A new physical model and experimental measurements of copper interconnect resistivity considering size effects and line-edge roughness (LER)

Abstract

A new closed-form effective resistivity (rhoeff) model as a function of line-edge roughness (LER), sidewall specularity p, and grain boundary scattering R is presented. There is improved physical insight to increasing resistivity than previous models. The model is validated against former simulation data and calibrated with electrical measurements of fabricated Cu interconnect test structures exhibiting an average of 14 nm LER for line widths ranging from 61 nm to 332 nm. Upon fitting the new model to experimental data, p and R are determined to be 0 and 0.79, respectively. The model is also used to interpret ITRS 2007 projections for local wire resistivity. ITRS projections for resistivity can only be achieved with very high quality interconnect structures that have nearly elastic sidewall collisions (p=0.95), low grain reflectivity (R=0.40), and no line edge roughness (LER=0nm). In fact, adding 6.0nm of LER increases rhoeff by ~20% for 2022 (11nm node). Finally, a projection with pessimistic values of p=0, R=0.5 and LER=1.0nm predicts an 87% greater rhoeff value than the ITRS 2007 projection for the 11 nm node.