Emission, thermocouple, and electrical measurements in SF6/Ar/O2 SiC etching discharges

Abstract

In SiC etching plasma devices, we have recorded plasma emission from Ar, F, and O atoms in SF6/Ar/O2 rf discharges as a function of pressure, input power, and mixture fraction. At fixed power, the emission intensities rise nearly linearly with increasing pressure between 100 and 300 mTorr; with pressure increases to 600 mTorr, the emission intensity rolls off due to the increase in collisional de-excitation. At fixed pressure, Ar and O atom emission shows a similar functional dependence on input power with a roll off at the higher powers due to decreasing reduced electric field strength (E/n, where n denotes the number density). In contrast, the F atom emission increase with increasing power is nearly linear. This reflects the fact that F atoms are produced by dissociative attachment of SF6 (for lower E/n conditions) in addition to direct electron impact dissociation. Electrical measurements, with a variable interelectrode gap discharge, indicate that the electric field to pressure (E/p) ratio does drop with increasing input power. Thermocouple measurements show that the ground electrode temperature increases with increasing power. The dissociative attachment of SF6 increases with increasing temperature as well. The SiC etch rate increases nearly linearly with input power up until the polymer buildup becomes the rate limiting process. At fixed pressure, the Ar emission from the 750 nm line decreases with increasing additions (up to 10%) of O2. This is due to resonantly enhanced quenching of the 4p (13.5 eV) manifold by oxygen atoms. In marked contrast, the F atom emission intensity increases suggesting F atom production by neutral species chemistry. Over the pressure (100–600 mTorr) and power (20–60 W) range studied, the Ar 750 nm emission line serves as a good actinometer for the 704 nm F line provided that there is not a high degree (or changing degree) of O2 dissociation. Resonant deactivation of the 750 nm line through collisional interaction with O atoms, can make the 750 nm line problematic. Under such conditions our previous work indicates that the Ar 641 nm line can provide an alternative actinometer. The excited state of the 641 nm transition lies above the O atom ionization limit making it immune from resonant quenching.