Thermal-Aware Modeling and Analysis of Cu-Mixed CNT Nanocomposite Interconnects

Growing only single-walled carbon nanotube (SWCNT) or only multi-walled carbon nanotube (MWCNT) in a bundle is not feasible practically. Thus, a random number of SWCNTs and MWCNTs whose diameters are also varied randomly (following a Gaussian distribution) should be considered in a bundle of mixed carbon nanotubes. A thermal aware electrical modeling of copper-mixed carbon nanotube (Cu-MCNT) composite interconnect is proposed here and its performance and reliability has been studied and compared with copper interconnect at 7 nm and 11 nm technology nodes. An increase in filling fraction (F_MCNT) of mixed carbon nanotubes in Cu-MCNT composites leads to a decrease in delay. But in some cases, if the fraction of SWCNTs is greater than that of MWCNTs in a bundle, then it leads to a higher delay. Variation in delay with respect to F_MCNT is more at higher temperature by $\sim$ 10%. Cu-MCNT composite interconnects with F_MCNT = 0.5 have $\sim$ 52% higher breakdown power than copper interconnects and are thermally more efficient. Scaling down the dimensions leads to degradation in performance and reliability combined but this degradation is more in copper interconnect when compared to Cu-MCNT composite interconnect by $\sim$ 14% Cu-MCNT composite interconnects attain $\sim$ 39% less temperature rise at steady state when compared to copper interconnects. The performance of Cu-MCNT composite interconnects in terms of meantime to failure at higher current densities is phenomenal as compared to copper interconnects. Our study promotes Cu-MCNT composite interconnects as an alternate candidate to copper interconnects considering their superior electro-thermal performance.