Mean field modeling of NO oxidation over Pt/Al2O3 catalyst under oxygen-rich conditions

Article abstract of DOI:10.1016/j.jcat.2004.11.029

This paper deals with catalytic NO oxidation into NO₂ over Pt/Al₂O₃ in oxygen-rich exhaust. Based upon kinetic measurements performed in a gradient free loop reactor and DRIFTS examinations, a mean field model is established to simulate catalytic data and calculate surface coverages. The kinetic model includes a network of 16 elementary reactions whereby NO oxidation is described by an Eley-Rideal-type mechanism. A comparison of measured and simulated data shows that the experimental results of the NO/O₂ reaction are well described by the model. Kinetic parameters for the elementary reactions were taken from the literature or determined from a fit of the model to experimental data. To reduce the number of free parameters, we estimated the activation energy of O desorption for zero coverage and the corresponding linear constant α2 by numerically simulating the O₂ TPD pattern. For the kinetic model of NO oxidation, the preexponential factors for NO₂ adsorption (68 m³ s^-1 m^-2) and NO oxidation (104 m³ s ^-1 m^-2); the activation energies for NO desorption (114 kJ mol^-1), NO₂ desorption (72 kJ mol^-1), NO oxidation (35 kJ mol^-1), and NO₂ dissociation (51 kJ mol^-1); and the linear constant α13 (14 kJ mol^-1) are fitted. Furthermore, the calculated surface coverages provide evidence for the effect of CO, CO₂, and H₂O on the rate of NO oxidation.

Full text of DOI:10.1016/j.jcat.2004.11.029

Journal of Catalysis 229 (2005) 480–489
www.elsevier.com/locate/jcat
Mean field modeling of NO oxidation over Pt/Al O catalyst
2
3
under oxygen-rich conditions
∗
Marcus Crocoll, Sven Kureti , Werner Weisweiler
Institut für Technische Chemie und Polymerchemie, Universität Karlsruhe (TH), Kaiserstr. 12, D-76128 Karlsruhe, Germany
Received 16 July 2004; revised 27 October 2004; accepted 24 November 2004
Available online 12 January 2005
Abstract
This paper deals with catalytic NO oxidation into NO over Pt/Al O in oxygen-rich exhaust. Based upon kinetic measurements performed
2
2 3
in a gradient free loop reactor and DRIFTS examinations, a mean field model is established to simulate catalytic data and calculate surface
coverages. The kinetic model includes a network of 16 elementary reactions whereby NO oxidation is described by an Eley–Rideal-type
mechanism. A comparison of measured and simulated data shows that the experimental results of the NO/O reaction are well described by
2
the model. Kinetic parameters for the elementary reactions were taken from the literature or determined from a fit of the model to experimental
data. To reduce the number of free parameters, we estimated the activation energy of O desorption for zero coverage and the corresponding
linear constant α by numerically simulating the O TPD pattern. For the kinetic model of NO oxidation, the preexponential factors for
2
2
3
−1 −2
3
−1 −2 −1
); the activation energies for NO desorption (114 kJ mol ), NO
2
NO adsorption (68 m s
m
) and NO oxidation (104 m s
m
2
−1
−1
−1
−1
desorption (72 kJ mol ), NO oxidation (35 kJ mol ), and NO dissociation (51 kJ mol ); and the linear constant α (14 kJ mol ) are
2
13
fitted. Furthermore, the calculated surface coverages provide evidence for the effect of CO, CO , and H O on the rate of NO oxidation.
2
2

2004 Elsevier Inc. All rights reserved.
Keywords: Kinetics; Mean field modeling; Mechanism; Catalytic oxidation; Nitric oxide; Nitrogen dioxide; Platinum; Diesel exhaust
1
. Introduction
in a large excess. Hence, new catalytic procedures were de-
veloped to convert NO under oxidizing conditions [1]. Cur-
rently, selective catalytic reduction (SCR) and NOx storage
catalyst (NSC) are the most favored techniques for this ap-
plication. To reduce the emissions of soot, the use of diesel
particulate filters (DPF) is taken into account. Such filter sys-
tems are already being applied in passenger cars and buses,
for instance, as part of CRT (Continuously Regenerating
Trap) systems.
The exhaust after-treatment procedures mentioned above
include the oxidation of NO into NO2 over platinum as a key
reaction. In SCR, nitrogen oxides are continuously reduced
by ammonia over TiO2-supported WO3/V2O5 catalysts [2],
whereby the NO reduction rate is substantially accelerated
in the presence of NO2 [3]. Nitrogen dioxide is produced
here with the use of a Pt/Al2O3 pre-catalyst. The NOx stor-
age catalysts contain platinum and basic adsorbents, such as
Diesel engines with direct fuel injection exhibit the high-
est efficiency for automotive applications. As a consequence,
the low fuel consumption of diesel vehicles leads to reduced
emissions of the greenhouse gas carbon dioxide (CO2).
However, a serious constraint of diesel engines is the produc-
tion of the pollutants nitric oxide (NO) and soot. Three-way
catalyst (TWC) technology is not suitable for the removal
of NO from diesel exhaust [1]. TWC systems have been
successfully applied to the simultaneous conversion of NO,
hydrocarbons (HC), and carbon monoxide (CO) in gasoline
cars since the 1980s. In the exhaust gas of diesel engines
only insufficient NO conversion takes places, as the reduc-
ing agents HC and CO react mainly with oxygen present
*
2 3
Ba(OH) /BaCO . Nitric oxide is catalytically oxidized on
Corresponding author. Fax: 0049-721-608-2816.
E-mail address: Kureti@ict.uni-karlsruhe.de (S. Kureti).
the Pt component, and then the formed NO2 is stored by the
0021-9517/$ – see front matter  2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.jcat.2004.11.029

M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
481
adsorbents. When the storage capacity is reached, NSC is
regenerated by engine management, resulting in the conver-
sion of the chemisorbed nitrogen oxides into nitrogen [4–6].
The CRT technique is based on mechanical filtration and
subsequent oxidation of the soot with the use of NO2 [7]. In
the first processing step the soot is separated by the DPF sys-
tems, and in the second step the oxidation of the deposited
soot is initiated by NO2 [8]. In this reaction, soot is mainly
converted into CO2, and NO2 is reduced to NO. In CRT,
NO2 is also formed over a Pt/Al2O3 pre-catalyst.
was established to be 2 wt%, whereas for DRIFTS analy-
sis a catalyst with a Pt content of 10 wt% was also used.
The loading of platinum was confirmed by AAS (AAS 4100,
Perkin–Elmer).
2.2. Catalyst characterization
The catalyst that contained 2 wt% Pt was characterized
by N2 physisorption, CO chemisorption, scanning electron
microscopy (SEM), and powder X-ray diffraction (PXRD).
N2 physisorption was carried out by with a Sorptomatic
Despite the technical importance of the Pt-catalyzed
NO/O2 reaction under oxygen-rich conditions, very little is
available on the details of the mechanism or kinetic model-
ing of this reaction in the literature. Burch and co-workers
◦
1990 from Porotec. The sample was pretreated at 400 C for
−4
◦
12 h in vacuum (3 × 10 mbar) and cooled to −196 C,
and then the N2 isotherm was recorded. Based upon these
data, the BET surface area, total pore volume, and mean
BJH pore diameter were determined. For CO chemisorption
also performed on the Sorptomatic 1990, the sample was
reduced in pure H2 at 300 C for 90 min and then evacu-
ated overnight in vacuum (10 mbar). After the sample was
cooled to 40 C, the CO adsorption isotherm was measured.
[
9] have shown that NO oxidation can be described with an
Eley–Rideal-type mechanism. They assume the oxygen to
be the dominating species on the active Pt sites, whereas NO
reacts from the gas phase. This mechanism was confirmed
by Olsson et al. [10], Chatterjee [11], and Deutschmann [12].
The latter authors [10–12] have studied the kinetics of NO
oxidation over Pt in detail and defined a complex network of
the respective elementary reactions.
The interaction between oxygen and active Pt surface un-
der atmospheric pressure conditions was examined by Ols-
son and co-workers [10] and by Zhdanov et al. [13]. The
rates of adsorption and desorption of oxygen were defined
with an Arrhenius-type equation, whereby a dissociative ad-
sorption of O2 is postulated. Furthermore, the activation en-
ergy for oxygen desorption was assumed to depend on the
oxygen coverage [14].
◦
−7
◦
The chemisorption of CO was carried out to evaluate both
the active Pt surface area and the dispersion of Pt. SEM pic-
tures (LEO Gemini) were taken to determine the size of the
Pt particles. PXRD analysis was conducted on a D501 from
Siemens with Ni filtered Cu-Kα radiation. A 2θ step size
was employed with an integration time of 3 s. The sample
was measured with a rotation frequency of 2 Hz.
2.3. Kinetic studies
The aim of this work has been to study the mechanism
and kinetics of catalytic NO oxidation over Pt in a realistic
diesel model exhaust. Based on the experimental data, a ki-
netic mean field model, including all elementary reactions,
was established to simulate catalytic results and calculate
surface coverages.
Kinetic studies were performed with a gradient free loop
reactor with an external gas cycle. The catalyst (1 g) was
fixed in a quartz glass tube (i.d. 11 mm) by a glass frit.
The feed gas flowed downward and was a blend of pure
components and synthetic air (Messer Griesheim). The lat-
ter was added at the reactor inlet to avoid NO oxidation in
◦
the stainless-steel lines that were heated to 150 C. All of
the gases were fed from independent flow controllers (MKS
Instruments). Water was dosed to the gas stream with a liq-
uid flow controller from Bronkhorst. The temperature was
monitored by two K-type thermocouples located directly in
front of and behind the catalyst bed. The total volume flow
2
. Experimental
2
.1. Catalyst preparation
3
−1
The catalyst used in this study was a Pt/Al2O3 sys-
was 400 cm min with a recycle ratio φ (φ = Floop/Fout)
◦
tem. We prepared the sample by coating γ -Al2O3 balls
of 80. The temperature was increased in steps of 25 C from
◦
(
(
d = 0.61 mm, Sasol) with an aqueous solution of Pt(NO3)2
150 to 500 C. The reactor effluents were analyzed after they
ChemPur). For impregnation an incipient wetness method
reached steady state. NOx was monitored with a chemilu-
minescence detector (CLD El-ht, Eco Physics), and N2O,
CO, and CO2 were measured by NDIR (Ultramat 5E from
Siemens for N2O; Binos 5 for CO and Binos 4b.1 for CO2,
both from Leybold Heraeus). Oxygen was detected with
a magnetomechanic analyzator (Magnos 6G, Hartmann &
Braun). To investigate the effect of O2, CO, CO2, and H2O
on the kinetics of the NO oxidation, several gas mixtures
were used (Table 1). Before the measurements were made,
was used, that is, a defined volume of the solution was taken
such that it was completely absorbed. With this technique the
load of platinum could be calculated in a simple way. The
coated alumina was dried at 20 C for 24 h and subsequent-
ly at 65 C for 2 h. After this, a gas mixture consisting of
◦
◦
3
−1
1
0 vol% H2 and 90 vol% N2 (1000 cm min ) was passed
over the sample to reduce the Pt precursor. The catalyst was
heated at a rate of 1.7 K min to 300 C and was held at
this temperature for 30 min. The sample was conditioned at
5
−
1
◦
◦
the catalyst was pretreated in nitrogen flow at 500 C to re-
◦
00 C for 2 h in air. For kinetic studies and TPD the Pt load
move possible impurities.

482
M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
Table 1
Composition of the feed gas used in the kinetic studies
in a N2 atmosphere after the same pretreatment as described
above. The DRIFTS data are only displayed in the region
from 1200 to 1900 cm , because at lower wavenumbers a
relatively high noise was observed that was associated with
strong lattice vibrations. At higher frequencies no features
of species adsorbed on Pt were detected.
−1
Mixture
NO
ppm)
CO
(ppm)
CO
(vol%)
H O
2
(vol%)
O
(vol%)
2
2
(
1
500
1.5, 3.0,
6.0, 15
2
3
4
5
500
500
500
500
500
500
6.0
6.0
6.0
6.0
6.0
6.0
6.0
6.0
3
. Results and discussion
Based upon the N2 isotherm, the BET surface of the
Mass transfer limitations were estimated by Mears [15]
and Weisz criteria [15,16]; the estimations indicated that nei-
ther film diffusion nor pore diffusion affects the catalytic
oxidation of NO or CO. The effect of pore diffusion is also
experimentally excluded. In these measurements different
particle sizes of the Pt/Al2O3 catalyst are used, whereby the
realistic diesel model exhaust gas is fed. The results show
that the same reaction rate for CO and NO oxidation is ob-
tained for the different catalyst sizes.
2
−1
Pt/Al2O3 catalyst was calculated to be 175 m g , and total
3
−1
pore volume was 0.70 cm g . The mean BJH pore diame-
ter was determined to be 8.6 nm. The active Pt surface area
(
Aact) and the dispersion of Pt [17,18] were deduced from
the CO adsorption isotherm with the use of Eqs. (1) and (2).
The CO:Pt stoichiometry (n) is assumed to be 0.7 [19],
2
whereas the surface area per Pt site (am) is taken as 8.07 Å
2
−1
[
17]. Eq. (1) leads to an active Pt surface of 5.3 m g , and
Pt
the dispersion of Pt is calculated to be 2.1% (Eq. (2)).
Temperature-programmed desorption (TPD) of O2 was
performed with the reactor system described above after the
external gas loop was separated. The catalyst (5.0 g) was
VadsNAnam
aPt =
× 100,
(1)
(2)
Vmmw
a M
◦
◦
heated in a N2 flow at 700 C for 1 h, cooled to 350 C, and
then treated with a gas mixture that consisted of 2 vol% O2
and 98 vol% N2. After saturation, the catalyst was flushed
with N2, and then TPD was started, with nitrogen as a car-
Pt
DPt =
× 100.
amNA
The denotations used in Eqs. (1) and (2) are as follows: Vads,
volume of chemisorbed CO; Vm, molar volume; NA, Avo-
gadro’s number; m, mass of catalyst; w, weight percentage
of Pt. The SEM micrograph shows that the mean size of the
Pt particles is about 200 nm. Furthermore, the PXRD data
prove the existence of crystalline Pt, which is indicated by
the respective (111) and (200) reflexes.
The effect of the O2 concentration on the kinetics of
the NO oxidation over the Pt/Al2O3 catalyst is displayed in
Fig. 1. The conversion of NO into NO2 rises in the whole
temperature range if the concentration of O2 is increased
from 1.5 to 6.0 vol%. However, a further increase in the O2
3
−1
rier gas (500 cm min ). In TPD the sample was heated to
7
◦ −1
00 C at a rate of 14 K min . Here O2 was analyzed by
CIMS (Airsense 500, V & F).
The surface compounds formed from reaction of Pt with
NO, NO2, CO, and CO2 were studied by vibrational spec-
troscopy. The analyses were performed on an ATI Matt-
son Galaxy 5020 FTIR spectrometer equipped with a MCT
detector (Thermo Mattson) and DRIFTS optics (Graseby
Specac). The stainless-steel IR cell contained a ZnSe win-
dow and was attached to a gas handling system. The spectra
−
1
were recorded in the range of 500 to 4000 cm with an in-
−1
strument resolution of 4 cm . Fifteen thousand scans were
accumulated for a spectrum. The temperature of the sample
was monitored with a K-type thermocouple placed ca. 3 mm
beneath the sample surface. Before the spectra were col-
lected, the catalyst was ground to powder in an agate mortar
for several minutes, charged in the sample holder, and then
heated for 5 h at 500 C in a nitrogen flow of 500 cm min
We exposed the catalyst to the reactive gases at 30 C by
passing a mixture (20 cm min ) of 500 ppm NO in N2 or
00 ppm NO2 in N2 through the cell. A dose of a mixture
◦
3
−1
.
◦
3
−1
5
that consisted of 500 ppm CO, 2 vol% CO2 and N2 as bal-
ance was also passed through the cell. We did not expose the
catalyst to O2 because respective bands are expected in the
spectral region of the strong lattice vibrations of the Al2O3
support. After 20 min of adsorption the cell was purged with
nitrogen, and then the analysis was started. The unreacted
catalyst was taken as a reference for the spectra (presented
in Kubelka–Munk units). Background scans were conducted
Fig. 1. Effect of O on the concentration of NO ((F) 1.5 vol% O ,
2
2
2
(
1) 3.0 vol% O , (Q) 6.0 vol% O ) in the oxidation of NO over the Pt/
2 2
Al O catalyst. With 6.0 and 15 vol% O the same results are obtained. In
2
3
2
all the measurements gas mixture 1 was used (Table 1).

M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
483
Fig. 2. Effect of CO (500 ppm) on the concentration of NO (1) and NO
Fig. 4. Concentration of NO (1) and NO (2) in the oxidation of NO over
2
2
(
2) in the oxidation of NO over the Pt/Al O catalyst. In this measurement
the Pt/Al O catalyst in the realistic diesel exhaust. In this measurement
2
3
2 3
gas mixture 2 was used (Table 1).
gas mixture 5 was used (Table 1).
Fig. 5. O TPD data of the Pt/Al O catalyst. The saturation with O was
2
2
3
2
Fig. 3. Effect of H O (6.0 vol%) on the concentration of NO (1) and NO
◦
carried out at 350 C.
2
2
(
2) in the oxidation of NO over the Pt/Al O catalyst. In this measurement
2 3
gas mixture 4 was used (Table 1).
the formation of N2O, a potential product on thermodynamic
grounds, is not observed.
concentration to 15 vol% does not result in a higher yield of
NO2. Concentrations of O2 exceeding 15 vol% are not estab-
lished because of a relatively high blank of NO2 associated
with NO oxidation in the gas phase. With 500 ppm CO in the
The profile of O2 TPD is presented in Fig. 5. The pattern
◦
exhibits a maximum at about 487 C (55 ppm) and a shoul-
◦
der at ca. 590 C (30 ppm). The first TPD peak is assigned
to the oxygen desorption from defect sites (e.g., steps), and
the latter one is associated with oxygen bonded to Pt terrace
sites [10,37].
Preliminary DRIFTS investigations show qualitatively
the same spectra for the catalysts that exhibit 2 and 10 wt%
Pt. As the latter sample provides a better signal-to-noise
ratio, this catalyst is employed in the following IR exami-
nations. The DRIFT spectrum recorded after NO exposure
◦
feed, the NO/O2 reaction is inhibited below 350 C (Fig. 2).
In this measurement no oscillations of the concentration of
the product CO2 are observed as the molar ratio of CO to O2
is very small, amounting to 0.008. From the literature it is
known that the oscillatory region starts from a CO/O2 ratio
of about 0.015 [20,21].
In presence of 6.0 vol% CO2 the rate of NO2 forma-
tion does not change (not depicted). As compared with CO,
a weaker decrease in NO conversion is observed when H2O
−
1
exhibits a very prominent band at 1714 cm and two less
−1
intense bands located at 1764 and 1787 cm (Fig. 6). These
features are associated with the stretching vibration of NO,
which is linearly coordinated by its N atom to Pt parti-
cles of different sizes [22–25]. The weak peaks observed at
(
6.0 vol%) is supplied (Fig. 3). With dosing of the realistic
diesel model exhaust that contains all of the gas compo-
nents, a stronger decrease in NO oxidation rate is obtained
than in the case where only CO is fed (Fig. 4). In all of
the kinetic measurements the outlet concentration of NOx
is always found to be 500 ppm. Hence, we conclude that
no NOx reduction takes place in these studies. Furthermore,
−
1
wavenumbers between 1200 and 1650 cm are referred to
nitrite and nitrato species originating from the reaction with
the Al O support [26]. The exposure to nitric oxide is also
2
3
◦
carried out at 175 and 200 C to examine the adsorption of

484
M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
Fig. 6. DRIFT spectrum of the Pt/Al O catalyst (10 wt% Pt) exposed to
NO at 30 C.
Fig. 8. DRIFT spectrum of the Pt/Al2O3 catalyst (10 wt% Pt) exposed to
CO and CO2 at 30 C.
2
3
◦
◦
elementary reactions is defined. In Eqs. (3)–(8), an asterisk
(
∗) labels a free Pt site and ri the respective reaction rate.
Reactions (3)–(8) represent the adsorption/desorption equi-
librium of each component, whereby dissociative adsorption
of O2 is included according to methods described in the lit-
erature [10,13].
r1
O2(g) + 2∗ ꢀ2O∗,
(3)
(4)
r2
r3
NO(g) + ∗ ꢀNO∗,
r4
r5
NO2(g) + ∗ ꢀNO2∗,
(5)
r6
r7
Fig. 7. DRIFT spectrum of the Pt/Al O catalyst (10 wt% Pt) exposed to
NO at 175 C (bottom) and 200 C (top).
2
3
CO(g) + ∗ ꢀCO∗,
(6)
◦
◦
r
8
r9
◦
CO2(g) + ∗ ꢀCO2∗,
(7)
NO on Pt. From the DRIFT spectrum recorded at 175 C
Fig. 7) it is obvious that the same type of NO species is
produced on the Pt surface as in the case of adsorption per-
r10
(
r11
H O + ∗ ꢀH O∗,
2
2
(8)
(g)
◦
◦
r12
formed at 30 C. However, at 200 C respective bands are
no longer observed in the spectrum. We therefore conclude
that the existence of NO bonded to Pt sites is rather un-
likely above 200 C. Furthermore, for NO2 adsorption the
same qualitative results are obtained as for NO exposure
r13
NO(g) + O∗ ꢀNO2∗,
(9)
r
14
◦
r
15
CO∗ + O∗ꢀCO2∗.
(10)
r16
(
not depicted). The absence of NO2 adsorbed on Pt sites is
also in agreement with the literature and is associated with
its fast desorption at ambient temperature [27]. After expo-
sure to the CO/CO2 mixture, an intense band is observed at
Reactions (9) and (10) show the reversible oxidation of
NO and CO, respectively. The formation of N2 and N2O
is neglected in the reaction network, since neither species
is formed in the kinetic experiments. For oxidation of NO
the Eley–Rideal-type mechanism is postulated (Eq. (9)), in-
cluding the reaction of dissociatively adsorbed oxygen with
gas-phase NO. This mechanism is supported by the litera-
ture [9] and by our kinetic and spectroscopic studies. The
kinetic results indicate first an increase and then a constancy
in the NO oxidation rate with increasing O2 concentration,
and in DRIFTS no further bands associated with NO ad-
− −1
1
2
092 cm with a shoulder at 2050 cm (Fig. 8). These fea-
tures are attributed to the stretching vibration of CO species
linearly bonded to different Pt sites [24,28,29]. Furthermore,
−1
a weak signal is found at 1849 cm , which is interpreted as
ν(CO) vibration of CO molecules bridging two Pt sites. The
bands located at lower wavenumbers are associated with car-
bonato surface complexes [28].
To describe the mechanism of the NO oxidation in the re-
alistic model exhaust, a network of 6 surface species and 16
◦
sorbed on Pt are observed above 200 C. For CO oxidation,

M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
485
ꢀ
ꢁ
E16(0) + α16θO
a Langmuir–Hinshelwood-type mechanism is implemented,
r16 = A16 exp −
θCO θ∗.
(26)
2
RT
as reported by Ertl and co-workers [30]. Furthermore, it must
be noted that in our mean field model the different types
of active Pt sites (e.g., terraces and steps) are supposed to
be equivalent. For every forward and backward reaction an
Arrhenius-based rate expression is defined (Eq. (11) to (26)),
where Ai is the preexponential factor, Ei is the activation
energy, Ei(0) is the activation energy at zero coverage, ci is
the gas-phase concentration of species i, θi is the respective
coverage, and θ∗ is the number of free Pt sites. Because of
repulsion of adsorbed species, a linear decrease in activation
energy with increasing oxygen coverage is assumed for the
desorption of O, NO, and NO2 (Eqs. (12), (14), (16)) [10,
For the modeling of the NO/O reaction and the calculation
of the surface coverage, some kinetic parameters are deter-
2
mined from literature sources and some are calculated in
order to reduce the number of free parameters in the esti-
mation procedure. The numeric simulation is based upon a
combination of the mass balance of each gas phase (Eq. (27))
and adsorbed species (Eq. (28)), leading to a system of six
algebraic and six nonlinear differential equations. The re-
spective surface coverages depending on the temperature are
then calculated with Matlab tool ods15s. The free parame-
ters, that is, activation energies and preexponential factors,
are estimated with a nonlinear regression method (Matlab
tool Isqcurvefit)
1
2]. For this purpose the constant α is introduced. A similar
relationship is used for desorption and oxidation of CO [31],
where activation energy decreases with the growth of CO
coverage (Eqs. (18) and (25)). However, oxygen is known to
inhibit the dissociation of CO2 (Eq. (26)), and hence it must
be taken into account that the activation energy increases
when oxygen coverage grows [12].
Nj
ꢂ
˙
˙
V ci,in − V ci,out + Aact
νij rij = 0,
(27)
(28)
j
Nj
ꢂ
dθ
AactΓcat
= Aact
νij rij .
ꢀ
ꢁ
dt
E1
2
j
r1 = A1 exp −
cO2(g)θ ,
(11)
(12)
(13)
(14)
(15)
(16)
(17)
(18)
(19)
(20)
(21)
(22)
(23)
(24)
(25)
∗
RT
The preexponential factor for the adsorption of NO, O2, CO,
CO2, and H2O is calculated with Eq. (29), and the corre-
sponding sticking coefficient for zero coverage (S ) is taken
from the literature [12,31–33]. Γ Pt denotes the Pt site den-
sity and is simply calculated from CO chemisorption data,
that is, aPt, n, and the CO monolayer volume. In Eq. (29)
we neglect the temperature dependence by using an average
ꢀ
ꢁ
E2(0)(1 − α2θO)
2
r2 = A2 exp −
θ ,
O
0
RT
ꢀ
ꢁ
E3
r3 = A3 exp −
cNO(g)θ∗,
RT
ꢀ
ꢁ
E4(0) − α4θO
r4 = A4 exp −
θNO,
RT
◦
temperature of 325 C
ꢀ
ꢁ
E5
NART
2πMiRT )1/2
r5 = A5 exp −
cNO2(g)θ∗,
0
Ai =
amΓPtS .
(29)
RT
(
ꢀ
ꢁ
E6(0)(1 − α6θO)
Furthermore, we neglect the activation energy for the ad-
r6 = A6 exp −
θNO2 ,
RT
sorption of all of the gaseous species, E1 = E3 = E5 =
ꢀ
ꢁ
−1
E7
E = E = E = 0 kJ mol , which is in good correspon-
7
9
11
r7 = A7 exp −
cCO(g)θ∗,
dence with the literature [12,33–35]. The preexponentialfac-
RT
ꢀ
ꢁ
tor of O2 desorption (A2) is taken from the literature [13,34].
For NO desorption the preexponential factor (A ) and cor-
E8(0) − α8θCO
r8 = A8 exp −
θCO,
4
RT
rection factor of the activation energy (α4) are assumed
ꢀ
r9 = A9 exp −
ꢀ
ꢁ
E9
16 −1 −1
[12,34,35] and 10 kJ mol [12], respec-
to be 10
s
cCO2(g)θ∗,
RT
tively. Moreover, the preexponential factor of NO desorp-
2
ꢁ
1
3
−1
tion (A6 = 10
s ) and α6 (0.075) are from other authors
E10
r10 = A10 exp −
θCO2 ,
[10,12,36]. The preexponential factor of NO decomposi-
2
RT
E11
RT
E12
RT
ꢀ
ꢁ
tion (A ), that is, the backward reaction of NO oxidation, is
supposed to be 10
CO desorption, A8 (10
(33 kJ mol ), are taken completely from Zerkle et al. [31].
For the activation energy of CO oxidation (E15), values in
1
4
1
3
−1
s [10,12]. The kinetic parameters of
r11 = A11 exp −
cH O(g)θ∗,
2
16 −1
−1
s
), E8(0) (146 kJ mol ), and α8
ꢀ
ꢁ
−1
r12 = A12 exp −
θH O,
2
ꢀ
r13 = A13 exp −
ꢀ
ꢁ
−1
the range of 100 kJ mol are reported in literature [12,30,
31,37]. In our kinetic model we have used the value deter-
E13(0) − α13θO
cNO(g) θO,
RT
ꢁ
−
1
mined by Deutschmann (108 kJ mol ), with α15 amounting
E14
−1
9
−1
to 33 kJ mol , and A15 is 10 s [12]. The activation
energy of the CO decomposition (E ) is from Campbell
r14 = A14 exp −
ꢀ
θNO2 ,
RT
2
16
ꢁ
−
1
E15(0) − α15θCO
et al., who calculated the value as 155 kJ mol [30,37]. The
further kinetic parameters of this reaction, α16 (45 kJ mol
r15 = A15 exp −
θCOθO,
−1
)
RT

486
M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
and A16 (1013
s
−1
), are also taken from Deutschmann [12].
The preexponential factors of the desorption of CO2 (A10)
and H2O (A12) are shown to be in the range of 1013
s
−1
[
12,31], and the corresponding activation energies are found
−1
to be 27 (E10) and 40 kJ mol (E12), respectively.
To obtain independent kinetic parameters for the adsorp-
tion and desorption of oxygen (Eq. (3)), we numerically sim-
ulated the spectrum of O2 TPD (Fig. 5). The interaction of
oxygen with the active Pt sites is expected to be a dominating
process in the NO/O2 reaction. The simulation is performed
in the same way as NO oxidation, whereby the mass bal-
ance of the gas phase and adsorbed oxygen (Eqs. (30) and
(
31)) results in a system of one algebraic (Eq. (32)) and one
nonlinear differential equation (Eq. (33)).
Fig. 9. Calculated surface coverage of oxygen in O TPD of the Pt/Al O
2 3
catalyst.
2
˙
˙
V cO ,in − V cO2(g),out − Aactr1 + 2Aactr2 = 0,
(30)
(31)
2
(g)
dθO
AactΓcat
= 2Aactr1 − 2Aactr2,
dt
E2(0)(1−α2θO)
2
AactA2 exp(−
V + AactA1 exp(− )(1 − θ0)
)θ
RT
O
cO2(g)
=
,
(32)
E
1
RT
˙
2
ꢀ
ꢁ
dθ
E1
2
AactΓcatβ
= 2AactA1 exp −
cO2(g)(1 − θ0)
dT
RT
ꢀ
ꢁ
E2(0)(1 − α2θO)
2
−
2AactA2 exp −
θ .
O
RT
(
33)
In the estimation procedure the parameters A1, E1, and
A2 are kept fixed, and E2(0) and α2 are fitted. The fit leads to
−1
an activation energy for the O2 desorption of 200 kJ mol
,
whereby α2 is determined to be 0.10 (Table 2). These re-
sults are consistent with data found in the literature. E2(0)
obtained from experiments on a Pt(111) single crystal sur-
face [38] and polycrystalline Pt [34] is reported to be 213
−1
Fig. 10. Comparison of measured (—) and fitted O TPD data (—) of the
and 200 kJ mol , respectively. Furthermore, Olsson et al.
10,34] have reported a value of 0.1 for α2. The calculated O
2
Pt/Al O catalyst.
2
3
[
coverage depending on temperature is displayed in Fig. 9.
The coverage is relatively high at the beginning of TPD
(
0.75) but decreases continuously with increasing temper-
◦
ature. At 700 C only 1% of the active Pt site is covered
with oxygen. Fig. 10 proves that the TPD curve is reason-
ably well fitted by the kinetic data implemented in the O2
adsorption/desorption model. As a result of the mean field
model, the TPD pattern is approximated by one peak only.
The kinetic parameters used for the modeling of the NO
oxidation are summarized in Table 2. In the estimation pro-
cedure these parameters are kept fixed, whereas the pre-
exponential factors for the NO2 adsorption (A5) and NO
oxidation (A13); the activation energies for NO desorption
(
E4(0)), NO2 desorption (E6(0)), NO oxidation (E13(0)),
and NO2 dissociation (E14); and the linear constant α13 are
fitted. The modeled results, including the respective 95%
confidence interval, are also presented in Table 2. Fig. 11
shows the comparison of the measured and simulated con-
centrations of NO and NO2. It is evident that the simu-
lated results correspond well with the experimental data, and
Fig. 11. Comparison of measured and fitted concentration of NO ((1) mea-
sured, (—) fitted) and NO ((2) measured, (—) fitted) in NO oxidation over
2
the Pt/Al O catalyst in the diesel model exhaust. In this measurement gas
2
3
mixture 5 was used (Table 1).

M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
487
Table 2
Kinetic parameters of the NO oxidation over the Pt/Al O catalyst
2
3
Reaction
Parameter
Value
95% confidence
interval
Reference
0
Units
Adsorption
3
−1 −2
m
O
+ 2∗ → 2O∗
A
1
11
0
138
0
68
0
141
0
0.67
0
157
0
S
= 0.07 [10,44]
(m s
)
)
)
)
)
)
2(g)
−1
E
1
A
3
E
3
A
5
E
5
[10,33]
0
(kJ mol
)
3
−1 −2
NO + ∗ → NO∗
S
= 0.85 [10]
(m s
m
(
g)
−
m
1
[10,33]
This work
(kJ mol
)
3
−1 −2
NO
+ ∗ → NO2∗
± 31
(m s
2(g)
−
m
1
[10,33]
0
(kJ mol
)
3
−1 −2
CO + ∗ → CO∗
A7
E7
S
= 0.84 [28]
(m s
(
g)
−
m
1
[10,33]
0
(kJ mol
)
3
−1 −2
CO
+ ∗ → CO2∗
A
9
S
= 0.005 [10,27]
(m s
2(g)
−
1
E
A
E
[10,33]
0
(kJ mol
)
9
3
−1 −2
H O + ∗ → H O∗
S
= 0.75 [10,27]
(m s
m
2
(g)
2
11
−1
[10,33]
(kJ mol
)
11
Desorption
O∗ → O2(g) + 2∗
−1
(kJ mol
(–)
−2
m
)
2
A
2 × 1010
200
[9,11]
(mol s
)
)
)
)
2
−1
E (0)
± 1
This work [45]
This work [45]
[9,31]
This work [45]
[10]
[8,10,32]
This work [41]
[8]
[27]
[27]
[27]
[10]
2
α
0.10
± 0.08
2
1
8
1
1
−1 −2
NO∗ → NO(g) + ∗
A
4
2 × 10
(mol s
m
−
1
E (0)
114
± 0.5
± 2
(kJ mol
(kJ mol
)
)
4
−
1
α
10
4
−1 −2
NO ∗ → NO
+ ∗
A
6
2 × 10
(mol s
(kJ mol
(–)
m
2
2(g)
−
1
E (0)
72
)
6
α
0.075
2 × 10
146
6
1
−1 −2
CO∗ → CO(g) + ∗
A
8
(mol s
m
−
1
E (0)
(kJ mol
(kJ mol
)
)
8
−
1
α
33
8
8
−1 −2
CO ∗ → CO
+ ∗
A
E
A
E
2 × 10
(mol s
m
)
)
2
2(g)
10
−
1
27
[10]
[10]
[10]
(kJ mol
)
1
0
8
−1 −2
H O∗ → H O + ∗
2 × 10
40
(mol s
m
2
2
(g)
12
−
1
(kJ mol
)
1
2
Surface reactions
3
−1 −2
m )
NO + O∗ → NO2∗
A
E
α
104
35
14
± 68
± 2
This work [45]
This work [45]
This work
[8,9]
This work
[10]
[10,26,33]
[10]
[10]
(m s
(
g)
13
−1
(0)
13
(kJ mol
)
± 3
13
8
4
−1 −2
m
NO ∗ → NO + O∗
A
E
A
E
2 × 10
(mol s
(kJ mol
(mol s
(kJ mol
(kJ mol
(mol s
(kJ mol
(kJ mol
)
)
2
(g)
14
−1
51
± 1
)
1
4
−1 −2
CO∗ + O∗ → CO ∗ + ∗
4 × 10
108
33
2 × 10
155
45
m
)
)
2
15
−
1
1
(0)
(0)
1
5
−
α
1
5
8
−1 −2
CO ∗ + ∗ → CO∗ + O∗
A
E
m
)
2
16
−
1
1
[26,33]
[10]
)
)
1
6
−
α
1
6
hence we conclude that our kinetic model (Eqs. (3) to (15))
describes satisfactorily NO oxidation over a Pt catalyst in
diesel model exhaust. The maximum difference between
simulated and experimental results is about 10%.
crystals. Grote and co-workers report that E4(0) is found to
−1
−1
be 105 kJ mol for a Pt(111) surface and 151 kJ mol for
a Pt(100) surface [35], whereas Morris et al. report a value of
−1
114 kJ mol for a Pt(111) surface [39]. Furthermore, Serri
et al. have reported an activation energy for desorption of
In the estimation procedure the preexponential factor of
3
−1 −2
−1
the NO2 adsorption (A5) is determined to be 68 m s
m
.
NO from Pt steps of 142 kJ mol [40]. A similar result
−1
For comparison, A5 is also calculated according to ki-
(145 kJ mol ) has been determined by Altmann for poly-
crystalline Pt [41]. These results agree well with our fitted
netic gas theory (Eq. (29)), leading to an average value of
3
−1 −2
−1
1
18 m s
m
, which is somewhat higher than the fitted
value (114 kJ mol ).
−1
result.
In this work we estimate a value of 72 kJ mol for the ac-
The activation energy for NO desorption (E4(0)) has
been investigated by several groups that used platinum single
tivation energy of the NO2 desorption E6(0). Bartram et al.
[42] have found E6(0) to be 80 kJ mol for a Pt(111) sur-
−1

488
M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
Fig. 12. Calculated surface coverage of the components of the diesel model
exhaust present in NO oxidation over the Pt/Al O catalyst. In the mea-
Fig. 13. Comparison of thermodynamic (—) and measured (2) concentra-
tion of NO in NO oxidation over the Pt/Al O catalyst. In this measure-
2
3
2
2 3
surement gas mixture 5 was used (Table 1).
ment the diesel model exhaust was used (mixture 5, Table 1).
face, and in presence of oxygen it is lowered to 46 kJ mol 1
−
.
affect the NO oxidation rate. The reason for this observation
−8
Wickham et al. [43] studied the NO2 desorption energy on
polycrystalline Pt and report to a result ranging from 59 to
5 kJ mol . From this comparison we deduce that our value
is in good agreement with the data available in the literature.
For the forward reaction of the NO oxidation, Olsson and
co-workers [10] have estimated an activation energy (E13)
is the extremely low CO2 coverage (< 10 %) indicated by
the surface calculation. This result is also supported by the
DRIFTS data, which show no bands of CO2 adsorbed to Pt
(Fig. 8). In contrast to CO2, CO leads to a clear decrease in
−1
7
◦
the oxidation rate, especially at temperatures below 250 C
(Fig. 2). In this temperature range CO surface species com-
pete strongly with oxygen, resulting in reduced O coverage
and deceleration of the NO oxidation, respectively. The high
carbon monoxide coverage is principally confirmed by the
DRIFTS results (Fig. 8), which indicate a very intense band
of CO coordinated with Pt. A much lower inhibition of the
−1
of 41 kJ mol , and for the backward reaction the activa-
−1
tion energy (E14) is 142 kJ mol . Deutschmann [12] has
presented results that differ significantly from those pub-
lished by Olsson. He has determined E13 and E14 to be
8 and 99 kJ mol , respectively. Our modeled result for
E13(0) (35 kJ mol ) agrees well with Olsson’s, whereas
E14 (51 kJ mol ) corresponds to the data estimated by
Deutschmann. The fitted linear constant α13 was found to
be 14 kJ mol . For comparison, Deutschmann’s simulation
−
1
9
−1
NO/O reaction is caused by water (Fig. 3), as the H O cov-
2
2
−
1
erage is about four magnitudes smaller than for CO.
The NO concentration recorded in diesel model gas is
2
−
1
compared with the thermodynamic concentration (Fig. 13)
−1
leads to a value of 45 kJ mol
.
calculated with HSC Chemistry Software from Outokumpu
Research. The formation of N and N O is neglected, as
Furthermore, the preexponential factor for NO oxida-
2
2
3
−1 −2
m ,
tion (A13), which we have estimated to be 104 m s
these species are not observed in the measurements (Fig. 4).
3
differs clearly from the values given by Olsson (3.36 m
The experimental data show that thermodynamic NO con-
2
−1
−2
3
−1 −2
◦
s
m
) and Deutschmann (0.38 m s
m
).
centrations are obtained above 375 C. At lower tempera-
The result of the calculation of the surface coverages in
tures the oxidation of NO is kinetically limited.
the realistic model exhaust is presented in Fig. 12. The data
show that the Pt surface is dominated by CO and O species in
◦
the whole temperature range. Up to 185 C CO is the most
4. Conclusion
prominent surface species, and at higher temperatures CO
◦
coverage decreases continuously. At 500 C only 0.1% of the
The simulations show that the experimental data for cat-
alytic NO oxidation over Pt/Al2O3 are well fitted by our
postulated reaction network. The network consists of eight
surface reactions, including forward and backward reactions.
In addition to reversible adsorption of all of the components
present in the feed, the NO/O and CO/O surface reaction is
implemented in the model. DRIFTS and kinetic experiments
indicate an Eley–Rideal-type mechanism for NO oxidation
involving dissociative adsorption of O2 and reaction of NO
from gas phase. To reduce the number of free kinetic para-
meters in the fitting procedure, E2(0) and α2 are determined
by numerical simulation of O2 TPD. For description of the
NO oxidation, A5, A13, E4(0), E6(0), E13(0), E14, and α13
Pt sites is covered by CO. In contrast, the amount of the O
surface species increases with rising temperature. At 185 C
oxygen becomes the dominant surface species, and above
00 C 99% of the Pt sites are covered by O. The cover-
age of NO amounts initially to 7% and decreases drastically
with temperature. In contrast to this, NO2 and H2O adsorb
to the Pt surface only to a small extent. For NO2 the cover-
◦
◦
3
−7
−6
age ranges from 10 to 10 , whereas for H2O it is about
−5
−8
.
1
0
. Furthermore, CO2 shows values smaller than 10
The calculation of the surface coverages provides evi-
dence for the effect of CO, CO2, and H2O on the rate of NO
oxidation. The kinetic experiments show that CO2 does not

M. Crocoll et al. / Journal of Catalysis 229 (2005) 480–489
489
are estimated, and other kinetic data are taken from the liter-
[21] V.P. Zhdanov, B. Kasemo, Surf. Sci. Rep. 20 (1994) 111.
[22] V.K. Agrawal, M. Trenary, Surf. Sci. 259 (1991) 116.
ature. Furthermore, the calculated surface coverages confirm
the rather low activity of the Pt/Al2O3 catalyst in the pres-
ence of CO and H2O.
[23] A. Bourane, O. Dulaurent, S. Salasc, C. Sarda, C. Bouly, D. Bianchi,
J. Catal. 204 (2001) 77.
[24] P.J. Levy, V. Pitchon, V. Perrichon, M. Primet, M. Chevrier, C. Gau-
thier, J. Catal. 178 (1998) 363.
[25] M. Primet, J.M. Basset, E. Garbowski, M.V. Mathieu, J. Am. Chem.
Soc. 97 (1975) 3655.
References
[
[
[
26] N. Apostolescu, T. Schröder, S. Kureti, Appl. Catal. B 51 (2004) 43.
27] K.I. Hadjiivanov, Catal. Rev.-Sci. Eng. 42 (2000) 71.
28] A.A. Davydov, Infrared Spectroscopy of Adsorbed Species on the Sur-
face of Transition Metal Oxides, Wiley, New York, 1990.
29] K. Nakamoto, Infrared and Raman Spectra of Inorganic and Coordi-
nation Componds, Wiley, New York, 1997.
[
[
1] P. Zelenka, W. Cartellieri, P. Herzog, Appl. Catal. B 10 (1996) 3.
2] W. Weisweiler, A. Burkardt, J.A.A. van den Tillart, A. Schäfer-Sind-
linger, E.S. Lox, Top. Catal. 1–4 (2001) 369.
[
[
[
[
[
[
[
[
[
3] W. Weisweiler, C. Walz, Chem. Ing. Tech. 74 (2002) 117.
4] M. Guyon, P. Blanche, C. Bert, I. Philippe, I. Messaoudi, SAE Paper
30] C.T. Campbell, G. Ertl, H. Kuipers, J. Segner, J. Chem. Phys. 73
2
000-01-2910.
5] L. Olsson, E. Fridell, M. Skoglundh, B. Andersson, Catal. Today 72
2002) 263.
(
1980) 5862.
31] D.K. Zerkle, M.D. Allendorf, M. Wolf, O. Deutschmann, J. Catal. 196
2000) 18.
32] C.T. Campbell, G. Ertl, H. Kuipers, J. Segner, Surf. Sci. 107 (1981)
07.
33] W.H. Weinberg, R.M. Lambert, C.M. Comrie, J.W. Linnett, Surf.
Sci. 30 (1972) 299.
34] L. Olsson, H. Persson, E. Fridell, M. Skoglundh, B. Andersson,
J. Phys. Chem. B 105 (2001) 6895.
[
(
(
[
[
[
6] L. Olsson, B. Andersson, Top. Catal. 28 (2004) 89.
7] B.J. Cooper, J.E. Thoss, SAE Paper 890404.
8] P. Ehrburger, J.F. Brilhac, Y. Drouillot, V. Logie, P. Gilot, SAE Paper
2
2002-01-1683.
[9] R. Burch, P. Fornasiero, T.C. Watling, J. Catal. 176 (1998) 204.
[10] L. Olsson, B. Westerberg, H. Persson, E. Fridell, M. Skoglundh, B. An-
dersson, J. Phys. Chem. B 103 (1999) 10433.
35] R.J. Grote, L.D. Schmidt, J.L. Gland, Surf. Sci. 109 (1981) 367.
[
[
11] D. Chatterjee, Ph.D. thesis, University of Heidelberg, 2001.
12] O. Deutschmann, Habilitation thesis, University of Heidelberg, 2001,
http://www.rz.uni-karlsruhe.de/~cf01/.
[36] M.E. Bartram, R.G. Windham, B.E. Koel, Langmuir 4 (1988) 240.
[37] K. Christmann, Introduction to Surface Physical Chemistry, Springer,
New York, 1991.
[38] C.T. Campbell, G. Ertl, H. Kuipers, J. Segner, Surf. Sci. 107 (1981)
220.
[39] M.A. Morris, M. Bowker, D.A. King, C.H. Bamford, Comprehen-
sive Chemical Kinetics – Simple Processes at the Gas–Solid Interface,
vol. 12, Elsevier, Amsterdam, 1984.
[40] J.A. Serri, J.C. Tully, M.J. Cardillo, J. Chem. Phys. 79 (1983) 1530.
[41] E.I. Altmann, R.J. Grote, J. Phys. Chem. 93 (1989) 1993.
[42] M.E. Bartram, R.G. Windham, B.E. Koel, Surf. Sci. 184 (1987) 57.
[43] D.T. Wickham, B.A. Banse, B.E. Koel, Surf. Sci. 223 (1989) 82.
[44] Y.Y. Yeo, L. Vattuone, D.A. King, J. Chem. Phys. 106 (1997) 392.
[45] M. Crocoll, Ph.D. thesis, University of Karlsruhe, 2003.
[13] V.P. Zhdanov, B. Kasemo, Surf. Sci. 415 (1998) 403.
[14] D.H. Parker, M.E. Bartram, B.E. Koel, Surf. Sci. 217 (1989) 489.
[15] D.E. Mears, Chem. Proc. Design Develop. 10 (1971) 541.
[16] M. Baerns, H. Hofmann, A. Renken, Chemische Reaktionstechnik,
Thieme, Stuttgart, 1992.
[
[
[
[
17] G. Bergeret, P. Gallezot, in: Handbook of Heterogeneous Catalysis,
vol. 4, Wiley, New York, 1997.
18] J.R. Anderson, Structure of Metallic Catalysts, Academic Press, Lon-
don, 1975.
19] P. Loeoef, B. Kasemo, S. Andersson, A. Frestad, J. Catal. 130 (1991)
181.
20] L.F. Razon, R.A. Schmitz, Catal. Rev.-Sci. Eng. 28 (1986) 89.