Failure analysis of beryllium tile assemblies following high heat flux testing for the ITER program

The following document describes the processing, testing and post-test anal ysis of two Be-Cu assemblies that have successfully met the heat load requi rements for the first wall and dome sections for the International Thermonu clear Experimental Reactor (ITER) fusion reactor. Several different joint a ssemblies were evaluated in support of a manufacturing technology investiga tion aimed at diffusion bonding or brazing a beryllium armor tile to a copp er alloy heat sink for fusion reactor applications. Judicious selection of materials and coatings for these assemblies was essential to eliminate or m inimize interactions with the highly reactive beryllium armor material. A t hin titanium layer was used as a diffusion barrier to isolate the copper he at sink from the beryllium armor. To reduce residual stresses produced by d ifferences in the expansion coefficients between the beryllium and copper, a compliant layer of aluminum or aluminum-beryllium (AlBeMet-150) was used. Aluminum was chosen because it does not chemically react with, and exhibit s limited solubility in, beryllium. Two bonding processes were used to prod uce the assemblies. The primary process was a diffusion bonding technique. In this case, undesirable metallurgical reactions were minimized by keeping the materials in a solid state throughout the fabrication cycle. The other process employed an aluminum-silicon layer as a brazing filler material. I n both cases, a hot isostatic press (HIP) furnace was used in conjunction w ith vacuum-canned assemblies in order to minimize oxidation and provide suf ficient pressure on the assemblies for full metal-to-metal contact and subs equent bonding. The two final assemblies were subjected to a suite of tests including: tensile tests and electron and optical metallography. Finally, high heat flux testing was conducted at the electron beam testing system (E BTS) at Sandia National Laboratories, NM. Here, test mockups were fabricate d and subjected to normal heat loads to 10 MW/m(2) (3 Hz) and abnormal heat loads to 250 MJ/m(2) (0.5 s) to determine their performance under simulate d fusion reactor conditions for first wall components. Both assemblies surv ived the normal heat loads with no visual damage. Optical and electron micr oscopy were used to evaluate the extent of the damage at the interfaces fol lowing the VDE simulations. (C) 2000 Elsevier Science B.V. All rights reser ved.