Computational investigation of dual-frequency power transfer in capacitively coupled plasmas

Summary form only given. Dual frequency capacitively coupled plasmas provide the microelectronics fabrication industry flexible control, high selectivity and uniformity. The spatial variation of the phases, magnitude and wavelength of the high frequency (HF) rf bias will affect electron density, electron temperature, sheath thickness and ion transit time through the sheath. These variations ultimately affect the ion energy and angular distributions (IEADs) to the substrate which are of critical importance for anisotropic etching or deposition. To optimize the separate control of rates of ionization and IEADs, the HF should be significantly different than the low frequency (LF), which results in the LF being few MHz. For classical sinusoidal rf biases applied on the same electrode, the HF/LF harmonic currents can be distinguished by their Fourier transforms. Recently, nonsinusoidal bias waveforms are being applied in etching recipes to control etching speed and selectivity, which then complicates separating the HF from LF since both now have high harmonic contents. In this paper, we report on a computational investigation of the rf power absorption, power coupling control and IEADs in a CCP resembling those industrially employed with dual-frequency biases both applied to the wafer substrate. The Hybrid Plasma Equipment Module (HPEM) was employed to predict plasma properties and obtain the harmonic contributions of voltage waveforms applied to the same electrode. The operating conditions are 20-50 mTorr in pure Ar and Ar/C4F8/O2 gas mixtures under with 2 MHz + 60 MHz rf biases. The ratio of the HF/LF power can be used to control plasma density, and provide extra control for the width and energy of the IEADs.