Carbon chemical erosion in high flux and low temperature hydrogen plasma

How long will the wall of a fusion reactor last: hours (i.e. much too short) or years? This question is investigated for the conditions foreseen in the fusion reactor ITER, the world’s joint experiment in the development of nuclear fusion as a clean, safe and inexhaustible energy source. In ITER, hot hydrogen plasma is confined by magnetic fields. Whereas the temperature in the center of the reactor is extremely high (200 million ??C), it is much lower where the plasma touches the wall. The contact area is situated in the so-called divertor. Here, the plasma temperature is a mere 1 eV (ten thousand ??C), but the flux density of hydrogen ions is very high, up to 1024 m-2 s-1, and the power flux can exceed 10 MW m-2. As a result of these extreme flux densities, the resilience of the material at the contact area is a critical factor for the success of ITER, and indeed, fusion energy. Carbon is presently the material selected for part of the ITER divertor. The aim of this thesis is to measure its chemical erosion rate in ITER-like conditions. The experiments were carried out in the linear plasma generator Pilot-PSI at the FOM Institute for Plasma Physics Rijnhuizen. In this device a ??1 cm diameter hydrogen plasma beam is produced, which is guided to a cooled target by a magnetic field of up to 1.6 T. The plasma density ne and temperature Te at the target can be varied in the range 1019 <ne <1021 m-3 and 0.1 <Te <4 eV. Like in ITER, Te is so low that no physical sputtering occurs, which would invalidate the experiment. Flux densities of more than 1025 hydrogen ions m-2 s-1, and tens of MW m-2 can be realized. Hence, the parameter regime accessible with Pilot-PSI fully covers the conditions relevant to the ITER divertor. Carbon targets were systematically exposed to scans of the plasma flux density at different plasma temperatures in Pilot-PSI. After a comparison of different carbon composites, which turned out to behave similarly, we selected fine grain graphite as the test material for our studies. Two methods were used to measure the erosion. Firstly, a spectroscopic measurement was used to determine the "gross" erosion, i.e. the instantaneous flux of carbon that leaves the surface in the form of hydrocarbon molecules. This method, which in present fusion experiments is commonly used at higher plasma temperatures, was extended to the temperature range studied here (Te ?? 1 eV). A different excitation process was shown to prevail: dissociative recombination of hydrocarbon ions rather than electron impact excitation. The latter is dominant at Te > 3 eV. Secondly, the "net" erosion, i.e. the net amount of material lost from an entire exposure, was determined by postmortem surface profilometry. The net erosion can differ substantially from the gross erosion because part of the eroded material can be redeposited. It was observed that redeposition occurs primarily in conditions with plasma temperatures Te > 0.5 eV. This can be understood as being the result of the ionization and subsequent magnetic trapping of the hydrocarbon molecules coming off the surface. At lower plasma temperatures (Te <0.5 eV) very little redeposition was observed. In those cases there was good correspondence between the gross erosion as measured with spectroscopy and the postmortem net erosion measurements. Having established and calibrated the methodology, the gross and net erosion of fine grain graphite was systematically assessed, for plasma flux densities GH varied from 1 · 1023 m-2 s-1 to 4 · 1024 m-2 s-1, at different plasma temperatures. The results for the gross erosion yield Ychem, i.e. the number of eroded carbon atoms per incoming hydrogen ion, can be summarized as follows: • The chemical erosion yield increases strongly with increasing Te: Ychem ?? 0.6% C/H for Te <0.5 eV, and ??6% C/H for Te > 1.0 eV, measured at GH = 1 · 1024 m-2 s-1. For Te > 1.0 eV no further increase of Ychem was observed. The Te dependence may in part be due to the surface temperature of the target. In the experiments with Te > 1.0 eV the surface temperature was around 500 ??C, at which the chemical erosion is at its maximum. For Te <0.5 eV the surface was cooler (200–300 ??C). • Both at Te ?? 0.5 eV and Te > 1.0 eV, Ychem is inversely proportional to the plasma flux density: G-1 H . This implies that the gross erosion rate – the amount of material eroded per second – is independent of the flux density. Thus, the central conclusion is that the gross erosion of carbon at the flux densities and temperature ranges expected in ITER is saturated: it cannot go any faster. A likely explanation is that the hydrogenation of the carbon at the surface is rate-limiting; the time it takes for a hydrocarbon molecule to form and leave the surface is longer than the average consecutive ion impact time (??10-4 s at a plasma flux density of 1 · 1023 m-2 s-1). Using this information to calculate the lifetime of a carbon wall in the ITER divertor we find for Te <0.5 eV a rate of ??6 · 1021 carbon atoms m-2 s-1 (??65 nm s-1). This measurement is not affected by redeposition since the gross erosion is equal to the net erosion, but it is an underestimation as the surface temperature is low. For Te > 1.0 eV the erosion rate is ??6 · 1022 carbon atoms m-2 s-1 (??0.65 µm s-1). This measurement is probably an upper estimate due to the recycling of carbon. It is therefore very likely that the effective erosion rate will be in between these two limiting estimates. This results in an estimated lifetime of 4 to 40 hours (corresponding to 40 to 400 discharges) for the ITER divertor if it is made of carbon. This implies that the carbon divertor is likely to be a limiting factor for ITER operation. This conclusion is based on a consideration of the gross erosion rate alone; other complicating factors, such as the retention of hydrogen and the formation of dust, have not been considered. The fact that the gross erosion rate is saturated indicates that there is little scope for measures that reduce this erosion rate. The redeposition of carbon does effectively reduce the net erosion in our experiment. The extent of this reduction will depend on material migration and the geometry of the divertor, and these effects have not been assessed in this thesis