Title :
Application of neural network method for process synthesis
Author :
Zhang, Wen-Jun ; Liang, Jian-Lin ; Xie, Xiao-Feng ; Tian, Li-Lin ; Yang, Zhi-Lian
Author_Institution :
Inst. of Microelectron., Tsinghua Univ., Beijing, China
Abstract :
Process synthesis is a top-down design methodology and can effectively reduce the process design time. In the paper the general method and the neural network (NN) package used for process synthesis are discussed. Then the characteristics of synthesizing some key process modules, including ion implantation and well formation, are analyzed, based on which the training set used to build the NN is generated. After that the NN model used for process modules synthesis is constructed and trained. Test results show that this model can fulfill the process modules synthesis.
Keywords :
feedforward neural nets; integrated circuit manufacture; ion implantation; learning (artificial intelligence); multilayer perceptrons; inverse model; ion implantation; multilayer feedforward neural network; neural network method; process synthesis; top-down design methodology; training set; well formation; Buildings; Circuit synthesis; Feedforward systems; Flowcharts; Integrated circuit modeling; Inverse problems; Mathematical model; Network synthesis; Neural networks; Packaging;
Conference_Titel :
Machine Learning and Cybernetics, 2002. Proceedings. 2002 International Conference on
Print_ISBN :
0-7803-7508-4
DOI :
10.1109/ICMLC.2002.1174559
