Research Projects

Thermoelectric power generation

Thermoelectric materials, used to convert heat into electricity, have applications ranging from industrial waste-heat recovery, to remote sensing, to powering space exploration.

Simple thermoelectric couple 

Here, two thermoelectric ‘legs’ made of n-type and p-type semiconductors are electrically in series and thermally in parallel. For power generation, the legs are placed across a temperature gradient, causing the electronic carriers to move towards the cold end, and generating a voltage across each leg. This behavior, known as the Seebeck effect, is the basis of thermoelectric energy conversion. In reverse, the same device operates as a Peltier cooler, in which an applied voltage generates a temperature gradient. 

Arguably the most important current application of thermoelectric power generation is radioisotope thermoelectric generators (RTGs), which have played a critical role in space exploration since the 1960’s.  RTGs are the sole-power source for NASA’s deep-space probes, they provided power on the lunar surface during the Apollo missions, and they are currently being used to power the Curiosity Mars rover.


An RTG consists of a cylindrical core filled with plutonium-oxide, which provides a hot side temperature above 1100 C for at least 30 years.  Multiple independent thermoelectric modules are wrapped around the core, taking advantage of the 900 C temperature delta.


Here on Earth, thermoelectric materials also have a major role to play.  More than 50% of the energy produced nationwide is lost as waste heat. The recovery of even a small fraction of this energy through the use of thermoelectric generators has the potential to significantly impact the global energy landscape.  However, further improvements in cost and efficiency of thermoelectric materials are required for thermoelectric generators to be widely implemented.   The efficiency of thermoelectric energy conversion depends on the temperature delta (via the Carnot efficiency) and the material figure of merit, zT. Traditional thermoelectric materials have zT near 1, which corresponds to about 10% of the Carnot efficiency. However there have been an increasing number of reports of larger zT through enhancements in both electronic and thermal properties


A high figure of merit requires (1) maximizing the power output, which is proportional to the Seebeck coefficient squared and (2) minimizing parasitic sources of voltage and heat loss, which are proportional to the electrical resistivity and thermal conductivity respectively.

We therefore need materials with high Seebeck coefficients, high electrical conductivity, and low thermal conductivity. However, this combination of properties is exceptionally difficult to achieve simultaneously – posing a fundamental dilemma for those developing materials with improved zT.


Figure 1.1b illustrates the balance that must be achieved in order to obtain high thermoelectric efficiency; a large Seebeck coefficient is found in low carrier concentration (n) insulators, while high electrical conductivity is found in high carrier concentration metals. As a consequence, most good thermoelectric materials are heavily doped semiconductors with carrier concentrations between 10^19 to 10^21 carriers/cm^3.

Single Crystal Growth

One of the most significant barriers to understanding the intrinsic electronic and thermal properties of complex semiconductors is the scarcity of large single crystal samples. This is particularly true for materials with highly anisotropic structures, for which measurements of bulk samples yield only characteristics of the average properties.

We use several high temperature techniques to grow single crystals of semiconducting and ceramic materials.  Optical floating zone growth is a powerful method for materials with very high melting temperatures and relatively low vapor pressures,  because of the very high temperatures achievable in the hot-zone and because it requires no crucible.


An optical floating zone furnace consists of several  halogen lamps set in elliptical mirrors (four, in our case).  The mirrors focus the light on the center of the chamber, where the sample feed rod and seed rod are suspended inside of an evacuated or inert gas-filled quartz tube. The feed rod hangs from a wire at the top, and the seed rod is mounted on an alumina rod at the bottom.  At the first stage of the heating, the feed and seed rod are separated by a few millimeters.  When heat is applied, the ends of the feed and seed rod wiII begin to melt, at which point, the upper feed rod is moved downwards until it touches the seed rod to form a stable melt zone. During growth, both rods are moved downwards, so that the melt zone slowly progressed up the length of the feed rod.

The floating zone method can also be adapted to materials with incongruent melting temperatures or higher vapor pressures by using a flux.  The flux (i.e., solvent material) must have a lower melting temperature than the target material, and it should not form any phases with the target material except for the desired phase.

Schematic illustration of the traveling solvent zone method.  The flux is melted first, and connected to the feed and see rods.  As the melt-zone moves upwards, the feed dissolves into the flux and precipitates out onto the seed rod.

Our lab’s floating zone furnace from Crystal Systems Corp. is expected to arrive in December 2016, allowing us to grow materials with melting temperatures as high as 2200 C.  We also have facilities for forming  and sintering the polycrystalline feed and seed rods, cutting single crystals, and orienting crystals using a Laue camera.

Crystal Systems Corp floating zone furnace (model FZ-T-2000-H-NI-VPO-PC)

Resonant Ultrasound Spectroscopy

Resonant ultrasound spectroscopy (RUS) is an elegant approach to characterize the elastic constants of bulk and single crystal samples by measuring the vibrational resonances (eigenfrequencies).

The RUS has a drive transducer that sweeps a range of frequencies and a pick-up transducer that records the response. Strong peaks occur only at the resonant frequencies, which are a function of the elastic constants, sample geometry, and sample density.   A complete analytical solution for the eigenfrequencies does not exist, so determination of elastic constants relies on the comparison of the experimental spectrum to computed values (predicted from the known sample density and dimensions)  in an iterative process until these discrepancies are minimized.


Resonant ultrasound spectroscopy requires no fixed contacts to the sample, making it convenient to probe changes in the elastic properties as a function of temperature and anomalies associated with phase changes.  RUS is ideal for measuring the full elastic tensor of anisotropic single crystals, and it is often the only approach to accurately measuring the elastic constants.


Schematic illustration of the high temperature RUS setup in our lab, which uses buffer rods to contact the sample at the center of a furnace (600 C max temp)

 Thermal and electronic transport properties

We collaborate with Dr. Morelli in the CHEMS department and Tim Hogan in ECE to measure the thermal and electronic transport properties of our materials.  We can measure the electrical conductivity and the Seebeck coefficients up to 800 C and the thermal diffusivity and heat capacity up to 1000 C.

Zintl phase thermoelectrics: Low thermal conductivity in complex crystals

Much of our research focuses on a fascinating class of thermoelectric compounds called Zintl phases.  The term “Zintl phase” does not refer to a single structural pattern, but rather, a class of intermetallic phases that follow similar bonding rules:


The number of covalent bonds formed by the anions in a Zintl phase depends on the number of valence electrons needed for filled orbitals minus the number of valence electrons available.  This rule