Optical emission spectroscopy was used to analyze the very-low-frequency cyclic evolution of the electron energy and density caused by repetitive formation and loss of dust nanoparticles in argon plasmas with pulsed injection of hexamethyldisiloxane (HMDSO, [CH3]6Si2O). After elaborating a Boltzmann diagram for Ar high-lying levels and a collisional-radiative model for Ar 2p (Paschen notation) states, temperatures characterizing the low- and high-energy parts of the electron population were calculated. Relative electron densities were also estimated from relative line emission intensities. Both temperatures increase when the dust occupation increases, and then decrease when dust is lost. The opposite trend was observed for the electron density. Such cyclic behaviors of the electron energy and electron density in the HMDSO-containing plasmas are in good agreement with the evolution processes in dusty plasmas, in which the formation of negative ions followed by an electron attachment on the surfaces of the nanoparticles is a critical phenomenon driving dust growth.

A combination of time-resolved optical emission spectroscopy measurements and collisionalradiative modeling is used to investigate the phenomena occurring over multiple time scales in the frequency domain of a low-pressure, axially asymmetric, capacitively coupled radiofrequency (RF) argon plasma with pulsed injection of hexamethyldisiloxane (HMDSO, Si 2 O(CH 3) 6). The collisional-radiative model developed here considers the population of argon 1s and all ten 2p levels (in Paschen's notation). The presence of HMDSO in the plasma is accounted for in the model by quenching of the argon 1s states by species generated by plasma processing of HMDSO, including HMDSO-15 (Si 2 O(CH 3) 5), acetylene (C 2 H 2) and methane (CH 4). Detailed analysis of the relative populations of Ar 2p states reveals cyclic evolutions of the electron temperature, electron density and quenching frequency that are shown to be linked to the kinetics of dust formation in Ar/HMDSO plasmas. Penning ionization of HMDSO and its fragments is found to be an important source of electrons for the plasma maintenance. It is at the origin of the cyclic formation/disappearance of the dust cloud, without attenuation of the phenomenon, as long as the pulsed injection of HMDSO is sustained. The multi-scale approach used in this study further reveals the straightforward relation of the frequency of HMDSO pulsed injection, in particular the HMDSO duty cycle, with the frequency of dust formation/ disappearance cycle.

The quality of the films obtained by magnetron sputtering depends on numerous parameters, including the energy of the ions impinging on the substrate. The energy distribution functions of Ar and Ag ions during magnetron sputtering of a silver target in rf argon plasmas are hereby reported. Measurements were carried out by plasma sampling mass spectrometry at (i) various bias voltages on the surface of the target at constant pressure and (ii) various operating pressures at constant bias voltage. A distinct high-energy tail is observed for the sputtered silver ions (ionized in the plasma) in comparison to the argon gas ions. The results indicate that the sputtered Ag atoms are not completely thermalized by collisions with background gas atoms over the range of experimental conditions investigated. To confirm such assertion, a model has been developed for the ejection of Ag atoms from the target, their transport in the gas phase, and their acceleration in the sheath at the surface of the mass spectrometer. Since sputtering occurs at low impinging ion energies, the energy distribution function of the extracted atoms cannot be represented by the usual Sigmund–Thomson distribution. It is rather assumed to be characterized by a bi-Maxwellian distribution, with one population related to the direct “classical” sputtering and the other one to indirect “2-step etching.” During the transport of Ag neutrals, both ionization and thermalization processes are considered. Finally, the rf sheath near the entry of the mass spectrometer oscillates at a period close to the transit time of the ions passing through it. This induces a complex energy gain also implemented in the model. An excellent agreement between the latter and experimental measurements is obtained. The results are used to probe the effect of the bias voltage and pressure on the fitting parameters, namely, the dc and rf components of the voltage drop in the sheath, the mean energy of the sputtered atoms, and the relative importance of the sputtered populations.

Plasma sampling mass spectrometry (PSMS) has been carried out to study the fragmentation kinetics of hexamethyldisiloxane (HMDSO) in a low-pressure, axially asymmetric argon rf discharge designed for the growth of nanocomposite thin films through a hybrid PVD/PECVD method. Experiments have been conducted with a pulsed injection of HMDSO over a 5-s period. Plasma conditions have been chosen to favor formation and disappearance of dust occurring in cycles of a few hundred seconds. The dissociation degree of HMDSO and the relative intensities of HMDSO-related fragments are reported and analyzed regarding these two specific time-scales. PSMS showed that formation of dust increases HMDSO dissociation. The same result can be deduced from the particle balance equation of HMDSO using the electron density and temperature obtained from optical emission spectroscopy as the only input parameters. For HMDSO, electron-impact dissociation is the dominant loss pathway over diffusion and recombination on the reactor walls. Small C x H y compounds and H 2 are mostly generated from surface recombination mechanisms and lost by electron-impact dissociation. K E Y W O R D S organosilicon precursors, physical vapor deposition, plasma diagnostics, plasma-enhanced chemical vapor deposition 2 of 12 | GAROFANO ET AL. F I G U R E 7 Temporal variation of the fragment-to-parent molecule PSMS ratios (m/z 2-133 / m/z 147) during the formation and disappearance of the dust cloud. Standard conditions. PSMS, plasma sampling mass spectrometry GAROFANO ET AL.

No abstract