University of Nebraska - Lincoln

DigitalCommons@University of Nebraska - Lincoln

Barry Chin Li Cheung Publications Published Research - Department of Chemistry

March 2006

Fabrication of nanopillars by nanosphere lithography

Chin Li Cheung
University of Nebraska at Lincoln, ccheung2@[Link]

R. J. Nikolic
Center for Micro and Nano Technology, Lawrence Livermore National Laboratory, Livermore, CA

C. E. Reinhardt
Center for Micro and Nano Technology, Lawrence Livermore National Laboratory, Livermore, CA

T. F. Wang
Directorate of Chemistry and Materials Sciences, Lawrence Livermore National Laboratory, Livermore, CA

Follow this and additional works at: [Link]

Part of the Chemistry Commons

Cheung, Chin Li; Nikolic, R. J.; Reinhardt, C. E.; and Wang, T. F., "Fabrication of nanopillars by nanosphere
lithography" (2006). Barry Chin Li Cheung Publications. 4.
[Link]

This Article is brought to you for free and open access by the Published Research - Department of Chemistry at
DigitalCommons@University of Nebraska - Lincoln. It has been accepted for inclusion in Barry Chin Li Cheung
Publications by an authorized administrator of DigitalCommons@University of Nebraska - Lincoln.
Published in Nanotechnology 17:5 (March 14, 2006), pp. 1339–1343; doi:10.1088/0957-4484/17/5/028
Copyright © 2006 IOP Publishing Ltd. Used by permission. [Link]
Submitted November 19, 2005; revised January 8, 2006; published online February 10, 2006.

Fabrication of nanopillars by nanosphere

lithography
C. L. Cheung1,*, R. J. Nikolić2, C. E. Reinhardt2, and T. F. Wang3
1 Department of Chemistry and Center for Materials Research and Analysis, University of Nebraska–Lincoln,
Lincoln, NE 68588, USA
2 Center for Micro and Nano Technology, Lawrence Livermore National Laboratory, 7000 East Avenue,

Livermore, CA 94550, USA

3 Directorate of Chemistry and Materials Sciences, Lawrence Livermore National Laboratory, 7000 East Avenue,

Livermore, CA 94550, USA

* Correspondence: C. L. Cheung, email: ccheung2@[Link]

Abstract
A low cost nanosphere lithography method for patterning and generation of semiconductor
nanostructures provides a potential alternative to the conventional top-down fabrication techniques.
Forests of silicon pillars of sub-500 nm diameter and with an aspect ratio up to 10 were fabricated
using a combination of the nanosphere lithography and deep reactive ion etching techniques. The
nanosphere etch mask coated silicon substrates were etched using oxygen plasma and a time-
multiplexed “Bosch” process to produce nanopillars of different length, diameter and separation.
Scanning electron microscopy data indicate that the silicon etch rates with the nanoscale etch masks
decrease linearly with increasing aspect ratio of the resulting etch structures.

1. Introduction [12]to the investigation of hydrophobicity and hydrophilicity

on nanostructured surfaces [13].
Device physics is known to change when structures of ma- Here we present the application of a low-cost NSL
terials are engineered at the nanoscale [1, 2]. Nanospheres method to generate high aspect ratio silicon nanopillars by
made of silica or organic polymer have been applied to gener- high density plasma reactive ion etching. The size and sepa-
ate various functional nanostructures, from photonic band gap ration of these nanopillars were controlled by using polysty-
materials [3]and surface plasmon sensor arrays [4]to nanome- rene nanospheres of different sizes and etching these spheres
ter-sized magnetic domains with orientation dependent mag- with oxygen plasma to tailor the diameter of the nanosphere
netic anisotropy [5]. Nanosphere lithography (NSL) is an eco- “lithography mask.” Since these “masks” are densely packed
nomical technique to generate single-layer hexagonally close with separation at the nanoscale, their packing density drasti-
packed or similar patterns of nanoscale features. Generally, cally affects the flow of reactive ion etchants and effluent ion
NSL applies planar ordered arrays of nanometer-­sized latex or species during the etching process. Thus, both the separation
silica spheres as lithography masks to fabricate nanoparticle of etched “resists” and the etch time control the etch rates of
arrays. NSL has been reported to generate nanowell arrays by silicon. An investigation of the as-fabricated structures with
conventional reactive ion etching [6, 7]. Though NSL and re- scanning electron microscopy (SEM) indicates that the etch
active ion etching has been recently used to pattern nanome- rates of the masked silicon substrates decrease linearly with
ter-sized metal masks to produce columnar and spike struc- increasing aspect ratio of the resulting etched structures.
tures [8–10], high aspect ratio pillar features with various
separation spacing have not yet been demonstrated using this 2. Experiment
technique. Nanopillar patterned surfaces are usually made by
deep ultraviolet lithography, electron beam lithography or na- Our strategy to fabricate the nanopillars consists of three ma-
noimprinting. These structures have been exploited for appli- jor steps: spin coating of polystyrene beads on silicon, reactive
cations ranging from microfluidic chips for separation of long ion etching (RIE) with oxygen to tailor the size of the polysty-
DNA molecules [11]and adhesion substrates for cell growth rene beads, and deep RIE of silicon to etch the pillar structures

1339
1340   Cheung, Nikolić, Reinhardt, & Wang in N a n o t e c h n o l o g y 17 (2006)

ters from 250 to 350 nm. After the NSL in step 1, SEM data
showed that the spin-coated beads were packed mostly as
monolayers of hexagonally closed packed (HCP) domains
of 5–150 µm in domain size on the substrates. Occasionally,
holes with diameters of 5–20 µm of empty space were found
between the domains. Though dip-drying is one of the more
common techniques for NSL [8], we found that spin-coating
Figure 1. Schematic diagram of nanopillar fabrication by nanosphere li- was a more controllable process, especially for patterning a
thography. Step 1: spin coat a hexagonally close packed monolayer of monolayer of beads on wafer-scale substrates.
polystyrene beads on substrates. Step 2: tailor the size of the beads’ “re- The tailoring of the size of polystyrene bead masks by
sist” by oxygen plasma etching. Step 3: etch the exposed semiconductor oxygen RIE shrunk the diameters of these beads from 500
areas by deep reactive ion etching using the “Bosch” process. to 250 nm linearly with etch time (figure 2, Row A & figure
3). Since the position of each bead remained the same after
(figure 1). For the NSL step, 2 × 2 cm2 pieces of silicon or 2 the oxygen RIE, the separations between each bead mask in-
inch silicon wafers (Prime grade, 2 mm thick, single and dou- creased as the sizes of the beads decreased. Though the etch-
ble sided polished silicon wafers, Recticon Enterprises, Inc., ing process caused a minor increase in surface roughening on
Pottstown, PA) were first cleaned by heating the samples in a the etched beads, the beads in general were etched evenly in
100 ml RCA solution (an aqueous mixture of 30% hydrogen the lateral dimension within the tested time periods of 0–150
peroxide (Mallinckrodt Baker, Inc., Paris, KY), concentrated s. The heights of the beads were also etched gradually dur-
ammonium hydroxide (FisherScientific, Pittsburgh, PA), and ing the RIE process. However, RIE runs with 180 s or longer
DI water with the volume ratio of 1:1:5.) at 70 °C for 30 min, often resulted in near to total removal of the beads, and thus
followed by a thorough rinse with DI water and drying in a they were used not in our experiments.
stream of nitrogen. Then, a polystyrene bead solution pre- The fabrication of silicon nanopillars from the bead
pared by a modified procedure in the literature [14] (50 µl of masks of different sizes and separation led to pillars of diame-
Triton X-100 diluted with methanol by 1:400 (Sigma-Aldrich, ters from 200 to 350 nm and height of about 400 nm–2 µm by
Milwaukee, WI) and 350 µl of polystyrene beads with diam- applying the “Bosch” time-multiplexed silicon etch process.
eter 500 nm (5050A, Duke Scientific, Palo Alto, CA)) was During each of the “Bosch” etch cycles, the exposed silicon
spin-coated on these samples with a commercial spin coater. areas were first etched with SF6 and then passivated with a
The spin-coating program consists of three stages: (i) 400 rpm polymer formed by C4F8. Since the bombardment of the reac-
for 10 s to spread the beads solution evenly; (ii) 800 rpm for tive ions was directional and perpendicular to the surface of
2 min to spin away the excess bead solution; (iii) 1400 rpm the substrate in the deep RIE etcher, the bottoms of the etched
for 10 s to spin off the excess materials from the edges. The wells were preferentially etched during the etch phase of the
surfaces of the resulting substrates are strongly diffracted un- etch cycle. The side walls were etched only minimally due to
der room light to reflect a purple opal color. The diameters of the parallel directionality of the reactive ions to the side walls
the polystyrene bead coated substrates were then tailored by and the protection of deposited fluorinated polymers. Con-
a parallel plate RIE etcher with 200 sccm of oxygen and 8.4 sequently, with appropriate etch masks and etch parameters,
sccm of tetrafluoromethane at a pressure of 200 mTorr and ra- this etch process had been shown to produce silicon structures
dio frequency (RF) power of 100 W with various etch periods with high aspect ratio features of 40 and above [17].
(0, 30, 60, 90, 120 s) (figure 2, row A). For silicon etch periods shorter than 22 cycles, the fab-
Nanopillars were fabricated with the polystyrene bead rication scheme yields short silicon pillars with structures as
coated silicon by applying the “Bosch” time-multiplexed RIE predicted. After five etch cycles, the diameters of the as-fabri-
process [15, 16] in a commercial inductive coupled plasma cated pillars were similar to the size of the five different bead
etcher. Briefly, in the process, alternated cycles of etching in etch masks used (figure 2, row B and row C). The “Bosch”
a flow of SF6 (12 sccm, 12s)and passivation in a flow of C4F8 process was shown to effectively etch the silicon substrate
(85 sccm, 9 s) were used to etch the unprotected areas and to though the gaps between the “bead masks,” which are of the
deposit fluorinated polymer to protect the side walls of the re- order of 20–150 nm. This further demonstrates that the chosen
sulting etched structures. The plasma with these chemical spe- etch process for the fabrication of small-size features is crucial
cies was generated with an RF power of 600 W and platen over the ordinary RIE process which had been used to fabri-
power of 14 W at a pressure of 4.5 mTorr. The temperature cate only low aspect ratio wells and pillars [7, 8]. Though the
of the substrate was kept at 25 °C by cooling with a stream of size of the bead mask shrunk about 5–10% in diameter during
helium. The samples were etched with different numbers of the silicon etch, the diameters of the pillars still matched the
etch cycles and examined by scanning electron microscopy to size of the corresponding masks because the diameter was de-
investigate the relationship between the etch rate and aspect fined during the beginning of the etch process. Moreover, the
ratio of the etched structures. structures and arrangements of ‘bead masks’ remained mostly
intact after the silicon etch.
3. Results and discussion The shapes of the cross sections along the silicon pillars
in general conform to the shapes of the bead masks. Along
The three-step fabrication scheme has been successfully car- the length of a silicon pillar, small but periodic variations of
ried out to yield “forests” of silicon nanopillars of diame- the pillar diameter indicate the etching and passivation cycles
F a b r i c at i o n of nanopillars by nanosphere lithography   1341

Figure 2. Scanning electron micrographs of nanospheres-on-silicon samples after the second and third step of the nanopillar fabrication scheme as
shown in figure 1. Row A: top views of the nanospheres’ “resist” on silicon substrates etched by oxygen plasma treatment of periods of 0–120 s. Row B
and row C: top and side views of samples in row A etched by deep reactive ion etching using the “Bosch” process for five cycles. All scale bars are 750
nm.

Figure 3. A plot of the diameters of oxygen RIE etched polystyrene Figure 4. A plot of etch rates of silicon substrates patterned with poly-
beads on silicon substrates versus the RIE etch time. The error bar is ± 3 styrene bead masks pre-etched with 0–120 s oxygen RIE etch periods
standard deviation of each etched bead diameter measurement. versus the measured aspect ratio of the corresponding etched wells using
the data in figure 2, row C.

during the “Bosch” process. This “scalloping” effect is usu- etchants and as-etched gaseous products is the major etch rate
ally a function of polymer quantity deposited for etch protec- limiting factor to etch these pillars. This is further supported
tion during the passivation period and the strength of the RF by the typical linear re­lationship between the silicon depth
power in the etching period. etch rate and the aspect ratio, which is commonly observed in
The etch rate of the silicon substrate was found to be a the ARDE analysis for micrometer-scale trench etching exper-
strong function of the aspect ratio of the etched well geome- iments limited by chem­ical transport (Figure 4). Since lower
try. In our case, the etched areas are of a complicated and pe- chamber working pres­sure will increase the mean free path of
riodic pattern. To simplify the estimation, we define the aspect the molecular species used and produced in the etch process,
ratio as the depth of the etched well (length of the pillar) di- the etch experiments were carried out at 4.5 mTorr to facil-
vided by the average of the minimum separation between the itate the transport and to increase the directionality of these
bead masks and ignore the effect of the etch mask pattern on species.
the etch rate. Using the data from figure 2, a graph of silicon Longer etch progress experiments were also performed to
substrate etch rates versus the aspect ratio (figure 4)indicates determine the effect of a prolonged etch period on the mor-
that the silicon etch rate drops as the aspect ratio increases. phology of the pillar structures and the durability of the bead
The aspect ratio etch rate dependence (ARDE) phe­ masks. HCP bead masks on silicon prepared by the above
nomenon in deep RIE etching is commonly observed in the procedure were tailored by oxygen plasma to yield average
micrometer and sub-micrometer scales when the aspect ratio etched bead diameters of 250 nm and minimum bead separa-
of the etched wells is above 1. The ARDE had been exten- tions of 250 nm. Both 15 and 22 cycles of the silicon “Bosch”
sively studied for trench etching in the µm range; however, etch process were used to etch these samples to generate pil-
little has been reported for high density nanoscale patterns. lar structures of different aspect ratio (Figure 5). After 15 cy-
Various fac­tors such as Knudsen transport of neutrals, ion cles of etching, the bead masks remained intact and the di-
shadowing and charging effects could contribute to the phe- ameter of the as-etched pillars is similar to those of the bead
nomenon [16]. In our case, since the separation between the masks. However, as the number of etch cycles increases from
bead masks is of the order of 10–150 nm, the transport of 15 to 22, the bead masks started to degrade. The tops of the
1342   Cheung, Nikolić, Reinhardt, & Wang in N a n o t e c h n o l o g y 17 (2006)

other gas phase nanowire synthesis, our process does not re-
quire the patterning of a metal catalyst and thus minimizes the
potential for unintentional doping of the wires.
The challenge to generate a large domain size of packed
nanospheres depends on the control of the phase of colloi-
dal packing in the solution media and the interaction between
the spheres and the substrates. Recently, Jiang et al have re-
ported their progress in generation of nearly single-domain
HCP nanosphere patterns on 4 inch wafers by adjusting the
viscosity of the silica nanosphere solution with appropriate
polymer content [7]. This provides a possible route to extend
Figure 5. Scanning electron micrographs of nanopillars fabricated the limitation of the present procedure to generate nanopillar
from a polystyrene bead patterned silicon surface pre-etched with a patterns with longer range order. Hence, the present demon-
120 seconds oxygen RIE etch period, followed by (a) 15 or (b) 22 cy- strated scheme provides an alternative inexpensive method for
cles of the silicon etch process. nanoscale, self-aligned pillar fabrication.

pillars were also etched 5–10% more due to their better ac-
cessibility to etchants in the process. Nevertheless, nearly uni- Acknowledgments
form silicon pillars of about 5 and 10 aspect ratio can be fab-
ricated using the present process parameters without further The authors are grateful for the support of the Laboratory Di-
optimization (Figure 5). Moreover, as the etch time increases, rected Research and Development grant from the Lawrence
the side walls of the as-etched pillars become smoother with- Livermore National Laboratory (grant # 04-ERD-107). CLC
out the “scallop features,” probably due to a local heating ef- thanks the support of the start up research fund from the Uni-
fect in the etching process. The tapering effect at the bottom versity of Nebraska–Lincoln. This work was performed under
of the etched areas is found to be minimal. The degrading of the auspices of the US Department of Energy by the Univer-
the polystyrene beads during the processing limits the unifor- sity of California, Lawrence Livermore National Laboratory
mity of the as-fabricated pillars. We are currently exploring under Contract No. W-7405-Eng-48.
the use of more robust beads made of cross-linked polymer
and silica and optimized etching conditions to make forest of
References
pillars with higher aspect ratios.
[1] Khann K V 2004 Physics of carrier-transport mechanisms and
4. Conclusion ultra-small scale phenomena for theoretical modelling of
nanometer MOS transistors from diffusive to ballistic re-
In short, a fabrication scheme that combines nanosphere li- gimes of operation Phys. Rep. 398 67–131
thography and reactive ion etching (both standard and deep [2] Yakushiji Y, Ernult F, Imanura H, Mitani S, Takanashi K, Taka-
hashi S, Maekawa S and Fujimori H 2004 Enhanced spin ac-
RIE) was developed to produce hexagonally close packed cumulation and novel magneto transport in nanoparticles
semiconductor nanopillars. The etch rates of these structures Nat. Mater. 4 57–61
were found to decrease with an increasing aspect ratio and [3] Van Blaaderen A 1998 Materials science—Opals in a new light
thus the etch process is chemical transport rate limited. Pil- Science 282 887–8
lars of smaller diameters and of different materials can be po- [4] Ormonde A D, Hicks E C, Castillo J and Van Duyne R P
tentially fabricated using bead masks of smaller sizes with the 2004 Nanosphere lithography: fabrication of large area Ag
nanoparticle arrays by convective self-assembly and their
present scheme. characterization by scanning UV–vis extinction spectroscopy
Since the patterns generated by nanosphere lithography Langmuir 20 6927–31
depend on the geometry of the sphere packing, the reported [5] Albrecht M, Hu G, Guhr I L, Ulbrich T C, Boneberg J, Leiderer
scheme allowed only a single layer of patterned features to be P, and Schatz G 2005 Magnetic multilayers on nanospheres
generated on a substrate at a time. Also, the presented proce- Nat. Mater. 4 203–6
dure yields only hexagonally close packed pillar patterns of [6] Whitney A V, Myers B D, and Van Duyne R P 2004 Sub-100 nm
triangular nanopores fabricated with the reactive ion etching
limited domain size with diameters of tens of microns. How- variant of nanosphere lithography and angle-resolved nano-
ever, applications which do not require perfect hexagonal pat- sphere lithography Nano Lett. 4 1507–11
terns such as adhesion growth substrates for cell growth [12] [7] Jiang P and McFarland M J 2005 Wafer-scale periodic nanohole
and the production of superhydrophobic surfaces [13] can be arrays templated from two-dimensional nonclose-packed col-
envisaged with further chemical functionalization of the pillar loidal crystals J. Am. Chem. Soc. 127 3710–1
surface. Moreover, epitaxially grown semiconductor wafers [8] Kuo C, Shiu J and Chen P 2003 Sub-10 nm fabrication of
with different semiconductor layers can be used directly to large-area periodic nanopillar Mater. Res. Soc. Symp. Proc.
EXS-2 M5.11.1–11.13
fabricate nanopillars with self-aligned and well-defined homo
[9] Rose J and Baugh D 2004 Nanorings, nanopillars and nano-
or hetero junction properties. Such structures can be used for spikes on Si(111) by modified nanosphere lithography: fab-
investigating proof of principle device architecture such as rication and application Mater. Res. Soc. Symp. Proc. 832
the vertical surround-gate field-effect transistors [18]. Unlike F7.14.1–14.11
F a b r i c at i o n of nanopillars by nanosphere lithography   1343

[10] Kuo C-W, Shiu J-Y, Chen P L and Somorjai G A 2003 Fab- [14] Hulteen C and Van Duyne R P 1995 Nanosphere lithography:
rication of size-tunable large-area periodic silicon nanopil- a materials general fabrication process for periodic particle
lar arrays with sub-10-nm resolution J. Phys. Chem. B 107 array surfaces J. Vac. Sci. Technol. A 13 1553–8
9950–3 [15] Blauw M A, Zijlstra T, and van der Drift E 2001 Balancing the
[11] Kaji N, Tezuka Y, Takamura Y, Ueda M, Nishimoto T, Nakani- etching and passivation in time-multiplexed deep dry etching
shi H, Horiike Y and Baba Y 2004 Separation of long DNA of silicon J. Vac. Sci. Technol. B 19 2930–4
molecules by quartz nanopillar chips under a direct current [16] Gottscho R A, Jurgensen C W, and Vitkavage D J 1992
electric field Anal. Chem. 76 15–22
Microscopic uniformity in plasma-etching J. Vac. Sci.
[12] Kim P, Kim D H, Kim B, Choi S K, Lee S H, Khademhosse- Technol. B 10 2133–47
ini A, Langer R, and Suh K Y 2005 Fabrication of nanostruc-
tures of polyethylene glycol for applications to protein ad- [17] Ayazi F and Najafi K 2000 High aspect-ratio polysilicon mi-
sorption and cell adhesion Nanotechnology 16 2420–6 cromachining technology Sensors Actuators 87 46–51
[13] Martines E, Seunarine K, Morgan H, Geadegaard N, [18] Ng H T, Han J, Yamada T, Nguyen P, Chen Y P, and Meyy-
Wilkinson C D W, and Riehle M O 2005 Superhydro- apan M 2004 Single crystal nanowire vertical surround-
phobicity and superhydrophilicity of regular nanopat- gate field-effect transistor Nano Lett. 4 1247–52
terns Nano Lett. 5 2097–103