Surface-enhanced Raman spectroscopy as an in situ real-time probe of catalytic mechanisms at high gas pressures: The CO-NO reaction on platinum and palladium

Article abstract of DOI:10.1006/jcat.1996.0305

Surface-enhanced Raman spectroscopy (SERS), combined with simultaneous mass spectrometric measurements, has been utilized to probe the reactive nature of surface species present during the reduction of NO by CO on Pt and Pd. As in our earlier studies, the SERS-active transition-metal surfaces are prepared by electrodepositing ultrathin films onto electrochemically roughened gold. These surfaces display remarkably robust SERS activity, enabling intense Raman spectra to be obtained over a range of reactant pressures (here up to 1 atm) and at temperatures up to at least 400°C. During nitric oxide adsorption at 1 atm on Pt, both terminal (240 and 470 cm^-1) and bridged (325 cm^-1) states of molecular NO were detected at lower temperatures (25-200°C), with some dissociation occurring at higher (ca 250°C) temperatures as evidenced by the presence of atomic nitrogen (295 cm^-1). Similarly, a bridged NO species (310 cm^-1) was observed on Pd under similar conditions, with dissociation detected in the form of atomic nitrogen (285 cm^-1) and surface oxide (450 and 665 cm^-1). Co-dosing of reactants on platinum produced a surface dominated by NO and CO (470 and 2080 cm^-1), whereas the former was adsorbed preferentially on Pd. Simultaneous SERS/MS measurements were performed during reaction of an equimolar reactant mixture at 1 atm of total pressure over both metals. Both CO₂ and N₂O were formed during reaction on Pt, with onset of detectable product formation correlating with depletion of adsorbed CO and NO, respectively. In contrast, CO₂ was the only product detected over Pd, with the depletion of surface oxygen suggesting that NO dissociation may be rate limiting at higher temperatures (ca 300°C). The extent of dissociation on these surfaces is compared and contrasted, with particular emphasis placed on its role in determining reaction selectivity. Furthermore, the overall behavior of these catalysts is compared with our former observations regarding this reduction process on rhodium.

Full text of DOI:10.1006/jcat.1996.0305

JOURNAL OF CATALYSIS 163, 63–76 (1996)
ARTICLE NO. 0305
Surface-Enhanced Raman Spectroscopy as an in Situ Real-Time Probe
of Catalytic Mechanisms at High Gas Pressures:
The CO–NO Reaction on Platinum and Palladium
�
�
�
�
Christopher T. Williams, Anish A. Tolia, Ho Yeung H. Chan, Christos G. Takoudis,
and Michael J. Weaver†
�
School of Chemical Engineering and †Department of Chemistry, Purdue University, West Lafayette, Indiana 47907
Received December 18, 1995; revised April 25, 1996; accepted April 30, 1996
bile emission standards which have been instituted over
Surface-enhanced Raman spectroscopy (SERS), combined with the years. The possibility of three products (CO2, N2O, and
simultaneous mass spectrometric measurements, has been utilized N ) raises issues regarding selectivity of different catalysts
2
to probe the reactive nature of surface species present during the
reduction of NO by CO on Pt and Pd. As in our earlier studies,
the SERS-active transition-metal surfaces are prepared by electro-
depositing ultrathin films onto electrochemically roughened gold.
and the mechanisms by which these reactions take place.
As a result, a large amount of effort has been expended
on the study of this reaction over platinum, palladium and
rhodium.
These surfaces display remarkably robust SERS activity, enabling
The ability of these catalysts to dissociate adsorbed NO
intense Raman spectra to be obtained over a range of reactant pres-
�
appears to lie at the heart of the issues of selectivity and
overall rates. While studies involving NO adsorption under
ultrahigh vacuum (UHV) conditions on both single crystal
sures (here up to 1 atm) and at temperatures up to at least 400 C.
During nitric oxide adsorption at 1 atm on Pt, both terminal (240
�
1
� 1
and 470 cm ) and bridged (325 cm ) states of molecular NO were
�
detected at lower temperatures (25–200 C), with some dissociation and polycrystalline Pt (1–11) and Pd (12–22) have revealed
occurring at higher (ca 250 C) temperatures as evidenced by the similar yet modest levels of NO dissociation, Rh is signifi-
presence of atomic nitrogen (295 cm ). Similarly, a bridged NO cantly more active in this regard (see Ref. (23) and cited
species (310 cm ) was observed on Pd under similar conditions,
with dissociation detected in the form of atomic nitrogen (285 cm
and surface oxide (450 and 665 cm ). Co-dosing of reactants on
platinum produced a surface dominated by NO and CO (470 and
�
�
1
�
1
literature therein). In the case of all three catalysts, more
highly stepped surfaces appear to facilitate NO dissociation
�
1
)
�
1
(
20, 24). The link between NO dissociation capability and
overall reaction rates is supported, for example, by investi-
gations of NO reduction over single crystal Pt surfaces (25–
�
1
2
080 cm ), whereas the former was adsorbed preferentially on Pd.
Simultaneous SERS/MS measurements were performed during re-
action of an equimolar reactant mixture at 1 atm of total pressure
over both metals. Both CO2 and N2O were formed during reaction
on Pt, with onset of detectable product formation correlating with
3
4). The Pt(111) surface was found to be practically inert,
while stepped Pt(410) was the most active face, probably
because of its greater ability to dissociate NO. However,
depletion of adsorbed CO and NO, respectively. In contrast, CO2 there is much controversy in the literature over whether
was the only product detected over Pd, with the depletion of sur- NO dissociation (25, 26, 34–36) or a bimolecular surface
face oxygen suggesting that NO dissociation may be rate limiting reaction between adsorbed NO and CO (37–42) is the rate-
�
at higher temperatures (ca 300 C). The extent of dissociation on
determining step. Related to these issues is the selectivity of
these materials towards CO2 and/or N2O formation. In our
recent study of the CO–NO reaction at atmospheric pres-
sure on Rh, CO2, and N2 were observed to be the only major
products (23). In contrast, both N2O and CO2 have been ob-
served during reaction on supported palladium (42, 43) and
platinum (35). However, other investigations involving Pt
these surfaces is compared and contrasted, with particular empha-
sis placed on its role in determining reaction selectivity. Further-
more, the overall behavior of these catalysts is compared with our
former observations regarding this reduction process on rhodium.
�
c 1996 Academic Press, Inc.
(
34, 37) and Pd (44) catalysts have reported CO2 and N2 as
INTRODUCTION
the major products.
The contradictory nature of many of these findings un-
The reduction of NO by CO over supported transition derscores the need for techniques that are able to exam-
metals has been investigated extensively due to automo- ine catalyst surfaces in situ under reaction conditions. With
63
0021-9517/96 $18.00
Copyright �c 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.

6
4
WILLIAMS ET AL.
its superior surface sensitivity, especially at high gas pres- controlled in order to limit the deposition to the desired
sures, surface enhanced Raman spectroscopy (SERS) can film thickness, usually less than five equivalent monolayers
offer a unique view of the surface species present under (45, 46).
technologically relevant conditions. We have demon-
strated, through earlier studies in our laboratory, that the of a 100 cm volume six-way cross, equipped with a turbo-
SERS effect can be extended to transition metals, includ- molecular pump (Balzers TPH 060) yielding a base pres-
The reactor utilized for the SERS experiments consists
3
�
6
ing Pt and Pd, by depositing them as thin films on rough- sure of 2 � 10 Torr (23). A combination of needle, gate,
ened gold electrodes (45, 46). Such films have been shown and butterfly valves allows operation of the reactor at pres-
�
3
to exhibit stable as well as intense SERS activity in the sures ranging from 10 to 760 Torr. The pressure over the
gas phase even at elevated temperatures (23, 47–52). As a entire range is measured using a combination of Pirani and
result, the wide frequency range along with the real-time cold-cathode gauges, controlled by a Balzers TPG 252 pres-
capability afforded by SERS may be exploited to yield sure controller. One of the reactor arms contains an optical
information regarding gas–surface interactions for reac- viewport, over which the specimen is mounted on a stain-
tive systems at high pressures that are not easily accessi- less steel cylindrical sample holder which can be heated to
�
ble through more conventional methods. By correlating 500 C. Inside the sample holder, a solid piece of stainless
the behavior of surface species with gas-phase reaction steel is surrounded by a coil of thin nichrome wire to pro-
products as monitored by simultaneous mass spectromet- vide resistive heating. Through the center of this heater is
ric (MS) analysis, information can be obtained regarding a K-type thermocouple which leads to an programmable
the possible role of the adsorbed molecules in the reaction temperature controller (Omega CN2041). Laser excitation
+
pathway (23).
is provided by a Kr laser at 647.1 nm (Spectra Physics)
We have recently reported results of a detailed examina- and scattered light is collected by a three-stage spectrome-
tion along these lines of the CO–NO reaction on Rh thin ter (SPEX Triplemate). The spectrometer is equipped with
films, utilizing this approach to elucidate the role of the a liquid nitrogen-cooled CCD (Photometrics), allowing ac-
adsorbed species (atomic nitrogen and CO) in the overall quisition of real-time spectral sequences typically even on
reaction (23). The study described herein for the same re- a seconds time scale with a good signal-to-noise ratio. Gas-
action on Pt and Pd films extends this effort to a pair of phase compositions are measured with a mass spectrome-
surfaces that, although being effective catalysts, dissociate ter (VG Instruments SX 200) interfaced with a Zenith Data
NO to a markedly smaller extent than on Rh. This forms Systems Z-248 computer. The gas is leaked into the mass
part of a continuing effort in our group to employ SERS spectrometer from a differentially pumped capillary line
combined with simultaneous mass spectrometric measure- that samples directly from the reactor.
ments of the gas phase to probe heterogeneous catalytic re-
Infrared reflection–absorption spectroscopic (IRAS)
actions. In order to gain more perspective on the CO–NO measurements were carried out in a Nicolet 800 FTIR
reaction, intercomparisons are made between this process spectrometer equipped with a MCT detector. The reac-
3
over Pt, Pd, and Rh. In addition, we present infrared spec- tor used is a small volume (5 cm ) chamber equipped with
tral data which provide an independent check on, as well KBr windows which could be evacuated to 0.1 Torr (Spec-
as a complement to, the SERS results.
tra Tech). The samples were heated using a heating ele-
ment identical to that described for the SERS reactor (vide
supra). Gases were mixed through a manifold and flow rates
were measured with a bubble flow meter. Spectra were
EXPERIMENTAL
The SERS-active surfaces were prepared utilizing 6 mm collected using near-grazing angle optics (Pike Technolo-
diameter discs cut out of 0.1 mm thick gold foil (Johnson gies) in a single-reflectance mode. Spectral features aris-
Matthey), and polished using 0.3 �m alumina powder. They ing from surface species were differentiated from their gas-
were then placed into an electrode holder that exposed phase counterparts by subtracting data obtained with s- and
2
Ca 7 mm of the surface and subjected to 25 oxidation– p-polarized light. Each spectrum was collected utilizing
�
1
reduction cycles from � 0.3 to 1.2 V versus SCE at 0.5 V 1500scansand 4cm resolution, and proportioned to back-
�
1
s
in 0.1 M KCl to produce the surface roughness neces- ground spectra obtained prior to the reactant dosage.
sary for Raman enhancement (53). The surface was rinsed
Both reactors behave as a well-mixed CSTR, as shown
thoroughly with deionized water and subsequently trans- by the transient response of the reactor composition with
ferred to another cell for electrochemical deposition of changes in gas flow conditions. The reactants are introduced
the Pt or Pd overlayer. Platinum was deposited from a 0.3 into the reactor through a manifold that allows mixing of
mM solution of H2PtCl6 in 0.1 M H2SO4 at 0.15 V ver- up to four inlet gases and flow rates are measured with a
sus SCE, while palladium was deposited from a 0.5 mM bubble flow meter. All reactants used in this study were of
solution of PdCl2 in 0.1 M HClO4 at 0.1 V versus SCE. ultra-high purity and obtained from Matheson and Airco
In both instances the cathodic charge was monitored and gases.

SERS OF THE CO–NO REACTION ON Pt AND Pd
65
RESULTS
ture (ca 3 min) for the spectra to attain time-independent
behavior.
Several features are apparent upon dosing of NO at room
temperature (bottom spectrum). Two sharp peaks are ob-
Platinum Films
�
1
Initial spectral examination of newly prepared platinum served at 240 and 325 cm along with two broader bands
�
1
thin films occasionally revealed SERS features at 330 and at ca 470 and 610 cm (Fig. 1A). Raman bands at 2180
�
1
� 1
2
60 cm associated with Pt–Cl and Au–Cl species remain- and 2250 cm are also evident in the corresponding higher
ing from the deposition process. In such cases, the sam- frequency spectra (Fig. 1B). The remaining spectra in the
3
� 1
�
ples were pretreated in 100 cm min H2 at 150 C for ca sequence, stacked in an upward direction, show the results
1
1
0 min, which resulted in a clean background in the 200– of heating the sample to the temperatures indicated. Upon
�
1
�
000 cm region. Nonetheless, broad bands resulting from heating to 150 C, the most noticeable change is the marked
�
1
surface carbonaceous impurities were commonly discerned intensity increase in the 2250 cm feature (Fig. 1B) which is
in the 1300–1600 cm region, removal of which required accompanied by milder intensity enhancements in the 240,
prior high-temperature oxidation. This, however, resulted 470, and 610 cm bands (Fig. 1A). At the same time, the
in severe disruption of the film with consequent partial loss 325 cm band is attenuated. Raising the temperature to
of SERS activity and was therefore not used here as part of 200 C results in the disappearance of the 2250 cm peak
the pretreatment procedure. and attenuation of the remaining features, along with the
�
1
�
1
�
1
�
� 1
�
1
The temperature-dependent adsorption of the individual appearance of a weak band at 295 cm . Further heating
�
reactants was examined with SERS prior to in situ real-time to 300 C caused attenuation of the remaining bands, with
probing of the surface under reaction conditions. Figure 1 final removal by 350 C.
�
shows a typical temperature-dependent spectral sequence
Assignment of these features to specific vibrations is not
3
� 1
for a Pt film exposed to 100 cm min of NO at 1 atm. entirely straightforward and is discussed below. In order to
The two spectral regions shown in Figs. 1A and B (ca 200– help the reader’s appreciation of the spectral results, how-
� � 1
1
8
00 cm and 1700–2400 cm , respectively) were obtained ever, we mention in advance our tentative assignment of the
�
1
separately in order to achieve satisfactory frequency reso- 240 and 470 cm bands to Pt–NO stretching and Pt–N–O
lution with the CCD detector employed. Prior to adsorp- bending modes, respectively, of NO adsorbed on atop (i.e.,
�
5
� 1
terminal) sites. The 325 cm feature is attributed to the
tion of the reactant, the reactor was evacuated to ca 10
Torr to purge the system of gas-phase contaminants. Fol- Pt–NO stretch of NO adsorbed on bridge sites.
lowing NO dosage, the sample was heated in a stepwise The adsorption of CO on platinum at room temperature
periodic fashion, allowing sufficient time at each tempera- resulted in spectra as shown in Figs. 2A, B. Two features can
3
� 1
FIG. 1. Temperature-dependent SER spectra for a Pt surface exposed to 100 cm min NO at 1 atm: (A) Low frequency spectra obtained with
2
0 s integration time; (B) High frequency spectra obtained with 60 s integration time.

6
6
WILLIAMS ET AL.
3
� 1
FIG. 2. SER spectrum of a Pt surface exposed to 100 cm min of CO at 1 atm: (A) Low frequency spectra obtained with 20 s integraton time;
(
B) High frequency spectra.
�
1
be seen at 470 and 380 cm in the low-frequency region The C–O stretch region shows a pair of peaks displaying
Fig. 2A) accompanied by bands at 2080 and 1890 cm� 1 in the marked “intensity transfer” expected for strong dipole–
the higher frequency spectra (Fig. 2B). Although heating dipole coupling: an intense high-frequency (�CO) feature
(
�
1
the sample in CO resulted in deteriorating signal quality (2060 cm ) and a markedly weaker lower frequency
�
1
due to loss of SERS activity, the bands appear to remain on (2010 cm ) band. As observed in Fig. 2B, the broad feature
�
the surface until at least 350 C. As in the SERS study of CO
and NO adsorption on Pt, Rh, and Ru by Wilke et al. (47),
�
1
the 470 and 2080 cm peaks are assigned to the Pt–CO
and C–O stretches of terminally adsorbed (i.e., atop) CO,
�
1
while the 380 and 1890 cm bands are similarly attributed
to CO bound to twofold bridging sites. A basis for these
assignments is that similar frequencies for both the Pt–CO
and C–O stretches have also been reported in FTIR (54)
and EELS (55) studies of CO adsorption on Pt surfaces
in UHV.
The form of the C–O and Pt–CO stretching bands for
13
12
adsorbed CO was also examined for various CO/ CO
isotopic mixtures in addition to dosing with 12CO as noted
above. These experiments were prompted primarily by the
insight that can be obtained into the extent of adsorbate
dipole–dipole coupling in this manner, and hence the ef-
fective local CO coverage present on the Pt surface (56).
Specifically, high local (i.e., microscopic) coverages of CO
will yield substantial dipole coupling for the C–O stretching
(
�CO) vibration. This is signaled in part by markedly higher
intensities of the higher-frequency (�CO) partner than ex-
pected on the basis of the relative 12CO/ CO coverages.
13
Typical data obtained during such an experiment, involv-
FIG. 3. High-frequency SER spectra obtained with 60 s integration
ing an equimolar 12CO/ CO mixture, is shown in Fig. 3. time of a Pt surface exposed to a mixture of 50% CO and 50% CO.
13
12
13

SERS OF THE CO–NO REACTION ON Pt AND Pd
67
ments at elevated temperatures. However, heating the Pt
�
film to temperatures greater than 150 C in vacuum (i.e.,
prior to reactant dosing) often resulted in an irreversible
loss of SERS activity, making it difficult to obtain repro-
ducible data under these conditions.
In order to obtain information linking the overall
temperature-dependent surface speciation with the cata-
lytic rates, the reaction for various CO and NO ratios on
the platinum films was probed by mass spectrometry (MS)
simultaneouslywith SERS. The generalprocedure followed
for these experiments was similar to the one used in our re-
cent study of the CO–NO reaction on Rh (23). The masses
monitored were 28 (CO and/or N2), 30 (NO), and 44 (CO2
and/or N2O), with data acquired continuously and recorded
every 5 s. Following exposure to each reactant mixture, sam-
ples were heated in a stepwise periodic fashion, waiting at
each temperature for a time-independent response to be
reached (ca 3 min) before spectra were acquired.
Figures 5A–C show such simultaneously obtained
temperature-dependent SERS (A, B) and MS measure-
ments (C) for a Pt surface exposed to an equimolar mixture
3
� 1
FIG. 4. High-frequency region of SER spectra obtained before and
after dosing NO on a CO-saturated surface.
of NO and CO, specifically50 cm min ofeach at 1 atm and
�
2
5 C. The ambient-temperature SER spectra show features
arising from both CO and NO, with low-frequency bands at
centered at 1830 cm 1 is indicative of CO adsorbed at 240, 325, 470, and 640 cm (Fig. 5A) and higher frequency
�
� 1
�
1
twofold bridging sites. This strong coupling effect confirms features at 2080, 2160, and 2230 cm (Fig. 5B). Upon heat-
�
the presence of a high local coverage of CO, an assertion ing to 125 C, the only noticeable change is the attenuation
�
1
�
which is supported by IRAS experiments discussed below. of the 640 cm feature. Heating further to 200 C, however,
As demonstrated in our earlier study of the CO–NO re- diminishes the intensities of the 325, 470, and 2080 cm
�
1
action on Rh (23), real-time SERS can be effectively uti- bands, the last indicating removal of adsorbed CO. Severe
�
lized to probe competitive adsorption of the two reactants, attenuation of all the SERS features is seen above 250 C,
�
1
which in turn can provide insight into reaction mechanisms. although traces of the 2230, 2160, and 470 cm features still
Since both NO and CO yield strong spectral features upon remain even at 350 C. The corresponding gas-phase com-
�
adsorption, experiments involving sequential reactant dos- position, as gleaned from the mass 44 intensity, is shown
�
ing were performed to assess the competitive nature of this in Fig. 5C. Raising the temperature to 200 C and beyond is
process. In each case, the sample was initially exposed to seen to trigger the onset of detectable reaction to form CO2
3
� 1
�
1
00 cm min of one reactant at 25 C until the spectral fea- and/or N2O (vide infra), with the signature mass 44 intensity
tures attained a constant intensity. The chamber was subse- exhibiting a monotonic increase with further temperature
quently evacuated to 10 Torr and then filled with 100 cm3 elevation, albeit somewhat muted above 250 C.
min of the other reactant, which essentially instanta-
neously (ca < 1 s) exposed the sample to 1 atm of pure gas. actant conditions using the CO isotope were also obtained
Typical results along these lines for a CO-predosed sur- in order to separate the contributions to the mass 44 inten-
face exposed subsequently to NO are shown in Fig. 4. Just sity arising from the two possible (CO2 and N2O) products.
�
5
�
�
1
Mass spectrometric measurements under equimolar re-
13
13
13
prior to NO addition, at t = 0 s (bottom spectra), a strong Masses 29 ( CO), 30 (NO), 44 (N2O), and 45 ( CO2) were
�
1
feature was observed at 2070 cm , indicative of CO ad- monitored together at 5-s intervals as the temperature was
sorbed at terminal sites. After ca 10 min exposure to NO raised in a stepwise fashion. Figure 6 shows a plot of the
(
top spectra), however, these features are no longer evident intensities of masses 44 and 45 for an equimolar mixture
�
1
3
� 1
and strong 2150 and 2240 cm bands have developed. The (10 cm min each) of CO and NO. The results indicate
loss of the CO spectral features indicates clearly that NO is detectable formation of both 1 CO2 (mass 45) above 200 C
3
�
�
able to displace CO under these conditions. The converse and N2O (mass 44) above 300 C, although the intensity of
experiment, involving CO dosing onto a NO-predosed sur- the latter is markedly smaller, suggesting that the major
face, yielded no detectable CO-induced changes in the NO product is CO2 under these conditions.
spectral features, showing that the reverse process is not
The reaction between CO and NO was also examined
facile. Attempts were also made to conduct similar experi- in an identical fashion to Fig. 5, but now utilizing mixtures

6
8
WILLIAMS ET AL.
3
� 1
FIG. 5. Simultaneous SERS and mass spectrometric measurements obtained during reaction of 100 cm min of 50% NO and 50% CO on Pt:
A) Temperature-dependent SER spectra in the low-frequency region obtained with 20 s integration time; (B) Temperature-dependent SER spectra
in the high-frequency region obtained with 60 s integration time; (C) Intensity of mass 44 vs time.
(
containing a fourfold excess of either CO or NO. Vibra-
Asalreadymentioned, a more cursoryexamination ofthe
tional features essentially similar to those observed under present system was also undertaken by means of infrared
equimolar conditions were also detected in those cases, reflection–absorption spectroscopy (IRAS). While IRAS
with SERS bands arising from both NO and CO adsorp- lacks the surface sensitivity and wide frequency range at-
tion being evident at room temperature. Reaction rates, as tainable by SERS, the former can yield surface vibrational
detected by the increase in mass 44 intensity, were found information that is a useful complement to the latter data.
to be considerably (ca threefold) lower than those for the This is partly because the infrared absorbances can pro-
13
equimolar ratio. Kinetics experiments utilizing CO failed vide semi-quantitative estimates of adsorbate coverages,
to detect any N2O formation, perhaps due to the lower over- since the IRAS responses should be free from any “special
all rates and hence weaker MS signals under these reaction surface-site” requirements that may affect the SER spec-
conditions.
tra. The samples used for these experiments were identical

SERS OF THE CO–NO REACTION ON Pt AND Pd
69
Palladium Films
Similar to platinum thin films, initial spectral examination
of the palladium overlayers often revealed a strong SERS
�
1
feature located at 300 cm associated with a Pd–Cl species
remaining from the deposition process. Samples exhibiting
�
3
� 1
this feature were pretreated at 100 C in 100 cm min of H2
for ca 5 min, which yielded a clean background in the 200–
�
1
1
000 cm region. Features associated with carbonaceous
�
1
impurities (1300–1600 cm ) were also observed but were
not removed by high-temperature oxidation (vide supra).
Following the protocol already outlined, before studying
the Pd surface under reaction conditions, the temperature-
dependent adsorption of each reactant was investigated
separately by utilizing SERS. Figure 8A shows a typi-
cal temperature-dependent spectral sequence in the low-
3
� 1
frequency region for a Pd film exposed to 100 cm min
of NO at 1 atm, with the experimental procedure the same
as detailed above. The adsorption of NO at room tempe-
rature (bottom spectra) yields an intense SERS feature at
3
� 1
FIG. 6. Isotopic kinetic experiment utilizing a flow of 20 cm min
1
3
of 50% CO and 50% NO at 1 atm. The temperature profile and the
intensities of masses 44 and 45 are presented as a function of time.
� 1 � 1
10 cm (with a shoulder at 285 cm ) and a broader band
3
�
1
at 570 cm . Unlike platinum, no spectral features were ob-
served in the 2100–2300 cm range. Upon heating to 100 C,
the intensity of the 310 and 570 cm features decreased and
�
1
�
�
1
1
�
�
to those utilized in the SERS experiments, except that the overlapping band at 285 cm became more prevalent.
they had a larger (1 cm ) surface area, obliged by the Raising the temperature to 150 C induced the disappear-
usual optical requirements of IRAS. Substantial gas-phase ance of the 310 cm feature and the severe attenuation of
interference, coupled with the low surface area of the cata- the 285 cm band, along with the appearance of a peak
lyst, precluded collection of atmospheric-pressure data, but at 245 cm . The 570 cm feature was also removed by
it was possible to obtain acceptable-quality surface signals this temperature. Further heating of the sample to 200 C
when using reactant pressures up to about 5 Torr. The best caused the removal of the 285 cm band, but without much
signals, however, were obtained when the gas-phase reac- change in the intensityofthe 245cm feature. Additionally,
2
�
1
�
1
�
1
� 1
�
�
1
�
1
�
1
tants were purged from the reactor prior to spectral acqui- two weak, broad bands were observed at 450 and 665 cm
sition. While such a procedure is obviously not convenient (Fig. 8A). Raising the temperature further resulted in the
for obtaining temperature-dependent spectral sequences, removal of the 245 cm band and a growth in the intensity
�
�
1
1
it is useful for examining atmospheric-pressure adsorption of the 450 and 665 cm features.
at ambient temperatures. This tactic was therefore used to
probe individual and co-adsorption of CO and NO at am-
bient temperatures.
An example of such data is presented in Fig. 7; this con-
sists of an infrared spectrum for a Pt thin film obtained
after exposure to one atmosphere of an equimolar mix-
ture of CO and NO. Along with the C–O stretch evident at
� � 1
1
2
080 cm is a peak at 2145 cm ; both these features are
similar to those observed by using SERS (Fig. 5B). Similar
experiments involving exposure to the different reactant
�
ratios (excess NO and excess CO) at 25 C also resulted in
spectra qualitatively similar to those observed with SERS.
Unfortunately, it was not possible in these preliminary ex-
periments to observe the N–O stretch, due largely to the
�
1
excessive spectral background in the 1700–1900 cm re-
gion arising from gas-phase water. However, exposing a
CO-presaturated surface to NO resulted in the removal of
the infrared band for adsorbed CO, a result concurring with
the SERS data in Fig. 4.
FIG. 7. Infrared absorption spectrum of 50% NO and 50% CO ad-
sorbed on a Pt film.

7
0
WILLIAMS ET AL.
3
� 1
FIG. 8. Temperature-dependent SER spectra for a Pd surface exposed to: (A) 100 cm min of NO at 1 atm and 20 s spectral integration time;
3
� 1
(
B) 100 cm min of O2 at 1 atm and 20 s spectral integration time.
In order to help diagnose which features formed during pure CO at room temperature (bottom spectra) and then
NO adsorption (Fig. 8A) may be associated with palladium subsequently heated. In the low-frequency region a band
3
� 1
� 1
was observed at 360 cm along with a weak shoulder at ca
oxide, a Pd surface was heated in pure O2 at 100 cm min
at 1 atm. Typical temperature-dependent SER spectra 470 cm 1. Two bands at 1960 and 2060 cm� 1 are also evident
�
resulting from this procedure are presented in Fig. 8B. No in the high-frequency region. Raising the sample temper-
�
spectral changes were observed until the sample was heated ature to 100 C resulted in the removal of the features at
�
� 1
to 200 C, where three features located at 240, ca 450 and 470 and 2060 cm , while the intensity of the other bands
65 cm appeared. Upon further heating to 250 C both decreased slightly. Upon further heating the latter bands
the broad ca 450 and sharp 665 cm features grew, while were markedly attenuated, however, being removed by ca
the 240 cm band was attenuated. Further heating of the 200 C. Carbon monoxide adsorption has been studied ex-
sample resulted in little change in the SER spectral inten- tensively on both polycrystalline (57) and single-crystal (58)
sities. On cooling the sample to 25 C, all three bands were Pd surfaces. Based on these studies, we assign the 360 and
present. These results contrast with those obtained for plat- 1960 cm bands to the Pd–C and CO stretching modes of
inum, where heating in pure oxygen under similar condi- CO bound to twofold bridging sites, with the 470 and 2060
tions yielded no SERS features. cm features being attributed to the Pt–C and CO stretch-
As for Pt, assignment of these features is not straight- ing vibrations of terminally bound CO.
�
1
�
6
�
1
�
1
�
�
�
1
�
1
forward and is discussed below. In order to aid the reader’s
Since both NO and CO adsorption again produced strong
appreciation of the spectral results, however, we again men- SERS features on palladium, experiments involving se-
�
1
tion some assignments in advance. The 310 cm band is quential dosing of reactants were performed to determine
assigned to the Pd–NO stretch of bridge-bound NO, while the nature of competitive adsorption. The protocol was
�
1
the 285 cm feature is attributed to the Pt–N stretch of identical to that utilized for platinum (vide supra). It was
adsorbed atomic nitrogen and is apparently similar to the observed that NO displaced CO from a palladium surface,
�
1
2
95 cm band observed on Pt at higher temperatures (vide as adjudged by the disappearance of adsorbed CO features
�
1
supra). The 245 cm feature is attributed to the Pd–O and their replacement by bands associated with NO adsorp-
stretch of a PdOx moiety. tion. The converse procedure indicated that CO could not
Typical temperature-dependent SER spectra arising replace NO. Interestingly, both these results are identical
from CO adsorption on palladium are shown in Figs. 9A to those obtained for platinum.
and B, again referring to low- and high-frequency segments,
respectively. The surface was exposed to 100 cm min of of these gases, the following transient experiment was
To gain further insight into the competitive adsorption
3
� 1

SERS OF THE CO–NO REACTION ON Pt AND Pd
71
3
� 1
FIG. 9. Temperature-dependent SER spectrum of a Pd surface exposed to 100 cm min of CO at 1 atm: (A) Low frequency spectra obtained
with 20 s integration time; (B) High frequency spectra obtained with 60 s integration time.
performed (Figs. 10A, B). A Pd surface was exposed to (note that the time values indicated above each spectrum
3
� 1
�
5
0 cm min of pure CO at 1 atm and 25 C until no further refer to the lapse between the addition of NO and the end
3
� 1
spectral changes were observed. At t = 0 s, 50 cm min of of each 10-s spectral acquisition period). The strong band
�
1
NO was added to the gas flow, so as to produce an equimolar at 360 cm , indicative of CO adsorption, is seen to be
CO/NO mixture, and SER spectra were obtained every 20 s slowly replaced (over ca 4 min) by the strong NO-associated
3
� 1
3
� 1
FIG. 10. (A) Time-dependent SER spectra of a Pd surface initially exposed to 50 cm min of CO, followed by addition of 50 cm min of NO to
�
the flow at t = 0 s at 1 atm and 25 C. Spectral integration time was 20 s. (B) High-frequency region of SER spectra obtained with 60 s integration time
before and 9 min after addition of NO to CO.

7
2
WILLIAMS ET AL.
platinum, no features were observed in the 2100–2300 cm 1
�
region.
Mass spectral measurements carried out under condi-
tions of equimolar reactants revealed an onset of detectable
�
reaction at 200 C, as evidenced by an increase in the mass
4
4 intensity. Experiments involving an equimolar mixture
13
of CO and NO indicated that only CO2 was being formed,
as evidenced by the appearance of only mass 45. The
temperature-dependent production of CO2 increased in a
manner similar to that observed for platinum (Fig. 5C), with
the overall change in mass 44 intensity (i.e., CO2 partial
pressure) indicating that the rates of reaction were similar
on both metals.
Unfortunately, the N–O stretching vibration for terminal
� � 1
1
(
1660–1800 cm ) and bridged (1400–1620 cm ) NO was
not observed on either Pt or Pd in this study. Similarly, we
were also unable to observe internal NO stretching modes
in our recent investigations of NO reduction on Rh (23, 52).
This was probably because of a low Raman scattering cross
section and interference from background features associ-
ated with carbonaceous impurities (vide supra). Indeed, the
internal N–O stretching vibration has only recently been
observed with SERS during an electrochemical study in-
volving NO adsorption on iridium (59). While this latter
FIG. 11. Temperature-dependent SER spectra obtained during reac-
tion of 100 cm min of 50% NO and 50% CO on Pd at 1 atm.
3
� 1
10 cm� 1 feature (Fig. 10A). Further evidence of the dis- case represented an optimal environment for the observa-
placement of CO is seen in the high-frequency spectra be- tion of the N–O stretch, due partly to its relatively high fre-
3
�
1
fore and 9 min after addition of NO (Fig. 10B), which re- quency (ca 1820 cm ) under these conditions, the band was
�
1
veals the two bands at 1965 and 2080 cm being completely weak in comparison with the analogous infrared spectra.
removed. These results show that NO can displace CO even
when the latter is given the advantage of continued pres-
ence in the gas phase as well as prior adsorption.
DISCUSSION
As with platinum, the reaction utilizing an equimolar
NO/CO mixture was studied on Pd with SERS and MS.
Identity of NO-related SERS Features
Figure 11 shows a typical temperature-dependent SER
Before discussingthe assignmentsofNO-related features
spectral sequence for a Pd surface heated in an equimo- on both Pt and Pd, it is useful to review our earlier find-
3
� 1
lar mixture of NO and CO, with 50 cm min of each ings regarding NO adsorption on rhodium (23). A single
�
1
at 1 atm. At room temperature (bottom spectrum) the SERS feature at 315 cm was induced at room tempera-
�
1
strong feature at 310 cm observed during NO adsorp- ture by NO adsorption on Rh, which grew in intensity as the
tion (Fig. 8A) is present, along with an overlapping band sample was heated, yet was attenuated beyond 300 C. This
at 285 cm . Heating the surface to 100 C resulted in the band was attributed to the Rh–N stretch of adsorbed atomic
diminution of the 310 cm band. Further heating to 150 C nitrogen, based partly on this temperature-dependent be-
caused severe attenuation of these features along with the havior (cf. Ref. (47)). Additionally, the surface was oxidized
appearance of a band at 245 cm . Bands at 665 and ca extensively at higher temperatures (>300 C) in a manner
50 cm appeared upon heating to 200 C, while the 245 similar to that reported elsewhere (51). These observations
cm peak remained unchanged and the 285 cm feature indicate the occurrence of extensive NO dissociation on
was completely removed. Subsequent heating caused the rhodium, even at room temperature. Based on literature
removal of all remaining bands by 300 C. It is interesting reports for NO adsorption of both Pt (1–11) and Pd (12–
�
�
1
�
�
1
�
�
1
�
�
1
�
4
�
1
� 1
�
that the observed spectral behavior mirrors closely that of 22), one would expect significantly less dissociation than
�
1
pure NO, except for the absence of a strong 570 cm band that observed for Rh. The present study shows that the
at room temperature. As might be expected from the re- temperature-dependent SER spectra of NO adsorption on
sults of the competitive adsorption experiment (Figs. 10A, Pt (Fig. 1) and Pd (Fig. 8A) at 1 atm are distinctly different
B), no features associated with adsorbed CO were discern- than for Rh. Keeping in mind the possible origins of these
able under these conditions. This is in contrast to the cor- features (such as NO, N, O), we now proceed with assigning
responding results for Pt (Figs. 5A, B). Also differing from the corresponding SERS bands observed on Pt and Pd.

SERS OF THE CO–NO REACTION ON Pt AND Pd
73
As detailed above, exposing a reduced Pt surface to 61–64), while the 2180 cm 1 band is assigned to either ad-
�
NO at room temperature yielded several Raman features sorbed cyanide (35) or isocyanate anion (60, 61). Both these
�
1
(
Figs. 1A, B). While we assigned the 325 cm band in an species are known to be formed by reaction of CO with
earlier report tentatively to the Pt–N stretch of adsorbed NO or its dissociation products, yet there is uncertainty re-
atomic nitrogen (52), that study lacked information on garding the location of these adsorbates with respect to
its temperature-dependent behavior. The attenuation the metal and support. In addition to supported catalysts,
of the band with increasing temperature as observed similar bands have been observed on polycrystalline (65)
here (Fig. 1A) is not suggestive of a dissociation process, and single-crystal Pt (4). This evidence suggests that the
�
� 1
especially since the removal temperature (ca 100–150 C) 2250 cm band observed in the present study (Fig. 1B)
is well below reported N2 desorption temperatures from Pt probably arises from an asymmetric stretching mode of ad-
(
1, 3, 7). It therefore seems more likely that this vibration sorbed isocyanate. The formation of this species during NO
arises from molecularly adsorbed NO. exposure, even in the absence of CO, indicates that NO,
Further insight into the nature of the species responsible or its dissociation products, reacts with surface carbona-
for this vibrational feature is gleaned from literature EELS ceous impurities. While the precise nature of the species
�
1
data obtained during NO adsorption on Pt single crystals. giving rise to the 2180 cm feature is not as clear (either
Ibach and Lewald (2) observed bands at 310 and 1510 cm
�
1
adsorbed cyanate ion or cyanide), it also appears to arise
at low coverages during NO adsorption on Pt(111). This from reaction between adsorbed NO and carbonaceous
�
bridging NO was observed to desorb by 150 C, the same impurities.
�
1
� 1
The assignment of the band at ca 610 cm is not im-
temperature at which the present 325 cm band is attenu-
ated. Similarly, Gland and Sexton (3) encountered 350 and mediately evident. One possibility is the N–C–O symmet-
470 cm bands for dilute NO adlayers, which were at- ric stretching mode of adsorbed isocyanate, which has
�
1
1
�
tributed to bridge-bound NO; this desorbed by ca 100 C. been reported in an EELS study of HNCO adsorption
These findings, therefore, support the assignment of the on Pt(111) (4). However, the lack of correlation between
�
1
�
1
3
25cm band observed here (Fig. 1A) to the Pt–NO stretch the temperature-dependent behavior of the 2250 cm fea-
�
1
of molecular NO possibly adsorbed on bridging sites.
ture with the ca 610 cm band casts doubt on this option.
By examining the temperature dependence of the 240 A more likely possibility is a surface carbonate formed
�
1
and 470 cm features (Fig. 1A), it is apparent that the through reaction between carbonaceous impurities and ad-
growth and eventual disappearance of these bands occur sorbed atomicoxygen formed byNO dissociation. The pres-
in tandem. Similarly related NO vibrations on Pt(111) have ence of such a carbonate species during reaction conditions
been observed with EELS by Gland and Sexton (3), who on supported Pt has been reported by Matyshak et al. [64].
recorded three peaks at high NO coverages (290, 450, and Lending credence to this assertion is our observation of a
�
1
1
710 cm ) associated with terminally bound NO. Pirug similar band when either Rh or Pt surfaces were heated in
et al. (9) have also assigned low frequency bands observed at CO/O2 mixtures.
�
1
�
2
30 and 380 cm upon dosing NO at 25 C on Pt(100) to Pt–
Nitric oxide adsorption on Pd surfaces yielded less com-
NO stretching and Pt–N–O bending modes, respectively, of plex spectra than for Pt, most notable being the absence of
terminalNO. These results, combined with vibrationalspec- higher-frequency bands attributable to cyanate-like moi-
tra of nitrosyl complexes (60), lead us to assign the 240 and eties. However, the room-temperature spectra (Fig. 8A)
�
1
4
75 cm features (Fig. 1A) to Pt–NO stretching and Pt–N– reveals several low-frequency vibrations associated with
O bending vibrations of terminally bound NO. The weak NO and its dissociation products. The attenuation of the
� � 1 � 1
1
�
2
95 cm feature that appears as the 240 cm band attenu- 310 cm band upon heating to 150 C is identical to the
�
� 1
ates at ca 200 C is probably not associated with a different behavior of the 325 cm NO band on platinum, suggesting
form of molecular NO, since TPD studies (1–6, 11) have in- that these features arise from similar species. Indeed, an
�
1
dicated that NO desorbs from Pt by this temperature. Since EELS study (13) of NO on Pd(100) reported a 310 cm
desorptive nitrogen recombination has commonly been ob- band at low coverage which was assigned to the Pd–NO
�
1
� 1
served at higher temperatures (1, 3, 4) and this 295 cm
stretch of bridged NO. The 285 cm feature, seen in Fig. 8A
�
band is attenuated by 300 C, we tentatively assign this vi- as a shoulder at room temperature and more clearly by
bration to the Pt–N stretch of adsorbed atomic nitrogen.
The high-frequency bands observed here on Pt at ca 2180 cates that the species is more strongly adsorbed than the
and 2250 cm during exposure to NO (Fig. 1B) have not 310 cm NO feature. The presence of adsorbed atomic
been reported in other vibrational studies. However, sev- oxygen (vide infra) also suggests that there is significant
eral infrared spectral reports of the NO–CO reaction on NO dissociation at room temperature. It is therefore likely
� �
100 C, is completely attenuated only at 200 C. This indi-
�
1
� 1
�
1
supported Pt have noted the occurrence of apparently simi- that this 285 cm vibration arises from the Pd–N stretch
�
1
� 1
lar bands (35, 61–64). The ca 2250 cm feature is commonly of adsorbed atomic nitrogen, similar to the 295 cm fea-
attributed to an adsorbed isocyanate species (–NCO) (35, ture observed on Pt at higher temperatures (vide supra).

7
4
WILLIAMS ET AL.
However, it is possible that this feature arises from a differ- even at room temperature (Figs. 1A, B). Further, the obser-
ent state of adsorbed NO.
vance of significant amounts of N2O formation during reac-
�
1
The 570 cm
band observed at room temperature tion over Pt at higher temperatures implies that significant
�
1
(
Fig. 8B) is similar to an EELS feature at 500–525 cm that dissociation is occurring, as adsorbed nitrogen is necessary
Ramsier et al. (20) reported when NO dissociated at around for this product to form (vide infra). The presence of atomic
�
2
00 C on both Pd(111) and Pd(112). This energy loss was oxygen and possibly nitrogen on Pd at room temperature
assigned to the Pd–O stretch of adsorbed atomic oxygen. indicates that dissociation has occurred (Fig. 8A), probably
While adsorption of oxygen at room temperature yielded to a greater extent than on Pt. The formation of palladium
no such SERS feature in this study, it has been reported oxide at higher temperatures provides a further indication
(
15) that NO adsorbed on Pd(111) can produce higher cov- that significant quantities of atomic oxygen are being pro-
erages of atomic oxygen than arise from dissociation of O2 duced by NO adsorption on Pd (Fig. 8A). As highly stepped
�
� 1
at 25 C. Furthermore, the 570 cm band appears to be ab- or kinked surfaces have been shown to be more active for
sent in the presence of CO (Fig. 11), indicating possible NO dissociation (20, 24), it is not surprising that signifi-
reactive removal (39, 66). We therefore assign this feature cant NO decomposition is occurring on the roughened sub-
to the Pd–O stretch of adsorbed atomic oxygen. The 665 strates utilized here.
�
1
�
and ca 450 cm bands observed above 150 C (Fig. 8A) are
similar to those observed during thermal O2 oxidation of
Pd (Fig. 8B), and are assigned to internal vibrations of a
Implications for the NO–CO Reaction Mechanism
Having discerned that both platinum and palladium dis-
PdOx moiety. In a related work, Weber et al. (67) investi- sociate NO to a certain extent, it is desirable to determine
gated oxidized Pd with Raman spectroscopy and observed the role that this process plays in determining the reaction
�
1
a strong band at 650 cm and weaker bands at 278, 423, pathway and selectivity. As we shall see below, the marked
� � 1
1
4
45, and 571 cm . The 245 cm feature observed at higher differencesbetween the NO decomposition activityofthese
temperatures during NO adsorption on Pd also appears at metals is consistent with the general notion that Rh is a bet-
similar temperatures during palladium oxidation (Fig. 8B). ter catalyst for NO reduction than Pt or Pd because of its
However, the temperature-dependent nature of this band greater ability to dissociate NO. Specifically, the observed
(
Fig. 8B) indicates that it is a behaviorally distinct oxide. selectivities towards CO2 and/or N2O can be, in part, ratio-
This is similar to the thermal oxidation of Rh (51), where nalized by the surface speciation detected under reaction
two distinct oxides were detected at higher temperatures.
conditions.
The commonly accepted reaction mechanism for NO re-
duction by CO over transition metals involves adsorption
Extent of NO Dissociation
Ascertaining the extent of NO dissociation on the cata- of reactants, dissociation of NO, and subsequent surface
lyst is important because it affects both the reaction path- reactions to form N2, N2O, and CO2. Generally, either NO
way as well as kinetics. As will be discussed later, this issue dissociation [1] or a bimolecular reaction between CO and
is central in explaining differences between Pt and Pd as NO [2]:
catalysts for the CO–NO reaction. Several studies have in-
NO(ads) + S ⇒ N(ads) + O(ads)
NO(ads) + CO(ads) ⇒ CO2 + N(ads) + S
[1]
[2]
dicated that NO dissociates to a modest extent on both
surfaces. At lower temperatures, only a small degree of dis-
sociation has been detected on Pt(111) (2–6) and Pt(110)
(
4, 7), while Pt(100) appears to be the most active low-index are thought to be rate limiting (S denotes a surface site).
face (4). Polycrystalline (9–11) as well as supported (62) Pt It has been suggested by several investigators that the NO
have also been shown to exhibit some activity towards NO dissociation reaction [1] is rate determining on supported
dissociation (10–15% ) at elevated temperatures. On palla- (34) as well as polycrystalline (35) Pt, while more recent ki-
dium, low levels of NO dissociation have been reported for netic (37) and spectroscopic (38) studies on polycrystalline
the (111) (14, 16, 20), (110) (14), (100) (12, 14), and (112) Pt have pointed to the bimolecular step [2] as being rate
(
20) faces, in addition to significant dissociation at higher limiting. There is also much debate on which step is rate
�
temperatures (>200 C) on polycrystalline surfaces (21, 22). determining on Pd. Support for the bimolecular RDS can
Our results appear consistent with these findings. We be found in studies on single crystal (39, 40), polycrystalline
have found rhodium to be especially active with regards (41), and supported (42) Pd. In contrast, Moriki et al. (36)
to NO dissociation (23), with more extensive dissociation have proposed that NO dissociation is rate limiting for re-
detected than for either platinum or palladium. However, action over polycrystalline Pd. As these studies were con-
Pd does appear to be more active than Pt in this regard. ducted over a wide range of pressures (from UHV up to
While molecularly adsorbed NO is the dominant surface 1 Torr), the RDS for this reaction appears to be distinctly
�
species on Pt below 250 C, the presence of carbonate and pressure-sensitive. Additionally, the temperature of reac-
isocyanate species suggests that some dissociation occurs tion can influence the RDS, since processes such as NO

SERS OF THE CO–NO REACTION ON Pt AND Pd
75
�
dissociation and surface oxidation are more facile at higher centration of these species at 300 C may be at least partly
temperatures (vide infra). Unland (60, 61) has also sug- rationalized, based on their removal via N2O formation.
gested that the formation of isocyanate species by the com-
In contrast with the product distribution observed for
bination of adsorbed atomic nitrogen and CO may be an platinum, CO2 was found to be the only detectable product
important step. However while isocyanate is a commonly during reaction over Pd (vide supra). Examining the species
observed surface species, it is not generally considered to be present on the surface during the reaction (Fig. 11) pro-
a significant intermediate (35). Indeed, lack of a frequency vides some insight into this process. As might be expected,
shift in the observed isocyanate vibrations during transient based on the transient replacement of adsorbed CO by ad-
13
CO isotope experiments over Pt in the present study pre- dition of gas phase NO (Figs. 10A, B), CO adsorption is
cludes this possibility. effectively blocked at all temperatures by NO and its dis-
Unlike the CO–NO reaction on Rh (23), where CO2 and sociation products. While this observation is different than
N2 were found to be the only products, reaction on Pt un- that observed for Pt above, it is very similar to the absence
der conditions of equimolar gas flow resulted in formation of adsorbed CO observed during reaction over Rh (23).
�
of both N2O and CO2 (Fig. 6). Therefore it is particularly At the onset of detectable reaction (200 C), molecular NO
�
1
interesting in this case to see if correlations can be made (310 cm ) has been removed and the surface is dominated
�
1
between catalyst selectivity and behavior of the surface by PdOx species (670, ca 450 and 245 cm ). As N2O for-
species. While NO can displace CO from a CO-saturated mation necessarily requires the presence of adsorbed NO
surface (Fig. 4), both reactants adsorb to a comparable ex- and atomic nitrogen at adjacent sites, oxidation of the sur-
tent on Pt over a wide range of reaction conditions if they face may hamper this process at these temperatures. The
are simultaneously present in the gas-phase (Figs. 5A, B). removal of oxide features by ca 300 C (Fig. 11) is a clear in-
Between 150 C and 200 C, the amount of CO on the sur- dication that CO is reactively removing oxygen, as this moi-
face decreased, as evidenced by a reduction in the intensity ety has been observed to desorb from palladium surfaces at
of the 2080 cm feature (Fig. 5B). It is unlikely that this much higher temperatures (Fig. 8B; Ref. (65)). Therefore,
is due to desorption, which would not be anticipated until the reaction between adsorbed CO and atomic oxygen is
higher temperatures (>350 C) at this pressure (68). Fur- occurring fast enough that the oxygen supply on the sur-
�
�
�
�
1
�
thermore, isocyanate formation from CO can also be ruled face is severely depleted. This dissipation of surface oxygen
out as a reason for these observations, based on the absence may indicate that NO dissociation is rate-limiting at these
13
13
of N CO formation when CO was used. Thus, the most temperatures.
probable explanation for CO removal from the surface lies
The fact that Pt shows some selectivity towards N2O at
in its reaction with adsorbed atomic oxygen from NO dis- higher temperatures, while Rh and Pd do not, appears to
sociation or with adsorbed NO in a bimolecular reaction be related to their relative abilities to dissociate NO. While
step. The simultaneous mass spectrometric measurements N2O formation would be facilitated by the availability of
(
4
Fig. 5C) show the first detectable appearance of the mass substantial quantities of molecular NO on Pt, it would be
4 peak at 200 C, the temperature corresponding to the hampered at lower temperatures by limited atomic nitro-
�
disappearance of adsorbed CO and decrease in the amount gen, thus resulting in detectable production only at higher
�
1
of terminally-bound NO (240 and 470 cm ). In addition, temperatures. Rhodium and palladium, on the other hand,
the 13CO isotope experiment under the same conditions favored NO dissociation more strongly, which led to a com-
(
Fig. 6) reveals that CO2 is the only product at this temper- pletely nitrogen and oxygen covered surface at higher tem-
ature. Such a correlation between the behavior of surface peratures that most likely hampered N2O formation in both
and gas-phase species indicates that CO is being reactively cases. Similarly, a recent report (69) involving this reac-
removed from the surface. However, it is difficult to say tion over Rh (111) and (100) at relatively high pressures
which, if not both, of the above reaction pathways might be (100 Torr) suggests that N2O selectivity decreases as the
predominant in the formation of CO2, since either appears extent of NO dissociation increases.
consistent with the observed product formation.
It is also desirable to similarly link the behavior of ad-
sorbed NO to the onset of reaction. It is clear that the ad-
CONCLUDING REMARKS
�
1
sorbed NO molecules evidenced by the 240 and 470 cm
Surface-enhanced Raman spectroscopy, combined with
�
features remain on the surface as high as 250 C but are seen mass spectrometry, has been employed to investigate NO
to decrease upon further heating (Fig. 5A). The equimolar reduction by CO over platinum and palladium under cata-
13
CO isotope experiment (Fig. 6) shows the onset of de- lytically relevant conditions. There are several similarities
tectable N2O production at these temperatures, indicating and differencesbetween the behavior ofsurface moietydur-
that there issome contribution from thisproduct in the mass ing both reactant adsorption and reaction. During NO ad-
4
4 peak observed by simultaneous mass spectrometric mea- sorption, molecular NO dominated both surfaces at lower
surements (Fig. 5C). Thus, the decrease in the surface con- temperatures, with a small degree of dissociation observed

7
6
WILLIAMS ET AL.
for each metal. However, while NO and its dissociation 32. Gardner, P., Martin, R., Tushaus, M., and Bradshaw, A. M., Surf. Sci.
2
69/270, 405 (1992).
3. Park, Y. O., Banholzer, W. F., and Masel, R. I., Surf. Sci. 155, 341
1985).
4. Banse, B. A., Wickham, D. T., and Koel, B. E., J. Catal. 119, 238
1989).
products effectively blocked adsorption of CO during re-
action over Pd, significant quantities of adsorbed CO were
detected on Pt under similar conditions. While CO2 and N2
were the only products observed during reaction over Pd,
3
3
(
(
N2O formation was detected at higher temperatures for Pt. 35. Lorimer, D., and Bell, A. T., J. Catal. 59, 223 (1979).
3
6. Moriki, S., Inoue, Y., Miyazaki, E., and Yasumori, Y., J. Chem. Soc.
Faraday Trans. 1 78, 171 (1982).
These findings, coupled with our previous observations dur-
ing the NO–CO reaction on Rh, suggest that the higher the
ability to dissociate NO, the lower the selectivity toward
N2O for these metals, at atmospheric pressure.
37. Klein, R. L., Schwartz, S., and Schmidt, L. D., J. Phys. Chem. 89, 4908
(
1985).
38. Scharpf, E. W., and Benziger, J. B., J. Catal. 136, 342 (1992).
39. Davies, P. W., and Lambert, R. M., Surf. Sci. 110, 227 (1981).
40. Yamada, T., Matsuo, I., Nakamura, J., Xie, M., Hirano, H., Matsumoto,
Y., and Tanaka, K., Surf. Sci. 231, 304 (1990).
ACKNOWLEDGMENT
4
1. Xi, G., Bao, J., Shao, S., and Li, S., J. Vac. Sci. Technol. A 10, 2351
This work is supported by a grant to CGT and MJW (CTS-9312008)
from the National Science Foundation.
(
1992).
4
2. Grill, C. M., Gonzalez, R. D., J. Phys. Chem. 84, 878 (1980).
3. Butler, J. D., and Davis, D. R., J. Chem. Soc. Dalton Trans. 21, 2249
(1976).
4
REFERENCES
44. El Hamdaoui, A., Bergeret, G., Massardier, J., Primet, M., and
1
. Comrie, C. M., Weinberg, W. H., and Lambert, R. M., Surf. Sci. 57, 619
1976).
Renouprez, A., J. Catal. 148, 47 (1994).
45. Leung, L.-W. H., and Weaver, M. J., J. Am. Chem. Soc. 109, 5113
(1987).
46. Leung, L.-W. H., and Weaver, M. J., Langmuir 4, 1076 (1988).
47. Wilke, T. E., Gao, X., Takoudis, C. G., and Weaver, M. J., Langmuir
7, 714 (1991).
48. Wilke, T. E., Gao, X., Takoudis, C. G., and Weaver, M. J., J. Catal. 130,
62 (1991).
(
2
3
4
5
6
7
8
9
. Ibach, H., and Lehwald, S., Surf. Sci. 76, 1 (1978).
. Gland, J. L., and Sexton, B. A., Surf. Sci. 94, 355 (1980).
. Gorte, R. J., Schmidt, L. D., and Gland, J. L., Surf. Sci. 109, 367 (1981).
. Campbell, C. T., Ertl, G., and Segner, J., Surf. Sci. 115, 309 (1982).
. Hayden, B. E., Surf. Sci. 131, 419 (1983).
. Gorte, R. J., and Gland, J. L., Surf. Sci. 102, 348 (1981).
. Bonzel, H. P., and Pirug, G., Surf. Sci. 62, 45 (1977).
. Pirug, G., Bonzel, H. P., Hopster, H., and Ibach, H., J. Chem. Phys. 71,
49. Tolia, A. A., Wilke, T. E., Weaver, M. J., and Takoudis, C. G., Chem.
Engr. Sci. 47, 2781 (1992).
9
3 (1979).
50. Tolia, A. A., Takoudis, C. G., and Weaver, M. J., J. Vac. Sci. Technol.
A 11, 2013 (1993).
51. Tolia, A. A., Smiley, R., Delgass, W. N., Takoudis, C. G., and Weaver,
M. J., J. Catal. 150, 56 (1994).
1
1
1
0. Mummey, M. J., and Schmidt, L. D., Surf. Sci. 109, 29 (1981).
1. Wickham, D. T., Banse, B. A., and Koel, B. E., Surf. Sci. 223, 82 (1989).
2. Jorgenson, S. W., Canning, N. D. S., and Madix, R. J., Surf. Sci. 179, 332
(
1987).
52. Tolia, A. A., Williams, C. T., Takoudis, C. G., and Weaver, M. J.,
Langmuir 11, 3438 (1995).
1
1
3. Nyberg, C., and Uvdal, P., Surf. Sci. 204, 517 (1988).
4. Sugai, S., Watanabe, H., Kioka, T., Miki, H., and Kawasaki, K., Surf. 53. Gao, P., Gosztola, D., Leung, L.-W. H., and Weaver, M. J., J. Elec-
Sci. 259, 109 (1991). troanol. Chem. 233, 211 (1987).
5. Conrad, H., Ertl, G., Kuppers, J., and Latta, E. E., Surf. Sci. 65, 235 54. Hoge, D., Tushaus, M., Schwiezer, E., and Bradshaw, A. M., Chem.
1
(
1977).
Phys. Lett. 151, 230 (1988).
1
1
6. Schmick, H.-D., and Wassmuth, H.-W., Surf. Sci. 123, 471 (1982).
7. Miyazaki, E., Kojima, I., Orita, M., Sawa, K., Sanada, N., Miyahara,
T., and Kato, H., Surf. Sci. 176, L841 (1986).
55. For example, see Steininger, H., Lehwald, S., and Ibach, H., Surf. Sci.
123, 264 (1982).
56. For example, see Hollins, P., and Pritchard, J., Prog. Surf. Sci. 19, 275
1
1
2
8. Bertolo, M., and Jacobi, K., Surf. Sci. 226, 207 (1990).
9. Wickham, D. T., Banse, B. A., and Koel, B. E., Surf. Sci. 243, 83 (1991). 57. Bradshaw, A. M., and Hoffman, F. M., Surf. Sci. 52, 449 (1975).
0. Ramsier, R. D., Gao, Q., Waltenberg, H. N., and Yates, J. T., Jr., 58. Bradshaw, A. M., and Hoffman, F. M., Surf. Sci. 72, 513 (1978).
(1985).
J. Chem. Phys. 100, 6837 (1994).
1. Matsumoto, Y., Onishi, T., and Tamaru, K., J. Chem. Soc. Faraday 60. For example, see Miki, E., Mizumachi, K., Ishimori, T., and Okuno,
Trans. 1 76, 1116 (1980). H., Bull. Chem. Soc. Japan 47, 656 (1974).
2. Miki, H., Nagase, H., Nagase, T., Kioka, T., Sugai, S., and Kawasaki, 61. Unland, M. L., J. Catal. 31, 459 (1973).
K., Appl. Surf. Sci. 33/34, 292 (1988). 62. Unland, M. L., J. Phys. Chem. 77, 1952 (1977).
3. Tolia, A. A., Williams, C. T., Takoudis, C. G., and Weaver, M. J., 63. Solymosi, F., Sarkany, J., and Schauer, A., J. Catal. 46, 297 (1977).
59. Zou, S.-Z., Gomez, R., and Weaver, M. J., in preparation.
2
2
2
J. Phys. Chem. 99, 4599 (1995).
64. Matyshak, V. A., Gazarov, R. A., Panchishnyi, V. I., and Kadushin,
A. A., Kinet. Catal. 29(6), 1382 (1988).
65. Solymosi, F., and Bansagi, T., J. Phys. Chem. 83, 552 (1979).
66. Stuve, E. M., Madix, R. J., and Brundle, C. R., Surf. Sci. 146, 155 (1984).
67. Weber, W. H., Baird, R. J., and Graham, G. W., J. Raman. Spec. 19,
239 (1988).
2
2
2
2
2
2
3
3
4. Morrow, B. A., Chevrier, J. P., and Moran, L. E., J. Catal. 91, 208 (1985).
5. Lambert, R. M., and Comrie, C. M., Surf. Sci. 46, 61 (1974).
6. Gorte, R. J., and Schmidt, L. D., Surf. Sci. 111, 260 (1981).
7. Bonzel, H. P., and Fischer, T. F., Surf. Sci. 51, 213 (1975).
8. Banholzer, W. F., and Mazel, R. I., Surf. Sci. 137, 339 (1984).
9. Lesley, M. W., and Schmidt, L. D., Chem. Phys. Lett. 102(5), 459 (1983). 68. Van Slooten, R. F., and Nieuwenhuys, B. E., J. Catal. 122, 429 (1990).
0. Schwartz, S. B., and Schmidt, L. D., Surf. Sci. 183, L269 (1987).
1. Schwartz, S. B., and Schmidt, L. D., Surf. Sci. 206, 169 (1988).
69. Peden, C. H. F., Belton, D. N., and Schmieg, S. J., J. Catal. 155, 204
(1995).