Capability of gas flow sputtering to coat non line of sight areas

Kurzfassung

Physical vapour deposition techniques (PVD) such as magnetron sputtering and electron-beam vapour deposition operate in high vacuum. The mean free path of the sputtered or evaporated atoms is much larger than the source to substrate distance. The atoms arrive nearly without collision at the substrate. As a result, only areas are coated which are in line of sight to the deposition source. In contrast to these PVD techniques gas flow sputtering (GFS) is know for its capability to coat non line of sight areas (NLOS) without substrate manipulation. Atoms are sputtered from a hollow cathode by glow discharge. A high inert gas flow streams through the cathode and provides transportation of sputtered material to the substrate. The mixture of inert gas and sputtered atoms performs a circulation around the contour of the geometry and reaches NLOS areas. Coating thickness distribution depends on fluid dynamics. In this investigation a basic understanding for the observed thickness distributions and a model for the mass transport was generated. Stripes, cylinders and turbine blades were coated with pure titanium. Thickness distributions and microstructure were examined by scanning electron microscopy. Some coating procedures were simulated by computational fluid dynamics to understand the correlation between gas flow and coating thickness distribution. All substrates, especially their non line of sight areas, were completely covered with a coating. The coating thickness is inhomogeneously distributed along the substrate contour. This observation can be explained by the successfully adapted mass transport model. In general, the particles diffuse perpendicularly to the main flow direction out of the gas flow. Due to the wall friction, a boundary layer is formed near the substrate surface and increases in the downstream direction. The deposition rate and the coating thickness respectively decrease with increasing boundary layer thickness. This model and the gas flow simulation enable predicting and optimizing the coating thickness distribution for any kind of complex substrate geometry.