Data-Driven Predictive Control Strategy for Rapid Thermal Processing Systems
Huaiqian Zhang
Tianjin Key Laboratory of Information Sensing and Intelligent Control, School of Automation and Electrical Engineering, Tianjin University of Technology and Education, Tianjin, China.
Yongli Zhang *
Tianjin Key Laboratory of Information Sensing and Intelligent Control, School of Automation and Electrical Engineering, Tianjin University of Technology and Education, Tianjin, China.
Aihua Jiang
Guangzhou Academy of Special Equipment Inspection and Testing, Guangzhou, China.
Guofeng Ji
Tianjin Key Laboratory of Information Sensing and Intelligent Control, School of Automation and Electrical Engineering, Tianjin University of Technology and Education, Tianjin, China.
Lihui Geng
Tianjin Key Laboratory of Information Sensing and Intelligent Control, School of Automation and Electrical Engineering, Tianjin University of Technology and Education, Tianjin, China.
*Author to whom correspondence should be addressed.
Abstract
Rapid Thermal Processing (RTP) is a key technology for semiconductor manufacturing. However, the strong coupling, nonlinearities, and multiple disturbances inherent in RTP render accurate physical modeling extremely challenging. This paper studies the application of subspace predictive control (SPC) to a wafer RTP system driven by standard tungsten halogen lamps. This control algorithm does not require an exact RTP physical model and employs a data-driven method for temperature control. The subspace identification algorithm is employed to construct a subspace-based predictor for the RTP system. A pseudo-random binary series (PRBS) is designed as the excitation signal for the RTP system to obtain the input and output data. The designed PRBS excitation signal produces a response that effectively captures the system dynamics. Finally, temperature control of RTP system is achieved through SPC. Experimental results show that the achieved wafer temperature uniformity and control accuracy satisfy the required RTP performance metrics. The SPC algorithm provides a feasible alternative to model-based strategies in complex industrial environments.
Keywords: Rapid thermal processing, subspace predictive control, data-driven method, Pseudo-random binary series