Interaction of cobalt with the Si(1 0 0)2 × 1 surface studied by photoelectron spectroscopy

Article abstract of DOI:10.1016/j.susc.2005.01.034

Cobalt adsorption and condensation on the Si(1 0 0)2 × 1 surface as well as solid-phase reaction of CoSi₂ formation have been studied by high-resolution photoelectron spectroscopy with synchrotron radiation. We have measured the Si 2p and valence-band spectra after the Co deposition from a submonolayer coverage to 6 ML thickness and a subsequent annealing to 600 °C. Room temperature Co adsorption results in the loss of the initial surface reconstruction, and the chemisorbed Co atoms appear to be embedded into the upper layer of Si(1 0 0); however, no stable CoSi₂ was observed. With further metal deposition, a discontinuous solid solution CoSi film was formed and the dissolved Si concentration decreased with the distance from the crystal surface. The formation of cobalt disilicide was found to occur in the range of 250-400 °C.

Full text of DOI:10.1016/j.susc.2005.01.034

Surface Science 578 (2005) 174–182
www.elsevier.com/locate/susc
Interaction of cobalt with the Si(100)2 · 1 surface studied
by photoelectron spectroscopy
a
M.V. Gomoyunova , I.I. Pronin , N.R. Gall ,
a,
a
*
b
D.V. Vyalikh , S.L. Molodtsov
c
a
c
A.F.Ioffe Physico-Technical Institute, Russian Academy of Sciences, 194021 St. Petersburg, Russian Federation
b
Institut f u¨ r Festk o¨ rperphysik, Technische Universit a¨ t Dresden, D-01062 Dresden, Germany
Institut f u¨ r Oberfl a¨ chen- und Mikrostrukturphysik, Technische Universit a¨ t Dresden, D-01062 Dresden, Germany
Received 28 November 2003; accepted for publication 20 January 2005
Available online 11 February 2005
Abstract
Cobalt adsorption and condensation on the Si(100)2 · 1 surface as well as solid-phase reaction of CoSi
2
formation
have been studied by high-resolution photoelectron spectroscopy with synchrotron radiation. We have measured the
Si2p and valence-band spectra after the Co deposition from a submonolayer coverage to 6 ML thickness and a subse-
quent annealing to 600 ꢀC. Room temperature Co adsorption results in the loss of the initial surface reconstruction, and
2
the chemisorbed Co atoms appear to be embedded into the upper layer of Si(100); however, no stable CoSi was
observed. With further metal deposition, a discontinuous solid solution CoASi film was formed and the dissolved Si
concentration decreased with the distance from the crystal surface. The formation of cobalt disilicide was found to
occur in the range of 250–400 ꢀC.
ꢁ
2005 Elsevier B.V. All rights reserved.
Keywords: Silicon; Cobalt; Solid phase epitaxy; Synchrotron radiation photoelectron spectroscopy
1
. Introduction
technological interest, so it has been studied exten-
sively for over 20 years [1–19]. This is mostly due
to the fact that CoSi2 formed at the interface
exhibits a metallic conductivity, a high thermal
stability, and, what is of especial importance, it
can produce an atomically sharp interface with a
Si substrate. The latter allows one to treat this sys-
tem as a model metal–semiconductor interface.
These properties of the system have provided its
The interaction between atomic Co and a Si sin-
gle crystal surface is of great fundamental and
*
Corresponding author. Tel.: +7 81 2247 9948; fax: +7 81
247 1013.
E-mail address: igor.pronin@mail.ioffe.ru (I.I. Pronin).
2
0
039-6028/$ - see front matter ꢁ 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.susc.2005.01.034

M.V. Gomoyunova et al. / Surface Science 578 (2005) 174–182
175
wide use in modern semiconductor electronics.
The formation of the CoSi /Si interface has been
discussion requiring a further research effort. The
aim of the present work was to investigate the ear-
liest stages of the interaction between Co and
Si(100)2 · 1, both at room and elevated tempera-
tures. The main technique used was core-level
photoelectron spectroscopy with synchrotron
radiation.
2
thoroughly studied for the Si(111)7 · 7 surface,
while the mechanism of the CoASi(100)2 · 1
interaction is less clear, although it is this surface
that is most important for silicon technology.
The initial interaction which determines the output
characteristics of the interface also remain poorly
understood. Indeed, the results obtained by differ-
ent researchers are contradictory in many aspects.
It was shown in Ref. [13] by coaxial impact col-
lision ion scattering spectroscopy that room tem-
perature adsorption of Co does not lead to any
atomic surface reconstruction. At least two Co
monolayers (MLs) can be adsorbed by an initially
reconstructed surface, occupying consecutively
three types of adsorption sites without the forma-
2. Experimental
The experiments were carried out with the Rus-
sian–German beamline, which was recently created
at the Berliner Elektronenspeicherring f u¨ r Syn-
chrotronstrahlung (BESSY), at a photon energy
(hm = 130 eV) providing a maximum surface sensi-
tivity of the Si2p core levels. The photoelectron
(PE) spectra were recorded in a wide solid angle
oriented around the normal to the surface of the
sample at room temperature. The overall energy
resolution of the analyzer and the monochromator
was set to 140 meV FWHM. The substrates used
were Si(100) P-doped wafers with the P concentra-
tion of CoSi . Recently, the Co adsorption on a
2
Si(100)2 · 1 surface without its additional recon-
struction was also considered in a theoretical study
[
12], where Co atoms were found to be adsorbed
preferably by high coordination sites. One should
note, however, that there are some discrepancies
between the binding energies of Co on Si(100)
1
7
3
tion of 10 at./cm . The accuracy of the (100) facet
was better than 0.1ꢀ. The samples were ex situ
prepared by a wet chemical treatment using the
method of Shiraki [20]. To clean the surfaces in
vacuum the samples were flashed at 1200 ꢀC and
cooled slowly with the rate of less than 50 ꢀC/min
to provide a well-defined (2 · 1) LEED pattern
and the surface uncontaminated with carbon and
oxygen.
2
· 1 calculated in [12] and the sequence of the sur-
face site filling found experimentally in [13].
A number of workers claim that adsorbed Co
reacts immediately with Si(100)2 · 1, giving rise
to CoSi nuclei. As reported in Ref. [14], the reac-
2
tion involves only Co atoms from the first mono-
layer, and a metal Co film with diffused Si atoms
grows at larger coverage. On the contrary, it was
found in Ref. [15] that Co atoms from 2 to
The chemical composition of the surfaces was
monitored using the photoemission spectra. Co-
balt was evaporated from a thoroughly degassed
source, in which a wire of highly purified deposited
material (99.99%) was heated by electron bom-
3
CoSi . The authors of Ref. [16] used a surface-ex-
MLs were involved in the reaction to produce
2
tended X-ray absorption fine-structure (SEXAFS)
technique to show that the interaction and the sil-
icide growth mechanism depend strongly on the
initial state of the surface, or on its preparation
and purification procedures. In particular, a loss
of the initial reconstruction was observed after
the adsorption of the submonolayer Co on a chem-
ically treated and annealed surface.
1
3
2
bardment. The Co flux was 2 · 10 at./cm s in
the standard operation mode. The Co source was
calibrated in two ways: by using a standard quartz
microbalance and by attenuating the W photoelec-
tron signal of a special tungsten sample pre-
exposed to a Co flux. We estimated the error of
the Co coverage to be less than 20%. Cobalt was
deposited at room temperature in amounts of
0.6–1 ML with Si2p and the valence-band spectra
were measured for each dose (1 ML is considered
Thus, the fundamental problems of CoASi
(
100)2 · 1 interaction, such as the changes in the
surface structure induced by adsorption, the type
of Co adsorption sites, and the production of
CoSi nuclei at room temperature, are still under
1
4
2
to be 6.8 · 10 at./cm , which is equal to the
2

176
M.V. Gomoyunova et al. / Surface Science 578 (2005) 174–182
concentration of Si atoms on the (100) surface).
After the total Co amount of 6 ML was deposited,
the sample was gradually annealed in short-term
cycles with increasing temperature. All the mea-
surements were performed in vacuum higher than
that the surface atoms appear to be in a special
state, as compared to bulk atoms, due to a break
in the translational symmetry on the crystal sur-
face. This produces the surface core-level shifts
[21] and gives rise to surface electronic states in
the valence band [22]. The core electron spectra
of complex reconstructed surfaces normally show
several surface modes of atoms with various local
geometry. The Si2p spectrum of the Si(100)2 · 1
surface, which is characterized by the presence of
asymmetric dimers along the h110i substrate
directions [23] (see Fig. 2a), was first found to ex-
hibit four surface modes [24], but later their num-
ber was increased to five [25,26]. In these works the
modes S and S were attributed to the upper and
ꢀ
10
2
· 10
Torr.
3
. Results and discussion
3
.1. Si(100)2 · 1 photoelectron spectra
A representative Si2p spectrum from a pure
sample obtained by subtraction of the background
fitted by the Shirley method is shown in Fig. 1. Be-
fore discussing its details, we would like to note
u
d
down dimer atoms, correspondingly. The energy
shift of the Su component is about ꢀ500 meV.
As for the component S there are two points of
d
view concerning the sign of its energy shift, and
this is the major difference between the various
studies. In Refs. [24,26], the S shift was found
d
to be positive and equal to 60 meV [24] or lying
in the range of 30–130 meV [26]. In Ref. [25], the
component Sd had a negative shift, ꢀ24 meV.
The third surface mode S was attributed to the
1
atoms of the second monolayer underlying the di-
mers. The interpretation of the other, less intensive
0
0
modes (S in [24], S and S in [25,26]) is still more
2
2
2
ambiguous. For example, the authors of [25] attri-
bute them to the third layer atoms located under
and between the dimers. However, the authors of
0
Ref. [26] believe that the component S is due to
2
Fig. 1. Si2p core-level spectrum with the subtracted back-
ground recorded from the Si(100)2 · 1 surface at room
temperature with the photon energy of 130 eV. The solid curve
through the data points shows the resulting fit. The decompo-
sition of the spectrum into bulk subspectrum B and five surface
subspectra is indicated only for the 2p3/2 component (thin lines).
The designations of the subspectra (see text) are shown in the
upper part of the figure.
Fig. 2. Schematic atomic structure of the Si(100)2 · 1 surface
(a) and the surface covered with 0.5 ML of Co (b). 1–4—Si
atoms of the first through fourth monolayers, respectively; 5—
chemisorbed Co atoms.

M.V. Gomoyunova et al. / Surface Science 578 (2005) 174–182
177
0
0
a half of the third layer and the component S is
2
likely to be associated with vacancy-type defects.
Taking the above results into account, we used
five surface components in addition to the bulk
one when analyzing the Si2p spectrum by means
of the least square fit procedure. The spectral fea-
tures were described by the Voigt functions [24]
commonly used for this purpose. Such a function
is a convolution of a Lorentzian function account-
ing for the lifetime of a core-level hole and a
Gaussian function simulating the phonon broad-
ening and the instrumental resolution. The
FWHM of the Lorentzian (70 meV), the branching
ratio of the 2p3/2 and 2p1/2 components (0.5) and
their splitting (608 meV) were taken to be con-
stant. The FWHM of the Gaussian, the relative
binding energies of the individual components as
well as their intensities were varied during the fit-
ting. The best agreement with the experiment was
achieved when the Gaussian width was 300 meV.
These findings are illustrated in Fig. 1. One can
see that the spectral analysis yielding one bulk
component and five surface ones provides a fairly
good fit between the calculations and measure-
ments. The energy shifts of the surface compo-
nents relative to the bulk one are found to be
Fig. 3. Valence-band spectra from samples with different
amounts of Co deposited at room temperature, hm = 130 eV.
Spectra are normalized to the initial photon flux.
both valence and Si2p electrons (Figs. 3 and 4).
It is clear from Fig. 4 that the 0.6 ML Co deposi-
tion leads to the disappearance of the step in the
low energy 2p3/2 sublevel slope of the Si doublet,
which is due to the S mode of the upper atoms
u
in asymmetric dimers. This mode is quite isolated
from the other spectrum, allowing one to follow
its behavior during the adsorption. It is important
to note that the upper dimer atoms are localized in
the uppermost substrate layer and that these atoms
are most sensitive to adsorption. For this reason,
our analysis of the spectrum clearly revealed only
the S mode and the conjugate S mode, the others
ꢀ
510 ± 20, ꢀ240 ± 20, 90 ± 20, 180 ± 20, and
330 ± 20 meV. These values are in good agreement
with the available data cited above, so we also
attribute the spectral features at ꢀ510 and
9
0 meV to the S and S modes, respectively. The
u d
large difference in the shifts is associated with the
charge transfer from the down dimer atoms to
the upper ones [27]. The component with the
u
d
being united into the B component. While this ap-
1
1
80 meV shift is the S mode. The last two modes
1
proach is somewhat simplified, it is often used to
study adsorption processes. For example, it was
successfully employed in [17] to study the system
concerned.
are due to the atoms in the third layer (Fig. 1).
According to Ref. [27], the difference in their shifts
is due to the variation in the final state contribu-
tion for the atoms underlying the dimers and for
those located between them.
The spectral analysis illustrated in Fig. 4 shows
that the 0.6 ML cobalt adsorption does lead to a
complete disappearance of the S mode. But simul-
A valence-band spectrum of the silicon surface is
shown in Fig. 3. It is similar to that reported in Ref.
u
taneously, a new S component arises with the en-
ergy shift of ꢀ290 meV, whose intensity is lower
than the initial S intensity. The S disappearance
[
17], being typical for the photon energy used.
u
u
3
.2. Cobalt adsorption on Si(100)2 · 1
may result from either the disappearance of dimers
or a strong chemical shift arising from the interac-
tion between Si atoms and the adsorbate. It seems
unlikely that asymmetric dimers persist on the
The Co adsorption on Si(100)2 · 1 at room
temperature changes appreciably the spectra of

178
M.V. Gomoyunova et al. / Surface Science 578 (2005) 174–182
the spectrum indicate an atomic rearrangement
of the Si surface due to Co adsorption involving,
at least, the upper dimer atoms.
Note that an atomic rearrangement has often
been registrated in the adsorption of metals on this
Si surface. Specifically, I-, II- and III-group metals
typically transform asymmetric dimers to symmet-
ric ones. For instance, authors [29] observed a
gradual disappearance of the Su mode and the
appearance of a new component between S and
u
the bulk mode due to the K adsorption. The inten-
sity of the new feature for a monolayer coverage
became twice as high as that of S , because the lat-
u
ter corresponds to a half of the Si monolayer,
whereas the symmetric dimer mode is determined
by the whole monolayer. In our case, the S inten-
sity is too low to identify it with a symmetric dimer
mode, although S is located between S and B .
u
1
Therefore, we believe that the Co adsorption on
the Si(100)2 · 1 surface leads to the loss of its di-
meric reconstruction. A similar idea was discussed
in Ref. [16].
As for the adsorption sites of chemisorbed Co
atoms, our findings fit well with the adsorption
model suggested in the work [16] using the SEX-
AFS method to study this system. In agreement
with this model, we think that Co atoms are incor-
porated into the upper monolayer of the unrecon-
structed crystal surface in between four Si atoms
to produce seven interatomic bonds with the sub-
strate. Two Si atoms from the first coordination
sphere of the chemisorbed Co are located in the sec-
ond layer and constitute bridge bonds. The other
five Si atoms are located in the second coordination
sphere; four atoms reside in the upper monolayer
around it, interacting with the Co atom to saturate
the surface dangling bonds. The last one is in the
third subsurface layer directly under the adatom.
One can find in the literature some additional
arguments in favor of this model. One is the con-
clusion from a theoretical study [12] that the
adsorption of Co atoms on the Si(100)2 · 1 sur-
face energetically favors the formation of a maxi-
mum number of interatomic bonds with Si (seven
in our case). Then, the model discussed is consis-
tent with the idea that all dangling bonds of sur-
face atoms are preferably saturated by adsorbed
particles. When the adatoms are localized at the
Fig. 4. Si2p core-level spectra taken at hm = 130 eV from
samples with different amounts of Co deposited at room
temperature. Spectra are normalized to the maximum intensity.
surface whereas Su disappears because of the
chemical interaction with the adsorbate, since this
mode experiences only a slight energy shift for a
chemically active metal like cesium, whose adsorp-
tion does not lead to the disappearance of asym-
metric dimers [28]. So, the observed changes in

M.V. Gomoyunova et al. / Surface Science 578 (2005) 174–182
179
sites, the saturation becomes possible if the adpar-
ticles are grouped into parallel atomic rows ori-
ented in the same way as the dimer rows on a
clean Si(100)2 · 1 surface (Fig. 2b). The (2 · 1)
diffraction pattern of Co adsorption obtained in
Refs. [16,17] is another evidence to support the
above adsorption model. It is also easy to see that
the filling of these sites with adatoms should occur
if the coverage H is equal to 0.5. The fact that we
differs strongly from the spectrum of CoSi (see
2
Section 3.4) showing once more that no clusters
of this stable phase arise during the Co adsorption
on Si(100)2 · 1 at room temperature.
3.3. Cobalt condensation on silicon at room
temperature
One can see in Fig. 4 that the relative intensity
of the B1 component decreases, whereas the
ꢀ300 meV feature monotonically rises in the
whole range of thicknesses, at least, up to 6 ML.
It is marked with the letter C in Fig. 4, since its ori-
gin seems to be different from that of the S mode
for chemisorbed atoms, which we observed in the
submonolayer coverage and which should become
weaker in polylayer coverages. The energy posi-
tion of C falls within the range of expected energy
shifts for atomic Si in the solid solution CoASi,
which was found in [14,17,18,31] to be ꢀ(200–
400) meV. The conclusion about the solid solution
formation was also supported by Auger electron
spectroscopy [14]. We attribute the C component
to the atomic Si in the solid solution CoASi as
well. The reason for the coincidence in the energy
positions of the C and S features requires a further
study. The negative sign of the C mode may result
from the enhancement of the interatomic relaxa-
tion due to the Co3d valence electrons.
did not observe the dimer S mode at H = 0.6 (Fig.
u
1
) close to a saturation coverage also supports this
model. Finally, we would like to comment on the
negative shift of the new S mode. It was pointed
out in Ref. [19] that a negative shift is characteris-
tic of adamantine disilicide, which is produced
when Co atoms find themselves in tetrahedral
voids of the Si lattice [30]. The adsorption sites
we are discussing correspond to the atomic Co
positions in the upper layer of just this kind of sil-
icide. One could keep in mind that the equilibrium
phase of CoSi with the CaF -type of structure
2
2
exhibits a positive energy shift [17–19,31]. The fact
that we did not observe such a spectral feature al-
lows the conclusion to be made that no nucleation
centers of a stable CoSi -like compound are
2
formed on the Si(100)2 · 1 surface at room tem-
perature. This point of view is supported by the
authors of [17] but it disagrees with the conclusion
in [14] that this kind of compound is indeed pro-
duced after the deposition of the first Co mono-
layer. A possible reason for this discrepancy,
which can also be found in studies made by other
techniques, is the sensitivity of the CoASi
It seems surprising at first sight that the intensi-
ties of the Si bulk component (B ) and the C com-
1
ponent should be close to each other after the
deposition of 6 ML of Co. Indeed, the bulk silicon
is expected to be buried under the film of solid
solution. We interpret this effect assuming that
CoASi solid solution films are not continuous,
and some of the surface remains to be covered only
with chemisorbed Co. However, this area cannot
be large considering the relation between the inten-
sities of the components and a low Si concentra-
tion in the solid solution. Note that the
abnormally high intensity of the bulk Si compo-
nent was also observed in [17] after the deposition
of 5 ML of Co on Si (100)2 · 1 at room tempera-
ture. Moreover, the intensity of this component
was even higher than that of the Si component in
the CoASi solid solution. The authors of Ref.
[17] explained this effect as being due to pinhole
(
100)2 · 1 interaction to the pretreatment of the
initial crystal surface [16]. Note also that the sam-
ple used in Ref. [14] was cut with the accuracy of
1
ꢀ only.
One can conclude from the analysis of the va-
lence-band spectra in Fig. 3 that the adsorption
of 0.6 ML of Co results in essential spectral
changes. There is a new broad peak located at
1
.3 eV relative to the Fermi level. As the cross-sec-
tion of photoexcitation of Co3d valence electrons
is much larger than those of Si3s and 3p electrons
at the photon energy used, the major contribution
to the observed maximum is made by the Co3d
electrons. This means that the spectrum in Fig. 3
(
0.6 ML) indicates the chemisorbed Co state. It

180
M.V. Gomoyunova et al. / Surface Science 578 (2005) 174–182
formation. Contrary to the authors of [14], we be-
lieve that the CoASi solid solution is not uniform
in depth, the Si concentration being decreased with
approaching the surface and with growth of the
film thickness. This idea is supported by the ab-
sence of surface components of this phase in the
spectra. Indeed, if the solid solution were uniform
in depth, the spectrum for 6 ML Co (Fig. 4) would
have a surface mode in addition to the intensive
bulk C mode. Its absence indicates that there are
no Si atoms at the surface and in the near-surface
layer, or their concentration is very low. The spec-
tra of valence electrons also supports the CoASi
solid solution nonuniformity (see below).
The valence-band spectra in Fig. 3 show that
the deposition of 0.6–6 ML of Co results in a grad-
ual growth of the photoelectron intensity over the
whole energy range, while the energy distribution
maximum shifts to lower BEÕs, i.e. to the Fermi le-
vel. When going from 0.6 to 1.2 ML of Co cover-
age, the main peak becomes broader and slopes
more and steeper toward the Fermi energy. With
further Co deposition, a new peak arises with the
energy a few tenths of an electron-volt lower than
the Fermi energy. Its position is very close to that
of Co [31] and is an evidence for a gradual conver-
sion of the solid–solution CoASi film to a pristine
cobalt film. This result shows that a film formed
due to the Co deposition at room temperature is
heterogeneous and the Si concentration decreases
toward the surface.
3
.4. Cobalt disilicide formation
At the second stage of the solid-phase epitaxy,
the samples with 6 ML Co were annealed by a
stepwise (each time for ꢁ2 min) heating. After
each anneal, the samples were cooled to room tem-
perature and the photoelectron spectra were taken.
The spectral features indicating a solid-phase reac-
tion were first found at ꢁ250 ꢀC, becoming more
pronounced with increasing temperature. The
Si2p spectra are presented in Fig. 5. Their changes
are opposite to those previously obtained after the
Co deposition. The broad flat feature becomes
gradually double-humped, and a step arises in its
right slope. Eventually it resembles the pure sub-
strate spectrum.
Fig. 5. Si2p core-level spectra taken after the annealing of a
sample covered with 6 ML of Co, hm = 130 eV.
The energy distribution curves were fitted with
the subspectra presented in Fig. 5. One can notice
two main trends in the behavior of the Si2p spec-
tra of the samples annealed at 280 ꢀC, as compared
with ‘‘as deposited’’ sample: (i) the intensities of
the fitting components B and C are reduced, while
1
the signal from the CoASi solid solution is de-
creased much greater, and (ii) a new component
D appears with a positive energy shift of 300 meV.

M.V. Gomoyunova et al. / Surface Science 578 (2005) 174–182
181
Similar positive shifts of the Si2p core-level fea-
tures were observed for cobalt disilicide with a sta-
ble CaF -type structure [14,17,18,31]. A positive
2
shift of 300–350 meV relative to the energy posi-
tion in the stable phase was found for metastable
CoSi2 with a CsCl-like structure [32,33]. From
these data, we attribute the component D to the
Si bulk contribution of cobalt disilicide with the
CaF structure, which is usually grown in a so-
2
lid-phase reaction. Note that its positive shift
may be due to the charge transfer from the Si to
Co sites, an effect known for CoSi2 with the
CaF structure [30]. Relaxation phenomena which
2
decrease the BE of electrons in photoemission
spectra may result in a positive shift, if they be-
come weaker in CoSi than in Si. It is known, how-
2
ever, that the main contribution to the relaxation
shift due to valence electrons is made by inter-
atomic relaxation, which is largest for metals.
Fig. 6. Valence-band spectra recorded after the annealing of a
sample covered with 6 ML of Co, hm = 130 eV.
Since CoSi is a metal (silicon is a semiconductor),
2
reaction were also observed after annealing at tem-
peratures higher than ꢁ250 ꢀC. Comparing the data
in Fig. 6, one can see that the strong peak at 0.4 eV
BE resulting from the metal Co phase becomes
nearly extinct after annealing. A trace of this peak
shows itself as a shoulder in the steep right slope.
There are two maxima of the electronic density
of states of CoSi with the CaF structure [3,35].
it is unlikely that the final state contribution is
smaller for CoSi than for silicon.
2
The temperature increase to 350 ꢀC suppresses
the pristine silicon component B and leads to fur-
ther growth of CoSi component D which becomes
1
2
dominant in the spectra. The component C, which
shows a negative shift, has a lower intensity and
varies its energy position to some extent that
may be interpreted as being due to a different ori-
2
2
The first one is due to Co3d antibonding states
and is located at ꢁ1.7 eV below the Fermi level;
it has a fine structure at a lower binding energy.
The second maximum at ꢀ3.7 eV is attributed to
the bonding states of mixed Si3p and Co3d char-
acter. The maximum with the fine structure was in-
deed observed in the spectra taken after annealing
at T P 350 ꢀC (Fig. 6), when the solid-phase reac-
tion between Co and Si was found to be accom-
plished. The presence of these features at ꢀ1.7,
ꢀ1.1 and ꢀ0.55 eV (indicated by the arrows in
gin. Its energy shift relative to bulk CoSi compo-
2
nent D is nearly ꢀ600 meV, which agrees well with
the value for the surface component S in a
D
CoSi (100) single crystal [34]. The above argu-
2
ments indicate the disappearance of the solid–solu-
tion phase and the termination of the solid-phase
reaction to produce an epitaxial CoSi (100) film.
2
Annealing the sample to 600 ꢀC modifies the spec-
trum again: the intensities of CoSi components D
and S are reduced, while the signal related to the
2
D
the figure) is an additional support for CoSi for-
2
substrate component B1 is increased. Simulta-
neously the surface component S appears indicat-
mation. The reason for the absence of the second
maximum in our spectra is the use of high-energy
photons. This peak was observed only at rather
low photon energies, between 20 and 40 eV
[3,14,35], while it was not monitored at 115 eV
[17], i.e. at a photon energy close to that used in
the present study.
u
ing the production of pinholes, areas of the
exposed reconstructed Si(100)2 · 1 surface. We
suppose that pinholes result from CoSi clustering.
2
The above results are supported by the valence-
band spectra measured in the same experimental
conditions as the Si2p spectra (Fig. 6). Here, the
drastic spectral changes indicating a solid-phase
The crystal annealing to 600 ꢀC reduces the
valence band spectral intensity while its shape is

1
82
M.V. Gomoyunova et al. / Surface Science 578 (2005) 174–182
[2] J. Bean, J. Poate, Appl. Phys. Lett. 37 (1980) 643.
practically preserved unchanged. This is also indic-
ative of a partial exposure of the crystal surface due
to the annealing. Indeed, the photoexcitation cross-
sections for Si s and p electrons are much smaller
than those for Co3d electrons. Therefore, the con-
tribution made by the exposed crystal surface ap-
pears to be small, and the valence band spectrum
[
[
[
[
3] J. Derrien, Surf. Sci. 168 (1986) 171.
4] R.T. Tung, J.Vac. Sci. Technol. A 7 (1989) 599.
5] R.T. Tung, Mater. Chem. Phys. 32 (1992) 107.
6] H. von K a¨ nel, Mater. Sci. Rep. 8 (1992) 193.
[7] C.W.T. Bulle-Lieuwina, Appl. Surf. Sci. 68 (1993) 1.
[8] S.H. Brongersma, M.R. Castell, D.D. Perovic, M. Zinke-
Allmang, Phys. Rev. Lett. 80 (1998) 3795.
[9] I. Goldfarb, G.A.D. Briggs, Phys. Rev. B 60 (1999) 4800.
is determined by the CoSi remaining on the surface,
2
[
[
[
10] R. Stadler, R. Podloucky, Phys. Rev. B 62 (2000) 2209.
11] A.P. Horsfield, H. Fujitani, Phys. Rev. B 63 (2001) 235303.
12] A.P. Horsfield, S.D. Kenny, H. Fujitani, Phys. Rev. B 64
which accounts for the preserved spectral shape.
(
13] W.S. Cho, J.Y. Kim, N.G. Park, I.W. Lyo, K. Jeong, S.S.
2001) 245332.
4
. Summary
[
Kim, D.S. Choi, S.N. Whang, K.H. Chae, Surf. Sci. 453
(
2000) L309.
14] J.M. Gallego, R. Miranda, S. Molodtsov, C. Laubschat,
G. Kaindl, Surf. Sci. 239 (1990) 203.
The use of core-level and valence-band photo-
[
[
electron spectroscopy has enabled us to suggest a
self-consistent physical mechanism of the interac-
tion between Co and Si(100)2 · 1. The adsorption
of Co submonolayers at room temperature leads
to the disappearance of the initial surface recon-
struction. We suppose that a similar effect may oc-
cur in other adsorption systems with a multivalent
adsorbate capable to saturate the substrate dan-
gling bonds. The deposition of larger Co coverages
in the range of 1–6 MLs produces a discontinuous
metallic film with dissolved Si atoms. No silicide
phases have been observed at room temperature.
Annealing of the films at temperatures between
15] U. Starke, W. Weiss, K. Heinz, G. Rangelov, Th. Fauster,
G.R. Castro, Surf. Sci. 352 (1996) 89.
[16] H.L. Meyerheim, U. D o¨ bler, A. Puschmann, Phys. Rev. B
4 (1991) 5738.
4
[
17] G. Rangelov, P. Augustin, J. Stober, Th. Fauster, Phys.
Rev. B 49 (1994) 7535.
[
18] G. Rangelov, P. Augustin, J. Stober, Th. Fauster, Surf.
Sci. 307 (1994) 264.
[19] G. Rangelov, Th. Fauster, Surf. Sci. 365 (1996) 403.
[20] A. Ishizaka, Y. Shiraki, J. Electrochim. Soc. 133 (1986)
6
66.
21] J.W.F. Egelhoff, Surf. Sci. Rep. 6 (1987) 253.
[
[
22] W.S. Yang, F. Jona, P.M. Marcus, Phys. Rev. B 28 (1983)
2
049.
[23] P. Koke, A. Goldmann, W. Monch, K. Wolfgarten, J.
Pollmann, Surf. Sci. 152/153 (1985) 1001.
250 and 350 ꢀC gives rise to cobalt disilicide in
agreement with [14,17]. At higher temperatures
[
24] E. Landemark, C.J. Karlsson, Y.-C. Chao, R.I.G. Uhr-
berg, Phys. Rev. Lett. 69 (1992) 1588.
(
ꢁ600 ꢀC), the area covered with CoSi becomes
2
smaller, showing exposed surface regions.
[
25] T.-W. Pi, C.-P. Ouyang, J.-F. Wen, L.-C. Tien, J. Hwang,
C.-P. Cheng, G.K. Wertheim, Surf. Sci. 514 (2002) 327.
Acknowledgements
[26] H. Koh, J.W. Kim, W.H. Choi, H.W. Yeom, Phys. Rev. B
7 (2003) 073306.
6
[
[
27] E. Pehlke, M. Scheffler, Phys. Rev. Let. 71 (1993) 2338.
28] Y.-C. Chao, L.S.O. Johansson, R.I.G. Uhrberg, Phys.
Rev. B 54 (1996) 5901.
The study was supported by the Russian Foun-
dation for Basic Research (Project No. 01-02-
1
7288), the Ministry for Industry, Science and
Technologies (State Contract No. 40.012.1.1.
152), the Bundesministerium f u¨ r Bildung und
[
[
[
[
[
[
29] Y.-C. Chao, L.S.O. Johansson, C.J. Karlsson, E. Lande-
mark, R.I.G. Uhrberg, Phys. Rev. B 52 (1995) 2579.
30] W.R.L. Lambrecht, N.E. Christensen, P. Bl o¨ chl, Phys.
Rev. B 36 (1987) 2493.
1
Forschung (Project No. 05-SF8OD1/4 and DLR-
IB RUS00/219), and by the bilateral Project ‘‘Rus-
sian–German Laboratory at BESSY’’.
31] F. Boscherini, J.J. Joyce, M.W. Ruckman, J.H. Weaver,
Phys. Rev. B 35 (1987) 4216.
32] C. Pirri, S. Hong, M.H. Tuilier, P. Wetzel, G. Gewinner,
R. Cortes, Phys. Rev. B 53 (1996) 1368.
33] K.-j. Kim, T.-H. Kang, K.-w. Kim, H.-j. Shin, B. Kim,
Appl. Surf. Sci. 161 (2000) 268.
References
34] R. Leckey, J.D. Riley, R.L. Johnson, L. Ley, B. Ditchek,
J. Vac. Sci. Technol. A 6 (1) (1988) 63.
[
1] S. Saitoh, H. Ishiwara, S. Furukawa, Appl. Phys. Lett. 37
1980) 203.
[35] C. Pirri, J.C. Peruchetti, G. Gewinner, J. Derrien, Phys.
Rev. B 29 (1984) 3391.
(