Abstract
High-end pack ages used for AI applications are becoming denser with the emergence of chiplets technologies. Among them,
the density of package substrates is also progressing, and the recent substrates tend to have an asymmetrical substrate structure.
However, such an asymmetrical substrate induces warping due to heating process of chip joining, and it is essential to control the
copper remaining ratio in the substrate in order to suppress it at the design stage.
In this paper, a genetic algorithm is used to optimize the copper remaining ratio, and an algorithm flow considering the allowable
warpage value at chip bonding is proposed. As a result of the actual optimization evaluation, the superiority of the proposed flow
was confirmed.
