Optimization of ultrasonic stress relief process parameters based on fuzzy neural network

摘要

The purpose of this work is to reduce the internal stress in cured SU-8 photoresist layer by ultrasonic stress relief technology, and the process parameters of ultrasonic stress relief were optimized. The stress relief mechanism of SU-8 photoresist layer was presented. Based on the principle of substrate curvature method, three factors and three levels orthogonal experiment was designed to measure and calculate internal stress relief rate of SU-8 photoresist with 9 groups of different ultrasonic stress relief process parameters. Fuzzy neural network was employed to optimize process parameters of ultrasonic stress relief. The fuzzy neural network was trained by orthogonal experiment datum, and the prediction model was built between internal stress relief rate and main process parameters: position, power input and relief time. Meanwhile, the network prediction results were consistent with the experimental results. Besides, the contrast experiments of ultrasonic stress relief technology and adjusting postbaking process parameters method were performed. Ultrasonic stress relief technology shows its advantages, and has certain practical value.