A Dynamic Negative-Ion Mixing Method
Using High-Current Heavy Ions and Electron Beam Evaporation
N. Kishimoto, J.P. Zhao1 N. Okubo2, Y. Takeda and V.T. Gritsyna3
National Research Institute for Metals, 1-2-1 Sengen, Tsukuba, Ibaraki 305-0047, Japan
Shanghai Institute for Metallurgy, Shanghai 200050, China
University of Tsukuba, Tsukuba, Ibaraki 305-8573, Japan
Kharkov National University, Kharkov 61077, Ukraine
Abstract-A new filmmaking technique with high-current also significant atomic introduction. We designate this
negative ions and electron beam evaporation has been developed method Dynamic Negative Ion Mixing (DNIM)  as a new
to fabricate optical insulating films, applying both ion-induced tool of surface modification or filmmaking.
dynamic processes and simultaneous formation of insulating Advantages of dynamic ion mixing methods, in general,
substrates. The dynamic negative-ion mixing (DNIM) method, are capability of film deposition with radiation-induced
with negative ions at high dose rates, alleviates surface charging diffusion and reaction, and ion implantation over a wide
on insulators and attains mass introduction of metal atoms. range irrespective of the ion penetration depth. However, a
Negative Cu ions of 60 keV irradiated an a-SiO2 substrate, possible concern of the method is performance of substrate
which simultaneously grew by electron beam evaporation of fabrication, particularly of compound materials. A material
SiO2. The dose-rate ranged from 5 to 50 µA/cm2 and the phase made by vacuum evaporation tends to be disordered
deposition rate was maintained constant at 0.2 or 0.4 nm/s and and less dense. It is also unknown how the growing substrate
controlled the Cu composition. The DNIM method succeeded in behaves under interaction with heavy ions. Again, the
fabricating thick uniform films which were finely dispersed with important are effects of high flux on insulating substrate
nanoparticles. The merit of high flux implantation is formation. Silica glass (a-SiO2) is one of the most common
spontaneous precipitation of Cu nanoparticles. The and superb materials for optical substrates, which has
nanoparticle-dispersed films showed intense linear and radiation-resistant transparency and is mechanically strong,
nonlinear optical properties. admitting considerable precipitates . The characteristics of
a-SiO2 are probably associated with the metastable behaviors
 under ion implantation, such as radiation-induced viscous
I. INTRODUCTION flow and compaction [9,10]. For dynamic mixing methods,
Negative ion implantation alleviates surface charging on surface morphology is particularly informative to inspect the
insulators down to several volts  and enables us to conduct matrix status, since a surface is successively incorporated into
accurate and efficient near-surface modification of insulators, the interior matrix.
particularly in the energy range less than about 102 keV . In this paper, we first describe specifications and
Negative metal ions of 1 mA are now available for material performance of the dynamic negative ion mixing and next
technology by virtue of progresses in the negative ion present morphology of interior nanoparticles and the exterior
technology, particularly Cs-assisted plasma-sputter type surface of the films. Merits and possible problems of the
sources . An important implication of the 1 mA-level is method are discussed based on the experimental results.
availability of ion fluxes comparable with atomic fluxes of
gas-phase deposition such as vacuum evaporation (typically II. EXPERIMENTAL
∼0.1 nm/s). We have focused on the merits of high flux and
little surface charging, and have applied negative Cu ions to Film deposition of a-SiO2 by electron-beam (EB)
insulators for novel optical properties of metal nanoparticles evaporation and ion implantation of 60 keV Cu- were
embedded in insulators [3-6]. The negative Cu implantation concurrently conducted onto an a-SiO2 disk substrate, whose
has demonstrated a high efficiency of atomic retention in the size is 15 mm in diameter and 0.5 mm in thickness. Fig.1
substrates . Up to now, we have presented kinetic methods shows a schematic diagram of a dynamic negative-ion mixing
to spatially control nanoparticles for a fixed substrate, i.e., method (upper) and appearances of a specimen deposited to
either two-dimensional (in-plane) or Gaussian-like-dispersed 500 nm thick (lower-left) and a specimen stage (lower-right).
nanostructures . However, the conventional implantation, Negative Cu ions of 60 keV produced by a plasma-sputter-
for a fixed substrate and ion energy, is subjected to limitation type ion source were mass-separated and introduced into the
of the total amount of nanoparticles, which results in vacuum chamber . To minimize interaction between a
limitation of optical signal intensity as a whole. negative-ion beam and molecules of SiO2 evaporated, a 1/4-
The merit of charging-free is further applicable to so-called spherical shield was installed surrounding the EB-furnace.
dynamic ion mixing for insulating substrates combining low- The incident angles of Cu ions and the SiO2-vapor flux were
energy negative-ion implantation and vacuum evaporation. +19 and -15 degrees from the normal to the substrate,
Usage of high-flux heavy ions provides us with attractive respectively. A Cu aperture mask with a 12 mmφ-hole
features different from the conventional mixing methods, i.e., clamped a substrate to the specimen stage via steel springs.
not only ion-induced mixing and/or ion-beam annealing but Dose rate of Cu- ranged from 5 to 50 µA/cm2. The deposition
Fig. 2 Cross-sectional TEM image of a specimen fabricated by
DNIM method. The evaporation rate and dose rate are 0.4
Fig.1 Schematic diagram of a Negative Ion Dynamic Mixing nm/s and 20 µA/cm2, respectively. The nanoparticle layer is
method (upper) and appearances of a specimen of 500 nm 500 nm thick.
thick (lower-left) and a specimen stage (lower-right).
0.4 nm/s and 20 µA/cm2, respectively, and Cu composition
rate of SiO2 was maintained constant at 0.2 or 0.4 nm/s. The corresponds to 10 at%. Nanoparticles of 5 nm in average
film thickness measured by a crystal-quartz monitor was set uniformly distribute throughout the thickness of 500 nm. In
to be 200 or 500 nm. For comparison, Cu ion implantation the conventional implantation without film deposition,
into a-SiO2 substrates without the SiO2 evaporation was nanoparticles were located, more or less, in the vicinity of the
carried out, at ion dose rates of 5- 30 µA/cm2. The substrate projected range (~45 nm) . The present thickness is more
temperature was monitored by using thermocouples attached than ten times as large as that of the projected range. The
to the aperture mask. Concentration of Cu in the films was thickness is virtually unlimited, as far as the film exfoliation
controlled by a ratio of the Cu dose rate and the SiO2 does not occur. The distributions of particle size and number
deposition rate, aiming at about 10 at%. Hereafter, a-SiO2 density are uniform along the thickness, as far as ion dose
films fabricated by the DNIM method are referred as Cu:SiO2 rate and deposition rate are stably maintained. The average
films. particle size of 5 nm in this case is smaller than that in the
After the DNIM processing, optical performance of the conventional implantation at the same dose rate (∼10 nm).
Cu:SiO2 films was evaluated: Absorbance was measured in a The smaller size in the DNIM process may be ascribed to
photon energy range from 0.5 to 6.5 eV, by a dual beam absence of energy accumulation at a given depth, or to
spectrometer. The third-order optical susceptibility, χ(3), of absorption of deposited energy by phase transformation of a
the films was measured by the DFWM (degenerate four wave deposited SiO2 film. The particle size in the DNIM
mixing) method using a wavelength of 532 nm at the plasmon processing is controllable, to some extent, by changing dose
resonance. Nonlinear time response of the films was detected rate. The average particle size gradually increases with
by a pump-probe method using a Ti: sapphire femto-sec laser increasing dose rate, under a fixed deposition rate.
system. Atomic force microscopy (AFM) in the tapping mode Another feature of the DNIM films is presence of depleted
and cross-sectional TEM were conducted to evaluate surface zones in depth around the front and the back ends. Since the
morphology and microstructure of the films, respectively. ion deposition profile is alike a Gaussian function, there
inevitably occur concentration-gradient regions of Cu
III. RESULTS AND DISCUSSION implants in the vicinity of the both edges. The widths are
determined by the ion profile and the matrix-deposition rate.
A. Nanoparticle and Surface Morphology of the DNIM The presence of the gradient regions may become either a
Process merit (buried nanoparticle structure, good adhesion to the
The dynamic negative-ion mixing method has succeeded in substrate) or a demerit (partial loss of uniformity).
fabricating thick and uniform films, dispersed with Cu Fig.3 shows dose-rate dependence of surface roughness of
nanocrystals. As seen in Fig. 1 (lower-left), the DNIM- Cu:SiO2 films that was taken from AFM images. Comparison
deposited area is macroscopically uniform colored with thick is made between the DNIM method (circle) and ion
brown and there were no signs to be exfoliated. Fig. 2 shows implantation alone (square). One of the most characteristic
a cross-sectional TEM image of a specimen fabricated with features of the DNIM method is a surface smoothing effect
the DNIM method, where evaporation rate and dose rate are on SiO2. The EB-evaporated SiO2 surface is considerably
Fig. 4 Optical absorbance of the DNIM films that were
implanted at various dose rates for a fixed evaporation rate
of SiO2 and total thickness of 500 nm.
Fig. 3 Ion dose-rate dependence of surface roughness of respectively. The absorbance reasonably increases with dose
Cu:SiO2 films fabricated by the DNIM (circle) and ion rate, since the Cu concentration in the film increases with the
implantation alone (square). The surface roughness of EB-
dose rate for a fixed period. More importantly, a surface
deposited SiO2 (without ion implantation) and that of
plasmon mode around 2.2 eV varies with does rate. The
unirradiated SiO2 substrates are also given for comparison.
absorption spectra up to about 20 µA/cm2 do not show a
distinct plasmon peak but an absorption edge near 2 eV. The
rough as compared to the virgin SiO2 of the optical grade. presence of the plasmon mode is consistent with the Cu
The least ion dose rate of the DNIM significantly smoothens nanoparticles in the TEM observation but the absorption peak
the surface of the Cu:SiO2 film which is close to the is less distinct than that of the sample produced by the
roughness level of unirradiated SiO2 substrates. Although a conventional ion implantation [3-6]. In the conventional
further increase in dose rate rather roughens the surface, the implantation, the surface plasmon peak was present more or
surface by the DNIM is kept much smoother than the EB- less, though the distinctness depended on dose rate. The
deposited surface. The insensitiveness to dose rate suggests indistinct plasmon peaks at low dose rates are attributable to
that the smoothening effect is not due to macroscopic insufficient Cu precipitation or defects in Cu nanoparticles
processes (e.g., beam heating) but to microscopic ones, such and/or the matrix. Therefore, the spontaneous formation of
as radiation-induced compaction [9,10] or reduction of free the surface plasmon peak is a merit of the high-flux DNIM,
volume in the amorphous phase. Contrarily, the conventional
ion implantation at the same energy gives a monotonic
increase of the surface roughness with increasing dose rate,
and the roughness level at high dose rates exceeds that of EB-
evaporated SiO2 surface. This surface roughening is due to
the enhanced sputtering at high dose rates .
It is noted that the both curves of the DNIM and ion-alone
cases do not converge to each other, in the dose rate region
measured. The moving-boundary (surface) condition of the
DNIM favors the surface smoothness, because a constant
supply of fresh surface avoids energy accumulation at the
specific surface. The surface smoothening effect of the
DNIM process is particularly important and advantageous,
because the surface morphology is directly incorporated into
the SiO2 matrix under the moving boundary condition. The
smoothing effect suggests formation of a dense and less-
B. Optical Performance of Cu:SiO2 films by the DNIM
Optical absorbance is significantly dependent on the
both particle fluxes of negative Cu- and SiO2 evaporation.
Fig.4 shows dose-rate dependence of absorption spectra of Fig. 5 Dose rate dependence of the third-order susceptibility
the Cu:SiO2 films, where deposition rate and the film normalized by optical absorbance, χ(3)/α, of the Cu-SiO2 films,
thickness are kept constant at 0.2 nm/s and 500 nm, that was measured at 532 nm (2.33 eV).
that is, high energy-deposition rate enhances the Ostwald IV. SUMMARY
ripening of nanoparticles and/or in-beam annealing.
The third-order susceptibility χ(3) enhanced by surface A new filmmaking technique with high-flux negative ions
and electron beam evaporation was developed and applied to
plasmon is of our main interest. The signal intensities
measured for the Cu:SiO2 films are much stronger than that fabricate a nonlinear optical material of nanoparticles
embedded in insulators. Usage of high-flux negative ions
of the conventionally ion-implanted a-SiO2 [13,14].
Accordingly, the primary goal to obtain massive attained well-defined ion implantation, and the dynamic
negative-ion-mixing method succeeded in fabrication of
nanoparticles has been fulfilled.
thick/uniform nanoparticle structures, together with surface
Here, efficiency of χ(3) per thickness or a figure-of-merit is smoothing of the a-SiO2 substrate, i.e., densification of the a-
also important for the thick Cu:SiO2 films. One of the SiO2 matrix. The high flux of ions kinetically effect on
important quantities is the third-order susceptibility spontaneous growth of metal nanoparticles. The thick
normalized by the optical absorbance, χ(3)/α. Fig. 5 shows Cu:SiO2 films by the DNIM processing showed superb
dose rate dependence of the χ(3)/α of the Cu:SiO2 films at 532 optical performance, not only for linear absorption but also
nm (2.33 eV) in comparison with that of conventional ion for nonlinear ultra-fast response due to the surface plasmon
implantation. Although the nonlinear optical signal as a resonance of nanoparticles. The high-flux DNIM method
whole is much stronger, the normalized susceptibility χ(3)/α demonstrated fabrication of thick uniform films for metal
of the DNIM films is lower than that of the conventional ion nanoparticles embedded in insulators and promising
implantation. The lower susceptibility may result from the applications for other compound materials in insulators.
smaller size or defects in the DNIM process. The χ(3)/α of
the Cu:SiO2 films is also dependent on film-deposition rate: ACKNOWLEGEMENTS
χ(3)/α of 0.2 nm/s is larger than that of 0.4 nm/s. The
dependence on deposition rate indicates that the χ(3)/α The authors are indebted to Dr. C.G. Lee of NRIM for part of
improves by optimizing deposition rate or thermal annealing. AFM measurement.
A transient variation of differential optical density (∆OD),
that was obtained by the pump-probe method, is shown in Fig. REFERENCES
6, where the Cu:SiO2 film was fabricated at a dose rate 20  J. Ishikawa, H. Tsuji, Y. Toyota, Y. Gotoh, K. Matsuda, M. Tanjyo and
µA/cm2, an evaporation rate of 0.2 nm/s, and the thickness S. Sakaki, "Negative-Ion Implantation Technique," Nucl. Instrum. &
was 200 nm. The transient decay of the ∆OD shows a time Methods Phys. Res. B96 (1995) pp. 7-12.
 Y. Mori, G.D. Alton, A. Takagi, A. Ueno and S. Fukumoto, "Further
response of 2-3 pico-sec, which is rather faster than the Evaluation of the High Intensity Plasma Sputter Heavy Negative Ion
specimen by the conventional ion implantation [13,14]. The Source," Nucl. Instrum. & Methods Phys.Res. A273 (1988) pp. 5-12.
ultra-fast response of metal nanoparticles, much faster than  N. Kishimoto, V.T. Gritsyna, K. Kono, H. Amekura and T. Saito,
typical semiconductors, was also demonstrated for the thick "Negative Copper Ion Implantation into Silica Glasses at High Dose
Rates and the Optical Measurements," Nucl. Instrum. & Methods Phys.
Cu:SiO2 film. Thus, not only the linear but also the nonlinear Res. B127/128 (1997) pp. 579-582.
optical performance was demonstrated for the films  N. Kishimoto, V.T. Gritsyna, Y. Takeda, C.G. Lee and T. Saito, “Dose-
fabricated by the DNIM method. Rate Dependent Implantation of Negative Copper Ions into Silica
Glasses and Effects on Colloid Formation,” Nucl. Instrum. & Methods
Phys. Res. B141, (1998) pp. 299-303.
 N. Kishimoto, Y. Takeda, V.T. Gritsyna, E. Iwamoto and T. Saito, “A
High-Current Negative-Ion Implanter and its Application for
Nanocrystal Fabrication in Insulators,” IEEE 98EX144, Proc. of 12th
Int. Conf. on Ion Implantation Technology (1999) pp.342-345.
 N. Kishimoto, N. Umeda, Y. Takeda, C.G. Lee and V.T. Gritsyna,
“Self-Assembled Two-Dimensional Distribution of Nanoparticles with
High-Current Cu--Implantation into Insulators,” Nucl. Instrum. & Meth,
Phys. Res. B 148 (1999) pp. 1017-1022.
 N. Kishimoto, Y. Takeda, N. Umeda, V.T. Gritsyna, C.G. Lee and T.
Saito, “Metal Nanocrystal Formation in Magnesium Alminate Spinel
and Silicon Dioxids iwht High-Flux Cu- Ions,” Nucl. Instrum. Methods
Phys. Res. B166/167 (2000) pp. 840-845.
 N. Kishimoto, Y. Takeda, K. Kono, “Film Fabrication Method and the
Apparatus with Negative-Ion Irradiation and Co-Evaporation,” JAP-Pat.
Applied No.245934 (1999) Aug.31.
 R.A.B. Devine, Nucl. Instrum. Methods in Phys. Res. B91, (1994) 378.
 M.L. Brongersma, E. Snoeks and A. Polman, Appl. Phys. Lett. 71,
 J.F. Ziegler, J.P. Biersack and U. Littmark, The Stopping and Range of
Ions in Solids (Pergamon Press, New York, 1985), Chap 8.
 N. Kishimoto, C.G. Lee, N. Umeda, Y. Takeda and V.T. Grytsyna,
“Sputtering Effects and Two Dimensional Arrangement of
Nanoparticles in Insulators under High Flux Cu- Implantation,” Mat.
Res. Soc. Symp. Proc. (1999) in press.
 Y. Takeda, V.T. Gritsyna, N. Umeda, C.G. Lee and N. Kishimoto,
“Linear and Nonlinear Optical Properties of Cu Nanoparticles
Fig. 6 Transient response of ∆OD pumped at 573 nm and Fabricated by High-Current Cu- Implantation in Silica Glass,” Nucl.
probed at 568 nm of the Cu:SiO2 film that was fabricated by Instrum. & Methods Phys. Res. B148 (1999) 1029-1033.
the DNIM method.  Y. Takeda, J.P. Zhao, C.G. Lee, V.T. Gritsyna and N. Kishimoto,
“Nonlinear Optical Properties of Cu Nanoparticles Embedded in
Insulators by High-Current Cu- Implantation,” Nucl. Instrum. &
Methods Phys. Res. B166/167 (2000) pp.877-881.