Electron bounce resonance heating in dual-frequency
capacitively coupled oxygen discharges
Yong-Xin Liu1, Quan-Zhi Zhang1, Jia Liu1, Yuan-Hong Song1, Annemie Bogaerts2,
and You-Nian Wang1,a)
1
School of Physics and Optoelectronic Technology, Dalian University of Technology,
Dalian 116024, China
2
Department of Chemistry, University of Antwerp, Campus Drie Eiken,
Universiteitsplein 1, BE-2610 Wilrijk-Antwerp, Belgium
Absract: The BRH in DF CCPs operated in oxygen has been studied by different
experimental methods and a particle-in-cell/Monte Carlo collision (PIC/MCC) simulation,
and compared with the electropositive argon discharge. All these experimental observations
are explained by PIC/MCC simulations, which show that in the oxygen discharge the bulk
electric field becomes quite strong and is out of phase with the sheath field.
Keywords: Dual-frequency, electron density, hairpin probe, oxygen discharge
a)
E-mail: [email protected]
1. Introduction
2. Experiment setup and simulation
Recently, the bounce resonance heating (BRH) effect has
been
experimentally
observed in
in
(2/60 MHz) capacitive discharge chamber [9]. When
electropositive argon, and it was found to markedly
conducting the optical measurements, the O2 discharge is
enhance the discharge performance under certain
diluted with 5% Ar, as it is difficult to probe the emission
conditions.
applications,
from the direct excitation of molecule O2. The measured
electronegative gases are most frequently employed.
electron density by the hairpin probe is corrected by
They are different from electropositive gas discharges in
using the fluid model described by Piejak [10]. In the
some basic, important features, such as the development
experiment, we will compare the different behavior of
of double layers [1, 2], a strong electric field and thus a
BRH in Ar and O2 and study the effect of high-frequency
high ionization rate in the bulk [3], more pronounced
(HF) power on the BRH in the O2 discharge.
However,
in
real
DF
CCPs
Our experiments are performed in a parallel-plate DF
modulation of the bulk electric field [4, 5], and a
discharge mode transition induced by voltage or
pressure[6-8]. These pronounced differences will have a
substantial impact on the BRH, and the resultant electron
energy probability function (EEPF) determines the basic
discharge dynamics, such as dissociation, excitation, and
ionization in electronegative discharges. Therefore, it is
of fundamental importance to study the features of BRH
in electronegative discharges.
Fig. 1 Schematic diagrams of the dual-frequency
capacitively coupled plasma reactor.
Our simulations, which serve as a numerical
explanation of the experiments of O2 and Ar discharges,
are based on the standard one-dimensional in space
the Ar discharge, the most significant BRH effect in the
three-dimensional in velocity (1D3V) electrostatic PIC
O2 discharge occurs at a somewhat larger gap, with the
MCC method. More details can be found in Ref. [9]. To
resonant peak broadened. The larger resonance gap and
shed light on the BRH behavior at different HF powers in
broader peak in the O2 discharge is related to the
the experiments of O2, we compare the spatiotemporal
electronegative discharge structure. In simulations, we
distribution of various plasma quantities (i.e., the
observed a narrower bulk region and higher bulk electric
electron impact ionization rate, bulk electric field,
field. Note that the position of the resonance peak is
trajectory of a typical bouncing electron and its energy)
almost independent of the pressure in both Ar [9] and O2.
for a range of HF voltages. In the case of the O2
(ii) At L > 2:5 cm, the positive ion and electron densities
discharge, the charged particles O2+, O, and the
drop monotonously in the O2 discharge upon increasing
electrons are traced using cross sections from Ref. 23.
L, while the plasma density rises in the Ar case. This is
For Ar, the cross sections are the same as in Ref. [9].
due to the more significant drop of the BRH with
increasing L in the O2 discharge. It is worth to mention
that our PIC MCC simulations have qualitatively
3. Results and discussion
reproduced the experimental density curves both in Ar
and O2.
3.1 electrode gap effect
3.2 HF power effect
0.4
0.8
-
e density in argon
e density in oxygen
by hairpin resonant probe
1
2
3
4
5
0.2
-3
cm )
+
1.2
O2 density (a.u.)
0.6
0.4
2.8
80 W
60 W
40 W
20 W
10
0.8
Electron density (10
-3
1.6
Ion density (10
2.0
10
cm /a.u.)
+
Ar density in argon
+
relative O2 density in oxygen
by floating double probe
2.4
2.4
2.0
1.6
6
Gap length (cm)
Fig. 2 Experimental measurements: positive ion and
electron density versus gap length, in Ar (black) and O2
1.2
1
2
3
4
5
6
Gap length (cm)
(red lines) discharges, at a fixed HF and LF power of
Fig. 3 Experimental measurements in the O2 plasma:
30W and 100 W, respectively. Note that the pressure
relative O2+ Ion density at the discharge center versus L,
increases from 1.3 Pa in Ar to 4.0 Pa to sustain the O2
at a range of HF powers PH for a fixed LF power PL of
discharge.
100 W, at a gas pressure of 4.0 Pa.
Fig. 2 illustrates the measured positive ion density and
The effect of HF power PH on the BRH in the O2
electron density versus gap length L, at a fixed HF and
discharge is shown in Fig. 3, which displays the
low-frequency (LF) power of 30W and 100 W, for the Ar
measured relative O2+ ion density versus L at different
and O2 discharge (black and red curves, respectively).
values of PH with a fixed PL of 100W and a pressure of
The Ar+ ion density is measured in absolute terms,
4.0 Pa. At lower PH, both the O2+ ion density peak
whereas the O2+ density is only obtained in relative
gradually drops and almost disappears at PH = 20 W,
terms. The BRH effect is illustrated by the peak in ion
which indicates a suppressed BRH. It should be noted
and electron density at a certain gap length, called the
that the results are quite different from the electropositive
resonant gap length. From this figure, the differences of
Ar discharge, where the BRH is hardly affected by PH
the BRH effect between Ar and O2 discharges can be
[9]. This is attributed to the fact that the O2 discharge
generally summarized as followings. (i) Compared with
experiences a mode transition from electropositive to
electronegative due to the lower amount of sheath
explained because at relatively low driving frequency the
heating with the drop of PH. This mode transition
electrons can respond to the rf electric field whereas they
induced by the driving voltage has been observed in
gradually fail to follow the rf field at higher frequencies,
CCPs with other electronegative gases as well [6, 7].
i.e. at the driving frequency f > 60 MHz. Note that this
3.3 Gas pressure
phase relationship between the plasma bulk field and the
Fig. 4 illustrates the measured positive-ion O2+ density
sheath field at different driving frequencies is consistent
versus electrode gap L at different pressures. At higher
with the results in [13].
pressures, the maximum around the resonance gap
0.500
diminishes gradually and almost disappears at 12 Pa,
electronegative discharges, two different pressure effects
are responsible for the BRH. First, as the electron mean
free path is inversely proportional to the pressure, the
resonance electrons simply cannot bounce long time
enough at high pressures, to gain sufficient energy before
being stopped or re-directed by a collision. This effect
also occurs in a pure argon discharge [12]. Moreover, an
additional pressure effect occurs in the oxygen discharge,
namely the enhanced bulk electric field reversal, which
4.0 Pa
6.0 Pa
8.0 Pa
12.0 Pa
+
O2 density (a.u.)
2.2
2.0
1.8
0.375
0.250
0.125
0.000
0
20
40
60
80
100
120
Driving frequency (MHz)
Fig. 5 Phase delay of the plasma bulk field with respect
to the sheath field, θ, as a function of the driving
suppresses the BRH at higher pressures.
2.4
Phase delay
/
which indicates again a suppression of the BRH. In
frequency.
4. Conclusion
1.6
1.4
We investigated the electron bounce resonance heating
1.2
(BRH) in dual-frequency capacitively coupled plasmas
1.0
operating in oxygen by different experimental methods
0.8
1
2
3
4
5
6
Electrode gap (cm)
and a particlein-cell/Monte Carlo collision (PIC/MCC)
simulation, and compared with the electropositive argon
Fig. 4 Measured positive-ion (O2+ ) density versus
discharge. In comparison with the electropositive argon
electrode gap L in the oxygen discharge at different
discharge, the BRH in the electronegative oxygen
pressures with a fixed HF and LF power of 30W and
discharge tends to happen at larger electrode gaps. This
100W
is related to the peculiar oxygen discharge structure, i.e.
3.4 HF frequency effect
the development of a double layer in the sheath and the
Fig. 5 illustrates the phase delay of the plasma bulk
narrower bulk region and higher reversal electric field in
field with respect to the sheath field, as a function of the
the bulk. At electrode gap L > 2.5 cm, the positive-ion
driving frequency. At the driving frequency of 13.56
(and electron) density drops monotonically in the oxygen
MHz, the bulk electric field is in phase with the sheath,
discharge upon increasing electrode gap, whereas it rises
so θ is zero. With the increase in the driving frequency,
(after an initial drop) in the argon case. This indicates
the phase of the bulk electric field will delay with respect
that the BRH drops significantly at electrode gaps larger
to the sheath field, the phase delay θ increases rapidly
than the resonance length, in the oxygen discharge.
and tends to reach its maximum π/2 when the driving
PIC/MCC simulation results show that in the oxygen
frequency approaches 120 MHz. This behavior can be
discharge, the bulk electric field reversal becomes quite
strong and is out of phase with the sheath field.
Therefore, it retards the resonance electrons when
Reference
traversing the bulk, resulting in a suppressed BRH.
[1] N. Nakano, N. Shimura, Z.-L. Petrovic´, and T.
Finally, in addition to the effect of the discharge gas
Makabe, Phys. Rev. E 49, 4455 (1994).
(oxygen versus argon), also the effect of HF power, gas
[2] V. A. Lisovskiy and V. D. Yegorenkov, Vacuum 80,
pressure and high frequency on the BRH were
458 (2006).
investigated, both experimentally and by simulations. It
[3] M. Shibata, N. Nakano, and T. Makabe, J. Appl. Phys.
appears that the resonance peaks in the positive-ion
77, 6181 (1995).
density become weaker with decreasing HF power. This
[4] M. Yan, A. Bogaerts, W. J. Goedheer, and R. Gijbels,
is attributed to the enhancement of the reversal bulk
Plasma Sources Sci. Technol. 9, 583 (2000).
electric field when the discharge undergoes a transition
[5] M. Yan, A. Bogaerts, and R. Gijbels, J. Appl. Phys.
from electropositive to electronegative mode with the
87, 3628 (2000).
drop in the HF power. Furthermore, in the pure oxygen
[6] J. Schulze, A. Derzsi, K. Dittmann, T. Hemke, J.
discharge, the BRH is suppressed with increasing
Meichsner, and Z.
pressure and it almost diminishes at 12 Pa. This is due to
Donko, Phys. Rev. Lett. 107, 275001 (2011).
a combined effect of more frequent electron–neutral
[7] O. V. Proshina, T. V. Rakhimova, A. T. Rakhimov,
collisions and enhanced bulk electric field reversal at
and D. G. Voloshin,
high pressures. Finally, the BRH was found to depend
Plasma Sources Sci. Technol. 19, 065013 (2010).
significantly on the high frequency applied, which can be
[8] K. Denpoh and K. Nanbu, Jpn. J. Appl. Phys., Part 1
explained based on the phase relation between bulk
39, 2804 (2000).
electric field and sheath electric field.
[9] Y. X. Liu, Q. Z. Zhang, W. Jiang, W. Q. Lu, and Y. N.
Wang, Plasma Sources Sci. Technol. 21, 035010 (2012).
[10] R. B. Piejak, V. A. Godyak, R. Garner, B. M.
5. Acknowledgement
Alexandrovich, and N. Sternberg, J. Appl. Phys. 95, 3785
This work was supported by the Important National
(2004).
Science and Technology Specific Project (Grant No
[11] V. Vahedi and M. Surendra, Comput. Phys.
2011ZX02403-001),
Commun. 87, 179 (1995).
the
National
Natural
Science
Foundation of China (NSFC) (Grand No 11075029), the
[12] Liu Y X et al 2012 Plasma Sources Sci. Technol. 21
International Science and Technology Cooperation
035010
Program of China (Grant No 2012DFG02150) and the
[13] Yan M, Bogaerts A, Gijbels R and Goedheer W J
joint research project in the framework of the agreement
2000 J. Appl. Phys. 87 3628
between MOST and FWO.