Investigation of Electromigration Stresses of Advanced Node Conductors through Pulsed Power Behavior
ORCID
https://orcid.org/0009-0006-7513-5622
Date of Award
Spring 2025
Language
English
Embargo Period
4-30-2025
Document Type
Dissertation
Degree Name
Doctor of Philosophy (PhD)
College/School/Department
Department of Nanoscale Science and Engineering
Program
Nanoscale Engineering
First Advisor
James Lloyd
Committee Members
Carl Ventrice, Hassaram Bakhru, Nathaniel Cady, Martin Gall
Keywords
electromigration, pulsed power testing, reliability, microelectronics, simulation of pulsed power electromigration, average current model for electromigration
Subject Categories
Nanoscience and Nanotechnology
Abstract
The microelectronics industry continues to put a heavy focus on preventing electromigration (EM) failures at advanced node conductors. The continuous scaling coupled with higher operating current densities further accentuate the problem. Most of the electromigration testing in the industry is performed with DC currents to evaluate the electromigration reliability of metal interconnects. However, most of the real-world applications operate using a pulsed current rather than a constant direct current. EM testing with pulsed direct current (PDC) show interesting results with how the current induced stress gradient evolves in absence of the electromigration driving force. It is generally assumed that the current induced stress acts opposite to electromigration driving force, but that is not always the case. During the nucleation phase of the EM failure, the current induced stress gradient does act opposite to the EM force but once the void is nucleated, the stress gradient temporarily switches direction and acts in the direction of the EM force causing damage even when the current is turned off.
In such a scenario, it is interesting to see how relevant the direct current EM testing is to the real-world applications, that mostly undergo pulsed operation. Black’s law is widely used by the industry to predict the lifetime of devices for EM failure. In view of this recent finding of additional damage during the periods of no current stress, Black’s law with a constant value of current density exponent (n) shows errors while calculating the lifetime of devices under test. It is of great interest is to see if this variation in the value of n is observed at different duty cycles of a pulsed current. And if yes, how disparate are the results from a conventional power law that is used for lifetime extrapolations for EM failure. A study of the current density exponent was therefore conducted as a function of duty cycle at accelerated test conditions. It was observed that in addition to the extended lifetime at lower duty cycles there was a gradual change in the value of the current density exponent as we went from a lower duty cycle to direct current stress.
Additionally, industry often uses the average current in design rules to estimate electromigration damage caused during a pulsed powering. It considers the use of average current passing through the interconnects over the duration of the test to calculate the lifetime of the interconnect. It is useful to see whether the use of average current remains relevant in light of the recent findings of additional damage incurred during the periods of no current stress. It was established that the present-day approach of direct current testing for EM is not entirely relevant to pulsed current operation. The good news is the extrapolated lifetimes calculated with a constant direct current stress are shorter than what is obtained with a pulsed direct current. The use of power law is therefore conservative but may be too restrictive. The duty cycle of operation should be considered to make more accurate predictions of failure times. Data analysis with this consideration may seem cumbersome, but extrapolation of constant stress data to pulsed operation is not as straightforward as we are made to believe.
The study of the physical phenomena behind EM damage has become more and more important since it can provide a deeper knowledge basis to anticipate the EM effect. In this context, experimental characterization coupled with mathematical simulations become a convenient way to understand the EM-induced failure. Therefore, the mathematical modeling has become a fundamental tool for explaining numerous experimental observations and, ultimately, can deliver a stronger basis for design and production of reliable metallization. The aim of this work is to merge the two different domains: on one side the experimental characterization of interconnect structures conventionally done in industry and on other hand side the development of a simulation setup, suitable for implementation in COMSOL, a tool for numerical calculations, as these two domains are often too disconnected. This original approach has brought excellent results, and it has allowed to confirm the hypothesis formulated by experimental analysis.
License
This work is licensed under the University at Albany Standard Author Agreement.
Recommended Citation
MATTOO, MOHD MUEEN UL ISLAM, "Investigation of Electromigration Stresses of Advanced Node Conductors through Pulsed Power Behavior" (2025). Electronic Theses & Dissertations (2024 - present). 205.
https://scholarsarchive.library.albany.edu/etd/205