New low thermal conductivity materials for thermoelectric applications

A low lattice thermal conductivity is one of the requirements for achieving high thermoelectric figures of merit. Several low thermal conductivity materials were identified and developed over the past few years at JPL, including filled skutterudites and Zn/sub 4/Sb/sub 3/-based materials. A study of the mechanisms responsible for the high phonon scattering rates in these compounds has demonstrated that materials with highly disordered or complex structures which can accommodate additional atoms in their lattice are likely to have low lattice thermal conductivity values. Several cluster compounds, including the Chevrel phases (Mo/sub 6/Se/sub 8/-type) and Re/sub 6/Te/sub 15/, are just such materials and are currently being investigated at JPL. The crystal structures of the Chevrel phases present cavities which can greatly vary in size and can contain a large variety of atoms ranging from large ones such as Pb to small ones such as Cu. These atoms are not localized in the structure and, depending on their size, can move between different sites and may produce significant phonon scattering. Although most of the Chevrel phases studied until now were reported to be metallic, it was found that semiconducting Chevrel phases can be engineered by controlling the number of electrons per [Mo/sub 6/] cluster. Initial results obtained on some cluster Chevrel phases and Re/sub 6/Te/sub 15/ are presented and discussed. These materials possess very low thermal conductivity values (/spl sim/10 mW/cmK at 300 K) but optimization of their electronic properties will be required to achieve high thermoelectric figures of merit.