A Real-Time Vibrational Spectroscopic Investigation of the Low-Temperature Oscillatory Regime of the Reaction of NO with CO on Pt{100}

Article abstract of DOI:10.1021/jp036487q

A new experimental method for obtaining real-time in situ IR reflection-absorption spectra of any reproducibly reversible process was applied to reveal the time-dependent coverages of adsorbates during the low-temperature oscillatory regime of the NO + CO reaction on Pt{100}. NO and CO were present at estimated coverages varying between 0.06 and 0.20 monolayers (ML) for CO and between 0.27 and 0.36 ML for NO at different points in the cycle. The relative phases of these two coverage oscillations differ very significantly. CO was present in both atop and bridging geometries, but the two species exhibit identical adsorption and reaction behavior, attributed to the high rate of equilibration between the two configurations. The results showed strong similarities to the predictions of the well-established model of Fink et al., but there were also notable differences.

Full text of DOI:10.1021/jp036487q

1708
J. Phys. Chem. B 2004, 108, 1708-1718
A Real-Time Vibrational Spectroscopic Investigation of the Low-Temperature Oscillatory
Regime of the Reaction of NO with CO on Pt{100}
J. H. Miners, P. Gardner, A. M. Bradshaw, and D. P. Woodruff*,§
†
‡
†,|
Fritz-Haber-Institut der Max-Plank-Gesellschaft, Faradayweg 4-6, D-14195 Berlin, Germany,
Department of Chemistry, Faraday Building, The UniVersity of Manchester Institute of Science and
Technology, Manchester M60 1QD, U.K., and Department of Physics, The UniVersity of Warwick,
CoVentry CV4 7AL, U.K.
ReceiVed: August 21, 2003
A new experimental method for obtaining real-time in situ infrared reflection-absorption spectra of any
reproducibly reversible process has been applied to reveal the time-dependent coverages of adsorbates during
the low-temperature oscillatory regime of the NO + CO reaction on Pt{100}. Simultaneous acquisition of
real-time mass spectrometry data is used to relate the coverages to the rates of reaction, providing new insight
into the mechanisms involved in the oscillation. NO and CO are shown to be present at estimated coverages
varying between 0.06 and 0.20 monolayers (ML) for CO and between 0.27 and 0.36 ML for NO at different
points in the cycle. The relative phases of these two coverage oscillations differ very significantly. CO is
present in both atop and bridging geometries, but the two species exhibit identical adsorption and reaction
behavior, attributed to the high rate of equilibration between the two configurations. Overall, the results show
strong similarities to the predictions of the well-established model of Fink et al. (J. Chem. Phys. 1991, 95,
2109), but there are also notable differences. The results indicate a significant role for the nonreactive
displacement of NO by CO not included in this model. Some of the other discrepancies may relate to
unavoidable surface inhomogeneity associated with reaction front propagation at key points in the oscillatory
cycle such as the “surface explosion” involving coadsorbed NO and CO. The coexistence of nonreacting CO
poisoned islands are also found to occur at the lowest temperatures.
8
1
. Introduction
Kinetic oscillations in the catalytic oxidation of CO were
discovered more than 30 years ago in experiments performed
{100} single crystal to 1 bar on a polycrystalline platinum
7
surface (for which the active face is {100}). The exact
temperature at which kinetic oscillations occur depends on the
total pressure of the reactants. The two regimes are, however,
separated by approximately 20 K and are thus generally referred
to as the high- and the low-temperature oscillatory regimes.9
Low-energy electron diffraction (LEED) measurements have
shown that the rate oscillations in the high-temperature oscil-
latory regime occur on a partially reconstructed surface, with
the transformation between the reconstructed “hex” surface
phase (the equilibrium clean surface structure at room temper-
ature in which the surface layer adopts a hexagonal close-packed
arrangement) and a nominally ideally terminated (1 × 1) surface
playing an important role in the bistability, whereas the surface
is entirely in the (1 × 1) phase in the low-temperature oscillatory
1-3
over polycrystalline Pt surfaces at atmospheric pressure. More
recent studies, in particular on the CO + O2 and CO + NO
reactions, have focused on single-crystal surfaces, where the
full arsenal of modern surface science techniques can be
applied. Single-crystal surface studies at low pressure (Ptot <
1
example, the adsorbed species can be expected to occupy
specific positions in a well-defined two-dimensional array and
one has very good control over the external parameters of
temperature and partial and total pressure and can ensure
homogeneity of all these parameters. The work of the Ertl group,
in particular, has revealed a variety of phenomena known from
other nonlinear systems, such as the transition to chaos, chemical
4,5
-
4
0
mbar) also offer some important simplifications. For
9
regime.
Fink et al. have modeled the reaction mathematically and have
reproduced the steady-state behavior and the lower temperature
kinetic oscillations, by assuming adsorbate island formation and
a vacant site requirement for NO dissociation. The individual
steps in the mechanism are shown schematically in Figure 1.9
Briefly, these are as follows: In step I, the surface is composed
of intermixed NO and CO having a high coverage which inhibits
NO dissociation due to lack of suitable vacant sites. In this state,
a chance desorption event leads to the creation of a vacant site,
allowing an adsorbed NO to dissociate, forming atomic O and
atomic N; the atomic O reacts with a CO molecule to form
CO2 and the atomic N combines with another N to form N2.
Both of these products desorb, thus creating more vacant surface
sites upon which NO can dissociate. The resulting exponential
waves, and the formation of Turing structures and of complex
spatio-temporal patterns.6,7
In the NO + CO reaction on Pt{100}, which we discuss here,
it is now well established that, using a mixture of NO and CO
with a partial pressure ratio in the range 0.8 > PNO-PCO >
1
.8, there are two distinct temperature regimes which exhibit
oscillatory behavior. Such reaction rate oscillations have been
-
9
observed for total pressures ranging from 10 mbar on a Pt-
*
Corresponding author. E-mail: D.P.Woodruff@Warwick.ac.uk.
Fritz-Haber-Institut der Max-Plank-Gesellschaft.
The University of Manchester Institute of Science and Technology.
The University of Warwick.
†
‡
§
|
Present address: Max-Planck-Insititut f u¨ r Plasmaphysik, Boltz-
mannstrasse 2, D-85748 Garching, Germany.
1
0.1021/jp036487q CCC: $27.50 © 2004 American Chemical Society
Published on Web 01/08/2004

Real-Time Vibrational Spectroscopic Investigation
J. Phys. Chem. B, Vol. 108, No. 5, 2004 1709
Figure 1. Schematic diagram of mechanism for the low-temperature NO + CO oscillatory reaction on Pt{100} proposed by Fink et al.9
creation of vacant surface sites leads to the so-called surface
explosion (step II), which rapidly reduces the surface coverage
by a factor of 2. Recently, Tammaro and Evans have proposed
a spatio-temporal model for the development of the surface
explosion and its passage, as a chemical wave, through the
adsorbate layer (steps I and II).10 In step III, any NO adsorbing
on the clean areas of the surface immediately dissociates and
the atomic nitrogen produced desorbs almost immediately as
N2, but because these experiments are conducted under condi-
tions in which the partial pressure of NO is greater than that of
CO, the oxygen titrates away any excess of adsorbed CO. As
the CO is removed, the surface oxygen coverage increases,
eventually reaching a state in which NO dissociation is hindered,
and the NO coverage increases again (step IV). The surface
oxygen is then removed by reaction with impinging CO
molecules (step V), allowing adsorption of NO and CO from
the gas phase at the free sites created to regenerate the mixed
NO + CO islands. Notice that this regeneration is almost
certainly aided by a separate process, proposed by Hopkinson
provide information on the nature and chemical environment
of the molecular species involved.
In a separate publication, we have presented IRAS and TPR
(temperature-programmed reaction spectroscopy) measurements
of the temperature dependence of NO and CO coadsorption
behavior on Pt{100} hex and shown that it is determined by
1
2
three processes, which also define the parameter space of the
two oscillatory regimes. At temperatures below those at which
NO dissociation (and thus reaction) occurs (below about 390-
400 K), the adsorption behavior is principally determined by
the nonreactive displacement of preadsorbed NO by CO,
attributed to the 0.23 eV greater heat of adsorption of CO.13
This displacement initially leads to the formation of mixed NO
+ CO islands but eventually produces an equilibrium state
comprising a CO-saturated surface (and thus a surface which
is CO-poisoned from the point of view of the CO/NO reaction).
At 390 K, and partial pressures of NO and CO equivalent to
those under which chemical oscillations are observed, exposure
to CO reduces the preabsorbed NO coverage by 50% in 30 s.12
Above 390-400 K, NO can dissociate (providing that there are
free sites available for the dissociation products), and the
resulting reaction of the atomic oxygen with CO leads to
desorption of CO2 and N2 and thus to vacant site creation.
Vacant sites are preferentially filled by NO because ((sNOpNO)/
(sCOpCO)) > 1 under the conditions of the experiment, where s
is the sticking coefficient and p is the partial pressure. The
reaction therefore results in an increase in NO coverage. It is
the competition between the nonreactive displacement of NO
by CO and the reactive replacement of CO by NO which
determines the parameter space of the low-temperature oscil-
latory regime and results in the formation of mixed, reacting
islands of NO + CO. Coincidentally, these coadsorption
1
1
and King, in which the CO also displaces NO nonreactively
due to its greater heat of adsorption). Although this nonreactive
(
displacement is not included in the modeling equations used
by Fink et al., the driving force for the cyclic behavior in both
models is considered to be the surface explosion and the
subsequent availability of vacant sites for NO dissociation. The
mathematical model of Hopkinson and King1 predicts es-
sentially the same behavior in the lower temperature existence
region but not in the upper one, where the reaction mechanism
is very similar to that of the oscillatory CO + O2 reaction on
the same surface.
1
These models, based on specific assumptions regarding the
reaction mechanisms, do appear to provide a reasonable
description of the kinetics as measured experimentally by
changes in the gas partial pressures. However, the more detailed
predictions of these models, which relate directly to the
underlying mechanisms, namely, the nature and concentration
of the species adsorbed on the surface during the reaction, have
not been challenged by experimental measurements. This
absence of experimental data stems, of course, from the
difficulty of applying many standard surface science techniques
at the high partial pressures of the reactants, which are present
during the reaction. In this regard, in situ optical vibrational
spectroscopy (photon-in, photon-out) offers the possibility of
overcoming this limitation. By use of infrared reflection-
absorption spectroscopy (IRAS), the intensities, spectral fre-
quencies (wavenumbers), and widths of the vibrational bands
1
2
studies show that both bridge and atop CO species exist on
the surface in dynamic equilibrium and their rapid rate of
interconversion leads to the two species exhibiting identical
behavior in terms of coverage changes.
While these measurements under steady-state conditions
provide valuable insight into the mechanisms which underlie
the CO/NO reaction over Pt{100}, a full understanding of the
processes occurring within the oscillatory reaction requires truly
time-dependent measurements which interrelate the adsorbate
coverage and the reaction rate. The challenge to achieve this
on single-crystal surfaces is to obtain sufficient sensitivity
combined with a temporal resolution significantly shorter than
the period of oscillation of the reaction. In principle, the rapid-
scanning facility of modern Fourier transform IR spectrometers

1
710 J. Phys. Chem. B, Vol. 108, No. 5, 2004
Miners et al.
can provide sufficient temporal resolution to follow an oscil-
latory cycle of the order of seconds, since a single scan at 2
extensive study of the coadsorption behavior of NO and CO
on Pt{100} under steady-state conditions.
1
2
-
1
cm resolution can easily be made in 100 ms or less. In a
typical surface science experiment, however, involving sub-
monolayer adsorbate coverages, several hundred such individual
interferograms must be added in order to obtain a sufficient
signal-noise ratio. By degrading the spectral resolution and/or
compromising the signal-noise ratio, temporal resolution can
be improved but inevitably important information is lost.
2. Experimental Details
The experimental procedure for collecting the time-dependent
IRAS and partial-pressure measurements has been outlined
1
5
above and described in more detail previously. The Pt{100}
sample was prepared in situ by the usual ion bombardment and
annealing cycles and then cooled to a temperature 10 K above
that of the existence range of the oscillatory regime to be studied.
Prior to initiating the oscillations, NO was then admitted to the
Of course, the fact that the time dependence of interest is
periodic means that all the measurements need not be completed
in the time-scale of a single oscillation, and data recorded at
equivalent times in the period cycle may be added. One possible
but rather complex approach to this problem is the synchronized
step-scan method, in which interferograms representing temporal
resolution elements of an oscillatory cycle are constructed by
adding single data points or groups of data points from
-
7
chamber and left to stabilize to a pressure of 7.5 × 10 mbar
over a period of several hours. CO was then introduced such
that the pNO/pCO ratio was 1.5. After a period of pressure
stabilization (∼60 min), during which time the partial pressures
of both gases were continuously monitored with the mass
spectrometer, the crystal was heated to 700 K and cooled to a
temperature ∼10 K above that of the existence range of the
oscillatory regime to be studied. A temperature modulation of
the desired period and amplitude (60 s and 2.0-2.5 K) was
then applied to the sample in order to initiate and sustain the
oscillations, and the mean temperature was gradually reduced
until the natural frequency of the system was in resonance with
the applied temperature oscillation. This condition was identified
by a large increase in the amplitude of oscillation of the CO2
partial pressure. Although this procedure is often referred to as
forcing the oscillations, extensive studies of periodic and random
temperature perturbations of this system have shown that it can
only be forced to oscillate at frequencies very close to the natural
frequency, i.e., the system acts as a narrow band-pass filter.18
To check that the system was behaving in the fashion
determined by the natural oscillation period of the reaction, tests
were conducted in which the applied temperature modulation
was turned off; no change was seen in the form or period of
the gas-phase partial-pressure traces, as measured by the mass
spectrometer. Previous work has shown that the natural frequen-
cies of the system at the upper and lower temperature extremes
of a modulation amplitude of 2.5 K may differ by a maximum
14
successive cycles. Although this method can achieve very high
temporal resolution (<1 µs), it is not a viable method for the
present problem. In particular, if individual data points are being
added to construct an interferogram, ∼8000 oscillation cycles
-
1
would be required to obtain a single interferogram of 2 cm
resolution for each temporal resolution element. Given that
typical oscillation periods for the reactions of interest here are
of the order of tens of seconds, the total data acquisition time
is unrealistically large. Moreover, even very small variations
in the oscillation period or loss of precise synchronization will
14
lead to spectral artifacts in the final transformed interferogram.
Recently, we have demonstrated the viability of a modifica-
tion of this approach that offers a more realistic compromise
between good temporal and spectral resolution and sensitivity,
enabling detailed information regarding the coverages and local
environment of the surface adsorbates, together with the partial
pressures of the reactants and products, to be obtained through-
out the oscillation cycle,15 and this approach has subsequently
been applied to studies of the high-temperature oscillatory
regime of the NO/CO reaction on Pt{100}.1
6,17
Briefly, it
involves sustaining the oscillatory cycle and continuously
recording interferograms over a period of many oscillatory
cycles. Interferograms recorded at equivalent points in the
oscillatory cycle are then added to create a single data set
representing a single oscillatory period. Since whole interfero-
grams are added, spectral artifacts are not introduced into the
system and the advantage of adding data over many oscillatory
periods is maintained. Notice that because IRAS samples a large
area of the surface, it is important that the state of the whole
surface oscillates in phase during the oscillatory reaction;
typically a temperature step can trigger the oscillations, but these
are damped at the low pressures used in our experiments due
to loss of macroscopic synchronicity.1 The application of an
external global coupling mechanism, in the present case a small
temperature modulation at the natural frequency of the oscil-
lation forces a homogeneous oscillation and is not thought to
change the mechanism of the reaction. The partial pressures of
the reactants and products are measured simultaneously with
the IR spectra so that the relative phases of the oscillations in
these signals can be related to those of the infrared data from
the adsorbates. Our original demonstration of this special form
of time-resolved IRAS was conducted on the Pt{100}/NO +
CO system, but a number of details of the results and their
interpretation proved ambiguous, possibly due in part to the
relatively low spectral resolution used.15 Here we present the
results of new and more extensive measurements performed at
higher resolution and discuss them in the light of our more recent
9
of 10%. This effect can also be minimized by ensuring that
the same degree of deviation from the applied frequency is
exhibited at the high and the low-temperature extremes of the
1
7
modulation. After these checks were performed, the temper-
ature modulation was reapplied and, following a second period
of stabilization (∼10-20 min), in which the amplitude and
shape of the CO partial pressure oscillation had become
2
completely reproducible, the IR data acquisition was started.
Figure 2 shows a schematic diagram of the experimental
control system and data acquisition procedure. A signal genera-
tor drives the temperature controller, which periodically modu-
lates the temperature of the sample. During the oscillations, the
8,19
partial pressure of NO (30 amu), CO + N (28 amu), and CO2
2
(44 amu) were continuously monitored with the quadrupole mass
spectrometer installed on the ultrahigh vacuum chamber.
Simultaneously, the FTIR spectrometer continuously records a
succession of interferograms over as many oscillatory periods
as required. Notice that there is no actiVe synchronization of
the IR spectra collection and the reaction oscillations, but the
relative timing is coordinated via the temperature of the sample.
The infrared spectrometer gives out a trigger pulse each time a
spectrum is acquired, which is fed into a PC. This records the
spectrum number, the temperature of the sample and the time
at which the spectrum was obtained. Another PC simultaneously
records the temperature of the sample and the data from the
mass spectrometer. Thus each data point in the mass spectrum

Real-Time Vibrational Spectroscopic Investigation
J. Phys. Chem. B, Vol. 108, No. 5, 2004 1711
these spectra the weak feature at 1750 cm- is an artifact
resulting from the polarizer and a small amount of drift in the
spectrometer.
1
3
. Results and Discussion
.1. Overview and the Role of Two Distinct Atop CO
3
Species. Figure 4 shows the set of time-resolved IRAS spectra
corresponding to 2 s intervals within a single period of
oscillation of the NO + CO reaction over Pt{100}, resulting
from our experiment conducted at an average temperature of
3
98 K within the low-temperature existence regime of the
oscillatory reaction. The overall behavior is similar to that
reported in the experiments conducted as part of the original
15
demonstration of this method, but there are important detailed
differences, in part due to the higher spectral resolution of the
-
1
present data (2 cm ) but also due to the slightly lower
temperature (398 K rather than 402 K). In particular, while the
-
1
main absorption bands are at 1628, 1875, and 2070 cm , this
highest wavenumber band is clearly split into two distinct
-
1
components at 2065 and 2085 cm in the new data of Figure
. From previous infrared studies of the reactant molecules on
4
Pt{100}, the three main bands can be assigned to the internal
stretching modes of NO, bridged CO, and atop CO, while the
two atop CO components can be assigned to CO in mixed NO
+
CO islands and CO in pure CO islands respectively.1
2,20-25
Notice that the exact frequencies associated with these vibra-
tional modes are influenced by dynamic dipole coupling, which
allows us to establish the nature of the environment of the
adsorbed molecules, and specifically the local coverage of
identical species, which determines the degree of vibrational
Figure 2. Schematic diagram of the control system for producing
forced oscillations and synchronously measuring FTIR spectra and
partial pressure data.
and each FTIR spectrum can be uniquely identified in terms of
the time and temperature at which they were recorded. Spectra
from equivalent points in the cycle are then combined and
averaged to give a set of spectra that represent a single period
of oscillation, as illustrated in Figure 2.
-
1
coupling. In particular, the frequencies of the 1628 cm (NO)
-
1
and 2065 cm (atop CO) absorption bands are shifted down
from those expected for pure islands of their respective
individual molecular species on the same surface; this has been
shown in steady-state coadsorption experiments to be associated
with the formation of domains of intermixed NO + CO on Pt-
{100}, which results in a reduction of the dipole-dipole
The temporal resolution of this method is determined by the
time it takes to record a single interferogram, although in the
case of the Biorad FTS 60 used here, the practical limit is the
time it takes to record two interferograms from the forward and
reverse directions of the moving mirror since these are
automatically added. However, since the times at which the
interferograms are collected are not actively synchronized to
the reaction oscillations, the effective temporal resolution is
actually twice the time required to record a single spectrum.
By use of the experimental setup described here, a single
spectrum resulting from the addition of a forward and a reverse
interferogram takes 250 ms. In a surface science experiment,
typically several hundred interferograms are added to obtain a
spectrum of sufficient signal-noise ratio. Thus, using the present
method, data from hundreds of oscillatory cycles would have
to be added to obtain spectra of comparable quality. Given that
our oscillation period is approximately 60 s, the total measure-
ment time to achieve the best possible time resolution is thus
many hours, leading to potential problems with accumulated
surface contamination. It was decided to degrade the temporal
resolution such that a set of spectra with a good signal-noise
ratio could be acquired over 20 oscillatory cycles by taking data
in blocks of 8 interferograms, giving a total of 160 interfero-
2
6
coupling between identical species. The frequency of the
bridged CO stretch is only slightly shifted from that observed
when the surface is prepared in an identical manner in the
2
3
absence of an applied NO partial pressure, so on the basis of
the frequency shifts alone it is not possible to establish whether
the bridged CO is also intermixed with the NO. As we shall
show, however, the variations of intensity of the atop and bridge
species during the reaction cycle do clearly establish that the
bridged CO is similarly intermixed. Notice, incidentally, that
the frequencies of these absorption bands are all indicative of
adsorption on the (1 × 1) surface, consistent with the low-
temperature oscillatory reaction occurring on an entirely un-
reconstructed surface as has been previously established in
diffraction studies.
The presence of the 2085 cm- C-O stretching mode
associated with islands of pure atop CO was also found at almost
the same temperature (400 K) in the steady-state experiments
1
1
2
presented previously. The interpretation in terms of pure CO
islands is further reinforced by the observation that this
adsorption band is much narrower than the lower frequency band
associated with atop CO intermixed with NO; for this latter state,
there will be a range of local environments and thus significant
inhomogeneous line broadening. Figure 5 shows an overlay of
the IRAS spectra in the C-O stretching range through the
reaction cycle. This shows very clearly that while the absorbance
-
1
grams in each summation (of 2 cm resolution) recorded at 2
s intervals, giving a temporal resolution of 4 s. Figure 3
illustrates how this summation is performed, showing the
individual spectra from measurements taken at an average
temperature of 401.7 K at equivalent relative times in four-
reaction oscillations, their sum, and the final spectrum for this
relative time after summation over all 20 cycles. Notice that in
-
1
of the 2065 cm band varies strongly during the cycle, that of
-
1
the 2085 cm is constant, as is its exact peak wavenumber

1712 J. Phys. Chem. B, Vol. 108, No. 5, 2004
Miners et al.
Figure 3. Illustration of how FTIR spectra, recorded at times of equivalent phase in the oscillation are combined and averaged to yield a single
spectrum of increased signal-to-noise ratio. The spectra are taken from a set of measurements of the CO/NO oscillatory reaction over Pt{100}
similar to those shown in Figure 4 but recorded at a slightly different temperature.
Figure 4. A series of 30 IRAS (FTIR) spectra at intervals of 2 s representing a single period in the CO/NO reaction over Pt{100} in the low-
-1
temperature oscillatory regime. Each spectrum is recorded with spectral resolution of 2 cm , comprised of 160 scans, and represents the combination
of spectra recorded over 20 oscillatory cycles, with a time resolution of 4 s (cf. Figure 3).
and width. This lack of variation implies very clearly that these
pure CO islands are not involved in the oscillatory reaction.
The spectra of Figure 5 also provide further evidence that the
two atop CO species are spatially distinct. Because the frequen-
cies of the two bands are very close, we would expect strong
vibrational coupling to occur if the associated molecules were

Real-Time Vibrational Spectroscopic Investigation
J. Phys. Chem. B, Vol. 108, No. 5, 2004 1713
Figure 5. Overlay of an expanded section of the IRAS spectra of Figure 4 in the spectral range corresponding to the atop C-O stretching vibrations.
close to one another; this would lead to intensity transfer to the
higher-frequency peak and thus to variations in its intensity due
to the variations of intensity of the lower-frequency peak.
The clear evidence that the pure atop CO islands exist on
the surface, not participating in the reaction, and unchanged
over many cycles of the reaction on the remainder of the surface
is puzzling. A completely CO-saturated surface ought to be
passive because there would be no free sites for NO adsorption
and dissociation, and NO arriving from the gas phase is known
not to displace adsorbed CO. However, one might still have
expected that reaction could occur at the edges of these islands,
especially as at these high temperatures the CO is very
mobile.2 On the other hand, atomic oxygen produced by NO
dissociation is strongly bound to the surface and is immobile
at this temperature,2 i.e., it cannot diffuse into these islands,
so this might limit the edge activity. We should also stress that
the relative contribution of these pure CO islands to the IRAS
data was found to be very sensitive to the conditions of the
experiment and especially to the temperature. Notice, for
example, that in Figure 3, recorded at the same spectral
resolution as the spectra of Figure 4 but an average temperature
28 (CO + N2), 30 (NO), and 44 (CO2). Also shown is the
temperature variation of the sample applied to ensure a spatially
synchronous reaction. Since this experiment is performed under
continuous flow conditions at high pumping speed, decreases
in the partial pressures of reactants and increases in the partial
pressures of the products are proportional to the rate of their
reaction and the rate of their production, respectively. Further-
more, we anticipate that the IRAS-integrated absorbances should
be linearly related to the coverage of the relevant species; this
is not, of course, universally true for IRAS, and in the case of
CO in Pt{100}, studies indicate that this linear relationship may
be lost due to site switching and possible tilting of mole-
7,28
22
cules for coverages above θCO ) 0.5 monolayers (ML), but
below this coverage the assumption is believed to be essentially
valid for both NO and CO, and is consistent with previous
9,30
1
2,15
results.
Figure 6 shows that the form of the temporal variations of
the CO2 partial pressure is essentially identical to the negative
of the NO partial pressure changes and is also clearly similar
to the variation of the time derivative of the absorbance of the
NO internal stretch. Unfortunately the coincidence in mass
number of CO and N2 make it difficult to deduce much detail
regarding the temporal variations of these individual reactant
and product species from the combined trace in Figure 6. To
overcome this limitation, a separate experiment was performed
-
1
3
.7 K higher (401.7 K), the pure atop CO peak at 2085 cm is
present only as a small high-wavenumber shoulder on the peak
-
1
at 2065 cm . Much the same appears to be true of the original
measurements conducted at 402 K although in view of the lower
resolution it is difficult to be sure.15 One possible underlying
reason for this narrow range of mutual existence of the reacting
and nonreacting regions of the surface may be related to the
temperature dependence of the rate of NO dissociation, the key
step in the NO/CO reaction; this is typically regarded as
occurring only above about 400 K, so very close to this
temperature the rate of dissociation will be delicately balanced
and could be significantly influenced by some subtle property
of the environment (including free sites for dissociation) at the
edge of the pure CO islands.
13
18
13
18
using high purity C O ( C, 99%; O, 98.4%), leading to a
CO mass of 31 amu with a negligible contribution (0.002%) at
3
1
28 amu. TPR measurements of Zemlyanov et al. show that
1
3 18
the kinetic isotope effect in using C O will not lead to a
significant change in oscillatory behavior. Representative results
from this additional experiment, recorded at 396 K with an
applied temperature modulation of 2 K, are displayed in Figure
7. The mass spectrometer signals are averaged over 5 oscillations
of 60 s; each point represents a 2.65 s segment of the oscillation
with an effective time resolution of 5.3 s. The mass 31 signal
1
(P
3 18
3
.2. Detailed Data Evaluation and Comparison with the
C
O) has been corrected to remove the mass spectrometer
1
3 18 16
Model Predictions of Fink et al. In Figure 6 is summarized
the variations of the integrated absorbance of the four IRAS
bands through the oscillatory reaction cycle together with the
associated variations in the partial pressures at mass numbers
cracking contribution of the C O O signal (5.7%), and the
1
4
16
12 18
mass 30 signal (P N O) has been corrected to remove the C O
1
3 18
contribution, which is 1.0% of the C O signal. To simplify
the comparison of the different partial pressure variations, Figure

1714 J. Phys. Chem. B, Vol. 108, No. 5, 2004
Miners et al.
Figure 7. The time dependence of individual peaks in the mass
spectrometer signal during the CO/NO oscillatory reaction over Pt-
100} resulting from the application of a 13C O/ N O mixture in the
18
14 16
{
-7
partial pressure ratio 1:2 at a total reactant pressure of 7.5 × 10 mbar,
average temperature 396 K with an applied temperature modulation of
2
K. Each plot is the result of an average over 5 oscillations of 60 s,
each point represents a 2.65 s segment of the oscillation with a time
resolution of 5.3 s. Displayed, from top to bottom, are the relative partial
pressures N (multiplied by a factor of 2), CO , and the NO and CO
2 2
Figure 6. The time dependence of the sample temperature, CO
and CO + N partial pressures, and the integrated absorbance of the
four IR absorption bands (N-O, atop C-O (2065 cm ), atop C-O
2
, NO,
2
-
1
partial pressures, the latter two multiplied by (-1).
-1
(
2085 cm ), and bridge C-O) during an oscillatory cycle, correspond-
ing to the experiment of Figure 4.
partial pressures implies that the production of both gases is
controlled by the same rate-limiting mechanism, presumably
that of NO dissociation. This lack of phase difference is in clear
disagreement with the model of Hopkinson and King, which
predicts that the N oscillations should lag those of CO by a
7
shows the CO and NO partial pressures multiplied by -1,
thus generating a signal which reflects the rates of CO and NO
adsorption; furthermore, the partial pressure of N2 is multiplied
by 2 in Figure 7 to allow for the fact that the surface reaction
2
2
1
90° phase difference ( / of an oscillation period) and be of a
4
1
11
takes the form CO + NO f CO2 + /2N2. The estimated
flatter, more extended form. In this regard the data are,
however, consistent with the assumption of Fink et al. 9 that
precision of these partial pressure plots is 5% of the maximum
values for CO and NO (the oscillation amplitudes are only 0.4%
of the total signal for these two reactants) and 1% for CO2 and
N2. While the signal-to-noise ratios of these data of Figure 7
are clearly inferior to those of the partial pressure plots of Figure
the formation of N is immediate upon NO dissociation. On
2
the other hand, the fact that the rate of reactant adsorption is
very similar to the rate of product formation implies that the
majority of the reaction occurs almost immediately on adsorp-
tion, i.e., that NO and CO adsorb on a cleared area of the surface,
and immediately react. This process correlates with stage III of
the oscillatory cycle (Figure 1), whereas in the model of Fink
et al., the highest rate of reaction occurs due to the surface
explosion and the reactive removal of the adsorbed layer, which
is identified as the earlier stage II of the reaction cycle. Careful
inspection of the data of Figure 7 suggests that the two product
oscillations do appear to lead slightly those of the reactant gas
pressures by about 2-3 s; bearing in mind that the pressures
used correspond to a rate of arrival of gas molecules of each
6, the important conclusion to be drawn from the results of this
additional experiment is that the pressure variations of NO and
CO are essentially identical in amplitude and phase, so we may
assume that the temporal variation of the partial pressure of
CO is closely similar to that of NO shown in Figure 6.
In fact, plotted in the form of Figure 7, it is clear that the
variations in partial pressure of all the gases are very similar,
with only very small differences in relative phase of their
oscillations. This result has two important implications. First,
the close similarity in phase of the variations in N2 and CO2

Real-Time Vibrational Spectroscopic Investigation
J. Phys. Chem. B, Vol. 108, No. 5, 2004 1715
species at the surface of very approximately 0.5 ML/s, this is
consistent with the idea that the reactant consumption is
governed by the creation of free sites by surface reaction. We
should, however, beware of reading too much into such subtle
details, as it is clear from the higher-quality partial-pressure data
of Figure 6 that there are subtle variations in the shape of the
temporal variations relative to the essentially sinusoidal varia-
tions which appear to dominate the data of Figure 7.
Consider now, the information provided by the IRAS data
in terms of temporal variations of surface coverage. One
interesting question is whether these data show any evidence
for the role of nonreactive displacement of adsorbed NO by
CO during the oscillatory reaction. We have already remarked
that there is ample evidence that this replacement process does
occur in coadsorption studies; it can be attributed to the higher
heat of CO adsorption, as first proposed by Hopkinson and
King,11 but this process is not included in the modeling of the
9
interaction by Fink et al. The data of Figure 6 show that during
the first 20 s or so of the oscillatory cycle (emboldened in Figure
6
and starting at a time of 30 s), the absorbance of the N-O IR
band falls steadily, those of the C-O IR bands rise, yet the
rate of CO2 production remains low and the NO partial pressure
actually rises very slightly. This combination of events can only
be understood in terms of nonreactive displacement of NO by
CO, so it is clear that this process does play a significant role
in part of the oscillatory cycle. By contrast, however, co-
adsorption studies also show that NO cannot displace adsorbed
CO in this way, so one may conclude that loss of surface CO
is a direct measure of surface reaction. Using this fact, one can
calculate a minimum value for the throughput of one oscillatory
cycle as being the variation in coverage of atop CO. To do so
we must relate the measured IR integrated absorbances to
absolute coverages, and the data of Martin et al.23 provide a
basis for this calibration. Notice that as the isolated CO species
-
1
giving the 2085 cm wavenumber band does not contribute to
the reaction, we ignore this component in the calculation. In
particular, at 90 K and a coverage of 0.5 ML corresponding to
a c(2 × 2) long-range ordered surface, they see essentially the
same ratio of integrated absorbances of the bridge-to-atop bands,
with both bands showing an integrated absorbance a factor of
2
Figure 8. The time dependence of the CO partial pressure and the
absorbance of the three main IR bands (N-O, C-O atop, C-O bridge)
during an oscillatory cycle, corresponding to the experiment of Figure
4
2
. The dashed curves are the predictions of the CO production and
9
NO and CO surface coverages given by the model of Fink et al. The
roman numerals refer to the five stages of this model of the reaction
shown schematically in Figure 1. The relative phases of the experimental
and theoretical results have been adjusted to align the main peaks in
3
larger than the peak values of our data. We therefore conclude
that the maximum total (reactive) CO coverage during the
2
the rate of CO production. The horizontal lines correspond to zero
1
reaction would be approximately /6 ML if these species were
values of integrated absorbance or coverage of the individual plots
indicated by the vertical arrows. On the basis of the calibration methods
described in the text, the estimated maximum coverages of CO and
NO represented by the integrated absorbance plots are 0.20 and 0.36
ML respectively.
distributed over the whole surface. However, we may estimate
that approximately 20% of the surface is occupied by the
nonreacting pure CO islands, so the peak coverage in the
reacting area is approximately 0.20 ML. During the reaction,
this value falls by a factor of 3, so the total amount of CO
consumed in one cycle is approximately 0.13 ML. This value
is consistent with estimates based on the measured changes in
partial pressures in the gas phase, calibrated using thermal
desorption of a known coverage of 0.5 ML of CO.
differences in the experimental and theoretical curves, although
differences in the relative phases of two plots in theory and
experiment should be meaningful.
Clearly the overall trends of the theoretical predictions and
experimental measurements shown in Figure 8 are very similar.
In particular, the relative phase of the variations of the partial
Figure 8 compares the partial pressure of CO2 and the
integrated absorbance of the NO, mixed atop CO, and bridged
CO taken from Figure 6 (solid lines) with the predicted CO2
yield and NO and CO surface coverages of the model of Fink
pressure of CO and the coverages of NO and CO on the surface
2
during the oscillatory cycle are in excellent agreement with the
model predictions, and many aspects of the shapes of the
9
et al. (shown schematically in Figure 1). Of course, to make
temporal variation of the CO coverage and CO partial pressure
2
this comparison, it is necessary to determine the relative phases
of the experimental and theoretical results; this has been
achieved by aligning the centers of the main CO2 partial pressure
peaks in the two data sets. The exact shapes of these peaks in
the experiment and theory differ, so there is some ambiguity in
this alignment. In considering this comparison, it is therefore
not possible to draw detailed conclusions from small phase
are also in agreement. The actual surface coverages of NO and
CO are also in rather good agreement with the model predic-
tions. Figure 8 shows the predicted coverages explicitly, but
the experimental integrated absorbances have not been converted
to coverages and only the locations of the respective zero
absorbances are marked on Figure 8 to allow comparison of
the relative amplitudes of the variations. We have already shown,

1716 J. Phys. Chem. B, Vol. 108, No. 5, 2004
Miners et al.
however, that the peak coverage of reactive CO (both intermixed
atop and bridging) is estimated to be 0.20 ML, in almost perfect
agreement with the model prediction. A similar calibration of
the N-O-integrated absorbance using spectra of the room-
1
2
temperature saturation coverage of NO (in the absence of CO)
with an estimated coverage of 0.5 ML21 allows us to convert
the average integrated N-O absorbance of 0.07 cm (Figure
) to a coverage of 0.32 ML, similar to, but slightly less than,
-
1
6
the predicted value of approximately 0.38 ML. Note, however,
that the magnitude of the measured temporal variations of both
the CO and NO coverages estimated from the IR absorbances
are somewhat less than those seen in the model predictions;
we will return to this aspect below.
Despite this excellent overall agreement, there are some
significant differences in detail between theory and experiment.
Notice, in particular, that in the experimental data, the minimum
of the NO coverage, as determined by the absorbance of the
N-O absorption band, occurs before the maximum in CO2
production, whereas the order is reversed in the model predic-
tions. This confirms our previously published result, recorded
at lower spectral resolution15 in an earlier report, it was
suggested that the IRAS showed some evidence of an unresolved
second N-O band which may correspond to a second species
not considered in the existing model. However, the new higher-
resolution spectra presented here show no evidence of such a
component. Notice, too, that the experimental data show the
NO surface coverage rising most steeply at a time in the cycle
at which the CO surface coverage is falling most steeply; in
the model predictions the CO coverage is almost constant and
has a very low value at the equivalent point in the cycle.
Figure 9. The time dependence of the peak wavenumber, fwhm, and
integrated absorbance of the three main IR bands (N-O, C-O atop,
C-O bridge) during an oscillatory cycle, corresponding to the
experiment of Figure 4. The dashed curves are the predictions of the
In consideration of the origins of these discrepancies, we
should note that our data clearly show the presence of bridging
CO as well as atop CO on the surface, whereas no distinction
between two sites is included in the model, so one might
speculate that a different reaction of these two species with
surface NO may play a role. However, the data clearly show
that the variations of the integrated absorbance of these two
species throughout the periodic cycle are essentially identical.
A more reasonable explanation for the detailed discrepancies
is that the model of Fink et al. fails to include the effect of the
nonreactive displacement of NO by CO. The rate of this
displacement process has not been determined at 400 K, but at
2
CO production and NO and CO surface. The roman numerals refer to
9
the five stages of the reaction according to the model of Fink et al.
shown schematically in Figure 1 aligned in phase as in Figure 8.
In view of the fact that the surface explosion lies at the heart
of the model of Fink et al., this discrepancy at first appears
rather serious. However, our comparison of theory and experi-
ment is based on the assumption that the experiment (like the
theory) corresponds to conditions in which the surface is
homogeneous and changes in the surface coverage with time
are perfectly synchronous. While the general methodology
adopted here is designed to ensure that this is basically correct,
we know that in this bistability the major transformation of the
surface occurs by the propagation of a reaction front, so at some
points in the cycle there must be local inhomogeneity. Clear
evidence that this is the case is contained in the variation of
the exact frequencies and widths of the IRAS absorption bands
through the periodic cycle, and these are shown, together with
the band absorbances, in Figure 9. We should recall that the
main factor influencing the exact frequencies of the N-O and
C-O stretching bands is dynamic dipole coupling, so higher
local coverages of a single species lead to higher frequencies.
There are also contributions from so-called chemical shifts in
which the vibrational frequency is influenced by modified
electronic interaction with the surface due to coadsorbed atoms
and molecules (including those of the same species) but these
effects are known to be small for CO and NO on Pt surfaces.
Bearing this in mind, one particularly striking aspect of the data
of Figure 9 is that the peak wavenumber of the intermixed atop
3
90 K, with equivalent partial pressures of NO and CO, CO
1
2
was found to displace 50% of the NO coverage in 30 s;
evidently this process should be very significantly faster at 400
K, ensuring that it is important in an oscillatory process with a
period of approximately 60 s. We have already shown that this
displacement process can account for the fall in NO coverage
and rise in CO coverage at the beginning of the oscillatory cycle
of Figure 8 when the rate of CO2 production is low (and not
increasing as in the model predictions). The NO coverage can
only increase by the reactive removal of CO, which creates free
sites on the surface for NO adsorption; such free sites are
preferentially occupied by NO because the NO partial pressure
is higher than that of CO, and their sticking factors at such sites
are both essentially unity. Indeed, the experimental data show
that this point in the cycle, when the CO coverage starts to fall
and the NO coverage starts to rise, corresponds to the highest
reaction rate; in the experiment, therefore, the evidence is that
this maximum rate arises from adsorption and immediate
reaction. By contrast, the model of Fink et al. shows the highest
reaction rate resulting from the surface explosion, which occurs
when the surface coverage of CO and NO are falling steeply
simultaneously.
-
1
C-O stretching mode around 2065 cm displays a sharp
maximum at the point in the cycle at which the integrated
absorbance of this band is minimum. In effect, therefore, this
implies that the local coverage of this species is greatest at the
condition when the aVerage coverage is lowest. Clearly this

Real-Time Vibrational Spectroscopic Investigation
J. Phys. Chem. B, Vol. 108, No. 5, 2004 1717
can only be reconciled with two different spatially separated
states on the surface. Notice too that this occurs at a point in
the cycle at which the average coverage of NO reflected by the
N-O absorbance is high (essentially maximum), but the full
width at half-maximum (fwhm) of this N-O band also shows
a narrow peak. The most obvious interpretation of this is that
there is a significant NO coverage in both the states of the
surface, but the local NO coverages in these two states differ,
leading to a kind of inhomogeneous broadening (or two nearby
unresolved components). Notice, incidentally, that the time
dependence of the N-O stretching frequency has an overall
form which is quite similar to that of its integrated absorbance,
indicating that the local and average NO coverages are similar
over most of the cycle, although the frequency rises more
sharply than the absorbance in the same part of the cycle we
have now identified with a two-phase regime, also consistent
with differences between local and average coverage. These
results therefore indicate that in the section of the reaction cycle
labeled IV, reaction fronts are moving over the surface,
transforming it from a state of relatively high CO coverage and
low NO coverage, to one of relatively low CO coverage and
high NO coverage.
oscillatory cycle, the average value of the ratio of the integrated
absorbance of bridged C-O to the intermixed component of
the atop C-O absorption band alone is 0.51 and varies between
0.81 and 0.42. These comparisons show rather clearly that there
are two bridged CO species, one intermixed with NO and
involved in the reaction and one that is in pure CO islands which
are not involved in the reaction. This conclusion is consistent
with independent steady-state coadsorption IRAS measure-
1
2
ments in which the presence of bridged CO in both environ-
ments is also observed. Notice that the almost constant value
of the ratio of the integrated absorbance of the bridged C-O to
the sum of the two atop C-O absorption bands is indicative
that the relative proportions of the bridged and atop CO species
is similar in both NO-intermixed and pure CO environments is
similar; this result is also consistent with IRAS studies of the
transient behavior of the coadsorption system when subjected
12
to thermal steps. On the basis of the conclusions of the earlier
23
study of pure CO on Pt{111}, it seems that the inactive bridged
CO species must occur in pure bridged CO islands which coexist
with the “spectator” atop CO islands responsible for the IR
-
1
absorption band at 2085 cm .
In considering the role of the intermixed bridged CO in the
NO/CO reaction, one further important observation is that
previous studies of coexistent atop and bridged CO on this
surface have shown that there is a high rate of interchange
between bridged and atop CO at these temperatures, the relative
occupation of the two species being in equilibrium.23 This same
effect has also been shown to occur in the NO/CO coadsorption
situation in which the NO and CO are intermixed on the surface,
although these studies also show that in the nonreactive
displacement of NO by CO, the actual displacement is effected
We should perhaps also remark that one consequence of this
kind of asynchronous transformation of the surface is that the
experiments measure average coverages over the two-phase
region, so one might expect that the experimental coverage
variations would be smaller than those predicted for homoge-
neous surface transitions. This may account, at least in part,
for the reduced modulation amplitudes which we have remarked
upon.
3
.3. The Role of the Bridging CO Species. So far we have
1
2
by atop CO. In particular, it has been shown, at 390 and at
00 K (both temperatures below that at which CO/NO reaction
discussed the reaction almost entirely in terms of the atop CO
3
-
1
species corresponding to the 2065 cm IRAS band, although
occurs), that if a surface which is initially NO covered is exposed
to CO, CO adsorption occurs at first in the atop site but with
subsequent occupation of bridge sites. At 390 K under similar
conditions to those used in this experiment, there is a 9.0 ( 4.5
s lag between the exposure of CO to NO + Pt{100} and the
equilibration between bridged and atop CO. At the slightly
higher temperatures of the oscillatory reaction investigated here,
this delay time will be significantly shorter. Strictly, the fact
that the nonreactive displacement of NO is effected by only
atop CO does not necessarily mean that only atop CO is
involved in the reactive removal of NO, but the rapid equilibra-
tion between the two CO species means that which species (if
indeed it is only one species) is involved in the reaction will
have little significance in influencing the behavior of the
oscillatory reaction cycle.
-1
in addition to the “spectator” atop CO species (2085 cm IRAS
band), which clearly does not participate in the reaction, we
-1
also see a bridging CO species (1875 cm IRAS band). Models
of the NO/CO reaction over Pt{100} have assumed that only
one CO species is present, but the variations of the integrated
absorbance of the bridged species through the reaction cycle
clearly shows that these CO species are involved in the reaction
in some way. We must therefore consider whether the presence
of this bridging CO plays an important role in the reaction. Of
course, the similarity of the variations of the integrated
absorbance of the bridging and intermixed atop CO species
during the reaction cycle (Figure 6) shows that at least some of
these species must clearly be intermixed with the NO on the
surface, and not in isolated islands as in the case of the 2085
-
1
cm atop CO species. In view of the fact that we have shown
that there are two distinct atop CO states, however, one in pure
CO islands and one intermixed with (and reacting with) NO,
we might ask whether the same is true for the bridging CO.
Notice that the bridging C-O band is weaker and broader than
the atop C-O band, so it is not possible to establish whether
there is any significant splitting of this band. However, if we
compare the changes of integrated absorbance of the bridge
C-O band through the reaction cycle we find that it displays a
very similar behavior to that of the sum of both atop C-O bands.
The minimum absorbance of the bridged C-O band is 52 (
4
. Conclusions
Overall, the results of our measurements of the temporal
variation of the CO and NO surface coverages, and the rate of
reactant uptake and product generation, are in very good
agreement with predictions of the model proposed by Fink et
9
al., although there are some detailed discrepancies. We attribute
many of these differences to two factors: (1) the fact that even
in an experiment designed to maintain a homogeneous surface
with synchronous changes in the state of the whole surface,
some transformations (such as the surface explosion) inevitably
involve reaction front propagation and thus inhomogeneity and
(2) the neglect in the model of the process of nonreactive
displacement of NO by CO.
5
% of its maximum value, similar to the value of 60 ( 2% for
the total intensity of both atop species and quite different from
-1
the value of 30 ( 3% for the intermixed (2065 cm ) atop C-O
band alone. Furthermore, whereas the ratio of the integrated
absorbance of the bridged and both atop C-O bands has an
approximately constant value of 0.26 ( 0.06 thoughout the
Our results also show that two CO species, atop and bridged,
are involved in the reaction within the intermixed NO + CO

1718 J. Phys. Chem. B, Vol. 108, No. 5, 2004
Miners et al.
regions of the surface. Although this factor has not been
considered in previous models of the reaction, bridged CO is
shown to play an equivalent role in the reaction mechanism to
atop CO, primarily because the two species rapidly equilibrate
with one another. For this reason, it is sufficient that a model
simply consider “adsorbed CO”. Somewhat surprisingly, the
coexistence of nonreacting pure atop CO islands is also observed
(4) Ertl, G. AdV. Catal. 1990, 37, 213.
(
(
(
(
5) Imbihl, R. Prog Surf. Sci. 1993, 44, 183.
6) Veser, J.; Imbihl, R. J. Chem. Phys. 1994, 100, 8492.
7) Sch u¨ th, F.; Wicke, E. Ber. Bunsen-Ges. Phys. Chem. 1989, 93, 191.
8) Singh-Boparai, S. P.; King, D. A. In Proceedings of the Fourth
International Conference on Solid Surfaces and the Third European
Conference on Surface Science; Degras, D. A., Costa, M., Eds.; SFV: Paris,
1
980.
9) Fink, Th.; Darth, J.-P.; Imbihl, R.; Ertl, G. J. Chem. Phys. 1991,
95, 2109.
(10) Tammaro, M.; Evans, J. W. J. Chem. Phys. 1998, 108, 7795.
(
(
and coexistent nonreacting pure bridge CO islands are also
inferred). The extent to which these “spectator” islands occur
appears to be very sensitive to the exact experimental conditions
and appears to be favored at the lowest temperatures of the “low-
temperature” oscillatory reaction regime at which the rate of
NO dissociation is probably only just sufficient to sustain the
reaction.
(
(
(
11) Hopkinson, A.; King, D. A. Chem. Phys. 1993, 177, 433.
12) Miners, J. H.; Gardner, P.; Woodruff, D. P. Surf. Sci., submitted.
13) Yeo, Y. Y.; Vattuone, L.; King, D. A. J. Chem. Phys. 1996, 104,
3
810.
(14) Willis, H. A.; Powell, D. B. In PerspectiVes in Modern Chemical
Spectroscopy; Andrews, D. L., Ed.; Springer-Verlag: Berlin, 1990.
15) Miners, J. H.; Martin, R.; Gardner, P.; Nalezinski, R.; Bradshaw,
A. M. Surf. Sci. 1997, 377-379, 791.
16) Magtoto, N. P.; Richardson, H. H. Surf. Sci. 1998, 417, 189.
(17) Miners, J. H.; Gardner, P. J. Phys. Chem. B 2000, 104, 10265.
18) Dath, J.-P.; Fink, Th.; Imbihl, R.; Ertl, G. J. Chem. Phys. 1992,
6, 1582.
19) Veser, J.; Mertens, F.; Mikhailov, A.; Imbihl, R. Phys. ReV. Lett.
993, 71, 935.
(20) Gardner, P.; T u¨ shaus, M.; Martin, R.; Bradshaw, A. M. Vacuum
990, 41, 385.
(
Clearly it would be of interest to extend the model of Fink
et al. to include the mechanism of nonreactive replacement of
NO by CO and make a more detailed comparison of the results
with our experimental data. It may, however, be even more
fruitful to go beyond a mean-field model (such as that used by
(
(
9
1
1
1
(
9
11
Fink et al. and Hopkinson and King). In particular, advances
in the time-dependent (dynamic) Monte Carlo3
2,33
simulation
of kinetic oscillations in heterogeneous catalytic systems can
be extended to predict pattern formation, as has been success-
(
21) Gardner, P.; Martin, R.; T u¨ shaus, M.; Bradshaw, A. M. Surf. Sci.
990, 240, 122.
22) Gardner, P.; Martin, R.; T u¨ shaus, M.; Bradshaw, A. M. J. Electron.
Spec. Relat. Phenom. 1990, 54/55, 619.
34
fully performed for the CO oxidation reaction by Gelten et al.
(
An improved model of the CO + NO reaction system might be
achieved by extending the lattice-gas model of Tammaro and
Evans10 of the surface explosion to include both the adsorption,
from an applied partial pressure of both gases, and the
displacement of NO by CO. This would allow modeling of the
stages of the oscillatory cycle after the surface explosion, and
the resulting spatio-temporal model may then reproduce the
various pattern formation observed by PEEM and LEEM as
well as our own experimental results.
(
(
(
(
23) Martin, R.; Gardner, P.; Bradshaw, A. M. Surf. Sci. 1995, 342, 69.
24) Crossley, A.; King, D. A. Surf. Sci. 1980, 95, 131.
25) Banholzer, W. F.; Masel, R. I. Surf. Sci. 1984, 137, 339.
26) Gardner, P.; Martin, R.; T u¨ shaus, M.; Bradshaw, A. M. Surf. Sci.
1992, 269, 405.
(
(
(
27) Seebauer, E.; Schmidt, L. Chem. Phys. Lett. 1986, 123, 129.
28) Kwasniewski, V.; Schmidt, L. Surf. Sci. 1992, 274, 329.
29) Norton, P. R.; Davies, J. A.; Creber, D. K.; Sitter, C. W.; Jackman,
T. E. Surf. Sci. 1984, 108, 205.
(30) Norton, P. R.; Binder, P. E.; Griffiths, K. J. Vac. Sci. Technol., A
984, 2, 1028.
1
(
31) Zemlyanov, D. Y.; Smirnov, M. Y.; Gorodetskii, V. V.; Vovk, E.
References and Notes
I. Catal. Lett. 1997, 46, 201.
(
(
(
1) Hugo, P. Ber. Bunsen-Ges., Phys. Chem. 1970, 74, 121.
2) Beusch, H.; Fieguth, P.; Wicke, E. Chem. Ing. Tech. 1972, 44, 445.
3) Scheintuch, M.; Schmitz, R. A. Catal. ReV. Sci. Eng. 1977, 15,
(32) Gillespie, D. T. J. Comput. Phys. 1976, 22, 403.
(33) Gillespie, D. T. J. Phys. Chem. 1977, 81, 2340.
(34) Gelten, R. J.; Jansen, A. P. J.; Van Santen, R. A.; Lukkien, J. J.;
Segers, J. P. L.; Hilbers, P. A. J. J. Chem. Phys. 1998, 108, 5921.
1
07.