Lead telluride is a reference high-performance thermoelectric which possesses a remarkable efficiency at intermediate temperatures in its pristine form, largely due to its low lattice thermal conductivity. Several studies have been proposed to analyse this thermal conductivity, but their range of application is usually limited: first-principles simulations do not allow for the study of intrinsic defects in large structures and the available classical potentials tend to overestimate the lattice thermal conductivity of the bulk material. Here, we present an optimised Coulomb-Buckingham potential that provides an improved description of the material when compared to other force fields in the literature, especially for the bulk lattice thermal conductivity, and that allows for the study of intrinsic defects. Using this potential, we studied how intrinsic defects influence the thermal conductivity, paying special attention to vacancies, interstitials, voids and grain boundaries. Our results show that the presence of both vacancies and grain boundaries can separately reduce the value of the thermal conductivity to $\sim 0.5$ W/mK, regardless of the temperature, i.e. a factor of 4 at room temperature. A similar decrease is observed in the presence of interstitials and voids, but in this case concentrations are more limited. However, small grains are not stable at high and intermediate temperatures, so in the second part of the thesis we study if grain boundaries survive or not against thermal coarsening under operational conditions. We attacked this problem by means of a multiscale dynamical modelling approach applied to porous, polycrystalline PbTe through a phase-field model informed by molecular dynamics simulations. The main hypothesis considered was that voids act as Zener pinning particles to stop grain growth and stabilise grain boundaries. We analyzed the stability of voids and determined metastable void sizes at different temperatures and observed that the presence of voids can stabilise grains with radii of the order of hundreds of nm. A simple serial model allowed us to calculate the thermal conductivity of these porous polycrystalline materials and showed that it can be reduced by a 35 %, with lower reductions at higher temperatures.
|Date of Award||Dec 2020|
- Queen's University Belfast
|Sponsors||Northern Ireland Department for the Economy & Science Foundation Ireland|
|Supervisor||Jorge Kohanoff (Supervisor), Myrta Grüning (Supervisor) & Tchavdar Todorov (Supervisor)|
- Thermal conductivity
- molecular dynamics