Ordering phenomena in materials often have a crucial impact on materials properties. They are governed by the competition between entropy and energy. Accordingly simulating these aspects requires the construction of models that enable an computationally efficient exploration of the relevant configuration space. Alloy cluster expansions are a technique that is particular suitable for this task as they can be trained to reach high accuracy while being computationally suitable for rapid sampling via Monte Carlo techniques.
In this thesis alloy cluster expansions have been applied in combination with Monte Carlo simulations to study the ordering behavior in various inorganic clathrates. Inorganic clathrates constitute a class of systems with a cage-like framework that can trap loosely bound atoms or even small molecules. These systems are small band gap semiconductors and have a very low lattice thermal conductivity, which gives rise to very good thermoelectric properties. Additionally the host atoms and cage framework can be occupied by a wide range of elements which provides extensive opportunities for property optimization. Inorganic clathrates are thus good examples for systems with a high degree of variability in composition, for which ordering phenomena play a crucial role.
In paper I we studied the ordering behaviour of Ba8Ga16Ge30. Configurations representative for different annealing temperatures were extracted from Monte Carlo simulations and further analyzed to obtain the temperature dependency of the thermoelectric power factor. These data was subsequently used to construct a cluster expansion for the power factor itself, which enabled us to optimize the chemical ordering that maximizes this property. The approach developed in this work is generalizable and can be adapted to other materials.
In paper II we studied the ordering behavior and related properties in the clathrate systems Ba8AlxSi46-x, Ba8AlxGe46-x, Ba8GaxGe46-x, and Ba8GaxSi46-x as a function of composition and temperature. We achieved very good agreement with the available experimental data for the site occupancy factors (SOFs). This enabled us to reconcile experimental data from different sources and explain the nonmonotonic variations of the SOFs. In particular, we provided a rationale for the extreme SOF behavior with varying composition observed in Al based clathrates.