THE ELECTROMIGRATION SHORT-LENGTH EFFECT IN TI-ALCU-TI METALLIZATION WITH TUNGSTEN STUDS

The electromigration short-length effect has been investigated by test ing a two-level structure with Ti-AlCu-Ti stripes and interlevel tungs ten (W) stud vias. This investigation represents a complete study of t he short-length effect using a technologically realistic test structur e. Lifetime measurements and resistance changes as a function of time were used to describe this phenomenon, where the latter approach provi des new insights into the electromigration behavior of multilayered me tallizations. A linear increase in resistance was followed by a resist ance change with time that approached zero. For the same product of cu rrent density and stripe length, longer stripes increased in resistanc e to higher values than shorter stripes. The sigma of the lognormal di stribution increased as the current density decreased and/or as the ma ximum allowed resistance change increased. The lifetime, or t(50), at relatively small current densities did not obey Black's empirical equa tion. Rather, the lifetime data obeyed a modified version of this equa tion that includes a critical current density j(c) as a new parameter. As an alternate approach to quantifying the short-length effect, we p ropose a novel and practical model for determining j(c) that focuses o n the apparent saturation of the resistance increase with time of the W stud chains. Unlike the modified Black's model, the resistance satur ation approach allows one to estimate lower bound limits of j(c). The threshold product, (jL)(c), is determined from the modified Black's mo del and from the resistance saturation model for stripe lengths of 50, 70, and 100 mu m. Both models indicate that j(c) strongly depends on the fail criterion, or the magnitude of the resistance change, but is independent of temperature in the range 175-250 degrees C. No evidence was found of a (jL)(c) below which no electromigration-induced damage occurs. (C) 1995 American Institute of Physics.