Pulsed Cl2/Ar inductively coupled plasma processing: 0D model versus experiments

Abstract

Comparisons between measurements and spatially-averaged (0D) simulations of low-pressure Ar and Cl2 pulsed-plasmas in an industrial inductively coupled reactor are reported. Our analysis focuses on the impact of the pulsing parameters (frequency f, duty cycle dc) on the chemical reactivity of the plasma and on the ion fluxes to the walls. Charged particle densities and ion fluxes are highly modulated when the plasma is pulsed at 1 kHz < f < 20 kHz. In rare gas Ar plasmas, the ion flux rise time is short (50 μs), therefore the dc has almost no influence on the ion flux value during the pulse. By contrast, in molecular electronegative Cl2 plasmas, both the value and rise/decay time of the ion flux during the on and off-periods depend strongly on the dc. This is because in Cl2 both the plasma chemistry and electronegativity depend on the dc. During the off-period, the electron density drops much faster than the negative ion density, leading to a large increase in plasma electronegativity. A minimum afterglow time (75 µs) is required for an ion-ion plasma to form and for the sheath to collapse, exposing the walls and wafers to a negative ion flux. The positive ion flux is 3 to 10 times smaller in Cl2 than in Ar for the same operating conditions. In contrast with charged species, the radical (Cl) kinetics are slow and thus the radical density is hardly modulated for f > 1 kHz. However, the dc strongly influences the Cl2/Cl density ratio and is an excellent knob for controlling the plasma chemical reactivity: the higher the dc the higher the Cl density. The trends and quantities in the 0D simulation are in close agreement with experiments. This proves the capacity of global models to reproduce the fundamental features of pulsed plasmas in simple chemistries and to assist the development of pulsed processes.