Sputtering from plasma-facing surfaces upon particle impact is an important process in material science. It is especially relevant in the diverter region of fusion devices, which nearly always consist of tungsten. Besides the main plasma components, argon is used in fusion devices to improve energy confinement. As a consequence, hot Ar atoms interact with W surfaces and can cause sputtering and other material degrading events. Atomistic simulations of the plasma-wall interactions make it possible to carry out a detailed analysis of sputtering, reflection, and retention processes. We report the results of molecular dynamics simulations with neural network potential energy expressions modelling the bombardment of tungsten samples by argon atoms in the energy range from 100 to 800 eV. The obtained sputtering results are in good agreement with available literature data. Furthermore, our data provide additional insight into atomic details of the processes involved in sputtering. We also investigate the effect of surface temperature on sputtering and reflection probabilities, which significantly affects the irradiation process at higher impact energies.
Graphical abstract