Predictive Validity for Sintered Die-Attach Performance Based on Cross-Sectional Morphology

The sintering of copper nanoparticles for die-attach bonding has emerged as a promising soldering alternative, with mechanical properties largely influenced by pore morphologies within the porous microstructures. This study evaluates the predictive validity in assessing the mechanical properties of sintered die-attach layers based on cross-sectional morphology. Over 150,000 pores were analyzed from 120 SEM images across three sample types, with features such as porosity, pore shape, and distribution extracted and analyzed. While nanoindentation measurements revealed significant differences in mechanical properties, Jensen-Shannon divergence indicated strong feature convergence across samples. An artificial neural network (ANN) model was developed to explore the influence of processing on pore features and their dispersion across samples. This study highlighted the importance of considering feature dispersion in predictive or simulation models, especially those based on local microstructures.