Access Official Technical PDF

Share Summary:LinkedInEmail

As integrated circuits scale, copper interconnects face increasing reliability challenges from high current density. Electromigration (EM)—the movement of metal atoms driven by momentum transfer from conducting electrons—can deplete copper and form voids. If a void grows enough to interrupt the conductive path, it can contribute to an open-circuit failure. FEM electromigration modeling offers a way to examine this process by tracking not only stress, but also the changing void, electrical current and temperature.

This matters to engineers and technical decision-makers working on semiconductor design, packaging and equipment that depends on reliable electronic systems. A model that captures interacting physical effects can provide more insight into how an interconnect evolves than an analysis that treats stress or current as an isolated factor.

Why post-voiding electromigration is difficult to analyze

Electromigration can cause copper depletion and open-circuit failures, as well as copper accumulation that may lead to short circuits. The post-voiding stage is especially challenging to simulate: the void changes shape and size, while the stress distribution in the remaining copper changes with it. Current density and Joule heating also respond to the evolving geometry.

These effects are coupled. A void can alter the path available for current, changing the local current-density distribution and the heat generated in the interconnect. In turn, temperature affects the conditions under which stress evolves. The resulting stress field is linked to the void’s volume through an atom-conservation relationship. A simulation that tracks stress but leaves void volume out of the calculation can therefore miss an important part of the physical picture.

What FEM electromigration modeling brings together

The finite element method (FEM) divides a physical domain into smaller elements, allowing governing equations to be solved numerically across a structure. In the approach described by Hengyang Zhao’s 2018 dissertation, the post-voiding problem is formulated as a coupled, time-dependent analysis of four physical systems:

  • Hydrostatic stress: Stress evolution in the confined copper interconnect is modeled using a form of Korhonen’s stress-diffusion equation.
  • Electrical current density: Electrical potential is solved across the conducting domain, which includes copper and liner material, to determine how current is distributed.
  • Void-boundary evolution: A phase-field method represents the changing copper–void boundary, allowing void shape to evolve within the modeled structure.
  • Temperature: A heat-diffusion equation accounts for Joule heating, with heat generation related to current density and electrical resistivity.

Each system has its own relevant physical domain and boundary conditions. For example, stress is solved in copper, while electrical current can flow through copper and liner. The thermal domain also includes surrounding materials such as capping layers and dielectric. Accounting for these distinct domains is part of representing how the interconnect and its surroundings influence the coupled problem.

Why phase field and conservation matter

Representing a moving void boundary is a central modeling decision. Explicitly moving a mesh boundary can distort finite elements as the void grows. It can also create difficulties when the boundary crosses labeled domain boundaries or when a void later shrinks and copper must occupy previously depleted space.

A phase-field representation avoids tracking the boundary as a sharply defined moving mesh edge. Instead, a field describes the transition between copper and void, allowing the interface to evolve across the computational domain. In the proposed method, the phase field is coupled with stress evolution and a regulated void-growth velocity. The regulation is intended to keep void volume consistent with the stress distribution according to the conservation relationship used in the model.

This coupling is significant for reliability analysis: void growth is not treated as an independent geometric change. Its volume and the mechanical state of the remaining copper are linked throughout the simulation. The method is designed for one-, two- and three-dimensional analysis, with dimensionality offering different trade-offs between computational detail and speed.

What the simulations can reveal

By updating the coupled systems over time, the analysis can show how stress and void shape evolve, how current density redistributes, and how temperature responds to Joule heating. It can also examine changes in interconnect resistance as a void grows. The dissertation reports that simulation results for stress, void-shape evolution and heating matched behaviors observed in silicon experiments. It also describes a transient resistance-change pattern associated with a void reaching a critical volume in the simulated upstream interconnect configuration.

These results illustrate the value of multiphysics analysis, but they should be understood in the context of the model and its assumptions. A simulation is a tool for examining physical behavior; its usefulness depends on how the geometry, material properties, boundary conditions and operating conditions represent the interconnect being assessed.

Practical implications for reliability engineering

For engineers evaluating copper interconnect reliability, coupled FEM can support a more detailed investigation of the post-voiding phase than stress-only approaches. It can help organize questions such as:

  • Where does current density concentrate as a void changes the conducting geometry?
  • How does the stress field evolve alongside void growth?
  • How does Joule heating affect the temperature distribution?
  • How are void volume and stress connected during the simulated process?

These insights can inform analysis of EM-induced failure mechanisms and provide a physics-based view of interconnect behavior. They do not eliminate the need to select appropriate model inputs or interpret results within the limits of the simulation.

Conclusion

Electromigration reliability is shaped by interacting electrical, thermal, mechanical and geometric effects. FEM electromigration modeling that couples stress, current density, void-boundary evolution and Joule heating provides a framework for studying these interactions over time. For semiconductor reliability teams and technical decision-makers, the practical value lies in examining how a growing void changes the wider interconnect system—not just the stress at a single location—and using that analysis to strengthen understanding of potential failure behavior.

Access Official Technical PDF

About The Author