PHYSICAL REVIEW B 74, 235412 共2006兲
STM study of hydroxyl formation at O / Ag„110…
L. Savio,1,2 M. Smerieri,1 L. Vattuone,1,2 A. Gussoni,3 C. Tassistro,3 and M. Rocca1,3
1Dipartimento
di Fisica dell’Università di Genova, Via Dodecaneso 33, 16146 Genova, Italy
Unità di Genova, Via Dodecaneso 33, 16146 Genova, Italy
3
IMEM—CNR, Via Dodecaneso 33, 16146 Genova, Italy
共Received 13 June 2006; revised manuscript received 15 September 2006; published 7 December 2006兲
2CNISM
We report here on a low temperature scanning tunneling microscopy study of the hydroxylation of oxygen
precovered Ag共110兲 at a crystal temperature of 250 K. For water doses comparable with the oxygen coverage,
we observe complete removal of the O-Ag added rows and the formation of an OH-induced superstructure in
the perpendicular 共具11̄0典兲 direction, in accord with previous results. OH sits in the channels of the unreconstructed surface, in agreement with the adsorption at threefold hollow sites proposed by K. Bange et al. 关Surf.
Sci. 183, 334 共1987兲兴. The OH layer disorders upon localized voltage pulses. The damage extends along the
具11̄0典 direction, suggesting a strong interaction among neighboring OH groups.
DOI: 10.1103/PhysRevB.74.235412
PACS number共s兲: 68.43.Jk, 68.37.Ef, 68.47.De
I. INTRODUCTION
Water interaction at surfaces is a topic of particular interest from the fundamental as well as from the applicative
point of view, as demonstrated by the extensive research in
this field.1 In particular, hydroxylation of metal substrates
occurs in several catalytic reactions in the heterogeneous
phase, acting, e.g., as an intermediary in the oxidation of
hydrogen to water on Pt surfaces.2 As an electronegative adsorbate, OH can affect also electrochemical processes. On
the other hand, being a fundamental component of the atmosphere, water corrosion is an important phenomenon for all
devices working in air.
Water dissociation at oxygen precovered Ag共110兲 was assumed as a model system for the hydroxylation of noble
metals. It is well known3,4 that above 200 K oxygen adsorbs
dissociatively and forms an added row superstructure consisting of O-Ag chains extending along the 具001典 direction
with 共n ⫻ 1兲 periodicity 共n = 2 , . . . , 7 depending on
coverage4兲. Low energy electron diffraction 共LEED兲 and He
diffraction indicate that hydroxylation of the 共n ⫻ 1兲
O-Ag共110兲 layer causes the formation of a 共1 ⫻ m兲 superstructure in the 具11̄0典 direction, with m = n / 2 for even n and
with domains of different periodicity for odd n.5,6 Thermal
desorption shows OH recombination temperatures ranging
from ⬃250 K to ⬃ 310 K, depending on coverage.5,7 Upon
partial desorption of the OH adlayer, the O / Ag共110兲 added
rows are restored and coexistence of the two superstructures
is detected.
In spite of these extensive experimental studies, some
controversies remain open. In particular, the adsorption site
for OH has been identified with the threefold hollow by electron stimulated desorption angular distribution 共ESDIAD兲.5
This assignment was supported by a more recent high resolution electron energy loss spectroscopy investigation,8
which shows two dipole active OH bending modes, as expected if the adsorbate occupies the low symmetry threefold
site. Angular resolved ultra-violet photoemission spectroscopy 共ARUPS兲 data, on the contrary, were interpreted as evidence for C2v symmetry, suggesting that OH sits at short
bridges on top of the atomic Ag rows.7 Indeed, this configu1098-0121/2006/74共23兲/235412共6兲
ration turned out to be slightly more stable than adsorption at
threefold hollows according to dipped adcluster model
calculations.9
We present here a low temperature STM study of the H2O
interaction with O / Ag共110兲 at a crystal temperature T
= 250 K. In accord with previous literature,5,6 we observe
that exposure to water causes the complete removal of the
O-Ag chains and that the OH groups generated in the reaction form a superstructure consisting of OH rows aligned in
the 具11̄0典 direction. We find hydroxyls to sit in the valleys
between the Ag rows, slightly displaced towards the threefold hollows. The OH rows are unstable under the STM tip
action above a critical voltage. Voltage pulses scatter the OH
groups around the surface and the damage propagates along
the OH row, leaving nearby rows unperturbed. The directionality of the destruction is indicative of a strong interaction
among OH groups in the 具11̄0典 direction.
II. EXPERIMENTAL
Experiments were carried out in an ultrahigh vacuum
共UHV兲 apparatus consisting of a main chamber, hosting a
commercial low temperature STM 共LT-STM, supplied by
Createc兲, and a preparation chamber. The latter is equipped
with valves for gas inlet, a quadrupole mass spectrometer
共QMS兲, an ion gun, and all other typical vacuum facilities.
The sample holder is mounted on a four degrees of freedom,
He-coolable manipulator and can be transferred from the manipulator into the STM. The sample can be resistively heated
to T = 800 K and cooled down to 7 K 共90 K兲 by liquid He
共liquid N2兲 flux.
Before each experiment, the Ag共110兲 surface is prepared
by sputtering followed by annealing to 740 K. High purity
O2 and bidistilled H2O are then dosed by backfilling the
chamber, while keeping the crystal, respectively, at 300 K
and 250 K. The gas inlet lines are distinct and are carefully
pumped before each preparation. After the exposure the
sample is cooled down below 120 K before being transferred
into the STM. This operation requires approximately 10 min,
during which some contamination from the residual gas
235412-1
©2006 The American Physical Society
PHYSICAL REVIEW B 74, 235412 共2006兲
SAVIO et al.
might occur. The lowest water dose performed here, 0.1 L, is
therefore chosen as the smallest quantity for which post-dose
adsorption is negligible. The adsorbate coverage was determined a posteriori from the periodicity of the superstructures
observed by STM.
The LT-STM is kept at 5 K by thermal contact with a
liquid He cryostat and is completely shielded from radiation
by a double screen 共inner part at liquid He, outer one at
liquid N2 temperature兲. At such a low temperature, the
sample can survive uncontaminated for several days after
preparation allowing for an extensive investigation. The
STM tip is produced in air by grazing cut of a 0.2 mm diameter Pt-Ir wire under strain. While measuring, the tip apex
is then reshaped by controlled crashes into the Ag sample, so
that tunneling occurs effectively through a silver tip. The
piezo constants are calibrated on the basis of the Ag lattice
parameter from atomically resolved images of the clean surface.
Images are acquired in constant current mode, using typical currents between 0.1 nA and 2.0 nA and bias voltages
between −1.20 V and 1.20 V applied to the sample.
III. RESULTS AND DISCUSSION
In Fig. 1 we show the Ag共110兲 surface after exposures to
95 L and 900 L of O2 at T = 300 K 关experiments 共a兲 and 共b兲,
respectively兴. The inset in panel 共a兲 reports an atomically
resolved image of the clean surface, which is employed for
calibrating the scan dimensions. As expected,3 we observe an
added row reconstruction with the formation of O-Ag chains
in the 具001典 direction. Atomic resolution on Ag adatoms can
be easily achieved 关see inset in panel 共b兲兴, while O adatoms
could not be imaged at any of the investigated voltages. The
identification of the bright features forming the added rows
with Ag is straightforward since they are in register with the
channels of the underlying Ag共110兲 substrate;3 moreover,
calculated STM images of the same structure predict an oxygen contribution to the STM intensity only for bias voltages
V ⬍ −1.0 V.10 Figure 1共c兲 reports line scans along the 具11̄0典
direction for the two surfaces 关dotted lines in 共a兲 and 共b兲兴.
The added-row periodicity depends strongly on oxygen dose;
in the present experiments, it is not uniform since exposures
were not calibrated to achieve the particular coverage corresponding to a well-defined overlayer. At T = 300 K the O-Ag
chains are indeed not mobile enough to allow for complete
surface ordering at low coverage. Statistical analysis 关performed over 13 and 15 overviews for preparations 共a兲 and
共b兲, respectively兴 indicates, however, that after 95 L most of
the surface is covered by 6 ⫻ 1 and 5 ⫻ 1 domains, while an
extended 3 ⫻ 1 superstructure forms for the larger dose. This
analysis allows us to deduce an average atomic oxygen coverage between 0.16 monolayer 共ML兲 and 0.20 ML for the
short exposure, of ⬃0.33 ML for the long one. We note that
the O-Ag rows end with a brighter spot, indicating that the
last Ag atom of the chain is electronically different from the
others. This effect was observed also by Zambelli et al. for
O-Ag chain formation at 190 K.11
In Fig. 2 we report STM images obtained upon exposing
the O precovered surface at T = 250 K to water. Figure 2共a兲
shows the Ag共110兲 substrate after exposure to 98 L at 300 K,
followed by ⬃0.1 L of H2O. Figure 2共b兲 reports the result of
the exposure to 1035 L of O2 followed by ⬃10 L of H2O.
With respect to Fig. 1, the surface structure has dramatically
changed, showing now dark, linear features in the 具11̄0典 direction. In accord with previous spectroscopic and He scattering results,5,7,8 we interpret this finding as indicative of
surface hydroxylation following the reaction H2O
+ O共ad兲→20H共ad兲. The OH groups organize themselves in
ordered rows extending in the 具11̄0典 direction. For the low O
coverage 关Fig. 2共a兲兴, the hydroxylation of the surface is complete already at the smallest H2O dose. In the intermediate
situation of water exposures insufficient to achieve a complete surface hydroxylation, the substrate consists of O-Ag
domains alternated to Ag-OH areas, as shown in Fig. 3共a兲.
Moreover, for an initially high O coverage 共3 ⫻ 1 phase兲, the
O-Ag chains compact into 2 ⫻ 1 domains 共not shown兲. The
complete corrosion of the O-Ag chains and the formation of
an extended OH overlayer occur, however, as soon as more
water molecules become available 关Fig. 2共b兲兴.
We note the following:
共1兲 Water dissociates with nearly unitary efficiency on
O / Ag共110兲, as witnessed by the very small doses involved.
Destruction of the O-Ag chains and formation of the Ag-OH
superstructure in the 具11̄0典 direction are expected from
previous diffraction results. Moreover we mention that an
STM image similar to those reported here was presented by
Zambelli et al.12 as a side result achieved during oxygen
adsorption experiments. They suggested a possible OH contamination but did not investigate the phenomenon further.
The OH superstructure is imaged dark in all tested tunneling conditions. Its identification with OH rows is quite
straightforward since their length and density scale with O
coverage and H2O exposure 共i.e., with OH coverage兲. Moreover, dark stripes appear as isolated features in O2 adsorption
experiments performed below room temperature, in which a
limited water/hydroxyl contamination is plausible. This assignment is reinforced by the observation that the dark lines
are unstable under the STM tip action above a critical voltage 共see below and Fig. 4兲, contrary to the brighter areas in
between them. The density of states close to the Fermi level
is thus lower for Ag-OH than for Ag atoms at the clean
surface.
共2兲 The extended hydroxyl induced superstructure is not
completely regular in the image reported in Fig. 2共a兲, where
areas of 1 ⫻ 3 periodicity are separated by flat areas four to
six lattice spacing wide. This is better evidenced in the line
scan reported in Fig. 2共c兲 and it is probably due to the mixed
periodicity of the starting O / Ag layer. The experiment reported in Fig. 2共b兲 exhibits indeed a more uniform 1 ⫻ 3
reconstruction with only small 1 ⫻ 2 areas. Statistical analysis 关over 52 and 10 overviews for preparations 共a兲 and 共b兲,
respectively兴 allows us to consider these images as representative of the whole sample and to estimate an OH coverage
between 0.25 ML and 0.33 ML for the former surface, and
slightly higher than 0.33 ML for the latter one. For the high
O coverage a larger OH density, in particular a coexistence
of 1 ⫻ 1 and 1 ⫻ 2 domains, would be expected. However,
the 1 ⫻ 1 layer desorbs at T ⬃ 240 K,6 so that this phase does
not form in our experimental conditions.
235412-2
PHYSICAL REVIEW B 74, 235412 共2006兲
STM STUDY OF HYDROXYL FORMATION AT O / Ag共110兲
FIG. 1. 共Color online兲 共a兲 STM image of the
Ag共110兲 surface after exposure to 90 L of O2 at T
= 300 K, showing the formation of an added row reconstruction in the 具001典 direction 共image size 207 Å
⫻ 207 Å, I = 1.5 10−9 A, V = 1.0 V兲. The inset reports an
atomically resolved image of the clean surface. The orientation of the surface lattice and the primitive unit cell
are marked 共image size: 24 Å ⫻ 24 Å, I = 1.0 10−9 A,
V = 0.87 V兲. 共b兲 Same as 共a兲 for an Ag共110兲 surface exposed to 900 L of O2 at 300 K. 共Image size 207 Å
⫻ 207 Å, I = 1.1 10−9 A, V = 0.80 V兲. The inset shows
atomic resolution on Ag atoms forming the added rows
共image size: 42 Å ⫻ 52 Å, I = 1.8 10−9 A, V = −1.1 V兲.
The white lines are a guide to the eye, underlining that
Ag adatoms are in register with the troughs of the underlying Ag共110兲 surface. 共c兲 Line scans cut along the
dotted lines in panels 共a兲 共top profile兲 and 共b兲 共bottom
profile兲 and showing the added rows periodicity in the
具11̄0典 direction.
235412-3
PHYSICAL REVIEW B 74, 235412 共2006兲
SAVIO et al.
FIG. 2. 共Color online兲 共a兲 STM image of the Ag共110兲 surface
after exposure to 98 L of O2 at 300 K and 0.1 L of H2O at 250 K
共image size: 189 Å ⫻ 189 Å, I = 3.0 10−10 A, V = 0.10 V兲. 共b兲 STM
image of the Ag共110兲 surface exposed to 1035 L of O2 at 300 K
and ⬃10 L of H2O at 250 K 共image size 189 Å ⫻ 189 Å, I
= 7.6 10−10 A, V = 0.40 V兲. 共c兲 Line scans along the dotted lines in
共a兲 共top profile兲 and 共b兲 共bottom profile兲, showing the periodicity of
the OH superstructure.
共3兲 An additional feature, consisting of a sequence of
bright spots sitting in the middle of OH domains, is visible in
Fig. 2共a兲. This structure has a two lattice spacing periodicity
in the 具11̄0典 direction and is observed more frequently upon
short water exposures. We tentatively identify it with residual
O-Ag units 共compacted in a 2 ⫻ 1 periodicity because of the
extending OH covered areas兲 that have not reacted with
H2O. Indeed their apparent height of ⬃0.5 Å compares well
FIG. 3. 共Color online兲 共a兲: STM image showing the intersection
between O-Ag chains 共bright horizontal atomic lines兲 and Ag-OH
rows 共dark vertical lines兲. Atomic resolution on Ag atoms could be
achieved on both structures. The surface corrugation in between the
OH-induced features is also evident 共image size: 54 Å ⫻ 40 Å, I
= 1.8 10−9 A, V = 0.29 V兲. 共b兲 Height profile along the white line in
共a兲. The low maxima corresponding to the Ag-OH groups are three
lattice spacings apart, marked by blue 共dark gray兲 arrows pointing
down. Three other maxima are found in between, corresponding to
rows of clean Ag atoms. The central one is higher 关red 共dark gray兲
arrow兴, while the apparent height of side maxima 关yellow 共light
gray兲 arrows兴 is slightly reduced. 共c兲 Simple model comparing OH
adsorption at Ag共110兲 threefold hollow 共I兲 or short-bridge 共II兲 sites.
Gray circles indicate Ag atoms 共dark gray: top layer; light gray:
second layer兲; light blue 共light gray兲 circles represent Ag 共big兲 and
O 共small兲 atoms forming O-Ag chains and blue 共dark gray兲 circles
are OH groups. As discussed in the text, only model 共I兲, i.e. adsorption in the channels, fits with experimental data.
with the one of ⬃0.4 Å estimated for the bright dots at the
O-Ag chains terminations 关see Fig. 1共b兲兴. The possibility of
an arrangement along the 具11̄0典 direction for residual O-Ag
groups has already been suggested in Ref. 5. The alternative
hypothesis that also this structure is due to OH or H2O molecules is at present under theoretical investigation.13
More details on the OH-induced superstructure can be
gained from inspection of Fig. 3, showing an atomically resolved image of the intersection between OH rows and still
unreacted O-Ag chains. The former ones are aligned with the
Ag atoms of the O-Ag added rows, which are well known to
235412-4
PHYSICAL REVIEW B 74, 235412 共2006兲
STM STUDY OF HYDROXYL FORMATION AT O / Ag共110兲
FIG. 4. 共Color online兲 共a兲–共d兲: Sequence of STM images showing the evolution of a hydroxylated surface 关prepared as in Fig. 2共a兲兴 under
the effect of voltage pulses applied both at the extremes or in the middle of long OH rows. Green 共gray兲 dots indicate the point at which a
pulse was applied. The corresponding voltage is written in green. Destruction of the ordered superstructure and scattering of OH groups are
evident in panel 共d兲. For all images: 105 Å ⫻ 105 Å, I = 3.8 10−10 A, V = 0.10 V. 共e兲 Probability of chain-breaking vs applied voltage. 共f兲
Fraction of rows broken at a fixed voltage with respect to the total amount of broken rows. In both cases circles refer to pulses applied at the
row terminations while squares to pulse applied in the middle of the rows. Since there is no dependence on polarity within experimental error
关see V+ and V− curves in 共e兲兴, we averaged data of both polarities to reduce the scatter of the data. Statistics were performed over 125
共circles兲 and 551 共squares兲 pulses in 共e兲 and over 51 共circles兲 and 48 共squares兲 successful events in 共f兲.
sit in fourfold hollow sites.3 This proves that OH adsorbs in
the channels of the Ag共110兲 surface. Such evidence is reinforced by a more detailed analysis of the surface corrugation
through the line scan in Fig. 3共b兲. The blue 共dark gray, pointing down兲 arrows indicate the position of the OH-related
features, which are three lattice spacings apart. In between
them, three maxima are present, the central one being
slightly higher. This line profile fits with model 共I兲 proposed
in Fig. 3共c兲, in which the OH structures are located in the
channels of the Ag共110兲 surface. The adjacent Ag rows
共marked in yellow/light gray兲 are slightly perturbed, so that
their apparent height is lower in STM images, while the central one 共red/dark gray兲 is not affected. Possibly also a slight
rearrangement of Ag surface atoms might have occurred. The
alternative model 共II兲, in which OH sits on the outermost
rows of atoms of the unreconstructed Ag共110兲 surface, implies only two maxima in the height profile and is therefore
ruled out by the experimental data.
Density functional theory calculations indicate indeed the
highly coordinated fourfold and threefold hollow sites as the
most stable for OH adsorption at Ag共100兲 and Ag共111兲,
respectively.9,14,15 For Ag共110兲 the assignment is more debated since both adsorption at the short-bridge site in a con-
figuration with the OH axis strongly tilted7,9 and adsorption
at threefold hollows5 have been proposed. Our observation
that OH sits in the channels of Ag共110兲 strongly supports
adsorption at threefold hollows. The shape of the high resolution height profile of Fig. 3共b兲, showing an asymmetric
location of the OH feature 共blue arrows兲, is also coherent
with the model of Bange et al.5
The OH rows are unstable under the STM tip action
above a critical voltage. As shown in Fig. 4, localized voltage pulses given by keeping the tip 4 Å away from the surface cause a local damage of the superstructure. OH groups
appear then randomly spread over the surface. Some statistic
on the phenomenon is reported in the bottom panels: Fig.
4共e兲 shows the probability that a pulse causes the breaking of
the OH row versus applied voltage while the istogram in Fig.
4共f兲 represents the fraction of rows broken at a given voltage
with respect to the total amount of broken rows. In both
cases, the analysis is performed separately for voltage pulses
applied at the end or in the middle of a row, since the different environment causes markedly different behaviors of the
OH structures. From an accurate inspection of Fig. 4, we
note that 共i兲 the effect is present only for pulses localized on
the dark features and both for negative and positive bias. 共ii兲
235412-5
PHYSICAL REVIEW B 74, 235412 共2006兲
SAVIO et al.
The measured threshold is 400– 500 mV for voltage pulses
close to the OH row terminations and ⬃800 mV for pulses
applied in the middle of long rows. 共iii兲 The interaction with
the tip is punctual, but the induced perturbation extends
along the 具11̄0典 direction for several lattice spacings, as evident from the sequence in Fig. 4. Since chain breaking does
not depend on polarity 关as demonstrated by comparison of
the V+ and V− curves in Fig. 4共e兲兴, we rule out that it is due
to electron injection into an empty state. Noteworthy, the
lowest critical voltage occurs close above the threshold of
the excitation of the OH stretch, which may thus mediate the
energy transfer to the translational degree of freedom. Coupling between internal and translational motion, causing lateral hopping of the adsorbed molecule, was observed, e.g.,
for CO on Pd共110兲 共Ref. 16兲 and for CH3S on Cu共111兲.17 In
our case the preferential direction suggests, however, a
strong substrate-mediated interaction between the aligned
OH groups, while neighboring stripes are not interacting
with each other. In this frame, the higher threshold measured
for pulses applied in the middle of long rows is explained in
terms of the higher coordination of OH groups and thus of
the higher energy required to break the ordered structure.
Henderson, Surf. Sci. Rep. 46, 1 共2002兲.
Beduerftig, S. Völkening, Y. Wang, J. Wintterlin, K. Jacobi,
and G. Ertl, J. Chem. Phys. 111, 11147 共1999兲.
3 T. Hashizume, J. E. Rowe, R. A. Malic, K. Motai, K. Cho, J.
Kishimoto, and T. Sakurai, J. Vac. Sci. Technol. B 12, 1809
共1994兲.
4
H. A. Engelhardt and D. Menzel, Surf. Sci. 57, 591 共1976兲.
5 K. Bange, T. E. Madey, J. K. Sass, and E. M. Stuve, Surf. Sci.
183, 334 共1987兲.
6 M. Canepa, P. Cantini, E. Narducci, M. Salvietti, S. Terreni, and
L. Mattera, Surf. Sci. 343, 176 共1995兲.
7 M. Canepa, P. Cantini, L. Mattera, E. Narducci, M. Salvietti, and
S. Terreni, Surf. Sci. 322, 271 共1995兲.
8
M. Borbach, W. Stenzel, H. Conrad, and A. M. Bradshaw, Surf.
Sci. 377–379, 796 共1997兲.
1 M.
2 K.
CONCLUSIONS
We have shown by low temperature STM investigation
that water exposure on the O / Ag共110兲 surface at 250 K
causes the complete removal of O-Ag chains and the formation of a regular superstructure in the 具11̄0典 direction. Our
data are indicative that the adsorption site for OH is the
threefold hollow, in the channels of the Ag共110兲 surface. The
OH rows are broken by voltage pulses applied with the STM
tip. The damage extends then along the OH row, leaving
nearby rows unperturbed and indicating a strong substratemediated interaction among OH groups within the same
channel.
ACKNOWLEDGMENTS
This research was funded through the FIRB2001
RBAU0157PZ project of the Italian Ministry of Research.
We also acknowledge financial support from Compagnia S.
Paolo and Fondazione Carige. We thank M. Canepa, R. Ferrando and J. Goniakowski for scientific discussion, A. Pozzo
for technical assistance, and L. Guillemot for showing unpublished data.
M. Hu and H. Nakatsuji, Surf. Sci. 425, 296 共1999兲.
Schimizu and M. Tsukada, J. Vac. Sci. Technol. B 12, 2200
共1994兲.
11 T. Zambelli, J. V. Barth, and J. Wintterlin, Phys. Rev. B 58,
12663 共1998兲.
12 T. Zambelli, J. V. Barth, and J. Wintterlin, J. Phys.: Condens.
Matter 14, 4241 共2002兲.
13 J. Goniakowski 共private communication兲.
14
C. Qin and J. L. Whitten, J. Phys. Chem. B 3109, 8852 共2005兲.
15 M. T. M. Koper and R. A. van Santen, J. Electroanal. Chem. 472,
126 共1999兲.
16 T. Komeda, Y. Kim, M. Kawai, B. N. J. Persson, and H. Ueba,
Science 295, 2055 共2002兲.
17
M. Ohara, Y. Kim and M. Kawai, Jpn. J. Appl. Phys., Part 1 45,
2022 共2006兲.
9 Z.
10 T.
235412-6