22nd International Symposium on Plasma Chemistry
July 5-10, 2015; Antwerp, Belgium
Particularities of particles propagation under high-power impulse magnetron
sputtering
N. Britun1, M. Palmucci1, S. Konstantinidis1 and R. Snyders1,2
1
Chimie des Interactions Plasma Surface (ChIPS), CIRMAP, Université de Mons, 23 Place du Parc, 7000 Mons,
Belgium
2
Materia Nova Research Center, Parc Initialis, 7000 Mons, Belgium
Abstract: The results on time-resolved 2-D characterization of the particles propagation in
the high-power impulse magnetron sputtering discharges are presented. Laser-induced
fluorescence Doppler-shift imaging technique is applied for this purpose. The particularities
of the velocity distribution for Ti neutral and singly-ionized atoms (Ti+) in an Ar-Ti
discharge during the plasma on- and off- time are presented and analysed.
Keywords: laser-induced fluorescence, LIF, HiPIMS, Doppler-shift, velocity distribution.
1. Introduction
High-power impulse magnetron sputtering (HiPIMS)
discharges belong to the class of ion-enhanced physical
vapour deposition (IEPVD) discharges [1]. They emerged
from the convenient direct current and pulsed magnetron
sputtering as a result of time re-distribution of the applied
power, achieved by squeezing the plasma on-time and
applying much higher power (>10 kW) during this
interval, while maintaining the average power at the level
of direct current magnetron sputtering (DCMS) [2, 3].
The pulse repetition frequency in HiPIMS normally varies
in the range from few tens Hz to few kHz, whereas the
pulse duration can be from ~ 5 to ~ 500 µs [4, 5].
The HiPIMS discharges are highly acknowledged in the
industrial domain due primarily to the useful properties of
the coatings which can be obtained using this technology,
such as high film density, controllable film phase
constitution, controllable structure and crystallinity, etc.
[6]. The scientific interest to the HiPIMS plasmas is
motivated by the variety of rather new dynamic physical
processes, which cannot be attained in simple glow or
hollow cathode discharges, and should be characterized in
µs time-scale [5]. The particles propagation, the evolution
of their velocity distribution, dynamic rarefaction of the
sputtering gas, secondary electrons motion, etc. are
among these processes.
This work is devoted to study of the particles
propagation and their velocity distribution function (VDF)
in Ar-Ti short-pulsed HiPIMS discharges using the timeresolved two-dimensional visualization of these processes
by laser-induced fluorescence.
2. Diagnostics method and setup
Laser-induced fluorescence (LIF) technique, together
with its analogues, such as LIF imaging, and LIF
spectroscopy represents a powerful (quasi) non-intrusive
plasma characterization tool applicable to variety of
gaseous discharges [7, 8]. Doppler-shift LIF (DS-LIF), as
well as DS-LIF imaging are the techniques capable to
P-I-2-4
selectively characterize velocity distribution of the
particles under interest, based on the well-known Doppler
effect (assuming narrow laser linewidth) [9]. The
unambiguous advantage of LIF technique is represented
by the fact that the fluorescence emission intensity, 𝐼𝐿𝐿𝐿 ,
is proportional to the population of the probed species in
the lower state 𝑁𝑗 [7]:
𝐼𝐿𝐿𝐿 ~ 𝑁𝑗 ∙ 𝐼𝑙𝑙𝑙 ∙
𝐴𝑖𝑖 𝐵𝑗𝑗
𝑄𝑖 +𝐴𝑖
,
(1)
where j, k, and i stand for the lower, intermediate, and
upper energy levels correspondingly, 𝐴𝑖𝑖 is the
fluorescence emission probability, 𝐵𝑗𝑗 is the absorption
probability on the level j, 𝑄𝑖 and 𝐴𝑖 are the quenching and
spontaneous emission coefficients corresponding to the
level i. Laser excitation corresponds to j→ i, whereas
fluorescence to i→ k spectral transition.
The dependence expressed by eq. (1) makes all LIFrelated diagnostics techniques unique for studying the
ground-state density of species in the gaseous discharges,
which is not attainable by optical emission spectroscopy
(OES), or OES imaging. This is especially useful for twodimensional visualization (mapping) of the ground states
and/or metastable species in plasma, as well as for
mapping of their velocity distribution.
The experimental setup used in this work which
contains the HiPIMS discharge reactor with a planar
circular target, along with the laser excitation system
utilized for LIF diagnostics are shown in Fig. 1. The
geometry of the laser beam introduced into the vacuum
reactor, and the studied discharge area above the target
are shown in Fig. 2. The additional experimental
parameters, related to both the HiPIMS plasma reactor
and LIF diagnostics are listed in Table 1.
1
Fig. 2. Side-view of the laser beam and the area of interest (≈
10×7 cm) in the studied HiPIMS discharge.
Table 1. HiPIMS and LIF experimental parameters.
Fig. 1. Top-view of the HiPIMS reactor with the dye laser used
for LIF discharge diagnostics.
HiPIMS discharge parameters:
3. Plasma diagnostics results
The experimental 2-D imaging results comparing the
sputtered Ti and Ti ions at low (5 mTorr) Ar pressure are
presented in Fig. 3. The measurements have been made at
different laser wavelengthλ, so the presented maps
correspond to different energetic groups of the discharge
particles. The wavelength (∆λ = λ - λ 0 ) as well as the
corresponding velocity shifts (∆v) are shown in the upper
right corners. The presented results have been obtained
during the plasma off-time and related to two different
delay times, ∆t = 25 µs (for Ti) and 40 µs (for Ti+). We
have to note that, the laser linewidth defines the typical
VDF uncertainty (in the velocity units) at the level of
≈ 0.75 km/s, which is less than the typical velocity of
sputtered species [10, 11], and can be considered as an
acceptable resolution for DS-LIF measurements.
From Fig. 3(a) one can clearly observe two groups of
particles sputtered in the opposite directions (which
correspond to the extremes of the wavelength shift). The
sputtered particles move toward the central discharge
region (the laser beam on each figure goes from left to
right), forming a density maximum at about 3 cm above
the target (not shown). The preferential angle of
sputtering is estimated to be about 70 degree, which does
not change much with increase of the Ar pressure. Even
though the existence of the high-energy particles in
HiPIMS is evident form Fig. 3, their relative contribution
to the total VDF is rather small (cannot be seen due to the
individual normalization of each density map).
At the same time, Ti ions demonstrate rather symmetric
velocity distribution, since all the DS-LIF density maps
have symmetric structure, without respect to the
wavelength shift, as shown in Fig. 3(b). Such behaviour
implies that, the ways of sputtering of neutral and ionized
particles in HiPIMS are different, so the ion VDFs in the
horizontal direction are symmetrical. The essential
velocity component for the ions laying out of the image
plane and appearing due to the Lorentz force, however,
should not be forgotten. The gyration effect, however, is
2
Target
99.99 % Ti , 10 cm in diameter
Working gas pressure
Ar, at 5 and 20 mTorr
Base pressure
< 10-6 Torr
HiPIMS pulse duration
20 µs
HiPIMS repetition frequency
1 kHz
HiPIMS peak current / voltage
~ 80 A / ~ 450 V
Energy per plasma pulse (typ.)
~ 0.26 J
Averaged power
~ 260 W
LIF diagnostics parameters:
Dye Laser used
Sirah Cobra Stretch
Dye laser linewidth
~ 2.3 GHz
Laser sheet cross-section
70 mm × 2 mm
ICCD camera used
Andor iStar DH740-18F
Optical width / number of pulses
20 ns / 100
Imaging lens
Nikkor 50 mm f/1.4 D
Ti λ excitation/fluorescence
320.58 nm / 508.70 nm
Ti+ λ excitation/fluorescence
314.80 nm / 456.38 nm
Optical filters used
510/10 nm (Ti), 460/10 (Ti+)
supposed to be minor for ions, since their Larmor radius
is in the cm range for our magnetron source.
Similar results, but corresponding to much higher Ar
pressure (80 mTorr) are presented in Fig. 4, where the
different velocity groups of sputtered Ti neutrals are
presented at two different time delays (∆t = 40 and
100 µs) in the off-time. As expected, in this case the highvelocity regions are much more squeezed toward the
target vicinity, which is due to the higher pressure and
thus faster particles thermalization. In spite of the
individual normalization of the density maps in Figs. 3
and 4, the zones corresponding to the high-energy
particles (associated with red colour) are roughly
proportional to the Ar pressure used (~ 3 cm at 5 mTorr
vs. ~ 0.5 cm at 80 mTorr), implying the dominance of the
thermalization processes during this time.
The high-energy particles are still visible at 80 mTorr,
and the asymmetry of the obtained density maps generally
reminds the case shown in Fig. 3 for Ti neutrals. As also
expected, the mentioned asymmetry between the left
P-I-2-4
Fig. 3. DS-LIF density maps for sputtered Ti (a) and Ti+ (b)
taken at 5 mTorr. The laser wavelength and velocity
displacements (∆λ, ∆v) are given in the right upper corners.
Positive ∆v corresponds to blue-shifted atoms. Each map is
normalized individually.
and right parts of the density maps vanishes at higher
time-delays (at 100 µs in this case, see Fig. 4(b)). The
propagation angle for the neutral particles in this case is
found to be close to the one at low pressure.
Another, rather strong difference from the low-pressure
case, is the localization of the high-energy particles closer
to the race-track region, whereas they are much more
focused in the central area at low-pressure. This effect in
our opinion can be explained by (i) the out-of-plane Ti
velocity components, (ii) the depletion of the Ti ground
state energy sublevels [12], and (iii) almost total
ionization of the sputtered material at low pressure. Note,
that the second and third reasons should be connected,
since the depletion of the lower energy sublevel of Ti,
studied in this work, should be strongly related to the
wave of the hot secondary electrons in plasma right after
the pulse, which must also strongly affect the
P-I-2-4
Fig. 4. DS-LIF density maps for sputtered Ti taken at 80 mTorr
at Δt = 40 µs (a) and 100 µs (b). The laser wavelength and
velocity (∆λ,∆ v ) displacements are given in the right upper
corners. Positive ∆v corresponds to blue-shifted atoms. Each
map is normalized individually.
ionization of the sputtered neutral species. This effect
vanishes with increase of the Ar pressure, due likely to
the faster cooling of the secondary electrons in this case,
which do not provoke strong sublevels depletion anymore.
This fact, however, has not been verified experimentally
yet. The first reason suggesting the existence of the strong
out of the image plane velocity components for the
sputtered neutrals might play a certain role in the
explanation as well, since the VDF of the sputtered
species in the low pressure case should retain its (large)
width during a longer time, i.e. inversely proportionally to
the working pressure in the reactor.
The additional Doppler-shift measurements, particularly
those performed in the plane above the race-track of the
target, as well as those comparing more different time
delays (including the plasma on-time) should clarify these
suggestions.
3
4. Conclusions
Time-resolved Doppler-shift laser-induced fluorescence
imaging is applied for two-dimensional characterization
of the physical processes in an Ar-Ti high-power impulse
magnetron sputtering discharge. The results of this study
illustrate several kinetic phenomena related to
propagation of the sputtered neutrals and ionized
particles, as well as those related to their thermalization in
the discharge volume at different working pressures in the
reactor.
Even though the observed results do not change the
known functionality of HiPIMS discharges, they do
clarify significantly the particularities of the particles
energy (velocity) distribution, via the direct discharge
particles visualization approach applied in this work. This
results in much deeper level of understanding of the shortpulsed HiPIMS discharges, which may be used for their
further investigations and improvements.
5. Acknowledgements
This work is supported by Belgian Government through
the ‘Pôle d’Attraction Interuniversitaire’ (PAI, P7/34,
‘Plasma-Surface Interaction’,Ψ). N. Britun is a postdoctoral researcher, and S. Konstantinidis is a research
associate of the Fonds National de la Recherche
Scientifique (FNRS), Belgium.
4
6. References
[1]. U. Helmersson,
M. Lattemann,
J. Bohlmark,
A. Ehiasarian, J.-T. Gudmundsson, Thin Solid
Films, 513, 1, (2006).
[2]. D. V. Mozgrin, I. K. Fetisov, G. V. Khodachenko,
Plasma Phys. Rep. 21, 400, (1995).
[3]. I.K. Fetisov, A.A. Filippov, G.V. Khodachenko, et
al. Vaccum 53, 133, (1999).
[4]. J. T. Gudmundsson, N. Brenning, D. Lundin, and
U. Helmersson, J. Vac. Sci. Technol. A 30, 030801,
(2012).
[5]. N. Britun, T. Minea, S. Konstantinidis, R. Snyders,
J. Phys. D: Appl. Phys. 47, 224001, (2014).
[6]. K. Sarakinos, J. Alami, S. Konstantinidis, Surf.
Coat. Technol. 204, 1661, (2010).
[7]. J. Amorim, G. Baravian, J. Jolly, J. Phys. D: Appl.
Phys. 33, R51–65, (2000).
[8]. S. Mazouffre, Plasma Sources Sci. Technol. 22,
013001, (2013).
[9]. K. Shibagaki, N. Nafarizal, K. Sasaki, J. Appl. Phys.
98, 043310, (2005).
[10]. M. Palmucci, N. Britun, S. Konstantinidis, and
R. Snyders, J. Appl. Phys. 114, 113302, (2013).
[11]. N. Britun, J. G. Han, S.-G. Oh, Appl. Phys. Lett.,
92, 141503, (2008).
[12]. N. Britun et al., Combined optical diagnostics of a
high-power
impulse
magnetron sputtering
discharge, ISSP-2013 Int’l Conf. Proc. (2013,
Kyoto, Japan), p. 198.
P-I-2-4