Investigation of void dynamics in metal interconnects under electrical mechanical thermal stimuli using physics based models augmented by machine learning

Electromigration (EM), stress induced voiding (SIV), stress migration (SM) and thermomigration (TM) have long been major reliability challenges for the microelectronics sector. Drastic increase of power densities and modern integration approaches have increased the intensity of electrical, mechanical and thermal stimuli that induce metallization voiding, thereby necessitating predictive physics-based models that can provide quantitative understanding of void dynamics. Nevertheless, the state of the art in physics-based void dynamics modelling is limited by simplifications due to the surfeit of involved phenomena and the computational cost associated with their comprehensive modelling. Therefore, such models have rarely been adopted by the industry where variability of involved fabrication parameters such as geometry and microstructure require highly efficient low-cost models for reliability predictions. To this aim, in this PhD an efficient and comprehensive physics-based void dynamics modelling framework will be developed which will simulate evolution and interaction of voids with electrical-mechanical-thermal stimuli and microstructure. Machine learning approaches will be employed to complement the standard numerical methods in favor of computational efficiency and thus applicability of the modelling framework. Experimental void dynamics analysis using in-situ microscopy will be used to calibrate and corroborate the simulation findings.