The efficiency of thermoelectric technology today is limited by the properties of available thermoelectric materials and a wide variety of new approaches to developing better materials have recently been suggested. The key goal is to find a material with a large ZT, the dimensionless thermoelectric figure of merit. However, if an analogy is drawn between thermoelectric technology and gas-cycle engines then selecting different materials for the thermoelements is analogous to selecting a different working gas for the mechanical engine. And an attempt to improve ZT is analogous to an attempt to improve certain thermodynamic properties of the working-gas. An alternative approach is to focus on the thermoelectric process itself (rather than on ZT), which is analogous to considering alternate cycles such as Stirling vs. Brayton vs. Rankine etc., rather than ‘merely’ considering alternative ‘gases’. Focusing on the process is a radically different approach compared to previous studies focusing on ZT. Aspects of the thermoelectric process and alternative approaches to efficient thermoelectric conversion are discussed.

In good thermoelectrics phonons have short mean free paths, and charge carriers have long ones. The other requirements are a multivalley band structure and a band gap greater than 0.1 eV for the 200 to 300 K temperature range. We discuss the use of solid state physics and chemistry concepts, along with atomic and crystal structure data, to select the new materials most likely to meet these criteria.

Characterization of thermoelectric materials can pose many problems. A temperature difference can be established across these materials as an electrical current is passed due to the Peltier effect. The thermopower of these materials is quite large and thus large thermal voltages can contribute to many of the measurements necessary to investigate these materials. This paper will discuss the characterization techniques necessary to investigate these materials and provide an overview of some of the potential systematic errors which can arise. It will also discuss some of the corrections one needs to consider. This should provide an introduction to the characterization and measurement of thermoelectric materials and provide references for a more in depth discussion of the concepts. It should also serve as an indication of the care that must be taken while working with thermoelectric materials.

When a new and promising thermoelectric material is discovered, an effort is undertaken to improve its “figure of merit”. If the effort is to be more efficient than one of trial and error with perhaps some “rule of thumb guidance” then it is important to be able to make the connection between experimental data and the underlying material characteristics, electronic and phononic, that influence the figure of merit. Transport and fermiology experimental data can be used to evaluate these material characteristics and thus establish trends as a function of some controllable parameter, such as composition. In this paper some of the generic-materials characteristics, generally believed to be required for a high figure of merit, will be discussed in terms of the experimental approach to their evaluation and optimization. Transport and fermiology experiments will be emphasized and both will be outlined in what they can reveal and what can be obscured by the simplifying assumptions generally used in their interpretation.

Some new guidelines are given that should be useful in the search for thermoelectric materials that are better than those currently available. In particular, clathrate and crypto-clathrate compounds with filler atoms in their cages offer the ability to substantially lower the lattice thermal conductivity.

Enhanced ZT has been predicted theoretically and observed experimentally in 2D quantum wells, with good agreement between theory and experiment. Advantages of low dimensional systems for thermoelectric applications are described and prospects for further enhancement of ZT are discussed.

We have measured the thermoelectric power (TEP) of MBE-grown epitaxial Bi and Bi1−xSbx alloy thin films and superlattices as a function of temperature in the range 20–300 K. We have observed that the TEP of a Bi thin film of 1 μm thickness is in good agreement with the bulk single crystal value and that the TEPs for superlattices with 400 Å and 800 Å Bi well thicknesses are enhanced over the bulk values. For x=0.072 and 0.088 in Bi1−xSbx thin films showing semiconducting behavior, TEP enhancement was observed by a factor of two. However as Bi or Bi1−xSbx well thickness decreases in superlattice geometry, the TEP decreases, which may be due to unintentional p-type doping.

Thin-film superlattice (SL) structures in thermoelectric materials are shown to be a promising approach to obtaining an enhanced figure-of-merit, ZT, compared to conventional, state-of-the-art bulk alloyed materials. In this paper we describe experimental results on Bi2Te3/Sb2Te3 and Si/Ge SL structures, relevant to thermoelectric cooling and power conversion, respectively. The short-period Bi2Te3/Sb2Te3 and Si/Ge SL structures appear to indicate reduced thermal conductivities compared to alloys of these materials. From the observed behavior of thermal conductivity values in the Bi2Te3/Sb2Te3 SL structures, a distinction is made where certain types of periodic structures may correspond to an ordered alloy rather than an SL, and therefore, do not offer a significant reduction in thermal conductivity values. Our study also indicates that SL structures, with little or weak quantum-confinement, also offer an improvement in thermoelectric power factor over conventional alloys. We present power factor and electrical transport data in the plane of the SL interfaces to provide preliminary support for our arguments on reduced alloy scattering and impurity scattering in Bi2Te3/Sb2Te3 and Si/Ge SL structures. These results, though tentative due to the possible role of the substrate and the developmental nature of the 3-ω method used to determine thermal conductivity values, suggest that the short-period SL structures potentially offer factorial improvements in the three-dimensional figure-of-merit (ZT3D) compared to current state-of-the-art bulk alloys. An approach to a thin-film thermoelectric device called a Bipolarity-Assembled, Series-Inter-Connected Thin- Film Thermoelectric Device (BASIC-TFTD) is introduced to take advantage of these thin-film SL structures.

Understanding the thermal conductivity and heat transfer processes in superlattice structures is critical for the development of thermoelectric materials and devices based on quantum structures. This work reports progress on the modeling of thermal conductivity of superlattice structures. Results from the models established based on the Boltzmann transport equation could explain existing experimental results on the thermal conductivity of semiconductor superlattices in both in plane and cross-plane directions. These results suggest the possibility of engineering the interfaces to further reduce thermal conductivity of superlattice structures.

We discuss ongoing work in three areas of thermoelectric materials research: 1) broad band semiconductors featuring anion networks, 2) filled skutterudites, and 3) polycrystalline Bi-Sb alloys. Key results include: a preliminary evaluation of a previously untested ternary semiconductor, KSnSb; the first reported data in which Sn is used as a charge compensator in filled antimonide skutterudites; the finding that Sn doping does not effect polycrystalline Bi1−xSbx as it does single crystal samples.

β-Zn4Sb3 was recently identified at the Jet Propulsion Laboratory as a new high performance p-type thermoelectric material with a maximum dimensionless thermoelectric figure of merit ZT of 1.4 at a temperature of 673K. A usual approach, used for many state-of-the-art thermoelectric materials, to further improve ZT values is to alloy β-Zn4Sb3 with isostructural compounds because of the expected decrease in lattice thermal conductivity. We have grown Zn4−xCdxSb3 crystals with 0.2≤x<1.2 and measured their thermal conductivity from 10 to 500K. The thermal conductivity values of Zn4−xCdxSb3 alloys are significantly lower than those measured for β-Zn4Sb3 and are comparable to its calculated minimum thermal conductivity. A strong atomic disorder is believed to be primarily at the origin of the very low thermal conductivity of these materials which are also fairly good electrical conductors and are therefore excellent candidates for thermoelectric applications.

TiNiSn, ZrNiSn and HfNiSn are members of a large group of intermetallic compounds which crystallize in the cubic MgAgAs-type structure. Polycrystalline samples of these compounds have been prepared and investigated for their thermoelectric properties. With thermopowers of about –200 μV/K and resistivities of a few mΩcm, power factors S2/ρ as high as 38 μW/K2 cm were obtained at 700 K. These remarkably high power factors are, however, accompanied by a thermal conductivity which is too high for applications. In order to reduce the parasitic lattice thermal conductivity, solid solutions Zrl−xHfxNiSn, Zrl−xTixNiSn, and Hfl−xTixNiSn were formed. The figure of merit of Zr0.5Hf0.5NiSn at 700 K (ZT = 0.41) exceeds the end members ZrNiSn (ZT = 0.26) and HfNiSn (ZT = 0.22).

Compositionally graded p-type Bi-Sb-Te thermoelectric material was synthesized by PIES (Pulverized and Intermixed Elements Sintering) method. The materials consisted of three segmented regions of different alloy composition, i.e. y = 0.8/0.825/0.9 in (Bi2Te3)1−y(Sb2Te3)y system. It was found that the electrical power output of the compositionally graded material was larger than that of the best single composition material when the temperature difference was the designed value.

Thermoelectric properties of polycrystalline (Bi1−xSbx)2Te3 (0.75 ≤ x ≤ 0.85), fabricated by mechanical alloying and hot pressing methods, have been investigated. Formation of (Bi0.25Sb0.75)2Te3 alloy powder was completed by mechanical alloying for 5 hours at ball- to-material ratio of 5: 1, and processing time for (Bi1−xSbx)2Te3 formation increased with Sb2Te3 content x. When (Bi0.25Sb0.75)2Te3 was hot pressed at temperatures ranging from 300°C to 550°C for 30 minutes, figure-of-merit increased with hot pressing temperature and maximum value of 2.8 × 10−3/K could be obtained by hot pressing at 550°C. When hot pressed at 550°C, (Bi0.2Sb0.8)2Te3 exhibited figure-of-merit of 2.92 × 10−3/K, which could be improved to 2.97 × 10−3/K with addition of 1 wt% Sb as acceptor dopant.

The p-type Te-doped Bi0.5Sb1.5Te3 and n-type SbI3-doped Bi2Te2.85Se0.15 thermoelectric compounds were fabricated by hot pressing in the temperature range of 380 to 440 °C under 200 MPa in Ar. Both the compounds were highly dense and showed high crystalline quality. The grains of the compounds were preferentially oriented and contained many dislocations through the hot pressing. The fracture path followed the transgranular cleavage planes, which are perpendicular to the c-axis. In addition, with increasing the pressing temperature, the figure of merit was increased. The highest values of figure of merit for the p- and n-type compounds, which were obtained at 420 °C, were 2.69 × 10−3/K and 2.35×10−3/K, respectively.

Thermoelectric power, electrical resistivity, and Hall effect of p-type Bi2−xSnxTe3 (0<x<0.03) singlecrystals have been measured in the temperature range 4.2–300K. By doping of Sn atoms into the host Bi2Te3 lattice, the enhancement in the thermoelectric power is observed in the intermediate temperature range 30–150 K for x≤0,0075. The activation type behaviour of Hall coefficient and resistivity are found which corresponds to the Sn-induced impurity band located above the second lower valence band.

The p-type Bi0.5Sb1.5Te3 compounds with Te dopant (4.0 and 6.0 wt%) and without dopant were fabricated by hot extrusion in the temperature range of 300 to 510 °C under an extrusion ratio of 20:1. The undoped and Te doped compounds were highly dense and showed high crystalline quality. The grains contained many dislocations and were fine equiaxed (˜ 1.0 μm) owing to the dynamic recrystallization during the extrusion. The hot extrusion gave rise to the preferred orientation of grains. The bending strength and the figure of merit of the undoped and Te doped compounds were increased with increasing the extrusion temperature. The Te dopant significantly increased the figure of merit. The values of the figure of merit of the undoped and 4.0 wt% Te-doped compounds hot extruded at 440 °C were 2.11×10−3/K and 2.94×10−3/K, respectively.

The thermoelectric properties of ZnSb films grown by metallorganic chemical vapor deposition (MOCVD) are reported. The growth conditions necessary to obtain stoichiometric ZnSb films and the effects of various growth parameters on the electrical conductivity and Seebeck coefficients of the films are described. The as-grown ZnSb films are p-type. It was observed that the thicker ZnSb films offer improved carrier mobilities and lower free-carrier concentration levels. The Seebeck coefficient of ZnSb films was found to rise rapidly at approximately 160°C. The thicker films, due to the lower doping levels, indicate higher Seebeck coefficients between 25 to 200°C. A short annealing of the ZnSb film at temperatures of ˜ 200°C results in reduced free-carrier level. Thermal conductivity measurements of ZnSb films using the 3-ω method are also presented.

Thermoelectric properties of polycrystalline Bi2(Te1−xSex)3 (0.05 ≤ x ≤ 0.25), fabricated by mechanical alloying and hot pressing, have been investigated. Formation of n-type Bi2(Te0.9 Se0.1)3 alloy powders was completed by mechanical alloying for 3 hours at ball-to-material ratio of 5: 1, and processing time for Bi2(Te1−xSex)3 formation increased with Bi2Se3 content x. Figure-of-merit of Bi2(Te0.9Se0.1) was markedly increased by hot pressing at temperatures above 450°C, and maximum value of 1.9 × 10−3/K was obtained by hot pressing at 550°C. With addition of 0.015 wt% Bi as acceptor dopant, figure-of-merit of Bi2 (Te0.9Se0.1)3, hot pressed at 550°C, could be improved to 2.1 × 10−3/K. When Bi2(Te1−xSex)3 was hot pressed at 550°C, figure-of-merit increased from 1.14 × 103/K to 1.92 × 10−3/K with increasing Bi2Se3 content x from 0.05 to 0.15, and then decreased to 1.30 × 103/K for x = 0.25 composition.

We have measured the thermal conductivity of TI2Mo6Se6, a quasi-one dimensional conductor which belongs to the family of M2Mo6X6 linear chain compounds. Using these results and our measurements of the Seebeck coefficient and the electrical conductivity we estimate the dimensionless figure of merit to be of the order of 0.08. This result suggest that this compound and other related compounds are good potential TE.