Thermoelectricity in Manganese Arsenides

Strongly correlated materials are on the forefront of fundamental research in condensed matter theory because of their fascinating and unique properties. Examples are heavy electron masses in actinide compounds, colossal magnetoresistance, or high-temperature superconductivity. These effects arise due to the strong interaction between electrons, rendering single-particle descriptions impossible. They show up in peculiar transport behavior, such as large thermopower and magnetoresistance, or non-Fermi liquid transport properties, well-known features of many correlated systems. The main objective of this project is a deeper understanding of the thermoelectric properties of manganese-based compounds that crystallize in the same structure as the iron-based pnictide superconductors and are, hence, qualitatively distinct from the better known manganese perovskites. Here, lacking large crystal-field splitting as compared to perovskites, the multi-orbital physics is qualitatively different. LaOMnAs for example, doped with charge carriers, shows an enormous Seebeck effect of 0.24 mV/K. Further estimates give power factors that are as high or even higher than for thermoelectric semiconductors. Understanding the magnetic ground state in those systems is also of big importance, because they determine largely the Fermi surface. There are lots of open questions, in particular why systems like LaOMnAs and BaMn2 As2 show different magnetic ordering patterns and why the ordering temperatures differ by a factor of two. In addition to the academic interest this has also impact on technological developments. In times of climate change and increasing prices of energy, its saving and alternative source of energy production are a very active and important research field. Understanding the transport properties of these materials will give us the possibility to apply the fundamental principles for materials design. For the practical calculations one needs highly accurate numerical methods. In recent years the combination of ab- initio methods with the Dynamical Mean-Field Theory was very successful in the description of strongly-correlated systems. The development of continuous-time Monte Carlo techniques brought a further revolution in the applicability of this method, because previously unthought-of parameter and temperature regimes became accessible. However, for the precise calculation of total energies and transport integrals the accuracy, in particular for multi-orbital systems, is still very difficult to achieve. Therefore, part of this project also aims on further development of the methodology. We will adopt the Variational Cluster Approach for a better description of total energies, and in collaboration with international partners, we will work on new developments for an efficient treatment of SU(2)-invariant interactions. This will allow a precise description of the low-energy features of the correlated systems.

In times of a drastic increase in energy consumption alternative sources of electric energy become increasingly important. Thermoelectric materials can intrinsically convert heat into electric energy, and therefore they are often considered as an alternative source of power generation. Otherwise wasted heat, e.g. from car engines or power plants, can be utilized in an environmentally friendly way. The fact that the power conversion through the Seebeck effect is essentially a material property opens the possibility of designing reliable, maintenance-free and long-living devices. Studying the underlying microscopic mechanisms that promote high thermoelectricity is a pivotal aspect not only in the search of new thermoelectric materials, but also in view of tuning and improving existing ones by targeted structural modifications, e.g. chemical substitution, external deformation or even heterostructure design. An interesting class of materials where large thermoelectric properties have been found are materials with strong electronic correlations. In these systems, the interaction between the individual electrons is so strong that the individual particles cannot move independently anymore, but the motion of every electron depends on all other electrons. Eventually this can lead to a state, where the electrons completely localize and stop moving through the crystal. This state is called a Mott insulator. The introduction of charge carriers into this electronic state leads to a very fragile metallic behavior, with reasonable electric conductivity only at very low temperatures. However, this bad metallicity is pivotal for large thermoelectric properties. The systems studied in this project are the two manganese pnictides LaMnAsO and BaMn2As2. As discussed above, these systems are insulators at their nominal chemical composition. Upon charge carrier doping, they exhibit large thermoelectric properties. We calculate these properties using state-of-the-art numerical methods, and our results are in excellent agreement with experimental data, confirming the assumption that electron correlations are important in these materials. We clarified in this project that the main reason for this is that these systems are doped Mott insulators. This situation allows for the large asymmetry in the electronic densities that leads to large thermopower. This mechanism does not only apply to the two systems that have been investigated in the course of this project, but we suggest that the doped Mott insulator is a very general set-up where large thermoelectric effects will occur. This is valuable insight for the guided search for new thermoelectric materials.