A Chemical Monitoring and Prediction System in Semiconductor Manufacturing Process Using Bigdata and AI Techniques

Abstract

Numerous chemical substances are used in the semiconductor manufacturing process, and homogeneity and quality control of surface treatment are performed through precise control of chemical substances in the process. The repeatability and reproducibility of each process is a fab's greatest concern, and even a slight deviation from specifications can lead to expensive equipment contamination and wafer scrap. In this study, we propose a real-time big data analysis system that integrates and manages the state of substances being measured at numerous points in a factory, and monitors them in real time, and delivers an alarm message to the manager when the preset upper/lower limit is exceeded. In addition, we propose an artificial intelligence prediction model that predicts the state of matter by using accumulated data as learning datasets. The data analysis and monitoring system and AI prediction model are designed to continuously improve accuracy through additional learning of related datasets in the future.