Dispersion and evaluation of thermal fatigue reliability for solder joint by using normal random numbers

This paper presents a method mining the worst thermal fatigue life of solder joints on chip components used in vehicle electronics. Thermal fatigue life of crack initiation and propagation were calculated by using the finite element method. Statistical analysis of thermal fatigue life in the solder layer below the chip component was carried out. A sample of dispersed fatigue lives was generated by using the Monte Carlo analysis and response surface methodology (RSM). As a result, it was confirmed that the fatigue lives follow Weibull distribution. Furthermore, case study was also conducted to examine the effect of solder shape on thermal fatigue reliability. It was confirmed that there is the complex interaction between each shape factor of the solder joint. The fatigue life in solder fillet drops when a crack propagates into a fillet of solder. It means that the shape of solder joints should be designed to control the failure mode. In order to study the possible cases of fillet failure in mass product, the boundaries of normal failure cases and failure cases for the combination of shape factors was estimated. And then the worst cases can be easily investigated by Monte Carlo analysis by considering the correlation coefficients between the shape factors. Considering scatter at 6¿ level, the reliability of such the worst cases can be studied and these results can be considered in the design stage.