Paul T. Vianco

A Practical Viscoplastic Damage Model for Lead-Free Solder

This paper summarizes the results of a program to construct an internal variable viscoplastic damage model to characterize 95.5Sn-3.9Ag-0.6Cu (wt.%) lead-free solder under cyclic thermomechanical loading conditions. A unified model is enhanced to account for a deteriorating microstructure through the use of an isotropic damage evolution equation. Model predictions versus experimental data are given for constant strain-rate tests that were conducted at strain rates of 4.2 X 10 -5 s -1 and 8.3 X 10 -4 s -1 over a temperature range from -25°C to 160°C; cyclic shear tests; and elevated-temperature creep tests. A description is given of how this work supports larger ongoing efforts to develop a predictive capability in materials aging and reliability, and solder interconnect reliability.

Validation of a General Fatigue Life Prediction Methodology for Sn-Ag-Cu Lead-Free Solder Alloy Interconnects

A general fatigue life prediction methodology, based on a unified creep plasticity damage (UCPD) model, was developed for predicting fatigue cracks in 95.5Sn-3.9Ag-0.6Cu (wt %) solder interconnects. The methodology was developed from isothermal fatigue tests using a double-lap-shear specimen. Finite element analysis model geometries, mesh densities, and assumptions were detailed for both a full model (an octant-symmetry slice of the entire ball grid array (BGA) assembly) and a submodel (the solder joint deemed most likely to fail and the surrounding package layers) to facilitate fatigue prediction. Model validation was based on the thermal mechanical fatigue of plastic BGA solder joints (250-4000 thermal cycles, -55°C to 125°C, and 10°C/min). Metallographic cross sections were used to quantitatively measure crack development. The methodology generally underpredicted the crack lengths but, nonetheless, captured the measured crack lengths within a ±2X error band. Possible shortcomings in the methodology, including inaccurate materials properties and part geometries, as well as computational techniques, are discussed in terms of improving both the UCPD constitutive model and the fatigue life prediction methodology fidelity and decreasing the solution time.