基于BP神经网络的SU-8光刻胶工艺参数优选研究

SU-8是一种性能优异的厚胶,广泛应用于高深宽比的MEMS微结构中。本文首先用正交试验研究了前烘时间、曝光剂量、后烘时间以及显影时间对SU-8光刻胶图形尺寸精度的影响,得到了优化的工艺组合。在此基础上,运用BP神经网络对试验数据进行分析处理,预测了较正交试验分析结果更为优化的工艺组合,并用试验验证了其正确性。结果表明,经正交试验数据训练过的BP神经网络,很好地映射了工艺参数与优化指标之间的复杂非线性关系,此时应用BP神经网络对工艺参数进行优选研究能够得到更全面、准确的结果。

Study of Optimizing SU-8 Photoresist Process Parameters Based on a BP Neural Network

SU-8 photoresist is widely used in high-aspect-ratio microstructures in MEMS for its excellent performanees. Using the orthogonal array technique, the paper performed an optimization experiment to study the influence of the process parameters of pre-bake time, exposure doses, post exposure bake time and development time on the accuraccy of SU-8 photoresist image sizes. The process parameters were optimized, on the basis of which the experimental data were analyzed by using a BP neural network. The process parameters with better resolution than the orthogonal array technique was predicted by the BP neural network and their correctness was verified by an experiment. The experimental results prove that after being trained by the data of orthogonal tests, the BP neural network has a good capability of mapping the complex nonlinear relationship between the process parameters and the optimization targets. Therefore it results in a more comprehensive and accurate optimization of process parameters.