Trajectory Planning For Capillary In IC Wirebonding

Wire loop profile and its strain, which greatly affect the density and stability of IC packaging, are determined by the capillary trajectory. Solving the mapping relationship between trajectory parameter and loop characteristic is a linchpin for trajectory planning. Conventional experimental statistical methods were expensive and time-consuming and can not meet the requirements for iterative operation in parameters optimization. Linkage-spring model and some finite element model were also applied to plan the capillary trajectory, but systematic study method and relevant results is still not available. This paper presents an integrated method in which finite element, BP neural network, and genetic algorithms are used to optimize the parameters for systemic and efficient trajectory planning. First, finite element method is used to simulate the loop profiling process with elastic-plastic beam solving the material and geometry nonlinearity. After the FEA model is validated by experiment, trajectory parameters analysis is made by grouping simulation to study the effects of trajectory parameters on wire loop profile and its interior strain. Second, a 3-9-3 structural BP neural network with single hidden layer, which is trained by samples from finite element model, is applied to explore the nonlinear multivariate relationship between parameters and responses. Finally, on the basis of target and restriction function from BP neural network, genetic algorithm is applied to optimize the trajectory parameters for optimum loop performances. Results in this study demonstrate the practicability of the proposed approach. Finally, summary has been given at the end of this dissertation, and some advices have been put forward for subsequent research.