Scanning tunneling microscopy and spectroscopy of CoO precursor and oxide layers on Ag(0 0 1)

Article abstract of DOI:10.1016/j.susc.2004.06.085

We have deposited Co in O₂ atmosphere onto Ag(0 0 1) in order to grow thin CoO films in (0 0 1) orientation. According to previous work CoO(0 0 1) layers are obtained for deposition at elevated substrate temperature (470 K) while at lower substrate temperature an oxide precursor develops. Our present work is devoted to a more systematic analysis of the precursor structure and the subsequent stages to formation of CoO(0 0 1) by scanning tunneling microscopy and its characterization by scanning tunneling spectroscopy, where we have measured I/U characteristics and determined the normalized conductance (dI/dU)/(I/U).

Full text of DOI:10.1016/j.susc.2004.06.085

Surface Science 566–568 (2004) 68–73
www.elsevier.com/locate/susc
Scanning tunneling microscopy and spectroscopy
of CoO precursor and oxide layers on Ag(0 0 1)
*
R. Shantyr , Ch. Hagendorf, H. Neddermeyer
Martin-Luther-Universit€at Halle-Wittenberg, Fachbereich Physik, D-06099 Halle, Germany
Available online 15 June 2004
Abstract
We have deposited Co in O2 atmosphere onto Ag(0 0 1) in order to grow thin CoO films in (0 0 1) orientation.
According to previous work CoO(0 0 1) layers are obtained for deposition at elevated substrate temperature (470 K)
while at lower substrate temperature an oxide precursor develops. Our present work is devoted to a more systematic
analysis of the precursor structure and the subsequent stages to formation of CoO(0 0 1) by scanning tunneling
microscopy and its characterization by scanning tunneling spectroscopy, where we have measured I=U characteristics
and determined the normalized conductance ðdI=dUÞ=ðI=UÞ.
ꢀ 2004 Elsevier B.V. All rights reserved.
Keywords: Scanning tunneling microscopy; Scanning tunneling spectroscopies; Cobalt oxides; Silver; Single crystal surfaces
1. Introduction
(0 0 1) orientation can be obtained by reactive
deposition of Co in O2 atmosphere on a hot (470
K) Ag(0 0 1) substrate [4]. A precursor state of
CoO(0 0 1) was identified qualitatively which con-
denses in islands of monolayer height. Subsequent
annealing leads to double layer CoO(0 0 1) islands.
A main question in this context is to which
extent the various phases, observed during the
growth, show oxidic behavior of the bulk oxide.
The optical band gap of bulk CoO is 2.3 eV [5] and
the development of the band gap and the elec-
tronic structure with increasing film thickness is of
crucial importance. For low thickness vertical
electric transport through the film will be possible
by tunneling even in the case of a fully developed
band gap. For a control of the uniformity of the
film, it will also be important to analyze the film
with respect to the existence of precursor and
oxidic structures. It might then become possible to
optimize the preparation conditions for the oxide
Uniform thin films of transition metal oxides
may become of importance in tunneling magneto
resistance (TMR) devices [1]. Moreover, well-
characterized oxide films are of interest for chem-
ical applications, i.e., to study adsorption and
reaction [2] and to use them as a model catalyst, in
particular, after further deposition of metal clus-
ters [3]. Since a number of years, we have therefore
made efforts to grow such films and to characterize
them by scanning tunneling microscopy (STM)
and spectroscopy (STS). One of the systems we
have been investigating is Co oxide and we have
already found that well-ordered CoO layers in
* Corresponding author. Tel.: +49-345-552-5572; fax: +49-
345-552-7160.
E-mail address: shantyr@physik.uni-halle.de (R. Shantyr).
0039-6028/$ - see front matter ꢀ 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2004.06.085

R. Shantyr et al. / Surface Science 566–568 (2004) 68–73
69
to obtain an ideal layer structure. Since the oxi-
dation state of Co may also be different [6], the
formation of rock salt CoO and Co3O4 in the
spinel structure is possible, in principle, and a
more detailed analysis of the grown structures in
this respect would be important.
films and Ref. [10] for CaF/Si. We found that the
various surface structures show large differences in
the break down of the potential barrier. Although
the detailed nature of this effect has still to be ex-
plained, it can unambiguously be used for identi-
fication of the grown surface species.
Previously, we have observed that the double
layer CoO(0 0 1) islands show a characteristic
contrast change with regard to the Ag(0 0 1) sub-
strate as a function of sample bias voltage [7].
Qualitatively, for small sample bias voltages which
only allow electric transport through the band gap
region of the oxide by tunneling the measured
height of the oxide in a constant current topog-
raphy must be much smaller than for large volt-
ages where electronic states of the valence and
conduction band of the oxide are available. In fact,
for small sample bias the islands are observed as
depressions below the uppermost Ag layer of the
substrate [7]. This can be explained by the more
localized states of the oxide and modifications of
the substrate states, which both must have a
smaller transmission coefficient for tunneling than
that of the more extended sp-like states above
uncovered Ag(0 0 1). It is clear that this picture is
only applicable for oxide layers thin enough to fit
into the potential barrier between the Ag surface
and the tip. Otherwise, contact between the oxide
surface and the tip will occur and lead to uncon-
trollable effects during the measurements. From
the contrast behavior, the band gap of the oxide
could be estimated and was found to be consistent
with the bulk band gap already for the double
layer CoO(0 0 1) islands [7].
In our present work we focus on the electronic
characterization of the grown structures by mea-
suring local I=U characteristics at a fixed sample
tip distance and by derivation of the normalized
conductivity ðdI=dUÞ=ðI=UÞ. The latter quantity is
normally related to the local density of states [8].
We found, however, that particularly for large
sample bias voltages, contributions to this quan-
tity are observed which have no counterpart in the
density of states but rather are related to the break
down of the tunneling barrier. Previously, this
contribution has not been considered in the inter-
pretation of spectroscopic data from thin insula-
ting layers, see, for example, Ref. [9] for thin MgO
2. Experimental
The experiments were performed in an UHV
chamber with a background pressure of better than
7 · 10ꢀ11 mbar. CoO precursors and the CoO layers
were prepared by deposition of Co with a typical
rate of 0.2 monolayer (ML) per minute in an O2
atmosphere of 10ꢀ6 mbar onto a clean Ag(0 0 1)
surface. Substrate temperatures between 300 K
(room temperature) and 470 K have been used. In
some cases the samples have been annealed subse-
quently in UHV or in O2 atmosphere as stated
below. In contrast to our previous experiments
[4,7], low-temperature STM instrumentation as
described in Ref. [11] has been used for the STM
and STS measurements. It turned out that the
stability of the tunneling conditions at low tem-
perature (100 K) was considerably better, com-
pared to 300 K. At room temperature, the large
sample bias voltages (between )6 and +6 V) needed
for the measurements of insulators frequently lead
to instabilities due to tip changes and desorption.
W tips were prepared by electrochemical etching in
NaOH solution and subsequent rinsing in isopro-
panol. In order to explore the influence of the tip
material on the STS measurements PtIr tips have
also been used. They were etched in a melt of
NaOH and NaNO3 by applying an alternating
current. After etching the PtIr tips were rinsed in
H2O, glowed in an ethanol flame and heated for
several hours at 390 K under UHV conditions.
For the STS measurements sample bias voltages
of at least 2–3 V (both negative and positive in
order to access the occupied and unoccupied states,
respectively) are needed to include the valence band
maximum and conduction band minimum of the
oxide [12]. As a consequence, very low tunneling
currents (below 10 pA) were observed in the band
gap region with a correspondingly large rela-
tive noise level. The I=U characteristics have been

70
R. Shantyr et al. / Surface Science 566–568 (2004) 68–73
recorded at each pixel of the sample surface at a
fixed sample tip distance, which was set by the
stabilization voltage U0. The ðdI=dUÞ=ðI=UÞ val-
ues were numerically determined from the I=U
dependencies and could be obtained in a confident
way only by averaging over many pixels of the
individual surface feature. For negative sample
bias voltages (tunneling current from the occupied
part of the electronic states of CoO) an influence of
the tip material was discernible and will be de-
scribed later. In the normal STM mode constant
current topographies (CCT) have been recorded as
usually.
latter islands increases with the annealing time (see
also Ref. [4]). The islands with the more rounded
shape have previously been identified as double
layer CoO(0 0 1) structures on top of Ag(0 0 1).
While the resolution of the image is not yet suffi-
cient to give a detailed analysis of atomic struc-
tures, it is nevertheless possible to differentiate
between rectangular islands showing a row struc-
ture on top (type A) and those, which have
developed a very flat surface (type B). Both types
of islands are transition states towards the for-
mation of CoO(0 0 1).
From the analysis of additional deposition and
annealing experiments we have to conclude that
islands of type A constitute the first step towards
the formation of CoO(0 0 1) and those of type B the
second one. Both structures already contain suffi-
cient O to develop double layer CoO(0 0 1) without
further O supply. More insight into the atomic
structure of the precursor states A and B can be
derived from measurements with better resolution,
which are reproduced in Fig. 2. In Fig. 2(a) the row
structure on top of the islands of type A is clearly
seen. Although the rows have not yet developed a
very regular pattern, the same kind of subunits was
found on many parts of the surface. They are ar-
ranged in pairs in a distance of 1.7 nm and
depending on the degree of lateral ordering form
stripes of protrusions. According to our analysis,
the observed islands showing the rows could be
explained as one monolayer of Co covered by
adsorbed O in sufficient quantity to form CoO by
further annealing in UHV. During the initial
growth stage of NiO(0 0 1) on Ag(0 0 1) we found a
similar feature of a Ni monolayer structure covered
with O, which did not yet correspond to NiO but
had also to be annealed further to develop oxidic
behavior [13]. In case of the NiO precursor a well-
ordered 2 · 1 pattern in registry with substrate
lattice could be observed. Due to the larger lattice
mismatch of CoO/Ag(0 0 1) compared to NiO/
Ag(0 0 1), the smaller degree of order for the CoO
precursor state is qualitatively expected.
3. Results and discussion
We first show and discuss CCTs of samples with
well-defined features obtained during the growth
process. Previously, we have already shown that
room temperature deposition of Co in O2 leads to
a fairly undefined surface morphology containing
both CoO species and Ag islands [4]. The latter
islands result from exchange processes of Ag
atoms from the substrate surface by Co and sub-
sequent formation of CoO species also within the
Ag surface layer. After shortly annealing such a
surface in UHV the islands on top of the substrate
surface mostly develop a rectangular and flat form
with edges along [1 1 0]-like direction. In Fig. 1
these islands are marked by A and B. A smaller
number of islands show a more rounded shape (see
island marked by C) with indications of straight
edges in [1 0 0]-like directions. The number of the
Upon further annealing type B islands develop.
As can be seen in Fig. 2(b), the inner parts of those
islands appear to be completely flat while at the
edges atomic features are recognized with a dis-
tance of 0.3 nm. In some experiments (for exam-
Fig. 1. Overview image. 0.3 ML deposited at 390 K and sub-
sequently annealed at 420 K for 15 min in UHV. The area is 50
nm · 28 nm, U ¼ ꢀ3 V, I ¼ 100 pA. Two types of precursor
states are visible (A and B), C corresponds to a double layer
CoO(0 0 1) island.

R. Shantyr et al. / Surface Science 566–568 (2004) 68–73
71
Fig. 2. High-resolution CCTs of precursor and oxidic structures towards the formation of CoO(0 0 1). (a) The area is 15 nm · 15 nm,
U ¼ ꢀ3 V, I ¼ 100 pA. (b) The area is 5 nm · 5 nm, U ¼ ꢀ0:25 V, I ¼ 100 pA. (c) The area is 6 nm · 6 nm, U ¼ ꢀ1 V, I ¼ 100 pA.
The images have been processed to enhance the contrast of the corrugation.
ple, Ref. [14] and during the present work) we
observed type A and type B structures in one is-
land. In the type B part of the measurements
an atomic corrugation expected for a CoO(0 0 1)
surface was identified. At the same time, Ag could
not be resolved atomically on uncovered parts of
the substrate surfaces. We therefore conclude that
the inner part of the type B islands does no more
correspond to a metallic surface. Otherwise it
would be difficult to explain why type B islands do
show an atomic corrugation and the Ag surface
does not. In Fig. 2(b) an atomic corrugation for
the inner part is not observed, which could be
explained by limitations in the imaging conditions
in this case. The type B islands also contain oxygen
in sufficient quantity as can be deduced from their
transformation to double layer CoO(0 0 1) without
additional O2 exposure.
A high-resolution double layer CoO(0 0 1) is-
land is displayed in Fig. 2(c). It has to be empha-
sized that the formation of double layer CoO(0 0 1)
islands is accompanied by a reduction of the sur-
face area covered by islands by a factor of 2. The
atomic corrugation on the surface of the island
shown in Fig. 2(c) is consistent with that expected
for CoO(0 0 1). The inner part of the surface is
perfectly ordered and does not show defects. The
edges are formed by protrusions in twofold peri-
odicity with regard to the inner part of the surface
plane. Their arrangement looks similar to that on
the edges of the type B monolayer precursor.
We now demonstrate how these Co/O species
can be distinguished from each other by their
electrical behavior in comparison to clean
Ag(0 0 1). For this purpose we have recorded local
I=U characteristics simultaneously to the mea-
surement of a CCT, which is reproduced in Fig.
3(a). In this particular measurement type A
(striped) islands and type B (flat) islands are present
on top of Ag(0 0 1). The averaged I=U character-
istics are shown in Fig. 3(c). We have included the
I=U characteristic from a double layer CoO(0 0 1)
island, which has been obtained on a different
sample. Reproducibility of the I=U characteristics
has been confirmed in various experiments.
For the measurements, the sample bias used for
fixing the sample tip distance has been set to a
value of )3.5 V with a tunneling current of 100 pA.
As a consequence, the I=U characteristics for
negative sample bias voltages can hardly be dis-
tinguished from each other, for U ¼ ꢀ3:5 V they
start at the common value of 100 pA (Fig. 3(c)).
For positive sample bias, the characteristics show
distinct differences for the different surfaces. For
Ag(0 0 1) a slight increase of the tunneling current
up to 300 pA is observed, which for the oxidic
surfaces is much more drastic. The steepest in-
crease is measured for the double layer CoO(0 0 1)
islands, where tunneling current of nearly 50 nA is
obtained at U ¼ 3:5 V. The type B precursor states
(dashed line) show a similar behavior. Compared
to double layer CoO(0 0 1), the onset of the tun-
neling current is slightly shifted towards a lower
sample bias, but the increase of I with U is also
very pronounced. The type A precursor (dotted
line) shows an intermediate behavior between the
aforementioned characteristics of the double layer
CoO(0 0 1) and the type B islands and the char-
acteristic of clean Ag(0 0 1). This already indicates
that the type A precursor has to be considered as

72
R. Shantyr et al. / Surface Science 566–568 (2004) 68–73
Fig. 3. STS results from CoO precursor and oxidic states. (a) CCT image of the area, where the STS maps were acquired. The area is
50 nm · 50 nm, U0 ¼ ꢀ3:5 V (stabilization voltage), I ¼ 100 pA. (b) STS map presenting the spatial distribution of the tunneling
current at U ¼ 2:7 V. (c) I=U spectra after averaging over the entire area of the specific structure. (d) Normalized tunneling con-
ductance ðdI=dUÞ=ðI=UÞ calculated from the I=U spectra shown in (c).
the first step towards the formation of an oxidic
film and that the type B islands cannot be far away
from an oxide. The apparent similarity of the
characteristics obtained on double layer CoO-
(0 0 1) and the type B islands with those of a
semiconductor pn junction [15] is striking.
The large differences in the I=U characteristics
can be used to identify the various surface struc-
tures by their current contrast, which has been
mapped in Fig. 3(b) for a sample bias of 2.7 V. As
a matter of fact, the type B islands develop the
highest current and are therefore displayed in the
brightest gray tone level. The type A islands show
an intermediate gray tone.
For negative sample bias voltages the curves show
a number of smaller peaks, which certainly lie
above the statistical uncertainty. Their physical
origin is not yet clear, however. For example, for
Ag(0 0 1) two distinct peaks are seen (at )2.4 V
and )1.2 V). These peaks are not directly related
to the density of states of Ag(0 0 1) which is fairly
constant in this energy range [16]. The data of the
oxidic structures are also difficult to explain. For
negative sample bias voltages, an influence of the
density of states of the tip cannot be excluded since
these structures below 0 V showed distinct changes
upon changing the tip material. For positive
sample bias voltages, the curves show more pro-
nounced maxima and the observed peaks are cer-
tainly beyond the experimental limitations by tip
material and state. Ag(0 0 1) shows a pronounced
From the I=U characteristics we have numeri-
cally evaluated the normalized conductivity
ðdI=dUÞ=ðI=UÞ. The results are shown in Fig. 3(d).

R. Shantyr et al. / Surface Science 566–568 (2004) 68–73
73
peak at around 1.6 V, which agrees very well with
the one-dimensional density of states due to bulk
transitions as observed in inverse photoemission
experiments [17]. For the oxidic islands a number
of pronounced peaks are found. To our opinion,
they result from both density of states properties
and from the behavior of the potential barrier
between sample and metallic tunneling tip as a
function of sample bias. For a more detailed
analysis of the observed peaks more theoretical
work is needed, which is not yet available.
support by R. Kulla is gratefully acknowledged.
This work has been supported by the Deutsche
Forschungsgemeinschaft through the For-
schergruppe ‘‘Oxidic Interfaces’’.
References
[1] J.S. Moodera, L.R. Kinder, T.M. Wong, R. Meservey,
Phys. Rev. Lett. 74 (1995) 3273–3276.
[2] H.-J. Freund, Phys. Status Solidi (b) 192 (1995) 407.
€
[3] M. Baumer, H.-J. Freund, Prog. Surf. Sci. 61 (1999) 127.
[4] I. Sebastian, Th. Bertrams, K. Meinel, H. Neddermeyer,
Faraday Discuss. 114 (1999) 129.
4. Conclusion
[5] J. van Elp, J.L. Wieland, H. Eske, P. Kuiper, G.A.
Sawatzky, F.M.F. de Groot, T.S. Turner, Phys. Rev. B 44
(1991) 6090.
We have demonstrated that during the growth
of thin CoO(0 0 1) films a number of well-defined
transition of precursor states are observed. Their
occurrence is related to the annealing stage of the
deposit, which indicates their nature as interme-
diates towards the formation of a thin oxidic film.
From an experimental point of view, the identifi-
cation of such structures is important, if one is
interested in a uniform film. We have shown that
the different kinds of islands can already be iden-
tified by their morphology and surface structure.
The oxidic nature of the observed precursor and
oxide islands can be analyzed by measuring the
local I=U characteristics. Although a full expla-
nation of these dependencies cannot yet been
given, we have shown that they are sensitive fin-
gerprints for their nature and can therefore be used
for identification.
[6] P.A. Cox, Transition Metal Oxides. An Introduction to
Their Electronic Structure and Properties, Clarendon
press, Oxford, 1995.
[7] I. Sebastian, H. Neddermeyer, Surf. Sci. 454–456 (2000)
771.
[8] R.M. Feenstra, J.A. Stroscio, A.P. Fein, Surf. Sci. 181
(1987) 295.
[9] S. Schnitke, S. Messerli, M. Pivetta, F. Patthey, L.
Libioulle, M. Stengel, A. De Vita, W.-D. Schneider, Phys.
Rev. Lett. 87 (2001) 276801-1.
[10] J. Viernow, D.Y. Petrovykh, A. Kirakosian, J.-L. Lin,
F.K. Men, M. Henzler, F.J. Himpsel, Phys. Rev. B 59
(1999) 10356.
[11] H. Wengelnik, D. Badt, H. Neddermeyer, Surf. Sci. 307–
309 (1994) 619.
[12] Ch. Hagendorf, R. Shantyr, K. Meinel, K.-M. Schin-
dler, H. Neddermeyer, Surf. Sci. 532–535 (2003) 346–
350.
[13] Th. Bertrams, H. Neddermeyer, J. Vac. Sci. Technol. B 14
(2) (1996) 2122.
€
[14] I. Sebastian, Ph.D. Thesis, Martin-Luther-Universitat
Halle-Wittenberg, 2000.
[15] C. Kittel, Introduction to Solid State Physics, 7 Ed., John
Wiley, New York, 1996.
Acknowledgements
€
[16] U. Konig, P. Weinberger, J. Redinger, H. Erschbaumer,
A.J. Freeman, Phys. Rev. B 39 (1989) 7492.
[17] F.J. Himpsel, J.E. Ortega, Phys. Rev. B 46 (1992) 9719.
We thank K. Meinel, K.-M. Schindler and W.
Widdra for fruitful discussions. The technical