Electrochemical deposition of semiconductors for thermoelectric devices
Thermoelectric microgenerators are an approach to power conversion which can deliver milliwatts to watts of power at a few to several tens of volts.
SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Thermoelectric microgenerators are an approach to power conversion which can deliver milliwatts to watts of power at a few to several tens of volts.
Multilayer thermoelectric devices now at the prototype stage of development exhibit a combination of desirable characteristics, including high figures of merit and high performance/cost ratios. These devices are capable of producing temperature differences of the order of 50 K in operation at or near room temperature. A solvent-free batch process for mass production of these state-of-the-art thermoelectric devices has also been developed. Like prior thermoelectric devices, the present ones have commercial potential mainly by virtue of their utility as means of controlled cooling (and/or, in some cases, heating) of sensors, integrated circuits, and temperature-critical components of scientific instruments. The advantages of thermoelectric devices for such uses include no need for circulating working fluids through or within the devices, generation of little if any noise, and high reliability. The disadvantages of prior thermoelectric devices include high power consumption and relatively low coefficients of performance. The present development program was undertaken in the hope of reducing the magnitudes of the aforementioned disadvantages and, especially, obtaining higher figures of merit for operation at and near room temperature. Accomplishments of the program thus far include development of an algorithm to estimate the heat extracted by, and the maximum temperature drop produced by, a thermoelectric device; solution of the problem of exchange of heat between a thermoelectric cooler and a water-cooled copper block; retrofitting of a vacuum chamber for depositing materials by sputtering; design of masks; and fabrication of multilayer thermoelectric devices of two different designs, denoted I and II. For both the I and II designs, the thicknesses of layers are of the order of nanometers. In devices of design I, nonconsecutive semiconductor layers are electrically connected in series. Devices of design II contain superlattices comprising alternating electron-acceptor (p)-doped and electron-donor (n)-doped, nanometer- thick semiconductor layers.
Nanowire thermoelectric devices, now under development, are intended to take miniaturization a step beyond the prior state of the art to exploit the potential advantages afforded by shrinking some device features to approximately molecular dimensions (of the order of 10 nm). The development of nanowire-based thermoelectric devices could lead to novel power-generating, cooling, and sensing devices that operate at relatively low currents and high voltages. Recent work on the theory of thermoelectric devices has led to the expectation that the performance of such a device could be enhanced if the diameter of the wires could be reduced to a point where quantum confinement effects increase charge-carrier mobility (thereby increasing the Seebeck coefficient) and reduce thermal conductivity. In addition, even in the absence of these effects, the large aspect ratios (length of the order of tens of microns diameter of the order of tens of nanometers) of nanowires would be conducive to the maintenance of large temperature differences at small heat fluxes. The predicted net effect of reducing diameters to the order of tens of nanometers would be to increase its efficiency by a factor of .3. Nanowires made of thermoelectric materials and devices that comprise arrays of such nanowires can be fabricated by electrochemical growth of the thermoelectric materials in templates that contain suitably dimensioned pores (10 to 100 nm in diameter and 1 to 100 microns long). The nanowires can then be contacted in bundles to form devices that look similar to conventional thermoelectric devices, except that a production version may contain nearly a billion elements (wires) per square centimeter, instead of fewer than a hundred as in a conventional bulk thermoelectric device or fewer than 100,000 as in a microdevice. It is not yet possible to form contacts with individual nanowires. Therefore, in fabricating a nanowire thermoelectric device, one forms contacts on nanowires in bundles of the order of 10-microns wide. The fill factor for the cross-section of a typical bundle is about 1/2. Nanowires have been grown in alumina templates with pore diameters of 100 and 40 nm.
Thermoelectric junctions made of semiconductors have existed in radioisotope thermoelectric generators (RTG) for deep space missions, but are currently being adapted for terrestrial energy harvesting. Unfortunately, these devices are inefficient, operating at only 7% efficiency. This low efficiency has driven efforts to make high-figure-of-merit thermoelectric devices, which require a high electrical conductivity but a low thermal conductivity, a combination that is difficult to achieve. Lowered thermal conductivity has increased efficiency, but at the cost of power output. An alternative setup is to use metallic junctions rather than semiconductors as thermoelectric devices. Metals have orders of magnitude more electrons and electronic conductivities higher than semiconductors, but thermal conductivity is higher as well. To evaluate the viability of metallic junction thermoelectrics, a two dimensional heat transfer MATLAB simulation was constructed to calculate efficiency and power output. High Seebeck coefficient alloys, Chromel (90%Ni-10%Cr) and Constantan (55%Cu-45%Ni), produced efficiencies of around 20-30%. Parameters such as the number of layers of junctions, lateral junction density, and junction sizes for both series- and parallel-connected junctions were explored.
A thermoelectric power generation device is disclosed using one or more mechanically compliant and thermally and electrically conductive layers at the thermoelectric material interfaces to accommodate high temperature differentials and stresses induced thereby. The compliant material may be metal foam or metal graphite composite (e.g. using nickel) and is particularly beneficial in high temperature thermoelectric generators employing Zintl thermoelectric materials. The compliant material may be disposed between the thermoelectric segments of the device or between a thermoelectric segment and the hot or cold side interconnect of the device.
Thermoelectric generators (TEG) are devices that generate energy by converting heat into electricity or provide cooling via the Peltier effect. This feature of thermoelectric devices originates from the Seebeck, Peltier, Thomson, and Joule heating effects. TEGs can be applied in energy and thermal management systems such as waste heat recovery and refrigeration, respectively. Thermoelectric device design is influenced by the material selection and the device's geometry operating conditions. Therefore, predicting, verifying, and validating thermoelectric device performance using simulations tools is essential to deploying thermoelectric devices in industry. The Multiphysics Object-Oriented Simulation Environment (MOOSE) Framework is an open-source simulation tool capable of modeling simple to complex systems. In this work, we demonstrate MOOSE's thermoelectric device modeling capabilities by simulating a unicouple, module, and exhaust gas recovery system. The Seebeck, Peltier, Thomson, and Joule heating physics are implemented into MOOSE. The MOOSE thermoelectric physics were thoroughly verified and validated using published COMSOL® results and experimental data. In addition, thermoelectric modules were integrated into an exhaust gas recovery system using the MOOSE MultiApp function as a demonstration of the model's ability. The verification and validation results and exhaust gas heat recovery system showcases MOOSE's capability to model thermoelectric devices and integrate these devices into practical energy systems.
Silica aerogels have been shown to be attractive for use as thermal-insulation materials for thermoelectric devices. It is desirable to thermally insulate the legs of thermoelectric devices to suppress lateral heat leaks that degrade thermal efficiency. Aerogels offer not only high thermal- insulation effectiveness, but also a combination of other properties that are especially advantageous in thermoelectric- device applications. Aerogels are synthesized by means of sol-gel chemistry, which is ideal for casting insulation into place. As the scale of the devices to be insulated decreases, the castability from liquid solutions becomes increasingly advantageous: By virtue of castability, aerogel insulation can be made to encapsulate devices having any size from macroscopic down to nanoscopic and possibly having complex, three-dimensional shapes. Castable aerogels can permeate voids having characteristic dimensions as small as nanometers. Hence, practically all the void space surrounding the legs of thermoelectric devices could be filled with aerogel insulation, making the insulation highly effective. Because aerogels have the lowest densities of any known solid materials, they would add very little mass to the encapsulated devices. The thermal-conductivity values of aerogels are among the lowest reported for any material, even after taking account of the contributions of convection and radiation (in addition to true thermal conduction) to overall effective thermal conductivities. Even in ambient air, the contribution of convection to effective overall thermal conductivity of an aerogel is extremely low because of the highly tortuous nature of the flow paths through the porous aerogel structure. For applications that involve operating temperatures high enough to give rise to significant amounts of infrared radiation, opacifiers could be added to aerogels to reduce the radiative contributions to overall effective thermal conductivities. One example of an opacifier is carbon black, which absorbs infrared radiation. Another example of an opacifier is micron- sized metal flakes, which reflect infrared radiation. Encapsulation in cast aerogel insulation also can help prolong the operational lifetimes of thermoelectric devices that must operate in vacuum and that contain SiGe or such advanced skutterudite thermoelectric materials as CoSb3 and CeFe3.5Co0.5Sb12. The primary cause of deterioration of most thermoelectric materials is thermal decomposition or sublimation (e.g., sublimation of Sb from CoSb3) at typical high operating temperatures. Aerogel present near the surface of CoSb3 can impede the outward transport of Sb vapor by establishing a highly localized, equilibrium Sb vapor atmosphere at the surface of the CoSb3.
An all-solid-state diffusion bonding process that exploits the eutectoid reaction between molybdenum and titanium has been developed for use in fabricating thermoelectric devices based on skutterudite compounds. In essence, the process is one of heating a flat piece of pure titanium in contact with a flat piece of pure molybdenum to a temperature of about 700 C while pushing the pieces together with a slight pressure [a few psi (of the order of 10 kPa)]. The process exploits the energy of mixing of these two metals to form a strong bond between them. These two metals were selected partly because the bonds formed between them are free of brittle intermetallic phases and are mechanically and chemically stable at high temperatures. The process is a solution of the problem of bonding hot-side metallic interconnections (denoted hot shoes in thermoelectric jargon) to titanium-terminated skutterudite n and p legs during the course of fabrication of a unicouple, which is the basic unit cell of a thermoelectric device (see figure). The hot-side operating temperature required for a skutterudite thermoelectric device is 700 C. This temperature precludes the use of brazing to attach the hot shoe; because brazing compounds melt at lower temperatures, the hot shoe would become detached during operation. Moreover, the decomposition temperature of one of the skutterudite compounds is 762 C; this places an upper limit on the temperature used in bonding the hot shoe. Molybdenum was selected as the interconnection metal because the eutectoid reaction between it and the titanium at the ends of the p and n legs has characteristics that are well suited for this application. In addition to being suitable for use in the present bonding process, molybdenum has high electrical and thermal conductivity and excellent thermal stability - characteristics that are desired for hot shoes of thermoelectric devices. The process takes advantage of the chemical potential energy of mixing between molybdenum and titanium. These metals have a strong affinity for each other. They are almost completely soluble in each other and remain in the solid state at temperatures above the eutectoid temperature of 695 C. As a result, bonds formed by interdiffusion of molybdenum and titanium are mechanically stable at and well above the original bonding temperature of about 700 C. Inasmuch as the bonds are made at approximately the operating temperature, thermomechanical stresses associated with differences in thermal expansion are minimized.
Thermoelectric (TE) devices heat, cool, and generate electricity when a temperature differential is provided between the two module faces. In cooperation with NASA, Chico, California-based United States Thermoelectric Consortium Inc. (USTC) built a gas emissions analyzer (GEA) for combustion research. The GEA precipitated hydrocarbon particles, preventing contamination that would hinder precise rocket fuel analysis. The USTC research and design team uses patent-pending dimple, pin-fin, microchannel and microjet structures to develop and design heat dissipation devices on the mini-scale level, which not only guarantee high performance of products, but also scale device size from 1 centimeter to 10 centimeters. USTC continues to integrate the benefits of TE devices in its current line of thermal management solutions and has found the accessibility of NASA technical research to be a valuable, sustainable resource that has continued to positively influence its product design and manufacturing
Thermal insulation and structural support for thermoelectric devices, discussing fabrication of diffusion bonded structures in conjunction with material selection
A thermoelectric device formed of nanowires on the nm scale. The nanowires are preferably of a size that causes quantum confinement effects within the wires. The wires are connected together into a bundle to increase the power density.
Application of thermoelectric devices as thermal control coatings for spacecraft
A material having useful thermoelectric properties was synthesized by combining indium-tin-oxide (ITO) with a NiCoCrAlY alloy/alumina cermet. This material had a very large Seebeck coefficient with electromotive-force-versustemperature behavior that is considered to be excellent with respect to utility in thermocouples and other thermoelectric devices. When deposited in thin-film form, ceramic thermocouples offer advantages over precious-metal (based, variously, on platinum or rhodium) thermocouples that are typically used in gas turbines. Ceramic thermocouples exhibit high melting temperatures, chemical stability at high temperatures, and little or no electromigration. Oxide ceramics also resist oxidation better than metal thermocouples, cost substantially less than precious-metal thermocouples, and, unlike precious-metal thermocouples, do not exert catalytic effects.
A technique for suppressing sublimation of key elements from skutterudite compounds in advanced thermoelectric devices has been demonstrated. The essence of the technique is to cover what would otherwise be the exposed skutterudite surface of such a device with a thin, continuous film of a chemically and physically compatible metal. Although similar to other sublimation-suppression techniques, this technique has been specifically tailored for application to skutterudite antimonides. The primary cause of deterioration of most thermoelectric materials is thermal decomposition or sublimation - one or more elements sublime from the hot side of a thermoelectric couple, changing the stoichiometry of the device. Examples of elements that sublime from their respective thermoelectric materials are Ge from SiGe, Te from Pb/Te, and now Sb from skutterudite antimonides. The skutterudite antimonides of primary interest are CoSb3 [electron-donor (n) type] and CeFe(3-x)Co(x)Sb12 [electron-acceptor (p) type]. When these compounds are subjected to typical operating conditions [temperature of 700 C and pressure <10(exp -5) torr (0.0013 Pa)], Sb sublimes from their surfaces, with the result that Sb depletion layers form and advance toward their interiors. As the depletion layer advances in a given device, the change in stoichiometry diminishes the thermal-to-electric conversion efficiency of the device. The problem, then, is to prevent sublimation, or at least reduce it to an acceptably low level. In preparation for an experiment on suppression of sublimation, a specimen of CoSb3 was tightly wrapped in a foil of niobium, which was selected for its chemical stability. In the experiment, the wrapped specimen was heated to a temperature of 700 C in a vacuum of residual pressure <10(exp -5) torr (0.0013 Pa), then cooled and sectioned. Examination of the sectioned specimen revealed that no depletion layer had formed, indicating the niobium foil prevented sublimation of antimony at 700 C. This was a considerable improvement, considering that uncoated CoSb3 had been found to decompose to form the lowest antimonide at the surface at only 600 C. Evidently, because the mean free path of Sb at the given temperature and pressure was of the order of tens of centimeters, any barrier closer than tens of centimeters (as was the niobium foil) would have suppressed transport of Sb vapor, thereby suppressing sublimation of Sb
Analysis for relating changes in heat input to power output of thermoelectric device
A class of thermoelectric compounds based on the skutterudite structure with heavy filling atoms in the empty octants and substituting transition metals and main-group atoms. High Seebeck coefficients and low thermal conductivities are achieved in combination with large electrical conductivities in these filled skutterudites for large ZT values. Substituting and filling methods are disclosed to synthesize skutterudite compositions with desired thermoelectric properties. A melting and/or sintering process in combination with powder metallurgy techniques is used to fabricate these new materials.
A class of thermoelectric compounds based on the skutterudite structure with heavy filling atoms in the empty octants and substituting transition metals and main-group atoms. High Seebeck coefficients and low thermal conductivities are achieved in combination with large electrical conductivities in these filled skutterudites for large ZT values. Substituting and filling methods are disclosed to synthesize skutterudite compositions with desired thermoelectric properties. A melting and/or sintering process in combination with powder metallurgy techniques is used to fabricate these new materials.
We report the results of fabrication and testing of a thermoelectric power generation module. The module was fabricated using a new "flip-chip" module assembly technique that is scalable and modular. This technique results in a low value of contact resistivity ( < or = 10(exp 5) Ohms-sq cm). It can be used to leverage new advances in thin-film and nanostructured materials for the fabrication of new miniature thermoelectric devices. It may also enable monolithic integration of large devices or tandem arrays of devices on flexible or curved surfaces. Under mild testing, a power of 22 mW/sq cm was obtained from small (<100 K) temperature differences. At higher, more realistic temperature differences, approx.500 K, where the efficiency of these materials greatly improves, this power density would scale to between 0.5 and 1 Watt/cm2. These results highlight the excellent potential for the generation and scavenging of electrical power of practical and usable magnitude for remote applications using thermoelectric power generation technologies.