Analysis of thermoelectric properties in arsenic-based structures using DFT

Thermoelectric materials are useful tools for energy conversion that are becoming
increasingly ubiquitous in the modern world thanks to their reliability, their autonomy
and their compact nature. Their primary limitation is usually their relatively low
efficiency, which struggles to be competitive with other energy conversion systems.
This thesis focuses on the computational analysis of semiconductors with complex
Fermi surfaces in order to identify properties that may be beneficial to thermoelectric
function. In doing so, it attempts to refine a methodology built on first principles cal-
culations so as to realistically evaluate the characteristics of highly anisotropic crystals,
in a field where this approach has traditionally been applied to isotropic materials.
Despite the goal specified of thermoelectricity, the properties studied included elec-
trical transport properties that may be of interest in multiple other fields of material
research. As such, a priority throughout this body of work is to test the strengths,
versatility and limitations of a new methodology while applying it on computationally
challenging materials.
These are the different phases of As2Te3 and As2S3, along with doped and alloyed
variants. These materials have, as of the time of writing, been subject to relatively
little computational research despite key similarities to well-known thermoelectrics.
As a result, a variety of key behaviours have been identified in correlation to ther-
moelectric function, and promising figure of merit values have been predicted for
As2Te3. While less promising overall, As2S3 has also shown that it can offer some
interesting thermoelectric characteristics if specifically selected for, while revealing
some of the computational and theoretical limitations of our models.
Our method was effective for materials that are very computationally challenging and
physically anisotropic, and involved steps that for the most part were quite systematic.
This suggests that it may be possible to incorporate its use into a high-throughput
evaluation of computationally cheaper materials for thermoelectric optimization.