Controlled particle generation in an inductively coupled plasma

Article abstract of DOI:10.1063/1.2193041

By injecting pulses of acetylene into an inductive argon/helium discharge, carbon clusters with diameters in the range of 10-50 nm are produced. These particles cause an instability of the plasma, which becomes visible as an oscillation of the emission intensity. The particles are analyzed ex situ using atomic force microscopy and scanning electron micrographs. A unique linear dependence between particle size and oscillation time period is found. Thereby the oscillation phenomenon can serve as monitor signal to control the size of plasma produced particles.

Full text of DOI:10.1063/1.2193041

Controlled particle generation in an inductively coupled plasma
M. Schulze, A. von Keudell, and P. Awakowicz
Citation: Applied Physics Letters 88, 141503 (2006); doi: 10.1063/1.2193041
View online: http://dx.doi.org/10.1063/1.2193041
View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/88/14?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Discharge mode transition and hysteresis in inductively coupled plasma
Appl. Phys. Lett. 102, 234104 (2013); 10.1063/1.4809925
Effect of capacitive coupling in a miniature inductively coupled plasma source
J. Appl. Phys. 112, 093306 (2012); 10.1063/1.4764333
Generation of high-power arbitrary-wave-form modulated inductively coupled plasmas for materials processing
Appl. Phys. Lett. 90, 071502 (2007); 10.1063/1.2696885
Stable and unstable behavior of inductively coupled electronegative discharges
J. Vac. Sci. Technol. A 21, 1849 (2003); 10.1116/1.1613953
Downstream instabilities of electronegative plasma discharges
Phys. Plasmas 10, 539 (2003); 10.1063/1.1535209
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
137.149.200.5 On: Sun, 30 Nov 2014 05:58:51

APPLIED PHYSICS LETTERS 88, 141503 ͑2006͒
Controlled particle generation in an inductively coupled plasma
M. Schulzea͒ and A. von Keudell
Arbeitsgruppe Reaktive Plasmen, Ruhr-Universität Bochum, Universitätsstrasse 150,
44780 Bochum, Germany
P. Awakowicz
Lehrstuhl für Allgemeine Elektrotechnik und Plasmatechnik, Ruhr-Universität Bochum, Universitätsstrasse
150, 44780 Bochum, Germany
͑Received 16 December 2005; accepted 22 February 2006; published online 6 April 2006͒
By injecting pulses of acetylene into an inductive argon/helium discharge, carbon clusters with
diameters in the range of 10–50 nm are produced. These particles cause an instability of the plasma,
which becomes visible as an oscillation of the emission intensity. The particles are analyzed ex situ
using atomic force microscopy and scanning electron micrographs. A unique linear dependence
between particle size and oscillation time period is found. Thereby the oscillation phenomenon can
serve as monitor signal to control the size of plasma produced particles. © 2006 American Institute
of Physics. ͓DOI: 10.1063/1.2193041͔
In the past decade, dust formation in reactive plasmas
has been investigated for many applications.1,2 This research
began with the need to reduce particle formation in process-
ing plasmas used in the semiconductor industry. In the mean-
time, also beneficial applications of particles have been es-
tablished. In particular, small particles with diameters below
10 nm can serve as quantum dots in electronic circuits or as
photonic devices. Embedded in an amorphous matrix, nano-
particles can induce photoluminescence3,4 or they can en-
hance the electronic properties of silicon thin film solar
cells.5
The adjustment of the particle diameter is often per-
formed empirically by varying the growth time in the
plasma. The in situ characterization of the properties of very
small particles, however, is an unresolved issue. Optical
methods such as laser scattering cannot be applied for par-
ticles with a diameter below dӍ20 nm, because the cross
section for light scattering scales with d6. Current methods to
control particle growth in rf driven plasmas work indirectly
by exploiting the feedback between particles and plasma per-
formance. A typical example are V-I probes that measure the
spectrum of the rf current to the powered electrode. The
formation of particles in a plasma causes its impedance to
change, which becomes apparent in the intensity of the har-
monics of the driving frequency.6 However, since the inten-
sity of the harmonics is influenced by the whole rf circuitry,
a frequent recalibration of the feedback is necessary.
A vessel resembling the Gaseous Electronics Conference
͑GEC͒ reference cell is powered by two electrodes: the one
at the top couples to the plasma capacitively and a pancake
antenna at the bottom can heat the plasma inductively ͓see
Fig. 1͑a͔͒. A movable sample holder is embedded into the top
electrode. It can be retracted into a load lock where it can be
loaded with 2ϫ2 cm2 sized flat silicon substrates. The gas
pressure is controlled by a butterfly valve that connects the
vessel with the pumping cabinet. Up to three gases can be
fed into the reactor by mass flow controllers via a ring
shaped gas shower. The reactor is equipped with several di-
agnostics such as a spectroscopic in situ ellipsometer, a
Langmuir probe system, a mass spectrometer, a photodiode,
and a high speed intensified charge coupled device ͑ICCD͒
In Ref. 7, a different method has been proposed to moni-
tor particle properties: an instability is triggered by injecting
particles in an inductive plasma, which becomes visible as an
oscillation of the plasma emission. It has been postulated that
a periodic movement of the particles within the plasma leads
to fluctuations in the plasma heating efficiency. The resulting
modulation of the plasma emission can easily be detected by
a photodiode. In this letter the frequency of the observed
plasma oscillations is correlated with the particle diameter. It
will be shown that the oscillation time period uniquely de-
pends on the particle diameter and can serve as a monitor
signal to control particle formation in reactive plasmas.
FIG. 1. ͑a͒ Sketch of the GEC-like ICP chamber. ͑b͒ Timing of the particle
generation.
a͒
Electronic mail: martin.schulze@rub.de
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
0003-6951/2006/88͑14͒/141503/3/$23.00 88, 141503-1 © 2006 American Institute of Physics
137.149.200.5 On: Sun, 30 Nov 2014 05:58:51

141503-2
Schulze, von Keudell, and Awakowicz
Appl. Phys. Lett. 88, 141503 ͑2006͒
in electronegative discharges.8 However, unlike this attach-
ment instability where the electronegative species experience
a periodic transition from neutral state to negative ion to
neutral to positive ion, the charge state of the particles in this
experiment depends predominantly on the particle size and
remains nearly constant during the oscillations. Hence, the
oscillations are not caused by charge fluctuations but by the
spatial transport of the particles and the charge they carry.
After consumption of all acetylene, the growth process
stops and a particle containing argon/helium discharge re-
mains. For a discharge volume of approximately 5 l, the
mean residence time of neutral species due to pumping
losses is half a second. Therefore, it is reasonable to assume
that the chemistry is finished within a few seconds after the
supply of acetylene is stopped. The particle confinement is
maintained by electrostatic force in the plasma sheaths,
which keeps the negatively charged particles inside the
plasma volume. The plasma is from then on purely inductive
and a characteristic frequency of the oscillation up to several
tens of hertz is established ͓phase ͑3b͒ in Fig. 2͔. The nature
of this oscillation will be discussed below. For the selected
external parameters, particles cause an oscillation with a time
period of 79 ms or a frequency of 12.5 Hz. Within the scatter
of the data, this frequency stays constant for some minutes.
Sporadically, a parasitic discharge in the dark room of the
chamber occurs and causes more violent variations in the
plasma emission ͑e.g., Fig. 2, 85–95 s͒. The reason for this
could not be clarified yet. By interrupting the plasma, par-
ticles can leave the discharge ͓phase ͑4͒ in Fig. 2͔. When the
antenna is switched on again, the inductive mode is stable
͓period ͑5͒ in Fig. 2͔.
FIG. 2. Optical plasma emission during a sequence of preparation ͑1͒,
acetylene injection ͑2͒, oscillations ͑3͒, plasma interruption ͑4͒, and stable
discharge ͑5͒. The discharge is powered by the capacitive electrode during
͑2͒ and by the inductive coil during ͑3͒ and ͑5͒.
camera with an acousto-optical wavelength filter.
The experiments are carried out as follows ͓see Fig.
1͑b͔͒: first ͑1͒ a flow of argon and helium as buffer gas is set
to 4 SCCM ͑SCCM denotes cubic centimeter per minute at
STP͒ ͑argon͒ and 15 SCCM ͑helium͒ at a pressure of 4 Pa.
Acetylene is added at a flow rate ␾
; next ͑2͒, a capacitive
C H
2
2
discharge at a forward power of 70 W is ignited for a time
period ⌬t1 to induce particle formation; then ͑3͒ the acety-
lene flow is switched off and the antenna is used to sustain
the plasma in the inductive mode at 100 W forward power
for a time period ⌬t2=120 s. The particles generated from
acetylene in ͑2͒ stay confined in the discharge during phase
͑3͒; afterwards ͑4͒ the plasma is switched off and the par-
ticles are collected on the silicon substrate for ex situ analysis
with an atomic force microscope ͑AFM͒ and a scanning elec-
tron microscope ͑SEM͒.
After measuring the oscillation time period during phase
͑3b͒, the particles are deposited on a silicon sample. The
experiment is repeated for different injection times
⌬t1=2,4, ... ,32 s. The SEM is used to obtain high resolu-
tion images of the samples. For cases ͑a͒ 4 s, ͑b͒ 8 s, and ͑c͒
32 s such images are shown in Fig. 3. Obviously, the isolated
particles are monodisperse and have an almost perfect
spherical shape. Their diameters increase with ⌬t1. Apart
from the isolated particles, also groups of two or more par-
ticles are visible on the images. However, a tendency to form
cauliflowerlike ensembles, as reported for much larger par-
ticles in other experiments,9 cannot be observed. Therefore,
it is assumed that the particle islands on the substrate are
formed during switching off the plasma. With the help of
edge detection and image segmentation techniques the par-
ticle density and the diameter distribution of the isolated par-
ticles on the samples have been determined. The diameter is
found to follow a Gaussian probability distribution with a
standard deviation between 3 and 5 nm for all experiments.
The mean diameter ͗d͘ is denoted on the images in Fig. 3.
Measurements with an AFM confirm the spherical shape
of the particles. The distribution of particle heights from
AFM matches also the diameter distribution from the SEM
images. While the resolution of the SEM is restricted to par-
ticle diameters above Ӎ10 nm ͑the contrast of smaller par-
ticles is too weak to detect them reliably and their boundary
is not sharp͒, the AFM would in principle allow us to mea-
sure even smaller particles. Due to problems with particle
sticking to the AFM probe tip, however, it has not been pos-
sible to get reproducible images for particles Ͻ10 nm.
Figure 4 shows the correlation between the measured
The overall optical emission of the plasma is measured
by a photodiode. The time evolution for an experiment with
␾
=4 SCCM and ⌬t1=4 s is shown in Fig. 2. During the
C H
2
2
capacitive plasma phase ͑2͒, the light emission is low com-
pared to the inductive phase ͑3͒ due to the much lower elec-
tron density. When powering the inductive antenna, the
plasma oscillations during phase ͑3͒ become clearly visible.
After interrupting the plasma ͑4͒ for a few seconds, the
plasma is switched on again and the emission remains stable
͓phase ͑5͒ in Fig. 2͔ from then on.
The principle mechanism driving the observed plasma
instability has been discussed in Ref. 7: during the injection
phase ͑2͒, particles are formed from acetylene. When the
inductive antenna is properly powered, the plasma imped-
ance is at the edge of the transition between capacitive and
inductive modes. Fluctuations in the spatial particle distribu-
tion cause strong variations in the heating efficiency, leading
to an oscillation between capacitive and inductive modes.
Consequently, a strong, chaotic modulation of the plasma
emission is observed ͓phase ͑3a͒ in Fig. 2͔. The situation is
comparable to an electron attachment instability with char-
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
acteristic frequencies of 102–104 Hz that has been observed
oscillation time period during phase ͑3b͒ and the mean par-
137.149.200.5 On: Sun, 30 Nov 2014 05:58:51

141503-3
Schulze, von Keudell, and Awakowicz
Appl. Phys. Lett. 88, 141503 ͑2006͒
FIG. 4. Correlation between particle diameter ͑͗d͒͘ and oscillation time
period. The bars in x direction indicate the uncertainty in the time period
while the bars in y direction equal the standard deviation of the particle
diameter distribution. The solid line is drawn arbitrarily to indicate the first-
order linearity of the dependence.
inertia of the particles. A correlation between oscillation pe-
riod and particle diameter results. Details of the Langmuir
probe measurements and the interpretation of these results
will be presented elsewhere.12
Particles have been generated by injecting acetylene in
an inductive plasma from argon. This triggers an oscillation
of the plasma emission. The time period of the oscillation is
found to depend linearly on the mean diameter of the par-
ticles. Thereby the oscillation phenomenon can serve as
monitor signal to control the size of plasma produced par-
ticles. The lower size limit of about 10 nm in the experi-
ments discussed here is determined by the current restric-
tions of the imaging tools. The extension of the experiments
to produce even smaller particles in a controlled manner is
currently under investigation.
FIG. 3. SEM pictures of particles deposited onto Si samples. The injection
times of C2H2 leading to pictures ͑a͒, ͑b͒, and ͑c͒ were 4, 8, and 32 s,
respectively.
ticle diameter. To first order, a linear dependence is found to
exist. But why does the oscillation frequency scale with the
particle diameter?
In order to identify the physical nature of the oscillation,
a Langmuir probe and a high speed ICCD camera have been
used to measure the electron density temporally and spatially
resolved. These data indicate that the oscillation consists of
an off-centered localized void or plasmoid, which rotates
around the symmetry axis of the reactor. This particle-
induced rotation of a plasmoid is similar to findings in litera-
ture: for a capacitive discharge in Refs. 10 and 11 the spon-
taneous formation of a void in a dusty plasma has been
observed, if the ion drag force becomes larger than the elec-
trostatic force. This void rotates with frequencies in the hertz
range for particles larger than 200 nm in diameter. In the
inductive plasma discussed here, the ion drag force is much
more intense due to the much higher electron density. As a
consequence, the condition for void formation is already met
for particles larger than 1 nm in diameter. It is believed that
this void formation takes place in the inductive phase ͓period
͑3͒ in Fig. 2͔. The minimum energy configuration is appar-
ently an off-centered void which rotates around the symme-
try axis of the reactor. Since the rotation of the void implies
also the movement of the dusty plasma around, it is straight-
forward that the period of this oscillation depends also on the
The authors acknowledge the support by the German
Science Foundation via the SFB 591.
1A. Bouchoule, Dusty Plasmas: Physics, Chemistry and Technological Im-
pacts in Plasma Processing, 1st ed. ͑Wiley, New York, 1999͒.
2C. Hollenstein, Plasma Phys. Controlled Fusion 42, R93 ͑2000͒.
3V. Suendo, G. Patriarche, and P. R. i Cabarocas, Opt. Mater. 27, 953
͑2005͒.
4Z. Shen, T. Kim, U. Kortshagen, P. McCurry, and S. Campbell, J. Appl.
Phys. 94, 2277 ͑2003͒.
5A. F. i Morral, H. Hofmeister, and P. R. i Cabarrocas, J. Non-Cryst. Solids
299–302, 1173 ͑2002͒.
6L. Boufendi, J. Gaudin, S. Huet, G. Viera, and M. Dudemaine, Appl. Phys.
Lett. 79, 4301 ͑2001͒.
7A. von Keudell and M. Bauer, Plasma Sources Sci. Technol. 13, 285
͑2004͒.
8P. Chabert, A. J. Lichtenberg, M. Lieberman, and A. Marakhtanov, Plasma
Sources Sci. Technol. 10, 478 ͑2001͒.
9I. Stefanović, E. Kovačević, J. Berndt, Y. Pendleton, and J. Winter, Plasma
Phys. Controlled Fusion 47, A179 ͑2005͒.
10G. Praburam and J. Goree, Phys. Plasmas 3, 1212 ͑1996͒.
11D. Samsonov and J. Goree, Phys. Rev. E 59, 1047 ͑1999͒.
12M. Schulze, A. von Keudell, and P. Awakowicz, Plasma Sources Sci.
Technol. ͑submitted͒.
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
137.149.200.5 On: Sun, 30 Nov 2014 05:58:51