that the NO escaping from the catalyst
surface has no time to react with the adsorbed NH3, and is
unable to produce N2.
On the other hand, the decomposition of surface nitrates
during the TPD produces mainly NO2, although this gas is
detected as NO due to its transient release. The NO2 mole-
cules are able to react with the adsorbed NH3 when escaping
from the surface, especially with the ammonia species as-
cribed to the MT and HT peaks of the NH3-TPD, whose
maximum desorption temperatures (262 and 370 ◦C, respec-
tively; Table 1) are above that for the HT peak of the NO-
TPD (231 ◦C). As can be seen in Fig. 3 the production rate
of N2 when NO + O2 is introduced into the system after
the adsorption of NH3 (curve k) is much faster than that ob-
served when only NO is passed through the catalyst. This
can be attributed to a faster SCR reaction by NO2 which is
produced by reaction (7). It can, therefore, be concluded that
NO2 is more reactive than NO toward the SCR reaction and
consequently can partly react with the adsorbed NH3 species
when it is formed by nitrate decomposition during the TPD
stage.
Fig. 6. Moles of desorbed NH per manganese moles (corrected for support
3
desorption) versus NH exposure time for: (I) LT peak of the NH -TPD,
3
3
(II) MT + HT peaks of the NH -TPD. The legends indicate the correspond-
3
◦
ing step-response sequences at 125 C performed over CCCM-Mn (codes
as in Table 3).
sient reaction can be written as:
Finally, the fact that the incoming NH3 causes the re-
lease of bridged nitrites from the catalyst surface explains
the peculiar evolution of this surface species during the SCR
process (codes d1 to d5 in Table 2), a matter that was left un-
explained in [11]. As can be seen, the initial increase in the
amount of surface nitrites is produced when the catalyst sur-
face is far from being saturated with weakly adsorbed NH3
(points d1 to d3 in Fig. 4). When the surface starts to be satu-
rated with weakly adsorbed NH3 (points d4 and d5 in Fig. 4),
the nitrites are displaced from the surface (codes d4 and d5
in Table 2). As happened with the surface nitrites formed on
the carbonaceous surface by the carbon-catalyzed oxidation
of NH3, the nitrites formed on the manganese oxide start to
be displaced by NH3 only when the NO-free sites have al-
ready been occupied by NH3. In this process the possibility
of a spillover phenomenon cannot be disregarded. This tran-
(12)
At this stage it is not possible to discriminate between the
options suggested by reaction (12), although considering the
minute contribution of (12) to the SCR process as a whole,
this point is of minor importance. A similar reaction can be
established for the competition of water with surface nitrites
(point f in Table 2), by merely substituting H2O for NH3 in
reaction (12).
4.2.4. NH3 desorbed in the low-temperature peak (ΦLT ):
NH3
steady-state SCR
LT
Fig. 6I plots the experimental corrected values of Φ
NH3
vs the NH3 exposure time for all the experiments performed

148
G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
◦
◦
Fig. 7. Evolution of gas-phase composition during the step-response stage of the SCR experiments indicated in the figure: (I) 175 C, (II) 125 C.
with NH3. Negative values imply that the NH3 species as-
cribed to the LT peak has been removed not only from
the active phase but also from the carbonaceous support.
As observed, the presence of water (point m) causes a de-
crease in the LT peak area. This can be attributed to the
fact that the NH3 species ascribed to the LT peak originates
from the previously weakly adsorbed NH3. This species
must fulfill the following requisites: (i) it must be formed
Several authors have proposed that the adsorption of NH3
over a vanadium-based catalyst leads to the formation of the
intermediate aminooxy groups, V–ONH2 [18,22,23,30,31].
In a recent work provided by Larrubia et al. [32] it was
stated that the spectroscopic features of adsorbed hydroxy-
lamine (oxyamine is a dissociated form of hydroxylamine)
are hardly distinguishable from those of coordinated am-
monia, so that spectroscopic confirmation or rejection of
the –ONH2 species is not straightforward. According to
Kapteijn et al. [5], coordinated ammonia species present
symmetric deformation in the 1100–1320 cm−1 region and
some of the peaks observed in the infrared spectra displayed
in a previous work [11] for CCCM-Mn subjected to SCR re-
action are located in this spectral region. We can therefore
assume that the bands at 805, 1097, and 1263 cm−1, as dis-
played in Fig. 7 of [11], are produced by the vibrations of
aminooxy groups.
without the participation of gas-phase oxygen (which ex-
LT
plains the similarity of the corrected Φ
values for the
NH3
125 ◦C
125◦C
NH3
and NH3 + O2
experiments (points h1 and
i1 in Fig. 6I)); and (ii) it must not occupy oxygen vacancies
on the manganese oxide, so that it does not compete with
adsorbed nitrosyls. These two premises are accomplished by
the oxyamine species formed via the following equilibrium:
As noted in the previous sections, the ammonia species
ascribed to the LT peak (aminooxy groups) react with NO
in the SCR process at 125 ◦C, which is consistent with the
position of points k, g, and c in Fig. 6I. However, under
pseudo-steady-state conditions (t > 60 min) the corrected
(13)

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
149
area of the LT peak is very similar both in the presence and
tion of N2:
in the absence of NO (points j, i1, h1, and d5 in Fig. 6I). Two
interesting conclusions can be drawn from this:
1. Under the TPD conditions, gas-phase NO2 reacts with
the NH3 species ascribed to the MT and HT peaks, but
not with the NH3 species ascribed to the LT peak, at
least not fast enough for the fugitive NO2 molecules.
This is deduced by comparing the TPD areas of the ex-
periments in which NO2 does not evolve during the TPD
(points h1 and i1 in Fig. 6I and Table 3) and those of the
experiment in which NO2 is released during the TPD
(point j in Fig. 6I and Table 3). However, as can be seen
from Fig. 3, when NO and O2 are continuously supplied,
the NO2 formed must react with –ONH2; otherwise the
long and gradual decay for times above 1000 s exhib-
ited by the curves in which only NO is supplied after
NH3 (curves c and g) would also be appreciated for the
curve in which NO + O2 is supplied (curve k). It can be
concluded that reaction of NO2 with the –ONH2 groups
(LT peak) is slower than the reaction of NO2 with the
NH3 species ascribed to the MT and HT peaks. The last
sector of curve k (slope between ∼ 300 and ∼ 1200 s)
corresponds, therefore, to the reaction of NO and NO2
with the –ONH2 groups. Thus, the acute N2 release peak
displayed by curve k in Fig. 3 is formed by the com-
bined contribution of the N2 produced by the reaction
of NO2 with the species ascribed to the MT and HT
peaks and the N2 that results from NO reduction on
the oxygen vacancies. The overlapping of both contri-
butions can be detected in the circled sector of curve k
in Fig. 3.
(15)
2. The reactions of NO/NO2 with –ONH2 are significantly
slower than the aminooxy formation reaction (13). Equi-
librium (13) is attained rapidly as can be deduced from
(16)
At 125 ◦C reaction (16) is faster than reaction (15), al-
though both are slow enough to ensure the saturation of the
oxide with aminooxy groups in the steady state. The desorp-
tion of small amounts of N2 and O2 was observed during
LT
iso
the roughly constant values of the Φ
/Φ
ratio
NH3
NH3
(corrected values) in the absence of water: 1 0.2. This
value therefore must be the equilibrium constant of re-
action (13) at 125 ◦C.
125◦C
the isothermal desorption stage of the NO + O2 + NH3
experiments performed after the reaction stage. This is con-
sistent with the mechanism proposed for reaction (16).
At this point we can try to interpret the transient processes
that produce the trends depicted in Fig. 3 in the light of
the reactions proposed for N2 formation. In the NH3 pre-
saturated samples (curves c, g, and k), when only a small
fraction of oxygen vacancies is available on the manganese
oxide (curve c), direct reaction on the vacancies cannot ini-
tially reduce all the incoming NO and so reaction (15) starts
at t = 0. This explains the differences between curves a2
(where the reaction only occurs on vacancies) and c (where
the reaction on vacancies and the reaction (15) overlap). On
the other hand, when a large amount of vacancies is avail-
able due to previous or simultaneous reduction via evolving
CO (curves g and k), two situations arise:
Aminooxy groups are known to react under ambient con-
ditions with ketones and aldehydes to form stable oximes.
This reaction has recently been used in the immobilization
of polymers and nanoclusters [33,34] and can be schema-
tized as follows:
R–O–NH2 + O=CH–Rꢁ → R–O–N=CH–Rꢁ + H2O. (14)
Similarly we propose the following reaction paths for
the reduction of NO and NO2 with the aminooxy groups
formed on the manganese oxide, by substituting Rꢁ–CH=O
for NO or NO2 in (14), and allowing the subsequent desorp-

150
G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
1. In the absence of O2 (curve g) all incoming NO is ini-
In the light of these considerations we propose that the va-
cancies are formed by surface dehydroxylation of the solid
product of reactions (15) and (16):
tially reduced on the oxygen vacancies and reaction (15)
starts to acquire relevance when NO becomes available
due to the progressive filling of the oxygen vacancies.
This is deduced from the shape of the curve of water re-
lease (inset) whose maximum rate is clearly displaced
at t > 0, and from the appearance of a second peak in
the curve of N2 release, corresponding to the onset of
N2 desorption formed via reaction (15). The fact that
this peak is slightly displaced with respect to the water
release peak suggests that N2 desorption in reaction (15)
is not an instantaneous process.
ꢀ
ꢁ
O
OH
O
2
Mn
O
Mn
ꢀ
ꢁ
ꢀ
ꢁ
O
O
O
O
O
→
+
O + H2O.
Mn
ꢀ
Mn
Mn
O
Mn
(17)
This step is a necessary part of most SCR mechanisms,
as reported by Busca et al. [27]. The special characteristic
proposed for reaction (17) is that one labile oxygen atom
of the octahedral phase removes two hydrogen atoms from
the adjacent hydroxyl groups. The dehydrogenation activity
of surface oxygen in the Mn3O4 catalyst has already been
studied by Baldi et al. [1] in the process of propane oxy-
dehydrogenation. Other possible paths should also take into
account the stoichiometry of reaction (17), the only restric-
tion being that gaseous oxygen cannot participate directly in
the reaction, at least at 125 ◦C, as, at this temperature, O2
does not dissociate on the vacancies.
In order to close the redox cycle and retain catalytic ac-
tivity in the steady state, the oxygen vacancies created by
reaction (17) must be refilled, or in other words, the octahe-
dral phase of the manganese oxide must be reoxidized. Two
situations can be encountered:
2. In the presence of oxygen (curve k) there is no appar-
ent restriction for the simultaneous occurrence of NO
reduction on the oxygen vacancies or for NO2 forma-
tion via nitrosyl oxidation, so that the SCR reactions
of NO2 with the adsorbed NH3 species (LT (16) and
MT + HT peaks) occur from the very beginning of
the step, as does the direct reaction of NO with the
aminooxy groups (15), though the latter takes place to
a lesser extent. This overlapping explains the difference
in the values of maximum rate of N2 release for curves
k (with preadsorbed NH3) and b2 (without preadsorbed
NH3).
In the light of the facts presented above, the initial val-
ues for reactions (15) and (16) at 125 ◦C can be estimated
by extrapolating to t = 0 the straight sections of curves c
and k, respectively, within a time interval in which NO re-
duction on the vacancies (and SCR reaction of MT + HT
peaks) is minimal (i.e. ∼ 700–1000 s). This procedure yields
the values displayed in Table 5, which are ∼ 5.0 × 10−5 and
1. In the absence of gas-phase oxygen there is steady-
state conversion of NO to N2 as observed in Fig. 7I.
Thus, for the experiment depicted in the figure (NO +
175◦C
175◦C
∼ 2.1 × 10−4 molN mol−1Mn s−1) for (15) and (15) + (16),
O2 + NH3
75 > NO + NH3
134) the values of
2
respectively, indicating that at 125 ◦C NO2 reduction by the
aminooxy groups is about 3 times faster than NO reduction
by the same groups.
steady-state conversion of NO to N2 were ∼ 84% for
the first step and ∼ 9% for the second step. The steady-
state level of water in the second step confirms the view
that the residual conversion of NO to N2 is a product
of reaction (15). In this situation there is only one way
to ensure the reoxidation of the manganese oxide so
that the catalytic cycle can be closed: the reduction of
nitrosyls on oxygen vacancies ((3) + (4)). In principle
this should be a rapid (noncontrolling) process, as can
be seen from the acute N2 peaks observed in Fig. 3.
In fact, the value for the N2 formation rate in the ab-
sence of oxygen evaluated at the steady state for the
In a previous work [11] it was suggested that the en-
hanced formation of nitrosyls and nitrites under the SCR
conditions for low NO exposure times could be due to an
increase in the creation of oxygen vacancies resulting from
the SCR mechanism itself. This can be clearly seen in the
◦
◦
NH3125 C > NO + O2125 C experiment (point k in Table 2),
in which the amounts of nitrosyls (ΦNisOo ), nitrites (ΦNLTO),
and nitrates (ΦNHOT) are remarkably high. The formation of
these species is favored by the presence of oxygen vacancies
((3) and (6)–(8)). However, for this experiment the excess of
surface NO species cannot be explained solely by the O2-
related creation of oxygen vacancies, which would induce
175◦C
175◦C
NO + O2 + NH3
> NO + NH3
exper-
134
75
iment is 7.1 × 10−5 molN molM−1n s−1 (Table 5), a value
2
that might well correspond to the reaction rate of (15)
at 175 ◦C, since at 125 ◦C this was estimated to be
an adsorption of NO species lower than the one observed
125◦C
5.0 × 10−5molN molM−1n s−1 (Table 5). All of these ar-
(see NO + O2
experiments: points b in Table 2), nor
2
guments imply that the controlling step of the whole
process is the SCR reaction (15). This is further sup-
ported by the high rate of the dehydroxylation reaction,
which is implied by the large amount of nitrosyls and
bridged nitrites on the catalyst surface for very low re-
action times (code d1 in Table 2).
by the previous NH3 step, in which the passage of NH3
by itself is not able to produce oxygen vacancies, as dis-
cussed above. Consequently, the oxygen vacancies on the
manganese oxides must be caused by certain reactions of the
SCR mechanism. Reactions (15) and (16) do not produce va-
cancies and so these must be formed in a subsequent stage.

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
151
Table 5
Values estimated for different reaction rates
b
Experiment
State
Reaction
T
Rate
(mol mol
−1
◦
−1
)
( C)
s
N
2
Mn
125 ◦
125 ◦
C
C
125 ◦
C
a
a
O
−5
−4
−4
NH
NH
> NO
Transient
Transient
(15)
125
125
125
5.0 × 10
2.1 × 10
3
3
97
90
94
125 ◦
C
O
O
> NO + O
((15) + (16))
2
100
125 ◦
C
T
T
NO + NH + O
Pseudo-steady state
((15) + (16)) + (23)
5.4 × 10 (30 min)
3
2
120
−4
4.6×10 (120 min)
175 ◦
175 ◦
C
C
175 ◦
175 ◦
C
C
O
−5
NO + O + NH
> NO + NH
> NO + NH
Steady state
Pseudo-steady state
(15)
175
175
7.1 × 10
2
3
3
75
75
3
3
134
134
O
−4
NO + O + NH
((15) + (16)) + (23)
7.3 × 10 (30 min)
2
−4
7.2 × 10 (75 min)
a
Initial rate evaluated by the extrapolation of interval 700–1000 s.
Superscripts indicate the manganese oxide phase in which reactions occur (O, octahedral; T, tetrahedral).
b
Summing up, in the absence of oxygen, the steady-state
SCR mechanism at 125 ◦C on the octahedral phase of
the manganese oxides comprises the following steps:
(i) NH3 adsorption on the oxygen atoms, which partly
competes with the preadsorbed water molecules (9),
(ii) formation of the aminooxy groups (13), (iii) SCR re-
action of these groups with gas-phase NO (15), (iv) sur-
face dehydroxylation (17), and (v) surface reoxidation
of oxygen vacancies via nitrosyl reduction to N2 ((3) +
(4)). The controlling step of the whole process is the
SCR reaction (15) itself (step (iii)).
In the absence of an SCR reaction, water is released
mainly in the initial transient period from the catalyst
surface (i.e., during a single NH3 passage; Fig. 2I). In
contrast, the SCR reaction causes the continuous re-
lease of water in the steady state, as observed in Figs. 7I
and 7II, where the evolution of water under SCR con-
ditions at 125 and 175 ◦C is displayed. As noted previ-
ously, some of this water may be adsorbed on the oxygen
atoms of the manganese oxides via the following equi-
librium:
with respect to the reaction in the absence of oxygen is
that now we have two possible oxidants: NO and NO2.
The latter is known to be a more active oxidant than
NO and O2 [35]. If NO2 oxidizes the oxygen vacancies,
then oxygen acts as a reagent in the SCR process on the
octahedral Mn3O4 phase. On the other hand, if NO is
responsible for closing the redox cycle, then gas-phase
oxygen is in fact acting as a catalyst, since the same
amount of oxygen that produces NO2 (reaction (7) is af-
terward released during the SCR reaction (6) (this is not
completely true since the formation of surface nitrates
consumes oxygen, though at a very slow rate). Under
these conditions the SCR reaction would progress ac-
cording to Eq. (19). In principle it is difficult to make
a definite decision on this issue, on the basis of the re-
sults available. One possible way to solve the problem is
to assess the ratio of consumed NH3 per consumed NO
during the SCR reaction. The value of this ratio is 0.67
in the absence of oxygen (reaction (19)) and 1.00 for the
classical SCR reaction in the presence of oxygen [27]:
4NO + 4NH3 + O2 → 4N2 + 6H2O.
(20)
ꢀ
ꢁ
O + H2O
O
O
Considering all the possible sinks and sources of NO
and NH3 (SCR plus side reactions as NH3 oxidation (Ta-
ble 4)), the experiment displayed in Fig. 7II yields a ra-
tio of gas consumption via the SCR reaction (NH3/NO)
of 0.82, which is somewhere between the values cor-
responding to reactions (19) and (20). As will be dis-
cussed in the following section, the SCR reaction on the
tetrahedral environmentof the manganese oxide initially
accounts for about half of the N2 produced at 125 ◦C
and operates according to reaction (20). This suggests
strongly that the SCR reaction on the octahedral en-
vironment is responsible for decreasing the NH3/NO
consumption ratio from 1.00 to 0.82. In this case it pro-
gresses according to (19). In other words, even in the
presence of oxygen at 125 ◦C NO performs the oxida-
tion step needed to close the catalytic cycle.
Mn
O
Mn


O
O
H
O










H
←→
.
(18)


O








Mn
O
Mn
Now, all the reactions needed to close the catalytic cycle
in the absence of oxygen on the octahedral phase of the
manganese oxides have been defined. This cycle is de-
picted in Fig. 8I in which the appropriate combination
of reactions is included. This cycle yields the expected
SCR stoichiometry in the absence of oxygen [27]:
6NO + 4NH3 → 5N2 + 6H2O.
(19)
2. In the presence of gas-phase oxygen other facts must
be considered. As oxygen cannot reoxidize the gener-
ated vacancies at 125 ◦C, reoxidation of the active phase
must be performed by a different agent. The difference
With this, the set of reactions for the SCR mechanism at
125 ◦C on the octahedral phase of the manganese oxides
is completed. Again, in the presence of O2 the control-
ling step is the SCR reaction (16). This is indicated by

152
G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
cussed below, the slow deactivation of the catalyst is related
to the NO/NH3/O2 interactions that occur in the tetrahedral
phase of the catalyst.
4.2.5. NH3 desorbed at the medium and high-temperature
peaks (ΦNMHT+HT): pseudo-steady-state SCR mechanism
3
The reaction rates values obtained for (15) and (16) at
125 ◦C (5.0 × 10−5 and 1.6 × 10−4 molN mol−M1n s−1, re-
2
spectively; Table 5) cannot explain the high nitrogen for-
mation rate observed under the SCR reactions. This can
be calculated at the pseudo-steady state from the slope of
the dashed line associated with sector 3 of the N2 evo-
lution curve during SCR at 125 ◦C (Fig. 7II). The val-
ues of the rate vary from 5.4 × 10−4 molN molM−1n s−1 at
2
30 min to 4.6 × 10−4 molN molM−1n s−1 at 120 min. This
2
means that a process different to reactions (15) and (16)
yields about 60% of the total N2 produced, this propor-
tion decreasing gradually with time to a point where it dis-
appears altogether. From the results analyzed in previous
sections it can be concluded that this process consists of
an SCR reaction of NO2 with the NH3 species ascribed
to the MT and HT peaks. These species, as previously
pointed out, are ammonium ions formed on the hydroxyl
groups that were produced by reaction (2) on the tetra-
hedral phase of the manganese oxides. The presence of
+
NH4 ions on the Brønsted sites of the metal oxide surfaces
is typically detected by asymmetric bending vibrations in
the 1400–1500 cm−1 region (V2O5/TiO2, 1428 cm−1 [36];
Mn2O3, 1459 cm−1 [9]; Fe-MOR, 1463 cm−1 [37]; Fe-
ZSM5, 1473 cm−1 [38]; MnOx/Al2O3, 1480 cm−1 [5]; etc).
Kapteijn+et al. [5] also report a vibration band at 1390 cm−1
for NH4 on MnOx/Al2O3. The infrared results displ+ayed
in [11] neither confirm nor refute the presence of NH4 on
the surface of the catalyst after the SCR reaction at 175 ◦C.
However, emerging through the noise of the IR signal there
appears to be a cons+picuous band at 1407 cm−1 which could
be assigned to NH4 ions.
Fig. 6II offers the variations with NH3 exposure time of
the corrected area values for the combined contribution of
the MT and HT peaks to the NH3-TPD. The MT and HT
peaks were analyzed as a whole and not separately due to
the close relation between the areas of both peaks (the cor-
Fig. 8. Mechanism of the steady-state SCR reaction: (I) in the absence of
oxygen; (II) in the presence of oxygen.
the low availability of oxygen vacancies in the steady
iso
NO
state (low value of Φ at high NO exposure times; Ta-
ble 2), which implies a comparatively high rate for the
oxidizing reaction (in principle (3) + (4)).
To summarize, in the presence of oxygen, the steady-
state SCR mechanism at 125 ◦C on the octahedral phase
of the manganese oxides, plotted in Fig. 8II comprises
the following steps: (i) NH3 adsorption on the oxy-
gen atoms, which partly competes with the preadsorbed
water molecules (9), (ii) formation of the aminooxy
groups (13), (iii) NO adsorption as nitrosyls on the oxy-
gen vacancies (3), (iv) oxidation of the surface nitrosyls
to gas-phase NO2 (7), (v) SCR reaction of aminooxy
groups with gas-phase NO2 (16), (vi) surface dehydrox-
ylation (17), and (vii) surface reoxidation of oxygen
vacancies via nitrosyl reduction to N2 ((3) + (4)). The
controlling step of the whole process is the SCR reac-
tion (16) itself (step (v)). It can be seen that the appropri-
ate combination of reactions yields the SCR stoichiom-
etry for reaction (19).
HT
MT
rected Φ values follow the general trend Φ
= 0.41Φ
;
NH3
NH3
R2 ≈ 0.8). These values suggest that similar processes are
involved in the formation of the species responsible for both
peaks. For the experiments in which NO and NH3 species
were present on the surface of the catalyst prior to the
TPD stage (c, d, j, k, in Fig. 6II), the normalized amount
of N2 released during the TPD was added to the ΦMT+HT
NH3
values (open symbols). As already noted this N2 comes
mainly from the SCR reaction of evolving NO2 with the NH3
species of the MT and HT peaks.
As can be seen in the figure, when NO is passed through
the catalyst after NH3 adsorption there is only a small de-
MT+HT
In principle the proposed mechanisms are thought to oc-
cur at a steady rate during the SCR process. As will be dis-
crease in the value of Φ
(point c), as compared to the
NH3
large decrease caused by the passage of NO + O2 (point k).

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
153
This is an indication of a higher rate of reaction of NO2 with
the related NH3 species compared to that of NO.
Under SCR conditions (points d in Fig. 6II), the val-
MT+HT
ues of Φ
initially increase up to approximately the
NH3
maximum possible value, and then start to decrease slowly.
MT+HT
The initial increase of the Φ
values with NH3 expo-
NH3
iso
sure time is parallel to that of Φ
. This suggests that the
NH3
MT-HT peak species are formed by the surface migration
of weakly adsorbed NH3. The decrease of ΦMT+HT caused
NH3
by the presence of water (point i1 vs point m) is consistent
with this assumption. On the other hand, the slow decrease
MT+HT
Fig. 9. Deactivation mechanism of the tetrahedral phase of the manganese
oxides.
of Φ
after they have reached the maximum value is
NH3
parallel to the decrease observed for the overall reaction rate
(Fig. 7II) and to the increasing trend in the amount of nitrates
adsorbed on the tetrahedral phase of the manganese oxides
oxygen excess x (Mn3O4+0.5x). Assuming that at 125 ◦C all
the available hydroxyls adsorb NH3 molecules according to
the stoichiometry of reactions (21) or (22), the oxygen ex-
HT
(Φ in Table 2).
NO
To recapitulate: (i) the reaction of NO2 with the NH3
species ascribed to the MT and HT peaks is faster than reac-
tions (15) or (16), (ii) the temperature range of the desorption
of the MT and HT peaks is similar to that of ammonium ions
cess can be roughly calculated as x = 3ΦNMHT+HT. For the
3
maximum value of ΦNMHT+HT (∼ 0.05) oxygen excess adopts
3
a value of x ≈ 0.15, which is well within logical expecta-
adsorbed on the acid hydroxyl groups of zeolites, (iii) under
tions (0 < x < 1).
MT+HT
SCR conditions the decreasing trend of the Φ
values
NH3
The reaction of NO2 with ammonium ions on a solid
surface is commonly used in the preparation of solid ox-
ides. With respect to the SCR process, this reaction has been
thoroughly studied by the group of Long and Yang [37–39]
with iron-exchanged zeolites, with the result that the reac-
tion rate with NO+O2 was much higher than with only NO,
which is in agreement with the results obtained in the present
work. We should therefore be able to apply the same reaction
mechanism for our catalytic system:
(Fig. 6II) is parallel to the decreasing rate of N2 formation
(stage 3 in Fig. 7II) and to the increasing amount of nitrates
adsorbed on the tetrahedral phase of the manganese oxides
HT
(Φ in Table 2).
NO
All of these results point to a singular mechanism in
which two types of ammonium ions of different strength
(thereby producing MT and HT peaks) bonded to the hy-
droxyls of the tetrahedral phase of the manganese oxide react
mainly with NO2 (more slowly with NO) to produce N2. In
the presence of NO + O2 the availability of ammonium ions
decreases with time due to the gradual formation of nitrates.
{(NH4+O−)(HO)Mn–O–Mn(OH)(O−NH4+)}T+NO
T
+ NO2 → {(HO)2Mn–O–Mn(OH)2} + 2N2 + 3H2O.
+
The following reactions for the formation of NH4 ions are
(23)
therefore proposed:
As expected, the appropriate combination of reactions ((3)+
(7) + (21)/(22) + (23)) yields reaction (20) for the whole
SCR process on the tetrahedral phase of the manganese ox-
ides.



T
HO
OH
OH

+ 2NH3
Mn O Mn


HO




T
The gradual deactivation of reaction (23) occurs through
the substitution of the hydroxyl groups for nitrates (reac-
tion (8)). This reaction competes with the SCR reaction,
as indicated in Fig. 9. Initially the active phase has an ar-
bitrary solid concentration of tetrahedral MnO. At 125 ◦C
this is in equilibrium with a given concentration of the prod-
uct of reaction (1) (oxygen excess formation) which de-
pends on its equilibrium constant [20]. In the absence of (8),
oxidized MnO (‘oxygen excess’) would continue to react
through (2), (21)/(22), and (23) in a steady SCR cycle. The
role of reaction (8) is to remove hydroxyl groups by the
formation of nitrates. It therefore acts as a sink for oxi-
dized Mn=O. Reaction (8) causes the progressive decrease
of the {Mn=O}T concentration by shifting equilibria (1) and
(2) to the right. This is reflected in the loss of activity of
the tetrahedral phase. Ideally, in the steady state the whole
{Mn=O}T phase should have been converted to manganese
NH4+O−
HO
O−NH4
OH
+
→
,
(21)
Mn O Mn





T
HO
OH

+ 2NH3
Mn O Mn


HO
OH


T
NH4+O−
NH4+O−
OH

→
Mn O Mn
.

(22)

OH
Of course these reactions are proposed only as an attempt
to explain the different thermal desorption ranges of the MT
and HT peaks. In any case, the general concept of ammo-
nium formation on the hydroxyl groups can be accepted
from the conclusions of the results discussed. These reac-
tions only occur on the manganese atoms associated with the

154
G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
nitrates. However, under SCR conditions the nitrate forma-
the whole process is due to the oxidation of weakly adsorbed
NH3 (11).
HT
tion rate is rather low (see Φ evolution for points d1–d5
NO
in Table 2) as a consequence of the diminished availability
of NO2 (reactions (16) and (23)) and hydroxyl groups (reac-
tion (21)/(22)) and so the deactivation progress is very slow,
hence the label pseudo-steady state. In fact, a comparison
of the reaction rates of ((15) + (16)) at 125 ◦C with that of
((15) + (16) + (23)) (Table 5) the rate for (23) should be-
come zero (complete deactivation) at ∼ 6 h; from this point,
the residual catalytic activity at steady state would be due to
reactions (15) and (16), though mainly the latter.
Reaction (1) implies that oxygen can cause a slight oxi-
dation of the tetrahedral MnO phase at 125 ◦C. This is con-
sistent with the stability of tetrahedral Mn3+–O bonds asso-
ciated with the oxygen excess, which was found to be higher
than that corresponding to the octahedral long Mn3+–O
bonds, the latter being responsible for the formation of oxy-
gen vacancies. Reaction (1) does not, therefore, contradict
the fact that oxygen does not dissociate on the octahedral
vacancies at 125 ◦C.
4.2.6. Temperature-based transition toward a more
conventional mechanism
Although this work has mainly focused on the SCR
mechanism at a low temperature (125 ◦C), the information
obtained from the experiments performed at a higher tem-
perature (175 ◦C) also gives rise to some interesting conclu-
sions. At 175 ◦C oxygen can be dissociated on the oxygen
vacancies of the manganese oxides. At the same time NO2
formation starts to be hampered by thermodynamic constric-
tions. This explains the inverse relationship between the con-
centration of surface nitrates and temperature (the value of
ΦNHOT decreases from ∼ 0.2 at 125 ◦C to ∼ 0.1 at 175 ◦C (Ta-
ble 2)). The increased temperature favors the oxidation of the
tetrahedral phase via gas-phase oxygen, as can be seen from
MT+HT
the greater increase of the Φ
value when oxygen is
NH3
added to the gas phase (codes 13 and 14 in Table 3: from
∼ 0.03 to ∼ 0.09). At 125 ◦C this increase is much smaller
(codes h1 and i1 in Table 3: from ∼ 0.06 to ∼ 0.08).
Thus there is a transition stage in which all the mech-
anisms discussed for the reaction at 125 ◦C start to evolve
toward a limiting mechanism in which: (i) NO replaces NO2
(no longer formed) as the main reagent in the SCR reaction,
(ii) the reoxidation of the active phase is performed by gas-
phase oxygen instead of NO/O2, (iii) nitrate formation does
not take place since NO2 is no longer available. This is con-
sistent with the conventional SCR mechanisms reported in
most of the literature. However, this discussion is irrelevant
for the system studied in this work, since a higher temper-
ature also implies extensive oxidation of the carbonaceous
support [6,7] and consumption of NH3 in unwanted oxida-
tion reactions (10) and (11).
We can now interpret the transient behavior of the dif-
ferent gases during the SCR reaction at 125 ◦C (Fig. 7II).
During the first stage (1), N2 evolution is rapid as a con-
sequence of NO reduction on the oxygen vacancies, their
number being significantly increased by the presence of oxy-
gen which gives rise to the evolution of CO [11]. The small
H2O step (∼ 0.02%) observed at this stage is mainly due to
the release of H2O from the oxidation of the support [11].
During stage (2) N2 evolves mainly through combined SCR
reactions. The steady increase in the concentration of N2
is a consequence of the increased availability of surface
aminooxy groups (Fig. 6I) and ammonium ions (Fig. 6II).
At this stage H2O release is significantly increased by the
combination of a variety of factors: (i) the SCR reactions
((15) + (16) + (23)), (ii) the displacement of H2O by weakly
adsorbed NH3 (9), and (iii) H2O released as a consequence
of nitrate formation (8). The maximum of NO concentration
observed at this stage is a consequence of the intersection
between the decreasing rate of NO reduction by oxygen va-
cancies (curve b2 in Fig. 3) and the increasing rate of NO re-
duction via SCR reactions as a consequence of the increased
availability of surface ammonia-derived species. Finally, in
stage (3) the concentration of H2O starts to decrease for two
reasons: (i) the reduced availability of weakly adsorbed wa-
ter for reaction (9), and (ii) the decreased SCR reaction rate
of NO2 with ammonium ions (23) due to the progressive de-
activation of the tetrahedral phase (8). The second reason
also explains the increasing trend of the NH3 and NO con-
centrations and the decreasing trend of N2. Surface nitrites
formed by reaction (6) are released at the beginning of this
5. Conclusions
This report analyzes the role of surface NH3 species in
the low-temperature (125 ◦C) SCR mechanism with the aim
of determining the mechanism of the SCR reaction in this
catalyst. The following conclusions can be drawn from what
has been discussed:
NH3 is weakly adsorbed on oxygen atoms of both the
carbonaceous support and the active phase. Ammonia ad-
sorbed by the support at 125 ◦C (∼ 30% of total adsorbed
ammonia) produces an insignificant amount of NO by re-
acting with gas-phase oxygen. The rest of the weakly ad-
sorbed ammonia is linked via hydrogen bonds to the oxygen
atoms of the octahedral phase of Mn3O4. This adsorption
displaces water and the not very significant linear (mon-
odentate) or bridged nitrites previously present on the same
oxygen atoms. In the presence of oxygen a small quantity
of the ammonia weakly adsorbed by the active phase is con-
verted to N2 and N2O. Weakly adsorbed ammonia is partly
converted to aminooxy groups by reaction with lattice oxy-
LT
NO
stage (reaction (12); Φ in Table 2). This phenomenon is
hardly noticeable in Fig. 7II since its contribution to total
NO release is very small (Fig. 5I), but it can be appreciated
in the enlarged section displayed in the inset. To conclude,
the small amount of N2O detected in the gas phase during

G. Marbán et al. / Journal of Catalysis 226 (2004) 138–155
155
gen. The aminooxy groups are an active species in the SCR
References
reaction that occurs on the octahedral phase. Furthermore,
NH3 can be adsorbed on the hydroxyls (oxygen excess) of
the tetrahedral phase as ammonium ions, these being the ac-
tive SCR species on the tetrahedral phase.
[1] M. Baldi, E. Finocchio, Ch. Pistarino, G. Busca, Appl. Catal. A 173
(1998) 61.
[2] A.M. Maltha, T.L.H. Favrem, H.F. Kist, A.P. Zuur, V. Ponec,
J. Catal. 149 (1994) 364.
[3] S. Cimino, S. Colonna, S. De Rossi, M. Faticanti, L. Lisi, I. Pettiti,
P. Porta, J. Catal. 205 (2002) 309.
[4] T. Valdés-Solís, G. Marbán, A.B. Fuertes, Catal. Today 69 (2001) 259.
[5] F. Kapteijn, L. Singoredjo, M. van Driel, A. Andreïni, J.A. Moulijn,
G. Ramis, G. Busca, J. Catal. 150 (1994) 105.
[6] G. Marbán, A.B. Fuertes, Appl. Catal. B 34 (1) (2001) 43.
[7] G. Marbán, A.B. Fuertes, Appl. Catal. B 34 (1) (2001) 55.
[8] T. Valdés-Solís, G. Marbán, A.B. Fuertes, Appl. Catal. B 46 (2003)
261.
At 125 ◦C the SCR reaction takes place via an ER mech-
anism, in which NO2 (or the less reactive NO) reacts from
the gas phase with the surface active NH3 species. Two dif-
ferent SCR mechanisms are operative depending on the NH3
species involved:
5.1. SCR by aminooxy groups
This is a steady-state mechanism. At 125 ◦C it comprises
the following stages: (i) ammonia adsorption on oxygen
atoms, (ii) the formation of aminooxy groups, (iii) SCR re-
action between the aminooxy groups and the gas-phase NO
(in the absence of oxygen) or NO2 (∼ three times faster
than with NO) formed via nitrosyl oxidation with gas-phase
oxygen, (iv) dehydroxylation of the octahedral phase, and
(v) surface oxidation of the oxygen vacancies (nitrosyl re-
duction to N2). The overall SCR process is 6NO + 4NH3 →
5N2 + 6H2O.
[9] W.S. Kijlstra, D.S. Brands, E.K. Poels, A. Bliek, J. Catal. 171 (1997)
208.
[10] W.S. Kijlstra, D.S. Brands, H.I. Smit, E.K. Poels, A. Bliek, J. Ca-
tal. 171 (1997) 219.
[11] G. Marbán, T. Valdés-Solís, A.B. Fuertes, Phys. Chem. Chem.
Phys. 6 (2) (2004) 453.
[12] G. Marbán, A. Cuesta, Energy Fuels 15 (2001) 764.
[13] J. Muñiz, G. Marbán, A.B. Fuertes, Appl. Catal. B 23 (1999) 25.
[14] W.H. Lee, P.J. Reucroft, Carbon 37 (1) (1999) 21.
[15] F. Xie, J. Phillips, I.F. Silva, M.C. Palma, J.A. Menéndez, Carbon 38
(2000) 691.
´
[16] G.S. Szyman´ski, Z. Karpin´ski, S. Biniak, A. Swia˛tkowski, Carbon 40
(2000) 2627.
[17] A.R. Chughtai, M.M.O. Atteya, J. Kim, B.K. Konowalchuk, D.M.
Smith, Carbon 36 (11) (1998) 1573.
5.2. SCR by ammonium ions
[18] U.S. Ozkan, Y. Cai, M.W. Kumthekar, J. Catal. 149 (1994) 390.
[19] G.M. Robb, W. Zhang, P.G. Smirniotis, Micropor. Mesopor. Mater. 20
(1998) 307.
[20] S. Mrowec, in: C. Laird (Ed.), Materials Science Monographs, vol. 5,
Elsevier, Amsterdam, 1980, p. 75.
[21] M. Inomata, A. Miyamoto, Y. Murakami, J. Catal. 62 (1980) 140.
[22] F. Janssen, F. van den Kerkhof, H. Bosch, J.J. Ross, Phys. Chem. 91
(1987) 5931.
[23] F. Janssen, F. van den Kerkhof, H. Bosch, J.J. Ross, Phys. Chem. 91
(1987) 6633.
[24] J. Marangozis, Ind. Eng. Chem. Res. 31 (1992) 987.
[25] M. Anstrom, N.-Y. Topsøe, J.A. Dumesic, J. Catal. 213 (2003) 115.
[26] D.F. Shiver, P.W. Atkins, Inorganic Chemistry 3.0, http://www.chem.
ox.ac.uk/InorganicChemistry3/Mn/molecules/mn3o41.mol.
[27] G. Busca, L. Lietti, G. Ramis, F. Berti, Appl. Catal. B 18 (1998) 1.
[28] G. Marbán, R. Antuña, A.B. Fuertes, Appl. Catal. B 41 (2003) 323.
[29] G. Marbán, A.B. Fuertes, Catal. Lett. 84 (1) (2002) 13.
[30] U.S. Ozkan, Y. Cai, M.W. Kumthekar, J. Catal. 149 (1994) 375.
[31] A.M. Efstathiou, X.E. Verykios, Appl. Catal. A 151 (1997) 109.
[32] M.A. Larrubia, G. Ramis, G. Busca, Appl. Catal. B 30 (2001) 101.
[33] E.W.L. Chan, D.-Ch. Lee, M.-K. Ng, G. Wu, K.Y.C. Lee, L. Yu, J. Am.
Chem. Soc. 124 (41) (2002) 12238.
[34] E.W.L. Chan, L. Yu, Langmuir 18 (2002) 311.
[35] S. Liu, A. Obuchi, J. Uchisawa, T. Nanba, S. Kushiyama, Appl. Catal.
B 37 (2002) 309.
[36] G. Ramis, G. Busca, F. Bregani, F. Forzatti, Appl. Catal. 64 (1990)
259.
At 125 ◦C this is a pseudo-steady-state mechanism which
initially accounts for ∼ 60% of the total NO reduction but it
is gradually deactivated by the nitrates formed on the same
hydroxyl groups that are available for ammonium formation.
It comprises the following stages: (i) ammonium formation
on the hydroxyl groups (oxygen excess) of the tetrahedral
phase and (ii) SCR reaction between the ammonium ions and
the gas-phase NO2 formed as a result of nitrosyl oxidation
with gas-phase oxygen. The overall SCR process is 4NO +
4NH3 + O2 → 4N2 + 6H2O.
At higher temperatures the mechanism evolves toward
a limiting mechanism in which NO replaces NO2 as the
main reagent in the SCR reaction; reoxidation of the active
phase is performed by gas-phase oxygen instead of NO/O2,
whereas nitrate formation does not take place at all since
NO2 is no longer available.
Acknowledgments
The authors gratefully acknowledge financial support
from FEDER (1FD97-1636) and Hidroeléctrica del Can-
tábrico S.A; T. Valdés-Solís thanks the Government of the
Principado de Asturias for the award of a predoctoral grant
with funds from the Plan de Investigación, Desarrollo Tec-
nológico e Innovación de Asturias 2001–2004.
[37] R.Q. Long, R.T. Yang, J. Catal. 207 (2002) 274.
[38] R.Q. Long, R.T. Yang, J. Catal. 194 (2000) 80.
[39] R.Q. Long, R.T. Yang, J. Catal. 207 (2002) 224.