Correlation of surface morphology with chemical structures of sulfurpassivated GaAs(100) investigated by scanning tunneling microscopy
and x-ray photoelectron spectroscopy
Jeong Sook Ha, Seong-Ju Park, Sung-Bock Kim, and El-Hang Lee
Electronics and Telecommunications Research Institute, Taejon 305-600, Korea
~Received 29 September 1994; accepted 6 March 1995!
The surface morphology and chemical structures of GaAs~100! after etching with sulfuric acid and
sulfur passivation with ammonium sulfide @~NH4!2Sx # solution were investigated using a scanning
tunneling microscope and x-ray photoelectron spectroscopy ~XPS!. Depending on the chemical
treatments which are routinely used in semiconductor processes, the surface morphology was
observed to be quite different. The effects of water rinse, HCl treatment, and sulfur treatment time
on the surface morphology were studied. Comparison of the surface morphology with chemical
structures obtained by XPS was used to explain the evolution of surface morphology during the
chemical treatments. In particular, a very flat surface with a surface undulation of 66 Å was
obtained by sulfur passivation with ~NH4!2Sx solution after chemical etching of a n-GaAs~100!
wafer. These studies enable a better understanding of a correlation of the chemical structures with
the surface morphology of the GaAs~100! surface on a nanometer scale. © 1995 American Vacuum
Society.
I. INTRODUCTION
GaAs has high electron and hole mobilities compared to
Si and thus is promising as a high-speed semiconductor device material. However, GaAs has several intrinsic characteristics unsuitable for application to large-scale integrated devices. One of the major problems is the high surface state
density, which induces Fermi-level pinning which makes the
control of Schottky barrier height difficult and also lowers
the current gain in the bipolar transistor with large surface
recombination velocity. In order to prevent the GaAs surface
from having the Fermi level pinned, it should be preserved in
the ultrahigh vacuum chamber or be passivated.
Recently, there have been many efforts to passivate GaAs
surfaces with Na2S, ~NH4!2S, or ~NH4!2Sx solution.1– 6
Among them, surface passivation with ~NH4!2Sx solution has
been proven to be the most effective in reducing surface state
densities and surface recombination velocities.1– 4 Spectroscopic techniques such as x-ray photoelectron spectroscopy
~XPS! and reflection high energy electron diffraction
~RHEED! were employed to clarify the bonding states in
passivation showing that the GaAs surface is S terminated.5,6
Some model calculations were performed to explain the surface passivation effect and showed that the surface state density was reduced by passivation moving the Fermi level.7 It
was also found that P2S5/~NH4!2S treatment could improve
the topographical and chemical uniformity of the GaAs
surface.8 In this case, the stability and uniformity of the
P2S5-passivated GaAs surface were attributed to the formation of an ordered ultrathin oxide rather than to the sulfur
termination.
Furthermore, nanometer-scale characterization of semiconductor surfaces and preparation of topographically
smooth and chemically stable surfaces are very important for
precise control of ultrafine structures which will be prerequisite to the realization of new devices using quantum effects. In order to use the passivated GaAs surface for the
646
J. Vac. Sci. Technol. A 13(3), May/Jun 1995
fabrication of ultrafine structures, complete understanding of
the surface topography before and after the surface treatment
is necessary since it may play a significant role in the formation of interface states during the fabrication of devices. Recently, there have been many reports on the morphology and
nanofabrication of H-passivated Si surfaces.9–13 However,
almost no experimental success in the surface modification
of passivated GaAs has been reported, presumably due to the
poor chemical stability of the sulfur-passivated GaAs surface. Also, no systematic investigation of surface morphology in nanometer scale after the routinely used sulfide solution treatment has been reported yet. In this article, we tried
to understand the chemical reactions such as etching and
passivation by investigating the surface morphology and
chemical structures of GaAs~100! using scanning tunneling
microscopy ~STM! and XPS.
II. EXPERIMENT
The STM used in this experiment was a home-built ultrahigh vacuum ~UHV!-STM and it was calibrated by taking
atomically resolved images from the highly oriented pyrolytic graphite and molybdenum disulfide surfaces.14 STM
tips were prepared by mechanically sharpening the Pt–Ir alloy wire ~Johnson-Matthey, 0.508 mm diameter!. All STM
images were taken in the constant current mode in air and the
scanning areas were considerably large, i.e., 3000 Å33000
Å. Sample bias voltage of V s 524 V and tunneling current
of I t 51 nA were used in taking STM images.
The chemical structures of the samples were investigated
by XPS. In the XPS measurements, the Al K a line was used
as an x-ray source. The photoelectron signal was detected at
a take-off angle of 10° from the surface in order to enhance
the surface sensitivity. The chemical shifts of Ga 2p 3/2,
As 3d, S 2 p, and O 1s core levels were investigated and the
binding energies were calibrated from the C 1s spectrum.
The reference binding energies and peak widths of Ga 2 p 3/2
0734-2101/95/13(3)/646/6/$6.00
©1995 American Vacuum Society
646
647
Ha et al.: Correlation of surface morphology with chemical structures of GaAs(100)
FIG. 1. STM image taken from the GaAs~100! surface after etching in
H2SO4 :H2O2 :H2O ~57:1:1! solution for 30 s and rinsing with flowing deionized water for 10 min. Experimental conditions are: sample bias voltage
V s 524 V, tunneling current I t 51 nA, scanning area53080 Å32960 Å.
and As 3d peaks were obtained from the ion sputter etched
pure GaAs~100! surface.
Samples were Si-doped n-GaAs~100! wafers with an electron concentration of ;1018 cm23. They were first cleaned
by successive treatments with trichloroethane, acetone, alcohol, de-ionized water, and blown dry in N2 gas. It was difficult to take a reproducible STM image from the GaAs~100!
sample after only organic cleaning due to a very unstable
tunneling current resulting from the thick oxide layer on the
surface. After organic cleaning, the sample surface was
etched in a room-temperature H2SO4 :H2O2 :H2O ~57:1:1!
solution and passivated in ~NH4!2Sx solution. Through the
basic sequence of etching and passivation, several different
kinds of surface treatments were made to better understand
the surface reactions. In particular, the effects of a water
rinse after etching and passivation on the surface morphology were investigated. The effect of HCl treatment after
etching on the surface morphology was also examined. For
all those differently prepared samples, the chemical structures were examined by XPS.
III. RESULTS AND DISCUSSION
The STM images taken after different etching durations
between 30 s and 5 min showed that the surface roughness
increased with etching time. Since etching for a shorter duration resulted in a smoother surface, all the GaAs~100!
samples used here were etched for 30 s, which corresponded
to etching of about 6000 Å in depth.15 Figure 1 shows the
STM image taken from the GaAs~100! surface after etching
for 30 s and postetch rinse with flowing de-ionized water
~FDIW! for 10 min. Stable and reproducible images could be
observed in successive scans. The surface is covered with
randomly distributed islandlike features, indicating that etching was not uniform. The maximum surface undulation was
estimated to be 619 Å over the scanned area.
In many semiconductor processes, HCl has been used to
clean the GaAs sample.5,8 The effect of HCl treatment on
JVST A - Vacuum, Surfaces, and Films
647
FIG. 2. STM image taken from the GaAs~100! surface after etching in
H2SO4 :H2O2 :H2O ~57:1:1! solution for 30 s, rinsing with flowing deionized water for 10 min, dipping in concentrated HCl solution for 10 min,
and briefly rinsing with water. Experimental conditions are: sample bias
voltage V s 524 V, tunneling current I t 51 nA, scanning area53080
Å32960 Å.
surface morphology was also investigated in this work. After
30 s etching and the postetch water rinse for 10 min, the
sample was dipped into the concentrated HCl solution for 10
min and rinsed with de-ionized water for a few seconds. As
shown in Fig. 2, the surface became much smoother than that
of the as-etched sample ~Fig. 1! to have maximum surface
undulation of 69 Å.
All the oxides of arsenic such as As2O3 and As2O5 are
known to be highly soluble in water. In contrast, gallium
oxides are insoluble in water even though they are slightly
soluble in acids.15 Therefore, we can expect that HCl treatment for a considerably long duration removed the residual
layers of gallium oxides resulting in a thinner residual layer
and smoother surface. Chemical composition estimated from
the XPS analysis showed that the relative ratio of As to Ga
increased twice with HCl treatment from the as-etched
sample, even though the absolute ratio could not be estimated. It was also observed that the shoulders of Ga 2p 3/2
and As 3d peaks at the higher binding energy side, which are
assigned as the gallium-oxide and arsenic-oxide peaks on the
as-etched sample, decreased with HCl treatment ~Fig. 3!. In
addition, Cl 2 p XPS spectra from the as-etched and HCltreated samples showed that small amount of Cl exist on the
surface after HCl treatment. This will be discussed later in
more detail. These XPS results support the interpretation of
STM results that HCl can remove the residual oxides left on
the sample rinsed with water after etching to result in a
smoother surface than that of the as-etched sample. However, the relative ratio of preferential removal of gallium oxide over arsenic oxide could not be estimated to explain the
recovery of stoichiometry with HCl treatment. Such morphological improvement was also observed in the recent atomic
force microscope ~AFM! experiment by Suemune and
Mukai.16 In their work, it was explained in terms of the selective etching effect of HCl on the steps and kinks, similar
to the case of hydrogen plasma etching of GaAs surfaces.17
Surface passivation with sulfur after etching for 30 s was
648
Ha et al.: Correlation of surface morphology with chemical structures of GaAs(100)
648
FIG. 4. S 2p XPS spectra taken from ~a! as-etched and sulfur-passivated
GaAs~100! samples ~b! with and ~c! without subsequent water rinse for 10
min after ~NH4!2Sx treatment, respectively.
FIG. 3. ~a! As 3d and ~b! Ga 2 p 3/2 XPS spectra of as-etched, HCl-treated,
and sulfur-passivated GaAs~100! samples from the bottom.
performed by dipping the sample in the ~NH4!2Sx solution at
room temperature. The passivation effects are shown in the
As 3d and Ga 2p 3/2 XPS spectra ~Fig. 3!. Clear differences
were observed depending on the surface treatments. On the
as-etched sample, a quite intense peak at the binding energy
higher than the As 3d peak by 3.0 eV is shown, which is
attributed to the oxidized As peak such as As2O3 .18,19 In the
J. Vac. Sci. Technol. A, Vol. 13, No. 3, May/Jun 1995
same way, a high-energy shoulder at the energy shifted from
the metallic Ga 2 p peak by 1.4 eV is obtained from the curve
fitting indicating the existence of gallium oxide on the surface. On the sulfur-treated GaAs~100! surface, however,
these peaks almost disappeared. These oxide peak intensities
also decreased significantly after HCl treatment. Furthermore, curve fitting showed that new peaks appear from the
sample with sulfur treatment. As 3d and Ga 2 p peaks which
shifted to higher binding energy side by 1.3 and 0.6 eV, are
attributed to As–S and Ga–S bonding, respectively.6 Such
observation implies that sulfur atoms are bonded to both surface Ga and As atoms after sulfur treatment with ~NH4!2Sx
solution. Subsequent water rinse after ~NH4!2Sx treatment
did not significantly influence the relative peak intensities of
the sulfur-treated samples. However, it was generally observed that the oxide peak intensity was minimum and
sulfur-related peak intensity was maximum when the sample
was not rinsed with water throughout the etching and sulfur
treatments.
The existence of sulfur on the surface was ascertained by
taking the S 2 p XPS spectra from as-etched and sulfurtreated samples, as illustrated in Fig. 4. The complexity observed in the spectrum was caused by the overlap of the
Ga 3s ~at the binding energy of 159.4 eV! and a bulk plasmon loss peak from arsenic vp ~As! to the S 2 p region consisting of SAsx , SGax , and Sn ~binding energies ranging
from 159 to 164 eV!.20 Due to the instrumental resolution, it
was not possible to resolve the bonding status of sulfur, i.e.,
S–Ga, S–As, or the contribution from polymerized sulfur
Sn . On the sample prepared without subsequent water rinse
Ha et al.: Correlation of surface morphology with chemical structures of GaAs(100)
649
(a)
649
(b)
FIG. 5. STM images and line plots taken from the ~NH4!2Sx -treated GaAs~100! surfaces ~a! with and ~b! without postetch water rinse. Duration in the sulfide
solution is 20 min. After sulfide solution treatment, samples were rinsed with flowing de-ionized water for 10 min. Experimental conditions are: sample bias
voltage V s 524 V, tunneling current I t 51 nA, scanning area53080 Å32960 Å.
after sulfur treatment, the relative contribution from S 2p
region was much higher than that of the sample with subsequent water rinse. In addition, there was a large peak at the
binding energy of 168.8 eV on the sample without water
rinse after ~NH4!2Sx treatment, which can be attributed to the
sulfate SOx .21 XPS data also showed that the arsenic–oxide
layer did not grow with time in the case of sulfur-treated
GaAs~100! in contrast to that of the as-etched one. This preliminary experiment on the aging effect suggests that the
~NH4!2Sx -treated GaAs~100! surface be passivated and be
chemically very stable.
With variation of surface treatment conditions during
etching and sulfur passivation, the surface morphology
showed dramatic differences. It is well known that GaAs
surfaces dissolve slowly in water. Therefore the influence of
a water rinse during etching and passivation processes on the
surface morphology was examined. Postetch water rinse was
chosen as a primary parameter. Figures 5~a! and 5~b! show
the STM images taken from the sulfur-passivated GaAs~100!
with and without postetch water rinse for 10 min, respectively. As can be clearly seen, the sulfur-treated GaAs~100!
surface without postetch water rinse showed a dramatic improvement in smoothness compared to the as-etched sample:
surface undulation was only 66 Å. The image did not show
any apparent surface topography. An atomic force microscope ~AFM! image taken over a larger surface area ~4.3
mm34.3 mm! also showed a very smooth surface, as shown
in Fig. 6. However, the sulfur-treated sample with postetch
water rinse for 10 min showed a surface topography quite
JVST A - Vacuum, Surfaces, and Films
similar to that of the as-etched sample with randomly distributed islandlike features.
Such a conspicuous morphological difference with variation of the surface treatment may be explained in terms of
chemical reactions on the GaAs surface. Especially, the effect of a water rinse seems to be very interesting. Postetch
rinse with de-ionized water is a standard molecular beam
FIG. 6. AFM image and a line plot taken from the ~NH4!2Sx -treated
GaAs~100! surface without postetch water rinse. Duration in the sulfide
solution is 20 min. After sulfide solution treatment, the sample was rinsed
with flowing de-ionized water for 10 min. Scanning area is 4.3 mm
34.3 mm.
650
Ha et al.: Correlation of surface morphology with chemical structures of GaAs(100)
FIG. 7. STM image taken from the ~NH4!2Sx -treated GaAs~100! surfaces for
20 min with postetch water rinse for 10 min. After sulfur passivation,
samples were blown dry without a subsequent water rinse. Experimental
conditions are: sample bias voltage V s 524 V, tunneling current I t 51 nA,
scanning area53080 Å32960 Å.
epitaxy ~MBE! procedure which produces an easily desorbed, gallium-rich amorphous oxide21 on GaAs prior to epitaxial growth. The gallium-rich oxide is a consequence of
the greater solubility of arsenic oxides at neutral pH.22 The
effect of a postetch water rinse on surface morphology was
also reported in the previous STM studies of
P2S5/~NH4!2S-treated GaAs~100! surfaces.8 In that work,
rough and cloudy features were observed in STM images of
the nonideally terminated surfaces after postetch water rinse,
which were attributed to the GaAs–oxide segregation
reaction.7
The effect of subsequent water rinse after sulfur treatment
on the surface morphology was also examined. Figure 7
shows the STM image taken from the GaAs~100! sample,
which was prepared in the sequence of etching, water rinse,
~NH4!2Sx treatment for 20 min, and blown dry without subsequent water rinse. In contrast to the STM image shown in
Fig. 5~a!, the surface was observed to be very rough: surface
undulation was 650 Å. Such a dramatic difference in the
surface morphology shown in Fig. 5~a! and Fig. 7 can be
explained by considering the passivation of the GaAs surface
by ~NH4!2Sx solution.22 The oxide film on the as-etched surface is removed and the top layer of GaAs is also etched to
reveal a fresh GaAs surface. The fresh surface is immediately covered with sulfur, and further precipitation of sulfur
may proceed to form an amorphous layer. Sublimation of the
amorphous sulfur layer is expected to occur in vacuum, resulting in a monolayer of sulfur on the GaAs surface. In the
same manner, water rinse after sulfur passivation is expected
to wash away deposited S in forms such as sulfate SOx and
polysulfur Sn .22 Therefore, there is a possibility that an
amorphous sulfur layer and sulfate exist on the sulfur-treated
GaAs surface without water rinse, giving rise to the rough
surface. This idea was supported by the S 2 p XPS spectra
shown in Fig. 4. A clear difference is noted in the spectra
from the sulfur-passivated surface with and without water
rinse after sulfur treatment. On the sample prepared without
J. Vac. Sci. Technol. A, Vol. 13, No. 3, May/Jun 1995
650
a water rinse after sulfur passivation, there was a large sulfate SOx peak at the binding energy of 168.8 eV whereas no
such peak was observed on the water rinsed sample. In addition, the relative contribution from the S 2p region was
higher without a water rinse even though it could not be
confirmed if only one monolayer of sulfur remains after a
subsequent water rinse.
Finally, the effect of sulfur treatment time on the surface
morphology was examined. It was found that dipping the
sample in ~NH4!2Sx solution for 16 h induced the surface
roughness leading to the surface undulation of 100 Å, which
greatly contrasts to the very smooth surface obtained by sulfur treatment for 20 min as shown in Fig. 5~b!. Such a dramatic difference in the surface morphology can be explained
by considering the etching capability of ~NH4!2Sx solution.
~NH4!2Sx is known to etch GaAs slowly.23,24 A reactive sulfur species, which is known to exist in excess in ~NH4!2Sx
solution, reacts to etch the GaAs surface. Therefore, the
GaAs surface can be oxidized in the ~NH4!2Sx solution and
then the oxide is removed and the surface is oxidized again.
As a result, surface roughness could be created as the dipping duration in ~NH4!2Sx solution increased. Similar experimental results from the ~NH4!2Sx -treated InP surface have
been reported.24
IV. CONCLUSION
The surface morphology and the chemical structures of
GaAs~100! after etching with sulfuric acid solution and passivation with ~NH4!2Sx solution were investigated using
STM and XPS. Depending on the chemical treatments during
these processes, the surface roughness was observed to be
quite different. The effects of postetch water rinse, subsequent water rinse after sulfur treatment, HCl treatment, and
sulfur treatment time on the surface morphology were discussed in terms of the chemical reactions on the GaAs~100!
surface in conjunction with chemical analysis results by
XPS. In particular, a very flat surface with a maximum surface undulation of 66 Å was obtained by sulfur treatment.
These studies enabled a better understanding of the correlation of the chemical reactions with surface morphology in
the nanometer scale.
ACKNOWLEDGMENTS
The authors would like to thank Professor I. C. Jeon of
the Jeon-buk National University for taking AFM images.
This work has been supported by Korea Telecom.
J. Fan, H. Oigawa, and Y. Nannichi, Jpn. J. Appl. Phys. 27, L1331 ~1988!.
M. S. Carpenter, M. R. Melloch, and M. S. Lundstrom, Appl. Phys. Lett.
52, 2157 ~1988!.
3
M. S. Carpenter, M. R. Melloch, and T. E. Dungan, Appl. Phys. Lett. 53,
66 ~1988!.
4
J. Fan, H. Oigawa, and Y. Nannichi, Jpn. J. Appl. Phys. 27, L2125 ~1988!.
5
H. Hirayama, Y. Matsumoto, H. Oigawa, and Y. Nannichi, Appl. Phys.
Lett. 54, 2565 ~1989!.
6
Z. H. Lu, M. J. Graham, X. H. Feng, and B. X. Yang, Appl. Phys. Lett.
62, 2932 ~1993!.
7
C. J. Spindt and W. E. Spicer, Appl. Phys. Lett. 55, 1653 ~1989!.
1
2
651
Ha et al.: Correlation of surface morphology with chemical structures of GaAs(100)
8
J. A. Dagata, W. Tseung, J. Bennett, J. Schneir, and H. H. Harary, Appl.
Phys. Lett. 59, 3288 ~1991!.
9
J. A. Dagata, J. Schneir, H. H. Harary, C. J. Evans, M. T. Postek, and J.
Benett, Appl. Phys. Lett. 56, 2001 ~1990!.
10
L. A. Nagahara, T. Thundat, and S. M. Lindsay, Appl. Phys. Lett. 57, 270
~1990!.
11
S. T. Yan, X. Zheng, and M. H. Nayfeh, Appl. Phys. Lett. 59, 2457
~1991!.
12
N. Barniol, F. Perez-Murano, and X. Aymerich, Appl. Phys. Lett. 61, 462
~1992!.
13
F. Perez-Murano, N. Barniol, and X. Aymerich, J. Vac. Sci. Technol. B 11,
651 ~1993!.
14
J. S. Ha, H. S. Roh, S. J. Park, J. Y. Yi, and E. H. Lee, Surf. Sci. 315, 62
~1994!.
15
M. J. Howes and D. V. Morgan, Gallium Arsenide—Materials, Devices,
and Circuits ~Wiley, New York, 1985!, pp. 119–160.
16
I. Suemune and H. Mukai, in Proceedings of the 1993 International Conference on Solid State Devices and Materials, 1993 ~unpublished! p. 745.
JVST A - Vacuum, Surfaces, and Films
17
651
I. Suemune, A. Kishimoto, K. Hamaoka, Y. Honda, Y. Kan, and M. Yamanishi, Appl. Phys. Lett. 56, 2393 ~1990!.
18
C. D. Wagner, in Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, edited by D. Briggs and M. P. Seah ~Wiley, New
York, 1983!.
19
Z.-H. Lu, C. Lagarde, E. Sacher, J. F. Currie, and A. Yelon, J. Vac. Sci.
Technol. A 7, 646 ~1989!.
20
Z. H. Lu, M. J. Graham, X. H. Feng, and B. X. Yang, Appl. Phys. Lett.
62, 2932 ~1993!.
21
D. Graf, M. Grundner, D. Ludeke, and R. Schulz, J. Vac. Sci. Technol. A
8, 1955 ~1990!.
22
Z. Liliental-Weber, C. W. Wilmsen, K. M. Geib, P. D. Kirchner, J. M.
Baker, and J. M. Woodall, J. Appl. Phys. 67, 1863 ~1990!.
23
Y. Nannichi, J. Fan, H. Oigawa, and A. Koma, Jpn. J. Appl. Phys. 27,
L2367 ~1988!.
24
K. Kurihara, Y. Miyamoto, and K. Furuya, Jpn. J. Appl. Phys. 32, L444
~1993!.