How much energy do we waste in America? A little over half of it, says Arun Majumdar, the Almy and Agnes Maynard professor of mechanical engineering at UC Berkeley.

The U.S. consumes about 100 quads of energy a year, he said. A quad is a quadrillion BTUs. A BTU is equivalent to the energy produced by a match.

"Of the 100 quads, about 55 to 60 go into waste heat," he said. "Fifty or 60 percent of the energy is wasted as heat."

The culprit is waste heat, the heat that is emitted into the atmosphere by everything from nuclear power plants and factories to car engines and that power brick on your laptop.

A new crop of startups, however, are tinkering with ways to make this useful. The companies can be divided into two categories. On one hand, there are the companies with mechanical systems, which capture heat, compress it and exploit the pressure to crank a turbine. ElectraTherm, for instance, recently raised $2.6 million to further test out its low-heat converter. ElectraTherm uses captured heat to boil a liquid refrigerant to create pressure. The company's Green Machine can produce up to 50 kilowatts of power with 200 to 400 Fahrenheit.

Recycled Energy Development (RED), founded by the father and son team of Tom and Sean Casten, is erecting a $45 million to $55 million plant at silicon producer West Virginia Alloys that could produce 40 to 44 megawatts of power when it becomes operational by 2010. Potentially, it could offset one-third of West Virginia Alloy's power. The Castens, who have participated in over 250 waste heat projects, have asserted that RED can reduce an industrial customer's power consumption by 20 percent.

Meanwhile, others are using solid state solutions to work with thermoelectric materials. These materials can convert heat into electricity and vice versa. Potentially, thermoelectric materials have an advantage over mechanical systems because there are fewer parts. Thermoelectrics can also potentially work at lower temperatures. Micron-Gap Thermal Photovoltaic (MTPV) is working on what it calls Thermal PV (i.e., a solar panel that can produce power from sunlight like other solar modules, but with an additional layer of material that can convert solar heat into electric power).

In a similar vein, GMZ Energy has a material that it says can accomplish the same heat-to-electricity conversion but with heat from industrial equipment. Going in a reverse direction, Promethean Power, which recently raised money from the Quercus Trust, wants to sell thermoelectric fridges to emerging nations. Electricity harvested from solar panels would be run through thermoelectric materials to produce hot and cold air streams. The cold would be used on the fridge.

And expect a startup to emerge from Berkeley soon. A team under Majumdar has devised a technique for making thermoelectric chips out of silicon nanowires, columns of pure silicon. Silicon ordinarily has a very low ZT, a number for calculating thermoelectric properties. Bismuth telluride, a thermoelectric material that researchers have examined for years, has a ZT of around one.

"The performance has been so low that it has not been cost effective," he said. Tellurium also now costs close to $100 a pound, thanks to the popularity of cadmium telluride solar cells.

Ordinary silicon has a ZT of 0.008. The silicon nanowire devices, however, may be able to achieve a ZT of 2, judging from early experiments. In turn, that could mean electric power at $1 a watt or less, which would make waste heat competitive with coal. "What is needed is a ZT of 2 or more," he said. "Can we get to a ZT of 2? I'm optimistic."

The Berkeley team is already talking to TSMC, the Taiwanese chip giant, about trying to manufacture nanowire thermoelectric chips.

Channel Changes

Not all waste heat can be converted to electrical power economically. Some gets lost merely in the process of delivering power. Nonetheless, there's a lot of heat out there. A 2005 survey from the Lawrence Berkeley Lab estimated that the U.S. at the time had 100 megawatts of untapped power in waste heat, garbage and petroleum byproducts. These waste products could ultimately supply 19 percent of U.S. power. About half of the power in the study would come from harvesting heat and pressure.

Expanding the waste heat industry, however, would likely require regulatory and grid changes. Right now, most factories that have installed waste heat systems consume the power themselves. But many of them want to sell power to third parties. Glass factories, for instance, can generate 6 megawatts of power with their 450 degree Celsius waste heat, said Majumdar. For that to happen state laws would likely have to be changed.

In the meantime, thermoelectric materials, following more extensive testing, could begin to be incorporated into a number of products. Engine and exhaust heat in cars, for instance, could be harvested if the material were cheap enough. Brake heat could help increase the power collected by regenerative braking systems.

But why convert at all? Why not just use the heat? In a word, location. Heat doesn't travel well.  

"Power plants would need to be in a city," said Majumdar.

Correction: An earlier version of this story incorrectly stated that RED could offset one-third of West Virginia's power when it is in fact West Virgina Alloy's power.