High-Pressure Oxidation of Ruthenium as Probed by Surface-Enhanced Raman and X-Ray Photoelectron Spectroscopies

Article abstract of DOI:10.1006/jcat.1997.1841

Surface-enhanced Raman spectroscopy (SERS) combined with X-ray photoelectron spectroscopy (XPS) has been utilized to study the oxidation of ruthenium at ambient pressure (1 atm) and elevated temperatures (25-300°C). The SERS probe provides in-situ vibrational information regarding surface oxide bonding. While the XPS probe necessarily involves ex-situ measurements (i.e., transfer to and from ultrahigh vacuum), it provides valuable complementary information on the metal and oxygen electronic states. Ruthenium surfaces were prepared by electrodepositing ultrathin films (about three monolayers) onto electrochemically roughened (i.e., SERS-active) gold substrates. Insight into the in-situ oxidation process was obtained by probing the changes of surface speciation by SERS upon heating Ru in flowing O₂. A pair of SERS bands at 470 and 670 cm^-1 appear in the spectrum acquired for a freshly electrodeposited film, which are assigned to different stretching modes of hydrated RuO₂ formed during sample transfer to the gas-phase reactor. However, a fully reduced Ru surface (i.e., devoid of oxide features) could be formed by adsorbing a protective CO adlayer in an electrochemical cell followed by heating to 200°C in vacuum so to thermally desorb the CO. While the initially oxidized (i.e., RuO₂) surface was stable to further oxidation upon heating in O₂, adsorbed atomic oxygen was detected at 200°C from the appearance of a SERS band at 600 cm^-1 and a XPS O(1s) peak at 531.7 eV. In contrast, the higher oxides RuO₄ and possibly RuO₃ were produced only upon thermal oxidation of the fully reduced Ru surface. Evidence for RuO₃ formation includes the appearance of a 800 cm^-1 SERS band at 200°C which correlates with the advent of a Ru(3d_5/2) peak at 282.6 eV. The surface was further oxidized to RuO₄ at 250°C, as deduced from the formation of a 875 cm^-1 band and a Ru(3d_5/2) peak at 283.3 eV. While RuO₃ and RuO₄ were exclusively formed at temperatures higher than 250°C, RuO₂ was produced upon cooling to room temperature, possibly via the decomposition of RuO₄. 997 Academic Press.

Full text of DOI:10.1006/jcat.1997.1841

JOURNAL OF CATALYSIS 172, 336–345 (1997)
ARTICLE NO. CA971841
High-Pressure Oxidation of Ruthenium as Probed by Surface-Enhanced
Raman and X-Ray Photoelectron Spectroscopies
�
�
,1
,2
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 June 3, 1997; revised July 23, 1997; accepted July 25, 1997
low overpotential for the electrooxidative generation of O2
Surface-enhanced Raman spectroscopy (SERS) combined with and Cl2 also makes ruthenium an important electrocata-
X-ray photoelectron spectroscopy (XPS) has been utilized to study lyst. The activity and selectivity of ruthenium catalysts de-
the oxidation of ruthenium at ambient pressure (1 atm) and elevated
temperatures (25–300 C). The SERS probe provides in-situ vibra-
tional information regarding surface oxide bonding. While the XPS
probe necessarily involves ex-situ measurements (i.e., transfer to
and from ultrahigh vacuum), it provides valuable complementary
information on the metal and oxygen electronic states. Ruthenium
surfaces were prepared by electrodepositing ultrathin films (about
three monolayers) onto electrochemically roughened (i.e., SERS-
pend greatly on the oxidation state of the metal, different
ruthenium oxides exhibiting distinct physical and catalytic
properties (4). For example, Bond et al. reported the effect
of oxidation followed by low-temperature reduction on a
�
supported Ru catalyst, enhancing the turnover frequency
�
of hydrogenolysis of ethane by a factor of 200 at 160 C (1).
The interaction between ruthenium and oxygen has
therefore been the subject of extensive investigation.
active) gold substrates. Insight into the in-situ oxidation process was
obtained byprobing the changesof surface speciation bySERS upon Chemisorption of oxygen in ultrahigh vacuum (UHV) on
�
1
heating Ru in flowing O2. A pair of SERS bands at 470 and 670 cm
ruthenium single crystal surfaces has been well documented
appear in the spectrum acquired for a freshly electrodeposited film, (5–13). Thomas et al. observed atomic oxygen adsorbed
which are assigned to different stretching modes of hydrated RuO2
formed during sample transfer to the gas-phase reactor. However,
on Ru(001) by using Electron Energy Loss Spectroscopy
(
EELS) (11). For low exposures, oxygen adsorbs dissocia-
a fully reduced Ru surface (i.e., devoid of oxide features) could be
formed by adsorbing a protective CO adlayer in an electrochemical
cell followed by heating to 200 C in vacuum so to thermally desorb
tively but incorporation into the bulk occurs to a limited
extent. The other mainstream approach involves the utiliza-
tion of surface-specific techniques to characterize different
stoichiometric ruthenium oxides. For instance, Mar et al.
� 1 studied the effect of various preparative routes to RuO2 on
the resulting Raman spectra (14).
�
the CO. While the initially oxidized (i.e., RuO2) surface was stable to
further oxidation upon heating in O2, adsorbed atomic oxygen was
�
detected at 200 C from the appearance of a SERS band at 600 cm
and a XPS O(1s) peak at 531.7 eV. In contrast, the higher oxides
RuO4 and possibly RuO3 were produced only upon thermal oxida-
Ruthenium, however, is not as noble as other platinum-
tion of the fully reduced Ru surface. Evidence for RuO3 formation group transition metals. From a technological standpoint,
�
1
�
includes the appearance of a 800 cm SERS band at 200 C which the formation of RuO , which is known to be a volatile
4
correlates with the advent of a Ru(3d5/2) peak at 282.6 eV. The sur-
face was further oxidized to RuO4 at 250 C, as deduced from the for-
mation of a 875 cm band and a Ru(3d5/2) peak at 283.3 eV. While
RuO3 and RuO4 were exclusively formed at temperatures higher
than 250 C, RuO2 was produced upon cooling to room tempera-
ture, possibly via the decomposition of RuO4.
oxide, is of primary concern as it is the origin of catalyst
loss in oxidative reaction systems (15). Given that most of
�
�
1
the fundamental studies have been limited to UHV condi-
tions in which ruthenium does not oxidize extensively, it is
important to extend such investigations closer to practical
reaction conditions, such as continuous-flow gaseous envi-
�
�c 1997 Academic Press
�
ronments at elevated temperatures (25–500 C) and pres-
sures (1 atm).
INTRODUCTION
Over the past decade, we have demonstrated that
Ruthenium is an important catalyst in chemical indus- surface-enhanced Raman spectroscopy (SERS) can pro-
try, having applications in hydrogenolysis (1), hydrogena- vide powerful information regarding adsorbate vibrations
tion of benzene (2), and Fischer–Tropsch synthesis (3). The for metal–gas as well as electrochemical interfaces under
technologically relevant conditions, especially at higher
1
(ambient) gas pressures. Although SERS is traditionally
considered applicable only to the coinage metal surfaces
Cu, Ag, and Au, following earlier work for electrochemical
Present address: Department of Chemical Engineering, University of
Illinois at Chicago, 810 S. Clinton Street, Chicago, IL 60607.
2
Corresponding author. Fax: (765) 494-0239.
336
0021-9517/97 $25.00
Copyright �c 1997 by Academic Press
All rights of reproduction in any form reserved.

SERS AND XPS STUDY OF Ru OXIDATION
337
+
surfaces by Leung and Weaver (16, 17), we have extended tation at 647.1 nm was provided by a Kr laser (Spectra
it to a number of catalytically important transition metals Physics). The scattered light was collected by a three-stage
such as Pt, Pd, and Rh in the gas phase (18–26) by similarly spectrometer (Triplemate) equipped with a charge-coupled
electrodepositing them as thin films on roughened gold. device (CCD) detector. The frequencies of the Raman scat-
The sensitivity of SERS to probe metal–oxygen vibrational tered light were calibrated by a neon light emitting diode.
modes, usually found in the low frequency region between
The XPS experiments were performed on similar Ru
�
1
2
00 and 1000 cm , also makes it a valuable technique for thin films. The sample (10 mm � 10 mm) was mounted on
monitoring surface oxidation of transition metals (21, 27). a molybdenum holder. The XPS instrument was a Perkin–
While SERS can yield insight into the nature of surface Elmer PHI 5300 connected to a PHI 04-800 high-pressure
bonding, X-rayphotoelectron spectroscopy(XPS) provides catalytic reactor, capable of heating the sample up to
�
a valuable complementary technique for assigning oxida- 600 C at 1 atm. After the surface oxidation was performed
tion states and stoichiometry of the oxides. There are a under the desired conditions, the sample was cooled down
�
number of XPS studies of various distinct Ru oxides; the to 25 C and then transferred to the UHV main chamber at
�
9
oxidation-state specificity of the emitted electron energies 10 Torr for spectral acquisition. Spectra were collected
facilitates the assignment of the detected SERS bands (28– using a magnesium X-ray source operated at 15 kV and
3
4). In contrast to SERS, however, XPS necessitates the 300 W, an analyzer pass energy of 8.95 eV and with a
use of UHV conditions, so that examination of metal ox- channel-plate detector. The spectrometer was calibrated
ides formed in high-pressure gas environments requires the from the known binding energy of the Au (4f7/2) level,
surface to be transferred into UHV prior to scrutinybyXPS. 84.0 eV. The XPS spectra were computer fitted in order to
Our group previously utilized combined SERS and XPS determine the chemical states of specific species present on
measurements to investigate the ambient-pressure ther- the samples. Curve-fitting was carried out using a nonlinear
mal oxidation of rhodium (21). Comparative inspection of least squares fitting program with a Gaussian/Lorentzian
the temperature-dependent SERS and corresponding XPS sum function, with the spin-orbit splitting of the Ru (3d)
spectra demonstrated the ability of this approach to identify doublet constrained to be 4.2 eV. The Ru XPS spectra were
specific oxides formed beyond initial oxygen chemisorp- further fitted with exponential tails to account for the peak
tion, especially Rh2O3. Such tactics should be even more asymmetry arising from core–hole coupling (36).
valuable for examining the thermal oxidation of ruthenium,
owing to the anticipated existence of a wider variety of
ruthenium oxides. We report herein such a study of ther-
mal oxidation of ruthenium by means of SERS combined
with XPS. Of central interest is to identify various ruthe-
nium oxides formed as a function of the elevated reaction
temperature under technologically relevant conditions, i.e.
RESULTS
Raman Spectroscopy
Freshly deposited ruthenium films examined by SERS in
�
�
6
the gas-phase reactor under vacuum conditions (10 Torr)
1
exhibited vibrational peaks at ca 470 and 670 cm , sug-
gesting the presence of surface impurities or an oxide film
formed during the transfer from the electrochemical cell
to the reactor (vide infra). Consequently, we attempted to
engender surface reduction by heating in H2. Figure 1A
1
atm of oxygen.
EXPERIMENTAL
The preparation of SERS-active thin film samples fol- shows typical temperature-dependent SER spectra of a
lowed a procedure used previously (17–19) involving elec- freshly electrodeposited Ru sample heated in a continuous
3
� 1
trodeposition of ruthenium on roughened gold. Gold discs flow of H2 at 100 cm min . (Note that the spectra shown
of diameter 6 mm were cut from 0.1-mm-thick gold foil in Fig. 1A and subsequently are displayed in an upward-
(
Johnson Matthey) and polished using alumina powder. stacked sequence.) No change was detected upon dosing H2
�
� 1
They were then put into an electrode holder that exposed at ca 25 C, the SERS bands at 470 and 670 cm remaining
1
oxidation–reduction cycles to produce a SERS-active gold progressively up to 150 C, holding at each temperature for
surface by following essentially the procedure described 3 min before the spectra was acquired. By 100 C, a SERS
in Ref. (35). Ruthenium thin films were deposited from a band at 600 cm appeared, whereas the 470 and 670 cm
5
saturated calomel electrode (SCE), the cathodic charge be- and convolved peak. No significant change was detected at
ing monitored so as to obtain the desired film thickness, usu- 150 C and further temperature increases were avoided as
2
0 mm of the surface in 0.1 M KCl solution, followed by (bottom spectrum). The surface temperature was increased
�
�
�
1
� 1
�
4
� 10 M RuCl3 solution in 0.1 M HClO4 at � 0.1 V versus bands were still present as shoulders, resulting in a broad
�
ally about 3 monolayers. The surface was then transferred some attenuation in SERS activity was experienced when
to an ambient-pressure gas reactor (ca 1 atm), mounting heating in H2 at higher temperatures. Subsequent cooling
�
1
on a sample holder which is capable of raising the surface to room temperature left the 600 cm band as the major
�
temperature up to 600 C by resistive heating. Laser exci- residual feature.

3
38
CHAN, TAKOUDIS, AND WEAVER
3
� 1
FIG. 1. (A) Temperature-dependent SER spectra for a Ru film in 100 cm min H2 at 1 atm. The film was prepared by electrodeposition on
3
� 1
roughened gold (Method I). (B) Subsequent temperature-dependent SER spectra obtained in flowing O2 (100 cm min ) at 1 atm.
The ruthenium was then reoxidized in flowing O2 at transferred through air without CO modification still
3
� 1
1
00 cm min ; Fig. 1B shows typical spectra as a function of exhibited partial oxidation as seen from the reappearance
�
1
� 1
temperature. No observable change to the 600 cm band of the 470/670 cm bands.) We will refer to this latter
was discerned at 25 C. After the sample was oxidized at pretreatment procedure as “Method II,” with the tactic
50 C, followed by cooling to room temperature, however, described above involving simple transfer of a freshly
�
�
2
several SERS peaks were detected at 470, 600, 670, and deposited Ru film being dubbed “Method I.”
�
1
8
4
00 cm (Fig. 1B, top spectrum). This reappearance of the
The effect of this Method II pretreatment on the SER
�
1
70 and 670 cm peaks suggests they originate from an ox- spectra, as well as the subsequent gas-phase oxidation, is
ide, since they are thermally removed in the presence of H2 shown in Fig. 2. After the sample had been transferred
but can be regenerated by heating in O2.
TABLE 1
Table 1 summarizes some literature vibrational spec-
troscopic data for several ruthenium oxides, including the
well-known RuO2 and RuO4 species. The 470/670 cm
Published Vibrational Modes of Various Ru Oxides, Oxyanions,
and Adsorbed Oxygen on Ru
�
1
pair observed here have comparable frequencies to the
characteristic normal Raman bands of RuO2, located at
�
(Ru–O)
�
�
1
Species
stretching modes (cm 1)
Reference
Technique
5
15 and 626 cm (14, 37). However, firmer assignment of
these and other bands requires parallel interpretation of
SERS and XPS results and will be addressed below in the
Discussion.
a
Adsorbed
atomic oxygen
600
585
SERS
EELS
EELS
EELS
10
5
11
a
5
5
75–595
20–600
Since H2 pretreatment of the freshly electrodeposited
sample leaves a residual species with a SERS band at
RuO2
470/670
510/630
SERS
SERS
Raman
Raman
a
�
1
6
00 cm (Fig. 1A), an electrochemical reduction step was
5
5
15/626
28/646
14
37
a
utilized as an alternative means to obtain an oxide-free Ru
surface. A pretreatment procedure achieving this objective
involves applying a reducing potential, � 0.3 V versus SCE,
to the freshly deposited Ru sample after transfer to 0.1 M
HClO4. The electrode was then sparged with CO so to
modify the surface, thereby protecting the sample from
atmospheric contamination during final transfer to the
gas-phase reactor. (Electrochemically reduced surfaces
RuO3
800
808
875
SERS
2
RuO
4
�
45
Raman
a
RuO4
SERS
Raman
Raman
8
8
78
81
44
45
a
Data from this work for comparison.

SERS AND XPS STUDY OF Ru OXIDATION
339
two behaviorally distinct features were detected at 875 and
� � 1
1
8
00 cm but the 600 cm band was absent. As seen in
�
1
Table 1, the 875 cm band has a similar frequency to the
normal Raman band of RuO4. However, the identification
�
1
of the 800 cm peak is not straightforward (see below).
Given these rich SER spectral changes induced by ther-
mal oxidation of the initially reduced ruthenium surface
prepared by Method II (Fig. 2), experiments were also
performed by similarly heating the initially partially oxi-
dized (i.e., Method I) Ru surface in oxygen, prepared as for
Fig. 1A. A typical temperature-dependent set of SER spec-
tra obtained in this fashion is shown in Fig. 3. The spectrum
was unaffected by dosing O2 at room temperature, but the
�
1
6
00 cm band is seen when the temperature was higher
�
than 200 C, remarkably similar to when the same surface
was heated in H2 (Fig. 1A). Furthermore, the features at
�
1
8
75 and 800 cm were not seen, contrasting the behav-
ior for oxidation of the initially reduced surface (Fig. 2).
Prompted by the broadness and asymmetry of the result-
ing band, the spectrum that was acquired after subsequent
�
cooling to 25 C was deconvolved (Fig. 3, top spectrum). It is
FIG. 2. Temperature-dependent SER spectra for thermal oxidation
of a metallic Ru surface. The initial oxide-free surface was prepared by
electrochemical reduction and protected by CO adsorption during the
transfer to the reactor (Method II): (a) CO-covered Ru surface; (b) re-
evident that both the features originally present, at 470 and
� 1
6
70 cm , also provide contributions to the final spectrum.
Besides studies aimed at the gas–solid interface, as al-
�
�
ready noted SERS is also invaluable (and indeed better
duced Ru surface at 25 C after CO was desorbed by heating to 200 C in
vacuum; (c–e) spectra obtained in O2 after ramping temperature to values known) for scrutinizing adsorption at metal–solution in-
�
indicated; (f) after cooling to 25 C.
terfaces (16, 17, 27). Consequently, related aqueous elec-
trochemical SERS experiments were performed to discern
any similarities, or differences, between the oxidation of
ruthenium in these two environments. The SERS-active Ru
to the reactor, two vibrational modes were observed
�
1
at 500 and 550 cm
(spectrum a), diagnostic of the
ruthenium–carbon stretching modes �(Ru–CO) of ad-
�
sorbed CO (17). The sample was then heated to 200 C
�
6
in a vacuum (10 Torr) to remove the CO; the result-
ing spectrum exhibits no discernable features in the 300–
00 cm region (spectrum b), indicative of the desired
�
1
9
formation of an oxide-free Ru surface. To initiate oxida-
tion, the surface temperature was raised incrementally in
3
� 1
�
flowing O2 at 100 cm min . At 200 C, three bands are
�
1
evident, centered at 510 and 630 cm with a more in-
tense feature at 800 cm (spectrum c). A new feature
emerged at 875 cm when the temperature was further
increased to 250 C (spectrum d). By 300 C, there was
an intensity increase of the 875 cm peak relative to
the 800 cm band (spectrum e). To facilitate compari-
son between corresponding XPS and SERS experiments,
�
1
�
1
�
�
�
1
�
1
�
the SERS sample was subsequently cooled down to 25 C,
mimicking the data acquisition procedure required for the
ex-situ XPS measurements outlined below. Interestingly,
�
1
along with the remnants of both the 875 and 800 cm peaks,
the resulting SER spectrum shows not only the reappear-
ance, but also an intensification, of the characteristic RuO2
�
1
bands at 510 and 630 cm (spectrum f). Significantly, the
results of this experiment (Fig. 2) are dissimilar to that in-
FIG. 3. Temperature-dependent SER spectra for thermal oxidation
of hydrated RuO2 (prepared by Method I) by exposing to 100 cm min
3
� 1
volving H2 pretreatment (Fig. 1B) in that in the former case O2 at 1 atm.

3
40
CHAN, TAKOUDIS, AND WEAVER
Photoelectron Spectroscopy
The XPS experiments were performed primarily on
ruthenium thin films deposited on smooth gold foils.
Roughened gold substrates, as in the SERS experiments,
were also used but the results showed no significant dif-
ferences from those with smooth surfaces. The sample
could be transferred without exposure to air between a
UHV chamber for spectral acquisition and a high-pressure
(
1 atm) reactor, where temperature-dependent surface ox-
idation was undertaken in a similar fashion to the SERS
experiments. XPS peaks arising from Ru(3d) and O(1s)
photoelectrons were examined, as they are the most intense
of the ruthenium and oxygen signals in the photoemission
spectrum. Unfortunately, there is an overlap between the
Ru(3d3/2) and the C(1s) peak at ca 284.5 eV, the latter being
due to surface carbonaceous contamination. In view of this,
�
all samples were initially heated in flowing O2 at 400 C for
3
0 min to remove the carbonaceous species, followed by
�
reduction in H2 at 300 C to obtain a metallic Ru surface.
Consequently, then, this surface can be compared most di-
rectly with that prepared by means of Method II above.
The XPS analysis of different Ru oxides has been well
FIG. 4. Electrochemical SER spectra obtained for a Ru film in 0.1 M
HClO4 as a function of applied potential (vs SCE). The surface was initially
reduced at � 0.3 V (bottom spectrum) prior to raising the potential to documented, and Table 2 summarizes selected published
where O2 evolution occurs (+1.16 V), and subsequently reduced again at results from other studies. Figure 5 shows typical curve-
�
0.3 V.
fitted Ru(3d) doublet and O(1s) singlet XPS spectra ob-
tained from a temperature-dependent oxidation experi-
film was prepared as described above, the surface then be- ment performed on an initially metallic ruthenium surface
ing transferred rapidly to a SERS cell with aqueous 0.1 M in 1 atm of O . After the gas-phase oxidation/reduction
2
HClO4 as supporting electrolyte. Figure 4 shows selected pretreatment, the Ru(3d5/2) peak was located at 280.3 eV
SER spectra for the ensuing electrooxidation of ruthenium (Fig. 5a); this value is comparable to the binding energy of a
as a function of electrode potential (vs SCE) which was al- metallic ruthenium surface (32–34). Corresponding O(1s)
�
1
tered in a stepwise fashion. The 470 and 670 cm bands, signals were found at 532.5 and 534.0 eV, possibly arising
initially present at the open circuit potential, were found to from surface impurities (vide infra).
be entirely removed when a reducing potential at � 0.3 V
As shown in Fig. 5, a noticeable attenuation of the
was applied, yielding a spectrally featureless surface (bot- Ru(3d5/2) peak at 280.3 eV was observed upon oxidation
�
tom spectrum). Upon increasing the potential to +0.6 V, at 200 C, along with the emergence of another two peaks
�
1
the simultaneous reappearance of the 470 and 670 cm
at 281.0 and 282.6 eV. The corresponding O(1s) XPS spec-
bands indicates that they are arising from electrooxidation trum consistsofa major newpeak at 531.7eV and another at
by water) of the Ru surface. Given that oxidation of an 530.0 eV. The latter is characteristic of lattice oxide (31, 33)
(
originally reduced Ru surface in dry O2 requires elevated and the coincidence with the binding energy upshift of the
temperatures (Fig. 2), this finding in Fig. 4 suggests that Ru(3d) peak suggests surface oxidation. Upon further in-
�
the initial room-temperature oxidation observed following creasing the oxidation temperature to 250 C, an additional
transfer through (wet) air (Fig. 1A, bottom spectrum) also Ru(3d5/2) peak was found at 283.3 eV but the 281.0 eV com-
involves an electrochemical mechanism, perhaps involving ponent remains dominant. The extent of surface oxidation
a water film on the surface. An additional SERS band at was greater, as indicated by the increase of the intensity
�
1
5
80 cm emerged at +1.16 V (Fig. 4) corresponding to the of the oxidic O(1s) peak at 530.0 eV. The surface was com-
point where O2 evolution reaction occurs on the Ru elec- pletely oxidized upon subsequently raising the temperature
�
trode in acidic medium (38). The surface was subsequently to 300 C as implied from the disappearance of the Ru metal
�
�
1
1
reduced by decreasing the potential. At +0.8 V, a 800 cm
peak at 280.3 eV. The relative intensity of the 283.3 eV to
the 282.6 eV peak also increased. Similarly to the SERS
feature was detected. In addition, the 470 and 670 cm
bands reappeared at +0.6 V, and they vanished when the data, XPS band assignments will be addressed further in
potential was lowered to � 0.3 V (Fig. 4), indicating that the the Discussion section below.
electrochemical oxidation/reduction processes are chemi-
cally reversible.
Additional XPS experiments (results not shown) were
also performed to correlate the SERS and XPS results in

SERS AND XPS STUDY OF Ru OXIDATION
341
TABLE 2
Literature XPS Binding Energies (eV) of Various RuOx Species
Stoichiometrically known
Ru oxides
Approach
Reference
Ru surface oxidation
(7)
This work
(34)
(40)
(32)
(28)
Reference
energy (eV)
Au(4f7/2) = 84.0
Ru(3d5/2) = 279.96
Au(4f7/2) = 84.0
Au(4f7/2) = 84.0
Au(4f7/2) = 84.0
Au(4f7/2) = 84.0
Ru(3d5/2)
Ru
RuO2
RuO3
RuO4
280.2
281.9
279.96
280.7
281.7
282.6
280.0
280.7
282.4
280.0
280.7
282.5
283.3
279.91
280.68
282.38
280.3
281.0
282.6
283.3
O(1s)
Adsorbed atomic
oxygen
Lattice oxygen
531.0
529.4
529.9
528.9
531.5
529.4
531.7
530.0
529.2
Remarks
High-pressure
oxidation of
ruthenium
powder.
NO2 was used
to obtain high
atomic oxygen
coverage.
Electrooxidation
Ru and RuO2
surfaces.
Characterization
of pure
stoichiometric
RuO4 and RuO2.
Characterization
of various
supported Ru
catalysts.
High-pressure
oxidation of
Ru thin film
in flowing O2.
order to aid identification of the 600 cm 1 SERS band found as already mentioned the two similar SERS bands at 510
�
�
1
upon heating initially oxidized Ru surfaces (cf. Fig. 1A). and 630 cm detected upon oxidizing a reduced (Method
Analogously to the SERS experiment shown in Fig. 3, an II) Ru surface to 200 C (Fig. 2c) are attributed to RuO2.
�
electrodeposited Ru sample (prepared by Method I) was This assignment is strongly supported by the corresponding
�
exposed to O2 at 250 C without the aforementioned ther- XPS data (Fig. 5b) in that the upshift of the major Ru(3d5/2)
�
mal oxidation/reduction pretreatment. The resulting XPS peak from the metallic value at 280.3 to 281.0 eV at 200 C in
O(1s) spectrum shows no difference to the one as shown in oxygen is consistent with the formation of RuO2, as seen by
Fig. 5c, which has the major 531.7 eV peak and a shoulder comparison with literature XPS data for bulk-phase RuO2
at 530.0 eV, but the presence of the carbon peak makes the (Table 2).
Ru(3d5/2) signal difficult to interpret.
It is known that RuO2 behaves as a hygroscopic oxide
4). Therefore, the 470 and 670 cm SERS bands formed
�
1
(
by Ru electrooxidation at +0.6 V are suggested to arise
from hydrated RuO2. Despite the lack of any vibrational
spectroscopic data for direct comparison, there are ultravi-
olet photoelectron spectroscopy (UPS) (39) and XPS (40)
studies of the same system which support the formation of
hydrated RuO2 under similar electrochemical conditions.
Recall that apparently similar 470 and 670 cm bands
were also detected upon thermal reoxidation of a partially
reduced Ru surface by H2 (Fig. 1B). Although it appears
less likely that hydrated oxide would form at the metal–gas
interface than in an aqueous electrochemical environment,
the presence of water may be accounted for by the reaction
between H2 and the lattice oxygen during the reduction of
the oxide. At first glance, one is also tempted to ascribe
the O(1s) peak observed at 532.5 eV (with a shoulder at
DISCUSSION
The data presented in this study demonstrates the inher-
ent complexity of the surface oxidation. The assignments of
various SERS and XPS peaks, deduced to arise from RuO2,
RuO3, RuO4, and an adsorbed oxygen species, will be dis-
cussed separately. Furthermore, several issues concerning
the oxidation processitselfwillbe addressed. These include:
�
1
(
1) In the presence of O2, what is the origin of the behav-
ioral difference between heating an initially partly oxidized
and a reduced Ru surface? (2) What is the role of adsorbed
oxygen in the selective formation of various Ru oxides? (3)
What is the thermal stability of the higher valent oxides?
RuO2
5
34.0 eV) following H2 reduction but prior to thermal oxi-
Ruthenium dioxide (RuO2) has a rutile structure and it
is also the most thermally stable species among the family
of Ru oxides (4, 15). Being a conductive oxide, it is infrared
inactive, but normal Raman spectra yield Eg and A1g modes
dation to water or possibly OH species in view of the high
binding energies. However, water should desorb well below
room temperature on ruthenium in UHV, eliminating this
possibility. While the assignment is certainly unclear, we
�
1
at 528 and 644 cm , respectively (14, 37). Consequently, note that a O(1s) band at a similarly high binding energy,

3
42
CHAN, TAKOUDIS, AND WEAVER
idizing condition (Fig. 3) is assigned to adsorbed oxygen.
The frequency of this vibrational mode is in agreement
with an EELS feature reported by Thomas et al. (11) when
O2 was adsorbed dissociatively on Ru(001). This band was
assigned to the metal–oxygen stretch of adsorbed atomic
oxygen and found to be coverage-dependent, varying from
� � 1
1
5
20 cm at low coverage to 600 cm at saturation. A sim-
�
1
ilar 600 cm SERS band emerging at elevated tempera-
tures (Fig. 3) can be interpreted as arising from dissociative
adsorption of oxygen on the oxide surface. The thermal sta-
bility of this surface oxygen is in harmony with the UHV
studies in that it is not desorbed in the latter environment
�
� 1
until ca 750 C (6, 8). Although the present 600 cm band
refers to an at least partly oxidized surface, rather than a
clean metal surfaces as studied in Refs. (6) and (8), our
assignment is strengthened by noting that this band also
appears on partially reduced Ru (Fig. 1A). The appearance
�
1
of a similar 600 cm band at the onset of reduction of hy-
�
drated RuO2 in H2 at 100 C (Fig. 1A) suggests that this
adsorbed oxygen may be involved in the reaction as an
intermediate.
Again, this tentative assignment is made firmer by com-
parison with the XPS data. The XPS results also show there
exists a corresponding major O(1s) peak component at
5
5
31.7 eV besides the one associated with lattice oxygen at
30.0 eV. This relatively high O(1s) binding energy, indi-
cating it originates from a less electron-rich oxygen, which
has been previously assigned to chemisorbed atomic oxy-
gen (32, 34). We therefore assign both the 531.7 eV O(1s)
�
1
XPS peak and the 600 cm SERS band to adsorbed atomic
oxygen, given that they were both detected upon heating
RuO2 in O2.
An additionalfeature isalso noticed in the higher-binding
energy side of the O(1s) peak of the oxidized surface
(
Figs. 5b–d). Its identity is not as clear. One possibility
ascribes it to remnant of the adsorbed species which was
evident prior to thermal oxidation (Fig. 5a). The other pos-
FIG. 5. Ru(3d) and O(1s) XPS spectra obtained upon progressive
thermal oxidation of a reduced Ru thin film deposited on a smooth gold
substrate. The deposited film in (a) was pretreated by oxidation in O2 at sibility is that it may arise from core–hole coupling which
� �
00 C to remove surface carbon followed by reduction in H2 at 300 C.
Spectra shown (b–d) obtained following oxidation in 100 cm min O2 at
atm at temperatures indicated and transferral to UHV in each case. The
uppermost trace in each spectrum represents overlapping of the data and
the best-fit curve. Exponential tail functions are shown as an example in
4
would yield an asymmetrical broadening of the XPS peak
towards higher binding energy, yielding so-called “shake-
up” peaks (41–43). This effect has been shown previously
in numerous spectra of adsorbate atoms on metal surfaces
3
� 1
1
Fig. 5a, accounting for the asymmetry of the Ru(3d) peaks due to core– (41). However, it is known that the stronger the adsorbate-
hole coupling (see text), but they are not shown in Figs. 5(b–d) for the sake
of clarity.
substrate interaction, the less marked this effect will be
(
41, 42). Considering the high thermal stability of the 530.0
and 531.7 eV peaks (Fig. 5d), they should be associated
with strongly bonded species. Moreover, a previous study
by Fuggle et al. has addressed the appearance of shake-up
lines in the XPS spectra for oxygen adsorption on Ru(001)
5
33.5 eV, was also observed in our earlier SERS/XPS study
of rhodium oxidation following heating in UHV (21). This
was tentatively attributed to “subsurface oxygen” or possi-
bly SiO2 formed from Si impurities in the metal.
(
43). A similar asymmetric broadening is evident in the
higher binding-energy side of the O(1s) peak, as well as a
Adsorbed Oxygen
satellite peak at ca 10 eV higher energy. The former was
The 600 cm 1 Raman band detected when a Method I assigned to another adsorbed species (43). On this basis,
�
Ru film was exposed to either reducing (Fig. 1A) or ox- then, we are inclined to similarly attribute the high-energy

SERS AND XPS STUDY OF Ru OXIDATION
343
asymmetry in our O(1s) peak to the presence of another the lower binding-energy peak at 281.0 eV corresponding
adsorbed state.
to RuO2 screened by metallic conduction electrons. This
asymmetric appearance of the XPS spectra is also consis-
tent with core–hole coupling which requires the atom to
RuO4
Ruthenium tetraoxide (RuO4) has been used as an oxi- have unfilled states above the Fermi level (49). As shown
dizing agent in organic chemistry (44). It is also identified in Fig. 5a, the Ru metallic peak was curve-fitted, including
as the major corrosion product involved in oxygen evolu- an exponential tail function to account for the asymmetry.
tion on ruthenium electrodes (38, 45). Several infrared (IR)
46) and Raman (47, 48) spectra of RuO4 have been re- d-metals, the spectra of most oxide surfaces being symmet-
ported, with the latter showinga symmetricstretchingmode ric. However, since RuO2 is a metallic oxide that there-
Core–hole coupling is most commonly observed on
(
�
1
at 880 cm . As already mentioned, this finding therefore fore also has a partially filled conduction band (4), core–
suggests an assignment of the SERS band observed here at hole coupling may also occur on this surface. Nonetheless,
�
1
8
75 cm to RuO4. This band is only discernable upon oxi- the increased intensity of the Ru(3d5/2) peak shoulder at
dation of a reduced Ru surface (prepared by Method II) at 282–283 eV upon oxidation (Figs. 5b–d) suggests that the
�
temperatures higher than 250 C (Fig. 2). Again, this asser- 282.6 eV feature in the deconvoluted spectra emerging by
�
tion is strengthened by inspecting the corresponding XPS 200 C signals the formation of another species, which has a
Ru(3d5/2) spectra (Fig. 5c), in which the additional 283.3 eV higher Ru valency than that of RuO2. Therefore, the curve-
�
peak emerging at 250 C is also characteristic of RuO4 (32). fitting procedure of the Ru(3d) peak of the oxidized sample
�
When the surface temperature was further in to 300 C, both includes an exponential tail as well as allowing for contri-
the SERS and XPS results show an increased production butions from other chemical components.
of RuO4. Once RuO4 was formed, it is stable at tempera-
Additional insight into this issue in the present case, how-
tures as high as 300 C, as well as upon subsequent cooling ever, can be gleaned by examining the corresponding SERS
�
to room temperature.
data. Significantly, these results favor the former interpre-
As far as the formation of RuO4 as a corrosion prod- tation, suggesting the presence of RuO3 as well as RuO2.
�
1
�
uct is concerned, the 875 cm band was not detected at Thus upon inspecting the SER spectrum at 200 C (Fig. 2c),
�
1
the potential of O2 evolution (+1.16 V vs SCE) in the a SERS band was detected at 800 cm besides the RuO2
�
1
electrochemical SERS experiment (Fig. 4). This observa- bands at 510 and 630 cm . Comparing the corresponding
�
tion apparently contradicts a recent in-situ IR spectroscopic SERS (Fig. 2c) and XPS (Fig. 5b) results at 200 C correlates
study (38) of the same system in which an IR band at the characteristic RuO2 peaks at 510/630 cm and 281.0 eV,
20 cm , assigned to RuO4, was observed under similar respectively, so that it is likely that the remaining peaks at
conditions. We tentatively account for this discrepancy by 800 cm and 282.6 eV both arise from another species.
suggesting that the infrared band observed in Ref. (32) While the 800 cm band is not present in the vibrational
arises from a solution species; note that SERS is an intrinsi- spectra of RuO2 and RuO4, it has a frequency slightly lower
cally surface-specific technique. The additional appearance than that of RuO4 (875 cm ), suggesting the presence of in-
�
1
�
1
9
�
1
�
1
�
1
�
1
of the 580 cm peak at +1.16 V is indicative of gold oxi- termediate Ru oxidation state, such as in RuO3.Indeed, the
dation (27), thereby providing evidence for Ru film disso- frequency of the symmetric stretches of several Ru tetraoxy
lution so to reveal the underlying gold substrate.
compounds has been shown to follow the sequence: RuO4
� � 1 � 1
880 cm ) > RuO4 (825 cm ) > RuO (810 cm ) (48).
4
1
�
2�
(
RuO3
�
1
An apparently similar 800 cm band was also detected
The existence of RuO3 has been the subject of contro- in the electrochemical SERS experiment (Fig. 4) which we
versy in many XPS studies (29–32). To our best knowledge, tentatively assign to RuO2 , a well-known Ru(VI) oxy an-
�
4
2�
there has not been any reported characterization of RuO3 ion (38, 48). In this regard, the formation of RuO4 at
as a discrete stoichiometric oxide. As shown in Fig. 5b, the +0.8 V may occur via reduction of the RuO4 produced in
Ru(3d5/2) peak has a dominant component at 281.0 eV as the oxygen evolution reaction (at +1.16 V). Interestingly,
well as a minor one at 282.6 eV. While the assignment of an analogous consecutive formation of different Ru oxy-
the former peak to RuO2 is well-accepted, that of the latter compounds is found by thermal means in the gas-phase
is not as clearcut. The higher binding-energy component system in that RuO4 (875 cm ) was only detected upon
at 282.6 eV has been previously proposed as arising from further raising the temperature after forming the 800 cm
RuO3 (29, 30, 32), which is believed to be present on RuO2 band species (Fig. 3). This leads to the suggestion that the
�
1
�
1
bulk-phase sample as a surface defect.
latter acts as a precursor to the formation of RuO4. How-
This interpretation, however, has been challenged by ever, while it is less likely that a charged species like RuO2
�
4
Cox et al. who suggest an alternative hypothesis involving will be present at the metal–gas interface, the similarity of
�
1
2�
core–hole coupling (31). Specifically, these authors claim the 800 cm band frequency to that of RuO4 suggests, in-
that both XPS spectral components are due to RuO2, with stead, that the two species share the same oxidation state.

3
44
CHAN, TAKOUDIS, AND WEAVER
Based on this line of reasoning, therefore, we tentatively that while SERS is an in-situ technique; this is not the case
�
1
assign the 800 cm feature to RuO3.
for XPS analysis where spectra were acquired at room tem-
perature after transfer from the high-pressure reactor cell.
On the other hand, however, inspecting the SER spectrum
in which a Ru metallic sample had been cooled down to
Behavioral Differences between Initially Reduced
and Partly Oxidized Ru Surfaces
toward Thermal Oxidation
�
room temperature after oxidation at 300 C (Fig. 2f), shows
the presence ofthree distinct oxides, includingRuO2, RuO3,
and RuO4, thereby agreeing nicely with the corresponding
XPS spectra (Fig. 5d).
Comparison of the SERS findings which involve thermal
oxidation of a reduced Ru and a RuO2 oxide surface display
remarkable differences in the ensuing surface composition.
While the initially RuO2 surface (prepared by Method I)
Asshown from our in-situ SERSresults, the higher-valent
oxides RuO and RuO were selectively formed at temper-
�
1
3
4
yielded primarilysurface oxygen (600cm ) upon oxidation
Fig. 3), metallic Ru (formed by Method II) is vulnerable
toward oxidizing to the higher valent oxides RuO3 or RuO4
Fig. 2). This behavioral difference is consistent with the
�
atures above 200 C. As far as the formation mechanism of
RuO2 is concerned, the decomposition of RuO4 upon cool-
ing is a likely candidate (32). This explains why the 281.0 eV
Ru(3d5/2) peak attributed to RuO2 tends to be the dominant
component in such XPS studies, even following vigorous
thermal oxidation. The thermal oxidation of the ruthenium
surface in 1 atm of oxygen can therefore be summarized by
the following reactions:
(
(
notion that RuO2 acts as a protective layer against further
oxidation (50). Considering that the formation of either
RuO3 or RuO4 involves the incorporation of oxygen into
the metal via place exchange, it would appear that oxygen
can penetrate more easily into a Ru metal than a RuO2
surface.
Oxidation
Further, comparing the results from oxidation of reduced
Ru (Fig. 2) and a sample previously covered with adsorbed
oxygen (Fig. 1B), it is found that RuO4 was formed in the
former experiment while in the latter one the highest ox-
ide detected was RuO3. Along with the observation, noted
above, that a RuO2 surface was covered with adsorbed oxy-
gen upon oxidation but without formation of higher oxides,
these experiments suggest that surface oxygen may act as a
barrier against further diffusion. This deduction is related to
the claim (8, 9) that dosed oxygen is trapped in the subsur-
face region without further oxidation on Ru(001) in UHV.
�
Ru(s) + O2(g) ⇒ RuO2(s) + RuO3(s) (200 C)
RuO3(s) + O2(g) ⇒ RuO4(s)
[1]
�
(250–300 C) [2]
Decomposition
�
RuO4(s) ⇒ RuO2(s)
(25 C)
[3]
CONCLUDING REMARKS
We have investigated the high-pressure oxidation of Ru
by using combined SERS and XPS. While the in-situ capa-
bility of SERS yields real-time information regarding the
change of surface speciation as a function of surface tem-
perature, supporting evidence such as oxidation states of
various Ru oxides is available from XPS analysis. The ma-
jor conclusions are as follows.
Thermal Stability of RuO3 and RuO4
As seen by inspecting Fig. 2, the exclusive presence of
�
RuO3 and RuO4 at temperatures above 250 C is interesting
as well as informative. This is surprising in that bulk-phase
�
RuO4 sublimes at 40 C (4). The thermal stability of the sur-
face oxide was therefore investigated further by prolonged
�
heating at 300 C. No significant attenuation of either the
8
(1) RuO2, RuO4, and possibly RuO3 were detected upon
�
1
� 1
�
00 cm and 875 cm bands was detected. Indeed, a re- thermal oxidation (25–300 C) of an initially reduced Ru
cent study reports that desorption of gas-phase RuO3 into surface in 1 atm of O2. The SERS results show that RuO3
�
UHV did not occur until above 500 C after forming a high and RuO4 were exclusively formed at temperatures above
�
coverage of atomic oxygen on Ru(001) (8).
250 C, with RuO4 decomposing upon cooling to form
While the complementary SERS and XPS spectra ap- RuO2. The latter gives rise to a XPS Ru(3d5/2) peak at
�
1
pears to be in harmony in terms of surface speciation, there 281.0 eV and a pair of bands at 510 and 630 cm in the
is a significant apparent difference in the distribution of SER spectrum.
different Ru oxides as probed by the two techniques. As
(2) In contrast to a Ru metallic surface, no significant
seen in Fig. 5c, the Ru(3d5/2) XPS spectra has a dominant formation of RuO3 and RuO4 occurred upon thermal ox-
�
2
81.0 eV component at 250 C, suggesting that the surface is idation of a RuO2 surface, suggesting that the latter oxide
composed primarily of RuO2. However, the corresponding acts as a protective layer against further oxidation. Instead,
�
1
SERS data show only weak 510 and 630 cm bands arising both the SERS and XPS experiments suggest the formation
�
from RuO2 (Fig. 2d). This discrepancy may arise from the of adsorbed oxygen on the RuO2 surface at 200 C, as diag-
�
1
fact that the XPS and SERS data were actually collected nosed by a 600 cm vibrational feature and a O(1s) peak
at different surface temperatures. It is important to realize at 531.7 eV, respectively.

SERS AND XPS STUDY OF Ru OXIDATION
345
(
3) For the first time we have shown that RuO3 and RuO4 20. Williams, C. T., Tolia, A. A., Chan, H. Y. H., Takoudis, C. G., and
Weaver, M. J., J. Catal. 163, 63 (1996).
may be responsible for the high-binding-energy compo-
nents of the Ru(3d5/2) peak of an oxidized Ru surface, based
on comparisons with the complementary SERS data. The
Ru(3d5/2) peak at 282.6 eV detected at 200 C correlates
with a SERS band at 800 cm ; both are tentatively as- 23. Tolia, A. A., Wilke, T., Takoudis, C. G., and Weaver, M. J., Chem. Engr.
signed to RuO3. The surface was further oxidized to RuO4
at 250 C, as deduced from the appearance of an 875 cm
SERS bands and a corresponding 283.3 eV Ru(3d5/2) peak.
2
2
1. Tolia, A. A., Smiley, R. J., Delgass, W. N., Takoudis, C. G., and Weaver,
M. J., J. Catal. 150, 56 (1994).
2. Williams, C. T., Tolia, A. A., Takoudis, C. G., and Weaver, M. J., Chem.
Eng. Sci. 51, 1673 (1996).
�
�
1
Sci. 47, 2781 (1992).
�
� 1
24. Tolia, A. A., Takoudis, C. G., and Weaver, M. J., J. Vac. Sci. Technol.
A 11, 2013 (1993).
25. Tolia, A. A., Williams, C. T., Takoudis, C. G., and Weaver, M. J., Lang-
muir 11, 3438 (1995).
ACKNOWLEDGMENTS
26. Tolia, A. A., Williams, C. T., Takoudis, C. G., and Weaver, M. J.,
J. Phys. Chem. 99, 4599 (1995).
The authors thank Tom Manz for assistance of the XPS experiments.
This work is supported by NSF Grant CTS-9312008.
2
2
2
7. Zhang, Y., Gao, X. P., and Weaver, M. J., J. Phys. Chem. 97, 8656
1993).
(
8. Bianchi, C. L., Ragaini, V., and Cattania, (M). G., Mater. Phys. 29, 297
(1991).
9. Atanasoska, Lj., O’Grady, W. E., and Atanasoski, R. T., Surf. Sci. 202,
REFERENCES
1
. Bond, C. G., Coq, B., Dutartre, R., Ruiz, J. G., Hooper, A. D., Proietti,
M. G., Sierra, M. C. S., and Slaa, J. C., J. Catal. 161, 480 (1996).
. Milone, C., Neri, G., Donato, A., Musolino, M. G., and Mercadante,
L., J. Catal. 159, 253 (1996).
. McQuire, M. W., and Rochester, C. H., J. Catal. 157, 396 (1995).
. Seddon, E. A., and Seddon, K. R., “The Chemistry of Ruthenium,”
Elsevier Science, New York, 1984.
142 (1988).
30. Atanasoska, Lj., Atanasoski, R. T., and Pollak, F. H., Surf. Sci. 230, 95
(1990).
31. Cox, P. A., Goodenough, J. B., Tavener, P. J., Telles, D., and Egdell,
R. G., J. Solid State Chem. 62, 360 (1986).
32. Kim, K. S., and Winograd, N., J. Catal. 35, 66 (1974).
33. Pedersen, L. A., and Lunsford, J. H., J. Catal. 61, 39 (1980).
2
3
4
5
. Mitchell, W. J., Xie, J., Lyons, K. J., and Weinberg, W. H., J. Vac. Sci. 34. Shen, J. Y., Adnot, A., and Kaliaguine, S., Appl. Surf. Sci. 51, 47
Technol. A 12, 2250 (1994).
(1991).
6
7
. Malik, I. J., and Hrbek, J., J. Vac. Sci. Technol. A 10, 2565 (1992).
. Hrbek, J., van Campen, D. G., and Malik, I. J., J. Vac. Sci. Technol. A
35. Gao, P., Gosztola, D., Leung, L-W. H., and Weaver, M. J., J Electroanal.
Chem. 233, 211 (1987).
1
3, 1409 (1995).
36. Proctor, A., and Sherwood, P. M. A., Anal. Chem. 54, 13 (1982).
37. Huang, Y. S., and Pollak, F. H., Solid State Comm. 43, 921 (1982).
8
9
. Malik, I. J., and Hrbek, J., J. Phys. Chem. 95, 10188 (1991).
. Praline, G., Koel, B. E., Lee, H.-I., and White, J. M., Appl. Surf. Sci. 5, 38. Bewick, A., Guti e´ rrez, C., and Larramona, G., J. Electroanal. Chem.
96 (1980). 332, 155 (1992).
0. Xie, J., Mitchell, W. J., Lyons, K. J., Wang, Y., and Weinberg, W. H., 39. K o¨ tz, R., Appl. Surf. Sci. 47, 109 (1991).
2
1
J. Vac. Sci. Technol. A 12, 2210 (1994).
40. Lewerenz, H. J., Stucki, S., and K o¨ tz, R., Surf. Sci. 126, 463 (1983).
1
1
1. Thomas, G. E., and Weinberg, W. H., J. Chem. Phys. 70, 954 (1979).
2. Conrad, H., Scala, R., Stenzel, W., and Unwin, R., Surf. Sci. 145, 1
41. Tillborg, H., Nilsson, A., and M a˚ rtensson, N., J. Electron Spectrosc.
Relat. Phenom. 62, 73 (1993).
(
1984).
42. Puglia, C., Nilsson, A., Hernn a¨ s, B., Karis, O., Bennich, P., and
M a˚ rtensson, N., Surf. Sci. 342, 119 (1995).
1
1
3. Shi, H., and Jacobi, K., Surf. Sci. 303, 67 (1994).
4. Mar, S. Y., Chen, C. S., Huang, Y. S., and Tiong, K. K., Appl. Surf. Sci. 43. Fuggle, J. C., Madey, T. E., Steinkilberg, M., and Menzel, D., Chem.
9
0, 497 (1995).
Phys. Lett. 33, 233 (1975).
1
1
5. Bell, W. E., and Tagami, M., J. Phys. Chem. 67, 2432 (1963).
6. Leung, L-W. H., and Weaver, M. J., J. Am. Chem. Soc. 109, 5113 45. K o¨ tz, R., and Stucki, S., J. Electroanal. Chem. 172, 211 (1984).
44. Lee, D. G., and Van Den Engh, M., Canad. J. Chem. 50, 2000 (1972).
(
1987).
46. Ortner, M. H., J. Chem. Phys. 34, 556 (1961).
47. Levin, I. W., Inorg. Chem. 8, 1018 (1969).
1
1
7. Leung, L-W. H., and Weaver, M. J., Langmuir 4, 1076 (1988).
8. Wilke, T., Gao, X. P., Takoudis, C. G., and Weaver, M. J., J. Catal. 130, 48. Griffith, W. P., J. Chem. Soc. A, 1663 (1968).
6
2 (1991).
9. Wilke, T., Gao, X. P., Takoudis, C. G., and Weaver, M. J., Langmuir 7,
14 (1991).
49. Briggs, D., and Seah, M. P., “Practical Surface Analysis by Auger and
X-Ray Photoelectron Spectroscopy,” Wiley, New York, 1983.
50. Cranstoun, G. K. L., and Pyke, D. R., Appl. Surf. Sci. 2, 359 (1979).
1
7