Computational design of carbon nanotube sensors for gas phase explosives detection

Gas phase detection of explosive molecules is a sensing application of wide interest. Light weight, low power sensors are needed for mobility and wide dissemination, however low vapor pressures and the presence of similar functional groups in a variety of explosive molecules make the development of sensitive and selective detection systems difficult. Experimental research has reported some success in the development of carbon nanotube based explosives sensors, however safety considerations and strict controls on the distribution of explosive materials hamper experimental progress. In this dissertation, ab initio computational models are developed for metallic and semiconducting carbon nanotube sensors, in a variety of device configurations. Their chemiresistive sensing performance is investigated in the detection of three common explosives. The effects of doping, lattice defects, and functionalizations on sensing performance are analyzed. Their performance in sensor arrays is also analyzed; array selectivity is improved by capitalizing upon the nonlinear current-voltage characteristics of the CNT sensors. A new ab initio molecular dynamics formulation is developed, for spin polarized systems. It employs a novel nonholonomic Hamiltonian modeling methodology to couple a quantum model of the electronic structure to a molecular model of the nuclear dynamics, and quantifies the modeled nanoscale systems interaction with the external thermal and electromagnetic environment. This theoretical model offers future opportunities for the simulation of finite temperature dynamics in carbon nanotube based sensors, under applied electric and magnetic fields.