Thermo-mechanical Analysis of Encapsulated Ball-Wedge Wire Bonds in Microelectronics, using Raleigh-Ritz Modeling

This study addresses encapsulated wire bonds in chip-on-board (CoB) multi chip modules, which provide a low cost option for dealing with the current trend towards compact microelectronic packages with increased I/O, higher reliability and lower cost. The focus is on thermomechanical stresses caused in the bond wires when the encapsulant is cooled from high curing temperatures and subsequently subjected to thermal cycling loading. The stresses generated in bond wires due to thermal expansion mismatches, in an encapsulated CoB are very complex and are driven by both global and local thermal expansion mismatches between: (i) glob-top encapsulant and the silicon die, (ii) encapsulant and the wire, and (iii) encapsulant and the substrate assembly. A 2D stress analysis model based on the variational Raleigh-Ritz (RR) method is developed, to estimate thermomechanical stresses in the bond wire, based on elastic analysis. The study focuses on detailed parametric investigation of different encapsulated CoB configurations. The initial wire profile, before encapsulation, is first modeled with RR 2-D trial functions based on cubic splines. This predicted geometry is then used for the subsequent thermomechanical stress analysis after encapsulation, based on trial functions composed of polynomials and exponential functions. The results are calibrated with Finite Element Analysis. Plastic deformations are ignored in the current analysis, as a first-order approximation. This model is therefore suitable for parametric design sensitivity studies and qualitative ranking of design options, but not for quantitative predictions of thermal cycling durability. The results show that the region above the ball bond is the predominant failure site. The RR 2-D model has a well-defined range of validity for CoB Ball-Wedge wire bond configurations with stiff encapsulants (E¬ >= 3 GPa) and thin wires (dia <= 2 mils). Also, the trend of maximum elastic strains obtained from the RR 2-D model is found to be in qualitative agreement with thermal cycling fatigue test data obtained from the literature