Speaker
Description
The plasma facing materials of fusion reactors are subjected to harsh environments, and especially to possible synergetic effects of several different factors at once. To understand both the effects of single factors and their synergistic effects, we
need to understand the phenomena at the atomistic level. Molecular dynamics (MD) simulations have been utilized to understand these phenomena, and with the increasing computational power, these simulations can be done on large and complex enough systems, to ultimately understand the sputtering effects. In the last decade, not only flat low index surfaces have been investigated, but also the effects of random and amorphous surfaces, and surface features. We showed that amorphous surfaces, usually used as a proxy for a random orientation, do not work. However, random surface orientations were studied, and they were shown to follow the Sigmund theory at high
energies [1]. Our combined experimental and computational study showed that surface structures can dramatically affect sputtering and its mechanisms [2].
In addition to surface structures and orientation, we can have impurities at the surface or just below the surface, e.g. deuterium or tritium decoration and boron from boronization. The study of the effects of hydrogen isotopes and boron on tungsten surfaces is enabled by newly developed and extremely accurate Machine Learning interatomic potentials. We studied the effects of decoration of the surface and found that not only is the tungsten sputtering affected, but also the fuel was seen to easily be recycled into the plasma. The sputtering of different elements followed their own trends, additionally affected by ion type. Sputtering of different molecules was also observed. Alloy surfaces showed preferential sputtering, and that at the steady state the surface composition is clearly different compared to the initial one. The preferential sputtering was not dictated by the elemental sputtering yields, but depending on other factors, such as mass. For all cases, decoration, boronization or alloying, the underlying mechanisms
were determined by the atomistic insight enabled by MD simulations.
Our results will not only give greater insight into the sputtering phenomena, but can be used to enable more accurate larger scale simulations. Additionally, the data obtained can be used to train surrogate models, which would permit having a fast model directly
integrated into the larger scale models, to enable very accurate simulations.
[1] Schlueter. et al. Physical Review Letters 125 (2020) 225502
[2] Cupak. et al. Physical Review Materials 7 (2023) 065406