Thermoelectric effects in carbon nanostructures

Heat from electrical devices, car engines, industrial processes and even our own body heat are an abundant, albeit difficult to harvest, energy source in today's society. As the extent of the climate crisis becomes clearer, recovering such waste heat can be an important step to design improved devices and reduce greenhouse gas emissions as well as the use of fossil fuels. Thermoelectric materials and devices are uniquely positioned to recover this waste heat and transform it to electricity due to their conversion characteristics and potentially small size. In addition, ever-increasing demands on computing power as well as further downscaling in chip sizes necessitates on-chip spot cooling and temperature sensing, which can be provided by thermoelectric devices. So far, whenever research in thermoelectrics seemed to be at a dead end, a new approach, such as for example most recently nanostructuring and reduced dimensionality, have reinvigorated the field. While this has lead to unprecedented advances in the figure of merit, now regularly reported above unity, thermoelectric devices are still lacking the required conversion efficiency to compete with conventional heat engines, leaving them an option mostly for niche applications. Particularly in low-dimensional devices and structures, theoretically predicted high and competitive efficiency values are hard to obtain due to environmental factors and inherent limits in device design. In this dissertation, influences on the thermoelectric properties of carbon nanostructures due to coupling configurations, defects, geometrical considerations and other local effects are studied. A first experiment investigates the power factor in C₆₀ molecules, contacted via electroburned graphene nanoconstrictions. The results suggest a threefold way to increase the achievable power factor in zero-dimensional structures: positioning the molecular energy levels close to the Fermi energy levels of the leads, ensuring the tunnel coupling is optimized for the desired operating temperature and, in order to achieve a maximum power factor, the tunnel couplings to the source and drain need to be equal. Building on these conclusions, a temperature dependent study of the power factor in graphene quantum dots was performed, suggesting that a quantum dot can work as a quantum heat valve especially when contacted by non-ideal heat conducting leads. Apart from highlighting the importance of leads that conduct heat well to achieve a high performance, the findings also again emphasize the potentially detrimental effect tunnel coupling has on the power factor. In order to analyse more localised effects, a Scanning Thermal Microscopy (SThM) approach was used to record thermoelectric maps (of both the Seebeck and Peltier effect) of graphene bow ties, revealing a geometrically dependent change in the Seebeck coefficient due to the impact of electron scattering at the graphene edges. This effect is exploited later to fabricate single-material graphene thermocouples that achieve an order of magnitude higher sensitivity than previously reported single-material metal thin film thermocouples. Lastly, Scanning Thermal Gate Microscopy (STGM) is developed, improving on the previously used SThM technique to achieve high resolution in thermovoltage maps at a fast scanning rate. STGM is then applied to graphene single-layer/bilayer junctions, demonstrating the influence of metallic contacts, layer thickness changes and local strain on the spatially distributed Seebeck coefficient in graphene. The effects studied in this dissertation are aimed at helping to further the understanding of local thermoelectric properties and opening the path to improve on current device design and performances as well as facilitating the elimination of parasitic noise caused by undesirable thermoelectric effects.