Modeling of Plasma-Enhanced Chemical Vapor Deposition of Nano-crystalline Silicon Carbide Films Using Neural Network

s The infrared absorption peak shift of PECVD nano-crystalline β-SiC films have been modeled using neural network which was trained using the error back-propagation algorithm. This ANN model can be built with process input– output experimental data. The PECVD process was characterized by modeling peak shifts of Fourier transform infrared (FTIR) spectroscopy as a function of varying seven controllable process parameters. The results obtained from the ANN-based model were verified with the experimental results not pertaining to the training data. It demonstrated an accurate prediction model with little error within the normal process input range. This model can then be applied to predict the input–output relationships of the process and to determine optimal operating recipe. 1. Background As one of the most promising wide bandgap semiconductors, Silicon carbide (SiC) is recognized to have considerable potential for elevated-temperature, high-power electronic application. In comparison with silicon, SiC thin film, which is easily n-type or p-type doped, shows high-breakdown electronic field of several 10 V cm, high saturation electron velocity and high thermal conductivity. Attention has also been recently paid on SiC films because of the increasing interests in using them as cold-cathode emitters in field emission devices (FED) [1-4], and the nanostructure films have been expected to show better field-emission property due to their small grain size and quantum dot effect [5-10]. In the previous experimental pursuits of nanostructure SiC films of high quality, Chemical vapor deposition (CVD) techniques have been developed and applied to prepare SiC films of high quality [11-14]. CVD is a process where one or more gaseous species react on a solid surface (substrate) and one of the reaction products is a solid film on that substrate. The reactant gases are introduced into a reaction chamber, decomposed, and finally reacted at a heated surface to form the thin film. A wide variety of thin films utilized in very large scale integration (VLSI) fabrication and Micro Electro Mechanical System (MEMS) are prepared by CVD. Based on their potential capabilities for satisfying demanding criteria, specific deposition methods have been developed to form such thin films. CVD 1 This research was done in conjunction with Prof. Bo Liao, Ms. J.J. Wang, Ms. S.S. Lu, Department of Electronic Engineering, BeiJing Institute of Technology, Prof. Hui Yan and Prof. Bo Wang, The Key Laboratory of Advanced Functional Materials of China Education Ministry, Beijing Polytechnic University.