Long before James Watt perfected his steam engine in 1776, people had been trying to convert heat into usable power. In today’s electricity-powered world, that generally is achieved by using steam or gas-fired turbines and attached generators. While steam turbines power about 90% of U.S. electricity demand, a lot of potential heat energy is wasted.

Thermoelectric generators short circuit some of the drawbacks of traditional power generation technologies. While the best barely reach 8% conversion efficiency, they are well suited for less intense heat sources like concentrated sunlight or vehicle exhaust gases.

What makes them work is the Seebeck effect, named after the German physicist Thomas Seebeck. He discovered that an electric current is generated when two different metals are joined in two places, with a temperature difference between them. Seebeck noticed that a compass needle could be deflected by the two metals – a thermocouple – but didn’t realize he also was generating an electric current. These days we know better.

Thermocouples also work in reverse. The Peltier Effect generates or absorbs heat at the junction of two conductors. For example, small electric coolers that plug into a car's dashboard demonstrate the Peltier effect at work. Around the home, thermocouples close gas valves when a pilot flame goes out and also help regulate oven temperatures. Although thermocouples are common, they typically put out only a few millivolts.

A thermoelectric generator can be made using bimetallic materials, but the result often is bulky and expensive. Current research is focused on solid state devices that use doped semiconductors made from bismuth or lead telluride, or calcium manganese oxide. Manufactured using a process much like photocells, they are expensive and hard to cool. The latter in particular is a drawback since maintaining a temperature difference across the device is critical. Solid-state thermoelectric generators need a temperature of 200 degree Celsius to generate more than a few watts. Thus, the heat sink often becomes more important, and more expensive, than the silicon.

At the Massachusetts Institute of Technology (MIT), the Solid State Solar-Thermal Energy Conversion Center – S3TEC for short – is working to bring down the cost and raise the efficiency of thermoelectric generation. Focusing largely on solar-heated generators, S3TEC has approached the problem from two angles: improving solid state performance and perfecting optical concentration technology.

By concentrating the incoming sunlight, researchers can increase efficiency from 4.6% to 5.3%. They still have a long way to go to catch up with photovoltaic efficiencies, which are in the 15% range for silicon-based cells and approach 30% using far more expensive gallium arsenide. Unlike solar cells, however, these devices can operate anywhere there is a heat source, day or night. The MIT team is evaluating phase change materials to store and release heat over long periods of time to stabilize output.

One of the biggest challenges to achieving higher conversion efficiencies is physics. Materials that conduct electricity well also tend to conduct heat equally well. The ideal material has the thermal conductivity of an amorphous material like glass, but the ability to pass electrons like a crystal.
Gang Chen, director of S3TEC, says, “We want the crystal and amorphous material in the same material. In the past, all thermoelectric materials used alloys to reduce thermal conductivity.” Chen's lab has developed a material that meets these requirements by crushing bismuth telluride into a powder and re-compacting it. The sintering process appears to improve electron flow without increasing heat transmission.

While MIT is focused on tweaking solid state physics, a California startup is exploring a more basic solution to large-scale thermoelectric generation. The firm, Alphabet Energy, is building generators from tetrahedrite, a naturally occurring copper antimony sulfosalt mineral commonly found in hydrothermal veins. Tetrahedrite is so common it is considered a minor ore of copper.

The mineral's possibilities were first tested at Michigan State University's materials laboratory in 2012 when researchers noticed that its chemical formula matched the thermoelectric compounds they were creating from scratch. The lab discovered that grinding the mineral into a powder and sintering it between two conducting plates creates a good thermoelectric generator with little or no additional processing. Unlike the rare earths and metals needed by other solid state generators, tetrahedrite costs about $4/kg. If the technology works, Alphabet predicts that it can deliver power at a level generation cost of $.03/kW.

Low-cost thermoelectric generation could open new opportunities to turn low-level waste heat into useful power. Replacing the alternator in a car with a thermoelectric exhaust pipe lining is one idea. Capturing the waste heat from refineries and other manufacturing processes is another option. So is converting the leftover heat from nuclear and coal-fired electric generating plants. Adding the technology to existing generating plants leverages the transmission infrastructure, and conceivably could reduce carbon emissions.

Another player in this space, TECTEG Manufacturing, is working with McMaster University to produce high amperage bismuth telluride cells for high-temperature environments. Its GENCELL devices are engineered to deliver 100-200 amperes, although the current specification claims 100 watts per 150 C of temperature differential and an efficiency of 6%. The company is building a proof of concept manufacturing facility with an annual device capacity of 10 MW. Future development plans include zinc/antimony (Zn4Sb3), magnesium silicide (Mg2Si), and manganese silicide (Mn2Si) cells designed to work in the 200–800C range at a theoretical efficiency of 9-10%.

According to MIT’s Chen, opportunities for low-level heat conversion are huge. Forty percent of the fuel energy going into internal combustion engines is wasted, and he points out that 3,000 terawatts of heat energy are lost to the environment each year from power plants alone.

One promising future technology may be 1D quantum nanowires, which combine high electrical transmission with the ability to scatter phonons – heat energy – off their inner and outer surfaces. Researchers believe that 20 nanometer wires seem to provide the best performance.
Although the technology provides high theoretical conversion rates, building a working human-scale generator has proven to be difficult. The nanowires must be aligned and compacted, their ends attached to conductors and heat-sinks added to create the necessary temperature differential. Nanocomposites, which use a polycrystalline matrix to encapsulate and align the nanowires, are considered the key to a commercially feasible nanowire generator.

Thermogeneration technology still has a long way to go. Work continues on designing materials with high electrical conductivity and low thermal conductivity. Thermal stability is also a challenge; the TECTEG high-temperature generators have a design life of less than 20 years, versus the decades of service expected from photovoltaics. On the plus side, however, waste heat is readily available and thermoelectric generators may be the most obvious way to convert it to something useful.