CO oxidation over a Pt(100) surface in a ternary gas mixture of CO, NO, and oxygen

Article abstract of DOI:10.1524/zpch.2009.6026

Catalytic CO oxidation in a ternary gas mixture of CO, NO and O₂ has been studied in the 10^-5-10^-4 mbar range using a Pt(100) sample as catalyst. Measurements of the reactive sticking coefficients of CO and NO are used to

Full text of DOI:10.1524/zpch.2009.6026

Z. Phys. Chem. 223 (2009) 89–104 . DOI 10.1524.zpch.2009.6026
© by Oldenbourg Wissenschaftsverlag, München
CO Oxidation over a Pt(100) Surface in a
Ternary Gas Mixture of CO, NO, and Oxygen
*
By R. Imbihl and E. Schütz
Institut für Physikalische Chemie und Elektrochemie, Universität Hannover, Callinstr. 3–3a,
30167 Hannover, Germany
Dedicated to Prof. Dr. Klaus Christmann on the occasion of his 65th birthday
(Received September 20, 2008; accepted October 22, 2008)
Catalysis . Carbon Monoxide . Nitrogen Monoxide . Oxygen Platinum
Catalytic CO oxidation in a ternary gas mixture of CO, NO and O2 has been studied in the 10–5
–10–4 mbar range using a Pt(100) sample as catalyst. Measurements of the reactive sticking
coefficients of CO and NO are used to decompose the overall CO oxidation rate into the
contributions from oxidation by NO and from oxidation by O2. The experiments are comple-
mented by simulations with a realistic 7-variable model. The surface phase transition hex 4
1!1 and the inhibition of NO dissociation and O2 adsorption at high adsorbate coverages lead
to significant hysteresis effects.
1. Introduction
One of the main challenges in designing better automotive catalytic converters
is to broaden the so-called λ-window, i.e. the range in air-to-fuel ratio, λ, where
both, the oxidation of CO and hydrocarbons, as well as the reduction of NOX,
proceed with high enough efficiency [1]. The simultaneous presence of oxidizing
and reducing reactions in the three-way-catalyst (TWC) reduces the usable range
of air-fuel ratios to a small window, the λ-window. As a simplified model system
for the TWC we consider here the a Pt(100) single crystal surface in a tenary
gas mixture of CO, NO, and O2. In this system essentially the reactions NO +
CO / CO2 + 1.2 N2 and CO + 1.2O2 / CO2 occur. Both systems, the NO +
CO reaction on Pt(100) and the CO + O2 reaction on Pt(100) have been exten-
sively studied because both systems exhibit interesting non-linear behavior in-
volving oscillatory reaction rates and chemical wave patterns [2–9]. The empha-
*
Corresponding author. E-mail: imbihl@pci.uni-hannover.de
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

90
R. Imbihl and E. Schütz
sis in this study, however, is not on nonlinear dynamics but on the two coexisting
reaction pathways which may compete but may also profit from another via
synergetic effects. The advantage of using these reactions is that both systems
can be considered as well understood as evidenced by realistic mathematical
models which describe rather well the experimentally observed behavior [5, 7].
The interesting nonlinear behavior of some catalytic reactions on Pt(100) is
largely due to the existence of an adsorbate-driven surface phase transition which
modulates the catalytic activity of Pt(100). The Pt(100) surface in its clean state
exhibits a quasi-hexagonal reconstruction (“hex”) of its topmost layer which can
be lifted reversibly upon adsorption of CO or NO [10–12]. This constitutes an
adsorbate-induced surface phase transition 1!1 4 hex which is controlled by
critical adsorbate coverages. A perfect hex reconstructed surface exhibits a very
low oxygen sticking coefficient and also a negligible efficiency in dissociating
NO [13, 14]. For the NO + CO reaction as well as for the O2 + CO reaction the
catalytically active phase is therefore the 1!1 phase. A structurally perfect hex
phase is inactive but the situation is different for the hex phase created by the
1!1 / hex phase transition without high temperature annealing. On such a
surface the NO reducing reactions are still quite fast and such a surface also
exhibits a much higher oxygen sticking coefficient than a perfect hex phase [7,
9, 13]. The remaining high activity of this surface is attributed to structural
defects in the hex phase created because the higher density of surface Pt atoms
in the hex phase as compared to the 1!1 phase requires the mass transport of
about 20% of the surface Pt atoms [12].
Under reaction conditions the surface phase transition shows up in a hystere-
sis of the reaction rate upon cyclic variation of a parameter. As will be shown
in this report in experiment and supported by mathematical modeling the pres-
ence of two reaction channels in combination with their nonlinear properties
leads to a number of nontrivial effects resulting from competition as well as
from synergetic interactions.
2. Experimental
The experiments are all conducted in a standard UHV systems equipped with a
photoemission electron microscope (PEEM), a four-grid LEED optics, a scan-
ning Auger electron spectrometer (SAES), and a differentially pumped quadru-
pole mass spectrometer (QMS) for rate measurements. Under reaction conditions
the system is continuously pumped by a turbomolecular pump with an effective
rate of app. 100 l.s. Gases are introduced via a feedback-stabilized inlet system.
For recording reaction rates the sample is brought in front of a cone with a
2 mm opening behind which the partial pressures are measured by a differentially
pumped QMS. Since only particles reflected from the sample surface can enter
the cone opening this method also allows a direct determination of the reactive
sticking coefficient of gases under reaction conditions. Denoting the signal of a
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

CO Oxidation over a Pt(100) Surface …
91
gas without reaction by I0 and during reaction with I we calculate the reactive
sticking coefficient sreac as
3. The mathematical model
For catalytic CO oxidation by NO and by O2 we can formulate the following
sequence of steps [2, 3]:
(* denotes a vacant adsorption site).
N2O formation has been neglected in this sequence because this product only
appears in larger quantities at higher pressure, above 1 mbar. The NO + CO
reaction is described by steps R1–R5 to which we have to add the dissociative
chemisorption of oxygen in step R6 in order that the reaction scheme also con-
tains the three steps of the LH scheme of catalytic CO oxidation (R1, R5 and
R6).
The mathematical model for the ternary system Pt(100).CO+NO+O2 is a
combination of two already existing models for Pt(100).CO + NO and Pt(100).
CO + O2 [5, 7]. The equations were only slightly modified in order to account
for the role of structural defects which are dynamically created by the phase
transition and which account for a substantial part of the catalytic activity in the
NO + CO reaction. The resulting 7-variable model describes the variation of the
CO and NO coverages on the two surface phases, the oxygen coverage on the
1!1 phase and the change in the substrate configuration caused by the 1!1 4
hex surface phase transition and by the creation and removal of structural defects.
For both reaction systems modified mathematical models have been proposed by
King et al. which differ essentially in the point that they use a power law depend-
ence for the kinetics of the hex / 1!1 surface phase transition [11, 16, 17]. In
an extended model which takes into account diffusion and reaction across the
1!1.hex phase boundary spatiotemporal pattern formation was treated in 2D-
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

92
R. Imbihl and E. Schütz
simulations [17]. Since the agreement with the experiment is not notably im-
proved by this modification here the original model is employed.
We have θ1!1 + θhex + θdef = 1 with θ1!1 and θhex denoting the fraction of
the surface which are in the 1!1 state and in the hex reconstructed state, respec-
tively, and θdef representing the amount of structural defects. We assume that the
structural defects are homogeneously distributed on the hex phase and that their
catalytic properties are identical to the 1!1 phase. Since we assume that their
catalytic properties are identical to the 1!1 phase we combine the two fractions
according to θdef + θ1!1 = θ‘1!1. The hex phase is catalytically inactive whereas
the 1!1 phase and the defects are active. We obtain:
(1)
(2)
(3)
(4)
(5)
(6)
(7)
with
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

CO Oxidation over a Pt(100) Surface …
93
For a detailed discussion of the equations we refer to refs. 5 and 7. In short, the
terms with k1 and k2 in Eq. 1 represent CO adsorption.desorption on the 1!1
phase, and the k4 term the reaction between CO ad oxygen (R5). The k3 term is
the so-called trapping term describing the unidirectional diffusion of CO and NO
adsorbed on the hex phase to 1!1 areas where they are trapped because of their
higher adsorption energy. If due to trapping and adsorption the local CO.NO
coverage in the 1!1 areas exceeds 0.5, then these islands grow thus maintaining
a constant local coverage (θgrow) and converting hex surface into 1!1 area.
As shown experimentally by their hysteretic behavior in adsorption.desorption
equilibrium CO and NO behave rather similar with respect to the surface phase
transition [7]. Eqs 2 and 4 describing CO and NO, respectively, on the hex phase
therefore have the same terms of adsorption, desorption, and trapping.
According to R3 NO adsorbed on the 1!1 phase can only dissociate (k8) if
an additional vacant site is available, i. e. if θempty > 0 [7]. Atomic oxygen on the
1!1 phase described by Eq. 5 therefore originates from two sources: adsorption
of O2 from the gas phase (k11) and NO dissociation (k8). Atomic oxygen is re-
moved either through reaction with CO in step R5 (k4 term) or, at high enough
temperature through thermal desorption (k12). As in earlier simulations we as-
sume that the inhibition of O2 adsorption by adsorbed CO is not perfect and that
a certain fraction of the surface, α, therefore remains active for O2 adsorption
despite a high CO coverage [5]. Atomic nitrogen formed by NO dissociation is
assumed to be bound only weakly and therefore the nitrogen will desorb immedi-
ately after formation.
The surface phase transition is treated in Eq. 6. The transition 1!1 / hex
proceeds when the local coverage in the 1!1 area falls below a critical coverage
necessary to stabilize the 1!1 phase. The transition in the reverse direction takes
place via the island growth mechanism discussed above in connection with CO.
NO trapping. The defects treated in Eq.7 are created as the surface undergoes
the phase transition and they are removed either via thermal annealing (k14) or
when the hex phase is converted back into the 1!1 phase. The proportionality
constant ꢀ determines how the rate of the phase transition 1!1 / hex affects
the amount of defects created. We can assume that a certain fraction of the
defects, θperm, is permanent and cannot be transformed into the hex phase. The
higher efficiency of defects in CO.NO trapping is denoted by teff. Since the
trapping efficiency will depend on the diffusion rate of CO and NO on the hex
phase and on the surface residence time of NO.CO, this constant should be
temperature dependent. In addition, teff will also vary depending on the spatial
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

94
R. Imbihl and E. Schütz
Table 1. Temperature independent constants of the model.
Constant
Reaction step
CO adsorption
NO adsorption
O2 adsorption
νi [mbar–1s–1]
2.21 ! 105
2.14 ! 105
4.15 ! 105
k1
k6
k11
The values of k1, k6, and k11 were calculated the expression for the impingement rate of gas
molecules from kinetic gas theory. The impingement rate λi is given by λi = pi.(2πmikT)1.2
with mi denoting the mass of the gas particle i. For the dissociative chemisorption in k11. an
additional factor of 2 has to be taken into account.
Table 2. Sticking coefficients.
hex
CO
1x1
hex
NO
1x1
Constant
Value
Ref.
s
s
s
s
sO1x1
CO
NO
2
0.78
11
0.91
16
0.91
7
0.91
7
0.3
13, 18
Table 3. Further temperature independent constants of the model.
inh
crit
Constant Erep
θOinh
θ
θOcrit
θ
θgrow
ꢀ
teff
θperm
α
CO, NO
CO, NO
Value
20 kcal. 0.34
mol
0.6
0.4
0.25
0.5 1!10–3 5!103 1!10–6 0.01
Refs.
fit to TDS 7, 16 7, 16 5, 13 5, 7, 10 5, 7, 10
fit fit fit fit
separation of the defects on the hex phase which again will change depending
on temperature. Since we do not know the diffusion rates on the hex phase and
since we also do not know the average distance of defects from one another we
neglect the temperature dependence of teff.
The constants in the simulation summarized in Table 1–3 and 4 were nearly
all identical with the constants used in earlier simulations of the separate reaction
systems. Only the trapping parameters and the constants concerning the defects
were fitted to experimental curves. The repulsion energy, Erep, between CO and
NO adsorbed on the 1!1 phase was determined by fitting thermal desorption
spectra of CO and NO. The adsorbate coverages in the Tables 1–3 which suffice
to completely suppress NO dissociation are denoted as θOinh for the oxygen cov-
erage and θNO, COinh for the combined CO and NO coverage. The critical covera-
ges of the adsorbates required to stabilize the 1!1 phase have been denoted as
crit
crit
θO for oxygen and θNO, CO for NO and CO.
4. Results
For quantifying the contributions from the two reaction channels
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

CO Oxidation over a Pt(100) Surface …
95
Table 4. Temperature-dependent constants of the model.
Constant
Reaction step
νj [s–1]
1!1015
1.6!106
2.0!109
3.7!1012
1!1016
Ej [kcal.mol]
Ref.
5, 10
fit
k2
CO desorption 1!1
CO trapping
36.8
8
k3
k4
CO2 formation
14
5
k5
CO desorption hex
NO desorption 1!1
NO dissociation
NO trapping
25.1
36.8
28.4
8
5, 10
7
k7
k8
2!1016
4!105
7
k9
7
k10
k12
k13
k14
NO desorption hex
O2 desorption
1.2!1013
3.0!1014
2.5!1011
2.0!109
25.1
52.1
25.3
31
7
5, 13
5
phase transition 1!1 / hex
defect annealing
fit
For describing the temperature dependence of constants we assume an Arrhenius depend-
ence ki = νi exp(–Ei.RT). The coverage dependence of the activation energy for CO and NO
desorption was approximated by Ead(θ) = Ead(0) – Erepθ2 where θ is the local combined cov-
erage of the molecular adsorbates with θ = (θCO + θNO).θ1!1and Erep is the repulsion energy
of the adsorbates.
we measure the reactive sticking coefficients of CO, sCO and NO, sNO, and de-
compose the reactive sticking coefficient of CO, sCO, according to the contribu-
tions of the two channels with sCO = sCO (NO) + sCO (O2) using sCO(NO) = sNO
.
Figure 1 displays the hysteresis we measure when a Pt(100) sample under-
goes a heating.cooling cycle in a CO + NO atmosphere. Instead of the reaction
rate, rCO2, we use here and in the following diagrams the reactive sticking coeffi-
cient of CO which is proportional to rCO2.The diagram in Fig. 1 shows a rate
hysteresis that is connected with the structural transformation 1!1 4 hex. The
hex spot intensity reflects directly the reconstructed fraction of the surface while
the intensity of the integral order spot is sensitive to the surface structure as well
as to the adsorbate coverages. Upon heating the reaction ignites at T = 440 K as
due to desorption vacant sites are created where NO can dissociate. As the cover-
age upon further heating falls below the critical coverage required for stabilizing
the 1!1 phase the surface starts to reconstruct into the hex phase which is the
case at 460 K. At 550 K the transformation into the hex phase is completed.
Upon cooling the surface remains in the inactive hex state until CO.NO adsorp-
tion below 500 K begins to lift the hex reconstruction.
We note that during heating the catalytic activity remains substantial between
550 K and 650 K despite the formation of the supposedly inert hex phase. The
upper rate branch in this T-range is metastable as one can easily demonstrate by
stopping the heating schedule in this range [15]. The rate then slowly returns to
the low rate branch. We therefore attribute the remaining activity in this T-range
to structural defects in the hex phase which are created as result of mass transport
in the 1!1 / hex phase transition. The defects are slowly removed by thermal
annealing thus explaining the metastable nature of this part of the rate branch.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

96
R. Imbihl and E. Schütz
Fig. 1. Hysteresis in the surface structure and in the catalytic activity upon cyclic variation of
the temperature of a Pt(100) sample in a NO.CO atmosphere. Shown are the hysteresis in the
LEED intensities of an integral and a fractional order beam monitoring the 1!1 4 hex phase
transition and in the CO2 production rate expressed here via the reactive sticking coefficient of
CO. The shaded bar indicates the T-range in which the upper rate branch of the heating cycle
is metastable. Heating.cooling rate 0.5 K.s.
In contrast to earlier measurements the catalytic activity on the well annealed
hex phase obtained after reaching a high temperature is in fact close to zero [7].
The difference to earlier measurements is explained by the method of measuring
the reactive sticking coefficient which excludes reaction from the backside of
the crystal.
The simulation displayed in Fig. 2 reproduces well the experimental data, in
particular, the connection of the phase transition with the rate hysteresis and the
remaining high activity on the heating branch beyond 500 K. One notes that at
low temperature the surface is almost completely covered by molecular NO.CO
adsorbate. Ignition of the reaction at 450 K is connected with a steep drop of the
coverage of the molecular adsorbates. Generated by NO dissociation simultane-
ously a substantial oxygen coverage develops. The structural defects created by
the phase transition are responsible for the high catalytic activity persisting even
after completion of the 1!1 / hex phase transition. The defects are removed
by thermal annealing, a process which becomes quite fast at higher temperature
thus limiting the temperature range in which defects can reach a substantial
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

CO Oxidation over a Pt(100) Surface …
97
Fig. 2. Simulation of the experiment in Fig. 1 of a heating.cooling cycle of Pt(100) surface in
a NO + CO atmosphere. Shown are the hysteresis in the combined NO and CO coverage and
the oxygen coverage on the 1!1 phase, the defect concentration, the amount of hex reconstruc-
tion, and the CO2 production rate. Extended black regions in the curves represent kinetic insta-
bilities. Heating.cooling rate ꢀ = 0.5 K.s.
population. The adsorbate coverage in this range of the heating branch from
540 K to 650 K is almost zero despite a high catalytic activity. Evidently the
defects have a very high efficiency of trapping NO and CO so that even with a
moderate defect concentration a large part of the CO and NO molecules imping-
ing on the surface can react to the products CO2 and N2. The stationary adsorbate
concentration remains low because NO dissociation at this high temperature is
fast and no longer rate-limiting. The dark regions on the cooling branch of the
reaction rate in Fig. 2 represent kinetic oscillations as described earlier [7].
When we keep the NO and CO partial pressures fixed but add oxygen with
nearly one order of magnitude larger oxygen partial pressure than pNO.pCO, we
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

98
R. Imbihl and E. Schütz
Fig. 3. Rate hysteresis of a Pt(100) sample in a ternary gas mixture of CO, NO and O2 during
a heating.cooling cycle with heating.cooling rate ꢀ = 0.5 K.s. The reaction rate of CO2
formation is expressed here via the total reactive sticking coefficient of CO, sCO(total), and the
contributions from oxidation via NO and O2 are represented by sCO(NO) and sCO(O2), respec-
tively. Experimental conditions: pNO = 4!10–6 mbar, pCO = 3!10–6 mbar, pO = 5.3!10–5
2
mbar.
obtain the rate curves of the ternary mixture reproduced in Fig. 3. The overall
reaction rate shown in the top section looks rather similar to that of the binary
mixture in Fig. 1 except some fine structure on the cooling branch of the binary
mixture around 450 K. The decomposition of the overall rate into the two reac-
tion channels, however, reveals that a sharp switching of the channels takes
place. The NO + CO channel is only dominant at the upper and at the lower
boundary of the reactive window while in between the CO + O2 channel takes
over control. The reaction rate of the NO + CO channel reaches almost zero at
550 K. The first switching from the NO + CO pathway to the CO + O2 channel
takes place during heating at 470 K; the second switching from CO + O2 back
to NO + CO occurs at 580 K.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

CO Oxidation over a Pt(100) Surface …
99
Fig. 4. Simulation of the rate hysteresis in catalytic CO oxidation over Pt(100) in a ternary gas
mixture of CO, NO and O2. The data are simulated with nearly the same parameters as used
in the experiment of Fig. 3. Shown are the defect concentration, θdef, the fraction of the surface
being present as hex phase, θhex, the CO2 production rate from the CO + O2 pathway, rO , the
CO2 production rate from the NO + CO pathway, rNO, and the total CO2 production rate,2rCO
The corresponding reactive sticking coefficients of CO for the different pathways of CO oxida-
tion are displayed on the right side of the diagram.
.
2
The simulation reproduced in Fig. 4 explains why this twofold switching of
the reaction channel takes place. The “surface explosion” of the NO + CO reac-
tion at 430 K creates a largely adsorbate free 1!1 surface where O2 can adsorb
uninhibitedly. Due to a nearly 10fold excess of oxygen in the gas phase which
overcompensates the much lower oxygen sticking coefficient of oxygen as com-
pared to NO, the CO + O2 channel takes control. The NO + CO channel gains
back control when the 1!1 / hex phase transition occurs at 570 K. The defects
on the hex phase are evidently not very active in the CO + O2 reaction while
due to trapping of both, NO and CO, the NO + CO reaction can still proceed
very efficiently on a hex reconstructed surface which has not been annealed.
Only when at high temperature the defect concentration becomes very small due
to thermal annealing the catalytic activity approaches zero. The defects on the
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

100
R. Imbihl and E. Schütz
Fig. 5. Variation of the overall reaction rate and of the different reaction channels in a ternary
mixture of CO, NO and O2 as pO2 is cycled. The hysteresis are probably not real but originate
from saturation of the chamber walls. Experimental conditions: T = 479 K, pNO = 4!10–6
mbar, pCO = 3!10–6 mbar, dp.dt = 9.3!10–7 mbar.s.
hex phase are therefore responsible for the switching from the CO + O2 pathway
to the NO + CO channel at 580 K. Since the defects on the hex phase presumably
exhibit an oxygen sticking coefficient comparable to that of the 1!1 phase one
should suspect that also the contribution from the CO + O2 channel remains
substantial in the T-range where the catalytic activity is controlled by defects.
The dominance of the NO + CO channel can only be explained by the very high
NO.CO trapping efficiency of defects. Compared to NO and CO the diffusivity
of atomic oxygen on Pt is much lower and consequently also the trapping effi-
ciency of oxygen will be much lower.
Upon cooling the activity remains very low until below 500 K CO.NO ad-
sorption starts to lift the hex reconstruction. Since the local adsorbate concentra-
tion in the 1!1 islands remains high dissociative O2 adsorption cannot compete
against the molecular adsorption of CO and NO. NO can still dissociate because
the inhibition coverage for NO dissociation, θinh, is with θinh = 0.6 above the
local adsorbate coverage in the growing 1!1 islands, θgrow which is 0.5 in the
simulation. Therefore the NO + CO channel dominates in this part of the T-
cycling experiment over the CO + O2 channel. When at even lower temperature
NO dissociation is finally suppressed by a high adsorbate coverage the surface
activity dies away.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

CO Oxidation over a Pt(100) Surface …
101
Fig. 6. Simulation of the experiment in Fig. 5 demonstrating the dependence of the overall
reaction rate and of the different reaction channels on the oxygen partial pressure. Shown are
the oxygen and NO coverage on the 1!1 phase, the total reactive sticking coefficient of CO
and the contribution of the NO + CO channel to the reactive sticking coefficient of CO. The
selectivity towards the NO + CO pathway is defined in the same way as in Fig. 5. Note that a
slightly higher temperature than the temperature in the experiment had to be used in order to
reproduce the experimentally observed behavior. Parameter values: T = 500 K, pNO = 4!10–6
mbar, pCO = 3!10–6 mbar, dp.dt = 5!10–7 mbar.s.
In order to systematically investigate the influence of O2 on the reaction the
oxygen partial pressure was cycled at fixed temperature T = 480 K and with pNO
and pCO being fixed at the values of the previous experiment. The resulting
diagram displayed in Fig. 5 shows an overall reaction rate that is nearly inde-
pendent of pO2. With increasing pO2 the contribution of the NO + CO channel
decreases monotonically from 100% at pO2 = 0 to only 10% when pO2 reaches
9!10–5 mbar; the corresponding CO + O2 fraction (not shown) rises from zero
to 90%. The simulation in Fig. 6 reproduces this behavior rather well. The con-
tinuous transition can be attributed to the absence of the phase transition. At
500 K with intermediate adsorbate coverages the 1!1 phase remains stable and
both reaction pathways are active in this coverage range. As a consequence, it is
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

102
R. Imbihl and E. Schütz
Fig. 7. Variation of the overall reaction rate and of the different reaction channels in a ternary
mixture with as pO2 is cycled. In comparison to the previous experiment in Fig. 5 CO is now
in excess compared to NO. Shown are the reactive sticking coefficient of oxygen, sO2, and the
different contributions to the reactive sticking coefficient of CO with sCO(NO) and sCO(O2),
representing the contributions from the NO + CO and the CO + O2 pathway, respectively.
Experimental conditions: T = 420 K, pNO = 2!10–6 mbar, pCO = 3!10–6 mbar, dp.dt =
9.3!10–7 mbar.s.
essentially the partial pressure ratio NO : O2 which dictates which channel is
preferred.
A qualitatively different behavior is found when CO in the CO + NO mixture
is in excess. For pNO : pCO = 2:3 and at lower temperature, at 420 K, one obtains
the diagram displayed in Fig. 7. A broad rate hysteresis is seen but in contrast
to the previous diagrams the hysteresis here is not caused by the phase transition
since the substrate remains in a 1!1 configuration all the time. Rather the hys-
teresis is caused by the inhibitory effects of large adsorbate coverages on the
reaction as demonstrated by the simulation shown in Fig. 8. At low pO2 NO
dissociation is blocked because on a surface fully packed with adsorbates no
vacant sites are available. With increasing pO initially the surface remains inac-
tive until at pO = 4.5!10–5 mbar the reac2tion ignites. In the simulation the
2
triggering is caused by the alpha-term in Eq. 5 supposedly representing surface
defects where adsorbed CO cannot inhibit O2 adsorption. The triggering of the
reaction lowers the adsorbate coverage far enough so that now both reaction
channels become active, the NO + CO and the CO + O2 pathway. Upon further
increasing pO the oxidation by O2 becomes more and more dominant. In the
direction of d2ecreasing pO2 the active state of the surface is maintained down to
5!10–6 mbar until the transition to an adsorbate poisoned surface occurs. The
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

CO Oxidation over a Pt(100) Surface …
103
Fig. 8. Simulation of the experiment in Fig. 7 showing the dependence of the overall reaction
rate and of the different reaction channels as pO is varied. Shown are the CO coverage on the
1!1 phase, the reactive sticking coefficient of 2oxygen, sO , and the different contributions to
the reactive sticking coefficient of CO with sCO(NO) and sCO2(O2), representing the contributions
from the NO + CO and the CO + O2 pathway, respectively. The total reactive sticking coeffi-
cient of CO is denoted as sCO(total). Parameter values: T = 418 K, pNO = 2!10–6 mbar, pCO
3!10–6 mbar, dp.dt = 2.5!10–8 mbar.s.
=
simulation in Fig. 8 which reproduces the experiment reveals that essentially the
build-up of a large CO coverage is responsible for the rate hysteresis.
5. Conclusions
The competition between two CO oxidation channels has been investigated with
a ternary reaction mixture of CO, NO and O2 and with a Pt(100) surface as
model catalyst. The adsorbate-induced surface phase transition and the inhibitory
effect of large adsorbate coverages on NO- dissociation and O2 adsorption deter-
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM

104
R. Imbihl and E. Schütz
mine the catalytic activity of the surface. Upon heating up the Pt(100) surface
the main reaction channel switches from oxidation by NO to oxidation by O2
and then back to oxidation to NO. The latter switching occurs because on a
freshly formed hex phase the catalytic activity in the NO + CO channel remains
high due to structural defects trapping CO and NO and dissociating NO. Syner-
getic effects between the two reaction channels have been observed as well.
Since both reaction channels are blocked by high adsorbate coverages ignition
of one reaction channel will automatically also start the reaction in the second
channel. The critical point for a further refinement of the model is clearly obtain-
ing a more detailed knowledge about the nature of the structural defects and their
spatial distribution.
References
1. E. S. Lox and B. H. Engler, Environmental Catalysis. Wiley-VCH, Weinheim
(1999).
2. R. Imbihl and G. Ertl, Chem. Rev. 95 (1995) 697.
3. R. Imbihl, in Chapter 9 in Handbook of Surface Science. Vol. 3, E. Hasselbrink and
B. Lundquist (Eds.), Elsevier, Amsterdam (2008).
4. R. Imbihl, M. P. Cox, and G. Ertl, J. Chem. Phys. 84 (1986) 3519.
5. R. Imbihl, M. P. Cox, G. Ertl, H. Müller, and W. Brenig, J. Chem. Phys. 83 (1985)
1578.
6. H. H. Rotermund, S. Jakubith, A. von Oertzen, and G. Ertl, J. Chem. Phys. 91, 4942
(1989).
7. T. Fink, J.-P. Dath, R. Imbihl, and G. Ertl, J. Chem. Phys. 95 (1991) 2109.
8. G. Veser and R. Imbihl, J. Chem. Phys. 96 (1992) 7155.
9. J. H. Miners, P. Gardner, J. Phys. Chem. B 104 (2000) 10265.
10. R. J. Behm, P. A. Thiel, P. R. Norton, and G. Ertl, J. Chem. Phys. 78 (1983) I 7438;
II (1983) 7448.
11. A. Hopkinson, J. M. Bradley, X.-C. Guo, and D. A. King, Phys. Rev. Lett. 71 (1993)
1597.
12. A. Borg, A.-M. Hilmen, and E. Bergene, Surf. Sci. 306 (1994) 10; P. van Beurden,
B. S. Bunnik, G. J. Kramer, and A. Borg, Phys. Rev. Lett. 90 (2003) 66106.
13. K. Griffiths, T. E. Jackman, J. A. Davies, and P. R. Norton, Surf. Sci. 138 (1984)
113; Guo, X.-C. , Bradley, J. M. Hopkinson, et al. , Surf. Sci. 310 (1994) 163.
14. R. J. Gorte, L. D. Schmidt, and J. L. Gland, Surf. Sci. 109 (1981) 367; J. M.
Gohndrone and R. I. Masel, Surf. Sci. 209 (1989) 44.
15. E. Schütz and R. Imbihl, Z. Phys. Chem. 216 (2002) 509.
16. A. Hopkinson, D. A. King, Chem. Phys. 177 (1993) 433.
17. R. B. Hoyle, A. T. Anghel, M. R. E. Proctor, and D. A. King, Phys. Rev. Lett. 98
(2007) 226102.
18. A. T. Pasteur, X.-C. Guo, T. Ali, M. Gruyters, and D. A. King, Surf. Sci. 166 (1996)
564.
Brought to you by | New York University Bobst Library Technical Services
Authenticated
Download Date | 12/18/16 7:37 PM