Electromigration and Local Field Effects in Mesoscopic Interconnects
Abstract
Samples were fabricated by the same processes as the previous ones, but with triangular e-beam pads. Validity of the test structure was first examined. After stressing groups of 40 nm and 50 nm wide lines, samples were checked in the FESEM. In the tested samples, most of the lines failed inside the lines and their pads were intact, and a small portion failed in the pads, as shown in Figure 10. The test results show that the new structure can withstand the high current density and survive the bamboo test lines. It is normal that a certain percentage of the failures took place in the pads section. Upon examination of the failure sites, it was found that the sites were always damaged, probably by arcing, similar to that shown in Figure 4. Voids were never found within the lines. There is a possible explanation for the damage where voids are expected. In the wide lines, where voids commonly form due to EM damage, at the final stage of EM, current is concentrated into a very small cross-sectional area, catastrophic damage occurs and the relatively large volume of metal at both sides of the void can be a good heat sink and conduct the heat away in time, leaving a voided open circuit failure. But in the very narrow lines, current density is very high, while the tine does not have good dissipation of the heat generated at the final catastrophic stage, resulting in arcing damage at the moment of the open circuit occurs. At present, we don't have direct evidence of what! actually happens at the failure sites. Direct in situ observation will help to get more information about it.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 14, 1999
- Accession Number
- ADA412254
Entities
People
- Gary Bernstein
Organizations
- University of Notre Dame