Influence of the dual-scale random morphology on the heat conduction of plasma-sprayed tungsten via image-based FEM

Vacuum-plasma sprayed tungsten (VPS-W) is a coating candidate for nuclear fusion reactors which exhibits a multimodal lamellar microstructure spanning two spatial scales. In the porous morphology, macroscopic porous inclusions (up to 100 μm in size) coexist with thin crack-like defects of sub-micrometric thickness. The crack-like defects develop between solidified tungsten droplets during deposition. They can be of two types: either fully open gaps (average thickness ∼1 μm) or inter-drop boundaries partially in contact showing sub-micrometric thickness (referred here as contact-zone – CZ). While perfect insulation can be assumed in vacuum for bulky pores and open cracks, a reduced thermal contact is expected across the CZ. However, the conductivity reduction in CZ and its impact on the overall coating conductivity are difficult to be estimated. In this study, an image-based finite-element method (IB-FEM) was adopted, which was able to numerically account for the dual-scale morphology of VPS-W. To this end, the dedicated software OOF2 was used for adaptive mesh generation in combination with image analysis techniques. In order to map the multimodal random morphology, high-resolution imaging of the microstructure was required on spatially extended domains. cal predictions of thermal diffusivity were in good agreement with experimental measurements in vacuum. The IB-FEM approach allowed a precise estimation of the reduced conductivity in CZ and the isolation of the contributions of the three different defect-types. It was found that the isolated CZ caused an average reduction of the coating diffusivity up to 30% when fully insulating. Effective conductivity in contact-zone turned to be less than 1% that of bulk tungsten. Compared to analytical estimations, IB-FEM predictions showed a better agreement with experimental data.