Johannes de Boor

Measuring thermoelectric transport properties of materials

In this review we discuss considerations regarding the common techniques used for measuring thermoelectric transport properties necessary for calculating the thermoelectric figure of merit, zT. Advice for improving the data quality in Seebeck coefficient, electrical resistivity, and thermal conductivity (from flash diffusivity and heat capacity) measurements are given together with methods for identifying possible erroneous data. Measurement of the Hall coefficient and calculation of the charge carrier concentration and mobility is also included due to its importance for understanding materials. It is not intended to be a complete record or comparison of all the different techniques employed in thermoelectrics. Rather, by providing an overview of common techniques and their inherent difficulties it is an aid to new researchers or students in the field. The focus is mainly on high temperature measurements but low temperature techniques are also briefly discussed.

On the True Indium Content of In-Filled Skutterudites

The incongruently melting single-filled skutterudite InxCo4Sb12 is known as a promising bulk thermoelectric material. However, the products of current bulk syntheses contain always impurities of InSb, Sb, CoSb, or CoSb2, which prevent an unbiased determination of its thermoelectric properties. We report a new two-step synthesis of high-purity InxCo4Sb12 with nominal compositions x = 0.12, 0.15, 0.18, and 0.20 that separates the kieftite (CoSb3) formation from the topotactic filler insertion. This approach allows conducting the reactions at lower temperatures with shorter reaction times and circumventing the formation of impurity phases. The synthesis can be extended to other filled skutterudites. High-density (>98%) pellets for thermoelectric characterization were prepared by current-assisted short-time sintering. Sample homogeneity was demonstrated by potential and Seebeck microprobe measurements of the complete pellet surfaces. Synchrotron X-ray diffraction showed a purity of 99.9% product with traces (≤0.1%) of InSb in samples of nominal composition In0.18Co4Sb12 and In0.20Co4Sb12. Rietveld refinements revealed a linear correlation between the true In occupancy and the lattice parameter a. This allows the determination of the true In filling in skutterudites and predicting the In content of unknown AxCo4Sb12. The high purity of InxCo4Sb12 allowed studying the transport properties without bias from side phases. A figure of merit close to unity at 420 °C was obtained for a sample of a true composition of In0.160(2)Co4Sb12 (nominal composition In0.18Co4Sb12). The lower degree of In filling has a dramatic effect on the thermoelectric properties as demonstrated by the sample of nominal composition In0.20Co4Sb12. The presence of InSb in amounts of ∼0.1 vol% led to a substantially lower degree of interstitial site filling of 0.144, and the figure of merit zT decreased by 18%, which demonstrates the significance of the true filler atom content in skutterudite materials.

Simulations for the Development of Thermoelectric Measurements

In thermoelectricity, continuum theoretical equations are usually used for the calculation of the characteristics and performance of thermoelectric elements, modules or devices as a function of external parameters (material, geometry, temperatures, current, flow, load, etc.). An increasing number of commercial software packages aimed at applications, such as COMSOL and ANSYS, contain vkernels using direct thermoelectric coupling. Application of these numerical tools also allows analysis of physical measurement conditions and can lead to specifically adapted methods for developing special test equipment required for the determination of TE material and module properties. System-theoretical and simulation-based considerations of favorable geometries are taken into account to create draft sketches in the development of such measurement systems. Particular consideration is given to the development of transient measurement methods, which have great advantages compared with the conventional static methods in terms of the measurement duration required. In this paper the benefits of using numerical tools in designing measurement facilities are shown using two examples. The first is the determination of geometric correction factors in four-point probe measurement of electrical conductivity, whereas the second example is focused on the so-called combined thermoelectric measurement (CTEM) system, where all thermoelectric material properties (Seebeck coefficient, electrical and thermal conductivity, and Harman measurement of zT) are measured in a combined way. Here, we want to highlight especially the measurement of thermal conductivity in a transient mode. Factors influencing the measurement results such as coupling to the environment due to radiation, heat losses via the mounting of the probe head, as well as contact resistance between the sample and sample holder are illustrated, analyzed, and discussed. By employing the results of the simulations, we have developed an improved sample head that allows for measurements over a larger temperature interval with enhanced accuracy.

Improvement of electrical conductivity in Pb0.96-yMn0.04SnyTe alloys for high temperature thermoelectric applications

Lead–tin–telluride is a well-known thermoelectric material in the temperature range 350–750 K. Here, this alloy doped with manganese (Pb0.96−yMn0.04SnyTe) was prepared for different amounts of tin. X-ray diffraction showed a decrease of the lattice constant with increasing tin content, which indicated solid solution formation. Microstructural analysis showed a wide distribution of grain sizes from <1 μm to 10 μm and the presence of a SnTe rich phase. All the transport properties were measured in the range of 300−720 K. The Seebeck coefficient showed that all the samples were p-type indicating holes as dominant carriers in the measurement range. The magnitude increased systematically on reduction of the Sn content due to possible decreasing hole concentration. Electrical conductivity showed the degenerate nature of the samples. Large values of the electrical conductivity could have possibly resulted from a large hole concentration due to a high Sn content and secondly, due to increased mobility by sp–d orbital interaction between the Pb1−ySnyTe sublattice and the Mn2+ ions. High thermal conductivity was observed due to higher electronic contribution, which decreased systematically with decreasing Sn content. The highest zT = 0.82 at 720 K was obtained for the alloy with the lowest Sn content (y = 0.56) due to the optimum doping level.

Nanostructured rocksalt-type solid solution series (Ge{sub 1−x}Sn{sub x}Te){sub n}Sb{sub 2}Te{sub 3} (n=4, 7, 12; 0≤x≤1): Thermal behavior and thermoelectric properties

Solid solutions (Ge1−xSnxTe)nSb2Te3 (n=4, 7, 12; 0≤x≤1) represent stable high-temperature phases and can be obtained as metastable compounds by quenching. High-resolution transmission electron microscopy reveals that the quenched (pseudo-)cubic materials exhibit parquet-like nanostructures comparable to, but especially for n=4 more pronounced than in (GeTe)nSb2Te3 (GST materials). The temperature-dependent phase transitions are comparable; however, substitution with Sn significantly lowers the transition temperatures between cubic high-temperature phase and the long range ordered layered phases that are stable at ambient conditions. In addition, the metrics of the quenched Sn-containing materials remains closer to cubic, especially for samples with n=7 or 12. For samples with high defect concentrations (n=4, 7), Sn-substituted samples exhibit electrical conductivities up to 3 times higher than those of corresponding GST materials. Since the difference in thermal conductivity is much less pronounced, this results in a doubling of the thermoelectric figure of merit (ZT) at high temperatures for (Ge0.5Sn0.5Te)4Sb2Te3 as compared to (GeTe)4Sb2Te3. Sn doping in (GeTe)7Sb2Te3 increases the ZT value by a factor of up to 4 which is also due to an increased Seebeck coefficient.