The Effect of Intercalation Doping on the Performance of Multi-Layer Graphene Nano Ribbons as VLSI Interconnects for Deep-Submicron technology nodes

Abstract

Performance of deep micron and sub-micron circuits is deeply impacted by the interconnect due to reduced pitch size. With the reduction of technology below 45nm, the resistivity of copper increases drastically. The surface grain boundary scattering and electro-migration problem causes serious degradation in performance of copper as interconnect material. With the invention of graphene, GNR’s are the major contenders of replacing copper as a VLSI interconnect. In this dissertation, the applicability of MLGNR as interconnect material is studied. With the advantage of relatively easy manufacturing of GNR’s when compared with copper and retaining high mean free path and thermal stability makes it a dominant contender for interconnect material. The impact of Fermi energy on the propagation delay and power dissipation is studied. The increase in Fermi energy is the direct impact of increasing doping in MLGNR interconnects. With the addition of suitable dopant atoms the Fermi energy of GNR interconnects increases. This result in reduced parasitic components involved with interconnects material. This decrease in RLC parameters directly causes reduce delay and power dissipation of MLGNR interconnects. The results obtained by varying Fermi energy of MLGNR are also compared with copper interconnects. The obtained results shows improvement in performance of graphene ribbons with increased doping The variation of delay and power dissipation of MLGNR is also discussed by varying the technology node. A suitable number of repeaters are inserted to obtain the performance analysis of GNR interconnects .As the scaling continues, the delay as well as power of MLGNR increases. The performance variation between copper and MLGNR interconnect is observed at higher global interconnect lengths. Thus MLGNR with suitable Fermi energy has a potential to replace copper at global interconnect lengths.