Table of contents

The International Workshop on Carbon Materials, held in Jülich on the 3rd and 4th of September 1998, was the eighth meeting in the series designed to highlight the progress in research and development of carbon-based materials for plasma facing components in controlled fusion devices. This time, the Workshop was held as a satellite meeting prior to the 20th Symposium on Fusion Technology (SOFT) in Marseille. The symposium, which was partly sponsored and excellently organised by the Forschungszentrum Jülich, gathered more than 70 registered participants from Germany, the United Kingdom, France, Japan, the United States, Italy, Austria, Sweden and Switzerland. The scientists represented several leading plasma physics, fusion-related materials research laboratories and industrial companies involved in development of plasma facing materials.

The scientific programme consisted of three major topics: erosion and deposition phenomena, high heat flux testing and neutron irradiation effects in materials. The programme included 14 invited talks and 14 posters. Ample time was reserved for discussions and `hot' topic contributions. The authors presented both an overview and detailed aspects of these vigorously developing interdisciplinary fields of science and technology. The most important aim and outcome of this very interactive meeting was to define and plan future activities to solve specific problems and to clarify many issues concerning the properties and behaviour of materials under high particle fluxes.

On behalf of the Programme Committee I would like to thank all the participants and organisers for their important contributions which made the workshop fruitful, interesting and pleasant.

Programme Committee H Bolt (Forschungszentrum Jülich) J Linke (Forschungszentrum Jülich) V Philipps (Forschungszentrum Jülich) J Roth (IPP Garching) M Rubel (KTH Stockholm) G Vieider (NET Garching)

Hydrogen retention in tokamaks is dominated by two mechanisms: implantation into plasma-facing surfaces and trapping in deposited layers. The amount of hydrogen implantation saturates at typically ~10²¹ atoms m^-2, giving ~2 × 10²³ atoms in the JET first wall (which is ~200 m²), whilst codeposition depends firstly on the quantity of carbon deposited, and secondly on the temperature history of the deposits. Generally, codeposition has dominated the retained H inventory in JET, which was typically ~10²⁴ atoms.

The installation of a divertor in JET has necessitated the presence of water-cooled components to protect the divertor field coils, whereas previously all plasma-facing components were normally at least 300°C. Thick carbon-based films are deposited on surfaces in the vicinity of the inner corner of the divertor on surfaces shadowed from the plasma. Because of the divertor cooling, these deposits are at low temperature and their H:C ratio is at least 0.5:1; as a result their contribution to the overall in-vessel inventory is increased. No comparable deposition is found at the outer divertor. In the Mk IIA divertor the majority of the deposition occurs on cool surfaces many centimetres from the plasma, although with a line-of-sight to the vicinity of the inner strike point, and the average amount deposited per pulse is much greater than previously observed. The number of carbon atoms forming the basis of the deposits amounts to several percent of the ion flux to the inner strike point. The build-up of material leads to spalling (probably on venting to air). The mechanism for the generation and transport of the carbon to form the films is unknown, but ELMs may have a role: it is important to fully understand the phenomenon because of the similarities between the JET and ITER divertor geometries.

Erosion and redeposition of eroded material was measured during the JET Mark I carbon divertor operational period from April 1994 until March 1995 on the walls of the main chamber, the poloidal inner and outer wall limiters and the divertor plates by quantitative surface layer analysis. Ni + Cr + Fe is eroded predominantly from the inner vessel wall and Be from the outer vessel wall. Carbon is eroded from the plasma exposed limiter faces and the plasma exposed parts at the divertor strike points. The material eroded at these areas is predominantly redeposited at the limiter sides and in the divertor in shadowed regions and the corners.

The paper describes in the first part the formation of carbon flakes up to 10–20 µm thickness (average growth rate 2 nm/s) on the graphite tiles of the toroidal belt limiter. This occurred as a consequence of a slight change of the geometry and turned parts of the surface area from net erosion into net deposition zones. The possible influence of the morphology on this behaviour is discussed in the second part by means of an erosion experiment where the gradual disappearance of a boron substrate could be discriminated from the simultaneous carbon deposition on the surface. The two counter-acting processes co-exist within 10–30 µm distance and lead to an extremely non-uniform carbon deposition even in net erosion zones. The carbon agglomeration coincides with surface imperfections, e.g. grooves, but agglomeration by temperature enhanced mobility is not excluded. The changeover from net deposition to net erosion averaged over larger distances can still be observed and is due to the hydrogen and carbon fluxes in the SOL. This is confirmed by Monte-Carlo code calculations.

The modifications of fine grain isotropic graphite surfaces after plasma exposure have been investigated using surface analysis techniques with high spatial resolution in area and depth. The samples are graphite target tiles of ASDEX-Upgrade and coated graphite collector samples exposed for special erosion/deposition experiments in the divertor plasma of ASDEX-Upgrade or in the scrape-off plasma of TEXTOR-94. In addition, a graphite sample was exposed to a low temperature, clean deuterium plasma to study the modifications of the surface morphology during plasma exposure. The results give clear indications of non-uniform erosion and deposition processes. The change of the surface morphology during these processes is discussed.

Redeposition of eroded carbon can form carbon-rich layers on the plasma facing components of experimental fusion devices. The trapping of hydrogen isotopes in these deposits may represent a potential safety hazard in T–D operated devices. Understanding the properties of these deposits, like composition, structure, chemical activity etc, is an important prerequisite to predict the hydrogen retention, and to develop in-situ removal techniques.

Thick layers (~150 µm), which tend to flake off, were found on different wall components of TEXTOR-94 after long term operation (6110 discharges). Flakes with a size of a few mm² had been removed from the main poloidal limiter. They have rough, bubbly appearance, and are porous but mechanically hard. The deposits were examined by SEM, NRA, EPMA, EDX, and OES. They consist of about 80%–90% carbon and incorporate non-uniformly distributed deuterium, Si, B and metals (mainly Fe, Cr, Ni, also as clusters). The D/C-ratio found in these flakes measured by TDS is 4 × 10^-4. Baking the flakes in the air at 250°C reduces their thickness at a rate of about 1.2 µm/h, baking at 450°C gives a rate of about 2.6 µm/h. After baking the flakes turn to ash. The main residues are metals, B and Si, elements that do not form volatile oxidation products.

The interactions of several types of mixed-materials with a bombarding deuterium plasma are described in this paper. The first type of mixed-material surface is a designed, or engineered, surface: a silicon-doped carbon-fiber composite (NS-31). The Si-doped CFC is compared to an identical, but undoped CFC. The net erosion rate, which under these experimental conditions should be dominated by chemical erosion, is reduced by an amount that is about the same as the concentration of the dopant material. Examination of the CFC surface shows that the dopant exists in macroscopic size zones and is not uniformly distributed throughout the CFC. The addition of a more uniformly distributed dopant, in this case beryllium deposited from the plasma on graphite, is shown to reduce the chemical erosion by more than the concentration of the dopant in the surface layer. Finally, the concentration of impurities in the plasma (and therefore the arrival rate of these impurities at the surface) is influential in determining the resultant chemical bonding on the surface. If the arrival rate of carbon at the surface is large, then a surface rich in carbon-carbon bonding develops. If the arrival rate of carbon at the surface is reduced, then the carbon in the surface exhibits preferential carbidic bonding. At low carbon concentration, carbidic bonding is observed in the surface layer regardless of the temperature of the sample during the plasma exposure.

The stationary plasma of the PSI-1 facility was used for studying the ion energy dependence of chemical erosion of different CFC materials in deuterium plasmas. A weak dependence on the bombarding energy was found having a maximum between 150 and 200 eV. Furthermore, long term exposure experiments with two different CFC materials were carried out showing that the reduction of chemical erosion can be explained by covering the surface with impurities. A study of the contribution of energetic neutrals to erosion revealed effects of the order of 50%; their influence, however, depends on the plasma conditions.

Chemical erosion of carbon via hydrocarbon formation is investigated by spectroscopic particle flux measurements and with Langmuir probes in the outer Divertor II of the ASDEX Upgrade tokamak. The measurements are made about 10 cm above the strike point of the separatrix at the vertical target plate for hydrogen ion flux densities up to 10²³ m^-2 s^-1 and up to reactor-relevant values of the power per major plasma radius, P/R ≤ 12 MW/m. The erosion yields obtained exhibit a pronounced hydrogen/deuterium flux dependence and an isotope effect, resulting in lower yields for higher fluxes and a factor of 2 higher yields for deuterium in comparison to hydrogen. Surface temperatures are relatively low and do not exceed 100°C at the position of the measurements documenting the advantageous power spreading features of carbon in combination with the Divertor II vertical target geometry. The interpretation of the measured CH molecular line emission in terms of underlying hydrocarbon molecular fluxes from the wall is hampered by the uncertainties in the atomic data and by the formation of heavier hydrocarbons. The simple model used for the interpretation of the CH band intensity in terms of flux is supported by the behaviour of carbon ion (C⁺, C²⁺) radiation. Owing to the high electron densities, carbon is found to be a very effective coolant for the divertor plasma.

In order to study the fluxes and the spatial distribution of hydrocarbons, which are formed by plasma interaction on limiters and walls of TEXTOR, both spectroscopic and mass analysis measurements have been performed. Spectroscopic diagnostic stations on specially designed limiter locks, where an easy replacement of limiters with different materials is possible, allowed the determination of the hydrocarbon production from carbon, silicon and titanium doped (SiC30, RGTi) graphites, B₄C as well as carbon coated metal limiters of copper, stainless steel, molybdenum, and tungsten. In addition a special carbon limiter was available, which could externally be heated up to temperatures of 1400 K and, therefore, allowed measurements on a homogeneously hot surface. It was found that the methane formation remained approximately constant up to 1050 K with a yield of about 3.5% and dropped well below 1% at 1350 K. Simultaneously the carbon ion fluxes into the plasma decreased by about 50% for ohmically heated plasmas. The fluxes onto the limiters could be increased by an order of magnitude by inserting the limiter deeper into the plasma, which resulted in a drop of the maximum methane production yield to 1% at 3 · 10²³/(m² s). Bulk doped graphites as well as Cu- and SS-limiters show a similar methane production behaviour as pure graphites, whereas for Mo- and W-limiters the hydrocarbon formation rate is low. A pronounced isotope effect could not be detected. Photon efficiencies for the CH/D-band emission were calculated with a local erosion and deposition model (ERO-TEXTOR) taking into account the measured plasma boundary parameters and the atomic data from the Langer model. These calculations justify the application of the factors used for the conversion of measured intensities into CH/D₄ particle fluxes for TEXTOR limiter conditions.

In the JET Mark I and Mark II divertors the in-situ chemical sputtering yield of the carbon target plates has been studied under a wide range of conditions. To evaluate the chemical yield, the incident hydrogenic flux is directly evaluated from Langmuir probe measurements while the methane production rate is inferred from the CD/CH molecular band emission calibrated using methane injection experiments. It is found that the chemical sputtering yield reduces at high flux densities (> 2 × 10²² m^-2 s^-1) and low impact energies (< 50 eV). In the Mark II divertor both the absolute yield and the flux dependence depend upon the tile temperature with Y_chem ∝ Γ^−0.4 at 450 K and Y_chem ∝ Γ^−0.7 at 360 K. In the Mark I divertor the yield exhibits an even stronger flux dependence (Y_chem ∝ Γ^−1.25) which is consistent with the relatively low tile temperature of 300 K. However, the increased flux densities are associated with reduced impact energies which may also contribute to the yield reduction. The chemical yield is also shown to depend strongly on the isotope mass (Y_chem ∝ m_ion) and is consistent with thermal reactions enhanced by radiation damage.

Tungsten test limiters of mushroom shape and a plasma facing area of approximately 100 cm² were exposed at the TEXTOR-94 tokamak to a number of deuterium fuelled discharges performed under various operation conditions. Two types of limiters were tested: a sole tungsten limiter and a twin limiter consisting of two halves, one made of tungsten and another of graphite. The exposed surfaces were examined with ion beam analysis methods and laser profilometry. The formation of some deposition zones was observed near the edges of the limiters. The deuterium-to-carbon concentration ratio was in the range from 0.04 to 0.11 and around 0.2 for the sole tungsten and the twin limiter, respectively. Significant amounts of the co-deposited tungsten and silicon atoms were found on the graphite part of the twin limiter indicating the formation of mixed W-C-Si compounds.

For carbon-based materials physical sputtering, chemical erosion, and radiation enhanced sublimation (RES) are the primary erosion processes due to ion impact. For H impact the influence of dopants (e.g. B, Si, Ti) on the radiation enhanced chemical erosion (Y_therm) at temperatures around 800 K is discussed. For B doping the reduction of the erosion yield could be described by a reduction of the activation energy for H release. At low surface temperatures and low H ion energies a reduction of the chemical enhanced physical sputtering yield (Y_surf) due to dopants is observed and discussed. It is unclear if the reduction is just caused by modification of the surface composition due to preferential erosion of carbon. The surface composition and the erosion yield depend on ion fluence, dopant distribution, production procedure and operational history of the target. On atomic scale the dopant distribution causes changes in the chemical bonds and reactions and, therefore, in the chemistry necessary for both chemical erosion regimes (Y_surf, Y_therm). Also, the basic processes of RES – interstitial production, migration, recombination, and thermal desorption – could be influenced. At high temperatures, at which RES dominates, the stability of the material and sublimation of the dopant has to be taken into account.

Silane (SiH4) has been injected directly into the scrape-off layer of TEXTOR-94 through 10 injection nozzles located at the toroidal edge of one limiter blade. The resulting temporal changes in the line emission of silicon and carbon due to the deposition and erosion of Si are spectroscopically monitored. Thus an erosion time can be defined, which is 5–10 seconds under ohmic heating and 2–3 seconds for auxiliary heated I-mode similar conditions. This method offers a possibility to deposit a hard, robust coating during discharges and is thus a possible coating technique for steady state devices.

The selection of the armour materials for the Plasma Facing Components (PFCs) of the International Thermonuclear Experimental Reactor (ITER) is a trade-off between multiple requirements derived from the unique features of a burning fusion plasma environment. The factors that affect the selection come primarily from the requirements of plasma performance (e.g., minimise impurity contamination in the confined plasma), engineering integrity, component lifetime (e.g., withstand thermal stresses, acceptable erosion, etc.) and safety (minimise tritium and radioactive dust inventories). The current selection in ITER is to use beryllium on the first-wall, upper baffle and on the port limiter surfaces, carbon fibre composites near the strike points of the divertor vertical target and tungsten elsewhere in the divertor and lower baffle modules.

This paper provides the background for this selection vis-à-vis the operating parameters expected during normal and off-normal conditions. The reasons for the selection of the specific grades of armour materials are also described. The effects of the neutron irradiation on the properties of Be, W and carbon fibre composites at the expected ITER conditions are briefly reviewed. Critical issues are discussed together with the necessary future R&D.

This paper describes the European technology programme on high heat flux components for the Next Step fusion reactor with main emphasis on carbon armoured mock-ups. The R&D included the development of high thermal conductivity 3D CFC composites, the manufacturing of monoblock and flat tile components, the development of suitable non-destructive methods and the investigation of the critical heat flux limits.

Development of divertor high heat flux components is one of the critical issues to realize next generation fusion experimental reactors, such as ITER. In ITER, the surface heat flux to the divertor is designed to be 5 MW/m² in normal operation and 20 MW/m² in transient operation. To withstand such high heat fluxes, the plasma-facing surface of the divertor components is covered with refractory armor materials. A carbon-fiber-reinforced carbon composite (CFC) is one of the candidate armor materials for the ITER divertor plate. JAERI has vigorously been developing the divertor high heat flux components with CFC armor materials. High heat flux experiments of various divertor mock-ups were carried out in a high heat flux test facility in JAERI. As a result of a thermal cycling experiment, a small-scale divertor mock-up with 3D-CFC armor tiles could withstand a cyclic heat flux of 20 MW/m² for 1000 cycles with no degradation of thermal performance. A silver-free braze technique using Cu-Mn braze material was found to be a promising solution for the ITER divertor application. As a result of thermal cycling experiments of full-scale ITER divertor mock-ups, the mock-ups with 1D-CFC armor tiles could endure a cyclic heat flux of 5 MW/m² up to 1000 cycles with no degradation of the braze interface of the 1D-CFC armor tiles. On the other hand, the mock-up with 3D-CFC armor tiles showed detachment of some armor tiles in an early stage of the experiment.

The emission of solid particles from strongly heated carbon surfaces may contribute to the erosion of plasma facing components. This process concurs with the thermal sublimation of carbon and is assumed to be dependent on the kind of carbon material. With regard to the application as plasma facing material it is of importance to investigate the mechanism and the process of particle emission with relation to the material structure and to quantify the loss of material due to particle emission depending on the material (e.g. graphite, CFC).

In the experiments an electron beam facility has been used to deposit surface heat loads onto specimens of different carbon materials, esp. fine and coarse grain graphites, CFCs, and pyrolytic carbon. The heat load was applied with power densities of 0.5 to 1.5 GW/m² for pulse durations of 0.1s to 1s. The eroded material was collected by catcher probes which were arranged to allow the quantitative determination of the angular dependence of the particle emission from the heated specimen surface. The evaluation of the probe surfaces by means of high resolution SEM and image analysis gave information on the size distribution of the eroded particles. A small fraction of the particles detected from heated fine grain graphite had sizes of the order of the graphite grains itself (a few µm), and a considerable amount of material was deposited on the probes in form of sub-µm particles. The structure of these particles was examined by TEM. In addition, photographs of the particle emission during heating were taken showing the trajectories of the emitted particles and allowing further comparison with the post-experimental analysis.

Thermal shock tests of neutron irradiated carbon fiber reinforced carbon composites (CFCs) were started with OHBIS to study the neutron irradiation effects on the erosion behavior. Specimens were one directional, two directional and three directional CFCs. These specimens were irradiated with a total neutron fluence of 0.3–0.4 dpa at 556-594 K using Japan Materials Testing Reactor. Heat load conditions were 20 MJ · m^-2 (800 MW · m^-2 × 25 ms and 500 MW · m^-2 × 40 ms). It was clear that the weight loss of the neutron irradiated specimens increased with neutron fluence, and that the erosion area of the neutron irradiated specimens were broader than that of the un-irradiated specimens.

In order to study degradation effects of neutrons on plasma-facing materials and joints, actively-cooled CFC monoblock mock-ups were irradiated in the High Flux Reactor in Petten up to 0.35 dpa at 350 and 750°C. Later, these samples were tested by means of an electron beam facility under static and cyclic heating conditions. The heat removal efficiency and the thermal fatigue behavior of these samples were compared to those of corresponding non-irradiated samples. A significant increase of surface temperature was observed for all CFC samples, due to a reduced thermal conductivity of the materials after neutron irradiation. This effect is less distinctive for samples irradiated at the higher temperature. Long term fatigue tests with 1000 heating cycles at 15 MW/m² did not create any failure of the plasma-facing material or the bond layer of the tested mock-ups.

The most widely accepted model for development of defect structure in neutron irradiated graphite has been such that following the first production of a pair of an interstitial and vacancy, di-interstitials and vacancies are formed and their subsequent growth would result in the production of an interstitial plane or loop in-between the basal planes and vacancy clusters, respectively, which could cause the loss of thermal conductivity and dimensional change.

Recently we have claimed that the formation of vacancy clusters and growth of the interstitial planes are not necessarily a unique interpretation of the damaging process. Instead, the damaging process is described by orientational disordering within the basal planes, i.e. fragmentation into small crystallites and rotation of their crystalline axes, change of stacking order and elongation of the interplanar spacing. The orientational disordering within the basal planes proceeds coordinately over a few layers with their layered correlation maintained. This process accompanies changes in bonding nature producing 5 member- and 7 member-atomic rings as appeared in fullerenes. This is so to speak "self-restoring or reconstruction" to maintain resonance bonds as strict as possible without the formation of dangling bonds.

This paper reviews irradiation effects in graphite such as increase of hydrogen retention, loss of thermal conductivity and dimensional change on the bases of our new model, taking account of the changes of the bonding nature in irradiated graphite.