Nanogratings containing sub-10-nm wide trenches by dimension reduction from sloped polymer profile

Abstract

Large area nanograting patterns are useful in many applications but difficult to fabricate. The authors demonstrate a low temperature dimension reduction method, as a cost-effective alternative to high resolution lithography, to define nanogratings as narrow as 8–10 nm . In this process, the slope of prepatterned polymer gratings, with pitch of 200 nm or larger and width of 100 nm or larger, is contrillably changed from the original straight to curved or sloped. Then, shadow metal evaporation is used to coat the sloped polymer profile to define a much narrower opening. This opening is then transferred to underlying material by plasma etching to form sub- 10- nm trenches. The width of trenches can be well controlled by both slope of the profile and angle of metal evaporation. Low processing temperature (as low as 55–85 ° C —depending on polymer) allows this method to be used with a wide variety of materials.

Full text

Page 1

Nanogratings containing sub-10-nm wide trenches by dimension reduction
from sloped polymer profile
Krutarth Trivedi and Walter Hua?
Department of Electrical Engineering, University of Texas at Dallas, Richardson, Texas 75080
?Received 1 July 2009; accepted 26 October 2009; published 3 December 2009?
Large area nanograting patterns are useful in many applications but difficult to fabricate. The
authors demonstrate a low temperature dimension reduction method, as a cost-effective alternative
to high resolution lithography, to define nanogratings as narrow as 8–10 nm. In this process, the
slope of prepatterned polymer gratings, with pitch of 200 nm or larger and width of 100 nm or
larger, is contrillably changed from the original straight to curved or sloped. Then, shadow metal
evaporation is used to coat the sloped polymer profile to define a much narrower opening. This
opening is then transferred to underlying material by plasma etching to form sub-10-nm trenches.
The width of trenches can be well controlled by both slope of the profile and angle of metal
evaporation. Low processing temperature ?as low as 55–85 °C—depending on polymer? allows this
method to be used with a wide variety of materials. © 2009 American Vacuum Society.
?DOI: 10.1116/1.3264683?
I. INTRODUCTION
Nanograting patterns are the building blocks of many de-
vices used in a variety of applications, e.g., nanofluidics,1
optoelectronics,2
nanowire sensors,3
engineering,5etc. Nanogratings are typically defined by
scanning beam lithography or advanced photolithography,6,7
followed by an etching or lift-off process to transfer patterns
from resist to substrate. As dimensions are controlled by
dose in electron-beam ?e-beam? and photolithography, it is
challenging to produce ultrasmall nanoscale patterns over
large areas consistently. More importantly, the sequential na-
ture of e-beam lithography or the high cost of advanced pho-
tolithography makes writing large areas of continuous nan-
ogratings prohibitively time consuming and/or expensive.
Nanoimprint lithography is often used as a cost-effective
process to produce nanostructures over large areas.8How-
ever, fabrication of imprint molds, particularly for ultra-
fine features, still demands conventional high resolution
lithography.
Reducing the size of larger patterns is one low-cost strat-
egy to define ultrasmall patterns without the use of high res-
olution lithography. One widely used scheme to reduce di-
mensions of silicon gratings is oxidation. Several novel, yet
unconventional methods such as edge lithography,9size re-
duction lithography,10and nanopattern oxidation11utilize
various forms of this technique to reduce dimensions during
or after the pattern transfer step. However, these techniques
are limited by the high temperatures required for silicon oxi-
dation, surface and edge variations, and oxidation nonunifor-
mities resulting in high line edge roughness ?LER?.11,12
Moreover, oxidation based techniques are not applicable to
materials other than silicon, e.g., III-V compound semicon-
ductors or even organic materials. Recently, a set of tech-
niques called self-perfection by liquefaction, or SPEL, has
solar cells,4
tissue
shown promising capability to reduce dimensions by melting
and reflowing nanoimprinted polymer structures.13,14Poly-
mer reflow may cause difficulty in controlling uniformity
over large areas for dimensions down to the sub-10-nm re-
gime. Another method to reduce grating dimension is
SAFIER, a temperature dependent ?as high as 160 °C?
shrinking process in which several coat and bake cycles with
a specialized coating material produce linewidths of as low
as ?20 nm.15Shadowed or oblique angle metal evaporation
has long been used to fabricate small gaps in materials.16,17
Its reliability largely depends on the geometry of structures
onto which the metal is evaporated.
In this work, we develop a simple dimension reduction
method to fabricate gratings with trenches as narrow as
sub-10-nm. Dimension reduction is achieved by controlling
the shape and slope of polymer grating sidewalls, facilitating
smooth and directed selective metal coating by oblique angle
evaporation, which effectively reduces pattern dimension.
Plasma etching is used to transfer the reduced gratings to
substrate materials. This low temperature process can be ap-
plied to a wide variety of materials without the need for high
resolution lithography.
II. PROCESS AND RESULTS
Initial grating patterns are defined in resist on top of a
substrate, either by photolithography or nanoimprint lithog-
raphy. Resist is chosen based on patterning method, as cer-
tain resists prefer either nanoimprint or photolithography.
Patterns in each resist can be calibrated to yield the desired
profile shapes; in our case, we have demonstrated control of
profile shape with poly?methyl methacrylate? ?PMMA? and
SU-8 for gratings defined by nanoimprint lithography as well
as S1813 for photolithographically defined gratings. As our
dimension reduction method is applicable to all substrates,
the choice of substrate is inconsequential; in our case, we use
oxidized silicon, as SiO2has sufficiently high selectivity to
silicon during plasma etching.
a?Electronic mail: walter.hu@utdallas.edu
28542854J. Vac. Sci. Technol. B 27„6…, Nov/Dec 2009 1071-1023/2009/27„6…/2854/4/$25.00©2009 American Vacuum Society

Page 2

The process flow is shown in Fig. 1. Polymer gratings of
larger size ?100 nm or larger? are first patterned and their
profile is controllably altered to have curvature or slope.
Metal is selectively coated onto sloped sidewalls of grating
patterns via shadow evaporation, producing small openings
in between the metal. Next, the patterns with reduced open-
ings are transferred by plasma etching into the underlying
substrate. In the case of nanoimprinted gratings, PMMA and
SU-8 were imprinted with a 100 nm line and space grating
mold ?Nanonex?. The mold was treated with perfluorodecyl-
trichlorosilane to prevent adhesion of polymer to the mold
during demolding. PMMA was imprinted at 175 °C for
15 min with a pressure of 60 bars. SU-8 was imprinted at
85 °C for 15 min with a pressure of 30 bars. For microscale
gratings, photolithography was used to make ?1 ?m line
and space grating in S1813 ?Shipley?.
Asloped profile of polymer lines is crucial, as it facilitates
uniform shadowed metal coating on the patterned polymer.
Without a sloped profile, the metal coating is not directed by
the underlying polymer, resulting in loss of control over final
trench dimension and significant LER in transferred patterns.
In the case of nanoimprinted patterns, a sloped and almost
sinusoidal profile was achieved by controlled melting or re-
laxation of the imprinted polymer grating directly after
demolding. The slight melting or relaxation of polymer grat-
ing would likely reduce the surface roughness, as indicated
by SPEL.13,14As shown in Figs. 2?a? and 2?b?, both the tem-
perature and time can significantly affect the shape of the
patterned polymer gratings, with higher temperature and
longer time resulting in decreasing sidewall slope. To mini-
mize the time, temperature was calibrated so that the profile
shape would change significantly in a matter of seconds or
minutes, as shown in Fig. 3?a?. PMMA ?Tg?100 °C? re-
quired a significantly higher temperature of 125 °C com-
pared to 55 °C for SU-8 ?Tg?55 °C? to produce similarly
sloped sidewalls; this difference could be attributed to differ-
ences in molecular weight, as the molecular weight of
PMMA used in this work is 950 000 Da compared to
?7000 Da of SU-8.18It is certainly possible that other com-
binations of time and temperature would produce the same
results, e.g., lower temperature and longer time. In the case
of SU-8, after initial heating at 55 °C to produce sloped
sidewalls, the patterns were cured under UV exposure
?450 mJ/cm2? and underwent a postexposure bake on a hot-
plate at 95 °C for 1 min. Cross-linking the SU-8 under these
conditions does not alter the profile shape further. Different
polymers will require different conditions to produce the de-
sired sidewall profile, as evidenced in the case of PMMA and
SU-8 in Figs. 2?a? and 2?b?. Such tunability is desirable as
certain applications may have low temperature requirements.
In the case of photolithography patterns, the dose is con-
trolled such that after photolithography and development, the
pattern profile is sloped, similar to Fig. 1?b?. For S1813,
underexposure at a dose of 50 mJ/cm2produces sloped side-
FIG. 1. Process flow for dimension reduction. ?a? Spin coated polymer is
imprinted with a mold containing a 100 nm line and space grating pattern.
?b? The pattern is heated to produce sloped sidewalls. ?c? Cr is evaporated at
an oblique angle, leaving small gaps in-between. ?d? The exposed polymer
residue is etched with oxygen plasma, followed by oxide etch in fluorine
plasma. ?e? The remaining Cr and polymer are removed by wet etching. ?f?
The patterns may optionally be transferred to Si and the SiO2mask may be
removed by wet etching.
FIG. 2. Cross sectional SEM images of ?a? a nanoimprinted PMMA grating
?i? before heating, ?ii? after heating at 125 °C for 30 s, ?iii? 1 min, and ?iv?
2 min. ?b?Ananoimprinted and cross-linked SU-8 grating ?i? before heating,
?ii? after heating at 55 °C for 15 s, ?iii? 1 min, and ?iv? 2 min. ?c? A S1813
grating defined by photolithography, exposed at a dose of 50 mJ/cm2.
FIG. 3. ?a? Slope angle of nanoimprinted grating sidewalls as a function of
heating time for both PMMA and SU-8; inset in the top right shows a cross
sectional SEM image of a sloped SU-8 grating. ?b? Trench opening size as a
function of metal evaporation angle for a typical sloped nanoimprinted grat-
ing; schematic inset in the bottom left illustrates the angle of metal evapo-
ration ??? while the image inset in the top right shows a cross sectional
SEM image of metal coating on the sloped grating.
2855
K. Trivedi and W. Hu: Nanogratings containing sub-10-nm wide trenches by dimension reduction2855
JVST B - Microelectronics and Nanometer Structures

Page 3

walls, as shown in Fig. 2?c?. These processes to control pat-
terned polymer shape, both with imprinted ?SU-8 and
PMMA? and photolithographically defined ?S1813? gratings,
have shown the polymer shape profile to be well controlled
at low temperatures ?as low as 55–85 °C for SU-8?, which is
desirable for a wide variety of materials, including plastic
substrates.
After a sloped sidewall profile is produced, the next step
is selective metal coating by shadow or oblique angle metal
evaporation. The metal is selectively coated onto both sides
of sloped polymer grating sidewalls to produce small open-
ings in between the metal, the size of which is determined by
the angle at which the metal is coated. The approximate size
of openings can be predicted by trigonometry, with higher
angles producing smaller openings. It is possible that shadow
evaporation over large areas could introduce global critical
dimension variations. The uniformity in size of openings
across large areas depends not only on the sample size but
also on the distance from metal source and configuration of
the metal source. For especially large samples, it may be
beneficial to use slightly deeper sloped grating patterns with
steeper slope so that higher angle shadow evaporation can be
used to create small gaps. Figure 3?b? shows the trench open-
ing width with respect to the angle of metal evaporation for
a typical nanoimprinted grating with sloped sidewalls, indi-
cating a linear relationship, which affords a high degree of
control in the final trench dimension ?down to 8–10 nm?.
Electron-beam deposition was used to evaporate the metal,
as a directional physical deposition method is required for
controlled shadowing. The quality of the metal coating can
play a role in final LER after pattern transfer. For uniform
metal coating with minimal roughness, a low chamber pres-
sure ??10−6Torr? and a low deposition rate ??0.5 Å/s?
were used. Cr was used as the etching mask, as it is easily
evaporated by e-beam deposition and provides sufficient se-
lectivity to SiO2in fluorine plasma.
After 15–30 nm Cr evaporation, the samples underwent
plasma etching in inductively coupled plasma ?ICP? for
minimal angular spread of the incoming ions to produce
straight sidewalls. First, the samples were etched in oxygen
plasma ?300 W ICP power, 100 W bias power, 5 mTorr,
25 °C chuck temperature, ?10 s? to remove the polymer
residue exposed in between the metal openings. After the
residue was etched, the pattern was transferred to the oxide
layer by etching in a mixture of C4F8, CHF3, and Ar with
conditions that have been optimized to produce straight side-
walls ?2000 W ICP, 500 W bias, 5 mTorr, 10 °C, ?10 s?. At
this point, the pattern can be further etched into the silicon, if
desired. Figures 4?a?–4?c? show cross sectional scanning
electron microscopy ?SEM? images of dimension reduced
nanotrenches transferred to the oxide layer. The Cr etch rate
in these plasma etching steps is sufficiently low, as evidenced
by the residual Cr that is clearly visible even after pattern
transfer to underlying oxide in Figs. 4?b? and 4?c?. Trench
dimensions of less than 10 nm have been produced using this
method, with reasonable uniformity, reduced from the origi-
nal 100 nm line and space gratings. This demonstrates a ten
times well-controlled size reduction. Figure 4?d? shows a
cross sectional SEM image of dimension reduced trenches in
S1813; the trench dimension has been reduced to ?200 nm
from the original ?1 ?m. With higher angle metal evapora-
tion, it is possible to scale the ?1 ?m gratings to well below
100 nm. The final transferred trench dimension can be tuned
not only by the angle of metal evaporation but also by the
slope of the grating sidewalls, providing two degrees of con-
trol. With such control, it is possible to produce uniform
dimension reduced nanogratings over large areas with preci-
sion, as shown in Figs. 4 and 5. As expected, the lines are
smooth and uniform and the cross sections exhibit excellent
profiles. Any defects or nonuniformities in the dimension
reduced lines are likely a result of original patterning, as seen
in Fig. 6, which shows several defects in a small area of the
original imprinted SU-8 patterns. Variations in smoothness
and height of original imprinted lines, both along a line and
among adjacent lines, will result in variations in final grating
slope and height, which, in turn, will directly affect shad-
owed Cr coating. Therefore, nonuniformity in the original
patterns would lead to LER in final dimension reduced pat-
terns. Small scale LER ?local roughness? largely depends on
the quality of metal coating and subsequent etching during
pattern transfer. Therefore, it is necessary to optimize metal
deposition and etching conditions if smaller LER is required.
FIG. 4. Cross sectional SEM images of dimension reduced gratings with
widths of ?a? ?50 nm, ?b? ?20 nm, ?c? 8–10 nm, and ?d? ?200 nm. ?a?,
?b?, and ?c? are dimension reduced nanoimprinted gratings ?scale bars are
100 nm; materials in ?a?, ?b?, and ?c? are the same? while ?d? is a dimension
reduced microscale grating originally defined by photolithography ?scale bar
is 650 nm?.
FIG. 5. Top view SEM images of dimension reduced gratings with widths of
?a? ?75 nm, ?b? ?50 nm, ?c? 8–10 nm, and ?d? ?200 nm. ?a?, ?b?, and ?c?
are dimension reduced nanoimprinted gratings while ?d? is a dimension
reduced microscale grating originally defined by photolithography.
2856
K. Trivedi and W. Hu: Nanogratings containing sub-10-nm wide trenches by dimension reduction2856
J. Vac. Sci. Technol. B, Vol. 27, No. 6, Nov/Dec 2009

Page 4

One of the limitations of this method is the inability to
reduce the pitch of the original grating, which, in our case,
was defined by a mold fabricated with interference
lithography19?in the case of nanoimprinted gratings? or by
the dimensions of patterns on the mask ?in the case of pho-
tolithographically defined gratings?. As defining ultrafine
gratings ??10 nm? is challenging and/or expensive for pho-
tolithography and interference lithography, combining our
dimension reduction method with these two techniques may
provide a unique means to define ultrasmall nanograting pat-
terns over large areas.
III. CONCLUSION
In this work, we have developed a simple dimension re-
duction process to make nanoscale gratings down to
sub-10-nm width over large areas. Dimension reduction is
achieved by controlling the shape and slope of patterned
polymer grating sidewalls, followed by oblique angle evapo-
ration and plasma etching to define openings with up to ten
times reduction in size. As we have shown, this process is
scalable to both microscale and nanoscale gratings, relaxing
the requirements for high resolution lithography. Low tem-
perature processing allows the use of a wide variety of ma-
terials. The controllability inherent to this process affords a
high degree of precision in dimension and uniformity of size
reduced nanogratings.
ACKNOWLEDGMENTS
This research was supported by the Air Force Office of
Scientific Research through the SPRING program ?Grant No.
FA9550-06-1-0403? and KETI through the international col-
laboration program of COSAR, “Next generation semicon-
ductor technology, equipments and materials.”
1L. J. Guo, X. Cheng, and C. F. Chou, Nano Lett. 4, 69 ?2004?.
2D. L. Brundrett, T. K. Gaylord, and E. N. Glytsis, Appl. Opt. 37, 2534
?1998?.
3E. Stern et al., Nature ?London? 445, 519 ?2007?.
4M. S. Kim, J. S. Kim, J. C. Cho, M. Shtein, L. J. Guo, and J. Kim, Appl.
Phys. Lett. 90, 123113 ?2007?.
5A. S. Crouch, D. Miller, K. J. Luebke, and W. Hu, Biomaterials 30, 1560
?2009?.
6B. D. Gates, Q. B. Xu, M. Stewart, D. Ryan, C. G. Willson, and G. M.
Whitesides, Chem. Rev. ?Washington, D.C.? 105, 1171 ?2005?.
7Y. Chen and A. Pepin, Electrophoresis 22, 187 ?2001?.
8S. Y. Chou, P. R. Krauss, and P. J. Renstrom, J. Vac. Sci. Technol. B 14,
4129 ?1996?.
9Y. Zhao, E. Berenschot, M. de Boer, H. Jansen, N. Tas, J. Huskens, and
M. Elwenspoek, J. Micromech. Microeng. 18, 064013 ?2008?.
10Y. K. Choi, J. Zhu, J. Grunes, J. Bokor, and G. A. Somorjai, J. Phys.
Chem. B 107, 3340 ?2003?.
11Q. F. Xia, K. J. Morton, R. H. Austin, and S. Y. Chou, Nano Lett. 8, 3830
?2008?.
12X. M. Yan, S. Kwon, A. M. Contreras, J. Bokor, and G. A. Somorjai,
Nano Lett. 5, 745 ?2005?.
13S. Y. Chou and Q. F. Xia, Nat. Nanotechnol. 3, 295 ?2008?.
14Y. Wang, X. G. Liang, Y. X. Liang, and S. Y. Chou, Nano Lett. 8, 1986
?2008?.
15X. Yang, H. Gentile, A. Eckert, and S. R. Brankovic, J. Vac. Sci. Technol.
B 22, 3339 ?2004?.
16D. L. Klein, R. Roth, A. K. L. Lim, A. P. Alivisatos, and P. L. McEuen,
Nature ?London? 389, 699 ?1997?.
17G. Philipp, T. Weimann, P. Hinze, M. Burghard, and J. Weis, Microelec-
tron. Eng. 46, 157 ?1999?.
18A. L. Bogdanov, Proc. SPIE 3999, 1225 ?2000?.
19T. A. Savas, M. L. Schattenburg, J. M. Carter, and H. I. Smith, J. Vac. Sci.
Technol. B 14, 4167 ?1996?.
FIG. 6. Tilted ?45°? SEM image of original imprinted SU-8 patterns.
2857
K. Trivedi and W. Hu: Nanogratings containing sub-10-nm wide trenches by dimension reduction 2857
JVST B - Microelectronics and Nanometer Structures