Study of physical and chemical assisted physical sputtering of beryllium in the JET ITER-like wall

The effective sputtering yield of Be $(Y_{\rm Be}^{\rm tot})$ was determined in situ by emission spectroscopy of low ionizing Be as function of the deuteron impact energy (Ein = 25–175 eV) and Be surface temperature (Tsurf = 200 °C–520 °C) in limiter discharges carried out in the JET tokamak. Be self sputtering dominates the erosion at high impact energies (Ein > 150 eV) and causes $Y_{\rm Be}^{\rm tot}$ far beyond 1. $Y_{\rm Be}^{\rm tot}$ drops to low values, below 4.5%, at the accessible lowest impact energy (Ein sime 25 eV) achievable in limiter configuration. At medium impact energies, Ein = 75 eV, two contributors to the measured $Y_{\rm Be}^{\rm tot}$ of 9% were identified: two third of the eroded Be originates from bare physical sputtering $(Y_{\rm Be}^{\rm phys})$ and one third from chemical assisted physical sputtering $(Y_{\rm Be}^{\rm chem})$ . The later mechanism has been clearly identified by the appearance of BeD A–X emission and quantified in cause of a temperature dependence at which the BeD practically vanishes at highest observed Be limiter temperatures. The recorded Tsurf dependence, obtained in a series of 34 identical discharges with ratch-up of Tsurf by plasma impact and inertial cooling after the discharge, revealed that the reduction of BeD is correlated with an increase of D2 emission. The release mechanism of deuterium in the Be interaction layer is exchanged under otherwise constant recycling flux conditions at the limiter.The reduction of $Y_{\rm Be}^{\rm chem}$ with Tsurf is also correlated to the reduction of the Be content in the core plasma providing information on the total source strength and Be screening. The chemical assisted physical sputtering, always present at the nominal limiter pre-heating temperature of Tsurf = 200 °C, is associated with an additional sputtering channel with respect to ordinary physical sputtering which is surface temperature independent. These JET experiments in limiter configuration are used to benchmark the ERO code and verify ITER first wall erosion prediction. The ERO code overestimates the observed Be sputtering in JET by a factor of about 2.5 which can be transferred to ITER predictions and prolong the expected lifetime of first wall elements