Converging Multi-Substrate Processing Technologies for 2.5D and 3D Heterogeneous Integration

Professor Peter Bermel of Purdue University presented at the Birck Nanotechnology Center Seminar Series.

The future of semiconductor technologies depends on continual increases in the density of components and functionality. As different types of chips have become specialized in different applications, 2.5D and 3D heterogeneous integration methods have been developed and implemented to squeeze different chips into very tight spaces. At the same time, most of these process technologies are targeted at a single substrate or interposer, even though different substrates such as silicon, ceramics, polymers, and glass offer significantly different strengths and capabilities, depending on the application space. In this work, a method of converging the processing methods of these different types of substrates to create identically functioning packages. Here, we present the results of several different recent experiments, and discuss achievements to date. Convergence is achieved in the embedding process, while partial convergence of Si and AlN is achieved in the back-filling process. Future work relating to completing process integration and performing initial testing is also briefly discussed.