Methanol Oxidation on Palladium Compared to Rhodium at Ambient Pressures as Probed by Surface-Enhanced Raman and Mass Spectroscopies

Article abstract of DOI:10.1006/jcat.1998.1955

The influences of surface speciation upon the catalytic oxidation kinetics of methanol on palladium under ambient-pressure flow-reactor conditions was examined by surface-enhanced Raman spectroscopy (SERS) combined with mass spectrometry (MS) and compared with corresponding data on rhodium. The former technique provides uniquely sensitive surface vibrational information under real-time (≈1 s) in situ conditions, by utilizing ultrathin catalyst films electrodeposited onto an inert SERS-active gold substrate. These transition-metal surfaces exhibit sufficiently robust SERS activity to enable temperature-dependent spectral measurements over the range 25-500°C. Parallel kinetic measurements undertaken with MS show the occurrence of methanol decomposition (to CO and H₂) in the absence of O₂ on both Pd and Rh. While the presence of a molar deficiency of O₂ yields methanol oxidation (to form CO₂ and H₂) in addition to methanol decomposition on Rh, only the latter occurred (at slower rates than Rh) on Pd. These dissimilar reaction selectivities are consistent with the absence of surface vibrational features on the latter surface and the observed presence of adsorbed CO on the former. The behavior can be rationalized by the paucity of adsorbed atomic oxygen, O_(ad), on Pd compared with Rh arising from the greater ability of the latter to dissociatively chemisorb O₂. Both catalysts induced exhaustive methanol oxidation (yielding CO₂ and H₂O) in a heavily O₂-rich reactant mixture, although Pd again yielded less facile reaction kinetics. In addition, a significant catalyst deactivation occurred upon heating Pd in this reactant mixture, which was entirely absent on Rh. The corresponding temperature-dependent SER spectra indicate the formation of palladium oxide (PdO) by 350°C, which was retained entirely upon subsequent cooling. While an oxide (Rh₂O₃) was also seen by SERS to form on Rh by 350° C under these conditions, this species was removed upon subsequent cooling. Transient SERS measurements following sudden exposure of such oxidized surfaces to a methanol gas stream revealed that PdO was entirely unreactive toward methanol even at 350° C, while, in contrast, Rh₂O₃ was removed entirely within ca. 5 s. This remarkable difference in oxide reactivity, which accounts for the Pd catalyst deactivation, was deduced to be due primarily to the inability of methanol to yield a suitable adsorbed oxygen scavenger by dissociative chemisorption on Pd. The possible involvement of a methoxy intermediate in the reaction on Pd under O₂-rich conditions, as suggested by the SERS data, is also discussed.

Full text of DOI:10.1006/jcat.1998.1955

JOURNAL OF CATALYSIS 174, 191–200 (1998)
ARTICLE NO. CA981955
Methanol Oxidation on Palladium Compared to Rhodium at Ambient
Pressures as Probed by Surface-Enhanced Raman
and Mass Spectroscopies
�
�
,1
,1
Ho Yeung H. Chan, Christopher T. Williams, Michael J. Weaver,† and Christos G. Takoudis‡
�
School of Chemical Engineering and †Department of Chemistry, Purdue University, West Lafayette, Indiana 47907; and ‡Department
of Chemical Engineering, University of Illinois at Chicago, Chicago, Illinois 60607
Received August 1, 1997; revised November 6, 1997; accepted November 17, 1997
ciative chemisorption on Pd. The possible involvement of a methoxy
The influences of surface speciation upon the catalytic oxidation intermediate in the reaction on Pd under O2-rich conditions, as sug-
kinetics of methanol on palladium under ambient-pressure flow- gested by the SERS data, is also discussed.
�c 1998 Academic Press
reactor conditions was examined by surface-enhanced Raman spec-
troscopy (SERS) combined with mass spectrometry (MS) and com-
pared with corresponding data on rhodium. The former technique
provides uniquely sensitive surface vibrational information under
real-time (� 1 s) in situ conditions, by utilizing ultrathin catalyst
films electrodeposited onto an inert SERS-active gold substrate.
These transition-metal surfaces exhibit sufficiently robust SERS
activity to enable temperature-dependent spectral measurements
INTRODUCTION
The catalytic oxidation of methanol on noble transition
metals has drawn considerable attention, due partly to en-
vironmental concerns regarding control of volatile organic
compound formation from industrial processes as well as
automobile emissionsstandards. Fundamentally, it servesas
a relatively simple model catalytic system, involving issues
such as how different bond cleavage routes (C–O or O–H)
�
over the range 25–500 C. Parallel kinetic measurements under-
taken with MS show the occurrence of methanol decomposition
(to CO and H2) in the absence of O2 on both Pd and Rh. While
the presence of a molar deficiency of O2 yields methanol oxida-
tion (to form CO2 and H2) in addition to methanol decomposition
may determine product selectivity. An understanding of the
on Rh, only the latter occurred (at slower rates than Rh) on Pd. oxidation pathways on different metals can also provide
These dissimilar reaction selectivities are consistent with the ab- mechanistic insight into the reverse reaction (i.e., CH OH
3
sence of surface vibrational features on the latter surface and the synthesis on supported noble metal catalysts from CO/CO2
observed presence of adsorbed CO on the former. The behavior can
be rationalized by the paucity of adsorbed atomic oxygen, O(ad), on
Pd compared with Rh arising from the greater ability of the latter
to dissociatively chemisorb O2. Both catalysts induced exhaustive
The decomposition of methanol on Pd (1–12) and Rh
methanol oxidation (yielding CO2 and H2O) in a heavily O2-rich
with H2) as well as other catalytic oxidation processes.
Two metals which have received interest with regard to
catalytic methanol oxidation are palladium and rhodium.
(
13–15) single crystals in ultrahigh vacuum (UHV) at low
reactant mixture, although Pd again yielded less facile reaction ki-
netics. In addition, a significant catalyst deactivation occurred upon
heating Pd in this reactant mixture, which was entirely absent on
Rh. The corresponding temperature-dependent SER spectra indi-
temperatures has been studied extensively, to gain funda-
mental insight into the initial steps of the reaction. It is
generally accepted that methanol decomposes readily at
�
cate the formation of palladium oxide (PdO) by 350 C, which was room temperature via the formation of a methoxy interme-
retained entirely upon subsequent cooling. While an oxide (Rh2O3) diate (metal–OCH ) to form adsorbed CO and H. In par-
3
�
was also seen by SERS to form on Rh by 350 C under these con- ticular, experiments involving the predosage of O prior to
2
ditions, this species was removed upon subsequent cooling. Tran-
sient SERS measurements following sudden exposure of such ox-
CH3OH exposure (attempting to mimic practical reaction
conditions) are common and have revealed that on most
idized surfaces to a methanol gas stream revealed that PdO was
transition metals the extent of methanol adsorption as well
�
entirely unreactive toward methanol even at 350 C, while, in con-
trast, Rh2O3 was removed entirely within ca. 5 s. This remarkable
difference in oxide reactivity, which accounts for the Pd catalyst
deactivation, was deduced to be due primarily to the inability of
methanol to yield a suitable adsorbed “oxygen scavenger” by disso-
as surface methoxy stability are increased on both Pd (16–
1
8) and Rh (19, 20). Considering the large differences be-
tween UHV experiments and those more relevant to tech-
nological catalysis, it is important to garner mechanistic and
surface chemical information under higher-pressure con-
ditions. However, there have been relatively few studies
of catalytic methanol dissociation and oxidation on Pd
1
To whom correspondence should be addressed.
191
0021-9517/98 $25.00
Copyright �c 1998 by Academic Press
All rights of reproduction in any form reserved.

1
92
CHAN ET AL.
(
21–24) and Rh (24–26), particularly with regard to in situ on the former surface as well as a weaker ability of Pd to
monitoring of adsorbed species (23). The highest reactant dissociatively chemisorb the methanol as well as oxygen
partial pressures that have been examined during kinetic reactants.
investigations are � 0.1–1 Torr and reveal that CO2 and
H2O are the major products (21–26) during reaction of
EXPERIMENTAL
mixtures containing a molar excess of oxygen. In contrast,
exhaustive methanol oxidation can be hampered in excess
The procedure for preparing SERS-active samples has
methanol gas (26). It is consequently of substantial inter- been detailed previously and involves electrodeposition of
est to account for these and other kinetic observations in Pd and Rh ultrathin films (about three to five monolayers)
terms ofthe surface speciation and reactivity under reaction on electrochemically roughened gold substrates (27, 28).
conditions.
Gold disks (10 mm in diameter) were cut from 0.1-mm-
With its superior surface sensitivity, especially at high thick gold foil (Johnson Matthey) and polished with alu-
(
ambient or beyond) gas pressures, surface-enhanced mina powder. They were subsequently placed into an elec-
2
Raman spectroscopy (SERS) can offer a unique view of trode holder that exposed 10 mm of surface in 0.1 M KCl
the surface species present under such technologically rel- solution, followed byoxidation–reduction cyclesto produce
evant conditions. In earlier studies from our laboratory, we a SERS-active gold substrate (41). Palladium films were
have demonstrated that the SERS effect can be extended to deposited from a 0.05 M PdCl solution in 0.1 M HClO4
2
adsorbates on transition metals, including Pd (27) and Rh at +0.1 V vs saturated calomel electrode (SCE). Similarly,
(
28), by electrodepositing them as thin films on roughened Rh films were deposited from a 0.05 M RhCl solution in
3
gold substrates. Such surfaces were subsequently shown 0.1 M HClO at +0.1 V vs SCE. The desired film thickness
4
to exhibit stable as well as intense SERS activity also in was controlled by monitoring the cathodic charge during
the gas phase, even at elevated temperatures (29–40). As deposition.
a result, the wide vibrational frequency range afforded by
SERS may be exploited to yield information regarding gas– with a turbomolecular pump (Balzers TPH 060) yielding
The high-pressure reactor is a six-way cross equipped
�
6
surface interactions for reactive systems at high pressures a base pressure of 10 Torr. The pressure over the en-
that are not easily accessible with more conventional meth- tire range is measured using a combination of Pirani and
ods. By correlating the presence and reactivity (or other- cold-cathode gauges (Balzers PKR 250), controlled by a
wise) of surface species with gas-phase reaction products Balzers TPG 252 pressure controller. Inside the sample
as monitored by mass spectrometric (MS) analysis, infor- holder, a solid piece of stainless steel is surrounded by a coil
mation can be obtained regarding the possible role of the of thin nichrome wire to provide resistive heating. Through
adsorbed moleculesin the reaction pathway(24, 37, 38, 40). the center of this heater is a K-type thermocouple which
The advent of charge-coupled device (CCD) technology is in intimate contact with the sample and leads to a pro-
along with the intrinsically intense SERS signals also al- grammable temperature controller (Omega CN2041). Re-
lows the acquisition of spectral sequences on a rapid (� 1 s) actants were delivered through a manifold which allows
time scale, thereby enabling us to probe the temporal evolu- mixing of up to four gases. Methanol vapor in argon carrier
tion of surface species by transiently perturbing the system gas was generated by sparging Ar through liquid methanol.
parameters (e.g., temperature, reactant ratio) in a gas-flow (This procedure yielded a maximum methanol partial pres-
reactor arrangement and monitoring the real-time spectro- sure in the gas stream of ca. 110 Torr.) The reactants utilized
scopic response.
were 99.9% pure MeOH (Fisher) and gases of ultrahigh pu-
We have recently reported a SERS study of methanol rity grade (Airco). Methanol oxidation experiments were
oxidation on Rh (39). In addition to detecting Rh2O3 for- performed at ambient pressure (1 atm) and elevated tem-
�
mation in situ under reaction conditions, it was shown that peratures (25–500 C). Two reaction mixtures with different
this species was highly reactive toward gas-phase methanol. O /CH OH ratios were utilized (one with large O excess
2
3
2
The role of the oxide along with other adsorbates detected and the other being marginally richer in CH OH).
3
by SERS during this reaction has also been investigated
Laser excitation at 647.1 nm for SERS was provided by
+
further in a detailed SERS study combined with reaction a Kr laser (Spectra Physics). The scattered light was col-
kinetics obtained by MS (40). Herein is presented a com- lected by a three-stage spectrometer (Triplemate) equipped
bined SERS–MS examination of methanol oxidation on Pd with a CCD detector (Photometrics). The Raman frequen-
in comparison with corresponding findings on Rh, with em- cies were calibrated by lines from a neon light-emitting
phasis placed on the effects of surface oxygen and lattice diode.
oxide on reaction rates and selectivity. The findings demon-
strate how the less facile oxidation kinetics on Pd compared performed separatelyin a small(ca. 4cm ) stainless-steelre-
with Rh, along with different reaction selectivities, can be actor which could be pumped down to 10 Torr. A bundle
understood on the basis of a relatively nonreactive oxide of gold wire (99.998% pure, Johnson Matthey) was used
Corresponding mass spectrometric experiments were
3
�
6

METHANOL OXIDATION ON PALLADIUM
193
as the substrate rather than the gold foil used in SERS ex- phase composition wasmonitored byMS. The experimental
�
periments, in order to produce a larger surface area (about protocol involved a series of ca. 50 C temperature ramps
7
2
�
cm ), to facilitate the kinetic measurements. [Although (starting from 25 up to 500 C), holding at each temperature
we have previously undertaken MS analysis simultaneously until the MS signals reached a constant level. These MS
with SERS (e.g., Ref. 37), the parallel MS experiments experiments revealed that methanol decomposes to yield
were employed instead here to optimize the sensitivity of exclusively CO and H2 on both Pd and Rh (40), with the
�
the kinetic measurements.] The Pd and Rh thin films on this onset of detectable product formation at 350 and 250 C, re-
�
gold substrate were prepared exactly as for the SERS ex- spectively. After excursions to 500 C, the temperature was
periments (vide supra). The gas-phase composition during decreased incrementally to detect any thermally induced
the reaction was measured by leaking the gas outlet into a deactivation of the catalyst. (The total time consumed for
quadrupole mass spectrometer (Lybold Inficon IPC-50).
the increasing and decreasing temperature segments was
about 40 min in both cases.) The CO production rates on
Rh were drastically decreased (to ca. 10% ) from the corre-
sponding values measured during the initial upward tem-
perature ramp, probably due to the progressive deposition
of carbon on the surface at higher temperatures (vide infra).
The methanol decomposition rates on Pd were decreased
to about 25% of the values measured with increasing tem-
perature during the subsequent cooling segment.
Corresponding SERS experiments involved essentially
the same protocol as that utilized for the MS measure-
ments. The surface temperature was increased in a stepwise
fashion, with SER spectra acquired using an integration
time of 20 s after a time-independent spectral response
was achieved (requiring ca. 2–3 min). Identical heating
procedures were utilized in all experiments described be-
low unless otherwise noted. No spectral featuress below
RESULTS AND DISCUSSION
Since results for methanol oxidation on Rh will be avail-
able elsewhere (40), we only summarize the most important
findings for this catalyst as they compare with those for Pd.
Both SERS and MS results obtained from corresponding
experiments on Pd and Rh surfaces are presented together
to facilitate such comparisons. Pure methanol decompo-
sition will be discussed first, followed by methanol oxida-
tion using two reactant mixtures, one deficient in O2 and
the other in excess O2. Steady-state MS and SERS results
are presented as a function of surface temperature and will
be interpreted in concert to link observed reaction rates
measured by the former technique with the presence of ad-
sorbates sensed by the latter. Finally we describe transient
SERS experiments performed on a seconds time scale to
determine the reactivity of the Pd and Rh oxides formed
during the reaction toward pure gas-phase methanol, to
gain further insight regarding their probable kinetic role
toward methanol oxidation.
� 1
�
ca. 1000 cm were evident upon methanol dosing at 25 C
on both Pd and Rh, suggesting the absence of adsorbed
species (at least at moderate to high coverages) under these
reaction conditions. (On both Pd and Rh, however, broad
� 1
SERS bands around 1350 and 1580 cm developed under
Freshly deposited samples used in both SERS and MS
these conditions, indicative of surface carbon formation.)
Furthermore, no additional SERS bands were observed on
either surface even upon increasing the temperature into
the range where gaseous CO production could be detected
�
experiments were first pretreated by heating in H2 at 150 C
for 5 min to remove traces of adsorbed chloride from the
deposition solution and obtain a reduced Pd or Rh surface.
This procedure produced a featureless SERS background
� �
350–500 C for Pd and 250–500 C for Rh).
(
�
1
in the 200–1000 cm region while leaving some weak SERS
�
1
bands between 1300 and 1600 cm due to residual carbona-
ceous contamination. Complete removal of these bands re-
quires thermal oxidation and subsequent reduction at high
temperatures, which typically results in severe loss of SERS
activity. Therefore, this latter pretreatment step was not
used here.
Methanol Oxidation: O2-Deficient Case
We next examine the occurrence of nonoxidative me-
thanol decomposition and oxidation by O2 in the pres-
ence of a molar excess of methanol. Figure 1A displays
an Arrhenius plot extracted from steady-state MS results
3
� 1
obtained for a Pd surface exposed to 100 cm min of a
methanol-rich mixture (13.5% CH3OH/76.5% Ar/10% O2
Methanol Decomposition
It is desirable to examine initially the interaction of both by volume) at 1 atm. The only products detected by MS
Pd and Rh with oxygen-free methanol gas in order to re- under these conditions were CO and H2, with the onset of
�
veal how this molecule decomposes in the absence of the detectable reaction occurring at 250 C. A constant activa-
�
1
oxidant. After samples (either Pd or Rh) were subjected to tion energy of 7 (� 1) kcal mol was obtained from 300 to
�
the H2 pretreatment, the reactor was evacuated and sub- 500 C. In contrast to the O2-free case, essentially identical
sequently flushed with Ar. Samples were then heated in a T-dependent kinetics were observed during the upward and
methanol-saturated Ar stream (15% CH3OH/85% Ar by subsequent downward temperature segments; i.e., no signi-
3
� 1
volume) at 100 cm min and 1 atm, during which the gas- ficant hysteresis was observed. The rates of CO formation

1
94
CHAN ET AL.
FIG. 1. (A) Arrhenius plot extracted from MS measurements of temperature-dependent CO production rates from a flowing mixture
3
� 1
� 1
(
2
100 cm min ) of 13.5% CH3OH/10% O2/76.5% Ar at 1 atm on Pd. (B) As for A, but for CO and CO2 production rates during exposure of Rh to a
3
50 cm min reactant flow.
�
1
� 1
1
are converted to turnover numbers (molecules site s ) observation of adsorbed CO contrasts the SERS findings
5
in Fig. 1A by assuming a surface site density of 1.8 � 10
for pure methanol exposure on Rh, in which no adsorbates
were detected, and indicates that additional surface chem-
�
2
sites cm
.
Similarly, Fig. 1B displays MS results obtained for expos- istry occurs as a result of the presence of oxygen. Signifi-
ing Rh to the same excess methanol mixture as described cantly, neither the Pd nor the Rh surfaces displayed SERS
3
� 1
� 1
above, but at a faster reactant flow rate, 250 cm min
.
bands at 1300–1600 cm in the presence of oxygen, indi-
While gaseous CO and H2 were detected on Rh beginning cating the lack of surface carbon formation under these
�
at ca. 250 C, in contrast to the Pd case, CO2 production conditions.
�
along with H2 (but not H2O) was also observed by 200 C.
Summarizing the influence of O2 on the kinetic data
At a given temperature, the rates of CO formation on Rh noted above, the major effect in the case of Pd is that
are about two to three timesfaster than the same reaction on the temperature required for (nonoxidative) methanol de-
�
Pd. Similarly to Pd, these kinetics displayed essentially no composition to CO is lowered by about 100 C in the pres-
T-dependent hysteresis. The activation energy for CO2 for- ence of 10% O2 (Fig. 1A), with the T-induced hysteresis
�
1
�
mation on Rh is12(� 1) kcalmol from 200to 350 C, above being removed. However, on Rh the presence of O2 also
which the reaction rates level off (Fig. 1B). However, the results in CO2 formation (Fig. 1B), although it has little
CO formation rates increase continuously throughout the influence on the temperature at which gaseous CO was
�
�
entire temperature range (250–500 C), although exhibiting first detected (ca. 250 C). Insight into such effects of O2 on
�
1
a lower activation energy of 8 (� 1) kcal mol .
the methanol decomposition kinetics on these metals can
Corresponding SERS experiments during reaction of this be obtained by comparing the results with previous stud-
methanol-rich mixture on both metals were also performed. ies involving methanol exposure on clean and O-covered
As for the oxygen-free case, no spectral features were de- surfaces in UHV. It has been shown that predosed oxy-
�
tected on Pd between 25 and 500 C. However, markedly gen assists methanol decomposition on both Pd (16–18)
different results were obtained for a Rh surface, in that and Rh (19, 20) for two reasons. First, oxygen provides an
�
1
two SERS bands at 465 and ca. 2000 cm were observed additional methanol decomposition pathway involving the
�
even upon dosing the methanol-rich mixture at 25 C. These abstraction of the methanolic hydroxyl hydrogen by coad-
features are assigned to the �Rh–CO and �C–O stretching vi- sorbed atomic oxygen, O(ad), to form surface hydroxyl and
brations, respectively, of atop (i.e., terminal) adsorbed CO methoxy (16–20):
�
(
40). Both bands remained until 350 C and disappeared at
higher temperatures (results not shown; see Ref. 40). This
CH OH
+ O(ad) ⇒ CH O
+ OH(ad)
.
[1]
3
(ad)
3
(ad)

METHANOL OXIDATION ON PALLADIUM
195
Second, the thermal stability of methoxy is increased by the The presence of 10% O2, however, lowers the apparent
presence of neighboring oxygen atoms [16–20]. activation energy for CO formation on Pd from roughly
�
1
This promotion effect of O(ad) on methanol decomposi- 14 (� 3) kcal mol during the upward temperature seg-
�
tion can partly explain the observed differences between ment (from 350 to 450 C) for pure methanol decompo-
�
1
Pd and Rh regarding selectivity and surface speciation un- sition, down to 7 (� 1) kcal mol . At least the observed
�
der reaction conditions. One can account for both the CO2 ca. 100 C diminution in temperature required for the de-
formation (Fig. 1B) and the presence of adsorbed CO ob- tectable onset of gaseous CO formation on Pd when O2 was
served on Rh, but not on Pd, in terms of the higher affinity present could therefore also be due to the influence of the
of the former surface for oxygen. A previous study of dioxy- O-assisted decomposition [1].
gen dissociation trends on noble transition metals by means
of temperature-programmed desorption (TPD) has shown
that oxygen adsorbs more extensively on Rh than Pd (42).
Methanol Oxidation: O2-Rich Case
Furthermore, atomic oxygen formed in such a fashion ex-
Figure 2A displays the Arrhenius plot extracted from
hibits a higher thermal stability on Rh. Consequently, then, steady-state MS results obtained while heating a Pd surface
3
� 1
the appearance of CO-related SERS features on Rh upon in a 100 cm min flow of a reaction mixture containing a
methanol-rich mixture dosage at room temperature (vide large excess of O2 (3% CH3OH/80% O2/17% Ar by vol-
supra) may well arise from the oxygen-induced methanol ume). Essentially an exhaustive oxidation of methanol was
decomposition pathway [1]. The smaller affinity of Pd for achieved so that CO2 and H2O were the exclusive products,
�
O2 should yield much lower O(ad) coverages and therefore with first detectable reaction occurring at 250 C (Fig. 2A,
can readily explain the lack of formation of adsorbed CO on open circles). This observation contrasts the kinetic results
Pd as a result of limited methanol decomposition. Further obtained for the excess methanol case (Fig. 1A), in which
supporting this notion, a given rate of gaseous CO forma- CO and H2 were the only products. Furthermore, the appar-
�
� 1
tion from methanol decomposition requires a roughly 50 C ent activation energy of the reaction is 5 (� 1) kcal mol
�
higher temperature on Pd than on Rh (Figs. 1A and 1B).
between 250 and 400 C during the upward temperature
Such higher O(ad) coverages on Rh can also account sequence (open circles, Fig. 2A), lower than that for CO
for the formation of CO2 (Fig. 1B) in the presence of O2 formation in the excess methanol mixture. Interestingly,
via oxidation of adsorbed CO. Along the same lines, the when the surface temperature of Pd was lowered incremen-
�
presence of only a very low O(ad) coverage on Pd can ac- tally from 500 C the reaction rates (filled circles) exhibited
count for the exclusive formation of gaseous CO (Fig. 1A). some hysteresis with respect to the preceding increasing
FIG. 2. (A) Arrhenius plot extracted from MS measurements of CO2 production rates from a flowing mixture (100 cm3 min ) of 13.5%
� 1
CH3OH/10% O2/76.5% Ar at 1 atm on Pd. Open and filled symbols refer to rate measurements obtained during stepwise heating and subsequent
cooling segments, respectively (See text for details). (B) As for A, but for Rh exposed to 250 cm3 min of reactants. (Note that the rates during
� 1
subsequent cooling on Rh were the same as those during the heating segment.)

1
96
CHAN ET AL.
temperature segment (open circles). As already noted, this mixture presumably yields relatively high O(ad) coverages,
was not observed in the excess methanol case (Fig. 1A). causing the O-assisted methanol decomposition pathway
Shown in Fig. 2B is the Arrhenius plot extracted from cor- [1] to become dominant. Since such a step would likely be
responding steady-state MS results obtained while heating more energetically favorable than the “unassisted” frag-
3
� 1
Rh in the excess O2 mixture. Again, a large (250 cm min
)
mentation route, this notion can explain the lower activa-
flow rate was used. As on Pd, an exclusive production of tion energies observed for CO2 production under excess
CO2 and H2O was observed, yielding an apparent activa- oxygen conditions (40). Furthermore, this interpretation
�
1
tion energy of 7.5 (� 1) kcal mol . These kinetics on Rh is supported by the exclusive detection of gaseous H2O
therefore show a different product selectivity and a lower product, rather than H2, along with CO2 (i.e., exhaustive
activation energy in comparison with reaction in the excess oxidation) under excess O2 conditions. The formation of
methanol mixture (Fig. 1B).
water during the reaction probably arises by combining ad-
However, there are several significant differences evi- sorbed H and OH, the latter being produced in O-assisted
dent between the kinetics of methanol oxidation on Pd and methoxy formation. Indeed, high selectivities toward H2O
Rh in the excess oxygen mixture. First, the latter surface versus H2 production have been observed in UHV stud-
is more catalytically active than the former, exhibiting a ies of methanol dissociation on oxygen predosed Rh(111)
two- to three-fold higher reaction rates at a given temper- surfaces (19, 20).
ature. Perhaps more intriguing, however, is the hysteresis
In the wake of the information regarding kinetics and
observed in the temperature-dependent CO2 formation ki- product selectivity obtained from the above MS experi-
netics on Pd (Fig. 2A), the effect not being observed for the ments, corresponding SER spectra were collected with an
corresponding measurements on Rh (Fig. 2B). This ther- eye toward correlation of the observed gas-phase product
mally induced catalyst deactivation on Pd suggests an alter- formation with surface speciation. Figure 3A shows typical
ation of the surface state and will be discussed in light of temperature-dependent SER spectra obtained for a Pd film
the corresponding SERS results (vide infra).
exposed to the excess O2 reaction mixture, again obtained
It is of interest to consider likely reasons for the sub- using a similar protocol to that used with the corresponding
�
1
stantially different product selectivities on both Pd and Rh MS measurements. A band at 255 cm was detected upon
�
observed in changing from an O2-deficient to an O2-rich dosing of reactants at 25 C (bottom spectrum), the inten-
�
reactant mixture. The observed gaseous reaction products sity of which begins to attenuate by 200 C, being removed
indicate the occurrence of the following three overall reac- completely at higher temperatures. While no spectral fea-
�
tions:
tures were observed between 250 and 300 C, raising the
temperature further yieldsthe appearance ofa pair ofSERS
�
1
CH3OH(g) ⇒ CO(g) + 2 H2(g)
CH3OH(g) + 1/2 O2(g) ⇒ CO2(g) + 2 H2(g)
CH3OH(g) + 3/2 O2(g) ⇒ CO2(g) + 2 H2O(g).
[2]
[3]
[4]
bands at 500 and 700 cm . These features were stable at
Thus, while exhaustive methanol oxidation [4] is observed
under excess oxygen conditions on both metals, only the
partial oxidation process [3] along with some nonoxidative
methanol decomposition [2] are observed in a molar excess
of methanol on Rh and only the nonoxidative reaction [2]
for the latter condition on Pd. As already noted, the ability
of Rh to catalyze reaction [3] under the latter condition is
probably due to the higher O2 affinity of this metal. Chang-
ing from excess methanol to a very low CH3OH/O2 molar
ratio also lowers the apparent activation energy from 12
�
1
to 7.5 kcal mol for CO2 production on Rh, suggesting a
mechanistic change. Although switching from the higher
to low CH3OH/O2 molar ratio also decreases the activa-
�
1
tion energy on Pd, from 7 to 5 kcal mol , the reaction also
changes from exclusive CO/H2 to CO2/H2O formation un-
der these conditions.
At least the results on Rh may be rationalized by consi-
dering the above interpretation regarding the possible ef-
fects that O(ad) exerts on the surface processes governing
FIG. 3. (A) Temperature-dependent SER spectra obtained duringex-
3
� 1
posure of Pd to 100 cm min flow of 13.5% CH3OH/10% O2/76.5% Ar
at 1 atm. The spectral acquisition time was 20 s. (B) As for A, but for Rh
3
� 1
methanol oxidation. Exposing either metal to an 80% O2 with 250 cm min reactant flow.

METHANOL OXIDATION ON PALLADIUM
197
�
temperatures as high as 500 C and remained upon cooling CO2 production rates on the former surface during the
�
to 25 C (Fig. 3A, top spectrum). Similar bands have been downward-temperature portion (Fig. 2A, filled circles) are
observed previously with SERS during thermal oxidation likely due to blocking of metal surface sites by PdO. In con-
of Pd with pure O2 (37) and are assigned to palladium oxide trast, the disappearance of the Rh oxide hands upon cool-
(
PdO) based partly on similarities with the normal Raman ing (Fig. 3B, top spectrum) indicates reactive removal at
spectra of the bulk material (43, 44). lower temperatures by methanol, resulting in regeneration
Typical temperature-dependent SER spectra obtained of the metallic sites for the reaction to proceed. Indeed, as
during similar exposure of the Rh surface to the excess we have recently shown (39, 40), the Rh2O3 formed in the
O2 mixture are displayed in Fig. 3B. Dosage of the reac- O2-rich mixture is extremely reactive with pure gas-phase
�
tion mixture at 25 C resulted in a weak SERS band at methanol; discussion of this issue is the topic of the next
�
1
2
50 cm . There was essentially no change upon heating section.
until 250 C, whereupon the intensity of this band increased
up to 350 C before decreasing somewhat at higher tem- Transient SERS Experiments
�
�
peratures (Fig. 3B). The thermal stability of the species re-
Prompted by the desire to elucidate whether the oxide
formed at higher temperatures is involved directly in the
methanol oxidation catalysis, the transient reduction kinet-
ics of Rh2O3 and PdO by pure methanol were investigated
by means of SERS. The experimental procedure has been
described in detail elsewhere (39, 40) and exploits the sub
second time resolution afforded by SERS. [While the SER
spectra cannot be used as a reliable monitor of oxide cov-
erage, this is likely to be high, at least circa one monolayer,
partly on the basis of XPS measurements for Rh2O3 (33).]
Oxide surfaces were initially formed by heating in the O2-
�
1
sponsible for this 250 cm feature on Rh is therefore much
higher than the adsorbate yielding the otherwise similar 255
�
1
cm band on Pd. Furthermore, the advent of the 510 and
5
�
1
�
60 cm bands on Rh at temperatures higher than 350 C
(
Fig. 3B) is indicative of the formation of another species.
This pair of SERS features are assigned to �Rh–O stretches
of Rh2O3, based on both a previous combined SERS and
X-ray photoelectron spectroscopy (XPS) study of Rh sur-
face oxidation (33) and the normal Raman spectra of bulk
rhodium oxide (33, 43). In the contrast to PdO, however,
both Rh2O3 SERS features disappeared when the sample
was cooled to room temperature (Fig. 3B, top spectrum).
The above SERS results obtained during reaction in the
O2-rich mixture on Pd and Rh therefore again show some
notable similarities and differences. No SERS bands associ-
ated with adsorbed CO were detected on either metal under
�
rich reaction mixture at 500 C. The flow was then switched
to pure Ar at this temperature, so to displace the gas mix-
ture from the reactor. The surface was then cooled to the
desired temperature, and the gas flow suddenly switched to
a methanol-containing Ar stream (39, 40). Figures 4A and
4
B show ensuing time-dependent SER spectra acquired
�
1
these conditions. While ca. 250 cm SERS bands were evi-
following exposure of PdO and Rh2O3, respectively, to
�
dent on both metals upon reactant dosing at 25 C, these
3
� 1
�
1
200 cm min of 15% methanol in Ar at 350 C.
features exhibit a marked difference in thermal stability.
Intriguingly, the Pd oxide is entirely nonreactive with
(
The assignments of the band on Pd, along with its possible
�
methanol even at 350 C as evidenced by the lack of any
change in the intensity of the oxide bands even over a
ca. 10 min time scale (Fig. 4A). In striking contrast, Rh2O3
shows very high reactivity toward gas-phase methanol, be-
ing removed entirely within 5 s following the onset of
methanol exposure under the same conditions (Fig. 4B).
This rapid reduction of rhodium oxide by methanol has
been attributed to the efficient removal of adsorbed oxy-
gen by methanol fragments, such as adsorbed CO, thereby
driving the conversion
role in the methanol oxidation reaction, will be discussed
below.) Another common feature of the results for Pd and
Rh is the emergence of oxide-related SERS bands at tem-
�
peratures higher than 350 C. These appear to be the first
in situ observations of oxide formation on such metals dur-
ing methanol oxidation. Interestingly, however, the temper-
ature at which both metals begin to oxidize under O2-rich
�
�
reaction conditions (ca. 350 C) is 150 to 200 C higher than
those observed during thermal oxidation of these surfaces
by pure O2 (33, 37). A plausible explanation is that, rather
than being incorporated into the metal lattices, surface oxy-
gen preferentially reacts with CH3OH and its dissociation
products. Significantly, the observed exclusive formation
O2 ⇒ O(ad)
�
[5]
of CO2 and H2O, i.e., reaction [4] at these temperatures irreversibly toward the right (40). (The lattice oxide is de-
noted by O2 in reaction [5].) Two distinct kinetic regimes
�
(
Figs. 2A and 2B), lends support to this supposition.
Interestingly, the complete retention of PdO upon cool- were identified by examining the Rh O removal kinetics
2
3
ing in the presence of the O2-rich reactant mixture, versus as a function of methanol partial pressure as well as tem-
the removal of Rh2O3 upon cooling, can account for the perature (40). At higher methanol pressures and/or lower
temperature-dependent kinetic hysteresis observed on Pd temperatures, reaction [5] appeared to be rate determining,
and its absence on Rh (Figs. 2A and 2B). Thus the lower yieldingan apparent activation energyof12(� 2) kcalmol�
1

1
98
CHAN ET AL.
periments that showed that PdO is rapidly removed (within
�
ca. 10 s) even at 250 C by dosing either with 1 atm of CO or
H2. [In the former case, removal of the characteristic PdO
SERS features is accompanied by the appearance of a band
�
1
at around 650 cm , tentatively attributed to adsorbed car-
bonate (cf. Ref. 38).] The Rh2O3 film was also found to be
reactive toward both CO and H2 under these conditions
(
50).
Given that Rh2O3 is found to be efficiently removed
by methanol, the question remains whether Rh2O3 partici-
pates importantly in the methanol oxidation reaction. This
issue has been investigated in detail elsewhere (40) by uti-
lizing a kinetic analysis to model the Rh2O3 removal by
methanol. It was found that the removal rates were about
two orders of magnitude lower than the rates of methanol
oxidation, suggesting that Rh2O3 is not a major reaction
intermediate, even though its reactive nature prevents the
oxide from acting as a surface “poison” in the manner of
FIG. 4. (A) Time-dependent SER spectra of an oxide-covered Pd PdO.
�
surface at 350 C following introduction of 1 atm 15% CH3OH/85% Ar
3
� 1
at 100 cm min . The oxides were initially formed in an O2-rich reac- Spectroscopic Evidence for Adsorbed Methanol Fragments
tion mixture and then flushed with Ar before suddenly introducing the
While the assignments of the SERS bands detected at
CH3OH/Ar stream (at time t = 0). The spectral acquisition time was 10 s.
�
(
B) Similarly to A, but for an oxide-covered Rh surface at 350 C following higher temperatures to oxides are straightforward, discern-
3
� 1
.
sudden introduction of 1 atm 15% CH3OH/85% Ar at 1200 cm min
The spectral acquisition time was 0.5 s.
ing the identity of the apparently reactive species responsi-
� 1
ble for the 255 cm SERS feature on Pd constitutes a signi-
�
ficant challenge. The band was removed by 250 C (Fig. 3A),
4
� 1
and a preexponential factor of 3.5 � 10 s . However, at with the corresponding MS data revealing first detectable
lower methanol pressures (ca. 10 Torr and below) and/or CO production at the same temperature (Fig. 2A). These
2
higher temperatures, the reaction rate became almost T in- results therefore suggest that the species responsible for the
�
1
dependent, attributed tentativelyto a rate-determiningstep 255 cm feature may be an active surface intermediate in
involving adsorptive formation of the methanol fragment methanol oxidation. One possibility is that it arises from the
or reaction with adsorbed oxygen (40).
�
Pd O
stretch of adsorbed atomic oxygen from dissociative
–
The nonreactivity of PdO toward methanol observed adsorption of O . However, we observed the absence of
2
18
here is remarkable since this species should be less stable a discernable frequency downshift upon O isotopic sub-
2
than Rh2O3 in terms of “dissociation pressures” [the O2 par- stitution, eliminating this possibility. Molecularly adsorbed
tial pressure in equilibrium with a dissociating bulk-phase methanol also seems an unlikely candidate due to its ex-
metal oxide at a given temperature (45, 46)]. This unex- tremely low thermal stability; it has been found to desorb
pected stability of the palladium oxide may be due to the in UHV by 150 K (11).
fact that it is a thin film (about three to five monolayers).
Surface methoxy (Pd–OCH ) is another possible candi-
3
Indeed, oxides grown on such films have been reported to date. This speculation is partly based on its comparable fre-
be more thermally stable than their bulk-phase analogues quency to the values obtained from EELS studies (1, 3, 8).
(
47, 48). A recent SERS study of Ru oxidation also revealed These UHV investigations find, however, that methoxy is
�
a marked difference in stability between thin-film and bulk- not stable on Pd at temperatures above 25 C, decompos-
phase RuO4 (49).
ing to adsorbed CO and H. Nevertheless, the fact that no
An alternative explanation for the nonreactivity of PdO SERS feature associated with adsorbed CO [which has a
�
1
toward methanol is that the surface concentration of the characteristic �
“
–
band at 360 cm (37)] was observed
Pd CO
�
adsorbed oxygen scavenger” species [i.e., methanol frag- at 25 C, coupled with the lack of detectable CO or CO
ment(s)] is very low on Pd, as a result of the relative im- formation at this temperature, suggests that the 255 cm
2
1
�
potence of this surface toward dissociative chemisorption band may be an adsorbate with a molecular structure be-
of methanol, perhaps together with lower coverages of tween CH OH and CO. Adsorbed methoxy therefore ap-
3
adsorbed oxygen itself (vide supra). These circumstances pears to be a prime candidate; possible alternative adsor-
would clearly yield very slow rates for reaction [2]. Signifi- bates such as Pd–CH O and Pd–CHO appear to be less
2
cantly, evidence strongly favoring the dominant importance stable (6). It is interesting to note that Pd is incapable of dis-
of the former factor was obtained from transient SERS ex- sociating methanol to form CO in aqueous electrochemical

METHANOL OXIDATION ON PALLADIUM
199
environments (51), contrasting the extensive dissociation should build up to a detectable concentration in the pres-
observed on Rh electrodes (52, 53). ence of coadsorbates, as may well be encountered with our
Indeed, the detection of surface methoxy on Pd even ambient-pressure reaction conditions.
at room temperature and above is not unprecedented. The observation of an apparently similar SERS band,
�
1
Christmann et al. (9) have reported the existence of a ther- centered at 250 cm , on Rh (Fig. 3B) should also be noted
mally stable methoxy on Pd(110) upon CH3OH exposure, in this context. However, the high thermal stability of the
�
which only decomposes to CO at 250 C under UHV condi- species responsible for this feature, as seen by its survival
�
tions. Moreover, a recent study using static secondary ion even up to 500 C (Fig. 3B) argues strongly against attribut-
mass spectrometry (SIMS) noted the presence of surface ing it to methoxy. While the assignment of the 250 cm
methoxy upon methanol dosage on Pd(111) even at 25 C band on Rh is uncertain (40), its markedly narrower band-
10), thereby exhibiting a greater thermal stability than that width than the 255 cm feature on Pd also suggests that
�
1
�
�
1
(
reported previously (1–4). It was found that maintaining these two vibrational modes arise from different methanol
a significant surface concentration of methoxy requires a fragments.
continuous gas-phase methanol supply, the species being
CONCLUDING REMARKS
removed when the dosing was terminated (10). This obser-
�
1
vation partly supports the assignment of the 255 cm band
to methoxy, since the present study involves a continuous
gas-flow environment that is very different than that uti-
lized for UHV investigations, which typically involve only
a finite initial dosing of reactants. However, the present
The present results demonstrate how the availability
of in situ surface vibrational information as furnished by
SERS, along with parallel information on the temperature-
dependent kinetics, can provide novel insight into the sur-
face chemical factors responsible for the differences in
catalytic activity and selectivity toward methanol oxidation
observed on Pd versus Rh. A critical virtue of the SERS
technique in this context is the ability to extract rapid (� 1 s)
sequences of surface vibrational spectra even under techno-
logically relevant catalytic conditions. Specifically, the chief
deductions are as follows.
�
1
2
55 cm band on Pd remained upon switching the gas flow
�
from O2-rich methanol to pure argon at 25 C. This feature
was subsequently removed only upon heating to ca. 200 C
�
in Ar, i.e., at a similar temperature to its disappearance in
the O2-rich methanol mixture (Fig. 3A).
This discrepancy regarding the stability of the methoxy
in UHV and in the present high-pressure gas environment
may be accounted for in terms of site availability. Any ad-
(a) The lack of methanol oxidation, in addition to de-
sorbed methoxy (Pd–OCH3, bonded through O) must find composition (CO/H formation), on Pd for the methanol-
2
an empty adjacent site in order to further dissociate and rich reaction mixture is consistent with the apparent paucity
form adsorbed CO (Pd–CO, bonded through C) (6). This of adsorbed methanol fragments as gleaned from the ab-
may explain why CO is readily observed upon methanol de- sence of detectable vibrational bands. In contrast, the facile
composition in UHV, where the adsorbates are presumably production of CO and H , as well as CO/H formation, on
2
2
2
surrounded by numerous free adjacent sites. On the other Rh is consistent with the presence of adsorbed CO as ob-
hand, the probable higher coverages obtained under our served by SERS under these conditions.
high-pressure conditions most likely hinder such a surface
(b) In the oxygen-rich reactant mixture, the hysteresis
reaction, resulting in a paucity of adsorbed CO (and hence in the temperature-dependent kinetics of CO and H O for-
2
2
the lack of detectable CO2 production on Pd) at tempera- mation observed on Pd, but not on Rh, can be accounted
�
tures below 250 C.
for by the nonreactive nature of the oxide seen by SERS to
This hypothesis is also supported by the findings of be formed on the former surface, versus the reactive prop-
Rebholz and co-workers (5, 10, 17), who reported that the erties of the corresponding oxide produced on the latter.
surface methoxy coverage on Pd(111) is directly propor-
(c) The ability of SERS to extract real-time kinetic in-
tional to the methanol partial pressure. This implies that formation for adsorbates, even on a ca. � 1 s time scale,
the present high-pressure reaction conditions should result following a sudden introduction of the methanol reduc-
in high coverages of this species. This is especially plausi- tant, enables the remarkable stability of surface PdO, as
ble given that the presence of excess oxygen facilitates the contrasted by the facile reactivity of Rh O , to clearly be
2
3
formation and aids the stability of methoxy (vide supra) delineated.
17). Second, an induction period for methoxy dissociation
(d) The differences in catalytic activity and selectiv-
(
was found, revealed by a ca. 200-s delay in the formation ity between Pd and Rh and on a given metal between
of adsorbed CO upon methanol exposure (10). Given that the excess methanol and O -rich reactant mixtures, along
2
methanol decomposes initially on the clean surface to form with corresponding dissimilarities in surface speciation as
CO, it was suggested that the latter species inhibits fur- gleaned from SERS, can be rationalized in terms of the dif-
ther decomposition (probably due to the aforementioned fering ability of these catalysts to dissociatively chemisorb
lack of adjacent sites) (10). As a consequence, methoxy oxygen.

2
00
CHAN ET AL.
(
e) The possible involvement of a methoxy (or related) 19. Houtman, C., and Barteau, M. A., Langmuir 6, 1558 (1990).
2
2
2
0. Solymosi, F., Tarnoczi, T. I., and Berk o´ , A., J. Phys. Chem. 88, 6170
(1984).
intermediate in the reaction mechanism on Pd, as gleaned
�
1
tentatively from the appearance of a 255 cm surface-
adsorbate vibration, is scrutinized. Likely reasons are
considered for the greater thermal stability of such species
1. Sharma, R. K., Zhou, B., Tong, S., and Chuang, K. T., Ind. Eng. Chem.
Res. 34, 4310 (1995).
2. Cordi, E. M., and Falconer, J. L., J. Catal. 162, 104 (1996).
observed in the present ambient-pressure reaction environ- 23. Rask o´ , J., Bontovics, J., and Solymosi, F., J. Catal. 146, 22 (1994).
2
2
2
2
2
2
4. McCabe, R. W., and Mitchell, P. J., Appl. Catal. 27, 83 (1986).
5. Papapolymerou, G. A., and Schmidt, L. D., Langmuir 3, 1098 (1987).
6. Zum Mallen, M. P., and Schmidt, L. D., J. Catal. 161, 230 (1996).
7. Leung, L.-W. H., and Weaver, M. J., J. Am. Chem. Soc. 109, 5113 (1987).
8. Leung, L.-W. H., and Weaver, M. J., Langmuir 4, 1076 (1988).
9. Wilke, T., Gao, X., Takoudis, C. G., and Weaver, M. J., Langmuir 7,
ment compared with the UHV conditions examined previ-
ously.
Overall, then, the combined application of SERS and MS
in this manner appears to be a potent as well as novel ap-
proach enabling the elucidation of long-sought-after rela-
tionships between catalyst performance and surface chem-
istry under technological relevant (i.e., high-pressure gas
flow) reaction conditions. Further applications along these
lines to other industrially as well as fundamentally signi-
ficant heterogeneous catalytic processes are under way in
our laboratory.
714 (1991).
3
3
3
3
0. Wilke, T., Gao, X., Takoudis, C. G., and Weaver, M. J., J. Catal. 130, 62
(1991).
1. Tolia, A. A., Wilke, T., Weaver, M. J., and Takoudis, C. G., Chem. Eng.
Sci. 47, 2781 (1992).
2. Tolia, A. A., Weaver, M. J., and Takoudis, C. G., J. Vac. Sci. Technol. A
11, 2013 (1992).
3. Tolia, A. A., Smiley, R. J., Delgass, W. N., Takoudis, C. G., and Weaver,
M. J., J. Catal. 150, 56 (1994).
ACKNOWLEDGMENTS
34. Tolia, A. A., Williams, C. T., Takoudis, C. G., and Weaver, M. J., J. Phys.
Chem. 99, 4599 (1995).
This work was supported by a grant to CGT and MJW (CTS-9312008)
from the National Science Foundation. We acknowledge the Dow Chem-
ical Company for providing funding for the Advanced Instrumentation
Laboratory in the School of Chemical Engineering at Purdue, which was
utilized in the MS kinetic experiments.
35. Tolia, A. A., Williams, C. T., Takoudis, C. G., and Weaver, M. J., Lang-
muir 11, 3438 (1995).
36. Williams, C. T., Tolia, A. A., Takoudis, C. G., and Weaver, M. J., Chem.
Eng. Sci. 51, 1673 (1996).
37. Williams, C. T., Tolia, A. A., Chan, H. Y. H., Takoudis, C. G., and
Weaver, M. J., J. Catal. 163, 63 (1996).
REFERENCES
38. Williams, C. T., Black, C. A., Takoudis, C. G., and Weaver, M. J., J. Phys.
Chem. B 101, 28 (1997).
1
2
. Davis, J. L., and Barteau, M. A., Surf. Sci. 235, 235 (1995).
. Kok, G. A., Noordermeer, A., and Nieuwenhuys, B. E., Surf. Sci. 135,
39. Williams, C. T., Takoudis, C. G., and Weaver, M. J., J. Catal. 170, 207
(1997).
6
5 (1983).
40. Williams, C. T., Takoudis, C. G., and Weaver, M. J., J. Phys. Chem. B
102, 406 (1998).
41. Gao, P., Gosztola, D., Leung, L.-W. H., and Weaver, M. J., J. Elec-
troanal. Chem. 223, 211 (1987).
42. Matsushima, T., Surf. Sci. 157, 297 (1985).
43. Weber, W. H., Baird, R. J., and Graham, G. W., J. Raman Spectrosc.
19, 239 (1988).
44. McBride, J. R., Hass, K. C., and Weber, W. H., Phys. Rev. B 44, 5016
(1991).
3
. Bhattacharya, A. K., Chesters, M. A., Pemble, M. E., and Sheppard,
N., Surf. Sci. 206, L845 (1988).
4
5
6
7
. Solymosi, F., Berk o´ , A., and T o´ th, Z., Surf. Sci. 285, 197 (1993).
. Kruse, N., Rebholz, M., and Matolin, V., Surf. Sci. 238, L457 (1990).
. Davis, J. L., and Barteau, M. A., Surf. Sci. 187, 387 (1987).
. Chen, J.-J., Jiang, Z.-C., Zhou, Y., Chakraborty, B. R., and Winograd,
N., Surf. Sci. 328, 248 (1995).
. Christmann, K., and Demuth, J. E., J. Chem. Phys. 76, 6308 (1982).
. Christmann, K., and Demuth, J. E., J. Chem. Phys. 76, 6318 (1982).
0. Rebholz, M., and Kruse, N., J. Chem. Phys. 95, 7745 (1991).
8
9
45. Barin, I., Knacke, O., and Kubaschewski, O., “Thermochemical Prop-
erties of Inorganic Substances,” pp. 631 and 645, Springer-Verlag,
Berlin, 1973.
1
1
1. Guo, X., Hanley, L., and Yates, J. T., Jr., J. Am. Chem. Soc. 111, 3155
(
1989).
2. Levis, R. J., Cheng, J. Z., and Winograd, N., J. Am. Chem. Soc. 111,
605 (1989).
3. Parmeter, J. E., Jiang, X., and Goodman, D. W., Surf. Sci. 240, 85
1990).
46. Simons, E. L., in “Scientific Foundations of Vacuum Technique”
(S. Dushman, Ed.), Chapter 11, Wiley, New York, 1962.
47. Orent, T. W., and Bader, S. D., Surf. Sci. 115, 323 (1982).
48. Bader, S. D., Richter, L., and Orent, T. W., Surf. Sci. 115, 501 (1982).
49. Chan, H. Y. H., Takoudis, C. G., and Weaver, M. J., J. Catal. 172, 336
(1997).
50. Williams, C. T., Chen, E. K.-Y., Takoudis, C. G., and Weaver, M. J.,
submitted for publication.
51. Leung, L.-W. H., and Weaver, M. J., Langmuir 6, 323 (1990).
52. Leung, L.-W. H., and Weaver, M. J., J. Phys. Chem. 93, 7218 (1989).
53. Chang, S.-C., Hamelin, A., and Weaver, M. J., J. Chim. Phys. 88, 1615
(1991).
1
1
1
4
(
4. Kruse, N., Chuah, G.-K., Abend, G., Cocke, D. L., and Block, J. H.,
Surf. Sci. 189/190, 832 (1987).
5. Solymosi, F., Berk o´ , A., and Tarnoczi, T. I., Surf. Sci. 141, 533 (1984).
6. Davis, J. L., and Barteau, M. A., Surf. Sci. 197, 123 (1988).
7. Rebholz, M., Matolin, V., Prins, R., and Kruse, N., Surf. Sci. 251/252,
1
1
1
1
117 (1991).
1
8. Gates, J. A., and Kesmodel, L. L., J. Catal. 83, 437 (1983).