The prediction of failure probability in anisotropic conductive adhesive (ACA)

Abstract

This paper develops a model for predicting the failure probability of flip chip packaging using anisotropic conductive adhesive (ACA). The proposed approach modifies the box model to improve the accuracy of the bridging probability estimation by considering bridging in all directions. In the current failure analysis, probability theory is applied to calculate the probability of both opening and bridging failure. The Poisson distribution is used to calculate the probability of opening in the vertical gap between the pads, while the modified box model is used to estimate the probability of bridging between the pads. The results indicate that the modified box model provides an effective means of accurately estimating the probability of failure. The failure analysis takes the volume fraction of the conductive particles as the major variable in defining the optimal ACA design condition. A V-shaped curve model can be employed to establish an appropriate value of the volume fraction as a function of conductive particle radius, pitch, and pad dimensions such that the overall probability of failure is minimized.