Graduate Students:  Scott Huxtable and Alexis Abramson

Thermoelectric devices have a wide variety of applications in many different fields.  For example, solid state thermoelectric refrigerators are used to cool high power integrated circuits, superconducting devices, and infrared detectors, while thermoelectric power converters are utilized in waste heat energy conversion and deep space probes.  The performance of a thermoelectric material may be quantified using the dimensionless figure of merit ZT=S²sT/k, where S is the Seebeck coefficient, s is the electrical conductivity, and k is the thermal conductivity.  Presently, the best thermoelectric materials are doped BiTe alloys, which have a ZT ~ 1 at room temperature. Previous work has suggested that when thermoelectric materials are nanostructured into quantum wells and wires, ZT can increase beyond 2 and even as high as 4 or 5.  This is due to both quantum confinement of electrons as well as higher impedance to phonon transport.  Thermoelectric refrigerators and power generators made out of such materials can have performance much superior to their vapor and gas-based counterparts, offering the promising prospects of fully solid-state and environmentally benign energy conversion devices in the future.

Recently, with advances in material science there has been a renewed interest in the search for materials with a higher ZT.  Our interest lies with superlattice heterostructures.  A superlattice is a periodic structure generally consisting of several to hundreds of alternating thin film layers where each layer is typically between 10 and 500 Angstroms thick.  Superlattices are grown either by Molecular Beam Epitaxy (MBE) or Metal Organic Chemical Vapor Deposition (MOCVD) to ensure high quality films.  Semiconductor superlattice structures have shown promise as thermoelectric materials for their high power factor (S²s) and low thermal conductivity.  While the power factor of a superlattice can be controlled through band gap engineering and doping, prediction and control of thermal conductivity has remained a challenge.  In moderately doped semiconductors, heat conduction is dominated by phonons (lattice vibrations) as electrons contribute very little to the thermal conductivity.  Phonon transport in semiconductor superlattices is controlled by two mechanisms: interface scattering by acoustic impedance mismatch, defects and/or roughness, and phonon filtering by Bragg reflection.  Reduction of superlattice thermal conductivity below its bulk or alloy values has, in the past, been attributed mainly to interface scattering. However, systematic experiments to distinguish between the two mechanisms have not yet been performed.  Currently, we are studying several systems including InGaAs/InP, InGaAs/InGaAsP, and Si/SiGeC.  The data on the InGaAs/InP superlattices proved to be interesting.  Although the thermal conductivity of InP is an order of magnitude higher than that of InGaAs, we observed that as the fraction of InP is increased within the superlattice layers, the overall thermal conductivity of the superlattice decreases (see following figure).  The reason for this trend is unclear at the present time, although work is underway to develop an understanding of this characteristic.

Through the use of systematic variations of the number of interfaces, periodicity, and thickness ratio of the superlattice samples we hope to gain a better understanding of phonon transport in these structures.  Our goal is to then utilize this experimental data along with some theoretical work in order to design of a new class of high ZT heterostructures.