Large Scale Simulations of Conduction in Carbon Nanotube Networks

Macroscopic wires fabricated from carbon nanotubes (CNTs) are an exciting new material that aims to replicate the exceptional electric properties of microscopic individual CNTs in a macroscopic scale material. Recently there have been significant experimental advances that allow these wires to be synthesised reliably and on an industrial scale. Their electrical conductance, however, is orders of magnitude lower than expected and the origin of this poor performance is not well understood. In these wires, the CNTs form a conducting network whereby current flows along individual CNTs and also between the CNTs. This dissertation presents theoretical investigations of mechanisms that affect conduction in these networks. First, the factors that affect tunnelling rates between axially-aligned CNTs are investigated. Using scattering theory and tight-binding calculations, a novel mechanism is identified whereby momentum scattering induced by weak perturbations can greatly improve conductances between CNTs with a large chirality mismatch. It is suggested that the disorder found in realistic networks is vital to the conductance of the network as a whole and, unintuitively, that purifying these networks by removing disorder will decrease rather than increase the overall network conductivity. This mechanism is then investigated using first-principles calculations for the specific structural perturbations of terminations and bends in CNTs. An optimised methodology is described that couples density functional theory with the Landauer-Buttiker transport formalism and allows for the unbiased study of the quantum conductance of systems containing thousands of atoms. Using this methodology, it is found that small, physically realistic bends can improve the conductance between CNTs by an order of magnitude, confirming the relevance of the momentum-scattering mechanism to realistic CNT networks. A comparison is made between improvements to the conductance and the induced backscattering due to bends and a range of termination geometries. Finally, the evidence for charge doping of CNTs by adsorbed water molecules is investigated. The conclusion of earlier calculations from the literature that water dopes carbon nanotubes is suggested to be unreliable due to the difficulty in isolating charge doping from charge polarisation effects due to the permanent electric dipole moment of water. Using large-scale first-principles simulations, novel semi-classical models and an analysis of the local electronic structures it is shown that the residual charge transfer between water and carbon nanotubes, once charge polarisation have been accounted for, is negligible.
The work was funded by the Engineering and Physical Research Council and a scholarship from British Telecommunications.