Reduction of native oxides on InAs by atomic layer deposited Al 2O3 and HfO2

Article abstract of DOI:10.1063/1.3495776

Thin high- κ oxide films on InAs, formed by atomic layer deposition, are the key to achieve high-speed metal-oxide-semiconductor devices. We have studied the native oxide and the interface between InAs and 2 nm thick Al ₂ O₃ or HfO₂ layers using synchrotron x-ray photoemission spectroscopy. Both films lead to a strong oxide reduction, obtaining less than 10% of the native As-oxides and between 10% and 50% of the native In-oxides, depending on the deposition temperature. The ratio of native In- to As-oxides is determined to be 2:1. The exact composition and the influence of different oxidation states and suboxides is discussed in detail.

Full text of DOI:10.1063/1.3495776

Reduction of native oxides on InAs by atomic layer deposited Al 2 O 3 and HfO 2
R. Timm, A. Fian, M. Hjort, C. Thelander, E. Lind, J. N. Andersen, L.-E. Wernersson, and A. Mikkelsen
Citation: Applied Physics Letters 97, 132904 (2010); doi: 10.1063/1.3495776
View online: http://dx.doi.org/10.1063/1.3495776
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/97/13?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Electrical characterization of high- k based metal-insulator-semiconductor structures with negative resistance
effect when using Al 2 O 3 and nanolaminated films deposited on p -Si
J. Vac. Sci. Technol. B 29, 01A901 (2011); 10.1116/1.3521383
Stability of HfO 2 / SiO x / Si surficial films at ultralow oxygen activity
J. Appl. Phys. 103, 114108 (2008); 10.1063/1.2937900
Atomic layer deposited TaC y metal gates: Impact on microstructure, electrical properties, and work function
on HfO 2 high- k dielectrics
J. Appl. Phys. 102, 104509 (2007); 10.1063/1.2817620
Atomic layer deposited HfO 2 / HfSi x O y N z stacked gate dielectrics for metal-oxide-semiconductor
structures
J. Vac. Sci. Technol. B 25, 1922 (2007); 10.1116/1.2811707
Influence of single and double deposition temperatures on the interface quality of atomic layer deposited Al 2
O 3 dielectric thin films on silicon
J. Appl. Phys. 99, 054902 (2006); 10.1063/1.2177383
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.134.145.190 On: Sat, 29 Mar 2014 10:07:58

APPLIED PHYSICS LETTERS 97, 132904 ͑2010͒
Reduction of native oxides on InAs by atomic layer deposited
Al2O3 and HfO2
R. Timm,a͒ A. Fian,b͒ M. Hjort, C. Thelander, E. Lind, J. N. Andersen, L.-E. Wernersson,
and A. Mikkelsenc͒
Department of Physics, The Nanometer Structure Consortium, Lund University, P.O. Box 118,
22 100 Lund, Sweden
͑Received 13 July 2010; accepted 8 September 2010; published online 28 September 2010͒
Thin high-␬ oxide films on InAs, formed by atomic layer deposition, are the key to achieve
high-speed metal-oxide-semiconductor devices. We have studied the native oxide and the interface
between InAs and 2 nm thick Al2O3 or HfO2 layers using synchrotron x-ray photoemission
spectroscopy. Both films lead to a strong oxide reduction, obtaining less than 10% of the native
As-oxides and between 10% and 50% of the native In-oxides, depending on the deposition
temperature. The ratio of native In- to As-oxides is determined to be 2:1. The exact composition and
the influence of different oxidation states and suboxides is discussed in detail. © 2010 American
Institute of Physics. ͓doi:10.1063/1.3495776͔
The continuous demand for higher performance and
lower power consumption in transistor technology has in
recent years lead to a strongly increased interest in III-V
metal-oxide-semiconductor ͑MOS͒ devices. Here the quality
of the interfaces between III-V semiconductors and gate ox-
ides with a large dielectric constant—so-called “high-␬”
materials—is of crucial importance, since interfacial defects
and native oxides on the semiconductor lead to frequency
dispersion and hamper the electrical properties of gate
transistors.1–4
With the utilization of atomic layer deposition ͑ALD͒,
the crystal and interface quality of a few nanometer thick
high-␬ layers such as Al2O3 or HfO2 on GaAs and InGaAs
have been significantly improved.5 Beside the total thickness
of the remaining native oxide2 also the exact stoichiometry
and oxidation state of the Ga and As ions within this native
oxide determine the device performance.6 Therefore, it is im-
portant to quantitatively investigate the chemical composi-
tion of the interface region between the III-V semiconductor
and the high-␬ oxide, which can best be done by core-level
spectroscopy using x-ray photoemission ͑XPS͒.
order to vary the energy of the photoemitted electrons and
thus the probing depth. By thoroughly fitting the high-
resolution XPS core-level spectra of InAs-high-␬ and InAs
reference samples, we can reveal the composition of the na-
tive oxide and obtain quantitatively how the deposition of
Al2O3 and HfO2 layers reduces the different oxide compo-
nents.
Nominally undoped InAs͑100͒ substrates from Wafer
Technology Ltd. were used for all samples. After removal of
the epiready layer by heat treatment, the substrates were
etched for 30 s in hydrofluoric acid, rinsed in isopropyl
alcohol, and within a few minutes loaded into the ALD
chamber, except for the reference samples. 24 cycles of ei-
ther HfO2 or Al2O3 were deposited in a Cambridge Nano-
Tech Savannah 100 ALD reactor using tetrakis ͑dimethyla-
mido͒hafnium ͑Hf͓͑CH3͒2N͔ ͒ and trimethylaluminum
4
͑Al͑CH3͒3͒, resulting in about 2 nm thick films. Deposition
was started with a single Hf pulse or a double Al pulse,
respectively. Standard temperatures were 250 °C for HfO2
and 350 °C for Al2O3. After ALD, the samples were kept in
air for a few days.
XPS was performed at beamline I311 of the MAX II
synchrotron ring.15 In 3d, In 4d, As 3d, O 1s, Hf 4f, and
Al 2p core-level spectra were taken at photon energies var-
ied between 70 and 1330 eV. All spectra were fitted self-
consistently with the FITXPS package assuming a Doniach–
Sunjic line form, resulting for the As 3d ͑In 3d͒ core-level
in a spin-orbit splitting of 0.67Ϯ0.01 eV ͑7.56Ϯ0.01 eV͒,
a branching ratio of 1.52Ϯ0.04 ͑1.50Ϯ0.02͒, a Lorentz
width of 0.15Ϯ0.03 eV ͑0.31Ϯ0.05 eV͒, an asymmetry
factor ␣=0.030Ϯ0.004 ͑0.003Ϯ0.001͒, and a Gauss width
of the InAs bulk peak between 0.4 and 0.5 eV ͑0.4 and 0.6
eV͒, in very good agreement with literature.16–18 Strong
background components are observed both in the As and In
spectra of all high-␬ samples due to inelastic electron scat-
tering within the 2 nm thick oxide layer. Many samples were
loaded in the XPS chamber simultaneously, so that the rela-
tive binding energies of their spectra can be compared with
high accuracy and energy resolution.
Numerous XPS studies on high-␬ gate dielectrics depos-
ited by ALD on GaAs ͑Refs. 2 and 7͒ or InGaAs ͑Refs. 8 and
9͒ underline the promising potential of this system for appli-
cation in high-speed MOS devices. However, although the
electron mobility is larger in GaAs than in Si, it is further
increased by about an order of magnitude in InAs.10 Addi-
tionally, InAs is an important binary reference material to the
InGaAs material system. Studies by capacitance-voltage
spectroscopy on high-␬ thin films on bulk InAs ͑Refs. 11 and
12͒ as well as on InAs nanowires13 have recently been pre-
sented, but only a few XPS studies on high-␬ thin films on
InAs have been reported yet.14
In this work, we present XPS results on the interface
between InAs and Al2O3 or HfO2 layers deposited by ALD
at different temperatures. A synchrotron source was used in
a͒
Electronic mail: rainer.timm@sljus.lu.se.
Also at Institute of Nanostructured Materials and Photonics, Joanneum
b͒
Figure 1͑a͒ compares As 3d spectra of a reference
sample, an InAs/Al2O3 sample, and an InAs/HfO2 sample,
Research, Franz-Pichler-Str. 30, 8160 Weiz, Austria.
Electronic mail: anders.mikkelsen@sljus.lu.se.
c͒
0003-6951/2010/97͑13͒/132904/3/$30.00
97, 132904-1
© 2010 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.134.145.190 On: Sat, 29 Mar 2014 10:07:58

132904-2
Timm et al.
Appl. Phys. Lett. 97, 132904 ͑2010͒
FIG. 1. ͑Color online͒ XPS As 3d spectra of a reference, an InAs/Al2O3,
and an InAs/HfO2 sample. ͑a͒ Sample comparison. The spectra are normal-
ized and shifted for clarity. ͓͑b͒–͑d͔͒ Fitted data. In ͑d͒ an additional Hf 5p
1/2 peak needs to be considered. A constant part of the actually larger
background in ͑d͒ is subtracted from the curves for clarity.
FIG. 2. ͑Color online͒ ͓͑a͒–͑d͔͒ XPS In 3d spectra of a reference, an
InAs/Al2O3, and an InAs/HfO2 sample. ͑a͒ Sample comparison. The spec-
tra are normalized and shifted for clarity. ͓͑b͒–͑d͔͒ Fitted data. ͑e͒ Fitted
XPS In 4d spectrum of a reference sample. Two different In-oxide peaks
can be distinguished.
showing the influence of the high-␬ film on the native oxide:
The doublet at a binding energy around 41 eV comes from
arsenic bound to indium, and the broad peak at about 44 eV
is caused by arsenic bound to oxygen. Thus, a strong
As-oxide signal is obtained for the reference sample, while
both high-␬ samples contain only very little As-oxide. The
amount of native oxide on the reference samples was ob-
served to be almost stable after less than ten minutes. There-
fore, also the InAs/HfO2 and InAs/Al2O3 samples must
have had a similar native oxide layer prior to ALD, which
was then reduced by the high-␬ material.
To study the oxide reduction in more detail, fitted spectra
are shown in Figs. 1͑b͒–1͑d͒ for the different samples. It can
be seen that two doublet components are necessary to fit the
native oxide peak, a larger one at 3.2 eV and a smaller one at
4.3 eV above the bulk InAs binding energy, with Gauss
widths of 0.9 eV and 1.2 eV, respectively. They can be iden-
tified as arsenic in a +3 oxidation state ͑like As2O3͒ and in a
+5 oxidation state ͑like As2O5͒.18,19 A substantial reduction
of both oxide peaks by the high-␬ material is obtained from
the spectra: The oxide peak of the InAs/Al2O3 as well as of
the InAs/HfO2 samples typically reaches less then 10% of
the corresponding oxide peak size at the reference samples,
corresponding to less than 0.1 nm layer thickness, for all
photon energies. The Al2O3 films seem, however, to be more
efficient than the HfO2 films to also reduce the As5+ oxide
component.
ing 50%–70% of the corresponding reference oxide size, as
summarized in Fig. 3͑a͒.
For both high-␬ materials, the separation in binding en-
ergy between the fitted In–As bulk and In-oxide peaks is
about 50 meV larger than for the reference sample. An In-
oxide consisting of two suboxides with different In oxidation
states, where the high-␬ films reduce that with the lower
oxidation state stronger, would lead to the observed shift of
the oxide peak energy. This might be important for applica-
tions, since in the well-known GaAs system the Ga3+ subox-
ide Ga2O3 is responsible for a strong decrease of device per-
formance, while the Ga1+ suboxide Ga2O has less influence.6
The oxides of the InAs material system are more complex
but by fitting an In 4d spectrum of a reference sample, as
shown in Fig. 2͑e͒, indeed two different In-oxide compo-
nents get apparent. However, the fitting uncertainty is too
large here to allow quantitative measurements of the oxide
reduction.
Concentrating now on the In peaks, a very thorough
spectra fitting is required to analyze the oxide composition,
since the bulk In-As and the In-oxide peaks overlap. Figures
2͑a͒–2͑d͒ show a comparison as well as individual In 3d
spectra of a reference sample, an InAs/Al2O3 sample, and an
InAs/HfO2 sample. Only one oxide peak is fitted, with a
Gauss width of about 1.2 eV, which is dominating the In–As
bulk peak at the reference samples, but gets significantly
reduced by the high-␬ layers: At the InAs/Al2O3 samples,
the oxide peak typically has about 40%–50% the magnitude
of the corresponding peak of a reference sample, while in the
InAs/HfO samples the oxide is slightly less reduced, show-
FIG. 3. ͑Color online͒ ͑a͒ Average oxide layer thickness of reference
samples, InAs/HfO2 samples, and InAs samples with Al2O3 films deposited
at 350 °C ͑standard͒ as well as 200 °C. ͓͑b͒ and ͑c͔͒ XPS As 3d and In 3d
spectra of the 200 °C InAs/Al2O3 sample. Peak components are indicated
as in Figs. 1 and 2. In ͑b͒ an additional shoulder needs to be considered.
This article is copy2righted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.134.145.190 On: Sat, 29 Mar 2014 10:07:58

132904-3
Timm et al.
Appl. Phys. Lett. 97, 132904 ͑2010͒
The stoichiometry of the native oxide is obtained from
the relative heights of the oxide peaks in In 3d and As 3d
reference spectra acquired at the same kinetic energy of the
photoelectrons, amounting to 62Ϯ5% In-oxide ͓see Fig.
3͑a͔͒. By simulating the relative oxide and bulk peak sizes
for each spectrum, the thickness of the native oxide film is
obtained, which varies between 1.1 and 1.4 nm in the differ-
ent reference samples. The As-oxide consists to about 80%
of arsenic with the oxidation state +3, as in As2O3, and to
about 20% of As5+, which can be assumed to As2O5 or the
mixed oxide InAsO4.18–20 InAsO4 as well as the stable In-
oxide is In2O3 consist of In3+. Since another In oxidation
In conclusion, we have studied the thickness and com-
position of the native oxide film on InAs. Strong oxide re-
duction is obtained by thin ALD films of Al2O3 or HfO2,
especially for As-oxides. These results underline the highly
promising perspective for InAs MOS devices and contribute
to an improved understanding of the complex processes tak-
ing place at the semiconductor high-␬ interface.
The authors acknowledge the valuable discussion with
Professor R. M. Wallace ͑UT Dallas͒ as well as substantial
support by K. Schulte at the MAX-lab synchrotron facility.
This work was performed within the Nanometer Structure
Consortium at Lund University, and was supported by the
Swedish Research Council ͑VR͒, the Swedish Foundation
for Strategic Research ͑SSF͒, the Crafoord Foundation, and
the Knut and Alice Wallenberg Foundation.
14
state is additionally observed, also In2O ͑with In1+͒ or InO
͑with In2+͒ could be present in the native oxide. Instead, also
the existence of a single phase metastable nonstoichiometric
oxide with microscopic units of different As and In oxidation
states could well explain the observed data.18 Calculations of
the Gibbs energy of formation have shown that InAsO4 and
In2O3 are energetically much more favorable than As2O3 or
As2O5,20 explaining the excess of In-oxide.
1M. Houssa, E. Chagarov, and A. Kummel, MRS Bull. 34, 504 ͑2009͒.
2G. K. Dalapati, Y. Tong, W.-Y. Loh, H. K. Mun, and B. J. Cho, IEEE
Trans. Electron Devices 54, 1831 ͑2007͒.
3H. Hasegawa and M. Akazawa, Appl. Surf. Sci. 254, 8005 ͑2008͒.
4E. J. Kim, L. Wang, P. M. Asbeck, K. C. Saraswat, and P. C. McIntyre,
Appl. Phys. Lett. 96, 012906 ͑2010͒.
At the InAs/high-␬ interface, the stoichiometry of the
oxide has changed to about 90% In-oxide, as shown in Fig.
3͑a͒. It has been reported that already the first cycle of Al2O3
deposition reduces most of the native oxide on GaAs,7,9
while on InAs an In-oxide peak related to In1+ remains.14
However, when we repeated our XPS measurements on ex-
actly the same InAs/HfO2 and InAs/Al2O3 samples after
several months, a much larger amount of both As- and In-
oxides was observed, showing a reoxidation through the only
2 nm thick high-␬ films under ambient conditions. Thus,
already the originally measured interfacial oxide might be
partly due to reoxidation after the sample preparation.
For high-␬ layers deposited at standard temperatures of
250 °C for HfO2 and 350 °C for Al2O3, no signs of alumi-
num or hafnium binding to indium or arsenic are observed in
the corresponding spectra. Varying the HfO2 deposition tem-
perature to either 150 or 350 °C slightly decreases the oxide
reduction but does not qualitatively alter the resulting spec-
tra. The temperature of the Al2O3 deposition, however, is
found to be of strong importance. Figures 3͑b͒ and 3͑c͒ show
As 3d and In 3d spectra of an Al2O3 film deposited on InAs
at 200 °C. In the As 3d spectrum, a shoulder at the high
binding energy side of the InAs bulk peak is obvious, which
can be fitted as an additional doublet with a 0.4 eV larger
binding energy, indicating arsenic bonded to aluminum.21 In
the In 3d spectrum, the In-oxide peak is significantly smaller
than at the standard InAs/Al2O3 samples, reaching only 10%
of the reference oxide peak size ͓Fig. 3͑a͔͒. A possible ex-
planation for both results could be the formation of a thin
interfacial AlAs layer during ALD at this reduced tempera-
ture, which then protects ͑at least partly͒ the underlying InAs
from being reoxidized later on.
5R. M. Wallace, P. C. McIntyre, J. Kim, and Y. Nishi, MRS Bull. 34, 493
͑2009͒.
6C. L. Hinkle, M. Milojevic, B. Brennan, A. M. Sonnet, F. S. Aguirre-
Tostado, G. J. Hughes, E. M. Vogel, and R. M. Wallace, Appl. Phys. Lett.
94, 162101 ͑2009͒.
7H. D. Lee, T. Feng, L. Yu, D. Mastrogiovanni, A. Wan, T. Gustafsson, and
E. Garfunkel, Appl. Phys. Lett. 94, 222108 ͑2009͒.
8M. Kobayashi, P. T. Chen, Y. Sun, N. Goel, P. Majhi, M. Garner, W. Tsai,
P. Pianetta, and Y. Nishi, Appl. Phys. Lett. 93, 182103 ͑2008͒.
9M. Milojevic, F. S. Aguirre-Tostado, C. L. Hinkle, H. C. Kim, E. M.
Vogel, J. Kim, and R. M. Wallace, Appl. Phys. Lett. 93, 202902 ͑2008͒.
10Silicon (Si), Electron Mobility; Gallium Arsenide (GaAs), Electron Mobil-
ity; Indium Arsenide (InAs), Carrier Mobilities, Landolt-Börnstein, New
Series, Group III, edited by O. Madelung, U. Rössler, and M. Schulz
͑Springer, New York, 2002͒.
11D. Wheeler, L.-E. Wernersson, L. Fröberg, C. Thelander, A. Mikkelsen,
K.-J. Weststrate, A. Sonnet, E. M. Vogel, and A. Seabaugh, Microelectron.
Eng. 86, 1561 ͑2009͒.
12E. Lind, Y.-M. Niquet, H. Mera, and L.-E. Wernersson, Appl. Phys. Lett.
96, 233507 ͑2010͒.
13S. Roddaro, K. Nilsson, G. Astromskas, L. Samuelson, L.-E. Wernersson,
O. Karlström, and A. Wacker, Appl. Phys. Lett. 92, 253509 ͑2008͒.
14A. P. Kirk, M. Milojevic, J. Kim, and R. M. Wallace, Appl. Phys. Lett. 96,
202905 ͑2010͒.
15See detailed description at http://www.maxlab.lu.se/beamlines/bli311/.
16D. M. Poirier and J. H. Weaver, Surf. Sci. Spectra 2, 224 ͑1993͒.
17L. O. Olsson, L. Ilver, J. Kanski, P. O. Nilsson, C. B. M. Andersson, U. O.
Karlsson, and M. C. Håkansson, Phys. Rev. B 53, 4734 ͑1996͒.
18G. Hollinger, R. Skheyta-Kabbani, and M. Gendry, Phys. Rev. B 49,
11159 ͑1994͒.
19D. Y. Petrovykh, M. J. Yang, and L. J. Whitman, Surf. Sci. 523, 231
͑2003͒.
20G. P. Schwartz, J. E. Griffiths, and G. J. Gualtieri, Thin Solid Films 94,
213 ͑1982͒.
21J. A. Taylor, J. Vac. Sci. Technol. 20, 751 ͑1982͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
142.134.145.190 On: Sat, 29 Mar 2014 10:07:58