Silicon NanoWires (NWs) and ultra-thin layers are promising building blocks for electronic and thermoelectric devices. Length scale and orientation provide additional degrees of freedom in engineering the electronic and thermal transport properties. We use the sp3d5s*-SO tight-binding model and Boltzmann transport theory, including all relevant scattering mechanisms, to investigate the thermoelectric properties of Si NWs. It is found that structural quantization below 10nm can severely affect the electronic properties of NWs by changing the curvature of the bands and altering degeneracies through valley and subband splitting. Specifically for p-type NWs, it was found that at large diameters, NWs oriented along the three principle orientations [100], [110] and [111], have a similar thermoelectric power factor. A large anisotropy in the thermoelectric power factor was found, however, for smaller diameters. As the diameter is scaled to 3nm, the power factor of the [111] and secondly the [110] NWs largely increases, whereas that of the [100] NWs remains low. This behavior originates from confinement-induced large curvature variations in the electronic subbands. In addition to electron transport we also investigated phonon transport in Si nanostructures. Si NWs with diameters in the range 1-10nm are considered. The lattice dynamics are modeled using the modified valence force field method, the ballistic thermal conductance is calculated using the Landauer transport formalism. The phonon group velocity and thermal conductance can vary by a factor of two depending on the geometrical features of the channel. Group velocity and thermal conductance is highest in the <110> NW and lowest in and the <111> NW. The <111> orientation is the most suitable for thermoelectric devices based on Si NWs. We also consider ultra-thin Si layers of major surface orientations {100}, {110}, {111}, and {112}. We find that the ballistic thermal conductance in the thin layers is anisotropic, with the {110}/<110> channels exhibiting the highest and the {112}/<111> channels the lowest thermal conductance. The resulting ratio is about two.
To model electronic transport in Quantum Cascade Lasers (QCL) and Quantum Cascade Detectors (QCD) we resort to the Pauli master equation. An efficient Monte Carlo (MC) simulator as part of the Vienna-Schrödinger-Poisson (VSP) simulation framework has been further enhanced. Several band structure models such as 2-band k·p or 4-band k·p can be combined with different in-plane dispersion relations used in the transport calculation. A model for stimulated emission and absorption of photons has been implemented. The simulator has been used to design and optimize the first functioning bi-functional QCL and QCD device. In mid-infrared and terahertz devices, novel types of optical guides and resonators are commonly found, such as ring cavities, micro-discs, photonic crystals, super-crystals and micro-antennas. Of particular interest are Photonic Crystal (PHC) cavities. This allows investigating the properties of a large finite PHC based on the analysis of a single unit cell. In our work a real-space approach is used. The periodicity is explicitly ensured by connecting the mesh vertices that lie on opposite surfaces of the unit cell. Thus, the unit cell can be made periodic only in certain spatial dimensions while different boundary conditions may be applied in the remaining dimensions. The real-space approach with mixed boundaries gives us the possibility to analyze PHC slabs. To capture this radiative dissipation effect we apply periodic boundary conditions in the horizontal plane and absorbing boundary conditions below and above the slab. Doing so, we obtain a complex photonic band structure which contains wavevector-dependent information about radiative losses for every PHC mode.
To model carrier transport in Graphene NanoRibbons (GNRs) the Non-Equilibrium Green's Function (NEGF) formalism and an atomistic tight-binding model have been employed. We investigate the effect of line-edge roughness using a non-perturbative approach. In this method, roughness is applied to many GNR samples using a Gaussian or exponential distribution and then ensemble averages are computed. Line-edge roughness at the two edges can have some degrees of correlation. In GNRs obtained by unzipping of carbon nanotubes the correlation coefficient between theses two edges is +1, whereas that for GNRs obtained from other methods is nearly zero. Our studies show that this correlation can play a significant role on the electronic properties of GNRs. In GNRs with un-correlated roughness, the electronic bandgap is strongly modulated by line-edge roughness.