High thermal conductivity for ultrapure semiconductors proven experimentally
Experimental verification of the long theorized effect of purity on the thermal conductivity of a semiconductor, specifically silicon-28, appears this week in the Journal of Applied Physics. By borrowing a piece of an ultra-high purity single crystal Si-28 sample made by the Avogadro Project to redefine the kilogram, researchers in Russia, the United States and Germany explored the effects of isotope content on heat transport in silicon.
Their work confirms that thermal conduction in crystal silicon can be improved by isotopic enrichment. While the enhancement in thermal conductivity is relatively small at room temperature (about 8 percent), it becomes much more dramatic at low temperatures — up to 1,000 percent, according to Alexander V. Inyushkin, one of the authors.
This ultrapure crystal has the highest thermal conductivity value ever measured for bulk dielectrics (including isotopically modified diamonds) at 450 Watts per centimeter-Kelvin at a temperature of 24 Kelvin. “Even highly conductive metals such as copper and silver have thermal conductivities about six times lower than that of silicon-28 at that temperature,” Inyushkin said.
This benchmarks the thermal performance of nearly perfect, defect-free, and isotopically pure crystalline silicon. “Our data, which captures the conductivity of silicon governed solely by the intrinsic phonon damping processes within a wide temperature range, is the new reference point for theoretical investigations of the heat conductivity of silicon.”
The group also discovered that below 31 Kelvin isotropic heat conduction becomes anisotropic, an effect associated with phonon focusing and the impact of boundary scattering, though it is not currently understood in terms of details.
Source: “Ultrahigh thermal conductivity of isotopically enriched silicon,” by Alexander V. Inyushkin, Alexander N. Taldenkov, Joel W. Ager III, Eugene E. Haller, Helge Riemann, Nikolay V. Abrosimov, Hans-Joachim Pohl, and Peter Becker, Journal of Applied Physics (2018). The article can be accessed at https://doi.org/10.1063/1.5017778 .