Modelling thermoelectric performance in nanoporous nanocrystalline silicon

TL;DR

This study theoretically investigates how nanostructuring, specifically porosity and grain boundaries, affects thermoelectric performance in silicon by balancing thermal conductivity reduction against charge carrier scattering.

Contribution

It provides a detailed analysis of the tradeoffs in thermoelectric performance due to nanostructuring in silicon using molecular dynamics and Green's function methods.

Findings

Thermal conductivity reduction exceeds power factor decrease at low porosities.

Including grain boundaries enhances ZT figure of merit.

Porosity and grain boundaries can be engineered to optimize thermoelectric efficiency.

Abstract

Introducing hierarchical disorder from multiple defects into materials through nanostructuring is one of the most promising directions to achieve extremely low thermal conductivities and thus improve thermoelectric performance. The success of nanostructuring relies on charge carriers having shorter mean-free-paths than phonons so that the latter can be selectively scattered. Nevertheless, introducing disorder into a material often comes at the expense of scattering charge carriers as well as phonons. In order to determine the tradeoff between the degradation of the lattice thermal conductivity and of the power factor due to this, we perform a theoretical investigation of both phonon and electron transport in nanocrystalline, nanoporous Si geometries. We use molecular dynamics for phonon transport calculations and the non-equilibrium Green's function method for electronic transport. We report on the engineering tradeoff that the porosity (number of pores and their in-between distance) has on the overall thermoelectric performance for the material optimization. We indeed find that the reduction in thermal conductivity is stronger compared to the reduction in the power factor, for the low porosities considered in this study (up to 5 %), and that the ZT figure of merit can experience a large increase, especially when grain boundaries are included, compared to just nanoporosity.