Effect of H2 Plasma Conditioning on Anodized Al Chamber and Quartz Discs

In this work, two projects related to semiconductor integrated circuit processing are investigated: (1) H2 plasma conditioning of anodized Al walls of a high density remote plasma source; (2) H2 plasma erosion and roughening of quartz discs.
In the first project, experiments were performed in a high power-density (10 – 40 W/cm3), purely inductive plasma in an anodized-Al coated, transformer coupled toroidal plasma (TCTP) source at a pressure of 0.8 Torr. Optical emission spectroscopy was employed to monitor the time-resolved evolution of species from the walls. 4%H2/Ar or 3%N2/9%H2/Ar plasmas, alternated with Ar, O2/Ar or NF3/Ar “cleaning” plasmas were investigated. In Ar plasmas, the H-to-Ar intensity ratio decayed to base line in a stretched exponential manner over ~0.5-1.5 min, reflecting out-diffusion of hydrogen from the prior hydrogen-containing plasma exposure, with larger signals observed after longer exposure to the hydrogen-containing plasma. Compared to Ar plasmas, the amount of evolved hydrogen increases up to ten-fold in NF3/Ar plasmas, while it decreases five-fold in O2/Ar plasmas with no dependence on the duration of the hydrogen-containing plasma. It was shown that formation of HF and OH in NF3/Ar and O2/Ar plasmas does not significantly affect H Balmer- emission intensity, and that the enhancement or suppression is due to F and O impingement on and diffusion into the anodized Al. It is proposed that F reacts with H in the layer to for HF, which diffuses out of the film, while O reacts with diffusing H and binds it as immobile AlOH. A time-dependent 1-D diffusion model was developed, and reproduces most of the observed effects with a H and/or H2 diffusion coefficient that increases as a function of distance from the surface, ascribed to the columnar, porous nature of anodized Al. The second project presents a study of the erosion of SiO2 in the same TCTP. At 0.5Torr, quartz samples were exposed to plasma densities of 1-3 x 1013cm-3 and H atom temperatures of 4000-8000K. Short (e.g. 1 min) H2/Ar plasma exposure followed by long (9 min) plasma-off cool-down periods resulted in higher etching rates and increased erosion, compared to long (e.g. 1 hr) continuous plasma operation, and was ascribed to the higher substrate temperatures reached with continuous plasma operation and a negative dependence of the etching rate on temperature. When exposure to H2/Ar plasmas were alternated with O2/Ar plasmas and plasma–off periods, the etching rate was reduced to near-zero and surface roughness was much reduced. N2/Ar plasma treatments were less effective in reducing the etching rate, while surface roughness was nearly eliminated. A proposed mechanism involves penetration of H below the surface and insertion into Si-O-Si linkages to form SiH and SiOH groups, and crack propagation that leads to shedding of small silica particles. Periodic exposure to O atoms hydroxalizes sub-surface and reforms Si-O-Si linkages, and generates H2O that presumably desorbs.