Comparison of n -type and p -type GaAs oxide growth and its effects on frequency dispersion characteristics

Article abstract of DOI:10.1063/1.2987428

The electrical characteristics of n - and p -type gallium arsenide (GaAs) capacitors show a striking difference in the accumulation capacitance frequency dispersion. This difference has been attributed by some to a variation in the oxide growth, possibly due to photoelectrochemical properties of the two substrates. We show that the oxide growth on n - and p -type GaAs substrates is identical when exposed to identical environmental and chemical conditions while still maintaining the diverse electrical characteristics. The difference in electron and hole trap time constants is suggested as the source of the disparity of the frequency dispersion for n -type versus p -type GaAs devices.

Full text of DOI:10.1063/1.2987428

Comparison of n -type and p -type GaAs oxide growth and its effects on frequency
dispersion characteristics
C. L. Hinkle, A. M. Sonnet, M. Milojevic, F. S. Aguirre-Tostado, H. C. Kim, J. Kim, R. M. Wallace, and E. M.
Vogel
Citation: Applied Physics Letters 93, 113506 (2008); doi: 10.1063/1.2987428
View online: http://dx.doi.org/10.1063/1.2987428
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/93/11?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Effect of atomic layer deposition growth temperature on the interfacial characteristics of HfO2/p-GaAs metal-
oxide-semiconductor capacitors
J. Appl. Phys. 116, 222207 (2014); 10.1063/1.4902963
An investigation of capacitance-voltage hysteresis in metal/high-k/In0.53Ga0.47As metal-oxide-
semiconductor capacitors
J. Appl. Phys. 114, 144105 (2013); 10.1063/1.4824066
Sub-nm equivalent oxide thickness on Si-passivated GaAs capacitors with low Dit
Appl. Phys. Lett. 99, 052102 (2011); 10.1063/1.3615680
Influence of the substrate orientation on the electrical and material properties of GaAs metal-oxide-
semiconductor capacitors and self-aligned transistors using Hf O 2 and silicon interface passivation layer
Appl. Phys. Lett. 92, 202908 (2008); 10.1063/1.2917823
Impact of surface chemical treatment on capacitance-voltage characteristics of GaAs metal-oxide-
semiconductor capacitors with Al 2 O 3 gate dielectric
Appl. Phys. Lett. 91, 063501 (2007); 10.1063/1.2764438
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:
161.6.141.187 On: Thu, 11 Dec 2014 01:41:40

APPLIED PHYSICS LETTERS 93, 113506 ͑2008͒
Comparison of n-type and p-type GaAs oxide growth and its effects
on frequency dispersion characteristics
C. L. Hinkle,1,a͒ A. M. Sonnet,1 M. Milojevic,2 F. S. Aguirre-Tostado,2 H. C. Kim,2 J. Kim,2
R. M. Wallace,2,b͒ and E. M. Vogel1,2,c͒
1Department of Electrical Engineering, The University of Texas at Dallas, Richardson, Texas 75080, USA
2Materials Science and Engineering, The University of Texas at Dallas, Richardson, Texas 75080, USA
The electrical characteristics of n- and p-type gallium arsenide ͑GaAs͒ capacitors show a striking
difference in the “accumulation” capacitance frequency dispersion. This difference has been
attributed by some to a variation in the oxide growth, possibly due to photoelectrochemical
properties of the two substrates. We show that the oxide growth on n- and p-type GaAs substrates
is identical when exposed to identical environmental and chemical conditions while still maintaining
the diverse electrical characteristics. The difference in electron and hole trap time constants is
suggested as the source of the disparity of the frequency dispersion for n-type versus p-type GaAs
devices. © 2008 American Institute of Physics. ͓DOI: 10.1063/1.2987428͔
Gallium arsenide ͑GaAs͒ metal oxide semiconductor
͑MOS͒ capacitors have long been known to display a fre-
quency dispersion of the “accumulation” capacitance.1–7
This phenomenon has been attributed to a high density of
interface traps ͑Dit͒ and the associated Fermi level pinning
arising from problems associated with native or deposited
oxides.8–11 This dispersion is significantly more pronounced
on an n-type GaAs surface than for a p-type substrate.12,13
Previous studies on molecular beam epitaxy grown GaAs
and air-grown native oxides on n- and p-type GaAs have
suggested that there is an inherent stability to maintain a low
surface-state density for the p-type substrates compared to
the n-type surfaces.14,15 The differences in the stabilities have
been attributed to photoelectrochemical properties of the two
surfaces in which illuminated GaAs surfaces oxidize
differently.12,15 In this letter, n-type and p-type GaAs sub-
strates were carefully treated and exposed to identical chemi-
cal and phototreatments in an effort to determine the oxida-
tion differences of the two surfaces and their effect on the
electrical characteristics of MOS capacitors.
tainers simultaneously and were subjected to identical atmo-
spheric, chemical, and UHV conditions at all times as com-
panion specimens until capped with a gate metal electrode.
The native oxides of freshly unpacked samples were ana-
lyzed using XPS to ensure that the starting surfaces were
chemically identical ͑not shown͒. The samples were then ex-
posed to atmospheric conditions and ambient fluorescent
lighting for 30 min prior to chemical treatment. After chemi-
cal exposure, the wafers were exposed to the same ambient
͑laboratory air and lighting͒ conditions for ͑a͒ 12 min or ͑b͒
5 min preceding pumpdown to UHV. The surfaces after de-
grease and chemical treatment were then analyzed using XPS
͑not shown͒ and also display the same chemical bonding to
within the level of detection of XPS for both n-type and
p-type substrates, regardless of ambient exposure. It is noted
that the 12 min ex situ exposure between chemical cleaning
and introduction to the UHV environment has been shown
previously to be long enough to reoxidize the surface to near
completion ͑assuming no passivation͒, while the 5 min ex-
posure time should have an incomplete reoxidation.18 This
previous work then implies that the indistinguishable oxides
of the n-type and p-type substrates are not due to minimal
atmospheric exposure or due to a saturation of the oxides
during reoxidation. After the initial ALD and analysis, an-
other 9 nm of Al2O3 was deposited ͑10 nm total͒ and the
samples were removed from UHV and annealed at 600 °C
for 60 s in 99.999% pure N2. The TaN gate electrodes were
deposited by rf sputtering through shadow masks. Ohmic
contacts were formed by depositing Ti/Au for p-type GaAs
and Ni/Au/Ge for n-type GaAs each annealed at 450 °C for
60 s in N2.19
The samples used in this work were n-type Si-doped
GaAs wafers and p-type Zn-doped GaAs wafers with doping
concentration of 1–4ϫ1017 cm−3. The GaAs surfaces were
prepared by degreasing the wafers in acetone, methanol, and
isopropyl alcohol for 1 min each, followed by a 3 min etch in
29% NH4OH.16 In situ, 1.0 nm Al2O3 thin films were depos-
ited on GaAs in a Picosun SUNALE® atomic layer deposi-
tion ͑ALD͒ reactor integrated to an ultrahigh vacuum ͑UHV͒
multitechnique deposition/characterization system ͑base
17
pressure=2ϫ10−11 mbar͒ at a substrate temperature of
300 °C and a base pressure of ϳ7 Torr. Trimethyl alumi-
num was used as the precursor for Al2O3 formation. The
oxygen source was de-ionized H2O. Analysis of the chemi-
cally treated GaAs surfaces and deposited films was done
using in situ monochromatic x-ray photoelectron spectros-
copy ͑XPS͒ using an Al K␣ ͑1486.7eV͒ x-ray source with a
line width of ϳ0.25 eV and pass energy of 15 eV. The
n-type and p-type samples were removed from sealed con-
The background-subtracted normalized XPS spectra of
the two companion GaAs substrates with chemical treatment
followed by 12 min atmospheric exposure and 1 nm deposi-
tion of ALD Al2O3 ͑Fig. 1͒ fully overlap, clearly showing
that the surfaces exhibit the same bonding environment to
within detectable limits. The NH4OH does not passivate the
GaAs surface and, therefore, does not inhibit oxide growth in
air ͓in contrast to ͑NH4͒2S treatment͔,20 and was chosen as a
chemical treatment specifically for this reason. Of particular
note are the ratios of the As–O and Ga–O bonding to their
respective bulk peaks in the As 2p and Ga 2p spectra. The
a͒
Electronic mail: chris.hinkle@utdallas.edu.
Electronic mail: rmwallace@utdallas.edu.
b͒
c͒
Electronic mail: eric.vogel@utdallas.edu.
0003-6951/2008/93͑11͒/113506/3/$23.00
93, 113506-1
© 2008 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:
161.6.141.187 On: Thu, 11 Dec 2014 01:41:40

113506-2
Hinkle et al.
Appl. Phys. Lett. 93, 113506 ͑2008͒
FIG. 3. ͑Color online͒ C-V data of the same samples used for XPS analysis
showing different ͑a͒ n-type and ͑b͒ p-type frequency dispersion character-
istics despite having chemically identical gate oxides and interfaces.
analyzing interfacial oxides. All other individual spectral
ranges analyzed ͑Al 2p, As 3d, Ga 3d, and C 1s͒ also
show identical chemical states for both n-type and p-type
substrates ͑not shown͒.
FIG. 1. ͑Color online͒ XPS data of ͑a͒ As 2p, ͑b͒ Ga 2p, and ͑c͒ O 1s
spectra for n- and p-type GaAs surfaces after chemical treatment, 12 min
atmospheric and fluorescent lighting exposure, and 1 nm Al2O3 deposition.
The XPS spectra shown have been aligned to the n-type
Ga–As bulk peak binding energy for direct comparison of
the chemical states of the two substrates and their overlayers.
The difference in binding energy between the same core
level electrons ͑for example, the Ga 2p photoelectrons͒ of
n-type versus p-type substrates arises due to the differences
of the Fermi levels as well as band bending associated with
oxide charge.21,22 As the binding energy of the photoelec-
trons is measured with respect to the Fermi level, the n-type
substrate shows a higher binding energy for the core levels
with respect to photoelectrons from the p-type sample, and
we have therefore shifted the p-type curves to a higher bind-
ing energy. The actual peak shifts were 0.15 eV for Fig. 1
͑As 2p positions at 1322.85 and 1323.0 eV, Ga 2p positions
at 1117.25 and 1117.4 eV, and O 1s positions at 531.7 and
531.85 eV͒ and 0.3 eV for Fig. 2 ͑As 2p positions at 1322.4
and 1322.7 eV, Ga 2p positions at 1116.8 and 1117.1 eV, and
O 1s positions at 531.3 and 531.6 eV͒. The curves shown
have all been normalized in counts as well in order to show
the relative ratios of the oxides to the bulk signals for each of
the spectra.
Having established that the n-type and p-type substrates
have indistinguishable interfacial oxides and bonding envi-
ronments, perusal of the corresponding MOS capacitor elec-
trical characteristics is quite revealing. Figure 3 shows the
capacitance-voltage ͑C-V͒ characteristics of the same n-type
and p-type GaAs substrates and dielectrics analyzed by XPS.
These characteristics are typical of oxides on GaAs when
compared to published literature.1–4 The fact that the fre-
quency dispersion is clearly distinct between the two starting
surfaces with the same interfacial chemistry strongly indi-
cates that the observed dispersion is not due to a different
surface-state density caused by dissimilar oxide growth on
the two types of substrates.
oxide growth and chemical states of those oxides are identi-
cal for both n- and p-type surfaces despite the fact that the
surfaces were illuminated, suggesting that there was no de-
tectable photoelectrochemical reaction from ambient light
among the two substrates. Moreover, the O 1s spectrum
shows the bonding environment and total amount of surface
oxygen to be identical for the two wafers. Figure 2 shows the
two types of GaAs substrates with chemical treatment fol-
lowed by 5 min atmospheric exposure and 1 nm deposition
of ALD Al2O3. The extent of oxidation on these samples,
while being noticeably different from the 12 min exposure
samples, is chemically identical to each other for an n-type
and a p-type substrate. This shows that there is no difference
detected in the oxidation rate or chemical state of the n- or
p-type substrates as long as they are exposed to identical
conditions. It is noted that the As 2p and Ga 2p spectra are
very surface sensitive due to the binding energies of those
core level electrons and are therefore particularly useful in
Rather, these data suggest that the source of the different
electrical characteristics is due to a difference in time con-
stants of electron and hole interface traps in this system.
Simulations of capacitance-voltage characteristics have been
performed to demonstrate the effects of various factors on
the electrical characteristics of MOS devices.3 Two separate
models have been employed in this effort. The first is a clas-
sical formulation using numerically calculated oxide, sub-
strate, and interface trap capacitances using surface poten-
tials derived from total semiconductor charge.23 The second
FIG. 2. ͑Color online͒ XPS data of ͑a͒ As 2p, ͑b͒ Ga 2p, and ͑c͒ O 1s
spectra for n- and p-type GaAs surfaces after chemical treatment, 5 min
atmospheric and fluorescent lighting exposure, and 1 nm Al2O3 deposition.
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:
161.6.141.187 On: Thu, 11 Dec 2014 01:41:40

113506-3
Hinkle et al.
Appl. Phys. Lett. 93, 113506 ͑2008͒
the FCRP Materials Structures and Devices ͑MSD͒ Center,
and the U.S./Korea FUSION project.
1T. Sawada and H. Hasagawa, Electron. Lett. 12, 471 ͑1976͒.
2G. Sixt, K. H. Ziegler, and W. R. Fahrner, Thin Solid Films 56, 107
͑1979͒.
3C. L. Hinkle, A. M. Sonnet, E. M. Vogel, S. McDonnell, G. J. Hughes, M.
Milojevic, B. Lee, F. S. Aguirre-Tostado, K. J. Choi, J. Kim, and R. M.
Wallace, Appl. Phys. Lett. 91, 163512 ͑2007͒.
4M. Passlack, in Materials Fundamentals of Gate Dielectrics, edited by A.
A. Demkov and A. Navrotsky ͑Springer, New York, 2005͒, pp. 403–467.
5L. G. Meiners, in Physics and Chemistry of III-V Compound Semiconduc-
tor Interfaces, edited by C. W. Wilmsen ͑Plenum, New York, 1985͒, pp.
230–236.
FIG. 4. ͑Color online͒ Simulated C-V data using a tunneling modified in-
terface state model. The Dit distribution is the same for both ͑a͒ n-type and
͑b͒ p-type simulations as suggested from the XPS data. The simulation
differences arise from trap time constant differences for electrons and holes.
6J. L. Freeouf, D. A. Buchanan, S. L. Wright, T. N. Jackson, J. Batey, B.
Robinson, A. Callegari, A. Paccagnella, and J. M. Woodall, J. Vac. Sci.
Technol. B 8, 860 ͑1990͒.
modifies the classical model by introducing a lower bandgap
interfacial region along the lines of the model suggested by
Hasegawa and Sawada.24 Using a Dit distribution similar to
those suggested in prior reports9,24,25 ͑although any distribu-
tion shows qualitatively similar results͒, the modified model
produces the C-V curves shown in Fig. 4. Based on these
models, the reason for the differences in the n- and p-type
simulated C-V curves stems from the differences in the cap-
ture time constants for electrons and holes, respectively.
These time constants vary with differences in the conduction
and valence band densities of states, capture cross sections of
electrons versus holes, and any energetic asymmetries in the
Dit distribution.23,24 We also note that reaching the assumed
value of oxide capacitance ͑Cox͒ is not necessarily evidence
of an accumulation layer. The high Dit results in a very small
number of free carriers in accumulation and an associated
small substrate capacitance as compared to the interface state
capacitance. Because of the extremely high interface state
capacitance at low frequency, the total capacitance in accu-
mulation is close to Cox despite a very small number of free
carriers present.
In summary, we have shown that oxide growth on n-type
and p-type GaAs surfaces are identical under matching envi-
ronmental conditions and cannot be the origin of the ob-
served difference in capacitance dispersion of devices made
on the two substrates. No photoelectrochemical reaction
variation between the two starting surfaces before or after
chemical treatment or after dielectric deposition was de-
tected. The difference in electron and hole trap time con-
stants is the source for the disparity of the capacitance fre-
quency dispersion for n-type versus p-type GaAs devices.
7A. Callegari, D. K. Sadana, D. A. Buchanan, A. Paccagnella, E. D. Mar-
shall, M. A. Tischler, and M. Norcott, Appl. Phys. Lett. 58, 2540 ͑1991͒.
8F. Gao, S. J. Lee, D. Z. Chi, S. Balakumar, and D. L. Kwong, Appl. Phys.
Lett. 90, 252904 ͑2007͒.
9W. E. Spicer, I. Lindau, P. Pianetta, P. W. Chye, and C. M. Garner, Thin
Solid Films 56, 1 ͑1979͒.
10K. Martens, C. O. Chui, G. Brammertz, B. De Jaeger, D. Kuzum, M.
Meuris, M. M. Heyns, T. Krishnamohan, K. Saraswat, H. E. Maes, and G.
Groeseneken, IEEE Trans. Electron Devices 55, 547 ͑2008͒.
11S. Koveshnikov, W. Tsai, I. Ok, J. C. Lee, V. Torkanov, M. Yakimov, and
S. Oktyabrsky, Appl. Phys. Lett. 88, 022106 ͑2006͒.
12T. Yang, Y. Xuan, D. Zemlyanov, T. Shen, Y. W. Wu, J. M. Woodall, P. D.
Ye, F. S. Aguirre-Tostado, M. Milojevic, S. McDonnell, and R. M. Wal-
lace, Appl. Phys. Lett. 91, 142122 ͑2007͒.
13G. K. Dalapati, Y. Tong, W.-Y. Loh, H. K. Mun, and B. J. Cho, IEEE
Trans. Electron Devices 54, 1831 ͑2007͒.
14M. D. Pashley, K. W. Haberern, R. M. Feenstra, and P. D. Kirchner, Phys.
Rev. B 48, 4612 ͑1993͒.
15D. Yan, E. Look, X. Yin, F. H. Pollak, and J. M. Woodall, Appl. Phys.
Lett. 65, 186 ͑1994͒.
16Y. Xuan, H.-C. Lin, and P. Ye, IEEE Trans. Electron Devices 28, 935
͑2007͒.
17P. Sivasubramani, J. Kim, M. J. Kim, B. E. Gnade, and R. M. Wallace, J.
Appl. Phys. 101, 114108 ͑2007͒.
18J. Massies and J. P. Contour, J. Appl. Phys. 58, 806 ͑1985͒.
19N. Braslau, J. Vac. Sci. Technol. 19, 803 ͑1981͒.
20T. Scimeca, Y. Muramatsu, M. Oshima, H. Oigawa, and Y. Nannichi,
Phys. Rev. B 44, 12927 ͑1991͒.
21Y. Lebedinskii, A. Zenkevich, and E. P. Gusev, J. Appl. Phys. 101,
074504 ͑2007͒.
22S. A. Chambers and V. A. Loebs, Phys. Rev. B 47, 15 ͑1993͒.
23E. H. Nicollian and J. R. Brews, MOS Physics and Technology ͑Wiley,
Hoboken, 1982͒.
24H. Hasegawa and T. Sawada, IEEE Trans. Electron Devices 27, 1055
͑1980͒.
25G. Brammertz, K. Martens, H. C. Lin, C. Merckling, J. Penaud, C. Adel-
mann, S. Sioncke, M. Caymax, M. Meuris, and M. Heyns, Presentation
9.2 at the European Materials Research Society ͑Materials Research Soci-
ety, Strasbourg, France, 2008͒.
This work is supported by the National Institute of Stan-
dards and Technology, Semiconductor Electronics Division,
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:
161.6.141.187 On: Thu, 11 Dec 2014 01:41:40