A Study of the restoration of Se/Si(1 1 1)-7 × 7 reconstructed surfaces: Preservation of the bulk-terminated state

Article abstract of DOI:10.1016/S0039-6028(02)01096-8

The role of absorbed Se on Si(111)-7×7 surface in the heat-induced 7×7 → 1×1 phase transition of the Si(111)-7×7 substrate was studied. The study took place in an ultra high vacuum utilizing low energy electron diffraction, thermal deposition spectroscopy (TDS) and work function measurements. Results showed that Si(111)-7×7 surface was preserved upon cooling to room temperature as long as the initial coverage of Se on Si(111)-7×7 surface was at least θ_Se=0.5 ML, while Se was strongly bound to the top substrate layer at E_b= 2.9eV/atom.

Full text of DOI:10.1016/S0039-6028(02)01096-8

Surface Science 504 (2002) L191–L195
www.elsevier.com/locate/susc
Surface Science Letters
A Study of the restoration of Se/Si(111)-7 ꢀ 7
reconstructed surfaces: preservation of the
bulk-terminated state
*
A.C. Papageorgopoulos , M. Kamaratos
Department of Physics, University of Ioannina, P.O. Box 1186, GR-45110 Ioannina, Greece
Received 10 December 2001; accepted for publication 4 January 2002
Abstract
In this work we study the role of adsorbed Se on Si(1 1 1)-7 ꢀ 7 surfaces in the heat-induced 7 ꢀ 7 ! 1 ꢀ 1 phase
transition of the Si(1 1 1)-7 ꢀ 7 substrate. The experiment took place in an ultra high vacuum utilizing low energy
electron diffraction, thermal desorption spectroscopy (TDS) and work function measurements. The maximum depos-
ited Se coverage was ꢁ1.6 ML (with the exception of 2.5 ML for some TDS measurements). Heating the Se/Si(1 1 1)-
7
E
1
ꢀ 7 surface at 1025 K causes the desorption of SiSe
2
, dominant at hSe > 0:5 ML, with a calculated binding energy of
b
¼ 2:7 eV/atom. Increasing the temperature to 1050 K brings about the phase transition: Si(1 1 1)-7 ꢀ 7 ! Si(1 1 1)-
ꢀ 1. The resulting restored Si(1 1 1)-1 ꢀ 1 surface is preserved upon cooling to room temperature as long as the initial
coverage of Se on the Si(1 1 1)-7 ꢀ 7 surface is at least hSe ¼ 0:5 ML, while Se is strongly bound to the top substrate
layer at E
b
¼ 2:9 eV/atom. The Se which preserves the Si(1 1 1) surface in its bulk terminated form saturates the free Si
bonds, forming a non-symmetric adlayer. Ó 2002 Published by Elsevier Science B.V.
Keywords: Low energy electron diffraction (LEED); Thermal desorption spectroscopy; Work function measurements; Surface
relaxation and reconstruction; Diffusion and migration; Silicon; Chalcogens
When the Si(1 1 1)-7 ꢀ 7 surface is heated to
(1 1 1)-7 ꢀ 7 surface and its phase transition are
of particular interest, in combination with vari-
ous adsorbates, due to the applications of surface
restoration and passivation in space technology,
nanotechnology, and Si device fabrication [5,6].
One of the few studies of Se on Si(1 1 1) sur-
faces was done by Dev et al. with X-ray standing
wave interference spectroscopy [7]. The authors
of the study claim that Se adsorbs on bridge sites
between adjacent surface silicon atoms. The sub-
strate exhibits a ‘1 ꢀ 1’ structure, while the
deposited Se forms a disordered overlayer. The
adsorbed surface, furthermore, is passivated, and
1
130 K, the surface exhibits a phase transition to a
Si(1 1 1)-‘1 ꢀ 1’ structure [1]. The transition is re-
versible upon slow cooling, unless the surface is
quenched (cooled rapidly), whereas it contains
other local structures as well as retaining a por-
tion of its bulk-terminated geometry [2–4]. The Si
*
Corresponding author. Tel.: +30-651-98570; fax: +30-651-
4
5381.
E-mail address: me00426@cc.uoi.gr (A.C. Papageorgopou-
los).
0
039-6028/02/$ - see front matter Ó 2002 Published by Elsevier Science B.V.
PII: S0039-6028(02)01096-8

L192
A.C. Papageorgopoulos, M. Kamaratos / Surface Science 504 (2002) L191–L195
remains stable for 12 h under atmospheric condi-
tions. A recent theoretical work by Natori et al.,
furthermore, maps out the surface potential of the
Si(1 1 1)-‘1 ꢀ 1’ clean surface with calculated val-
ues for possible adsorption sites [8]. In this study,
atomic scale structural and electronic processes are
described using a combination of low energy
electron diffraction (LEED) with an approximate
ing at 1300 K, and the temperature of the sample
was measured by a Cr-Al thermocouple calibrated
by an infrared pyrometer in a temperature range of
900–1300 K.
The WF changes were taken after deposition
of ꢁ1.6 ML (20 doses) of elemental Se on the
Si(1 1 1)-7 ꢀ 7 surface, which was subsequently
heated to 1300 K (with measurements taken in 50
K intervals between heatings). The results are de-
picted in the top portion of Fig. 1 in correlation
with LEED observations. The room temperature
changes of the WF of the Si(1 1 1)-7 ꢀ 7 surface vs.
Se coverage do not concern us here. Suffice it to
say that 1.6 ML of adsorbed Se results in a ꢁ0.4
eV increase of the substrate WF, and that LEED
4
0 eV electron beam energy, work function (WF)
measurements where we used the diode method
and thermal desorption spectroscopy (TDS) with a
heating rate of b ¼ 17 K/s. The influence of ad-
sorbed Se on the Si(1 1 1)-7 ꢀ 7 surface, as it is
heated to transition temperatures, is described
using the above techniques, and for the first time
Se is verified as a RT structural preservative of the
bulk-terminated state of the restored Si(1 1 1)-
1
ꢀ 1 surface.
The experiments were performed in an ultra
high vacuum (UHV) chamber, equipped the afore-
mentioned techniques. Elemental Se was evapo-
2
rated by thermal dissociation of WSe single crystal
flakes enclosed in envelopes made by thin Ta plates
which were heated by passing current. During
2
dissociation of WSe the W remained in the Ta
envelope, while Se was evaporated through holes
made at the edge of the envelope. By plotting QMS
signal intensity vs time for both atomic and mo-
lecular Se, with a constant current running through
the Ta envelope and compare the areas under these
graphs we measured the atomic Se to be 30% of the
total yield. The Si(1 0 0) single crystal substrate was
used as a reference surface to calibrate surface
coverage. Selenium adsorbed on the Si(1 0 0)-2 ꢀ 1
surface causes the latter to undergo RT restoration
to its bulk-terminated state when hSe ffi 1:0 ML [9].
The corresponding surface atomic density is then
considered to be equal to that of the outmost layer
14
2
of Si: 6:9 ꢀ 10 atoms/cm . For this study 1 dose
14
2
was calculated at 0:63 ꢀ 10 atoms/cm . Assuming
the sticking coefficient S ffi 1 for Se on Si(1 1 1)-
7
ꢀ 7, we require 12 doses to attain a surface
14 2
density of ꢁ7:6 ꢀ 10 atoms/cm , approximately
Fig. 1. Top: The WF changes on the Se(1.6 ML)/Si(1 1 1) sur-
face after heating to 1300 K (in 50 K intervals), in correlation
with LEED observations. Bottom: Thermal desorption spectra
equivalent to 1.0 ML. One dose, therefore, equals
0
was retained at a constant amount throughout the
.08 ML of Se on Si(1 1 1)-7 ꢀ 7 surfaces. Dosage
(
TDS) of Se (79 amu), after deposition of Se on the Si(1 1 1)-
7 ꢀ 7 surface. The heating rate was 17 K/s, and was constant
throughout the whole temperature range of 300–1300 K. Peaks
experiment while both Si(1 1 1)-7 ꢀ 7 and Si(1 0 0)-
2
ꢀ 1 single crystal surfaces were cleaned by heat-
b , b and region b are depicted accordingly.
1
2
3

A.C. Papageorgopoulos, M. Kamaratos / Surface Science 504 (2002) L191–L195
L193
indicates a high degree of surface disorder, re-
sulting from either an induced structural disorder
and/or a disordered Se adlayer.
doses of Se thereupon reaching a saturation limit.
The second peak b desorbs at 1090 K and begins
forming after the 6th dose. It maximizes at about
the 20th dose, whereupon it also saturates. The
2
The gradual heating that follows at the 20th
dose (1.6 ML), reveals mild WF variations to 700
K, followed by a decrease in the surface WF to 900
K. Up to 1000 K (confirmed by TDS measure-
ments shown at the bottom portion of Fig. 1),
there is no desorption from the surface. The pat-
tern of the WF curve to the latter temperature is,
in all probability, a reflection of an increasing
surface atomic mobility, most likely similar to that
which has been observed to cause a structural
rearrangement of the clean Si(1 1 1)-7 ꢀ 7 sur-
face between 700 and 900 K [4]. Specifically, a
pronounced dimer and adatom surface mobility,
would certainly cause the behavior observed in the
WF, only in our case there would be a strong Se
presence bound to the mobile silicon complexes,
contributing to the surface diffusion mechanism.
At about 1000 K the 1=7th order spots of the
room temperature LEED pattern begin to disap-
pear, coinciding with is a dramatic WF increase,
which peaks at 1050 K with the establishment of
the 1 ꢀ 1 structure on the screen of the LEED
apparatus. The WF peak indicates that the surface
dipole moment is at its strongest, most likely oc-
curring when adsorbate adatoms occupy the top-
most positions on the substrate surface. In other
words, the above data confirm that the remain-
ing Se adatoms are most likely occupying the top-
most positions on the Si(1 1 1)-1 ꢀ 1 surface, which
maximize their dipole moment. We propose that
this adatom structural configuration is responsible
for the observed suppression of the 7 ꢀ 7 recon-
struction. The Se adatoms, in particular, according
region b desorbs between 500 and 900 K, and first
3
appears when Se quantities increase to about 20
doses. Between 20 and 30 doses the latter region
undergoes a dramatic increase in contrast to the
stabilized b and b high energy peaks. We can
1
2
safely say that b represents loosly bound Se on the
3
Si(1 1 1)-7 ꢀ 7 surface which is probably responsi-
ble for the increased backround, and eventual
disappearance, of surface periodicity as observed
with LEED. The first two peaks represent much
higher binding states of Se on the Si(1 1 1) surface
than does b . We notice, first of all, that Se de-
3
sorbed above 1050 K, is actually removed from the
restored 1 ꢀ 1 surface, which is where we need to
look for its possible binding states. This is con-
firmed for b , which desorbs at the same temper-
1
ature where the LEED pattern resumes its 7 ꢀ 7
form. The latter peak, therefore, represents the
portion of deposited Se which preserves the Si-
(1 1 1)-1 ꢀ 1 surface structure at room tempera-
ture. Peak b , on the other hand, represents the Se
2
desorbing at a temperature range which coincides
with the 7 ꢀ 7 ) ‘1 ꢀ 1’ phase transition itself.
For a more comprehensive picture the above
data requires comparison with that of Fig. 2.
2
The latter depicts the TDS spectra of SiSe (186
amu). The resulting peak (b ), detected at about
4
1025 K, has a very low intensity up to 5 doses (6
doses ꢁ 0:5 ML). It thereafter increases dramati-
cally, especially after hSe approaches 1 ML (10
doses). The SiSe of peak b4 is removed from
2
the surface at the same temperature after which the
clear 1 ꢀ 1 LEED pattern begins to appear. The
beginning of the 7 ꢀ 7 ) 1 ꢀ 1 phase transition,
therefore, most likely coincides with the desorp-
ꢀ
to Dev et al. maintain a 2:07 Æ 0:09 A from the
surface plane, which would account for the in-
crease in the dipole moment [7].
The bottom portion of Fig. 1 shows the thermal
desorption spectra (TDS) of Se (79 amu) after
deposition of Se on the Si(1 1 1)-7 ꢀ 7 surface up to
tion of SiSe from the Si(1 1 1)-7 ꢀ 7 surface. By
2
the time the materials of peak b have been re-
4
moved the LEED pattern shows a clear 1 ꢀ 1
pattern with about zero background intensity at a
40 eV (very low penetration depth) electron beam
energy. From Fig. 1(bottom), however, we notice
3
0 doses (hSe ffi 2:5 ML). The heating rate for all
spectra was 17 K/s, and was constant throughout
the whole temperature range of 300–1300 K. In the
indicated portion of Fig. 1 we identify peaks b , b
that the Se of peak b has only partially desorbed
1
2
2
and a broader region labeled b . The peak desig-
from the substrate, indicating the presence of Se,
in all probability residing below the top surface
3
nated b desorbs at 1180 K, and maximizes at 6
1

L194
A.C. Papageorgopoulos, M. Kamaratos / Surface Science 504 (2002) L191–L195
E
b
¼ 1:3 eV/atom at 550 K, and may indicate
loosly bound Se, adsorbed on the 7 ꢀ 7 surface at
higher coverage levels (hSe > 1:5 ML). Peak b ,
4
representing SiSe
of E
energy for peak b is very close to the latter
2
, has a calculated binding energy
b
¼ 2:7 eV/atom, while the calculated binding
2
at E ¼ 2:8 eV/atom. The peak desorbing at the
b
highest temperature of the spectra of Fig. 1(bot-
tom), b has a binding energy also close to b ,
1
2
where E
b
¼ 2:9 eV/atom. Upon thermal desorp-
tion the highest energy peak b is the only peak
1
visible at hSe 6 0:5 ML up to 6 doses, where this
peak saturates. The disappearance of this high
energy peak by heating the system above 1190 K,
coincides with the reappearance of the 7 ꢀ 7 pat-
tern with LEED. This shows that the Se removed
from the Si surface at this temperature is that
which is directly responsible for the suppression of
the 1 ꢀ 1 ) 7 ꢀ 7. The WF reflects this as a rapid
decrease, (Fig. 1(top)), as the suppressing element
is removed. It is interesting to note here that there
are no indications that Se remaining below the top
surface layer influences the binding of the top layer
Se. It is, therefore, highly likely that at an initial
coverage of only 0.5 ML the preservation of the
bulk-terminated state would still occur after
heating to 1050 K without the presence of Se
below the top surface layer. This may make Se a
useful surfactant in film growth applications. At
hSe > 0:5 ML, however, it is highly likely that the
strong nature of the Si–Se bond causes an etching
of the top silicon layers during annealing.
2
Fig. 2. The TD spectra for SiSe (186 amu), designated as b4,
for various Se dosages, at a constant heating rate of 17 K/s.
layer of the Si substrate. From the above data we
2
may suggest that as SiSe is removed from the
substrate the Si:Se stoicheometry changes from 1:2
to 1:1, and that this change results in a higher Si–
Se binding state. When this is attained the phase
transition occurs and the top substrate surface
layer is preserved in its bulk-terminated state with
a quantity Se still residing below the surface.
For the calculation of the TD peak binding
The proposed structural model for the Se/
Si(1 1 1)-1 ꢀ 1 surface is given in Fig. 3, where the
top and side views of the aforementioned surface
are depicted. The figure depicts Se bound to a
bridge site between two top Si surface atoms. This
site is favored for three reasons: (a) the interatomic
energies we used Seebawer’s formula [10]:
ꢀ
distance between the Si surface atoms is 4.4 A,
!
small enough to accommodate a single Se atom;
(b) the 1 ꢀ 1 ‘top’ atoms contain broken bonds
easily available to the divalent Se atom and (c) the
site, referred to as site ‘M’ (center of the 1 ꢀ 1 unit
mesh), by Natori et al. has an adsorption energy of
Ed
1
2
ln Y ln Y
þ Y À ln Y þ Y
¼
p
À
2 À ln Y
;
2
k T
b
2Y
where E
equal to the binding energy E
is the preexponential factor, equal to 10
d
is the desorption energy, considered
b
, Y ¼ lnðv
0
T
p
=bÞ, v
0
À13 À1
s , T
p
E
ad ¼ 2:9 eV, equal to the 2.9 eV/atom value of the
the thermal desorption peak temperature, and b
the constant heating rate, equal to 17 K/s. The
binding energy of TD peak b3 is, accordingly,
b peak, as calculated using Seebawer’s equation
1
[8]. According to Ref. [8], this value is very close to
that of the T and H adsorption sites. The T and
4 3 4

A.C. Papageorgopoulos, M. Kamaratos / Surface Science 504 (2002) L191–L195
L195
adatoms and the Si(1 1 1) substrate, reported by
Dev et al. [7]. When the M sites are filled on the
1
ꢀ 1, even in a disordered manner, two Si free
bonds correspond to each Se adatom. The surface
coverage, therefore, amounts to 0.50 ML.
It is necessary at this point in our discussion
to add that the 1.0 ML surface density of 7:8 ꢀ
1
4
2
10
atoms/cm for the Si(1 1 1)-1 ꢀ 1 unit cell is
equivalent to 49 atoms/unit mesh. According to
our proposed model, we have a total of 25 Se
adatoms/unit mesh on the restored Si(1 1 1)-1 ꢀ 1
surface, which agrees with the calibrated coverage
corresponding to peak b . We may conclude,
1
therefore, that to suppress the surface reconstruc-
tion the requirements are a 0.50 ML saturation of
the top Si(1 1 1)-7 ꢀ 7 double layer at RT by Se
adatoms, and a subsequent heating to 1050 K. In
addition to this, it must be noted that an ordered
Se adlayer would alter the observed clear 1 ꢀ 1
LEED pattern into 2 ꢀ 1 configuration. This was
not observed, however, supporting the conclusion
that the preserving Se overlayer is disordered, as
indicated by the model in Fig. 3.
References
Fig. 3. The proposed model for the Se/Si(1 1 1)-‘1 ꢀ 1’ structure
top and side views) The Se is bound in a disordered array to a
bridge site between two Si surface top atoms, maintaining an
[
[
1] J.J. Lander, Surf. Sci. 1 (1964) 125.
2] W. Telieps, E. Bauer, Surf. Sci. 162 (1985) 163.
(
ꢀ
interatomic distance of 4.4 A.
[3] M. Hoshino, Y. Shigeta, K. Ogaua, Y. Homma, Surf. Sci.
65 (1996) 29.
3
[
[
[
4] T. Ishimaru, K. Shimada, T. Hoshino, I. Ohdomari, Surf.
Sci. 433 (1999) 401.
3
H sites, however, do not meet the first two criteria
5] R.K. Jain, G.A. Landis, 22 IEEE Photovoltaic specialists
conference, Las Vegas NV, October 7–11, 1991.
6] H. Yanagi, D. Schlettwein, H. Nakayama, T. Nishino,
Phys. Rev. B 61 (2000) 1959.
favoring site M.
The bonding, furthermore, of Se to two adja-
cent dangling bonds over site M would also ac-
count for the WF peak (Fig. 1(top)) observed after
heating to 1050 K, and indicating that the Se
surface adatoms on the 1 ꢀ 1 surface reside at el-
evated positions with respect to the substrate. This
latter observation is also in agreement with the
[7] B.N. Dev, J. Thundat, W.M. Gibson, J. Vac. Sci. Technol.
A 3 (1985) 946.
[
8] A. Natori, T. Suzuki, H. Yasunaga, Surf. Sci. 367 (1996)
6.
9] A.C. Papageorgopoulos, M. Kamaratos, Surf. Sci. 466
2000) 173.
[10] E.G. Seebawer, Surf. Sci. 316 (1994) 331.
5
[
(
ꢀ
measured distance of 2:07 Æ 0:09 A between Se