Characterizing Microwave Power in a MPCVD System using Gaussian Mixture Modeling

Microwave Plasma Chemical Vapor Deposition (MPCVD) can be used to grow various kinds of diamond films and carbon nanotubes at various temperatures. Issues concerning plasma modeling and control play an important role in MPCVD systems. One crucial factor in controlling the plasma shape and position is the tunable reflected microwave power of the MPCVD system. However, modeling the tunable reflected power is highly complex and remains a poorly developed. In this paper, the microwave power distribution corresponding to the adjustable electromagnetic field is modeled by 2-D Gaussian Mixture Modeling (GMM). The simulations using the model show that microwave power data can be simplified to a linear combination of some Gaussian functions, allowing predictable control for tuning manufacturing parameters and plasma sharp in real-time. The experimental results show that each E-H tuner position can fabricate the Multiwall Carbon Nanotubes (MWCNTs) films with high reproducibility after GMM modeling.