APPLIED PHYSICS LETTERS 87, 252104 共2005兲
Surface passivation of III-V compound semiconductors using
atomic-layer-deposition-grown Al2O3
M. L. Huang, Y. C. Chang, C. H. Chang, Y. J. Lee, and P. Chang
Department of Materials Science and Engineering, National Tsing Hua University, Hsinchu, Taiwan,
Republic of China
J. Kwo
Department of Physics, National Tsing Hua University, Hsinchu, Taiwan, Republic of China
T. B. Wu and M. Honga兲
Department of Materials Science and Engineering, National Tsing Hua University, Hsinchu, Taiwan,
Republic of China
共Received 16 August 2005; accepted 18 October 2005; published online 13 December 2005兲
Al2O3 was deposited on In0.15Ga0.85As/ GaAs using atomic-layer deposition 共ALD兲. Without any
surface preparation or postthermal treatment, excellent electrical properties of Al2O3 / InGaAs/ GaAs
heterostructures were obtained, in terms of low electrical leakage current density 共10−8 to
10−9 A / cm2兲 and low interfacial density of states 共Dit兲 in the range of 1012 cm−2 eV−1. The
interfacial reaction and structural properties studied by high-resolution x-ray photoelectron
spectroscopy 共HRXPS兲 and high-resolution transmission electron microscopy 共HRTEM兲. The depth
profile of HRXPS, using synchrotron radiation beam and low-energy Ar+ sputtering, exhibited no
residual arsenic oxides at interface. The removal of the arsenic oxides from Al2O3 / InGaAs
heterostructures during the ALD process ensures the Fermi-level unpinning, which was observed in
the capacitance-voltage measurements. The HRTEM shows sharp transition from amorphous oxide
to single crystalline semiconductor. © 2005 American Institute of Physics.
关DOI: 10.1063/1.2146060兴
III–V
compound
semiconductor
metal-oxidesemiconductor field-effect-transistors 共MOSFETs兲 would
give several advantages over their Si-based counterparts, due
to the III-V’s intrinsic characteristics in high electron mobility, semi-insulating substrates, and high breakdown fields,
which lead to high-speed devices, prevent cross talks during
the high-frequency operation, and offer high-power 共and/or
high-temperature兲 applications, respectively. Efforts in
searching insulators on GaAs with a low interfacial density
of states 共Dit兲, one of the key challenges in the III-V devices
over the past four decades,1,2 have found a solution in ultrahigh vacuum 共UHV兲 deposition of amorphous
Ga2O3共Gd2O3兲 and single-crystal Gd2O3 on GaAs
surfaces.3,4 It was speculated5 that the impingement of
Gd2O3 molecules 共and/or clusters containing Gd2O3兲 on the
GaAs surface has forced As to be detached from the GaAs
surface 共with oxygen occupying the As site兲 and diffused to
the oxide surface during the process of film deposition.
The employment of Ga2O3共Gd2O3兲 as a gate dielectric
along with an ion-implantation process led to the demonstration of the first enhancement mode GaAs MOSFETs with
inversion on semi-insulating GaAs substrates in both n- and
Depletion-mode
GaAs
p-channel
configurations.6,7
MOSFETs of 0.8 ␮m gate length have been shown a drain
current of 450 mA/ mm and a transconductance of
130 mS/ mm, and moreover, have exhibited negligible drain
current drift and hysteresis, the first achievement in this class
of transistors and an important technological advance for
manufacturing consideration.8 A power GaAs MOSFET exa兲
Electronic mail: [email protected]
hibited excellent performance and is a promising candidate
for microwave applications.9
Recently, a 1 ␮m gate-length depletion-mode n-channel
GaAs MOSFET with an 8 nm thick Al2O3 gate oxide has
shown a maximum transconductance of 120 mS/ mm and a
drain current of 400 mA/ mm.10 The Al2O3 gate oxide, ex
situ deposited on GaAs, was grown using atomic-layer deposition 共ALD兲, a technique commonly used for depositing
high-␬ gate dielectrics for Si. A study has been carried out by
Frank et al.11 to characterize the structure and composition of
the ALD-Al2O3 / GaAs and HfO2 / GaAs heterostructures. It
was found that a native oxide remains intact underneath
HfO2 during growth, while thinning of the native oxide occurs during Al2O3 deposition. Hydrofluoric acid etching
prior to growth minimizes the final interlayer thickness.
Thermal treatments at 600 °C are found to be necessary to
decompose arsenic oxides and remove interfacial oxygen.
The work presented here has been taken to investigate
the mechanism of III-V surface passivation by studying compositional, electrical, and structural characteristics of the
ALD-Al2O3 / III-V 共InGaAs兲 semiconductor heterostructures,
particularly those at the interfaces. The high-resolution x-ray
photoelectron spectroscopy 共XPS兲 using synchrotron radiation has detected no residual arsenic oxides in the oxide or at
the oxide/ InGaAs interface after the deposition of ALD
Al2O3. The native arsenic oxide, As2O3, on top of the
molecular-beam epitaxy 共MBE兲 grown InGaAs after being
exposed to air was revealed using XPS. After the ALD process, a small amount of As2O5 共note that not As2O3兲 was
found to be on the Al2O3. The removal of arsenic oxides
from the oxide/ InGaAs heterostructures ensures the effective
passivation 共i.e., Fermi-level unpinning兲, with a Dit of
⬃1012 eV−1 cm−2 deduced from the capacitance-voltage
0003-6951/2005/87共25兲/252104/3/$22.50
87, 252104-1
© 2005 American Institute of Physics
Downloaded 17 Feb 2009 to 140.114.12.89. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
252104-2
Huang et al.
Appl. Phys. Lett. 87, 252104 共2005兲
FIG. 1. HR-TEM picture of Al2O3 on In0.15Ga0.85As after nitrogen annealing
at 500 °C.
共C-V兲 measurements. Current density-field 共J-E兲 curves
show a leakage current density of about 10−8 to 10−9 A / cm2
at biasing fields of ⬍3 MV/ cm for both as-deposited and
postannealed samples, where J is the current density 共current
divided by the measured area兲 and E is the electrical filed
共voltage divided by the oxide film thickness兲. Highresolution cross-section transmission electron microscopy
共HRTEM兲 reveals a smooth interface between the oxide and
InGaAs in both the as-deposited and annealed conditions.
In this letter, ALD-Al2O3 was grown on native oxidecovered MBE-grown InGaAs/ GaAs. Deposition of Al2O3
was performed using an ALD reactor manufactured in
Taiwan.
In0.15Ga0.85As/ GaAs strained epilayers with a Si doping
of 5 ⫻ 1017 cm−3 were grown on Si-doped GaAs 2 in. substrates in a solid-source GaAs-based MBE chamber. After
the semiconductor epilayer growth, the wafers were ex situ to
the ALD system for oxide growth. Al2O3 films with thickness ranging from 4 to 14 nm were grown with a wafer
temperature of 300 °C and a chamber pressure of 1 Torr
using alternating pulses of Al共CH3兲3 共TMA兲 and H2O as
precursors. High-purity Ar 共99.999%兲 was used as diluted
and purged gas. To take the process within a self-limited
region, the pulse period was maintained at 3-3-3-3 s percycle 共i.e., 3 s each for TMA, Ar, H2O, and Ar兲 with 0.1 Torr
for a partial pressure of the precursors. The estimated deposition rate of roughly 0.084 nm per cycle was measured using a spectroscopic ellipsometer, x-ray reflectivity, and transmission electron microscopy 共TEM兲. The TEM sample
analytical work was performed using a Philips TECNAI-20
FEG type TEM. The metal-oxide-semiconductor 共MOS兲 diode structure was fabricated by evaporating Au dots 0.1 mm
in diameter. J-E and C-V characteristics were measured using Agilent 4156C and 4284, respectively.
High-resolution x-ray photoelectron spectroscopy 共HRXPS兲 using synchrotron radiation was performed to determine the chemistry at surface and bulk of the Al2O3 films
and at the oxide/semiconductor interface. The XPS data were
taken at the U5 undulator beam-line of National Synchrotron
Radiation Research Center in Hsinchu, Taiwan. The pass energy of electron analyzer was fixed at 5.85 eV, and overall
energy resolution was better than 0.15 eV.
Figure 1 shows a cross-section HRTEM of the
Al2O3 / In0.15Ga0.85As/ GaAs structure which was after
500 °C nitrogen annealing. The oxide thickness measured by
the TEM is 8.5 nm and a sharp transition from crystalline
InGaAs to amorphous oxides was observed. The oxide thickness measured by TEM is similar to that by x-ray reflectivity
FIG. 2. As 3d core-level spectra recorded from two samples,
Al2O3 / In0.15Ga0.85As/ GaAs 共top兲 and native oxide/ In0.15Ga0.85As/ GaAs
共bottom兲: 共a兲 At the surface of Al2O3; 共b兲 immediately below the Al2O3
surface; 共c兲 in the bulk of Al2O3; 共d兲 at the interface of
Al2O3 / In0.15Ga0.85As; 共e兲 at the surface of air-exposed In0.15Ga0.85As; and 共f兲
in the bulk of In0.15Ga0.85As.
共not shown兲. As-grown and oxygen-annealed samples have
shown a similar morphology in terms of the amorphous nature of the oxide and the surface and interfacial roughness.
The interface shows good thermal stability after nitrogen annealing at 500 °C. The oxide appeared to be pinholes free.
This is consistent with the J-E measurement that shows a
low leakage current density. Note that the interfacial layers
contain In2O3 and Ga2O3, as was found using HR-XPS, and
will be discussed below.
For
the
XPS
studies,
two
samples,
Al2O3 / In0.15Ga0.85As/ GaAs
and
native
oxides/ In0.15Ga0.85As/ GaAs, were studied with the latter
served as a reference 共Fig. 2兲. The As 3d spectra for the
as-grown Al2O3 / In0.15Ga0.85As/ GaAs 共top of Fig. 2兲 showed
a small, but very profound peak of As2O5. The peak quickly
disappeared with a slight Ar+ sputtering, indicating a very
small amount of arsenic oxides on top of the as-grown
sample. There was no detection of any arsenic oxides during
the continuous sputtering. After the removal of Al2O3 with
sputtering, the peak belonging to InGaAs was revealed. In
comparison, the native oxide on the reference sample was
found to be As2O3 共bottom of Fig. 2兲 different than the arsenic oxide on the ALD grown Al2O3 on InGaAs. The XPS
studies on the above two samples clearly showed that: 共i兲
There is a native arsenic oxide on the MBE-grown InGaAs
after being exposed to air, which is As2O3; 共ii兲 the arsenic
oxide on top of Al2O3 after ALD deposition is As2O5; 共iii兲
there is no detectable residue of arsenic oxides within the
Al2O3 and at the Al2O3 / InGaAs interface; and 共iv兲 existence
of In2O3 and Ga2O3, residues of the native oxide, was detected at the Al2O3 / InGaAs interface.
It is possible that As2O3, the native oxide on InGaAs
surface, first interacts with Al共CH3兲3 to be transformed to
become arsenic and Al2O3. 关Note that As2O3 was reduced by
Al共CH3兲3. However, it is not known that the arsenic form is
As, As2, or As4.兴 The arsenic then reacts with H2O to be-
Downloaded 17 Feb 2009 to 140.114.12.89. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
252104-3
Appl. Phys. Lett. 87, 252104 共2005兲
Huang et al.
FIG. 3. 共a兲 Leakage current density J共A / cm2兲 vs E共MV/ cm兲 for
Al2O3 / In0.15Ga0.85As heterostructure in different thermal processes. 共b兲
C-V curves of a MOS diode made of Au/ Al2O3共8.5 nm兲 / In0.15Ga0.85As after
two-frequency corrections in different thermal processes.
come As2O5, which then reacts again with the next incoming
pulse of Al共CH3兲3 to become As and Al2O3. As2O5 has a
lower Gibbs free energy than As2O3. This reaction has repeated itself during the ALD process. Most of the arsenic
oxides evaporated since the melting point of As2O5 is low
around 280 °C with a small amount of residual As2O5 left on
the skin-depth top of Al2O3. The removal of arsenic oxides
from the Al2O3 / InGaAs heterostructures during the ALD
process ensures the Fermi-level unpinning, which was observed in the C-V measurements as discussed below. In2O3
and Ga2O3 in the original native oxides were left at the
Al2O3 / InGaAs interface due to their higher thermal stability.
Figure 3共a兲 of the J-E curves shows a leakage current
density of 8.5 nm Al2O3 on In0.15Ga0.85As/ GaAs. The Al2O3
dielectric films are highly electrically insulating, both asdeposited and postannealed samples showing very low leakage current densities of ⬃10−8 to 10−9 A / cm2 at a bias of 1
V. This indicates that our ALD Al2O3 is stoichiometric and
dense in the as-deposited condition and the postannealing
under oxygen and nitrogen has not improved the oxide leakage. The result differs from the previous report, in which
annealing the sample under oxygen ambient at ⬃600° C was
necessary to improve the ALD-Al2O3 in stoichiometry and
density.
As illustrated in Fig. 3共b兲, C-V curves were obtained
with frequencies varying from 1 kHz to 1 MHz. Both preand postannealing C-V measurements showed the accumulation and inversion. A transition from accumulation to depletion mode occurs at ⬃2 V. The C-V measurement results
proved the good quality of Al2O3 / InGaAs interface without
thermal treatment. The dispersion in the C-V curves of different frequencies was reduced markedly using equivalent
circuit of complex impedance.12 The Dit was calculated to be
around 1012 cm−2 eV−1 at the midgap using the Terman
method, a value comparable to what was reported earlier.10
In comparison, the Dit from Ga2O3共Gd2O3兲 deposited on
GaAs in an ultrahigh vacuum 共UHV兲 is one order of magnitude lower in the range of ⬃1011 cm−2 eV−1.12 The higher Dit
in the ALD-Al2O3 on GaAs or InGaAs is probably caused by
the existence of the native oxides of In2O3 and Ga2O3 at the
interface, while there are no such native oxides at the UHV
prepared Ga2O3共Gd2O3兲 on GaAs or InGaAs.3,4,12
In conclusion, we have achieved good electrical properties in ALD-Al2O3 / InGaAs heterostructures without any surface preparation such as HF dip prior to the Al2O3 deposition
or postthermal treatment such as annealing in O2 at
⬃600 ° C. The high-resolution XPS using synchrotron radiation has shown that no residual arsenic oxides exist at the
interface of the oxide/ InGaAs heterostructures, without postdeposition annealing in O2. This indicates the removal of the
arsenic oxides during the ALD-Al2O3 deposition, thus ensuring the Fermi-level unpinning as measured from the C-V
characteristics. Moreover, the HR-XPS allows us to distinguish As2O3 from As2O5. Our findings here are different
from what were reported in Ref. 11 in which thermal treatments at ⬃600 ° C were needed to decompose arsenic oxides
and remove oxygen from the oxide/semiconductor interface.
The reduction of arsenic oxides, thus removing them from
the interface during our ALD process is an improvement
over the ALD in Ref. 11, in which a postthermal treatment is
needed and some defects may be generated with the diffusion of arsenic and oxygen from the interface to the oxide
surface.
The authors wish to acknowledge the support from Department of Natural Sciences at National Science Council
共under Grant No. NSC-93-2120-M-007-007兲, Taiwan, Republic of China.
1
M. Hong, C. T. Liu, H. Reese, and J. Kwo, in Encyclopedia of Electrical
and Electronics Engineering, edited by J. G. Webster 共Wiley, New York,
1999兲, Vol. 19, p. 87, and references therein.
2
Physics and Chemistry of III–V Compound Semiconductor Interfaces, edited by C. W. Wilmsen 共Plenum, New York, 1985兲, and references therein.
3
M. Hong, J. Kwo, A. R. Kortan, J. P. Mannaerts, and A. M. Sergent,
Science 283, 1897 共1999兲.
4
M. Hong, M. Passlack, J. P. Mannerts, J. Kwo, S. N. G. Chu, N. Moriya,
S. Y. Hou, and V. J. Fratello, J. Vac. Sci. Technol. B 14, 2297 共1996兲.
5
A. R. Kortan, M. Hong, J. Kwo, J. P. Mannaerts, and N. Kopylov, Phys.
Rev. B 60, 10913, 共1999兲.
6
F. Ren, M. Hong, W. S. Hobson, J. M. Kuo, J. R. Lothian, J. P. Mannaerts,
J. Kwo, Y. K. Chen, and A. Y. Cho, Tech. Dig. - Int. Electron Devices
Meet. 1996, 943; Solid-State Electron. 41, 1751 共1997兲.
7
Y. C. Wang, M. Hong, J. M. Kuo, J. P. Mannaerts, J. Kwo, H. S. Tsai, J. J.
Krajewski, J. S. Weiner, Y. K. Chen, and A. Y. Cho, Mater. Res. Soc.
Symp. Proc. 573, 219 共1999兲.
8
Y. C. Wang, M. Hong, J. M. Kuo, J. P. Mannaerts, J. Kwo, H. S. Tsai, J. J.
Krajewski, Y. K. Chen, and A. Y. Cho, Tech. Dig. - Int. Electron Devices
Meet. 1998, 67; IEEE Electron Device Lett. 20, 457 共1999兲.
9
Y. C. Wang, M. Hong, J. M. Kuo, J. P. Mannaerts, H. S. Tsai, J. Kwo, J. J.
Krajewski, Y. K. Chen, and A. Y. Cho, Electron. Lett. 35, 667 共1999兲.
10
P. D. Ye, G. D. Wilk, B. Yang, J. Kwo, S. N. G. Chu, S. Nakahara, H.-J.
L. Gossmann, J. P. Mannaerts, M. Hong, K. K. Ng, and J. Bude, Appl.
Phys. Lett. 83, 180 共2003兲.
11
M. M. Frank, G. D. Wilk, D. Starodub, T. Gustafsson, E. Garfunkel, Y. J.
Chabal, J. Grazul, and D. A. Muller, Appl. Phys. Lett. 86, 152904 共2005兲.
12
Y. L. Huang, P. Chang, Z. K. Yang, Y. J. Lee, H. Y. Lee, H. J. Liu J. Kwo,
J. P. Mannaerts, and M. Hong, Appl. Phys. Lett. 86, 191905 共2005兲, and
references therein.
Downloaded 17 Feb 2009 to 140.114.12.89. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp