Accurate estimation of DLC thin film hardness using genetic programming

Abstract

Abstract In the current research, diamond-like carbon thin films are deposited on silicon substrates by plasma-enhanced chemical vapor deposition. The effect of argon-C2H2 flow rate, hydrogen flow rate and deposition temperature on the thin film hardness is investigated. Morphology of the DLC films is investigated by scanning electron microscopy and atomic force microscopy, while the nano-hardness is investigated using nanoindentation. Raman spectroscopy is used for the characterization of the structural properties of the film. A metamodel of the DLC deposition process with argon- C2H2 flow rate, H2 flow rate and deposition temperature as the regressor variables and coating hardness as the response is built by using a novel symbolic regression approach. A state-of-the-art machine learning approach - genetic programming (GP) - is used for the symbolic regression. By carefully evaluating the performance of the current GP metamodel against a classical RSM (response surface methodology) metamodel, it is seen that the GP significantly outperforms RSM.