Interfacial reliability analysis and structural optimization of System-in-Package modules

The System-in-Package (SiP) module is a significant component for the next generation terminal product and system in accordance with the electronics packaging driven demands of miniaturization and multi-functionality. Nevertheless, the SiP technology also faces multiple challenges to satisfy the requirements of increased performance, smaller factor, and lower power. The interfacial reliability of SiP module is one crucial issue during the entire life-cycle of devices, especially the packaging and reflows soldering process. The interfacial delamination is driven due to the mismatch of the material properties, such as the coefficient of thermal expansion (CTE) and Young’s modulus, even the crack propagation will be sped up due to the vapor and oxygen permeation. In this paper, the mechanical-thermal loads are applied to shed light on the interface reliability of laminated SiP structures. And the interfacial behaviors are systematically investigated using hybrid analytical method and finite element model, where the interfacial stresses between different components are calculated, the maximum von Mises stress of solder joints is obtained, and the warpage deformation of PCB-substrate is also included. Then the machine learning (ML) method is further applied to the structural optimization of SiP modules based on above interfacial reliability analysis, and the relationships between the properties of SiP modules and the thermal-mechanical behaviors are obtained. The effect of material properties and geometry of packaging components on the thermal-mechanical behavior are investigated. The interfacial stress increases with the increasing of CTE and Young’s modulus of EMC, and the maximum warpage deformation also increase with the increasing of material properties of EMC, but it decreases with the increasing of EMC thickness. The computational efficiency and accuracy of the proposed machine learning-based model are validated, and the application of ML model, in particular, is considered as a crucial useful technology to predict the thermal cycling response of laminated SiP structures and provide a helpful guidance for the geometry and material selection (structural optimization) of packaging design.