N2 Product Internal-State Distributions for the Steady-State Reaction of NO with H2 and NH3 on the Pt(100) Surface

Full text

Page 1

N2Product Internal-State Distributions for the Steady-State Reactions of NO with H2and
NH3on the Pt(100) Surface†
Alexander J. Hallock, Carl M. Matthews,‡Frank Balzer,‡and Richard N. Zare*,§
Department of Chemistry, Stanford UniVersity, Stanford, California 94305
ReceiVed: March 5, 2001; In Final Form: May 16, 2001
The steady-state reaction of NO with H2and NH3on Pt(100) is investigated over a temperature range of
340-570 K, and the rotational and vibrational states of the N2product are probed by resonance enhanced
multiphoton ionization. For NO + NH3the N2reaction product leaves the surface rotationally and vibrationally
excited, whereas for NO + H2the desorbing N2is rotationally equilibrated with the surface and no vibrational
excitation can be detected. We conclude that although these two surface reactions both yield N2, the actual
reaction mechanisms must be different.
Introduction
For reactions on well-defined surfaces knowledge of the
internal energy distribution of one or all reaction products might
give valuable hints to clarify transition states and verify models
for potential energy surfaces. Hydrogen recombination reactions
on various surfaces1provide examples of the information that
can be gained from such studies. For heavier molecules, such
as N2, the situation is expected to be more complex but still
can give insight into the shape of potential energy surfaces.2
In this study we examine the internal energy distribution of
the N2 product for two different precursor reactions on Pt-
(100): NO + NH3 and NO + H2. These catalytic reactions
exhibit several notable features such as surface explosion,3
hysteresis in reaction rates, and kinetic oscillations.4It is
believed that for both systems the recombination of two nitrogen
atoms on the Pt(100) surface leads to formation of N2, Na+ Na
f N2,g(with the subscript a for adsorbed species, and subscript
g for gaseous). The N2subsequently desorbs from the surface.
We show that the internal energy distributions of N2resulting
from these two precursors are not the same, indicating that
simple recombination models are inadequate. These differences
might be interpreted as different transition states or different
coverages for the two reactions.
Experimental Apparatus
In an ultrahigh vacuum (UHV) chamber5,6a constant pressure
of reactant gases is maintained. The chamber achieves a base
pressure of 5 × 10-10Torr by means of a liquid nitrogen trapped
diffusion pump. A Pt(100) single crystal is fixed to a three-
axis manipulator. This setup allows accurate positioning for the
use of Auger, low-energy electron diffraction (LEED), and time-
of-flight mass spectrometry (TOF MS). The sample can be
heated to 1100 K by means of a filament located behind it. The
sample can also be cooled to 170 K using liquid nitrogen. The
temperature of the sample is monitored by a chromel-alumel
thermocouple, spot-welded to its front face. The temperature
of the Pt(100) single crystal is cycled, and product molecules
are detected both by a quadrupole mass spectrometer (QMS,
Stanford Research Systems RGA200) and by resonance en-
hanced multiphoton ionization (REMPI) followed by time-of-
flight (TOF) mass spectrometry. The latter scheme provides
internal state identification. An advantage of our setup is that
the TOF scans are run simultaneously with the QMS to ensure
that the background gas pressures are approximately constant
during the course of an experiment and to correct REMPI spectra
with regard to small fluctuations in the reaction rates.
A Pt(100) surface adopts one of three arrangements depending
on preparation conditions, temperature, and surface adsor-
bates: (1 × 1), hex, and hex-R0.7°.7,8The hex-R0.7° face is
the most stable configuration of a clean Pt(100) surface. The
hex face is metastable and converts to the hex-R0.7° phase at
temperatures above 1100 K. For both arrangements the surface
atoms form a hexagonal structure, which is slightly buckled.
The presence of adsorbates such as NO, CO, and O2revert the
hex or hex-R0.7° to the bulk-terminated (1 × 1) configuration.9
These three arrangements can be easily distinguished by
LEED.10In our experiments only the hex- and (1 × 1)-face
were observed, as expected from our preparation conditions.
The platinum surface was cleaned by sputtering with Ar+
ions at 1 keV while annealing at 900 K for 30 min. Auger
spectra demonstrated that carbon is the most frequent contami-
nant. The carbon was removed by cooling the surface from 900
K (postannealing) to 470 K in 1.5 × 10-5Torr of O2to oxidize
the carbon on the surface to CO and CO2. After this procedure,
the surface was flashed to 900 K to desorb any remaining
oxygen. Subsequent Auger scans verified that the carbon had
been removed. In this way a clean Pt(100)-hex surface was
prepared. We were able to reproduce the published LEED
pattern of the hex reconstructed Pt(100) surface,10usually with
one predominant domain, and match previously published O2,
NO, and H2desorption.11
To detect N2 molecules in a state-selective manner, we
employed a (2+1) REMPI scheme.12,13Molecular nitrogen ions
were produced by two-photon excitation from the X1Σ+g
electronic ground state to the a′′1Σ+gRydberg state, followed
by one-photon ionization. We probed the (0,0) and (1,1)
Q-branch lines because this branch has the highest intensity and
is insensitive to molecular alignment. Tunable laser light around
202 nm was produced using a Nd:YAG pumped dye laser and
two BBO crystals,6guided into the UHV chamber, and focused
†Part of the special issue “Royce W. Murray Festschrift”.
* Corresponding author. Fax: +1-650-725-0259. Email address: zare@
stanford.edu.
‡Present address: Intel Corporation, Santa Clara, CA.
§Present address: Odense Universitet, Fysisk Institut, Campusvej 55,
DK-5230 Odense M, Denmark.
8725
J. Phys. Chem. B 2001, 105, 8725-8728
10.1021/jp0108216 CCC: $20.00 © 2001 American Chemical Society
Published on Web 06/23/2001

Page 2

about 2 cm in front of the surface. Nitrogen ions are generated
at the entrance of a TOF tube and steered by various charged
plates to a 40 mm multichannel plate array. The signal is
acquired and sent to a computer via a digital oscilloscope.
Experiments show that the laser power needs to be normalized
with an exponent of 1.5, which is the typical behavior for a
(2+1) REMPI transition showing geometric saturation.14Peak
heights are then corrected for nuclear spin statistics gJ, rotational
degeneracy (2J+1), and Bray-Hochstrasser line strength fac-
tors.15If the distribution is characterized by a rotational
temperature, the Boltzmann plot (the natural logarithm of
corrected peak heights against rotational quantum number) is a
straight line. We find that all data recorded are well described
by linear Boltzmann plots, and therefore we report rotational
temperatures.
We used high-purity gases for the experiments (H299.999%,
NO 99.5%, NH399.999%). The major contaminants in the NO
were N2(<3000 ppm) and CO2(<1000 ppm). A certain gas
ratio was mixed in a stainless steel cylinder, and a total pressure
of 3.7 × 10-7Torr (uncorrected for differences in ion gauge
sensitivity) was put into the chamber with the clean Pt(100)
hex surface held at 900 K. Immediately afterward, the sample
was cooled to 230 K. In this way we prepare a surface that is
almost completely covered with NO. The ratios are chosen to
maximize N2production. We used a 20% NO/80% H2mixture
in the NO + H2case resulting in a sensitivity-corrected pressure
as measured by ion gauge of 5.9 × 10-7Torr, and 50% NO/
50% NH3in the NO + NH3case resulting in 2.9 ×10-7Torr.
The rotational excitation of the desorbing molecules is defined
in terms of a mean rotational energy per molecule for each
rotational distribution. In the case where a Boltzmann plot is
linear, a characteristic rotational temperature can be assigned
to the molecules. We found our distributions to be linear,
although they can be prone to errors arising from noise in the
high-J populations. To avoid this issue, the laser scan was made
over a sufficiently large number of states, often out to J ) 24
or further. During the experiments the surface temperature was
cycled from 230 to 620 K (heating rate 0.5 K/s) and back again
(cooling rate 1 K/s), while a constant pressure of adducts was
maintained. To obtain rotational and vibrational distributions
the temperature was selected and held constant while a REMPI
spectrum was taken.
Results
A typical N2 REMPI spectrum yields ion intensity as a
function of dye fundamental wavelength. In Figure 1a the lower
curve shows a typical REMPI spectrum for a mixture of NO
and H2in the presence of the Pt(100) surface, probing N2in
the vibrational ground state V ) 0. The surface temperature was
held at 273 K, i.e., no catalytic activity is present. The N2stems
mainly from N2contamination of NO (approximately 3000 ppm)
and from ambient gas in the chamber. Figure 1b presents the
Boltzmann plot, from which a linear regression yields a
rotational temperature of Trot) 293 ( 25 K. The upper curve
in Figure 1a represents a REMPI spectrum for a surface
temperature of 503 K, i.e., with catalytic activity present (circles
in Figure 1b). After subtracting the background a rotational
temperature can be extracted. Vibrationally excited states are
only sparsely populated at room temperature; hence, reactive
measurements for V ) 1 or higher are essentially background
free.
NO + NH3. The primary reaction products for this reaction
are N2and H2O, as shown in Figure 2. Catalytic activity starts
around 400 K, where the QMS shows a sharp rise in both the
N2signal at mass 28 and the H2O signal at mass 18. As the
surface temperature increases both signals remain on a high level
up to 500 K, then drop and almost disappear around 600 K.
This behavior is similar to previously observed temperature
dependencies.16The cooling curve is not shown in the figure,
because the catalytic activity remains on a low level and we
did not perform any REMPI measurements. This behavior is
demonstrative of the reaction hysteresis. Product nitrogen is only
formed in significant quantities while heating the surface over
a certain temperature range in the presence of reactants. A
typical spectrum took on the order of 2 h to acquire, during
which time the gas concentrations were all monitored to ensure
that the reaction was indeed steady-state.
In Figure 3a rotational temperature (circles) of desorbing N2
molecules in V ) 0 are presented as a function of surface
temperature. Only between 416 and 485 K was it possible to
find a stable reaction rate with enough product ion intensity to
acquire a complete REMPI spectrum. The solid line denotes
the rotational temperature that would result from a complete
equilibration of the molecules with the surface. The rotational
temperatures for V ) 0 are far above thermal and increase with
Figure 1. (a) The upper curve is a typical REMPI spectrum of N2
from the NO + H2reaction with a surface temperature of 503 K. The
lower curve is a spectrum of background N2 in the chamber. (b) A
Boltzmann plot of raw data (circles), the background showing a
temperature of 293 ( 25 K (triangles), and the resulting subtraction
(squares).
Figure 2. Rates of product formation for the quasi steady-state (heating
rate 0.5 K/s) reaction NO + NH3on Pt(100) as a function of surface
temperature. The traces represent mass 28 and 18, respectively.
8726 J. Phys. Chem. B, Vol. 105, No. 37, 2001
Hallock et al.

Page 3

surface temperature. Vibrational excitation of N2 is clearly
visible, too. From the population ratio P(V)1)/P(V)0) ≈ 0.2
we estimate a vibrational temperature on the order of 2000 K.
Rotational temperatures for V ) 1 are presented in Figure 3b.
We clearly see that the rotational temperature increases with
surface temperature, but the energies are closer to thermal as
compared to V ) 0.
NO + H2. For the NO + H2reaction three main reaction
products occur, namely N2, NH3, and H2O, as shown in Figure
4. N2production shows two peaks on the heating cycle. We
clearly see competition between N2 and NH3 formation-
ammonia production starts with a decrease in nitrogen signal
and slows down as nitrogen slowly returns. Other investigators
have also observed this change in selectivity.17
Rotational distributions of the N2 product are taken for
different surface temperatures ranging from 343 to 566 K.
Although in Figure 4 catalytic activity is only observed between
400 and 600 K, this range can be extended to lower temperatures
by starting the cooling cycle before the transition to the hex
phase takes place; see the inset in Figure 4. Whereas catalytic
activity is hindered during the heating cycle because of the high
surface coverage with NO (see below), once the reaction started
the surface coverage remains small and the reaction can proceed
even at low temperatures.
The reaction exhibits hysteresis over the entire temperature
range investigated. We wanted to see if N2production in the
cooling direction looked rotationally different than N2production
when heating. N2was produced in barely detectable amounts
which showed up as a broad peak around 150 K. The extremely
small amount of product produced lead to noisy spectra. We
conclude that there is no large change from a thermal distribution
so the formation mechanism might be the same.
The squares in Figure 3a show the temperature dependence
of the rotational temperature of N2. In contrast to N2from NO
+ NH3the molecules seem to be rotationally accommodated
with the surface. After extensive searching for both the V ) 1
and V ) 2 band heads, we conclude if vibrational excitation
exists, it is below our ability to detect it. We estimate that this
would make the vibrational temperature well below 1000 K.
This result is surprising because a simple recombination of N
atoms to form N2 would be expected to produce much
vibrationally excited molecules.
Discussion
Previous experiments on the NO + H2 and NO + NH3
reactions have revealed largely similar mechanisms.4,18In
contrast, our study of the internal state distributions of the
resulting N2product reveals quite different behavior. Indeed,
this work establishes the value of being able to observe in detail
the escaping products of a well-defined surface reaction.
Two key steps dictate the temperature dependence of the
reaction rates: NO dissociation and the (1 × 1) T hex phase
transition. Up to 350-400 K, a high NO coverage stabilizes
the (1 × 1) Pt(100) surface. NO dissociation, which occurs
above 380 K,19is the rate-limiting step for the reactions. The
reactivity of the NO covered surface is low, because for NO
dissociation a free adsorption site is necessary. Free adsorption
sites occur only above 400 K for which temperature desorption
of adsorbed NO and NH3or H2takes place. Consequently, the
reactions can only start above this temperature. The reaction
product water is weakly bound to the surface and desorbs
immediately after formation.20The only important pathway for
forming N2seems to be the recombination of N atoms on the
surface.18Atomic N has a finite residence time on the surface.21
Additional free adsorption sites are created via desorption of
reaction products, and the reaction proceeds rapidly. The surface
coverage with NO decreases, whereas the coverage with
intermediates like NHx(x)1-3) (known from work function
measurements) increases. Between 500 and 600 K the phase
transition from the (1 × 1) to the hex surface takes place. The
hex phase is not active in NO dissociation3with the exception
of edge defects.22Consequently the reactivity ceases.
Figure 3. (a) Mean rotational energies as a function of surface
temperature for desorbing N2molecules for the vibrational ground state
V ) 0. The white squares denote the reaction of H2with NO (1.18 ×
10-7Torr and 4.72 × 10-7Torr respectively), the filled circles the
reaction of NH3with NO (both 1.45 × 10-7Torr). (b) Mean rotational
energy for the reaction NH3with NO for V ) 1. In both cases the gray
line represents surface temperature (fully accommodated molecules).
Figure 4. Same as Figure 2, but for the reaction NO + H2. The traces
represent mass 28, 18, and 17, respectively. The insert demonstrates
how to get catalytic activity for surface temperatures lower than 400
K by cooling before the surface phase transition to the Pt hex phase
takes place.
N2Product Internal-State Distributions
J. Phys. Chem. B, Vol. 105, No. 37, 2001 8727

Page 4

N2 itself does not adsorb on the Pt(100) face at room
temperature.21King et al.23,24demonstrated that the sticking
probability for N2on Pt(100) is zero for translational energies
up to 3 eV, and they concluded that recombinative desorption
on Pt(100) proceeds directly through a late barrier, without
trapping and equilibration to the surface temperature. This
mechanism results in vibrationally excited gaseous N2. In
agreement with this mechanism, Foner et al.25found evidence
for excitation up to V ) 9 for N2 recombination after NH3
dissociative adsorption on polycrystalline platinum.
Vibrationally excited N2has been observed in recombination
reaction on several other metals like sulfur covered polycrys-
talline iron,26on Cu(111),27Ru(001),28and Ag(111).29In each
of these systems, the vibrational temperatures are much higher
than the surface temperature. The transition states of these
reactions must feature atoms that are separated by distances
greater than equilibrium bond length. Furthermore, the very large
vibrational spacing in the N2 molecule (approximately 2360
cm-1) precludes rapid equilibration with the surface.
Because the N atoms are thought to recombine for both
precursors as they might do in a simple recombination scenario
we expected to observe vibrationally excited N2product in the
NO + H2and the NO + NH3reactions. We also expected to
observe rotationally excited N2products because of the large
nitrogen-nitrogen distance in the transition state. For the
reaction between NO and NH3these expectation were met for
this case, but for NO + H2surprisingly they were not. Instead
the N2seems to be completely equilibrated (at least rotationally)
with the surface. We conclude that two different mechanisms
are operative but presently we can only speculate what might
be the reasons for this behavior. It is obvious that the source of
N atoms on the Pt surface can be different for both systems.
For NO + H2 the N atoms stem solely from dissociation of
NO, but for NO + NH3two possible origins exist. Imbihl et
al.16studied the NO + NH3reaction via isotopically labeled
adducts and demonstrated that the desorbing N2 is produced
entirely from NO only for temperatures above 500 K. For lower
temperatures, N2stems either from dissociated NH3alone, or
from NO and NH3. For this reaction we only could take REMPI
spectra to a surface temperature of 485 K due to decreasing N2
production and increasing backgrounds. This means we probe
N2that is made of at least one nitrogen from dissociated NH3,
not entirely from NO. Consequently, the possibility exists that
the N2we observe comes from another reaction like Na+ NOa
f N2,g+ Oa, which would account for the different internal
state distribution. For N2formation during NO reduction at Pd-
(110)30these two N2forming reactions have been identified to
lead to different angular distributions of desorbing N2and to
different state distributions, although on Pt(335) Na+ NOahas
not been seen.31Another possibility could be that different
coverages of the surface with reaction adducts or intermediates
as NHx, structural surface defects especially in the vicinity of
adduct islands or the formation of complexes like NOa-NHx,a16
influence the height of the barrier to the Na+ Narecombination.
This second possibility cannot be ruled out but seems to be
less attractive than the first.
Conclusions
Our study shows that the NO + H2and NO + NH3reactions
behave in similar fashions but they reach their respective final
states through different intermediates. We base this conclusion
on the different distributions of energies in the N2 product.
Traditional methods for probing surface chemistry are blind to
this type of effect. The precise mechanism accounting for the
different intermediates still needs clarification, but the measure-
ment of N2product internal energy distributions for NO + H2
and NO + NH3on Pt(100) shows unequivocally the need for
further study of these reaction systems.
Acknowledgment. C.M.M. thanks the Hertz-Foundation,
and F.B. thanks the Deutsche Forschungsgemeinschaft (Ba 1858/
1-1 and Ba 1858/1-2) for financial support. This study was
supported by the U.S. National Science Foundation under Grant.
CHE-9900305.
References and Notes
(1) Bent, S. F.; Michelsen, H. A.; Zare, R. N. Laser Spectroscopy and
Photochemistry on Metal Surfaces; World Scientific: Singapore, 1995.
(2) Hodgson, A. Prog. Surf. Sci. 2000, 1, 63.
(3) Lesley, M. W.; Schmidt, L. D. Surf. Sci. 1985, 155, 215.
(4) Lombardo, S. J.; Fink, T.; Imbihl, R. J. Chem. Phys. 1993, 98,
5526.
(5) Ellison, M. D.; Matthews, C. M.; Zare, R. N. J. Chem. Phys. 2000,
112, 1975.
(6) Matthews, C. M.; Balzer, F.; Hallock, A. J.; Ellison, M. D.; Zare,
R. N. Surf. Sci. 2000, 12, 460.
(7) Kuhnke, K.; Kern, K.; Comsa, G.; Moritz, W. Phys. ReV. B 1992,
45, 14388.
(8) Ritz, G.; Schmid, M.; Varga, P.; Borg, A.; Ronning, M. Phys. ReV.
B 1997, 56, 10518.
(9) Norton, P. R.; Davies, J. A.; Creber, D. K.; Sitter, C. W.; Jackman,
T. E. Surf. Sci. 1981, 108, 205.
(10) Hagstrom, S.; Lyon, H. B.; Somorjai, G. A. Phys. ReV. Lett. 1965,
15, 491.
(11) Norton, P. R.; Griffiths, K.; Bindner, P. E. Surf. Sci. 1984, 138,
125.
(12) Lykke, K. R.; Kay, B. D. J. Chem. Phys. 1991, 95, 2252.
(13) Hanisco, T. F.; Kummel, A. C. J. Chem. Phys. 1991, 95, 8565.
(14) Zakheim, D. S.; Johnson, P. M. Chem. Phys. 1980, 46, 263.
(15) Bray, R. G.; Hochstrasser, R. M. Mol. Phys. 1976, 31, 1199.
(16) Lombardo, S. J.; Esch, F.; Imbihl, R. Surf. Sci. Lett. 1992, 271,
L367.
(17) Madden, H. H.; Imbihl, R. Appl. Surf. Sci. 1991, 48/49, 130.
(18) Slinko, M.; Fink, T.; Lo ¨her, T.; Madden, H. H.; Lombardo, S. J.;
Imbihl, R.; Ertl, G. Surf. Sci. 1992, 264, 157.
(19) Fink, T.; Dath, J. P.; Bassett, M. R.; Imbihl, R.; Ertl, G. Surf. Sci.
1991, 245, 96.
(20) Ibach, H.; Lehwald, S. Surf. Sci. 1980, 91, 187.
(21) Schwaha, K.; Bechthold, E. Surf. Sci. 1977, 66, 383.
(22) Miners, J. H.; Gardner, P. J. Phys. Chem. B 2000, 104, 10265.
(23) King, D. A. Faraday Discuss. 1993, 96, 79.
(24) Bradley, J. M.; Hopkinson, A.; King, D. A. J. Phys. Chem. 1995,
99, 17032.
(25) Foner, S. N.; Hudson, R. L. J. Chem. Phys. 1984, 80, 518.
(26) Thorman, R. P.; Bernasek, S. L. J. Chem. Phys. 1981, 74, 6498.
(27) Murphy, M. J.; Skelly, J. F.; Hodgson, A. J. Chem. Phys. 1998,
109, 3619.
(28) Murphy, M. J.; Skelly, J. F.; Hodgson, A.; Hammer, B. J. Chem.
Phys. 1999, 110, 6954.
(29) Carter, R. N.; Murphy, M. J.; Hodgson, A. Surf. Sci. 1997, 387,
102.
(30) Haq, S.; Hodgson, A. Surf. Sci. 2000, 463, 1.
(31) Wang, H.; Tobin, R. G.; DiMaggio, C. L.; Fisher, G. B.; Lambert,
D. K. J. Chem. Phys. 1997, 107, 9569.
8728 J. Phys. Chem. B, Vol. 105, No. 37, 2001
Hallock et al.