Thermoelectricity can be defined as the art of coupling thermal and electric powers.
Looking for a way to fulfill harvesting specifications with thermoelectrics leads to well-known materials and technologies.
Considering room temperature applications, e.g. ambient air being the cold source and human skin the hot source, doped bismuth and antimony tellurides seem the best options for P & N type materials choice.
They present the highest figures of merit (ZT) in the considered temperature range, hence the best achievable conversion yield. The scientific community has been and is still extensively working on eco-friendly, abundant and low cost alternatives to tellurides, but viable breakthroughs are still to come. Considering the minimum voltage new applications ask for, it comes to light that bulk technologies are not able to fulfill the specifications, because of their relatively low density of thermoelectric legs. ‘High density” technologies claim today about one thousand legs per square centimeter with very limited options to increase significantly this figure. From an application point of view, the lower the available temperature difference is, the better the thermal coupling must be at all interfaces, considering that all constituting materials (thermoelectrics, substrates, packaging…) have already been optimized in that way. Many previous attempts to integrate thermoelectric generators, in various areas as automotive or space applications, focused on materials development (new compositions, nano-structurations) & their integration at the device level, but under-estimated the global thermal coupling optimization. Hence, an accurate versatile and multi-scale modeling for designing such materials and their thermal coupling is still lacking. Such a modeling should be able to answer the modeling methodology issue, namely that the thermal coupling determines the system. Based on an Onsager type approach, the schematic modeling of figure 1 can lead to the following system of equations linking electric and thermal currents to potentials.
Figure 1
The necessary hybrid boundary conditions introduce strong modifications in the design induced by thermal to electric feedbacks.