Role of Nanostructures in Reducing Thermal Conductivity below Alloy Limit in Crystalline Solids Woochul Kim, Suzanne Singer and Arun Majumdar Department of Mechanical Engineering University of California Berkeley, CA 94720, USA Joshua Zide and Arthur Gossard Department of Materials University of California Santa Barbara, CA 93106, USA Ali Shakouri Department of Electrical Engineering University of California Santa Cruz, CA 95064, USA Abstract Atomic substitution in alloys can efficiently scatter phonons, thereby significantly reducing the thermal conductivity in crystalline solids to the “alloy limit”. It has been difficult to beat the alloy limit without creating defects, dislocations, and voids, which also reduce electrical conductivity, making it ineffective for increasing the material’s thermoelectric figure of merit. Using In 0.53 Ga 0.47 As containing epitaxially embedded ErAs nanoislands a few nm in size, we demonstrate thermal conductivity reduction by almost a factor of two below the alloy limit, and corresponding increase in thermoelectric figure of merit by more than a factor of two. A theoretical model suggests that while point defects in alloys efficiently scatter short wavelength phonons, the ErAs nanoislands provides additional scattering mechanism for the mid to long wavelength phonon – the combination reduces the thermal conductivity below the alloy limit. Introduction The performance of thermoelectric energy conversion devices depends on the thermoelectric figure of merit (ZT) of a material, which is defined as ZT = S 2 σT/k where S, σ, k, and T are the Seebeck coefficient, electrical conductivity, thermal conductivity and absolute temperature, respectively. Low thermal conductivity and high power factor (S 2 σ) are essential for efficient operation of thermoelectric devices. Over the past five decades, it has been challenging to increase ZT > 1, since modifying one parameter in ZT affects the others due to their interdependence [1, 2]. Recent reports have shown, however, that it is possible to ZT > 1 by nanostructuring thermoelectric materials [3-5]. While the original goal for nanostructuring was to increase S 2 σ due to quantum confinement of carriers [6, 7], experiments [3-5] have shown that the key reason for ZT > 1 was the reduction of thermal conductivity. Yet, the fundamental reasons for how and why nanostructuring reduces thermal conductivity in crystalline materials are not fully understood. In this paper, we experimentally and theoretically show that it is possible to reduce thermal conductivity by a factor ~ 2 below the “alloy limit” in crystalline materials, thus laying down some principles of designing nanostructured thermoelectric materials. Historically, it has been difficult to reduce the thermal conductivity of crystalline solids below that of an alloy without creating defects, dislocations, and voids – often called the “alloy limit” of thermal conductivity in crystalline solids. For example, thermal conductivity of pressure-sintered Si 0.8 Ge 0.2 alloy was shown to be less than that of the crystalline alloy due to heavy point defects [8]. However, the figure of merit was not increased due to proportional reduction in electrical conductivity. There have been reports that the thermal conductivity of Si/Ge superlattice can be lower than that of Si x Ge 1-x alloy [9, 10]. However, because of the large lattice mismatch (~ 4%) between Si and Ge, the strain between Si and Ge in Si/Ge superlattices produces defects and dislocations when the layer thickness exceeds the critical value. Such approaches also have not led to ZT > 1, thus suggesting that the electrical conductivity also reduces proportionally. More recently, despite systematically increasing the interfacial acoustic impedance mismatch in Si/Si x Ge 1-x or Si y Ge 1-y /Si x Ge 1-x superlattices, Huxtable et al. [11, 12] failed to reduce the thermal conductivity below that of Si x Ge 1-x alloy without creating significant defects in the superlattice. There are very few instances, however, where the thermal conductivity was reduced below the alloy limit [13, 14], while maintaining the crystalline structure of the material. Using GaAs/AlAs superlattices, Capinski et al. [13] showed that only when the period thickness was in the range of a few nm, the cross – plane thermal conductivity was less than that of an Al 0.5 Ga 0.5 As alloy. Venkatasubramanian [14] measured the cross – plane thermal conductivity of Bi 2 Te 3 /Sb 2 Te 3 superlattices and found the lattice conductivity of short-period (a few nm) superlattices to be less than those of solid solution alloy. It has been theoretically proposed that the thermal conductivity in such periodic structures is reduced due to the formation of phonon bandgaps [15], akin to the formation of 0-7803-9552-2/05/$20.00 ©2005 IEEE 9 2005 International Conference on Thermoelectrics