The present work investigates the effect of carbon nanotube (CNT) inclusions on the electrical and thermal conductivity of a thermoset epoxy resin. The characterization of electrical and thermal conductivity of CNT/epoxy composites is presented. Pristine, oxidized, and fluorine-functionalized unpurified CNT mixtures ("XD grade") were dispersed in an epoxy matrix, and the effect of stirring rate and pre-curing of the epoxy on the dispersion of the CNTs was evaluated. In order to characterize the dispersion of the CNTs at different length scales, Optical Microscopy (OM), Raman Spectroscopy, and Scanning Electron Microscopy (SEM) was performed. Samples of varying CNT weight fractions were fabricated in order to find the effect of CNT weight fraction on thermal and electrical conductivity. Electrical conductivity was measured using a dielectric spectrometer, and thermal conductivity was determined by a transient plane source thermal analyzer. It was found that electrical conductivity increases by orders of magnitude for the pristine and oxidized XD CNT composites relative to the neat epoxy matrix, while fluorinated XD CNT composites remain electrically non-conductive. A small, but significant, increase in thermal conductivity was observed for pristine, oxidized, and fluorinated XD CNT composites, showing a linear increase in thermal conductivity with increasing CNT weight fraction. Pristine XD CNTs were ball-milled for different times in order to reduce the degree of agglomeration and entanglement of CNTs, and composites were fabricated using the same technique as with non-milled XD CNTs. Using ball-milled CNTs shows improved dispersion but results in an electrically non-conductive composite at the CNT weight fractions tested. The thermal conductivity of the ball-milled CNT samples shows an initial increase higher than that of non-milled pristine, oxidized, and fluorinated XD CNTs, but remains constant with increasing CNT weight fraction. A micromechanics model based on the composite cylinders method was implemented to model the electrical and thermal conductivity of the CNT/epoxy composites. Nanoscale effects in electrical and thermal conduction, such as electron hopping and interface thermal resistance, respectively, were incorporated into the model in order to accurately predict the acquired results. Modeling results show good agreement with acquired experimental results.

abstract: Although high performance, light-weight composites are increasingly being used in applications ranging from aircraft, rotorcraft, weapon systems and ground vehicles, the assurance of structural reliability remains a critical issue. In composites, damage is absorbed through various fracture processes, including fiber failure, matrix cracking and delamination. An important element in achieving reliable composite systems is a strong capability of assessing and inspecting physical damage of critical structural components. Installation of a robust Structural Health Monitoring (SHM) system would be very valuable in detecting the onset of composite failure. A number of major issues still require serious attention in connection with the research and development aspects of sensor-integrated reliable SHM systems for composite structures. In particular, the sensitivity of currently available sensor systems does not allow detection of micro level damage; this limits the capability of data driven SHM systems. As a fundamental layer in SHM, modeling can provide in-depth information on material and structural behavior for sensing and detection, as well as data for learning algorithms. This dissertation focusses on the development of a multiscale analysis framework, which is used to detect various forms of damage in complex composite structures. A generalized method of cells based micromechanics analysis, as implemented in NASA's MAC/GMC code, is used for the micro-level analysis. First, a baseline study of MAC/GMC is performed to determine the governing failure theories that best capture the damage progression. The deficiencies associated with various layups and loading conditions are addressed. In most micromechanics analysis, a representative unit cell (RUC) with a common fiber packing arrangement is used. The effect of variation in this arrangement within the RUC has been studied and results indicate this variation influences the macro-scale effective material properties and failure stresses. The developed model has been used to simulate impact damage in a composite beam and an airfoil structure. The model data was verified through active interrogation using piezoelectric sensors. The multiscale model was further extended to develop a coupled damage and wave attenuation model, which was used to study different damage states such as fiber-matrix debonding in composite structures with surface bonded piezoelectric sensors. / Dissertation/Thesis / Ph.D. Mechanical Engineering 2012