Thermoelectric cooling devices

Cooling in miniature, without bulky machines, conventional fluids or moving parts, is goal of materials research at Cornell.

provided by Cornell University

t's possible that one day all the cooling power of a noisy, bulky household refrigerator will be available on a small device that is lightweight and has no moving parts. And the same device, when given a heat source like a car's exhaust pipe, could be used to generate electricity.

Such thermoelectric devices already exist in consumer products like plug-in auto beverage coolers, where energy efficiency is less important than portability and low weight. The challenge facing researchers is to find new materials that could bring the technology to the next level in which the efficiency would rival that of conventional coolants in air conditioners as well as refrigerators. Also in the future might be miniature cooling devices directly on computer chips.

"As we increase the efficiency of thermoelectric devices, we create another tool in the arsenal for choosing the most efficient way to do things. You can think of some applications pretty quickly, and others would come up once the technology is available," said Francis DiSalvo, professor of chemistry and chemical biology, who is attempting to develop new thermoelectric materials. DiSalvo described the status of research in the field in a recent issue of Science.

In conventional cooling devices, heat is carried away by a working fluid, such as a chlorofluorocarbon, which involves the moving parts that cause most equipment breakdowns, environmental damage and bulkiness. In thermoelectric devices, the "working fluid" is electrical current that runs through a junction between differently doped semiconductors and pulls heat away from that junction, producing cooling without any moving parts.

Current thermoelectric materials operate at roughly 10 percent of Carnot efficiency, the theoretical maximum allowed by the laws of thermodynamics, compared with about 30 percent for an average household refrigerator. The theory behind thermoelectric devices has been around for more than 40 years, but current materials don't rival the efficiency of compressor-based devices. "The theory is not specific enough to say, 'If you could make this kind of material -- that is, this composition with atoms in a particular crystalline arrangement -- that will give you the high thermoelectric efficiency.' We have to go find the materials by an empirical process, test them one at a time and say, 'Is our understanding good enough that we can predict from what we're learning today about what's the next best thing to try to synthesize after that?'" DiSalvo said.

The search, funded by the US Office of Naval Research, is complex because, according to DiSalvo, with the exception of designing organic molecules based on carbon, the ability to predict the composition and structure of materials made from three or more elements is completely lacking. "For most of the elements in the periodic table, we don't know what will happen when we put them together. If we knew how to do that, then we could calculate from the structure what the thermoelectric properties might be like. And the theory is good enough now that the results would be fairly accurate," he said.

The search is focused on uniform bulk materials, which can be prepared in large amounts by traditional synthetic methods, and on compositionally modulated films, which require expensive nanofabrication. Bulk materials, the object of DiSalvo's research, primarily have applications in large devices like home refrigeration and recovering power from car heat exhaust, while modulated film research might be applicable to niche markets like on-chip cooling.

Researchers know they need a material with low thermal conductivity and high electrical conductivity, which has led them to look at compounds of heavy elements like lead, antimony, bismuth and tellurium. "We have some compounds that looked promising based on platinum. No way you're going to build devices like that out of platinum -- way too expensive," DiSalvo said. But, he said, "if we can do the proof of principle, then we're off and running; we've got our foot in the door."