Driven grounded electrode

As previously mentioned in section 2.3, the DC component of a capacitively coupled plasma can be short-circuited by using a λcable/4 long cable grounded at its end (3.65 m for 13.56 MHz,

see Table 4.4). By doing so, the driven electrode self-bias, normally responsible for accelerating ions, is obviously equal to zero. In case of a conducting electrode, the self-bias is effectively zero and a high plasma potential is expected (hypothesis I). On the other hand, an insulating surface will act as a capacitor and a negative bias can develop on the surface, although the RF generator will indicate 0 V (hypothesis II). Cleaning experiments and plasma measurements were both done in ESCA-1 facility using 13.56 MHz excitation frequency.

The first etching trial was done on a 50 nm thick pure and dense Al2O3 film deposited on

a Mo mirror, with Ar as a process gas at a pressure of 1.5 Pa with 50 W RF power. After 1 h of etching, XPS measurements revealed a pure Mo mirror, free from any Al contaminants. Following this good result, another test with similar conditions was conducted on a metallic coating, namely 10 nm of Rh deposited on a Mo mirror. If the plasma potential would remain as usual (approx. 25 V), the ion energy should be below the damage threshold of Ar on Rh (see Table 2.2), and hence no sputtering of the Rh film was expected. After 40 min of etching in the aforementioned conditions, no more Rh was measured with XPS indicating a total removal of the conducting film, in agreement with a plasma potential increase. Measurements with Langmuir probe on the ETS mock-up and plasma simulations were additionally performed to verify the hypothesis I and II.

A plasma was sustained at RF power equivalent to the one needed to obtain_{−200 V self-bias} in the normal CCP configuration. In the case of the anodized Al electrode, plasma potential similar to the normal case were measured, around 20–30 V. When the electrode was replaced by the conducting version, the estimation of the plasma potential was only possible at low RF power. The Langmuir probe voltage range is in fact limited from_{− 100 to 100 V. The evaluation} of the plasma parameters was therefore restricted to low RF power due to the positive charging of the plasma with a higher power. The evolution of the plasma potential with respect to the input RF power is depicted in Figure 4.9, with experimental values displayed up to a power of

30 W. 0 4 8 12 16 1.00E+019 2.00E+019 3.00E+019 4.00E+019 Ion flux [.m -2 .s -1 ] Position [cm] 10 20 30 40 50 60 70 80 90 60 80 100 120 140 160 180 200 Experimental points Linear Fit Vp [V] Power [W]

Figure 4.9: Ion flux (left) and plasma potential Vp (right) measured along the surface with

Langmuir probe from the center (0 cm) to the edge of the SS ETS mock-up. The discharge was sustained at 13.56 MHz, 0.5 Pa Ar, 80 W RF power for 0 V bias. The graphs include standard error of the mean.

Assuming a linear increase of the plasma potential with increasing RF power, similarly to the evolution of the self-bias in Figure 4.4, a linear fit of the experimental data was performed. The plasma potential extrapolated from this fit corresponding to the RF power requested to obtain −200 V in the normal case, e.g. 80 W, was approximatively 180 V. The same figure also shows that the ion flux along the surface of the mirror is almost constant, indicating a homogeneous etching process.

Plasma simulations were performed by Z. Donko and P. Hartmann from the Wigner Research Centre for Physics of the Hungarian Academy of Sciences in Budapest using particle in cell Monte Carlo code (PIC-MCC). The aim of this collaboration was to simulate the specific case of driven grounded electrode in CCP discharges and to compare the simulation with experimental results obtained in Basel. The method employed for the simulation is thoroughly described in [97]. The simulations were run in a strong asymmetric system, the driven electrode being  84 mm with a counter grounded electrode of  200 mm. The inter-electrode gap was fixed to 114 mm. In the simulations the electrons and Ar+ _{ions are traced from their creation by gas}

phase electron impact ionization or secondary electron emission from surfaces to their absorption on the wall or electrode. The elementary processes included are elastic scattering of electrons on Ar atoms, electron-impact excitation and ionization, and elastic scattering of Ar+ _{ions on}

Ar atoms. Surface processes include absorption and reflection of electrons and SE emission by impinging Ar+ _{ions. In the case of the insulating electrode, a 13.56 MHz harmonic voltage}

waveform with amplitude V0and a uniform self-bias voltage Vbias is allowed to develop. Vbias

is iteratively approximated by the simulation code by finding a value that provides a balance between electron and ion fluxes to the electrode area. For the conducting electrode, Vbias is

forced to be equal to 0 V. The following parameters were used for the simulations, close to the one used experimentally:

• Ar pressure: p = 1 Pa.

• RF (13.56 MHz) amplitude: V0 = 180 V.

• Gas temperature: T = 350 K.

• Electron reflection coefficient: αe= 0.2.

• SE emission yield: γ = 0.1.

4.2. CCP: discharge properties

The main results obtained from the simulations are depicted in Figure 4.10 and are listed in Table 4.5. It should be noted that the vertical scales are not the same for the different configurations. It can be seen that for driven grounded electrode configuration with an insulating substrate (a, b and c), the plasma potential does not increase and a negative bias can develop on the electrode. The ion energy distributions obtained in this case by simulations are almost identical to the one obtained in the normal CCP mode (see Figure 4.5). When the driven grounded electrode is conducting (d, e and f), the plasma potential increases up to 180 V while the electrode stays at 0 V in simulations, similar to what was observed experimentally (see Figure 4.9). The maximum ion energy is spread around 175 and 150 eV for the insulating and conducting case respectively. In simulations, it appears that the ion energy is slightly lower for the driven grounded electrode case and was also confirmed experimentally in section 6.3. Figure 6.11.

Table 4.5: Self-bias and average plasma potential extracted from PIC-MCC simulations.

Electrode Self-bias Plasma potential (V) (V) Conducting − 164 21 Insulating 0 155 Half-insulating/ _{− 52} 126 half-conducting

The performance of RF discharge with driven grounded electrode was investigated for both extreme cases, fully insulating or conducting. An intermediate case, where the surface of the  84 mm electrode is 50 % conducting and 50 % insulating was explored in simulations. In the inner part (insulating) the self-bias is adjusted to reach flux balance of electrons and ions absorbed at the inner electrode, while at the outer ring Vbias= 0 V is set constant. The results

are displayed in Figure 4.10 (h) and (i), and show a charging of the plasma potential to lower values than in the case of the pure conducting electrode while the insulating surface develops a bias voltage. The ion energy distribution, computed for the whole electrode therefore displays two defined energy peaks, one around 170 eV, the other around 120 eV. They respectively correspond to the insulating (Vbias + Vp) and conducting part of the electrode (0 + Vp). The

results obtained from the simulations are, to our knowledge, the first ever showing that different ion energy distributions can be obtained on one single electrode exhibiting different material properties locally. Tests and measurements are still ongoing in Basel to support the simulation results.

(a) (b) (c) Ion d ensity Vp and V bia s

Insulating Conducting Half insulating-half conducting

(e) (d) (f) (g) (h) (i)

Figure 4.10: PIC-MCC simulation of CCP plasma discharge using Ar at 1 Pa, 13.56 MHz excitation frequency. The simulations were done on insulating (a, b, c), conducting (d, e, f ) and half insulating-half conducting (g, h, i) driven grounded electrode. The spatial ion density in m−3 (a, d, g), plasma and electrode potentials in volts (b, e, h) as well as the ion energy distribution on the driven electrode (c, f, i) are shown. The simulation geometry and an illustration of the half insulating / half conducting sample are depicted.

4.2. CCP: discharge properties