The thermoelectric performance of nanostructured low dimensional silicon and silicon-germanium has been functionally compared device-wise. The arrays of nanowires of both materials, grown by a VLS-CVD (Vapor-Liquid-Solid Chemical Vapor Deposition) method, have been monolithically integrated in a silicon micromachined structure in order to exploit the improved thermoelectric properties of nanostructured silicon-based materials. The device architecture helps to translate a vertically occurring temperature gradient into a lateral temperature difference across the nanowires. Such thermocouple is completed with a thin film metal leg in a unileg configuration. The device is operative on its own and can be largely replicated (and interconnected) using standard IC (Integrated Circuits) and MEMS (Micro-ElectroMechanical Systems) technologies. Despite SiGe nanowires devices show a lower Seebeck coefficient and a higher electrical resistance, they exhibit a much better performance leading to larger open circuit voltages and a larger overall power supply. This is possible due to the lower thermal conductance of the nanostructured SiGe ensemble that enables a much larger internal temperature difference for the same external thermal gradient. Indeed, power densities in the μW/cm2 could be obtained for such devices when resting on hot surfaces in the 50–200 °C range under natural convection even without the presence of a heat exchanger.
Photonic crystals (PCs) are built to control the propagation of light within their structure. These can be used for an assortment of applications where custom designed devices are of interest. Among them, one-dimensional PCs can be produced to achieve the reflection of specific and broad wavelength ranges. However, their design and fabrication are challenging due to the diversity of periodic arrangement and layer configuration that each different PC needs. In this study, we present a framework to design high reflecting PCs for any desired wavelength range. Our method combines three stochastic optimization algorithms (Random Search, Particle Swarm Optimization and Simulated Annealing) along with a reduced space-search methodology to obtain a custom and optimized PC configuration. The optimization procedure is evaluated through theoretical reflectance spectra calculated by using the Equispaced Thickness Method, which improves the simulations due to the consideration of incoherent light transmission. We prove the viability of our procedure by fabricating different reflecting PCs made of porous silicon and obtain good agreement between experiment and theory using a merit function. With this methodology, diverse reflecting PCs can be designed for any applications and fabricated with different materials.