JOURNAL OF CHEMICAL PHYSICS

VOLUME 120, NUMBER 19

15 MAY 2004

High density gradients in the 3 3 R 30-CO layer on Ru0001

P. Jakoba)
Physik-Department E20, Technische Universitat Munchen, 85747 Garching, Germany

Received 8 September 2003; accepted 24 February 2004

The coverage regime just beyond 0.33 ML, representative of a perfectly ordered ()
))R30-CO layer on Ru0001, has been investigated using infrared-absorption spectroscopy.
Different isotopic mixtures of CO have been employed to derive a profound understanding of
structural properties of such layers. It is found that extra CO molecules incorporated into the ()
))R30-CO layer affect their nearest neighbor molecules only, and the associated density
gradient extends over no more than a few angstroms. Contrary to existing belief, the model system
CO on Ru0001 does not represent a case of an unusually shallow adsorption potential corrugation.
Rather, CO experiences an exceptionally strong site preference when adsorbed on Ru0001.
Annealing causes the local distortion of the overlattice to propagate laterally, most probably in a
density wave-like manner. Incipient motion on the atomic scale thereby has been detected by means
of isotopic labeling of inequivalent molecules within the high density areas. All major conclusions
are based on observations of isotopically labeled minority CO species which feature negligible
dynamical lateral coupling. The majority CO species, on the other hand, provide laterally averaged,
unspecific information on the status of the layer. 2004 American Institute of Physics.
DOI: 10.1063/1.1710852

INTRODUCTION

summarize the structural properties of CO on Ru0001 at a

coverage CO0.33 ML we state:

The ()))R30 CO layer on Ru000113 is a very

well-understood system with stable long range order. It is
therefore ideal to investigate the effect of point defects, as
produced by extra particles, starting with this layer. The main
focus in this study will be the lateral dimensions of such
local distortions of the two-dimensional 2D lattice. Naturally, the primary effect of an extra particle on the existing
adsorbate superlattice will be a structural distortion. In addition to the intrinsically defect related heterogeneity the electronic structure of nearby preadsorbed species will also be
subject to some change chemical effect. Such static lateral
interactions decay rapidly with distance from the defect, very
much like chemical shifts in nuclear magnetic resonance
NMR spectra.4 A sensitive tool to detect tiny variations in
the electronic structure of adsorbates is vibrational spectroscopy using optical methods.5,6 In addition to the abovementioned static short range interactions a dynamic long
range coupling of the oscillators dipole moments occurs
which can be used to extract complementary information.
The literature related to the ()))R30 CO overlayer
on Ru0001 is extensive. Whereas the early papers focus on
the general characterization of the layer in terms of geometry
and lateral ordering,1,2,7,8 phase transitions,9 vibrational
spectroscopy,3,10 and electron stimulated desorption of ions
and their angular distribution ESDIAD,1114 the more recent work simply makes use of this molecular adsorbate
layer as a highly homogeneous model system to study the
various aspects of dynamics in adsorbate layers.1518 To

i
ii

iii

Little is known about the microscopic structure of the

CO/Ru0001 layer at coverages CO0.33 ML.
i

ii

Permanent address: Fachbereich Physik der Philipps-Universitat Marburg,

Renthof
5,
35032
Marburg,
Germany.
Electronic
mail:
peter.jakob@physik.uni-marburg.de

0021-9606/2004/120(19)/9286/11/$22.00

CO exclusively occupies on-top sites and is oriented

perpendicular to the surface,8,19
neighboring CO molecules experience an attractive
interaction for d)d RuRu next nearest neighbor
sites.20,21 For nearest neighbor sites a strongly repulsive force exists, while for larger distances d)
d RuRu the interactions are negligible. This is why,
depending on temperature, ()))R30 CO islands
may form already at CO0.05 ML; and
at CO0.33 ML the long range order is very stable
and is retained up to the onset of desorption (T
400 K). 9 It has been speculated that the strong repulsive nearest neighbor interactions are responsible
for this particular behavior. The strong localization of
CO within the ()))R30 lattice leads to extremely high prefactors in thermal desorption spectroscopy TDS.20,22

9286

From high resolution electron energy loss spectroscopy HREELS and infrared absorption spectroscopy
IRAS it has been concluded that the linear bond of
adsorbed CO is retained at all coverages.3,10 Specifically, a single band has been reported which shifts
continuously from 2030 to 2060 cm1 as CO increases 0.330.68 ML.
Complex LEED patterns (2)2))R30 or (5)
5))R30 overstructures suggest that the arrangement of CO molecules within the large unit cells and
their local binding geometry must be nonuniform.13,7
2004 American Institute of Physics

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

iii
iv

A sudden drop of the thermal desorption prefactor at

CO0.33 ML has been interpreted as an enhanced
lateral mobility of CO.20,23
Quantitative LEED studies found that the ()
))R30 LEED spots quickly loose intensity at CO
coverages beyond 0.33 ML and they are gone at
CO0.42 ML. 9

In this paper we investigate the coverage regime CO

0.33 ML CO , with CO0.01 0.06 ML. 24 Extra
12 16
C O or 13C16O molecules in excess of 0.33 ML were adsorbed at low T35 K such that they more or less remain at
their landing positions until annealing enhances mobility
within the layer, leading to a randomization of the isotopic
species deposited sequentially. The respective quantities of
extra CO, CO , have been calibrated using TDS with the
peak at 470 K, associated with desorption of the ()
))R30 CO layer ( CO0.33 ML), serving as a reference. This way the error margins of CO could be kept low.
Vibrational coupling effects in an inhomogeneous
environment

Vibrational frequencies are known to display characteristic shifts in the vicinity of extra species, structural defects,
or upon substrate lattice strain.6,25,26 In this work IR spectroscopy of naturally abundant 13C16O 1.1% has been used
to accurately and sensitively probe such chemical shifts of
CO vibrational-modes. The term chemical shift summarizes
interactions due to electron correlations among the coadsorbates, either directly adsorbateadsorbate or mediated by
the substrate adsorbatesubstrateadsorbate. What makes
things complicated in dense layers is that, besides static
chemical shifts, adsorbateadsorbate interactions and their
influence on vibrational modes may also have a dynamic
origin.27 Such a dynamical interaction of two oscillators by
means of their oscillating dipole fields displays a 1/r 3 dependence and additionally requires a resonance condition to
be fulfilled which is naturally the case for identical species
with identical binding geometries. In nondilute layers specific information on the chemical environment of adsorbates
is difficult to extract from vibrational modes subject to strong
dynamical coupling, i.e., associated with the majority 12C16O
species. Lateral coupling quite naturally leads to the formation of delocalized vibrational modes;28,29 local variations of
the chemical environment of a surface species then will be
completely averaged out. The standard procedure to distinguish static and dynamic influences on the vibrational frequency therefore is to determine the respective line positions
of two isotopically labeled species with identical binding
geometry, the one being strongly dilute minority species,
surrounded by molecules of the majority type.29 For CO the
1.1% natural abundance of 13C16O represents an ideal admixture to determine chemical frequency shifts largely independent from dynamical shifts, which predominantly affect the
majority regular 12C16O molecules. The extraction of specific information on the chemical environment of adsorbates
requires vibrational modes to be localized. This is easily obtained in weakly heterogeneous systems, produced either by
adding chemical impurities or, as in this work, using iso-

High density gradients in the CO layer on Ru(0001)

9287

topic labeling. For homogeneous systems dynamic localization may be attained through anharmonic coupling among
vibrational modes.30,31
Strictly speaking, dynamical dipole coupling for the isotopically labeled minority species is not exactly zero. First,
their spatial separation is large but not infinite, so that resonant coupling among identical minority molecules does occur producing a weak blueshift; second, and more important, decoupling of the minority species from the vibrational
motion of the surrounding majority species is complete only
for very large isotopic shift factors, which is usually not
achievable. This is why the vibrational bands of the minority
13 16
C O species are in fact slightly influenced by the surrounding regular CO molecules15 causing i an intensity
transfer 29 in favor of the modes of the majority 12C16O
molecules, and ii a slight redshift of the 13C16O derived
band due to antiphase motion of neighboring regular 12C16O
molecules when 13C16O is vibrationally excited. These effects are larger the closer the two CO isotopic species, both
spatially and spectrally.

EXPERIMENT

Infrared absorption spectroscopy IRAS has been performed in an UHV chamber base pressure 5
1011 mbar) equipped with additional facilities for low
energy electron diffraction LEED, thermal desorption spectroscopy TDS, and x-ray photoemission spectroscopy
XPS. The Fourier transform IR spectrometer Bruker IFS
66v used in conjunction with a MCT mercury cadmium
telluride, cutoff at 600 cm1 or an InSb indium antimonide,
cutoff at about 1800 cm1 detector is highly stable and
evacuable. For IR measurements, the sample was placed in a
specially designed cell which allows simultaneous gas dosing via a microcapillary array and IRAS data taking. In addition, a titanium sublimator has been installed in the IR cell
in order to keep contaminations low. The Ru crystal was
oriented within 0.1 of the 0001 direction. The surface
was cleaned by Ar ion-sputtering and oxygen cleaning
cycles. The sample could be cooled to 30 K and heated resistively to 1570 K using a computer controlled power supply.

RESULTS AND INTERPRETATION

The starting point of our experiments will always be a

perfect ()))R30 layer of CO on Ru0001. Its isotopic composition, however, has been varied in order to
properly assign the various spectral features appearing after
adsorbing small amounts of extra CO. This latter process is
usually performed at low temperatures in order to freeze out
the status of the resulting layer for in situ spectral analysis.
Frequency shifts of the internal CO stretch mode
( CO) associated with isotopic substitution of either
12
C 13C or 16O 18O amount to about 45 cm1 the individual shifts have been determined experimentally32,33. In
addition, line positions may change if the local environment
is modified; these chemical shifts can be as large as 50

9288

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

P. Jakob

FIG. 1. IR absorption spectra obtained from a well-ordered ()))R30 12C16O/Ru(0001) layer, a, which has been exposed to 0.01 ML 12C16O at T
35 K, b. Annealing to about 400 K desorbs postadsorbed CO so that the original spectrum is regained, c. The CO stretch mode of the 12C16O majority
species is displayed in panel A. Panel B shows the spectral region of the 13C16O minority species 1.1% natural abundance at an enlarged vertical scale. In
panel B the ratio of spectrum b and those at CO0.33 ML have been calculated in order to highlight the spectral changes associated with extra CO within
the perfect ()))R30 CO overlayer d. All spectra were obtained at T35 K at a spectral resolution of 2 cm1 using an InSb detector cut off at about
1800 cm1.

cm1, depending on the distance and nature of the neighboring species.34,35 For altered adsorption site geometries the
associated frequency shifts are much larger.
In Fig. 1, a well-ordered ()))R30 CO/Ru0001
layer, a, has been exposed to an additional small amount of
12 16
C O ( CO0.01 ML) at T35 K, b. The layer then
has been annealed to about 400 K to desorb CO in excess of
0.33 ML, so that the original layer is restored, c.20 Panel A
shows the CO stretch mode of the 12C16O majority species

which is subject to considerable dynamical lateral coupling,

inducing a dispersion on the order of 50 cm1. In panel B,
the spectral region of the 13C16O minority species is
depicted.36 Contrary to the delocalized nature of the CO
mode of the majority CO species in panel A, the respective
vibrational modes associated with the minority CO molecules are more localized due to their low abundance and
accordingly weak dynamical lateral coupling.
Both types of mode therefore behave very differently

FIG. 2. IR absorption spectra obtained from a well-ordered ()))R30 12C16O/Ru(0001) layer, a, which has been exposed to 0.0150.02 ML 12C16O at
T35 K, b. The original layer in a contained a 10% fraction of 13C16O so that the associated bands at about 1941 cm1 become more prominent. Annealing
to about 400 K desorbs postadsorbed CO in excess of 0.33 ML, c. The CO stretch mode of the 12C16O majority species is displayed in panel A. Panel B
shows the spectral region of the 13C16O minority species at an enlarged vertical scale. In panel B the ratio of spectrum b and those at CO0.33 ML has been
calculated in order to highlight the spectral changes associated with extra CO within the perfect ()))R30 CO overlayer d. All spectra were obtained
at T35 K at a spectral resolution of 2 cm1 using an InSb detector.

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

under the influence of local heterogeneities as produced by

extra CO molecules postadsorbed into the perfect ()
))R30 layer Fig. 1b. In panel A, we observe a continuous blueshift of CO( 12C16O) 2030.32032.4 cm1 .
Due to the heterogeneity introduced by CO in excess of 0.33
ML this band is broadened and developed a tail towards the
low frequency side. Since the change of the spectral line
shape is severely influenced by lateral dipole coupling it cannot be interpreted in an easy and straightforward way.
In panel B, an additional band at 1960.5 cm1 is identified after 0.01 ML 12C16O has been added to the ()
))R30 CO layer. We note that, besides the growth of
this new band, the CO( 13C16O) mode at 1941 cm1 slightly
shifts to 1941.3 cm1 and, more importantly, loses intensity
minus 20%; in addition, a weak feature at 1925.3 cm1
appears. In order to highlight the spectral changes the ratio of
spectra b with respect to a or c is plotted in an 5
enlarged vertical scale, d; more precisely we calculated
2(b)/ (a)(c) .
Before we start assigning the various modes in Fig. 1,
we present a very similar experiment Fig. 2. Here the original ()))R30 CO/Ru0001 layer has been enriched
with 13C16O so that its abundance was about 10%. As a consequence the dynamical lateral interaction among the 12C16O
molecules is slightly reduced, leading to a redshift more
precisely, it is a reduced blueshift of CO( 12C16O),
2030.32029 cm1 . Most significantly, the CO( 13C16O)
band at 1941 cm1 is now much more intense compared to
Fig. 1 and it has received a sideband at 1944.5 cm1 due to
the coincidental neighborhood of two 13C16O molecules and
the associated dynamical frequency shift.31 An 11% fraction
of 13C18O within our 13C16O gas manifold generates yet another band at 1895.5 cm1; its total amount is only 0.003 ML
so that coincidental pairing of these isotopic species remains negligible.
Upon adsorption of small amounts of 12C16O at 35 K
CO0.015 0.02 ML, Fig. 2b the bands at about 1961
and 1925 cm1 grow, as has been observed already in Fig. 1.
However, while the intensity of the 1961 cm1 band remains
similar to Fig. 1b, the larger fraction of 13C16O within the
original ()))R30 12C16O layer of Fig. 2a causes the
band at 1925 cm1 to grow substantially. This leads us to
conclude that the 1961 cm1 band must be associated with
the majority 12C16O entity, while the mode at 1925 cm1 is
due to minority 13C16O. In both cases the surrounding 12C16O
molecules at ())) lattice sites play a crucial role which
will be further discussed below.
In agreement with the observations in Fig. 1, the original
bands of minority 13C16O as well as of 13C18O both occupying ())) lattice sites lose intensity when extra 12C16O
is adsorbed. Annealing to about 400 K to desorb CO in excess of 0.33 ML, c, reverses this effect, not to 100%, since
the relative content of 13C16O and 13C18O within the layer
has been reduced due to the extra 0.0150.02 ML 12C16O
dose; this observation actually proves that the entire layer
i.e., the various isotopically labeled species becomes intermixed and randomized before desorption occurs. The spectral ratio 2(b)/ (a)(c) is again displayed in d. When
comparing Figs. 1d and 2d, we notice that the ratio of the

High density gradients in the CO layer on Ru(0001)

9289

FIG. 3. IR absorption spectra obtained from a well-ordered ()

))R30 13C16O/Ru(0001) layer, a, which has been exposed to 0.03 and
0.06 ML 13C16O at T115 K, b and c, respectively. Note that the used
13 16
C O gas contained an 11% fraction of 13C18O. Shown is the spectral
region of the 13C18O minority species which allows direct comparison to the
spectra in Fig. 2, after the isotopic frequency shift of about 45 cm1 has
been taken into account. All spectra were obtained at T115 K at a spectral
resolution of 4 cm1 using a MCT detector.

intensity loss of the 1941 band with respect to the gain of the
1925 cm1 band is independent of the absolute 13C16O content. It is therefore concluded that a conversion from 1941
cm1 associated with 13C16O in a ()))R30 environment to 1925 cm1 ( 13C16O modified by extra 12C16O, most
likely at nearest neighbor sites occurs. From a quantitative
analysis of the intensity loss of the 1941 cm1 band versus
the amount of extra CO in excess of 0.33 ML from TDS we
find that 0.02 ML CO converts about 0.08 ML of the original
()))R30 CO layer to the modified state, i.e., one
postadsorbed CO molecule seems to affect about four preadsorbed CO molecules.
Next we look into the postadsorption of small amounts
of 13C16O with an 11% fraction of 13C18O) into a ()
))R30- 13C16O/Ru(0001) layer at 115 K. For this set of
spectra the spectral resolution was 4 cm1 and the splitting
of the 13C18O minority band due to 13C18O pairs could not
be resolved. Adding 0.03 or 0.06 ML 13C16O to a wellordered ()))R30 13C16O layer Fig. 3a leads to the
spectra in Figs. 3b and 3c, respectively. The characteristic
modes and spectral changes are similar to those of Fig. 2,
e.g., the band at 1878 cm1 equivalent to 1925 cm1 in Fig.
2b is due to 13C18O in the neighborhood of extra 13C16O
molecules. This band is barely visible in Fig. 2d 1881
cm1 band which contained a mere 1% fraction 13C18O in
the original ()))R30 layer. The mode at 1917 cm1 is
assigned to 13C16O of the original ()))R30 layer close
to postadsorbed 13C16O and it is equivalent to the band at
1961 cm1 in Fig. 2. Note that the 1917 cm1 band in Fig.
3b is different from the 1925 cm1 band observed in Fig.
2b, despite their closeness in frequency and both originating from 13C16O molecules: While the 1917 cm1 mode is
associated with a layer containing 90% 13C16O, the 13C16O
entity amounts to only 10% or less with 90% 12C16O)
when the 1925 cm1 band is observed. The different frequencies are due to a dissimilar dynamical lateral coupling of

9290

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

P. Jakob

FIG. 4. IR absorption spectra of the

thermal evolution of a ()))R30
CO/Ru0001 layer, a, which has
been exposed to 0.0250.03 ML
13 16
C O at T ads35 K, b. The layer
has been annealed to increasingly
higher temperatures: c 75 K; d 100
K; e 128 K; and f 400 K. The CO
stretch mode of the 12C16O majority
species is displayed in panel A. Panel
B shows the spectral region of the
13 16
C O and 13C18O minority species at
an enlarged vertical scale. All spectra
were obtained at T35 K at a spectral
resolution of 2 cm1 using an InSb detector.

13 16

C O with neighboring molecules of the ()))R30

CO layer. According to dynamical dipole coupling
theory15,28,29 a redshift is generally induced in case a molecule adsorbed on a metal surface is surrounded by oscillators with higher frequencies. The closer the two frequencies the stronger is the coupling and the associated redshift:
the redshift of CO( 13C16O) is therefore stronger in Fig. 3.
Analyzing the data of Figs. 1 and 2 and others not
shown here, we find a direct correlation of the intensity loss
of the vibrational bands associated with the minority CO
species, as well as of the line shift of the majority CO with
the amount of extra CO postadsorbed into the perfect ()
))R30 CO/Ru0001 layer. More precisely, a discrete
line shift is introduced for weakly coupled localized modes
which is paralleled by an intensity loss of the original minority band, while for strong lateral coupling leading to
delocalized vibrational modes a continuous line shift occurs.
For small amounts of extra CO the observed relative intensity losses and frequency shifts approximately scale linearly
with the CO coverage in excess of 0.33 ML. The respective
extracted values are I/ CO1/I 0 1300%/ML (I 0 denotes the original intensity of the minority 13C16O band at
1941 cm1 and / CO200 cm1 /ML.
In order to investigate the mobility of the extra CO molecules within the ()))R30 CO layer, yet another
mixed isotope experiment has been performed. Thereby
13 16
C O has been postadsorbed at T ads35 K on a perfect
()))R30 12C16O/Ru(0001) layer and subsequently annealed to increasingly higher temperatures; all spectra were
taken after recooling to 35 K. The starting layer in Fig. 4a
therefore is equivalent to Fig. 1. Quite surprisingly, postadsorption of 0.0250.03 ML 13C16O has virtually no effect,
both on the majority 12C16O band as well as the weak
CO( 13C16O) band at 1941 cm1 Fig. 4b. However, as
the annealing sequence Figs. 4c 4f will demonstrate
the close reproduction of the original spectrum is both amazing and deceiving. In fact, a closer inspection of Figs. 4a
and 4b identifies tiny, but significant and most informative
differences: at 1897 cm1 a weak band has appeared. Also,

extra CO should reduce the 1941 cm1 band which is due to

naturally abundant 13C16O within the original ()
))R30 layer as has been observed in Figs. 1 and 2.
According to the estimate above, 0.0250.03 ML should influence about 35% of the CO molecules of the original ()
))R30 layer and the 1941 cm1 band should be reduced
accordingly. Nevertheless, this band almost retains its intensity a mere 5% reduction is found. This means that 13C16O
molecules postadsorbed into ()))R30 12C16O are not
invisible but they contribute to the 1941 cm1 band as well;
the similar frequencies for both configurations is an unfortunate coincidence. This conclusion is corroborated by the observation of the 1897 cm1 band due to the 11% 13C18O
isotopic admixture in our gas manifold used in the postadsorption of 13C16O) in Fig. 4b. Its line position overlaps
almost perfectly with the well-known CO( 13C18O) band of
minority 13C18O within the ()))R30 CO layer at 1895
cm1 see Figs. 2a and 3a.37
Annealing to 75 K slow heating with 1 K/s and recooling to 35 K has no effect on the spectrum Fig. 4c; annealing to 100 K Fig. 4d, on the other hand, produces
dramatic changes in a separate, similar experiment heating
to 90 K induces an effect 31 as large as shown in Fig. 4d:
i
ii

the band at 1896 cm1 and the doublet at 1941 cm1,

which are indicative of 13C18O and 13C16O molecules
within a ()))R30 overlayer, sky-rocket; and
the bands at 1925 and 1962 cm1 which are wellknown from Figs. 1 and 2 appear. This process continues and reaches its maximum after annealing to
about 120 K.

The 1925 cm1 band becomes pretty strong as the layer

now contains a 10% fraction of 13C16O. The mode at 1962
cm1 appears since the layer closely resembles the one in
Fig. 2b. This similarity of the spectra in Figs. 4d and 2b
is not surprising since their overall isotopic composition is
very similar (90% 12C16O10% 13C16O). After complete
intermixture of pre-and post-adsorbed species in Fig. 4d

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

High density gradients in the CO layer on Ru(0001)

9291

FIG. 5. Possible arrangements of CO, postadsorbed into a well-ordered ()))R30 CO/Ru0001 layer indicated by light gray dots. In A the extra
molecule dark gray dot is located in the center of three CO molecules medium gray dots; in B it is assumed that one of these three off-center CO has moved
to a neighboring site due to the impact of postadsorbing CO. In A B one extra CO affects three four CO of the ())) lattice.

most of the special sites initially taken up by postadsorbed

13 16
C O will be occupied by the 90% 12C16O fraction. Note
that in Fig. 2 these special sites have been filled by 12C16O
in the first place.
The observation that the annealed layer in Fig. 4d has
become virtually identical to the frozen-in layer of Fig. 2b
actually proves another thing: Annealing retains the overall
structure of the layer; in particular it causes no homogeneization of the layer or smoothening of the density gradient in the vicinity of postadsorbed CO. Annealing, however,
does induce a rapid intermixing of postadsorbed CO with
preadsorbed molecules.
A straightforward model would place a single CO molecule in the center of three molecules within the ()
))R30 overlayer Fig. 5A. An alternative model
which would be in better agreement with our observation
that each extra CO postadsorbed into the ()))R30
layer affects four others is presented in Fig. 5B; here one of
the CO within the ()))R30 CO layer has been displaced laterally by one substrate lattice site, most probably as
a result of the impact of postadsorbed CO.
As will be detailed in the following, model B has implications at variance with experimental observations and can
be ruled out: The two molecules in Fig. 5B marked by
obviously are equivalent from the point of view of their local
geometry. Only one of them, however, has been postadsorbed, the other one has moved from its original position by
one substrate lattice site. For adsorption sequences involving
different isotopic species as in Fig. 4 a clarification of this
issue can be attained. According to model B of Fig. 5 the
spectral changes associated with adding extra CO to the
()))R30 overlayer should depend to 50% on the nature of postadsorbed CO and to 50% on the isotopic composition of the original ()))R30 CO layer. Using either
12 16
C O or 13C16O in the postadsorption step should therefore
always reproduce a common subset of new bands; more precisely since the original layers in Figs. 1 and 4 are 100%
12 16
C O) the spectral changes reported in Fig. 1 e.g., growth
of the 1960.5 cm1 band should be observable in Fig. 4b,
too. No such thing. The arrangement displayed in model B is
therefore discarded, i.e., lateral displacements of this type,
induced by postadsorbing CO, are virtually absent; they
amount to 5%10% at most estimate from the noise level in
Fig. 4b. It is therefore concluded that extra CO molecules
in excess of 0.33 ML are arranged according to model A,

with the CO within the ()))R30 overlayer retaining

their original sites. Nevertheless, the arrangement according
to Fig. 5B very likely plays an important role in explaining
mobility within the layer at CO0.33 ML and will be realized temporarily during thermally induced lateral motion of
the high density area see Discussion.
I therefore like to make a definite statement in favor of
model A; deviations from the finding that one extra CO
seems to affect four CO of the original ()))R30 overlayer are attributed to lateral coupling effects involving
nearby CO molecules, thereby producing slight intensity
variations of the various vibrational bands.
The suggested model A in Fig. 5 assumes that within the
high local density areas the CO molecules bind to nearest
neighbor Ru atoms. This is primarily based on the observation that the band of the minority 13C16O within the perfect
()))R30 overlayer at 1941 cm1 which is extremely
sensitive to variations of its local environment retains its
narrow line shape when extra CO is postadsorbed. A gradual
relaxation of strain within the adsorbate layer would inevitably lead to substantial broadening of this band which is not
observed; instead an intensity loss of this mode and the
growth of a new, discrete, and narrow mode at 1925 cm1 is
found. The local environment of randomly distributed 13C16O
within the ()))R30 CO layer therefore is either modified substantially or not at all. This finding is totally incompatible with a gradual release of strain in the vicinity of
postadsorbed CO and it fully supports the existence of highly
strained, high local density areas, as sketched in Fig. 5A.
There is actually a high chance that the three off-center
CO become tilted due to the repulsion imposed by the central
CO. This hypothesis is based on the fact that the maximum
attainable coverage of CO on Ru0001 is 0.68 ML;7,10 even
high doses of several thousand langmuirs could not increase
this value significantly. The lateral dimensions of perpendicularly oriented CO therefore are such that 1 ML is not
attainable, i.e., d COd Ru . Placing CO at on-top positions of
nearest neighbor Ru atoms therefore should lead to considerable lateral repulsive forces which are not compensated by
equivalent forces from surrounding molecules, i.e., the offcenter CO in Fig. 5A very likely will be bent away from
the perpendicularly oriented central CO molecule.
Still the question remains, how the frequencies of the
individual CO molecules within the high density complex
displayed in Fig. 5A are modified with respect to the values

9292

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

P. Jakob

FIG. 6. Arrangements of CO at CO
0.33 ML CO according to Fig.
5A. The sites of the individual CO
isotopes have been chosen according
to the layer compositions in Figs. 1 or
2 and 4 in A and B, respectively. In
A an extra 12C16O molecule has been
added to the ()))R30 CO/
Ru0001 layer close to a minority
13 16
C O, whereas in B the extra species in the center of three 12C16O molecules is 13C16O. Coloring as in Fig. 5.

within a perfect ()))R30 layer, and how they relate to

the modes observed in our spectra. Apparently, the various
sites of CO within the high density areas, i.e., of the central
CO, and his immediate neighbors are inequivalent. Qualitatively, we expect the central CO will be considerably affected by his three close neighbors off-center CO, whereas
the reverse effect, the central CO acting on each of his immediate neighbors should be accordingly weaker. In order to
obtain a proper assignment, we will proceed in four steps.
1 We first consider the effect of an extra 12C16O molecule
on a nearby off-center minority 13C16O or 13C18O molecule, arranged according to Fig. 6A.
2 Similarly, the effect of an extra 12C16O molecule on a
nearby off-center majority 12C16O is examined; arrangement according to Fig. 6A.
3 Next, the line position and frequency shift of postadsorbed center 12C16O embedded in a ()))R30
12 16
C O layer, arranged according to Fig. 6A, is discussed.
4 Finally, the line position and frequency shift of extra
13 16
C O embedded in a ()))R30 12C16O layer, arranged according to Fig. 6B, is considered.
1 We know from Figs. 1 and 2 that the configuration in
Fig. 6A individual 13C16O molecules holding off-center
positions within the high density complex leads to the
growth of the band at 1925 cm1. Its intensity scales with the
13 16
C O content within the original ()))R30 12C16O
layer. It is therefore concluded that postadsorbed 12C16O
causes a CO redshift of the surrounding off-center COs by
approximately 15 cm1; specifically, for 13C16O this shift
reads like 19411925 cm1 . For 13C18O a weak band at
1881 cm1 is observed in Fig. 2 shift: 18951881 cm1 ).
Since these shifts are very similar for both isotopic species
its origin is mostly static. Resonant or near-resonant coupling
by means of dynamic dipole coupling with the central 12C16O
molecule is small due to a considerable frequency mismatch;
the respective dynamic frequency shifts are close to zero for
13 18
C O and about 2 cm1 for 13C16O.
2 The effect of postadsorbed 12C16O on an off-center
12 16
C O molecule will be considered next see Fig. 5A. Due
to the absence of isotopic shifts, dynamical dipole coupling
among the central and off-center 12C16O molecules within
the high density complex will be substantial, despite the

static 15 cm1 redshift imposed on the off-center 12C16O

see previous paragraph. As a consequence of this dynamical coupling the bands associated with off-center 12C16O will
be additionally shifted towards the red by several cm1; a
rough estimate based on the lateral distances and the frequency mismatch of 15 cm1 gives approximately 10
cm1 dynamic shift; this latter number also includes coupling to undisturbed, nearby 12C16O of the ()))R30
layer. Since the line position of 12C16O within the ()
))R30 layer is located at 1988.6 cm1,33 the frequency
shifted mode associated with off-center 12C16O is therefore
expected at 198815(10)1963 cm1 . This estimated
value agrees reasonably well with the 1961 cm1 band observed in Figs. 1 and 2 and of Fig. 4, after annealing. The
intensity of the 1961 cm1 band is substantially attenuated
due to an antiphase motion of neighboring 12C16O molecules
within the ()))R30 layer with respect to the off-center
12 16
C O molecules which reduces the dynamical dipole moment associated with this normal mode.38 As a general rule
of thumb dynamical dipole coupling in adsorbate layers on
metal surfaces with electric dipoles oriented perpendicular
to the surface always lead to an apparent intensity transfer in
favor of the higher frequency mode.
3 The situation is kind of reversed if we now consider
the central 12C16O molecule Fig. 6A: Due to the 15
cm1 redshift imposed on the surrounding off-center CO the
central 12C16O molecule will experience a notable blueshift
due to dynamical dipole coupling with his immediate neighbors; a static, chemical shift due to the presence of offcenter CO at nearest neighbor sites will contribute to this
blueshift in addition. At present, both effects cannot be quantified properly; taken together they take responsibility for the
observed blueshift of the dominant IR-absorption band
2030.32032.4 cm1 in Fig. 1a and equivalently in Fig.
2a.
4 How about placing 13C16O in the center of three
12 16
C O molecules Fig. 6B as has been done in Fig. 4.
Now the resonance frequency of the central 13C16O will be
lower than of the surrounding 12C16O due to its 45 cm1
isotopic shift. Dynamical coupling with the three off-center
12 16
C O therefore causes a slight redshift of CO( 13C16O),
which, as the experimental observations in Fig. 4 show, is
compensated by a static blueshift of similar magnitude.
Taken together, the band of central 13C16O stays right at

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

about 1941 cm1, i.e., it overlaps with the CO( 13C16O)

minority band of the original ()))R30 CO phase and
therefore seems to remain invisible.
To conclude this section on mode assignments, a clear
picture of the local structure of the high density CO complex
and associated frequency shifts is derived based on a detailed
analysis of the various lateral interaction mechanisms in layers of different isotopic composition. It is found that, in particular, dynamical dipole coupling among the inequivalent
CO molecules is important to derive a consistent interpretation; depending on the frequency mismatch between the various CO isotopes, the individual dynamical frequency shifts
may add substantially to the static chemical and isotopic
shifts.
DISCUSSION

Through the years, there has always been a discussion on

how to interpret the fact that CO adsorbed on Ru0001 retains its linear bonding configuration as deduced from vibrational spectroscopy at all coverages while the lateral dimensions of perpendicularly oriented CO does not allow nearest
neighbor on-top site occupation on this surface maximum
CO coverage on Ru0001 is 0.68 ML Refs. 7 and 10. One
may argue that vibrational spectroscopy probably is not the
most reliable method to determine binding sites since electron scattering of CO on Ni111 proved the original assignment based on CO vibrational frequencies wrong 39. However, while vibrational bands at 18001950 cm1, as
observed for CO on Ni111 at 0.5 ML, may not provide
clear signatures for one or the other adsorption site geometry
due to considerable frequency shifts associated with dynamical lateral coupling40, there has never been any doubt
that the frequency range above 2000 cm1 can be associated
with an on-top bound linear CO. Thomas and Weinberg3 as
well as Menzel and co-workers23 tried to resolve the issue by
questioning a direct correlation between the linear bonding
configuration and the on-top adsorption site geometry for CO
on Ru0001: A linear bonding configuration of CO, emphasizing the hybridization state of the C atom leading to a
CO frequency at about 2000 cm1, should characterize
adsorbed CO better than its coordination to the surface
atoms. 23 A high order commensurate, fluid-like phase with
nearest neighbor distances determined by the repulsive interaction between CO molecules has been suggested or supported by various authors.2,3,10,17,20,23 The proclaimed model
thereby assumed a weak corrugation of the RuCO interaction potential which should render the correlation of molecular structure and adsorption site likewise weak. CO is therefore expected to retain its linear molecular structure despite
being located at bridge or hollow sites or, at least a certain
lateral displacement off the on-top site should not affect adsorbed CO too much. This model has been applied all along
and there seemed no indication and no need to question it.
This is the more surprising as soon after the described model
has been suggested, an alternative, three domain model with
a nonuniform arrangement of CO at neighboring on-top sites
was suggested to explain the observed (2)2))R30 and
(5)5))R30 LEED patterns.41 Also Fig. 5b of Ref. 3

High density gradients in the CO layer on Ru(0001)

9293

seemed to suggest that lateral strain produced by extra CO

within the ()))R30 layer does not gradually relax with
distance but remains localized within nearest neighbor CO.
From our spectroscopic data we have now definite evidence that the considerable strain produced by the arrangement of Fig. 5A remains localized and that a lateral displacement to higher coordinated sites does not occur. It is
found that all neighboring CO influenced by postadsorbed
CO display an identical signature in terms of vibrational frequency. A gradual release of strain would inevitably lead to a
variety of bands, or, strongly broadened bands, at variance
with the observed spectra. In particular, the narrow linewidth
of the 13C16O minority band at 1941 cm1 within the ()
))R30 12C16O layer should be affected, which is not the
case: as can be seen from Fig. 1 and for 13C18O in Fig. 2, CO
in excess of 0.33 ML induces an intensity loss but no line
broadening. It is therefore concluded that, contrary to the
existing belief, the site preference of CO on Ru0001 for
on-top sites is quite strong and that CO tries to keep those
sites, even at the expense of a tilted geometry.
High density gradients within otherwise well-ordered adsorbate layers probably are quite common whenever adsorbates display a pronounced preference for a particular lattice
site and ordered overlayers are formed at max . Extra
particles deposited into such an ordered array of adsorbates
then produce considerable lateral stress which may or may
not cause a lateral displacement of preabsorbed species. Especially for layers with a strong tendency to retain their long
range order such displacements should not occur and local
high density areas will form. In any case the repulsive forces
will destabilize the neighboring preabsorbed species in
their lattice sites which facilitates their lateral displacement
at elevated temperatures, thereby enhancing the mobility
within the adsorbate layer. A prominent example is the
Ru(0001)-(22)-O overlayer with oxygen atoms occupying hcp hollow sites which is very stable order-disorder
phase transition at 754 K Ref. 42. As soon as the oxygen
coverage exceeds 0.25 ML the mobility within the layer is
dramatically enhanced, according to STM images taken at
room temperature.43,44 CO on Ru0001 apparently behaves
similarly, as demonstrated in this work and as already
pointed out in a more general way by Pfnur et al. on the
basis of thermodynamic data.20 The latter authors thereby
established a correlation between the repulsive interactions
within the adsorbate layer as the CO coverage exceeds 0.33
ML and a sudden drop in adsorption energy by 55 kJ/mol:
Upon desorption the leaving molecule can pick up the energy gained when the neighbor molecules snap back into
optimal sites. 20 Their model assumed that molecules in the
vicinity of the desorbing less strong bound CO do not sit on
nearest neighbor sites but are laterally displaced which also
included further relaxation in the neighborhood. The present
work now clearly shows that, while the general ideas of
Pfnur et al. had been correct, their structural model needed
revision in terms of site specifity of CO, e.g., within the high
density areas. This seems a minor issue only; however, the
previous model assumed an exceptionally easy lateral displacement of CO on Ru0001, while this is explicitly ruled
out on the basis of the present work.

9294

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

P. Jakob

FIG. 7. IR absorption spectra obtained from saturated CO/Ru0001 layers. The three spectra were obtained from layers with different isotopic compositions:
a 99% 12C16O1% 13C16O; b 95% 12C16O5% 13C16O; and c 89% 13C16O11% 13C18O. Panel A shows the dominant modes of the respective
majority species. Panel B shows the modes of the respective minority species vertical scale enlarged by 17. All spectra were obtained at T30 35 K at
a spectral resolution of 2 cm1. The vibrational bands of spectrum c are somewhat broader due to tiny admixtures of 12C16O, with dynamical dipole coupling
producing a relatively strong high frequency tail. Also, a relatively high content of the heavier isotopic species leads to broadening of the various vibrational
bands. The base line distortion in spectrum b is due to the usage of an InSb detector with cutoff at about 1800 cm1.

In this work an enhanced mobility within the adsorbate

layer at CO coverages CO0.33 ML has been found and it
has been correlated with repulsive interactions among neighboring CO molecules within the high density areas. In general, mobility in adsorbate layers may be associated either
with single particle motion or with propagation of density
waves. From an energetic point of view the effect of the
repulsive interaction between the three off-center CO and the
central CO is twofold: i The adsorption energy of the central CO is reduced which helps to surmount kinetic diffusion
barriers; and ii the height of the diffusion barrier associated
with the central CO jumping straight across two off-center
CO is increased. Since both effects cannot be quantified, an
unequivocal decision on the diffusion process cannot be
made based on these considerations alone. Clear evidence for
a density wave-like lateral motion, however, is provided by
our data: Specifically, we find a fast isotopic exchange as
soon as lateral motion starts at about 90 K; a classical, single
particle motion process, on the other hand, will preserve the
character of an isotopically labeled central CO as such. The
extra CO molecules therefore do not move themselves within
a lattice of immobile CO molecules at ()))R30 lattice
sites. The temperature of 90 K needed to induce mobility
implies activation energies even lower than for diffusion of
single CO molecules on Ru0001. In particular, this temperature is substantially lower than T400 K which is
needed to cause any mobility within a perfect ()
))R30 CO/Ru0001 layer. Such a dramatically increased lateral mobility also reflects itself in the strong decline of the desorption prefactor for CO0.33 ML, as has
been pointed out by Pfnur and Menzel,20,23 based on thermal
desorption kinetics.
In the case of density wave-like propagation the repulsive forces within the high density area induce the displace-

ment of one of the three off-center CO to a neighboring site;

this way an arrangement as displayed in Fig. 6B is temporarily formed. Once the original central CO is likewise
pushed aside by the two remaining off-center CO, then an
equivalent high density configuration of CO molecules is
restored, two substrate lattice spacings away from the original position. Interestingly, the relative positions held by preand post-adsorbed CO within the high density area are going
to be interchanged; such an exchange of sites is easily detected using isotopically labeled CO, as has been demonstrated in Fig. 4. As it will always be one of the off-center
CO which gives directions to the lateral motion, a zigzag
type of path will result note that the orientation of the three
off-center CO rotates by 60 every time the high density area
propagates.
CONCLUSIONS

Using isotopic labeling, CO overlayers on Ru0001 at

coverages CO0.33 ML CO have been studied using
IR spectroscopy. It is found that high density areas, as produced by postadsorption of CO into the ()))R30-CO
layer remain highly localized to within a few angstroms. It is
concluded that CO adsorbed on Ru0001 experiences an
exceptionally strong site preference for on-top sites. This
finding in particular invalidates the existing belief of an unusually weak site preference and accordingly shallow
chemical corrugation potential of this paradigmatic model
system. According to isotopic exchange processes lateral
mobility is dramatically enhanced as soon as the CO coverage exceeds 0.33 ML. It is further suggested that lateral motion of the localized high density areas at CO0.33 ML
proceeds via a density wave-like manner rather than by a
single particle hopping of postadsorbed CO.

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

APPENDIX

As a supplement to data covering the coverage regime

CO0.33 ML CO , I would like to present IR spectra
obtained from fully saturated layers of CO on Ru0001.45
Even though a fair number of papers so far have discussed
the vibrational properties of such layers, they all have missed
an important feature, some due to insufficient spectral resolution, others due to bad signal-to-noise ratios. Agreement
exists that a saturated layer of CO on Ru0001 leads to a
strong mode at about 2060 cm1.3,10 The band is quite narrow and, despite twice as many CO molecules, it is slightly
weaker than the equivalent band associated with the ()
))R30 CO/Ru0001 layer. This loss in spectral intensity has been attributed to electronic screening of the molecules dynamic dipole moment which is particularly effective at high densities.10
In Fig. 7, IR spectra taken from saturated layers of CO
on Ru0001 with different isotopic compositions are displayed. The layers have been prepared at temperatures below
100 K and annealed to 250 K thereafter to ensure proper
ordering without annealing the spectral features remained
the same but they were significantly broader. Besides the
dominant modes, apparent in panel A, we clearly identify a
second peak at 1986 1989 cm1 a and b and at 1943
c in panel B. Contrary to our very first thought they are
not due to the respective minority isotopic species since they
do not scale with the relative fraction of the respective
heavier CO isotope within the layer. It is therefore evident
that the quoted bands at 1988 and 1943 cm1 must be
associated with the majority CO species and that the saturated CO/Ru0001 layers contain at least two inequivalent
on-top CO species.
The question remains, where are those vibrational bands
due to minority 13C16O? A closer inspection of the spectra
taken from layers with a high content of minority CO 5%
fraction of 13C16O in b and 11% fraction of 13C18C in c
of panel B in fact identifies a set of modes at 19101920
cm1 Fig. 7b, and 18651875 cm1 Fig. 7c, respectively. Their intensities are extremely low despite their relatively high abundance within the layer. This is attributed to
the extraordinary strong lateral coupling in high density layers such as in Fig. 7: modes with frequencies lower than
those of the surrounding molecules then will lose intensity in
a dramatic way intensity transfer.29
Two questions arise immediately:
i
ii

Are there even more than two different CO species

present at high CO?
Are these two or more species distributed homogeneously throughout the CO/Ru0001 layer?

Our answer to question i is: most probably yes, to question

ii: definitely not.
Regarding i we state that close-ups of the displayed spectra
see panel B reveal bands at 1913 cm1 spectrum c and
1958 1959 cm1 spectra a and b with similar intensities.
Following the same argumentation line as above it is concluded that their presence must be associated with the majority CO isotope as well.

High density gradients in the CO layer on Ru(0001)

9295

Regarding the second question I like to state that a homogeneous distribution of different species in dense layers
will render all but one of them virtually invisible due to the
strong dynamical coupling; this has been verified for dense
layers similar to the ones in Fig. 7 using 50%: 50% mixtures of different CO isotopic species data not shown. A
dense striped phase arrangement of CO, as suggested by Biberian and van Hove,41 on the other hand, would be in agreement with the present spectroscopic data, since screening of
vibrational dipole moments is less pronounced for extended
one-dimensional structures.
ACKNOWLEDGMENTS

The author thanks D. Menzel for continued support and

helpful discussions. Financial support of the Deutsche Forschungsgemeinschaft through Sonderforschungsbereich 338
is gratefully acknowledged.
1

T. E. Madey and D. Menzel, Jpn. J. Appl. Phys., Suppl. 2, Pt 2, 229

1974.
2
E. D. Williams and W. H. Weinberg, Surf. Sci. 82, 93 1979.
3
G. E. Thomas and W. H. Weinberg, J. Chem. Phys. 70, 1437 1979.
4
C. P. Slichter, Principles of Magnetic Resonance, 3rd ed. Springer, New
York, 1990.
5
Y. J. Chabal, Surf. Sci. Rep. 8, 211 1988.
6
A. Schiffer, P. Jakob, and D. Menzel, Surf. Sci. 465, 198 2000.
7
H. Pfnur and D. Menzel, J. Chem. Phys. 79, 2400 1983.
8
G. Michalk, W. Moritz, H. Pfnur, and D. Menzel, Surf. Sci. 129, 92
1983.
9
H. Pfnur and D. Menzel, Surf. Sci. 148, 411 1984.
10
H. Pfnur, D. Menzel, F. M. Hoffmann, A. Ortega, and A. M. Bradshaw,
Surf. Sci. 93, 431 1980.
11
T. E. Madey, Surf. Sci. 79, 575 1979.
12
P. Feulner, H. A. Engelhardt, and D. Menzel, Appl. Phys. 15, 355 1978.
13
W. Riedl and D. Menzel, Surf. Sci. 163, 39 1985.
14
W. Riedl and D. Menzel, Surf. Sci. 207, 494 1989.
15
P. Jakob, Appl. Phys. A: Mater. Sci. Process. 75, 45 2002.
16
M. Bonn, C. Hess, S. Funk, J. H. Miners, B. N. J. Persson, M. Wolf, and
G. Ertl, Phys. Rev. Lett. 84, 4653 2000.
17
J. Braun, K. L. Kostov, G. Witte, and C. Woll, J. Chem. Phys. 106, 8262
1997.
18
S. Wurm, P. Feulner, and D. Menzel, Phys. Rev. Lett. 74, 2591 1995.
19
M. Gierer, H. Bludau, H. Over, and G. Ertl, Surf. Sci. 346, 64 1996.
20
H. Pfnur, P. Feulner, and D. Menzel, J. Chem. Phys. 79, 4613 1983.
21
H. Pfnur, P. Feulner, H. A. Engelhardt, and D. Menzel, Chem. Phys. Lett.
59, 481 1978.
22
U. Leuthausser, Z. Phys. B 37, 65 1980.
23
D. Menzel, H. Pfnur, and P. Feulner, Surf. Sci. 126, 374 1983.
24
CO has been kept low in order to stay in the linear regime of spectral
changes.
25
P. Jakob, Y. J. Chabal, and K. Raghavachari, J. Electron Spectrosc. Relat.
Phenom. 6465, 59 1993.
26
A. Schlapka, M. Lischka, A. Gro, U. Kasberger, and P. Jakob, Phys. Rev.
Lett. 91, 016101 2003.
27
R. M. Hammaker, S. A. Francis, and R. P. Eischens, Spectrochim. Acta 21,
1295 1965.
28
G. D. Mahan and A. A. Lucas, J. Chem. Phys. 68, 1344 1978.
29
B. N. J. Persson and R. Ryberg, Phys. Rev. B 24, 6954 1981.
30
S. Flach and C. R. Willis, Phys. Rep. 295, 181 1998.
31
P. Jakob and B. N. J. Persson, J. Chem. Phys. 109, 8641 1998.
32
P. Jakob and B. N. J. Persson, Phys. Rev. B 56, 10644 1997.
33
P. Jakob, Phys. Rev. Lett. 77, 4229 1996.
34
R. L. C. Wang, H. J. Kreuzer, P. Jakob, and D. Menzel, J. Chem. Phys.
111, 2115 1999.
35
F. M. Hoffmann, N. D. Lang, and J. K. Nrskov, Surf. Sci. Lett. 226, L48
1990.
36
In addition to the 1.1% natural abundance of 13C16O there is another 0.2%
contribution due to 12C18O. The CO line positions of both isotopic species overlap within about one wave number.

9296
37

J. Chem. Phys., Vol. 120, No. 19, 15 May 2004

The apparent oscillator strength of extra 13C16O molecules embedded in

the ()))R30 layer is much smaller than of 13C18O at the same site:
while their relative number ratio amounts to 11% according to the manufacturer and our mass-spec analysis, the observed IR-intensity ratio is
40%50% i.e., the CO oscillator strength is 4 times larger for 13C18O
with respect to 13C16O). This IR-intensity loss is attributed to dynamical
dipole coupling with neighboring 12C16O molecules. Since CO( 13C16O)
is much closer in frequency to CO( 12C16O) than CO( 13C18O) is, this
effect is much more pronounced for the 13C16O isotope. Note that for a
not very dense perfect ()))R30 12C16O layer containing tiny
amounts of the heavier 13C16O and 13C18O isotopes this ratio of apparent
oscillator strengths is approximately 1:2. Since the extra 13C16O and
13 18
C O molecules in Fig. 4b have been squeezed into the ()
))R30 layer, the dipole coupling and intensity transfer to neighboring
12 16
C O naturally is much more pronounced than for the perfect ()
))R30 12C16O layer, e.g., in Fig. 4f.
38
For off-center 13C16O there will be dynamical coupling as well. Since the
frequency mismatch with respect to the central 12C16O molecule as well as

P. Jakob
to the surrounding 12C16O of the ()))R30 layer is much larger for
13 16
C O than for 12C16O, the associated additional redshift of CO( 13C16O)
at about 1925 cm1 ( 2 cm1 ) as well as the associated intensity
transfer will be relatively weak.
39
K.-M. Schindler, Ph. Hofmann, K.-U. Wei et al., J. Electron Spectrosc.
Relat. Phenom. 6465, 75 1993.
40
A. Eichler, Surf. Sci. 526, 332 2003.
41
J. P. Biberian and M. A. Van Hove, Surf. Sci. 138, 361 1984.
42
P. Piercy and H. Pfnur, Phys. Rev. Lett. 59, 1124 1987.
43
M. Gsell and P. Jakob unpublished results.
44
K. Meinel, H. Wolter, Ch. Ammer, A. Beckmann, and H. Neddermeyer, J.
Phys.: Condens. Matter 9, 4611 1997.
45
Our IR data related to saturated CO/Ru0001 layers now exist for about 9
years. So far we restrained from publishing them because there still existed some open questions regarding the structural arrangement of the
various species and details of their binding geometry. In order to avoid
confusion in the literature and to straighten up a few points related to these
layers, we decided to publish them along with a less detailed interpretation
of our observations.

The Journal of Chemical Physics is copyrighted by the American Institute of Physics (AIP). Redistribution of
journal material is subject to the AIP online journal license and/or AIP copyright. For more information, see
http://ojps.aip.org/jcpo/jcpcr/jsp