Molecular beam measurements and Monte Carlo simulations of the kinetics of N2O decomposition on Rh(111) single-crystal surfaces

Article abstract of DOI:10.1063/1.2335845

The kinetics of N₂O decomposition on Rh(111) single-crystal surfaces were investigated both experimentally by isothermal molecular beam measurements and theoretically using a Monte Carlo algorithm. The present work was directed to the understanding of two unusual observations derived from our previous work on this system, namely, (1) the lower rates for N₂O decomposition seen at higher reaction temperatures, and (2) the lower total nitrogen yields and final oxygen surface coverages that accompany that behavior. Experimentally, it was determined here that after the rhodium surface is rendered inactive by N₂O decomposition at high (520 K) temperatures, significant activity is still possible at lower (350 K) temperatures. The Monte Carlo simulations explain these observations by assuming that the surface sites required for the activation of adsorbed N2O increase in size with increasing reaction temperature.

Full text of DOI:10.1063/1.2335845

THE JOURNAL OF CHEMICAL PHYSICS 125, 074705 ͑2006͒
Molecular beam measurements and Monte Carlo simulations of the kinetics
of N2O decomposition on Rh„111… single-crystal surfaces
Rodolfo Omar Uñac and Victor Bustos
Laboratorio de Ciencias de Superficies y Medios Porosos, Universidad Nacional de San Luis,
5700 San Luis, Argentina
Jarod Wilson
Department of Chemistry, University of California, Riverside, California 92521
Giorgio Zgrablich
Laboratorio de Ciencias de Superficies y Medios Porosos, Universidad Nacional de San Luis,
5700 San Luis, Argentina
Francisco Zaeraa͒
Department of Chemistry, University of California, Riverside, California 92521
͑Received 24 March 2006; accepted 13 July 2006; published online 18 August 2006͒
The kinetics of N2O decomposition on Rh͑111͒ single-crystal surfaces were investigated both
experimentally by isothermal molecular beam measurements and theoretically using a Monte Carlo
algorithm. The present work was directed to the understanding of two unusual observations derived
from our previous work on this system, namely, ͑1͒ the lower rates for N2O decomposition seen at
higher reaction temperatures, and ͑2͒ the lower total nitrogen yields and final oxygen surface
coverages that accompany that behavior. Experimentally, it was determined here that after the
rhodium surface is rendered inactive by N2O decomposition at high ͑520 K͒ temperatures,
significant activity is still possible at lower ͑350 K͒ temperatures. The Monte Carlo simulations
explain these observations by assuming that the surface sites required for the activation of adsorbed
N2O increase in size with increasing reaction temperature. © 2006 American Institute of Physics.
͓DOI: 10.1063/1.2335845͔
I. INTRODUCTION
into the kinetics of N2O decomposition on Rh͑111͒ directly
using our molecular beam approach.24 The decomposition of
pure N2O on Rh͑111͒ single-crystal surfaces was determined
to occur at temperatures as low as 120 K, and to lead to the
stoichiometric production of N2͑g͒ and adsorbed atomic oxy-
gen. This was to be expected, and is consistent with the
model mentioned above for the reduction of NO. However,
two surprising observations also derived from that work: ͑1͒
the initial rate of N2O decomposition decreases with increas-
ing temperature, and ͑2͒ the total N2 yields and final ad-
sorbed O coverages are also lower at higher temperatures. At
that time we justified the negative apparent activation energy
implied by the first statement via a competition between N2O
desorption and a less activated N2O decomposition. How-
ever, the second conclusion was harder to explain. One pos-
sibility is for N2O decomposition to be facilitated by adja-
cent adsorbed oxygen atoms, in a similar fashion as in the
NO conversion model, which involves a N2O intermediate at
the edges of N islands.15 If this were the case, it could be
speculated that at low temperatures the long residence time
of N2O on the surface would allow it to diffuse farther, and
to therefore be more likely to find an oxygen surface atom to
help with its decomposition. This would lead not only to
higher decomposition rates, but also to the growth of oxygen
surface islands. At higher temperatures the decomposition
would be expected to occur more randomly, on the empty
sites encountered by the incoming gas-phase N2O molecules,
and to result in a slower rate of ͑oxygen-unassisted͒ decom-
The reduction of nitrogen oxides from polluting emis-
sions has become one of the central environmental issues of
our times. This problem is typically dealt with by using
rhodium-based catalysts, one of the central components of
the so-called three-way catalysts used in automobiles and
other applications.1,2 A wealth of scientific literature already
exists on the intimate details of the kinetics and thermody-
namics of the interactions of the major gaseous species
present in the exhaust of internal combustion engines over
Rh surfaces. Extensive experiments have also been carried
out on both realistic supported catalysts3–5 and model single-
crystal samples.6–10 The treatment of N2O, in particular, has
garnered increased interest because of its serious greenhouse
potential,11 and also because a N–NO intermediate is be-
lieved to be a key participant during the reduction of
NO.12–15 Our molecular beam studies on the kinetics of
NO+CO conversion strongly suggest that the atomic nitro-
gen produced by NO dissociation forms islands on the
rhodium surface,16–20 and that the atoms in the perimeter of
those two-dimensional structures are the ones converted to
N2,10,21 via the formation of
a
N–NO surface
intermediate.10,15,22,23
To corroborate this hypothesis, we have recently looked
a͒
Author to whom correspondence should be addressed. Electronic mail:
zaera@ucr.edu
0021-9606/2006/125͑7͒/074705/8/$23.00
125, 074705-1
© 2006 American Institute of Physics

074705-2
Omar Uñac et al.
J. Chem. Phys. 125, 074705 ͑2006͒
position and in less surface oxygen aggregation. If the sites
needed for N2O decomposition are large ͑an ensemble of
several Rh atoms͒, the randomly distributed adsorbed oxy-
gen atoms would poison the surface faster, and would lead to
lower overall N2 yields and oxygen surface coverages.
Here we report on results from molecular beam experi-
ments and Monte Carlo simulations carried out to test these
ideas. It was found that no kinetic model incorporating a
competition between N2O desorption and surface mobility
could lead to significant oxygen island formation, or account
for the decreases in nitrogen and oxygen yields with increas-
ing temperature seen experimentally. Sequential kinetic mea-
surements for the conversion of N2O on Rh͑111͒ at different
temperatures also yielded results inconsistent with our initial
proposal. Instead, evidence was obtained for an inherent in-
crease in the size of the surface site needed for N2O decom-
position with increasing temperature. Such a conclusion may
reflect interesting changes in the dynamics of adsorption and
reaction at different temperatures, which, in our opinion, de-
serves further investigation.
supply line behind the doser. Postmortem titration experi-
ments are then carried out by saturating the surface with CO
and following the desorption of CO2 while ramping the tem-
perature of the sample.
III. KINETIC MODEL AND SIMULATION METHOD
A lattice-gas model was used in our Monte Carlo kinetic
simulations where the surface of the Rh͑111͒ single crystal is
represented by a triangular lattice of N=LϫL adsorption
sites with periodic boundary conditions. The gas phase is
assumed to be pure N2O at a fixed pressure. Our reaction
scheme takes into account the following elementary pro-
cesses:
͑a͒ Adsorption of N2O molecules from the gas phase onto
the surface;
͑b͒ surface diffusion of adsorbed N2O and CO species ͑O
is considered to be strongly bonded to the surface, so
its diffusion is neglected͒;
͑c͒ desorption of adsorbed N2O and CO species from the
surface into the gas phase;
͑d͒ dissociation of N2O, when the presence of nearest
neighbor ͑nn͒ vacant sites makes it possible, followed
by immediate desorption of N2 into the gas phase; and
͑e͒ formation of CO2, when CO is available on the surface,
followed by immediate desorption.
II. EXPERIMENT
The experiments reported herein were carried out in a
molecular beam UHV apparatus described previously.25,26
The 6 l stainless-steel chamber is pumped with a 170 l/s
turbomolecular pump ͑Pfeiffer TPU 170͒ to a base pressure
of less than 2ϫ10−10 Torr, as measured using a Bayert-
Alpert ion gauge, and equipped with an ion gun ͑Leybold͒
for sample cleaning and a quadrupole mass spectrometer
͑UTI 100C͒ for gas analysis. Data acquisition is done with a
personal computer interfaced to the mass spectrometer,
which allows for the simultaneous monitoring of up to 15
different masses and the temperature of the Rh crystal during
each experiment. A rectangular-shaped ͑1.10ϫ0.56 cm2͒
Rh͑111͒ single crystal was mounted on a manipulator pos-
sessing rotation and translational motion along all three or-
thogonal axes. The sample is cooled via liquid-nitrogen-
cooled feedthroughs and heated resistively in order to
achieve any desirable temperature between 100 and 1200 K.
The temperature is monitored by a chromel-alumel thermo-
couple spot welded to the back side of the crystal, and set by
a homemade temperature controller. The Rh͑111͒ single-
crystal sample was cleaned before each experiments by se-
quential Ar+ ion sputtering, heating in a 1ϫ10−7 O2 atmo-
sphere, and annealing at 1200 K. The gases, argon ͑Liquid
Carbonic 99.999%͒, oxygen ͑Nellcor Purtain Bernett,
99.5%͒, and nitrous oxide ͑Matheson, 99%͒, were all used as
provided.
For the reaction of N2O on clean Rh͑111͒, a mechanism
was used that included the following elementary reaction
steps:
N2O͑g͒ + site → N2O͑ads͒,
͑1͒
͑2͒
͑3͒
N2O͑ads͒ → N2O͑g͒ + site,
N2O͑ads͒ + site → N2͑g͒ + O͑ads͒ + site.
In the case of N2O exposures on CO preadsorbed
Rh͑111͒, two additional steps were added,
CO͑ads͒ → CO͑g͒ + site,
͑4͒
͑5͒
CO͑ads͒ + O͑ads͒ → CO2͑g͒ + 2 sites.
Diffusion of adsorbed N2O and CO species, although not
listed above, was also taken into account by allowing for
activated jumps to nn vacant sites.
The occupancy of each site is described by an appropri-
ate label, and the state of the system determined by a set of N
such labels. The time evolution of the system is assumed to
occur as a Markovian stochastic process, where every el-
ementary reaction step has a rate constant associated with its
probability of occurrence per unit time, and a master equa-
tion describes the time evolution of the probability distribu-
tion of the states of the physical system,27
The kinetic experiments are carried out using a
1.2-cm-diameter multichannel microcapillary array doser.
The beam flux is controlled by setting the backing pressure
with a leak valve, and blocked and unblocked at will by
using a stainless-steel flag.25 The polished face of the
Rh͑111͒ crystal is first placed in front of the doser and heated
to a fixed temperature. The gas supply is turned on, and the
beam unblocked by removing the flag at t=0 s. The kinetics
of nitrogen production is followed versus time by mass spec-
trometry until the run is stopped via evacuation of the gas
ץP␣͑t͒
ץt
=
͓W␤→␣P␤͑t͒ − W␣→␤P␣͑t͔͒.
͑6͒
͚
␤
Here P␣͑t͒ and P␤͑t͒ are the probabilities of finding the sys-
tem in given configurations ␣ and ␤ at time t, respectively,
which can be considered as components of a vector P repre-

074705-3
N2O decomposition on Rh͑111͒
J. Chem. Phys. 125, 074705 ͑2006͒
TABLE I. kinetic parameters used in the Monte Carlo simulations.
senting the probability distribution of the whole set of system
configurations. W is the transition probability per unit time of
the process indicated in the subscript.
Preexponential
Activation energy
E ͑kcal mol−1
͒
Elementary step
factor ␯ ͑s−1
͒
This master equation can only be solved analytically in
very simple situations.28 In the case of the reaction consid-
ered here, where, as shown below, some processes depend on
the configuration surrounding a given site, it must be solved
through a dynamic Monte Carlo simulation. The simulation
can be performed through the “random selection
method,”29,30 which, as adapted to systems such as ours
where elementary step transition rates depend on temperature
and on the configuration around a given site, goes as follows:
N2O dissociation
N2O diffusion
N2O desorption
CO diffusion
1.3ϫ1010
5.0ϫ108
1.0ϫ1010
1.0ϫ1013
1.0ϫ1012
1.0ϫ1012
8.5
3.5
7.5
11.5
15.0
15.0
CO desorption
CO+O reaction
N2O initial sticking coefficient, S0 =0.5
͑a͒ r=͚Wi is assumed to be the total transition rate of the
system for all possible elementary steps, and R
=max͓r͔ the maximum total transition rate;
͑b͒ a surface site is selected at random with probability
1/N;
͑c͒ a given i-type reaction step ͑i.e., adsorption, desorp-
tion, dissociation, diffusion, etc.͒ is chosen at random
with probability Wi/r;
͑d͒ if the selected i-type reaction step is viable for the cho-
sen site, it is then executed with probability r/R;
͑e͒ after each attempt to modify the state of a given site,
the time is increased by ⌬t, as calculated by the for-
mula,
N2O dissociation. The dissociation of N2O, followed by
the immediate desorption of N2, is considered as an activated
process with a rate defined as in Eq. ͑10͒. This is the key step
in the whole reaction process. As we shall see in the follow-
ing section, it is not possible to reproduce the measured
amounts of adsorbed O at different temperatures unless an
additional condition is introduced for this elementary step.
This condition, as we shall see, is that the dissociation can
only take place if an increasing amount of nn vacant sites is
available to a given adsorbed N2O species as the temperature
is increased.
CO2 production. This process is possible when two nn
sites are occupied by CO and O species, respectively. It is
considered as an activated process with a rate given as in Eq.
͑10͒.
Table I shows the values of the parameters used for all
the steps in our simulations. These values are similar to those
reported in the literature.31 A lattice size of L=100 was cho-
sen to make sure that finite size effects are negligible ͑a fact
verified by independent simulations͒. Simulated quantities
were averaged in each time interval over 104 simulation runs.
ln ␰
⌬t = −
,
͑7͒
R
where ␰ is a random number selected according to a
uniform probability distribution in a ͑0,1͒ interval. This
equation renders the real time evolution caused by a
system transition.
The following elementary reaction steps and their corre-
sponding rate constant are considered in our simulations.
Adsorption. The impinging flux of molecules of species
i, Ji, from the gas phase onto the surface is given by
IV. RESULTS AND DISCUSSION
As stated in Sec. I, the present work was designed to try
to explain some surprising observations from our previous
molecular beam kinetic measurements on the decomposition
of N2O on Rh͑111͒ single-crystal surfaces. The key results
were presented in Figs. 2 and 5 of our past publication on
this subject.24 Figures 1 and 3 in this report display similar
raw data for the production of N2 as a function of reaction
time. In Fig. 1 two traces are displayed for each of two
temperatures, 350 and 520 K ͑the top curves in each of the
pairs in that figure͒. They are consistent with the previous
reported data, and highlight the two trends addressed by the
Monte Carlo simulations described below, namely, ͑1͒ the
initial rate of N2 production, indicated by the initial signal
rise at t=0 s, is approximately twice as large at 350 K than
at 520 K ͑within our experimental error, 5%͒, and ͑2͒ the
total N2 yield, obtained by integration of these traces over
the whole time interval of the study, also decreases at the
higher temperature, in this case to about 70% of the value at
350 K.
p
Ji =
,
͑8͒
1/2
͑2␲mikT͒
where p is the pressure in the gas phase, mi the mass of
species i, k the Boltzmann constant, and T the absolute tem-
perature. The adsorption rate Wi of particles impinging on
the surface per site and per second is then obtained as
Wi = JiAS0,
͑9͒
where A is the area per site on the surface and S0 the initial
sticking coefficient of the adsorbing molecule.
Desorption. Desorption is considered as an activated
process with a rate given by
Wi = ␯i exp͑− Eai/kT͒,
͑10͒
where ␯i is the frequency factor and Eai the activation energy
for species i.
Diffusion. Diffusion jumps are allowed for adsorbed
N2O and CO species to vacant nn sites. These jumps are
calculated as activated processes with rates defined as in Eq.
͑10͒.
The bottom traces in each set in Fig. 1 correspond to
subsequent kinetic runs carried out on the surfaces produced
by the first ͑top traces͒ without further modification, that is,
without cleaning of the surface. Those second runs were per-

074705-4
Omar Uñac et al.
J. Chem. Phys. 125, 074705 ͑2006͒
some to induce the formation of O islands. However, simu-
lations using those mechanisms resulted in a clear loss of
quality in the reproduction of the temporal behavior of the
N2O decomposition rate while not resolving the issue of this
high-temperature surface poisoning. Eventually, it was con-
cluded that the only way to correctly reproduce the observed
N2O decomposition rates, the overall nitrogen yields, and the
final oxygen coverages at different temperatures was to as-
sume an additional condition for the N2O dissociation: that
this elementary step can only occur if a mean number of nn
vacant sites, V͑T͒, is available to the adsorbed N2O mol-
ecule. It was then found that this number needs to increase
with temperature to reproduce the experimental values. V͑T͒
was used as the only fitting parameter in our simulations; all
other parameters were determined independently or taken
from the literature, and fixed at the values listed in Table I.
Interestingly, there is no longer a need to assume that N2O
desorption is more activated than its decomposition;
good results were obtained here even with an activation
energy difference between the two of Ea͑act͒−Ea͑des͒
= +1 kcal/mol.
Again, the general conclusion that the decrease in reac-
tivity with increasing temperature is the result of the dynam-
ics of the decomposition reaction, and not of a kinetic com-
petition with other steps, is borne by the data in Fig. 1. We
argue that if the time evolution of the kinetic behavior at any
given temperature depends on the evolution of the surface as
the reaction proceeds, preparing the surface at different tem-
peratures first should affect any subsequent rates in ways not
easily predicted by the first-order kinetics observed for this
reaction.24 Specifically, if oxygen islands were to form at low
temperatures but not at high temperatures ͑perhaps because
of a competition among mobility, decomposition, and de-
sorption steps͒, N2O decomposition could require the same
surface ensembles ͑sites͒ at all temperatures, only the surface
oxygen would not block the surface randomly at low tem-
peratures, and would leave more open sites for reaction.
However, in that case the surfaces obtained after reactions at
high temperatures should lose their subsequent activity at all
temperatures, since at that point the oxygen atoms would
already be randomly distributed on the surface. This is not
what is observed in Fig. 1. On the other hand, if there is a
need for larger surface ensembles for N2O decomposition at
high versus low temperatures, as the Monte Carlo simula-
tions indicate, a small coverage of randomly dispersed oxy-
gen would be enough to kill the reaction at high temperatures
but not at low temperatures; there should still be space for a
low-temperature reaction requiring smaller ensembles. Fig-
ure 1 shows that this is indeed the case.
FIG. 1. Isothermal molecular beam kinetic data for the decomposition of
N2O on a Rh͑111͒ single-crystal surface. Shown are the rates of molecular
nitrogen production as a function of time for four sets of double-run experi-
ments. In each pair an initial run was carried out on the clean surface at
either one of two temperatures, 350 ͑two bottom pairs͒ or 520 ͑two top
pairs͒ K; the results are reported in the top traces of each pair. A second run
was then recorded immediately afterwards, without cleaning the surface, but
after changing the temperature. Those results are shown in the lower traces
of each pair. The data in this figure indicate that N2O decomposition is
slower, produces less N2, and deposits less atomic oxygen on the surface at
higher temperatures. It also points to the fact that the inactive surfaces
obtained at the end of the N2O conversion at high temperatures can still
promote additional decomposition at lower temperatures.
formed in order to probe the temperature dependence of the
reactivity of the surfaces saturated with adsorbed oxygen
from N2O decomposition at different temperatures, to test
the hypothesis presented in Sec. I. The data indicate that
while the surface prepared at 350 K displays no significant
subsequent reactivity at either 350 or 520 K ͑two bottom
cases in Fig. 1͒, the one resulting from saturation with N2O
at 520 K is still quite active at 350 K: the data in the third set
from the bottom in Fig. 1 indicate a similar initial rate for the
second run at 350 K as for the first at 520 K. In other words,
surfaces saturated with oxygen atoms from N2O decomposi-
tion at high temperatures are still active for that reaction at
lower temperatures. On the other hand, the opposite is not
true: surfaces saturated at low temperatures are not active at
higher temperatures. This observation is central to our dis-
cussion below. More generally, the reaction rates obtained in
the second runs scale quite nicely with the fraction of empty
sites left after the first runs ͑as calculated from the N2 yields;
data not shown͒, as expected from the first-order kinetics
followed by this reaction.24
The experimental results reported before for the kinetics
of N2O decomposition on Rh͑111͒ ͑Ref. 24͒ could be satis-
factorily reproduced by simulations with the values of V͑T͒
shown in Fig. 2. A value of V͑T͒=0 was obtained at 335 K,
indicating that N2O can dissociate on the same site on which
it is adsorbed, without the need of any additional nn vacant
sites. V͑T͒ then increases with temperature until reaching a
maximum of six around 795 K, which in a triangular lattice
such as that in Rh͑111͒ represents the maximum number of
nn atoms around a given site. A continuous line has been
Monte Carlo simulations were carried out to try to repro-
duce and explain the experimental findings obtained by us
here and by Wehner et al.24 First, to explain the decrease in
nitrogen yield and oxygen deposition seen with increasing
reaction temperature, several kinetic mechanisms were at-
tempted based on the idea of larger N2O desorption prob-
abilities before dissociation at higher temperatures, including

074705-5
N2O decomposition on Rh͑111͒
J. Chem. Phys. 125, 074705 ͑2006͒
FIG. 2. Mean number of nn vacant sites, V͑T͒, necessary for N2O dissocia-
tion at different temperatures, as estimated from Monte Carlo simulations
͑symbols͒. The solid line represents a fit to the data, and is provided only to
help visualize the trend observed. Higher values of V͑T͒, that is, larger
surface ensembles, are needed for N2O decomposition on Rh͑111͒ at higher
temperatures.
FIG. 3. Time evolution of isothermal rates of N2 production from N2O
decomposition on clean Rh͑111͒. Data from Monte Carlo simulations ͑solid
lines͒ are contrasted with results from molecular beam experiments ͑Ref.
24͒ for different temperatures. Good agreement is seen in all cases. A N2O
pressure of 1.5ϫ10−7 torr in the gas phase was used in the simulations.
fitted to the calculated values of V͑T͒ obtained in our simu-
lations for the five temperatures used in those experiments
and included in Fig. 2, although it is not clear from our
calculations if the surface ensemble size required for reactiv-
ity increases smoothly or in a stepwise manner as a function
of temperature. Nevertheless, based on our calculations, we
favor the former possibility. In most of our simulations the
number on nn required for reactivity was kept constant
throughout the time evolution of the reaction ͑although a
different value was used for each different temperature͒, but
in the case of 639 K a different tactic was employed where
the number of near neighbors required for dissociation was
chosen randomly to be either 5 or 6; that is how the frac-
tional number of 5.5 was obtained. This opens the possibility
of obtaining other fractional numbers for V͑T͒ at other tem-
peratures, and justifies the smooth nature of the curve in Fig.
2. Unfortunately, we do not have a good explanation for this
behavior ͓nor for one where the changes in V͑T͒ are dis-
crete͔.
lower N2O coverage on the surface and therefore fewer mol-
ecules available for dissociation ͑see Fig. 4͒, ͑2͒ N2O desorp-
tion, which increases with temperature and contributes to the
same trend as above, and ͑3͒ N2O dissociation, which in-
creases with temperature and counterbalances that behavior.
As it can be seen in Fig. 5, the prevalence of a combi-
nation of the first two effects leads to an effective N2O de-
sorption rate that decreases with increasing temperature.
Semilogarithmic plots of the N2 production rate versus time
such as that shown in Fig. 6 for T=335 K evidence the first-
order behavior of this reaction, as also found experimentally.
Once the variable reaction site size behavior was estab-
lished, the kinetics of the N2O conversion reaction was suc-
cessfully simulated. The results obtained are described next.
A. N2O decomposition on clean Rh„111…
Figure 3 shows the simulated N2 production rate as a
function of time obtained at different temperatures ͑solid
lines͒. These traces follow reasonably well the experimental
results,24 which are shown for the three lowest temperatures
as dots in Fig. 3. The agreement is particularly good consid-
ering that only one adjustable parameter, V͑T͒, was used in
all these calculations. Closer inspection of the simulated sur-
faces provided some insights into the reasons for the ob-
served behavior. For one, the decreases in production rate
with increasing temperature seen at all times during the re-
action were determined to be entirely due to a balance
among three elementary step rates, namely, ͑1͒ N2O adsorp-
tion, which decreases with increasing temperature, yielding a
FIG. 4. Time evolution of the N2O coverage on the surface at different
temperatures. These data correspond to the same simulations as those in Fig.
3. They point to a decrease in N2O adsorption rate with increasing
temperature.

074705-6
Omar Uñac et al.
J. Chem. Phys. 125, 074705 ͑2006͒
FIG. 5. N2O desorption rates as a function of time from the same simula-
tions. These data attest directly to the increase in N2O desorption rate with
increasing temperature.
FIG. 7. Calculated evolution of the coverage of atomic oxygen on the sur-
face as a function of time at different temperatures. These coverages are
always lower at higher reaction temperatures.
The behavior of the oxygen coverage as a function of
time for the different temperatures studied is shown in Fig. 7.
From these data, the final oxygen coverages for the different
temperatures can be obtained and compared to the experi-
mental values, as obtained by both integration of the N2 yield
over time and directly by CO temperature programmed de-
sorption ͑TPD͒ titrations after the isothermal kinetic runs.24
This is done in Fig. 8, which indeed shows an excellent
agreement between simulations and experiments. It can be
seen there that the final oxygen coverage decreases signifi-
cantly with increasing temperature. Also, only a small
amount of oxygen is sufficient to poison the reaction at high
temperatures, as also found experimentally ͑Fig. 1͒.24 This is
entirely consistent with our fundamental hypothesis. The
spatial distribution of the deposited oxygen in the final stages
of the reaction at different temperatures is visualized in the
simulation snapshots of the surface displayed in Fig. 9. Ran-
dom distributions are observed in all cases, but perhaps more
clearly at higher temperatures. The higher coverages ob-
tained at lower temperatures are also evident in that figure.
B. N2O decomposition on Rh„111… with preadsorbed
carbon monoxide
The same model, with the same parameters, was also
tested for the decomposition of N2O on Rh͑111͒ surfaces
predosed with carbon monoxide. The N2 and CO2 production
rates calculated in these systems are shown as a function of
time in Fig. 10 for different reaction temperatures. These
data also exhibit quite good agreement with the previous
experimental results,24 and provide further credence to our
model. Again, these simulations involved no additional ad-
justable parameters.
FIG. 8. Comparison between experimental and simulated values for the final
oxygen coverages after N2O decomposition on Rh͑111͒ at different tempera-
tures. The lower saturation coverages seen at higher temperatures is one of
the surprising observations that could be explained by the Monte Carlo
kinetic modeling.
FIG. 6. Semilogarithmic plot of the rate of N2 production from N2O decom-
position on Rh͑111͒ at 335 K. These data, also calculated from the simula-
tions discussed above, corroborate the first-order kinetics seen experimen-
tally. The parameters from a straight-line fit to the points are also provided.

074705-7
N2O decomposition on Rh͑111͒
J. Chem. Phys. 125, 074705 ͑2006͒
FIG. 9. Surface snapshots from the Monte Carlo simulations at the end of
the N2O conversion. Pictures of the spatial distributions of the adsorbed
oxygen and N2O are shown here for four different temperatures. Random
distributions, without island formation, were obtained in all cases, but sig-
nificantly higher oxygen coverages were obtained at lower temperatures.
FIG. 10. N2 and CO2 production rates as a function of time at different
temperatures, calculated using the same simulation parameters as above.
These data correspond to N2O dosing at 2.5ϫ10−8 torr on surfaces presatu-
rated with CO. Excellent agreement was once again obtained with the ex-
perimental data ͑Ref. 24͒ without the need of any additional adjustable
parameters.
V. CONCLUSIONS
A model for the kinetics of the decomposition of N2O on
Rh͑111͒ single-crystal surfaces has been proposed, and dy-
namic Monte Carlo simulations have been performed to re-
produce and explain the behavior observed in past and
present experiments. The experimental data had led to the
identification of two interesting and puzzling observations,
namely, ͑1͒ that the rate of decomposition of N2O͑ads͒ into
N2͑g͒+O͑ads͒ decreases with increasing temperature at all
times during the reaction, and ͑2͒ that both the total nitrogen
yield and the coverage of O͑ads͒ on the surface at the end of
the reaction decrease with increasing temperature. The simu-
lations reported here have pointed to the necessity of intro-
ducing a strong hypothesis in any kinetic model to be able to
fit the experimental data: N2O dissociation cannot take place
at higher temperatures unless vacant sites of increasing area
are available on the surface. This severe condition was tested
both by new molecular beam kinetic experiments specifically
designing to discard other hypotheses and a number of
Monte Carlo simulations. On the basis of this hypothesis,
satisfactory agreement was obtained between the simulated
and experimental kinetics seen for this system.
on kinetics alone. Following our past experience with this
type of reactions,18,32 we paid particular attention to any pos-
sibility of clustering of adsorbates, island formation, or other
interadsorbate interactions that may result in their heteroge-
neous distribution on the surface, all without success. It
could very well be that the incoming N2O molecules require
larger surface sites at higher surface temperatures to dissipate
their excess translational and/or internal energy. This is
needed for adsorption and reaction. It is known that such
dissipation is often aided by larger differences between the
temperatures of the gas molecules and the surface,33,34 and
could therefore be more efficient ͑and therefore require
smaller number of surface atoms͒ at low temperatures. Un-
fortunately, this idea is quite difficult to test experimentally
in this system. We hope that our results may encourage the-
oretical quantum mechanical calculations to explain the rea-
sons for this variable reaction-site size for N2O dissociation
on Rh͑111͒ from first principles.
We are somewhat reluctant to advance a model to ex-
plain the preceding conclusions at the present time. Based on
our simulations, we believe that the effects seen experimen-
tally need to be explained by arguments based on the dynam-
ics of the adsorption of N2O on Rh͑111͒. Certainly, we were
not able to come up with a satisfactory model for this based
ACKNOWLEDGMENTS
This work was supported by the U.S. National Science
Foundation and by the Consejo Nacional de Investigaciones
Científicas y Técnicas ͑CONICET͒ of Argentina.

074705-8
Omar Uñac et al.
J. Chem. Phys. 125, 074705 ͑2006͒
1 K. C. Taylor, Automobile Catalytic Converters ͑Springer-Verlag, Berlin,
1984͒.
18 F. Zaera, S. Wehner, C. S. Gopinath, J. L. Sales, V. Gargiulo, and G.
Zgrablich, J. Phys. Chem. B 105, 7771 ͑2001͒.
2 R. J. Farrauto, R. M. Heck, and B. K. Speronello, Chem. Eng. News 7,
34 ͑1992͒.
19 C. S. Gopinath and F. Zaera, J. Catal. 186, 387 ͑1999͒.
20 F. Zaera and C. S. Gopinath, Phys. Chem. Chem. Phys. 5, 646 ͑2003͒.
21 F. Zaera and C. S. Gopinath, J. Chem. Phys. 116, 1128 ͑2002͒.
22 V. Bustos, R. Uñac, F. Zaera, and G. Zgrablich, J. Chem. Phys. 118, 9372
͑2003͒.
3 W. C. Becker and A. T. Bell, J. Catal. 84, 200 ͑1983͒.
4 B. K. Cho, J. Catal. 148, 697 ͑1994͒.
5 C. T. Williams, A. A. Tolia, H. Y. H. Chan, C. G. Takoudis, and M. J.
Weaver, J. Catal. 163, 63 ͑1996͒.
23 L. A. Avalos, V. Bustos, R. Uñac, F. Zaera, and G. Zgrablich, J. Mol.
Catal. A: Chem. 228, 89 ͑2005͒.
6 S. B. Schwartz, L. D. Schmidt, and G. B. Fisher, J. Phys. Chem. 90,
6194 ͑1986͒.
24 S. Wehner, M. T. Paffett, and F. Zaera, J. Phys. Chem. B 108, 18683
͑2004͒.
7 S. H. Oh, G. B. Fisher, J. E. Carpenter, and D. W. Goodman, J. Catal.
100, 360 ͑1986͒.
25 J. Liu, M. Xu, T. Nordmeyer, and F. Zaera, J. Phys. Chem. 99, 6167
͑1995͒.
8 C. H. F. Peden, D. W. Goodman, D. S. Blair, P. J. Berlowitz, G. B. Fisher,
and S. H. Oh, J. Phys. Chem. 92, 1563 ͑1988͒.
9 M. J. P. Hopstaken, W. J. H. van Gennip, and J. W. Niemantsverdriet,
Surf. Sci. 435, 69 ͑1999͒.
26 F. Zaera, Int. Rev. Phys. Chem. 21, 433 ͑2002͒.
27 N. G. van Kampen, Stochastic Processes in Physics and Chemistry
͑North-Holland, Amsterdam, 1992͒.
10 F. Zaera and C. S. Gopinath, J. Mol. Catal. A: Chem. 167, 23 ͑2001͒.
11 J. Perez-Ramirez, F. Kapteijn, K. Schoffel, and J. A. Moulijn, Appl.
Catal., B 44, 117 ͑2003͒.
28 H. Hinrichsen, Adv. Phys. 49, 815 ͑2000͒.
29 K. Binder, Monte Carlo Methods in Statistical Physics ͑Springer, Berlin,
1986͒.
12 B. K. Cho, J. Catal. 138, 255 ͑1992͒.
30 J. L. Sales, R. O. Uñac, M. V. Gargiulo, V. Bustos, and G. Zgrablich,
Langmuir 12, 95 ͑1996͒.
13 H. Permana, K. Y. S. Ng, C. H. F. Peden, S. J. Schmieg, D. K. Lambert,
and D. N. Belton, J. Catal. 164, 194 ͑1996͒.
31 V. P. Zhadanov and B. Kasemo, Surf. Sci. Rep. 29, 31 ͑1997͒.
32 M. Xu, J. Liu, and F. Zaera, J. Chem. Phys. 104, 8825 ͑1996͒.
33 J. A. Barker and D. J. Auerbach, Surf. Sci. Rep. 4, 1 ͑1985͒.
34 S. T. Ceyer, J. D. Beckerle, M. B. Lee, S. L. Tang, Q. Y. Yang, and M. A.
Hines, J. Vac. Sci. Technol. A 5, 501 ͑1987͒.
14 P. Granger, J. J. Lecomte, C. Dathy, L. Leclercq, and G. Leclercq, J.
Catal. 175, 194 ͑1998͒.
15 F. Zaera and C. S. Gopinath, Chem. Phys. Lett. 332, 209 ͑2000͒.
16 F. Zaera and C. S. Gopinath, J. Chem. Phys. 111, 8088 ͑1999͒.
17 C. S. Gopinath and F. Zaera, J. Phys. Chem. B 104, 3194 ͑2000͒.