Micro-Raman spectroscopy of disordered and

Applied Surface Science 252 (2006) 7642–7646
www.elsevier.com/locate/apsusc
Micro-Raman spectroscopy of disordered and ordered sulfur
phases on a passivated GaAs surface
T. Błachowicz a,*, G. Salvan b, D.R.T. Zahn b, J. Szuber a
a
Department of Electron Technology, Silesian University of Technology, PL-44100 Gliwice, Poland
b
Institut für Physik, Technische Universität Chemnitz, D-09107 Chemnitz, Germany
Available online 9 May 2006
Abstract
The depletion layer width and band bending of passivated n-type Sn doped GaAs(1 0 0) between subsequent steps of chemical treatment as well
as after a single run treatment were investigated by micro-Raman light scattering by longitudinal optical phonons and coupled phonon–plasmon
modes. Experiments were carried out ex situ at room temperature. We conclude that all observed lineshape changes are due to band bending and to
an amorphous surface phase represented by a broad spectral component. We applied two passivation methods. One was based on (NH4)2Sx solution
and lasted 30 min. The second was based on the S2Cl2 solution and lasted 10 s. These enabled identification of surface regions of different
amorphousness and for faster passivation places of enlarged and completely reduced band bending.
# 2006 Elsevier B.V. All rights reserved.
PACS: 78.30.j; 81.65.b; 63.22.+m; 71.55.Jv
Keywords: Gallium arsenide; Sulfur passivation; Raman spectroscopy; Surface morphology; (NH4)2Sx; S2Cl2
1. Introduction
The problem of reduced electronic quality of the GaAs
surface has been evaluated for many years as important for
practical applications [1,2]. Many types of chemical and optical
methods were used to manage this task [3,4]. Among these
methods we applied those based on the sulfur water or alcoholic
solutions [5–7]. In particular those based on (NH4)2Sx and
S2Cl2 compounds have been used for slow passivation, running
in minutes, and fast passivation, going in seconds, respectively
[8–11].
Many experimental techniques can be applied for testing the
passivation efficiency. Among others there are X-ray photoelectron spectroscopy [12], photoluminescence [13], Kelvin
probe [14], and finally Raman spectroscopy [15]. Raman
spectroscopy performed in the micro-Raman configuration
probes a region on the sample of about 1 mm in diameter.
Raman spectroscopy can provide in a non-destructive way
information about surface and sub-surface regions up to about
* Corresponding author. Tel.: +48 32 2372071; fax: +48 32 2372057.
E-mail address: [email protected] (T. Błachowicz).
0169-4332/$ – see front matter # 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.apsusc.2006.03.058
100 nm below the top surface depending on the penetration
depth of the laser beam. This is why it is specially suited to
compare the influence of surface and sub-surface modifications
in the depletion zone.
The application of Raman spectroscopy to the study of sulfur
passivated surfaces relies on the comparison of integral
intensities stemming from scattering by longitudinal optical
(LO) phonons and coupled phonon–plasmon (L and L+)
modes [15]. The scattering on LO phonons comes from the
depletion layer, while the phonon–plasmon scattering takes
place in the bulk where free carriers exist and where the
phonon–plasmon coupling is possible. After chemical treatment the LO line intensity can be reduced due to reduction of
the depletion layer thickness if the chemical treatment leads to a
decrease in band bending.
In the paper, we provide Raman spectroscopy results on the
depletion layer modifications before and after each step of
chemical wet treatment, that is, degreasing, etching, rinsing,
and finally followed by dry nitrogen blowing. We repeat this
procedure for both (NH4)2Sx and S2Cl2 compounds. The results
were compared also with treatments done in a single run. Due to
different dynamics of the two passivation methods employed,
which results from different chemical compositions and mainly
T. Błachowicz et al. / Applied Surface Science 252 (2006) 7642–7646
Fig. 1. Raman spectroscopy results for the GaAs treated with S2Cl2 solution
after a complete passivation procedure described in the text (a), and after the
same treatment with DI water and acetone and ethanol in rinsing step omitted
(b).
from the time of chemical treatment, the sample surface
presented regions of different surface morphology. In some of
these regions the compound seemed to have an amorphous
nature, judging from the occurrence of a broad spectral
component (background) located near LO and phonon–
plasmon peaks (Fig. 1). Additionally, the broad component
was enhanced for S2Cl2 passivation after rinsing only in water.
This resulted in many different local surface morphology states
and passivation efficiencies. Obtained modifications were
easily tested by the micro-Raman technique. Results of band
bending and depletion layer width for all chemical procedures
were obtained utilizing the new approach which takes into
account a broad amorphous state component.
2. Experimental
The micro-Raman spectroscopy experiments were carried
out in the backscattering geometry (zðxyÞz̄ in Porto notation),
using a Dilor XY spectrometer and a Argon-ion laser working
with l = 488 nm and incident power of 6 mW. We used a
7643
microscope objective, which probed an area of approximately
1 mm in diameter. This enabled analysis of local features
associated with different surface morphologies. The radiation
penetration depth was estimated to be 76.4 nm for the laser line
employed. During typical data accumulation a single scan of a
single full spectral range lasted 20 s and usually this was
repeated 20 times. The band bending and the depletion layer
depth were calculated from data measured between each step of
chemical treatment as well as after all steps done in a single run.
The measurements were done ex situ.
In all experiments we measured n-type Sn doped
GaAs(1 0 0) with a donor concentration of 1018 cm3.
Passivation was carried out using (NH4)2Sx or S2Cl2 solutions.
Steps of treatment consisted of degreasing, etching, and rinsing
followed finally by dry nitrogen blowing. The degreasing in
both passivations was done using acetone (in 5 min), ethanol
(5 min), and completed by 18 MV deionised (DI) water
(5 min). The degreasing process removed a layer of oxides of
1.7 nm thickness, as measured by ellipsometry. Thus, etching
for the (NH4)2Sx solution lasted 30 min at a temperature of
313 K. The next rinsing step was made using DI water for 10 s.
The (NH4)2Sx passivation is a rather well-elaborated method. It
was additionally improved using alcohol solution of the
ammonia sulfide [5–8]. The etching for the S2Cl2 solution
passivation (CCl4 + S2Cl2, 3:1) lasted 10 s, i.e. much faster than
for the (NH4)2Sx passivation. The rinsing process consisted here
of four steps: CCl4 (5 s), acetone (5 s), ethanol (5 s), and DI
water (5 s). All steps in this procedure were carried out at RT.
The second method of passivation is rapid. This can cause a
disturbance of the sample surface morphology and random
orientations of molecular bounding, especially those of Ga–S
and As–S. This resulted in occurrence of the broad spectral
component which additionally contributed to the LO and
phonon–plasmon modes well described in literature [15]
(compare Fig. 1a with Fig. 1b). To enhance this effect of surface
morphology modifications we made test measurements for the
S2Cl2 passivation using only DI water rinsing as a final step
before dry nitrogen blowing. The spectra obtained and the
parameters derived are analyzed in the next section.
3. Discussion of results and conclusions
Information about band bending and depletion layer width
was derived from the ratio of integrated intensities from
scattering by longitudinal optical phonons, which exist in the
depletion region, and from scattering by bulk phonon–plasmon
coupled modes. In the fitting we did not use the second
plasmon–phonon mode (L+) as it was not observed for most
measurement or was registered on a noise level (compare
Fig. 1). The algorithm of band bending calculations was
described in detail elsewhere [16–18]. Thus, the relation
between the depletion layer width d and the band bending VB
can be expressed as follows:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2e0 eS VB
d¼
;
e2 n
(1)
7644
T. Błachowicz et al. / Applied Surface Science 252 (2006) 7642–7646
Table 1
Band bending and depletion layer width for different steps of chemical passivation and for treatment done in a single run
Step of a passivation procedure
(NH4)2S solution
Band bending
(eV)
Depletion layer
width (nm)
Before treatment
After degreasing
After etching
After DI water rinsing
0.84 0.06
0.81 0.05
0.72 0.05
0.73 0.09
34.8 1.2
34.3 1.1
32.2 1.2
32.5 2.0
Treatment using all
steps in a single run
0.63 0.04
30.1 1.0
where e0 is the vacuum dielectric constant, eS the static, relative
dielectric constant, e the elementary electron charge, and n is
the donor concentration in the material. Table 1 provides values
of band bending and depletion layer for both (NH4)2Sx and
S2Cl2 passivations. The data clearly reveal three important
facts. First, the band bending and depletion layer decreases
upon subsequent steps of treatment. Second, the chemical
procedure made in a single run is more effective than those
performed in steps. In this case the difference in the final values
of the band bending between the treatment performed in steps
and that in a single run is equal to 13.7% for the (NH4)2Sx
etching and 7.1% for the S2Cl2 etching. Third, the S2Cl2
passivation is more effective than the (NH4)2Sx passivation.
The difference between final band bending values is equal to
about 0.11 eV.
A relation between band bending and depletion layer width
is given in Fig. 2 along with experimental errors derived from
fitting. In the figure underlined numbers represent subsequent
steps in the (NH4)2Sx passivation (open symbols), while those
Fig. 2. Band bending as a function of depletion layer width after subsequent
steps of chemical wet passivations. The (NH4)2Sx solution passivation steps are
numbered from 1 to 4 and represented by open circles. The S2Cl2 passivation is
represented by primed numbers and filled circles.
Step of a passivation
procedure
S2Cl2 solution
Band bending
(eV)
Depletion layer
width (nm)
Before treatment
After degreasing
After etching and rinsing
0.84 0.06
0.84 0.06
0.56 0.04
34.8 1.2
35.0 1.2
28.4 0.9
Treatment using all
steps in a single run
0.52 0.03
27.5 0.9
primed are related to the S2Cl2 passivation (closed symbols).
For the latter case the etching and final rinsing are treated as
single step, represented by the single point (30 ), due to short
time of both procedures.
All results given in Table 1 and those shown in Fig. 2 were
derived from fitting the experimental data assuming the
existence of a broad component associated with amorphous
phase on a surface. Importantly, this approach can give more
insight into the understanding of the influence of different steps
involved during the chemical passivation procedure (see
Fig. 3). Experimental points from Fig. 2 representing a sample
before degreasing (1, 10 ) and after degreasing (2, 20 ) are
overlapping within experimental errors range. The same points
drawn in Fig. 3 are clearly separated. Fig. 3 informs about two
fundamental issues. Firstly, the degreasing improves surface
morphology. This can be concluded from the integral intensity
values of the broad component, which provides a good
estimation of surface quality. Thus, after degreasing one obtains
the improvement falling approximately in the range of 7–14%.
Secondly, the slower passivation method, namely that based on
the (NH4)2Sx solution, reduces the broad component more
efficiently than those running in seconds during the S2Cl2
passivation. This is visible through relative changes of the
integral intensity of the broad component. For the (NH4)2Sx
passivation the relative change in the integral intensity
calculated after degreasing, etching, and rinsing, related to
the untreated sample, is equal to 30.1%. The same parameter
for the S2Cl2 procedure is equal to 8.5%. These facts are
obviously related to the time of treatment and dynamics of
chemical influence.
Going forward with the analysis we can easily notice one
important difference between both passivation procedures used
here. For the fast method (Fig. 3b) the band bending and the
background obtained after all steps of passivations done in a
single run have the lowest values. Another situation exists for
the slow method. While after each chemical step, in the step by
step procedure, the sample was blown with dry nitrogen before
a Raman measurement, the single run procedure was
completely wet and the sample was blown with dry nitrogen
only once before the measurement. This resulted in better
morphology for the step by step method, but a lower reduction
in the band bending (see point 4 in Fig. 3a) for the (NH4)2Sx
passivation. The single run process resulted in the maximum
band bending reduction. For the fast process, on the other hand,
T. Błachowicz et al. / Applied Surface Science 252 (2006) 7642–7646
7645
Fig. 4. Raman spectra obtained during the S2Cl2 passivation terminated by
rinsing only into DI water in order to obtain enhanced amorphous phase on the
surface. Two examples of spectra are shown: with maximum width of amorphous spectral component (a) and minimum band bending (b).
Fig. 3. Integral intensity of the broad band spectral component centered at
350 cm1 related to the disturbed surface morphology as a function of band
bending for the (NH4)2Sx passivation (a) and the S2Cl2 passivation (b).
the single run process is optimal and proper. The breaks in a
treatment due to the Raman measurements certainly disturb the
surface morphology and passivation efficiency (see point 3 in
Fig. 3b).
To finally prove that we deal with an amorphous phase
represented by a broad component in the Raman spectra we
prepared samples with highly disturbed surface regions. This
was effectively possible for the fast S2Cl2 procedure, however,
with rinsing restricted to 5 s treatment of DI water. On a surface
irregular shapes were obtained and on the Raman spectra we
observed enhanced broad components for some regions.
Among others, in Fig. 4a and b two interesting cases are
shown: that with increased band bending (0.96 0.11 eV), and
that with the LO phonon completely hidden below background,
respectively. The result from Fig. 4a was obtained from a small
region of 1–2 mm size, while Fig. 4b result was obtained from
much larger zone surrounded by irregularities of the 1–2 mm
size. The centre frequency of the broad component lies in the
330–350 cm1 range for all the investigated regions on the
samples. Table 2 provides positions for the LO phonons,
Table 2
Spectral peak positions of the coupled plasmon mode (L), the optical phonon
(LO), and amorphous broad component (BC) for chosen steps in chemical
treatment
Chemical treatment
L (cm1)
LO (cm1)
BC (cm1)
Before treatment
(NH4)2S passivation
S2Cl2 passivation
S2Cl2 solution (Fig. 4a)
S2Cl2 solution (Fig. 4b)
271.7 0.1
273.6 0.1
273.3 0.1
266.9 0.1
266.4 0.1
289.9 0.1
291.0 0.1
290.7 0.1
283.1 0.3
288.3 0.6
345.0 3
319.0 3
361.0 3
332.0 1
343.4 0.4
7646
T. Błachowicz et al. / Applied Surface Science 252 (2006) 7642–7646
coupled plasmon–phonon modes, and those for the broad
components. Data are given for the sample before treatment,
after treatments in a single run for both passivation methods,
and finally for the results seen in Fig. 4. The interesting feature
of these results is that we obtained completely passivated
surface and that this effect is accompanied by a strong
amorphous phase. The occurrence of CCl4 lines results from
water rinsing with acetone and ethanol omitted.
In summary, we conclude that the reduction of band bending
naturally influenced by the sulfur solution chosen is additionally influenced by other steps in chemical wet procedure, that is,
degreasing, rinsing, and drying. Thus, for some experimental
conditions it was possible to obtain enhanced amorphous phase
of the Ga–S and As–S chemical compounds with strongly
reduced band bending. In other words, there exists a
compromise between surface quality and the efficiency of
the band bending reduction as the enhanced amorphousness of
the not perfect surface is qualitatively more similar to bulk than
for the ideal surface. This is why the situation related to not
perfect surface morphology reflected by a broad spectral
component should be taken into account in all typical
passivation procedures. The approach enabled observation of
surface morphology modifications during all subsequent steps
of chemical treatments and gave evident and clear sequence of
data. The data show that dynamics of surface morphology
depends on many chemical factors, but also on the time of
treatment and the thermodynamic conditions applied.
Acknowledgement
This work was performed within V FPEC Project of Centre
of Excellence in Physics and Technology of Semiconductor
Interfaces and Sensors—CESIS, under the Contract: G6MACT-2002-04042.
References
[1] D. Shoji, T. Miura, M. Niwano, N. Miyamoto, Appl. Surf. Sci. 130–132
(1988) 441.
[2] T.K. Oh, C.H. Baek, B.K. Kang, Solid State Electron. 48 (2004) 1549.
[3] V. Augelli, T. Ligonzo, A. Minafra, L. Schiavulli, V. Capozzi, G. Perna, M.
Ambrico, M. Losurdo, J. Luminesc. 102–103 (2003) 519.
[4] V.L. Bekovits, V.P. Ulin, M. Losurdo, P. Capezzuto, G. Bruno, G. Perna, V.
Capozzi, Appl. Phys. Lett. 80 (2002) 3739.
[5] V.N. Bessolov, E.V. Konenkova, M.V. Lebedev, Mater. Sci. Eng. B44
(1997) 376.
[6] V.N. Bessolov, M.V. Lebedev, J. Appl. Phys. 82 (1997) 2640.
[7] V.N. Bessolov, A.F. Ivankov, M.V. Lebedev, J. Vac. Sci. Technol. B13
(1995) 1018.
[8] V.N. Bessolov, M.V. Lebedev, N.M. Binh, M. Friedrich, D.R.T. Zahn,
Semicond. Sci. Technol. 13 (1998) 611.
[9] S.-H. Sa, M.-G. Kang, H.-H. Park, K.-S. Suh, Surf. Coat. Technol. 100–
101 (1998) 234.
[10] Y. Wu, Y. Liu, X.M. Ding, E.G. Obbard, X.Z. Wang, H.J. Ding, X.Y. Hou,
X.B. Li, Appl. Surf. Sci. 228 (2004) 5.
[11] D.N. Gnoth, D. Wolfframm, A. Patchet, S. Hohenecker, D.R.T. Zahn, A.
Leslie, I.T. McGovern, D.A. Evans, Appl. Surf. Sci. 123 (1998) 1248.
[12] J. Szuber, E. Bergignat, G. Hollinger, A. Polakowska, P. Kościelniak,
Vacuum 67 (2002) 53.
[13] S. Anantathanasarn, H. Hasegawa, Appl. Surf. Sci. 190 (2002) 343.
[14] E.T. Yu, Mater. Sci. Eng. R17 (1996) 147.
[15] J. Guerts, Surf. Sci. Rep. 18 (1993) 1.
[16] V.N. Bessolov, M.V. Lebedev, D.R.T. Zahn, Semiconductors 33 (1999)
416.
[17] G. Abstreiter, R. Trommer, M. Cardona, Solid State Commun. 30 (1979)
703.
[18] L.A. Farrow, C.J. Sandroff, M.C. Tamargo, Appl. Phys. Lett. 51 (1987)
1931.

Download Report