A Neural Network-Based Prediction Model in Embedded Processes of Gold Wire Bonding Structure for Stacked Die Package

The trend in the consumer electronics market is to offer lighter, smaller outline, and more functional products, especially for portable products. This trend has pushed electronic packages to be thinner, with a lower profile and with multiple chips in one package. When the package becomes thinner, the space of the package correspondingly becomes an important issue. In order to obtain higher density and thinner package, we have to develop an embedded gold wire bonding assembly technology. It provides a simple structure with a thick adhesive layer to fix the upper layer die and bottom layer die gold wire. The developed technology can use exactly the same die size as for wire bonding interconnections without any additional processing. All electrical connections of the upper and the lower die are achieved by wire bonding to the substrate, independently. We have performed this stacking assembly by precise control of the die attach film layer thickness and low wire loop shape.