A Fuzzy Adaptive Network Model for Waviness Removal in Grinding of Wire-Sawn Silicon Wafers

Abstract

Silicon is the primary semiconductor material used to fabricate microchips, and the quality of microchips depends directly upon the quality of the starting silicon wafers. One of the manufacturing problems in silicon wafer manufacturing is the presence of waviness on the surface as a result of wire-saw slicing. Various factors influence the waviness reduction capacity during soft-pad grinding; the grinding process is very complicated and difficult to define. In this research, fuzzy adaptive network, which is ideally suited to the modeling of vague phenomena, is used to model the waviness problem. Fuzzy adaptive network has the learning ability of a neural network and the linguistic representation of a complex, not well understood phenomenon. Simulation data are used to illustrate the applicability of fuzzy adaptive network. The results, even though based on some very limited data, indicate the influences of the independent parameters clearly.