Skip to Main Content
Summary form only given. In the semiconductor manufacturing industry, fluorocarbon-based plasmas are increasingly being replaced by NF3 plasmas for cleaning of CVD chambers. The clear advantages in doing so include faster clean times, reduced chamber damage and substantially lower emissions of gases having high global warming potentials (GWPs). Atomic fluorine (F), known for its ability to etch silicon-based compounds is the main product of dissociation of NF3 in a remote plasma source (RPS). It is then delivered using a transport tube into the cleaning chamber. The current work is aimed at understanding and improving the key processes involved in the production and transport of F atoms. To achieve faster cleaning, the system is operated at relatively higher pressures 1-10 Torr. The RPS system used is the MKS Astronregi2 that is capable of achieving >95% dissociation of NF3 for flow rates up to 3 std. liter per minute. For study purposes, the dissociation products (mostly F, Ar and N2) are transferred into a cylindrical reaction chamber and etch rates of SiO2 samples (indicative of F conc.) are monitored at the different locations. The obtained results are used for validation of the two proposed models. Two models, for the remote plasma source (RPS) and reaction chamber (RC) are proposed. A zero-dimensional chemical kinetic model is chosen for the RPS. The governing reactions are taken from previous works. The results obtained match up well with those experimentally reported in literature. We consider a reduced mechanism for the quasi one-dimensional RC model, which incorporates volume and surface recombination, adsorption and desorption of F atoms. The RC modeling predictions also compare well to both our experimental results and those reported in literature.