Apply genetic algorithm to minimize the overkills in wafer probe testing

In this paper, an ordinal optimization (OO) based algorithm is applied to minimize the overkills under a tolerable level of re-probes in a wafer probe testing process, which is formulated as a constrained stochastic simulation optimization problem that consists of a huge input-variable space formed by the vector of threshold values in the testing process. First, we construct a crude but effective model based on a shorter stochastic simulation with a small amount of test wafers. This crude model will then be used as a fitness function evaluation in the genetic algorithm to select N good enough solutions. Then, starting from the selected N good enough solutions we proceed with the goal softening searching procedures to search for a good enough solution. Applying to a real semiconductor product, the vector of good enough threshold values obtained by the proposed algorithm is promising in the aspects of solution quality and computational efficiency. We also demonstrate the computational efficiency of the proposed algorithm by comparing with the genetic algorithm and the evolution strategy.