In search of the perfect materials for fusion reactors

Nuclear fusion could be an ideal solution to mankind’s energy problem, guaranteeing a virtually limitless source of power without greenhouse gas emissions. But there are still huge technological challenges to overcome before getting there, and some of them have to do with materials. Fusion reactors need materials that can be used at the interface with plasma, in conditions that are nothing short of extreme. The design of the experimental European reactor ITER being built in the south of France, in particular, includes a component called a divertor, that extracts heat and ash produced by the fusion reaction and directs the flow of heat and particles from the plasma to specific surfaces for cooling. On the divertor, the plasma-facing materials not only withstand extremely high temperatures, but are constantly bombarded by a flux of neutrons, electrons, charged ions and high-energy radiation.

In the design for the ITER project, the divertor is made of tungsten, a metal known for its excellent heat resistance. But alternatives such as carbon fibers or ceramic materials were considered in the past, and it is still not certain if, for future reactors, tungsten would really the best option.

Can theory and computation methods help the search for the best divertor material and thus contribute to making fusion a reality? Scientists in Nicola Marzari’s MARVEL laboratory at EPFL decided to answer the question, and in a new article in PRX Energy they present a method for a large-scale screening of potential plasma-facing materials, and a shortlist of the most promising ones.

First of all, the scientists had to find a way to make computations treatable. “A realistic simulation of the dynamics at the plasma-material interface would require simulating the behavior of thousands of atoms over several milliseconds, that would not be feasible with ordinary computational power” says Andrea Fedrigucci, a PhD student in the THEOS lab and first author of the paper. “So we decided to select a few key properties that a plasma facing material needs to have, and use them as an indication of how well the material may perform on the divertor”.

First, the scientists looked at the Pauling file database, a large collection of known inorganic crystal structures, and created a workflow to find the ones that have enough resistance to survive the temperatures found in the reactor. This can be understood by looking at their thermal capacity, thermal conductivity, melting temperature and density. Because the surface temperature of a material layer depends on its thickness, they also computed the maximum thickness that each material can have before melting and ranked the materials accordingly. In the case of materials for which information on the maximum thickness could not be computed, they used a Pareto optimization method to rank them according to the previously mentioned properties.

The result was a first shortlist of 71 candidates. At this stage, a very non-computational and old-school method had to be used. “I patiently looked up the literature on each of them to check if they had already been tested and discarded, of if there were properties that would prevent their use in a fusion reactor and that were not in the database, such as a tendency to erosion or degradation of their thermal properties under plasma and neutron bombardment”.  Interestingly, this part of the study led to discard as divertor materials some innovative materials that have recently been  proposed for application in fusion reactors, such as high-entropy alloys.

In the end 21 materials remained, on which a DFT workflow was applied to calculate two key properties that a good plasma fusion material should have: the surface binding energy, that is a measure of how easy it is to extract an atom from the surface, and the formation energy of a hydrogen interstitial, that measures a proxy of tritium solubility in the crystal structure. “f a divertor material is excessively eroded during its operational lifetime, the released atoms disperse into the plasma, leading to a reduction in its temperature” says Fedrigucci. “In addition, if the material is chemically reactive with tritium, it can subtract the tritium available for fusion and cause an accumulation of tritium inventory that exceeds the safety limits imposed for this type of technology.”.

In the end, the final ranking based on all the key properties includes some usual suspects that have been extensively tested: tungsten itself in metallic (W) and carbide forms (WC and W₂C), diamond and graphite, boron nitride, and transition metals, such as molybdenum, tantalum and rhenium. But there were also a few surprises, such as  a peculiar phase of tantalum nitride or other ceramics based on boron and nitrogen, that have never been tested for this application.

In the future, says Fedrigucci, the group hopes to leverage neural networks to better simulate what really happens to materials in the reactor, including the interaction with neutrons that could not be simulated here.