Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
9670362 | Microelectronic Engineering | 2005 | 7 Pages |
Abstract
Semiconductor industry has chosen copper (Cu) interconnects because of their enhanced electromigration performance compared to Al. In order to evaluate their electromigration lifetime, either very long test times or higher acceleration factors like, e.g. higher test temperatures have to be chosen. A bimodal distribution of fail times has been reported [1] in such highly accelerated tests. In this paper, we present a quantitative model which is based on Ta diffusion out of the barrier to explain the bimodality. The diffused Ta atoms are assumed to block vacancies at the fastest Cu diffusion path and thus lead to a change of the failure location and the failure time, as experimentally observed. The quantitative interpretation results in a modification of Black's equation which allows to predict the lifetime change and to extrapolate experimental results to operation conditions. The model predicts an increase of the interconnect lifetime by at least a factor of ten.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Hardware and Architecture
Authors
Oliver Aubel, Wolfgang Hasse, Martina Hommel, Heinrich Koerner,