Thermal fatigue life prediction for solder joints with the consideration of damage evolution

Abstract

Thermal fatigue of solder joints is critical to electronic package performance and service life. It is well known that the fatigue life is rather difficult to be estimated because of the complex interaction between creep and fatigue of solder materials. Conventional life prediction method such as Coffin-Manson relation or its modifications may give erroneous results at low strain range. In this paper, a nonlinear finite element based model is introduced for predicting the thermal fatigue life of solder joints. This method not only considers the damage evolution in the solder, but also saves substantial computational efforts. In the present model, the evolution of damage is determined by the collapse of hysteresis loops of shear stress-strain of solder joints in the double-beam joint specimen. The agreement in results between FEM and the experiment is very encouraging. This method is used to estimate the solder joint reliability of an SON CSP assembly. The proposed model is validated by experimental testing data.

Microstructure dependent viscoplastic damage modelling of tin–lead solder

“…Various time-dependent models are available and have been employed to describe the material behaviour of solder, e.g. the Anand model (Anand, 1985;Amagai, 1999;Cheng et al, 2000;Wang et al, 2001) and the Dorn equation (Bird et al, 1969;Guo and Conrad, 1993;Akay et al, 1997;Zhang and Lee, 1998;Anderson et al, 2000). Here, the viscoplastic strain rate function defined by Perzyna (1966) is selected to describe the mechanical response (Wang et al, 1997;Van der Sluis et al, 1999;Ju et al, 1994;.…”

“…In the literature, solder is usually considered as a single phase material when modelling damage (Basaran and Chanderoy, 2000;Basaran and Tang, 2002;Zhang and Lee, 1998;Towashiraporn et al, 2004;Yang and Nassar, 2005). The damage parameter is in most of these cases only influenced by the microstructure through the inclusion of an averaged grain size parameter in the constitutive model.…”