Enthusiasts spend a lot of time and money getting heat out of their systems, but what if you could tap in-to those scalding processor surfaces and turn that wasted heat into energy?

Thermoelectrics is nothing new. The problem, though, has been that the conversion efficiency that different experiments yield has been too low and the costs too high to make the technology feasible. Professor Heiner Linke of the University of Oregon and Tammy Humphrey, a post doctorate fellow at the University of New South Wales in Australia, have made a breakthrough that may make thermoelectrics a popular reality.

The pair’s discovery hinges on the laws of thermodynamics, which state that heat will always flow to cold, until a balance of temperature is achieved. In the process, the energy expended by electrons is lost as waste. Conventional thermoelectric materials have a hard time capturing this waste because the heat flow is uncontrolled. However, the nano-structured materials that Linke and Humphrey have been able to develop—semiconductors that are riddled with nanowires—can bring order to the electrons’ movement. Only electrons with a specific energy are allowed to flow through the material, meaning electrons can’t shuffle heat to cold areas. Thus, a sort of equilibrium is forced between the two regions. In fact, because the regions are in equilibrium, the electron flow is reversible, making the device much more efficient than prior approaches.

“We have a semiconducting material with an energy gap in it that electrons jump over,” says Humphrey. “They are excited by the heat emitter to a high energy, they jump over this gap, and then they can be used to do useful work against a voltage. In computers you would need a really high-efficiency material to make it worthwhile to recycle waste heat. If you could make a material that operates with an efficiency of, say, half the Carnot limit, then you could in principle recoup roughly 17% of the heat from a CPU.”

Linke and Humphrey believe their design will achieve about 50% of the Carnot limit. Previous CPU-sized designs have realized less than 15%. Humphrey hopes that by combining their work with similar research projects, the technology can be commercialized in three to five years in such areas as computing, car engines, and geothermal hot springs.


Philips Out-Flashes Flash

Whenever you spin up a CD or DVD, you are using phase-change memory. Phase-change materials are what deform in the discs when a laser hits them, shifting from amorphous to crystalline phases and exhibiting different reflective properties accordingly. Philips is now working on making solid-state memory by applying a thin film layer of phase-change material on a silicon substrate. Electrical charges deform the material while resistance reads its state.

Unlike other memory technologies that have trouble scaling down in size, Philips’ “line-cell” approach requires less voltage as feature sizes decrease. Better yet, phase changes can occur in about 30ns, more than 100 times faster than a flash memory cell can change states, and line-cell chips can be made almost entirely using conventional chip fabrication methods.

“The holy grail of the embedded memory industry is a so-called unified memory that replaces all other types, which combines the speed of SRAM with the memory density of DRAM and the nonvolatility of Flash. Philips’ new phase-change line-cell technology is a significant step toward this goal,” says Dr. Karen Attenborough, project leader of the Scalable Unified Memory project at Philips Research, in a statement.

by William Van Winkle