Statistical Model for Prediction of Life-Reduction in SAC Lead-free Interconnects during Long-Term High Temperature Storage using Principal Component Regression and Ridge Regression

Abstract

Electronics in military and defense applications may be stored for prolonged period of time prior to deployment in mission critical applications. In addition, electronics in automotive applications may be used underhood, mounted directly on-engine and on-transmission with expectation to survive in excess of 10 years, 100,000 miles of usage in a variety of environments. Previous researchers have studied the microstructure, mechanical response and failure behavior of leadfree solder alloys when subjected to elevated isothermal aging and/or thermal cycling. The effects are most pronounced in the widely used SnAgCu based alloys including SAC105, SAC205, SAC305 and SAC405 solders. Lower silver solders such as the SAC105, often touted for their resistance to transient dynamic shock and vibration, are the most susceptible to thermal aging amongst the SAC solders. The effects have been verified in the solder alloys at both lower strain rates in the neighborhood of 1e-4 sec-1 to 1e-5 per sec typical of thermal cycling, and at 1-to-100 per sec typical of shock and vibration. Degradation in the neighborhood of 50% has been measured at low temperature exposures. In this thesis, accelerated test thermal cycle data collected on SAC leadfree assemblies subjected to high temperature thermal aging for period of up to 1-year has been used for development of model for prediction of life reduction from long term storage at elevated temperatures. The input parameters to the model include geometry parameters including chip size, mold compound thickness, package size, board thickness, solder joint height, pad diameter, die attach thickness, and package pitch. In addition, material parameters considered include coefficient of thermal expansion, elastic modulus, and the glass transition temperature for all the package elements in the electronic assembly. Principal component regression in conjunction with stepwise regression and Ridge Regression have been used to identify the influential variables, remove the multi co-linearity between the predictor variables, and calculate the sensitivities of the life reduction due to elevated temperature exposure on the predictor variables. The life reduction model has been used to predict the expected life reduction after prolonged storage of 20-40 years.