< Back to previous page

Project

Electromigration Mechanics in Scaled Interconnects

Electromigration (EM) is a mass transport phenomenon resulting from the momentum transfer between the electrons and metal ions in a conductor.  This mass transport may eventually lead to voids in the metal line, which can result in failure of the entire electronic component.  Electromigration is of large concern for the reliability of electronic interconnects; understanding the mechanisms by which it is caused is of foremost importance in order to mitigate or inhibit its effects. The standard EM test method is based on accelerated testing (high current density and elevated temperature), followed by extrapolation of the failure times to operation conditions. This methodology, however, has some important drawbacks: it is destructive, time consuming, provides only limited physical understanding and the extrapolation to lower stress conditions cannot always be justified. Therefore, in this PhD thesis, a new electromigration test method, based on low-frequency noise measurements, is developed. We show that this methodology allows the calculation of activation energies, early electromigration damage detection and qualitative lifetime prediction. In addition to the experimental validation of the low-frequency noise methodology by identifying empirical correlations with standard electromigration tests, new physics-based models are developed to explain why this strong correlation exists.

A first model is an expansion of the existing phenomenological model of Dutta et al., but provides the underlying theoretical assumptions that the original model is missing. A second model is based on the thermodynamic equilibrium properties of vacancies and as such can be used to extract these parameters from the low-frequency noise temperature dependence. A successful experimental validation of both models is shown. Finally, it is demonstrated that a direct analysis of the interconnects’ current fluctuations in the time domain can provide interesting insights. The current fluctuations in nano-electronic interconnects are found to exhibit fluctuation scaling (Taylor’s law). This is a scaling law, originally used in biology to describe the spatial distribution of animals in a habitat but applied here to study the temporal distribution of the amplitude of current fluctuations. We find that studying relative changes in the scaling exponent can provide information about EM-induced voiding and EM activation energies, again by means of non-destructive testing. This type of time-domain study could further reduce test times and provide additional fundamental understanding.

The new test solution is then applied to investigate electromigration mechanisms in state-of-the-art interconnects. Due to the continuous downscaling of interconnect linewidths, copper is starting to reach its limits as the material of choice. Possible solutions to extend the use of copper are scaling the barrier and introducing new barrier, liner and cap materials to mitigate EM. In this thesis we therefore first study the impact of such different metallization schemes in sub-30 nm wide copper lines. Using the low-frequency noise test method, it is shown that below 30 nm linewidth, grain boundary diffusion becomes a dominant EM failure mechanism. Eventually, at dimensions below 20 nm, copper will have to be replaced by alternative metals due to its unacceptable resistivity increase and insufficient electromigration lifetime.  Potential candidates are cobalt, ruthenium and tungsten. Here, their electromigration performance is studied and benchmarked against copper.  

Date:22 Sep 2015 →  31 Dec 2019
Keywords:Electromigration, Low-frequency noise, Interconnects, Diffusion mechanisms, Activation energy, Reliability
Disciplines:Metallurgical engineering
Project type:PhD project