Automatic Reaction Modeling in Chemical Vapor Depositions Using Multiple Process Simulators
- PDF / 157,800 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 57 Downloads / 220 Views
JJ9.15.1
Automatic Reaction Modeling in Chemical Vapor Depositions Using Multiple Process Simulators Takahiro Takahashi1, Kimito Funatsu, and Yoshinori Ema1 Department of Knowledge-based Information Engineering, Faculty of Engineering, Toyohashi University of Technology, Hibarigaoka, Tempaku, Toyohashi, Aichi 441-8580, JAPAN 1 Department of Electrical and Electronic Engineering, Faculty of Engineering, Shizuoka University, 3-5-1 Jouhoku, Hamamatsu, Shizuoka 432-8561, JAPAN ABSTRACT The identification of appropriate reaction models is very helpful for developing chemical vapor deposition (CVD) processes. In this paper we propose a novel system to analyze experimental data of various CVD reactors and identify reaction models automatically using Genetic Algorithms (GA) with multiple process simulators and modeled functions. We demonstrate that this system is able to adequately model reaction systems, and that complex analysis of various experimental data increased the reliability of the reaction modeling results. INTRODUCTION CVD is one of the most important manufacturing processes used in the semiconductor industry. One important goal of CVD research and development efforts is to identify the reaction model (i.e., the reaction mechanism) that reveals the reaction path from reactants (source gases) to products (films), both quantitatively and qualitatively. In previous work, we proposed a system that analyzed deposition profiles (i.e. film thickness non-uniformities in a macroscopic cavity (hereafter denoted as Macrocavity [1])), and identified the reaction models automatically using GA [2]. In this study, we improve the system in order to handle not only the deposition profiles but also the deposited shape in a microstructured cavity or Microcavity [1] (e.g., the cross sections of the deposited films Flow Diffusion on nano- or micro-structured Substrate substrates observed by a scanning electron microscope). Spacer The system was able to Macrocavity Microcavity construct more reliable reaction Diffusion models by employing complex Figure 1. Schematic structure of Micro- and Macrocavities analysis of experimental data.
JJ9.15.2
This method corresponds to the automation of the analyzing and modeling procedure of Micro/Macrocavity method [1]. (Figure 1) On the other hand, the calculation cost to reproduce the shape of the cross sections in the Microcavity by process simulators using the Monte Carlo method can be enormous, and represents an obstacle to reaction modeling by the GA. Therefore we investigated the correlation between the sticking probabilities of the deposition species and the stepcoverage, which is the typical index used to characterize the shape of the film cross section, and replaced the simulations in the system with the modeled functions. By this approach, we could reduce the calculation time dramatically and make the system practical. THEORETICAL METHODS AND CALCULATED CONDITIONS Figure 2 shows the schema of the automatic modeling system. The system consists of three devices, that is, a user interfa
Data Loading...
