CVD diamond as a heat spreader in X-ray anode assemblies : microstructure and performance

Abstract

Maximising X-ray flux into a focussed spot during analysis by X-ray photoelectron spectroscopy (XPS) is often beneficial. Excitation X-rays for the analysis are usually generated by directing an electron beam at a thin aluminium film. The limiting factor for X-ray flux is potential damage to the thin film by heat from the electron beam. XPS anodes are therefore water cooled. Some commercial instruments also employ a diamond heat spreader beneath the anode film. Diamond has the highest thermal conductivity of any known material, and a cost-effective source is polycrystalline material grown by chemical vapour deposition (CVD). CVD diamond has been shown to have a coefficient of thermal conduction (CTC) that is inhomogeneous and anisotropic. The CTC varies with depth beneath the growth face of the diamond and is greater perpendicular to the face than parallel to it. The CTC can vary by a factor of 5 or more. This is caused by the microstructure of the CVD diamond. This thesis explores the effect of this CTC variation on the performance of the material as a heat spreader in XPS anodes. Detailed finite element analysis of an XPS anode is undertaken, which for the first time incorporates diamond CTC that is temperature dependent, inhomogeneous and anisotropic, and heat delivery into the aluminium film that is distributed over a realistic depth. CVD diamond samples from different suppliers were characterised by white light interferometry (WLI), atomic force microscopy (AFM), scanning electron microscopy (SEM), laser scanning confocal microscopy (LSCM) and Raman microscopy. A test facility was commissioned to perform destructive testing to compare (i) anodes with and without a diamond heat spreader (ii) CVD diamond from different manufacturers and (iii) to compare the relative performance of the nucleation and growth faces. This testing was not completed. The nucleation and growth faces were easily distinguishable by WLI. Surface voids were examined in detail by SEM and appeared to result from polishing damage, incomplete growth or local contamination. LSCM revealed apparent internal voids in some samples. The volume fraction of internal voids is small, so heat transport is unlikely to be affected, but they might be an indicator of lower quality CVD diamond. The material with the most voids was the material with the most non-diamond peaks in the Raman spectrum. Surface voids could cause hot spots if the aluminium film does not have good contact with the underlying diamond. LSCM also identified a highly reflective distinct layer within some samples that could potentially impede heat transport within the diamond. The FEA results indicate that a CVD diamond heat spreader reduces the aluminium temperature from 613 °C to 301 °C with a 20 W load. Having the anode material applied to the nucleation face instead of the growth face gave a temperature of 575 °C. Attempts to improve cooling by making the anode tip wall thinner, or by having the water directly cool the diamond actually caused a small increase in the temperature. The maximum temperature attained is highly sensitive to the aluminium film thickness, the aluminium CTC and the diamond CTC. 65% of the heat was removed by the cooling water from the walls of the anode rather than the tip, regardless of whether a heat spreader is employed. Physical testing was hampered by repeated equipment failures. The performance of the first electron gun suffered due to poor component quality and poor design choices. Even so, an electron beam was extracted from the prototype and moderately focussed onto a phosphor screen. It proved impossible to apply more than 7.5 kV to the electron gun without causing a flashover. A new electron gun design is under way. The FEA results have important implications for anode design. A diamond heat spreader is shown to be effective, but the choice of diamond face is critical. As most of the cooling occurs at the walls of the anode, the return route of the cooling water must fully exploit the area available by passing over as much wall area as possible. A narrow conduit for the effluent could miss important cooling opportunities. Attempts to bring the cooling fluid closer to the hot spot by thinning the tip impedes heat flow to the area where most heat is removed and leads to a temperature increase.