Author: Dr. Hans Maier, IPP In the beginning of 2005, the ITER-like Wall Project was initiated at JET. The ambitious goal of this project is to investigate for the first time in a real tokamak-environment the plasma-facing materials combination that is intended to be used in ITER. Since JET is currently the largest tokamak experiment in the world, it is best suited for this purpose from a physics point of view, for example when the response of the materials to tokamak-specific pulses of high heat flux, socalled edge-localized modes (ELMs), is of concern.

Figure 1: A pair of tungsten coated carbon fibre compound tiles during heat loading with a hydrogen beam in the GLADIS facility at IPP Garching. The bright horizontal lines correspond to the fibre direction. The vertical features on tile on the right are tensile cracks in the coating. As described in the text, they are oriented perpendicular to the fibres.

Unlike ITER, JET will start this programme with a full tungsten divertor. This allows the investigation of a full metal first wall without carbon contamination, a configuration which ITER might be willing to use for its D-T experiment. The divertor could later be changed to the presently foreseen ITER configuration by installing Carbon Fiber-reinforced Carbon (CFC) targets. For the outer horizontal divertor target plate, the socalled load-bearing septum replacement plate, a solution employing bulk tungsten lamellae has been developed at Forschungszentrum Jülich (see EFDA Newsletter Vol 2006/1). This rather complex approach is, however, expensive and could not be applied to the whole divertor. Instead, it was decided to go for the less expensive approach of tungsten coatings.

As a substrate material the currently employed two-directional carbon fibre-reinforced carbon (CFC) was chosen, since the qualification of a new material with respect to all relevant properties and tile design details would not have been possible within the given tight time frame. Due to this boundary condition, thermo-mechanical problems were to be expected, since the anisotropic CFC material and tungsten are not particularly well adapted with respect to thermal expansion. For this reason, a rather large effort was undertaken to investigate a variety of tungsten deposition methods. Three different coating thicknesses were investigated; two thin types of 4 μm and 10 μm and one thicker type of 200 μm. This resulted in a total of fourteen different sample types, each a combination of a tungsten deposition method and an applied coating thickness. This strategy was chosen to maximise the chance for finding a viable solution.

Research and development was performed at five Euratom Associations: CEA Cadarache, ENEA Frascati, IPP Garching, MEdC Bucharest with support from CEA Cadarache, and TEKES Finland. With the exception of the Romanian Association, this was done in cooperation with industry. The position of task coordinator was assigned to IPP Garching, taking into account the fact that, in the frame of the ASDEX Upgrade tungsten program, IPP had already gathered experience in the field of tungsten coatings dating as far back as 1995 (see EFDA Newsletter 2003/5).

After a short R&D period in 2005, the Associations delivered their optimised coatings to IPP Garching. They were deposited on specially designed identical test tiles incorporating some features of actual JET divertor tiles, which had been manufactured at IPP Garching and, after baking at JET, had been distributed to all participants. At IPP Garching all coating types were subjected to a test program, including for example metallographic investigation and the analysis of the coatings impurity content. The latter was also supported by Forschungszentrum Jülich.

Thermal expansion mismatch

The main part of the test program was a high heat flux investigation conducted in two steps in IPP’s new high heat flux facility GLADIS (see EFDA Newsletter Vol 2005/5). The first step consisted of a thermal screening procedure, where samples were subjected to a number of successively increasing central power densities and pulse durations to determine the ultimate performance limits of the individual coating sample types. This was done up to a central power density of more than 23 MW/m2, which led to peak surface temperatures in excess of 2000°C after a pulse duration of 1.5 s. Based on the results of this screening procedure, a number of coating types were selected for the second step, a thermal cycling procedure.

Given the risk of thermo-mechanical problems mentioned above, this step was considered to be of great importance, since the thermal expansion mismatch between CFC and tungsten might lead to fatigue phenomena. A program of 200 cycles of 10.4 MW/m2 for 5 seconds was adopted. With a typical cooling time between pulses of 2-3 minutes, this took two experimental days per sample pair. The waiting time was determined to allow the tungsten coatings to cool down below the ductile to brittle transition temperature. A third heat load qualification step was performed in the JUDITH facility at Forschungszentrum Jülich, where 1000 ELM-like pulses of 350 MW/m2 were applied for a duration of 1 ms. In parallel to these tests, the possibility of non-destructive testing of the tungsten coatings was investigated at CEA Cadarache and partly also at UKAEA Culham.

Figure 1 shows a pair of samples during thermal loading with a hydrogen beam in the GLADIS facility during the thermal cycling step. In this step, two tiles are placed next to each other while the power density profile is centred in the middle so that two tiles are loaded simultaneously. On some samples the coatings developed partial delamination and subsequent local melting during the screening program.

A general result, however, was the formation of tensile cracks in the coatings during heat loading, due to the thermal expansion mismatch of tungsten coatings and CFC substrate tiles. The cracks occurred in the fibre reinforced direction of the CFC material, in which the thermal expansion of the substrate is smaller than that of the tungsten coatings. From the orientation of the cracks it must be concluded that the failure occurred after the coatings had become stress-relaxed at high temperature. In the nonreinforced direction of the substrate material, coatings which are stress-relaxed at high temperature experience compression upon cool-down. The second principal result was that tungsten films in the micrometer range of thickness are not stiff enough. Upon repeated compressive stress this insufficient stiffness leads to instability against delamination and buckling. This is accompanied by fatigue cracks. This buckling and cracking is illustrated in the scanning electron image shown in figure 2.

Except for the samples produced by the Romanian Association MEdC, this buckling instability failure occurred on all samples with coatings in the μm range. Only the thick 200 μm coatings proved to be stable against buckling in the cyclic loading program. Therefore the 200 μm coating thickness was selected for application in the JET tungsten divertor. Since some of the tested 200 μm coatings displayed local failures, it was decided that a test at a moderate heat flux is mandatory for all tiles prior to installation in JET. This will be done in the JUDITH II facility in Forschungszentrum Jülich with support from IPP Garching for realtime data processing.

JET is equipped with a plasma heating system based on neutral beam injection. Energetic neutral hydrogen atoms are produced and injected into the plasma. These particles collide with plasma particles, are then ionized and deposit their energy into the plasma. Some of them can, however, pass through the whole plasma diameter without collisions – these are the “shinethrough” particles. They hit the wall at full energy which can correspond to a considerable heat load. Therefore some locations of the main chamber inner wall will also require tungsten coatings.

Figure 2: Scanning electron microscopic image of the buckling failure described in the text. This failure occurred on thin coatings during cyclic heat loading. The image is taken such that the coating looks like being illuminated by a light source. In the centre of the image a buckle therefore appears with a bright and a shadowed flank. The cracks running vertically in the image, i.e. perpendicular to the fibres, are the tensile cracks described in the text. The horizontal cracks running parallel to the buckle flanks are a fatigue phenomenon.

For this application, 10 μm coatings are sufficient and are now being produced by the Association MEdC Bucharest. They are made by a combination of magnetron sputtering and energetic ion beam implantation for simultaneous stress relaxation. They are equipped with an intermediate layer of molybdenum. These samples did not develop the above-mentioned buckling failure in the GLADIS cycling tests and failed only under the extreme conditions of ELM-like loading which are not expected to occur on the “shine-through” protections. A setup for coating large tiles will be constructed and commissioned in Bucharest. After a further qualification step, this setup will be used for large-scale production with accompanying support and testing at IPP Garching.