Global migration of carbon impurities in the ASDEX Upgrade tokamak Euratom-Tekes Fusion Seminar Tartu, May, 2012 Antti Hakola VTT Technical Research Centre of Finland Collaborators: S. Koivuranta, J. Likonen: VTT M. Groth, T. Kurki-Suonio, V. Lindholm, T. Makkonen, J. Miettunen: Aalto University A. Herrmann, K. Krieger, M. Mayer, H. W. Müller, R. Neu, V. Rohde: IPP-Garching P. Petersson: KTH

Material migration: why is it important? Material migration is important because it is net, rather than gross, erosion which is of practical consequence (P. C. Stangeby) In other words, necessary step between erosion of plasma-facing components and deposition of the eroded material & retention of plasma fuel (particularly T) in fusion reactors Carry out tracer injection experiments (e.g. 13 C) just before venting the vessel for maintenance Analyze a comprehensive set of first-wall components for their surface densities of the tracer elements Try to obtain – and predict – the resulting deposition profiles numerically How can the migration mechanisms in tokamaks be elucidated?

Migration can be studied both globally and locally: Here the focus in on the global scale isotopically labelled methane ( 13 CH 4 ) (and recently also 15 N 2 ) injected into the torus from one valve at the outer midplane results modelled using the ASCOT, DIVIMP, and SOLPS codes This presentation concentrates on migration studies in ASDEX Upgrade (AUG) AUG is an ITER- and DEMO-relevant environment for migration studies all the first-wall structures W- coated graphite tiles since 2007

CampaignDischarges/ configuratio n GasB t (T)I p (MA)n e (×10 19 cm -3 ) P aux (MW) 13 C/ 15 N injected (×10 22 at.) Injection rate (×10 21 s -1 ) Flat-top time (s) L-mode, LSN H (NBI) 0.9 (ECRH) (mixture) L-mode, LSN D (NBI,< 1 s) 0.9 (ECRH) L-mode, LSN H H-mode, USN H H-mode, LSN H See A. Hakola et al., Plasma Phys. Control. Fusion 52, p , 2010 A. Hakola et al., Journal of Nuclear Materials (submitted) Master table of the AUG global 13 C/ 15 N experiments

The latest injection experiment took place in 2011: global injection of both 13 C and 15 N High-density, lower single-null L-mode discharges in hydrogen Altogether 11 shots (#27382-#27392), two reference shots (#27366, #27371) 13 C and 15 N injected simultaneously, atomic ratio of 13 C and 15 N was 1:1 Midplane manipulator in operation during the reference shots T e, n e, and flow velocities at the outer midplane Additional data for modelling purposes: T e and n e distributions at the divertor from Balmer emission lines j sat, T e, and n e data from fixed Langmuir probes (LP) at the strike-point zones thermography data at the divertor plates (temperature and power density) bolometer data to estimate radiated power in different regions

keV O 2 + primary ion beam, current 500 nA, analysis area 300×430 m 2 Tiles analyzed using SIMS at VTT – first extensive analysis program after the 2007 experiment SIMS = Secondary Ion Mass Spectrometry Si samples In 2007, 13 C distribution measured from selected W-coated graphite tiles (3-5 m or 200 m) uncoated graphite regions of marker tiles (divertor) small Si samples (remote areas)

RegionDeposition Inner divertor0.8% Roof baffle0.1% Outer divertor0.1% Limiter region0.3% PSL0.04% Upper divertor1 - 6% Heat shield0.6% Remote areas1.3% Only 4 - 9% of injected 13 C found experimentally Why to bother anymore: everything was clear after the 2007 experiment!? Totally different deposition behavior on W and on C, especially at the outer divertor Is the applied assumption of toroidally symmetric deposition valid? Well, not really…

Predictive ASCOT modelling indicates strong deposition hot spots at certain toroidal locations Is this really the case? Must be checked experimentally!

wall tiles removed for post mortem analyses after the 2011 injection experiment Tiles removed for analyses marked in red Especially: tiles removed from many different toroidal locations at the outer midplane: two ICRH antennas, two different poloidal limiters samples taken from the side faces of the removed tiles deposition in tile gaps! 17 mm Gap samples Plasma-facing samples Tile A1/2RIGHT

Results (1): remarkable deposition at the outer midplane largest surface densities (up to at/cm 2 ) localized to the vicinity of the injection valve deposition decays from the peak values to at/cm 2 within 100 mm large differences in 13 C levels between tiles from different ICRH antennas and poloidal limiters toroidally symmetric deposition not a valid assumption

Results (2): deposition profiles in different regions of the torus UD: rather uniform deposition, results in line with the 2007 profile HS: situation dramatically changed from 2007 to 2011 plasma flows different? lower divertor minor deposition region OD: deposition profiles in 2011 qualitatively different from the 2007 case due to different magnetic configuration? or due to eroded tile surfaces? tile gaps account for considerable 13 C surface densities Deposition at (a)upper divertor (UD) (b)heat shield (HS) (c)inner divertor (ID), roof baffle (RB), and outer divertor (OD) (d)tile gaps

Results (2): deposition profiles in different regions of the torus

Approximately 35% of the injected 13 C found Inner divertorRoof baffleOuter divertorUpper divertorHeat shieldOuter midplane 1.5% (4% of 13 C found) 0.3% (1% of 13 C found) 0.4% (1% of 13 C found ) 4% (11% of 13 C found) 15% (41.5% of 13 C found) 15% (41.5% of 13 C found) toroidally symmetric deposition assumed for the inner divertor, roof baffle, outer divertor, upper divertor, and heat shield at the outer midplane, the average 13 C surface density multiplied the total surface area of the different limiter and ICRH antenna tiles Main chamber is the main deposition region for 13 C

SOLPS modelling: range of background plasmas and flow profiles obtained SOLPS = 2D plasma fluid code shot #27385 with pure H plasma, n e,sep as a free parameter decent match for n e and T e at the OMP and OD – but not simultaneously background plasma corresponding to a fit for n e at OMP selected for ASCOT simulations (top) SOLPS predicts weak plasma flow and stagnation point at the OMP in contradiction with typical situations in tokamaks n e,sep = 2.25 × m -3 n e,sep = 1.5 × m -3 ASCOT

ASCOT modelling: imposed flow profile required ASCOT = 3D Monte Carlo code shot #27385, magnetic equilibrium at 2.8 s test particles (300,000) followed until their deposition ASCOT predicts strong, localized deposition at OMP imposed flow profile required to shift deposition away from outer divertor SOLPS flow profile predicts 50% at the outer divertor (experimentally: 1% of the 13 C found)

Conclusions Global 13 C and 15 N injection experiment successfully carried out in AUG in 2011 A comprehensive set of toroidally and poloidally distributed tiles analyzed for their 13 C content Experimental highlights: Main chamber of AUG is the most prominent sink for 13 C: almost 35% of the injected 13 C found there Gaps between tiles contain significant 13 C inventories Lower divertor is a minor deposition region for 13 C Status of numerical simulations SOLPS simulations provided a set of background plasmas and poloidal flow profiles but flows generally rather weak and stagnation points occur at wrong places ASCOT simulations with the weak SOLPS plasma flow would deposit 50% of the particles at the outer divertor imposed flow profile required to reproduce the observed localized deposition peaks at the outer midplane