Effects of surface roughness on sputter yield of Mo under keV D ion bombardment, experiment and SDTrimSP-3D simulations

An important issue in the development of thermonuclear fusion is the lifetime of the reactor inner wall. Bombardment by energetic ions and neutrals from the plasma will lead to continuous erosion of the plasma-facing surface. Inside the plasma chamber of a fusion device, particles coming from the plasma impinge on the components at different angles depending on both local plasma parameters and on the orientation of the magnetic field lines. In the past, work has been done to study the effects of incidence angle on sputter yields for smooth surfaces [1] and was expanded on model surfaces with well-defined periodicity of surface roughness, which was around 25 nm [2]. However, to predict the erosion of components in the reactor first wall their surface morphology needs to be taken in to account. In this work the influence of surface roughness on the sputter yield of Mo under keV D ion bombardment was investigated. For this purpose, thin films of Mo (~120 nm) were deposited by pulsed laser deposition onto graphite substrates with varying surface roughness from 5 nm to 2-3 ?m. For samples with surface roughness of 2-3 ?m, we tried to imitate surface finish produced after manufacturing. Samples were irradiated by 1 keV/D ions at different angles of incidence between 0° and 70° to D ion fluence in the range of 1023 D/m2 . For all the surfaces, we observed a strong angular dependence of the sputter yield. For smooth and intermediate surface roughnesses, up to Ra ~ 280 nm, we obtained an increase of the sputter yield with the angle up to a factor of five compared to 0°. In contrast, for surface roughness of 2-3 ?m, the sputtering yield decreases with increasing impact angle. The results served as validation of the Monte Carlo code SDTrimSP-3D [3]. We obtained good agreement between the simulated and experimental sputter yield for surfaces for which we could provide high resolution atomic force microscopy (AFM) surface representations. It turns out that SDTrimSP-3D code is sensitive on input data of sample surface morphologies. As high-resolution surface mapping was not possible for surface roughness of 2-3 ?m, we found large deviation of calculation from the measured data. Also the reasons and possible solutions will be discussed in detail together with possible extrapolations to lower ion energies relevant for divertor conditions in a fusion device.