Low-energy electron diffraction structure determination of an ultrathin CoO film on Ag(0 0 1)

Article abstract of DOI:10.1016/j.susc.2009.06.020

The atomic structure of a four layer thick film of CoO on a Ag(0 0 1) substrate has been determined by comparing experimental low-energy electron diffraction (LEED) I(V) curves with multiple scattering calculations. The CoO film has been prepared using re

Full text of DOI:10.1016/j.susc.2009.06.020

Surface Science 603 (2009) 2658–2663
Contents lists available at ScienceDirect
Surface Science
journal homepage: www.elsevier.com/locate/susc
Low-energy electron diffraction structure determination of an ultrathin CoO film
on Ag(001)
*
K.-M. Schindler , J. Wang, A. Chassé, H. Neddermeyer, W. Widdra
Institut für Physik, Naturwissenschaftliche Fakultät II, Martin-Luther-Universität Halle-Wittenberg, D-06099 Halle, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
The atomic structure of a four layer thick film of CoO on a Ag(001) substrate has been determined by
comparing experimental low-energy electron diffraction (LEED) I(V) curves with multiple scattering cal-
culations. The CoO film has been prepared using reactive evaporation of Co in an oxygen atmosphere
leading to almost layer-by-layer growth. Contrary to the surface of CoO crystals an outward relaxation
of the two outermost CoO layers as well as rumpling in the top layer has been found. The supposed driv-
ing force of this relaxation is the in-plane compressive stress, which results from the pseudomorphic
growth of the CoO film on the Ag(001) substrate and the lattice mismatch of the two materials.
Ó 2009 Elsevier B.V. All rights reserved.
Received 11 December 2007
Accepted for publication 24 June 2009
Available online 30 June 2009
Keywords:
Low-energy electron diffraction (LEED)
Thin film structure
Metal-oxide interface
Cobalt oxide
1. Introduction
the energy associated with the dipole of polar surfaces increases
with thickness and that reconstructions of the surface are a way
Physical properties, such as magnetism and electronic structure,
depend on the geometric structure. Although a correct description
of the electronic structure of transition metal oxides is difficult, al-
ready the simplistic model of ligand field is sufficient to show qual-
itatively that the splitting of the metal d states strongly depends on
the metal oxygen distance. This effect has been studied with X-ray
absorption on CoO films [1]. Whereas the X-ray absorption of bulk
CoO is isotropic, it was found that the strain resulting from the
lattice mismatch between substrate and film gives rise to a polar-
ization contrast at the Co L2;3 edge. With the help of a cluster
calculation, the influence of magnetic ordering and symmetry low-
ering through strain is deduced from the spectra. In order to use a
more sophisticated theoretical model for the calculation of the
electronic structure, such as density functional theory, the geomet-
ric structure is needed quantitatively. In particular in ultrathin
films, which are candidates in tunneling magneto resistance
devices, the large surface to bulk ratio gives surface relaxations
an even stronger influence on physical properties than in the films
mentioned above.
to lower the overall energy of the system. The (2 ꢀ 2) reconstruc-
tion of the NiO(111) surface is such an example [4,5]. However,
lateral strain from the lattice mismatch between film and substrate
also increases the energy of the system and may lead to relax-
ations. In films with polar surfaces the contributions from the film
dipole and the strain are difficult to distinguish. Investigating non-
polar surfaces of thin films and comparing them to bulk surfaces,
however, offers the unique opportunity to isolate the strain effect
on relaxations from the dipole effect.
Additionally, the geometric arrangement of atoms at the inter-
face of film and substrate is of prime importance when discussing
interactions across the interface, such as mechanical, electrical or
magnetic couplings between film and substrate. The late 3d transi-
tion metal oxides exhibit interesting physical properties to make
them promising candidates for new electronic devices, for example
magneto-electric coupling across according interfaces. Scanning
probe microscopy proved to be very useful to study the growth
mode and morphology of such films, in particular the complex ini-
tial developments at low coverages [6,7]. Much less is known about
quantitative structural deviations from the bulk structure in thin
films of these oxides. A multi-method study of NiO films on
Ag(001) investigated their stress/strain behaviour and found
relaxation to the bulk structure for thicknesses above 10 ML and
an upper limit of the a/c ratio of the tetragonally distorted film
of 0.96 [8]. Thinner NiO films on Ag(001) with a thickness of 2
ML have been studied with X-ray photoelectron spectroscopy
(XPS), low-energy ion scattering (LEIS) and LEED intensity analysis
[9]. Contrary to the above expectations no major deviations from
A structure determination of the (001) surface of a CoO single
crystal revealed no significant deviation from a bulk terminated
surface [2]. Compared to metal surfaces two additional origins of
surface relaxations have to be considered in thin films of oxides.
According to the scheme of Tasker [3], polar and non-polar surfaces
are distinguished. For the discussion here, the main point is that
* Corresponding author. Tel.: +49 345 5525558; fax: +49 345 5527160.
E-mail address: karl-michael.schindler@physik.uni-halle.de (K.-M. Schindler).
0039-6028/$ - see front matter Ó 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2009.06.020

K.-M. Schindler et al. / Surface Science 603 (2009) 2658–2663
2659
the bulk structure of NiO were found. The reason for this discrep-
ancy remains unclear. For a 15 ML thick layer of CoO on Ag(001)
bulk like properties have been found in photoemission and Auger
electron diffraction [13], similar to NiO. On Au(111) a 10 ML thick
CoO(111) film also showed bulk-like angle-resolved photoemis-
sion spectra [14]. Very recently, Torelli et al. showed with a grazing
incidence X-ray diffraction (GIXD) study that CoO films grow
pseudomorphically up to a thickness of 5 ML and gradually relax
to the bulk lattice with increasing thickness [16].
the (10) and (11) reflections. The arrows indicate the unit cell of
the reciprocal lattice and point to the (10) diffraction spots. The
orientations of the two patterns are identical, which shows how
well defined the CoO layer is oriented relative to the Ag(001) sub-
strate with fourfold rotational symmetry. The spots of the CoO
layer are broader than the ones from the Ag(001) surface. Obvi-
ously, the long range order in the CoO film is less perfect than that
of the Ag(001) surface. In summary, the LEED pattern shows that
our preparation results in a film comparable to the ones investi-
gated previously and that the deposition of Co in an O2 atmosphere
has resulted in a film of cobalt oxide growing pseudomorphically
on the Ag(001) substrate.
2. Experiment
Fig. 2 shows the experimental LEED I(V) curves of the clean
Ag(001) surface and the 4 ML thick CoO film on Ag(001). The
two sets of curves are completely different, which is most promi-
nently seen at energies, where a spot of the LEED pattern has max-
imum intensity for one system and minimum intensity for the
other system. Example energies are 80 eV, 115 eV and 225 eV for
the (10) diffraction spot, 120 eV, 145 eV, 230 eV, 290 eV and
330 eV for the (11) spot and 250 eV and 300 eV for the (20) spot.
Obviously, the majority of well ordered Ag(001) parts of the sub-
strate are covered by a uniform layer of CoO thick enough, so that
electrons scattered in the Ag substrate do not reach the surface.
This is in line with previous findings of a nearly layer-by-layer
growth mode of the ultra thin film of CoO. The absence of
Ag(001) LEED spots, however, does not preclude the presence of
small or disordered patches of Ag protruding through the CoO film,
since the scattering of electrons would not result in a diffraction
pattern. However, it has been shown with scanning tunneling
microscopy (STM) that such Ag protrusions are absent in films pre-
pared under these conditions (deposition rate, oxygen partial pres-
sure and most importantly substrate temperature) [6].
A comparison of the LEED I(V) curves from the film and from the
(001) surface of a CoO single crystal gives additional evidence that
the preparation conditions give CoO with rocksalt structure and no
other oxide phase, which cannot a priori be excluded. Fig. 3 com-
pares the (11), (20) and (21) I(V) curves of the cobalt oxide film
on Ag(001) and the (001) surface of a CoO single crystal as pub-
lished [2]. The similarity of the two sets of curves confirms that
the oxide in the film is actually CoO with rocksalt structure. Com-
paring the experiments of the two systems (bulk and thin film) re-
veals another advantage of thin films of insulating materials. In the
case of the bulk single crystal charging effects prevented the
recording of I(V) curves below 170 eV [2], whereas in the case of
the thin film, charges created in the film present no problem as
they are easily transferred to the metallic Ag substrate. As a result
the intensity of the (10) spot could be recorded from an energy as
low as 30 eV, as evident from Fig. 2. For all spots together about 6
additional intensity maxima could be included in the data analysis.
The additional low-energy parts of all spots sum up to 250 eV, a
substantial part of the whole dataset.
Experiments were performed in a standard UHV chamber (Omi-
cron Full Lab) with a base pressure below 5 ꢁ 10ꢂ9 Pa (5 ꢁ 10ꢂ11
mbar). The Ag(001) sample had previously been used for numer-
ous investigations. Therefore, the success of preparations regarding
cleanliness and long-range order of this crystal had been checked
extensively with different methods, including scanning tunneling
microscopy (STM). Prior to the investigations described here, the
Ag sample was prepared by several cycles of Arþ ion sputtering
(600 eV ion energy and typically 2 l
A=m2 ion current) and anneal-
ing at 600 K until sharp LEED spots were obtained.
CoO films were prepared by reactive evaporation of Co metal
from a Knudsen cell in an oxygen atmosphere of 10ꢂ4 Pa (10ꢂ6
mbar) similarly as described previously [13,15]. The deposition
rate was 0.2 ML of CoO per minute as determined using a quartz
microbalance. In order to reduce the roughness of the film, the
Ag substrate was kept at
evaporation.
a temperature of 470 K during
Normal incidence low-energy electron diffraction I(V) curves
were recorded at room temperature with a standard 4-grid LEED
optics (Omicron), a video camera and the Erlangen LEED software.
The orientation of the sample in front of the LEED optics was ad-
justed until minimal differences between symmetry equivalent
beams were achieved. The agreement between the beams was
quantified in terms of Pendry’s r-factor Rp [17]. The orientation of
the sample was adjusted for normal incidence of the electron beam
until an r-factor of below 0.1 was achieved.
3. Multiple scattering calculations
Multiple scattering calculations were performed using the sym-
metrized automated tensor LEED package (SATLEED) of Barbieri
and van Hove [18]. The structure determination of the ultrathin
film of CoO on Ag(001) is based on five independent I(V) curves
with beam indices (1,0), (1,1), (2,0), (2,1), (2,2), energy ranges
from 30 to 400 eV and a total energy range of 1025 eV. Phase shifts
up to an lmax of 9 were derived from the Barbieri–van Hove phase
shift package. The imaginary part of the inner potential was fixed
to 5 eV, the real part was initially set to 10 eV and optimized in
the search procedure. The Debye temperatures of Co, O, and Ag
were initially set to 385, 500, and 225 K, respectively. After a best
fit model has been determined with the Debye temperatures kept
at these values, a second pass of optimization including the Debye
temperatures has been performed. The structural parameters re-
mained basically unchanged in this second pass. However, the er-
ror bars of structural parameters became smaller due to an
increased sensitivity on structural changes.
Unfortunately, no LEED I(V) curves have been published for
other cobalt oxides, in particular Co3O4, the next stable bulk phase
with some Co in the oxidation state 3þ. However, Co3O4 can safely
be excluded by considering the details of its structure, since its spi-
nel structure differs strongly from the rocksalt structure of CoO.
First of all, each side of the unit cell of Co3O4 is about two times
as long as those of CoO, which at least should give rise to a unit cell
doubling and a (2 ꢀ 2) superstructure or an even more complex
pattern. Secondly, there are differences in the Co and O sublattices.
However, it is sufficient to discuss the Co sublattice, since the dif-
ferences are larger than in the O sublattices and the scattering fac-
tors of Co are higher than those of O. The first difference is that the
number of Co atoms in octahedral sites in Co3O4 is only half of that
in CoO. Secondly, an additional quarter of Co atoms occupies tetra-
hedral sites in Co3O4, whereas such sites are not at all occupied in
4. Results and discussion
Fig. 1 shows the LEED patterns of a clean Ag(001) surface and a
CoO film on Ag(001). For easier comparison both patterns have
been recorded at an electron energy of 75 eV and with the sample
at the same position. The bright spots of the patterns correspond to

2660
K.-M. Schindler et al. / Surface Science 603 (2009) 2658–2663
Fig. 1. LEED patterns: (a) 4 ML thick CoO film on Ag(001), (b) clean Ag(001) surface. The energy of the electron beam is 75 eV. The arrows indicate the basis vectors of the
reciprocal unit cell.
surface would be completely different from those of the
CoO(001) surface.
4ML CoO/Ag(100)
Ag(100)
(2 0)
(1 1)
In summary, the comparison of the LEED I(V) curves in Fig. 3
provides overwhelming evidence that the deposition of Co in an
O2 atmosphere gives a CoO layer, which grows well ordered and
pseudomorphically on the Ag(001) substrate, fully in line with
previous investigations. Finally, there are small differences be-
tween the curves of the film and the (001) surface of a bulk crystal.
The analysis of the I(V) curves below will show that these differ-
ences can actually be attributed to small structural deviations of
the film structure from a bulk terminated structure.
In order to determine the structure of the CoO film quantita-
tively, the experimental curves were compared to theoretical ones
from multiple scattering calculations. Prior information allows to
restrict the search to models of CoO rock salt structure with
(001) orientation. The following structural properties were exam-
ined in detail and optimized until best agreement between exper-
imental and calculated I(V) curves was achieved:
(1 0)
Energy (eV)
– Thickness of the CoO film.
Fig. 2. Experimental LEED I(V) curves of the clean Ag(001) surface and the 4 ML
thick CoO film on Ag(001). Both are taken with a normal incident electron beam.
– Registry of the CoO film versus the Ag(001) substrate.
– Distance of the CoO film versus the Ag(001) substrate.
– 1st to 2nd and 2nd to 3rd layer distances in the CoO film.
– Rumpling between Co and O in both the 1st and 2nd outermost
layer.
4ML CoO/Ag(100)
CoO single crystal
(2 1)
(2 0)
Fig. 4 shows the experimental and best-fit calculated LEED I(V)
curves of the 4 ML thick CoO film. The corresponding best-fit struc-
ture is shown in Fig. 5.
4.1. Thickness of the film and registry at the interface
Investigations of CoO films with high resolution spot profile
analysis LEED, scanning tunneling microscopy and GIXS have re-
vealed that CoO grows pseudomorphically on Ag(001) [6,16]. This
means that the length of the in-plane vectors of the unit cells are as
those of bulk Ag, i. e. 2.888 Å. Compared to the bulk value of CoO
(3.012 Å) this is a compression of 4% in both lateral directions. As
this compression has been shown convincingly by two indepen-
dent investigations, the calculations of LEED I(V) curves were only
performed for unit cells with a lateral mesh of 2.888 ꢀ 2.888 Å2. In
order to determine the thickness dependence, calculations were
performed for 2, 3, 4, and 5 layers of CoO. In addition, three differ-
ent registries at the CoO – Ag interface were considered:
(1 1)
Energy (eV)
Fig. 3. Experimental LEED I(V) curves of the 4 ML thick CoO film on Ag(001) and
the (001) surface of single-crystalline CoO [2]. Both are taken with a normal
incident electron beam.
(A) Co on-top Ag, O in hollow sites.
(B) Co in hollow sites, O on-top Ag.
(C) Co and O in bridge sites.
CoO. The occupation of tetrahedral sites results in additional layers
of atoms which in turn give rise to additional intererence condi-
tions. Therefore, it is not even necessary to do a simulation of an
assumed Co3O4 surface to see that the LEED I(V) curves of such a
In all three models the top two CoO layers were allowed to relax
until best agreement with experiment was achieved. The

K.-M. Schindler et al. / Surface Science 603 (2009) 2658–2663
2661
Table 1
Pendry’s r-factor Rp obtained from calculations performed for pseudomorphic
CoO(001) films of various thickness. For each thickness, calculations were performed
for three registries of the CoO film with respect to the Ag(001) substrate. (A) Co
atoms directly above Ag atoms and O atoms in hollow sites. (B) O atoms directly
above Ag atoms and Co atoms in hollow sites. (C) Co and O atoms in bridge positions.
Registrynthickness
2 ML
3 ML
4 ML
5 ML
A (Co on-top, O hollow)
B (Co hollow, O on-top)
C (Co and O bridge)
0.47
0.45
0.65
0.35
0.31
0.64
0.31
0.27
0.63
0.30
0.30
0.63
is at the higher end of the expected range. We attribute this pro-
nounced sensitivity to the stronger scattering of electrons at Ag
compared to scattering at Co and O.
Comparing the thickness dependence (shown in Table 1), mod-
els A and B show reasonable agreement for 3 and 5 ML. Although
this could be mere coincidence, previous STM investigations have
shown that some parts of the surface are actually covered by three
and five layers after a deposition of nominal four layers. Conse-
quently, the experimental curves are expected to also contain
some contributions from 3 ML and 5 ML structures. In principle,
it is possible to determine the fraction of each thickness. However,
calculations show that the I(V) curves mainly depend on relax-
ations in the first two layers and much less on changes in deeper
layers. In addition, the relaxations in the top layers are identical
for 3, 4 and 5 ML. Therefore, increasing the thickness of a layer
simply amounts to exchanging the top substrate Ag layer by an
additional CoO layer. This actually prevents the determination of
the individual fractions and justifies to limit further structure
refinements to model B, i.e. O on-top of Ag and Co in the fourfold
hollow site with a thickness of 4 ML.
Fig. 4. Experimental and best-fit calculated LEED I(V) curves of the 4 ML thick CoO
film on Ag(001). The best-fit structure of the multiple scattering calculations is
given in Fig. 5.
agreement between experimental and calculated I(V) curves was
quantified using Pendry’s r-factor Rp [17]. Model C with all atoms
in bridge sites as well as films with two layers can clearly be ruled
out. The best agreement with Rp ¼ 0:27 has been obtained for a
thickness of 4 ML and a film – substrate registry where the O atoms
at the interface are directly above Ag atoms. The film thickness as
determined with LEED is in good agreement to the nominal thick-
ness as derived from the deposition conditions and again confirms
the almost layer-by-layer growth mode. Model B with O on-top of
Ag agrees better than model A for 3 and 4 ML and is therefore the
favored one. However, for films with a thickness of 5 ML models B
and C agree equally well. Considering the attenuation length of
electrons and other LEED studies a sensitivity limit of 4–5 layers
4.2. Layer distance at the interface
An interesting detail of the interface structure is the distance
between the top Ag layer and the CoO layer at the interface. The
corresponding Ag–O bond length is presumably the essential factor
for the adhesion of the film at the substrate. Elaborate calculations
were performed for a series of distances from 1.93 Å to 2.93 Å in
steps of 0.10 Å. Best agreement was achieved for a distance of
2:4 ꢃ 0:1Å. This is much longer than the Ag–Ag layer distance
(2.04 Å) or the CoO–CoO layer distance (2.13 Å). It is also much
longer than Ag–O bond lengths in adsorbate systems. In
Fig. 5. Best-fit structure of the 4 ML thick CoO film on Ag(001) as found by multiple scattering calculations.

2662
K.-M. Schindler et al. / Surface Science 603 (2009) 2658–2663
Ag(110)(2 ꢀ 1)O the Ag–O bond length has been determined to be
2.044 Å, 2.045 Å and 2.05 Å with PED [10], LEIS [11] and SEXAFS
[12]. With 2.0441 Å, the bulk oxide Ag2O also exhibits a consider-
ably shorter Ag–O bond length [19]. Interestingly, Ag–O distances
of basically the same size were found on Ag(001) with MgO and
NiO films (2.39 Å [9] and 2.43 Å [21]). The GIXS study of a CoO film
of the same thickness also found the same registry (O atoms on top
of Ag atoms, Co atoms in fourfold hollow sites) and basically the
same interface distance of 2:32 ꢃ 0:07 Å [16]. In summary, the long
Ag–O bond length at the interface seems to be a general property of
such systems. The increased length corresponds to a particular
weak bonding between Ag and O. A plausible explanation is based
on the difference in bonding of Co and Ag to O. Due to the high
electronegativity of oxygen, a strong metal–O bond always implies
electron transfer from the metal to the oxygen. This results in a
competition in electron donation between Co and Ag to the O at
the interface. Since Co can donate electrons much easier than Ag,
the charge transfer from the five surrounding Co atoms is about
as strong as in bulk CoO. Consequently, the charge on the oxygen
is higher than when coordinated by Ag atoms only. This explains
that the bond between Ag and O at the interface is actually weaker
than in the bulk oxide or Ag–O adsorbate systems. Together with
the weak interaction between the positively charged Co and Ag
atoms this is prerequisite for the carpet-like coverage of substrate
steps with a CoO film as seen in STM investigations [6,20]. A quan-
titative determination of the bonding with theoretical calculations
would be interesting.
ones due to tensile stress. The s11 and s12 components of the stress
tensor are 6.43 TPꢂ1 and ꢂ2.31 TPꢂ1 [22], which in turn yield a
s12
Poisson ratio
v
(v
¼ ꢂ s ) of 0.36. With this Poisson ratio the com-
11
pression of 4.2% in both lateral directions would result in an expan-
sion of 3%. Whereas the expansion of the second to third layer
distance (2.3%) fits quite well to this bulk value, the expansions
in the outmost layer (8.5% and 5.6%) are substantially larger. This
clearly demonstrates the necessity to study these relaxations on
an atomic level, in particular to address the question whether the
coincidence found with NiO [8] or the difference found here with
CoO is the more general case.
Photoemission spectra of the valence band of CoO films did not
show any significant difference from bulk [13,14,23], contrary to
what is expected from the deviations of the film geometry from
bulk, as found in this investigations. The changes in bond lengths
and the lowering of symmetry should lead to shift of states and
splitting of degenerate states. As already discussed in the refer-
ences, the widths of the photoemission lines are considerably
broadened beyond the instrumental resolution. Most likely, this
is the reason, why no effect could be detected in the photoemis-
sion spectra, whereas the NEXAFS investigations [1] revealed clear
differences in the spectra, which indicate changes in the unoccu-
pied electronic states. It would be interesting to see from the
determination of the electronic structure of such a system by
theoretical calculations how strongly the occupied states are
actually influenced or whether the changes only affect the unoc-
cupied states.
4.3. Layer distance expansions and rumplings in the two outermost
layers
5. Summary
The best agreement between experimental and calculated LEED
I(V) curves has been obtained with an expansion of the two outer-
most interlayer distances as well as a rumpling between the Co
and O atoms in the outermost layer. Rumpling in the second layer
has also been considered, but found to be small and with a large er-
ror bar (0:01 ꢃ 0:04 Å). Therefore, rumpling in the second layer will
be neglected in the following discussion and the Co and O atoms in
the second layer are assumed to reside in the same plane. This plane
(the second outermost layer), however, was found to still have a
slightly larger distance from the 3rd layer (2:18 ꢃ 0:04Å) than bulk
layers of CoO (2.130 Å), an expansion of 2.3%. However, this is close
to the sensitivity limit of the method as obvious from an error bar of
nearly equal size (1.9%). The positions of the atoms in the outermost
layer deviate much stronger from their bulk positions. The first layer
Co atom is 2:25 ꢃ 0:04 Å above the second layer, which is an expan-
sion of 5.6% and well above the sensitivity limit. In addition to the
expansion a rumpling has been found in the outermost layer, with
the oxygen atoms 0:06 ꢃ 0:04Å higher than the first layer Co atoms
i.e. 2.31 Å (8.5% expansion) above the second layer.
As mentioned above, a previous surface structure determination
of the (001) surface of a CoO single crystal did not find any relax-
ations, although with a larger error bar of ꢃ3% [2]. We attribute
this obvious difference to the fact that the CoO film grows pseudo-
morphically on Ag(001), which results in a compression in both
lateral directions. Therefore, the vertical expansion of the film is
an expected reaction. Such behaviour has previously been found
with films of NiO [8] and MgO [21] on Ag(001) as well as in the
GIXS study of CoO films on Ag(001) [16]. The upper limit of such
an expansion can be derived from the assumption of a constant
volume. Assuming a lateral compression of 4.2% in both directions
results in an expansion of 9%. This clearly fulfills the claim of an
upper limit, since it is larger than 8.5% (top layer oxygen), 5.6%
(top layer cobalt) and 2.3% (second layer oxygen and cobalt).
Although the lateral compression leads to a complex relaxation
in the film it is valid to compare the expansions to macroscopic
The atomic structure of a four monolayer thick film of CoO on a
Ag(001) substrate has been determined by comparing experimen-
tal normal incidence LEED I(V) curves with theoretical ones from
multiple scattering calculations. The analysis shows that the
evaporation of Co in an O atmosphere results in pseudomorphic
layer-by-layer growth of a CoO film. The structure in the film devi-
ates substantially from the bulk structure of CoO. The Co and O
atoms of the top layer exhibit a rumpling of 0.06 Å as well as
increased distances from the second layer (Co: 2.25 Å, O: 2.31 Å).
Compared to the bulk distance of 2.13 Å the distance between
the second and third layer is also expanded (2.18 Å). Finally, the
registry of atoms at the interface between film and substrate has
been determined. At the interface, O atoms are located directly
above Ag atoms and the Co atoms are in fourfold hollow sites.
The Ag–O distance of 2.4 Å is fairly large, but very similar to that
of other oxide films. The vertical relaxations are supposed to result
from lateral strain due to the lattice mismatch between film and
substrate. A comparison of the deviations from the bulk structure
to macroscopic elastic behaviour of CoO indicates particular behav-
iour of the ultrathin film. The relaxations as found show clearly
that they can already be induced by the strain from the
substrate-film lattice mismatch and not necessarily from a polar
termination of the film. Furthermore, it appears interesting to
follow the implications for the electronic structure in such films
by experimental and theoretical methods.
Acknowledgement
The authors are grateful to thank S. Sindhu and Ch. Ammer for
supporting discussions and to the members of our group for the
stimulating atmosphere. Special thanks goes to Ch. Jeckstiess for
her valuable technical support. This work was financially
supported through the Forschergruppe FG 404 ‘‘Oxidische
Grenzflächen”.

K.-M. Schindler et al. / Surface Science 603 (2009) 2658–2663
2663
[12] L. Becker, S. Aminpirooz, A. Schmalz, B. Hillert, M. Pedio, J. Haase, Phys. Rev. B
44 (1991) 13655.
References
[13] M. Heiler, A. Chassé, K.-M. Schindler, M. Hollering, H. Neddermeyer, Surf. Sci.
454–456 (2000) 36.
[14] S. Sindhu, M. Heiler, K.-M. Schindler, H. Neddermeyer, Surf. Sci. 541 (2003)
197.
[15] I. Sebastian, M. Heiler, K. Meinel, H. Neddermeyer, Appl. Phys. A 66 (1998)
525.
[16] P. Torelli, E.A. Soares, G. Renaud, S. Valeri, X.X. Guo, P. Luches, Surf. Sci. 601
(2007) 2651.
[17] J.B. Pendry, J. Phys. C. Solid St. Phys. 13 (1980) 937.
[18] A. Barbieri, M.A. Van Hove, Symmetrized Automated Tensor LEED Package,
from M.A. Van Hove, Lawrence Berkeley Laboratory, 94720 Berkeley, CA, USA,
<http://www.lbl.gov.software>.
[19] A. Sanson, F. Rocca, G. Dalba, P. Fornasini, R. Grisenti, M. Dapiaggi, G. Artioli,
Phys. Rev. B 73 (2006) 214305. and references therein.
[20] I. Sebastian, H. Neddermeyer, Surf. Sci. 454–456 (2000) 771.
[21] C. Giovanardi, A. di Bona, T.S. Moia, S. Valeri, C. Pisani, M. Sgroi, M. Busso, Surf.
Sci. 505 (2002) L209.
[1] S.I. Csiszar, M.W. Haverkort, Z. Hu, A. Tanaka, H.H. Hsieh, H.-J. Lin, C.T. Chen, T.
Hibma, L.H. Tjeng, Phys. Rev. Lett. 95 (2005) 187205.
[2] R.C. Felton, M. Prutton, S.P. Tear, M.R. Welton-Cook, Surf. Sci. 88 (1979) 474.
[3] P.W. Tasker, J. Phys. C. Solid St. Phys. 12 (1979) 4977.
[4] H. Hannemann, C.A. Ventrice Jr., Th. Bertrams, A. Brodde, H. Neddermeyer,
Phys. Stat. Sol. (a) 146 (1994) 289.
[5] C.A. Ventrice Jr., Th. Bertrams, H. Hannemann, A. Brodde, H. Neddermeyer,
Phys. Rev. B 49 (1994) 5773.
[6] I. Sebastian, Th. Bertrams, K. Meinel, H. Neddermeyer, Faraday Discuss. 114
(1999) 129.
[7] Ch. Hagendorf, R. Shantyr, K. Meinel, K.-M. Schindler, H. Neddermeyer, Surf.
Sci. 532–535 (2003) 346.
[8] C. Giovanardi, A. di Bona, S. Altieri, P. Luches, M. Liberati, F. Rossi, S. Valeri, Thin
Solid Films 428 (2003) 195.
[9] M. Caffio, B. Cortigiani, G. Rovida, A. Atrei, C. Giovanardi, A. di Bona, S. Valeri,
Surf. Sci. 531 (2003) 368.
[10] M. Pascal, C.L.A. Lamont, P. Baumgärtel, R. Terborg, J.T. Hoeft, O. Schaff, M.
Polcik, A.M. Bradshaw, R.L. Toomes, D.P. Woodruff, Surf. Sci. 464 (2000) 83.
[11] M. Canepa, P. Cantini, F. Fossa, L. Mattera, S. Terreni, Phys. Rev. B 47 (1993)
15823.
[22] H. Landolt, R. Börnstein, New Series, vol. III-29a, Springer Verlag, 1979. p. 68.
[23] S. Sindhu, M. Heiler, K.-M. Schindler, W. Widdra, H. Neddermeyer, Surf. Sci.
566–568 (2004) 471.