Study the Temperature Dependent Performance Analysis of Multilayer Graphene Nanoribbon (MLGNR) As VLSI Interconnect

Abstract

As the technology is evolving the performance of interconnects in a circuit that was once neglected, is now turning out to be major concern for the high performance integrated circuits. It has been realized that with the scaling of technology in deep submicron (DSM) region, the resistance of consistently used copper (Cu) interconnects has started to increase due to surface roughness and grain boundary scattering. However, the resistance and capacitance play an important role in determining the performance of interconnects in terms of propagation delay, power dissipation and crosstalk. Due to continuous scaling of technology, the high performance circuits are becoming highly sensitive to temperature variations. Therefore, temperature is turning out to be one of the major factors that determine the performance of an interconnect material. The allotropes of carbon such as Carbon nano tubes (CNTs) and Graphene nanoribbons (GNRs) are being recommended as next generation interconnect materials to reduce the performance limiting factor like power, PDP, delay and the crosstalknoise voltage in VLSI interconnects.This thesis work briefly presents the temperature dependent modeling and performance analysis of multilayer graphene nanoribbons (MLGNRs) based interconnects. The temperature dependent performance analysis is done, using SPICE simulations, in terms of propagation delay, power dissipation, frequency spectrum of the output pulse, crosstalk induced noise voltage and its frequency spectrum for MLGNR interconnects at 14nm technology node. A comparative performance analysis is done between MLGNR and conventionally used copper interconnects over a temperature range from 300K to 500K at 14nm technology node. According to the results, with the rise in temperature from 300K to 500K, the performance ofMLGNR is superior as compared to copper interconnects.Also the performance comparison is done between the temperature independent model of MLGNR with the temperature dependent model. The simulation results reveal that the delay and power dissipation of MLGNR, obtained through temperature dependent model are lower as compared to conventionally (temperature-independent) used model of MLGNR, at different interconnect lengths from 400μm to 1000μm. Also the results gives the effect of temperature on the frequency spectrum of MLGNR interconnects. It is found that the increase in temperature cause the loss in signal power and decrease in bandwidth. Further the time duration of victim output waveform of temperature dependent model of MLGNR is compared with that of temperature independent model at different interconnect lengths, ranging from 400 μm to 1000μm. An average improvement of 38.9% is noticed in the time duration of temperature dependent model of MLGNR when compared with temperature independent model. Therefore, it is important to take temperature-dependent models into consideration for the performance optimization of high speed integrated circuits. Also, the performance superiority of MLGNR compared to copper based interconnects at 14nm technology, makes it a possible replacement for copper based interconnects in deep submicron (DSM) region.