Nuclear fusion offers a promising path to a nearly limitless, clean energy source, but creating fusion power is no simple task. Fusion reactors operate under extreme conditions, where the materials in direct contact with the plasma must withstand immense heat, neutron bombardment, and high-energy radiation. One of the most critical components in a fusion reactor design is the divertor, a part that extracts the heat and ash generated by the fusion reaction, managing the flow of heat and particles to keep the plasma in check. In ITER, the European experimental fusion reactor being built in France, the divertor uses tungsten due to its high heat resistance. However, questions remain as to whether tungsten is the best option, and researchers are exploring alternatives to find materials that might better endure the harsh environment.
Scientists in Nicola Marzari’s MARVEL laboratory at EPFL are tackling this materials challenge with a new approach using computational methods. In a study published in PRX Energy, they presented a large-scale screening process for potential divertor materials and identified a shortlist of promising candidates. This effort shows how theory and computation could accelerate the search for optimal materials in fusion technology.
A realistic simulation of plasma-material interaction is extremely complex, requiring simulations of thousands of atoms over several milliseconds—computationally unfeasible with current capabilities. To overcome this, the researchers focused on identifying a few critical properties that could serve as indicators of a material’s potential as a divertor candidate. Andrea Fedrigucci, the lead author and Ph.D. student in the THEOS lab, explains that instead of trying to simulate everything, they honed in on properties like thermal capacity, thermal conductivity, melting temperature, and density, which are vital for understanding a material’s ability to endure reactor temperatures.
The team began by analyzing data from the Pauling File, a comprehensive database of inorganic crystal structures, to assess potential candidates based on thermal resistance and stability. They also calculated the maximum layer thickness for each material, as thickness affects surface temperature. When data was unavailable for some materials, they used Pareto optimization, balancing various properties to rank the candidates’ suitability.
This initial screening produced a shortlist of 71 materials, which the team further narrowed down by meticulously reviewing scientific literature. Fedrigucci’s extensive search through historical studies helped eliminate materials with known issues, such as high erosion rates or thermal degradation under neutron exposure. Surprisingly, this manual review led to the elimination of some newer materials, including high-entropy alloys, that were initially thought to be promising but had drawbacks for fusion applications.
After further refinement, the team narrowed the list down to 21 materials, applying density functional theory (DFT) calculations to assess two additional key properties: surface binding energy and the formation energy of hydrogen interstitials. These properties are crucial because divertor materials that erode easily will disperse atoms into the plasma, lowering plasma temperature and reducing fusion efficiency. Additionally, if the material is reactive with tritium, it can deplete tritium needed for the fusion process and lead to tritium accumulation, posing safety risks.
The final list of promising materials included some familiar contenders—tungsten in metallic and carbide forms, graphite, diamond, boron nitride, and certain transition metals such as molybdenum, tantalum, and rhenium. However, there were also a few surprising candidates, such as a unique phase of tantalum nitride and boron-based ceramics, which had not yet been tested in fusion settings. These materials could open new avenues for research in fusion technology, potentially offering improved performance and durability over currently used options.
Fedrigucci notes that the next step involves advancing simulations with neural networks, which could allow researchers to more accurately model neutron interactions—a crucial factor that couldn’t be addressed in the current study. As these computational techniques improve, they could significantly enhance our understanding of material performance in fusion reactors, helping bridge the gap between theoretical predictions and real-world fusion applications.
This work from EPFL highlights the potential of computational methods in the quest to make fusion energy a reality. By narrowing down materials that can withstand the unique demands of fusion, scientists are one step closer to developing reactors that can reliably produce clean, virtually limitless power. As researchers continue refining their models and testing new materials, fusion energy’s potential as a transformative energy source for the future becomes increasingly tangible.