Full text of DOI:10.1016/0038-1098(77)90358-1

Solid State Communications, Vol. 21, PP. 895—897, 1977.
Pergamon Press.
Printed in Great Britain
ADSORPTION OF H2S, H20 AND02 ON Si(111) SURFACES
K. Fujiwara, H. Ogata and M. Nishijima*
Central Research Laboratory, Mitsubishi Electric Corporation, Amagasaki, Hyogo,Japan
(Received 30 August, 1976 by A.A. Maradudin)
Electron energy-loss spectroscopy has been applied to the study of Si(1 11)
surfaces covered with H~S,H20 and 02 at room temperature and the sur-
faces annealed at 600 C. The experimental results strongly suggest that
H2S and H20 adsorb in the molecular states at room temperature. It is
proposed that 02 is first adsorbed in a molecular state, then adsorbs as
atoms, and finally oxidizes forming SiO2.
ELECTRON energy-loss spectroscopy (ELS) of solid
The energy-loss spectrum of a clean surface, curve
(a), shows all of the characteristic loss peaksin agree-
ment with those reported by other workers.’3 When
this surface is exposed to H2S at room temperature,
new peaks appear at 3.7, 4.8 and 8.8 eV, while the peaks
attributed to the intrinsic surface-state transitions dimin-
ish. The result is shown in curve (b) of Fig. 1. The peak
surfaces is a very sensitive tool for the study of the
intrinsic and extrinsic surface electronic states.13
l’hispermits us to investigate the chemical bonds of the
adsorbates, i.e. whether gas molecules adsorb dissociat-
ively or non-dissociatively on solid surfaces. In this
paper, we present ELS results for thermally cleaned
Si(1 11) surfaces covered with H2 S, H20 and 02 at room amplitudes are saturated with the exposures of— 100
temperature and the surfaces heated to 600°C.The
observed differences in the loss spectra strongly suggest
that these gases are adsorbed in the molecule-like states
on Si(1 11) surfaces at room temperature.
Langmuir (1 L = 10—6 torr sec). The peak positions
are independent of the exposures and the primary elec-
tron energies between 50 to 200 eV within the experi-
mental errors of ± 0.3 eV. These indicate that the three
peaks are due to the extrinsic surface states which
characterize H2S adsorption on Si(1l 1) surfaces. The
peaks near 10 and 18eV are due to surface and bulk
plasmons, respectively, because they become prominent
on increasing the primary electron energy. No evidence
has been obtained that multiple adsorption states are
involved. The room temperature adsorption of H20 has
been reported elsewhere,3 and the loss spectrum is
shown in curve (c). This result is quite similar to curve
(b), which is not surprising due to the similarity of the
structural and the electronic properties of the two
molecules. However, a difference is found in the elec-
tron induced effect, i.e. for H20, an electron beam
irradiation of 30 mm changes the spectra drastically
suggesting the decomposition of the molecules, whereas
much smaller changes are observed for H2 S in the same
period.
There has been a controversy about the models
for the adsorption of H2S, H20 and 02 on Si(lll) sur-
faces at room temperature. For H2S and H20 covered
surfaces, the dissociative adsorption model has been
proposed by Meyer and Vrakking4 using Auger electron
spectroscopy (AES) and ellipsometry. Fujiwara and
Nishijima3 suggested that H20 adsorbs as molecules by
ELS. For oxygen covered surfaces, the low energy-
electron scattering experiments by Ibach et a!.5 indicate
that 02 is adsorbed as molecules at the initial stage of
the oxidation. Ludeke and Koma2proposed that 02 is
adsorbed as atoms using ELS. Recent theoretical cal-
culation by Goddard III eta!.6 indicates that 02 is
adsorbed as peroxy radical, or in a molecular state.
The electron spectrometer consists of an electron
gun, and two 127°cylindrical analyzers used for the
monochrometor and the energy analyzer. The energy-
loss spectra were measured in the second derivative
mode with a constant resolution of 0.8 eV. The same
The loss spectrum of H2S adsorption is observed
to change substantially on heating the surface to above
spectrometerwas also used to analyze the surface species 500°C.Curve (d) shows the spectrum after three cycles
by AES. Clean Si(l 11) surfaces were obtained by heat-
ing the samples to 1200°Cin the vacuum chamber.
of the successive 50 L H2Sexposure and heating at
600°C.The observed peaks are at 3.4, 5.1, 7.4, 8.9,
Special care was taken to minimize the beam induced
effects. The readers are referred to reference 3 for a
further detail.
10.3, 13.5, 18.1 amd 21 eV. The 2.2 eV peak is either
spurious or due to the residual surface-state transition.
The above change is attributed to the decomposition
of H25 and the subsequent desorption of hydrogen at
*
Present address: Dept. of Chemistry, Faculty of
Science, Kyoto University, Japan.
—
500°C,7 and the curve (d) is interpreted to correspond
to the loss spectrum of a Si(ill) surface covered with
895

896
ADSORPTION OF H2S, H20 AND 02 ON Si(1l1) SURFACES
Vol. 21, No.9
I
I
I
I
I
observed peaks are at 3.2, 4.9, 7.1, 9.6, 13.2, 17.9 and
20.6 eV, which are in general agreement with the pre-
72
48
(a)
vious reports.”2 This result is interpreted to indicate
that 02 is adsorbed as molecules in agreement with
Thach and Rowe.1 When the surface was heated to
700°Cfor 30 sec, the peak at 9.6 eV disappears. We
find a peak at 10.1 eV and a small hump at 8.6 eV as
shown in curve (f). It should be emphasized that these
78
03
uJ
(b)
((1-))
~l2Ioo87~7:
p aksdo not exist at room temperature, b cause their
center-of-gravity is not at 9.6 eV. The same result is
>Cr~
(C)
obtained by the electron irradiation for about 10 mm
Cr
is the loss spectrum for 02 adsorbed as atoms, because
the curve is essentially the same as that obtained for a
mH
Cd) ~79
(e)
35
~Ci:r
~~22
(without thermal treatment). We believe that curve (I)
by three cycles of the successive H20 exposure and
c’J
LU
178 ~
Si(111) surface covered withatomic oxygen produced
207
132 10186
49
heating at 700°C.3Curve (0 is not the spectrum for
(f)
32
3
SiO2, which is also shown in curve (g), because the peak
z
at 10.1 eV does not have the same origin as the 10.5 eV
-u
24 21
715035
I ( g ) ~ E7p4 l 4 o I a 6
psereicamtkioaonryfs,Seiin.0ee.2rgt.hyTehdlieasptitesencrdosenhnfociwremsoeafdtdhbeecyreetxhacseeitdaoitffifoetnhreecnrceoelsasitnivtehe
cross sections as the primary electron energy is lowered
I
I
I
I
I
30
25
20
5
CO
5
0
to 30eV, although the former does not. The survival of
ENERGY LOSS, EL (eV)
the bulk plasmon peak at 18 eV indicates also that the
substrate is not oxidized into 5i02 yet. An additional
Fig. 1. Negative second derivatives of the electron energy- evidence is the similarity of curves (d) and (f). In con-
loss spectra (primary energy = 80eV) of the Si(lll)
surfaces: (a) clean surface; (b) after H2Sexposure of
2000 Langmuir; (c) after H2O exposure of 100 Langmuir;
(d) after three cycles of successive 50 Langmuir H2S
exposure and heating at 600°C;(e) after 02 exposure
of 1.2 Langmuir; (f~after 02 exposure of 100 Langmuir
and heating at 700 C; (g) SiO2 surface,
nection with this, we want to point out that the sur-
face covered with atomic oxygen is oxidized into
SiO2 by a prolonged electron bombardment, whereas
the formation of Si52 is comparatively difficult.
Recently, Ludeke and Koma2suggested, from the
similarity of the observed energy-loss spectra with the
excitation spectra of SiO molecules, that oxygen is
sulfur atoms. A similar result is observed for H20
covered surfaces.3
chemisorbed into a monoxide-like bound state. If
this identification is correct, a similar interpretation is
applicable to the case of sulfur atoms chemisorbed on a
silicon surface. However, the location of the loss peaks
for this surface is quite different from the excitation
energies of SiS molecules.8 For example, the main loss
peaks at 7.4eV can not be related to theE’ r--x’ ~
transitions at 4.8—6.2 eV. Therefore, the interpretation
by Ludeke and Koma of oxygen adsorption is question-
able. Our conclusion is supported by the low energy-
electron scattering experiments of Ibach et al.5 and the
theoretical calculation of Goddard III et al.6
As a summary,we have shown that H2S, H2O and
02 are adsorbed as molecules on Si(1 11) surfaces at
room temperature by the comparison of the energy-loss
spectra for gas-covered surfaces and the surfaces heated
to 600°C.The three stages of the oxidation, i.e. initial
adsorption as molecules, then as atoms, and finally as
Si02, have been distinguished.
Meyer and Vrakking4 have proposed that H2S ad-
sorbs dissociatively, and that sulfur and hydrogen atoms
form covalent bondings withthe dangling bonds of a
Si(lll) surface. If this is true, the loss spectrum should
be a superposition of those corresponding to the Si(lll)
surfaces covered with sulfur and hydrogen atoms, pro-
vided that the lateral interactions of the adsorbates
can be neglected. However, no peak is observed at 7.4
eV which is the most prominent of the sulfur covered
surface. The la~terailnteraction is thought to be neg-
ligibly small, since no shift is observed of the loss peaks
as the surface coverage of H2S is varied. Therefore, our
experimental results show that H2S adsorbs as mole-
cules at room temperature. A similar discussion has
previously been reported for H20 adsorption . ~
Curve (e) shows the loss spectrum of a Si(111)
surface exposed to oxygen at room temperature. The

Vol. 21, No. 9
ADSORPTION OF H2S, H20 AND 02 ON Si(1 11) SURFACES
REFERENCES
897
1. IBACH H. & ROWE J.E., Phys. Rev. B9, 1951 (1974); ibid. 10,710 (1974).
2. LUDEKE R. & KOMA A.,Phys. Rev. Lett. 34, 1170 (1975).
3. FUJIWARA K. & NISHIJIMAM., Phys. Lett. SSA, 211(1975).
4. MEYER F. & VRAKKING J.J., Surf. Sci. 38, 275 (1973).
5. IBACH H., HORN K., DORN R. & LUTH H., Surf Sci. 38,433 (1973).
6. GODDARD III W.A., REDONDO A. & McGILL T.C., Solid State Commun. 18,981(1976).
7. BOONSTRA A.H.,Philips Res. Rept. Suppl. 3 (1968).
8. PEARSE R.W.B. & GAYDON A.G., The Identification o fMolecular Spectra. Chapman & Hall, London (1965).