AVS 46th International Symposium
    Plasma Science and Technology Division Monday Sessions
       Session PS-MoP

Paper PS-MoP31
Investigation of the TEOS Dissociation Coefficient by Electron Impact

Monday, October 25, 1999, 5:30 pm, Room 4C

Session: Poster Session
Presenter: C. Vallee, University of Nantes, France
Authors: C. Vallee, University of Nantes, France
A. Rhallabi, University of Nantes, France
A. Granier, University of Nantes, France
A. Goullet, University of Nantes, France
G. Turban, University of Nantes, France
Correspondent: Click to Email

Although O@sub 2@/tetraethoxysilane (TEOS) PECVD has been extensively used to deposit SiO@sub 2@-like films, the dissociation cross section of TEOS by electron impact is still unknown. In fact, most of the effort has been devoted to the determination of the TEOS fragmentation by O atoms (ko) since it is generally assumed that oxygen atoms are responsible for the dissociation of TEOS. Nevertheless, thanks to the experimental results obtained in a helicon reactor, it was suggested that TEOS fragmentation in a low pressure O@sub 2@/TEOS plasma was mainly achieved by electron impact while the main role of oxygen atoms was the etching of the organic part of the growing film. To get better insight into the TEOS dissociation by electron impact, we tried to determine the TEOS dissociation coefficient ke in a rf helicon reactor by developing a simple model which describes the experimental evolution of the growth rate with the distance from the TEOS injection. In this experiment, the TEOS gas is introduced at a point of the wall reactor instead of using a dispersal ring. A radial evolution of the deposition rate is obtained which can be related to a radial consumption of the TEOS. The model which gives ke at 3 eV (2 mTorr), is based on the following assumption: i) the deposition rate is proportional to the flux of fragments coming from the TEOS fragmentation by electron and oxygen atoms; ii) the spatial variation of the TEOS concentration is determined using a one-dimensional chemistry model. The first hypothesis requires the use of a sticking coefficient s. Dependence of ke with the value of s introduced in the model is also investigated. Finally, we found ke to vary from a few 10@super -8@ cm@super 3@s@super -1@ to a few 10@super -7@ cm@super 3@s@super -1@.