As described in Section 1.4.1, the electromigration TTF of a given interconnect is determined by monitoring its resistance change with time. The electromigration lifetime test of interconnect lines is the most commonly used method for evaluating their resistance change [48]. Other test methods related to the industrial assessment of electromigration reliability of the interconnects are the classic resistometric method (CRM) [128], the temperature-ramp resistance analysis to characterize electromigration (TRACE) method [121], the breakdown energy of metal (BEM) method [80], the standard wafer-level electromigration acceleration test (SWEAT) method [127], and the wafer-level isothermal Joule heated electromigration test (WIJET) method [86].
Electromigration lifetime-testing includes precise resistance measurements under accelerated test conditions. The electrical resistance of a metal line changes with time under the influence of an electric current. Such changes come primarily from the material transport due to electromigration along the conductor which leads to the nucleation of voids. When some of them reach a certain size required to form a stable void, which is able to grow under continued electromigration conditions, the interconnect resistance increases. This increase is due to the cross sectional area of the line being reduced due to the presence of the newly-created void, forcing the current to flow through the resistive surrounding layer. In performing an electromigration test, a given percentage change in resistance of an interconnect under stress conditions forms the failure criterion. Typically, interconnect failure occurs when a resistance increase of 20% is attained [152]. The electromigration test is stopped after reaching the given failure criterion and the TTF of the given test structure is therefore determined by the time necessary to achieve this threshold value. Since the electromigration test is carried out for a sample of similar interconnect structures in order to ensure the credibility of test data in a reasonable time frame [149], the results of the experiments are usually given in terms of mean-time-to-failure (MTTF), which is the time needed to reach the failure of the half of the group of similar structures [79], and the standard deviation of a lognormal distribution of electromigration lifetimes for the given sample. The distribution of lifetimes obtained from accelerated test has to be related to the use conditions, in order to estimate the interconnect lifetime under real operating temperatures and current densities.