Neural networks for the design and fabrication of integrated circuits for aerospace applications

A great effort has been underway in recent years to develop microwave and millimeter wave monolithic integrated circuit (MIMIC) technology for phased array radar systems for airborne and space based applications. Solid-state MIMICs have made a dramatic impact on capability, reliability, and cost of these sensor systems. However, the computer-aided engineering (CAE) tools necessary for the modeling and simulation of such complex systems has not kept pace. This paper presents a methodology for the utilization of neurocomputing technology to achieve enhanced yield and efficient yield prediction in integrated circuit (IC) manufacturing. Specifically, artificial neural networks (ANNs) are employed to model complex relationships between material and device characteristics at critical stages of the semiconductor fabrication process. Additionally, a methodology for the development of ANN models of MIMIC passive elements based on the elements physical parameters is described. Results are presented which illustrate that a ANN can predict the equivalent circuit parameters (ECPs) and/or the scattering-parameters of these passive elements to nearly the same degree of accuracy as that afforded by electromagnetic (EM) simulation