Experimental Investigation of the Correlation between a Load-Based Metric and Solder Joint Reliability of BGA Assemblies on System Level

Abstract

A fast and accurate method based on experimentally obtainable parameters for the solder joint reliability prediction under thermo-mechanical load can add significant value in the product development process. On system level, for products with over thousand components, the evaluation of every individual solder joint is not always feasible. This is first and foremost true for products with many large BGA components that have very high ball counts. In addition, especially for these components, board level lifetime test results cannot be easily transferred to the product due to system level effects like stress triaxiality. The CTE mismatch between PCB and housing leads to PCB warpage and complex solder joint loading conditions for bending sensitive components under temperature cycling. The work presented in this paper advances the state of the art in solder joint reliability prediction by using only experimentally obtainable parameters such as PCB strain, PCB curvature and component warpage as metric for the reliability prediction. With this metric, a fast pre-qualification for BGA assemblies on system level with easier accessible parameters is possible. In order to verify the method, two fine pitch BGA packages with SAC solder joints, identical footprint and lateral package dimensions but different warpage behavior over temperature are investigated. The number of cycles to failure of the solder joints are experimentally determined in passive temperature cycling test under varying PCB strains and curvatures by superimposing a 3P-bending load. In this bending test, well defined PCB strain and curvature can be applied on the components in order to investigate the influence of these loads on the solder joint reliability. The correlation of component warpage, PCB strain and curvature with the solder joint reliability of these BGA packages is assessed.