Sub-10-nm nanolithography with a scanning helium beam
Vadim Sidorkin
a兲
Kavli Institute of Nanoscience, Delft University of Technology, Lorentzweg 1, Delft 2628 CJ,
The Netherlands
Emile van Veldhoven
TNO Science and Industry, Stieltjesweg 1, Delft 2628 CK, The Netherlands
Emile van der Drift, Paul Alkemade, and Huub Salemink
Kavli Institute of Nanoscience, Delft University of Technology, Lorentzweg 1, Delft 2628 CJ,
The Netherlands
Diederik Maas
TNO Science and Industry, Stieltjesweg 1, Delft 2628 CK, The Netherlands
共Received 13 April 2009; accepted 23 June 2009; published 24 July 2009兲
Scanning helium ion beam lithography is presented as a promising pattern definition technique for
dense sub-10-nm structures. The powerful performance in terms of high resolution, high sensitivity,
and a low proximity effect is demonstrated in a hydrogen silsesquioxane resist. © 2009 American
Vacuum Society. 关DOI: 10.1116/1.3182742兴
I. INTRODUCTION
With the ongoing trend toward miniaturization and ever
higher device performance, the need for controlled realiza-
tion of sub-10-nm dimensions is becoming increasingly im-
portant. For several decades, the mainstream lithography of
choice for defining nanostructures has been electron beam
lithography. In a long sequence of developments 共e.g., the
field emission gun, 100 kV acceleration voltage, / 4096 la-
ser interferometry, and 16-bit or higher digital-to-analog con-
verter technology兲, the electron beam technique has reached
the nanometer precision needed for sub-10-nm performance
in resist. Rather than the control and accuracy of the beam
writing process, the limiting factors are turning out to be the
electron-resist interactions in conjunction with the resist de-
velopment. We propose that scanning helium ion beam li-
thography 共SHIBL兲 is the next step. In particular, low prox-
imity effects, in combination with the demonstrated high
sensitivity and resolution, make SHIBL a promising technol-
ogy for nanopatterning in the sub-10-nm regime, although
He ion impact on semiconductor nanodevice structures may
not always be harmless.
1
In the past, focused ion beam lithography 共FIBL兲 has been
a somewhat less important player in the field of nanolithog-
raphy. Most work in the field of FIBL has been performed
using either a scanning beam or an ion projection approach,
which is based on Ga
+
ion sources. Even fewer beam studies
deal with light ions such as Be
+
,H
+
,H
2
+
, and He
+
.
2–4
An
early overview was given by Melngailis.
5
Besides the advan-
tages of higher sensitivity and lower proximity effects
6
com-
pared to the electron beam approach, major disadvantages of
the earlier FIBL work are a lower resolution due to a larger
beam diameter and 共certainly with respect to Ga+ion expo-
sure兲 potential damage or contamination from the ion impact
共Ga implantation
7
兲. High-resolution FIB tools have been
available for about 2 decades. Features including a 12–15 nm
linewidth in poly共methyl methacrylate兲共PMMA兲 resist
8
and
a 30 nm dot size in polyphenylsilsesquioxane resist
9
have
been reported. Today, the smallest probe size of Ga
+
ion
beams is typically around 5 nm, and direct engraving in a 20
nm thick membrane material yields nanopores as small as 3
nm in diameter.
10
Recently, scanning helium ion microscopy
11
with a he-
lium probe size of 0.75 nm in diameter was launched in the
market. In the work described here, we used the scanning
He
+
ion beam setup as a beam writing tool for nanolithogra-
phy and compare its performance with electron beam expo-
sure behavior. The expected advantages of SHIBL over elec-
tron beam lithography 共EBL兲 comprise of reduced proximity
effects and the ability to write smaller features given the
subnanometer probe size. The former is due to a more direc-
tional scattering profile of He
+
ions and the different second-
ary electron 共SE兲 generation mechanisms with ions in resist,
which causes a high yield of slow SEs.
11–14
Altogether, these
characteristics greatly suppress the blurring background that
arises when writing dense patterns with EBL.
II. EXPERIMENT
A Carl Zeiss Orion™ Plus scanning helium ion micro-
scope and an FEI Strata DB 235 scanning electron micro-
scope 共SEM兲, both operating at 30 kV, were used for helium
ion beam and electron beam exposures, respectively. The FEI
setup and a Hitachi S4800 SEM were used for the inspection
of the developed structures. Especially, the latter imaging
setup is superior in contrast and resolution due to its sophis-
ticated system to handle the secondary and backscattered
electrons. A concise discussion of high-resolution scanning
electron and helium ion microscopy was published
recently.
15
Dose variations were achieved by means of the
beam current and by controlling the dwell time per pixel,
which varied from 20 to 2000
s.
a兲
Electronic addresses: v.a.sidorkin@tudelft.nl and sidorkin_vadim@mail.ru
L18 L18J. Vac. Sci. Technol. B 27„4…, Jul/Aug 2009 1071-1023/2009/27„4…/L18/3/$25.00 ©2009 American Vacuum Society
Downloaded 14 May 2013 to 131.180.131.253. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvstb/about/rights_and_permissions
This work deals with exposures in hydrogen silsesquiox-
ane 共HSQ兲 resist from Dow Corning 共FOx-12 product兲 on
silicon substrates. Resist thickness was achieved by tuning
the spin speed 共Karl Suss RC 5/8兲 and by appropriate dilu-
tion of HSQ with methyl isobutyl ketone. Typically, 1:10 and
1:1 dilutions are used for thicknesses of 5 and 55–70 nm,
respectively. All resist films were dried for 30 min in vacuum
at room temperature in order to minimize film roughness.
16
Two types of experiments were conducted. In the first
type, the exposure sensitivity is measured for both EBL and
SHIBL in HSQ films of ⬃70 nm thick using a defocused
beam in a dose range from 0.1
C/ cm
2
to 1 mC/ cm
2
. Pat-
terns used for that purpose were 50 ⫻ 50
m
2
squares. After
development in MF 322 developer 共Rohm & Haas兲 for 1
min, the height of the structures was measured by means of
profilometry 共Tencor兲. Sensitivity values were obtained as
the dose D
s
necessary to achieve 50% of its original thick-
ness after development. Contrast values were determined ac-
cording to the procedure of Thompson.
17
In the second type of experiments, high-resolution He
+
beam spot exposures were performed in 5 and 55 nm thick
layers of HSQ. Exposures were performed with a fine probe
size,a1pAbeam current, and a dwell time of 100
sat
20
m aperture and 7 mm working distance. A single imag-
ing raster scan was used, resulting in the formation of arrays
of dots with a variable pitch. The field of view 共FOV兲 was
25⫻25
m
2
and the numbers of pixels were 256⫻ 256,
512⫻512, and 1024⫻ 1024, resulting in pitches of 98, 47,
and 24 nm, respectively. In one exposure, the FOV was 15
⫻15
m
2
with 1024⫻1024 pixels, resulting in a pitch of
14 nm. A droplet of Au nanoparticles 共⬃20 nm in diameter兲
was placed directly on the resist surface, which allowed the
ion beam to be focused accurately in situ prior to the actual
writing in resist. After development for 5 min 共MF351 from
Rohm & Haas兲 and immersion in a “stopper” solution
共MF351: H
2
O=1:9兲, samples were rinsed with de-ionized
water and blown dry in nitrogen.
III. RESULTS AND DISCUSSION
A. Contrast and sensitivity
Residual HSQ film thickness dependencies after develop-
ment on helium ion beam and e-beam exposure dose are
shown in Fig. 1. The results demonstrate that HSQ resist is
4.4 times more sensitive for helium ions 共D
s
=31⫾3
C/ cm
2
兲 than for electrons 共D
s
=137⫾5
C/ cm
2
兲, whereas the contrast values are practi-
cally the same 共
␥
=2.1⫾0.1兲 for both types of exposure. The
difference in sensitivity between SHIBL and EBL can be
partly due to the higher yield of SE for helium ions com-
pared to electrons of the same energy.
11
Additionally, the
resist sensitivity is further enhanced because in He
+
ion ex-
posure the fraction of low-energy SE is higher. The same
contrast for both EBL and SHIBL indicates similar molecular
weight distributions of cross-linked resist monomers. It is
solely the yield and energy distribution of the SE that differ.
The higher sensitivity of HSQ for He
+
than for electron ex-
posure is relatively moderate compared to the sensitivity en-
hancement by factors of 100–300 in PMMA under Ga
+
,He
+
,
H
+
, and Ar
+
exposure
18
and 16–50-fold enhancement in a
range of other resist products under He
+
ion projection
lithography.
19
Recent work on H
+
beam writing of HSQ at
MeV ion energy exposure shows a sensitivity of
3.2
C/ cm
2
,
20
which drops to about 20
C/ cm
2
sensitivity
upon HSQ resist aging. The difference between the values
from these studies and those obtained in our experiment with
helium ions could be due in part to differences in experimen-
tal conditions. Beam type and energy, initial thickness,
21
de-
velopment time, developer type, concentration, and
temperature
22
have been known to have a substantial impact
on the sensitivity of the resist. However, additional impact
from a different exposure mechanism cannot be ruled out.
B. Minimal feature size and proximity effect
SEM images of an array of isolated dots at a pitch of 98
nm in 5 and 55 nm thick HSQ films are shown in Figs. 2共a兲
and 2共b兲, respectively. The average dot diameters are 6 ⫾ 1
and 14⫾1 nm, respectively. These results prove the He
+
ion
beam capability for nanostructuring in ultrahigh resolution
mode. In both cases the exposure dose 共100
s dwell time
per pixel兲 and the development time 共5 min兲 were the same.
FIG. 1. Normalized thickness of HSQ resist film as a function of dose for
helium ion beam and electron beam exposures at 30 keV. The thickness was
normalized prior to the development. The sensitivity D
s
and contrast values
␥
s
are shown in the inset.
FIG. 2. SEM images of arrays of dots written in 共a兲 5 nm and 共b兲 55 nm
thick HSQ layers at 98 nm pitch using scanning helium ion beam lithogra-
phy. Field of view is 900 nm in SE mode at 20 kV. Average dot diameters:
共a兲 6⫾ 1 nm and 共b兲 14⫾ 1nm.
L19 Sidorkin et al.: Sub-10-nm nanolithography with a scanning helium beam L19
JVSTB-MicroelectronicsandNanometer Structures
Downloaded 14 May 2013 to 131.180.131.253. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvstb/about/rights_and_permissions
Nevertheless, the dot size for the thicker layer is about twice
as large as the thinner one. This observation corresponds
with previous results for electron beam exposure of HSQ
films of different thicknesses. Lines 2 pixels wide written
with an e-beam in 5 and 55 nm thick HSQ have linewidths of
11 and 24.5 nm, respectively,
21
when processed under iden-
tical exposure and development conditions 共but different
from the current He ion experiments兲. To our knowledge, the
best results in e-beam exposure of HSQ are currently 6 nm
lines on a pitch of 20 nm written at 100 keV.
23
The dot size as a function of pitch is shown in Fig. 3.
Remarkably, an identical dot size of 6 ⫾ 1 nm was achieved
down to a pitch of 14 nm 关see Fig. 3共d兲兴. This observation
indicates an undetectably small proximity effect for helium
ion beam exposure. Several factors may be involved in this
low proximity effect. First, the scattering profile of helium
ions is known to be confined within a narrow cone penetrat-
ing relatively deep into the material with very low
backscattering.
11
Second, helium ions produce mostly low-
energy SE 共⬍20 eV, when derived from H
+
data on CO
2
,
scaled to the same velocity
14
兲, which do not travel far in the
resist.
Finally we need to address two other aspects in this work.
First, the exposure dose of 1 pA for 100
s corresponds to
about 600 ions/dot. The inherent shot noise implies a dose
uncertainty of 4%/dot. Although the insets in Figs. 3共a兲–3共c兲
show dot-to-dot size irregularities, a possible relation to shot
noise is not likely, given the contrast curve in Fig. 1. The
observed nanometer scale irregularities touch on the second
aspect related to the minimum achievable feature size in
HSQ resist, where several factors play a role. Irrespective of
the writing probe size, HSQ studies in literature
23
and this
work indicate absolute lower limits of about 5–6 nm for
HSQ structures. Adhesion to the substrate becomes too low
with decreasing feature size in order to withstand the force
interactions during the wet development step. In addition, the
number of cross-linked resist monomers decreases with
shrinking feature size. Consequently, at some minimum fea-
ture size the contrast in the exposed area is too low to “sur-
vive” the development. Future work to test this size limit
hypothesis could deploy special patterns, wherein the ul-
trafine features are attached to larger scale structures that will
provide sufficient mechanical stability. With its negligible
proximity effect, SHIBL seems to be the best exposure tech-
nology by far to realize the ultimate limit in this matter.
IV. CONCLUSION
Scanning He
+
ion beam lithography on HSQ is demon-
strated to have very a high resolution and a superior low
proximity effect. Furthermore, He
+
ion exposure is several
times more effective than electron beam exposure at the
same acceleration voltage, whereas the contrast is equal.
Overall, He
+
ion beam lithography is a very promising tech-
nique for the formation of ultrahigh resolution structures of a
high density and having feature sizes in the sub-10-nm
range.
ACKNOWLEDGMENTS
This research is part of NanoNed, a national research pro-
gram on nanotechnology funded by the Ministry of Eco-
nomic Affairs in The Netherlands. Ping Cheng, Kees Hagen,
and Pieter Kruit are acknowledged for their contributions to
this work.
1
E. van der Drift, R. Cheung, and T. Zijstra, Microelectron. Eng. 32,241
共1996兲.
2
C. W. Slayman, J. L. Bartelt, and C. M. Mckenna, Proc. SPIE 333,168
共1982兲.
3
J. N. Randall, D. C. Flanders, N. P. Economou, J. P. Donnelly, and E. I.
Bromley, J. Vac. Sci. Technol. B 1,1152共1983兲.
4
I. Adesida, C. Anderson, and E. D. Wolf, J. Vac. Sci. Technol. B 1, 1182
共1983兲.
5
J. Melngailis, Nucl. Instrum. Methods Phys. Res. B 80-81,1271共1993兲.
6
S. Matsui, Y. Kojima, Y. Ochiai, and T. Honda, J. Vac. Sci. Technol. B 9,
2622 共1991兲.
7
D. C. Ferranti, J. C. Morgan, W. B. Thompson, and W. C. Joyce, Proc.
SPIE 2194, 394 共1994兲.
8
R. L. Kubena, J. W. Ward, F. P. Stratton, R. J. Joyce, and G. M. Atkinson,
J. Vac. Sci. Technol. B 9, 3079 共1991兲.
9
R. L. Kubena, R. J. Joyce, J. W. Ward, H. L. Garvin, F. P. Stratton, and R.
G. Brault, J. Vac. Sci. Technol. B 6,353共1988兲.
10
J. Gierak et al., Microelectron. Eng. 84,779共2007兲.
11
J. Morgan, J. Notte, R. Hill, and B. Ward, Microscopy Today 14,24
共2006兲.
12
R. Ramachandra, B. Griffin, and D. Joy, Ultramicroscopy 109,748
共2009兲.
13
S. Matsui, Y. Kojima, and Y. Ochiai, Appl. Phys. Lett. 53,868共1988兲.
14
W. Q. Cheng, M. E. Rudd, and Y. Y. Hsu, Phys. Rev. A 40, 3599 共1989兲.
15
A. Vladar, M. Postek, and B. Ming, Microscopy Today 17,6共2009兲.
16
V. Sidorkin, A. Grigorescu, H. Salemink, and E. van der Drift, Microelec-
tron. Eng. 86, 749 共2009兲.
17
L. F. Thompson, Solid State Technol. 17,27共1974兲.
18
H. Ryssel, K. Haberger, and H. Kranz, J. Vac. Sci. Technol. 19, 1358
共1981兲.
19
S. Hirscher, R. Kaesmaier, W.-D. Domke, A. Wolter, H. Loschner, E.
Cekan, C. Horner, M. Zeininger, and J. Ochsenhirt, Microelectron. Eng.
57-58, 517 共2001兲.
20
J. A. van Kan, F. Zhang, C. Zhang, A. A. Bettiol, and F. Watt, Nucl.
Instrum. Methods Phys. Res. B 266, 1676 共2008兲.
21
V. A. Sidorkin, A. van Run, A. van Langen-Suurling, A. Grigorescu, and
E. van der Drift, Microelectron. Eng. 85, 805 共2008兲.
22
J. Yang and K. K. Berggren, J. Vac. Sci. Technol. B 25,2025共2007兲.
23
A. E. Grigorescu, M. C. van der Krogt, C. W. Hagen, and P. Kruit, J. Vac.
Sci. Technol. B 25, 1998 共2007兲.
FIG. 3. SEM images of arrays of dots written ina5nmthick HSQ layer
using SHIBL at pitches of 共a兲 48 nm, 共b兲 24 nm, and 共c兲 14 nm, and 共d兲 the
average dot size vs pitch. The insets are SEM images at a higher magnifi-
cation. The average dot size for all pitches is 6⫾ 1nm.
L20 Sidorkin et al.: Sub-10-nm nanolithography with a scanning helium beam L20
J. Vac. Sci. Technol. B, Vol. 27, No. 4, Jul/Aug 2009
Downloaded 14 May 2013 to 131.180.131.253. Redistribution subject to AVS license or copyright; see http://avspublications.org/jvstb/about/rights_and_permissions
