The interaction of water with silica thin films grown on mo (1 1 2)

Article abstract of DOI:10.1016/j.susc.2004.06.213

The adsorption of water on ultrathin SiO₂ films at low temperatures has been studied with metastable impact electron spectroscopy (MIES) and ultraviolet photoelectron spectroscopy (UPS (HeI)). High-resolution electron energy-loss spectroscopy (

Full text of DOI:10.1016/j.susc.2004.06.213

Surface Science 565 (2004) 107–120
www.elsevier.com/locate/susc
The interaction of water with silica thin films grown
on Mo(112)
a
S. Wendt , M. Frerichs , T. Wei , M.S. Chen , V. Kempter ,
a,b
a
a
b
a,
*
D.W. Goodman
a
Department of Chemistry, Texas A&M University, College Station, TX 77842-3012, USA
Institut f u¨ r Physik und Physikalische Technologien, Technische Universit a¨ t Clausthal, Leibnitzstrasse 4,
D-38678 Clausthal-Zellerfeld, Germany
b
Received 31 March 2004; accepted for publication 11 June 2004
Available online 29 July 2004
Abstract
The adsorption of water on ultrathin SiO
2
films at low temperatures has been studied with metastable impact elec-
tron spectroscopy (MIES) and ultraviolet photoelectron spectroscopy (UPS (HeI)). High-resolution electron energy-
loss spectroscopy (HREELS), work function measurements (Du), and temperature programmed desorption (TPD)
were also utilized to study the interaction of water with silica. Evidence for molecular absorption of water on low-
and high-defect silica surfaces is presented. The data are consistent with the growth of 3-D water clusters even at
low coverage, i.e., the water–water hydrogen bonding is stronger than the water–silica interaction. No evidence for dis-
sociation of water was found in contrast to previous UPS results.
ꢀ
2004 Elsevier B.V. All rights reserved.
Keywords: Water; Metastable induced electron spectroscopy (MIES); Visible and ultraviolet photoelectron spectroscopy; Electron
energy loss spectroscopy (EELS); Thermal desorption
1
. Introduction
therefore is vitally important to disciplines from
corrosion to geochemistry. Furthermore, water
plays a key role in electrochemistry, catalysis and
electronic devices. Therefore, a molecular-level
understanding of the adsorption and reaction of
water on solid surfaces is a goal of modern surface
science [1,2]. An important general question is
whether water adsorbs molecularly or whether it
dissociates. Many studies during the last 25 years
Water is the most abundant compound in our
environment, covers most solid surfaces, and
*
Corresponding author. Tel.: +1 979 845 5151; fax: +1 979
8
45 6822.
E-mail address: goodman@mail.chem.tamu.edu (D.W.
Goodman).
0
039-6028/$ - see front matter ꢀ 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2004.06.213

108
S. Wendt et al. / Surface Science 565 (2004) 107–120
have addressed the adsorption and reaction of
water on single crystalline metal surfaces. For
Ru(0001) the adsorptive behavior of water is in
dispute [3] even though this system has been exten-
sively investigated [1,2].
than ideal. Only recently, well-ordered SiO thin
2
films with a thickness of 0.5–0.8 nm (one to two
monolayers) have been successfully prepared on
Mo(112) [16,17]. Based on HREELS data, the
ultrathin SiO thin film can be characterized as a
2
The adsorption of water on metal oxide sur-
faces has also received considerable attention
Mo(112)-c(2 · 2)-[SiO ] monolayer where all four
oxygen atoms bind directly to Mo substrate atoms
[18]. Moreover, it has been shown with STM that
4
[
1,2]. The investigation of oxide surfaces is more
complex than metals since most metal oxides are
not conducting and the preparation of well-defined
surfaces is often problematic. Generally it has been
assumed that water preferentially dissociates
on defect sites of oxide surfaces [1]. For TiO2-
ultrathin SiO thin films grown on Mo(112) are
2
remarkable flat and homogeneous [19].
Previously, the adsorption state of water has
been studied on amorphous films prepared by oxi-
dation of Si(100) and Si(111) surfaces, respec-
tively [20,21]. Using infrared reflection–adsorption
spectroscopy (IRAS), Chabal and Christman
found evidence for molecular physisorbed water
at 80 K on Si(100) covered with its native oxide
[20]. In contrast, Nishijima et al., based on EELS
measurements, proposed that water partially dis-
(
110)––the most extensive studied oxide sur-
face––earlier studies report a small amount of
water dissociation supporting the view that point
defects are the active sites for water dissociation
[
4,5]. This view was recently strengthened by scan-
ning tunneling microscopy (STM) and UPS exper-
iments together with density functional theory
sociates on a thin SiO layer over Si(111) at 300
x
(
DFT) calculations [6,7]. For MgO(100), however,
K [21]. Several theoretical studies have focused
on the interaction of water with a-quartz [22,23]
and with amorphous SiO [24,25]. Thermodynam-
several theoretical studies recently have proposed
that a mixed (water + hydroxyl) layer is energeti-
cally more stable than pure water layers [8–11].
To probe this important question, whether water
partly dissociates on ideal oxide surfaces, addi-
tional studies are required on well-ordered oxide
surfaces with very low-defect concentrations.
Because of its extensive technological impor-
tance numerous studies have been directed toward
2
ically the dissociation of water on a-quartz and on
amorphous SiO is favored. However, dissociation
of water may be inhibited on SiO since dehy-
2
2
drated silica (gel, powder or silica glass fracture
surfaces) is not readily hydrolyzed when exposed
to air or water [26–28]. It is known that the reacti-
vity of silica surfaces strongly depends on the pres-
ence of hydroxyl (silanol [29]) groups [30–32]. Sneh
and George, for instance, investigated the reactiv-
ity of 5 lm thick silica films as a function of hyd-
roxyl coverage. It is noteworthy that in this study
the study of silica (SiO ). To a large extent these
2
studies have focused on high-surface-area silicas
such as powders or silica gels, although to a lesser
extent some studies have addressed the initial film
growth of amorphous SiO on Si surfaces and on
2
a H O plasma discharge was used to produce a
2
the SiO /Si interface. However, only a few studies
2
hydroxylated SiO2 surface [32]. To explain the
hydrophobic or hydrophilic properties of vitreous
silica surfaces, the role of defect sites has been
addressed in several theoretical studies [25,33,34].
However, many questions remain regarding the
interaction of water with silica surfaces.
have used ultrahigh vacuum (UHV) methodologies
to address the interaction of adsorbates with well-
defined silica surfaces. The insulating properties
of oxides limit the use of charged particle surface
analytical probes because of sample charging. Such
difficulties can be circumvented by using thin films
grown on refractory metal substrates [12]––a meth-
odology used successfully for a number of insulat-
ing metal oxides including MgO, TiO and Al O
3
In this paper we present MIES/UPS, HREELS,
Du, and TPD results addressing the interaction of
water with ultrathin, SiO surfaces in UHV. Evi-
2
dence is provided for molecular adsorption of
water on various SiO2 thin films grown on
Mo(112). The data can be explained best by 3-D
growth of water clusters even at low coverage.
2
2
[
13–15]. Silica films have been prepared, and
although stoichiometric, these films were amor-
phous rather than single crystalline and thus less

S. Wendt et al. / Surface Science 565 (2004) 107–120
109
Under the conditions used in these experiments no
evidence for the dissociation of water is found. This
conclusion conflicts with a previous report based
on MIES/UPS data [35]. In Ref. [35] it was con-
cluded that hydroxyl groups form at the silica/
water interface in the presence of multilayer water.
However, the results upon which this conclusion
was reached are in error due to an artifact in the
method used to separate the MIES–UPS data.
the MIES signal arises. In this case, a plot of the
intensity of the ejected electrons versus their ki-
netic energies yields the surface density of states
(SDOS) for the topmost layer of the surface [37].
In the spectra reported here, all binding energies
are referenced to the Fermi level of the Mo sub-
strate. A Mo(112) crystal, used as the substrate
for the SiO thin films, was spot-welded on an u-
2
shaped tantalum wire allowing resistive heating.
The temperature of the sample could be varied
from 90 K (by cooling with liquid N ) to 2200 K
2
2
. Experimental
(by e-beam heating). The sample temperature
was measured by a (W–5%Re/W–26%Re) thermo-
couple spot welded to the back side of the sample.
The Mo(112) sample was cleaned by multiple
flashes to 2100 K, and the cleanliness verified with
AES. The clean Mo(112) surface showed a rectan-
gular LEED pattern with intense spots and low
background intensity.
The experiments were performed in two turbo-
pumped UHV chambers. One chamber is equipped
with Auger electron spectroscopy (AES), high-
resolution electron energy-loss spectroscopy
(
HREELS, LK-2000), electron optics for low en-
ergy electron diffraction (LEED), and a quadru-
pole mass spectrometer (QMS) for temperature
programmed desorption (TPD). The primary elec-
tron beam in the HREELS study was 4 eV and the
angle of incidence 60ꢁ with respect to the surface
normal of the sample. Under the chosen conditions
the spectral resolution of the instrument was 64–96
cm (8–12 meV). The second chamber is equipped
with a cold-cathode gas discharge source produc-
ing both ultraviolet photons (He I: E = 21.2 eV)
and metastable He 2 S (E = 19.8 eV) atoms with
thermal kinetic energy [36]. The axis of the cylindri-
cal mirror analyzer (CMA) was perpendicular to
the incoming photon/metastable beam. MIES and
UPS spectra were acquired with photon/metastable
beams incident at 48ꢁ with respect to the surface
normal. The MIES/UPS spectra were measured
simultaneously using a mechanical chopper (time-
of-flight method) [36]. Auger spectra were acquired
using an electron gun integral to the CMA. The res-
olution of the analyzer, estimated from the width of
the Fermi edge, is ꢀ0.3 eV. The work functions
were estimated from the low-energy onset of the
secondary electrons in the UPS spectra. This cham-
ber was also equipped with LEED optics and TPD.
The QMS was differentially pumped with an 8 mm
aperture facing the sample at a distance of 1–2 mm.
For wide bandgap materials or metals with
work functions lower than ꢀ3.5 eV, Auger-deexci-
tation (AD) is the dominant mechanism by which
The preparation of the SiO2 thin films on
Mo(112) was carried out as follows: (i) prior to
À8
Si deposition oxygen was dosed at 5 · 10 Torr
with the Mo(112) substrate at 850 K for ꢀ10
min. Following this procedure, a p(2 · 3)O surface
was obtained [38]; (ii) after deposition of Si at
room temperature, the surface was heated to 800
À1
À7
K in an oxygen atmosphere (p = 1 · 10 Torr);
*
and (iii) the SiO films were further annealed at
the same oxygen pressure at 1050 and 1150–1200
K, respectively. The AES spectra of the SiO thin
2
3
*
2
0
4
+
films showed a Si feature at 76 eV and no Si fea-
ture at ꢀ90 eV, indicating that the silicon surface
was completely oxidized. Attenuation of the Mo
MNN (187 eV) AES feature was used to compute
the film thickness. Assuming a mean free path of
0.95 nm for 187 eV electrons in SiO [39], the
2
thickness of the SiO thin films is estimated to be
2
0.5–0.7 nm for an anneal at 1050 K, and 0.3–0.5
nm for an anneal at 1150–1200 K. LEED showed
a sharp c(2 · 2) structure, indicating the formation
of well-ordered surface [16,18]. MIES and UPS
spectra for low- and high-defect SiO surfaces have
2
been described previously [40]. Briefly, the low-de-
fect SiO surface produced by an anneal at 1150 K
2
in oxygen shows in MIES two O(2p) features at 6–
9 eV, and features at 10–12 eV related to the Si–O
bond (lowest spectrum in Fig. 1A). In UPS, the
O(2p) peaks of SiO are evident at 6 and 8 eV
2

110
S. Wendt et al. / Surface Science 565 (2004) 107–120
W3
W3
MIES spectra
UPS spectra
D O /SiO /Mo(112)
D O /SiO /Mo(112)
2
2
2
2
W2
T = 90 K;
W1
0.35 L per
W2
spectrum
EF
W1
EF
Si-O
18 16 14 12 10
Si-O O(2p)
8 16 14 12 10
O(2p)
1
8
6
4
2
0
8
6
4
2
0
(
A)
binding energy (eV)
(B)
binding energy (eV)
W3
5.25 L D O / SiO
2 2
UPS
MIES
W2
W1
EF
0
18 16 14 12 10
8
6
4
2
(
C)
binding energy (eV)
Fig. 1. (A) MIES spectra and (B) UPS (He I) spectra collected from a low-defect SiO
2
/Mo(112) surface as a function of water
exposure at 90 K. The MIES and UPS spectra were acquired simultaneously. The shift toward higher binding energies as a function of
water exposure (charging) has been removed from all spectra. Enlarged region of the UPS spectra near at the Fermi edge was used to
estimate the degree of shifting. (C) Direct comparison of the MIES and UPS spectra after an exposure of 5.25 L water corresponding to
multilayer coverage.
and the Si–O bonding states at 10–13 eV (lowest
spectrum in Fig. 1B). Shapes of the MIES and
UPS spectra are different for oxidized SiO2/
Mo(112) surfaces which were annealed at only
study. Furthermore, no indication for the forma-
tion of any Mo silicide was found under the condi-
tions used to synthesize the SiO thin films on
2
Mo(112).
1
050 K. In this case, the O(2p) features at 6–9
eV in MIES and UPS are not resolved, consistent
with a higher density of extended defects in com-
parison to the SiO thin films which were annealed
3. Water adsorption and desorption experiments on
various SiO surfaces
2
2
at 1150–1200 K [40,41]. The SiO thin films have a
2
low density of point defects regardless of the
annealing temperature. Point defects should yield
states in the bandgap region [15]. No such states
3.1. Low-defect SiO surfaces
2
Fig. 1A shows MIES spectra collected as a
function of water exposure to a low-defect SiO2/
were detected for the SiO thin films used in this
2

S. Wendt et al. / Surface Science 565 (2004) 107–120
111
Mo(112) of 0.3–0.5 nm thickness at 90 K. The sur-
face was exposed successively to 0.35 L water for
of the 3a MO (partly bonding and partly non-
1
bonding (lone pair)), and the W3 feature at
ꢀ13.5 eV is due to deexcitation of the 1b MO
each spectrum (L = Langmuir, with 1 L = 1 ·
2
À6
1
0
Torrs). Upon water exposure, the substrate
(bonding orbital). The 3a /1b separation for the
1
1
signal attenuates and concomitantly three new
peaks appear, labeled as W , W and W . In
molecular species is 2.35 eV, very similar to the sep-
aration for gas-phase H O of 2.2 eV [42]. The MIES
1
2
3
2
UPS (Fig. 1B) the same trend is evident, i.e. dimi-
nution of the substrate features (including Mo fea-
tures between 0 and 5 eV) and appearance of three
water-induced features. The positions of the fea-
tures W and W in MIES and UPS are very sim-
spectra in Fig. 1 are consistent with molecular
adsorption of water on ultrathin, low-defect SiO2
films on Mo(112).
In Fig. 2 the MIES intensities corresponding to
the molecular orbitals 1b , 3a and 1b are shown
1
3
1
1
2
ilar, suggesting that these features have the same
origin. In contrast, the W feature with the highest
as a function of water exposure at 90 K. The silica
surface is completely covered after a 3 L water
exposure since the intensities of the 1b and 3a
2
intensity occurs at different energies in MIES and
UPS. A direct comparison of the MIES and UPS
is shown in Fig. 1C acquired after a 5.25 L water
exposure. At this exposure the three water-induced
features are fully developed (Fig. 2). In MIES a
1
1
features are maximized at this exposure. Further-
more, no contributions are apparent from the
silica substrate in MIES. That the 1b induced
2
feature does not maximize is related to the increas-
ing number of secondary electrons from the addi-
tional water adsorption. It is noteworthy that
upon water exposure up to 10 L the c(2 · 2) LEED
broad W feature is visible at 9.6 eV, and a shoul-
2
der at 11.3 eV is discernible. The shoulder in MIES
roughly coincides with the maximum of W2 in
UPS. Nevertheless, the MIES spectrum in Fig.
pattern of the SiO /Mo(112) surface was still
2
1
C resembles the gas-phase photoemission spec-
visible. No new water-induced LEED spots
were evident, consistent with there being no long-
range order in the water-covered SiO /Mo(112)
trum of molecular water [42], allowing identifica-
tion of the characteristic features W , W and
W . The W feature at 7.25 eV is due to deexcita-
1
2
2
surface.
3
1
tion of the 1b lone pair molecular orbital (MO).
1
In Fig. 3 MIES and UPS difference spectra are
shown for the data of Fig. 1. Prior to subtraction
of the SiO -spectrum from the D O/SiO -spec-
The W feature at 9.6 eV is due to deexcitation
2
2
2
2
trum, the SiO -spectra were attenuated taking into
2
account the D O-induced damping of the SiO -
2
2
features. In addition, the background due to sec-
ondary electrons was subtracted from the UPS
spectra. Since the number of secondary electrons
in MIES is substantially smaller than in UPS, a
background subtraction was not performed for
the MIES difference spectra. Therefore, there is a
sharp feature due to the secondary electrons at
MIES-intensities
1
b2
a1
1b1
15 eV in the MIES spectra that is absent in the
3
UPS difference spectra. In the MIES data, the
positions of the water features are identical to
those in the spectra of Fig. 1. On the other hand,
0
1
2
3
4
5
6
in the UPS data the W feature shifts from 9.4
2
D O exposure (L)
2
eV at low water coverage to 10.4 eV at high cover-
age, whereas the W and W features are invariant.
The shifting of the W feature in the UPS differ-
ence spectra is not readily apparent in the UPS
spectra of Fig. 1B. At low coverage (corresponding
1
3
Fig. 2. Intensities of the water induced features in the MIES
spectra from Fig. 1 as a function of water exposure. The
orbitals of molecular water, which causes the features seen in
MIES, are noted at the curves.
2

112
S. Wendt et al. / Surface Science 565 (2004) 107–120
W3
W3
MIES difference spectra
UPS difference spectra
D O / SiO / Mo(112)
D O / SiO / Mo(112)
2
2
2
2
W2
W1
W2
W1
EF
0
EF
0
1
8 16 14 12 10
8
6
4
2
18 16 14 12 10
8
6
4
2
(
A)
binding energy (eV)
(B)
binding energy (eV)
Fig. 3. (A) MIES difference spectra and (B) UPS difference spectra of the data shown in Fig. 1. Before subtracting the SiO
from the D O/SiO -spectrum, the SiO -spectrum was attenuated taking into account the D O-induced damping of the SiO -features. In
addition, the background was subtracted from the UPS spectra using an exponential approach.
2
-spectrum
2
2
2
2
2
to a water exposure of 0.35 L) all water-induced
features appear in the MIES and UPS difference
spectra at very similar energy positions.
Fig. 4A shows water TPD spectra for a low-
defect SiO2 thin film after water adsorption at
water on solid surfaces [1,2]. TPD therefore cannot
be used for coverage determination. Even for
higher temperatures than those shown in Fig.
5A, no additional features are evident. The activa-
tion energy for desorption was estimated using a
simple exponential relation for the rate of desorp-
tion (Fig. 4B). The TPD trace corresponding to 5
L was used for the Arrhenius-plot in Fig. 4B. It
is noteworthy that 63.9 ± 2 kJ/mol is higher than
9
0 K. For a water exposure of 0.1 L, a small peak
at 160 K is evident. For a higher water exposure
the maximum of the peak shifts to a higher tem-
perature consistent with zero-order desorption
kinetics. There is no discernible transition from
the monolayer to the multilayer typically seen for
the heat of sublimation of D O-ice estimated to
2
be as high as 44–50 kJ/mol [1,2].
temperature (K)
1
72
168
164
160
156
152
m/e = 20
heating
rate 4.5 K/s
8
6
4
2
5
1
0
L
L
.1 L
D2O / SiO2/ Mo(112)
a
des
E
= (63.9 ±2) kJ/mol
5
.8
6.0
6.2
6.4
6.6
150
160
170
180
190
-
1
(
A)
temperature (K)
(B)
1000 / T ( K )
Fig. 4. (A) Water TPD spectra of a low-defect SiO
desorption rate (logarithmic) versus the reciprocal temperature for the TPD data taken after exposure of 5 L water. Such a water
exposure corresponds to multilayers of water on SiO /Mo(112).
2
/Mo(112) surface for 0.1, 1.0 and 5 L water exposure at 90 K. (B) Plot of the
2

S. Wendt et al. / Surface Science 565 (2004) 107–120
113
data for various substrates (see for example Refs.
44–47]).
-1
1
048 cm
H2O / SiO2/ Mo(112) at 90 K
[
8
20 cm-1
1650 cm-1 3360 cm-1
3
690 cm-1
3.2. SiO surfaces with extended defects
2
a)
b)
c)
x40
x 300
x 300
x 80
x 15
x 15
x 15
Because defect sites likely play a decisive role in
the dissociation of water, MIES/UPS adsorption
experiments were also performed on thicker silica
films which exhibit broad O(2p) features in MIES
and UPS at 6–9 and 5–9 eV, respectively (see the
lowest spectrum in Fig. 7A and B). The broad
O(2p) features are interpreted to be indicative of
extended defects, e.g., steps, corners, etc. [40].
Broad O(2p) features are typical for amorphous
silica films [41,48], consistent with the presence of
extended defects. Recent STM and SPALEED re-
sults of Schroeder et al., describe the defect struc-
x 1500
x 1500
d)
e)
x 300
x 800
x 1500
x 2000
x100
x1000
f)
x 4000
x 2000
0
1000
2000
3000
-1
4000
electron energy loss (cm )
Fig. 5. HREELS spectra collected from a low-defect SiO
Mo(112) surface as a function of water exposure at 90 K: (a)
bare SiO surface, (b) after exposing to 0.5 L, (c) 1.0 L, (d)
.0 L, (e) 5.0 L and (f) 10 L H O. The sensitivity of the ion
2
/
2
2
2
ture of the SiO /Mo(112) surface as steps and
2
gauge in the HREELS chamber is comparable to that in the
UHV chamber in which the MIES/UPS measurements were
performed.
antiphase domain boundaries [17]. These results
are consistent with the interpretation of the MIES
and UPS spectra, observed for SiO films which
2
were annealed at different temperatures [40]. In
this respect it is noteworthy that the defect density
is a strong function of the film thickness. Narrow
O(2p) features indicative of a flat surface with a
very low density of extended defects were only
In Fig. 5 HREELS spectra are shown collected
as a function of water exposure at 90 K on a low-
defect SiO film of 0.3–0.5 nm thickness. Spectrum
2
(
has been discussed previously [18]. The spectrum is
a) is that from the bare SiO /Mo(112) surface and
found for ultrathin SiO
broad O(2p) features, i.e. a high density of ex-
tended defects, were found for 0.6 nm thick and
2
films (ꢀ0.4 nm), whereas
2
À1
dominated by a loss at 1048 cm , assigned to the
asymmetric Si–O stretch mode. Further losses ap-
thicker SiO
film we performed the same adsorption experiment
as for a low-defect SiO film presented in Fig. 1.
The data for water exposures up to 6 L at 90 K
to a SiO film with extended defects (not shown)
are very similar to those for the low-defect SiO
films [41]. For a ꢀ0.6 nm thick SiO
2
2
À1
pear at 2112 cm (overtone of the asymmetric Si–
O stretch mode) as well as doublets at 630–830 and
2
À1
1
acquired for a SiO (0001) single crystal [43], the
680–1870 cm . Comparing this spectrum to that
2
2
À1
loss at 1048 cm is red-shifted by approximately
À1
2
1
26 cm . This shift has been explained by the
thin film. Even at low water coverages only fea-
tures originating from molecularly adsorbed water
molecules are apparent. This is evident in Fig. 6A
where the difference spectra for a water exposure
of 0.35 L of the two films are shown. For an easier
comparison of the spectra the background was
subtracted from the UPS spectra. A substantially
dominance of Si–O–Mo linkages over Si–O–Si
linkages in the one ML thin film [18]. Curves b,
c, d, e and f are for exposures of 0.5, 1.0, 2.0, 5.0
and 10 L, respectively. With increasing water
exposure the substrate losses attenuate and fea-
À1
tures at 3360 cm appear (spectra b–f). For high
water exposures the spectral intensity dramatically
decreases—thus complicating the interpretation of
the spectra. Nevertheless, for an exposure of 10 L
larger broadening of the 3a
with the 1b and 1b features can be seen from the
difference spectra. The lower intensity of the UPS
difference spectrum of the film with extended de-
fects (curve b) compared to the low-defect film
(curve a) is due to the slightly different pressure
feature in comparison
1
1
2
H O (spectrum f) new features are evident at 3690,
2
À1
1650, and at 820 cm . These losses are assigned to
condensed water considering previous HREELS

114
S. Wendt et al. / Surface Science 565 (2004) 107–120
UPS
0
.35 L D O on SiO at 90 K
2
2
0
.0
0.1
-0.2
MIES
D O/SiO /Mo(112) at 90 K
1
b2
3a1
1b1
2
2
-
extended
EF
b)
defects
-
-
-
0.3
0.4
0.5
a) low defect surface
low-defect
b) extended defects
a)
8
surfaces
1
16 14 12 10
8
6
4
2
0
0
1
2
3
4
5
binding energy (eV)
(
A)
(B)
D O exposure (L)
2
Fig. 6. (A) Initial water adsorption (0.35 L D
2 2 2
O) at 90 K on two differently prepared SiO thin films: (a) Low-defect SiO film
produced by deposition of Si at room temperature followed by a 1150 K anneal; (b) SiO
2
film prepared by annealing at 1050 K with a
high density of extended defects at the surface. The MIES and UPS difference spectra are compared. The background was subtracted
from the UPS spectra using an exponential approach. (B) Work function change Du upon water exposure at 90 K for the two
differently prepared SiO thin films on Mo(112): (a) Low-defect SiO film and (b) SiO film with extended defects.
2
2
2
conditions used in the cold-cathode He discharge
source for these measurements.
data set shown in Fig. 1B and curve (b) from the
second UPS data set where a ꢀ0.6 nm thick SiO
2
The work function changes Du upon water
film was used. For both SiO thin films the work
2
adsorption on the different SiO films are displayed
in Fig. 6B. Curve (a) was extracted from the UPS
function change upon water adsorption is nega-
tive. For the low-defect, ultrathin SiO film the
2
2
1
b2
1
b2
MIES
UPS
1
b1
6 LD O on SiO
2
2
3
a1
at 90 K and after
annealing
3
a1
1
b1
EF
EFF
9
0 K
90 K
1
60 K
63 K
66 K
69 K
160 K
1
63 K
1
1
1
O(2p)
8
Si-O
166 K
1
69 K
2
O(2p)
Si-O
18 16 14 12 10
binding energy (eV)
1
8
16 14 12 10
6
4
2
0
8
6
4
0
(
A)
binding energy (eV)
(B)
Fig. 7. (A) MIES spectra and (B) UPS spectra collected from a SiO
2
/Mo(112) surface initially covered with 6 L water at 90 K as a
function of the anneal temperature. The SiO
extended defects.
2
surface was prepared by annealing in oxygen at 1050 K and posses a high density of

S. Wendt et al. / Surface Science 565 (2004) 107–120
115
work function decreases gradually then remains
constant after ꢀ2 L exposure of water. The work
and hydroxyl groups appear in the same energy
range it is generally difficult to exclude water disso-
ciation completely. Therefore, it is useful to esti-
mate the upper limit of detectable hydroxyl
groups using MIES and UPS. To estimate this
limit we performed a 3 curve data-fit using Gauss
curves to reproduce the MIES difference spectra of
Fig. 6A in the range from 5 to 11.5 eV (Fig. 8). The
three curves represent the broad 3a , the 1b of
function of a ꢀ0.6 nm thick SiO film with a high
2
density of extended defects decreases more slowly
than that of a low-defect SiO film. Furthermore,
2
the work function remains constant only after a
3
.5 L exposure of water. For a 0.6 nm thick SiO2
film with a high density of extended defects, the
work function change is 0.5 eV, approximately
1
1
0
low-defect SiO thin film.
.1 eV greater than for water adsorption on a
molecular water, and the 1p orbital feature of
the hydroxyl species. The 1p orbital feature typi-
cally appears at ꢀ1 eV lower binding energy than
2
Fig. 7 shows MIES (A) and UPS spectra (B)
from a water-covered SiO film with a high density
the 1b feature of molecular water, and the 3r fea-
2
1
of extended defects as a function of the annealing
temperature. The upper spectra are those of the
water-covered surface at 90 K after exposure of 6
L water. This exposure corresponds to multilayer
of water evident when compared with Fig. 1C.
After a brief anneal at 160 K, the spectra still
resemble multilayer water. After a second brief an-
neal at 163 K, significant changes occur related to
the onset of multilayer sublimation. Annealing to
even higher temperatures indicated in Fig. 7, the
spectra more resemble the MIES/UPS spectra of
ture (not considered here because of the lack of
intensity of the 1p feature) appears at ꢀ4 eV high-
er binding energy (see for comparison Refs.
[1,2,49,50]). Hence, for a mixed (water + hydroxyl)
layer a shoulder in the low-binding energy side of
the 1b feature is expected. In the present case
1
we did not detect such a shoulder even for the low-
est water exposure. In spite of this fact, a small
fraction of dissociated water molecules can be
overlooked as one can see from Fig. 8. For the
low-defect SiO surface we estimate an upper limit
2
a bare SiO film with a high density of extended
2
of 9% of detectable hydroxyl groups using MIES.
For the surface with extended defects the limit is
only 3%. To estimate these values it was assumed
defects. In each case all detected features in the
spectra can be assigned to the electronic structure
of molecular water or SiO . Note that at the low
that the FWHM of 1b feature is 1.3–1.4 eV and
1
2
energy edge of the MIES spectra, where the 1p fea-
ture of hydroxyl was apparent in the MIES spectra
for water on other oxide surfaces [49,50] (see
arrow), no feature is evident.
the (hypothetical) 1p feature is at 0.9 eV lower
binding energy than the 1b feature. Furthermore
1
it was assumed that the deexcitation cross sections
of water and hydroxyl are comparable. For UPS
the limits are likely similar. Our assumptions are
conservative. This is especially true for the impor-
tant assumption regarding the FWHM of the 1b1
feature.
4
. Discussion
First we focus on the results obtained with var-
ious techniques. Subsequently, we discuss our
main conclusions with respect to previous work.
Our result that there is no apparent feature in
the MIES and UPS spectra that can be assigned
to hydroxyl groups at the water/SiO interface is
2
at variance to the results by Kim et al., where
multilayer-induced dissociation of water on SiO2/
Mo(112) at 90 K was reported [35]. Kim et al.,
reached this conclusion based on data that showed
four, rather than three, features in the UPS spec-
trum following water adsorption [35]. The MIES
data in Ref. [35] are in good agreement with
the data presented here. The discrepancy in the
UPS data of Ref. [35] is believed to be due to a
4
.1. MIES and UPS data
We start our discussion with the MIES and
UPS difference spectra for small coverages (Fig.
A). The apparent features can be assigned to
6
molecularly adsorbed water. However, since fea-
tures in UPS and MIES are rather broad, and
since additionally features of molecular water

116
S. Wendt et al. / Surface Science 565 (2004) 107–120
difference spectrum
MIES
difference spectrum
curve data-fit
MIES
3
curve data-fit
3
low-defect SiO film
2
extended defects
difference spectrum
data-fit 3 curve
difference spectrum
data-fit 3 curve
0
.35 L D O
0.35 L D O
2
2
12
11 10
9
8
7
6
5
12 11 10
9
8
7
6
5
(
A)
(B)
binding energy (eV)
binding energy (eV)
Fig. 8. (A) Data-fit to the MIES difference spectrum for initial water adsorption (0.35 L D
2
O) at 90 K on low-defect SiO
2
: single curves
O)
2
surface possessing extended defects. In all panels the MIES difference spectrum (5–11.5 eV) is shown for
(lower panel), and 3 curve data-fit (upper panel). (B) Data-fit to the MIES difference spectrum for initial water adsorption (0.35 L D
2
at 90 K on the SiO
comparison.
MIES contribution to the UPS spectra. One of the
four features in the UPS data, located at identical
kinetic energy as the 1b feature in MIES, was as-
film has more step sites than the 0.4 nm thin film,
the shape of the difference spectra for relative low
water exposures may be different for these two
surfaces.
1
signed to correspond to the 1b orbital of molecu-
1
larly adsorbed water. The actual 1b feature at a
1
The shift of 3a feature (W ) in the UPS differ-
1
2
binding energy of ꢀ7 eV in the UPS spectra was
erroneously assigned to the 1-p orbital of hydroxyl
groups.
ence spectra as a function of exposure (Fig. 3B) is
likewise consistent with the formation of hydro-
gen-bonded water clusters. It is noteworthy that
the peak positions of the other features do not
The substantially larger broadening of the 3a1
feature in comparison with the 1b and 1b fea-
shift. The 3a feature in the UPS (He I) begins to
1
1
2
tures (Figs. 1, 3 and 6A) is caused by hydrogen
bonds between neighboring water molecules, i.e.,
clustering of water molecules [51,52]. The 3a fea-
ture is more strongly influenced when forming a
hydrogen network since hydrogen orbitals as well
as the oxygen lone pairs contribute to the 3a1
molecular orbital of water (partly bonding and
shift after the lowest water exposure of 0.35 L
and continues to shift monotonically with further
exposure. This finding indicates that clustering or
multilayer growth occurs even at low coverages,
and is consistent with facile diffusion of water on
1
SiO /Mo(112) at 90 K. It is also noted that a shift
2
of the 3a feature in the UPS (He I) difference spec-
1
partly non-bonding), whereas the 1b orbital is
1
tra as a function of water coverage has been found
for other oxide surfaces [53].
purely non-bonding and the 1b orbital is a purely
2
bonding orbital [1]. For small coverages (Fig. 6A),
the network of hydrogen-bonded water molecules
on the low-defect silica surface (a) is more exten-
sive than on the silica surface with extended de-
fects (b). A possible explanation for this is the
initial adsorption of water molecules along step
sites whereby the forming of a hydrogen network
is partly obstructed. Since the 0.6 nm thick SiO2
The importance of hydrogen bonds for explain-
ing the spectra is further confirmed considering the
different energy positions of the 3a feature (W ) in
1
2
MIES and UPS for water multilayers (Fig. 1C).
The 3a feature appears in UPS (He I) at ꢀ1 eV
1
higher binding energy than in MIES. Similar peak
positions were found in previous water adsorption
studies that are summarized in Table 1. This indi-

S. Wendt et al. / Surface Science 565 (2004) 107–120
117
Table 1
Energy positions of the three valence orbitals for water ice as observed with MIES and UPS
System
Technique
Molecular orbitals
Reference
1
b
1
3a
1
1b
2
D
2
2
O/SiO
2
/Mo(112)
MIES
UPS (He I)
7.25
7.05
9.6
10.5
13.5
13.5
This work
[49]
H
O/TiO
2
(110)
MIES
UPS (He I)
UPS (He II)
7.8
7.9
7.8
10.0
11.2
9.8
13.2
13.5
13.2
D
2
O/MgO(100)
MIES
UPS (He I)
7.2
7.0
9.2
10.0
13.6
13.6
[50]
Binding energies are in eV with respect to the Fermi level. In all cases the sample temperature during water exposure was 90–100 K.
cates that the relative peak positions of the 3a or-
1
supported by the UPS measurements of water
multilayers at various photon energies carried
out by Schmeisser et al. (Ref. [51], Fig. 4). Using
photons of 21 eV (relatively less surface sensitive),
bital in MIES and UPS (He I) are related to water
multilayers in general regardless of the substrate.
The different peak positions can be explained by
considering the difference in surface sensitivities
of MIES and UPS (He I) with respect to hydrogen
bonding in water clusters (ice). As observed previ-
the broad 3a feature maximizes at relatively high
1
binding energy. However, at photon energies >30
eV (more surface sensitive), maxima of the 3a fea-
ture appear at lower binding energy values.
1
ously, the broad 3a feature in UPS consists of at
1
least two contributions [51]. Schmeisser et al. [51],
explained a splitting of the 3a orbital up to ꢀ2 eV
4.2. TPD data
1
using a model of flat lying linear water dimers in
conjunction with calculations [54], taking into ac-
count hydrogen bonding. In the linear water di-
mers one water molecule acts as proton-donor
and the other as proton acceptor. For multilayer
The TPD data are consistent with molecular
desorption from clustered water whereby the
water–water hydrogen bonding is stronger than
the water–surface interaction. Typical for molecu-
lar desorption from clustered water is pseudo-
zeroth-order behavior in the monolayer range
(Ref. [2], p. 39). Furthermore, no separation be-
tween TPD features stemming from monolayer
and multilayer coverage ranges is evident, i.e.,
water desorption occur exclusively below 200 K.
water, i.e. ice, a broadening of the 3a feature
1
can be explained by the equivalence of the water
molecules, i.e. the molecules act as proton-donors
as well as acceptors [51]. Adapting this model for
the data presented here, we conclude that the out-
ermost surface of water multilayers, i.e. ice, act
more as donors whereas the in bulk ice, the water
molecules act as donors and acceptors. It is note-
worthy that MIES probes exclusively the outer-
most surface [37] whereas UPS (He I) spectra
contain contributions from the bulk. Donator-like
properties of an ice surface means that the proton
accepting oxygen ends of the water molecules are
on the average directed into the bulk and are not
visible by the MIES probe. In any case, the differ-
ence in the surface sensitivities of MIES and UPS
The latter is clearly the case for SiO /Mo(112),
2
suggesting that this system is comparable with
Ag [55], Au [56] and Cu [57,58] where such a
behavior was previously observed. The similarity
with these surfaces is also appropriate when com-
paring the activation energy of desorption. Hinch
and Dubois observed for Cu(111) that dynamic
adsorption/desorption of water at 155 K results
in the formation of small metastable water islands
at low coverages that coalesce into large crystalline
clusters with increasing coverage [57,58]. The
rate of desorption from the water islands at low
(He I) is the key to understanding the various peak
positions of the 3a features. This conclusion is
1

118
S. Wendt et al. / Surface Science 565 (2004) 107–120
coverages was found to be noticeably greater than
that from crystalline clusters at high coverages. A
similar result was found for Ag addressing water
desorption in the submonolayer range [59]. Con-
sidering these results we conclude that the interac-
b and c) also shows that water adsorbs molecularly
on SiO /Mo(112) and, further, that water mole-
cules cluster even at low coverages.
2
4.4. Work function data
tion of water with SiO thin films is indeed similar
2
to that of water with Ag, Au and Cu. Further-
more, we speculate that 5 L D O in our system
The work function change Du upon water
adsorption onto low- and high-defect SiO thin
2
2
correspond to a coverage not significant higher
than one monolayer in agreement with the data
shown in Figs. 2 and 6B. We emphasize that water
desorption was only found in the temperature
range 160–180 K. However, it is known that hyd-
roxyl groups (silanol groups) are strongly bound
to silica surfaces [32] and hence should desorb
films (Fig. 6B) is likewise evidence for molecular
adsorption. A negative work function change is
generally associated with the adsorption of mole-
cular water [1]. A work function decrease can be
caused by charge donation from the adsorbed
water to the substrate, by polarization of the water
molecule, and by an adsorption geometry with the
oxygen atoms oriented towards the substrate.
Since the work function decreases of 0.4 and 0.5
eV, respectively, are rather small, it is not likely
that all of the above mentioned effects contribute
significantly to the bonding of water to SiO2/
Mo(112). Discontinuities of the Du curves may
indicate the complete covering of the SiO2/
Mo(112) surface. In this regard it is understanda-
ble that curve (b) in Fig. 6B remains constant at a
higher value compared to curve (a) since the sur-
(
after recombination to water) at much higher
temperatures than 160–180 K. Therefore, the
TPD results support our conclusion that water
adsorbs molecularly on SiO /Mo(112).
2
4
.3. HREELS data
The SiO2 thin film, which was used for the
HREELS measurement, is comparable with the
low-defect film in the other chamber where
the MIES/UPS measurements were performed. In
both cases the thickness was estimated to be 0.3–
face area should be larger on a SiO film having
2
extended defects in comparison to a flat, low-
defect film.
0
.5 nm and sharp c(2 · 2) LEED patterns were ob-
served. The loss spectrum of Fig. 4f resembles the
HREELS spectrum of water multilayers (ice) in
superposition with features of the substrate. The
4.5. Comparison to previous results
À1
broad loss at 3360 cm is due to OH stretching
À1
Comparing the results found here that water
adsorbed molecularly rather than dissociatively
on SiO /Mo(112) with previous results for silica
modes (OH) and the loss at 1650 cm due to
the scissor mode d(HOH). The librational modes
of molecular water are expected at around 820
2
is difficult considering that surface structure of
typical high surface area silicas is unknown or
hypothetical. Since Nishijima et al., found partly
dissociated water on a thin SiOx layer over
Si(111) [21], we assume dissociation is not kineti-
cally limited. One explanation of the differences
in the results in the literature and the results pre-
sented here is that defect sites on silica surfaces
are extremely important yet are variable from sil-
ica to silica. First of all, it can be argued that the
exclusively non-dissociative adsorption found in
the present work and that by Chabal and Christ-
man [20] can be traced to the uniform stoichio-
metry of the silica films. Whereas the films used
À1
cm and may also contribute to the spectrum.
À1
The frequency of the small loss at 3690 cm is
too large for a hydrogen-bonded OH species and
is in the region of the stretching mode for free
OH groups [2]. However, a feature at this fre-
quency with low intensity in comparison to the
À1
bands at around 3360 cm is generally seen in
HREELS data for multilayer ice surfaces [4,44–
4
hydrogen-bonded OH stretches of H O molecules
6]. This loss feature has been assigned to non-
2
[
3
3
44,45]. The occurrence of broad losses around
À1
360 cm for low water exposure and no loss at
690 cm for these low coverages (Fig. 4, curves
À1

S. Wendt et al. / Surface Science 565 (2004) 107–120
119
by Nishijima et al., were non-stoichiometric silica
film grown on Si(111), the SiO2 thin films in
the present study grown on Mo(112) films are sto-
ichiometric. In this regard, theoretical work by
Feuston [33] and Bakaev [34] deserve special con-
sideration. Feuston et al. [33], concluded there to
be a considerable concentration of topological
and bonding defects in vitreous silica surfaces. It
was proposed that non-bridging oxygen atoms
constitute a large component of these defects.
Later Bakaev and Steele concluded that single
and triple coordinated oxygen atoms as well as dis-
thin films prepared under UHV conditions and an-
nealed at elevated temperatures. This result is con-
sistent with previous studies where hydrophobic
properties of dehydroxylated silica surfaces were
reported. Furthermore, the non-dissociative ad-
sorption of water is consistent with theoretical
studies that purport point defects to be crucial
for the hydroxylation of silica surfaces. Extended
defects, i.e. steps and corners, are not sufficient
to cause water dissociation on SiO thin films pre-
2
pared on Mo(112). The results indicate that clus-
tering of water, i.e. 3-D growth, begins at very
low coverages. The water–water interaction via
hydrogen bonding is stronger than the interaction
of water with the silica thin film. Varying peak
torted SiO tetrahedra at the surface of vitreous
4
silica create a strong electrostatic field and that
these defects increase the hydrophility of the sur-
face [34]. It is important to note that a simulated
surface which consists only of bridging oxygen
atoms was found to be completely hydrophobic
positions of the 3a orbital in MIES and UPS
1
(He I) for multilayers of water (ice) can be ex-
plained invoking different surface sensitivities of
MIES and UPS (He I).
[
34]. Considering this and those studies in which
the hydrophobic character of pure silica surfaces
not containing OH groups, i.e. silanol groups)
(
have been reported [26–28], we conclude that de-
fect sites are decisive. The silica thin films prepared
on Mo(112) are not only well-ordered, but most
probably are free of defects leading to water disso-
ciation. The presence of extended defects, i.e. steps
and corners, do not cause detectable amounts of
OH groups upon water adsorption as is evident
in Figs. 6A, 7 and 8. Point defects such as single
and triply coordinated oxygen atoms may lead to
hydroxylation of silica surfaces. In view of this
Acknowledgments
The funding for this work was provided by the
Department of Energy, Office of Basic Energy Sci-
ences, Division of Chemical Sciences and Robert
A. Welch Foundation, and the Texas Advanced
Technology Program under Grant 010366-0022-
2001. M.F. thanks for the kind hospitality at
Texas A&M University during his six-month stay.
The stay of M.F. was possible in the frame of the
cooperation agreement between Texas A&M Uni-
versity and the Institute of Physics at the Technical
University Clausthal, Germany.
point, it is not surprising that a H O plasma dis-
2
charge was necessary to transform a hydrophobic
SiO surface into a hydroxylated silica surface as
reported by Sneh and George [32].
2
5
. Conclusions
References
[
[
[
1] P.A. Thiel, T.E. Madey, Surf. Sci. Rep. 7 (1987) 211.
2] M.A. Henderson, Surf. Sci. Rep. 46 (2002) 1.
3] (a) P.J. Feibelman, Science 295 (2002) 99;
(b) D. Menzel, Science 295 (2002) 58.
Based on MIES, UPS (He I), HREELS, TPD
and Du measurements we present conclusive evi-
dence that water adsorbs molecularly on SiO thin
2
films prepared on Mo(112). Considering the lower
limit of the detectability of hydroxyl-features by
MIES and UPS, we estimate that the fraction of
dissociated water molecules is at most 9%. How-
ever, since there is no evidence for water dissocia-
tion it is likely that water does not react with SiO2
[4] M.A. Henderson, Surf. Sci. 355 (1996) 151.
[5] M.A. Henderson, Langmuir 12 (1996) 5093.
[
6] R. Schaub, P. Thostrup, N. Lopez, E. Lægsgaard, I.
Stensgaard, J.K. Nørskov, F. Besenbacher, Phys. Rev.
Lett. 87 (2001) 266104-1.
[7] I.M. Brookes, C.A. Muryn, G. Thornton, Phys. Rev. Lett.
87 (2001) 266103-1.

120
S. Wendt et al. / Surface Science 565 (2004) 107–120
[
[
8] R.M. Lynden-Bell, L. Delle Site, A. Alavi, Surf. Sci. 496
2002) L1.
9] L. Delle Site, A. Alavi, R.M. Lynden-Bell, J. Chem. Phys.
13 (2000) 3344.
[33] B.P. Feuston, S.H. Garofalini, J. Chem. Phys. 91 (1989)
564.
(
[34] V.A. Bakaev, W.A. Steele, J. Chem. Phys. 111 (1999) 9803.
[35] Y.D. Kim, T. Wei, J. Stultz, D.W. Goodman, Langmuir
19 (2003) 1140.
1
[
[
10] M. Odelius, Phys. Rev. Lett. 82 (1999) 3919.
11] L. Giordano, J. Goniakowski, J. Suzanne, Phys. Rev. Lett.
[36] (a) W. Maus-Friedrichs, M. Wehrhahn, S. Dieckhoff, V.
Kempter, Surf. Sci. 237 (1990) 257;
8
1 (1998) 1271.
12] (a) X. Xu, D.W. Goodman, Appl. Phys. Lett. 61 (1992)
74;
b) X. Xu, D.W. Goodman, Surf. Sci. 282 (1993) 323.
[
(b) W. Maus-Friedrichs, S. Dieckhoff, V. Kempter, Surf.
Sci. 249 (1991) 149.
7
(
[37] Y. Harada, S. Masuda, H. Ozaki, Chem. Rev. 97 (1997)
1897.
[
[
13] D.W. Goodman, J. Vac. Sci. Technol. A 14 (1996) 1526.
14] S.C. Street, C. Xu, D.W. Goodman, Annu. Rev. Chem. 48
[38] (a) T. Schroeder, J.B. Georgi, A. Hammoudeh, N. Magg,
M. B a¨ umer, H.-J. Freund, Phys. Rev. B 65 (2002) 115411;
(b) A.K. Santra, B.K. Min, D.W. Goodman, Surf. Sci. 513
(2002) L441.
(
1997) 43.
15] S. Wendt, Y.D. Kim, D.W. Goodman, Prog. Surf. Sci. 74
2003) 141.
[
[
(
16] (a) T. Schroeder, A. Hammoudeh, M. Pykavy, N. Magg,
M. Adelt, M. B a¨ umer, H.-J. Freund, Solid-State Electron.
[39] S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interf. Anal. 17
(1991) 911.
4
5 (2001) 1471;
b) T. Schroeder, M. Adelt, B. Richter, M. Naschitzki, M.
B a¨ umer, H.-J. Freund, Surf. Rev. Lett. 7 (2000) 7;
c) T. Schroeder, M. Adelt, B. Richter, M. Naschitzki, M.
B a¨ umer, H.-J. Freund, Microelectron. Reliab. 40 (2000)
41.
[40] Y.D. Kim, T. Wei, D.W. Goodman, Langmuir 19 (2003)
354.
(
[41] S. Wendt, E. Ozensoy, T. Wei, M. Frerichs, Y. Cai, M.S.
Chen, D.W. Goodman, to be submitted.
(
[42] D.W. Turner, C. Baker, A.D. Baker, C.R. Brundle,
Molecular Photoelectron Spectroscopy, Wiley-Inter-
science, New York, 1970.
8
[
[
[
17] T. Schroeder, J.B. Georgi, M. B a¨ umer, H.-J. Freund,
Phys. Rev. B 66 (2002) 165422.
18] M.S. Chen, A.K. Santra, D.W. Goodman, Phys. Rev. B 69
[43] P.A. Thiry, M. Liehr, J.J. Pireaux, R. Sporken, R.J.
Caudano, Vac. Sci. Technol. B 3 (1985) 1118.
[44] H. Ibach, S. Lehwald, Surf. Sci. 91 (1980) 187.
[45] E.M. Stuve, R.J. Madix, B.A. Sexton, Surf. Sci. 111 (1981)
11.
(2004) 155404.
19] E. Ozensoy, B.K. Min, A.K. Santra, D.W. Goodman, J.
Phys. Chem. B 108 (2004) 4351.
[
[
20] Y.J. Chabal, S.B. Christman, Phys. Rev. B 29 (1984) 6974.
21] M. Nishijima, K. Edamoto, Y. Kubota, S. Tanaka, M.
Onchi, J. Chem. Phys. 84 (1986) 6458.
[46] J.A. Schaefer, J. Anderson, G.J. Lapeyre, J. Vac. Sci.
Technol. A 3 (1985) 1443.
[47] M.-C. Wu, C.A. Estrada, D.W. Goodman, Phys. Rev.
Lett. 67 (1991) 2910.
[
22] N.H. de Leeuw, F.M. Higgins, S.C. Parker, J. Phys. Chem.
B 103 (1999) 1270, and references therein.
[48] H. Ishii, S. Masuda, Y. Harada, Surf. Sci. 239 (1990) 222.
[49] S. Krischok, O. H o¨ fft, J. G u¨ nster, J. Stultz, D.W. Good-
man, V. Kempter, Surf. Sci. 495 (2001) 8.
[50] J. G u¨ nster, S. Krischok, V. Kempter, J. Stultz, D.W.
Goodman, Surf. Rev. Lett. 9 (2002) 1511.
[51] D. Schmeisser, F.J. Himpsel, G. Hollinger, B. Reihl, K.
Jacobi, Phys. Rev. B 27 (1983) 3279.
[
[
[
23] N.H. de Leeuw, J. Phys. Chem. B 105 (2001) 9747.
24] M. Wilson, T.R. Walsh, J. Chem. Phys. 113 (2000) 9180.
25] T.R. Walsh, M. Wilson, A.P. Sutton, J. Chem. Phys. 113
(2000) 9191, and references therein.
[
[
[
[
[
26] V. Bolis, B. Fubini, L. Marchese, G. Martra, D. Costa,
J. Chem. Soc., Faraday Trans. 87 (1991) 497.
27] D.W. Sindorf, G.E. Maciel, J. Am. Chem. Soc. 105 (1983)
[52] S. Casassa, P. Ugliengo, C. Pisani, J. Chem. Phys. 106
(1997) 8030.
1
487.
28] A.S. DÕSouza, C.G. Pantano, J. Am. Ceram. Soc. 85 (2002)
499.
[53] Y. Joseph, W. Ranke, W. Weiss, J. Phys. Chem. B 104
(2000) 3224.
1
29] Various OH configurations bonded to Si are classified as
silanol. For an overview see, for example, Ref. [30].
30] H. Kn o¨ zinger, in: P. Schuster, G. Zundel, C. Sandorfy
[54] H. Umeyama, K. Morokuma, J. Am. Chem. Soc. 99 (1977)
1316.
[55] M. Klaua, T. Madey, Surf. Sci. 136 (1984) L42.
[56] R.S. Smith, B.D. Kay, Surf. Rev. Lett. 4 (1997) 781.
[57] B.J. Hinch, L.H. Dubois, J. Chem. Phys. 96 (1992) 3262.
[58] B.J. Hinch, L.H. Dubois, Chem. Phys. Lett. 181 (1991) 10.
[59] K.J. Wu, L.D. Peterson, G.S. Elliott, S.D. Kevan, J.
Chem. Phys. 91 (1989) 7964.
(Eds.), The Hydrogen Bond, vol. 3, North-Holland,
Amsterdam, 1976, p. 1263, Chapter 27.
31] B.A. Morrow, I.A. Cody, L.S.M. Lee, J. Phys. Chem. 80
[
[
(1976) 2761.
32] O. Sneh, M. George, J. Phys. Chem. 99 (1995) 4639.