METI/NEDO Projects on Cluster Ion Beam
Process Technology
Isao Yamada(1),(3), Jiro Matsuo(2) and Noriaki Toyoda(3)
(1)
Collaborative Research Center for Cluster Ion Beam Process Technology, Osaka Science and Technology Center
(2)
Ion Beam Engineering Experimental Laboratory, Kyoto University
(3)
Laboratory of Advanced Science and Technology for Industry, Himeji Institute of Technology
Abstract. Since the initial study of gas cluster ion beams (GCIB) was started in the Ion Beam Engineering
Experimental Laboratory of Kyoto University, more than 15 years have passed. Some of the results of that study have
already been applied for industrial use. Unique characteristics of gas cluster ion bombardment have been found to offer
potential for various other industrial applications. The impact of an accelerated cluster ion upon a target surface imparts
very high energy densities into the impact area and produces non-linear effects that are not associated with the impacts
of atomic ions. Among prospective applications for these effects are included shallow ion implantation, high rate
sputtering, surface cleaning and smoothing, and low temperature thin film formation.
fundamental research was conducted on these subjects.
In 2000, 5 years GCIB R&D project aimed at
development of basic industrial technology was started
under a contract from New Energy and Industrial
Technology Development Organization (NEDO). The
project is comprised of three groups (highly functional
semiconductor surface processing, high accuracy
surface processing and high-quality thin film
formation) which are organized into a Collaborative
Research Center.
In 2002, a new GCIB project with special emphasis
on nano-technology applications has been started
under a contract from the Ministry of Economy and
Technology for Industry (METI). This METI R&D
project is currently involves efforts in the following
areas: (1) development of size-selected GCIB
equipment and (2) GCIB processes for very high rate
etching and for no damage etching of magnetic
materials and compound semiconductor materials.
INTRODUCTION
Gas Cluster Ion Beam (GCIB) processing of
materials is based on the use of electrically charged
cluster ions consisting of a few hundreds to a few
thousands of atoms or molecules of gaseous materials.
Individual gas atoms are first condensed into neutral
clusters and subsequently ionized and accelerated.
When an energetic cluster ion impacts upon a surface,
it interacts simultaneously with many target atoms and
deposits high energy density into a very small volume
of the target. The concurrent energetic interactions
between many atoms comprising the cluster and target
atoms result in highly non-linear sputtering and
implantation effects, which are fundamentally different
from those associated with binary collisions induced
by monomer ion impacts.
Research on GCIB processes was started at Ion
Beam Engineering Experimental Laboratory of Kyoto
University in 1988 after intense gas cluster formation
had been confirmed. Figure 1 shows a basic GCIB
configuration. In order to establish new industrial
technology employing GCIB, investigations were first
required for formation cluster beam and of ion-solid
interactions with these beams. For more than 10 years,
Ionization
CHARACTERISTICS OF GCIB
EQUIPMENT AND CLUSTER IONSOLID INTERACTIONS
The study of cluster beam formation was initially
the most important topic. It has been well known that
small numbers of clusters are produced by ejection of
metal vapors from heated Knudsen cells. While a
heated Knudsen cell was a candidate for a cluster
beam source, achievable beam intensities have been
too low for practical experimental use[1]. Becker et al.
first studied cluster formation from gas through
supersonic nozzles for thermonuclear fuel applications
[2]. The supers successful for producing cryogenic
beams containing onic expansion approach had been
Acceleration
Skimmer
Gas
Nozzle
Figure 1. Schematic of GCIB equipment
CP680, Application of Accelerators in Research and Industry: 17th Int'l. Conference, edited by J. L. Duggan and I. L. Morgan
© 2003 American Institute of Physics 0-7354-0149-7/03/$20.00
727
System Performance:
Substrate Size: 75 - 200mm
300mm optional
Smoothing Limit:
< 2Å Ra 1µm sq. field(Beam
scanned)
Uniformity/Reproducibility:
< 2% variation
Ar Beam Current: 200 µA
Max. Beam Energy: 25keV
Figure 2. Image and performance of 25 keV GCIB
UltraSmoother™ systems
very large numbers of clusters. However, it was
thought to be difficult to use it at room temperature for
producing intense cluster beams. Our studies showed
that supersonic nozzles with convergence-divergence
shapes could produce strong cluster beams[3]. Our
new cluster ion source, which employing these
supersonic nozzles operated at room temperature, has
successfully produced intense cluster ion beams. With
this, we were able to start studies of new ion-solid
interactions by cluster ion impacts. It has shown that
GCIB offers unique ion/solid interaction processes and
can open a new field in atomic and molecular ion
beam process technology.
GCIB are now being used for producing materials
with superior properties, for developing new chemical
compounds and for altering, refining and machining of
materials and surfaces. Examples include: low-damage
atomic-scale surface smoothing for metals[4], for
superconductors[5], for diamond films[6], for nonspherical plastic lens molds, and for SiC surfaces of
synchrotron radiation mirrors; shallow implantation
for junction formation[7]; high-rate and low-damage
anisotropic etching for MR sensors[8]; assistedformation of thin multi-layer film for reliable and
durable optical filters[9]. Some of unique processes
and recent results are shown below.
EQUIPMENT DEVELOPMENT
Earliest equipment developed at Kyoto University
for investigations of gas cluster ion interactions with
solid surfaces was operational by 1988. Then several
GCIB systems were constructed at Kyoto University
for fundamental studies and investigation of pocess
applications. During collaboration with Epion Co., as
part of work sponsored by Japan Science and
Technology Corporation (JST), development of
commercial GCIB system was begun in 1995. GCIB
systems designed for research and development now
exist in a number of laboratories in Japan and the U.S.
The first GCIB production system was introduced in
late 1999, and a number of GCIB systems are now
operational in Japan and the U.S.
Figure 2 shows an image and specification of
25keV GCIB Ultra-Smoother™ system for high
throughput automated processing of 300mm wafers.
Because gas flows used for cluster generation are high
(several hundred sccm), large pumps are employed for
the nozzle and ionization chamber. Beam current is a
parameter of concern for GCIB applications for
production system to satisfy adequate throughput.
Cluster ion currents in early equipment were low,
sometimes only a few nA, and the factors responsible
for beam intensity limitations were not well
understood at that time. Improvements of neutral
cluster formations, ionization efficiency and beam
transport of GCIB without increasing gas consumption
and pumping requirements have been successfully
demonstrated in the project.
Theoretical study of supersonic nozzles is being
conducted to understand the dependence of beam
intensity on nozzle configurations for designing an
ideal shape of nozzle so as to generate intense beam
and to facilitate generation of controllable cluster sizes.
For this study, a Direct Simulation Monte Carlo
(DSMC) method, originally introduced by Bird in
1970, has been further developed [10]. The DSMC
method is a probabilistic method employing a number
of simulating “particles” in the model which is many
orders of magnitude less than the real number of atoms
in a gas. A large number of nozzle sizes, apex angles
and shapes, with diffuse and secular atomic reflection
at nozzle walls, have been studied. Strong effects of
the nozzle shape were found on the flow variables:
beam intensity, flow density, and nucleation rate at
constant gas temperature. Figure 3 shows comparison
between a Laval nozzle (a) and a conical nozzle(b).
The lines on these two plots show the streamlines that
reflect the flow intensity and direction. It has shown
that an intense directional beam can be formed in the
conical nozzle.
Cluster ion beam currents of several hundred µA
have been demonstrated. At these levels, many
production applications are economically available.
Other applications will require beam currents as high
Figure 3. Comparison between two nozzle shapes: (a) Laval
nozzle, (b) conical nozzle
728
as 1mA or more. Current development programs are
expected to introduce commercial 1mA GCIB system
within about two years.
2)Sputtering and Smoothing
A characteristic of bombardment by large cluster
ions is lateral sputtering effect. The angular
distribution of sputtered atoms by Ar cluster ion shows
laterally ejected one. This was the first experimental
evidence of “lateral sputtering” effect. Very high
sputtering yields on various materials by cluster ion
have been observed in experiments and in computer
simulation studies.
During the project, smoothing of CVD diamond
films deposited on Si has been studied by Mitsubishi
Materials Co. to use them for X-ray lithography mask
membranes. Diamond films (32mmx32mm in area)
with initial roughness of 70nm have been smoothed to
3nm by 20keV Ar cluster ions at a dose of
3x1017ions/cm2. At GCIB current of 50µA, it took 160
min. With a conventional mechanical polishing, it will
take about 54 hours. Figure 5 shows phase-contrast
microscope images of diamond surfaces before and
after GCIB treatment.
PROCESS DEVELOPMENT
1) Shallow ion implantation
Implantation characteristics of monomer ions,
small cluster ions and large cluster ions are very
different even at the same energy per atom, not only
with respect to implant range, but also in terms of
damage formations. For semiconductor devices,
especially for CMOS fabrication, major emphasis has
been placed on development of very low energy ion
implantation to introduce dopant into ultra-shallow
depths. In the case of B, only a few hundred eV are
required. As has been noted above, low-energy
implantation is an ideal application for ions containing
many B atoms in one ion.
Figure 4 shows cluster size dependence on the
number of displacements in Si and energy deposition
on Si target by a 20keV Ar cluster ion. In the case of
cluster size larger than 10,000, there is no displaced Si
atom although 6keV of energy is deposited. It is also
seen that the largest number of displacements are
caused by clusters with size about 3000.
In 1996, first p-MOSFETs were fabricated by
Fujitsu /Kyoto univ. group to demonstrate B10H14
implantation for shallow source/drain formation[11].
In the following year, 40nm p-MOSFETs were
fabricated[9]. B10H14 ion implantation for p-type
source/drain (S/D) junctions was performed at
acceleration energy of 30keV with a dose of 1x1013
ions/cm2 and was followed by annealing at 1000oC for
10sec. A junction depth of 20 nm was achieved. For
S/D extensions, B10H14 at 2 keV was implanted with a
dose of 1x1012ions/cm2 followed by annealing at
900oC for 10sec. A 7nm ultra-shallow junction without
TED or TD was achieved. Equipment development is
in progress under a program of the Japan Science and
Technology Corporation to realize B10H14 implantor
with beam current of 3mA at 3keV of acceleration
energy.
2) Thin film formation
GCIB can be utilized for thin film deposition at low
temperatures. The dense energy deposition produced
by individual cluster impacts significantly enhances
the chemical reactions even at low temperatures. High
chemical reactivity of GCIB and control over film
microstructure has been demonstrated using O2 cluster
ion assisted deposition conducted by an Adachi new
industry co. and Himeji Inst. of Tech. group. Multilayer structures of Ta2O5 and SiO2 have been made
using O2-GCIB bombardment during Ta2O5 and SiO2
evaporation by e-beam source[9]. Figure 6 shows a
cross-sectional SEM image of a Ta2O5/SiO2 multilayer
and AFM images of the interface. The 2nd, 3rd, 4th
and 7th layers from the bottom were formed with
7keV O2-GCIB assisted deposition. The 5th and 6th
layers were formed without GCIB irradiation.
Structure of O2-GCIB assisted was uniform with flat
interfaces and there were no porous or columnar
structures. In contrast, the layers produced without
GCIB assist shows porous grains and rough interfaces.
By using GCIB assisted deposition, very uniform and
dense films without porosity or columnar structure
were obtained.
2k 1k
200 100
20 10
2
1
20000
18000
16000
14000
12000
10000
8000
6000
total 20keV
Displacements
Energy Deposition
10
100
1000
4000
2000
10000
Energy Deposition to Target [eV]
Number of Displacements
Energy per Atom [eV]
15000
14000
13000
12000
11000
10000
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
0
Cluster Size
(a) Before irradiation
(b) After Ar-GCIb irradiation
Figure 5. Phase-Contrast Microscope images of the CVD
diamond surfaces (a) before and (b) after Ar-GCIB irradiation.
Figure 4. Cluster size dependence on the number of displacements
and the energy deposition on Si by 20keV Ar cluster ion
729
O2-GCIB
Without
O2-GCIB
Ta2O5
SiO2
Ta2O5
SiO2
O2-GCIB
Ta2O5
SiO2
7th
Ra=0.7nm
6th
5th
Top surface : O2 GCIB assist deposition (Ta2O5)
4th
3rd
2nd
SEM cross-sectional view of Ta2O5/SiO2
Ra = 1.5nm
Without O2 GCIB assist (SiO2)
Figure 6. SEM cross-sectional image of Ta2O5/SiO2 multiplayer film and AFM images at the interfaces between layers
The Nomura Plating Co./Himeji Inst. Tech. group
has studied very hard DLC films, which are formed by
Ar-GCIB assisted deposition during the evaporation of
C60. When DLC film was formed with 7keV Ar cluster
ions, hardness of DLC film reached 5,000kg/mm2
(50GPa). Hardness of DLC films formed by other
methods (RF plasma, ion plating and ECR plasma)
was around 2000kg/mm2. Also friction coefficients of
the GCIB-assisted DLC films were approximately 0.1,
which was almost 1/5 lower than those formed by
other methods. The sp2 contents of DLC films were
evaluated by near-edge X-ray absorption fine structure
(NEXAFS) spectra using synchrotron radiation. By
integrating the intensity of carbon K-edge peak, which
corresponding to the transition from C1s orbital to p*
orbital, relative sp2 contents of various DLC films
were estimated. The sp2 contents of DLC films formed
by GCIB assisted deposition were lower than those
formed by other methods. The hardness measured with
a nano-indentation technique was found to have strong
correlation to sp2 contents in DLC films. This result
suggests that Ar GCIB assisted deposition was able to
form hard DLC films with a high fraction of sp3
bonding.
REFERENCES
7. K. Goto, J. Matsuo, Y. Tada, T, Tanaka, Y. Momiyama, T.
Sugii, and I. Yamada, IEDM Tech. Dig., IEEE (1997), p.
471.
8. J.J. Sun, K. Shimazaw, N. Kasahara, K. Sato, T. Kagami,
S. Saruki, S. Araki, and M. Matsuzaki, J. Appl. Phys.
89(2001)6653.
9. K.Shirai, Y.Fujiwara, R. Takahashi, N. Toyoda, S. Matsui,
T. Mitamura, M.Terawawa and I. Yamada, Jpn. Appl.
Phys. 41 (2002) 4291.
10. Z. Insepov, I. Yamada, submitted in Proc. of Int. Conf.
on Comp. Simul. of Rad. Effects in Solids (COSIRES2002), Dresden, Germany, 2002.
11. K. Goto, J. Matsuo, T. Sugii, H. Minakata, I. Yamada,
IEDM Tech Dig. 1996, IEEE (1996) p. 435.
SUMMARY
The present status of the research and development
programs and the characteristics of cluster ion beam
processing have been presented. Potential applications
of GCIB processes in industry have been suggested.
Non-linear and non-equilibrium effects due to
bombardment of large cluster ions are now attracting
attention as new technology in the area of ion beam
processing. GCIB processing is an advanced approach
which will contribute to further progress in this field.
This work is supported by New Energy and
Industrial Technology Development Organization
(NEDO) and by Ministry of Economy and Technology
for Industry (METI).
1. W.L.Brown, M.F.Jarrold, R.L.McEachern, M.Sosnowski,
G.Takaoka, H. Usui and I. Yamada, Nucl. Instr. and
Meth. B, 59/60 (1991) 182.
2. E.W. Becker , ‘On the history of cluster beams’, in F.
Trager and G. zu Putlitz, eds, Proc. Internat. Symp. On
Metal Clusters - 1986, Springer Verlag, Berlin (1986) p1.
3. I. Yamada, Radiation Effects and Defects in Solids, 124
(1992) 69.
4. I. Yamada, Nucl. Instr. and Meth. B, 112 (1996) 242.
5. W. K. Chu, Y. P. Li, J. R. Liu, J. Z. Wu, S. C. Tidrow, N.
Toyoda, J. Matsuo, and I. Yamada, Appl.Phys.Lett., 72
(1998) 246.
6.A.Nishiyama,
M.Adachi,
N.Toyoda,
N.Hagiwara,
J.Matsuo, and I.Yamada, Proc. Application of Accelerators
in Research and Industry ’98, AIP Press, New York (1999),
p. 421.
730