|FEATURE FOCUS: MICRO AND NANOTECHNOLOGY |
Thermoelectric materials have long promised to outshine ordinary heat engines. Now, researchers have discovered that the key to success is nanoscale engineering.
by Jeffrey Winters, Associate Editor
An automobile’s air conditioning system is one of the biggest drags on its performance. In fact, it’s been estimated that running the A/C can knock as much as 15 miles per gallon off the efficiency of a hybrid electric vehicle such as the Honda Insight. Cars also lose almost 30 to 35 percent of their fuel’s original energy in heat vented through their exhaust systems. And another 30 percent is lost through the engine block.
Whenever thermodynamics is in play, there are going to be losses. But with gasoline prices soaring—together with concerns about the viability of continually increasing oil production—engineers are examining energy-saving technology like never before. One technology in particular has the potential to both recover energy from waste heat and more efficiently cool the passenger compartment.
Thermoelectric materials have been known for decades, along with their ability to effortlessly convert heat to electricity. But while thermoelectrics have been used in some high-profile niches, they have never been cheap enough or efficient enough for use in applications where traditional heat engines were available.
That situation could soon change, however. Thanks to developments in mechanical engineering labs, researchers have begun to fabricate high-efficiency thermoelectric materials with features as small as a few dozen nanometers. These nanoscale materials behave differently from bulk solids with the same chemistry, and in some cases are easier to produce.
“The material we have created in the laboratory could get to 8 or 9 percent efficiency,” said Gang Chen, a professor of mechanical engineering at the Massachusetts Institute of Technology in Cambridge. “And for high temperatures, there are reports of 12 to 14 percent efficiency.”
To improve its thermoelectric performance, MIT researchers ground up a material to jumble its crystalline structure.
While no one is predicting that nanoscale thermoelectrics could entirely replace familiar heat engines, they could find their way into everyday objects. “I don’t think we’re going to see thermoelectric power plants,” Chen said. “But they are going to make inroads on mainstream applications.”
A thermoelectric material in action is quite an odd thing. It consists of a semiconductor containing metals such as bismuth or telluride and little else. Attach wires to the front and back, apply a heat source to one side and a heat sink to the other, and current starts to flow across the wire.
This effect was first discovered in the 19th century by German physicist Thomas Seebeck. In 1821, working with a circuit made up of two different metals, copper and bismuth, Seebeck accidentally found that if there was a temperature difference across the junction, a nearby compass needle would be deflected. Further investigation showed that this was due to a current induced by a voltage created by the temperature difference. The larger the proportional temperature difference, the greater the produced voltage.
About a decade later, French scientist and watchmaker Jean Peltier ran the experiment in the opposite direction: Running current through a circuit, Peltier found that he could heat or cool a bimetal junction, depending on which way the electrons were traveling. Physicists today call the phenomenon the Peltier-Seebeck effect.
Lab-grown silicon nanowires (below) were strung across a gap on a semiconductor chip. While electrons sped across the wire, thermal vibrations were slowed.
With no moving parts, thermoelectric-based applications are prized for their durability. Thermocouples, for instance, use stacks of bi-metal junctions to measure temperature differences; they are often used in steel mills to track the temperature of the molten metal. Thermocouples also can be found in older gas furnaces to monitor the pilot light.
Some small electric beer and wine coolers now also use thermoelectric elements, drawing heat from the interior of an insulated compartment. Such coolers have no moving parts and draw as little as 12 volts, which means they can be run off a car’s electrical system.
Another, much more exotic use of thermoelectric materials is in powering deep space probes. Called radioisotope thermoelectric generators, or RTGs, these devices convert the heat from the steady decay of plutonium into as much as a few hundred watts of electricity. RTGs are especially important for deep-space probes, such as NASA’s Galileo and Voyager spacecraft, that operate far from the Sun and therefore can’t use solar panels for power.
ELECTRONS AND PHONONS
For all their utility, thermoelectric materials have always had one huge drawback—inefficiency. The amount of electricity produced through the Peltier-Seebeck effect is relatively small: Unlike a simple-cycle engine that can be upward of 30 percent efficient, thermoelectric generators are mired in the single digits.
The efficiency varies by the type of thermoelectric material used. The key term for comparing different materials is the figure of merit, which is proportional to the ratio of electrical conductivity of a material to its thermal conductivity. In essence, engineers want a material that allows electrons easy passage through it while resisting the passage of heat or thermal vibrations (which, in an analogy with electrons, physicists think of as “phonons”). The higher the figure of merit, or ZT, the more electricity can be obtained from a temperature difference.
“People have been able to find materials with good electronic properties,” said Bruce White, a physicist at Binghamton University in New York, “but the thermal conductivity has also been high. They tend to go together.”
The highest ZT yet measured is 2.4 for a material constructed with atomic precision, but the most commonly used thermoelectric materials, alloys of bismuth and tellurium, show bulk ZTs around 1 or less. According to White, ZTs in the neighborhood of 3 or higher are needed for thermoelectric generators to compete with traditional heat engines. Some experts believe ZTs several times that number are possible in specially engineered material. “Some researchers say that if we got to the 5 to 8 range, it would revolutionize energy generation,” White said. “I don’t know if we can get there.” But while calculating ZT for a material is relatively easy, engineering the stuff in bulk that can live up to the promise has been difficult.
Several research groups are making progress, however, by investigating how nanostructures can affect the conduction of heat.
NASA’s Cassini probe, which has been studying Saturn and its moons continuously for four years, is powered by thermoelectric generators.
Gang Chen and his colleagues have been looking to improve on a material that had been developed some years earlier at the Research Triangle Institute in North Carolina. There, researchers, led by Rama Venkatasubramanian, had created a high-ZT material made of careful depositions of antimony telluride and bismuth telluride selenide upon a substrate of bismuth telluride. Thanks to the way the atomic lattices of the various materials failed to align, the phonons that carried heat through the solid were scattered, much the way that light traveling through a fog dissipates. The result was a material that had one of the highest ratios of electrical conduction to heat conduction that had been measured.
Unfortunately, making the material was expensive and time-consuming, involving vapor deposition in a vacuum chamber. Chen’s group wanted to find a way to get similar results with much less fuss. “We’d been using nano-structured materials to reduce thermal conductivity,” Chen said. “But thin films are hard to work on—both in terms of measurement and material optimization.”
After many false starts—“I have been working on improving thermoelectrics for 10 or 11 years,” Chen said—his team hit upon a ridiculously simple solution.
Instead of careful layering, Chen’s team simply ground up the material. “We discovered from the theoretical work that we didn’t need thin films,” Chen said. “A random nanostructure will give you similar results.” His team took inch-wide ingots of bulk bismuth antimony telluride—a well-known thermoelectric material with a ZT of about 1—and ground them into an ultrafine powder that was loaded into a die. The powder was then hot pressed into disks that could be tested.
The result was a marked improvement, with ZTs as high as 1.4 recorded—a huge leap. To further confirm their results, the team rigged up a thermoelectric cooler using the nanostructured material on one side and a commercially available thermoelectric material on the other. The nanomaterial cooler brought the temperature down some 30 degrees more than one made with two commercially available alloys could.
What’s more, Chen said, while the thin film superlattice takes as much as a week to grow a few hundred micrometers, the ground material can be produced in the lab by the gram, even the kilogram.
Another approach was taken by a group of researchers at Lawrence Berkeley Laboratory in California. Led by Arun Majumdar, director of the lab’s environmental energy and technology division and an ASME Fellow, the team looked at how to turn a material that was a fairly poor thermoelectric choice into something much better.
One potential problem with thermoelectrics is the availability of the raw material. Tellurium, bismuth, and other frequently used materials are fairly rare; indeed, tellurium is one of the least common naturally occurring elements on Earth. If thermoelectric generators and coolers are going to become mainstream applications, they need to be both efficient and cheap, which means looking for inexpensive materials.
The Majumdar team looked at the thermoelectric properties of something decidedly common: silicon. Unfortunately, silicon’s ZT is about .01, or one percent of bulk bismuth telluride. To come up with a form of silicon with lower thermal conductivity, the team synthesized rough-sided nanowires made of silicon.
The wires ranged from 10 nm to 100 nm in diameter—large enough to enable electrons to travel easily but so thin as to create a barrier to heat transport. As a result, the thermoelectric ability of the silicon increased substantially. When measured on a wafer coated with a thin film of silicon nitride, its ZT rose to a level close to that of the best thermoelectric materials in common use.
The Majumdar team expects that the nanowire technique could produce similar increases in other materials. “The trick,” White said, “is getting the phonons to scatter without getting the electrons to scatter. And no one’s quite gotten that yet.”
RUNNING HOT AND COLD
There’s still a way to go before thermoelectric devices become commonplace, but the recent advances have experts thinking about what they could be used for.
One obvious use would be in cooling computer equipment, which must shed more and more heat as it processes more and more data. Unlike traditional cooling fans, which not only must remove heated air from the entire computer enclosure but also have moving parts with a tendency to fail over time, a solid state thermoelectric cooler could be scaled to about the same size as a processor and thus work more efficiently.
The same logic applies, on a larger scale, to the use of thermoelectric cooling inside automobiles. Although air conditioners are expected to remain more efficient than thermoelectrics, they suffer from the need to cool the entire passenger compartment. A thermoelectric cooling system could be designed to operate close to individual passengers, much like a reading light. This could make such cooling competitive with traditional A/C in larger vehicles.
Thermoelectric cooling has proven useful in small refrigerators. More efficient material may open up applications in computers and automobiles.
Another potentially widespread application is thermoelectric power generation from hot exhaust. Chen said that using his nanostructure material, one could construct a thermoelectric generator that was 8 to 9 percent efficient; intercepting a fraction of the heat from a car’s exhaust stream and converting it to electricity could recover 1 to 2 percent of the fuel’s original energy. Considering that automobiles effectively utilize roughly 20 percent of that energy today, such a system could boost fuel efficiency by 5 to 10 percent. “That would make it attractive for automobile use,” Chen said.
An efficient-enough thermoelectric material could even find its way into solar cells. Instead of converting light into electricity, a solar thermoelectric cell would concentrate sunlight onto a thermocouple, heating it up to generate electricity.
Caution, it should be said, is more than warranted. Much like nuclear fusion, thermoelectricity has been a technology of the near future since about the 1950s. Back then, fueled by the success of materials science research in the post-World War II era, physicists expected that thermoelectric efficiencies would soon approach the Carnot limit. That standard of progress is still somewhere over the horizon. But the breakthroughs seen of late in nanoengineered materials suggests that thermoelectricity’s day in the sun is closer than ever before.
“It’s a lot of hard work, but it feels good to finally make some progress,” Chen said.