Table of contents

The implications for the use of carbon based PFCs in next step devices are reviewed, with ITER as the reference example device. Issues with respect to steady-state operation concerning the carbon source at the divertor target in realistic conditions (including impurity seeding and mixed material formation caused by interaction of the plasma with the main chamber PFCs) and their implications for tritium retention are discussed taking into account other restrictions for next step plasma operation, such as the achievement of detached divertor operation with peak power loads <10 MW/m². The effect of transient power loads on the divertor CFC target caused by ELMs are discussed and the implications for the use of CFCs at the divertor target for regimes with repetitive ELMs, acceptable from the divertor lifetime point of view, are reviewed.

In the new CIEL configuration of Tore Supra, all the plasma facing components are actively cooled. The surface area of the wall that is covered with carbon measures about 15 m² (2 to 4 m² of which in close interaction with the plasma). Steady-state plasma conditions up to 4 min 25 s have been maintained in this configuration. In these experiments, the required gas injection to maintain the prescribed density remains constant during the whole discharge. The exhausted flux is also constant and equal to 40 ÷ 50% of the injected flux. Therefore, 50 ÷ 60% of the injected particles remain trapped in the vessel, the total retention being proportional to the plasma duration. Since the amount of gas recovered between shots or by He-glow discharges does not always balance the injected gas, it follows that a quantity of deuterium remains indefinitely trapped in the vessel, which appears as an infinite reservoir. This reservoir is believed to be dominated by co-deposited layers, as observed in several places of the vessel. The thickest deposits (up to 800 µm) are observed on the leading edge of the neutralizers of the pump limiter. They display a column-like shape (typical growth rate ~20 nm/s) and have a graphite-like structure. Their deuterium concentration is D/C ~ 1%. Conversely, in regions that are shadowed from the direct plasma flux, the deposits show a smoother shape and their deuterium content is typically ~10 ÷ 15%.

The Large Helical Device (LHD) has employed a graphite helical divertor in a stainless steel vacuum vessel since the beginning of its third experimental campaign in 1999. Initially, installation of the carbon divertor resulted in a dramatic decrease in iron impurities as evidenced by the hollowing out of the radiation profiles and by spectroscopic measurements compared to the previous campaign using a stainless steel divertor. However core radiation and radiation from iron spectral lines gradually increased during the third campaign, which was attributed to sputtering of the stainless steel walls during glow discharge cleaning and redeposition on the carbon divertor tiles.

Studies have shown that the graphite tiles were eroded at the divertor leg strike points to a depth of 5 µm during one experimental campaign (~10,000 shots), which is in agreement with numerical calculations using the EDDY code. Carbon is redeposited at and near the strike points along with metallic impurities forming a mixed layer, which may be contributing to iron impurities in the plasma.

In LHD radiative collapse resulting from an edge thermal instability caused by radiation of gasses desorbed from the divertor during long pulse discharges terminates the discharge at high density. This radiative collapse is one of the limitations for long pulse discharges up to 150 s.

A series of quiescent L-mode discharges have been used with varying degrees of divertor attachment to study carbon erosion in the DIII-D tokamak divertor. Spectra of atomic and molecular carbon plasma emissions are measured across the divertor. Predictions of carbon emission resulting from chemical and physical sputtering at the outer strikepoint are provided by the WBC transport code. For attached ionizing plasmas (T_e ~ 20 eV and carbon surface T ~ 350 K) the CD and C₂ carbon radical emissions are consistent with a total chemical sputtering yield, Y_chem ~ 0.3%. During sweeps of the inner divertor leg, CD and C₂ emissions indicate the main-wall tiles has a six times higher Y_chem than the divertor tiles despite identical incident plasma conditions. Emission of atomic (CI) and singly ionized carbon are dominated by physical sputtering with a measured yield Y_phys ~ 2% as expected from laboratory data, indicating that chemical sputtering plays a minor role as a divertor carbon source. The Doppler broadening of the CI emission agrees with the physical sputtering model, but the predicted Doppler shift is a factor three larger than the experiment. CD and C₂ emission are reduced below detection during divertor detachment (T_e ~ 1–2 eV), an indication of the suppression of chemical erosion with Y_chem < 10^-4 based on expected photon intensity from WBC.

The paper focusses on the conversion of photon fluxes into particle fluxes by means of photon efficiencies in tokamaks. Both flux contributions to the chemical erosion yield, the deuterium flux, as well as the eroded hydrocarbon flux, are critically investigated, and possible uncertainties in the data evaluation are identified. Inverse photon efficiencies, so-called S/XB or D/XB values, are critical factors in the conversion of the fluxes. For deuterium, the flux determination is influenced by the presence of a molecular source. In contrast, the hydrocarbon flux is also uncertain due to the indirect conversion from photon to particle fluxes. In particular, the conversion of CD photon fluxes into CD₄ particle fluxes can lead to misinterpretation, because C₂D_y and CD₄ can finally contribute to the flux of CD radicals. A correction formula for the total hydrocarbon flux which includes the contribution of different hydrocarbon sources is introduced.

We also present results from spectra simulations of the A-X transition of CD, which show that the usual observation of the CD band (429.4 nm–430.9 nm) may lead to an underestimation of the total photon flux from the whole A-X transition of more than a factor of two. Up to now, the modelling with erosion and deposition codes only takes the complete electronic transition into account. A unification of the quantities is necessary to compare modelling and experiment.

Today carbon is the most common first wall material in fusion experiments, whereas the first wall of the next step device will consist of a mixture of elements. Especially tungsten has been shown to be an alternative to low-Z materials. However, even with 40% of tungsten coated plasma facing components, carbon is still the dominant impurity at ASDEX Upgrade. A consistent picture of the carbon migration in ASDEX Upgrade has been achieved. Primary carbon sources are the protection limiters at the low field side of the main chamber. Eroded carbon is distributed all over the main chamber. So, the initially tungsten coated central column acts as the main carbon source during discharges, even though a considerable amount of tungsten surfaces persists. Carbon coverage of the central column can significantly change on a shot to shot basis. The divertor target plates act as a strong carbon sink. Deposits are found at the inner and outer divertor, which may be re-eroded forming precursors for layer production at remote areas. In ASDEX Upgrade, deposits on the subdivertor structure are formed by hydrocarbons with a high effective sticking coefficient. A parasitic plasma at these locations may enhance the surface loss probability by surface activation. At more remote areas, such as the pump ducts, a very small deposition is found. Non sticking hydro-carbons are effectively pumped by the cryopump and turbo molecular pumps.

This paper summarizes observations about net carbon erosion and re-deposition in the limiter tokamak TEXTOR, and the divertor tokamaks ASDEX Upgrade and JET. The major carbon source in TEXTOR is the toroidal graphite belt limiter. Eroded carbon is mainly re-deposited on net deposition dominated areas of that limiter, and on areas perpendicular to the magnetic field in the scrape-off layer. In ASDEX Upgrade and JET the major carbon sources are located in the main chamber. Eroded carbon is re-deposited on the divertor tiles in the inner and to a lesser extent outer divertors. Carbon migration to remote areas without direct plasma contact like the JET louvres, or the area below the divertors II and IIb of ASDEX Upgrade, is observed. Soft, hydrogen rich hydrocarbon layers with a deuterium to carbon ratio of 0.7–1.4 are formed in these areas. Precursor for this layer formation is sticking of hydrocarbon radicals like CD₃ or C₂D_x (x = 1, 3, 5) created by re-erosion of deposited hydrocarbon layers on the divertor tiles.

Two regimes of the chemical erosion of carbon materials under hydrogen bombardment have been separated: (i) the thermally activated regime, Y_therm, with the maximal erosion yield in the temperature range between 550 and 850 K, and (ii) the so-called "surface" regime, Y_surf, at low temperatures (~300 K) and low impact energies (< 100 eV). Doping carbon materials largely reduces their chemical reactivity with hydrogen and their chemical erosion. In addition, dopant enrichment at the surface due to preferential sputtering of carbon contributes to a reduction of the erosion yield. Erosion measurements with 30 eV and 1 keV D for various doped carbon materials with dopant concentration between 0.25 and 13 at.% were performed at temperatures between 77 and 1100 K. For Y_surf at high ion fluences (> 10²⁵ D/m²), a reduction of the erosion yield by one order of magnitude is observed for fine-grain carbide-doped graphites. Scanning electron microscopy (SEM) allows to associate these fluence dependencies with the evolution of a rough surface morphology of several µm in the erosion area. For Y_therm an almost complete suppression of the CD₄-production yield is observed for Ti-doped C layers. This reduction due to the doping on atomic scale exceeds all previously observed reductions of materials with a coarser dopant distribution. For all investigated carbon materials, the yield below RT does not depend on temperature.

Multi-species impact is of high relevance for fusion reactors, where, in addition to the hydrogen fuel, the He ash, impurity species (e.g., C and O), and in some cases injected elements are also present. Combined irradiation of graphite with relatively low-energy H⁺ (typically below 100 eV) and energetic (above 300 eV) hydrogenic and non-hydrogenic ions leads to an enhancement of the chemical erosion yield compared to the low-energy H⁺-only case. This is due to the additional energy deposition and associated lattice damage by the added energetic ions. Studies of combined species irradiation also have the potential to enhance our understanding of the mechanisms leading to erosion and hydrogen transport in graphite. Mixed isotope experiments with simultaneous H⁺–D⁺ irradiation of graphite led to the conclusion that thermalized H and D move and recombine on internal surfaces within the implantation zone. In contrast, during H⁺–D⁺ irradiation, methane molecules are formed at the end of range of the incident ions. Combined H⁺ and O⁺ irradiation of graphite leads to the formation of hydrocarbons, carbon oxides and water, with water being formed at the end of the O⁺ implant range.

Experiments at UC San Diego PISCES-B, in collaboration with EFDA, are investigating the influence of beryllium impurities on deuterium plasma erosion of graphite material. The experiments are designed to reduce uncertainties in the prediction of tritium retention in redeposited mixed-materials expected in future burning plasma devices. Earlier PISCES-B experiments hinted that small amounts of beryllium surface impurities on graphite could reduce the rate of chemical erosion. In the present experiment, beryllium is seeded into a deuterium plasma to simulate the flow of the scrape-off layer plasma from the first wall to the divertor in ITER. Plasma containing low levels of Be impurities have so far been investigated and a greater reduction of carbon erosion is found. For beryllium concentrations above ∼ 0.1%, a thin coating of beryllium is observed to form on the surface of the graphite. This beryllium layer suppresses chemical and physical erosion of the underlying carbon.

The ITER divertor plasma is expected to contain percent levels of beryllium impurities, so similar beryllium deposition may occur on the ITER divertor plates. Future experiments at PISCES will quantify the formation rate of redeposited carbonaceous films and their suppression, and aim at better simulating ITER conditions including higher operating temperature ranges for the graphite targets, and larger beryllium impurity fractions.

A series of experiments dedicated to the analysis of transport and sticking of hydrocarbon molecules have been performed in the plasma generator PSI-2. As a source of hydrocarbons defined amounts of CH₄ and C₂H₄ have been injected. After injection the molecules undergo a series of ionisation and dissociation reactions. The thickness change of deposited layers on temperature controlled collector surfaces (outside the plasma column) was measured in-situ by means of an optical spectroscopy for different plasma conditions (density, working gas, injection position) and as function of the collector temperature. It was found that the flux of atomic hydrogen mainly determines whether net-deposition or net-erosion occurs. For the simulation of the experiments the 3D Monte Carlo code ERO was applied allowing a numerical description of the emission, transport, deposition and re-erosion cycle of hydrocarbon molecules.

In this paper we discuss the formation and erosion of multi-component materials ('mixed materials') on metals with respect to carbon deposition from the vapor phase. The carbide formation and carbon accumulation at the surface are governed both by chemical reactions and by carbon diffusion into the bulk metal. Carbide formation at room temperature is restricted to several monolayers at the interface between carbon and supporting metal. Additional energy entry into the system either by ion bombardment or by annealing enhances the carbide formation. The carbon amount present at the surface decreases after a material-dependent temperature threshold for diffusion in the case of metallic carbides of transition metals. Taking into account oxygen as a reactive plasma impurity, e.g. by starting with an oxidized surface before carbon deposition, new erosion channels open up in these ternary systems due to chemical interactions, rendering interaction simulations by kinematic Monte Carlo models insufficient. Chemical erosion mechanisms additional to mere kinematic processes are also responsible for carbon erosion by deuterium ion impact, shown for film thicknesses smaller than the ion range. They lead to enhanced carbon film erosion rates compared to erosion by sputtering.

Tritium removal is a major unsolved development task for next-step devices with carbon plasma facing components. The 2–3 order of magnitude increase in duty cycle and associated tritium accumulation rate in a next-step tokamak will place unprecedented demands on tritium removal technology. The associated technical risk can be mitigated only if suitable removal techniques are demonstrated on tokamaks before the construction of a next-step device. This article reviews the history of codeposition, the tritium experience of TFTR and JET and the tritium removal rate required to support ITER's planned operational schedule. The merits and shortcomings of various tritium removal techniques are discussed with particular emphasis on oxidation and laser surface heating.

Infrared emission from plasma exposed and virgin CFC, as well as virgin fine grain graphite was investigated in the laboratory with a spatial resolution of 30 µm. The surface structure, as measured with SEM, was different for both CFC probes, but the infrared emission show a comparable spatial structure. The error for a large area temperature measurement is estimated from a two component model. It decreases with increasing temperature and is below 10% at 800 K for the probes under investigation.

In order to study the influence of the impurity sources on line emissions, gas fuelling experiment were performed with controlled CD₄ gas inlet into the private flux region (PFR) of the inner and outer divertor legs, as well as into the outer scrape-off layer (SOL) of JET. Comparison discharges were performed with D₂ injection into the outer SOL, with the same deuterium source rate as for the CD₄ pulse. Changes in the divertor spectroscopic signals have been analysed and show a similar behaviour for CD, CII and CIII emission during the inner divertor fuelling into either SOL or PFR, indicating that these signals are independent of the location of methane sources. In contrast to the inner divertor puffing, CD₄ fuelling into the PFR of the outer divertor shows the strong dependence of CD-emission on the location of injection to the divertor leg. The evolution of carbon impurity profiles in L-mode discharges will be discussed, with respect to its dependence on the location of methane fuelling (inner/outer divertor leg) and on the position of the strike points.

The temperature measured on the surface of particle neutralisers of Tore Supra, i.e. the target plates under the toroidal pump limiter that channel particles into the pumping ducts, is significantly higher than expected because of thick carbon deposits formed on those plates. With growing deposit thickness their temperature and cooling time constants increase. Lower hybrid current drive heating and gas puffing seem to enhance the layer growth. The deposits form at first isolated hillocks giving rise to an enhanced luminance in the near infrared range. Later on, they may form continuous films (with black-body like spectra) which may flake off again distorting the spectra. The thermal behaviour of thick continuous films can be described by a radiative model. The conducted power flux density to the neutralisers for which this model is in accordance with the measurements is of the order of 0.1 MW/m² in ohmic discharges. This requires characteristic scrape-off layer decay lengths of 7 to 8 mm as far as 5 cm outside the last closed magnetic flux surface. The thermal diffusivities deduced with this model from observations in Tore Supra and from laboratory experiments show values 2 to 3 orders of magnitude below the values characteristic for carbon fibre composites.

Plasma facing components from JET and TEXTOR were studied. The emphasis was on the comparison of co-deposition, material mixing and fuel inventory on plasma facing and side surfaces of tiles, i.e. in gaps separating the tiles. Integrated fuel content in gaps of the Mk-I JET divertor floor was approximately two times greater than detected on the plasma facing surfaces. Taking into account similarities between the Mk-I structure and the castellation in the ITER divertor, the impact of the tile shaping on the tritium inventory is addressed. Deposition on the side of limiter tiles in TEXTOR was around 3–5% of that on the plasma facing surfaces. Experiments aiming at a deeper insight into the deposition on ITER-relevant components are also proposed.

Formation of layers deposited in remote areas of TEXTOR are discussed in detail. Their composition, thickness and optical properties were examined. These deposits form soft, semitransparent films consisting mainly of carbon, hydrogen and deuterium. The content of hydrogenic species depends on temperature of the location, where the deposit had grown. The D/C fraction varies between ~ 0.07 at the hot part of the duct inside the vacuum vessel and ≥ 0.7 in cold sections of the pump duct. The mean deposition rate of carbon in the pump ducts was ~ 1 × 10²¹ C/h, which corresponds to about 0.1% of carbon eroded from the blades of the poloidal belt limiter. Despite the low deposition rate of carbon and, consequently, also of deuterium, these deposits might be decisive in view of tritium retention. Deposition pattern observed in the ducts are compared with computer simulations assuming neutral hydrocarbon radicals as originators of the films.

Carbide-doped and pure graphites were eroded by energy- and mass-separated hydrogen ion beams and by hydrogen containing planar inductively coupled RF plasmas. The erosion yield for 30 eV D impact on doped graphites and its fluence dependence are in the same order of magnitude for both exposure types. The D flux from the plasma onto the specimens (~3.1 × 10²¹ m^-2 s^-1) contains energetic D ions and thermal D (650 K) in the ratio 1 to 5. The contribution of the thermal D to the erosion yield reaches at most the same magnitude as the erosion yield by ions only (30 eV). This contribution depends on the structure of the graphite. In the case of pure graphites the observed erosion yield is independent of the fluence for both types of exposure, although a rough surface morphology is created. For doped graphites reductions of the erosion yield with fluence by a factor of about 10 are observed, while the surfaces get rough and strongly enriched with dopants.

A Quartz Microbalance (QMB) system was implemented in the inner divertor region of JET in order to measure in situ and time resolved (minimum exposure time ≥ 0.1 s) material fluxes (mainly carbon) and layer deposition. The system has been developed to operate at temperatures up to 200°C. The aim is to investigate carbon transport to the remote areas, and hence the tritium retention in dependence on plasma conditions. This question is still a major concern for the ITER operation. The mass sensitivity of the system is S_m = 1.5 × 10^-8 [g/Hz cm²]. First reliable measurements were made during the C5 campaign (March–May 2002; ≈ 1000 plasma discharges). The results presented are based on 74 selected exposures (694 s) under various conditions (strike point position, input power, neutral pressure, ELM frequency). Most influencing on the carbon deposition in the remote area seems to be the geometry i.e. the strike point position on the divertor tiles. In average 1.9 × 10^-4 C-atom are deposited per deuterium ion flowing into the inner divertor.

Understanding the mechanisms of the chemical erosion of carbon by hydrogen is of great importance for the application and development of plasma-facing materials, as well as the macroscopic scale modeling of plasma-surface interactions in fusion devices. In the recent years there has been a number of atomistic modeling studies, namely molecular dynamics and electronic structure calculations, on these processes. This article is an overview of these investigations, with comparison to experimental measurements as well as current theoretical understanding of the chemical erosion of carbon. Implications of the modeling are discussed along with an outlook on future studies.

The material mixing effects on the erosion and deposition processes of W under simultaneous bombardment with deuterium (D) ions and a small amount of carbon (C) ions are investigated by using a dynamic Monte Carlo simulation code, EDDY. The transition from deposition to erosion occurs abruptly when the plasma temperature is increased. Below the transition temperature, a thick C layer appears as a result of the low reflection coefficient of the deposited C so that the W is perfectly protected against sputter erosion. Above the transition temperature, a balance between the incident and released C on the W surface is reached where the surface is eroded only due to the reduced physical sputtering of the W-C mixed layer.

The erosion and deposition patterns observed on the W side of the W-C twin test limiter exposed to TEXTOR edge plasmas are reproduced by using the modified EDDY code. On most parts of the surface, the enhanced physical sputtering of the deposited C suppresses the formation of a thick C layer, whereas near the limiter edge the C deposition dominates where the thickness increases with increasing C fluence. The extent of the deposited area which is broader on Ta than on W is simulated due to higher surface temperature (stronger diffusion) or lower chemical sputtering yield of the deposited C.

Reflection coefficients for carbon and hydrocarbon atoms/molecules on carbon-based surfaces are critically needed for plasma-surface interaction analysis in fusion devices, as carbon will continue to be used in next step devices like ITER. These have been calculated at different energies and angles with a molecular dynamics code using the Brenner hydrocarbon potential. Hydrogen saturated graphite was prepared by bombarding a graphite lattice with hydrogen, until a saturation at ~0.42 H:C. Carbon at 45° has a reflection coefficient (R) of 0.64 ± 0.01 at thermal energy, decreasing to 0.19 ± 0.01 at 10 eV. Carbon dimers (R_thermal = 0.51, R_{>1 eV} ~ 0.10) tend to stick more readily than carbon trimers (R_thermal = 0.63, R_{10 eV} = 0.16). Hydrocarbons reflect as molecules at thermal energies and break up at higher energies. The total reflection via these fragments decreases with energy, the number of unpaired electrons, and changing hybridization from sp³ to sp² to sp. The results compare reasonably well with binary collision modeling for higher energies and experimental sticking data at thermal energies. A second surface, representing a "soft" redeposited carbon layer formed by the deposition of hydrocarbons onto a graphite surface, is also analyzed. In general, reflection is lower from the "soft" surface by 0.1–0.2. This reflection data can and has been incorporated in erosion/redeposition codes to allow improved modeling of chemically eroded carbon transport in fusion devices.

A significant R&D effort was carried out in EU to develop suitable CFC materials with a three-directional fibre structure. The main goal was to obtain a high thermal conductivity, which is required to remove the expected heat load with a sacrificial thickness of at least 15–20 mm, with a proper balance of the mechanical properties. Two CFC materials have been developed and are both suitable for the ITER divertor construction, namely NB31 by the French company Snecma Propulsion Solide and Concept 2 by the British company Dunlop Aerospace. The main R&D effort is now focussed in developing cheaper grades and optimising the manufacturing process.

The development of suitable technologies for the manufacturing of carbon armoured high heat flux components resulted in the construction of a medium-scale divertor prototype, which was successfully tested well above the ITER design requirements. This component was followed by the manufacturing of a full-scale prototype, which is now being high heat flux tested. In parallel to the technology development, suitable non-destructive methods have been developed to inspect the delicate CFC/Cu joint. These techniques include: infrared examinations, lock-in thermography and ultrasonic examinations.

An important topic that will be addressed in the near future is the definition of suitable acceptance criteria for carbon armoured components in view of the series production for ITER.

Two experimental campaigns (2001 and 2002) have been carried out with the tokamak Tore Supra since the complete upgrade of all plasma facing components. The toroidal pumped limiter (TPL) allows reliable steady state operation at significant injected power (up to 8.5 MW peak, 4.3 minutes at 3 MW). One of the first and main result is related to the abundance and diversity of carbonaceous deposits on plasma facing surfaces. They have been observed on the TPL on surfaces close to the plasma frontier. On neutralisers located at pumping ducts under the limiter, a layer builds up rapidly and the thickness reaches 840 µm after 2.3 hours of plasma. These deposits and layers distort the infrared measures and the deduction of the incident heat flux from the temperature is difficult. A simple 1D thermal calculation tool linked to the database has been developed and is used to allow flexible analysis. On the TPL, the aim is to evaluate the ageing of the bond between the tile and the metallic cooling structure from the surface temperature measurement. So far, no evidence of ageing has been observed after the two first years of operation.

New diagnostics were installed to observe dynamical erosion processes under intense transient heat loads. Particle release from graphite and CFC was studied. Released particles from graphite consist of small and large particles. These were observed in signals of photodiode as quasi-continuum and sharp peaks. Large particle release from graphite was significantly enhanced at elevated temperatures. This implies that there is a threshold temperature for the particle releases from graphite, whereas, the particle release from CFC did not change very much at elevated temperature. The mean speed of small particles from graphite was dependent on the applied power density. This might be due to the increase of the particle size. The maximum speed of large particles was estimated to be 330 m/s in this condition. The large particles released from graphite appeared to rotate, rapidly, slowly sublimating and leaving away at a rather shallow angle.

Erosion mechanisms for different carbon based materials (graphite, carbon fiber composites (CFCs), Si-doped CFC) have been studied under brittle destruction under intense transient thermal loads (ELMs, plasma disruptions, VDEs) with respect to material erosion in different particle emission regimes, characterization of emitted particles, and behavior of preheated samples. Furthermore, the experimental data were compared with 3-D numerical simulation on the onset of brittle destruction. From a morphological point of view, the resulting erosion patterns on the test samples and ejected particles differ significantly for the three materials. The isotropic graphite shows a homogeneous erosion profile with flat craters, while the CFC forms no crater and only preferential erosion in localized spots in the PAN fiber area while the pitch fiber strands remain almost undamaged. The particles originating from graphite samples which have been collected on TEM grids are composed of nano sized amorphous carbon. CFCs have been the source for sub µm sized agglomerated fragments of crystalline carbon or silicon particles with ~50 nm diameter. Preheating of the test samples to 500 or 800°C results in a remarkable increase of the erosion depth and weight loss compared to the samples loaded at room temperature and identical heat fluxes. In particular, melting phenomena in the Si-doped CFC materials became essential at elevated temperatures.

In recent years, a series of multi-element doped graphite materials has been developed for PFC (plasma facing components) in China. Detailed investigation of the composition and full characterization of the microstructure of the doped graphite were carried out in the last several years and are now still in progress. These research activities were mainly concentrated on multi-element (B, Si, Ti) doped graphite and thick gradient SiC coatings. Investigations on the applicability of doped graphite as PFC under high heat flux, and further evaluation under HT-7 limiter plasma irradiation have been carefully investigated. From a 0.3 m² poloidal limiter to a 1.2 m² new belt limiter plus two poloidal limiters with actively water-cooled, this changes to the new carbon limiter in the HT-7 device turned out to be very successful. The plasma performance has been significantly improved and the longest discharge duration has been more than 60 s. These results have demonstrated that new carbon armoured PFCs will be an attractive choice making them competitive with other candidate PFC for the first wall of steady state fusion devices.

Infrared inspections of a CFC mock-up designed for ITER were performed using the IRINA test facility (Infrared Inspection for Nondestructive Analysis). The impact of local differences in emissivity of the CFC material on temperature measurements and the detection limit of the presented analysis was studied. The tested CFC mock-up showed an area with poor heat transfer properties. The same CFC mock-up was loaded under high heat flux in JUDITH. Here, overheating was observed in the area with poor heat transfer properties. Thus the infrared inspection can be seen as successful tool to verify the quality of CFC mock-ups. However, further studies are necessary to define the types of defects.

Graphite (with and without different tungsten layers) and CFC have been investigated in bending and tensile experiments accompanied by sound emission recording. Attempts have been made to excite surface acoustic waves (SAWs) with the aim to characterize the different materials as well as to test new diagnostic possibilities.

Graphite is an elastic and brittle material whereas CFC shows clearly quasi-plastic behaviour. Sound emission from evolving failures is easily detectable and in the case of CFC originates from the collective rupture of all fibres in a bundle. Young's modulus differs remarkably when it's deduced either from the stress-strain curves or from the sound velocity.

Fatigue experiments with CFC reveal a relaxation process in an early period characterized by numerous sound events, preferably produced by friction between the fibre bundles and the matrix, followed by a very calm period, finally leading to a rupture accompanied by massive sound emission again. The capability of sound measurements to diagnose such events is clearly demonstrated.

On graphite samples covered by a tungsten layer purposely excited Rayleigh-waves can be used to extract information about the surface. In a first attempt the dispersion relations of the SAWs observed on samples with W-layers of different thickness have been measured and compared with model calculations assuming the known thicknesses. The promising results reveal that the determination of thickness, Young's modulus, and density of a layer from experimental dispersion relations should be possible.

The influence of several graphitization parameters (temperature, dwell time, HIPing subsequent to graphitization) on the final properties of doped isotropic graphite has been investigated. The aim of this work is to obtain doped isotropic graphite with reduced chemical erosion by hydrogen bombardment, high thermal conductivity and large thermal shock resistance. As starting material, a self-sintering mesophase carbon powder and different metallic carbides (TiC, VC, ZrC and WC) as dopants has been used. Longer dwell time results in a remarkable increase of thermal conductivity, depending on the dopant and on the graphitization temperature. However, it leads also to carbide coarsening and local carbide agglomeration and thus to degradation of the mechanical properties. HIPing subsequent to graphitization leads to a significant reduction of porosity for the materials doped with VC and WC and thus to an improvement of their mechanical properties. A solid–liquid–solid model for metal catalysts can be applied to our experimental observations of graphitization in the presence of metallic carbides.

Ceramic materials produced on the basis of SiC and SiC/SiC composites are considered, due to their high-temperature strength, pseudo-ductile fracture behavior and low-induced radioactivity, as candidate materials for fusion reactors. The radiation resistance of ceramic materials under neutron irradiation is one of the key problems, which determines the use of these materials in fusion reactor environment. The dimensional stability (radiation swelling) of SiC is an important problem for degradation of radiation resistance of SiC materials. In fusion reactor environment helium atoms will be produced in SiC in the first wall region up to very high concentrations (15,000–20,000 at.ppm) and therefore it is very important to understand the helium effect on radiation swelling of SiC.

In the present paper the effect of helium on radiation swelling of SiC is investigated. Recent experimental results concerning the helium effect on radiation swelling of SiC under neutron, single-ion and dual-beam irradiation are presented. A new theoretical model is suggested for an explanation of the effect of helium on radiation swelling in ceramic materials. Point defects in ceramic materials can have an effective charge (e.g., an F⁺_centre, vacancy with a single trapped electron). The suggested theoretical model is based on kinetic consideration of charged point defect accumulation and kinetic growth of dislocation loops in a matrix of ceramic material (SiC) taking into account the effect of internal electric fields formed under irradiation near dislocation loops on diffusion processes of charged point defects. In this theoretical model, vacancies and small vacancy clusters are considered as additional traps for helium atoms. This results in enhanced growth rate of dislocation loops and finally swelling increase. The obtained numerical and analytical theoretical results for radiation swelling in the presence of helium atoms are compared with the existing experimental data for irradiated SiC material containing helium atoms. It has been shown that helium atoms increase the radiation swelling of SiC.

Carbon fibre composite (CFC) flat tile armours for actively cooled plasma facing components (PFC's) are an important challenge for controlled fusion machines. Flat tile concepts, water cooled by tubes, were studied, developed, tested and finally operated with success in Tore Supra. The components were designed for 10 MW/m² and mock-ups were successfully fatigue tested at 15 MW/m², 1000 cycles. For ITER, a tube-in-tile concept was developed and mock-ups sustained up to 25 MW/m² for 1000 cycles without failure. Recently flat tile armoured mock-ups cooled by a hypervapotron tube successfully sustained a cascade failure test under a mean heat flux of 10 MW/m² but with a doubling of the heat flux on some tiles to simulate missing tiles (500 cycles). This encouraging results lead to reconsider the limits for flat tile concept when cooled by hypervapotron (HV) tube. New tests are now scheduled to investigate these limits in regard to the ITER requirements. Experimental evidence of the concept could be gained in Tore Supra by installing a new limiter into the machine.

Transient thermography testing is an established method for testing the joint quality of CFC tiles in metallic heat sink systems. With this method hot pressurized water is introduced into the cooling channels of the elements and the transient heating of the CFC surface is recorded by an infrared camera. CFC tiles with a lack of bonding can be detected since they heat more slowly than well joined tiles. To perform this inspection a water flow in the tubes must be established and therefore the connection of the component to a hot water circuit is necessary. A new approach to this inspection problem is the application of lock-in thermography. With this method periodic illumination with powerful lamps of the CFC surface causes periodic heating of the CFC material. The phase shift between the periodic surface temperature evolution and the periodically operating heat source is recorded and evaluated. The lock-in thermography can be applied immediately after joining of the tiles to the component because the presence of a cooling channel system is not required for the application of this method. The method is suited to test the bonding of CFC tiles onto the component and the integrity of CFC tiles regarding possible internal cracks.

The divertor armour material for the tokamak ITER will probably be carbon manufactured as fibre composites (CFC) and tungsten as either brush-like structures or thin plates. Disruptive pulse loads where the heat deposition Q may reach 10² MJ/m² on a time scale τ of 3 ms, or operation in the ELMy H-mode at repetitive loads with Q ~ 3 MJ/m² and τ ~ 0.3 ms, deteriorate armour performance. This work surveys recent numerical and experimental investigations of erosion mechanisms at these off-normal regimes carried out at FZK, TRINITI, and IPP-Kharkov. The modelling uses the anisotropic thermomechanics code PEGASUS-3D for the simulation of CFC brittle destruction, the surface melt motion code MEMOS-1.5D for tungsten targets, and the radiation-magnetohydrodynamics code FOREV-2D for calculating the plasma impact and simulating the heat loads for the ITER regime. Experiments aimed at validating these codes are being carried out at the plasma gun facilities MK-200UG, QSPA-T, and QSPA-Kh50 which produce powerful streams of hydrogen plasma with Q = 10–30 MJ/m² and τ = 0.03–0.5 ms. Essential results are, for CFC targets, the experiments at high heat loads and the development of a local overheating model incorporated in PEGASUS-3D, and for the tungsten targets the analysis of evaporation- and melt motion erosion on the base of MEMOS-1.5D calculations for repetitive ELMs.

An analysis by means of numerical simulation of brittle destruction of carbon-based materials under experimental conditions at the electron beam test facility JUDITH is carried out. For the simulations a 3-dimensional model based on the destruction threshold enthalpy of 10 kJ/g is developed, in which a significant decrease of heat conductivity due to crack formation in CFC is taken into account. Calculated dependences of onset of brittle destruction as functions of absorbed power density as well as the depths of graphite erosion are in a rather good agreement with experimental results for isotropic fine grain graphite and carbon-fiber composite. The brittle destruction threshold of 10 kJ/g is confirmed for pulse loads lasting less than 100 ms, both for isotropic fine grain graphite and CFC.

The dependence of effective thermal diffusivity on temperature caused by volumetric cracks is modelled for macroscopic graphite samples using the three-dimensional thermomechanics code Pegasus-3D. At high off-normal heat loads typical of the divertor armour, thermostress due to the anisotropy of graphite grains is much larger than that due to the temperature gradient. Numerical simulation demonstrated that the volumetric crack density both in fine grain graphites and in the CFC matrix depends mainly on the local sample temperature, not on the temperature gradient. This allows to define an effective thermal diffusivity for graphite with cracks. The results obtained are used to explain intense cracking and particle release from carbon based materials under electron beam heat load. Decrease of graphite thermal diffusivity with increase of the crack density explains particle release mechanism in the experiments with CFC where a clear energy threshold for the onset of particle release has been observed in J. Linke et al, Fusion Eng. Design, in press, Bazyler et al, these proceedings. Surface temperature measurement is necessary to calibrate the Pegasus-3D code for simulation of ITER divertor armour brittle destruction.

A list of participants and their email addresses from the 10th International Workshop on Carbon Materials for Fusion Applications, held at Jülich, Germany on 17–19 September 2003.

There is little doubt that elemental carbon in all its allotropic forms has fundamental importance and countless applications in various traditional and modern technologies because properties of carbon-based materials can be tailored to demand. Fusion reactor technology is not an exception. The carrier of carbon as a material for the first wall components in controlled fusion devices began in the early eighties. This was due to low atomic mass and excellent power-handling capabilities of graphite and carbon fibre composites. It was soon—and not surprisingly—recognised that carbon was also an "enfant terrible" because of its "addiction" to hydrogen. All aspects of carbon properties and behaviour under fusion conditions have been widely studied and the series of International Workshops on Carbon Materials has become a discussion forum for experts from research institutions and industry. The initiative came from the Forschungszentrum Jülich and the first Workshop was organised in 1985. The next five symposia were also organised there in the period between 1986 to 1993. These were followed by meetings in Stockholm (1995), Jülich (1998) and Hohenkammer (2000). Until then, the Workshop was organised as a satellite meeting to a major conference related to fusion reactor materials and technology held somewhere in Europe. In 2003, for the first time, the meeting was a "detached" three-day symposium organised in co-operation with the EU Task Force on Plasma-Wall Interactions whose programme is vastly focused on appealing issues of carbon behaviour in fusion devices. A key objective for this meeting was to provide timely input for crucial decisions on the selection of first wall materials for ITER. Our community realises both big advantages and serious drawbacks of carbon. As a result, a strong discussion is in progress whether to use carbon or eliminate it and replace it by tungsten components. In this sense, the aim of the Workshop was not only to discuss hot topics but also to identify the most important research areas that need urgent resolution. The goal is also to achieve full integration of physics and technology. Therefore, the invited talks and contributed presentations dealt with aspects of fundamental processes, experimental findings, advanced modelling and technology of fusion reactor components. The presentations were given by 89 participants from research laboratories and industrial companies from various countries. They covered several major topics selected as the mainframe of the Workshop:

• Material erosion, migration and deposition

• Tritium retention

• Modelling and predictions for future machines

• Material performance

• Effects of neutron irradiation

The Workshop was held and financially supported by the Forschungszentrum Jülich. We would like to thank the Board of Directors very much for their encouragement and strong support. We are very grateful for the financial support received from the Deutsche Forschungsgemeinschaft (DFG). It allowed us to invite scientists from Russia. We are very thankful to the staff of the Forschungszentrum who helped with the organisation matters: Mrs Erika Wittig, Dr René Gail, Mrs Michaele Freisinger, Mrs Annemarie Winkens, Mrs Sabine Küper and Dipl. Math. Ulrich Stegelmann. And our most cordial thanks and gratitude are going to Mrs Angelika Hallmanns for all her kindness and efficiency in the organisation that helped all of us to enjoy the meeting. We thank all participants for their contributions and we thank the referees of submitted papers. Thank you all for your hard work and co-operation. We are looking forward to the next meeting. However, when and where it will be held depends on the wish of our community when we feel there is substantial new material to be discussed.