A Deep Sub-Wavelength Process For The
A Deep Sub-Wavelength Process For The
A deep sub-wavelength process for the formation of highly uniform arrays of nanoholes and
nanopillars
This content has been downloaded from IOPscience. Please scroll down to see the full text.
(http://iopscience.iop.org/0957-4484/18/48/485302)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 90.147.26.244
This content was downloaded on 02/09/2015 at 11:18
E-mail: [email protected]
1. Introduction of the holes are coupled, and hence these properties cannot be
independently controlled [18]. Finally, NSL requires spheres
There is an increasing demand for high-throughput and parallel to be formed directly on the substrate surface [14], which is
nanofabrication techniques. In particular, there is a high not possible for many materials.
demand for a low-cost process capable of producing highly Here we present a novel photolithography technique,
uniform arrays of nanopillars and nanoholes, since these utilizing the self-assembled planar array of spheres as optical
patterns have found a wide range of applications in many lenses to generate deep sub-wavelength regular patterns over
devices such as solar cells [1], photodetectors [2], surface large areas on a photoresist. Previous studies showed that
plasmonics [3], photonic crystals [4], memory devices [5], spheres with correct optical index values can focus light
nanofiltration [6], fuel cells [7] and artificial kidneys [8]. into deep sub-wavelength dimensions [19]. We noticed that
Conventional photolithography techniques cannot satisfy the both silica and polystyrene (PS) have optical indices [20]
small dimension requirements in many of these applications, that are close to the optimum value at the wavelengths
due to the light source’s wavelength limit [9]. Novel techniques used by conventional UV lithography. Interestingly, our full
such as x-ray [10], electron-beam [11], focused ion beam [12] three-dimensional finite difference time domain (3D-FDTD)
and nanoimprint [13] lithography can achieve the desired calculations show that the beam waist is a very weak function
resolution, but are either too slow or expensive for fabrication of the sphere diameters and hence an extremely uniform
over large areas. Nanosphere lithography (NSL) [14–17] pattern size can be achieved even for relatively poor sphere
is a promising approach that uses planar ordered arrays of size uniformity. Figure 1 shows our simulation results for
micro/nanospheres as a lithography mask to generate ordered conventional UV lithography i -line (wavelength = 365 nm)
nanoscale arrays on the substrate. However, there are several and silica spheres of sizes from 0.5 to 5 μm. Using a linear fit,
limitations associated with this method. First is that the the relation between the sphere diameter d and FWHM (full
monolayer of spheres can always contain dislocations resulting width at half-maximum) w of the focused light is found to be
in agglomerations of particles after metal evaporation and w = 0.0056d + 160 (nm). This means that the variation of
prevent successful lift-off [5]. Second, the size and spacing the FWHM of the focused light is about 0.6% of the change in
Figure 1. (a) 3D-FDTD simulations of the electrical field profile for silica micro/nanospheres with diameters of D = 0.5, 1, 2 and 5 μm at
λ = 365 nm. (b) The normalized light intensity cross section shows a very small change of the beam FWHM despite a 10-time change in the
sphere diameter.
(a) (b)
(c)
(d)
Figure 2. A schematic illustration of the process for the fabrication of periodic nano-arrays: (a) spin-on deposition of photoresist on a
substrate, (b) a monolayer of silica or PS spheres formed on top of the photoresist, (c) UV light exposure of the photoresist covered by the
monolayer of HCP microspheres. The two-dimensional intensity of the UV light in the photoresist plane, calculated by our three-dimensional
FDTD, is shown on the right, (d) sub-wavelength patterns obtained after removing the spheres’ photoresist development. An AFM image of a
developed photoresist is shown on the right.
the sphere’s diameter. Similar simulation results could also be process steps were performed in a class 100 clean-room. Two
obtained for polystyrene spheres. FWHM is a good measure of types of spheres, silica and PS, were used to form hexagonal
the photoresist exposure, since the etching rate usually changes close packing (HCP) arrays on top of the photoresists. We used
by almost an order of magnitude for a 50% optical intensity 10 wt% aqueous suspension of transparent silica or polystyrene
change around the photoresist threshold dose [21]. spheres of 0.97 μm diameter. Before usage, the aqueous sus-
pension was diluted by DI water down to 0.1 wt% for both
2. Experimental details types of spheres. A higher concentration of the spheres forms
multilayers of spheres and a low concentration would not be
We prepared GaAs substrates with two kinds of commonly enough to form a complete layer. The fabrication process is
used photoresists (PR), AZ 5214-E and Shipley 1805. All schematically illustrated in figure 2. Both photoresists were
2
Nanotechnology 18 (2007) 485302 W Wu et al
solution for a few seconds. This step made the photoresist sur-
face hydrophilic enough to promote formation of a monolayer
of HCP silica or PS spheres directly on top of the photoresist.
To form a good monolayer, we carefully studied the effects
of different concentrations of the spheres, the amount of the
sphere suspensions added, the evaporation rates of water and
different tilt angles. Next, the samples were UV exposed us-
ing a conventional photolithography tool (Quintel Q-4000) at a
lamp power of about 11.4 mW cm−2 and center wavelength of
400 nm for about 0.8 s. The short time was not enough to ex-
pose the photoresist except at the focused areas. Silica spheres
were removed by dipping the sample into hydrofluoric acid for
30 s, while sonication was used to remove PS spheres. Pho-
toresists were developed in AZ-300 MIF developer for about
30 s, rinsed by DI water and dried by nitrogen.
Figure 4. SEM images of uniform two-dimensional HCP lattices of nanoholes with different hole diameter and lattice period.
3
Nanotechnology 18 (2007) 485302 W Wu et al
Figure 5. (a) SEM top image of HCP arrays of gold nanopillars. The inset shows the 2D Fourier transform, demonstrating the excellent
coherence of the pillars’ position and size. (b) High-resolution side view SEM image of the gold nanopillars. The inset shows the nanopillar
size distribution, which is currently limited by our image resolution.
shows the top view of the AZ5214 photoresist pillars, while Acknowledgments
figure 3(f) shows the top view of the Shipley 1805 photoresist
pillars. This work is partially supported by National Science
We could change the sizes of the holes, as well as the Foundation grant no. ECCS-0621887 and Defense Advanced
hole-to-hole distance in the arrays precisely and independently. Research Project Agency grant no. N00014-07-1-0564.
Figure 4 shows an example of such flexibility with uniform
HCP arrays of holes with hole diameters of about 300, 500 References
and 700 nm with lattice periods of about 500, 1000, 2000 and
[1] Chiu W L, Alkaisi M M, Kumaravelu G, Blaikie R J,
4000 nm. The hole diameter could be controlled with varying Reeves R J and Bittar A 2006 Adv. Sci. Technol. 51 115–20
exposure time, and the hole-to-hole distance (lattice period) [2] Rogalski A 2003 J. Appl. Phys. 93 4355–91
was controlled by different spherical diameters. Obviously, [3] Dintinger J, Klein S and Ebbesen T W 2006 Adv. Mater.
an array of holes with a lattice period smaller than the hole 18 1267–70
diameter is not feasible (e.g. 700 nm holes with a 500 nm [4] Masuda H, Ohya M and Nishio K 2000 J. Appl. Phys.
period). 39 1039–41
[5] Weekes S M, Ogrin F Y and Murray W A 2004 Langmuir
Figure 5(a) shows an example of HCP arrays of gold 20 11208–12
nanopillars resulting from the lift-off process. The inset shows [6] Baker R W 2004 Membrane Technology and Applications
a 2D Fourier transform of the SEM image of the nanopillars. 2nd edn (Chichester: Wiley)
The Fourier image shows narrow hexagonal patterns up to [7] Shah K, Shin W C and Besser R S 2003 J. Power Sources
high satellite orders, due to a highly coherent structure. Such 123 172–81
[8] Gadegaard N, Martines E, Riehle M O, Seunarine K and
coherence is particularly important for the photonic crystal
Wilkinson D W 2006 Microelectron. Eng. 83 1577–81
applications. Figure 5(b) shows the magnified side view of the [9] Martin O J F 2003 Microelectron. Eng. 67-8 24–30
gold pillars. The inset shows the histogram size distribution, [10] Silverman J P 1997 J. Vac. Sci. Technol. B 15 2117–24
showing a standard deviation of about 13 nm in diameter. [11] Corbierre M K, Beerens J and Lennox R B 2005 Chem. Mater.
Unfortunately, this value is limited by our image digitization 17 5774–9
resolution with ∼15 nm per pixel. Higher magnification [12] Melngailis J, Mondelli A A, Berry I L and Mohondro R 1998
J. Vac. Sci. Technol. B 16 927–57
images show much smaller standard deviation values, while
[13] Chou S Y, Krauss P R and Renstrom P J 1996 J. Vac. Sci.
the number of pillars in those images is not statistically large Technol. B 14 4129–33
enough. [14] Zhang X Y, Whitney A V, Zhao J, Hicks E M and
Van Duyne R P 2006 J. Nanosci. Nanotechnol. 6 1920–34
[15] Whitney A V, Myers B D and Van Duyne R P 2004 Nano Lett.
4. Summary 4 1507–11
[16] Kosiorek A, Kandulski W, Chudzinski P, Kempa K and
In conclusion, we have demonstrated a novel maskless sub- Giersig M 2004 Nano Lett. 4 1359–63
wavelength photolithography technique with a potentially high [17] Kosiorek A, Kandulski W, Glaczynska H and Giersig M 2005
throughput for the formation of HCP arrays of deep sub- Small 4 439–44
wavelength holes and pillars. This method provides accurate [18] Vossen D L J, Fific D and Blaaderen A V 2005 Nano Lett.
and independent control of the hole’s diameter and the lattice 5 1175–9
[19] Chen Z, Taflove A and Backman V 2004 Opt. Express
period. Using a broadband UV source, we demonstrated 12 1214–20
highly uniform features that are about 0.6 times the exposure [20] Mal X, Lu J, Brock R, Jacobs K, Yang P and Hu X 2003 Phys.
wavelength, while theoretical calculation predicts feature sizes Med. Biol. 48 4165–72
that are about half, for a narrow line UV source down to a [21] See for example, Shipley 1800 series photoresist development
wavelength = 193 nm. curves at http://cmi.epfl.ch/materials/Data S1800.pdf