Electromigration Risk Assessment and Circuit Optimization using Innovative Multiphysics Modeling

Abstract

With smaller and denser transistors, the physical flow of electrons may inhibit the performance of the device over time by forming voids and cracks at interconnects due to Electro-Migration (EM). Circuit designs that fail to meet EM specifications may lead to catastrophic failures and SI/PI performance degradation. One way of mitigating EM is to use multiple vias between layers of copper traces to reduce the current crowding effect. However, the quantities of vias may affect the current density and current redistribution inside critical joints. Current studies mainly focus on predicting the EM time-to-failure (TTF) based on the empirical Black’s equation. However, this method may not give enough insights about void formation and crack propagation and reflects the current redistribution that could impact the TTF. In this study, we compared the EM lifetime of Ball Grid Array (BGA) test vehicles with different structural designs and developed a methodology to consider the diffusion of atoms in solder joints based on Multiphysics field migration to study the current redistribution influence of vias. Moreover, crack propagation was also simulated to understand the failure mechanism. BGA traces without vias and with 8 vias are stressed under 5A, 7A, and 9A at 150C to compare the EM performance. Moreover, each test structure is manufactured with two different surface finishes: A and B. Based on the experimental results, Finite Element Analysis (FEA) simulations based on Atom Flux Divergence (AFD) were performed to compare with the experiment results. It was found that the current crowding effect could be significantly reduced by 8 vias compared to daisy chained traces. The study shows better EM resistance with 8 and 4 vias than no-via traces and helps predict the EM life of different structures to provide guidance for design optimization