Effects of scaling on microstructure evolution of Cu nanolines and impact on electromigration reliability

Scaling can significantly degrade the electromigration (EM) lifetime for Cu interconnects, raising serious reliability concerns. Different methods have emerged to enhance the EM resistance of Cu by suppressing the interface diffusion (the historically fastest diffusion path), notably using CoWP metal cap and Mn alloying. With further scaling of Cu interconnects, EM reliability becomes increasingly complex due to changes in Cu microstructure. In ultra-fine Cu lines a large population of small grains mix with bamboo-type grains, resulting in an additional contribution of grain boundary diffusion to EM degradation. With the interface diffusion suppressed by CoWP or Mn alloying, the grain structure effect becomes even more important. The objective of this study is to investigate the EM reliability of ultra-fine Cu interconnects, focusing on the scaling effect on grain structure and mass transport. First, the detailed microstructure information of Cu interconnects down to the 22 nm node was analyzed using a transmission electron microscope (TEM)-based high resolution diffraction technique. A dominant sidewall growth of {111} grains was observed for 70 nm Cu lines (45 nm node), reflecting the importance of interfacial energy in controlling grain growth. The strength of the {111} texture was found to significantly increase as line width was reduced to 40 nm (22 nm node), while the length fraction of coherent twin boundaries was reduced to ~1%. Secondly, the results from microstructure together with the deduced interfacial and grain boundary diffusivities were used to identify flux divergent sites for void formation and to analyze the grain structure effect on EM statistics using a microstructure-based kinetic model. Finally, based on the analysis of Cu grain structure evolution with downscaling, the scaling behavior of EM drift velocity was investigated for Cu interconnects with CoWP capping and Mn alloying. This enables us to project the EM lifetime and statistics for future technology nodes. The Mn alloying effect on mass transport in combination of grain structure control was found to provide an effective means to improve EM reliability especially with further scaling. In summary, this study establishes a correlation between the microstructure of Cu nanolines, void formation kinetics, and EM statistics.