Análise de projetos de encapsulamentos do tipo System-In-Package aplicando fluidodinâmica computacional (CFD) no processo de moldagem por compressão

Abstract

Integrated circuits of the System-In-Package (SiP) type typically exhibit high component density in their construction, which may pose challenges in the integrated circuit encapsulation process. The aim of this study was to develop a computational model to simulate the compression molding process, with the purpose of validating SiP encapsulation designs, specifically in hard-to-reach regions such as the 16 Land Grid Array (LGA) device area, which was the focus of this study. This work aimed to improve the flow of the molding compound, also known as Epoxy Molding Compound (EMC), reducing the incidence of voids, a defect formed after molding where air becomes trapped in a specific region of the chip, potentially causing various quality issues such as short circuits, electrical defects, component delamination, and more. The methodology involved the study of different designs, with "A" and "B" being pre-existing SiP device designs, along with proposed designs "C" and "D" specifically for this study. These alternatives aimed to enhance the EMC flow. The main difference between designs is essentially the internal design of the integrated circuit, where the central terminal, responsible for heat dissipation, was modified. Existing samples underwent acoustic microscopy analysis, cross-sectional cutting, and statistical analysis. Designs "A" and "B," along with designs "C" and "D," were evaluated through numerical simulations using the COMSOL Multiphysics® program. Each design involved a change in the channel layout, aiming to improve the EMC flow. Design "A" lacked a proper entry area for EMC, and its channel construction hindered flow. On the other hand, the other designs provided an extra entry area for EMC, and their channels were reconfigured for improved fluidity, each with its own characteristics. Results showed that design "B" exhibited significant improvements compared to design "A," used as a reference. Design "B" demonstrated a 19.28% improvement in mass flow (fluidity), resulting in a substantial reduction in void formation, an undesirable flaw in semiconductor encapsulation processes. Designs "C" and "D," developed as alternatives to designs "A" and "B," showed simulation improvements compared to design "A" but not as much as the results obtained with "B." Design "C" had a 6.06% increase in mass flow, while design "D" showed a 16.80% increase compared to design "A." This approach provided insights to enhance designs subjected to the compression molding process, highlighting that design "B" is the alternative with the best channel flow and consequently has a lower probability of void formation.