As electrons whiz around the nanosized circuits inside computer chips, they generate a lot of heat. And that waste heat creates a major headache for the electronics industry. It shortens the lifespan of devices and requires significant amounts of energy to combat. For example, data centers in the U.S. devote about half of their electricity use just to keeping processor chips cool. Finding new ways to dissipate all that heat could make our electronics more energy efficient.
One solution could be working with materials that conduct heat more readily. Now three teams of researchers, in separate papers, report that a semiconductor, boron arsenide, has a thermal conductivity far greater than other commonly used electronic materials (Science 2018, DOI: 10.1126/science.aat5522, 10.1126/science.aat7932, and 10.1126/science.aat8982).
All three teams synthesized BAs crystals and measured the semiconductor’s thermal conductivity to be at least 1,000 watts per meter per kelvin (W m–1 K–1) at room temperature. The highest value was 1,300 W m–1 K–1, measured by a team at the University of California, Los Angeles.
The measured values are nearly 10 times as great as that of the most common semiconductor, silicon, which has thermal conductivity of about 150 W m–1 K–1. Even metals are less efficient; for example, copper’s thermal conductivity is roughly 400 W m–1 K–1. The only known material with a higher thermal conductivity is diamond, with a value of 2,000 W m–1 K–1. But diamond is expensive, difficult to work with, and isn’t a semiconductor.
Theoretical calculations suggested BAs should have a high thermal conductivity, but these three studies are the first in which researchers have made crystals big enough to measure the effect. “It confirmed a prediction my colleagues and I made several years ago,” says David Broido, a physicist at Boston College and an author of one of the papers.
The key to these measurements was figuring out how to make crystals with the necessary purity to see the predicted effect, says Yongjie Hu, a mechanical engineer at UCLA and a member of the team that recorded the highest value. Thermal conductivity depends on the interaction of phonons, which are vibrations in a material’s crystal lattice. “If there’s any imperfection of the crystal lattice, that’s going to reduce the thermal conductivity,” Hu says. It took his team several years to learn how to control the synthesis precisely enough to remove all the defects, he says. The other teams worked on ways to minimize defects to get their results.
The studies also help refine the theoretical understanding of the effect, Hu says. Earlier theory predicted BAs would have thermal conductivity closer to 2,000 W m–1 K–1 due to the interaction of three different types of phonons. All three groups eliminated the crystal defects and still couldn’t get values higher than 1,300 W m–1 K–1. Hu says these results instead fit nicely with predictions that rely on a rarer four-phonon interaction.
Potential applications for these findings are very broad, says Zhifeng Ren, a physicist at the University of Houston and a coauthor with Broido. “They could be anywhere you need thermal management,”
Hu points out that there is more work to do on BAs synthesis. To find practical applications in electronics, the material will need to be grown at wafer-scale, about 300 mm across. The three groups made crystals that were a few millimeters across.
“This is very impressive work with important scientific impact,” says Thomas Reinecke, a nanoscience researcher at the U.S. Naval Research Laboratory. Back in 2013, he and Broido had predicted the 2,000 W m–1 K–1 value, though he was not involved in this current work.
Though the teams worked independently, Ren points out that many of the authors of the three papers have collaborated in the past, some serving as postdocs in the same lab.
CORRECTION: This story was updated on July 19, 2018, to correct the spelling of Zhifeng Ren’s name and to correct the units for thermal conductivity. Thermal conductivity is measured in units of watts per meter per kelvin —W m–1 K–1, not W/mK.