Modelling and analysis of randomly distributed optimised structure of mixed CNT bundle interconnects-impact on crosstalk induced delay

Modern VLSI devices are manufactured using deep submicron technology. On-chip interconnects represent a major performance bottleneck for high-speed VLSI architectures. Nowadays, carbon nanotube (CNT) bundles have been projected as a possible nano interconnect material for increasing the operational speed and adaptable functionality of a system-on-chip. In this paper, a number of CNT bundles and their spatial orientations are considered, including single wall CNT (SWCNT), multi-wall CNT (MWCNT), and mixed CNT bundles (MCB). A good way to reduce the delay can be created by randomly distributing SWCNTs and MWCNTs of different diameters in an MCB. This can be a possible solution for a high speed VLSI connection. Although bundled SWCNTs and MWCNTs are easy, it is almost difficult to create an MCB with accurate SWCNT and MWCNT configurations. The main goal of this work is the optimized use of randomly distributed carbon nanotubes (RMCNTs) as nanocompounds. Furthermore, a new ATSO (Adaptive Transient Search Optimization) algorithm with Huang’s corner placement algorithm is studied to solve problems related to the placement of randomly distributed CNTs with different diameters in an MCB. This algorithm maximizes the tube density by distributing the CNTs randomly. Also, the temperature-dependent circuits are modelled for analyzing the crosstalk performance in the capacitively and inductively coupled MCB and RMCNT bundle (RMCNTB) interconnects at the far end of the victim line. The proposed optimized RMCNTB interconnect outperforms existing MCB interconnects in terms of propagation delay and power dissipation within a temperature range of 300 to 500 K. Furthermore, with an interconnect length of 1.5 mm, the proposed optimized RMCNTB structure exhibits a relative delay of 3.22% and a power dissipation of 10.51% less than the best MCB structure. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.