Reactivity of stoichiometric and defective TiO2 (110) surfaces toward DCOOD decomposition

Article abstract of DOI:10.1021/jp030511i

The effect of defects on TiO₂ (110) on the decomposition of DCOOD was investigated by temperature-programmed reaction spectroscopy (TPRS). Different concentrations of anion vacancy defects were created on the TiO₂ (110) surface by el

Full text of DOI:10.1021/jp030511i

J. Phys. Chem. B 2003, 107, 11709-11720
11709
Reactivity of Stoichiometric and Defective TiO2 (110) Surfaces toward DCOOD
Decomposition
Qinggen Wang, Jurgen Biener, Xing-Cai Guo, Enrique Farfan-Arribas, and Robert J. Madix*
Department of Chemical Engineering and Chemistry, Stanford UniVersity, Stanford, California 94305
ReceiVed: April 29, 2003
The effect of defects on TiO2 (110) on the decomposition of DCOOD was investigated by temperature-
programmed reaction spectroscopy (TPRS). Different concentrations of anion vacancy defects were created
on the TiO2 (110) surface by electron beam irradiation in order to minimize structural damage to the surface.
For comparison, defects were also created by argon bombardment. The oxidation state produced by the defects
and their concentrations were characterized by X-ray photoelectron spectroscopy (XPS). The selectivity of
the reaction for the production of carbon monoxide, formaldehyde, and D2 was increased significantly by the
defects. An appreciable decrease in the temperature for CO and D2CO formation on the defective surfaces
produced by electron beam irradiation was observed, compared to the stoichiometric surface, suggestive of
changes in the kinetics of the decomposition. The presence of isotopic exchange in the products provides
evidence for the formation of formyl intermediates during DCOOD decomposition.
I. Introduction
coupling reactions, presumably via multiple coordination of
reactants to the same cation, as suggested by Kim and Barteau.26
TiO2 is widely used as a catalyst support in industry and as
a photocatalyst for water dissociation. Its surface structure and
reactivity have been studied extensively in recent years utilizing
single crystals.1-17 Among the low-index single-crystal surfaces,
the (110) face is the most stable, and it has received considerable
attention in the surface science literature. The TiO2 (110) surface
exposes rows of coordinatively unsaturated titanium cations,
which, along with neighboring oxygen anions, form potential
Lewis acid-base pairs. The Lewis acid sites can bond electron-
donor molecules and/or act as coordination sites for the
conjugate base of an acidic reactant, such as formic acid. Indeed,
formate ions have been shown by scanning tunneling micro-
scopy to coordinate to these cations, forming highly ordered
structures along the [001] direction.3
Formic acid decomposition has been studied on stoichiometric
TiO2 (110) surfaces;37,38 formaldehyde was observed in one case
and the production of formaldehyde was proposed to be related
to oxygen vacancies on the surface.38 In this work the effects
of defects created by both electron-beam irradiation and argon-
ion bombardment on the kinetics of the decomposition of formic
acid have been studied further by controlling the defect
concentration. A reasonable mechanism to explain their effects,
which involves the participation of formyl as an intermediate
in the formation of both CO and formaldehyde, is formulated.
Furthermore, important differences in the effects of Ti3+ defects
introduced by ion bombardment and electron irradiation are
noted.
Point defects on surfaces are believed to play an important
role in catalysis.18 For example, on TiO2 (110) defects are re-
quired for water dissociation, and they also lead directly to the
reduction of formaldehyde.19 The most common methods used
to create defects on TiO2 surfaces are high-temperature anneal-
ing, electron-beam irradiation, and argon-ion bombardment.19-26
On the TiO2 (110) surface, high-temperature annealing produces
a variety of surface structures due to the migration of titanium
cations to the surface,27 and argon bombardment creates
substantial disruption of the surface order.3 We have chosen
electron irradiation for defect generation because it is expected
to generate anion vacancy defects, generating less structural
disorder, at least at low defect concentration.28
Several studies of the effects of defects on TiO2 surfaces,
generated by different methods, on the adsorption and/or reaction
of small molecules have been reported in the litera-
ture.19,23,24,26,29-36 Generally, these studies show two effects: (1)
oxygen vacancy defects can cause reduction (e.g., water, NO,
and CH2O) or dehydration (e.g., alcohols); (2) high degrees of
coordinative unsaturation of the titanium cations can lead to
II. Experimental Section
The experiments were performed in a UHV system that
consists of adjoined preparation and spectroscopy chambers. The
preparation chamber is equipped with a sputter-ion gun and gas
dosers. The spectroscopy chamber is equipped with LEED
optics, a Perkin-Elmer 04-548 dual anode X-ray source, an EA
10 Plus hemispherical energy analyzer from SPECS, a UTI 100c
quadrupole mass spectrometer, and a syringe-needle gas doser.
The base pressures of the preparation chamber and the spec-
troscopy chamber were 6 × 10-10 Torr and 3 × 10-10 Torr,
respectively. During the XPS and TPRS measurements the
preparation chamber was isolated from the spectroscopy cham-
ber.
The polished TiO2 (110) rutile single crystal (10 × 10 × 1
mm3, supplied by Atomergic Chemetals, Corp.) was supported
by a tantalum foil. The foil was resistively heated via two
tantalum wires spot-welded to the back of the foil, and the
crystal was heated by conduction. The temperature was moni-
tored by a Chromel-Alumel thermocouple glued to the side of
the crystal with ceramic adhesive (Aremco Products, Inc.). After
cycles of sputtering and annealing, the crystal appeared light
* Corresponding author. E-mail: rjm@chemeng.stanford.edu.
10.1021/jp030511i CCC: $25.00 © 2003 American Chemical Society
Published on Web 09/27/2003

11710 J. Phys. Chem. B, Vol. 107, No. 42, 2003
Wang et al.
TABLE 1: Surface Composition of TiO2 (110) As
Characterized from Ti 2p3/2 XPS Peak on Surfaces of
Different Exposure Times to Electron Irradiation at 300 eV
and Ar Bombardment for 1 min
TiO2 E5 E7 E10 E15 Ar1
Ti4+ percentage
100% 97% 90% 85% 81% 81%
Ti3+ percentage
0%
0
3% 10% 15% 19% 19%
0.06 0.21 0.32 0.40 0.40
oxygen defect concentration (ML)
surfaces.24,39,40 Hence the surface prepared was assumed to be
stoichiometric, though the existence of a small concentration
of anion defects is possible.
Electron-beam irradiation introduces defects to the surface.
It is believed that the defects produced are anion vacancy defects
produced by electron-stimulated desorption (ESD) of surface-
bridging oxygen ions, though desorption of in-plane oxygen ions
cannot be ruled out completely. Further, at least at low
concentration of defects, the overall surface structure is relatively
undamaged during the process.20,28 For the exposure times
studied here, only Ti3+ defects were produced on the surface,
as indicated by the additional Ti 2p3/2 peak centered at 457.4
eV in Figure 1. It was also observed on other surfaces that the
Ti3+ 2p3/2 peak had a binding energy ∼1.8 eV lower that that
of Ti4+ 2p3/2 peak.21,41
By varying the exposure time, surfaces with a wide range of
defect concentrations were prepared. The area of the Ti3+ 2p3/2
peak increased from 0 to 19% of the total area of the Ti 2p3/2
peaks as the exposure to the electrons was increased. The
percentage areas of Ti3+ and Ti4+ 2p3/2 peaks for the surfaces
prepared in this manner are listed in Table 1. For simplicity, in
this paper the stoichiometric surface is designated as TiO2, the
surface irradiated by electron beam for 5 min as E5, and the
surface bombarded by Ar+ for 1 min as Ar1. Other surfaces
were designated in analogous fashion. Table 1 also lists the
estimated defect concentration at the surface. The concentration
was estimated using the published value for the escape depth
of electrons from the TiO2 (110) surface,39 assuming that Ti3+
cations reside only on the top layer. In this way the defect
concentration following 15 min of irradiation by electrons is
estimated to be approximately 40% of a monolayer (1 monolayer
) ∼1.04 × 1015 sites cm-2). For the surface of such high defect
density we may anticipate some surface reconstruction to occur
to accommodate the high polarity on the surface due to the
existence of Ti3+. Extensive reconstruction is not likely as the
(1 × 1) LEED pattern is still discernible on such surface, though
with a high background (see below).
Argon-ion bombardment creates Ti3+ point defects through
preferential sputtering of oxygen ions. After Ar+ bombardment,
only Ti3+ defects were detected by XPS; the Ti3+ 2p3/2 peak
was slightly sharper than those observed on electron-irradiated
surfaces. The Ti3+ 2p3/2 peak area was 16% of the total area of
the Ti 2P3/2 peaks after 1 min sputteringsa concentration close
to that observed for the E15 surface. The XPS spectrum of the
Ar+ sputtered surface is shown in Figure 2 as an illustration of
the peaks used for peak fitting for the Ti XP spectra obtained
in this set of experiments. Scanning tunneling microscopy
reveals that Ar+ bombardment severely disrupts the surface
structure. The bombarded surface is much rougher than the
original surface; small islands ca. 4 nm in diameter are formed,
and long-range order is severely reduced.3
Figure 1. Ti 2p core-level XP spectra of TiO2 (110) surfaces:
stoichiometric surface (9), surfaces irradiated by electron beam for 5
min (b), 7 min (2), 10 min ([), and 15 min (x).
blue. Before each experiment the TiO2 (110) surface was cleaned
by sputtering for 30 min with 500 eV argon ions, followed by
annealing in a vacuum at 850 K for 15 min.
Defects were created both by irradiation with electrons and
by argon-ion bombardment. For electron irradiation the crystal
surface was biased at 300 V and exposed to electrons emitted
from a nearby hot filament. Currents measured were typically
0.7 mA, but as the sample holder was also exposed to the
electron beam, the estimated current on the sample is about one-
tenth of that measured. Argon ions were generated by a
conventional sputtering gun (Physical Electronics) with 5 × 10-5
Torr argon and 500 eV high voltage, the current density at the
sample was ∼6 µA/cm2.
The cleanliness and composition of the surface were checked
by XPS using nonmonochromatic Mg KR X-rays. The photo-
electrons emitted normal to the sample surface were collected
by the energy analyzer at a pass energy of 25 eV. Binding
energies were calibrated by the Au 4f peak (84.00 eV). DCOOD
(Aldrich, 99+ atom % D) was used for temperature-programmed
reaction studies. DCOOD liquid was outgassed by freeze-
pump-thaw cycles before use each day. In each TPRS
experiment, identical saturation DCOOD doses were adminis-
tered through the needle doser at room temperature. TPR spectra
were recorded by a computer coupled with the mass spectrom-
eter, whose ionizer was enclosed in a glass cap with a small
hole facing the crystal surface.
III. Results
A. Surface Characterization. The surface composition was
first characterized by XPS (Figure 1). In the absence of Ar+
bombardment or electron-beam irradiation there existed only
one symmetric Ti 2p3/2 peak centered at 459.2 eV, consistent
with the binding energies reported for Ti4+ 2p3/2 on rutile (110)
The structure of these surfaces was also studied by LEED.
The images are shown in Figure 3. The stoichiometric surface
gave a sharp 1 × 1 pattern. When the surface was subjected to
electron irradiation, the LEED spots became more diffuse, and
the background became brighter with longer irradiation time.

TiO2 (110) Surfaces and DCOOD Decomposition
J. Phys. Chem. B, Vol. 107, No. 42, 2003 11711
Figure 2. Peak fitting of Ti 2p core-level XP spectrum of TiO2 (110)
Figure 4. Rates of the products in steady-state DCOOD decomposi-
tion: CO (9) and CO2 (b). Dashed line is the fitting for CO reaction,
dotted line for CO2 reaction. ‚ - ‚ line is the fitting using the activation
energy determined by Henderson.38
surface bombarded by Ar+ for 1 min.
But even after electron irradiation for 15 min, the 1 × 1 LEED
pattern persisted. For the surface sputtered with Ar+ for 1 min,
no LEED pattern could be detected. The LEED results confirm
that the TiO2 (110) surface was not disrupted by electron
radiation significantly, while Ar+ sputtering destroyed the long-
range structure on the surface completely.3
B. Steady-State Reaction on a Stoichiometric Surface. For
reference we first examined the steady-state catalytic activity
of stoichiometric TiO2 (110) for the DCOOD decomposition
in a fashion similar to that conducted by Onishi et al.37 and
Bowker et al.9 The pressure of formic acid used was 2 × 10-7
Torr. Our results closely paralleled the results of Onishi et al.
Two channels for DCOOD decomposition to CO and CO2 were
observed, and the decomposition of DCOOD to CO has a higher
activation energy than that for CO2 formation. The effect of
temperature on the parallel reaction routes to CO and CO2 is
plotted in Figure 4. By extrapolating the fit of the experimental
data, the crossover in reactivity to form CO or CO2 was
estimated to be 525 K; this result is qualitatively in agreement
with Onishi et al. who reported that the rate of the dehydrogen-
ation reaction exceeded that of the dehydration reaction below
500 K. The activation energies deduced for CO and CO2
formation are 63.5 kJ mol-1 and 27 kJ mol-1, respectively. For
comparison, a dot-dash-dot line is drawn through our data
for CO using the activation energy determined by Henderson.38
The agreement is quite good. In contrast, Bowker et al.9 did
not observe the dehydrogenation reaction in their molecular
beam experiments. The origin of this difference is unknown.
Since the surface must be “conditioned” in the beam for a period
of time before steady-state is reached, there may be subtle
differences in surface structure among the three studies that
influence reactivity.
Figure 3. LEED pattern observed on stoichiometric surface, surfaces
irradiated by electron beam for 5, 7, 10, and 15 min.

11712 J. Phys. Chem. B, Vol. 107, No. 42, 2003
Wang et al.
Figure 5. TPR spectra for saturation exposure of DCOOD on
stoichiometric TiO2 (110) at room temperature. The signals are corrected
by QMS sensitivity.
Figure 6. CO TPRS spectra from DCOOD decomposition on TiO2
(110) surfaces: stoichiometric surface, surfaces irradiated by electron
beam for 5 min, 7 min, 10 min, 15 min, and surface bombarded by
Ar+ for 1 min.
C. TPRS on Stoichiometric and Defective Surfaces. In
agreement with previous work,38 we observed DCOOD to
decompose into CO, D2CO, CO2, D2O, and D2 on both the
stoichiometric and defective TiO2 (110) surfaces. Methanol was
not observed. The TPR spectrum for DCOOD decomposition
on the stoichiometric surface following the adsorption of formic
acid near room temperature is shown in Figure 5. The CO2
spectrum was corrected by subtracting the cracking fraction from
DCOOD. Among the decomposition products, CO and D2O
were the major products (∼60%), D2CO and CO2 were minor
products (∼10%), and a small amount of D2 (∼2%) was
observed as well.
TABLE 2: Peak Temperature of the Products from
DCOOD TPRS on the Surfaces Studied
D2 (K) D2O (K) CO (K) D2CO(K) CO2 (K) DCOOD(K)
TiO2 444 575 438 577
E5
E7
E10 428 535 404 575 470 546
E15 428 517 408 538 474 541
Ar1 428 591 410 606
-
567
427 560 438 579 462 555
436 566 410 583 466 553
575
528
487
479
472
556
475 585 411 564
456 580 410 554
476 577 417 562
456 590 408 543
496 580 410 508
436 613 410 564
-
567
The effects of defects on the formation of each of the different
products are now described separately. General discussion of
the implications of the effects of the defects then follows in the
Discussion section.
TABLE 3: Area Percentage of CO from Low- and
High-Temperature Channels on Electron-Irradiated Surfaces
(i) CO. Figure 6 shows CO/DCOOD TPR spectra for each
surface. The peak temperatures for CO, as well as for other
products, on each surface studied are listed in Table 2. CO was
evolved at 566 K in a single peak on the stoichiometric surface.
The peak shape and peak temperature resemble those reported
by Henderson.38 Assuming a preexponential factor of 2 × 109
TiO2
E5
E7
E10
E15
low temp
high temp
0%
100%
33%
67%
42%
58%
66%
34%
83%
17%
is obvious from Figure 6 that the area of the high-temperature
CO peak shrinks with the irradiation time. This implies that
the high-temperature channel is short-circuited by the existence
of the low-temperature channel. CO desorption from the CO-
dosed E10 surface occurred at 425 and 465 K, suggesting that
CO evolution at 560 K from the electron-irradiated surfaces is
reaction limited in the case of formic acid.
The existence of two CO desorption channels may be related
to the adsorption configuration of formate on a defective surface.
On the stoichiometric surface one formate occupies two
neighboring 5-fold coordinated Ti4+ cations (Figure 7, config-
uration A, R-bridging).42-44 Recent studies by HREELS, STM,
and NC-AFM have confirmed the bidentate nature of this
formate species, in good agreement with theoretical calcula-
tions.11,15-17 On a defective surface a formate may take other
s-1 37
the activation energy for the formation of gaseous CO
,
was estimated from the TPR spectrum to be 112 kJ mol-1 for
the stoichiometric surface, which is in good agreement with the
result of Onishi et al. (120 kJ mol-1).37
After the stoichiometric surface was irradiated with the
electron beam, an additional CO peak appeared around 465 K.
This peak is broader than the peak on the stoichiometric surface,
and its intensity increases with the irradiation time. The low-
temperature channel becomes the dominant one on E10, and
on E15 most of the CO originates from this low-temperature
channel. The area percentage of the CO produced from each
channel on the different surfaces is listed in Table 3. The area
was calculated by fitting each spectrum with two Gaussian
peaks; a sample fit is shown in Figure 6 for the E10 surface. It

TiO2 (110) Surfaces and DCOOD Decomposition
J. Phys. Chem. B, Vol. 107, No. 42, 2003 11713
TABLE 4: Selectivity to CO, D2CO, and CO2 of DCOOD
Decomposition on Different Surfaces
TiO2
E5
E7
E10
E15
Ar1
CO
D2CO
CO2
58%
10%
32%
63%
20%
16%
67%
22%
11%
62%
34%
5%
65%
32%
3%
73%
22%
5%
for the low-temperature channels30 kJ mol-1 lower than the
high-temperature channel. Defects introduced by electron-beam
irradiation are much more active at decomposing DCOOD to
CO than the 5-fold coordinated Ti4+ sites exposed on the
stoichiometric TiO2 (110) surface.
However, compared to the stoichiometric surface, the CO
desorption peak on the defective surface created by argon
bombardment did not move, but broadened and exhibited an
increase in peak area. Therefore the defects introduced by Ar+
bombardment increased the selectivity toward CO but not the
reactivity. Considering that the extent of reduction on the Ar1
surface is similar to that on the E15 surface, this shows that
DCOOD decomposes more easily in the presence of anion
vacancy defects in the regular TiO2 (110) structure; i.e., on the
surface after electron exposure. The local environment of the
defects clearly plays a significant role in determining the
reactivity. That the reactivity of DCOOD decomposition to CO
is sensitive to the structure of the substrate surface can also be
exemplified by comparing reactions on TiO2 (110) and TiO2
(100) surfaces.38,46 It has been shown that, under similar
experimental conditions, formic acid decomposed to CO on the
stoichiometric TiO2 (100) (1 × 1) surface at 590 K, while on
the TiO2 (110) (1 × 1) surface it decomposed at ∼560 K.
Finally, the difference observed between the effects of the argon
bombardment and electron irradiation make it clear that the
structures of the two resulting surfaces differ appreciably.
CO remains the dominant decomposition product on all
surfaces studied, as illustrated by the selectivity to CO listed in
Table 4. When calculating the selectivity, it was assumed that
CO, D2CO, and CO2 were the only carbon-containing decom-
position products, and both channels of the CO formation were
taken into account. The signals were corrected for the mass
spectrometer sensitivity. Table 4 also shows that the overall
selectivity to CO did not vary much on the defective surfaces
produced by electron irradiation.
(ii) D2O. By stoichiometry the formation of CO would be
expected to lead to similar amounts of water. Overall, however,
the amount of D2O formed decreases with electron irradiation
time, which is consistent with oxygen being used to fill the
oxygen vacancies. The D2O TPRS peaks from the different
surfaces are shown in Figure 8. In contrast to the results for
CO, on the stoichiometric surface D2O is formed in two separate
reaction channels; one occurs at 440 K, and the other occurs at
577 K, neither of which corresponds to the temperature of CO
formation. These two channels are observed on all electron-
irradiated surfaces, with the high-temperature peak changing
from the major to the minor feature as irradiation time increases.
Interestingly, the temperature of these two peaks on all the
surfaces shows little shift, except that the high-temperature peak
on the E15 surface was about 40 K lower than on all the other
electron-irradiated surfaces.
Figure 7. Schematic representations of the various configurations for
formate on an electron-irradiated TiO2 (110) surface.
configurations. The possible adsorption geometries of for-
mates on a defective surface with oxygen vacancies in the
bridging oxygen rows are shown in Figure 7. Among the
configurations, one of the oxygen atoms in a formate may
occupy one oxygen vacancy (coordinating with two Ti3+ sites)
in the bridging oxygen row, while the other oxygen atom either
dangles (configuration B, monodentate) or bonds with a Ti4+
site in the neighboring row (configuration C, â-bridging). A
formate can also take configuration D in which it bonds to two
neighboring Ti3+ sites which originated from one oxygen
vacancy. Configuration D is similar to A, but it is presumably
less stable since one oxygen vacancy may not provide enough
space to accommodate the two oxygen atoms of the formate.
The existence of two neighboring oxygen vacancies in the
bridging oxygen row of significant concentration was excluded
by the fact that no Ti2+ was detected by XPS. Likely the most
stable form is â-bridging, as it has been suggested by theoretical
calculations24 and detected by absorption IR spectroscopy on
the TiO2 (110) surface.45 Different studies using HREELS,15
NEXAFS,14 and STM9 have provided additional evidence of
the existence of a second formate species. Monodentate and
â-bridging configurations will reduce the number of pairing Ti4+
sites available for formate adsorption in R-bridging configura-
tion, hence less CO would appear in the high-temperature peak.
The generation of oxygen vacancies in the in-plane oxygen rows
may be another source of the reduction of the amount of formate
in R-bridging configuration; however, the number of such
oxygen vacancies is expected to be insignificant.20,28 Using the
same preexponential factor for CO production on the stoichio-
metric surface, we estimated an activation energy of 92 kJ mol-1
On the Ar+-bombarded surface both the low- and the high-
temperature features are reduced to a similar magnitude, but
the high-temperature peak is much wider than the low-
temperature one. Compared with the stoichiometric surface, the
peak temperature of the low-temperature channel shifts to lower
temperature and becomes sharper, while the second peak shifts
to a higher temperature.

11714 J. Phys. Chem. B, Vol. 107, No. 42, 2003
Wang et al.
Figure 10. CO2 TPRS spectra from DCOOD decomposition on TiO2
(110) surfaces: stoichiometric surface, surfaces irradiated by electron
beam for 5 min, 7 min, 10 min, 15 min, and surface bombarded by
Ar+ for 1 min.
Figure 8. D2O TPRS spectra from DCOOD decomposition on TiO2
(110) surfaces: stoichiometric surface, surfaces irradiated by electron
beam for 5 min, 7 min, 10 min, 15 min, and surface bombarded by
Ar+ for 1 min.
(iii) D2CO. The D2CO TPRS from all surfaces studied is
shown in Figure 9. On the stoichiometric surface the peak
temperature was only slightly higher than that of CO (Table
2). With the increase of the electron irradiation time, the D2CO
peak shifted steadily to lower temperature, from 575 K on the
stoichiometric surface to 471 K on the E15 surface. On all these
surfaces the peak temperature for D2CO formation was signifi-
cantly lower than that for CO evolution in the higher temperature
channel. The selectivity for D2CO increased with defect
concentration, from 10% on the stoichiometric surface to 34%
on the surface irradiated by electrons for 10 min (Table 4). On
the Ar+-bombarded surface the D2CO peak temperature was
only slightly lower than that for CO. The selectivity to D2CO
was higher than the stoichiometric surface, lower than the E15
surface, and similar to the E7 surface. The increase in D2CO
production appears to be at the expense of CO2 formation.
(iV) CO2. Figure 10 shows the CO2 product spectra, corrected
for the cracking fraction of DCOOD. On the stoichiometric
surface the CO2 spectrum shows peaks at 475 and 585 K, with
both peaks showing similar intensity. The area of both peaks
decreases as the electron irradiation time increases, but the low-
temperature peak decreases faster, and it almost disappears on
the surface irradiated for 15 min. The presence of defects created
by electron beam suppressed the decomposition of formate to
CO2. For the electron-irradiated surfaces there are variations in
the temperature of both the high- and low-temperature peaks,
but variations appear to be small.
When the surface was bombarded with Ar+, the amount of
CO2 decreased in both the low- and high-temperature range.
The low-temperature peak shifted to lower temperature, while
the high-temperature peak shifted to higher temperature. This
behavior differs sharply from that observed on the electron-
beam-irradiated surfaces.
Figure 9. D2CO TPRS spectra from DCOOD decomposition on TiO2
(110) surfaces: stoichiometric surface, surfaces irradiated by electron
beam for 5 min, 7 min, 10 min, 15 min, and surface bombarded by
Ar+ for 1 min.

TiO2 (110) Surfaces and DCOOD Decomposition
J. Phys. Chem. B, Vol. 107, No. 42, 2003 11715
Figure 11. D2 TPRS spectra from DCOOD decomposition on TiO2
(110) surfaces: stoichiometric surface, surfaces irradiated by electron
beam for 5 min, 7 min, 10 min, 15 min, and surface bombarded by
Ar+ for 1 min.
Figure 12. DCOOD TPRS spectra from DCOOD decomposition on
TiO2 (110) surfaces: stoichiometric surface, surfaces irradiated by
electron beam for 5 min, 7 min, 10 min, 15 min, and surface bombarded
by Ar+ for 1 min.
(V) D2. Figure 11 shows the TPR spectra of D2 for the
different surfaces. On the stoichiometric surface D2 desorbed
at 440 and 573 K. The low-temperature peak was smaller in
intensity than the high-temperature one. After the surface was
irradiated by electrons, both the low- and high-temperature peaks
increased in magnitude with the defect concentration, and the
high-temperature peak shifted slightly to lower temperature. On
the Ar+-bombarded surface the two-peak feature still remained,
but the high-temperature peak had a higher intensity and shifted
to higher temperature compared with the stoichiometric surface,
the low-temperature peak had almost no change. The amount
of D2 desorbed was significantly less than that on the E15
surface, which had a Ti3+ concentration similar to the Ar+-
bombarded surface.
(Vi) DCOOD. The amount of DCOOD desorbed from the
stoichiometric surface is small, though peaks with similar
intensities are discernible at 406 and 566 K (Figure 12). These
features have previously been assigned to reaction of formate
and hydroxyl species.38 After defects were introduced by
electron irradiation, the DCOOD formation appeared to be
slightly suppressed at the low-temperature peak, and there is
no obvious trend in the change of the high-temperature peak.
After Ar+ bombardment, neither peak shifts, but the low-
temperature peak shrinks and the other peak shows a large
increase in intensity.
HCOOD was used as a reagent on the E10 surface to ascertain
whether the H-C bond breaks when forming formaldehyde.
No D2CO (m/e ) 32) was detected in this experiment, revealing
that intermediates necessitating cleavage of the H-C bond are
not involved in formaldehyde formation.
E. Oxidizability of the Defective Surfaces. The left panel
of Figure 14 shows the Ti 2p X-ray photoelectron spectra of
the E15 surface after being dosed with DCOOD at room
temperature and subsequently flashed to 450 and 600 K in less
than 1 min. The spectra were taken at room temperature. It took
about 5 min for the sample to cool from 600 K to room
temperature. As the figure shows, even at room temperature
the DCOOD reoxidized the defects to a small extent. The surface
was oxidized further after the sample was flashed to 450 K, at
which temperature formate started decomposing. At this point
about half of the Ti3+ was oxidized to Ti4+. Most of the Ti3+
(∼80%) was oxidized to Ti4+ at 600 K, at which temperature
most of the formate on the surface has decomposed. For the
Ar+-bombarded surface the oxidation was noticeable (∼20%,
Figure 14, right) only after the DCOOD-dosed surface was
flashed to 600 K. Most of the formate on the surface has
decomposed at this temperature. The oxidation of the surface
defects was mainly due to the presence of formate, the diffusion
of Ti3+ from the surface to the bulk did not play a major role.
Without DCOOD adsorption, no oxidation was detectable of
the Ar1 surface after it was flashed to 600 K. Only about 35%
Ti3+ was oxidized to Ti4+ on the E15 surface under the same
conditions, while no oxidation was observed after the surface
was flashed to 450 K. The difference between the oxidizability
of the electron irradiated and the Ar+-bombarded surfaces again
suggests that the defect structures of Ti3+ cations on these
surfaces are not the same.
D. Isotopic Effects. A mixture of H13COOH and DCOOH
was dosed onto the stoichiometric and the E15 surfaces. Peaks
for CO (m/e ) 28) and 13CO (m/e ) 29) are shown in Figure
13. On both surfaces the 13CO peak leads the CO peak by about
8 K. On the stoichiometric surface the 13CO and CO peaks show
similar intensity, while on the E15 surface, the CO peak is about
30% higher than the 13CO peak.

11716 J. Phys. Chem. B, Vol. 107, No. 42, 2003
Wang et al.
Figure 13. 12CO (9) and 13CO (b) TPRS spectra from a mixture of D12COOH and H13COOH decomposition on stoichiometric TiO2 (110) surface
(left) and TiO2 (110) surface irradiated by electron for 15 min (right).
Figure 14. Left: Ti 2p core-level XPS spectra of TiO2 (110) surfaces irradiated for 15 min (9), after dosing DCOOD (b) at room temperature
followed by flashing to 450 K (2) and 600 K ([). Right: Ti 2p core-level XPS spectra of TiO2 (110) surfaces bombarded by Ar+ for 1 min (9),
after dosing DCOOD at room temperature followed by flashing to 600 K (b).
The influence of anion vacancy defects on the decomposition
of the formate was explored with oxygen post-dosing experi-
ments. The experiments consisted in adsorbing HCOOH on the
E10 surface and subsequently exposing the surface to oxygen

TiO2 (110) Surfaces and DCOOD Decomposition
J. Phys. Chem. B, Vol. 107, No. 42, 2003 11717
Figure 15. TPR spectra for a saturation exposure of HCOOH on the
E10 TiO2 (110) surface after 120 L of oxygen post-dosed at room
temperature. The signals are corrected for QMS sensitivity.
Figure 16. Integrated peak area of D2 (9), D2O (high-temperature
peak, b), CO (2), D2CO ([), CO2 (1), DCOOD (*) from TPRS of
DCOOD versus the defect concentration on the TiO2 (110) surfaces
after electron irradiation.
at room temperature. The TPRS spectra for such an experiment
is shown in Figure 15. Several differences were observed
between this spectra and the one obtained from the E10 surface
in the absence of oxygen. On the oxygen post-dosed surface,
CO desorption occurred in two states as it did on the defective
surfaces when oxygen was not post-dosed. However, the low
temperature peak, which was characteristic of the defective
surfaces and dominated on the E10 surface, decreased substan-
tially in favor of the high-temperature peak when oxygen was
post-dosed. The area of the low-temperature peak was calculated
to be 26% of the total, as opposed to the 66% obtained for the
E10 surface in the absence of oxygen. This behavior suggests
that oxygen has refilled some of the defects, and that these
defects were indeed responsible for the low-temperature CO
desorption state. Also, the H2CO desorption peak decreased by
35% on the oxygen post-dosed surface and shifted back to higher
temperature (505 K). This is also consistent with the fact that
oxygen is refilling the vacancies since, as shown previously,
the H2CO peak progressively increases in area and shifts to
lower temperatures with increasing number of defects. No new
products were detected in these experiments that could be
attributed to the reaction of formic acid with oxygen. In addition,
when 18O2 was used for the post-dosing experiment, 18O2 was
not incorporated into the products, suggesting that the main role
of oxygen is to refill vacancies and not to react with the adsorbed
species.
formed (D2, CO, and formaldehyde) from DCOOD decomposi-
tion increase with defect density, while CO2 formation is
suppressed by the defects. This effect is clearly shown in Figure
16, which relates the peak areas for the different products to
the defect density.
On the stoichiometric surface the majority of the CO is
believed to be the product of unimolecular decomposition of
adsorbed formate:37,38
DCOO-(a) f CO(g) + OD-
(3)
(a)
This mechanism is consistent with the observation by STM that
formate is formed on the TiO2 (110) surface and that basic
oxygen anions are adjacent to the coordinatively unsaturated
Ti4+ cations to which formate binds. However, a study by
Henderson38 shows that surface oxygen is exchanged readily
into the CO product, and reaction 3 does not account for this
occurrence. According to Henderson, “oxygen atoms in adsorbed
formate species readily scrambled with lattice oxygen atoms in
the TiO2 (110) surface.” The exchanged oxygen appears in CO
and formaldehyde in a significant amount above 400 K, at which
temperature oxygen vacancies are introduced onto the surface
by condensation of bridging hydroxyl groups. This active role
of the surface in the reaction has been demonstrated in an STM
study by Bowker et al.9 that showed the formation of a “(1 ×
1/2)” structure in islands at the temperature of formate decom-
position. This structure was shown to correspond to a (1 × 1)
structure in which several bridging oxygen atoms had been
removed.
IV. Discussion
In summary, the defects created by electron irradiation
significantly reduce the temperature for the evolution of CO
and formaldehyde. A low-temperature reaction channel grows
in importance with increasing defect density, ultimately domi-
nating product formation. The amounts of the reduced products
The exchange of surface oxygen into formate and, subse-
quently, formaldehyde and CO is consistent with interconversion

11718 J. Phys. Chem. B, Vol. 107, No. 42, 2003
Wang et al.
small, as only a single C(1s) peak (289.0 eV), assignable to
formate, was detected in the XP spectrum. Assuming the C1s
binding energy of formyl is no greater than that of formaldehyde,
an additional peak about 1 eV below that of formate51,52 should
be observed if the concentration of formyl is significant. The
isotopic effect observed for carbon monoxide formation suggests
that C-H bond scission, not the C-O bond breakup, is the
rate-limiting step in the formation of carbon monoxide on both
the stoichiometric and the electron-irradiated surfaces.
Clearly D2CO can only be a product of DCOO reduction.
Therefore, the amount of D2CO formed is expected to increase
as the number of point defects grows. As shown in Figure 17,
no isotopic effect was observed for D2CO and DHCO production
from the decomposition of DCOOH on the E15 surface. This
indicates that the rate of formation of formaldehyde on the E15
surface is not limited by the rate of transfer of hydrogen to
formyl. The slowest reaction in this process is probably the
formation of formyl, though the reaction of formyl with hydroxyl
is also a possible rate-limiting step, as no isotopic effect was
detected for the liberation of water from this surface either.
When the number of defects was decreased by post-dosing
oxygen on the defective surfaces, the amount of formaldehyde
produced was smaller, supporting the idea that point defects
are responsible for the increase in formaldehyde production
observed on the defective surfaces. Similarly, the lower-
temperature CO peak that had developed in the presence of
defects was seen to decrease in the oxygen post-dosing
experiments as well, suggesting that oxygen post-dosing did in
fact result in filling of anion defects. The filling of anion vacancy
defects by exposure to O2 has been previously reported in the
literature.6,53 Diebold has suggested that the oxygen released
in formaldehyde formation may not only be used to fill
vacancies but also could react with Ti3+ interstitials.8 An STM
study by Bennett et al. on the adsorption of formic acid on the
cross-linked (1 × 2) reconstructed surface has shown that
oxygen atoms are inserted into the surface from formate during
decomposition, resulting in the formation of new TiO2 (1 × 1)
islands.10
Figure 17. D2CO (9) and DHCO (b) TPRS spectra from DCOOH
decomposition on TiO2 (110) surface irradiated by electron for 15 min.
(DHCO signal was multiplied by 1.6.)
between formate and formyl on this surface, according to
DCOO(a) + OV h DCO(a) + O(lattice)
(4)
where OV designates an oxygen anion vacancy. The oxygen
anion defects serve as sites for the oxygen to be released from
the formate anion. Formyl may further decompose to CO or
extract a hydrogen atom from a hydroxyl group to form
formaldehyde. With increasing defect concentration more formyl
is formed, more reoxidation of the surface by formate occurs,
and more CO and formaldehyde are formed.
In addition to the oxidation state of the titanium, surface
structure also plays an important role in determining the
selectivity of the decomposition of the acid to formaldehyde.
The selectivity of the reaction for formaldehyde on the Ar+-
bombarded surface is lower than that on the E15 surface, which
has a similar defect concentration. Further, the peak temperature
on the two surfaces is dramatically different, being significantly
lower on the electron-irradiated surface. A dependence of the
selectivity to formaldehyde on the surface structure was also
noticed on the TiO2 (001) surface by Idriss et al.29 They
observed that the yield of formaldehyde from formic acid
decomposition first decreased and then increased with the
annealing temperature of the Ar+-bombarded TiO2 (001) surface.
The effect of defects created by Ar+ bombardment on the
TiO2 (110) surface on the steady-state decomposition of formic
acid has been previously investigated.54 No difference in the
catalytic activity of the Ar+-bombarded partially annealed and
the well-ordered TiO2 (110) surfaces for the DCOOD decom-
position was detected. Our TPRS results confirm the finding
of Onishi et al. that the activity of DCOOD decomposition to
CO was not altered by Ar+ bombardment, as the temperature
at which CO evolved from both the stoichiometric and the Ar1
surfaces is the same. However, our results show a significant
drop in selectivity to CO2 after Ar+ bombardment. That they
did not see a change in the production of CO2 is probably due
to reoxidation of the ion-bombarded surface during the exposure
The formyl intermediate has been proposed to participate in
many reactions, e.g., reactions between CO and H2 on rhodium
catalysts,47 decomposition of formic acid on nickel surfaces,48
and the adsorption of CO on hydrogen-reduced cerium oxide,49
among others. Its existence on other surfaces has been sug-
gested.49,50 Its formation on the TiO2 (110) surface is further
supported indirectly from the decomposition of DCOOH on the
E15 surface. Shown in Figure 17 is the TPR spectra of D2CO
(m/e ) 32) and DHCO (m/e ) 31) from DCOOH decomposi-
tion. D2CO and DHCO were formed in a ratio of 3:2; when
scaled, the two spectra track each other very well, indicative of
a common rate-limiting reaction intermediate in their formation.
The intermediate is very likely to be formyl. D2CO is formed
from the combination of DCO with OD, the decomposition
product of DCOO, and DHCO is formed from combination of
DCO and residual OH, which results from the dissociation of
DCOOH during adsorption and would otherwise disproportion-
ate to form water at about 525 K. Greater intensity of D2CO
than DHCO is reasonable, as most of the OH which resulted
from the dissociative adsorption of DCOOH is channeled into
H2O formation at ∼410 K (Figure 8). The formyl concentration
on the defective surfaces at room temperature is apparently

TiO2 (110) Surfaces and DCOOD Decomposition
J. Phys. Chem. B, Vol. 107, No. 42, 2003 11719
to formic acid required to reach steady state at the elevated
temperatures of the experiments; the reactivity detected may
no longer reflect the defective state of the original reduced
surface.
The low-temperature peak for water desorption from the
stoichiometric surface is believed to originate from the recom-
bination of hydroxyl groups on the surface, which have been
formed from formic acid dissociation during adsorption:38
during reaction and have identified it as a carbonate species
(CO32-). This species may be responsible for some of the CO2
production. The suppression of the production of CO2 by defects
could be because Ti3+ has a higher affinity for oxygen atoms
than the 5-fold coordinated Ti4+, and the oxygen in formate is
prone to extraction by Ti3+
.
Clearly the increase in the amount of D2 formed with
increasing Ti3+ concentration indicates that the anion defects
promote D2 formation. Chemisorption of H2 on defective TiO2
(110) surfaces produced by high-temperature annealing has been
DCOOD + O2-(lattice) f DCOO-(a) + OD-
(1)
(2)
(lattice)
reported, and it was suggested that H2 dissociates to form Ti4+
-
2OD-(lattice) f D2O (g) + O2-(lattice) + Ov
H- bonds during the process.32,55,56 Recently, bright features
on a STM image of a TiO2 (110) surface annealed at 900 K
were assigned to Ti-H species.57 On defective surfaces the D
atom may be extracted from DCOO-/DCO by Ti3+ sites on
the surface to form Ti4+-D- pairs; the D atoms then combine
to form D2. D2 also could be the product of the condensation
of the hydroxyl groups on the surface generated during DCOOD
dissociative adsorption, as observed by Lu and co-workers.19
The existence of the second channel is supported by the
production of H2 from DCOOH decomposition on both sto-
ichiometric and E15 surfaces. In either case, higher defect
concentrations on the surface should produce larger amounts
of D2, as observed. The complete decomposition of formate to
yield D, C, and lattice oxygen is unlikely, as no residual carbon
was detected by XPS.
where “Ov” refers to an oxygen vacancy site, “a” denotes an
adsorbed species and “g” a gaseous species. It is reasonable to
assume that the lower temperature water peak on the defective
surfaces has the same origin. It appears that the reactivity of
water formation from hydroxyl groups at this low temperature
channel is not significantly affected by the presence of point
defects. Lu and co-workers19 also observed that water desorbed
at the same temperature on stoichiometric and defective TiO2
(110) surfaces. The decrease in the magnitude of the high-
temperature water peak with the electron irradiation time is
consistent with observations that (1) the oxygen from the
adsorbed formate is retained by the surface to reoxidize the Ti3+
cations, and (2) hydrogen from formic acid is diverted to
dihydrogen and formaldehyde, as the amounts of both products
formed show significant increase on the defective surfaces.
Regarding the minor products, several reactions have been
proposed for the production of CO2 from formate on TiO2 (110)
surfaces, including:7,37,38
The defects generated by electron irradiation are different
from those generated by Ar+ bombardment in several ways.
First, the TPR spectra for DCOOD decomposition on the
surfaces with similar defect concentration are quite different.
Second, their reoxidizability by formic acid differs significantly.
The differences may result from the local structure around the
Ti3+ cations. On the surfaces irradiated by electron beam, the
defects are believed to be created by removing the bridging
oxygen on the surface, at least at low defect concentration. The
surface structure may exhibit little rearrangement, the defects
assuming a distorted pyramidal coordination structure. On the
Ar+-bombarded surface, there is significant surface reconstruc-
tion, and the surface is believed to be similar to that of Ti2O3,58
which has the structure of corundum with Ti3+ in an octahedral
coordination.
OD-(a) + DCOO-(a) f D2(g) + CO2(g) + O2-
(5)
(a)
DCOO-(a) + DCOOD(g) f CO2(g) + D2(g) + DCOO- (6)
(a)
2DCOO-(a) + Ov f CO2 + D2CO + O2-
(7)
(lattice)
Reactions 5 and 6 were proposed in accord with the assumption
that equivalent amount of D2 and CO2 should be produced from
DCOOD decomposition on the TiO2 (110) surface. However,
experimental results show this assumption to be incorrect, as
the amount of CO2 exceeds that of D2 produced on the
stoichiometric surface. But this fact alone does not rule out the
existence of reactions 5 and 6. Reaction 7 is not likely to occur,
at least as an elementary step, as the CO2 and D2CO desorption
peaks are not concomitant, and the introduction of oxygen
vacancies suppresses CO2 production.
V. Summary
When oxygen anion defects are introduced into the stoichio-
metric TiO2 (110) surface by electron-beam irradiation in
different concentration, new active sites for DCOOD reduction
to CO and formaldehyde are generated which lower the
activation energy and increase the selectivity for the reaction
products. The pattern of reactivity is strongly suggestive of the
participation of formyl as the rate-limiting reactive intermediate.
Ti3+ defects introduced by Ar+ bombardment show a very
different reactivity at the same defect concentration, suggesting
significant differences in the nature of the active sites on the
defective surfaces generated by the different methods.
Other reaction channels could exist, e.g.:
O(lattice) + DCOO(a) f OD(a) + CO2(g)
DCOO(a) f D(a) + CO2(g)
(8)
(9)
OD(a) + DCOO(a) f D2O(g) + CO2(g)
(10)
Acknowledgment. The authors gratefully acknowledge the
support of the National Science Foundation through NSF CTS
0000283.
The inclusion of these reactions can account for the inequiva-
lence of D2 and CO2 in the decomposition products. CO2 desorbs
in two temperature ranges on all the surfaces studied. The two
desorption peaks may originate from two different reactions or
one reaction which proceeds under different local conditions.
It is hard to distinguish these two cases from the present
experiments, as the reaction mechanisms involved are still
unclear. Iwasawa et al.11,12 have reported a different adsorbate
on the bridging oxygen rows which was observed by STM
References and Notes
(1) Chung, Y. W.; Lo, W. J.; Somorjai, G. A. Surf. Sci. 1977, 64, 588.
(2) Rohrer, G. S.; Henrich, V. E.; Bonnell, D. A. Surf. Sci. 1992, 278,
146.
(3) Onishi, H.; Fukui, K.; Iwasawa, Y. Bull. Chem. Soc. Jpn. 1995,
68, 2447.

11720 J. Phys. Chem. B, Vol. 107, No. 42, 2003
Wang et al.
(4) Diebold, U.; Lehman, J.; Mahmound, T.; Kuhn, M.; Leonardelli,
G.; Hebenstreit, W.; Schmid, M.; Varga, P. Surf. Sci. 1998, 411, 137.
(5) Bennett, R. A.; Stone, P.; Price, N. J.; Bowker, M. Phys. ReV. Lett.
1999, 82, 3831.
(6) Epling, W. S.; Peden, C. H. F.; Henderson, M. A.; Diebold, U.
Surf. Sci. 1998, 412/413, 333.
(7) Sherrill, A. B.; Barteau, M. A. Principles of Reactivity from Studies
of Organic Reactions on Model Oxide Surfaces. In Oxide Surfaces, 1st ed.;
Woodruff, D. P., Ed.; Elsevier Science B. V.: Amsterdam, 2001; Vol. 9,
p 409.
(8) Diebold, U. Surf. Sci. Rep. 2003, 48, 53.
(9) Bowker, M.; Stone, P.; Bennett, R.; Perkins, N. Surf. Sci. 2002,
511, 435.
(10) Bennett, R. A.; Stone, P.; Smith, R. D.; Bowker, M. Surf. Sci. 2000,
456-456, 390.
(11) Iwasawa, Y.; Onishi, H.; Fukui, K.; Suzuki, S.; Sasaki, T. Faraday
Discuss. 2000, 114, 259.
(12) Iwasawa, Y.; Onishi, H.; Fukui, K. Top. Catal. 2001, 14, 163.
(13) Sasahara, A.; Uetsuka, H.; Onishi, H. J. Phys. Chem. B 2001, 105,
1.
(14) Gutie´rrez-Sosa, A.; Mart´ınez-Escolano, P.; Raza, H.; Lindsay, R.;
Wincott, P. L.; Thornton, G. Surf. Sci. 2001, 471, 163.
(15) Chang, Z.; Thornton, G. Surf. Sci. 2000, 462, 68.
(16) Ka¨ckell, P.; Terakura, K. Appl. Surf. Sci. 2000, 166, 370.
(17) Ka¨ckell, P.; Terakura, K. Surf. Sci. 2000, 461, 191.
(18) Henrich, V. E.; Cox, P. A. The Surface Science of Metal Oxides;
Cambridge University Press: Cambridge, 1994.
(31) Linsebigler, A.; Lu, G.; Yates, J. T., Jr. J. Chem. Phys. 1995, 103,
9438.
(32) Go¨pel, W.; Rocker, G.; Feierabend, R. Phys. ReV. B 1983, 28, 3427.
(33) Muryn, C. A.; Hardman, P.; Crouch, J. J.; Raiker, G. N.; Thornton,
G.; Law, S. D.-L. Surf. Sci. 1991, 251/252, 747.
(34) Wang, L.-Q.; Baer, D. R.; Engelhard, M. H.; Shultz, A. N. Surf.
Sci. 1995, 344, 237.
(35) Kim, K. S.; Barteau, M. A. Langmuir 1990, 6, 1485.
(36) Farfan-Arribas, E.; Madix, R. J. J. Phys. Chem. B 2002, 106, 10680.
(37) Onishi, H.; Aruga, T.; Iwasawa, Y. J. Catal. 1994, 146, 557.
(38) Henderson, M. A. J. Phys. Chem. B 1997, 101, 221.
(39) Sambi, M.; Sangiovanni, G.; Graznozzi, G.; Parmigiani, F. Phys.
ReV. B 1997, 55, 7850.
(40) Price, N. J.; Reitz, J. B.; Madix, R. J.; Solomon, E. I. J. Electron
Spectrosc. Relat. Phenom. 1999, 98/99, 257.
(41) Go¨pel, W.; Anderson, J. A.; Frankel, D.; Jaehnig, M.; Phillips, K.;
Scha¨fer, J. A.; Rocker, G. Surf. Sci. 1984, 139, 333.
(42) Thevuthasan, S.; Herman, G. S.; Kim, Y. J.; Chambers, S. A.;
Peden, C. H. F.; Wang, Z.; Ynzunza, R. X.; Tober, E. D.; Morais, J.; Fadley,
C. S. Surf. Sci. 1998, 401, 261.
(43) Chambers, S. A.; Henderson, M. A.; Kim, Y. J.; Thevuthasan, S.
Surf. ReV. Lett. 1998, 5, 381.
(44) Iwasawa, Y. Surf. Sci. 1998, 402-404, 8.
(45) Hayden, B. E.; King, A.; Newton, M. A. J. Phys. Chem. B 1999,
103, 203.
(46) Henderson, M. A. J. Phys. Chem. 1995, 99, 15253.
(47) Breault, R.; Hindermann, J. P.; Kiennemann, A.; Laurin, M. Stud.
Surf. Sci. Catal. 1984, 19, 489.
(19) Lu, G.; Linsebigler, A.; Yates, J. T., Jr. J. Phys. Chem. 1994, 98,
11733.
(20) Wang, L.; Baer, D. R.; Engelhard, M. H. Surf. Sci. 1994, 320, 295.
(21) Pe´tigny, S.; Moste´fa-Sba, H.; Domenichini, A.; Lenieswska, E.;
Steinbrunn, A.; Bourgeouis, S. Surf. Sci. 1998, 410, 250.
(22) Pe´tigny, S.; Domenichini, A.; Moste´fa-Sba, H.; Lenieswska, E.;
Steinbrunn, A.; Bourgeouis, S. Appl. Surf. Sci. 1999, 142, 114.
(23) Pan, J. M.; Maschhoff, B. L.; Diebold, U.; Madey, T. E. J. Vac.
Sci. Technol. A 1992, 10, 2470.
(24) Wang, L. Q.; Ferris, K. F.; Shultz, A. N.; Baer, D. R.; Engelhard,
M. H. Surf. Sci. 1997, 380, 352.
(25) Eriksen, S.; Egdell, R. G. Surf. Sci. 1987, 180, 263.
(26) Kim, K. S.; Barteau, M. A. J. Catal. 1990, 125, 353.
(27) Murray, P. W.; Condon, N. G.; Thornton, G. Phys. ReV. B 1995,
51, 10989.
(48) Raupp, G. B.; Dumesic, J. A. J. Phys. Chem. 1986, 90, 3359.
(49) Li, C.; Sakata, Y.; Arai, T.; Domen, K.; Maruya, K.; Onishi, T. J.
Chem. Soc., Faraday Trans. 1 1989, 85, 1451.
(50) Mitchell, W. J.; Wang, Y.; Xie, J.; Weinberg, W. H. J. Am. Chem.
Soc. 1993, 115, 4381.
(51) Vohs, J. M.; Barteau, M. A. Surf. Sci. 1986, 176, 91.
(52) Idriss, H.; Kim, K. S.; Barteau, M. A. Surf. Sci. 1992, 262, 113.
(53) Henderson, M. A.; Epling, W. S.; Perkins, C. L.; Peden, C. H. F.;
Diebold, U. J. Phys. Chem. B 1999, 103, 5328.
(54) Yamaguchi, Y.; Onishi, H.; Iwasawa, Y. J. Chem. Soc., Faraday
Trans. 1995, 91, 1663.
(55) Knotek, M. L. Surf. Sci. Lett. 1980, 91, 17.
(56) Knotek, M. L. Surf. Sci. 1980, 101, 334.
(57) Suzuki, S.; Fukui, K.; Onishi, H.; Iwasawa, Y. Phys. ReV. Lett.
2000, 84, 2156.
(28) Cocks, I. D.; Guo, Q.; Williams, E. M. Surf. Sci. 1997, 390, 119.
(29) Idriss, H.; Lusvardi, V. S.; Barteau, M. A. Surf. Sci. 1996, 348,
39.
(30) Wang, L.-Q.; Ferris, K. F.; Skiba, P. X.; Shultz, A. N.; Baer, D.
R.; Engelhard, M. H. Surf. Sci. 1999, 440, 60.
(58) Henrich, V. E.; Dresselhaus, G.; Zeiger, H. J. Phys. ReV. Lett. 1976,
36, 1335.