Reliability analysis based on cascaded-Foster thermal networks for Systems-in-Package (SiP)

This paper presents a fast and comprehensive method for reliability prediction of 3D System-in-Package (3D SiP) technologies. The proposed approach accounts for both critical wear-out failure mechanisms and mission-specific profiles. A novel reliability assessment framework is introduced to address the limitations of traditional methods, which often overlook the variability in failure mechanisms and wear-out rates across different layers within the same mission profile. A key contribution of this work is the coupling between the mission profile and dominant wear-out rates, enabling simultaneous consideration of multiple failure mechanisms, such as those affecting through-silicon vias (TSVs) and solder joints, under real-time operating conditions. The framework offers a multilayer analysis that uses unified units for each layer and incorporates a precise thermal model, enabling the rapid prediction of thermal behavior throughout the system's lifetime. Additionally, the proposed method can be easily integrated into circuit simulators and utilized as a real-time reliability estimator. This capability enables researchers and engineers to further investigate interactions between failure mechanisms across different layers of the SiP under realistic and dynamic operational conditions. By identifying the dominant failure mechanism in each layer, the method supports early-stage design decisions to mitigate potential reliability issues. The reliability estimation process involves selecting the mechanism with the shortest predicted lifespan for each layer and constructing the overall reliability curve using a series configuration of the reliability block diagram. Long-term mission profiles are translated into thermal loads through a cascaded Foster thermal network, with Monte Carlo simulations applied to determine the system's failure distribution.

Mots clés

• 3D SiP
• Failure mechanisms
• Foster model
• Monte-Carlo simulation
• Reliability
• System-in-package (SiP)