Supercomputer Network Simulates Material That Might Not Melt in a Sunspot

Simulations may lead to materials with higher melting points for better energy production

3 min read
Supercomputer Network Simulates Material That Might Not Melt in a Sunspot
Image: Brown University/American Physical Society

imgImage: Brown University/American Physical Society

Computer simulations can save time and money when investigating new technical designs but also when looking for new materials. Some serious supercomputing helped two scientists at Brown University in Providence, RI, to—virtually—break a melting-point record.

The current record is held by a mixture of hafnium, tantalum, and carbon (Ta4HfC5), which melts at 4,215 K (3942 oC).  The Brown scientists predicted that a material made up from a mixture of hafnium, nitrogen, and carbon, could have an even higher melting point, 4,400 K (4127 oC), which is about two thirds the temperature at the sun's surface. (Sunspots range from 3000 – 4500 K, so such a material would probably stay solid in one.) At that temperature, the theoretical material would emit light with an intensity about one third of the sun’s surface. The researchers, Axel van de Walle and Qi-Jun Hong, published the results of the computer simulations of this compound in the journal Physical Review B on in June.

They used a computational approach called “Density Functional Theory” (DFT). With this method one can compute the location of electrons in the crystal lattice of a compound using the laws of quantum mechanics, explains van de Walle. This electron distribution then allows the computation of the strength of the bonds between the different atoms in the crystal. It is the bond strength that determines the melting point.

DFT has been intensively used for calculating melting temperatures and other properties of compounds. Previous computer simulations are essentially static—capturing only the instant when the weakest intermolecular bond breaks. Here the researchers used a dynamic approach on a simulated sample of 100 atoms. This approach allowed more precision in the computations, because it showed how each incremental increase in temperature altered the structure of the crystal in ways that lead to melting. But it also required processing time on the National Science Foundation's XSEDE  computer network and on Brown's “Oscar” high-performance network. "You have to let the system evolve over time, and that requires thousands of steps, and for each step you compute the energy,” says van de Walle.

They also had to try out these simulations for compounds containing different proportions of hafnium, carbon, and nitrogen. "We were somewhat fortunate that in this system the simplest members, hafnium carbide and hafnium nitrate, both had a known crystal structure, and we conjectured that it would remain the same,  and that was what we put” in the simulation, says van de Walle. “If we had been wrong, the simulation would have told us.” 

Pinning down the right structure is important. "At such a high temperature, atoms move a lot, and if they are not happy in one configuration, they will find another configuration in the liquid phase,"  says van de Walle.

Their calculations showed that a hafnium compound in which 20 percent of the atoms are nitrogen and 27 percent of the atoms are carbon would beat the current record holder, Ta4HfC5,  by  200 Co.

The proof, of course, is in the  synthesis of this compound, which is the next project.

Van de Walle and Hong are confident. "We liked sending out the announcement  before the material was made so that we could have a proof that we predicted something, otherwise it sounds like we retrofitted our story after the fact," says van de Walle.

Measuring the melting point of an object at a temperature where everything else melts requires a few tricks. For this they turned to Alexandra Navrotsky, who heads the Peter A. Rock Thermochemistry Laboratory at the University of California, Davis. Navrotsky’s lab has equipment that levitates hot materials in a gas jet and heats the material by laser, so no container is needed, says van de Walle. Measuring the temperature will not melt the thermometer either. They’ll use a spectropyrometer, a device that looks at the spectrum of electromagnetic radiation a hot body emits, and based on the relative intensity of different wavelengths, the pyrometer infers temperature, explains van de Walle.

Materials with high melting temperatures are actively sought after by turbine engineers. (And also by Venus probe designers.) "By increasing  the temperature in a turbine, to produce electricity for example, you increase the efficiency, as predicted by the second law of thermodynamics.  So there is always a drive to go to high temperature materials,"  van de Walle concludes.

The Conversation (0)