Contact electromigration: a method to characterize test structures for reliability parameter estimation

The combined use of single chain test structures and of cross bridge Kelvin resistors has shown that while the resistances of the diffusions increase monotonically with temperature, different behaviors are obtained for contacts, depending on the metal-silicon interface configuration. The TiN/Ti/n/sup +/Si interface can be modeled as an ohmic resistor, and therefore the life-test data can be normalized to the actual temperature of the life test, and correct activation energies can be extracted. The Al-1%Si contact's behavior, on the contrary, deviates from linearity above 150 degrees C. Since the life tests are preformed above that temperature, no correction of the data is possible. A way to avoid this limitation is found through statistical considerations on the data concerning the metallurgy with the barrier.<
>