| The ever increasing demand for faster and more reliable chips depends largely on the performance of on-chip interconnects. Over ten years ago, copper became the preferred metal over aluminum as the on-chip interconnects material in the most advanced silicon chip technology. However, as copper interconnects scale down, its resistivity increases exponentially due to the degrading effect of surface scattering and grain-boundary scattering. Currently, the scaled-down copper interconnects suffer from reliability concerns due to electromigration in the sub-30 nm technology regime. In order to keep up with the International Technology Roadmap for Semiconductors (ITRS), copper must be replaced. Because of their superior properties, carbon-based nanostructures are the preferred choice to replace copper in next-generation integrated circuits.;The principal objective of this work is to develop a methodology to investigate the high-frequency electrical conduction in one-dimensional carbon nanostructures, in particular, carbon nanofibers (CNFs) as a potential replacement for copper in next-generation on-chips interconnects. This work is divided into four parts. First, a high-frequency test structure with low transmission is designed and fabricated, allowing capacitances less than 1fF to be measured from 0.1 to 50 GHz. Second, S-parameters from 0.1 to 50 GHz are measured for the test structure and for the test structure connected by a CNF. Third, based on this data a frequency-independent parallel-RC network that models the CNF is proposed that matches the measured S-parameters. In this model the capacitance is the only fitting parameter. Finally, an analytical approach is developed that validates the frequency-independent RC -model, whose parameters are obtained directly from measurements without fitting. This simple RC-model takes into account the effect of both the contact impedance between the signal pads and the nanofiber, as well the impedance of the nanofiber. |
