Impact of mechanical material modeling on the solder joint fatigue analysis of a leadless package mounted at different positions inside a generic aluminum ECU

The solder joint fatigue lifetime of new microelectronics packages is usually experimentally evaluated using generic and unconstrained printed circuit boards (PCBs). In the later application, the PCB carrying the package may be constrained due to the mounting in a housing, which can adversely affect the solder joint fatigue lifetime. Solder joint fatigue simulation can be used as tool along the value chain to anticipate in an early phase a potential degradation of the solder joint lifetime and to evaluate countermeasures, such as a different placement of the package on the PCB inside the housing. The accuracy of the simulation will depend on the quality of the package model passed along the value chain. This paper studies for a discrete power package different types of material modeling which could potentially be communicated along the value chain and their impact on the accuracy of the prediction. All investigated descriptions correctly identify the most critical mounting positions of the discrete power package inside a generic model of an electronics control unit (ECU). The highest accuracy of the prediction is obtained with a non-linear and temperature-dependent material description of the package, followed by an approach using homogenized blocks with effective non-linear material properties are used. For linear-elastic properties, high deviations w.r.t. the non-linear case are observed in case constant properties over temperature are used and the glass transition region of polymers is passed though in the temperature profile of interest. Linear-elastic temperature-dependent properties are identified to be more accurate than linear-elastic constant properties, but are found to deviate in their estimations up to 25% compared to the non-linear material modeling.