The role of carbon deposition from CO dissociation on platinum crystal surfaces during catalytic CO oxidation: Effects on turnover rate, ignition temperature, and vibrational spectra

Article abstract of DOI:10.1021/jp014679k

The process of ignition during Pt catalyzed CO oxidation was studied using sum frequency generation (SFG) vibrational spectroscopy, AES, GC, and TPD. CO dissociation at 40 Torr of CO was observed on Pt(100), Pt(557), and Pt(111) at 500, 548, and 673 K, respectively. The rate of reaction below ignition was significantly higher on the initially carbon covered Pt surface when the reaction was performed in excess CO or equal pressures of CO and O₂. The carbon oxide species, proposed to be a carboxylic anhydride, decomposed near 900 K with similar concentrations of both CO and CO₂. Additionally, it gasified near this temperature under oxidation conditions. When CO oxidation was performed on an initially clean surface, the SFG background increased, indicating CO was dissociating. The oxidation of surface carbon increased the amount of chemical energy evolved by the system, resulting in a decrease in the ignition temperature. This observed carbon oxide species could be important in other oxidation processes such as methane oxidation.

Full text of DOI:10.1021/jp014679k

10854
J. Phys. Chem. B 2002, 106, 10854-10863
The Role of Carbon Deposition from CO Dissociation on Platinum Crystal Surfaces during
Catalytic CO Oxidation: Effects on Turnover Rate, Ignition Temperature, and Vibrational
Spectra
Keith R. McCrea, Jessica S. Parker, and Gabor A. Somorjai*
Department of Chemistry, UniVersity of California at Berkeley, Berkeley, California 94720
ReceiVed: December 31, 2001; In Final Form: May 3, 2002
Experiments were performed to understand the process of ignition during platinum catalyzed CO oxidation
using sum frequency generation (SFG) vibrational spectroscopy, Auger electron spectroscopy (AES), gas
chromatography (GC), and temperature programmed desorption (TPD). Both CO dissociation and CO oxidation
ignition studies on the (100), stepped (557), and (111) surfaces of platinum are presented. Rapid CO dissociation
on the Pt(100), Pt(557), and Pt(111) occurred in narrow temperature ranges ((10 K) at 500, 548, and 673 K,
respectively. The CO ignition temperature at a pressure of 40 Torr of CO and 100 Torr of O2 is lower on
Pt(100) than on Pt(111) and Pt(557). Thus, both CO dissociation and the ignition of CO oxidation are structure
sensitive. An in depth study of CO oxidation on Pt(557) was performed on both initially clean and initially
carbon covered platinum surfaces to investigate the role of carbon obtained from CO dissociation in CO
oxidation. Under excess O2 and excess CO conditions, a clean platinum surface will remain carbon free
below and above ignition. However, a carbon oxide species was formed on initially carbon covered platinum
surfaces once oxygen was added at a high temperature (548 K). This carbon oxide species results in a large
SFG background signal, allowing us to measure the formation and reactivity of this species during oxidation
reactions. The carbon oxide species also formed on initially clean platinum below the ignition temperature
when the Pt crystal was exposed to equal partial pressures of CO and O2. The turnover rates on the carbon
oxide covered platinum surfaces were higher than on the initially clean surfaces below ignition. The ignition
temperature on the carbon covered surface (648 K) was lower than on the clean platinum (700 K) surface at
equal pressures of CO and O2. All of this evidence indicates the surface carbon oxide species is better at
oxidizing CO than platinum under certain pressure and temperature conditions. CO dissociation is an important
step during the onset of ignition, when surface carbon oxidation provides a new exothermic reaction channel
in addition to the Pt surface catalyzed oxidation of molecular CO.
Introduction
recognized by the observation of a spontaneous temperature
increase.1,2 Above the ignition temperature, a new, mass
transport controlled steady state reaction rate is established.1,2,13
Using platinum wires, Rinnemo and co-workers showed the
ignition temperature is dependent on the CO and O2 partial
pressure.5 As the CO/O2 ratio increases, the ignition temperature
increases. Ignition temperatures on Pt single crystals range from
500 to 600 K.5,7 Above ignition, the surface is oxygen covered,
and the reaction is mass transport limited by either CO
approaching or CO2 leaving the surface.
The oxidation of carbon monoxide (CO) on catalyst surfaces
is one of the most studied heterogeneous catalytic reactions.
At certain pressures of CO and O2 (10-6 to 760 Torr), the
reaction proceeds under two different kinetic regimes separated
by an ignition temperature. Below the ignition temperature, the
reaction rate is governed by surface reaction kinetics.1,2 In this
regime, the CO oxidation reaction mechanism follows Lang-
muir-Hinshelwood kinetics, and the surface is primarily CO
covered.3-10
CO(g) + O(a) f CO2(g)
CO(a) + O(a) f CO2(g)
This reaction is positive first order in O2 and negative first
order in CO pressure.10,11 These results agree for both low-
pressure and high-pressure studies on platinum single crystals.7
The activation energy for this reaction was found to be between
3010,12 and 42 kcal/mol.7 In a narrow temperature range, a
transition occurs where the surface, once both CO and atomic
oxygen covered, becomes exclusively oxygen covered.6 The rate
of CO2 production becomes high, causing a rapid increase in
temperature usually called ignition. Ignition occurs when the
heat losses in the system can no longer keep pace with the
amount of chemical energy evolved by the exothermic surface
reaction.1,2,13,14 Experimentally, the ignition temperature is
The rate of reaction above ignition is extremely high with
turnover rates well above 1000 (molecules/site)/s on Pt single
crystals. Su and co-workers found the activation energy to be
14 kcal/mol on Pt(111) at a pressure of 40 Torr of CO and 100
Torr of O2,7 similar to the activation energy found using
molecular beams on an oxygen covered Pt(111) surface (11.7
kcal/mol).15 For both CO and O2, a positive half-order depen-
dence in partial pressure was observed under high-pressure
conditions.7
In this paper we report on studies of the surface structure
sensitivity of CO dissociation and ignition temperatures over
platinum single crystal surfaces. Sum frequency generation
10.1021/jp014679k CCC: $22.00 © 2002 American Chemical Society
Published on Web 10/01/2002

CO Dissociation on Platinum Crystal Surfaces
J. Phys. Chem. B, Vol. 106, No. 42, 2002 10855
(SFG) surface vibrational spectroscopy and Auger electron
spectroscopy (AES) were used to monitor the effect of exposing
platinum single crystals to high pressures of CO at high
temperatures. Monitoring the shift of the CO top-site peak by
SFG as a function of temperature allowed us to determine when
CO dissociation occurs. We define the CO dissociation tem-
perature as the temperature where we can detect significant
amounts of carbon on our platinum surface using AES. This
temperature coincides with a dramatic change in the CO spectral
feature as monitored by SFG. The dissociation of CO at a
pressure of 40 Torr of CO has been studied on three different
crystal faces of platinum: the (100), the stepped (557), and the
(111) orientation surfaces. Carbon deposition occurs at different
temperatures depending on the surface structure of the platinum
crystals as they are heated at constant CO pressure: 500 K for
the (100) surface orientation of platinum, 548 K for Pt(557),
and 673 K for Pt(111). Thus, CO dissociation is surface structure
sensitive, and its mechanism is related to roughening of the metal
surfaces by platinum carbonyl formation, leading to the Bou-
douard reaction.
under most experimental conditions of CO oxidation, resulting
in an oxygen covered metal surface.
Experimental Section
Sum Frequency Generation. SFG is a powerful technique
for studying high-pressure catalytic reactions on single crystal
surfaces because it is surface specific, nondestructive, and highly
sensitive with good spectral resolution. SFG has been described
in detail elsewhere,16-20 so only a brief discussion will be given
here. During an SFG experiment, two laser beams are over-
lapped in time and space on a single crystal surface. The first
beam is a fixed visible 532 nm green beam (ωvis), and the second
beam is a tunable infrared (IR) beam (ωIR) in the wavelength
range between 2 and 10 µm (1000-4000 cm-1). The visible
and IR beams mix on the surface to drive an oscillating dipole,
emitting a coherent beam of photons at the sum of the visible
and IR frequencies (ωSFG ) ωvis + ωIR). This beam is easily
detected.
Under the electric dipole approximation, the intensity of the
sum frequency signal is proportional to the square of the second-
order nonlinear surface susceptibility (I |ø(2)|2). The suscep-
tibility is described by
2CO f C + CO2
The dramatic decrease of the CO vibrational frequency to 2054
cm-1 just before dissociation takes place, as detected by SFG,
led us to this conclusion.
AR
ø(2) ) ANR
+
∑
(ωIR - ω0 - iγ)
R
The ignition temperatures for CO oxidation on the three
platinum crystal faces also show structure sensitivity. The
ignition of CO oxidation, when the combustion reaction evolves
more heat than the system can dissipate, is correlated with
carbon deposition from CO dissociation by our experimental
findings. The oxidation of surface carbon to CO2 represents a
new exothermic reaction channel, increasing the amount of
chemical energy evolved by the oxidation reaction.
where ANR is the nonresonant contribution, γ is the line width,
ωo is the resonant vibrational frequency, and ωΙR is the IR
frequency. The resonant strength, AR, is proportional to the
number and orientational average of molecules on the surface
and the IR and Raman transition moments. As observed in this
equation, when ωIR is equal to ωo, ø(2) is maximized and so a
surface vibrational spectrum can be obtained by scanning ωIR
through a frequency range of interest.
The role of carbon deposited by CO dissociation during
catalytic CO oxidation over platinum was investigated on Pt-
(557). The discovery of a new SFG signature, indicating the
presence of both carbon and oxygen on a platinum surface, gives
insight into the role of carbon during combustion. When carbon
is deposited on a Pt single crystal during reaction at high
temperatures (>500 K), it reacts immediately with oxygen to
produce a surface carbon oxide species. This carbon oxide
species increases the level of the SFG background, and this
background is dependent on the surface carbon-to-oxygen ratio.
This allows us to detect carbon deposition on an initially carbon
free surface during the oxidation reaction. Carbon cannot be
detected on the platinum surface when combustion is carried
out at high pressures (40-100 Torr) either in excess oxygen or
in excess CO. The metal surface remains carbon free both below
and above the ignition temperature in these circumstances.
However, both carbon and oxygen are detectable on the platinum
surface when it is exposed to equal partial pressures of CO and
O2, indicating the formation of a carbon oxide species.
The ignition temperature on the initially carbon covered Pt-
(557) surface is lower than on the clean platinum surface at
equal partial pressures of CO and O2. This is additional evidence
that carbon deposition by CO dissociation is an important step
for the onset of ignition, and the carbon oxide covered platinum
surface is a better CO oxidation catalyst than platinum. The
rapid gasification of surface carbon
Because AR is proportional to the IR and Raman transition
moments, the selection rules for both IR and Raman spectros-
copy must be obeyed. Hence, media must be both IR and Raman
active to generate SFG. Only media lacking inversion symmetry
will satisfy this requirement, as shown by group theory. Usually,
bulk materials are centrosymmetric and do not generate SFG.
Isotropic gases and liquids also do not generate SFG. Only at
surfaces or interfaces where the centrosymmetry is broken can
SFG be produced. This is the reason SFG is surface specific.
Thus, the technique can be used to probe any interface as long
as the media the laser beams pass through do not interfere with
the light.
The visible and tunable IR laser beams were generated using
a Nd:YAG laser to pump a commercial optical parametric
generation/amplification (OPG/OPA) system provided by La-
serVision. This system utilized the fundamental output of the
Nd:YAG laser of 1064 nm to pump a KTP crystal to generate
a visible beam of 532 nm. This beam was split, with one portion
being sent to the experiment and the second portion being sent
to a pair of KTP crystals to generate a beam in the near-IR
(710-880 nm). This near-IR beam was then difference fre-
quency mixed with a 1064 nm beam through two KTA nonlinear
crystals, generating a tunable IR beam between 1950 and 4000
cm-1
.
The p-polarized IR and visible beams are both spatially and
temporally overlapped on a single crystal mounted in an
ultrahigh vacuum (UHV) chamber. The visible beam makes an
angle of 50° with respect to surface normal, whereas the IR
beam is at 55° with respect to surface normal. The SFG beam
C + O2 f CO2
must be the reason for the absence of surface carbon on platinum

10856 J. Phys. Chem. B, Vol. 106, No. 42, 2002
McCrea et al.
is sent through a monochromator, and the signal intensity is
detected by a photomultiplier tube and integrated by a gated
integrator.
Ultrahigh Vacuum/High-Pressure Reaction Chamber. A
UHV system with a base pressure below 1 × 10-10 Torr was
used to perform the experiments.21 The system was equipped
with a rearview retarding field analyzer (RFA) for AES and
low-energy electron diffraction (LEED) studies. A quadrupole
mass spectrometer monitored the background gas composition
in the system under vacuum. For high-pressure experiments,
the bell jar was separated from the vacuum pumps, and gases
were introduced through a manifold system. Through a recir-
culation loop attached to the chamber, the gas composition under
high-pressure catalytic conditions could be monitored by gas
chromatography (GC), allowing us to calculate reaction kinetics.
After a high-pressure experiment, the chamber was pumped
down to UHV. At that time, the surface composition of the
platinum crystal was analyzed using AES and, in certain cases,
temperature programmed desorption (TPD) studies were per-
formed.
Figure 1. Comparison of the adsorbed CO vibrational frequency as a
function of temperature for Pt(111), Pt(557), and Pt(100) at a pressure
of 40 Torr of CO. The Auger spectrum before and after heating Pt-
(557) in the presence of 40 Torr of CO is included to show surface
carbon can be detected after the experiment.
The Pt(557) single crystal used for this study was mounted
in the UHV chamber. A Pt(557) surface is prepared by cutting
a platinum single crystal 9.5° relative to the (111) orienta-
tion.4,22,23 This prepares a stepped surface with six atom wide
terraces of (111) orientation and single atom high steps. Prior
to each experiment, the single crystal was cleaned by two cycles
of argon ion bombardment followed by annealing in 5 × 10-7
Torr of oxygen at 1123 K for 2 min. The oxygen was pumped
out, and the crystal was annealed at a pressure below 2 × 10-9
Torr at 1133 K for 1 min. Once the crystal was shown by AES
to be clean, the sample was exposed to high pressures of CO
(Scott Specialty 99.99%) and O2. The CO was further purified
by flowing it through a liquid nitrogen trap before introducing
it to the sample. When catalytic experiments were performed,
the chamber acted as a batch reactor. For SFG experiments,
the chamber was equipped with a CaF2 window at the end of
an inverted flange to allow the IR light to reach the sample
with minimal gas phase absorption. The path length from the
window to the crystal is approximately 4 cm. To normalize for
any gas phase absorbance of the IR beam, an IR cell was
attached to the sample loop to allow for the acquisition of gas
phase IR transmission spectra of gas mixtures.
cooled to room temperature, and more SFG scans were acquired.
For each surface, an irreversible process occurred where the
peak position shifted to lower frequency. The chamber was
quickly evacuated, and AES showed the surface was covered
with carbon.
When the Pt(557) was exposed to 40 Torr of CO, a single
CO peak was observed at 2100 cm-1. As the sample was heated,
the CO spectral feature shifted to lower frequencies. This
occurred up to 523 K where the CO peak was observed at 2082
cm-1. Once the sample was heated to 548 K, the peak shifted
to 2077 cm-1 and the intensity of this peak decreased over time.
The crystal was cooled to 300 K. The post-heating cycle
frequency of the CO peak was red-shifted to 2087 cm-1, whereas
the fitted amplitude remained essentially the same as before
the heating cycle. The chamber was evacuated to 5 × 10-8 Torr,
and an Auger spectrum was acquired. The spectrum was
dominated by carbon, indicating CO decomposed at 548 K on
Pt(557) at a pressure of 40 Torr of CO.
The same experiment was performed on the Pt(111) and Pt-
(100) crystal faces, and CO dissociation was observed in both
cases.24 To summarize the results for CO dissociation on the
three single crystal surfaces, Figure 1 shows the frequency as a
function of temperature for these surfaces. The frequency of
the CO spectral feature decreases until a critical temperature is
reached where dissociation and carbon deposition occur. CO
dissociation is obviously structure sensitive, because the dis-
sociation temperatures for Pt(111), Pt(557), and Pt(100) are 673,
548, and 500 K respectively.
Carbon can be deposited on Pt(557) by heating to 548 K in
high-pressure CO.21 In this work, CO oxidation was performed
on both initially clean and initially carbon covered surfaces.
The initially carbon covered surfaces were prepared by heating
the crystal in the desired pressure of CO before introducing O2.
Results
Surface Structure Sensitivity of CO Dissociation on Pt-
(557), Pt(111), and Pt(100). In this section, the results for
experiments that explore the properties of Pt single crystals
under high-pressure CO and high temperatures are discussed.21
The pressure of CO used in these experiments (40 Torr) is the
same as that used during the CO oxidation reactions with excess
oxygen. These high-pressure CO experiments are done to help
understand the interaction of CO with platinum as a function
of temperature and the mechanism of ignition under oxidation
conditions. After a crystal was cleaned using the procedure
described above, it was exposed to 40 Torr of CO. SFG spectra
were acquired at 300 K. The samples were heated sequentially
to higher temperatures, and at each temperature SFG scans were
obtained and averaged. At a particular temperature for each
crystal, the SFG spectra evolved with time. The sample was
Surface Structure Sensitivity of CO Oxidation Ignition
with Excess O2 on Pt(557), Pt(111), and Pt(100). CO oxidation
experiments with excess pressure of O2 were performed on the
(557), (111), and (100) faces of platinum. Only the results for
Pt(557) will be discussed due to the qualitative similarities
among all three crystal faces. After the Pt(557) single crystal
surface was cleaned as described above, it was introduced to
40 Torr of CO, 100 Torr of O2, and 630 Torr of He at 300 K.
SFG spectra at different temperatures during the experiment are
shown as a stacked plot in Figure 2. Each spectrum is the
average of three consecutive scans. At 300 K, a single resonance
is observed centered at 2094 cm-1, due to a saturated coverage
of top-site CO. The sample was heated to 473 K, and the CO
peak red-shifted to 2085 cm-1. The CO peak retained the same
frequency and intensity as the temperature was increased to 623

CO Dissociation on Platinum Crystal Surfaces
J. Phys. Chem. B, Vol. 106, No. 42, 2002 10857
Figure 2. SFG spectra of Pt(557) at a pressure of 40 Torr of CO and
100 Torr of O2 as a function of temperature. The CO peak was absent
once ignition was reached, indicating the surface was no longer CO
covered above ignition.
Figure 4. SFG spectrum of a clean platinum surface at a pressure of
40 Torr of CO is compared to the SFG spectrum of the surface at 300
K after the oxidation reaction at a pressure of 40 Torr of CO and 100
Torr of O2. The oxidation reaction took place on an initially carbon
covered Pt(557) surface. The large SFG background indicates the
surface is carbon oxide covered. The SFG background from a clean
platinum crystal is given as a reference.
TABLE 1: Temperature Ranges (K) for the Onset of Rapid
CO Dissociation and the Transition from Surface Reaction
Controlled Kinetics to Mass Transport Controlled Kinetics
(Ignition)
Pt(100)
Pt(557)
Pt(111)
CO dissociation temp
(40 Torr CO)
oxidation ignition temp
(40 Torr CO, 100 Torr O2)
500 ( 10 550 ( 10 673 ( 10
500 ( 20 640 ( 20 620 ( 20
Even though the SFG spectra were very similar for all three
crystal faces during CO oxidation at these pressure conditions,
the ignition temperature was considerably different for the Pt-
(100) surface as compared with the Pt(111) and Pt(557) surfaces.
Table 1 compares the CO dissociation and the CO oxidation
ignition temperatures for the three surfaces. As evident from
the table, a similar trend for the structure sensitivity of both
CO dissociation and ignition is observed. Both ignition and CO
dissociation occur at a higher temperature for the (111) surface
than Pt(100). The CO ignition temperature for Pt(111) and Pt-
(557) are very similar to each other within experimental error,
indicating CO oxidation occurs mainly on the (111) terraces.
This agrees with other studies, which indicated the (111) terrace
sites are more important for CO oxidation than step sites.4,25,26
To determine how carbon affects the reaction under excess
O2, a carbon covered Pt(557) surface was prepared as described
above, and the reaction was performed. As soon as 100 Torr of
O2 was added at 548 K, any residual CO peak disappeared,
and the SFG background increased significantly. The implication
of this large background will be discussed below. Once the
ignition temperature was reached (640 ( 20 K), the background
increased as a function of temperature and did not decrease as
the sample was cooled back to room temperature. The CO peak
does not recover when the sample is cooled to room temperature
under these pressure conditions, as shown in Figure 4. The
Figure 3. SFG spectra of Pt(557) before and after CO oxidation at a
pressure of 40 Torr of CO and 100 Torr of O2. Changes in the
vibrational spectra are reversible, indicating the platinum surface is
unchanged by this reaction.
K. The turnover rate for the production of CO2 at 623 K was
20 (molecules/site)/s. Once the sample was heated to the ignition
temperature, 640 ( 20 K, the temperature jumped to 723 K.
The turnover rate was calculated to be 1400 (molecules/site)/s,
and the SFG spectrum was featureless. If the sample was cooled
below the ignition point, the reaction was quenched and the
CO resonant peak recovered to the same frequency and intensity
observed before ignition, as shown in Figure 3.
For all three surfaces, the qualitative features of the SFG
spectra are essentially the same. Below the ignition temperature,
the CO peak slowly shifts as a function of temperature until
the ignition temperature is reached. At the ignition temperature,
the crystal temperature increases rapidly, and the CO peak
decreases rapidly. Once above the ignition temperature, no
spectral feature is observed for these pressure conditions.

10858 J. Phys. Chem. B, Vol. 106, No. 42, 2002
McCrea et al.
Figure 6. SFG spectra of Pt(557) at a pressure of 100 Torr of CO and
40 Torr of O2 as a function of temperature. The CO top site peak red-
shifted in frequency as the temperature was raised. No distinct ignition
temperature is seen in this case. Instead, a smooth change in temperature
is observed at all temperatures. Thus, the system was able to change
from surface reaction controlled kinetics to mass transport controlled
kinetics without experiencing the instability associated with ignition.
reaction reveals both a large carbon peak and a large oxygen
peak. Thus, the SFG background originates from an oxidized
carbon species.
CO Oxidation with Excess CO on Pt(557). After the Pt-
(557) crystal was determined by AES to be clean, 100 Torr of
CO, 40 Torr of O2, and 630 Torr of He were introduced to the
chamber. At 300 K, a single SFG peak at 2100 cm-1 was
observed. The crystal was heated gradually to 1023 K. As the
crystal was heated, the SFG peak slowly shifted to 2040 cm-1
at 1048 K, as shown in Figure 6. Detectable CO2 production
began at 673 K, although the turnover rate did not become high
(990 (molecules/site)/s) until 723 K. The crystal was allowed
to cool, and the post-reaction SFG spectrum was identical to
the pre-reaction spectrum. As previously reported, the ignition
temperature is dependent on the CO to O2 ratio.5 As the ratio
increases, the ignition temperature increases. No distinct ignition
temperature is seen in the excess CO case. Instead, a smooth
change in temperature is observed at all temperatures. Thus,
the system was able to change from surface reaction controlled
kinetics to mass transport controlled kinetics without experienc-
ing the instability associated with ignition.1
Figure 5. Temperature profile and SFG signal as a function of time
for Pt(557) at a pressure of 40 Torr of CO and 100 Torr of O2. The
temperature and SFG signal were monitored as the crystal temperature
was raised to the ignition temperature. (A) On initially clean platinum,
the CO peak was monitored at 2085 cm-1. The temperature of the
crystal was slowly ramped to the ignition temperature as we monitored
the CO top site peak at 2085 cm-1. Once the ignition temperature was
reached, the intensity of the CO peak immediately decreased. As the
temperature of the crystal was cooled below the ignition temperature,
the intensity of the CO peak immediately recovered. The SFG
background did not change during the experiment, indicating the
platinum surface remained clean. (B) On initially carbon covered
platinum, the SFG background was monitored at 2075 cm-1. The SFG
background increased as ignition occurred, indicating the presence of
carbon oxide on the platinum surface under these reaction conditions.
CO oxidation under excess CO was also performed on an
initially carbon covered platinum surface by first heating the
Pt(557) single crystal in 100 Torr of CO at 573 K. After 20
min of heating, 40 Torr of O2 and 630 Torr of He were
introduced into the chamber. Immediately, any residual CO peak
vanished, and the SFG background increased substantially. The
SFG background was monitored as a function of temperature,
as shown in Figure 7. A high turnover rate for CO2 production
began at 673 K, 50 deg lower than observed on the initially
clean surface. As the sample temperature was increased to 723
K, the SFG background increased substantially. A small SFG
peak at 2070 cm-1 was also observed in addition to the large
background at 723 K. With a continuing increase of the
temperature above 723 K, the SFG background began to
decrease, and the small peak decreased in frequency as well.
By 1023 K, the SFG spectrum was flat without any features or
large background. The spectrum at 300 K after the crystal was
heated to 1023 K is very similar to the spectrum at 300 K before
dissociating CO, which indicates the surface is free of the carbon
oxide species.
turnover rates below (548 K) and above ignition (723 K) were
20 and 1280 (molecules/site)/s, respectively. Figure 5 shows
the temperature and SFG intensity traces as a function of time
for both the initially clean (A) and initially carbon covered (B)
surface under excess O2. As shown in (A), the CO peak
disappears immediately upon ignition and no change in the SFG
background is observed. On an initially carbon covered surface
(B), the SFG background increases dramatically upon ignition
and stays large as the reaction proceeds.
Auger spectra of both the initially clean and initially carbon
covered platinum surfaces were acquired after CO oxidation.
The spectrum of the initially clean surface after CO oxidation
revealed a small carbon peak in addition to the platinum peaks.
This is attributed to electron beam dissociation of CO. The
spectrum of the initially carbon covered platinum surface after

CO Dissociation on Platinum Crystal Surfaces
J. Phys. Chem. B, Vol. 106, No. 42, 2002 10859
Figure 7. SFG spectra of an initially carbon covered Pt(557) surface at a pressure of 100 Torr of CO and 40 Torr of O2. A large SFG background,
due to carbon oxide on the surface, was seen at temperatures as low as 573 K. As the surface was heated, the SFG background intensity went
through a maximum at 723 K. The SFG background intensity then decreased as the temperature was raised even further. At 1023 K, we observed
a flat, featureless spectrum very similar to the spectrum of clean platinum (shown in Figure 4). This indicates the carbon oxide species responsible
for the large SFG background can be removed from the surface at high enough temperatures at these reaction conditions.
and SFG intensity traces as a function of time for both the
initially clean (A) and initially carbon covered (B) surfaces. The
temperature was ramped linearly and no sudden jump to indicate
ignition occurred. On the initially clean surface (A), the spectral
feature decreases in intensity linearly (although peak position
does red-shift) and no appreciable change in the background is
observed. The initially carbon covered surface (B) shows the
background reaches a maximum around 723 K and then
decreases until the background is negligible around 1023 K,
indicating the surface is free of the carbon oxide species.
CO Oxidation with Equal Pressures of CO and O2 on Pt-
(557). The Pt(557) crystal was cleaned as described above and
introduced to 70 Torr of CO, 70 Torr of O2, and 630 Torr of
He at 300 K. The sample was gradually heated to just below
the ignition temperature. Interestingly, the SFG background
began to increase around 623 K, unlike the other two pressure
regimes where the SFG background does not increase under
oxidation conditions for an initially clean crystal. As the sample
Figure 8. Comparison of the turnover rates for both initially clean
and initially carbon covered Pt(557) as a function of temperature at a
is heated higher, the background increases. As stated previously,
the background is due to a carbon oxide species on the surface.
pressure of 100 Torr of CO and 40 Torr of O2. The carbon oxide
Because carbon was not deposited before the reaction began,
covered Pt surface had a higher turnover rate at low temperatures,
CO is actually dissociating under these oxidizing conditions.
indicating the carbon oxide covered Pt surface is a better CO oxidation
The reaction rate at 673 K, which is below the ignition
temperature for these conditions, is 900 (molecules/site)/s. The
ignition temperature was 700 K, and upon ignition the temper-
ature increased to 790 K. The turnover rate at 790 K was 4300
(molecules/site)/s. After the reaction, the sample was cooled,
and the SFG spectrum revealed an irreversible change happened
during reaction (Figure 10). The peak position for both spectra
is 2100 cm-1, although CO is adsorbed on a modified platinum
surface after ignition. The modified surface has a carbon oxide
species coadsorbed with CO, which explains the increased
background.
catalyst than the initially clean Pt surface at these pressures and at
temperatures below 873 K.
Figure 8 compares the turnover rates versus temperature for
the initially clean and initially carbon covered platinum surfaces
under excess CO. The turnover rate on the initially carbon
covered surface is much higher at low temperatures than on
the initially clean surface. Around 873 K, the turnover rates
become essentially the same. The rates converge for the two
different surfaces when the SFG background decreases for the
initially carbon covered surface, indicating any residual carbon
oxide is being gasified away. Figure 9 shows the temperature

10860 J. Phys. Chem. B, Vol. 106, No. 42, 2002
McCrea et al.
Figure 10. SFG spectra of Pt(557) at a pressure of 70 Torr of CO and
70 Torr of O2 before and after reaction. Although the frequency of the
top site CO peak is the same in both spectra, the peak is superimposed
upon a large SFG background in the post-reaction spectrum. This
indicates the platinum surface was modified during the oxidation
reaction and now contains a carbon oxide species.
(molecules/site)/s. Because the SFG background was still
considerable, the carbon was not oxidized away as in the case
with excess CO. Figure 11 shows the temperature and SFG
intensity traces as a function of time for both the initially clean
(11A) and initially carbon covered (11B) surfaces. On the
initially clean surface, the background is observed to increase
as a function of temperature and reaches a maximum shortly
after ignition. Also, the CO peak disappears immediately upon
ignition.
Increased SFG Background Signal from a Carbon and
Oxygen Containing Surface Species. In addition to SFG signal
enhancement when the IR beam is at the same frequency as a
vibrational mode, there can also be SFG enhancement when
the visible beam is near an electronic resonance of a surface
species. If this is the case, the SFG background increases and
is considerably higher than the normal SFG nonresonant
background. In this study, under certain conditions, the SFG
background was found to increase. To determine the cause of
this background, control experiments were performed.
The sample was heated in pure CO or pure O2 to high
temperatures to determine whether either one of these species
was responsible for the observed background. When the Pt-
(557) crystal was heated in high pressures of CO, CO dissociated
and deposited carbidic carbon, as determined by AES.21 The
addition of this carbon species did not contribute to the SFG
background, as seen in previous studies. When a clean prepared
platinum surface is heated in a high pressure of O2, the SFG
background remains at the same level as the nonresonant
background for clean platinum. AES reveals a clean platinum
surface, and no evidence of atomic oxygen on the surface is
observed in the spectrum. This result is expected because
platinum does not easily oxidize under these conditions, unlike
other transition metals.
Figure 9. Temperature and SFG signal as a function of time for Pt-
(557) at a pressure of 100 Torr of CO and 40 Torr of O2. (A) On an
initially clean surface, the temperature is ramped linearly, and we do
not observe a distinct ignition temperature. The CO peak decreases
linearly, and the SFG background remains negligible, indicating carbon
deposition is not occurring under these conditions. (B) On an initially
carbon covered surface, a distinct ignition temperature is not observed.
The SFG background reaches a maximum at 723 K before decreasing.
At 1023 K, the SFG background resembles that of clean platinum,
indicating the carbon oxide surface species gasifies above 723 K.
The same pressures of CO and O2 were also explored on an
initially carbon covered Pt(557) crystal. After the crystal was
cleaned, 70 Torr of CO was introduced into the chamber, and
the sample was heated to 573 K. After 20 min, 70 Torr of O2
was added along with 630 Torr of He. Immediately, any residual
CO peak disappeared and the SFG background increased. At
623 K, below the ignition temperature, the turnover rate was
800 (molecules/site)/s. This rate is substantially higher than that
observed on the initially clean Pt surface under the same
conditions at 623 K. The sample was gradually heated and
ignition was observed at 648 K. This ignition at 648 K is 50
deg lower than that observed for the initially clean crystal. The
SFG background increased as a function of temperature below
ignition but did not increase appreciably after ignition, indicating
the relative concentrations of carbon and oxygen on the surface
were not changing. Above ignition, at 790 K, the rate was 4400
Because the SFG background increased in the presence of
both CO and O2, an experiment was performed where the sample

CO Dissociation on Platinum Crystal Surfaces
J. Phys. Chem. B, Vol. 106, No. 42, 2002 10861
Figure 11. Temperature and SFG signal as a function of time for Pt-
(557) at a pressure of 70 Torr of CO and 70 Torr of O2. The temperature
and SFG signal were monitored as the crystal temperature was raised
to the ignition temperature. (A) On an initially clean surface, a sharp
jump in temperature is observed at 700 K, and the CO peak immediately
decreased. The SFG background is observed to increase beginning at
573 K and reaches a maximum just after ignition. This indicates carbon
is being deposited under oxidation conditions and a carbon/oxygen
species is being formed. (B) On an initially carbon covered surface,
the ignition temperature was observed at 648 K, 50 deg lower than on
the clean Pt surface under these conditions.
was heated first in 40 Torr of CO to deposit carbon onto the
surface. Once the carbon was deposited, the system was
evacuated. AES revealed a carbon covered platinum surface.
The carbon covered surface was then titrated with 5 × 10-7
Torr of O2 as a function of temperature starting at 300 K. After
being heated at various temperatures, the crystal was cooled
and three SFG spectra were acquired and averaged. Figure 12A
shows the absolute intensity SFG spectra of Pt(557) under
vacuum after titration in O2, and Figure 12B shows the
corresponding Auger spectra. It is important to note that Figure
12A is not a stacked plot. There is a direct correlation between
the intensity of the SFG background and Auger oxygen-to-
carbon ratio, as shown in Figure 12C. Clearly, as the O/C ratio
increases, the SFG background increases. The species giving
rise to the large SFG background is an oxidized carbon species.
Figure 12. (A) Change in SFG spectra from an initially carbon covered
Pt(557) surface as the crystal was heated in 5 × 10-7 Torr of O2. The
SFG background increased as the temperature was increased, indicating
we were forming a carbon oxide species on the surface. Please note
this is an absolute intensity plot and not a stacked plot. (B) Auger spectra
of carbon covered Pt(557) as a function of temperature as it was heated
in 5 × 10-7 Torr of O2. (C) SFG background at 2140 cm-1 plotted
against the corresponding oxygen to carbon Auger peak ratio. The
intensity of the SFG background was found to increase as the oxygen/
carbon Auger peak ratio increased.

10862 J. Phys. Chem. B, Vol. 106, No. 42, 2002
McCrea et al.
Various carbon oxides exist on activated carbon. These
species have been characterized using diffuse reflectance infrared
Fourier transform spectroscopy (DRIFTS) and TPD experi-
ments.27-29 Depending on the surface preparation, different
oxide species of carbon can form. A TPD experiment was
performed on the carbon oxide species found on Pt(557). CO
and CO2 were monitored by mass spectrometry as a function
of temperature to determine the decomposition temperature of
this unknown species. Both CO and CO2 evolved between 1000
and 1100 K with similar pressures, indicating a single type of
species was decomposing. Because the pressures were similar,
the carbon oxide is likely a carboxylic anhydride species
(-(CO)O(CO)-).
Figure 13. Proposed structure for carboxylic anhydride on the
surface.27 This species is thought to be responsible for the large SFG
background seen under certain pressure and temperature conditions.
the intensity of the SFG background. The rate of CO2 production
through the reaction of CO and oxygen on the carboxylic
anhydride covered platinum surface is faster than the rate of
production of CO2 from the reaction of adsorbed CO and O on
platinum.
Discussion
Both CO dissociation and CO oxidation ignition are structure
sensitive.24 Pt(100) showed the lowest dissociation and ignition
temperatures (500 K), whereas Pt(111) exhibited the highest
dissociation and ignition temperatures (>620 K). Thus, CO
dissociation is an important step for ignition.
Once above 723 K, the SFG background begins to decrease
as a function of temperature, indicating the carboxylic anhydride
species is gasifying at these temperatures. By 900 K, the
background is essentially gone, and the rate is the same as
observed on an initially clean Pt(557) surface, indicating the
surface carbon oxide was completely gasified. At these tem-
peratures, the reaction kinetics are mass transport controlled.
When CO oxidation was performed at equal pressures of CO
and O2, we found evidence for CO dissociation on an initially
clean Pt surface under oxidation conditions. Also, the ignition
temperature in this pressure regime is 50 deg lower for the
initially carbon covered Pt surface than the initially clean
surface. CO dissociation is an important mechanistic step for
the ignition of CO oxidation, as supported by the above
evidence.
Carbon and Oxygen Surface Species. When CO is dissoci-
ated, carbidic carbon islands are formed on the surface. When
the initially carbon covered platinum surface is heated in the
presence of O2, the carbon islands become terminated with
oxygen, which give rise to the SFG background and also allow
oxygen detection in the Auger spectrum. The background is
due to an electronic resonance of the 532 nm beam with the
carbon oxide species, and the size of the carbon islands will
determine the amount of oxygen observed in the Auger
spectrum. The intensity of the SFG background is also depend-
ent on the concentration of carbon and oxygen on the surface.
As the carbon-to-oxygen ratio decreases, the SFG background
increases.
Activated carbon is a common support used in catalysis. There
have been many studies on activated carbon surfaces.27-29 When
activated carbon is introduced to oxidizing conditions, it
becomes oxygen terminated, and many techniques have been
used to identify the types of carbon oxide species. One such
technique is TPD. Otake and co-workers oxidized an active
carbon either by heating in air or by exposing the carbon to
concentrated nitric acid.27 When the carbon was heated in air,
they found the surface began to oxidize significantly at 573 K,
and upon further heating, the amount of oxidation became much
more considerable. These temperatures are similar to those used
in this study where the large SFG background is observed to
grow under oxidizing conditions.
Once the carbon was oxidized, Otake and co-workers
performed TPD experiments on the system.27 The amounts of
CO and CO2 and temperatures at which they are evolved classify
the oxide species present on the surface. They observed two
CO peaks at 900 and 1175 K. One CO2 peak at 900 K was also
observed. Otake and co-workers determined this surface oxide
to be carboxylic anhydride. When this species decomposes, both
Much is known about the rates of CO oxidation above and
below ignition, but what remains unclear is the mechanism
causing the sudden transition from a CO and O covered surface
to an atomic oxygen covered surface. One explanation describes
the ignition as the point where CO coverage decreases below a
critical coverage, no longer inhibiting oxygen adsorption.6 In
this view, the CO coverage can decrease either by desorption
or reaction. Using molecular beams on Pd(110), Bowker and
co-workers showed a surface under CO oxidation conditions is
already covered with oxygen to about one-fourth of the
saturation value even below ignition.6 Ignition would only occur
when the CO coverage decreased to below 0.4 monolayer.
In addition to this observation about the onset of ignition,
CO dissociation induced carbon deposition must also contribute
to the decrease in CO coverage at the ignition temperature. The
ignition temperature trend for CO oxidation on the three
platinum crystal faces correlates with the CO dissociation
temperature trend. The ignition of CO oxidation is controlled
by carbon deposition from CO dissociation and is surface
structure sensitive.
Effect of Surface Carbon on CO Oxidation and Ignition.
CO oxidation experiments were performed and turnover rates
were measured on both initially clean and initially carbon
covered Pt(557) above and below the ignition temperature. The
reaction on an initially clean Pt surface under excess O2 is well-
known. On the basis of previous studies, the reaction follows
Langmuir-Hinshelwood kinetics below ignition, and CO2 is
produced by the reaction of adsorbed CO and adsorbed atomic
oxygen.5,6,10 Above ignition, mass transport kinetics control the
reaction.1,2,13
When CO oxidation is performed on an initially carbon
covered Pt(557) surface under excess oxygen, the reaction
kinetics are the same as those on the initially clean Pt(557) sur-
face. However, the SFG background increases under these
reaction conditions. This large SFG background is due to a sur-
face carbon oxide species. On the basis of our kinetic data, this
species is as active for O2 dissociation as the initially clean Pt.
When the reaction was performed on an initially carbon
covered Pt(557) surface under excess CO, the rate of CO2
production below ignition is significantly higher than on an
initially clean platinum surface. The surface oxide species,
believed to be carboxylic anhydride (Figure 13), appears to be
better at oxidizing CO than an initially clean Pt surface. The
amount of the carbon oxide surface species is proportional to

CO Dissociation on Platinum Crystal Surfaces
J. Phys. Chem. B, Vol. 106, No. 42, 2002 10863
CO and CO2 evolve simultaneously in equal concentrations. The
900 K temperature is very similar to the temperature of CO
and CO2 evolution in our TPD experiments and also of the
gasification of the carbon oxide species under oxidation
conditions during catalytic experiments. The concentrations of
CO and CO2 evolved in our experiments are essentially equal.
Thus, an important carbon oxide species on the Pt surface must
be carboxylic anhydride.
The electronic resonance with the 532 nm beam may be due
to a metal complex involving platinum atoms and the anhydride
species similar to what is seen for d-d transitions in organo-
metallic clusters.30 Calculations modeling the system are cur-
rently being performed and will be published at a later date.
After the carbidic carbon has been deposited through the
dissociation of CO, carbon islands are formed. When these
carbon islands are exposed to O2 near 600 K, the carboxylic
anhydride species forms from the reaction of surface oxygen
and carbon.
surface than the initially clean platinum surface. Hence, CO
dissociation is important for ignition. The oxidation of surface
carbon increases the amount of chemical energy evolved by
the system, resulting in a decrease in the ignition temperature.
This observed carbon oxide species could be important in
other oxidation processes such as methane oxidation. The
concentration of this species is dependent on temperature,
making it possible to control the amount of this species on the
surface. Further experiments will be performed to gain a better
understanding of this carbon oxide species. Experiments and
calculations will be used to assign the electronic SFG resonance
observed during the SFG experiments. A study correlating the
intensity of the SFG background and the concentration of carbon
and oxygen will also be performed using AES, so reaction
kinetics can be related to the concentration of carbon and oxygen
during reaction. This could greatly help us to understand the
nature of combustion reactions.
Acknowledgment. This work was supported by the Director,
Office of Energy Research, Office of Basic Energy Sciences,
Materials Science Division, of the U.S. Department of Energy.
Conclusions
There is a similar trend for CO dissociation and CO oxidation
ignition temperatures for the (111), (557), and (100) crystal faces
of platinum, indicating both processes are structure sensitive.
CO dissociation at 40 Torr of CO was observed on Pt(100),
Pt(557), and Pt(111) at 500, 548, and 673 K, respectively. The
ignition of CO oxidation at 40 Torr of CO and 100 Torr of O2
on the three crystal faces was observed at 500, 640, and 620 K,
respectively. The similarity in the ignition temperatures for the
(557) and (111) crystal faces implies the terrace sites on the
(557) crystal are more important for the onset of ignition than
the step sites. CO dissociation is important for the onset of
ignition as surface carbon oxidation increases the chemical
energy evolved by the reaction.
References and Notes
(1) Rinnemo, M.; Fassihi, M.; Kasemo, B. Chem. Phys. Lett. 1993,
211, 60.
(2) Fassihi, M.; Zhdanov, V. P.; Rinnemo, M.; Keck, K.-E.; Kasemo,
B. J. Catal. 1993, 141, 438.
(3) Hong, S.; Richardson, H. H. J. Phys. Chem. 1993, 97, 1258.
(4) Ohno, Y.; Sanchez, J. R.; Lesar, A.; Yamanaka, T.; Matsushima,
T. Surf. Sci. 1997, 382, 221.
(5) Rinnemo, M.; Kulginov, D.; Johansson, S.; Wong, K. L.; Zhdanov,
V. P.; Kasemo, B. Surf. Sci. 1997, 376, 297.
(6) Bowker, M.; Jones, I. Z.; Bennett, R. A.; Esch, F.; Baraldi, A.;
Lizzit, S.; Comelli, G. Catal. Lett. 1998, 51, 187.
(7) Su, X.; Cremer, P. S.; Shen, Y. R.; Somorjai, G. A. J. Am. Chem.
Soc. 1997, 119, 3994.
(8) Wartnaby, C. E.; Stuck, A.; Yeo, Y. Y.; King, D. A. J. Chem. Phys.
1995, 102, 1855.
The rate of reaction below ignition was considerably higher
on the initially carbon covered Pt surface when the reaction
was performed in excess CO or equal pressures of CO and O2.
For the excess O2 condition, ignition was very close to the start
of CO2 production and so pre-ignition kinetics could not be
studied as in depth as in the other two cases. In addition to the
higher turnover rates on the initially carbon covered surface,
the SFG background was significantly larger than observed on
the initially clean surface. The background was due to the
presence of a carbon oxide surface species with an electronic
resonance in the visible region near 532 nm, the frequency of
the visible beam used in these SFG experiments. The magnitude
of the SFG background has been correlated with the amount of
carbon oxide on the Pt surface. Because of the higher turnover
rate and the existence of the carbon oxide species, this surface
is better at oxidizing CO than initially clean Pt. The carbon
oxide species is proposed to be a carboxylic anhydride, as
observed in studies on activated carbons. This species decom-
poses near 900 K with similar concentrations of both CO and
CO2. Additionally, it gasifies near this temperature under
oxidation conditions.
(9) Imbihl, R.; Cox, M. P.; Ertl, G. J. Chem. Phys. 1986, 84, 3519.
(10) Berlowitz, P. J.; Peden, C. H. F.; Goodman, D. W. J. Phys. Chem.
1988, 92, 5213.
(11) Zhdanov, V. P.; Kasemo, B. Appl. Surf. Sci. 1994, 74, 147.
(12) Langmuir, I. Trans. Faraday Soc. 1922, 17, 672.
(13) Kulginov, D.; Rinnemo, M.; Kasemo, B. J. Phys. Chem. B 1999,
103, 3170.
(14) Frank-Kamenetskii, D. A. Diffusion and Heat Transfer in Chemical
Kinetics, 2nd ed.; Plenum: New York, 1969.
(15) Cambell, C. T.; Ertl, G.; Kuipers, H.; Segner, J. J. Chem. Phys.
1990, 73 (11), 5862.
(16) Shen, Y. R. Surf. Sci. 1994, 299/300, 551.
(17) Du, Q.; Superfine, R.; Freysz, E.; Shen, Y. R. Phys. ReV. Lett. 1993,
70, 2313.
(18) Johal, M. S.; Ward, R. N.; Davies, P. B. J. Phys. Chem. 1996,
100, 274.
(19) Conboy, J. C.; Messmer, M. C.; Richmond, G. L. J. Phys. Chem.
1996, 100, 7617.
(20) Shen, Y. R. Nature 1989, 337, 519.
(21) McCrea, K. R.; Parker, J. S.; Chen, P.; Somorjai, G. A. Surf. Sci.
2001, 494 (3), 238.
(22) Lin, T. H.; Somorjai, G. A. Surf. Sci. 1981, 107, 573.
(23) Somorjai, G. A.; Joyner, R. W.; Lang, B. Proc. R. Soc. London A
1972, 331, 335.
(24) McCrea, K. R.; Parker, J. S.; Somorjai, G. A. Topics in Catal., in
Additional evidence for CO dissociation being important for
the onset of ignition was found when CO oxidation reactions
were performed at equal pressures of CO and O2. The SFG
background increases when a carbon oxide species is on the
surface. When CO oxidation was performed on an initially clean
surface, the SFG background increased, indicating CO was
dissociating. The deposited carbon reacted with oxygen to
produce carboxylic anhydride. Thus, CO can dissociate under
oxidizing conditions to deposit carbon. In addition, the ignition
temperature was 50 deg lower for the initially carbon covered
press.
(25) Szabo´, A.; Henderson, M. A.; Yates, T., Jr. J. Chem. Phys. 1992,
96 (8), 6191.
(26) Akiyama, H.; Moise, C.; Yamanaka, T.; Jacobi, K.; Matsushima,
T. Chem. Phys. Lett. 1997, 272, 219.
(27) Otake, Y.; Jenkins, R. G. Carbon 1993, 31, 109.
(28) Dandekar, A.; Baker, R. T. K.; Vannice, M. A. Carbon 1998, 36,
1821.
(29) Haydar, S.; Moreno-Castilla, C.; Ferro-Garcia, M. A.; Carrasco-
Marin, F.; Rivera-Utrilla, J.; Perrard, A.; Joly, J. P. Carbon 2000, 38, 1297.
(30) Shriver, D. F.; Atkins, P.; Langford, C. H. Inorganic Chemistry,
2nd ed.; W. H. Freeman and Co.: New York, 1994.