Study of ruthenium-based barrier layer for copper metallization

Abstract

Bilayer of Ta/TaN is the common diffusion barrier for Cu metallization in microelectronics. However, this bilayer is not conducive for electrochemical plating of Cu as it has a poor seeding ability. Consequently, an additional seed layer becomes inevitable. As the feature size of interconnects has become smaller, the composite barrier cum seed layer has become disproportionately thick in comparison to via/trench dimensions and hence the search for a thinner composite layer is taking technological importance. One obvious avenue to explore is a single layer that could provide adequate barrier function while serving as an effective seed for Cu nucleation. Some recent work indicates that Ru is a probable candidate for this purpose. Hence, this project was launched to conduct a systematic investigation into the barrier properties of Ru. The performance of pure Ru was examined in the first phase with and without Ta, in a Cu/low-k system. Pure Ru was found to promote the nucleation of the beneficial Cu(111) texture in the Cu overlayer when compared to seeding on pure Ta. Adhesion tests indicated sufficient adhesion strength of ~6 J/m2 in Ru/low-k interface which is comparable to that in Ta/low-k interface (~6.5 J/m2) . The barrier performance was assessed in Cu/Si, Ru/Si, and Cu/Ru/Si systems. Barrier failure could be detected at relatively low temperatures in the resistivity of the films, which ultimately led to detectable silicide formation at higher temperatures. Attempts were made to improve the barrier performance in the second phase of this project. It was envisaged to obstruct the easy grain boundary diffusion path by dissolving N atoms in Ru, in order to "stuff' the grain boundaries.