Power coupling mode transitions induced by tailored voltage waveforms in capacitive oxygen discharges

The experimental setup is shown schematically in figure 2. A reactor with an inter-electrode gap of L = 2.5 cm was made geometrically symmetric by adding a thick Teflon ring (internal diameter 12 cm, external diameter 19 cm), with a window comprising a glass block 2.5 cm × 2.5 cm × 10 cm to allow optical access. The voltage waveforms were generated using a computer-controlled arbitrary function generator (AFG Tektronix 3150) and a wideband power amplifier (Prana AP32GN310, 0.01–200 MHz, 1 kW) as described previously [36]. The true delivered waveform is monitored by a high voltage probe (Tektronix P5100), and the waveform is corrected for distortion using a feedback loop as proposed in [37]. The intensified charge-coupled device (ICCD) camera (Andor iStar) is synchronized to the AFG, and the phase is varied using a digital delay generator (Stanford Research Systems DG 645). An intensifier gate of the ICCD camera is set to 2 ns or 5 ns depending on the fundamental frequency of the waveform being investigated. We used phase resolved optical emission spectroscopy (PROES) to observe the 844.6 nm line, which corresponds to emission from the O(3p³P) excited state, as a function of position between the two electrodes and phase within the fundamental RF cycle. Under the conditions investigated in this study, where the dissociation degree is expected to be low [38], the O(3p³P) state is predominantly formed through dissociative excitation of molecular oxygen [39] (given by reaction 15 in table 1). Emission at 844.6 nm is isolated using a custom narrow bandpass filter (LOT—QuantumDesign: 844.79 ± 0.95 nm). The excitation rate of the O(3p³P) at each position and phase is derived from the measured emission, using the method proposed in [40]. Collisional quenching of the O(3p³P) excited state by molecular oxygen is accounted for in the calculation of the excitation using the quenching coefficient (9.4 × 10⁻¹⁶ m³ s⁻¹) measured in [41].

Zoom In Zoom Out Reset image size

Table 1.  List of collision processes considered in the simulation.

#	Reaction	Process	References
1	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {{\rm{O}}}_{2}+{{e}}^{-}$	Elastic scattering	[42]
2	${{e}}^{-}+{{\rm{O}}}_{2}({r}=0)\longrightarrow {{e}}^{-}+{{\rm{O}}}_{2}({r}\gt 0)$	Rotational excitation	xpdp1
3	${{e}}^{-}+{{\rm{O}}}_{2}({v}=0)\longrightarrow {{e}}^{-}+{{\rm{O}}}_{2}({v}=1)$	Vibrational excitation	xpdp1
4	${{e}}^{-}+{{\rm{O}}}_{2}({v}=0)\longrightarrow {{e}}^{-}+{{\rm{O}}}_{2}({v}=2)$	Vibrational excitation	xpdp1
5	${{e}}^{-}+{{\rm{O}}}_{2}({v}=0)\longrightarrow {{e}}^{-}+{{\rm{O}}}_{2}({v}=3)$	Vibrational excitation	xpdp1
6	${{e}}^{-}+{{\rm{O}}}_{2}({v}=0)\longrightarrow {{e}}^{-}+{{\rm{O}}}_{2}({v}=4)$	Vibrational excitation	xpdp1
7	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {{e}}^{-}+{{\rm{O}}}_{2}({{a}}^{1}{{\rm{\Delta }}}_{g})$	Metastable excitation (0.98 eV)	xpdp1
8	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {{e}}^{-}+{{\rm{O}}}_{2}({{b}}^{1}{{\rm{\Sigma }}}_{g})$	Metastable excitation (1.63 eV)	xpdp1
9	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {\rm{O}}+{{\rm{O}}}^{-}$	Dissociative attachment	xpdp1
10	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {{e}}^{-}+{{\rm{O}}}_{2}$	Excitation (4.5 eV)	xpdp1
11	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {\rm{O}}{(}^{3}{\rm{P}})+{\rm{O}}{(}^{3}{\rm{P}})+{{e}}^{-}$	Dissociation (6.0 eV)	xpdp1
12	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {\rm{O}}{(}^{3}{\rm{P}})+{\rm{O}}{(}^{1}{\rm{D}})+{{e}}^{-}$	Dissociation (8.4 eV)	xpdp1
13	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {\rm{O}}{(}^{1}{\rm{D}})+{\rm{O}}{(}^{1}{\rm{D}})+{{e}}^{-}$	Dissociation (9.97 eV)	xpdp1
14	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {{\rm{O}}}_{2}^{+}+{{e}}^{-}+{{e}}^{-}$	Ionization	[43]
15	${{e}}^{-}+{{\rm{O}}}_{2}\longrightarrow {{e}}^{-}+{\rm{O}}+{\rm{O}}(3{{\rm{p}}}^{3}{\rm{P}})$	Dissociative excitation (14.7 eV)	xpdp1
16	${{e}}^{-}+{{\rm{O}}}^{-}\longrightarrow {{e}}^{-}+{{e}}^{-}+{\rm{O}}$	Electron impact detachment	xpdp1
17	${{e}}^{-}+{{\rm{O}}}_{2}^{+}\longrightarrow {\rm{O}}{(}^{3}{\rm{P}})+{\rm{O}}{(}^{1}{\rm{D}})$	Dissociative recombination	xpdp1
18	${{\rm{O}}}_{2}^{+}+{{\rm{O}}}_{2}\longrightarrow {{\rm{O}}}_{2}+{{\rm{O}}}_{2}^{+}$	Elastic scattering	[43]
19	${{\rm{O}}}^{-}+{{\rm{O}}}_{2}\longrightarrow {{\rm{O}}}^{-}+{{\rm{O}}}_{2}$	Elastic scattering	[43]
20	${{\rm{O}}}^{-}+{{\rm{O}}}_{2}\longrightarrow {\rm{O}}+{{\rm{O}}}_{2}+{{e}}^{-}$	Detachment	[43]
21	${{\rm{O}}}^{-}+{{\rm{O}}}_{2}^{+}\longrightarrow {\rm{O}}+{{\rm{O}}}_{2}$	Mutual neutralization	[43]
22	${{\rm{O}}}^{-}+{{\rm{O}}}_{2}({{a}}^{1}{{\rm{\Delta }}}_{g})\longrightarrow {{\rm{O}}}_{3}+{{e}}^{-}$	Associative detachment	[44]

The numerical investigation of O₂ plasmas is based on a bounded, one-dimensional in space and three-dimensional in velocity space (1D3v) particle-in-cell simulation code incorporating a Monte Carlo treatment of collision processes (PIC/MCC) [45]. This code is suitable to describe geometrically symmetric plasma reactors, without taking into account 2D effects. The core of the code has been benchmarked against other implementations of the same algorithm within an international collaboration [46]. The reactor geometry is simplified by assuming plane and parallel electrodes with a gap of L = 2.5 cm between them, one driven by a RF voltage waveform composed of a maximum of four consecutive harmonics generated according to equations (1) or (3), while the other electrode is grounded. Three different values of the fundamental frequency are used: 5, 10, and 15 MHz. The gas temperature is fixed at ${T}_{{\rm{g}}}$ = 300 K. Both the electron reflection coefficient at the electrodes and the secondary electron yield are set to zero. The gas pressure is 50 mTorr  ≤ p ≤ 700 mTorr, and the maximum applied peak-to-peak voltage is 400 V. The DC self-bias, η, which develops due to the amplitude or slope asymmetry of the applied waveforms is determined in an iterative way in the simulations, by adjusting its value so as to achieve equal losses of positive and negative charges at each electrode over one period of the fundamental driving frequency [47].

In the simulations, the following charged species are traced: ${{\rm{O}}}_{2}^{+}$, O⁻ and electrons. The collision-reaction model underlying the simulations is the same as that introduced in [26]. In that work a detailed analysis of the numerous sets of processes and related cross sections/rate coefficients proposed in the literature for the numerical description of O₂ plasmas [43, 44, 48, 49, 50] has also been performed. In [26] a thorough test of the model is also provided: good agreement was found between various discharge characteristics (DC self-bias voltage, power absorbed by the discharge, ion flux, and ion flux energy distribution) obtained from PIC/MCC simulations and experiments. The set of elementary processes considered in the model is given in table 1. Further details of the model can be found in [26]. We note the relatively low number of different plasma species and elementary processes included in this model. However, the good agreement between the simulation and experimental results for a wide range of discharge conditions indicates that our model describes reasonably well the complex oxygen discharge under the conditions studied.

The density of oxygen singlet delta ${{\rm{O}}}_{2}({a}^{1}{{\rm{\Delta }}}_{g})$ molecules in the discharge is largely controlled by their surface quenching probability, α. The value of this coefficient is not well known, it shows a large variation for different electrode materials [51, 52]. Our choice of $\alpha =6\times {10}^{-3}$ resulted in the best overall agreement between the experimental and simulation data for the ion fluxes at the electrodes in a previous work [26]. To reveal the influence of α on the spatio-temporal profiles of light emission, computations were carried out over a wide range of this parameter.

All computations have been carried out using a spatial grid with N_x = 200–1000 points and N_t = 2000–10 000 time steps within the fundamental RF period. These parameters have been selected to fulfil the stability criteria of the computational method. The simulation results have been obtained by averaging over 20 000 cycles of the fundamental RF frequency.

In order to study the AAE in O₂ plasmas, driving voltage waveforms of the form given by equation (1), with voltage amplitudes according to equation (2) are used. Three different fundamental frequencies, f₁, are studied (5, 10, and 15 MHz), considering single-(N = 1), dual-(N = 2), triple-(N = 3), and quad-(N = 4) frequency cases at each value of f₁. The phase shifts of all the odd harmonics are set to zero, while for all the even harmonics the phase shifts are set to π, resulting in valleys-type voltage waveforms (see figure 1(a)). The peak-to-peak voltage is kept constant at ${\phi }_{\mathrm{PP}}$ = 400 V for all cases. The neutral gas pressure is set to p = 150 mTorr. Study of the effect of the gas pressure on the excitation dynamics is also carried out for three harmonics (N = 3) by performing measurements and simulations at 50 mTorr as well as at 700 mTorr.

Figure 4 shows the spatio-temporal distribution of the excitation rate of the O(3p³P) excited state derived from PROES measurements (${R}_{{\rm{exc}},{\rm{EXP}}}$) and compared to the rate of dissociative excitation obtained from PIC/MCC simulations (${R}_{{\rm{exc}},{\rm{SIM}}}$). Different values of the base frequency are used: f₁ = 5 MHz, 10 MHz, and 15 MHz, in parts I, II and III, respectively. In each part of figure 4, the experimental results are presented in the first row, while the simulation results are shown in the second row for different numbers of harmonics: N = 1 (first column), 2 (second column), 3 (third column), and 4 (fourth column). All plots show the whole gap between the powered electrode (x = 0 cm) and grounded electrode (x = 2.5 cm) for one period of the given fundamental frequency (${T}_{{\rm{RF}}}=1/{f}_{1}$). The sheath edge positions are shown as white lines in the simulation plots. The spatio-temporal distributions of the electric field obtained from the simulations for the above conditions are displayed in figure 5.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The agreement between the spatio-temporal distribution of the excitation rates obtained from the experiments and those from simulations is very good. The patterns of the plasma emission observed experimentally are well reproduced by simulations for all three values of the fundamental frequency as the applied voltage waveform is changed by increasing the number of harmonics, N. However, the emission is somewhat more widely spread across the gap in the experimental results compared to the simulations. At the lowest fundamental frequency studied here (f₁ = 5 MHz), for N = 1, strong excitation in the plasma bulk is observed. The excitation rate is maximal near the edge of the collapsing sheaths, i.e. the discharge operates in DA-mode. We note the presence of a second local excitation maximum in the bulk when the sheath collapses at either electrode. This specific excitation profile is caused by electrons accelerated by significant electric fields in the bulk plasma. Several local maxima of the electric field in the plasma region can be seen in the spatio-temporal maps of the electric field obtained from simulations for f₁ = 5 MHz fundamental frequency in the first row of figure 5. Such spatio-temporal structure in the electric field and striations in the ion production rate have been recently reported for electronegative CF₄ plasmas in [54]. For N = 2 at f₁ = 5 MHz, unequal maxima of the excitation rates are found at the collapsing sheath edges, with stronger excitation at the side of the grounded electrode. Further increase of N leads to strong excitation at the collapsing sheath edge at the powered electrode, while at the grounded electrode the excitation maximum shifts from the collapsing sheath edge to the expanding sheath edge. Thus, at this low frequency of 5 MHz, the discharge operation mode changes from the DA-mode to a hybrid α-DA mode as the number of harmonics is increased.

At 10 MHz, for N = 1 the discharge operates in DA-mode with strong excitation at the collapsing sheath edges. With 2 harmonics, excitation can be observed at both sides. During sheath expansion at the grounded electrode side, the excitation is significantly stronger compared to the powered electrode side. Under these conditions the discharge operates in α-DA mode. For $N\gt 2$, the discharge operates in pure α-mode characterized by strong excitation at the grounded electrode side. Similar behavior of the excitation dynamics is found for 15 MHz, the highest fundamental frequency studied here: while the discharge operates in α-DA mode at N = 1, increasing the number of harmonics enhances excitation at the grounded electrode during sheath expansion at that side, leading to pure α-mode operation of the discharge. These transitions of the discharge operation mode are well confirmed by the spatio-temporal distributions of the electric field under these conditions (shown in figure 5, second and third rows): for 10 MHz high electric field in the plasma bulk can be seen for N = 1 and 2, while for 15 MHz a high electric field outside the sheath is found only in the case of N = 1. At 15 MHz, for $N\gt 1$, the plasma series resonance effect [55] is also observed.

In figure 6 the effects of the number of harmonics in valley-type voltage waveforms on the DC self-bias, ${{\rm{O}}}_{2}^{+}$ ion flux at the electrodes, mean ion energy at the electrodes, and electronegativity of the discharge are presented for different fundamental frequencies. In panel (a) normalized values of the DC self-bias, $\overline{\eta }=\eta /{\phi }_{\mathrm{PP}}$, obtained experimentally and from PIC/MCC simulations are plotted. The low values of η measured for N = 1 ($| \eta | \lt 10\,{\rm{V}}$) prove that the experimental setup is nearly symmetrical. For $N\gt 1$ a positive DC self-bias is formed as a result of the amplitude asymmetry. η increases with increasing the number of harmonics, both in simulations and experiments. The experimental trends are well reproduced by simulations, even the higher values of the self-bias obtained for the lowest fundamental frequency for $N\gt 2$, compared to the values obtained for higher fundamental frequencies. However, differences between the calculated and measured values of the DC self-bias are found, the values obtained from simulations being higher by a factor of up to 1.73. The highest value of η is obtained at 5 MHz for N = 4 ($\overline{\eta }$ = 27% in the experiment, compared to $\overline{\eta }$ = 42% in the simulation). For $N\gt 2$, the DC self-bias decreases as the fundamental frequency is increased. This is in agreement with previous findings for electronegative CF₄ discharges [27], where the absolute value of the negative DC self-bias, obtained for peaks-type voltage waveforms for N = 4 and constant amplitude asymmetry, becomes much larger for lower base frequencies.

Zoom In Zoom Out Reset image size

In figure 6(b) the simulated ${{\rm{O}}}_{2}^{+}$ ion flux at the powered and grounded electrodes is shown. The flux of ions is higher at the grounded electrode compared to the powered electrode and increases with increasing N, except with a 5 MHz base frequency, where a slight decrease of the ion flux is observed as the number of harmonics is increased. At 15 MHz, the ion flux can be tuned by a factor of 1.5 and 3.5 at the powered and grounded electrodes, respectively, by changing N. At 10 MHz the ion flux remains nearly constant at the powered electrode, while it changes by a factor of about 1.8 at the grounded electrode by increasing the number of harmonics from 1 to 4; at 5 MHz a small change (by a factor of about 0.7) is seen at both electrodes by varying N. The mean ion energy can also be controlled by varying N (see figure 6(c)): an increase by a factor of about 2.5, and a decrease by a factor of about 1.8 can be achieved at the grounded and powered electrode, respectively, by increasing N.

Figure 6(d) shows the electronegativity of the discharge, obtained from the simulations. At 5 MHz the electronegativity increases (from 30 to 45) as N is increased from 1 to 4. At 10 and 15 MHz, the discharge is highly electronegative at N = 1 (with electronegativity of 55 and 42, respectively). With increasing N, a strong decrease of the electronegativity is induced (to 6 at 10 MHz, and to 0.5 at 15 MHz). This change of the electronegativity at 10 and 15 MHz reflects the change of the discharge operation mode discussed above: in case of low electronegativity the discharge operates in α-mode, while at high values of the electronegativity α-DA mode is found.

The effect of gas pressure on the excitation patterns is illustrated in figure 7. In the first row the experimentally observed excitation rate of the O(3p³P) excited state (derived from the 844.6 nm PROES measurements) are displayed, while in the second row the rates of dissociative excitation obtained from the simulations are plotted, for f₁ = 15 MHz. Three different pressures are investigated: 50 mTorr (first column), 150 mTorr (second column), and 700 mTorr (third column), for three harmonics (N = 3); the peak-to-peak voltage amplitude is 400 V in all cases. For all pressures, the excitation maxima occurs at the grounded electrode side. Strong excitation is observed in front of this electrode during sheath expansion. Further excitation at the powered electrode side is seen near the expanding sheath edge, which is enhanced at higher pressures, but remains significantly lower compared to the excitation at the grounded electrode side. Under these conditions excitation induced by electrons near the expanding sheath edge is dominant, thus the discharge operates in the α-mode. However, at the highest pressure (700 mTorr) excitation in the bulk is also observed in the experimental results, but this is not observed in the simulations.

Zoom In Zoom Out Reset image size

The SAE in O₂ plasmas is studied here using the driving voltage waveforms given by equation (3). The same discharge conditions are investigated as for the preceding AAE study: three different fundamental frequencies (f₁ = 5, 10, and 15 MHz) are considered and up to four harmonics ($N\leqslant 4$) are used at each fundamental frequency, f₁. Equation (3) is used here with positive sign to generate voltage waveforms with sawtooth-down shape (see figure 1(b)). The peak-to-peak voltage is kept constant at ${\phi }_{\mathrm{PP}}$ = 400 V for all cases. The neutral gas pressure is set to p = 150 mTorr. A study of the effect of the gas pressure is carried out for N = 3 by performing measurements and simulations at 50 and 700 mTorr.

Figure 8 shows the distribution of the excitation rate of the O(3p³P) excited state compared to the rate of dissociative excitation obtained from PIC/MCC simulations. Different values of the base frequency are used: f₁ = 5 MHz, 10 MHz, and 15 MHz in parts I, II and III, respectively. In each part of the figure, in the first and second rows the experimental and simulation results are shown, respectively. Results obtained for different numbers of harmonics are plotted: N = 1 (first column), 2 (second column), 3 (third column), and 4 (fourth column). All plots show one period of the given fundamental frequency (${T}_{{\rm{RF}}}=1/{f}_{1}$). The sheath edge positions are shown as white lines in the simulation plots. The spatio-temporal distributions of the electric field obtained from the simulations for the above conditions are displayed in figure 9.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The agreement between the measurements and the simulations is very good, especially for the lowest fundamental frequency (f₁ = 5 MHz). However, even at higher f₁, the main features of the observed plasma emission are well reproduced by the simulations for all N. At f₁ = 5 MHz, for N = 1, the discharge operates in DA-mode. High electric fields are observed in the bulk plasma region (see figure 9, first row) which are created to ensure current continuity for low electron density conditions, inducing significant excitation in the bulk plasma and at the collapsing sheath edges. Similarly to the case of O₂ discharges driven by valley-type voltage waveforms generated by using a low fundamental frequency of 5 MHz (see previous section), specific distributions of the bulk electric field and spatial striations in the excitation rates are also found here. Increasing the number of harmonics leads to stronger excitation at the powered electrode side compared to the grounded electrode sheath at times of sheath collapse at both electrodes (N = 2). For $N\geqslant 3$ the excitation is mainly concentrated near the powered electrode collapsing sheath edge. Therefore, at f₁ = 5 MHz, the discharge operation mode is not changed when the number of harmonics is increased for the sawtooth voltage waveforms under the conditions studied here, and the DA-mode is dominant for all N. However, the excitation maximum vanishes at the grounded electrode side as N is increased, and strong excitation builds up at the powered electrode side.

At 10 MHz, for N = 1 the discharge operates in the DA-mode with excitation maxima at the collapsing sheath edges. Increasing the number of harmonics to 2, excitation at the expanding sheath edge at the grounded electrode is generated, with strong excitation at the powered electrode at the same time. Therefore, the discharge operates in hybrid α-DA mode. Further increase of N enhances the excitation near the expanding grounded electrode sheath (N = 3) and the α-mode becomes dominant at N = 4. In this case excitation is also generated at the powered electrode during sheath expansion. In agreement with these findings, a high electric field is observed in the plasma bulk (figure 9, second row) for low N, while at high values of N the electric field is low in the bulk. A similar scenario is seen at f₁ = 15 MHz (see figure 8, part III, and figure 9, third row), where the transition from α-DA mode to α-mode takes place at lower N.

In figure 10 the DC self-bias, ${{\rm{O}}}_{2}^{+}$ ion flux at the electrodes, mean ion energy at the electrodes, and electronegativity of the discharge are presented as a function of the number of harmonics at different fundamental frequencies. Normalized values of the DC self-bias, $\overline{\eta }=\eta /{\phi }_{\mathrm{PP}}$, obtained experimentally and from PIC/MCC simulations are plotted in panel (a) of figure 10. The experimental and simulation results are in good agreement. For f₁ = 5 MHz, a positive DC self-bias is formed for all N; η increases with increasing N up to 3, followed by a decrease of η at N = 4. Similar trends were seen in CF₄ discharges excited by sawtooth-down waveforms at 5.5 MHz base frequency, where a small decrease in the DC self-bias is observed when increasing N from 3 to 5 [33]. The dependence of the DC self-bias on N for f₁ = 10 MHz is similar to that observed for f₁ = 5 MHz. However, at this higher base frequency η changes sign and drops below 0 for N = 4. For f₁ = 15 MHz, negative DC self-bias values are obtained, and η decreases with increasing N.

Zoom In Zoom Out Reset image size

Figure 10(b) shows the ${{\rm{O}}}_{2}^{+}$ ion flux at the powered and grounded electrodes obtained from the simulations. The flux of ions is higher at the powered electrode, except for a 15 MHz base frequency, where the ion flux is slightly higher at the grounded electrode for $N\gt 1$. At f₁ = 5 MHz the ion flux remains nearly constant at the powered electrode as N is increased, while the mean ion energy decreases by a factor of about 1.4 (figure 6(c)). At the grounded electrode the ion flux increases by a factor of about 4.5 as N is increased from 1 to 4 at f₁ = 5 MHz, while the mean ion energy remains nearly constant. Such separate control of the ion properties is not possible at higher f₁: at 10 MHz only small changes in the ion flux and mean ion energy are induced at both electrodes by changing N; at f₁ = 15 MHz both the ion flux and mean ion energy increases by increasing N.

In figure 10(d) the electronegativity of the discharge is shown as a function of N for different fundamental frequencies. For f₁ = 5 MHz, the electronegativity increases from 31 to 47 as N increases from 1 to 4. For f₁ = 10 MHz, the electronegativity decreases from 56 to 37 with increasing N. In case of f₁ = 15 MHz, increasing N causes the electronegativity to drop from 40 at N = 1 to 7 at N = 4. At high electronegativity, the discharge operates in the DA-mode: at f₁ = 5 MHz, where the discharge is highly electronegative for all N, excitation occurs in the bulk and at the collapsing sheath edge (figure 8, part I); for f₁ = 10 MHz and 15 MHz, a transition from DA- to the α-mode can be observed as the electronegativity of the discharge decreases with increasing N (figure 8, part II and III).

Figure 11 presents the effect of the gas pressure on the spatio-temporal excitation. The experimental results are shown in the first row, while in the second row the rates of dissociative excitation obtained from the simulations are plotted for N = 3 and f₁ = 10 MHz at three different pressures: 50 mTorr (first column), 150 mTorr (second column), and 700 mTorr (third column). The peak-to-peak voltage amplitude is 400 V in all cases. The agreement between the simulations and the experiment is acceptable, with a significant difference appearing between the excitation patterns only at the highest pressure (700 mTorr), where an excitation maximum at the powered electrode side is seen in the experiment, while in the simulation the strongest excitation is located in front of the grounded electrode. At 50 mTorr, the discharge operates in the DA-mode and strong excitation at the powered electrode collapsing sheath edge is found. At the higher pressure of 150 mTorr, the excitation maximum is shifted to the grounded electrode side. Strong excitation in the plasma bulk and at the collapsing sheath edge are also found. Under these conditions the discharge operates in a hybrid α-DA mode. At the highest pressure studied here (700 mTorr) excitation due to both α- and DA-modes are found.

Zoom In Zoom Out Reset image size