We built and tested a subterranean thermoelectric power source that converts diurnal heat flow through the upper soil layer into electricity. This paper describes the operation, design, and performance of the device. Key features of the power source include the use of bismuth-telluride thermopiles optimized for small ΔT and aerogel insulation to minimize thermal losses. The device weighs 0.24 kg and was designed with a flat form factor measuring 12 × 12 × 1.7 cm to facilitate modularity, packing, and assembly into larger arrays. One full year of field testing was performed between June 2009 and May 2010 in Albuquerque, New Mexico where the device generated an average power output of 1.1 mW. The season with the highest performance was spring (March–May) while the season of lowest performance was winter (November–January). During May 2010, the device generated an average power of 1.5 mW and a peak power of 9.8 mW at 9.3 V. Ten years of continuous operation at 1.1 mW would yield an energy density and specific energy of 1384 W h/L and 1430 W h/kg respectively, which is competitive with chemical batteries and is orders of magnitude greater than published subterranean and ambient thermoelectric harvesters. Numerical simulations show that performance is sensitive to the thermal properties of the soil and environmental conditions. This class of energy harvester may provide an option for supplemental power, or possibly primary power, for low power remote sensing applications.
In order to design a thermoelectric (TE) module suitable for long-term elevated temperature use, the Department 8651 has conducted parametric experiments to study material compatibility and thermal aging of TE materials. In addition, a comprehensive material characterization has been preformed to examine thermal stability of P- and N-based alloys and their interaction with interconnect diffusion barrier(s) and solder. At present, we have completed the 7-days aging experiments for 36 tiles, from ambient to 250 C. The thermal behavior of P- and N-based alloys and their thermal interaction with both Ni and Co diffusion barriers and Au-Sn solder were examined. The preliminary results show the microstructure, texture, alloy composition, and hardness of P-(Bi,Sb){sub 2}Te{sub 3} and N-Bi{sub 2}(Te,Se){sub 3} alloys are thermally stable up to 7 days annealing at 250 C. However, metallurgical reactions between the Ni-phosphor barriers and P-type base alloy were evident at temperatures {ge} 175 C. At 250 C, the depth (or distance) of the metallurgical reaction and/or Ni diffusion into P-(Bi,Sb){sub 2}Te{sub 3} is approximately 10-15 {micro}m. This thermal instability makes the Ni-phosphor barrier unsuitable for use at temperatures {ge} 175 C. The Co barrier appeared to be thermally stable and compatible with P(Bi,Sb){sub 2}Te{sub 3} at all annealing temperatures, with the exception of a minor Co diffusion into Au-Sn solder at {ge} 175 C. The effects of Co diffusion on long-term system reliability and/or the thermal stability of the Co barrier are yet to be determined. Te evaporation and its subsequent reaction with Au-Sn solder and Ni and Co barriers on the ends of the tiles at temperatures {ge} 175 C were evident. The Te loss and its effect on the long-term required stoichiometry of P-(Bi, Sb){sub 2}Te{sub 3} are yet to be understood. The aging experiments of 90 days and 180 days are ongoing and scheduled to be completed in 30 days and 150 days, respectively. Material characterization activities are continuing for the remaining tiles.
We have built and tested a miniaturized, thermoelectric power source that can provide in excess of 450 {micro}W of power in a system size of 4.3cc, for a power density of 107 {micro}W/cc, which is denser than any system of this size previously reported. The system operates on 150mW of thermal input, which for this system was simulated with a resistive heater, but in application would be provided by a 0.4g source of {sup 238}Pu located at the center of the device. Output power from this device, while optimized for efficiency, was not optimized for form of the power output, and so the maximum power was delivered at only 41mV. An upconverter to 2.7V was developed concurrently with the power source to bring the voltage up to a usable level for microelectronics.