Boron arsenide surpasses diamond, silicon in thermal conductivity

In a potential seismic shakeup for materials science, a team of researchers led by Zhifeng Ren, a physics professor and director of the Texas Center for Superconductivity at the University of Houston, recently overturned earlier theory work that predicted boron arsenide (BAs) can’t compete with the heat conduction of diamond.

Thermal conductivity is a measure of how fast heat is conducted away from Point A to Point B, in the unit of watts per meter-Kelvin (W/(m·K)). Ren’s team discovered high-quality boron arsenide crystals can conduct heat at more than 2,100 W/(m·K), which is as good as diamond.

This is incredibly significant for the semiconductor and electronics realm because boron arsenide is easier and cheaper to manufacture than diamond—it removes the need for extreme temperatures/pressures. It’s also an exceptional thermal conductor and an effective semiconductor—with potential to outperform silicon with its high thermal conductivity, wider bandgap, higher carrier mobility in both electrons and holes, and a well-matched coefficient of thermal expansion.

Why boron arsenide (BAs)?

Boron arsenide was “surprisingly predicted to have thermal conductivity as high as diamond back in 2013, but it was then revised back down to lower than diamond in 2017,” says Ren. “Now, our experiments show that BAs indeed has thermal conductivity as high as diamond. Many applications need high thermal conductivity to effectively take heat out of the devices so that they work properly.”

Diamond isn’t ideal because it has problems as a heat conductor for a wide range of applications. And while silicon is a good semiconductor, its thermal conductivity is too low and it requires the use of other high thermal conductivity materials to dissipate the heat silicon devices generate. “BAs can, of course, be used for optical applications due to its wide bandgap and very high carrier mobility,” Ren adds.

The most important aspect of this discovery? “Not fully believing in limits set by the theory,” says Ren. “If we had believed in the theory, we would never have tried to look into the possibility of thermal conductivity above the theoretically predicted value of about 1,360 W/(m·K). When we actually saw the indication of thermal conductivity at about 2,000 W/(m·K), we felt a bit lost because now we don’t know what the limit is. But it gives us incentive to pursue it further and hope we’ll reach a much higher record.”

Challenges remain—including figuring out how to grow the crystals with uniform high thermal conductivity, as well as how to grow large crystals up to a few centimeters and eventually a few inches. “We believe we can reach these goals if we can get the necessary financial support,” Ren says. “This material is extremely important because it’s the future of the semiconductor industry, due to all of its desirable properties.”

What does this mean for materials science?

Achieving thermal conductivity above 2,000 W/(m·K) in BAs is a significant breakthrough. “First, a better thermal conducting material than diamond means thermal management has found a solution, and it’ll have significant implications for high-power electronics and data centers,” says Ren. “And since the theory didn’t predict the value would be so high, it means either a revision to the theory or a new theory is needed, so it may lead to prediction of new materials with even higher thermal conductivity under the revised or new theory. This will be a gamechanger for materials science.”

Wanted: Funding

Depending on the application, such as small-area ones, “we’re ready now,” says Ren. “But for large-area applications, we need to grow uniform crystals up to a few inches. The next step is to grow uniform and large crystals with high thermal conductivity. How fast it will happen depends largely on how fast we can find financial support—I hope significant financial support happens sooner rather than later so the U.S. maintains its lead.”

FURTHER READING

A. B. Niyikiza et al., Mater. Today (2025); https://doi.org/10.1016/j.mattod.2025.09.021.