Investigation on new Coatings and an in-situ Bake-Out for the SIS18 Beam Loss Collimators

For an optimal functionality and a high durability of the planned SIS18 beam loss collimators a low ion-induced desorption yield of the collimator material is required. Gold coated copper was found to be the material of choice: copper has a high thermal conductivity, a high specific heat and can be produced with high purity for reasonable prizes. The gold layer prevents the copper from oxidation. The influence of the oxidation on the desorption yield was reported to be significant [1]. But we have also shown that copper and gold diffuse during a typical UHV bake-out cycle [2]. In 2007 we have studied nickel diffusion barriers using Rutherford Backscattering Spectroscopy (RBS). As reported [3] desorption turned out to be predominately a surface effect with a strong link to the bulk properties. Additionally there was always a cleaning observed of the target under ion irradiation. The total amount of desorbed gas corresponds always to roughly one surface monolayer. To observe best cleaning of the target with 1.4 MeV/u Xe ions a flux of a few 1 · 10 ions per cm is necessary. Since a beam cleaning of the loss regions in SIS18 is not possible we have tried to minimize the surface coverage by a special bake-out treatment of the target. The idea is to keep the target during UHV bake-out always higher in temperature compared to the environment in order to minimize the readsorption of gas on the surface. Nickel Diffusion Barrier To avoid diffusion of gold and copper, which is stimulated by the UHV bake-out procedure, nickel diffusion barriers were investigated using RBS. The measurements were performed with 1.4 MeV/u C ions. Using RBS we gain sensitivity to heavy elements compared to ERDA. Additionally, sensitivity and depth resolution are increased by a comparable heavy projectile like C with the disadvantage to lose any information about low Z target components. Fig. 1 shows the RBS spectra for a nickel-gold coating on copper before and after bake-out. The gold layer thickness is around 400 nm and the nickel layer is about 1 μm. Because of the total layer thickness of 1.4 μm the copper substrate is not detected before bake-out (Fig. 1, blue spectrum) due to the limited penetration depth of the C ions. The red spectrum shows the same sample after bake-out. Essentially the shape of the nickel distribution is changed since copper starts to diffuse across the nickel layer to the sample surface indicated by the dashed line. Additionally, the gold layer is broadened due to the diffusion of gold into nickel. However, the diffusion is