BGA Board Level Reliability Analysis & Optimization

Abstract

The solder joint reliability of semiconductor package interconnects to printed circuit boards is critical for product durability. A dominant failure mode is solder fatigue due to the CTE mismatch between the BGA component and PCB at thermal cycling. However, it is well known that other factors can impact fatigue behavior and time to failure such as solder joint geometry, die geometry, solder system, etc. Finite element modeling (FEM) and simulation can play an integral role in providing deeper insight into the impact of these package parameters on the overall assembly. However, a major challenge of accurately modeling these systems includes simulation of multiple length scales from the package, substrate, and solder joints. The FEM approach addressing these can lead to reduced cycle time, accurate simulation, and improved package performance. In this work, the finite element modeling and simulation procedure is demonstrated for a BGA package at accelerated temperature cycling conditions. At the component level, key details regarding the properties and constituents of the BGA package mold compound and substrate are established by coupling measured experimental warpage data and finite element modeling. Comparison of simulated & Thermoire measurements shows excellent agreement at the package level, with warpage correlation achieved over the entire temperature range. At the assembly level, the truncated sphere model is used to arrive at precise solder joint profiles for accurate representation to tie the package to the board. The combined validated package-level results and solder joint profiles are employed towards a subsequent thermo-mechanical analysis of the full BGA assembly. The entire simulation procedure is demonstrated for a BGA design, where inelastic creep and reliability test data are compared. High strain regions in the solder joint array are shown to compare closely with regions of failure from experimental reliability test data. The validated FEM model allows for extrapolating to similar package conditions allowing faster design cycle time and less time consuming experimental work.