New bulk thermoelectric compounds are normally discovered with the aid of simple qualitative structure-property relationships. Most good thermoelectric materials are narrow gap semiconductors composed of heavy elements with similar electronegativities. The crystal structures are usually of high symmetry (cubic, hexagonal, and possibly tetragonal), and often contain a large number of atoms per unit cell. In the present work a new structure-property relationship is discussed which links atomic displacement parameters (ADPs) and the lattice thermal conductivity of clathrate-like compounds. For many clathrate-like compounds, in which one of the atom-types is weakly bound and “rattles” within its atomic cage, room temperature ADP information can be used to estimate the room temperature lattice thermal conductivity, the vibration frequency of the “rattler”, and the temperature dependence of the heat capacity. ADPs are reported as part of the crystal structure description, and hence APDs represent some of the first information that is known about a new compound. For most ternary and quaternary compounds, all that is known is its crystal structure. ADP information thus provides a useful screening tool for the large and growing crystallographic databases. Examples of the use and limitations of this analysis are presented for several promising classes of thermoelectric materials.

The partially filled skutterudite structure is a candidate thermoelectric material with the capacity for phonon scattering by the decoupled rattling of filling ions. In this transmission electron microscopy investigation of a 1.6%Ce, 1.6%Ni, 4.9%Ge, 22.8%Co, and 69.1%Sb alloy, the structure is found to be that expected of a partially-filled skutterudite, but with a varied assortment of structural defects. These defects are characterized and their effect on thermoelectric properties is discussed.

We have used controlled crystallization of elementally modulated reactants to prepare a series of kinetically stable, crystalline skutterudites (M'xM4Sb12 where M' = vacancy, RE, Hf,…; M = Ni, Fe, Co) and crystalline superlattices composed of promising thermoelectric materials. For the bulk synthesis of skutterudites, low angle diffraction data demonstrates that the elemental layers interdiffuse at temperatures below 150°C. Nucleation of the skutterudite structure occurs with at large exotherm on annealing at temperatures below 200°C regardless of the ternary metal. All of the metastable ternary compounds and the new metastable binary compounds were found to decompose exothermically on higher temperature annealing. The decomposition temperature ranged from 250°C for the binary compound NiSb3 to above 550°C for the rare earth containing cobalt compounds. The occupation of the ternary site was found to depend on the composition of the initial reactant and was varied from 0 to 1. Full occupancy typically required an excess of the filling cation. The lattice parameters of the compounds prepared at low temperatures are distinctly smaller than those prepared using traditional synthetic approaches. High temperature annealing converts the lattice parameters of the low temperature compounds to those prepared at higher temperatures using traditional synthetic approaches. Diffraction patterns of crystalline superlattices containing skutterudites prepared using elementally modulated reactants show splitting of high angle diffraction maxima as well as the presence of the expected low angle diffraction pattern from a supperlattice. The skutterudite superlattices are stable with respect to low temperature annealing.

The thermoelectric power factor has been calculated for a two-constituent composite medium, where one constituent is a “high quality thermoelectric” while the other constituent is a “benign metal”, with large electrical and thermal conductivities but poor thermoelectric properties. It was recently discovered that, in such a mixture, the power factor could be greatly enhanced by an appropriate choice of microstructure. Here we report on a study of three periodic microstructures with cubic point symmetry under rotations: simple cubic (SC), body centered cubic (BCC), and face centered cubic (FCC) arrays of identical spheres of the benign metal embedded in the high quality thermoelectric host. We show detailed results for these microstructures in the case where the benign metal constituent is Copper, while the high quality thermoelectric constituent is the thermoelectric alloy (Bi2Te3)0.2 (Sb2Te3)0.8.

The skutterudite family of compounds continues to be of interest for thermoelectric applications due to the low thermal conductivity obtained when filling the voids with small diameter, large mass interstitials such as trivalent rare-earth ions. In the last few years there has been a substantial experimental and theoretical effort in attempting to understand the transport properties of these compounds in order to optimize their thermoelectric properties. One such approach involves partially-filling the voids in attempting to optimize the power factor while maintaining low thermal conductivity. In this report experimental research on skutterudites with the voids partially filled with heavy mass lanthanide and alkaline-earth ions is reported.

Samples of CoP3, CoAs3 and CoP1.5As1.5 have been synthesized and their thermoelectric properties measured. All three samples show semiconducting behavior. The Seebeck coefficients of CoP3 and CoAs3 are weakly dependent on temperature and are relatively small with maximum values of about 40 and 50 μV/K, respectively. The Seebeck coefficient of the solid solution gradually decreases with increasing temperature and the values are larger than those of CoP3 and CoAs3 in the temperature range investigated, with a maximum value of about 89 μV/K near room temperature. The thermal conductivity of CoP3 and CoAs3 are higher than that of CoSb3, as can be expected from the effect of anionic size on lattice vibration. A substantial reduction in thermal conductivity was observed for the solid solution compared to the constituent binary compounds due to additional phonon scattering from lattice disorder and other possible point defects such as vacancies. Other compositions in the CoP3-xAsx system have also been synthesized and their thermoelectric properties are currently being investigated to provide essential information about lattice thermal conductivity reduction by point defect scattering and to further develop strategies for optimizing the thermoelectric properties of skutterudite materials.

The temperature-dependent cross-plane thermal conductivity of a 1-μm thick 50Å Bi / 50Å Sb superlattice on a (111) CdTe substrate was measured, using a differential 3-ω method. This method uses the temperature difference between the superlattice sample and a reference sample to calculate its cross-plane thermal conductivity. However, the substrate thermal conductivity is comparable to or smaller than the superlattice thermal conductivity near room temperature. This results in a very small or negative temperature difference, making the existing data reduction method inapplicable. Based on an improved model, the temperature-dependent thermal conductivity of the Bi/Sb superlattice is obtained and is about half of the literature value of Bi0.5Sb0.5 bulk alloy.

In order to find the next generation thermoelectric (TE) material, we are focussing on the parameter Nv, the degeneracy of the band extrema in semiconductors near the Fermi energy. We attempt to synthesize ‘multivalley’ semiconductors by incorporating tetrahedral anions to introduce structural complexity while maintaining high crystallographic symmetry.

The synthesis and crystal structures of two new compounds that partially fulfill our requirements for potential TE materials are reported. Pb3(PS4)Br3 is monoclinic, space group P21/m with a=9.1531(1)Å, b=10.9508(3)Å, c=12.7953(1)Å and β=111.024(2)°. Pb3(PS4)I1.75Te0.625 crystallizes in the space group R 3 with a=9.4876(4)Å, c=46.189(3)Å. Both structures are built from alternating layers of lead halide and the thiophosphate ion (PS4)3-.

The synthesis, physicochemical, spectroscopic, and structural characterization of the compound β-K2Bi8Se13 has been previously reported. The results indicated that this material should be investigated further for possible thermoelectric applications. β-K2Bi8Se13 exhibits excellent electrical conductivity values at room temperature while maintaining high Seebeck coefficients. In this work, the optimization of the compound β-K2Bi8Se13 is continued by the introduction of varying concentrations of several different dopants. The value of x in K2Bi8– xSbxSe13 was varied in order to find the composition with minimum thermal conductivity. Where possible, transport measurements were carried out on both single crystal and polycrystalline ingot material. From these data, the trends in the key parameters were identified for optimizing the power factor and figure of merit.

Widespread applications of thermoelectric materials are limited due to low efficiency. Currently, the most widely used thermoelectric devices consist of alloys based on Bi2Te3. In such devices, the thermoelectric figure-of-merit (ZT) of bulk Bi2Te3 has been increased through doping. It is postulated that further enhancements in ZT may be attained by engineering the microstructure of the material to enhance carrier mobility while suppressing the phonon component of the thermal conductivity. This may be achieved by fabricating Bi2Te3 in the form of one-dimensional (1D) nanowires. We have deposited nanowires of Bi2Te3 with two different diameters (200 nm and 40 nm) by electrodeposition into porous anodic alumina. Characterization of the Bi2Te3/porous Al2O3 composite materials has been accomplished using X-ray diffraction (XRD), scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Energy dispersive X-ray spectroscopy (EDS) has been used to determine the stoichiometry of the wires.

A new version of a segmented thermoelectric unicouple incorporating advanced thermoelectric materials with superior thermoelectric figures of merit has been recently proposed and is currently under development at the Jet Propulsion Laboratory (JPL). This advanced segmented thermoelectric unicouple includes a combination of state-of-the-art thermoelectric materials based on Bi2Te3 and novel materials developed at JPL. The segmented unicouple currently being developed is expected to operate between 300 and about 975K with a projected thermal to electrical efficiency of up to 15%. The segmentation can be adjusted to accommodate various hot-side temperatures depending on the specific application envisioned. Techniques and materials have been developed to bond the different thermoelectric segments together for the nand p-legs and low contact resistance bonds have been achieved. In order to experimentally determine the thermal to electrical efficiency of the unicouple, metallic interconnects must be developed for the hot side of the thermocouple to connect the n- and p-legs electrically. The latest results in the development of these interconnects are described in this paper. Efforts are also focusing on the fabrication of a unicouple specifically designed for thermal and electrical testing.

Many compounds with the spinel structure type have been analyzed for their thermoelectric properties. Published data was used to augment experimental results presented here and to select promising thermoelectric spinels. Compounds studied here include Cu0.5Al0.5Cr2Se4, Cu0.5Co0.5Cr2Se4, Cu0.5In0.5Cr2Se4, and CuIr2Se4. Many exhibit low lattice thermal conductivity of about 20 mW/cmK, independent of temperature. Preliminary results are given for two series of compounds that were selected for further study: GaxCu1-xCr2Se4 and ZnxCu1-xCr2Se4.

Most optoelectronic devices for long haul optical communications are based on the InP/InGaAsP family of materials. Thin film coolers based on the same material system can be monolithically integrated with optoelectronic devices such as lasers, switches, and photodetectors to control precisely the device characteristics such as wavelength and optical power. Superlattice structures of InGaAs/InP and InGaAs/InGaAsP are used to optimize the thermionic emission resulting in a cooling behavior beyond what is possible with only the Peltier effect. A careful experimental study of these coolers is undertaken. Mesa sizes, superlattice thickness, and ambient temperature are all varied to determine their effect on cooling performance. A three-dimensional, self-consistent thermal-electric simulation and an effective one-dimensional model are used to understand the experimental observations and to predict what will occur for other untested parameters. The packaging of the coolers is also determined to have consequences in the overall device performance. Cooling on the order of 1 to 2.3 degrees over 1-micron thick barriers is reported.

We have calculated the room temperature thermal conductivity in semiconductor germanium clathrates using statistical linear-response theory and an equilibrium molecular dynamics (MD) approach. A key step in our study is to compute a realistic heat-current J (t) and a corresponding auto-correlation function < J (t) J (0) >. To ensure convergence of our results and to minimize statistical fluctuations in our calculations, we have constructed large super-cell models (2944 atoms) and have performed several independent long time simulations (>1,500 ps in each simulation). Our results show an unexpected “oscillator” character in the heat-current correlation function of the guest-free Ge clathrate frameworks. This is absent in the denser diamond phase and other with simple structural frameworks. We seek to interpret these results using lattice dynamics information. A study of the effects of the so-called “rattling” guest atoms in the open-framework clathrate materials is in progress.

We have synthesized large single crystals of clathrate-II compounds with frameworks of silicon and germanium by employing mixed alkali metal countercations. The combinations of alkali metals are rationally selected in order to fit the different cages of the clathrate-II structure. This approach leads to the following stoichiometric and fully “stuffed” compounds: Cs8Na16Si136, Cs8Na16Ge136, Rb8Na16Si136 and Rb8Na16Ge136. The structures and the corresponding Si-Si and Ge-Ge distances are elucidated and established with high accuracy from extensive single crystal X-ray diffraction work. The compounds are stoichiometric, metallic, and are very stable at a variety of extreme conditions such as heat, concentrated acids, hydrothermal treatment etc. No evidence was found for vacancies in the silicon and germanium networks or partial occupancies of the alkali metal sites. The stoichiometry of these fully “stuffed” clathrates is consistent with the measured temperature independent Pauli paramagnetism, supported also by the conductivity measurements on single crystals and thermopower measurements on pellets. A new compound with novel clathrate-like structure forms when small and large cations are combined with tin. The new materials, A6Na18Sn46 (A = K, Rb, Cs), are made of clathrate layers and the interlayer space filled with Sn4-tetrahedra and alkali-metal cations. Its formula can be rationalized as A6Na6Sn34 + 3·Na4Sn4 (one clathrate layer and three tin tetrahedra). The compound is stable in air and is being currently tested at other conditions. Detailed measurements of its transport properties are under way.

The pentatelluride materials (Hf1-XZrXTe5) have recently garnered much interest as a potential low temperature thermoelectric material. Their power factor exceeds that of the current Bi2Te3 materials over the temperature range 150 K < T < 350 K. A formidable challenge has been the capability of measuring the thermal conductivity of small needle-like samples (2.0 × 0.05 × 0.1 mm3) such as pentatellurides (HfXZr1-XTe5) due to heat loss and radiation effects. However in order to fully evaluate any material for potential thermoelectric use, the determination of the thermal conductivity of the material is necessary. We have recently developed a new technique called the parallel thermal conductance (PTC) technique to measure the thermal conductivity of such small samples. In this paper we describe the PTC method and measurements of the thermal conductivity of the pentatelluride materials will be presented for the first time. The potential of these materials for low temperature thermoelectric applications will be further evaluated given these results as well as future work and directions will be discussed.

Partially due to their lack of periodic structure, quasicrystals have inherently low thermal conductivity on the order of 1 - 3 W/m-K. AlPdMn quasicrystals exhibit favorable room temperature values of electrical conductivity, 500–800 (Ω-cm)-1, and thermopower, 80 μV/K, with respect to thermoelectric applications. In an effort to further increase the thermopower and hopefully minimize the thermal conductivity via phonon scattering, quartenary Al71Pd21Mn8-XReX quasicrystals were grown. X-ray data confirms that the addition of a fourth element does not alter the quasiperiodicity of the sample. Al71Pd21Mn8-XReX quasicrystals of varying Re concentration were synthesized where x had values of 0, 0.08, 0.25, 0.4, 0.8, 2, 5, 6, and 8. Both thermal and electrical transport property measurements have been performed and are reported.

Filled skutterudite compounds possess very low thermal conductivities due to the scattering of a wide range of phonon modes caused by a loosely bound cation incorporated in a cavity of the structure. The inclusion of such a filler cation causes several synthetic difficulties since the desired compounds are thermodynamically unstable with respect to disproportionation. Modulated elemental reactants were used in this study to circumvent these difficulties. SnxCo4Sb12 samples with x=0.5 and nearly 1.0 were synthesized using this method. To prevent nucleation of unwanted binary compounds, the repeat unit made up of elemental layers was less than 20 angstroms 500mg of each sample were produced, allowing for the samples to be hot pressed into a pellet. Structural analysis as well as measurements of the physical properties are presented.

Past studies on the thermal conductivity suggest that phonon confinement and the associated group velocity reduction are the causes of the observed drop in the cross-plane thermal conductivity of semiconductor superlattices. In this work, we investigate the contribution of phonon confinement to the in-plane thermal conductivity of superlattices and the anisotropic effects of phonon confinement on the thermal conductivity in different directions, using a lattice dynamics model. We find that the reduced phonon group velocity due to phonon confinement may account for the dramatic reduction in the cross-plane thermal conductivity, but the in-plane thermal conductivity drop, caused by the reduced group velocity, is much less than the reported experimental results. This suggests that the reduced relaxation time due to diffuse interface phonon scattering, dislocation scattering, etc, should make major contribution to the in-plane thermal conductivity reduction.

The use of Multi Quantum Well structures has been shown to provide a promising strategy for improving the thermoelectric figure of merit. In a recent paper the concept of carrier pocket engineering has been applied to strain symmetrized Si/Ge-superlattices leading to a ZT of 0.96 at room temperature for (111) orientation. Since the strain of the individual layers is crucial for the desired modification of their band structures, their experimental determination will be of importance. We have prepared a series of (111) oriented, 100 period (Si 2nm / Ge 2nm) superlattices on a graded Si0.5Ge0.5-buffer by sputter deposition. Deposition temperature and buffer thickness have been varied, the superlattices were characterized by AFM and XRD. The technique of XRD reciprocal space mappings of asymmetric reflections has been applied to describe the strain state of the superlattice. We found a buffer thickness of 1.1μm sufficient for more than 90% strain relaxation. XRD-data of 4nm-period superlattices are consistent with complete strain symmetrization.