Co-induced restructuring of Pt(110)-(1×2): Bridging the pressure gap with high-pressure scanning tunneling microscopy

Article abstract of DOI:10.1063/1.1540611

CO-induced structural transformations occurring on the reconstructed Pt(110)-(1×2) was investigated. The structural changes were measured on the atomic scale as a function of CO pressure over 12 orders of magnitude by the use of high-pressure scanning tunneling microscope (HP-STM). The transition between the low-coverage and saturation-coverage structures was found to proceed through local displacements of substrate Pt atoms.

Full text of DOI:10.1063/1.1540611

CO-induced restructuring of Pt(110)-(1×2): Bridging the pressure gap with high-
pressure scanning tunneling microscopy
P. Thostrup, E. Kruse Vestergaard, T. An, E. Lægsgaard, and F. Besenbacher
Citation: The Journal of Chemical Physics 118, 3724 (2003); doi: 10.1063/1.1540611
View online: http://dx.doi.org/10.1063/1.1540611
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/118/8?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Dynamics of water adsorption on Pt { 110 } - ( 1 × 2 ) : A molecular dynamics study
J. Chem. Phys. 131, 064703 (2009); 10.1063/1.3204700
Origin, symmetry, and temperature dependence of the perturbation induced by Si extrinsic defects on the
Sn/Si(111) α surface: A scanning tunneling microscopy study
J. Vac. Sci. Technol. A 18, 1946 (2000); 10.1116/1.582451
A molecular beam study of nonlinearity in the CO-induced surface restructuring of Ir{100}
J. Chem. Phys. 109, 10996 (1998); 10.1063/1.477738
Scanning tunneling microscopy of equilibrium crystal shapes
J. Vac. Sci. Technol. A 16, 1059 (1998); 10.1116/1.581233
Elevated temperature scanning tunneling microscopy study of formic acid adsorption and reaction on oxygen
(2×1) covered Cu(110)
J. Chem. Phys. 108, 6916 (1998); 10.1063/1.476106
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 147.143.2.5
On: Sun, 21 Dec 2014 00:20:33

JOURNAL OF CHEMICAL PHYSICS
VOLUME 118, NUMBER 8
22 FEBRUARY 2003
CO-induced restructuring of Pt„110…-„1Ã2…: Bridging the pressure gap
with high-pressure scanning tunneling microscopy
P. Thostrup, E. Kruse Vestergaard, T. An, E. Lægsgaard, and F. Besenbachera)
Institute of Physics and Astronomy, CAMP, and Interdisciplinary Nanoscience Center (iNANO), University
of Aarhus, DK-8000 Aarhus C, Denmark
͑
Received 15 May 2002; accepted 3 December 2002͒
We present an extensive investigation of CO-induced structural transformations occurring on the
reconstructed Pt͑110͒-͑1ϫ2͒ surface while bridging the so-called pressure gap between surface
science and industrial catalysis. The structural changes are followed on the atomic scale as a
function of CO pressure over 12 orders of magnitude, up to 1 bar, by the use of a novel
high-pressure scanning tunneling microscope ͑HP-STM͒. The transition between the low-coverage
and saturation-coverage structures is found to proceed through local displacements of substrate Pt
atoms. The structural transformations of the Pt surface as observed by STM can be explained within
a very simple picture governed by the gain in CO binding energy when CO binds to low-coordinated
metal atoms. © 2003 American Institute of Physics. ͓DOI: 10.1063/1.1540611͔
I. INTRODUCTION
pressure gap is necessarily rendered invalid, if the surface
science approach fails to emulate processes occurring at high
pressures and temperatures. From cases in the literature
A major issue to the catalytic community concerns the
validity of applying surface science data obtained under rari-
fied ultra high vacuum ͑UHV͒ conditions to ‘‘real’’ catalytic
processes occurring at atmospheric pressure and above, often
where relations between the two sides of the pressure gap
have been investigated so far,3
,6–10
it can only be concluded
that the high-pressure response is system specific. There is
thus a clear need to add to the number of systems considered
and from specific cases try to extract general trends.
The system we have chosen to study in detail is CO/
Pt͑110͒. CO adsorption on Pt͑110͒ has been a favorite pro-
totype system for adsorbate-induced structural transforma-
tions in the modern surface science era and has been
1
referred to as the surface science approach. Here one typi-
cally performs experiments at low temperature ͑relative to
relevant adsorbate desorption temperatures͒ and low pres-
Ϫ10
Ϫ5
sures (10 –10 mbar͒. A number of so-called gaps be-
tween surface science conditions and high-pressure/high-
temperature conditions in heterogeneous catalysis has been
identified, among these:2
investigated extensively with almost any conceivable
theoretical1
1–15
and experimental UHV technique: low en-
͑
͑
1͒ pressure gap;
2͒ materials gap.
1
6–23
ergy electron diffraction ͑LEED͒,
ultraviolet photoelec-
thermal desorption spectros-
2
4,19
tron spectroscopy ͑UPS͒,
2
17–20,22
The former gap appears due to the high (1–10 bar͒ pres-
sures and temperatures employed in industrial catalytic pro-
cesses while the latter gap describes the disparity between
oxide-supported cluster catalysts and the corresponding
single-crystal model catalysts traditionally investigated in
surface science. Model catalysts have to be introduced due to
the prohibitively complex structure of real catalysts, which
impedes atomic-scale characterization. A general trend in
surface science has been—and currently is—to approach, in
small steps, real conditions from a direction defined by one
copy ͑TDS͒,
high-resolution electron energy-loss
spectroscopy ͑HREELS͒,1
9,25–27
angle resolved ultraviolet
2
5,28,29
photoemission spectra ͑ARUPS͒,
Rutherford back-
scattering ͑RBS͒,3 nuclear microanalysis ͑NMA͒, reflec-
0,20
20
2
6,22,31,32
tion absorption infrared spectroscopy ͑RAIRS͒,
2
1
x-ray photoelectron spectroscopy ͑XPS͒, work function
1
7,21,23
measurements,
scanning tunneling microscopy
͑STM͒,3
3–35,15
inverse photoemission spectroscopy ͑PES͒,
36
3
7
reflection high energy electron diffraction ͑RHEED͒, sum-
frequency generation ͑SFG͒, microcalorimetry, as well as
high-pressure ͑HP͒ STM. In Sec. IV we will show that our
2
7
38
3
,4
2,5
6
of these gaps or even several directions at once.
At the high pressures and temperatures employed in real
catalysis, new and unexpected phenomena have been re-
ported as compared to what is observed through the conven-
tional surface science UHV approach.6–8 This approach is
used extensively to mimic high-pressure situations where a
high incoming molecular flux facilitates high adsorbate cov-
erages even well above the adsorbate desorption tempera-
tures. The application of UHV data on the other side of the
HP-STM results differ to some extent form the findings of
McIntyre et al.6
Beside the fundamental interest in platinum, the metal
has also been the subject of extensive investigations over the
years due to its widespread use of oxide-supported platinum-
cluster catalysts in, e.g., the automotive three-way catalyst.39
Here platinum has the effect of oxidizing carbon monoxide
and uncombusted hydrocarbons.
Crystalline materials usually expose planar low-Miller-
index facets at low temperatures due to the high energy re-
quired to form structural defects. In the order of 0.25 eV is
a͒Author to whom correspondence should be addressed. Electronic mail:
fbe@phys.au.dk
0021-9606/2003/118(8)/3724/7/$20.00
3724
© 2003 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 147.143.2.5
On: Sun, 21 Dec 2014 00:20:33

J. Chem. Phys., Vol. 118, No. 8, 22 February 2003
CO-induced restructuring of Pt(110)-(1ϫ2)
3725
needed to break nearest-neighbor ͑nn͒ platinum bonds so
very high temperatures ͑Ͼ1080 K͒ are needed for a sizable
fraction of Pt–Pt bonds to be broken spontaneously.40 How-
ever, with the introduction of CO adsorbates, surface rough-
ening proceeds readily even at room temperature. Here we
report on the atomic-scale response of the Pt͑110͒ surface to
CO adsorption at temperatures in the 300–400 K range and
Ϫ9
pressures ranging from 10 to 1000 mbar. Depending on
temperature and CO pressure, the surface structure changes
dramatically. In particular, we observe an extremely rough
equilibrium structure at intermediate coverages, which trans-
forms into a well-ordered saturation structure under high CO
pressure. The seemingly complex behavior of the CO/
Pt͑110͒ system can be explained by a very simple model at
all pressures. Furthermore, we shall see that the structure
observed at high pressures could have been predicted form
UHV data only.
FIG. 1. Missing-row reconstructed Pt͑110͒ surface and the effect of CO
adsorption. ͑a͒ Model of the missing-row reconstructed Pt͑110͒-͑1ϫ2͒ sur-
face. Pt coordination numbers for the perfect surface as well as for the
various defects are also shown. ͑b͒ STM image showing the clean Pt͑110͒-
͑1ϫ2͒ structure. A monatomic step is seen in the upper-left corner ͑141
2
ϫ155 Å ).
up to a total pressure of 1 bar. Ultrapure argon is commer-
cially available ͑no further cleaning is necessary͒ and is used
for backfilling due to its unreactive nature.
Sample temperatures in the range 90–400 K are acces-
sible in our UHV-STMs and the sample temperature can be
varied in the 300–400 K range in the HP-STM by heating
and equilibrating the entire HP cell.
In situ HP-STM proves more difficult than conventional
UHV-STM in that a higher noise level is evident, mainly due
to chemical reaction between gas and tungsten tip, causing
tip changes, or stronger coupling of acoustic noise through
the gas. Nonetheless, we are indeed able with our setup to
record STM images with atomic resolution in both unreac-
tive and reactive gases at pressures up to 1 bar, as illustrated
in Sec. III C.
II. EXPERIMENT
The experimental results presented below were obtained
on an experimental system combining standard UHV equip-
ment, including UHV-STMs, and a high-pressure ͑HP͒ cell
containing a STM. Here in situ STM investigations at pres-
sures up to 1 bar of pure gases or gas mixtures can be carried
out as described in detail in a previous publication.41
In the present context it is important to point out that
special care has been taken to use ultrapure gases. Impor-
tantly, a CO-cleaning system designed to avoid nickel–
carbonyl ͑Ni͑CO͒ ) formation was added. First, exposure of
4
CO to stainless steel components is minimized by using cop-
per tubing. Second, just prior to the inlet to the gold-plated
high-pressure cell, the CO is passed through an activated
carbon filter heated to ϳ400 K. Any Ni-carbonyls will dis-
sociate on the high surface-area powder and the Ni is left
behind. This cleaning system has proven effective in that no
Ni has been detected by AES on the surface under investiga-
tion, even after extended CO exposure.
III. EXPERIMENTAL RESULTS
The Pt͑110͒ surface exhibits the well-known ͑1ϫ2͒
missing-row reconstruction in its clean state,4
2,43
see Fig. 1.
Gas adsorption alters the delicate energy balance between a
bulk truncation and the missing-row reconstruction ͓the total
energy per atom of the ͑1ϫ2͒ structure is only 0.27 meV/
The setup facilitates standard sample cleaning proce-
ϩ
dures ͑30 min 800 V Ar sputtering, 10 min annealing at 800
Ϫ7
44
K in 10
mbar O to remove residual carbon impurities
atom lower than in the ͑1ϫ1͒ structure ͔ so the adsorption
2
of CO induces a lifting of the ͑1ϫ2͒ reconstruction.3
3,15
The
followed by 2 min annealing at 1000 K in UHV͒. Once the
sample is deemed clean by STM, experiments can be carried
out in the UHV chamber at CO pressures up to 10 mbar
͑1ϫ2͒→͑1ϫ1͒ transition is readily observable at the atomic
scale with STM through the local motion of the Pt atoms.
Note that at all coverages below full monolayer coverage
͑␪ϭ1͒, the CO molecules are not detected by STM at room
temperature. This has been attributed to the CO molecules’
Ϫ3
͑this relatively high pressure does not affect the chamber
base pressure significantly͒.
Experiments at pressures up to 1 bar are carried out in
the high-pressure cell equipped with a specially designed
HP-STM. A gate valve is closed between the main chamber
and the high-pressure cell and the cell is pressurized to 1 bar
of pure gas or gas mixtures with a particular CO partial pres-
sure. We always use 1 bar total pressure in order to avoid
electric discharges between the closely-spaced piezocontrol
electrodes. These electrodes are held at potentials of a few
hundred volts relative to each other and total pressures in the
1
–100 mbar range are insufficient to dampen an ionization
FIG. 2. STM movie showing homogeneous nucleation and growth of 4–6
avalanche between the electrodes. Gas mixtures are obtained
by first letting in the desired CO partial pressure, pumping
out the gas inlet line and adding argon through a check valve
2
Ϫ9
membered holes. Size 300ϫ300 Å . Pϭ2•10 mbar. The relative to the
introduction of CO is indicated below each still. The entire movie can be
viewed at our web site: http://www.phys.au.dk/camp/movies/co01.mpg
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 147.143.2.5
On: Sun, 21 Dec 2014 00:20:33

3726
J. Chem. Phys., Vol. 118, No. 8, 22 February 2003
Thostrup et al.
FIG. 5. STM movie recorded at 373 K demonstrating relaxation towards
2
Ϫ9
thermodynamic equilibrium. Size 300ϫ300 Å . Pϭ2•10 mbar. The en-
tire movie can be viewed at our web site: http://www.phys.au.dk/camp/
movies/co02.mpg
critical exposure of about 2 L is needed for hole-creation
onset.
FIG. 3. Appearance of a completely lifted reconstruction at room tempera-
ture. The color scale has been repeated to show the emerging atoms of the
The question now arises whether this rough structure
represents the thermodynamical equilibrium structure. This
can be investigated either by simply keeping the sample tem-
perature at an elevated temperature during CO exposure or
by flash-annealing the sample after room-temperature expo-
sure, thereby allowing the Pt atoms to equilibrate.15
second layer. As expected, these are shifted by half a ͓001͔ spacing, cf. Fig.
2
1
. Image size 138ϫ154 Å .
rapid diffusion and/or the molecules’ frustrated motion,45 i.e.,
the ability of the on-top-adsorbed CO molecule to vibrate
while still staying bound to a Pt atom. These kinds of motion
require very little energy and thus proceed rapidly even com-
pared to the fairly high scanning speed obtainable with our
STMs.
B. CO exposure at elevated temperatures
To resolve this issue, we investigated the structural re-
sponse of the Pt substrate towards CO adsorption at elevated
temperatures, here 373 K. Figure 5 shows the evolution,
which proceeds in two regimes: First, kinetic limitations
dominate through Pt translations akin to those observed at
room temperature. Holes form in the close-packed rows, i.e.,
atoms are displaced from ridges to trough, but the size of the
holes is no longer limited to 4–6 atoms since higher thermal
activation enables more far-reaching Pt displacements. We
note in passing that at 373 K, hole creation commences at a
lower exposure ͑ϳ1 L͒ than at room temperature.
The creation of the rough structure is followed by relax-
ation towards equilibrium. We see how short close-packed Pt
strings line up perpendicular to the close-packed rows, see
last still in Fig. 5. The same kind of equilibrium structure can
be produced by flash-annealing the sample in UHV after
room-temperature exposure, as seen in Fig. 6.15 Again a
channel-like structure lined up in the ͓001͔ direction is
formed. In both cases discussed above, desorption will occur,
as the lower of the two CO desorption peaks lie at ϳ375 K.19
The CO coverage will thus be below saturation coverage
A. CO exposure at room temperature
We find a room-temperature behavior similar to that re-
3
3,34
ported previously.
Homogeneously nucleated ‘‘holes’’ in-
volving 4–6 atoms are created in the close-packed rows. A
STM movie, i.e., a series of sequential STM images,46 Fig. 2,
shows the homogeneous nucleation and growth of these moi-
eties. The resulting structure of room-temperature CO expo-
sure is seen in Fig. 3. Notice the emergence of second-layer
Pt atoms, which become accessible to CO adsorption only
after the reconstruction is lifted. The end result is a very
rough structure with no kind of ordering except a retained
preference for Pt atoms to reside in fourfold hollow sites.
We have studied the evolution of the CO-induced lifting
of the reconstruction by varying the CO background pressure
and counting the number of holes as a function of time in
STM movies. The results are shown in Fig. 4. We see that a
͑
␪ϭ1͒.
FIG. 4. Number of holes as a function of time for different CO background
Ϫ9
2
pressures. Number of holes as a function of time. ᮀ: PCOϭ5.0ϫ10 torr.
FIG. 6. The effect of a flash-anneal to 400 K. ͑a͒ 200ϫ200 Å ; ͑b͒ 500
ϫ500 Å2
Ϫ9
2
छ: PCOϭ1.5ϫ10 torr. Image size: 275ϫ307 Å .
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 147.143.2.5
On: Sun, 21 Dec 2014 00:20:33

J. Chem. Phys., Vol. 118, No. 8, 22 February 2003
CO-induced restructuring of Pt(110)-(1ϫ2)
3727
FIG. 7. Imaging of CO molecules in the ͑2ϫ1͒-p2mg-2CO structure. ͑a͒
2
5
0ϫ50 Å . ͑b͒ Ball model of the structure. The substrate atoms diameter
͑2.78 Å͒ and CO molecule van der Waals diameter ͑3.2 Å͒ are shown in
correct relative proportions. Unit cells of the bare substrate and ͑2ϫ1͒ over-
layer are indicated. The glide line and mirror symmetries belonging to the
p2mg space group are indicated by dashed and full lines, respectively.
It is, however, also possible to form an equilibrated satu-
ration structure in UHV. Cooling down from 600 K to room
Ϫ7
temperature in a 10 mbar CO background creates a ͑2ϫ1͒-
p2mg-2CO structure also found for several other transition-
metal ͑110͒ terminations, e.g., Ni.45 In this case, the CO
overlayer is so densely packed that the molecules’ frustrated
motion is impeded and the CO molecules, opposite to the
low-coverage case, themselves are imaged as also discussed
in the beginning of this section. STM images show a char-
acteristic zig–zag CO structure, identical to that observed
previously by STM,35 covering the entire surface, see Figs.
7͑a͒ and 7͑b͒ for a simple ball model.
From the results presented above, one realizes that a
substrate temperature of 373 K is high enough to sufficiently
invigorate substrate Pt atoms to approach their equilibrium
structure at a given coverage. Intermediate coverage ͑0
Ͻ␪Ͻ1͒ structures are easily accessible but a special treat-
ment has to be implemented to produce the full coverage
structure in UHV.
Ϫ7
FIG. 8. STM images of CO–Pt͑110͒ at different CO pressures: ͑a͒ 10
2
Ϫ6
2
Ϫ5
mbar CO, ͑300ϫ300 Å ), ͑b͒ 10 mbar CO, ͑300ϫ300 Å ); ͑c͒ 10 mbar
CO, ͑1000ϫ1000 Å ); ͑d͒ 10 mbar, ͑700ϫ700 Å ); ͑e͒ 1000 mbar CO
900ϫ900 Å ); and ͑f͒ 1000 mbar CO ͑28ϫ28 Å ). All images were ob-
2
Ϫ2
2
2
2
͑
tained at 373 K.
Another question then arises as to the effect of even
higher CO pressure than the maximum of 10 mbar em-
Ϫ7
ments reported in the following were carried out at 373 K.
This was accomplished by heating the entire HP cell exter-
nally.
ployed above. To this end, our novel HP-STM design allows
us to follow the changes in the atomic-level surface morphol-
ogy at progressively higher pressures up to 1 bar, as dis-
cussed in Sec. II. We are thus able to bridge the pressure gap
at the atomic scale. This will be the subject of the next sec-
tion.
Even though CO desorption takes place at a sample tem-
perature of 373 K, we expect to reach progressively higher
adsorbate coverages when increasing the CO background
pressure. Raising the CO pressure, and thus the incoming
molecular flux, puts the adsorption system in surface-gas
phase equilibrium where a free site created by a desorbed
molecule is quickly reoccupied by an incoming molecule.
This is where high-pressure experiments differ from UHV
experiments at low temperature. At low temperature, no re-
plenishment from the gas phase is needed to maintain a high
adsorbate coverage.
C. Elevated-temperature high-pressure CO exposure
The entire pressure range from UHV pressures to 1 bar
Ϫ3
could be covered by going to 10 mbar in the UHV cham-
Ϫ2
ber and from 10 mbar to 1 bar in the HP cell. As described
in Sec. II, a high-pressure experiment is always conducted at
Figure 8 shows the resulting structures at a range of CO
partial pressures. We see how at low pressures the displaced
Pt atoms order into the same channel-like structure ordered
along the ͓001͔ direction described above. The ͓001͔ steps
are characterized by a low kink density. Higher CO pres-
sures, however, result in kink formation, and consequently
1
bar total pressure composed of a given CO partial pressure
backfilled with Ar. The CO partial pressure can thus not be
adjusted while performing STM measurements so only batch
experiments are possible. To ensure that only thermo-
dynamic-equilibrium structures are created, all HP experi-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 147.143.2.5
On: Sun, 21 Dec 2014 00:20:33

3728
J. Chem. Phys., Vol. 118, No. 8, 22 February 2003
Thostrup et al.
step edges smoothen, see Fig. 8͑d͒. The transition from kink-
free step edges is found to take place at a CO partial pressure
Ϫ3
of ϳ10 mbar where also the first hint of the high-coverage
͑2ϫ1͒ p2mg zig–zag structure is observed ͑not shown͒. At a
CO pressure of 1 bar the channel-like structure has been
¯
replaced by islands elongated in the ͓110͔ direction. A
zoom-in ͓in Fig. 8͑f͔͒ reveals the ͑2ϫ1͒-p2mg zig–zag struc-
ture also observed in UHV on both first and second-layer
areas. As will be discussed further below, it thus appears that
the CO structure observed at 373 K and high pressures from
Ϫ3
10
mbar to 1 bar is indeed identical to the one observed at
low pressure.
FIG. 9. Ball model of the kink formation process showing the coordination
number of the Pt atoms involved. ͑Left͒ Kink-free step edge. When the CO
coverage is less than 1 ML, CO is only adsorbed on the low-coordinated
sites ͑Ͻ8 nearest neighbors͒, and the kink-free configuration is therefore the
energetically favored configuration. ͑Right͒ Kink formation. When the CO
coverage is close to 1 ML the energy difference between the two configu-
rations is close to zero and the higher entropy of the kinked step edge is
therefore favored at finite temperatures.
IV. DISCUSSION
To disentangle the multitude of details presented above,
the following discussion is divided into two natural parts
treating low and high-coverage structures, respectively.
The observed structures and accompanying transforma-
tions depend on CO coverage in a fairly complex way. How-
ever, as will be shown below, a simple model in which the
CO binding energy depends in a linear fashion on the coor-
dination number of the Pt atom to which it binds, is able to
account for all the STM results.
Fermi energy. The CO adsorption energy depends strongly
on the relative positions of the molecular orbitals involved in
bonding ͑here the 5␴ HOMO and 2␲* LUMO͒ and the en-
ergy of the metal d-states. The closer in energy, the stronger
are the orbital overlap and adsorbate binding. In absolute
numbers, CO binds more strongly to a 5-coordinated adatom
than to a 7-coordinated ridge atom by 0.45 eV,15 incidentally
exactly equal to the energy required to rupture two Pt nn
bonds. This explains the preference and ability of CO to
create its own low-coordinated adsorption site and lift the
͑1ϫ2͒ reconstruction.
As an aside to the above energy considerations, we men-
tion that most previous studies of the CO-induced lifting of
the Pt͑110͒-͑1ϫ2͒ reconstruction,32 have been interpreted in
terms of a CO adsorption energy difference between the
͑1ϫ2͒ and ͑1ϫ1͒ clean Pt substrate phases. This is concep-
tually different from the discussion above where the driving
force behind the lifting process is interpreted in terms of
single CO molecules gaining binding energy. It should be
noted, however, that the microscopic explanation presented
here in the end has the same ͑macroscopic͒ thermodynamic
implications as the broader view proposed earlier.
A. Low-coverage behavior
The overall effect of CO exposure is the creation of low-
coordinated Pt atoms. The van der Waals diameter of CO is
ϳ3.2 Å, significantly larger than the Pt nn distance of 2.78
Å, so ␪ϭ1/4 ͑one CO per two top-layer Pt atoms͒ is the
highest coverage possible when imposing the criterion of
preferred on-top adsorption in the top Pt layer. This is in
accord with previous studies in that at RT lifting is found to
commence when ␪Ϸ0.2. The concept of a critical coverage is
also found in Fig. 4 where we measure a critical exposure of
ϳ2 L since the CO molecule elude detection at submono-
layer coverage.
For CO adsorption to continue, CO/Pt complexes have
to ‘‘squeeze out’’ from the ridges to accommodate more CO
and indeed previous studies show that the lifting process is
completed—no ͑1ϫ2͒ areas are detected—when ␪Ϸ1/2.
These steps only require displacements ranging a few lattice
spacings and are readily observed at room temperature. As
we have seen above, however, at room temperature the re-
sulting structure is metastable and higher temperatures are
needed to form the equilibrium structure. As shown in Sec.
III B and a previous publication,15 a temperature of 373 K is,
however, sufficient to facilitate the creation of thermally
equilibrated structures.
B. High-coverage behavior
The considerations above only consider coordination
number changes in top-layer Pt atoms. When the ͑1ϫ2͒ re-
construction is lifted, however, the next Pt layer is made
available for CO adsorption. At most half of the second layer
is exposed because the top layer Pt atoms effectively consti-
tute half a monolayer when compacted into the ͑1ϫ1͒ phase.
The exposed part of the second layer is a full ͑1ϫ1͒ layer
with a Pt coordination number of 7; higher coordination
numbers ͑8–10͒ are found at step edges, as seen in Figs. 1͑a͒
and 9. Even while the lifting process occurs, adsorption must
take place on both first and second-layer atoms, initially at
coordination number 7 sites. At ever higher coverage, ad-
sorption is forced to take place at less energetically favored
8–10-coordinated sites, see Figs. 1͑a͒ and 9. Here we recog-
nize the driving force behind the line-up of short, close-
The driving force behind the lifting of the ͑1ϫ2͒ recon-
struction can be explained within the so-called d-band
1
3
model. Density functional theory ͑DFT͒ calculations show
that CO binds more strongly to low-coordinated metal atoms.
The qualitative trends in this observation are simple to un-
derstand: When any atom lowers its coordination number,
the orbital overlap with neighboring atoms is lowered as
well—in other words, the atom turns more ‘‘free-atomlike.’’
Lower orbital overlap with neighboring atoms results in band
narrowing and in order to maintain the band filling degree,
the d-band center shifts upwards in energy, towards the
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 147.143.2.5
On: Sun, 21 Dec 2014 00:20:33

J. Chem. Phys., Vol. 118, No. 8, 22 February 2003
CO-induced restructuring of Pt(110)-(1ϫ2)
3729
packed Pt rows along the ͓001͔ direction observed at inter-
mediate coverages ͑see Fig. 6͒: The lined-up structure does
not contain a lower number of step ͑coordination number
5
–6 in the top layer͒ atoms than would a structure with no
kind of ordering of the short close-packed rows. It rather
minimizes the average coordination number of occupied Pt
sites in the second layer.
For coverages close to saturation, the energy gained in
the kink-free channel structure vanishes due to adsorption on
second-layer, high-coordinated Pt atoms. Entropic effects
͑
higher disorder͒ are then manifested as kinked or ‘‘rounded’’
Ϫ2
step edges, as was observed at a pressure of 10 mbar in
Fig. 8͑d͒. The formation of a kink is illustrated in Fig. 9,
where we recognize the higher average coordination number
of a kinked structure. At even higher pressures, channels coa-
lesce to form extended ͑1ϫ1͒ substrate areas with the char-
acteristic zig–zag ͑2ϫ1͒-p2mg-2CO pattern seen in Figs. 7
and 8͑f͒.
FIG. 10. Step density calculated from Monte Carlo simulations at 300 K
based on DFT parameters from a previous publication ͑Ref. 15͒ shown as a
function of CO coverage. A step atom is defined as an atom with coordina-
tion number 5 or 6. Reproduced from Ref. 15.
Our results shed new light on a previous high-pressure
6
STM study of the CO/Pt͑110͒ system. This being a pioneer
experiment in the area of high-pressure STM, the design did
not enable atomic resolution. Furthermore, McIntyre et al.
made use of a transfer cell between crystal cleaning and HP
imaging which raised the background pressure during trans-
͑
1ϫ1͒ substrate areas with a close-packed ͑2ϫ1͒-p2mg-2CO
layer on top. This transition to a ͑1ϫ1͒ substrate is also
explained by the strong dependence of the CO binding en-
ergy on the coordination number of the Pt atom to which it
binds.
Ϫ5
fer to 10 mbar. After heating at 425 K for 5 h in 1 bar CO,
McIntyre et al. observed large flat terraces displaying no
missing-row reconstructions separated by multiple-height
steps and groups of steps with a corrugation amplitude of 42
Å. Due to the lack of atomic resolution, no further details
could be extracted of the atomic structure on the flat terraces.
We also observe extended terraces at high pressures and
the mechanism for their creation was presented above.
Atomically-resolved images of the ͑2ϫ1͒-p2mg-2CO struc-
ture created on the flat terraces allow us to confidently con-
clude that the pressure gap has indeed been bridged for this
system. Contrary to the findings of McIntyre et al. we do,
A temperature of 373 K was shown to facilitate the for-
mation of equilibrated structures. At intermediate coverages,
a rough structure was observed but the Pt substrate heals at
saturation ͑␪ϭ1͒. At pressures above ϳ10 mbar, a satu-
rated ͑2ϫ1͒-p2mg-2CO layer adsorbed on a ͑1ϫ1͒ substrate
Ϫ3
is formed. This structure can also be formed in UHV by
Ϫ7
cooling from 600 K in a 10 mbar CO background. As a
common feature at all coverages, substrate transformations
are found to proceed through minimal displacements. In par-
ticular, a given Pt atom is found to remain on its native
terrace.
however, not observe any large-scale changes, i.e., the well-
known mesoscopic ‘‘corrugated-iron’’ structure4
7,48
In conclusion, we emphasize that no new CO phase was
observed when bridging the pressure gap, i.e., by raising the
CO pressure to 1 bar. The Pt͑110͒ surface thus responds
identically to CO under UHV and high-pressure conditions.
For this system, however, raising the pressure is not exactly
equivalent to lowering the temperature since at low tempera-
ture the rearranging Pt substrate atoms would not be suffi-
ciently thermally activated to form the equilibrium structure
at a given coverage. The high-coverage equilibrium structure
can, however, be formed under UHV conditions by cooling
the sample in a CO background pressure as discussed above.
In the present case, UHV data are adequate to predict the
surface structure at high pressure and elevated temperature.
In general one must thus take into account pertinent activa-
tion barriers, e.g., substrate equilibration when making con-
jectures as to the high-pressure response of an adsorption
system.
is not
modified even after overnight exposure to 1 bar CO at 373
K. The level of gas cleanliness may be the point where the
study of McIntyre et al. differs from the present study and
could explain the discrepancies.
V. CONCLUSIONS
The classic prototype system for adsorbate-induced sub-
strate transformations, CO/Pt͑110͒, was investigated in detail
with STM in the 300–400 K temperature range and
Ϫ9
10
–1000 mbar pressure range. The overall behavior is
summarized well by the step-density plot in Fig. 10: At first,
adsorption at top-layer Pt sites dictates a step density in-
crease through the adsorbate’s preference for low-
coordinated sites. When second-layer Pt atoms become ex-
posed, adsorption continues here as well, and adsorption on
high-coordinated second-layer sites is minimized by forming
the low-kink-density channel structure. This continues until a
CO coverage of ␪Ϸ0.7 where adsorption on high-
coordinated second-layer Pt sites removes the preference for
the low-kink-density structure. Finally, the energetically fa-
vored structure at saturation consists of low-step-density
ACKNOWLEDGMENTS
We acknowledge the financial support from the Danish
National Research Foundation through the Center for
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 147.143.2.5
On: Sun, 21 Dec 2014 00:20:33

3730
J. Chem. Phys., Vol. 118, No. 8, 22 February 2003
Thostrup et al.
2
2
2
2
4
5
6
7
Atomic-Scale Materials Physics ͑CAMP͒ and from the Dan-
ish Research Council through the program ‘‘New Design
Strategies for Catalysts.’’
S. R. Bare et al., Surf. Sci. 120, 367 ͑1982͒.
P. Hoffmann, S. R. Bare, and D. A. King, Phys. Scr., T T4, 118 ͑1983͒.
S. R. Bare, P. Hofmann, and D. A. King, Surf. Sci. 144, 347 ͑1984͒.
C. Kl u¨ nker, M. Balden, S. Lehwald, and W. Daum, Surf. Sci. 360, 104
͑
1996͒.
1
28
G. Ertl, Surf. Sci. 299Õ300, 742 ͑1994͒.
P. Hofmann, S. R. Bare, and D. R. King, Solid State Commun. 42, 645
͑1982͒.
D. Rieger, R. D. Schnell, and W. Steinmann, Surf. Sci. 143, 157 ͑1984͒.
T. E. Jackson et al., J. Phys. C 15, L99 ͑1982͒.
R. K. Sharma, W. A. Brown, and D. A. King, Chem. Phys. Lett. 291, 1
͑1998͒.
R. K. Sharma, W. A. Brown, and D. A. King, Surf. Sci. 414, 68 ͑1998͒.
T. Gritsch, D. Coulman, R. J. Behm, and G. Ertl, Phys. Rev. Lett. 63, 1086
͑1989͒.
T. Gritsch, D. Coulman, R. J. Behm, and G. Ertl, Appl. Phys. A: Solids
Surf. 49, 403 ͑1989͒.
J. Wintterlin and R. J. Behm, in Scanning Tunneling Microscopy I, Vol. 20
in Springer Series in Surface Sciences, 2nd ed., edited by H.-J. G u¨ ntherodt
and R. Wiesendanger ͑Springer-Verlag, New York, 1994͒, pp. 39–82.
E. Bertel, N. Memmel, G. Rangelov, and U. Bischler, Chem. Phys. 177,
337 ͑1993͒.
S. Schwegmann, W. Tappe, and U. Korte, Surf. Sci. 334, 55 ͑1995͒.
C. E. Wartnaby, A. Stuck, Y. Y. Yeo, and D. A. King, J. Phys. Chem. B
100, 12483 ͑1996͒.
R. J. Farrauto and R. M. Heck, Catal. Today 51, 351 ͑1999͒.
I. K. Robinson, E. Vlieg, and K. Kern, Phys. Rev. Lett. 63, 2578 ͑1989͒.
E. Lægsgaard et al., Rev. Sci. Instrum. 72, 3537 ͑2001͒.
P. Fery, W. Moritz, and D. Wolf, Phys. Rev. B 38, 7275 ͑1988͒.
T. Gritsch, D. Coulman, R. J. Behm, and G. Ertl, Surf. Sci. 257, 297
͑1991͒.
2
H.-J. Freund et al., Top. Catal. 15, 201 ͑2001͒.
3
4
5
¨
29
30
31
L. Osterlund et al., Phys. Rev. Lett. 86, 460 ͑2001͒.
S. Helveg et al., Phys. Rev. Lett. 84, 951 ͑2000͒.
J. Wang, C. Y. Fan, K. Jacobi, and G. Ertl, J. Phys. Chem. B 106, 3422
͑2002͒.
6
7
8
32
B. J. McIntyre, M. Salmeron, and G. Somorjai, J. Vac. Sci. Technol. A 11,
964 ͑1993͒.
X. Su, P. S. Cremer, Y. R. Shen, and G. A. Somorjai, Phys. Rev. Lett. 77,
858 ͑1996͒.
3
3
3
3
4
5
1
3
J. A. Jensen, K. B. Rider, M. Salmeron, and G. A. Somorjai, Phys. Rev.
Lett. 80, 1228 ͑1998͒.
P. Stoltze and J. K. No”rskov, Phys. Rev. Lett. 55, 2502 ͑1985͒.
D. W. Goodman, J. Phys. Chem. 100, 13090 ͑1996͒.
V. P. Zhdanov, Surf. Sci. 257, 344 ͑1991͒.
9
0
1
11
36
37
1
1
2
3
A. V. Myshlyavtsev and V. P. Zhdanov, Langmuir 9, 1290 ͑1993͒.
B. Hammer, Y. Morikawa, and J. K. No”rskov, Phys. Rev. Lett. 76, 2141
38
͑1996͒.
1
1
1
1
4
5
6
7
Q. Ge and D. A. King, J. Chem. Phys. 111, 9461 ͑1999͒.
P. Thostrup et al., Phys. Rev. Lett. 87, 126102 ͑2001͒.
R. M. Lambert, Surf. Sci. 49, 325 ͑1975͒.
39
40
41
C. M. Comrie and R. M. Lambert, J. Chem. Soc., Faraday Trans. 2 72,
42
1
659 ͑1976͒.
1
1
2
2
8
9
0
1
43
S. Ferrer and H. P. Bonzel, Surf. Sci. 119, 234 ͑1982͒.
P. Hofmann, S. R. Bare, and D. R. King, Surf. Sci. 117, 245 ͑1982͒.
T. E. Jackson et al., Surf. Sci. 120, 389 ͑1982͒.
N. Freyer, M. Kiskinova, G. Pirug, and H. P. Bonzel, Appl. Phys. A:
Solids Surf. 39, 209 ͑1986͒.
44
45
K.-D. Shiang and T. T. Tsong, Phys. Rev. B 51, 5522 ͑1995͒.
P. Sprunger, F. Besenbacher, and I. Stensgaard, Surf. Sci. 324, L321
͑1995͒.
F. Besenbacher, Rep. Prog. Phys. 59, 1737 ͑1996͒.
P. Hanesch and E. Bertel, Phys. Rev. Lett. 79, 1523 ͑1997͒.
K. Swamy, E. Bertel, and I. Vilfan, Surf. Sci. 425, L369 ͑1999͒.
2
2
2
3
46
B. E. Hayden, A. W. Robinson, and P. M. Tucker, Surf. Sci. 192, 163
47
͑1987͒.
48
R. Imbihl, S. Ladas, and G. Ertl, Surf. Sci. 206, L903 ͑1988͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 147.143.2.5
On: Sun, 21 Dec 2014 00:20:33