Microfabrication and Testing of a Thermoelectric Device for Generating Mobile Electrical Power

This technology can be used to power robotics, enable portable/wearable power, and provide power from vehicle waste heat energy.

Several attractive features of thermo-electric (TE) technology include no moving parts, light weight, modularity, covertness, silence, high power density, low amortized cost, and long service life with no required maintenance. Many of the potential uses for mounted/ dismounted power, such as recharging batteries, are therefore ideal for TE technologies. However, these applications will require more interconnected, smaller-scale modular devices than are currently available. Most commercial off-the-shelf (COTS) TE devices are optimized for cooling, not for generating power, so new device structures with materials and geometries better optimized for power generation are needed for broader use of TE technologies.

The Thermoelectric Device: (A) break-out view of the individual components, (B) view of the separate hot- and cold-junctions, and (C) the final device after flip-chip assembly.
New miniaturization and fabrication techniques exploit recent developments in materials with improved ZT, or index of power conversion efficiency. Well-known materials with good efficiency in TE power generators that can be miniaturized were chosen. N-type lead telluride (PbTe) and p-type antimony telluride (Sb2Te3) were identified for the initial device, with longer-term interest in alternate materials whose thermal conductivity may be reduced by incorporating nanostructures, or whose electrical properties could be improved with quantum-confinement heterostructures. Although PbTe and Sb2Te3 are generally much more efficient at higher temperatures, they are suitable for showing that in this quite moderate temperature range, the new miniaturization concepts are highly effective for the applications of interest.

The PbTe and Sb2Te3 TE legs were polished to a specular finish to reduce uncertainty in geometry and thus, reduce measurement error. Each sample had similar geometries: 0.40 cm in length, 0.25 cm in width, and 0.10 cm in depth. All geometrical dimensions were measured using a micrometer thickness gauge, which yields measurements that are accurate to about 2%.

The metal junctions of the device were fabricated from oxygen-free copper stock so that the contribution to the total electrical resistivity from the hot and cold junctions is negligible. Further, the thermal conductivity is significantly better in copper than either of the semiconductors, so the junctions remain isothermal. To maintain permanent electrical isolation between the different sections of the TE module, the metal junctions were mounted on a thin square aluminum nitride (AlN) platen having an area of 6.25 cm2.

The components are shown in the figure at the various stages of assembly. In (A), the large copper heat sink, the AlN top isolation with serpentine trace, one of the metal junctions (positioned next to the copper heat-sink), and one of the Sb2Te3 components (positioned next to the length scale) are shown. In (B), the PbTe and Sb2Te3 components are shown fully integrated with copper hot-junctions permanently mounted to the top AlN isolator. Between that and the length scale are the three copper cold-junctions permanently mounted onto the bottom AlN isolator.

To proceed with the “flip-chip” assembly, a special mechanical x-y translation stage was designed to accurately manipulate and strategically position the hot-junction onto the cold junctions permanently mounted onto the copper heat sink. Once in position, the metallurgical junctions were formed by application of pressure and heat. Fabricated in this way, the obtained device shown in (C) is mechanically rugged and fully operational.

The initial characteristics of the device were determined immediately after assembly. The device’s total electrical resistance was experimentally found to be 0.432 Ω and remained stable throughout thermal testing. The resistivity of the oxygen-free copper is 1.7 × 10-6 Ω-cm, so the contribution to the total device resistance from the sum of the resistance from all six copper junctions is more than almost five orders of magnitude less than that from the semiconductor materials, so the dominant contributions would be only from the semiconductors, and potentially from the average contact resistance, Rc.

For higher-temperature tests, the device was mounted in a vacuum system to prevent oxidation of the materials and reduce error from convective heat flow. Two type-K thermocouples were used to independently determine the temperature hot-junction at the top of the device, and a third thermocouple was used to measure the temperature of the cold-junction on the large heat sink on the bottom.

To test the device for power generation, the hot-junction was heated radiatively by a 30 W “HotWatt” electrical power resistor. The resistor was not in contact with the TE device so as to minimize thermal mass. The output power was measured from the two braided copper terminals (shown in C in the figure). To assure that the device had reached a steady-state temperature difference, the range of temperature difference was slowly scanned from ΔT = 0 to 100 K. To normalize this electrical power to the area from which it was generated, the cross-sectional area of all the elements of the device was calculated and the power density was determined.

The high power density also points to the utility of this technology for producing even more highly miniaturized and integrated state-of-the-art power sources. Because this fabrication assembly yields low contact resistance, it can be scaled to microscopic dimensional scales consistent with integrated circuits (ICs) and micro electromechanical systems (MEMS). Such a demonstration would show excellent potential for providing power for auton omous microsystems, robotics, and portable/wearable power for mounted/dismounted units. Although this device is not perfectly impedance-matched, its performance under thermal testing is excellent and can be used to qualify its potential for power generation in more realistic applications. One application could be to provide electrical power from the waste heat energy from vehicles that would otherwise be dissipated and lost to the environment.

This work was done by Patrick J. Taylor, Brian Morgan, and Bruce Geil of the Army Research Laboratory; and Nibir K. Dhar of the Defense Advanced Research Projects Agency (DARPA). ARL-0055



This Brief includes a Technical Support Package (TSP).
Document cover
Microfabrication and Testing of a Thermoelectric Device for Generating Mobile Electrical Power

(reference ARL-0055) is currently available for download from the TSP library.

Don't have an account? Sign up here.