Efficient Thermoelectrics based on Silicon Nanomeshes

Thermoelectric devices convert heat flow into useful electrical power. They are characterized by the figure of merit ZT, which is around unity for some of the best thermoelectric bulk materials such as bismuth telluride (Bi2Te3) and lead telluride (Pb2Te3). Efficiency of these thermoelectrics, however, is still too low to enable wide spread use in energy harvesting and cooling applications. Another problem of these materials is the rareness of Te. Nanostructures, on the other hand, provide the opportunity to design the properties of materials such that high thermoelectric performance can be achieved. This is also feasible for abundant, cost effective, and even poor thermoelectric starting materials such as silicon with a ZT value of around 0.01. A ZT value around one was already demonstrated for Si-based nanocomposites, 1D nanowires, and 2D superlattices. A novel nanostructure proposed recently is the Si nanomesh, also known as nano-porous Si. This structure benefits from well established and less expensive fabrication processes. In this project, we theoretically investigate thermoelectric transport in Si-based nanomeshes. An optimized thermoelectric material needs to have low thermal and high electrical conductivity. Important tasks of the project are, therefore, the analysis of i) the electronic and phononic bandstructures, and ii) the electron and phonon transport properties in nanomeshes. Geometries with feature sizes from a few hundred down to a few nanometers are considered. Accordingly, we employ k*p models on the continuum level and the tight-binding model sp3d5s* on the atomistic level for electronic bandstructure calculation. For the phonon bandstructures we also employ both continuum methods and the atomistic valence-force-field method. For electronic transport we resort to both semiclassical (Boltzmann transport) and quantum mechanical (non-equilibrium Green`s functions) approaches, for phonon transport to diffusive as well as coherent methods depending on the device length scales. We investigate design concepts for the electronic and phononic properties of these artificial lattices to maximize thermoelectric efficiency. In particular, we address fundamental open questions such as: i) The possibility of relaxing the usual interdependence of the Seebeck coefficient and the electrical conductivity that limits the power factor, ii) Development of understanding towards electronic bandstructure engineering in nanomeshes, iii) Possible ways to engineer energy bands to alter the density of states by proper filling of the pores of the nanomeshes to improve performance, and iv) Design directions of the phonon modes of the nanomesh using the "phononic crystal" concept for drastically reducing the thermal conductivity. An expected outcome is to theoretically demonstrate designs that will achieve ZT > 3 in a large range of operating temperatures, a value required for broad, economic application.

An estimated 70% of all power generated in the world is lost as waste heat. Thermoelectrics are a special class of materials that have the capability to harvest usable renewable energy from this heat. However, the efficiency of this power generation has remained stubbornly low for decades, which has relegated the technology to only niche applications. To enhance the power harvesting efficiency of these materials, there has been ongoing research into the potential of nano-technology and material design. However, for most materials, such nano-structuring is expensive and the best ways of structuring, to obtain maximal increases in efficiency, are not fully understood. In general, a high thermoelectric efficiency requires a material with low thermal conductivity and simultaneously a high electrical conductivity. The goals of this project were two-fold: i) to explore the potential of silicon for thermoelectric application. Silicon is the most widely used material in semiconductor industry, and nano-structuring approaches are already well developed and cheap; and ii) to develop material design strategies to maximize thermoelectric performance. During the project a number of specific nano-structuring strategies were explored and their potential for enhancing thermoelectric energy harvesting efficiency assessed. First, the effect of the insertion of tiny, regular arrays of pores intended to hamper heat flow has been explored by means of Monte Carlo calculations. It was found that this approach is quite effective provided the pores were large and their surfaces rough. Second, the effect of electrostatic doping and modulation doping of silicon nano-wires on the thermoelectric efficiency was investigated. It was shown that these methods can theoretically increase the so-called thermoelectric power factor five times. Another set of studies explored the potential for efficiency enhancement of inserting thin alternating layers of a different material into a system. It was found, for optimal structures, that moderate enhancement is possible, but that for rough and imprecise structures that this enhancement is destroyed completely. Further exploration revealed that this degradation from non-ideal structures could be mitigated to some extent by choosing insertion materials with low thermal conductivity.