Sub- 100 nm nanolithography and pattern transfer on compound semiconductor using sol-gel-derived TiO2 resist

Article abstract of DOI:10.1149/1.2883730

Typical methods for sub- 100 nm compound semiconductor patterning are based on indirect pattern transfer using Si O2 or Si N4 as intermediate masks, which inevitably increase the complexity of pattern transfer and cause potential damage to samples. We present an approach of direct pattern transfer using Ti (O Bun)4 sol-gel-derived Ti O2 resist as mask. The optimal dose of Ti O2 resist for E-beam lithography is ～220 mC cm2. The sample InP compound semiconductor etching selectivity to Ti O2 resist is as high as 9:1 with aspect ratio of over 30:1. Various sub- 100 nm scale inductively coupled plasma etching patterns are obtained with a high-quality etching profile. The smallest feature is as small as 20 nm wide with a depth of over 600 nm.

Full text of DOI:10.1149/1.2883730

Journal of The Electrochemical Society, 155 ͑5͒ P57-P60 ͑2008͒
P57
0013-4651/2008/155͑5͒/P57/4/$23.00 © The Electrochemical Society
Sub-100 nm Nanolithography and Pattern Transfer on
Compound Semiconductor Using Sol-Gel-Derived TiO2 Resist
Boyang Liuz and Seng-Tiong Hoz
Department of Electrical Engineering and Computer Science, Northwestern University, Evanston,
Illinois 60208, USA
Typical methods for sub-100 nm compound semiconductor patterning are based on indirect pattern transfer using SiO2 or SiN4 as
intermediate masks, which inevitably increase the complexity of pattern transfer and cause potential damage to samples. We
present an approach of direct pattern transfer using Ti͑OBun͒4 sol-gel-derived TiO2 resist as mask. The optimal dose of TiO2 resist
for E-beam lithography is ϳ220 mC/cm2. The sample InP compound semiconductor etching selectivity to TiO2 resist is as high
as 9:1 with aspect ratio of over 30:1. Various sub-100 nm scale inductively coupled plasma etching patterns are obtained with a
high-quality etching profile. The smallest feature is as small as 20 nm wide with a depth of over 600 nm.
© 2008 The Electrochemical Society. ͓DOI: 10.1149/1.2883730͔ All rights reserved.
Manuscript submitted November 28, 2007; revised manuscript received January 8, 2008. Available electronically March 11, 2008.
Compound semiconductor-based devices have a variety of appli-
cations in communications, photonics, and optoelectronics, such as
transistors, lasers, diodes, photodetectors, etc. Recently, semicon-
ductor devices with a sub-100 nm scale have been drawing more
and more attention due to their high integratability and various
unique applications, such as very-large-scale integration, quantum
disk confinement structure, and photonic bandgap.1-4 For those pur-
poses, various instruments and techniques have been developed to
reduce the size of compound semiconductor devices to such a small
scale.
n-butoxide. They reacted in ethanol solvent with an equal molar
amount. All reactions were carried out inside a glove box ͑Ͻ14%
relative humidity͒ for 3 h at room temperature. The solution was
clear and no precipitation was observed. Then the solution was fil-
tered through a 0.2 ␮m filter to remove large particles and improve
the resist spin-coating quality. The desired thickness was obtained
by varying the spin rate and dilution ratio using 1-petonal. A Wool-
lam variable angle spectroscopic ellipsometer and atomic force mi-
croscope were used for resist thickness measurement.
Due to the vulnerability of the titanium atom in Ti͑OBun͒ to
4
However, currently sub-100 nm patterning of a compound semi-
conductor is limited by the lithography resist. For example, for InP
semiconductor compound etching, the etching selectivity of
poly͑methyl methacrylate͒ ͑PMMA͒ is as low as 2–5.5 In addition,
PMMA could not resist high temperatures over 200°C, which is
required by chlorine-based InP dry etching to remove the by-product
during the etching process. Hence, SiO2 or SiN4 are typically used
as intermediate masks for indirect pattern transfer. First, nanopat-
terns are formed on resist using nanolithography. Next, the patterns
are transferred to SiO2 or SiN4 by dry etching. Finally, after remov-
ing the resist and subsequent pattern transfer to the compound semi-
conductor by etching using intermediate SiO2 or SiN4 mask, this
intermediate mask has to be removed for the subsequent fabrication
process. These additional mask transfer and etching away processes
inevitably increase the complexity of InP submicrometer patterning,
especially for sub-100 nm size patterns. Therefore, it is desirable to
find an easier and effective method of compound semiconductor
pattern transfer. One method is to use the recently reported metal
oxide nanolithography resist,6-8 which is expected to have harder
structures and higher resistance to temperature and etching chemi-
cals.
In this paper, we present a method of sub-100 nm direct pattern
transfer of compound semiconductor using sol-gel-derived spin-
coatable TiO2 resist as mask. Here, we specifically demonstrate the
technique on InP compound semiconductor using inductively
coupled plasma reactive ion etching ͑ICP-RIE͒. This technique can
be readily extended to GaAs, InGaAs, InGaP, InAlAs, AlGaAs,
InGaAsP, and other compound semiconductors with similar etching
chemistry. Using spin-coated TiO2 as mask, we show that 20 nm
direct pattern transfer with high selectivity and high aspect ratio
etching profile can be obtained.
nucleophilic attack through the electronegative butoxyl groups, tita-
nium n-butoxide shows a fast hydrolysis when exposed to moisture.
To protect the titanium atom from the nucleophilic attack, we che-
lated the Ti͑OBun͒ with a ␤-ketoesters chelating agent, BzAc. Fig-
4
ure 1 shows the reaction mechanism of the chelation of Ti͑OBun͒
4
with BzAc. Like most ␤-diketones, BzAc is capable of keto-enol
tautomerism. When Ti͑OBun͒ is present, the enol form BzAc can
4
easily react with it and form a chelated compound, which is more
stable due to the steric hindrance effect. As a result, hydrophilicity
of titanium n-butoxide is significantly reduced.
A Thermo Nicolet Nexus 870 FT-IR spectrometer was used to
study the chemical bonds, atmospheric stability, and thermal prop-
erties of BzAc-stabilized TiO2 sol-gel. Figure 2 shows the Fourier
transform infrared ͑FTIR͒ spectrum of the as-spin-coated ͑at 0 h͒
TiO2 resist. Two primary absorption peaks at 1593 and 1511 cm−1
were attributed to ͑CvO͒ and ͑CvC͒ vibrations of the chelate
v
v
ring formed between the Ti core and BzAc, respectively, which in-
dicates the formation of the expected chelate complex. The TiO2
film obtained was then exposed to the atmosphere ͑25°C, 30% hu-
midity͒ for different time durations and then characterized by FTIR
spectra to study the air stability. The absorption peaks corresponding
to ͑CvC͒ and ͑CvC͒ were almost unchanged even after 48 h,
v
v
which indicated a high environmental stability of the chelated TiO2
sol-gel and showed that the chelate ring of Ti and BzAc can effec-
tively stabilize the TiO2 sol-gel-derived resist.
The influence of prelithography baking temperature on TiO2 re-
Experimental
Resist making and characterization
.— Titanium n-butoxide was
used to make sol-gel-derived spin-coatable TiO2 resist, which could
be used in both photolithography and E-beam lithography applica-
tions. Benzonyl acetone ͑BzAc͒ was used to stabilize titanium
z E-mail: BoyangLiu2008@u.northwestern.edu; sth@northwestern.edu
Figure 1. Reaction mechanism of BzAc with Ti͑OBun͒4.
Downloaded on 2015-06-26 to IP 137.99.31.134 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

P58
Journal of The Electrochemical Society, 155 ͑5͒ P57-P60 ͑2008͒
Figure 2. FTIR absorption spectra of spin-coated TiO2 resist after different
lengths of time.
Figure 4. Developed TiO2 E-beam lithography pattern with dose of
219 mC/cm2; feature size is 20 nm.
sist was studied using AXIC JetFirst 150 rapid temperature process-
ing ͑RTP͒ system and FTIR spectra measurements. The baking tem-
perature varied from 100 to 700°C with increments of 100°C for
5 min, respectively. As shown in Fig. 3, the FTIR spectra indicate
that when the baking temperature goes up to 300°C, the absorption
peaks corresponding to vibrations of the chelate ring have almost no
change, which indicates a good thermal stability of the TiO2 resist.
This good thermal stability gives the derived TiO2 resist more ap-
plications than conventional lithography resists in semiconductor
device fabrications. When the temperature goes above 400°C, the
absorption peaks corresponding to vibrations of the chelate ring start
to decrease and eventually disappear above 500°C, which means the
chelate rings are all broken down. When the temperature goes above
600°C, pure TiO2 metal oxide is formed.
thickness of TiO2 resist was ϳ300 nm. The E-beam writing dose
ranged from 40 to 900 mC/cm2 to study the exposure characteristics
of TiO2 resist. The dose of ϳ220 mC/cm2 showed an overall good
lithography quality. Finally, a dose range from 100 to 500 mC/cm2
was used for this study. As shown in Fig. 4, the TiO2 lithography
pattern with the exposure doses of 219 mC/cm2 showed a good
profile quality and development properties.
ICP-RIE etching.— To study the thermal characteristics and per-
formance of TiO2 resist patterns as an etching mask, the patterned
sample InP compounds were transferred to the AXIC JetFirst 150
RTP system again for postlithography annealing. The postannealing
temperature varied from 200 to 500°C with an increment of 100°C.
Typically, two chemistries were used for InP compound pattern-
ing by dry etching, chlorine-based etching, and methane-based etch-
ing. The former provided a high etching speed over 1 ␮m/min. The
latter provided a lower etching rate of 50–100 nm/min with polymer
by-products, which is good for shallow etching with good depth
control. Various instruments are generally used for dry etching, in-
cluding RIE,9 chemically assisted ion beam etching,10 electron cy-
clotron resonance-reactive ion beam etching,11 and ICP-RIE.12-15
Among them, ICP-RIE is most intensively studied due to its capa-
bility of obtaining a high etching rate and low physical damage to
etching samples at the same time. Hence, chlorine-based ICP-RIE is
utilized to study the direct pattern transfer using TiO2 as a mask.
In this study, a PlasmTherm SLR 770 ICP-RIE system was used
for TiO2 pattern transfer. The coil power was set at 800 W for a
relatively fast etching speed. The physical etching by etching radi-
cals and undercut of the profile primarily depended on radio fre-
quency ͑rf͒ power. The increased rf power improved the anisotropy,
whereas it decreased the etching selectivity. Therefore, an optimal rf
power of 120 W was used to balance the undercut and etching se-
lectivity. The anisotropy of dry etching greatly depended on ICP
chamber pressure;16,17 thus, 1.7 mTorr was used for chamber pres-
sure to optimize etching anisotropy after calibration tests. The by-
product of chlorine-based InP ICP etching was InClx, which was
generally involatile below 150°C. In the study, we used an elevated
substrate temperature of 250°C to remove etching product, improve
the etching rate, and smooth the surface. In addition to standard
Cl2/Ar gas mixture, H2 was added to the etching chamber to balance
the physical and chemical processes. An optimal Cl2/Ar/H2 ratio
range from 2/3/1.5 to 2/3/2.5 was chosen for etching due to its best
overall etching profile. In this range, almost all Cl2 reacted with H2
in the etching chamber, resulting in the smallest amount of radicals
responsible for chemical etching and, therefore the etching was bal-
Nanolithography
.— The 100 kV high-resolution Leica VB6-HR
E-beam lithography system was used to study the dose characteris-
tics of TiO2 resist at a 5 nA beam current, and sub-100 nm nano-
lithography was performed using the JEOL JBX-9300FS electron
beam lithography system at 100 kV with 2 nA beam current. Pat-
terned resist was developed in acetone and rinsed in isopropyl alco-
hol ͑IPA͒ for 10 s. Zeiss ultrahigh-resolution field-emission scan-
ning electron microscopy ͑SEM͒ and atomic force microscopy
͑AFM͒ were used to study the high-resolution TiO2 patterns.
Before being transferred to E-beam writing, the TiO2 spin-coated
sample wafers were soft-baked at 80°C for 30 min. The measured
Figure 3. FTIR absorption spectra of spin-coated TiO2 resist under different
prelithography baking temperatures.
Downloaded on 2015-06-26 to IP 137.99.31.134 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, 155 ͑5͒ P57-P60 ͑2008͒
P59
Figure 5. InP compound etching control sample under optimal etching con-
ditions by ICP-RIE.
anced by Ar physical etching. Figure 5 shows the control sample for
InP compound semiconductor etching under optimal etching condi-
tions.
TiO2 patterned sample wafers were then etched using the above
optimal conditions. Figure 6 shows the selectivity of sample InP
compound semiconductor to TiO2 spin-coatable resists at different
postlithography annealing temperatures. The selectivity increased
from 2 to ϳ 9 when the annealing temperature was increased from
200 to 300°C, remained at 9 until 400°C, and dropped to 2 at
500°C. The increase in etching selectivity after the postannealing
processes was primarily due to the formation of pure metal oxide.
High-temperature annealing breaks down chelate rings in stabilized
TiO2 resist and causes crystallization of TiO2 resist to pure metal
oxide, which is known to have a much harder structure and higher
resistance to chlorine-based etching chemicals. This process in-
creased the etching selectivity with the increase of TiO2 postanneal-
ing temperature. As a result, the etching selectivity was increased
from 2 to 9 and kept at 9 when the annealing temperature was
increased from 200 to 400°C. The possible reason for the drop in
etching selectivity after 400°C was the thermal decomposition of
bulk InP surfaces at temperatures over ϳ480°C.18,19 Here, the de-
composition temperature shift from 480 to 400°C could be due to
Figure 7. ICP-RIE etching using sol-gel-derived TiO2 resist as a mask. ͑a͒
SEM pictures of 100 nm diameter pillars etched by ICP-RIE. ͑b͒ SEM pic-
tures of 40 nm wide grid etching pattern; etching depth is over 1200 nm. ͑c͒
SEM pictures of grid-shaped etching patterns. The line is as small as 20 nm
wide and over 600 nm deep.
the fast heating rate of 500°C/min in RTP processing, which has up
to 15% overheating when the annealing temperature arrives at the
preset temperature. As a result, the actual temperature experienced
by the TiO2 resist coated InP wafer at the beginning would be tens
of degrees higher than the preset temperature. During the thermal
decomposition of InP surface at high temperatures, many macro-
scopic etch pits are formed, which degraded the subsequent etching
process due to the macromasking effect. In addition, due to P deple-
tion and In accumulation on the InP surface during decomposition,
the chemical etching in ICP etching was affected, which led to a
decrease of etching selectivity as well.
As shown in Fig. 7, we achieved various sub-100 nm dry-etching
patterns of InP compound semiconductor with good profile quality
using spin-coated TiO2 resist as a mask. Figure 7a shows an etching
of 100 nm diameter pillars. Figure 7b shows a 40 nm feature etch-
ing, where the etching depth is over 1200 nm ͑aspect ratio Ͼ 30:1͒.
Figure 6. Selectivity of InP to TiO2 spin-coatable resist at different postan-
nealing temperatures.
Downloaded on 2015-06-26 to IP 137.99.31.134 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

P60
Journal of The Electrochemical Society, 155 ͑5͒ P57-P60 ͑2008͒
A smooth etching sidewall has been achieved. Figure 7c shows a
grid-shape etching pattern, where etching features as small as 20 nm
wide and over 600 nm deep ͑aspect ratio of over 30:1͒ have been
successfully obtained.
National Nanotechnology Infrastructure Network, which is sup-
ported by the NSF ͑grant no. ECS-0335765͒.
Northwestern University assisted in meeting the publication costs of this
article.
Conclusion
References
We have realized a nanolithography using sol-gel-derived TiO2
resist and demonstrated its new applications to compound semicon-
ductor direct nanopatterning using TiO2 as a mask. The optimal dose
of sol-gel-derived TiO2 as E-beam lithography resist is
ϳ220 mC/cm2. A thermal stability study of spin-coated TiO2 shows
good performance as a lithography resist even at 300°C, which will
have wider applications than conventional resists. Postlithography
annealing at different temperatures is performed to study tempera-
ture dependence of patterned TiO2 resist as a dry-etching mask. The
etching selectivity of sample InP compound semiconductor to TiO2
resist is as high as 9. A variety of sub-100 nm dry-etching patterns
have been obtained. SEM pictures show good profile qualities for
those nanometer-scale etching patterns, such as a smooth etching
sidewall, high aspect ratio of etching patterns, and ultrasmall etching
features. The aspect ratio of etching patterns is over 30:1, and the
smallest feature is as small as 20 nm and over 600 nm deep. This
sol-gel-derived TiO2 spin-coatable nanolithography resist with high
etching selectivity and a high-aspect-ratio etching profile provides a
convenient way to directly pattern compound semiconductor mate-
rial for various challenging nanometer-scale photonic, electronic,
and optoelectronic applications.
1. J. K. W. Yang, V. Anant, and K. K. Berggren, J. Vac. Sci. Technol. B, 24, 3157
͑2006͒.
2. G. T. Edwards, A. Sobiesierski, D. I. Westwood, and P. M. Smowton, Semicond.
Sci. Technol., 22, 1010 ͑2007͒.
3. J. P. Zhang, D. Y. Chu, S. L. Wu, W. G. Bi, R. C. Tiberio, R. M. Joseph, A. Taflove,
C. W. Tu, and S. T. Ho, IEEE Photonics Technol. Lett., 8, 1041 ͑1996͒.
4. H. Cao, J. Y. Xu, W. H. Xiang, Y. Ma, S.-H. Chang, S. T. Ho, and G. S. Solomon,
Appl. Phys. Lett., 76, 3519 ͑2000͒.
5. J. Daleiden, K. Eisele, R. Keller, G. Vollrath, F. Fiedler, and J. D. Ralston, Opt.
Quantum Electron., 28, 527 ͑1996͒.
6. K. R. V. Subrammanian, M. S. M. Saifullah, E. Tapley, K. D. Joon, M. E. Welland,
and M. Butler, Nanotechnology, 15, 158 ͑2004͒.
7. M. S. M. Saifullah, K. Kurihara, and C. J. Humphreys, J. Vac. Sci. Technol. B, 18,
2737 ͑2000͒.
8. M. S. M. Saifullah, K. R. V. Subramanian, E. Tapley, D.-J. Kang, M. E. Welland,
and M. Butler, Nano Lett., 3, 1583 ͑2003͒.
9. I. Adesida, K. Nummila, E. Andideh, J. Hughes, C. Caneau, R. Bhat, and R.
Holmstrom, J. Vac. Sci. Technol. B, 8, 1357 ͑1990͒.
10. C. Youtsey, I. Adesida, J. B. D. Soole, M. R. Amersfoort, H. P. LeBlanc, N. C.
Andreadakis, A. Rajhel, C. Caneau, M. A. Koza, and R. Bhat, J. Vac. Sci. Technol.
B, 14, 4091 ͑1996͒.
11. T. Suzuki, N. Haneji, K. Tada, Y. Shimogaki and Y. Nakano, Jpn. J. Appl. Phys.,
Part 1, 41, 15 ͑2002͒.
12. S. L. Rommel, J.-H. Jang, W. Lu, G. Cueva, L. Zhou, I. Adesida, G. Pajer, R.
Whaley, A. Lepore, Z. Schellanbarger, et al., J. Vac. Sci. Technol. B, 20, 1327
͑2002͒.
13. J. Etrillard, P. Ossart, G. Patriarche, M. Juhel, J. F. Bresse, and C. Daguet, J. Vac.
Sci. Technol. A, 15, 626 ͑1997͒.
Acknowledgments
14. N. O. V. Plank, M. A. Blauw, E. W. J. M. van der Drift, and R. Cheung, J. Phys. D,
The work was supported by the NSF under award no. ECS-
0501589 and ECCS 0622185, by the NSF MRSEC program under
grant no. DMR-0520513, by the NASA Institute for Nanoelectronics
and Computing under award no. NCC 2-1363, and by the National
Center for Learning and Teaching in Nanoscale Science and Engi-
neering ͑NCLT͒ under the NSF grant no. 0426328. The work was
performed in part at the Cornell NanoScale Facility, a member of the
36, 482 ͑2003͒.
15. D. W. Kim, H. Y. Lee, S. J. Kyoung, H. S. Kim, Y. J. Sung, S. H. Chae, and G. Y.
Yeom, IEEE Trans. Plasma Sci., 32, 1362 ͑2004͒.
16. E. W. Berg and S. W. Pang, J. Electrochem. Soc., 146, 775 ͑1999͒.
17. E. W. Berg and S. W. Pang, J. Vac. Sci. Technol. B, 16, 3359 ͑1998͒.
18. S. N. G. Chu, C. M. Jodlauk, and W. D. Johnston, J. Electrochem. Soc., 130, 2398
͑1983͒.
19. F. Riesz, L. Dobos, and J. Karanyi, J. Vac. Sci. Technol. B, 16, 2672 ͑1998͒.
Downloaded on 2015-06-26 to IP 137.99.31.134 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).