Local and non-local shot noise in multiwalled carbon nanotubes

Abstract

We have investigated shot noise in multiterminal, diffusive multiwalled carbon nanotubes (MWNTs) at 4.2 K over the frequency f = 600 − 850 MHz. Quantitative comparison of our data to semiclassical theory, based on non-equilibrium distribution functions, indicates that a major part of the noise is caused by a non-equilibrium state imposed by the contacts. Our data exhibits non-local shot noise across weakly transmitting contacts while a low-impedance contact eliminates such noise almost fully. We obtain F tube < 0.03 for the intrinsic Fano factor of our MWNTs.Multiwalled carbon nanotubes (MWNTs) are miniscule systems, their diameter being only a few nanometers. Yet, in surprisingly many cases their transport properties can be described with incoherent theories, interference effects showing up only through weak localization [1,2]. This is in contrast to single-walled tubes, where interference effects dominate, and give rise for example to Fabry-Perot resonances with distinctive features in conductance and current noise [3].When interference effects are washed out, semiclassical analysis based on non-equilibrium distribution functions is adequate, and the circuit theory of noise becomes a powerful tool in considering nanoscale objects [4,5]. This theory makes it straightforward to calculate current noise of incoherent dots and wires, and to relate the current noise to the transmission properties of the corresponding section of the mesoscopic object. Semiclassical analysis provides a way to make a distinction between sample and contact effects, and thereby it allows one to investigate contact phenomena, of which only a little is known in carbon nanotube systems.We have investigated the influence of contacts on the shot noise in multiterminal, diffusive carbon nanotubes. We have made four-lead measurements on MWNTs in which two middle probes have been employed for noise measurements. We show that quantitative information can be obtained from such measurements using semiclassical circuit theory in the analysis. We find that probes with contact resistance R C < 1 kΩ act as strongly inelastic probes, resulting in incoherent, classical addition of noise of two adjacent sections, while "bad" contacts (R C ∼ 10 kΩ) act as weakly perturbing probes which need to be analysed on the same footing as the other parts of the sample. We also find that good contacts eliminate noise that couples to the probe from a nonneighboring voltage biased section. In addition, we find from our analysis that the tubes themselves are quite noise-free, with a Fano-factor F tube < 0.03. As far as we know, our results are the first shot noise measurements addressing the contact issues in carbon nanotubes.To clarify the results of our multi-probe noise measurements, let us consider the three-terminal structure depicted schematically in Fig. 1. Assume that the aver-age current I flows between 1 and 2, and the average potential of the terminal 3 adjusts to the potential of the node. In our work, terminal 3 is disconnected from the ground at low frequenc...

Shot noise suppression and hopping conduction in graphene nanoribbons

“…Indeed, even in the VRH regime, the apparent Fano factor could be formed by other means, for example by noise from the graphene islands at the ends, and that the actual shot noise from the ribbon nearly vanishes. This would be reminiscent to carbon nanotubes where very small F have been observed in various configurations [26,27]. Nearly total suppression of shot noise indicates very effective energy relaxation at finite bias which could be realized by disorder-enhanced electron-phonon coupling [28] or by relaxation via new degrees of freedom provided by the edges of the ribbon.…”