Speaker
Description
In fusion devices, the plasma-facing wall is primarily eroded via physical sputtering, which introduces impurities into the plasma core that degrade device performance. Aiming to gain a fundamental understanding of the sputtering processes relevant for plasma-wall interactions with machined plasma-facing components, we investigate angular distributions of sputtered particles in a controlled laboratory environment. Already in 1955 G. K. Wehner [1] discovered that in single crystals, sputtered particles are predominantly ejected along certain symmetry axis. However, plasma-facing materials typically have more complex structural compositions and surface topologies. Surface roughness is also known to significantly influence sputtering; recently, the first moment of the surface inclination angle distribution was found to be a strong predictive parameter [2, 3].
A rough tungsten model system served as a substitute for industrially machined tungsten first-wall components used in modern fusion devices, such as WEST, ASDEX-Upgrade or ITER. To achieve a specific surface roughness, a silicon substrate was etched under controlled conditions, resulting in pyramid-like surface structures on the micrometer scale. A thin tungsten coating was grown onto it by magnetron sputter deposition, preserving the pyramidal substrate topology. Roughness parameters and topography were quantified and investigated via atomic force microscopy (AFM). The sputtering behavior of one flat and two rough samples was studied, their root mean square roughness values being approximately 15 nm, 690 nm and 1.2 µm, respectively. Irradiation was performed with with a low flux ($\Gamma\approx10^{16}Ar^{+}m^{-2}s^{-1}$), energy selected 2 keV argon ion beam. For the angular resolved measurement of the sputtered particles, a precise quartz crystal microbalance (QCM) was used. Mounted on a 5-axis manipulator, the QCM enabled measurement of angular-resolved mass gain rates, corresponding to the local intensity of sputtered particle flux relative to the sample surface normal.
Simulations were carried out using the in-house developed “SPuttering simulation via RAYtracing of particles” – SPRAY code. It utilizes a data repository of sputtering data generated by state of the art binary collision approximation codes, such as SDTrimSP-7, and arbitrary input geometries, including AFM image data. First comparisons between experiment and simulation show that only accounting for the surface roughness in modeling is not sufficient, implying that further characteristics, such as e.g. crystalline texture, have to be considered to accurately model and understand the observed sputtering process.
[1] G.K. Wehner, Journal of Applied Physics, 26 1056 (1955).
[2] C. Cupak et al., Applied Surface Science, 570 (2021)
[3] A. Uccello et al., Nuclear Fusion, 65 (2025)