Atomic layer deposition of Al2O3 thin films using trimethylaluminum and isopropyl alcohol

Article abstract of DOI:10.1149/1.1470659

Al₂O₃ thin films were deposited on Si substrates by atomic layer deposition (ALD) using Al(CH₃)₃ (trimethylaluminum, TMA) as a metal source and (CH₃)₂CHOH (isopropyl alcohol, IPA) as an oxygen source at 250°C. The film growth rate is saturated at 0.8 A/cycle, and it is slightly lower than that for the procedure which uses water vapor instead of IPA. The as-deposited film has a stoichiometry close to Al₂O₃ (Al/O ratio ～2:3.1) and the residual carbon content of the films is below the detection limit of secondary ion mass spectroscopy. An interface oxide between the Al₂O₃ film and Si-substrate was not detectable on the as-deposited film by transmission electron microscopy. However, after annealing at 800°C for 5 min, an interface oxide is newly formed even under the neutral ambient of Ar, and it grows thicker under the oxidizing ambient of O₂. That is, oxygen, which is needed for forming the interface oxide during the annealing process is supplied from both of the ambient oxygen and the excess oxygen in the films. Thus, although the loss of dielectric constant is inevitable during the annealing process, it can be minimized by using IPA in Al₂O₃ ALD.

Full text of DOI:10.1149/1.1470659

C306
Journal of The Electrochemical Society, 149 ͑6͒ C306-C310 ͑2002͒
0013-4651/2002/149͑6͒/C306/5/$7.00 © The Electrochemical Society, Inc.
Atomic Layer Deposition of Al2O3 Thin Films
Using Trimethylaluminum and Isopropyl Alcohol
a,z
Woo-Seok Jeon,a Sung Yang, Choon-soo Lee,b, and Sang-Won Kang
*
aDepartment of Materials Science and Engineering, Korea Advanced Institute of Science and Technology,
Taejon 305-701, Korea
bGenitech, Incorporated, Taejon 306-230, Korea
Al2O3 thin films were deposited on Si substrates by atomic layer deposition ͑ALD͒ using Al(CH3)3 ͑trimethylaluminum, TMA͒
as a metal source and (CH3)2CHOH ͑isopropyl alcohol, IPA͒ as an oxygen source at 250°C. The film growth rate is saturated at
0.8 Å/cycle, and it is slightly lower than that for the procedure which uses water vapor instead of IPA. The as-deposited film has
a stoichiometry close to Al2O3 ͑Al/O ratio ϳ2:3.1͒ and the residual carbon content of the films is below the detection limit of
secondary ion mass spectroscopy. An interface oxide between the Al2O3 film and Si-substrate was not detectable on the as-
deposited film by transmission electron microscopy. However, after annealing at 800°C for 5 min, an interface oxide is newly
formed even under the neutral ambient of Ar, and it grows thicker under the oxidizing ambient of O2 . That is, oxygen, which is
needed for forming the interface oxide during the annealing process is supplied from both of the ambient oxygen and the excess
oxygen in the films. Thus, although the loss of dielectric constant is inevitable during the annealing process, it can be minimized
by using IPA in Al2O3 ALD.
© 2002 The Electrochemical Society. ͓DOI: 10.1149/1.1470659͔ All rights reserved.
Manuscript submitted May 16, 2001; revised manuscript received December 11, 2001. Available electronically April 12, 2002.
As the size of microelectronic devices is scaled down, oxide
wall and gas delivery line, which would generate particles and elon-
gate the pulse time of inert gas to purge out H2O vapor from the
reactor. It has been clearly shown that IPA is an efficient reactant for
trimethylaluminum ͑TMA͒ in forming aluminum oxide, and it
should also be useful in ALD of other metal oxides.
films with high dielectric constants are required for gate and capaci-
tor dielectrics. Even though SiO2 is an excellent dielectric material
in many features, the preparation of reliable SiO2 films thinner than
30 Å is difficult, because of the quantum mechanical tunneling of
electrons, which generates a large leakage current across the films.1
Thus, an alternative oxide film with higher dielectric constant than
SiO2 is needed.2 In this respect, Al2O3 has been considered as a
possible candidate for gate oxides as well as Ta2O5 , HfO2 , and
ZrO2 .3-5 Also, Al2O3 is considered as a promising capacitor dielec-
tric, which can replace silicon oxynitride ͑SiON͒ in future dynamic
random access memories ͑DRAM͒.6 In addition, Al2O3 has been
recently regarded as a good passivation layer, which can prevent
hydrogen penetration into the active region of semiconductor
devices.7
In preparing high-dielectric metal oxides with a thickness in the
subhundred angstrom range, atomic layer deposition ͑ALD͒ attracts
considerable interest owing to its inherent merits in thin film depo-
sitions, such as a digital controllability of film thickness on an
atomic scale, a superior conformal deposition on three-dimensional
structures, and no particle generation, etc.8-10 In typical ALD of
metal oxide thin films, H2O has been most widely used as the oxy-
gen source.11-16
However, by using H2O an unwanted interface oxide may be
formed between the metal oxides and Si substrates, and it is thought
that the interface oxide is composed mainly of SiOx , as the result of
the reaction between H2O and the Si substrates during the initial
stage of the ALD. Hence, the interfusion of the interface oxide
would be expected to reduce the dielectric constant of the deposited
thin films.17-19 Therefore, the suppression of the formation of the
interface oxide is required in order to retain the dielectric constant of
the metal oxide itself. Although Gusev et al. reported that no inter-
face oxide was present between H-terminated Si and Al2O3 ,20 Fuji-
nami et al. reported that rinsing in pure water for 30 min or more
after HF treatment initiated oxidation of the Si surface even at room
temperature.21
Experimental
Al2O3 films were grown on Si͑100͒ substrates at 250°C using
TMA and IPA under a deposition pressure of 133 Pa. The growth
experiments were carried out using a hot-wall-type ALD reactor.
The initial Si͑100͒ wafer surface was treated in 50% HF solution for
3 min, rinsed in deionized ͑DI͒ water, and blown with a nitrogen
gun. TMA and IPA were alternately supplied into the reaction cham-
ber by the sequential injection of a TMA pulse, an Ar pulse, an IPA
pulse, and then another Ar pulse. This sequential injection is one
deposition cycle, and the growth of films is accomplished by repeat-
ing a cycle. The Ar pulse isolates the reactants from each other.
During ALD, the partial pressure and the pulse time of TMA were
controlled in the range of 1.33-13.3 Pa and 0.3-5.0 s, respectively. In
case of IPA, the pulse time was also changed from 0.3 to 5 s, but the
partial pressure was fixed at 13.3 Pa.
An annealing process was performed on the as-deposited films
by a rapid thermal process ͑RTP͒ at 800°C for 5 min at 40,000 Pa
under the neutral ambient of Ar or the oxidizing ambient of O2 .
Before and after the annealing, the surface smoothness, stoichiom-
etry, and crystallinity of the films were evaluated with atomic force
microscopy ͑AFM͒, X-ray photoelectron spectroscopy ͑XPS͒, and
X-ray diffraction ͑XRD͒, respectively. A single-crystal sapphire was
used as a standard sample to correct the relative atomic sensitivity
factor ͑RASF͒ of XPS for aluminum and oxygen. Carbon incorpo-
ration in the films was investigated with SIMS and the film thick-
ness was measured with an ellipsometer ͑Rudolph AutoEL-II͒. In
addtion, the interface between the deposited Al2O3 films and the Si
substrate was investigated by TEM analysis. The current-voltage
(I-V) and the capacitance-voltage ͑C-V at 1 MHz͒ characteristics of
the films were measured on metal oxide semiconductor ͑MOS͒ ca-
pacitors of Pt/Al2O3 /p-type Si͑100͒ structure, in which 1000 Å
thick Pt dots with a diameter of 160 ␮m were sputtered on the
Al2O3 films.
In this study, isopropyl alcohol ͑IPA͒ was used instead of H2O in
Al2O3 ALD in order to prevent the interface oxide from being
formed during the film deposition. Moreover, the use of IPA avoids
problems caused by the condensation of H2O vapor onto the reactor
Results and Discussion
Figure 1 shows the growth kinetics of Al2O3 ALD depending on
the pulse time and partial pressure of TMA and IPA at a deposition
temperature of 250°C. The deposition of Al2O3 shows typical char-
* Electrochemical Society Active Member.
z E-mail: swkang@kaist.ac.kr
Downloaded on 2015-06-20 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 149 ͑6͒ C306-C310 ͑2002͒
C307
Figure 1. Thickness/cycle of deposited Al2O3 films on Si͑100͒ vs. reactant
pulse times at 250°C. ͑a͒ The dependence of growth thickness/cycle of
Al2O3 film on TMA pulse time, the partial pressure of TMA was 1.33 Pa, the
pulse time of IPA was 3 s. ͑b͒ The dependence of growth thickness/cycle of
Al2O3 film on TMA pulse time, the partial pressure of TMA was 13.3 Pa, the
pulse time of IPA was 3 s. ͑c͒ The dependence of growth thickness/cycle of
Al2O3 film on IPA pulse time, the partial pressure of TMA was 1.33 Pa, the
pulse time of TMA was 3 s.
Figure 2. XPS depth profile through 110 Å Al2O3 films on Si͑100͒, com-
paring ͑a͒ no anneal vs. ͑b͒ Ar, ͑c͒ O2 anneal. The atomic percents of Al and
O were ͑a͒ 37.39 and 58.33%, ͑b͒ 37.35 and 58.58%, and ͑c͒ 37.34 and
58.58%. The extent of carbon impurities were minimal.
Before and after annealing at 800°C for 5 min under Ar or O2
ambient, the chemical composition of the films was analyzed by
XPS using a single-crystal sapphire as a reference. Figure 2 illus-
trates the XPS depth profiles for aluminum, oxygen, and carbon in
the films. The chemical composition of the films remains unchanged
before and after annealing and, regardless of the annealing ambient,
it is slightly oxygen-rich with the composition ratio of Al:O
ϭ 1:1.56. The XPS analysis also shows that carbon impurities are
negligible in the bulk of Al2O3 film as well as at the Al2O3 /Si
interface. In order to confirm the carbon impurities in the films,
SIMS depth profiles of carbon were measured, and are shown in Fig.
3. Except at the film surface, the carbon impurities in the films were
about 102 counts/s, almost the detection limit of SIMS, before and
after the annealing, and are independent of the annealing conditions.
That is, Al2O3 films deposited by ALD using IPA are nearly carbon-
free. This extremely low carbon incorporation is significant, when
compared to results previously reported in other ALD24 and metall-
organic chemical vapor depositions ͑MOCVD͒ of Al2O3 .18,25,26 In
their work, the carbon impurities were incorporated into the films up
acteristics of ALD, in which the film thickness deposited during one
cycle ͑thickness/cycle͒ is self-limited. From the deposition charac-
teristics, it is clear that IPA reacts with TMA and forms aluminum
oxides at 250°C. When the pulse time of IPA exceeds 1 s, the
thickness/cycle is saturated at 0.8 Å/cycle. In addition, the saturated
pulse time of TMA is shortened from 2 s to 1 s as the partial pres-
sure of TMA increases from 1.33 Pa to 13.3 Pa. However, the
thickness/cycle at the saturated condition is not affected by the par-
tial pressure of TMA. This implies that, under the saturated condi-
*
tion, the amount of TMA ͑the asterisk denotes all possible surface
species remaining after the TMA pulse͒ adsorbed on the film surface
is limited by its molecular size owing to steric hindrance. The satu-
rated thickness/cycle of 0.8 Å/cycle, which corresponds to 0.37
monolayer/cycle ͑ML/cycle͒, is lower than that of 1.0 Å/cycle re-
ported by George et al., in which H2O was used instead of IPA at
the same deposition temperature of 250°C.12 The difference in the
saturated thickness/cycle between two Al2O3 ALDs can be under-
stood by adopting the concept of secondary adsorption, which has
been proposed in our previous papers to explain the saturated
thickness/cycle exceeding 1 ML/cycle in ALD.22,23 During IPA or
*
H2O pulse, the oxygen sources will react with TMA adsorbed on
the film surface and form Al2O3 . The oxygen sources are then likely
to adsorb on both of the outermost surfaces of the newly formed
Al2O3 and the newly exposed surface, which had been screened
*
with TMA before. This is referred to as the secondary adsorption.
*
As a result, the surface of the film will be covered with IPA or
*
H2O after the pulse of the oxygen sources, and then these second-
ary adsorbed oxygen sources will also react with TMA during the
following TMA pulse. However, at the saturation conditions in the
two Al2O3 ALDs, where the adsorption and desorption rates are at
the steady-state condition, the quantity of the secondary adsorbed
*
*
chemical groups would be much less in case of IPA than H2O ,
*
because IPA has a much larger molecular size. As a result, the total
amount of TMA that will be involved in Al2O3 formation during a
cycle will be reduced. Thus, when IPA is used as an oxygen source
in Al2O3 ALD, the saturated thickness/cycle will be reduced as com-
pared to the situation, where H2O is used as an oxygen source.
Figure 3. SIMS depth profiles of carbon in Al2O3 films. Except at the films
surface, the films were nearly carbon-free before and after annealing.
Downloaded on 2015-06-20 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C308
Journal of The Electrochemical Society, 149 ͑6͒ C306-C310 ͑2002͒
Figure 5. TEM cross-sectional images of Al2O3 film using TMA and H2O
as-deposited at 250°C. Interlayers were present at the interfacial regions of
Al2O3 /Si in as-deposited films.
roughness of the films analyzed by AFM is a few angstroms in root
mean square values and shows no difference between before and
after the annealing.
It is well known that the Al2O3 films formed by conventional
MOCVD yield SiO2 at the interfacial region of the Al2O3 /Si sub-
strate before annealing.18,19 In order to study the formation of inter-
face oxide in Al2O3 ALD, cross-sectional TEM images were inves-
tigated before and after the annealings. Figures 4a and 5 illustrate
the formation of an interfacial layer on the as-deposited Al2O3 films
depending on the oxygen sources used. When IPA is used, no inter-
facial layer is detectable. However, an interfacial layer with a thick-
ness of 12 Å, as shown in Fig. 5, is generated on the as-deposited
film that is grown by ALD using H2O instead of IPA. The interfacial
layer should be formed during an early stage of Al2O3 ALD, be-
cause the surface of the Si substrate will be directly exposed to H2O
until it is completely covered with an Al2O3 layer, and the exposed
surface will almost certainly be oxidized at a temperature of 250°C,
which is the deposition temperature of the film. Hence, the interfa-
cial layer should be an oxide having a composition of SiAlxOy ,
even though it is not pure SiO2 . On the other hand, the data herein
show that IPA does not thermally oxidize Si at a temperature of
250°C, and thus, although the surface of the Si substrate is also
exposed to IPA during an early stage of Al2O3 ALD, there is no
chance to form silicon oxides at the interface. It would be helpful to
obtain a higher effective dielectric constant in the films by suppress-
ing the formation of SiAlxOy , which should have a lower dielectric
constant than Al2O3 . However, as shown in Fig. 4b and c, the in-
terface oxide is newly generated during the annealing processes, and
is inevitable, even though the annealing process is performed under
the neutral ambient of Ar. The thickness of the interface oxide ap-
pears to be about 10 and 25 Å after annealing under the ambient of
Ar and O2 , respectively. This implies that the interface oxide is
grown by the self-reaction between excess oxygen in the Al2O3 film
and the Si substrate at the annealing temperature of 800°C, and
further by the thermal oxidation of the Si substrate due to the diffu-
sion of the ambient oxygen through the film. Therefore, the film
annealed in O2 ambient has a thicker interface oxide. But, surpris-
ingly, before and after annealing the increment of the film thickness
including the interface oxide, is almost negligible. From this, it
would be expected that the annealing process would make the Al2O3
layer become dense, leading to reduction in the thickness of the
Al2O3 layer. In order to clarify the existence of silicon oxides at the
interface, XPS depth profiles were performed on the films which
were deposited using IPA, and the results for Si 2p near the interface
Figure 4. TEM cross-sectional images of Al2O3 films using TMA and IPA
͑a, top͒ as-deposited at 250°C, ͑b, center͒ after annealing at 800°C for 5 min
in Ar, and ͑c, bottom͒ after annealing at 800°C for 5 min in O2 . Interlayers
were present at the interfacial regions of Al2O3 /Si after annealing.
to the order of atomic percent, and were piled up especially at the
Al2O3 /Si interface. This suggests that IPA is very efficient reactant
for TMA without leaving a hydrocarbon residue. From the XRD
analysis, not shown here, it is revealed that the Al2O3 films have an
amorphous structures before and after the annealing process, while
the Al2O3 films as-deposited by MOCVD using TMA and N2O at
720-1050°C have ␥-Al2O3 structure.26,27 This difference in crystal
structure may be due to the deposition temperature. The surface
Downloaded on 2015-06-20 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 149 ͑6͒ C306-C310 ͑2002͒
C309
deposition, and that the interface oxide which is newly generated
during annealing is inevitable owing to the self-reaction between
Al2O3 and the Si substrates.
The I-V and C-V characteristics were measured on the MOS
capacitor with Pt/80 Å thick Al2O3 /Si substrate structure before and
after annealing. The leakage current density measured at 1 MV/cm
decreases from ϳ10Ϫ6 A/cm2 to 5 ϫ 10Ϫ8 A/cm2 and 1 ϫ 10Ϫ8
A/cm2 after annealing under Ar and O2 ambients, respectively. The
reduction in the leakage current after annealing can be partly attrib-
uted to the densification of Al2O3 , as previously mentioned in Fig.
4. Regardless of the annealing ambient, the time zero dielectric
breakdown ͑TZDB͒ of the annealed films is у7 MV/cm. However,
as expected from the growth of the interface oxide, which is depen-
dent on the annealing ambient, the film, which has a bilayer struc-
ture with the Al2O3 layer and the interfacial layer of SiAlxOy ,
shows differences in its effective dielectric constant after annealing.
For the case of the Ar-annealed film, the effective dielectric constant
is about 11, while that of the film annealed in O2 ambient is about 9.
The film thickness equivalent to SiO2 , tox , was 30 Å for the Ar
annealed Al2O3 film. The trend in effective dielectric constants, as
measured by the C-V method is qualitatively in good agreement
with the analysis obtained from the cross-sectional TEM imagines in
Fig. 4; the thicker the interface layer, the lower the effective dielec-
tric constant. Further studies are needed to develop a quantitative
approach to this issue, however.
Conclusions
Amorphous Al2O3 thin films were deposited by ALD using TMA
and IPA at 250°C. The fully saturated deposition thickness/cycle of
0.8 Å/cycle was determined using optimum precursor pulse times.
AFM images revealed exceptionally flat and conformal Al2O3 films
with a surface roughness of only 3.5 Å ͑rms͒ after a deposition of
ϳ80 Å film. In spite of the use of carbon-based oxygen precursor,
the films were nearly carbon-free and the stoichiometry of the films
was close to Al2O3 . The tox was about 30 Å. As evidenced by TEM
and XPS analyses, the SiAlxOy was formed at the interfacial regions
of the Al2O3 /Si after the annealing treatments. This had an effect on
the tox . However, the as-deposited film has no interfacial layer at the
interfacial regions. Contrary to H2O, IPA does not thermally react
with Si at a temperature of 250°C. The Al2O3 film prepared by ALD
using TMA and IPA appears to be an excellent candidate material
for sub-0.2 ␮m design rule applications.
Acknowledgments
This work was supported by the National Research Laboratory
program of the Ministry of Science and Technology.
Korea Advanced Institute of Science and Technology assisted in meeting
the publication costs of this article.
References
1. A. Masaki, IEEE Circuits Devices Mag., 8, 18 ͑1992͒.
2. V. Mikhaelashivili, Y. Betzer, I. Prudnikov, M. Orenstein, D. Ritter, and G. Eisen-
stein, J. Appl. Phys., 84, 6747 ͑1998͒.
3. A. Pignolet, G. M. Rao, and S. B. Krupanidhi, Thin Solid Films, 258, 230 ͑1995͒.
4. J. Aarik, A. Aidla, A. A. Kiisler, T. Uustare, and V. Sammelselg, Thin Solid Films,
340, 110 ͑1999͒.
Figure 6. XPS analysis for Si 2p peak near the interfacial regions of
Al2O3 /Si for ͑a, top͒ as-deposited at 250°C, ͑b, center͒ after annealing at
800°C for 5 min in Ar, and ͑c, bottom͒ after annealing at 800°C for 5 min in
O2 .
5. E. T. Kim and S. G. Yoon, Thin Solid Films, 227, 7 ͑1993͒.
6. L. Manchanda, W. H. Lee, J. E. Bower, F. H. Baumann, W. L. Brown, C. J. Case,
R. C. Keller, Y. O. Kim, E. J. Laskowski, M. D. Morris, R. L. Opila, P. J. Silver-
man, T. W. Sorsch, and G. R. Weber, Tech. Dig. - Int. Electron Devices Meet., 1998,
605.
7. Y. K. Kim, S. H. Lee, S. J. Choi, H. B. Park, Y. D. Seo, K. H. Chin, D. Kim, J. S.
Lim, W. D. Kim, K. J. Nam, M. H. Cho, K. H. Hwang, Y. S. Kim, S. S. Kim, Y. W.
Park, J. T. Moon, S. I. Lee, M. Y. Lee, Tech. Dig. - Int. Electron Devices Meet.,
2000, 369.
are illustrated in Fig. 6. As shown in Fig. 6a, on the as-deposited
film, which has no interface oxide, there is only one Si 2p peak
emitted from the Si substrate at the interfacial regions of Al2O3 /Si.
However for the film annealed in O2 ambient, two Si 2p peaks are
found at 97.2 and 101.4 eV, which correspond to peaks emitted from
the Si substrate and SiOx , respectively. This indicates that the inter-
face oxide is partly composed of silicon oxides. From the results of
TEM and XPS analyses, it is clear that the growth of the interface
oxide can be prevented by the use of IPA instead of H2O during film
8. M. Ritala and M. Reskela, Nanotechnology, 10, 19 ͑1999͒.
9. T. Suntola, in Handbook of Crystal Growth, 3-part B, D. T. J. Hurle, Editor, p. 605,
Elsevier, Amsterdam ͑1994͒.
10. T. Suntola and J. Antson, U.S. Pat. 4,058,430 ͑1977͒.
¨
11. M. Ritala, H. Saloniemi, M. Leskela, T. Prohaska, G. Friedbacher, and M. Grass-
erbauer, Thin Solid Films, 286, 54 ͑1996͒.
Downloaded on 2015-06-20 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C310
Journal of The Electrochemical Society, 149 ͑6͒ C306-C310 ͑2002͒
12. A. W. Ott, J. W. Klaus, J. M. Johnson, and S. M. George, Thin Solid Films, 292,
135 ͑1997͒.
I. J. R. Baumvol, and G. N. Parsons, Appl. Phys. Lett., 75„25…, 4001 ͑1999͒.
20. E. P. Gusev, M. Copel, E. Cartier, I. J. R. Baumvol, and M. A. Gribelyuk, Appl.
Phys. Lett., 76„2…, 176 ͑2000͒.
13. L. Hiltunen, H. Kattelus, M. Leskela, M. Makela, L. Niinisto, E. Nykanen, P.
Soininen, and M. Tiitta, Mater. Chem. Phys., 28, 379 ͑1991͒.
14. H. Kattelus, M. Ylilammi, J. Saarilahti, J. Antson, and S. Lindfors, Thin Solid
Films, 225, 296 ͑1993͒.
21. M. Fujinami and N. B. Chilton, Appl. Phys. Lett., 63„25…, 3458 ͑1997͒.
22. J. W. Lim, J. S. Park, and S. W. Kang, J. Appl. Phys., 87, 4632 ͑2000͒.
23. J. W. Lim, H. S. Park, and S. W. Kang, J. Appl. Phys., 88, 6327 ͑2000͒.
¨
15. K. Kukli, J. Ihanus, M. Ritala, and M. Leskela, J. Electrochem. Soc., 144, 300
¨
24. K. Kukli, M. Ritala, and M. Leskela, J. Vac. Sci. Technol. A, 15, 2214 ͑1997͒.
͑1997͒.
25. A. C. Jones, D. J. Houlton, S. A. Rushworth, and G. W. Critchlow, J. Phys. IV, 5,
C5 ͑1995͒.
16. G. S. Higashi and C. G. Fleming, Appl. Phys. Lett., 55„19…, 1963 ͑1989͒.
17. D. A. Chang, P. Lin, and T. Y. Tseng, J. Appl. Phys., 78, 7103 ͑1995͒.
18. T. W. Kim, W. N. Kang, Y. S. Yoon, S. S. Yom, J. Y. Lee, C. Y. Kim, H. Lim, and
H. L. Park, J. Appl. Phys., 74, 760 ͑1993͒.
26. M. Ishida, I. Katakabe, T. Nakamura, and N. Ohtake, Appl. Phys. Lett., 52„16…,
1326 ͑1988͒.
19. T. M. Klein, D. Niu, W. S. Epling, W. Li, D. M. Maher, C. C. Hobbs, R. I. Hegde,
27. K. Sawada, M. Ishida, and T. Nakamura, Appl. Phys. Lett., 52„20…, 1672 ͑1988͒.
Downloaded on 2015-06-20 to IP 128.192.114.19 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).