Experimental investigation of standing wave effect in dual-frequency capacitively coupled argon discharges: role of a low-frequency source

Abstract

It is well known that the plasma non-uniformity caused by the standing wave effect has brought about great challenges for plasma material processing. To improve the plasma uniformity, a low-frequency (LF) power source is introduced into a 100 MHz very-high-frequency (VHF) capacitively coupled argon plasma reactor. The effect of the LF parameters (LF voltage amplitude ϕL and LF source fL) on the radial profile of plasma density has been investigated by utilizing a hairpin probe. The result at a low pressure (1 Pa) is compared to the one obtained by a 2D fluid-analytical capacitively coupled plasma model, showing good agreement in the plasma density radial profile. The experimental results show that the plasma density profile exhibits different dependences on ϕL and fL at different gas pressures/electrode driven types (i.e., the two rf sources are applied on one electrode (case I) and separate electrodes (case II)). At low pressures (e.g., 8 Pa), the pronounced standing wave effect revealed in a VHF discharge can be suppressed at a relatively high ϕL or a low fL in case I, because the HF sheath heating is largely weakened due to strong modulation by the LF source. By contrast, ϕL and fL play insignificant roles in suppressing the standing wave effect in case II. At high pressures (e.g., 20 Pa), the opposite is true. The plasma density radial profile is more sensitive to ϕL and fL in case II than in case I. In case II, the standing wave effect is surprisingly enhanced with increasing ϕL at higher pressures; however, the center-high density profile caused by the standing wave effect can be compensated by increasing fL due to the enhanced electrostatic edge effect dominated by the LF source. In contrast, the density radial profile shows a much weaker dependence on ϕL and fL in case I at higher pressures. To understand the different roles of ϕL and fL, the electron excitation dynamics in each case are analyzed based on the measured spatio-temporal distributions of the electron-impact excitation rate by phase resolved optical emission spectroscopy.