-
This content has been downloaded from IOPscience. Please scroll
down to see the full text.
Download details:
IP Address: 129.78.233.211This content was downloaded on
09/11/2014 at 10:12
Please note that terms and conditions apply.
Mass spectrometry diagnostics of short-pulsed HiPIMS
discharges
View the table of contents for this issue, or go to the journal
homepage for more
2013 J. Phys. D: Appl. Phys. 46 215201
(http://iopscience.iop.org/0022-3727/46/21/215201)
Home Search Collections Journals About Contact us My
IOPscience
-
IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS
J. Phys. D: Appl. Phys. 46 (2013) 215201 (11pp)
doi:10.1088/0022-3727/46/21/215201
Mass spectrometry diagnostics ofshort-pulsed HiPIMS
dischargesMaria Palmucci1 Nikolay Britun1 Tiago Silva1 Rony
Snyders1,2and Stephanos Konstantinidis1
1 Chimie des Interactions Plasma-Surface, CIRMAP, UMONS Research
Institute for Materials Scienceand Engineering, Universite de Mons,
Place du Parc 23, B-7000 Mons, Belgium2 Materia Nova Research
Center, Parc Initialis, Avenue N. Copernic 1, B-7000 Mons,
Belgium
E-mail: [email protected]
Received 7 December 2012, in final form 2 April 2013Published 9
May 2013Online at stacks.iop.org/JPhysD/46/215201
AbstractTime-resolved mass spectrometry (MS) study of a
high-power impulse magnetron sputteringdischarge (HiPIMS) operating
in the short-pulse regime (5 s) at 1 kHz of the repetitionfrequency
is undertaken. Several time-resolved effects related to both Ti+
and Ar+ ion energydistribution functions (IEDF) are found. In
particular, the dynamics of both the low- (05 eV)and high- (530 eV)
energy regions presented in Ti+ IEDFs is clarified. According to
ourresults the sputtered and ionized Ti arrive at the virtual
substrate position in the form of twowaves, with the first one
representing the high-energy Ti+, and the second one responsible
forthe low-energy Ti+. An essential decrease in the population of
the energetic Ti+ group isobserved at the moment of the arrival of
the low-energy group, which is explained by thecharge exchange
processes, as well as by the refilling process afterwards. The role
of Armetastables presumably generated at the end of the plasma
pulse for further Ti ionization isstressed as well. The
time-averaged IEDFs for Ti+ and Ar+ are additionally analysed.
Theeffective ion temperatures are calculated for these species for
the above-mentioned energyranges. A considerable increase in the
effective ion temperature for the high-energy Ti+ isfound, which is
directly related to the elevation of the high-energy tail in the
time-averagedIEDFs with increasing discharge voltage. Possible
mechanisms for such an elevation arediscussed.
(Some figures may appear in colour only in the online
journal)
1. Introduction
High-power impulse magnetron sputtering (HiPIMS) is aphysical
vapour deposition (PVD) process which has attractedconsiderable
interest over the last few decades since itsinvention by Mozgrins
group [1], and its further improvementsby Kouznetsov et al in 1999
[24]. HiPIMS plasmas arecharacterized by a high electron densityne
(up to 1019 m3 [5]),which in turn leads to a higher ionization
degree of the sputteredparticles (typically above 50%) compared
with that observedin conventional direct current magnetron
sputtering (DCMS)discharges (about 1% [6]) where ne is typically up
to 1016 m3[5]. Different discharge parameters such as pulse width,
peakcurrent, target voltage to name but a few allow controlling
theion-to-neutral ratio, the metal-to-argon ion ratio, as wellas
the energy of the ions bombarding a growing film. Theseparameters
are of key importance since they strongly influence
the physico-chemical characteristics of the deposited
materialssuch as their density, crystallinity, microstructure, etc
[7, 8].
In HiPIMS, which is an example of a fast time-changingplasma
discharge, the use of time-resolved diagnostics iscrucial in order
to gain information about dynamics of thedischarge. Following this
purpose, several groups haveemployed time-resolved Langmuir probes
for characterizationof HiPIMS discharges. The corresponding results
generallydemonstrate the existence of several groups of electrons
withdifferent kinetic energies, i.e. cold and hot electrons
whichare generated at different parts of the HiPIMS period [9,
10].In fact, at the beginning of the on-time, the ions impingingthe
cathode result in secondary electron emission from itssurface.
These hot electrons gain their kinetic energy upto the level
equivalent to that of the cathode potential andgive rise to a
rather high electron temperature (Te) duringthe discharge ignition.
As the discharge current increases,
0022-3727/13/215201+11$33.00 1 2013 IOP Publishing Ltd Printed
in the UK & the USA
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
Te decreases and exhibits a bimodal distribution suggestingthe
dominance of argon species at the early stages of thedischarge,
which is followed by the dominance of metallicspecies due to gas
rarefaction in the cathode vicinity [10, 11].Similar observations
have also been obtained by means ofoptical emission spectroscopy
(OES) when a decrease in thesignal related to argon species
correlates with an increase inthat related to the metallic species
[6, 12]. It is to be noted thatsuch a variation of Te throughout
the period induces a variationof the plasma potential (Vp) from a
few to about ten volts.
Among the other plasma diagnostics techniques, massspectrometry
(MS) is proven to be a powerful tool to monitorthe ion dynamics in
HiPIMS discharges by examining theion energy distribution function
(IEDF). However, if thebasics of the IEDF for DCMS and
bipolar-pulsed magnetronsputtering (BPMS) discharges are rather
well known [1316],our knowledge is sparse in the domain of HiPIMS
discharges,which remain unclear in many aspects [3, 17]. For
example, atypical IEDF obtained in HiPIMS energy-wise reveals
severalgroups of particles corresponding to different ionized
specieswith low-, mid- and high-energies, which in addition
areproduced at different stages of the discharge period [5]. Itis
accepted that the low-energy peaks corresponding to argonand
metallic ions, which generally dominate intensity-wise,are
attributed to the thermalized ions reaching the groundedmass
spectrometers orifice with kinetic energies close to theplasma
potential [3, 1720]. The peak corresponding to thehigh-energy
sputtered particles is generally attributed to (i)back-reflected
ions from the target surface and/or (ii) metallicatoms ejected from
the target with the so-called SigmundThompson (ST) distribution
[21], and further ionized (e.g. byelectron impact) in the dense
plasma region [20, 22].
Several hypotheses have been presented in the literatureto
explain the origin of the peak of the IEDF appearing atintermediate
energies (around 38 eV) [20, 23]. In fact, thecorresponding ion
energy group has not always been detectedfor both argon and
metallic ions. According to Hecimovicet al [23], the second peak of
the Cr+ IEDF (i) might beattributed to Cr2+ ions which undergo a
charge exchange withargon atoms, or (ii) might be the result of a
wave of sputteredparticles due to the sharp drop in both Vp and Te
observedduring the discharge ignition. In the other works, the
mid-energy peak of Ar+ IEDF has been attributed to the
partiallythermalized high-energy particles, as well as to the
presenceof gas rarefaction [17, 24]. On the other hand, Greczynski
andHultman [20] suggested that the energy of the mid-energy ionsin
the Ar+ and Ti+ IEDFs (which is roughly equal to 34 eV)is close to
that of the main ion peak (0.51.3 eV). This peakin turn might be
assigned to the thermalized ions acceleratedin the
plasma-spectrometer sheath during the variation of Vpthroughout the
period, as discussed in [20]. Such a mid-energyion peak has also
been reported before where its existence forC+ IEDFs has been
explained by the positive values of Vp,which increase additionally
during the on-time [25].
It is especially important to note that most of the worksdevoted
to the ion dynamics in HiPIMS discharges deal withthe pulse
duration ( ) ranged from 70 to 200 s or higher. Insuch long-pulse
discharges, gas rarefaction, which favours
Figure 1. Schematic diagram of the experimental setup.
self-sputtering, has been pointed out to be partially
responsiblefor generally low deposition rates in HiPIMS compared
withDCMS [26]. It has also been reported that some of HiPIMSpower
supplies are not capable of maintaining a constant targetvoltage
during the entire pulse resulting sometimes in twodischarge regimes
(dc-like and HiPIMS-like), and furthercomplicating the
understanding of HiPIMS discharges [20].
Summarizing the above-mentioned argumentation, inorder to
skilfully monitor the ignition of a HiPIMS discharge,this work
deals with the short-pulse ( = 5 s) HiPIMS case,when the target
current does not reach a plateau, and the plasmais still in a
current rise phase, as described in [27]. Havingthe goal to enhance
the level of understanding of the plasmadynamics in HiPIMS
discharges, both energy- and time-resolved MS measurements of an
ArTi HiPIMS discharge areconsidered and discussed in detail in the
following sections.
2. Experimental
All the experiments were carried out in a cylindrical
stainlesssteel chamber, 450 mm long, with 250 mm diameter, pumpedto
a residual pressure of about 106 Torr by a turbomolecularpump
combined with a membrane pump. The pumpingspeed was adjusted
through a closed-loop throttle valve tokeep the Ar pressure
constant (5 mTorr or 0.67 Pa). Argon(99.999% purity) was introduced
into the chamber at a constantflow rate of 40 sccm by means of a
mass flow controller(Brooks Instrument 5850 TR). Figure 1
represents a diagramof the experimental setup where the position of
the massspectrometer is shown relatively to the magnetron
cathode(target). The magnetron system itself consists of a
planarbalanced magnetron source with a circular titanium target
(Ti,99.99% purity), 100 mm in diameter. The pulsed discharges
inthis study were sustained using a high-power pulse
generatordescribed elsewhere [28]. The used power supply ensured
ashort rise time of the discharge current, as well as generally
2
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
Figure 2. Typical magnetron voltage and current waveforms in
aHiPIMS discharge operated at 5 mTorr of Ar pressure. The
pulseduration is 5 s, the repetition frequency is 1 kHz.
Table 1. ToF calculated according to the m/z ratio and the
kineticenergy of ions.
Kinetic energyIon m/z range (eV) ToF (s) ToF average (s)Ar+ 40
125 84.4384.10 84.3Ti+ 48 130 92.8392.03 92.4
stable and reproducible performance of the HiPIMS discharge.The
power supply without a pre-ionization option is utilizedin this
work. The pulse frequency () was varied in the range0.510 kHz in
order to keep the average power (P ) constantat 300 W, which is
defined according to the following relation:
P = 1
0
I (t)U (t) dt (1)
where I (t) and U(t) are the discharge current and
targetvoltage, respectively. I (t) was measured by a
high-currentprobe (Tektronix AM503B) and U(t) by a high-voltage
probe(Tektronix P5100). The typical I (t) and U(t)
waveformsrecorded by an oscilloscope (Tektronix TDS2024B)
arepresented in figure 2. In the experimental results shown in
thefollowing sections the target voltage during the pulse was
keptat the level of about 1.2 kV, unless mentioned otherwise.
Basedon the measured waveforms, U(t) remains nearly
constantthroughout the entire pulse duration.
MS measurements were performed with an energy-resolving mass
spectrometer HAL 7 EQP 1000 (HidenAnalytical, UK). The grounded
orifice of the instrument waspositioned at 80 mm in front of the
target surface, i.e. in thevirtual substrate position (see figure
1). Taking into account theorifice geometry, the solid angle was
estimated to be equal toaround 102 srad. For time-resolved
measurements, the massspectrometer circuit was triggered by a
transistor-transistorlogic (TTL) pulse sequence generated at the
output of theutilized power supply. The data were recorded during
theentire period which implies that the measurement time wasmuch
longer in the time-resolved mode compared with thetime-averaged
one. A 2 s gate width was used during thepulse. The gate width was
set at 20 s, 120 s after the pulse
Figure 3. Ti+ (a) and Ar+ (b) IEDFs measured as a function of
timein the ArTi HiPIMS discharge during its entire period.
and at 50 s after 300 s. The dwell time was set to collectall
the data during 1000 consecutive pulses. In order to beindependent
from varying the pathway of ions in the massspectrometer, their
time-of-flight (ToF) from the orifice to thedetector was
additionally calculated considering the equipmentparameters
according to [3]. The ToF values for each sort ofions are given in
table 1 for lowest and highest values of thecorresponding kinetic
energy range. As the variations in ToFdo not exceed 1%, all the
data for a given sort of ion werecorrected by its average ToF
value.
3. Results and discussion
3.1. Time-resolved mass spectrometry results
A general image of the short-pulse HiPIMS dischargewhich is a
focus of this study can be obtained from the 3Drepresentation of
the time-resolved IEDFs shown in figure 3.As one can observe, the
studied ions are represented by twogroup of particles, one at
energies of about 1.5 eV (referredto as Ar+low, Ti
+low below), whereas the second representing
the range of about 1030 eV (Ti+high) (figure 3(a)). The
high-energy group of Ti+ appears abruptly in the MS spectra aftera
definite time delay (t) and quickly disappears afterwards.At the
same time, a similar high-energy group of Ar+ is muchless numerous
compared with the one of Ti+, as presentedin figure 3(b). The
presence of high-energy particles maypoint out the influence of the
energetic ionized particles (Ti+and Ar+), which might be
accelerated and elastically reflected
3
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
Figure 4. Energy mass spectra of Ti+ measured in the
HiPIMSdischarge at different t during its period.
from the target surface [29], differently for Ar+ and Ti+.
Moreremarks on this effect are given below.
More particularities of the described time-resolvedevolution
become available after the comparison of theTi+ IEDFs taken at
different t , as shown in figure 4,which confirms the previous
observations. As we can seeadditionally, the position of the second
broad maximum maydrift by several eV sometimes demonstrating the
splitting ofthe maximum (referred to as peak 2 and peak 3 in this
work).
A detailed analysis of the time evolution of both peakpositions,
as well as the particles density associated with them,etc, is given
below. The corresponding data are presented infigure 5. The maxima
found experimentally were fitted by theGaussian curves and further
analysed in a time-resolved way.
Positions of the energy maxima. As shown in figure 5(a),the
energy corresponding to the low-energy group of particlesevolves
similarly in time for both Ar+ and Ti+. The visiblediscrepancies at
the beginning of the plasma period arerelated to the uncertainties
of the peaks fitting at low timedelays. The evolution of the Ar+
low-energy peak positionis often associated with the plasma
potential (Vp). It isobserved that the position of the intensity
maximum of Ar+lowchanges throughout the period, which is confirmed
by thetime-resolved Langmuir probe measurements [10, 11].
Theseresults, however, have to be interpreted properly taking
intoaccount the particularities of each plasma diagnostics tool.
Infact, Langmuir probe measurements give the instantaneousVp values
which match the pulse shape, whereas in MS, the Vpvalues are
shifted and somewhat extended in time out of theplasma pulse. The
discrepancy between both techniques mayalso come from the
difference between the electrons and ionsmobility [30].
The data in figure 5(a) corresponding to the 1224 eVenergy range
represent the second and third (when visible)energy maxima found in
our measurements. These maximaappear in figure 3(a) at t 27 s, in
the form of a singlepeak, whereas the third maximum becomes
distinguishableafter t 50 s. The positions of these energy maxima
havea trend to be gradually separated as a function of time,
having
Figure 5. Time evolution of the position (a), value (b) and the
areaunder the curve (c) corresponding to three energy maxima
visible infigure 4, representing low- and high-energy Ti+. Two
waves of thesputtered Ti+ particles coming to the MS detector after
a certaindelay time are visible. Ar+ time evolution is given for
comparison.
final energies of about 12 eV and 23 eV at the end. However,due
to the large width of these peaks, as well as their proximity,the
described maxima are considered together in the frameworkof this
study, especially taking into account their similar timeevolution
(shown in figures 5(b) and (c)).
Values of the energy maxima. The time evolution of the
peakmaxima found based on the corresponding fit is representedin
figure 5(b). This quantity, however, does not representthe number
of particles that can be associated with the foundenergy peaks and
is given for an illustrative purpose. The areasunder the curves
corresponding to the particle population fora chosen energy range
are more suitable parameters for thedischarge characterization, as
described below.
Populations of different energy groups of species. The
totalamount of species (in the arbitrary units, though) presentedin
the HiPIMS discharge at a certain moment of time andcorresponding
to a certain energy range can be representedby the area under the
curve in a measured IEDF. The timeevolution of this parameter for
the species considered in thisstudy is given in figure 5(c). It
should be noted that sometime corrections should be made in order
to understand the
4
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
real timing based on this figure, namely the position of
thedetector (80 mm from the target) should be taken into account,as
well as the fact that the increase in the electron densityand the
density of the sputtered species normally happens atthe end of the
on- or at the right beginning of the off-time inHiPIMS [9, 23].
Several important observations can be madebased on the
time-resolved results presented in figure 5(c).(a) There are two
(high- and low-) energy groups of Ti+
participating in the resulting time-resolved shape of theionized
sputtered species coming to the detector.
(b) These two groups have different arrival times, as well
asdifferent growth times at the moment of their
detection,representing two waves of Ti+, as indicated in figure
5(c).
(c) The arrival time for the first Ti+ wave is about t =t1 27 s,
whereas for the second one it is aboutt = t2 38 s.
(d) The first Ti+ wave increases abruptly at the
beginningshowing a total growth time of about t2t1 10 s,afterwards
it starts to decay. The growth time for thesecond wave is equal to
t3t2 32 s.
(e) The decay for the high-energy group of Ti+
(high-energygroup) is nearly exponential. Its overall
measurabledensity drop is more than three orders of magnitude.
(f) At the same time, the low-energy group of Ti+ reveals
yetanother broad maximum at t 400500 s. Finally, itspopulation
returns back to the background value detectedat the on-time
beginning (106 counts in figure 5(c)).
(g) The argon ions are detected much earlier (at tAr 12
s)revealing much longer growth time (about 80100 s).
(h) The total growth of the Ar+ population accounts for
aboutthree orders of magnitude.
(i) The background values (e.g. at t = t0) of the Ar+density are
about one order of magnitude lower than thatdetected for Ti+.
3.2. Analysis of the time-resolved resultsIn this section, the
time-resolved data shown above areanalysed. This is first of all
related to the time behaviour ofthe Ti and Ar ions during the
HiPIMS discharge period, assummarized in the observations (a)(i)
made previously.
Relation to the ST energy distribution. The presence of
twogroups of the sputtered species in the discharge with
verydifferent mean energies (1.5 eV versus 1223 eV) pointsout the
differences between a classical sputtering process,when the energy
distribution of the sputtering species canbe described by a ST
distribution, and the HiPIMS process.The ST distribution mainly
characterizes the DCMS-likesputtering, and can be presented in the
form [21]
dNdE
E(E + Eb)
3 , (2)
where E is the energy of a sputtered particle, and Eb is
thesurface binding energy of the sputtered cathode.
A comparison between the ST distribution calculated forTi (Eb is
taken to be 4.9 eV [21]) and the Ti+ IEDFs measured
Figure 6. Comparison between the normalized ST
distributioncalculated for Ti sputtered atoms (Eb = 4.9 eV), and
the Ti+ IEDFsmeasured in HiPIMS in this work at t = 30, 60 and 100
s(mutually normalized in the low-energy range).
in HiPIMS at differentt is presented in figure 6. In spite of
thefact that the ST distribution describes neutral atoms,
whereasthe presented IEDFs are related to Ti+, the IEDFs
determinedin this work coincide well with the ST distribution in
the energyrange of about 02 eV. This fact confirms that the Ti+low
grouporiginates from the low-energy sputtered Ti (
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
Figure 7. Ti+ IEDFs measured at two different target voltages
Uapplied during the HiPIMS pulse. The mean ion energies
Emeancalculated by equation (3) are given for comparison; t = 44
s.
Ti+. Figure 7 shows that Emean corresponding to the high-energy
Ti+ changes roughly proportionally toU , supporting thehypothesis
of the back-reflected ions. These ions, after beingaccelerated in
the sheath, should have an energy proportionalto the target voltage
before transmitting it to the otherrelatively slow discharge
species by collisions. According tothe additional calculation based
on the time-resolved IEDFsgiven in figure 3, the total population
of the high-energy Ar+appears to be about 30 times less than that
of Ti+ at themoment of their maximum appearance in the discharge
(att 45 s). This value is comparable to the differencein the
background ion populations for these species beforethe plasma pulse
(observation (i), section 3.1). Assumingthe presence of rarefaction
in the discharge, the amount ofback-reflected Ar+ (during the whole
pulse duration) mightfurther decrease compared with Ti+, confirming
our suggestionfor the high-energy tails of the observed IEDFs.
Followingthese considerations, the third energy peak sometimes
visiblein the measured IEDFs (see also [31]) might be due to
theback-reflection of the double-charged Ti ions (also presentin
HiPIMS), since the energy corresponding to the thirdmaximum (when
it is visible) is always approximately twice ashigh as the energy
of the second one (24 eV versus 13 eVsee figure 5(a)). These
arguments, however, require additionalverifications.
Based on our measurements, the high-energy Ti+ havethe mean
energy Emean equal to about 15 eV. In the firstapproximation, the
net velocity of these species is about8 km s1, giving the flight
time to the detector (80 mm formthe target) equal to about 10 s.
However, the actual arrivaltime measured in this work is about 27 1
s (see figure 5).Assuming that most of the sputtered species are
produced at theend of the 5 s pulse, the final flying time is
reduced to 22 s.This value is still about twice as large as that
calculated solelybased on the particle energy. Since the mean
energy valueof these species measured at 80 mm from the target
alreadyaccounts for a possible thermalization during the flight,
wefinally arrive at the conclusion that this group of ions
originates
at t 22 s (arrival) 10 s (flight time) = 12 s, i.e. afterthe end
of the HiPIMS pulse3.
It should also be noted that the growth time of the first
Ti+wave at the detector position is about 10 s (see figure
5(c)),i.e. twice longer than the pulse duration. This fact
additionallypoint out certain energy dissipation during the flight
for thisgroup of species. This is also true for the second Ti+
wave.Since the initial time profile for the sputtered species (and
itsduration) is unknown, and it may not coincide with the
HiPIMSpulse, the role of the energy dissipation during the flight
timecannot be determined precisely.
If the low-energy component in the IEDFs taken atdifferent t
follows the ST distribution, the intensity of thehigh-energy one
changes essentially, as shown in figure 6.According to figure 5(c),
the population of the high-energyTi+ surpasses the low-energy
population between time t1 andt2, nearly approaching the level of
Ar+ (at t = 30 s), anddecreases afterwards. The same rate of
decrease for peaks 2and 3 implies the same dissipation kinetics for
these high-energy species.
Low-energy Ti+. The low-energy Ti ions have an averageenergy of
1.5 eV, which decays gradually down to about0.7 eV at the end of
the plasma off-time, where these speciescan be considered as
thermalized [23]. At the beginning ofthe off-time the measured
energy of 1.5 eV corresponds to thenet velocity of 2.5 km s1. This
value is in good agreementwith the previous observations in HiPIMS
made using opticalspectroscopy [12], and Langmuir probe
measurements [32].
The measured flight time for this group of species isabout 32 s,
which is in very good agreement with theexperimental results
(flight time t25 s 38 s5 s =33 s). This conclusion is also
consistent with both STdistribution predicting the velocity of the
sputtered Ti particlesto be about 3 km s1, and with the numerous
measurements ofthis velocity presented elsewhere [3337].
Low-energy Ar+. Based on our measurements, the Ar ions aremainly
distributed within the low-energy peak (see figure 3(b)),having the
average energy close to that of Ti+low. The timeevolution of the
population density for these species have tobe described as
well.
First, the measured background level of Ar+ is about 10times
lower than that of Ti+ low, as shown in figure 5(c). Thisis
presumably due to the presence of relatively cold electrons(Te <
1 eV) in the discharge till the end of the HiPIMS off-time [27],
which are still capable of ionizing the sputtered Ti(ionization
threshold is 6.8 eV), whereas their energy might betoo low for Ar
ionization (ionization threshold is 15.76 eV).The low Ar+
population might also be partially related to therarefaction in
this time interval.
Second, as mentioned above, the growth of Ar+ densityis much
slower than that of both groups of Ti+ and takes3 This energy, as
well as the estimated net velocity are given here forillustrative
purpose. Theoretically, having very different energies, the
speciesfrom the whole high-energy region in the found IEDF should
arrive to thedetector at different time delays, which does not
happen in our case, however(figure 3(a)). The observed
contradiction happens likely due to the net velocityof the detected
species may be different from that determined by their
kineticenergy.
6
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
Figure 8. Ratio between the Ar and Ti electron impact
ionizationrates calculated for different Te based on the Maxwellian
EEDF.Zone I corresponds to the Te range typically reached in
HiPIMSduring the off-time, whereas zone II represents possible Te
values atthe end of the pulse. Inset: three normalized EEDFs
calculated atdifferent Te along with the electron impact ionization
cross sectionsfor Ar and Ti [42]. The overlapping interval (grey
filling) isproportional to the ionization rate Qion.
a much longer time. The moment of time when the Ar+population
starts to increase should be assigned to the arrival ofthe
electrons non-confined in the targets vicinity resulting inthe Ar
ionization wave observable by the mass spectrometer.This suggestion
is supported by the growth of the electrondensity at the end of the
on-time reported in [38, 39], as wellas by the velocity of the bulk
gas excitation wave recentlyfound by OES, which stays in the range
724 km s1 dependingon the bulk gas [12]. Giving the Ar ionization
arriving timeequal to 7 s, this velocity is estimated to be equal
to80 mm/7 s = 11 km s1 in our case.
Population of Ti+ compared with Ar+. In terms of the arrivaltime
for different groups of particles, as well as their
relativepopulations, the above-mentioned results can be illustrated
bycalculations of the relative ionization rate for Ar and Ti.
Takingthe expression for ionization rateQion in the form (e.g. [40,
41])
Qion =
Eion
ion(E)
2Eme
f (E) dE, (4)
where Eion is the electron ionization threshold, ion(E) is
theelection impact ionization cross section, E is the
electronenergy, f (E) is the EEDF, a curve representing the
ratiobetween the Ar and Ti ionization rates can be obtained. Sucha
curve is shown in figure 8 where a Maxwellian EEDF isassumed in the
discharge [27], and the values of ion(E) for Arand Ti are taken
from [42]. At the end of the HiPIMS pulse, Temay increase up to 26
eV due to a wave of energetic electrons[12, 27] which are
especially favourable in the presence ofrarefaction [4]. Afterwards
the electrons cool down to typically0.51 eV, as a result of
numerous collisions with the sputteringmaterial [38, 39, 43]. This
fact is represented by two Te zones
in figure 8, with zone I representing the cold, and zone II
hotelectrons in the discharge. As we can see, the cold
electronscannot ionize Ar efficiently, resulting in an Ar/Ti
ionizationrate ratio of about (15) 105. Provided the Ar density
inthe reactor is about 34 orders of magnitude higher than thatof Ti
[44], the Ar+/Ti+ ion density ratio is 0.1, which is veryclose to
the result shown in figure 5(c).
Furthermore, the Ti+ high population in its maximumnearly
reaches the one of Ar+ (at t 30 s). Sucha proximity implies that
the ratio between the Ar and Tiionization rates is nearly equal to
the Ti/Ar density ratio inthe discharge at that moment of time.
Assuming the Ti/Ardensity ratio is about 103 (based on the
time-resolved opticalabsorption measurements performed in [44]), we
obtain thatTe is not surpassing the limit of about 2 eV
corresponding tothe beginning of zone II in our case (see figure
8). In spite ofsuch a small Te change, this conclusion is in good
agreementwith several observations performed in this pressure range
inHiPIMS [9, 38, 39].
Timing for different groups of particles. In this section,we
will try to give the explanation for the time-resolvedphenomena
summarized in figure 5. As we can see, the waveof ionization
protruding with a speed of about 11 km s1 isfaster than the
velocity of both low- and high-energy Ti afterthe sputtering. In
addition, it should be taken into account thatin the used balanced
magnetron source a dense electron cloudconfined in the target
vicinity (at 3 cm [27]) is formed, aswas verified additionally by
OES for our magnetron source(not shown). As a result, the sputtered
Ti undergoes ionizationwhen flying through this region. Based on
the estimatedarrival and flight times for the measured species, as
wellas on the change of their population in time, a simplifiedchart
reproducing the main processes taking place in the areaadjacent to
the detector can be built. Presented in figure 9, thischart shows
mainly the wave of electrons, both high- and low-energy groups of
Ti+, as well as Ar+ coming to the detectoralong with possible
plasma kinetic reactions which might beresponsible for a certain
process.
The most critical reactions for the found kinetics of thestudied
discharge representing the ionization, recombination,charge
exchange processes and so on are summarized below:
Ti + e Ti+ + 2e (R1)
Ar + e Ar+ + 2e (R2)Ar + e Armet + e (R3)
Ti+high + Tilow Tihigh + Ti+low (R4)Ti+high + Ar Ti+low + Armet
(R5)Ti+high + Ar Tihigh + Ar+ (R6)
Tilow + Armet Ti+low + Ar + e (R7)X+ + e X, (X = Ar, Ti)
(R8)
These reactions tend to illustrate only the main
processesobserved in this study by MS measurements in HiPIMS. Fora
better understanding of the contribution of each reaction
7
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
Figure 9. Schematic chart illustrating the population change for
themain groups of particles at the position of the detector (80 mm
awayfrom the cathode) as a function of t . The population
behaviourcorresponds to the log-vertical scale, representing about
three ordersof magnitude of the total change in the case of Ar+ and
Ti+. Time t4is 400500 s. The reactions with higher importance are
markedin bold.
(as well as the other possible reactions),
correspondingcalculations should be undertaken accounting for the
reactionrates as well as the other quantitative data, which is
outside thescope of this paper.
A few additional remarks should be made based onfigure 9. First,
since the electron confinement zone locatedclose to the target is
supposed to be the main ionization regionin the HiPIMS discharges
with a balanced magnetron source,the evolution of the electron
density sketched in figure 9 shouldbe associated with the
non-confined electrons only. In addition,since the high-energy
groups of sputtered Ti cross the electronconfinement zone before
the low-energy ones, the high-energyTi gets ionized first, which in
turn results in the essentialcooling of the electrons in the
confinement zone [12, 45]. Suchan electron cooling (i.e. the Te
decrease) should result in aweaker ionization of the following
groups of the sputteredatoms passing the confinement zone, which
have lower energyand longer arrival time.
Second, the production of the Ar metastables (Armet)in the
discharge volume should be mentioned, since it isthe essential
factor for Penning ionization of the low-energygroup of Ti at the
position of the detector. This process issupposed to be essential
after refilling starts (roughly after50100 s according to [23]). In
fact, Armet can particularlybe produced as a result of collisions
with the energetic Ti+(reaction (R5)) and be used for further Ti
ionization (reaction(R7)). According to our recent study involving
resonant opticalabsorption [44], performed under somewhat different
HiPIMSconditions, Armet are mainly produced right after the
dischargepulse, resulting in further generation of Ti metastables
(seefigure 10), and in the slow increase of the Ti+ population
due
Figure 10. Time evolution of the absolute density of Ti+, as
well asTi and Ar metastables during a HiPIMS pulse measured by
opticalabsorption spectroscopy. The Ar pressure is 20 mTorr, the
pulseduration is 20 s, distance from target is 50 mm; adapted from
[44].
to the Penning ionization (reaction (R7)). This is also
observedin this study in the time interval between t3 and t4 (see
figure 5and figure 9).
Third, the population exchange between the high- andlow-energy
groups of Ti+ during the time interval t2t3, visiblein figure 5(c)
and figure 9, should be stressed. Such a processis likely to happen
as a result of the efficient charge exchangebetween these two
groups of Ti+, after the low-energy (andrather weakly ionized) Ti+
group arrives at the detector vicinity.Such an exchange results in
about one order of magnitudedecrease in the high-energy Ti+ and in
roughly a similarincrease in the Ti+low population. The further
decrease (withthe same rate) of the Ti+high group after time t3 is
likely due tothe combination of several factors, in particular
represented byreactions (R4) and (R5).
Finally, the presence of the described charged
particlesthroughout the entire HiPIMS period is very important
asthey play a role of pre-ionization for the next pulse, sothe
additional pre-ionization is not required for most of theHiPIMS
discharges with comparable repetition rates [46, 47].Furthermore,
the metal ions are transported in a medium whichis still filled
with electrons and ions. This situation might,therefore, enhance
the ion transport as suggested previouslyin [6].
3.3. Time-averaged mass spectrometry results
Having rather high-repetition rate as well as
short-pulseduration, the considered HiPIMS discharge cannot
becharacterized rigorously by the time-resolved
temperaturecorresponding to the heavy particles presented in the
gasmixture (e.g. ions or neutrals) at each moment of time.The
reason for this is that the thermalization time constant(defining
the thermalization rate of heavy particles in thedischarge, and
thus the characteristic time of thermalization)for the
translationaltranslational energy relaxation, being inthe range of
several hundred s at 5 mTorr, is roughly one
8
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
Figure 11. Results of the time-averaged MS measurements of
Ti+(a), (b), and Ar+ (c), (d) in the HiPIMS discharge. The
changesinduced by the different applied voltages U result in
changing theeffective Ti+ temperatures shown for Ti+ (b). Inset:
the effective Ti+temperatures calculated based on the Maxwellian
fit of the datashown in (b).
Table 2. Effective ion temperatures calculated as a result
ofMaxwellian distribution fits based on the actual
dischargeparameters in the corresponding energy ranges.
Teffi (eV)Ar+low Ti
+low Ti
+high
U (V) Ip (A) Jp (A cm2) 59 eV 59 eV 1230 eV800 48 0.15 0.89 1.61
4.661000 122 0.39 0.76 1.27 5.801100 166 0.53 0.80 1.31 6.231200
192 0.61 0.81 1.28 6.67
order of magnitude higher than the HiPIMS pulse
durationconsidered in this work [48, 49]. This difference might
beeven larger taking into account the rarefaction further
reducingthe Ar pressure. Since the gas refilling itself may
takeabout 100 s [50], which is also much longer than the
pulseduration, the only time-averaged effective ion temperature(T
effi ) perceptible during a relatively long time in HiPIMScan be
considered.
The results of the time-averaged argon and titanium
IEDFmeasurements are given in figure 11. The ion temperatures
aredetermined for two energy ranges, as shown in figure 11(a).Since
the total measured IEDF has a bi-Maxwellian nature,the Maxwellian
fits are applied for each energy range,corresponding to the low-
and high- energy group of Ar+or Ti+ (see figure 11), similarly to
the considerations describedin the previous sections. The obtained
T effi for the low-energyspecies are close for Ar+ and Ti+, whereas
T effi found for thehigh-energy Ti+ are much larger. Moreover, in
the last caseT effi reveals the dependence on the applied voltage
(see table 2and the inset in figure 11(a) and (d)). This fact can
be observedin figure 11(b) where the high-energy tail of Ti+
extends as thetarget voltage increases from 800 to 1200 V.
Considering the last result, a suggestion about the increasein
the total amount of the sputtered species in the whole energyrange
[51] with increase in the voltage is barely acceptablein our case,
since only the tail of the IEDF is affected. Atthe same time,
considering the DCMS discharges, a simulatedenergy distribution
function for the back-reflected Ar+ showsthe appearance of a
high-energy component increasing from50 to 200 eV while the
incident ion energy increases from200 to 800 eV [29]. In the HiPIMS
discharges, since thegas rarefaction can be rather strong due to
the high targetcurrent values compared with DCMS [4, 52], the
sputteredmetal density may exceed several times the gas density in
thetarget vicinity. In such a situation, the sputtering
probabilityof titanium target by Ti+ (self-sputtering) is enhanced
ascompared with DCMS processes [26, 53], which may furtherpromote
the effect of back-reflected ions. On figure 11(a),the
time-averaged contribution of Ti+ with energy higher than30 eV is
23 orders of magnitude lower compared with the low-energy peak
intensity meaning roughly that during the wholeHiPIMS period less
than 1% of the particles have energy above30 eV. This result is in
good agreement with the probability ofthe so-called back-attracted
Ti+ () estimated as 0 0.3in [53], and with the SRIM simulations
showing that only3% of atoms are back-reflected from the target
[54]. Itis noteworthy that the SRIM simulations do not take
intoaccount the post-ionization of these reflected neutrals in
theplasma. Moreover, the mass spectrometer orifice only acceptsthe
ions within a rather small solid angle (102 srad) whichmay explain
the discrepancy between the experimental andtheoretical data.
It is known that a drop in the deposition rate in HiPIMScan be
observed when both the ionization rate of the metallicatoms and the
self-sputtering processes are favourable [26, 53].Therefore, an
increase in U , and hence Ip, leads to an increasein the
self-sputtering process and consequently to the loss ofthe
deposited material at the substrate. In this study the
dataextracted from the x-ray fluorescence analysis (XRF,
BrukerS4-Pioneer spectrometer) performed on titanium films show
a39% decrease of the amount of deposited material when U
isincreased from 800 to 1200 V. This effect, widely observedin the
literature [52, 55], provides additional proof that theobserved
tail elevation effect might be attributed to the back-reflected
ions in our case.
Another possible explanation for the found high-energytails in
the time-averaged IEDFs can be related to theazimuthally
accelerated ions, as was mentioned above anddescribed in [31] and
in [56], where the high-energy tails (up to80 eV) were observed
when a mass spectrometer was locatedperpendicular to the target
surface. The origin of the foundenergy tails in these works is
explained by the existence ofthe so-called modified two-stream
instability (MTSI) drivenby the relative drifts between the fast
electrons and slow ionsin the presence of a magnetic field. This
effect leads to theappearance of azimuthal forces caused by the
azimuthal currentand might be virtually responsible for the
observed elevationof the high-energy tails, along with the
back-reflected Ti+, asmentioned in the previous sections.
9
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al
4. Summary
In this work, the plasma dynamics of a short-pulse
HiPIMSdischarge, during sputtering of a Ti target in Ar gas,
isinvestigated using MS diagnostics.
As a result of the time-resolved measurements thepresence of
high- and low-energy groups for Ti ions isconfirmed. The time
evolution of the average energy andthe population of these groups
of species are analysed. Asa result of this analysis, two waves of
the incoming Ti+ arefound. The first wave represents the sputtered
and ionizedTi with a mean energy Emean of about 15 eV, whereas
thesecond one corresponds to the low-energy Ti+ group (Emean 1.5
eV) arriving at the virtual substrate position with
certainretardation.
The estimated arrival times for the various groups ofspecies
correlate with the measured mean energies of thesespecies, whereas
their somewhat extended growth time at themoment of detection
compared with the pulse duration pointsout a definite energy
dissipation of the arriving species. Lowerpopulation of the
low-energy Ti+ group at the moment of arrivalmight be mainly
explained by cooling of the confined electronsin the target
vicinity during the on-time, as a result of collisionswith the
energetic sputtered Ti passing this region previously.
The population of the energetic group of the detected Ti+decays
fast (nearly exponentially) after reaching its maximum,whereas the
second low-energy Ti+ reveals the additional widemaximum
corresponding to the middle of the plasma off-time. Several
particularities, such as the population exchangebetween two groups
of Ti+ found after the arrival of the low-energy Ti+, as well as
the further slow increase in the low-energy Ti+ population are
clearly detected. While the firsteffect may happen due to the
charge exchange mechanism(reactions (R4) and (R5)), the second
might be due to theformation of Ar metastables, detected previously
in HiPIMSby time-resolved absorption spectroscopy. The influence
ofAr metastables, however, is supposed to be weak, taking
intoaccount rather minor elevation of the Ti+low population in
thetime interval between t3 and t4. At the same time, furtherfast
decay of the energetic Ti+high population, until it
becomesundetectable, is attributed mainly to the collisions with
theslow particles, e.g. Ar, especially at the moment when
refillingeffect starts (reactions (R4)(R6)).
The time-averaged IEDF measurements performed in thesame
discharge for Ti+ and Ar+ indicate that the effective
iontemperatures for low-energy Ti+ and Ar+ are close to each
other(1.5 eV), whereas this temperature is voltage-dependent forthe
high-energy group of Ti+. The last effect is directly relatedto the
elevation of the high-energy Ti+ population (a tail ofIEDF) with
voltage, which is explained by a combination ofseveral factors in
this work.
Finally, the role of the back-reflected ions in the
observedeffects needs to be mentioned. The proportionality between
thecalculated mean ion energy in the Ti+ IEDF and the
dischargevoltage U (figure 7) points out a possible contribution
ofthese energetic particles to the energy transfer, which is
alsoconfirmed by the high-energy IEDF tail elevation in the
time-averaged case. On the other hand, the population of
high-energy Ar+ found in this study is roughly 30 times less
than that of Ti+ at t 45 s where these populationstake their
maxima. The presence of rarefaction inherent inHiPIMS, however,
additionally supports the determined lowAr+ population. The exact
measurement of the rarefaction (e.g.in terms of Ar or Ar+ density
changes during the dischargeperiod) was not the goal of this work.
Such measurementsmay be performed, e.g. by time-resolved optical
absorptionspectroscopy in HiPIMS in the future.
Acknowledgments
The authors would like to acknowledge the financial supportof
the Belgian Government through the IAP program (P06/08)as well as M
Michiels and D Walrave (Materia Nova) forthe technical contribution
to this work. N Britun andS Konstantinidis are postdoctoral
researcher and researchassociate of the Fonds National de la
Recherche Scientifique(FNRS), respectively.
References
[1] Mozgrin D V, Fetisov I K and Khodachenko G V 1995
PlasmaPhys. Rep. 21 400
[2] Kouznetsov V, Macak K, Schneider J M, Helmersson U andPetrov
I 1999 Surf. Coat. Technol. 122 290
[3] Bohlmark J, Lattemann M, Gudmundsson J T, Ehiasarian A
P,Aranda-Gonzalvo Y, Brenning N and Helmersson U 2006Thin Solid
Films 515 1522
[4] Horwat D and Anders A 2010 J. Appl. Phys. 108 123306[5]
Bohlmark J, Gudmundsson J T, Alami J, Latteman M and
Helmersson U 2005 IEEE Trans. Plasma Sci. 33 346[6]
Konstantinidis S, Ganciu M, Ricard A, Dauchot J P and
Hecq M 2006 J. Appl. Phys. 99 13307[7] Petrov I, Barna P B,
Hultman L and Greene J E 2003 J. Vac.
Sci. Technol. A 21 S117[8] Samuelsson M, Lundin D, Sarakinos K,
Bjorefors F,
Walivaara B, Ljungcantz H and Helmersson U 2012 J. Vac.Sci.
Technol. A 30 031507
[9] Pajdarova A D, Vlcek J, Kudlacek P and Lukas J 2009
PlasmaSources Sci. Technol. 18 022008
[10] Poolcharuansin P and Bradley J W 2010 Plasma Sources
Sci.Technol. 19 025010
[11] Poolcharuansin P, Liebig B and Bradley J W 2010 IEEE
Trans.Plasma Sci. 38 3007
[12] Hala M, Viau N, Zabeida O, Klemberg-Sapieha J E andMartinu
L 2010 J. Appl. Phys. 107 043305
[13] Martin N, Santo A M E, Sanjines R and Levy F 2001
Surf.Coat. Technol. 138 77
[14] Kadlec S, Quaeyhaegens C, Knuyt G and Stals L M 1997
Surf.Coat. Technol. 97 633
[15] Bradley J W, Backer H, Aranda-Gonzalvo Y, Kelly P J
andArnell R D 2002 Plasma Sources Sci. Technol. 11 165
[16] Kim Y M, Chung Y M, Jung M J, Vlcek J, Musil J andHan J G
2005 Surf. Coat. Technol. 200 835
[17] Hecimovic A, Burcalova K and Ehiasarian A P 2008 J. Phys.D:
Appl. Phys. 41 095203
[18] Misina M, Bradley J W, Backer H, Aranda-Gonzalvo Y,Karkari
S K and Forder D 2002 Vacuum 68 171
[19] Vlcek J, Kudlacek P, Burcalova K and Musil J 2007
Europhys.Lett. 77 45002
[20] Greczynski G and Hultman L 2010 Vacuum 84 1159[21] Thompson
M W 1968 Phil. Mag. 18 377[22] Kadlec S, Quaeyhaegens C, Knuyt G
and Stals L M 1997 Surf.
Coat. Technol. 89 177
10
-
J. Phys. D: Appl. Phys. 46 (2013) 215201 M Palmucci et al[23]
Hecimovic A and Ehiasarian A P 2009 J. Phys. D: Appl. Phys.
42 135209[24] Hecimovic A and Ehiasarian A P 2010 J. Appl.
Phys.
108 063301[25] Poolcharuansin P and Bradley J W 2011 Surf. Coat.
Technol.
205 S307[26] Christie D J 2005 J. Vac. Sci. Technol. A 23
330[27] Gudmundsson J T, Brenning N, Lundin D and Helmersson U
2012 J. Vac. Sci. Technol. A 30 030801[28] Ganciu M, Hecq M,
Konstantinidis S, Dauchot J P,
Touzeau M, de Poucques L and Bretagne J 2004 WorldPatent No
WO2005/090632
[29] Matsui H, Toyoda H and Sugai H 2005 J. Vac. Sci. Technol.
A23 671
[30] Grill A 1993 Cold Plasma in Materials Fabrication(New York:
IEEE Press) p 18
[31] Lundin D, Larsson P, Wallin E, Lattemann M, Brenning N
andHelmersson U 2008 Plasma Sources Sci. Technol.17 035021
[32] Gylfason K B, Alami J, Helmersson U and Gudmundsson J T2005
J. Phys. D: Appl. Phys. 38 3417
[33] Britun N, Han J G and Oh S-G 2008 Appl. Phys. Lett.92
141503
[34] Britun N, Han J G and Oh S-G 2008 Plasma Sources
Sci.Technol. 17 045013
[35] Dullni E 1984 Nucl. Instrum. Methods Phys. Res. B 2 610[36]
Husinsky W, Betz G and Girgis I 1984 J. Vac. Sci. Technol. A
2 698[37] Goehlich A, Niemoeller N and Doebele H F 2000 Phys.
Rev. B
62 9349[38] Gudmundsson J T, Sigurjonsson P, Larsson P, Lundin D
and
Helmersson U 2009 J. Appl. Phys. 105 123302[39] Vetushka A and
Ehiasarian A P 2008 J. Phys. D: Appl. Phys.
41 015204[40] Fantz U 2006 Plasma Sources Sci. Technol. 15
S137[41] Zhu X-M and Pu Y-K 2010 J. Phys. D: Appl. Phys. 43
403001[42] National Institute for Fusion Science. Atomic and
Molecular
Numerical Databases http://dpc.nifs.ac.jp/aladdin
[43] Ehiasarian A P, Vetushka A, Hecimovic A andKonstantinidis S
2008 J. Appl. Phys. 104 083305
[44] Britun N , Palmucci M , Konstantinidis S and Snyders
R,Time-resolved evolution of the sputtered species groundstate
densities in a HiPIMS discharge Proc. 11th Int. Symp.for Sputtering
(ISSP-2011) (Kyoto, Japan 68 July 2011)p 198
[45] de Poucques L, Imbert J-C, Boisse-Laporte C, Bretagne
J,Ganciu M, Teule-Gay L and Touzeau M 2006 PlasmaSources Sci.
Technol. 15 661
[46] Vasina P, Mesko M, Ganciu M, Bretagne J, Boisse-Laporte
C,De Poucques L and Touzeau M 2005 Europhys. Lett. 72 390
[47] Vasina P, Mesko M, Imbert J C, Ganciu M, Boisse-Laporte
C,De Poucques L, Touzeau M, Pagnon D and Bretagne J 2007Plasma
Sources Sci. Technol. 16 501
[48] Fridman A and Kennedy L 2011 Plasma Physics andEngineering
2nd edn (New York: Taylor and Francis) p 134
[49] Bray C A and Jonkman R M 1970 J. Chem. Phys. 52 477[50]
Lundin D, Brenning N, Jadernas D, Larsson P, Wallin E,
Lattemann M, Raadu M A and Helmersson U 2009 PlasmaSources Sci.
Technol. 18 045008
[51] Bunshah R F 1994 Handbook of Deposition Technologies
forFilms and Coatings-Science, Technology and Applications2nd edn
(Park Ridge, NJ: Noyes Publications) p 273
[52] Sarakinos K, Alami J and Konstantinidis S 2010 Surf.
Coat.Technol. 204 1661
[53] Brenning N, Huo C, Lundin D, Raadu M A, Vitelaru C,Stancu G
D, Minea T and Helmersson U 2012 PlasmaSources Sci. Technol. 21
025005
[54] The Stopping and Range of Ions in Matter
softwarewww.srim.org
[55] Nouvellon C, Michiels M, Dauchot J P, Archambeau
C,Laffineur F, Silberberg E, Delvaux S, Cloots R,Konstantinidis S
and Snyders R 2012 Surf. Coat. Technol.206 3542
[56] Poolcharuansin P, Liebig B and Bradley J W 2012
PlasmaSources Sci. Technol. 21 015001
11
1. Introduction2. Experimental3. Results and discussion3.1.
Time-resolved mass spectrometry results3.2. Analysis of the
time-resolved results3.3. Time-averaged mass spectrometry
results
4. Summary Acknowledgments References