ANNUAL PROGRESS REPORT 2003

Transcription

1 Dok. EURFZJ 976 Nuclear Fusion Project Association EURATOM / Forschungszentrum Jülich ANNUAL PROGRESS REPORT 2003 including the contributions of the TEC Partners ERM/KMS Brussels and FOM Nieuwegein and the IEA Partners Forschungszentrum Jülich GmbH August 2004

2 /RPS 2

3 CONTENTS PAGE A. Introduction... 5 A.1. Nuclear Fusion and Plasma Research... 7 B. General Programme on TEXTOR... 9 B.1. Main Topic I Plasma Wall Interaction... 9 B.2. Main Topic II Confinement B.3. Main Topic III Impurity Transport and Radiation B.4. Main Topic IV Magnetohydrodynamics B.5. Main Topic V Advanced Tokamak Scenarios B.6. Main Topic VI Dynamic Ergodic Divertor (DED) B.7. Main Topic VII Theory and Modelling B.8. Operation and further development of TEXTOR Commissioning of TEXTOR with the Dynamic Ergodic Divertor (DED) Data Acquisition and Processing B.9. Plasma Diagnostics B.10. Contributions to ITER B.11. Contributions to Wendelstein 7-X B.12. Characterization of Materials and Components for Plasma/Wall Interaction C. Technology Programme C.1. Characterization of Materials and Components for Plasma/Wall Interaction C.2. Corrosion Resistance of Fusion relevant C-based Material C.3. Mechanical Properties of Fusion Materials D. Partners of the IEA TEXTOR Implementing Agreement D.1. Japan D.2. Canada D.3. United States of America E. Summary on results of the main projects in the framework of "Projects for enhancing the mutual co-operation between Associations" F. Structure of the Fusion Programme and Related Figures G. Scientific Publications, Talks and Posters

4 4

5 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ A. INTRODUCTION r.wolf@fz-juelich.de u.samm@fz-juelich.de The Institute takes part in international fusion research with the long-term aim of imitating on earth the sun's method of producing energy and thus harnessing a practically inexhaustible energy source with favourable safety and environmental features for mankind. The progress achieved in recent years at fusion devices provides a solid data base today for extrapolation to a fusion machine with tenfold power gain. This decisive step is to be made by implementing the ITER experiment planned in international cooperation, which will furnish a fusion power of 500 MW for a burning time of approx. 8 minutes per plasma pulse and will be the last intermediate stage prior to the construction of a continuously operating demonstration power plant (DEMO). The Institute's research programme is oriented to the strategy of the European research programme (EURATOM/EFDA), which pursues four parallel lines: a) the implementation of ITER in global cooperation, b) an ITER-accompanying research programme at smaller devices, c) the development of the necessary fusion technologies for DEMO and d) further research into alternative confinement concepts. For the research programme accompanying ITER the TEXTOR tokamak is available in Jülich. In 1996, the EURATOM-associated fusion laboratories in the three-frontier region, Institute of Plasma Physics of Research Centre Jülich (D), Instituut voor Plasmafysica Rijnhuizen of FOM (NL) and Laboratoire de Physique des Plasmas of ERM/KMS Brussels (B) have joined forces forming the Trilateral Euregio Cluster (TEC) with the aim of carrying out a joint research programme at the large central TEXTOR device. TEC allows resources to be combined (e.g. the radiofrequency heating schemes are provided and operated by the TEC partners), favourably combines different expertises complementing each other and provides a centre of attraction for the universities in the region. The Institute additionally cooperates in the use of TEXTOR with Japan, the USA and Canada under an IEA Implementing Agreement. Apart from TEXTOR, experimental facilities outside Jülich are also used to an increasing extent. This includes above all the JET tokamak used under the European Fusion Development Agreement (EFDA). At the national level, the Max Planck Institute of Plasma Physics, Garching, Research Centre Karlsruhe and Research Centre Jülich have joined as Helmholtz centres in the Nuclear Fusion Development Association to coordinate their work. Within Research Centre Jülich all fusion-relevant activities at the institutes are coordinated by the Nuclear Fusion Project. In order to proceed from ITER to DEMO, the continuous operation of a fusion reactor must be implemented. To this end, it is above all necessary to achieve a sufficient lifetime of the wall components under strong load. The TEXTOR tokamak experiment will contribute in the years to come with the Dynamic Ergodic Divertor (DED) pioneering experiment towards exploring the fundamental possibilities of reducing wall exposure with the aid of external electromagnetic perturbation fields. Moreover, the basic concept of TEXTOR with in part unique provisions for experiments allows detailed research into fundamental processes, so that more reliable models for predicting the lifetime of wall components can 5

6 be made e.g. by a better understanding of the plasma-wall interaction. In this field, close cooperation also takes place with the material-oriented investigations performed at IWV-2. For the planning and construction of ITER, the European associations will have to furnish their contributions in accordance with existing expertise. The Institute of Plasma Physics aims at tackling problems from the fields of plasma diagnostics and plasma heating. Due to its inherently steady-state plasma operation, the stellarator is considered to be the most promising alternative to the tokamak. With the Wendelstein 7-X stellarator in Greifswald to become operational in about 2010 Germany will have a worldwide leading experiment in this field. The Institute of Plasma Physics will contribute to the construction and scientific use of the new stellarator by solving electrotechnical problems and developing and providing diagnostics. 6

7 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ A. INTRODUCTION A.1. NUCLEAR FUSION AND PLASMA RESEARCH r.wolf@fz-juelich.de Tasks and Objectives The institute participates in the further development of magnetic confinement concepts for the realization of nuclear fusion as a new primary energy source. The central facility is the tokamak TEXTOR, which is operated in collaboration with the partners from the Trilateral Euregio Cluster. With its highly flexible instrumentation TEXTOR is oriented towards the investigation of fundamental processes in fusion plasmas. Main element of reporting year 2003 was the commissioning of the newly installed Dynamic Ergodic Divertor (DED). This unique experiment serves, amongst other things, for the improvement of the heat load distribution on the wall of the plasma vessel and for the control of plasma confinement and stability. Additional emphasis is focussed on the participation in the scientific programme of the European fusion experiment JET, the construction and subsequent utilization of the stellarator Wendelstein 7-X, and on the planning of the next step tokamak experiment ITER, which for the first time shall demonstrate a burning fusion plasma. Main Achievements The DED was successfully taken into operation with direct and alternating currents up to a value of 7 ka and with frequencies up to 7 khz, producing rotating magnetic fields in the dynamic mode. With the DED, a novel and unique tool has been provided for dynamically influencing the magnetic field topology by ergodisation with external coils. In combination with specialised diagnostic methods already installed at TEXTOR, new chances and potentials for basic studies have thus been opened, addressing the enhanced optimisation of the heat and energy transport as well as the stability of magnetic confinement. The relevance of ergodic structures has also been shown during the reporting year by means of collaborative experiments at the DIII-D tokamak in the USA. Here, already by static ergodisation in a mode of operation also being planned for ITER, negative effects due to boundary instabilities (so-called ELMs or edge localized modes) could be reduced considerably. Now the challenge is to achieve a better understanding of the basic mechanisms and as a result to contribute to an improved operation of ITER. The set-up of plasma diagnostic systems still was of major importance during the reporting year. First measurements in combination with the DED confirmed predictions of the heat load distribution on wall elements caused by the magnetic field perturbation. Beside the expected smearing and the subsequent reduction of the heat load, a manifold of modifications of the magnetohydrodynamic features of the plasma could be observed even though this is not understood yet. The development of a 3-dimensional hydrodynamic Monte Carlo code for the description of plasma transport in the complex magnetic field geometry of the DED was completed. 7

8 In collaboration with a group at the University of Marseille and the Max-Planck-Institut für Plasmaphysik first attempts have been made to incorporate the DED into turbulence models. For the further development of the ITER reference scenario, the H-mode, comprehensive studies have been conducted at JET. In the field of plasma stability the institute made important contributions to the control of neoclassical tearing modes, which in ITER would result in undesirable deterioration of the plasma confinement, and in particular to the understanding of the ELM physics. Here, new properties were discovered which are not in agreement with the presently accepted explanation of the ELM cycle. Based on earlier TEXTOR results, a new feedback control scheme was developed for the quasi-stationary sustainment of JET discharges at high density. However, theoretical modelling of these plasma conditions (with the RITM code) shows that in JET the improvement of confinement does not take place, which in TEXTOR was observed under similar conditions. In order to achieve a sufficient life time for materials, which are in direct contact with the plasma, for their later use in ITER or a fusion reactor, extensive studies of the plasma wall interaction are necessary. To this end, in JET and TEXTOR the carbon erosion and re-deposition has been investigated in-situ. Numeric models such as the ERO code play an important role for the interpretation of the experimental data. The observed carbon balances can only be understood in case an unexpectedly high erosion rate is assumed for the freshly deposited layers. The mechanism still is unclear up to now. Furthermore, the review and improvement of the description of atomic and molecular processes play an important role. An important spin-off has resulted from a unique cooperation between the IPP, the University of Düsseldorf and Philips Lightning. In this project, the similarities of the physical properties in the ITER edge plasma and gas discharge lamps are utilized. Measurements with gas discharge lamps serve as a validation of the Monte Carlo code EIRENE, developed in Jülich, which subsequently will be used to describe parts of the ITER edge plasma. Emphasis is also put on the development of plasma diagnostics and heating methods, also for their later use in Wendelstein 7-X or ITER. Worth mentioning are the further development of vacuum- UV spectroscopy, charge exchange recombination spectroscopy and motional Stark effect diagnostics. The construction of a combined electron cyclotron emission imaging and microwave imaging reflectometer system has been completed. A new 140 GHz gyrotron with 3 s pulse length and a power of 800 kw for heating the plasma electrons successfully went into operation thus enriching the spectrum of heating methods available at TEXTOR with a flexible new instrument. For the characterization and test for materials for their use in future devices, such as Wendelstein 7- X and ITER, the electron beam test facility JUDITH was employed, providing thermal loads of up to 20 MW/m 2. Various tests included the study of thermal expansion behaviour of graded materials, dust generation at intense transient heat loads, neutron induced degradation of the thermal expansion coefficients and different target designs after neutron irradiation with fluences of up to 1 dpa. The institute will also take over extensive work packages for the construction of the stellarator Wendelstein 7-X, consisting of the development of diagnostic systems, tasks concerning the design and construction of components for the superconducting coils and bus bar system, the electrical joints as well as supporting tasks in welding technology. 8

9 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.1. PLASMA WALL INTERACTION v.philipps@fz-juelich.de Plasma Wall Interaction affects the energy release and fuel dilution in the plasma core by impurities released from the walls, the lifetime of wall components by erosion and the long term retention of the fuel gas in the walls. This year, the activities were concentrated on the analysis of both deuterium and tritium recycling, hydrocarbon signal evaluation and radical formation both in the JET divertor (within the Task Force E exhaust and edge physics) and in front of carbon test limiters in TEXTOR. Additionally, carbon migration was studied in JET using a shot resolved deposition monitor in the inner divertor (QMB). Analysis of the effect of erosion and carbon deposition on mirror properties was started in TEXTOR and during in situ laser desorption/ablation with the aim to develop an in situ diagnostic for material deposition and fuel retention in ITER. Hydrogen recycling properties A way of studying the properties of released fuel components is the determination of their velocity distribution and atomic and molecular composition. Isotope exchange experiments are helpful to identify the local and global fuel recycling in more detail. Fulcher-band spectroscopy can support these measurements by measuring the variation of the molecular composition of the hydrogen isotopomeres in front of PFCs. This could be shown in TEXTOR (see report 2002) and has been extended to tritium containing compounds in JET. Fig. 1 shows the identified T 2 molecular lines detected in front of the T 2 gas inlet valve. Hydrocarbon formation The release of carbon has been intensively studied both in the JET divertor and in front of carbon test limiters in TEXTOR. At high surface temperatures (around C) carbon will be strongly released in form of C 2, which may both stem from higher order hydrocarbons and direct molecular release. Fig. 2 shows such a molecular release around the strike point region in the JET divertor. Because of the complicated dissociation chain and the restricted diagnostic methods of passive spectroscopy it is vital to study the emission pattern of the accessible C 2 and CD (+) emission both in identical discharges and as detailed as possible. Therefore, in TEXTOR gas blow experiments have been performed both through nozzles and extended compact graphite limiters. Fig. 3 displays a spectrum under these conditions obtained with a highly resolving Echelle spectrometer (see report 2002), which reveals all of the spectrally interesting regions simultaneously. 9

10 Fig 1: Molecular plasma mixture during T 2 gas blow experiments in JET. Fig 2: Molecular C 2 ablation at the inner JET divertor. 10

11 Fig 3: Molecular spectra during a CD 4 puff through a nozzle into a TEXTOR boundary plasma. Atomic and molecular data For the interpretation of the measured spectral line intensities in terms of fluxes and densities the conversion factor S/XB (ionisations per photon) has to be known. Various tools exist for the determination of these values codes like GKU, R-matrix, databases (ADAS) etc. TEXTOR offers also the possibility of comparing these theoretical data with experimentally found results. S/XB values for B I & B II (Fig. 4) have been obtained by B(CH 3 ) 3 puffing and are compared with code calculations. Important tools (codes ATImpactParameterMethod and ATCloseCoupling) for the calculation of atomic data were developed. They extend the package ATOM developed earlier. They were tested on the example of the 3s-3p-3d system in He I. Further development of these codes are the inclusion of other elementary processes and the extension of the region of application. The corresponding D/XB values for the methane family have also been tested in TEXTOR and JET and show the complexity of a transfer of the values derived from different fusion edge plasma configurations. Erosion, deposition and fuel inventory Material deposition has been studied in the inner divertor of JET using the quartz microbalance technique. The most prominent parameter determined the material deposition on the measuring position is the plasma geometry in the divertor. With the strike point on the vertical target and the QMB in the private flux region the deposition is small (typically nm/shot) but increases with moving the strike point in the direction to the QMB. Much more deposition is measured with the strike point on the horizontal target and the QMB in the SOL. A history effect has been observed showing large deposition during the first discharge with the plasma strike point touching a fresh area. The rate strongly decreases in successive shots keeping the strike point fixed. Therefore the carbon deposition pattern found after long term plasma operation is the integral result of a many different plasma configurations. 11

12 Fig. 4: Theoretical S/XB values for B II obtained by the code GKU from the P.N. Lebedev Institute, Moscow; experimental IPP D/XB values for transition 1 are a factor of 2 to 3 smaller! Performance of mirrors after plasma exposure The performance of optical mirrors from polished polycrystalline molybdenum has been analysed after exposure in the SOL of TEXTOR in erosion and C-deposition dominated regions. The optical reflectivity has been measured before exposure in a wide wave length range and polarisation angle followed by post mortem analysis. A strong degradation on the deposition areas is observed due to carbon deposition, while the degradation is smaller on the erosion areas. In particular areas the Mo surface was molten leading to strong grain grow and an enhanced optical reflectivity. Another set of mirrors has been mounted for long term exposure in a periscope system with controlled temperature and exposed to the SOL plasma for about 500 plasma seconds. A first analysis indicates that in this special geometry the carbon deposition is reduced in plasma shadowed regions of the periscope. Fig. 5: Appearance of a Mo optical mirror after exposure to the SOL plasma of TEXTOR. 12

13 Laser induced desorption The hydrogen content on a test limiter surface during plasma pulses of TEXTOR has been determined in situ by the laser desorption technique. A single pulse ruby laser operated in a free generation mode with a maximum energy of 15 J and a pulse duration of about 0.3 ms was used with a spot size on the limiter surface of about 8 mm. The temperature of the limiter surface in the centre of the laser spot was measured by a fast optical pyrometer(15 µs) and the light emitted by thermally desorbed particles in the plasma was detected by means of a five channel polychromator with a temporal response of about 5 µs. For a graphite limiter the surface temperature in the laser spot during laser pulse increases up to about 800 C. The increase of Balmer line emission during surface heating exceeds the background level caused by recycling by one order of magnitude. The total number of desorbed hydrogen and deuterium atoms was deduced from the number of the H α and H β photons emitted during the laser pulse using the S/XB coefficients (photon per ionisation) from the ADAS database. Values corresponding to hydrogen surface concentrations in the range of (H+D)/cm 2 are found for different experimental conditions. Similar experiments have been performed with the edge Lidar laser system on JET using a long laser pulse of about 0.3 ms. Desorbed hydrogen and carbon has been detected using the fast KS3 spectrometer during the limiter phase of the plasma before formation of the X point. Hydrogen and C II signals are clearly correlated with plasma loading of the desorbed surface but the amount of desorbed particles saturates quickly. This indicates that under the low laser power conditions in JET (18 kw/cm 2 ) only a shallow surface layer is depleted by the laser pulse. Ne2*10 12 cm -3 S/XB Te50 ev CII(514nm) 6.97 Hβ x10 15 C H+D 2x x shot Fig. 6: Amount of hydrogen and carbon evaluated from C II and Η β light released by means of laser desorption from a graphite test limiter. Modelling Modelling of the transport of 13 CH 4 molecules injected from a declined test limiter has been continued using newest available data for rate coefficients of the CH 4 reaction chain and an enhanced chemical erosion (Y e ) of re-deposited carbon layers compared to the erosion of substrate graphite. An enhanced chemical erosion yield for re-deposits of about Y e 8% by the background hydrogen the hydrogen returning with the hydrocarbons is needed. The modelling therefore supports a picture of long-range carbon transport where carbon can be re-eroded much more effectively after re- 13

14 deposition on plasma-wetted areas which can explain the carbon transport to shadowed areas in JET and TEXTOR. Modelling of carbon erosion and re-deposition at the TEXTOR ALT shows a qualitative agreement with erosion-deposition patterns after several months of plasma operation. First calculations of the beryllium and carbon transport in the linear plasma simulator PISCES-B have been done with a main emphasis on the observed Be emission profile. Predictive calculations of tritium retention and target lifetime in ITER have been performed taking into account a revised formula for chemical erosion (including surface temperature, energy and flux dependencies) and atomic hydrogen flux as additional eroding species. 14

15 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.2. CONFINEMENT b.unterberg@fz-juelich.de Plasmas with a radiating plasma mantle to allow for acceptable power exhaust have been a main subject of investigation. While no experiments were conducted on the tokamak TEXTOR in Jülich because of the shut down to install the Dynamic Ergodic Divertor, extensive work has been done on JET in Abingdon, UK. To understand the impact of the radiating impurities on transport in the plasma core and the global confinement properties, transport modelling has been an important element in 2003 as described in this report. Experiments and modelling on radiating plasmas at JET under EFDA The experiments at JET, the world largest tokamak device, have been jointly conducted under the European Fusion Development Agreement (EFDA) by the various EURATOM-Associations. The Trilateral Euregio Cluster (TEC) is participating in these activities. One of the main subjects is the exploration of plasma regimes, where the power is exhausted from the plasma in form of radiation onto large wall areas by radiation of impurities. Thereby, excessive heat loads on plasma facing components can be reduced at locations where the magnetic field directs charged particles on relatively small areas. Two aspects have been especially investigated during last years campaign: the development of a radiative scenario at high density with good confinement which is realised by means of a sophisticated feedback system, and the investigation of the impact of the seeded impurities on turbulence driven transport in the confined plasma. For the latter the comparison to results obtained on the medium sized TEXTOR device is of special importance. The plasma regime chosen for the impurity seeding experiments has been the so-called High confinement mode (H-mode) where a transport barrier at the edge is formed and energy and particles are released in periodic, burst-like events (so called edge localised modes, ELMs). The ELMy H- mode is the reference scenario presently foreseen for the next step fusion experiment ITER. The seeding of argon into such kind of plasma discharges allows to establish high energy confinement at high plasma densities (close to the operational density limit of tokamak discharges, the socalled Greenwald density) in combination with a radiating plasma mantle. Previous experiments have shown, that the energy confinement at the high plasma density is very sensitive to the external gas fuelling, a characteristic which is also seen in discharges with high energy confinement and high density in TEXTOR as described later on. This experience has led to the development of a dual feedback system for JET, where at the same time the fraction of the radiated power with respect to the total input power is controlled by the external argon fuelling rate and the energy confinement time expressed by the enhancement factor with respect to the H-mode scaling law H98(y,2) is controlled by the deuterium fuelling rate. Once the energy confinement starts to degrade, the D2-15

16 fuelling rate is reduced and vice versa. In this way the plasma density is determined by the effective particle confinement time at the given gas flow. Discharges with a radiating mantle are characterised by improved particle confinement with respect to un-seeded reference pulses, which explains the possibility to obtain highest plasma densities in excess of the empirical Greenwald density limit in the presence of the radiating impurities. Fig. 1 shows an example of a high triangularity discharge in the JET tokamak with dual feedback on the radiated power fraction and the H98 factor. During the phase, where the feedback is active, simultaneously high density (density normalised to the empirical Greenwald density limit f Gwd 1.15), good H-mode performance (energy confinement time normalised to the ELMy H-mode scaling, H98(y,2)1) and a radiated power level f rad P rad /P tot 60% could be realised under quasistationary conditions. # MA, 2.7T Feedback Time (s) Fig. 1: Time traces of global plasma parameters in a JET discharge with dual feedback control of the radiated power fraction and the normalized energy confinement time. In parallel to the development of the experimental discharge scenario with the feedback scheme mentioned above much effort has been spent to improve the modelling tools which allow to study the physical mechanisms in impurity seeded, high density H-mode discharges. For this purpose the RITM code has been extended to predict the transport reduction in the edge barrier region of the H- mode plasma. In addition to the transport caused by unstable drift waves owing to ion temperature gradient, dissipative trapped electron and drift resistive ballooning modes, transport related to drift Alfvén modes have been incorporated. This transport model allows to predict the transition between L- and H-mode in dependence of the power flow to the separatrix. Fig. 2 illustrates the capability of the code to predict the formation of an edge pedestal by depicting the temperature at the top of the edge barrier and the normalised pressure gradient as a function of the heating power for the discharge scenario shown in Fig. 1. Note, that the power required to establish the barrier corresponds fairly well with the experimental threshold for the power flow to the separatrix of 9 MW for the discharge conditions under consideration, if we take into account that 25% of the total input power is radiated from the confined volume. 16

17 T i ( ρ 0.95 ), ev, ( ) L-mode P tot, MW H-mode α ( ρ 0.95) ( ) Fig. 2: Normalised pressure gradient and ion temperature taken at ρ 0.95 as a function of the total heating power. With this extended version of the RITM code, predictive and self-consistent modelling of the impact of argon on high density H-mode discharges from the magnetic axis to the separatrix has been performed. Fair agreement between modelling results and experimental data was obtained. It was found that in this plasma regime the injected impurities only weakly affect the background transport. In particular, a transition to a state with peaked density profiles and improved core confinement as observed in L-mode plasmas is hindered by the very flat q-profile in these H-mode plasmas. Experiments and modelling on radiating plasmas at TEXTOR In order to achieve high plasma density necessary in a future fusion reactor, puffing of neutral working gas is ordinarily applied in fusion devices. However, a too intensive gas puff normally leads to a confinement degradation. Prior to the shut down to install the Dynamic Ergodic Divertor, dedicated experiments had been performed to study the effect of external gas injection on the global confinement properties of discharges with a radiating plasma mantle at high density. Detailed spectroscopic studies have revealed clear indications that intense gas injection leads to the formation of a cold and dens plasma cloud at the injection zone. This process is facilitated by localised recycling owing to the local cooling of the plasma and results in a significant amplification of the neutral influx. The global energy confinement at densities well above the empirical Greenwald density limit is found to be closely correlated with the total localised neutral influx as indicated in Fig. 3. Furthermore, the isotope effect of the fuelling gas on the overall plasma performance had been studied. It has been found that the degradation takes place at lower external fuelling rates in the case of hydrogen compared to deuterium injection for a given plasma density. This fact has been explained by a stronger decrease of the effective plasma charge Z eff during the hydrogen puffing. Within the core plasma the reduction of Z eff during the hydrogen puff leads to a decrease of the collision frequency and a proportional increase of the transport driven by unstable drift waves (here the dissipative trapped electron mode). This hypothesis has been supported by self-consistent modelling with the RITM code. 17

18 1.2 τ E / τ RI Γ tot < 5.0 x s x s -1 < Γtot < 1.0 x s x s -1 < Γtot < 1.5 x s -1 Γ tot > 1.5 x s n e / n GW Fig. 3: Global energy confinement time as a function of normalized density in discharges with a radiating plasma boundary using the total neutral influx to the injection zone as a parameter (including the localized recycling flux). Transport processes at the plasma edge The 2D multi-fluid code TECXY was used to study the impact of the Dynamic Ergodic Divertor on the formation of MARFEs in TEXTOR. The increase of the effective transport coefficients in the stochastic magnetic field have been calculated according to a model prescribing optimal paths, which leads to analytical formulae involving the field line diffusion coefficient and the Kolmogorov length. According to the calculations, the increased heat conductivities from DED transport tend to weaken the MARFEs, whereas the increased diffusion and radial convection via enhanced recycling rather tend to feed and strengthen the MARFEs, in particular by enhanced sputtering of carbon impurities. Thus, in the presence of carbon no MARFE stabilisation has been found. However, the plasma core is found to be more efficiently screened from carbon impurities owing to their rapid outward convection. Study of self-organized-criticality (SOC) behaviour of edge plasma fluctuations in TEXTOR In recent years, to understand the widely found non-diffusive transport behaviour characterized by long-range time and spatial dependencies in fusion plasmas, models based on self-organized criticality (SOC) were proposed suggesting the existence of avalanche-like transport propagating from the plasma core towards the edge. The concept of SOC brings together the ideas of self-organization of nonlinear dynamical systems with the often observed near-critical behaviour of many natural phenomena, which exhibit remarkable spatial and temporal self-similarities and long-range correlations. In such systems, scale lengths can be described by fractal geometry and time scales that lead to a 1/f-like power-law frequency dependence of the spectrum. 18

19 To deepen our understanding on these issues, we have studied the SOC-relevant properties on the fluctuation data in ohmic discharges and edge biasing experiments on TEXTOR. The results are as follows: For the normal ohmic heated plasmas, the floating potential and also the density fluctuations exhibit an f -1 power-law dependence in their spectra. The autocorrelation function displays slowly decaying long tails, the rescaled range (R/S) analysis shows Hurst parameters well larger than 0.5, indicating an existence of long-range correlations at all measured locations, and a radial propagation of avalanche events is manifested in the edge plasma region. All these results are consistent with SOC behaviour and plasma transport mechanisms based on avalanches. During the edge biasing experiments, where an edge radial electric field E r and thus an E r x B flow shear is generated, the Hurst parameters are substantially enhanced in the negative E r shear region and in the scrape-off layer as well. Nevertheless, it is found that the local turbulence is well de-correlated by the E r x B velocity shear in the negative E r shear zone, being consistent with theoretical predictions. The increase of long-range dependence and concomitant decrease of local decorrelation time of turbulence in the negative E r shear region appear to support the statements given by Carreras et al. that The dynamics governing the de-correlation of the local fluctuations and the long-range time dependencies are probably different, one being the de- correlation of the turbulence and the other being the de- correlation of the transport events (avalanches). 19

20 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.3. IMPURITY TRANSPORT AND RADIATION m.von.hellermann@fz-juelich.de 1. Impurity survey during the DED start-up phase During the start-up phase of TEXTOR after the DED installation, a variety of plasma impurities was detected during the discharges. Two different types of impurity events can be clearly distinguished and the elements involved can be identified by means of VUV spectroscopy, which measures the intensity of the characteristic spectral lines of the impurities. The first case: Small metal dust particles, which were remaining in the vessel after the long-term in-vessel work. These particles were released to the plasma mainly due to the action of the magnetic AC fields during DED operation, which lead to a mechanical vibration of vessel components. Since the number of stored particles in the vessel is limited, the number of these impurity events per discharge decreased significantly in the course of the experimental campaigns. The number of the detected particles is in the range of to impurity atoms released to the plasma, where the lower limit is defined by the sensitivity of the spectrometers, while larger impurity events may lead to fast disruptions of the discharge induced by radiation cooling. A typical example is shown in figure 1, where the spectra from three different cases are displayed: a) the typical impurity spectrum from a discharge with oxygen as the dominating impurity species; b) the spectrum during an impurity event releasing Ni and Cr into the plasma, and c) the spectrum from a Cu based impurity event. The second type of impurity events is caused by the material release due to the direct plasma-wall contact. During normal plasma operation the released material is mainly carbon, from which the plasma-facing components are made. However, in case of technical and physics problems such as loss of plasma position control, electrical problems or overheating of in-vessel components, also other materials apart from carbon may come into contact with the discharge and may reach significant concentrations in the plasma. As an example we show the spectrum from Zr, which is used on TEXTOR in form of ZrO 2 as insulator behind the DED carbon tiles. 20

21 Figure 1: Impurity spectra dominated by spectral lines from oxygen, nickel/chromium and copper Figure 2: Zirconium spectrum from a TEXTOR discharge 21

22 2. X-ray spectroscopy on TEXTOR The spectra of He-like argon have been revisited to improve the fit of the theoretical models to the experimental spectra. During the analysis it has been found, that the cascades within the system of doubly excited Li-like ions can play a significant role in the level population. Especially levels with small auto ionization rates and hence small population due to dielectronic recombination, can be strongly affected by cascades from higher lying levels within the Li-like system. The intensity of the q and r satellites can increase up to a factor 2 relative to the direct dielectronic recombination. As these lines have strong Ar 15+ exp /Ar15+ cor without cascades new: with cascades Electron Density, N e (10 13 cm -3 ) Figure 3: Density of Li-like ions relative to corona predictions. contributions due to inner shell excitation by electron collisions, they are applied to the determination of the density of Li-like ions. As the contribution by dielectronic recombination increases if the cascades are taken into account, the measured density of the Li-like ions decreases. In Fig. 3 we show the density of Li-like ions in ohmically heated plasmas relative to the coronal predictions. The density of the Li-like ions is increased due to transport effects and due to charge exchange recombination. The ion transport limits the duration of stay in the hot plasma and prevents the ions to arrive at their final equilibrium stage, and charge exchange with atomic hydrogen reduces the charge state as well. Both effects increase for low plasma density, for high densities, the charge state distribution approaches the coronal equilibrium. For the strong lines with high population by dielectronic recombination, the cascades are unimportant. The electron temperature, which is obtained from the intensity of the large dielectronic satellites relative to the resonance lines, does not change if the cascades are taken into account, therefore the previous data are still valid. Density ratio Li-like to He-like relative to the coronal predictions for traces of argon in a deuterium plasma with ohmic heating as a function of the plasma density in the centre. Circles: cascades not included, crosses: cascades included. At high plasma densities, the coronal equilibrium is approached. 3. TEC participation in ITER diagnostics: Active Beam Spectroscopy A comprehensive package of active beam based spectroscopy tools for ITER has been developed and evaluated. The feasibility study encompasses CXRS (Charge Exchange Recombination Spectroscopy) for the measurement of the main impurity ion densities (including helium ash), ion temperatures and toroidal as well as poloidal plasma rotation. Beam Emission Spectroscopy is proposed as indispensable cross-calibration tool for absolute local impurity density measurements 1 and also for the continuous monitoring of the neutral beam power deposition profile. Finally, a full exploitation of the Motional Stark Effect pattern is proposed to deduce local pitch angles, total magnetic fields and possibly radial electric fields. 1 M. von Hellermann et al.: Proc. Adv. Diag. for Magn. & Inert. Fus., Varenna 2001, , Plenum Publ. N.Y. 22

23 More recently 2, a further promising application has been proposed, that is a study of slowing-down alpha particles in the energy range of 0.1 to 0.6 MeV and 1.6 to 2.4 MeV, respectively, making use of the 2.2 MW 100 kev/amu DNB (Diagnostic Neutral Beam) and the 17 MW 500 kev/amu HNB (Heating Neutral Beam). An important asset of the proposed slowing-down scheme is the potential investigation of anisotropic features in the alpha velocity distribution function making use of top and equatorial observation periscopes. Performance studies and estimates of expected spectral signal-to-noise ratios are based on atomic modelling 3 of neutral beam stopping and emissivities for CXRS, BES and background continuum radiation as well as extrapolations from present CXRS diagnostic systems. Single high-étendue and high-resolution spectrometers 4 for each radial channel are proposed for thermal CXRS and BES/MSE and high-étendue broad-band spectrometers for CXRS on slowing-down features. The concept of integrated data evaluation procedures plays a pivotal role for a feasible and successful application of active beam diagnostics on ITER. The issue of measuring the helium ash content is strongly interwoven with the ability of assessing at the same time all complementary ion densities (including bulk ions) and ion temperatures. The consistency of diamagnetic energy data with spectroscopic reconstructions from electron and ion pressure profiles, or the consistency of measured thermo-nuclear neutron rates with modelled predictions from measured ion temperature and deuteron/triton density profiles belong to standard Figure 4: Proposed periscope system on ITER for CXRS and BES (MSE) prediction packages (TRANSP, CHEAP) in present day experiments (e.g. JET, TEXTOR). A key issue for the viability of the proposed active beam spectroscopy package is the survival probability of the first mirror in its periscopes. Studies of metallic mirrors deposition 5 and today s experiments on sputtering processes addressing reflection values and polarisation characteristics have given first results 6. 2 M. von Hellermann et al.: Proc. 4th ITPA Topic Group Meeting on Diagnostics, Padova, February, c.f. H.P. Summers et al.: Physica Scripta T92, 80 (2001) and 4 S. Tugarinov et al.: Rev. Sci. Instr., 74, 2075, A. Malaquias et al.: Proc. 5th ITPA Topic Group Meeting on Diagnostics, St. Petersburg, July, K. Vukolov et al.: Proc. 4th ITPA Topic Group Meeting on Diagnostics, Padova, February,

24 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.4. MAGNETOHYDRODYNAMICS h.r.koslowski@fz-juelich.de The work of this topic group concentrated on sawtooth stabilization experiments with respect to the triggering of neoclassical tearing modes (NTM), the direct stabilization of NTMs by ion-cyclotron current drive (ICCD), and the analysis of MHD modes in conjunction with edge localized modes (ELM) at the JET tokamak. The work performed on TEXTOR comprised sawtooth and mode stabilization experiments with electron cyclotron heating (ECRH) and current drive (ECCD), and first investigations of the influence of error fields induced by the dynamic ergodic divertor (DED) on the MHD characteristics of the plasma. The control of sawteeth is an important issue for a next step fusion experiment like ITER. It is well known that a large fraction of high energetic particles leads to the stabilization of the central sawtooth instability. The time between sawteeth collapses becomes longer and the sawteeth get much larger amplitudes. This large sawteeth trigger NTMs more easily and at much lower beta values. Once the NTM is excited it causes a considerable reduction of confinement or even leads to a disruption of the plasma current. It is therefore desirable to have means to control the sawtooth period and thus its amplitude in the presence of a high fraction of fast particles. TEC scientists participated in JET experiments where central ion cyclotron heating (ICH) was applied in order to create longer sawtooth periods due to the enhancement of the fast particle population. A second ICH system was then successfully utilised for ICCD at the q 1 surface and to destabilize the sawteeth again. The reverse B campaign on JET allowed to study the influence of counter neutral beam injection (NBI) on the sawtooth period. The main effect of counter injection is a change of the plasma rotation. A detailed scan of the injected power was made in L-mode plasmas. Compared to the pure ohmic and co-injection cases a clear reduction of the sawtooth minimum was observed at around 4 MW of counter NBI. A candidate mechanism to explain the experimental findings is the dependence of the internal kink stability on the sheared rotation profile. Previous experiments made at TEXTOR have shown a minimum in the sawtooth period at low values of co-nbi. Work to model the JET results and a comparison with the TEXTOR results is underway. Once the NTM is excited in the plasma, various means to stabilize this mode again have to be explored. An experiment at JET aimed on the stabilization of a NTM by ICCD at the mode rational surface, i.e. at q 3/2. A series of discharges where the ion cyclotron resonance was moved across the q 3/2 radius by varying the toroidal magnetic field was performed. Only a marginal effect on the mode amplitude was observed. The new powerful 140 GHz gyrotron at TEXTOR will be used in the future to stabilize modes induced by the DED under well controlled conditions and to study the physics of mode stabilization by heating and current drive in detail. Preparatory experiments on TEXTOR aimed on a modulated 24

25 operation of the gyrotron in a well defined phase with the DED current which creates the mode. The milestone of modulated ECRH at 2 khz controlled by the external DED current signal was successfully achieved. Another plasma instability of concern for future fusion reactors are ELMs. The related modes lead to periodic collapse events which release a considerable part of the plasma stored energy as well as particles to the plasma facing components and the divertor target plates, leading to unacceptable high heat loads. Work performed at JET aimed on a better understanding of various types of MHD modes at the plasma edge which may be connected to energy and particle losses or the ELM crash itself. Especially so-called type-ii ELMs, where the energy losses are driven by much more benign instabilities between the ELM crashes are currently of great interest. Observations made on ASDEX Upgrade have shown that high confinement plasmas with acceptable divertor heat loads have been achieved. A similar scenario has not been obtained at JET up to date, but mixed type-i/ii phases where the ELM frequency dropped were observed. In these plasmas, part of the energy is transported via Fig. 1: Spectrogram calculated from the signal of a an additional loss mechanism magnetic pick-up coil. which is still not definitely explained. In-between the ELMs a characteristic change of the turbulence due to so-called washboard modes, rotating in the electron diamagnetic direction, was observed. These modes show a rather complex interplay with the ELM precursor modes observed recently, and could be a good candidate to explain the additional losses. The analysis showed that the electron temperature at the edge pedestal could be kept constant when these modes occurred at high amplitudes, but the electron density was still rising, leading finally to the triggering of an ELM crash. The experimental observations of ELM precursor and washboard modes suggest that the standard peeling-ballooning ELM model has to be revisited. TEXTOR resumed operation after a long shutdown for the integration of the Dynamic Ergodic Divertor (DED). The 16 coils were linked up to produce an m 12, n 4 mode structure which does not show deep penetration into the plasma, but merely influences the plasma edge. During the shutdown many diagnostic systems were upgraded as well. Figure 1 shows a spectrogram (windowed Fourier analysis) of a signal from one of the new Mirnov coils. The n 4 mode induced by the DED is clearly visible. A more striking feature is that during operation of the DED the sawtooth postcursor mode, situated deep in the plasma core at the q 1 surface, is not visible. This behaviour is presently not understood and the analysis is ongoing. The new pick-up coil design yields a large bandwidth and allows to measure the whole range of frequencies which are expected to be important. More interesting measurements of MHD features in DED plasmas were made with the correlation reflectometry system. There is a strong influence of the DED perturbation on the rotation frequency of modes located at the plasma edge. This has been observed for both directions of rotation (co and counter) of the external ac DED field. 25

26 After the 2003 campaign the DED coils a re-linked to the m 3, n 1 mode configuration. This mode of operation is much more favoured for MHD investigations. Experiments on error field amplification and ECRH/ECCD mode stabilization are planned for

27 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.5. ADVANCED TOKAMAK SCENARIOS r.jaspers@fz-juelich.de The goal of the "advanced tokamak scenarios" (ATS) is to reach higher pressure at a given plasma current and approaching steady state operation with a large fraction of bootstrap current, i.e. a noninductive, self generated current. These scenarios can be realized by modification of the current profile, which has been shown on many tokamaks to lead to internal transport barriers, thus improving the confinement quality. Moreover, these transport barriers lead to regions of high temperature and pressure gradients which in turn lead to a high bootstrap current. The main aim of the TEC topic group advanced tokamak scenarios for the moment is to create such an operational scenario with internal transport barriers and then focus on the role of rational values of the helical winding number q (directly related to the current density) and the electron transport in these regimes. Furthermore, by local heating or current drive with the electron cyclotron heating system, an active manipulation of the current density profile is foreseen. After a longer shutdown, TEXTOR became operational in the reporting year. This shutdown period to install the Dynamic Ergodic Divertor (DED) was utilized as well to improve some hardware of relevance for the investigations of the advanced scenarios. The experimental campaign on TEX- TOR was first used to bring these into operation. Aside from that, some physics results on the topics of transport barriers and electron transport could be made. Finally, similar experiments were performed in collaboration on different tokamaks, such that comparative studies could be undertaken. Hardware Undoubtedly the most eminent hardware improvement is the new 140 GHz, 800 kw gyrotron. This provides the possibility for local electron heating, thus creating large temperature gradients and e- ven off-axis temperature maxima, which could change the current density profile as well. This latter effect can be directly influenced by the current drive capability of the gyrotron, i.e. launching the electron cyclotron waves oblique to the magnetic field. Furthermore, the gyrotron can also deliver modulated pulses to the plasma. From the analysis of the evolution of these heat pulses the electron transport can be examined. This gyrotron was successfully taken into operation. Showing a record pulse of 800 kw during more than 2.7 s, it outperformed the previous gyrotron by far, see Fig.1. From the diagnostic side, developments in three different areas, being relevant for the ATS programme, were undertaken. The MSE, the Multi Pulse Thomson scattering and the 2D ECE-Imaging systems came into operation. 27

28 Fig. 1: The effect of the new 140 GHz gyrotron on the central electron temperature, compared with the old 110 GHz gyrotron: longer pulse length, more power, leading to higher electron temperatures. MSE: the clear relation between the advanced scenarios and the current density profiles urged for a good diagnostic of the q-profile. Exploiting the motional stark effect of the injected neutral hydrogen atoms can reveal this q-profile, as has been demonstrated in the last decade on the big tokamaks. Before the DED-shutdown a prototype of an MSE system based on measuring the ratio of intensities of two orthogonal polarised lines of the Balmer-α spectrum was already tested. This has now been improved and became operational in the last experimental campaign (although not all 20 radial channels yet). The first results are depicted in Fig. 2. Fig. 2: First results of the MSE diagnostic: On the left a typical spectrum is shown. From the ratio of the π and σ components and the Doppler shift, the q-value can be directly obtained. This q- profile is shown on the right for the 8 operational channels. Thomson scattering: The high resolution Thomson scattering system has been upgraded to provide time information as well. In the new design the laser cavity encompasses the plasma, so that the laser pulse passes many times through the plasma. In this way, a fast repetition of laser pulses is obtained; typically three bursts of laser pulses with a repetition frequency of up to 10 khz can be generated (see Fig. 3). Newly developed fast CMOS detectors are used to record measurements of the temperature and density profile for each individual pulse. The spatial resolution and accuracy are the same as that for the existing double-pulse Thomson scattering system. 28

29 Pumping power, MW Inversion, J Probing energy, Joules Probing power, MW Inversion without osc., J Inversion with osc., J t, ms a b c Exp. probing energy, J Calc. probing energy, J d 5 4 Pumping power, MW Inversion, J Probing energy, J t, ms 6 Fig. 3: Output of the multi-pulse laser system for Thomson scattering. The left graphs show in blue the output that has been obtained thus far with an improvised capacitor bank. The red circles show the calculated values. Due to the short duration of the pumping power from the capacitor, the pulse energy quickly decreases. Still almost 20 laser pulses are obtained. The right figure shows the output energy of the laser with an improved capacitor bank as calculated with the same model being used in the left graphs. 2D-ECE-imaging: This system images a two-dimensional part of the poloidal plane onto an MMIC array with in total 8 horizontal and 16 vertical measuring channels. The system has a spatial resolution of approximately one cm in the poloidal plane, which is better than that of most standard heterodyne ECE systems. This makes it possible to measure the detailed 2D profile of electron temperature fluctuations. The ECE-I system is combined with the Microwave Imaging Reflectometer (MIR) system and has been recently installed at TEXTOR. Physics results The physics experiments were concentrated on the issue of electron transport barriers. Before the shutdown, using the older gyrotron, L-mode discharges were made which exhibited a strong barrier associated with the q1 surface. These experiments could reasonably well be described by the RTP q-comb model, in which the heat conduction is supposed to be a function of q only with a constant high value interspersed with narrow regions of low conductivity located near the low rational values. Now, this q-comb model has been applied to a density scan and a current scan. Given the experimental density profile, the heat deposition profile, Z eff and the heat diffusivity χ e (q) according to the q-comb model (taken the same as on RTP, except for an L-mode scaling factor, thus no fitting parameters were used), the electron temperature profile was iterated until a stationary solution was obtained. In all cases the experimental data was insufficient to provide an accurate test of the model. Nevertheless, the model reproduced to a certain extent the main features of the experiment: the central temperature and the width of the profile. Some examples are shown in Fig

30 Fig. 4: Application of the same q-comb model to centrally heated ECRH discharges at low [n e (0) 1.0 x m -3, 87638] and higher density [n e (0) 2.5 x m -3, 86912], both at I p 200 ka, and to two discharges differing in plasma current: 87650, I p 200 ka and 90405, I p 305 ka, both with n e (0) 2.0 x m -3. Although not all details are provided by the experimental data, some features such as the central temperature and the width of the profiles are reasonably well described. A further confirmation of the existence of at least the q1 barrier was found in modulated ECRH experiments in which a modest change of the deposition radius around the inversion radius caused a sharp transition in the phase profile of the perturbation. Apart from these discharges with central ECRH heating, an extensive set of experiments was made in which the deposition radius for the ECRH was dynamically scanned through the plasma. Contrary to RTP results, no transitions or crashes were observed when a rational q surface was crossed. TEC proposed and was involved in the same kind of experiments performed on other tokamaks. On ASDEX and DIII-D, part of the phenomenology of RTP was found as well: off-axis heating of ECRH led to off-axis maxima in the electron temperature. However, the transition in confinement when the heating crossed a rational surface was not observed on ASDEX. Also a collaboration with T10 was initiated. A very specific feature that has been observed in T10 plasmas, in which ECRH power is deposited just outside the q1 radius, is that after switching off the ECRH gyrotrons the central temperature does not decrease immediately (see Fig. 5). Instead, it stays constant for several tens of ms before it starts to decay. The effect in T10 was most pronounced in case the ECRH power was just high enough to stabilize the sawtooth activity in the plasma core. From these results the statement that the necessary condition for the formation of a transport barrier was a low shear close to a rational value of the safety factor could be confirmed. Similar attempts were undertaken at TEXTOR, but due to the limited shot time, only a marginal effect could be observed up to now. Finally, the TEC involvement in the JET advanced scenario s programme was concentrated on the realization of internal transport barriers at high density, equal ion and electron temperatures and a high fraction of bootstrap current. This could be achieved in a scenario in which there is first a preheat phase by lower hybrid heating to form a hollow q-profile, followed by a short ohmic phase, in which pellets were injected to rise the density, after which the main heating phase started and the internal barrier was formed in a high density plasma. 30

31 T ECE (kev) #32913 r3cm T e-in T e-out 20ms Fig. 5: Result from T10: After switch off of ECRH the central temperature stays approximately constant for 20 ms, whereas outside the barrier the temperature immediately drops. 1.0 r11cm 0.8 ECRH Time (ms) 31

32 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.6. DYNAMIC ERGODIC DIVERTOR k.h.finken@fz-juelich.de The primary goal of the topical group is the application of the DED to the TEXTOR plasma and the understanding of the resulting physics. Electric currents in the coils of the DED superimpose a resonant perturbation magnetic field to the equilibrium field of the plasma; this additional field weaves the magnetic field lines such that a field line fills a whole volume instead of a surface. This ergodisation increases the plasma transport at the boundary such that the deposited heat will be distributed over a relatively large wall area in a future fusion reactor, the high heat flux density to the walls is one of the critical issues. In addition, the DED of TEXTOR is unique in so far as the perturbation field can rotate with a velocity up to the order of the ion drift velocity in the plasma edge. By this rotation, new possibilities may open up for improving the plasma confinement. After two years of installation work, the DED has been taken into operation in June In about 250 discharges the DED was applied successfully. In about half of the discharges the DED was operated in the DC mode and the rest is shared in the 2 Hz mode (55), the 2 khz mode (58) and in the 7 khz mode (15). During 2003, the DED was operated in the fine m/n 12/4 current distribution mode which is relevant in particular for questions of plasma-wall interaction. End of 2003, the DED configuration has been switched to the coarse m/n 3/1 mode; aim of this operation is the investigation of the interaction of islands deeply within the plasma. So far the following measurements have been performed: IR thermography in order to visualize the characteristic stripe pattern in front of the DED coils and expected from the near field of the DED; this strip pattern is shown in the figure. Un-uniformities of the heating arise from the imperfection of the alignment of the DED divertor target tiles. The distribution of the heat flux has been related to results of field line tracing and it is being analysed to which path the power takes through the rather complicated ergodic boundary layer to the target plates. Probe measurements for obtaining the local values of the electron density and temperature at the target plate. Diverse spectroscopic measurements yielding the fluxes of recycling deuterium, and the released impurity distributions. Analysis of the screening efficiency of the ergodic zone against impurity influx to the plasma core for different plasma conditions. 32

33 Atomic beam measurements for deriving the local transport coefficients in the ergodic and laminar zones. These data are the basis for tests of different transport theories based on nonlinear dynamics. Development of a new Hall probe for obtaining the current pattern induced into islands of the ergodic zone. Electron temperature and density profile measurements for analyzing the global confinement properties. Determination of the structure of the rotating magnetic field waves by sets of Mirnov coils. It has been found that the rotating mode spectrum is modified by the presence of the DED field with characteristic differences during static and dynamic operation. In addition to the specialized diagnostics, the whole set of standard diagnostics has been applied to investigate the influence of the DED on the plasma. 0 TEXTOR-DED - toroidal/poloidal grid 0 TEXTOR-DED - toroidal/poloidal grid s (mm) s (mm) φ (deg) φ (deg) 20 Thermographic image for the case of low edge ergodization Thermographic image for the case of high edge ergodization For the analysis of the DED, the following theoretical work has been performed during the recent year: For investigating the particle transport in the ergodic area, a new mapping method for the guiding centre motion of ions and electrons has been derived and improved. It has been shown that the ergodisation is more pronounced for counter-rotating particles than for corotating ones. In order to characterise the perturbed edge zone, the code Atlas has been prepared showing the ergodic zone, the field line connection length of the laminar zone and the strike zones of the magnetic field lines at the divertor target plate for different plasma conditions (I p, β pol ). The atlas is a major tool for referring a specific measurement with respect to the complicated 3-D structure of the ergodized edge plasma. The 3-D code EMC3 for the plasma transport in the laminar and ergodic zones of the DED plasma has been developed. The code has converged for a series of interesting plasma conditions. 33

34 The penetration of the rotating perturbation field into the plasma is still an unsolved problem in case of an ergodic background field pattern. With respect to the DED, several groups (TEC partners, groups in Graz/Austria, Sao Paulo/Brazil, Kharkov/Ukraine and Novosibirsk/Russia) are contributing. The groups are using different techniques, such as the analysis of low frequency wave propagation, an analytical linear model or a non-linear numerical code. The different methods calculate the shielding current induced in the edge of the plasma due to the rotating magnetic field, the radial decay of the perturbation field into the plasma and the force transferred from the external coil currents to the plasma. Collaborations The group participates in an international effort of exploiting positive effects of ergodisation on fusion devices. The efforts concentrate on the mitigation of Edge Localized Modes (ELMs); these ELMs are a prominent feature of so called High Confinement Discharges (H-modes) and are linked with high transient power losses. It has been shown on DIII-D (San Diego / USA) that the ELMs were suppressed or largely reduced by the application of an external resonant perturbation field. For a fusion reactor, methods are investigated to reduce these extreme heat fluxes. The group collaborates in this respect with the fusion groups in Cadarache/France (Tore Supra) and San Diego/USA (DIII-D). Finally, an international workshop on Stochasticity in Fusion Edge Plasmas (SEP) has been established which held a meeting in October in Jülich. In the workshop, recent efforts and developments on stochasticity were presented; an aim is the coordination of related activities in the different laboratories. The papers will be published in the highly recognized journal Nuclear Fusion. It is planned to organize the workshop biannually. 34

35 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.7. THEORY AND MODELLING r.koch@fz-juelich.de A substantial fraction of the theoretical activity of the TEC, closely related to experimental research, is reported in the corresponding Topic Group sections. The present review covers complementary activity. Transport and confinement Modelling of L-H transition and H-mode characteristics with the code RITM The 1-D transport code RITM has been amended by including a model for anomalous transport contribution driven by drift-alfvén (DA) turbulence. This allows simulating H-mode plasmas in a self-consistent way, from the plasma axis to the separatrix including the edge transport barrier. It is demonstrated that the formation of edge pedestals requires suppression both of DA and of ion temperature gradient (ITG) unstable modes. The former occurs due to decreasing plasma collisionality and increasing beta with growing heating power. The stabilization of ITG turbulence is caused mostly by sharp density gradients, which develop at the edge due to ionization of incoming neutrals when DA transport is reduced. On the one hand, this explains the proportionality of the radial pedestal width to the penetration depth of charge-exchange neutrals, which scales as the square root of the ion temperature. On the other hand the temperature gradient, which drives ITG instability, is controlled in the barrier by the neoclassical heat conductivity and, therefore, the pedestal width scales inversely with the plasma current. All together this leads to a proportionality of the barrier width to the poloidal Larmor radius, widely observed in experiments. The seeding of argon can lead to a significant modification of the edge transport, in particular, to a widening of the edge barrier and an increase of the plasma density, thus preserving a good confinement quality. However, a too intensive edge radiation from impurities, which cools the plasma edge too strongly, results in a resumption of DA activity and can trigger a back H-L transition. Studies of plasma micro-instabilities and their contribution to anomalous transport The effect of gyro-viscosity on ion dynamics and improved mixing length approximation have been taken into account by considering dissipative trapped electron (DTE) instability. It is demonstrated that DTE contribution to anomalous transport significantly decreases with increasing hydrogen isotope mass in agreement with numerous experimental observations. A theory of the ITG PVS (ion temperature gradient / parallel velocity shear) drift instability, driven by a synergy of the gradients of the ion temperature and the parallel velocity, has been elaborated. On the one hand, the theory predicts that the threshold of the ITG instability is reduced in the presence of the parallel velocity shear. On the other hand, the threshold of the PVS instability becomes intrinsically dependent on the temperature gradient. The features of the ITG PVS instability appear 35

36 to explain (i) the origin of the quasi-coherent (QC) mode which occurs at the transition from ELMfree to EDA H-mode in ALCATOR C-Mod; (ii) the reduction of the particle (but not of the energy) confinement time which is observed in EDA discharges; (iii) the strong role of q (the safety factor) and, to a lesser extent, of δ (the triangularity parameter) in obtaining EDA behaviour. The scenario assumes that the H-mode pedestal (where the ITG PVS instability and enhanced particle transport are localised) is impermeable for neutrals, since charged particles issued from ionisation would otherwise accumulate in the core. That requirement, together with the ITG PVS instability condition, might be difficult to satisfy in tokamaks with low or moderate magnetic fields. L-H transition and Reynolds stress In view of understanding the L to H mode transition, a theoretical model has been developed to investigate whether the Reynolds stresses could be responsible for it. This model retains as driving mechanism the momentum from the turbulence that is transferred to the main flow via the Reynolds stress and includes neutral particle friction and parallel viscosity as damping mechanisms. Neoclassical theory and H-mode transition in presence of additional heating The development of a new neoclassical theory taking into account the non-maxwellian distributions generated by the external heating of the plasma by electromagnetic waves has progressed along the following lines: a) definition of a non-maxwellian stationary state in presence of heating, b) inclusion of this state in the method of solution of the drift-kinetic equation in presence of a turbulent field, c) determination of the turbulent transport coefficients. At this stage, various models of turbulence are investigated. For the investigation of the relation between the L-H transition and the plasma impurity content, the radial electric field bifurcation due to different loss and input mechanisms including impurity injection was considered. This work is in the continuation of previous investigations by considering the time evolution of the electric field during the transition. One of the main results is that, within the framework of the model, the stationary radial electric field strongly depends on the fraction of expelled impurity ions. Investigation of mechanisms for the density limit in tokamaks Normally the radiation losses from impurities are considered as the main cause of the density limit in fusion devices, although a significant increase of particle convection was recorded as a precursor of the density limit in various machines. It has been demonstrated that namely the synergy of these two channels for energy losses plays an extraordinary role for the density limit. As the plasma density is ramped up, e.g. by gas puffing, the edge temperature drops. The process is, however, dramatically accelerated when at a critical level n cr the turbulence driven by the so called drift ballooning instability starts to dominate anomalous transport at the plasma edge. As a result the particle convection increases dramatically and strongly cools the plasma edge. The latter cooling activates the radiation channel of energy losses, which finally leads to the density limit at a certain n max (see Fig. 1). The calculations demonstrate that due to a very non-linear temperature dependence of the radiation, n ame only slightly exceeds n cr and, as it is observed in experiments, varies very weakly with the impurity concentration. Therefore, the transition in the edge turbulence, taking place at n cr, actually sets the density limit. The parametric dependence of n cr that is obtained is very close to that given by the experimental Greenwald scaling. 36

37 The scenario described above does not exhaust all possibilities and often, the density limit is caused by development of MARFEs, a region with very low temperature and high density at the high field edge of the device. A model for the MARFE threshold, which takes into account plasma interaction with the inner wall, has been developed. Both local release of impurities and recycling of hydrogen neutrals are considered. It is shown that, by leading the so called recycling instability, main neutral particles play a more important role in the formation of MARFEs than impurities. However, the situation can be changed if the recycling properties of the wall are modified, e.g., by wall covering with specific materials. This explains, in particular, a significant increase of the density limit in TEXTOR after wall boronization. The Dynamic Ergodic Divertor (DED), operating now on TEX- TOR, can lead to an essential modification of the plasma wall interaction pattern, predominantly in its intensification at the high field side. According to the modelling performed this should result in a systematic decrease of the MARFE density limit. Fig. 1: Edge temperature versus core density computed without (curve 1), with (curve 2) impurity radiation and under the assumption that drift resistive ballooning instability is suppressed in the whole parameter range (curve 3). (The broken branches of curves 2 and 3 correspond to unstable states). Plasma heating physics Modelling of ion-cyclotron antennas In the domain of ion cyclotron heating, the plasma wave and Fokker-Planck codes able to describe self-consistently the evolution of the distribution function of the heated species and the propagation of waves in the presence of this distorted distribution function continue to be developed. The effort in the modelling of the antennas and the coupling of power to the plasma has focused on the use of the Micro Wave Studio (MWS) code. This was applied not only to the JET ITER-like antenna under development but also to the existing JET A2 antennas. Fig. 2 shows the degree of detail of the numerical model. The predictions of the code were found to be in agreement with the electrical characteristics of the A2 antennas measured in vacuum. For simple antenna cases, the input matrix produced by this code was compared to that obtained from the code ICANT, which allows considering coupling to plasmas. 37

38 Fig. 2: MWS model of one-half of the JET-A2 antenna. Impurity pump-out with RF The effective pump out of a large fraction of impurity ions by off-axis ion cyclotron heating (ICRH) was observed in the TFR tokamak. The energy gained by the cyclotron interaction can be high enough to throw the impurity ions out of the plasma column. The code RITM has been modified by including the heat balance equations for the impurity ions in order to allow calculating the response to RF heating of selected impurity species. The numerical simulations, incorporating results of a trajectory code were performed for the JET plasma. The outgoing ion flux created by this method can generate a radial electric field, which may give rise to an L-H transition. This behaviour is studied analytically when the radial electric field is driven by the combination of this outgoing ion flux and an anomalous nonambipolar flux. In this way, a microscopic theory of the L- H transition is obtained, based on a model of non-linear mode interaction provided by a system of coupled Langevin equations for the Fourier modes of the radial electric field. A neoclassical theory of turbulence in presence of RF heating is developed, based on a) the definition of a usually non-maxwellian stationary state in presence of heating, b) insertion of this state in the method of solution of the drift-kinetic equation in presence of a turbulent field, c) determination of the turbulent transport coefficients. At this stage, it is attempted to go deeper into the transport coefficients by assuming various models of turbulence through the dispersion relation. Electron cyclotron current drive A mechanism for the acceleration of electron populations resulting from the effect of crossing electromagnetic waves propagating in a dispersive medium is studied, in view of possible application to non-inductive plasma current generation in the electron cyclotron wave domain. To analyse this mechanism, the resonant moments (RM) of the distribution, i.e. velocity moments computed in the resonant layer only, are evaluated. Although the RM approach has to be considered as an approximation, this prediction is reasonably confirmed by direct statistical simulations. It is shown that the two-wave scheme allows to raise the mean electron velocity up to one order of magnitude when compared to the one-wave scheme because of collective effects. A power dependence of the current drive efficiency has been reported some years ago for the electron cyclotron current drive (ECCD) problem. The effect is observed in simulations when the absorbed power is larger than some threshold value, being close to 1.5 W/cm 3 for typical tokamak plasma parameters, but was not seen experimentally. Similar scenarios have been simulated with the Fokker-Planck solver developed in SPP-ULB, showing that in the framework of the quasilinear theory, the effect is found only for off-axis cyclotron resonance position. 38

39 Nonlinear plasma dynamics, turbulence and plasma behaviour in stochastic magnetic fields Electromagnetic tokamak turbulence simulations Numerical simulations of turbulent transport in tokamak experiments with intense electron heating have been performed using the CUTIE code. These simulations model the plasma turbulence at scales above the ion gyroradius and include three-dimensional electrostatic and magnetic perturbations, various types of MHD modes, neo-classical transport effects and other "toroidal plasma physics" in a cylindrical co-ordinate system. The code is developed by A. Thyagaraja at UKAEA- Culham and further optimized at Rijnhuizen to run on the supercomputing facility SARA. Typical simulations describe how plasma turbulence develops and saturates when (localized) heating and particle sources are applied to the tokamak. Focus of the studies have been the transient transport effects caused by perturbations. Experiments have been simulated in which fast plasma edge cooling is caused by e.g. injection of hydrogen ice pellets. The introduction of cold material causes a so-called cold pulse to travel inward. Remarkably, in the former RTP tokamak such injection has been found to have also non-diffusive consequences, such as a transient rise of the central electron temperature. This suggests a reduction of turbulence driven transport in an area that is larger than the area that is cooled by the pellet. CUTIE simulations do indeed show such a reduction. CUTIE simulations of the Russian tokamak T-10 have been started in the framework of the NWO collaboration between Rijnhuizen and the Kurchatov institute in Moscow. As part of the TEC collaboration CUTIE is being prepared for transport studies in the TEXTOR tokamak. Electron-magneto-hydrodynamic turbulence The fundamental question of how free energy, pumped into a plasma, is dissipated and transported by driving plasma turbulence, was addressed in numerical studies of high frequency plasma turbulence. The rapid fluctuations were captured in a two-dimensional model (with all three magnetic field components) of the plasma electron dynamics: electron-magnetohydrodynamics. The strategy was to start with perturbations with specific wavelengths and to simulate the ensuing turbulence while it decays into a (usually self-similar) state that is insensitive to details of the initial perturbations and therefore has a number of universal characteristics. The simulations have shown how the turbulent energy is redistributed over large and small scale fluctuations The resulting energy spectra (distributions of energy over modes with different scales) have been monitored during the turbulence decay. The spectra are found to be different for fluctuations that are larger and smaller than an intrinsic scale in the electron fluid, its inertial skin depth. At scales exceeding the skin depth, the turbulent energy is found to follow a direct cascade to smaller scales, while other conserved quantities (mean square momentum and helicity) followed an inverse cascade to larger scales. The time evolution of the ratio between axial and poloidal magnetic energies depends on their initial values. When initially the energy was unevenly distributed over the axial and poloidal magnetic fluctuations, a rapid energy exchange took place. Subsequently the turbulence reached a self-similar decay state in which the energy and square momentum spectra are specific powers of the mode number. The total energy is found to decay as t -2/3, consistent with the selective decay of the energy at constant mean square momentum. The maximum of the energy spectrum shifts towards low mode numbers and decays as 1/t, as expected from the infrared scaling of the turbulence. In the opposite regime, where the skin depth is large compared to the fluctuations, both energy and mean square momentum exhibit direct cascades. The dynamics reduces to a Navier-Stokes equation for the axial field fluctuations (the poloidal field being passively advected), but only if the poloidal kinetic energy is larger than or equal to the axial kinetic energy. A different energy spectrum is found now, with different power laws for mode numbers larger and smaller than the initial perturbations. 39

40 Further insight in the numerical results has been gained by making a systematic analysis of the scaling exponents of the two hierarchies of structure functions, generalizing the method of equivalent reference fields in scale-invariant Navier-Stokes turbulence. It has been found that these scaling exponents are determined by six free parameters that need to be determined numerically, two of which are related to the nature of the most intermittent structures. Collisionless magnetic reconnection Magnetic reconnection is a key mechanism in transport processes both in (virtually collisionless) astrophysical plasmas and magnetic fusion experiments. It is this absence of collisional dissipation to challenge our understanding of high reconnection rates and release of magnetic energy. In weakly collisionless plasmas with a strong magnetic guide field, reconnection due to electron inertia is accelerated by electron compressibility to reconnection rates comparable with those estimated from tokamak plasma instabilities. Numerical two-fluid simulations have shown that the high reconnection rate involves the acceleration of electrons in an exceedingly thin current layer. However, in this fluid model of reconnection, smaller and smaller length scales develop: current and vorticity gradients become singular. Surprisingly, it was found that a more accurate, kinetic, description of the electrons during this process resolves the singularity and still predicted fast reconnection rates. Hence, the singular behaviour in the fluid description is an avoidable artefact of the fluid equation of state. This kinetic electron description introduces wave-particle resonances as an additional collisionless reconnection mechanism besides electron inertia. It turns out to be possible to separate the two effects by studying quasi-steady reconnection in which the Landau resonances do not play a role. Remarkably, such quasi-steady solutions to the kinetic equations are as tractable as their fluid counterparts and so fully nonlinear analytic solutions can be obtained. Such solutions show a fine but smooth structure of the current density in the reconnection layer, also found in particle simulations of collisionless reconnection and resembling the x-shaped current distributions found in the two-fluid model. In magnetically confined plasmas, reconnection of field lines that are at different temperatures play a key role in collisionless electron heat transport. The kinetic description resolves this process. It has been found that the effect of a temperature difference between the pre-reconnection regions is to perturb the vorticity and current distributions and thus the entire x-point geometry. Formation of internal transport barriers (ITBs) The hypothesis that ITBs form as a result of a reduced coupling between magnetic islands is investigated using the map technique TOKAMAP. Simulations show that the magnetic topology is strongly dependent on the value of q min. When it is smaller than 1.5, the two rational island chains m/n 3/2 are well separated radially. When it increases above this value, the two island chains are displaced to the low shear region until they overlap, creating there a wide stochastic belt. All the robust magnetic surfaces are lost in this structure. With a further increase in q min, a wide low shear region with robust magnetic surfaces an internal transport barrier is formed. Zonal flows In a tokamak (but also in the terrestrial atmosphere), under certain circumstances the drift wave turbulence is able to generate a shear flow spontaneously. The latter is characterized by a correlation length that is much longer than that of the original drift wave turbulence. This large-scale turbulent poloidal flow is called a zonal flow. Its effect is a tearing apart of the drift wave structures and their fragmentation. This phenomenon, and its implications for the theory of the H-mode, is approached 40

41 using a new treatment, based on an extension of the decorrelation trajectory method introduced earlier. Large eddy simulation (LES) of turbulence Part of the activity in hydrodynamics and magnetohydrodynamics (MHD) is focused on the numerical simulation of turbulent phenomena. Two types of approaches are considered. For moderately turbulent systems, the evolution equations of the turbulent fields can be computed directly using the so-called direct numerical simulations (DNS). However, for highly turbulent systems, numerical approaches mixing simulation and modelling are required. In that case, the large-eddy simulation (LES) technique is considered as a valuable alternative. The LES equations are obtained by applying a filter to the evolution equations that separates the large scales of the turbulence from the small scales. Though the large scales are still simulated directly, the influence of the small scales is taken into account through a model. This kind of investigation is performed under a variety of forms, ranging from the study of simple physical systems, which allow to investigate the accuracy of the hypotheses up to realistic systems with a complicated geometry. As an example of the former, Fig. 3 compares the evolution of a turbulent channel flow obtained on one side by direct numerical simulation, and on the other side by the LES decomposition technique. The quality of the assumptions made to model the small-scale turbulence is reflected in the closeness of the two results and serves as validation for using the same hypotheses in more complex situations. The latter require sophisticated numerical methods to represent complex geometrical situations. One approach is represented on Fig. 4, where spectral representation is used in one direction (assumed periodic) while a non-uniform triangular 2-D mesh is used in the perpendicular planes. Fig. 3: Comparison of energy contours obtained for channel flow from the DNS and LES description. The initial flow (left) is isotropic and evolves towards a state (right) where the flow is elongated in the direction of the intense magnetic field. Another code has been written that computes the relativistic trajectories of particles in a static turbulent electromagnetic field. When this code will be interfaced with that computing the turbulent field distribution, statistics on particle trajectories will be obtained and the acceleration mechanism will be studied. Hamiltonian maps in a stochastic magnetic field The study of mapping models for magnetic field lines in a toroidal system has been continued and it was demonstrated that symmetric mapping models are compatible with field line equations. This symmetry means invariance with respect to a reversal of the toroidal direction. "Tokamap", a map 41

42 proposed earlier by Balescu et al. for a description of field line behaviour in a tokamak, has been generalized as a "symmetric tokamap". A systematic and rigorous method for the construction of symplectic maps near the separatrix of generic Hamiltonian systems has been developed. This method was applied to study the structure of magnetic field lines in tokamaks with a poloidal divertor in the presence of non-axisymmetric magnetic perturbations. Some peculiarities of this structure can be responsible for the stabilizing effect of asymmetric magnetic perturbations on edge localized modes observed in DIII-D. Fig. 4: A typical representation incorporating combined spectral and finite-element approaches. Penetration of the DED field into the plasma An ongoing study of the penetration of the DED field into TEXTOR uses a linearized two-fluid warm plasma description of eighth order in slab geometry, incorporating thermal effects and an inhomogeneous equilibrium. Wave interaction in the Alfvén resonance region is crucial to DED operation. Although unexpected from the dispersion relation, a first numerical integration through this resonance region shows probable existence of coupling between the cold and warm wave modes. A study of the energy fluxes in the different waves is being carried out with a view to connecting a highly localized full-wave integration across the Alfvén resonance zone to short wavelength WKB modes carrying energy away from the resonance. Modelling of transport induced by the DED The code DALF3 developed at IPP Garching for 3-D simulations of the edge turbulence has been applied to model the modification in anomalous transport caused by the DED. First calculations demonstrate that the magnetic field perturbations from the DED should significantly modify the pattern of the electric field at the plasma edge and, thus, intensify drift flows of charged particles. Also the intrinsic magnetic fluctuations associated with drift-alfvén turbulence increase noticeably when the DED is activated. All together this results in a significant augmentation of transport in the region influenced by the DED. 42

43 The theory groups in the TEC The Jülich group is mainly active in transport modelling, edge and plasma wall interaction codes and DED modelling. In the Netherlands at FOM, the theoretical research is carried out within the different research groups of the plasma physics department. In Belgium, the ERM/KMS team investigates ion cyclotron heating, DED field penetration and edge flows. The ULB team focuses on turbulence, transport theory, plasma heating and current drive. 43

44 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.8. OPERATION AND FURTHER DEVELOPMENT OF TEXTOR b.giesen@fz-juelich.de Commissioning of TEXTOR with the Dynamic Ergodic Divertor (DED) Technical Concept of the DED The Dynamic Ergodic Divertor (DED) has been installed in TEXTOR in order to influence transport parameters in the plasma edge and to study the resulting effects on heat exhaust, edge cooling, impurity screening, plasma confinement and stability. The DED consists of four sets of coils at the inboard side of the TEXTOR vacuum vessel, each with four helical conductors (Fig. 1). Two additional conductors at the top and at the bottom are necessary for the compensation of edge effects due to the combination of four coils each at the vacuum feed-throughs. Fig. 1: Location of the DED coils inside the TEXTOR vacuum vessel. For the installation of the coils the TEXTOR liner has been removed, a 120 degrees section cut out poloidally and re-inserted. A support structure welded to the vacuum vessel clamps the coils and also the target plates to the vessel. Coaxial vacuum feed-throughs include connections for coil currents and cooling media. The coil sets are energized by DC or 4-phase currents at selected frequencies (50 Hz and between 1 khz and 10 khz) with amplitudes of up to 15 ka. 44

45 DED Prototype Testing and Simulation Numerous tests of the support structures, the coils, the feed-throughs, the eddy current heating and the assembly of DED components have been performed using a specially built mock-up, which is equipped with a spare toroidal field coil of TEXTOR for generation of relevant magnetic fields and a prototype power converter, allowing for the operation of the DED coils at the full frequency range (DC and 50 Hz to 10 khz). Integration of the DED During the major shut down all preparations for the DED have been performed and the components have been integrated. In parallel TEXTOR components have been modified to allow for the integration of the DED and for improvements of the TEXTOR performance. General maintenance has been carried out, subsystems have been improved and wear parts exchanged. New platforms have been planned, manufactured and set up in the TEXTOR bunker in order to allow for the installation of DED-components as well as diagnostics. The main steps of the DED installation have been: opening of the TEXTOR bunker (removal of the roof and parts of a wall) removal of diagnostics (control cabinets and diagnostic equipment including flanges to the vacuum vessel) dismantling of TEXTOR (after splitting the vacuum vessel, the liner, poloidal coils, transformer yokes and other components are removed) removal, modification and reintegration of the modified liner adjustment of the modified liner inside the vacuum vessel (using specially designed turnable supports) construction, assembly, adjustment of a laser measuring system and an alignment rail system used for precise positioning of DED in-vessel components assembly and installation of the coaxial feed throughs and high frequency shielding welding of pads for DED coil supports onto the vacuum vessel installation of the coil clamps and DED coils installation of the divertor supports and divertor target plates Electrical Systems Only two separate power supplies are required to produce 4-phase currents. Nine power supply units with 750 kw each (Fig. 2) supplied by a single rectifier transformer are foreseen to limit the unit size, to have 9 identical units and to allow for special asymmetrical patterns of current distribution. Each power supply unit feeds a load unit of two coils. The details of the power supply concept have been defined together with industry. 45

46 Fig. 2: Arrangement of the DED power supply system. The maximum operating frequency of 10 khz led to the choice of IGBT inverters. Between 1 khz and 10 khz the reactive load is about 20 times the active load. For compensation capacitor banks are connected in series to the load thus forming resonant circuits. Two capacitor banks in three different configurations each, plus the parallel connection of both, yield 7 resonant frequencies. The DED power supply system has been completely installed and commissioned. After testing with a full size dummy load and final modification of the control software the system has passed the final acceptance test. Commissioning of TEXTOR Apart from usual commissioning procedures for the vacuum system, the heating system, the cooling system, the power supplies, the control and data acquisition, the main diagnostics the situation after integration of the DED required particular attention to the new operating characteristics of TEXTOR. During shut down the TEXTOR liner with 120 degrees poloidally cut out has been modified in order to redistribute the liner heating current. The heating of the liner has carefully been observed and analyzed in order to be sure to have a proper surface conditioning and to avoid partial overheating. With the DED coils and their support structure as well as the divertor a huge amount of additional surfaces had to be conditioned many of them being heated via radiation from the liner and the vacuum vessel only. Even if due to outgassing of polyimide the base pressure in TEXTOR for a couple of weeks did not reach the level observed before the integration of the DED, plasma operation started. 46

47 The start-up phase of the plasma also had to be adjusted after major modifications. TEXTOR start up works at maximum pre-magnetization of the iron core in order to maximize the shot length (up to more than 10 s). The ionization process at application of the break down voltage is very sensitive to the presence of stray fields even in the order of 0.1 mt. Using the stray field diagnostics appropriate start-up conditions could be found. Glow discharge cleaning and boronization of TEXTOR finally led to first successful shots and to improved performance (Fig. 3). Fig. 3: Plasma current of TEXTOR shot #92358 at improved performance after boronization. The black trace shows reference shot #91580 before DED installation. The shot length is influenced by auxiliary heating. After improvement of the vacuum condition the feed-back control systems have been adjusted to the new situation resulting in a plasma performance comparable to that before DED integration. According to experiences during commissioning of the DED power supply system on a dummy load induced error signals have to be expected at medium frequency operation. In order to prevent this, a fibre optic transmission was developed. More than 50 diagnostic signals being sensitive to disturbances have been equipped with this. A couple of machine diagnostics have been modernized during shut down. One of them being the magnetic diagnostic of the TEXTOR iron core had to be replaced because repair of the old system was impossible. After appropriate adjustment and calibration the performance and accuracy of the old system could be reached (Fig. 4). Thus, the diagnostic is available again for pre-control of the plasma current feed back control resulting in an improved performance. 47

48 Magnetizing of the TEXTOR iron core BK / T ThetaIP+294*IBM-48*IBV-64*IBF / kaw Fig. 4: Magnetizing of the TEXTOR iron core comparison of results of the old (black and new (blue, green) diagnostic. Commissioning of the DED The DED power supply system has been commissioned on a full size dummy load before cabling with the DED coils. Thus, necessary control parameters already were available. Since the dummy load could not simulate several inductive couplings as well as mirror currents in the TEXTOR vacuum vessel an additional commissioning with the real DED coils was required. After tuning the power supply control parameters to fit the real load and tuning of the de-coupling transformer optimum, AC operation without mutual disturbances and without influence on the plasma position could be achieved (Fig. 5). Fig. 5: DED current traces. Operation at 2 khz and 3 ka in shot # degree system (black, inverted), 90 degree system (blue), compensation coils (green). In addition to the originally planned operation of the DED at DC and AC between 1 khz and 10 khz a low frequency mode was prepared and successfully applied. Slow ramping of the DED currents in DC mode results in low frequency field with a DC component overlaid (Fig. 6). 48

49 After finishing the DED commissioning at all frequencies the DED was applied to the plasma for the first time. As expected the influence of the magnetic perturbation field could clearly be seen in a certain pattern of radiation from the divertor target plates (Fig. 7). Fig. 6: DED current traces for operation at 2 Hz and 7 ka in shot # Low frequency operation is produced by ramping in DC mode. 0 degree system (black, inverted), 90 degree system (blue), compensation coils (green). Fig. 7: Influence of the perturbation field on the distribution of radiation in front of the divertor in TEXTOR shot #

50 Improvement of the DED First operating experience showed the need of minor improvements of the DED. At DED operation with plasma arcing occurred to the DED coils. In order to avoid damage of the outer tubes of the DED coils additional insulation has been inserted in a short shut down in In addition to this, grounding resistances have been connected to all tile support structures. This defines the electric potential during glow discharge cleaning on one hand and allows for voltage measurement on the other hand. The value of the resistance has been chosen for limiting the currents flowing during disruptions. Large holes in the liner have been mechanically reinforced by stainless steel bridges during DED installation. Experiences with liner heating showed the necessity to relieve those bridges from the liner heating current. This has been done by adding copper strips in parallel. Preparation of the DED 3/1 mode operation In parallel to the experimental programme with the DED in the 12/4 mode all preparations have been done to allow for first experiments in the 3/1 mode in For this purpose a second set of coaxial cables had to be routed, configured and connected. At the end of TEXTOR operation in 2003 all cables of the 12/4 mode have finally been replaced by the 3/1 mode cables. This was followed by a measurement of the complete inductance matrix of the DED coils including cabling. These results were required in order to adjust the power supply system control to the new load. Estimates of the parameters could only be roughly made in advance, since in 3/1 mode uncompensated feeding of the currents adds considerable couplings to the inductance matrix. Data Acquisition and Processing In 2003, first experiments with the Dynamic Ergodic Divertor (DED) have been performed on TEXTOR. Furthermore, a large part of the diagnostic systems moved from the old fashioned VMS based data acquisition software to the new platform independent Java based architecture JDAQ. The diagnostic data is now acquired under JDAQ control and subsequently transferred to the common storage facility (CSF) for every discharge. Even during experiments with a large number of participating scientists and many diagnostics being online, the access to the data once having been stored at the CSF is very fast. The available disk space of the CSF amounting to 1.5 TByte was not enough for the envisaged aim to store one complete year of TEXTOR experiments. In consequence, additional disk space of 3 TByte has been added. The huge increase of the data volume per discharge stems from the fact that a lot of diagnostics have been upgraded during the TEXTOR shutdown (more channels per diagnostic system and a better time resolution). Furthermore, several new diagnostics each producing additional data in the order of ~ 100 MByte per discharge (and even more) a have been developed and became operational. For the detailed analysis of the data mainly IDL and MATLAB licenses have been bought and installed on the PCs in the TEXTOR control room. These packages are used because the application of these high level data analysis tools drastically reduces the time for programme development due to the large number of ready-to-use functions. Several client libraries have been developed to allow for raw and pre-processed data access using these software packages. For a simple and oscilloscope-like visualisation of raw and pre-processed data a Java based software from Padua University (jscope) is used while being under continuous development. 50

51 Several programmes to calculate physical quantities from the raw data of various diagnostic subsystems were developed. These codes have been implemented into the so called "Chain1 automated intershot data analysis". The intershot control programme launches these programmes and stores the derived physical quantities into the TEXTOR Physics Database (TPD), an Oracle database running on a dedicated server machine. The main process of the intershot data analysis is based on Unix script languages like sh, awk, and make. However, the individual analysis programmes can be written in any standard programming language, like C, C++, Fortran or Java, respectively. The intershot main process starts the various analysis programmes in parallel, making the whole analysis quite fast. For a better control and fault tracking of the intershot analysis log files for every discharge and every calculation step are written automatically and can be accessed with a standard internet browser. For those diagnostics which still need a large amount of manual control a web interface to allow the upload of data files has been developed. A second command line driven interface based on tcl/tk is available, too, and can easily be integrated into the user-side analysis codes. The TPD itself is discharge oriented, being able to store 1- and 2-dimensional data. Support of 3- or more-dimensional data can be implemented easily. Post-processing of already available data is foreseen in future. A version control scheme is available, which keeps track of the various data uploads. The most recent version of a signal will be returned from the TPD server, as long as no special version number is specified. For a fast access, the data is indexed by the discharge number. The number of available signals from the TPD has steadily increased over the year and the TPD as one of the main data stores is now widely used by scientists working at TEXTOR. 51

52 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.9. PLASMA DIAGNOSTICS u.samm@fz-juelich.de Edge Diagnostics Traditionally many of the diagnostics for the investigation of the plasma boundary and surface interaction region in TEXTOR are based on optical techniques. Existing methods in astronomy, low temperature plasma discharges and surface analysis have been tested in laboratory experiments and then adapted to TEXTOR. Moreover, TEXTOR also allows the use of active techniques, where either atoms or laser light is introduced into the region to be diagnosed. intensity / arb. units spectral fit Trot p v0: 3478 K v1: 3058 K v2: 3920 K intensity / arb. units # CD + CII D γ % in the standard range λ / nm Fig. 1: High resolution measurement and modelling of a CD 4 gas puff through a nozzle into a TEXTOR boundary plasma. The passive spectroscopy of molecules and line profile measurements has been considerably improved by the use of an Echelle spectrometer, which allows the simultaneous recording of a wavelength range between 375 nm and 700 nm with a high resolution of λ/ λ Fig. 1 gives an example concerning measurements of molecular emission during a discharge on TEXTOR, which 52

53 demonstrates the possibilities of the instrument. The accuracy allows for a reliable modelling of the spectrum. The visible spectroscopy was considerably upgraded by the installation of several small rack spectrometers, which allow the simultaneous observation of a number of volumes near the limiters and close to the inner bumper in wavelength regions ranging from the air UV (λ 200 nm) up to the near IR (λ 1000 nm). The very flexible test limiter spectroscopy equipment can now also be used for the observation of the spatial impurity distribution during the operation of the Dynamic Ergodic Divertor (DED, see fig. 2) Fig. 2: Relative changes of particle emission intensities on the DED via imaging spectroscopy. A supersonic helium beam diagnostic was developed for fluctuation measurements of the electron temperature and density. The beam injection system that provides a collimated beam with a pulse length of 0.1 s and 2 Hz repetition rate is attached on top of the TEXTOR vessel. The radial profiles of three emission lines are observed by two complementary detector units with a temporal resolution of 10 ms and 10 µs, respectively. The radial observation range of the slow unit covers about ±50 mm around the last closed flux surface (LCFS) at 460 mm. The emission profiles are used for normalisation of the fluctuation signals obtained with the fast unit which passed the final tests in the laboratory. First electron temperature and density profiles were measured with the slow detection systems. Charge exchange recombination spectroscopy (CXRS) applied on a high energetic (50 kev) neutral H diagnostic beam is used for the measurement of radial profiles of ion temperature, impurities and poloidal velocities. The full radial range from centre to 500 mm at the low field side is observed by two spectrometers. The light of the central channels is conducted via fibres onto an entrance slit of a spectrometer. With optical lenses the edge region from 350 mm to 500 mm is imaged onto the slit of the second spectrometer. Additionally, the red and blue Doppler shifted CXR line intensities from top and bottom of the vessel are measured simultaneously by a high resolution spectrometer with 20 spatial channels distributed over the last close flux surface (LCFS). An in-situ measurement of the species distribution has been developed and applied at the hydrogen beam line allowing the determination of the beam density distribution inside the plasma. 53

54 DED diagnostics The R&D work in 2003 mainly concentrated on the implementation and the commissioning of new diagnostics relevant for the DED operation. A set of 18 Langmuir probes were implemented into the divertor target plates of TEXTOR (see Fig. 3). They measure the particle flux and the electron temperature in front of the plates. Thermocouples with high time resolution were implemented into the divertor tiles to measure the heat fluxes. A set of five Hall probes has been mounted on a slow probe drive at the low field side and first measurements were performed. These probes measure the modification of the magnetic field at the plasma edge induced by the DED. A fast IR camera observing the divertor target region was taken into operation. It measures the heat deposition pattern imposed by the DED. The camera has a time resolution of about 75 µs and is therefore suited for higher frequency operation of the DED. The pellet injection system was installed and taken into operation in the laboratory. In collaboration with CEA-Grenoble/France three of the nine barrels were replaced in order to produce pellets of larger size (2.5 mm instead of 1.5 mm). Preparations have been done for the new pellet guiding system, which will be installed in 2004 in collaboration with the Technical University of Applied Physics at St. Petersburg/Russia. A fast valve which was developed in recent years was implemented at TEXTOR. The valve opens within less than a millisecond and releases some hundred millibar-liter of gas into the discharge chamber. First experiments to mitigate disruptions by massive helium puffs were performed successfully. Fig. 3: Langmuir probe for the DED tiles. 54

55 Core Diagnostics A variety of diagnostics is operated at TEXTOR in order to determine the properties of the hot plasma, especially now under the conditions imposed by the Dynamic Ergodic Divertor (DED). These include magnetic diagnostics to measure and control the position of the plasma, the plasma current and the internal structure of the magnetic fields, as well as measurements of the plasma radiation ranging from the microwaves up to the x-rays. In addition, active diagnostics in the microwave and far infrared spectral range are applied in order to determine the plasma density by interferometry and to derive the density profile by reflection of microwaves at the cut-off density. The radiation is measured integrally over the appropriate spectral range in order to determine the total emitted power by bolometry. Furthermore, it is recorded in broad bands in the x-ray region with energies above approximately 1 kev to determine the structure of the hot plasma in the core, and in narrower bands using instruments such as the SPRED VUV spectrometers and solid state detectors in order to measure the impurity content of the plasma. The plasma radiation is also studied by means of high resolution x-ray spectroscopy to determine the plasma parameters in the plasma core. Experience gained with these instruments at TEXTOR has been applied to the design and construction of diagnostics for the next generation of magnetic fusion machines, such as W7-X and ITER. Many of the diagnostics dedicated to the plasma core are used not only for the characterisation of the plasma, but they are also incorporated into the online control system of TEXTOR. The most important ones are the magnetic and interferometric real time control systems for the vertical and horizontal plasma position as well as for the plasma density. Due to the installation of the DED, many of the diagnostics had to be modified and recommissioned in order to ensure a proper operation of TEXTOR. The magnetic systems for the plasma position depend on the measurements of the magnetic fields outside the plasma column. The magnetic fields are determined from long-term integrated signals of magnetic loops. Due to the DED, both the positions of the magnetic pick-up coils, as well as the feedthroughs had to be changed. In addition, the level of electromagnetic interference significantly increases during operation of the DED, as its frequency range directly coincides with the frequencies of the signals to be recorded. The online control systems have been hardened to withstand this new quality of interference and they have been recalibrated. It has been demonstrated that reliable real time positioning of the plasma is obtained even under high electromagnetic interference levels. Especially challenging was the reinstallation of the compensated magnetic loop to measure the plasma energy. In this device, the diamagnetism of the plasma is determined by comparing the magnetic flux in a loop enclosing the plasma with the flux in a loop outside the plasma. Due to the installation of the Dynamic Ergodic Divertor, the geometry of the compensated loop had to be modified and thus, its properties have changed considerably. Taking into account all the unavoidable effects such as misalignments and bendings of the loops due to temperature changes in TEX- TOR, the signal can be corrected using data of the other magnetic field measurements. Now, the signal delivering the total plasma energy is available again. A new set of fast Mirnov-coils has been designed and installed on TEXTOR. The coils are made of molybdenum wire and the windings are isolated by a high temperature spray ceramics. These pickup coils are arranged as close to the plasma as possible. Two different kinds of coils have been mounted in TEXTOR: One stack of coils consists of two coils, one measuring the radial magnetic field component and the other one measures the poloidal field component. 55

56 Fig. 4: Pick up coil with a graphite limiter ring. One coil stack is located in front of the liner close to the plasma and another one is placed behind the liner. An array of eight poloidal coils is fixed at the top of the torus, serving for the toroidal mode number determination, while another array of coils is located in a poloidal cross-section in order to analyse the m-number of MHD instabilities. The coils facing to the plasma are shielded by a small carbon limiter. The highest sampling frequency at the moment is limited to 100 khz and will be increased next year. Most coils have been sampled at a frequency of 25 khz. Fig. 5 shows the effect of different heating methods on the behaviour of the MHD-instabilities. In this figure, the influence of low-power NBI co-injection on the frequency of the 2/1 mode can be seen. The slowing down is in good agreement with a reversal of plasma rotation with co-injection, as has been analysed in the past. The spectrogram shows a broad-band turbulent feature during the phase with ECRH heating. Fig. 5: Spectrogram of a pick-up coil during a TEXTOR shot with NBI and ECRH heating. 56

57 The new Mirnov coils complement the ability of the core diagnostics like ECE, interferometry, and SXR cameras to measure instabilities like sawteeth, sawtooth precursor and postcursor modes, and tearing modes. The aim of the ongoing research is the analysis of the influence of external error fields induced by the DED on the MHD characteristics of the plasma, as well as detailed studies of the penetration and amplification of error fields. For the bolometer system to determine the total radiated power from the plasma, a fully digital coupling between the analog amplifiers and the digital data acquisition system has been introduced. The noise level has been reduced, enabling measurements under the harsh conditions of the DED. The software codes to analyse the data have been rewritten for the new computer systems at TEX- TOR. An example for the local emission of radiation is shown in Fig. 6. The image has been obtained from Radon transformed bolometer signals during a discharge with DED operation, showing a highly radiating spot in front of the DED. Fig. 6: Radiation pattern of a TEXTOR discharge with DED operation. The DED coils are located on the left-hand side, where a highly radiating spot is formed. The joint experiment tangential x-ray camera, which was planned and built in close cooperation with our Japanese collaborators from the National Institute of Fusion Sciences (NIFS) and the Princeton Plasma Physics Laboratory (PPPL) in USA, has been reinstalled at TEXTOR. The major components such as the light guide with a large diameter of 100 mm, the image converter tube to compress the image to the size of the CCD detector and the fast camera system have been supplied by NIFS. Using a newly developed CCD camera, the framing rate could be increased up to about 10,000 frames per second. Impressive pictures of the modes within the plasma, especially the mode with number m2/n1 have been obtained and locking to the externally applied frequency of the DED has been detected. The analysis of the data is still going on. In the meantime, the system has been sent back to NIFS and has been reinstalled to the Large Helical Device (LHD) experiment. The poloidal x-ray camera system has been used for a long time to determine the mode structure and the islands at magnetic surfaces with rational q in the plasma. It has been upgraded to higher time resolution in order to detect islands caused by the DED with a sufficiently high sampling frequency. The positions and the widths of the modes could be detected, both for the operation of the DED with DC and AC current. It is now being equipped with new front end amplifiers to reduce 57

58 noise and pickup of electromagnetic interference. A new data acquisition system based on PXI modules is currently being installed to provide a sampling rate of up to 200 khz over the full discharge. A prototype of the CO 2 dispersion interferometer for TEXTOR has been built at the Budker institute in Novosibirsk and has been tested at the gas dynamic trap (GDT). On TEXTOR, a further developed instrument will be installed in the second half of Diagnostics FOM-Instituut voor Plasmafysica In this field projects have been concluded that received preferential support by EURATOM under the special rules for equipment of smaller associations applied on preferentially supported larger devices of another association. Three systems were accepted in 1998 by EURATOM: 1. A double-pulse high-resolution Thomson scattering diagnostic; 2. An upgraded version of the ECE imaging diagnostics tried out at RTP; 3. A wavelength-selective ultra soft x-ray tomography (USXT) diagnostic. The Thomson scattering system was already operational in 2000 and contributed significantly to the scientific output of the FOM-group at TEXTOR. The success was so convincing that a completely new project has been started in 2002 together with the German colleagues of IPP and Russian colleagues from the Ioffe Institute, St. Petersburg. The aim of the project is the development of a 10 khz burst-mode operated Thomson scattering system. A new laser system has been developed that is able to generate three bursts of pulses per plasma discharge. The repetition rate of the pulses within a burst is up to 10 khz and the energy for each individual pulse is J. The spectrometer has been equipped with fast CMOS cameras that can cope with the high repetition rate. The whole system has been set up in such a way that 120 points along the laser chord are sampled with a spatial resolution of 7.5 mm. The laser system has been installed at TEXTOR and successfully tested for one burst of pulses. First Thomson scattered light from TEXTOR has been observed with the new system in December The ECE imaging system in its originally planned form has been successfully concluded and gave a lot of input to a PhD thesis that was defended in November The original ECE-I system monitored an array of 16 sample volumes that are aligned along a vertical chord. In 2003 the system was upgraded to a system featuring a two-dimensional coverage of the plasma. It is now possible to monitor the electron temperature profile and its fluctuations in a two-dimensional matrix of sample volumes (8 radial x 16 vertical). The spatial resolution of the system is approximately 1 cm in the poloidal plane and the time resolution is up to 1 MHz. The ECE-I system is fully integrated with a Microwave Imaging Reflectometer that monitors the density fluctuations at 16 positions along the cut-off surface. Both systems were installed and tested at the end of A short movie has been produced about the evolution of the electron temperature profile during a sawtooth in TEXTOR. The work in the field of microwave imaging is done in tight collaboration with physicists from UC Davis and Princeton, USA. In 2003 the ultra soft x-ray tomography (USXT) system was almost completed. The system that has been procured by a consortium of the Dutch consultancy firm Phystex, the Ioffe Institute (St. Petersburg) and the Institute for Applied Physics (Nizhny Novgorod) and by the end of the year it was almost ready for installation onto TEXTOR. The system will deliver the 2-D emission distributions of several spectral impurity lines at the same time. The spectral window can be varied to detect lines in the wavelength range from 0.3 to 12.5 nm. The system will be used for the study of the penetration of impurities into the plasma in space and time. 58

59 Another exiting diagnostic that was installed in 2003 with normal EURATOM support is a 20 channel Motional Stark Effect (MSE) system to measure the evolution of the q-profile as a function of time. The system has been developed in close collaboration with Phystex and has come into operation by the end of the year. The first data that have been taken are quite promising and we are confident the system will play an important role in the transport physics programme of the FOM team at TEXTOR. Diagnostics LPP ERM / KMS The diagnostic for the determination of Zeff from bremsstrahlung in the visible range has been upgraded. The detection system was changed from a tilting mirror with a photomultiplier (repetition time 150 ms) to a CCD camera with 21 viewing chords. In contrast to the previous system, the simultaneity of the detection of bremsstrahlung along all the 21 chords now allows Zeff to be determined also for transient phases. A few channels have been left available for future use with photomultipliers in order to allow for a data acquisition with a higher time resolution than being possible with cameras. From the line-integrated signal of a central chord a line-averaged Zeff is computed, presently every 100 ms (this can be reduced by a factor of two), using also density and temperature profiles from the HCN-interferometry and ECE diagnostics. The line-averaged Zeff is available online. The system also allows the Zeff profile to be derived, and work is being done to improve the accuracy of the profiles via statistical signal processing techniques. A new diagnostic for the detection of C III and C V spectral lines has been installed. The system is now operational at TEXTOR. The brilliance of the two spectral lines is detected simultaneously along nine lines of sight, being located on the high field side of the tokamak. The repetition time for the 18 signals (9 for C III and 9 for C V) at present is 100 ms, but work is being done to increase the time resolution to about 1 ms. This diagnostic has been especially designed to evaluate the influence of the Dynamic Ergodic Divertor on the radiation and transport properties of the intrinsic carbon. The electrical probe system has been upgraded to measure potential and density fluctuations with a bandwidth of 1 MHz. The system is built in such a way that various schemes are available to concomitantly measure fluctuations of the density, temperature and the plasma potential. In a first application long space correlation and zonal flows were measured using the existing rake probe. Furthermore, a new probe head to measure Reynolds stress has been designed and is presently under construction. Together with the new electronic set-up, questions on sources and sinks of rotation can now be tackled. The detection of both slowing down and escaping alpha particles remains one of the crucial and most delicate issue in reactor grade plasmas. The lack of established techniques to measure these particles has been recognized as one of the diagnostic weaknesses of the tokamak community, requiring further R&D in the ITER perspective. At the moment no diagnostic is available to measure the alpha particle losses. We have continued our developments in this field. Our approach is mainly based on the activation technique. To obtain some preliminary confirmation of the technique's capability, we did test experiments on the JET tokamak during the C9 campaign. Here, the particle drift was opposite to the normal direction, and thus the particle fluxes collected should be particularly high. To this end, the reciprocating probe shield cap has been suitably prepared and exposed to D-D plasmas using the vertical manipulator on top of the machine. The shield cap material is made of 59

60 boron which is a suitable material for the subsequent activation reaction 10 B(p,α) 7 Be. Protons escaping from the plasma with energies above about 1.5 MeV produce the element 7 Be, which is a radioisotope with a half-life of 53.3 days. Preliminary results from the JET C9 campaign are very promising. Gamma spectra taken after exposure during the whole campaign clearly show a line at 477 kev which is associated with the radioactive 7 Be decay. The preliminary signal analysis reveals that the 7 Be line intensity is found to be equal to the calculated intensity expected from classical losses of MeV fusion protons within a factor of two. For the development of position sensitive detectors for TEXTOR new measurements with nuclear track detectors have been performed in a collaboration with the Soltan Institute at Warsaw, Poland. Clear distinction is made between impact of protons and Tritium. The analysis of position dependence is in progress. 60

61 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.10. CONTRIBUTIONS TO ITER w.biel@fz-juelich.de Within the TEC collaboration, IPP Jülich is working on several scientific tasks which are directly related to the development of ITER diagnostics. Furthermore, IPP Jülich is preparing to take over significant contributions to the detailed design of ITER heating systems and work is going on for the development and qualification of materials for plasma facing components (see also the report of the Institute for Materials and Processes in Energy Systems, IWV). Diagnostics In the course of 2003, the available expertise and resources for the planned ITER diagnostic contributions have been reviewed within the TEC, and the ongoing ITER diagnostic planning as performed by the international diagnostic working group (DWG) has been supported. TEC has confirmed its continued interest to take over the responsibility for the construction of several ITER diagnostic systems or to participate therein, including the possibility to take over the assembly of an ITER diagnostic port plug. The actual selection of diagnostic packages systems to be worked on by TEC will depend on the results of the DWG negotiations among the ITER partners. In 2003 a number of tasks have been devoted to the ITER diagnostic design. One of them was a design study of the poloidal polarimetry system for ITER, while another required the evaluation of the potential performance of active (CXRS) spectroscopy using the Diagnostic Neutral Beam (DNB) in combination with upper port viewing in ITER. A third task addressed the Motional Stark Effect measurements using the DNB for current density profile determination in ITER. Within the frame of the fourth task, work has been performed on the optical design of the VUV and X-Ray spectroscopy for ITER. Finally, tests for a new technique for alpha-particle measurements at ITER were performed at JET. The polarimeter task was aimed at resolving remaining design issues associated with the conceptual design developed during studies performed under earlier EFDA contracts, in particular assessing the compatibility of the in-vessel optics with the ITER port plug. Apart from the retro-reflector, the issues that have been considered were the possibility to have a transmission line with only two focussing components per line, of which one (the second mirror) is placed inside the port plug. It has been investigated whether or not it is possible to have an automatic alignment procedure to direct the laser beam onto the retro-reflector by a mirror that is positioned as far as possible from the plasma. More particularly, it was studied which mirrors would be the most ideal to be used for scanning. A design study was done concerning the alignment and calibration methodologies for initial installation and during subsequent operation. Specifically, the tolerable angular and lateral misalignments of the various optical components were studied to see whether the methodologies foreseen are realistic. Furthermore, a conceptual design of a double vacuum barrier was made. Finally, 61

62 in cooperation with V. Voitsenya (Kharkov, Ukraine) specific laboratory tests have been done to check the reflective properties of retro-reflectors at micrometer wavelengths (in particular at 118 µm, which is the envisaged wavelength for the polarimeter) after having been exposed to deuterium ions from a plasma source. The work done in the field of active CXRS using the Diagnostic Neutral Beam included an optimisation of the upper port viewing system based on physics arguments, a demonstration of the feasibility to implement such a viewing system, a motivation for an equatorial viewing port system for poloidal and toroidal velocity measurements, an assessment of the physics merits of a negative ion source based DNB compared to a positive ion source system, a study of the physics merits for a tilted DNB reaching the ITER magnetic axis, and a description of pilot experiments and drawing-up of a plan for further development work. Layout of top and equatorial CXRS and MSE periscopes (courtesy IT Garching) In the field of MSE for ITER an initial feasibility study and assessment was made of its implementation in ITER. In collaboration with a number of other Associations an assessment was made of the relative merits and limitations of MSE on the DNB and on the heating neutral beam. Within the VUV-/X-ray spectrometer task, the optical design of 6 VUV spectrometers for ITER has been finished, which in total cover the wavelength range from 2.3 nm to 160 nm, divided into 6 different wavelength ranges with some overlapping. The use of MCP detectors in the ITER VUV spectrometers has been analysed based on experience from spectrometers at JET and TEXTOR and this detector type is found to be appropriate, provided that the detector location is behind the radiation shielding. The wavelength ranges chosen for the VUV instruments allow to monitor many different ionization stages of all relevant impurities in ITER with sufficiently high wavelength resolution and to identify all impurities in the plasma which have significant concentrations. The physics measurement possibilities for the X-ray diagnostic have been analyzed in detail. It has been found that the design for the x-ray spectroscopy at ITER can be made such as to allow for a simultaneous determination of the ion temperature profile, the ion rotation profile, the profile of the relative abundance of different ionization stages and the profile of the electron temperature. Considerations about the optimization of mirror/crystal materials and reflection angles have been started. A new technique was proposed in 2002 to measure alpha particle losses on ITER, based on charged particle induced activation. To document this technique, a probe shield cap was exposed to D-D plasmas using the vertical manipulator on JET. The results are very promising as the preliminary 62

63 signal analysis of the radiation emitted by the exposed sample agrees in magnitude with that calculated assuming classical losses of fusion protons. Heating For the ITER Ion Cyclotron Heating and Current Drive system, in line with the continued activity of LPP-ERM/KMS in this field, an original antenna and matching system has been proposed and studied. Contrary to the reference system, which requires the installation of tuneable vacuum capacitors inside the vacuum vessel, the system proposed here is tuned externally, therefore requiring no movable parts inside the vessel. This matching option, based on the same resonant double loop also called conjugated T tuning concept as the reference ITER system, is shown to perform better and to have a number of substantial advantages when compared to it. The design group of IPP Jülich provided the 3D-CAD models for the conceptual design of this new ICRF antenna. The task also included adaptation of the new system in the ITER environment. Closely related to the previous topic, LPP-ERM/KMS is also leading the ITER-like ICRF Antenna project on JET. The design phase of the antenna system is now complete and the procurement phase is under way. The new antenna is planned to be installed in JET and operated in Still related to the ITER antenna activity, a new antenna system has been designed and constructed for TEX- TOR to be compatible with the inlet of a diagnostic beam. This antenna is conceived such as to allow testing the conjugated-t mode of operation. The new antenna was tested electrically and will be installed on TEXTOR early in For the development of the ECRH system for ITER, a team of 6 full-time professionals of the Dutch TEC partner FOM is involved in the design of the upper-port launcher. Contrary to the conventional front-steering system, where mirrors rotate very close to the plasma, FOM is developing the remotesteering launcher. Here, the scan in the vertical plane is achieved by means of rotating a mirror far away from the plasma, launching the mm-wave beam into a square waveguide, resulting at the end of the waveguide in the same scanning angles. The scan range is now of ± 12 degrees at the input and output of the square waveguide resulting in a scan range of ± 6 degrees to ± 8 degrees in the ITER plasma, depending on the focusing strength of the end mirror. Modelling As an ongoing long term support of the ITER team the B2-EIRENE plasma edge and divertor modelling code is permanently supported by IPP Jülich. In 2003 in particular the implementation of radiation transport in the EIRENE code was further developed. It must be expected that due to its size the dynamics of the ITER divertor will be significantly affected by opacity effects of the Lyman line radiation (distinct from present experiments). Implementation of these extensions into the ITER-team version of B2-EIRENE is currently underway. Due to its strong B-field also the effects of Zeeman line splitting on resonance line opacity has to be investigated, rendering emission and absorption highly anisotropic. The related code extensions have been started. Several edge modelling studies in the last years have pointed out several rather critical divertor design aspects related to a rich chemistry of the hydrogen private flux plasma (e.g. in-out asymmetries in the divertor flow, pumping, etc.) This has led to a major revision of related databases for H, H -, H +, H 2, H 2 + and H 3 + in 2003, involving now a large number individual reaction channels and their energetics. The new database has been established and published (FZJ-report Jül-4105, Dec. 2003). It will shortly be made available to the ITER team in electronic from through the EIRENE code database formats ( 63

64 Plasma surface interaction (PSI) Besides the ongoing physics programme special ITER related PSI topics are under evaluation in the framework of EFDA technology tasks. One task is the development of ITER relevant diagnostics to measure material deposition and fuel retention, in particular in the non-activated phase, to assess the fuel retention that gives the basis for the material choice in the later T-phase. The possibilities of laser desorption/ablation combined with in-situ spectroscopy detection of desorbed fuel and ablated material is tested in laboratory experiments and TEXTOR test limiter experiments. Special PSI effects associated with the use of castellated surfaces as foreseen in ITER to reduce the mechanical stress (10 x 10 mm with 0.5 mm gaps) are the subject of investigations in TEXTOR using macro brush ITER like wall structures. The analysis concentrates on edge effects with respect to power loading, arcing, material deposition and fuel retention in gaps, both in erosion and deposition dominated regions. The ERO code has been applied to ITER divertor conditions and geometry to model and predict erosion, deposition and T retention. In addition in a special task the change of optical properties of mirrors by plasma erosion and material deposition is investigated. Molybdenum and tungsten mirrors have been used in the SOL of TEXTOR in erosion and deposition dominated zones and the change of their optical properties was measured by means of reflectometry. 64

65 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.11. CONTRIBUTIONS TO WENDELSTEIN 7-X a.pospieszczyk@fz-juelich.de The stellarator concept is the most promising alternative to the tokamak because of its inherent stationary plasma operation. The prospect of stationarity opens new possibilities to investigate reactorrelevant physics issues. However, it also requires additional solutions for the accompanying technical problems which are for instance related to the superconducting field coils, the durability and cooling issues of wall elements as well as the control and data analysis of diagnostics. FZJ participates in the design and construction of diagnostics for the stellarator Wendelstein 7-X which is presently being built at Greifswald by taking over a large work package. In the future, FZJ will also participate in the scientific exploitation of the project. Within a co-operation between the Max-Planck-Institute for plasma physics (IPP) and the Forschungszentrum Jülich (FZJ) the task for the superconducting bus system (the superconducting current connections between the solenoid coils and to the current supplies) was adopted. The technical specifications (1-AAH-S0002.1) are the basis for the construction, manufacturing and assembly of the superconductors and the appropriate holders and supports. For the performance of these tasks an appropriate hall space was rented and prepared according to the requirements. An overall concept of the project was prepared with the goal, to optimize the working steps and simplify the assembly. For a checking of the bending results as well as the examination of assembly a 1:1 model was developed. The individual parts are presently manufactured. For an optimum compensation of the magnetic fields due to the current flow as well as for simplification of assembly a new topology was developed. A partition of the bus systems results in avoiding collisions with other parts, reduced space requirements for integration and facilitates transport. For the qualification of different working steps test setups were manufactured and first examinations were accomplished. These are setups for - insulation checks including measurements in the Paschen minimum, - thermal tests of the behaviour under cryo-temperatures at 77 K, - mechanical bending loads as well as high helium pressure tests to check quench situations, and - vacuum compatibilities of the materials and methods used. The detailed design work for a set of new VUV/XUV spectrometers has been performed, which shall be used for impurity monitoring and impurity transport studies at the stellarator Wendelstein 7-X. The optical design for numerically optimised toroidal holographic diffraction gratings is finalised and the mechanical design for the spectrometers is developed in great detail. Many peripheral components of the spectrometers such as vacuum systems, detectors and calibration light sources are specified and procurement is ongoing. Delivery of many components is expected during 2004, while the construction and laboratory testing of the spectrometers is scheduled to be finalised in

66 A high energetic hydrogen beam for diagnostic is foreseen on Wendelstein 7-X for the measurement of ion temperature profiles. A diagnostic injector will be developed, the beam of which provides an equivalent current of more than 5 A at 60 kev. During the whole duration of injection (10 s) the beam properties (divergence < 0.5º, particles with full energy > 70 %) should be maintained. The pulse duration shall also cover the phases with additional neutral particle heating and will with a pulse frequency of at least 0.5/min also allow measurements during very long discharges. An additional beam modulation of 100 Hz will help to discriminate active and passive signal intensities and to improve the quality of the data. The European tender action for the high voltage power supply has been finished. The contract, together with the fabrication of the neutralizer chamber will be given to the Budker Institute of Nuclear Physics (BINP) in Novosibirsk, Russia. Additionally, the grid structure for higher current densities and lower beam divergence of the ion optics has been optimised. The divergence of 0.35 o for a beamlet was verified in experimental tests at BINP. In 2003 the disposition of the available budget for Wendelstein 7-X diagnostics was again discussed and updated. A main part of the scientists and technicians have been moved to the construction department so that as a result the reduced resources led to another concentration on the most necessary parts, which are either necessary for the start-up, time critical or already in contract form. In particular there was a reduction of the present efforts in the edge and divertor diagnostics. Shifted to a low level, however not cancelled, were the high-resolution X-ray imaging spectrometer and the Target Tile Manipulator. It was agreed that the devices can be fitted later to the vessel without too much additional effort and that for the latter, the necessary preparations would be made in the divertor modules. A thermo-stress analysis of various materials proposed as a window for spectroscopic observation systems from the hot plasma for the stellarator W7-X has been performed (IWV). Window materials studied include fused silica, crystal quartz, magnesium fluoride, calcium fluoride, zinc selenide and sapphire. The calculations have shown that even for a large window (about 13 cm of diameter) sapphire appears to be the appropriate choice for visible/infrared optical systems which can successfully survive under a maximum radiation power load of 50 kw/m 2 during about 20 minutes discharge length which is expected for W7-X. Other materials such as fused silica, MgF 2, and ZnSe can only be used for small diameters such as 5 cm because of their high temperature radiation which may disturb the measurements. CaF 2 is unacceptable for such windows because of a strong in-plane distortion during the long heat load. However, the investigated optical materials may be usable for a short pulse duration or low power load. The plasma facing components in thermonuclear fusion devices are subjected to intense fluxes of charged and neutral plasma particles and radiation. Resulting from these plasma-wall-interaction processes the materials will be degraded with respect to their thermal and mechanical properties. A major aim of the activities is to develop and fabricate new materials for Wendelstein 7-X and to characterise and to test them under simulated operation conditions, i.e. at thermal loads up to 20 MWm -2. To evaluate the component behaviour and the resulting material damage under Wendelstein 7-X relevant conditions high heat flux simulation tests are continuously being performed with a powerful electron beam (JUDITH, hot cells at FZJ) and with ion beam test facilities (MARION, IPP at FZJ). These experiments are focussed on different design options of high heat flux components with carbon, pure B 4 C, Si-doped B 4 C and tungsten armour. 66

67 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ B. GENERAL PROGRAMME ON TEXTOR B.12. CHARACTERIZATION OF MATERIALS AND COMPONENTS FOR PLASMA/WALL INTERACTION j.linke@fz-juelich.de E , E , E , E The plasma facing components in thermonuclear fusion devices are subjected to intense fluxes of charged and neutral plasma particles and radiation. Resulting from these plasma-wall-interaction processes the materials will be degraded with respect to their thermal and mechanical properties; in addition wall erosion is another critical issue which has significant impact on the lifetime of plasma facing components and on the contamination of the fusion plasma. The plasma facing materials in future fusion devices are primarily based on beryllium, boron, carbon or silicon as well as tungsten in combination with copper as a heat sink. A major aim of the R&D activities is to develop and fabricate new materials for future fusion devices such as ITER or Wendelstein W7-X and to characterise and to test them under simulated operation conditions, i.e. at thermal loads up to 20 MWm -2 and at neutron fluences up to approx. 1 dpa. Development of graded W-Cu joints The character of the interface between plasma facing material (PFM) and heat sink has significant influence on the lifetime of plasma facing components under intense cyclic heat fluxes. In particular these are the promising material candidates: tungsten (PFM) and copper (heat sink) which exhibit substantial differences in their thermal and mechanical properties. To reduce inherent stresses which originate from these mismatches, graded interface structures have been developed. Two different processes have been selected to produce functionally graded materials (FGM), namely laser sintering using the blown powder process and low pressure plasma spraying. For both processes composite materials with a wide variation of the W/Cu-ratio have been manufactured successfully. The resulting test samples have been utilized to provide a database with thermal and mechanical properties; these data were used to optimise the geometry and therefore the performance of graded interfaces in divertor-components by finite element methods. Hence, beside the materials selection and the geometry the type of joining technology has strong impact on the quality and the robustness of the joint. To face these challenges the choice was made on HIP-technologies combined with electro-chemical methods for adding diffusion-bonding layers. Test modules, based on experimental and computational results, have been produced; they have been analyzed by non-destructive testing methods and will be exposed to high heat fluxes applied by electron beam scanning. The outcome of these tests will provide information for further optimisation. 67

68 Dust formation during transient thermal loads Intense energy is deposited on localized areas of the plasma facing materials under transient thermal loads such as edge localized modes (ELMs), plasma disruptions or vertical displacement events (VDEs). The generation of dust during these events is a critical concern both for carbon based and for metallic components. Isotropic fine grain graphites and multidirectional carbon fibre composites are damaged by brittle destruction; this process results in a macroscopic erosion forming carbon dust particles with diameters up to approx. several ten microns. The threshold for the onset of the brittle destruction process has been investigated in electron beam simulation experiments for pulse durations ranging from ms. In addition, erosion scenarios have been evaluated on pure tungsten and other refractory alloys. For these materials the thermal erosion is initiated by strong evaporation, the convection of melt and boiling processes. Melt layer instabilities can generate significant amounts of activated W-dust in future fusion reactors. In addition, the formation of a vapour cloud above the heat affected surface has been observed in electron beam simulation experiments. From screening tests on different high- Z materials, pure tungsten was found to show the highest resistance against thermal shocks; castellated structures are very beneficial to reduce the formation of thermally induced cracks compared to monolithic structures. Fig. 1: Recrystallized surface of a pure sintered W sample and particle trajectories under transient thermal loading (E abs 6.2 MJm -2, t 4.4 ms). Non-destructive testing of high heat flux components by IR methods New diagnostics have been developed to determine the integrity of joints between the plasma facing material and the heat sink in actively cooled divertor components by non-destructive methods. Infrared thermography was chosen as a relatively cheap and fast, easy to handle non-destructive method for the detection of defects resulting from the manufacturing procedure. A special facility IRINA (IR Inspection for Non-destructive Analysis) was configured. This facility is based on the principle of infrared thermography. It consists of a heating system, an IR camera and several control units. The water flow through the component to be tested is supplied by two water circuits with well defined and controlled temperatures (95 C and 20 C) and flow rates. Several actively cooled plasma facing components were investigated in order to study the heat transfer peculiarities of different component configurations: monoblock and flat tile geometry, different combinations of armour and heat sink materials, geometric effects such as component size, 68

69 and arrangement of the cooling tubes. It was found that the component design limits the application of infrared observation of heat transfer differences caused by structural defects. The method of infrared heat transfer inspection was elaborated and criteria to judge the modules quality based on comparative analyses of temperature gradients on the modules surface during heating were worked out. The optimal parameters for the detection of areas with lower heat transfer properties were defined (e.g. areas A1 and A2 in Fig. 2). A2 top A1 side face 2 Fig. 2: Areas with lower heat transfer properties in a CFC-monoblock component (FT113-1) observed in the facility IRINA after a heating period of 3 s (top: side face 1, centre: plasma facing surface, bottom: side face 2). The electron beam facility JUDITH was applied to investigate the degradation of the heat transfer from the plasma facing surface to the heat sink under fusion specific thermal loads up to approximately 20 MWm -2. This method also allows to study the evolution of structural imperfections caused by thermal fatigue. Intense cyclic thermal loads applied to CFC flat tile modules result in an irreversible thermal fatigue damage of the module; the resulting surface temperature of the components changed in two steps: firstly, the surface temperature increases gradually with cycle number. Secondly, thermal cycling leads to a rapid temperature increase (during a few cycles). Beryllium flat tile modules in general, did not shown the heat transfer reduction during thermal fatigue loading. The steady state temperature remained almost constant during the cyclic loading until complete failure occurred within one single additional cycle. Thermal response of windows for Wendelstein 7-X In the stellarator experiment Wendelstein 7-X, maximum continuous radiation power loads of 50 kwm -2 on plasma-facing components of spectroscopic observation systems take place during about 20 minutes. Under these thermal loads some of the foreseen materials may degrade or achieve unacceptable temperatures which may disturb the spectroscopic measurements. 69

70 Temperature, C SiO 2 (fused) MgF 2 ZnSe Al 2 O L10 mm diameter, mm Fig. 3: Diameter dependence of the maximum temperature of window materials for 20 minutes of 50 kwm -2 of power load. Design parameters are: diameter varied, thickness L 10 mm. In order to develop and demonstrate the thermal performance of a window designed to operate with visible and IR sensors under long thermal loads of 50 kwm -2 from short-wavelength radiation, finite element method calculations have been carried out. Window materials considered in this study include fused silica, crystal quartz, magnesium fluoride, calcium fluoride, zinc selenide and sapphire. Calculations show that large quartz, CaF 2 and MgF 2 windows may be degraded or destroyed due to the intense incident-radiated power for a long pulse duration. The absorption of a large fraction of the radiated power in a window material for a long pulse duration together with a low thermal conductivity of the window lead to unacceptable temperatures and thermal stresses. Among the considered optical materials only sapphire remains within an acceptable temperature during high heat flux loading. Thermo-stress analyses do not indicate any crack formation of sapphire during the above mentioned power load even for a large window (about 13 cm of diameter). Other materials such as fused silica, MgF 2, and ZnSe can only be used for small diameter windows such as 5 cm because of their high temperature radiation which will disturb the measurements. CaF 2 is unacceptable for the protective window because of strong in-plane distortions during long lasting thermal loads. However, all investigated optical materials may be usable for shorter plasma pulses or reduced power densities. New diagnostics for JUDITH As mentioned above, the generation of dust during off-normal plasma scenarios is a critical concern in future confinement experiments such as ITER. Melt layer instabilities e.g. during plasma disruptions can generate significant amounts of activated W-dust. On the other hand, carbon based materials are damaged by brittle destruction. Brittle destruction results in a significant material loss which is associated with the formation of carbon dust particles. To diagnose the mechanism of macroscopic erosion such as brittle destruction, optical measurements have been applied to fine grain graphite and CFC (Carbon Fibre reinforced Composite) during thermal shock experiments in the electron beam test facility JUDITH. A photodiode array detected thermally radiating particles passing through observation volumes that were situated along the beam axis at 52 and 69 mm above the surface. The time evolution of light emission from released particles indicated that the particle release processes were clearly different between graphite 70

71 and CFC. The signals in graphite show continuum and sharp peaks that correspond to small (i.e. binder phases of graphite) and large particles (i.e. graphite grains) released from the loaded surface, whereas, the signals from CFC show only one species of released particles. Taking into account the time lags between the signals from two photodiodes, the velocity of released particles was derived. The mean speed of the small particles released from graphite was estimated to be about 300 m/s at 2.4 GW/m 2, increased with decreasing power density and decreased at elevated temperatures. This results from the variation of the mean size of the released particles. The maximum speed of the large particles was estimated to be 330 m/s under these particular conditions. According to the detailed analysis, the large particles are released from graphite at rather shallow angles and they seem to be rotating rapidly and slowly sublime during their flight. The release-angle and rotation of the released particles should be related to the release processes from the surface. Design of a new electron beam test facility JUDITH 2 Plasma-facing materials and components in future tokamaks (such as ITER) will suffer higher heat loads than in existing machines. In order to study the effect of these high heat loads, the electron beam testing facility JUDITH is currently utilised for testing of high heat flux components of next step fusion devices. However the progress in the understanding of plasma physics have revealed new kinds of short term events (ELMS), which can only insufficiently be simulated in the existing facility. Additionally the demand of electron beam testing has strongly increased. For these reasons a decision was taken to build up a second electron beam facility JUDITH II. The electron beam of JUDITH II will have a maximum power of 200 kw. The acceleration voltage, important to reduce volumetric heating, can be stepwise chosen between 30, 45 and 60 kv. Maximum power density at a beam diameter of approx. 5 mm will be around 10 GW/m². The beam deflection of ± 250 mm allows investigations on very large components (0.5 x 1 m). Short term events can be simulated by either using short pulse lengths (< 1 ms) or by optimising the local retention period. This very flexible beam control additionally allows the combination of cyclic static and transient (disruptions, VDEs and ELMs) heat loads within one experiment. In order to increase the capacity, a 150 kw experimental cooling circuit is designed to allow parallel testing (screening and thermal fatigue) of up to four mock-ups. The new facility is planned to be installed in September Starting of regular experimental campaigns is foreseen for the beginning of

72 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ C. TECHNOLOGY PROGRAMME C.1. CHARACTERIZATION OF MATERIALS AND COMPONENTS FOR PLASMA/WALL INTERACTION j.linke@fz-juelich.de E An essential material related issue of the technology programme is the development of plasmainteractive components for next step devices (ITER) and future electricity generating fusion reactors. Major aim of the research programme at FZJ is the characterization of plasma facing materials and actively cooled components and their assessment with respect to their thermo-mechanical and neutron irradiation behaviour. Physical and mechanical properties of plasma facing materials High-Z, high temperature alloys (W or Mo) are considered to be primary material candidates for divertor structural applications. However, these materials suffer from embrittlement at low temperatures after irradiation although the DBTT depends strongly on the interstitial element concentrations and the material processing. Investigations to improve, produce, characterise and join this type of materials have to be carried out. For the testing of creep properties two optimised tungsten grades (pure W and La 2 O 3 -doped WL10) have been ordered and provided by Plansee AG, Reutte; cylindrical creep test specimens have been machined. The microstructure of the as received material was analysed. The WL10 exhibited a very homogeneous intragranular distribution of the La-oxides. The mean grain size number in cross section perpendicular to the forging direction was between 4 and 5 for WL10 and 7 for W (DIN 50601, res. ISO ). Creep tests at 1000 and 1300 C in a helium/hydrogen atmosphere up to 5000 hours have started Fig. 1: Tensile strength of the 3-directional carbon fibre composite NB31 for different fibre orientations in dependence on local density variations. tensile strentgh [MPa] pitch-direction specified min. strength value PAN-direction needling-direction 0 1,77 1,82 1,87 1,92 1,97 density [g/cm³] 72

73 Undoped 3-directional CFC grade NB31 has been characterized with respect to its mechanical and thermal parameters for all three orientations. Flat tensile test specimens have been used for the pitch- and PAN-fibre direction; short cylindrical samples had to be used for the so-called needling direction. The obtained data both for tensile strength and Young s modulus show a clear correlation with local density variations of the carbon fibre composite. Measurements on Si-doped CFCs (NS31) are in preparation. (TW3-TVM-CFCQ2, TW3-TTMA-00) High heat flux tests of Be coated primary FW mock-ups High heat flux tests have been performed on two Primary First Wall (PFW) mock-ups with beryllium as a plasma facing material, a water cooled copper alloy heat sink, and a SS316L stainless steel back-plate. In these samples the beryllium covered a surface area was 86 x 96 mm 2 and had a thickness of 10 mm. In order to reduce the stresses, the Be plate was segmented into four tiles. The joint between plasma facing material and heat sink was produced by hot isostatic pressing (HIP). Different HIPing parameters (temperature, pressure, time) and interlayers were used for both mockups and for some of the individual tiles. Both mock-ups were tested in the electron beam facility JUDITH. They withstood thermal fatigue for 1000 cycles at 1.5 MW/m 2, but when the power density was increased to 2.0 MW/m 2 they failed after a few cycles. These failure limits were similar to the ones found for other designs of first wall components, before. Additionally, the effect of neutron irradiation on the thermal fatigue performance of Be/CuCrZr joints in actively cooled PFW mock-ups is investigated in thermal fatigue experiments in the electron beam facility JUDITH. After finishing the irradiation campaign, the samples will be high heat flux tested up to 3 MW/m 2 to check the remaining engineering margins of the irradiated Be/CuCrZr joints. In a first step, five PFW mock-ups underwent screening heat flux tests to check the quality of the joints. On the basis of the results, some samples were selected for the neutron irradiation campaign. Fig. 2: Primary First Wall (PFW) mock-ups consisting of beryllium armour tiles, an actively cooled CuCrZr-heat sink and a stainless steel back-plate (316L). In addition to the screening, two mock-ups (cf. Fig. 2) were tested under thermal fatigue conditions as a reference to the irradiation samples. On both mock-ups, all tiles withstood 500 cycles at 1.5 MW/m 2 and additional 200 cycles at 2.0 MW/m 2. But when the loads were increased to

74 MW/m 2, one of the small tiles failed in both experiments. Although the tests were continued on the remaining tiles, the failure limits were reached between 2.5 and 2.7 MW/m 2. (TW2-TVP/PFCFT, TW2-TVV-FWMUHF, TW3-TVB-HFTEST) Thermal fatigue and thermal shock behaviour Worldwide, several electron beam and ion beam facilities are involved in high heat flux testing of plasma facing components for next step fusion devices. Up to a certain degree these machines are comparable, but differences concern e.g. beam generation, beam sweeping, calibration techniques and diagnostics. In order to assess the influence of these different machine parameters on the results of the high heat flux experiments, a round robin test has been initiated by the EFDA team. Experiments were supervised by FZJ; they have been carried out on the following testing facilities: FE200 (Framatome, Le Creusot, France) EBTS (Sandia Nat. Lab., Albuquerque, NM, USA) TSEFEY (Efremov Inst., St. Petersburg, Russia) JET NBI Testbed (JET, Culham, UK) JUDITH (FZJ, Jülich, Germany) In this testing campaign, a set of specially produced actively cooled CFC monoblock mock-ups has been loaded in the different facilities at comparable power densities. The temperature response during these loadings on the surface (IR cameras, pyrometers) and inside the mock-ups (thermo couples) has been registered and used as a criterion for comparison. Furthermore finite element calculations have been carried out for the temperature response with respect to the power densities. Most of the surface temperatures were found in a relatively narrow scatter band. Only EBTS showed somewhat higher temperatures compared to the other machines. At higher power densities, the JET-NBI data are on the upper side of the scatter band. This may be explained by the peaked beam profile in this machine. (DW0-TV1/02, JWX-FT-6.3) Mechanical characterization of CuCrZr-SS-joints Tubular joints between CuCrZr and 316L steel with a galvanic deposited interlayer produced by ANSALDO are tested in tensile internal pressure and fatigue tests. Eight samples were tested under internal water pressure (cf. Fig. 3). Failure pressures were found to be in the range between 475 and 725 bars, according to hoop stresses between 260 and 400 MPa. In most cases the samples failed in the joint between CuCrZr and the nickel adapter. Only the two samples with the highest failure pressures failed in the copper. 74

75 (TW3-TVD-CUSS) Fig. 3: Pressure test samples with tubular joints between CuCrZr and 316L stainless steel. Neutron induced material degradation In order to study the influence of fast neutrons on the behaviour of high heat flux components for ITER, miniaturized samples have been neutron irradiated in the High Flux Reactor (HFR) in Petten. Neutron fluences were 0.2 dpa in carbon / 0.15 dpa in tungsten (irradiation campaign PARIDE 3) and 1 dpa in carbon / 0.6 dpa in tungsten (irradiation campaign PARIDE 4). Irradiation temperatures were 200 C approximately for all samples. The irradiated mock-ups as well as un-irradiated reference samples were tested under thermal fatigue conditions in the electron beam facility JU- DITH. They were loaded up to several thousand cycles and later they were cut and investigated by hot metallography. The failure limits of CFC flat tile and tungsten macrobrush mock-ups are slightly reduced after irradiation. This is especially valid for the tungsten macrobrush. Due to its geometry, the monoblock design has proven to be the most robust solution for plasma facing components even after neutron irradiation. In addition to these thermal fatigue experiments, the degradation of thermal conductivities for plasma facing materials has been investigated. Samples from CFCs and tungsten alloys have been tested. Thermal diffusivities were measured by a laser flash method, heat capacities by differential scanning calorimetry (DSC). The thermal conductivities are strongly reduced after neutron irradiation (especially for carbon). The degradation starts at a relatively low irradiation level and increases with the neutron dose. At higher temperatures the loss of thermal conductivities is less serious. (GB8 - DV6) 75

76 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ C. TECHNOLOGY PROGRAMME C.2. CORROSION RESISTANCE OF FUSION RELEVANT C-BASED MATERIAL IN PRESENCE OF AIR AND STEAM k.kuehn@fz-juelich.de PES-FZJ Task GB9-V65 Postexaminations of oxidized carbon based first wall materials were performed [1]. Fig. 1 contains a view of unoxidized and oxidized NS31 (siliconized 3D CFC) with its fibre and matrix; the oxidation was done within the facility THERA in steam at 980 C up to an overall burn-off of 80 %. There are indications, that the matrix is preferently oxidized in case, that external mass transfer limitations are not present. This is probably due to the comparatively low temperature treatment of the matrix in NS31 (and NB31, too), which enhances to some extent the oxidation rates, as already reported in the Annual Progress Report A material similar to NS31 but improved with respect to matrix temperature treatment was developed shortly by the manufacturer Snegma and should be examined concerning oxidation resistance in next future; a drawback of this promising new material concerning application in fusion facilities is however is its still extraordinarily high price. unoxidized oxidized Fig. 1: Appearance of NB31 (upper row) and NS31 (bottom row) before and after oxidation in steam at 980 C. 76

77 Some limited experiments on ozone interaction on TEXTOR a-c:d flakes were performed in order to find out the optimum temperature for removal of tritium containing a-c:d layers in future fusion devices: There is a competition between oxidation rates of the flakes by ozone and ozone thermal decay rates, both increasing with temperature. For the facility OZOX used, we found that the optimum temperature is with 200 C only slightly higher than used in earlier experiments (185 C), documented in the Annual Progress Report It has to be noted, however, that in real environments catalytic influences by metals may lead to faster ozone decay. Further on, ozone formation in presence of air may result in additional NO x which changes the oxidation behaviour. In addition, one has to bear in mind that ozone decay on the way between formation and application may be different in OZOX compared to a real fusion device. Altogether, additional experimental parameter studies are required in this field. [1] Kühn, K.; Hinssen, H.-K.; Moormann, R. (2003): Behaviour of C-based materials in contact to oxidising gases. Proceedings of ICAPP '03, Cordoba, Spain, May 4-7, 2003, Paper 3031, CD-published. 77

78 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ C. TECHNOLOGY PROGRAMME C.3. MECHANICAL PROPERTIES OF FUSION MATERIALS p.jung@fz-juelich.de These tasks comprise investigations on the effects of the transmutation products hydrogen and helium on elastic and mechanical properties and microstructure of materials in the first-wall and blanket of a fusion reactor. The materials investigated are iron, reduced activation martensitic stainless steels, tungsten and ceramic materials. Implantation of hydrogen and helium ions by the Jülich compact cyclotron CV28 is used to simulate the loading by transmutation. In the case of hydrogen also loading from plasma and from gas phase is employed. Long Term Programme / Task Area: Materials Development Subtask: TW3-TTMS-003b (Compatibility with Hydrogen and Liquids) D5: H-He effects on mechanical properties (tensile, hardness) and fracture surface TEM examination dose He [dpa] EUROFER Fig. 1: Ultimate tensile strength of helium implanted EUROFER97 effect of additional hydrogen implantation. σ UTS [MPa] appm H (0.008 dpa) appm H Timpl 150 C (0.08 dpa) 250 C 325 C C 550 C open:ttest25 C 200 filled:ttesttimpl c He [atppm] 78

79 ε UTS [%] EUROFER97 Timpl 150 C 250 C 325 C C 550 C open:ttest25 C 3 filled:ttesttimpl appm H (0.08 dpa) 300 appm H (0.008 dpa) dose [dpa] Fig. 2: Strain at ultimate tensile strength of helium implanted EURO- FER97 effect of additional H implantation c He [appm] EUROFER97 specimens of 100 µm thickness are homogeneously pre-implanted with helium at temperatures from 125 C to 500 C to concentrations of 300 appm. Subsequently or after an annealing treatment at 550 C or 750 C (to induce clustering of the helium atoms), hydrogen is implanted at 125 C or 250 C to concentrations of 300 and 3000 appm. The specimens are tensile tested at temperatures from room temperature to 500 C and fracture surfaces are analysed by scanning electron microscopy. Figs. 1 and 2 show the effect of hydrogen on ultimate tensile strength σ UTS and corresponding strain ε UTS : While after hydrogen implantation to 300 appm no change in σ UTS is observed, 3000 appm H cause a significant decrease. ε UTS is reduced also at the lower H content, mainly when tested at higher temperatures. Investigation of He-implanted plus annealed specimens is in progress. EUROFER97 shows after implantation of helium at 250 C to 2500 appm increased strength and reduced ductility. An important result is, that after annealing for 10 hours at 550 C or 750 C the increased strength is almost retained, while ductility recovers almost to the value before implantation. Underlying Technology Field: 2. TV Vessel in Vessel Task Area: 2.1.TVP Plasma Facing Components Tungsten as a candidate material for high heat flux devices, will experience radiation damage from neutrons, inducing atomic displacements and compositional variations by nuclear transmutation. The latter will produce large quantities of the light elements hydrogen and helium. Helium, atomic or clustered, will be effectively retained up to temperatures close to melting and will have strong effects on hydrogen retention and mechanical properties. To simulate this introduction of helium, sheets of 99.96% purity tungsten were polished to about 140 to 148 µm and were cut to dog-bone shaped tensile specimens by spark erosion. The specimens were homogeneously implanted with 3 He ions by using a degrader wheel with Al foils of appropriate thicknesses which varies the energy in front of the specimens from 0 to 34. MeV Tensile testing was performed under 10-3 Pa vacuum at a strain rate of /s at 25 C and at 450 C. Fracture surfaces were analysed by scanning elec- 79

80 tron microscopy (SEM). Reduction of area ( A/A ) of the foil specimens was approximated by the ratio of neck width w neck to total thickness d by: A/A 1 w neck / d. Fig. 3 shows stress-strain curves of specimens implanted at 900 C to various He concentrations and tested at room temperature and 450 C, respectively. A decrease in strength is also observed, when the specimen is hold at 900 C without implantation. In this case also a loss of ductility occurs which is seen in Fig. 3 only above 160 appm He. Fig. 4 gives a compilation of strains at ultimate tensile strength after implantation of 3 He and tensile testing. The most intriguing result is the increase of ε UTS at low He concentrations. Fracture analysis by SEM shows only faint, if any, changes by implanted helium. σ [MPa] W T impl 900 C c He [appm] 600 T 400 test [ C] : : ε [%] Fig. 3: Tensile stress-strain curves of tungsten after implantation at 900 C to 3 He concentration up to 500 appm and tensile testing at 25 C (solid lines) and 450 C (dashed), respectively. ε UTS [%] dose [10-2 dpa] W Fig. 4: Strain at ultimate tensile strength of tungsten after implantation of 3 He and tensile testing at 25 C ( ) or 450 C ( ) for implantation temperatures of 125 C (empty), 300 C (empty-crossed), 450 C (filledcrossed) and 900 C (filled), respectively. Lines indicate trends at both testing temperatures for implantation at 125 C (dashed) and at 900C (dotted), respectively c He [appm] 80

81 In conclusion, the strain maximum at low helium concentration deserves further investigation. Given the negligible effect of implanted helium on fracture surfaces, no significant changes of microstructure are expected, which could be perceivable by TEM. The results on helium implantation will be compared to implantation of deuterium and of deuterium in material pre-implanted with helium. A detailed comparison of the results after implantation to results from neutron irradiation on mechanical properties will be performed when the results from deuterium and from He + D implantation are available. 81

82 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ D. PARTNERS OF THE IEA TEXTOR IMPLEMENTING AGREEMENT D.1. JAPAN noda@nifs.ac.jp Measurement of DED-induced magnetic islands with a tangentially viewing soft x-ray camera K. Toi 1, S. Ohdachi 1, G.Fuchs 2, S. von Goeler 3 1 NIFS, 2 TEC, 3 PPPL Aims A tangentially viewing soft x-ray (SX) camera system is a useful tool to study detailed structures of magnetic islands and perturbations caused by MHD instabilities and/or externally applied magnetic fields such as the DED field. This diagnostic technique has a high potentiality concerning measurements of island structures and the island-induced transport. Work performed In 2003, the tangential viewing soft x-ray camera (TSXC) succeeded in getting two dimensional images of MHD instabilities in the Large Helical Device (LHD). This has clearly demonstrated a potentiality of a powerful diagnostic tool for studies of MHD instabilities and magnetic islands. In the middle of this fiscal year, the TSXC was installed on a tangential port in TEXTOR for studies of sawtooth events and plasma perturbations induced by DED fields. The soft x-ray images with photon energies above 1 kev were measured with TSXC, having a framing rate of 4.5 khz (full-frame mode) and 13.5 khz (half-frame mode). This high speed SX-camera can detect low frequency MHD modes such as sawtooth precursors and plasma perturbations induced by DED fields rotating with up to several khz. TSXC measurements were performed with DED modes of m/n 12/4 and m/n 3/1 (m, n: poloidal and toroidal mode number). Although TSXC did not detect clear perturbations during the m12/n4 mode of operation, a rotating m/n 2/1 structure having the same frequency ( 1 khz) with the DED coil current was detected in the m/n 3/1 mode. The amplitude of observed SX perturbations is not in proportion to the DED coil current, but suddenly enhanced when the applied DED field exceeds a certain threshold. A preliminary estimation of the full width of the observed m2/n1 rotating islands suggests them to be larger than 10 cm. A more detailed analysis of the SX images is being carried out incorporating other diagnostics, in order to determine the island size and to understand the formation process and the impact on plasma transport. Work planned The TSXC system will be applied to LHD plasmas in the next fiscal year, for a test of further upgrading the system. We have a plan to replace the present fast framing camera of 8 bits dynamic 82

83 range of sensitivity with a new high framing rate camera with 10 to 12 bit dynamic range. If this plan is realized, the upgraded TSXC will be able to measure more detailed structures of perturbations. After testing of the upgraded camera in LHD during the experimental campaign in 2004, it will be applied to various DED experiments planned for the year Application of BIXS to the analysis of tritium retention in JET divertor tiles M. Matsuyama and Y. Torikai, Toyama University Aims (1) To examine the detailed distribution of tritium retained in a complete poloidal set of divertor tiles in JET. (2) To examine the effects of x- and γ-rays on a measuring technique of β-ray-induced x-ray spectrometry (BIXS). Work performed (1) Tritium distribution in a complete poloidal set of divertor tiles A complete poloidal set of 10 divertor tiles transferred from JET after the deuterium-tritium experimental campaign (DTE1) was analyzed. Nine spots for each divertor tile (1IN1 to 1ON10) were investigated systematically. The effect of tile location in the divertor on the tritium retention was particularly investigated. These results are compared to the combustion measurements obtained from surface and bulk specimens from the same set of tiles. Surface and bulk (up to 1 mm in depth) tritium concentrations are obtained by BIXS. The results showed that the tritium distribution is different not only from tile to tile but also between the surface and the bulk tritium concentrations at the same tile. The tritium distributions over the surface obtained by BIXS are different from the corresponding ones obtained by the combustion method. On the other hand, the bulk tritium concentration obtained by BIXS correlates well with the corresponding ones obtained by the combustion method. The data of the present work show that for the 1 mm depth range, BIXS provides a more accurate tritium distribution profile than combustion. It was concluded that the results give a good overall image of the tritium distribution over the surface and bulk of each tile. (2) Effects of X- and γ-rays on BIXS measurements The effects of a massive γ-radiation background on the measurements of x-ray spectra were examined. As the first test, two γ-sources were used, i.e. 137 Cs and 60 Co. Measurements with both sources showed that the presence of the γ-field has no influence on the shape of the x-ray spectra, although the whole intensity of the observed x-ray spectra increased with activity of a γ-source. To simulate the in-vessel environment of the tiles more accurately, one of the divertor tiles still having its metallic part was subjected to BIXS. Although the metallic part was gamma active, no significant effects were observed for the recorded x-ray spectrum. This indicates that BIXS can be used without a special shielding of the x-ray detector, even in presence of a strong γ-field as being expected in ITER. 83

84 Plasma-wall interaction of low activation ferritic steel K. Tsuzuki, JAERI Aims Low activation ferritic steel is one of the candidates for the structure materials of a fusion demonstration reactor. However, it is a ferromagnetic material and its vacuum properties are worse compared to stainless steel. For this reason, the compatibility of ferritic steel with plasmas had to be investigated in the JFT-2M tokamak at JAERI. In the case of JFT-2M, global parameters such as the total radiation loss were not affected by the installation of ferritic steel. However, the local impurity release behavior has not been investigated. In addition, local magnetic fields may affect plasmawall interaction. The main purpose of this work is as follows: (1) To examine the response of low activation ferritic steel under plasma exposure with spectroscopic measurements. (2) To examine the influence of (local) magnetic fields on plasma-wall interaction with heat and H emission profile measurements. Work performed The limiter head was fabricated in 2002 at JAERI and sent to TEXTOR. The low activation ferritic steel, F82H, was buried into the limiter head made of stainless steel (SUS304) in order to compare the plasma-wall interaction of SUS304 and F82H directly. It was installed on TEXTOR's limiter lock no 3 and then exposed to plasma shots ranging from no to (17 discharges in total). Two-dimensional measurements with D β, Fe I, C II, and W I filters were performed to measure the distribution of influx and impurity release. Surface temperatures were also measured by IRTV. During the experiments, the bulk temperature of the limiter was continuously raised from 420 C to 620 C in order to investigate the effect of ferromagnetism on the flux distribution. Work planned The detailed analysis has just started. Line intensities of Fe I will be compared for F82H and SUS304, normalized to the C II line or to the D β line. Flux distributions on the limiter will be investigated as a function of bulk temperature to show the effect of ferromagnetism on the flux distribution. The behavior of melted layers will also be studied. As for the additional experiments, there are no plans at present. PMI studies related high-z materials in TEXTOR T. Tanabe, Nagoya University Aims (1) Investigation of the behavior of high-z impurities in main and boundary plasmas. (2) Understanding of local phenomena (sputtering, reflection, redeposition, etc.) in front of a high-z limiter surface. (3) Examination of high-z materials behavior exposed to plasma heat load. 84

85 (4) Influence of the DED on impurity production and its local transport. (5) Tritium distribution analysis on PFM tiles by the imaging plate technique. (6) Simulation of PMI. Work performed (1) CVD/W limiter heat load test CVD/W (0.5 mm) on a Cu limiter was examined in the TEXTOR tokamak as a test limiter to study the performance under high heat loads. The influence of the limiter on the plasma was observed to be the same as a sole tungsten limiter. Owing to an unintentional high heat load (probably more than 20 MW/m 2 was absorbed at the limiter), local melting of the Cu substrate was observed, which also did not show an appreciable effect on the plasma. Analysis of spectroscopic data for hydrogen recycling and impurity transport is now proceeding. (2) Gas puffing through a small hole in a W limiter To study PSI in front of a high-z limiter, gas puffing through a small hole of 1 mm diameter in a W limiter was performed. Gases of D 2, O 2 and 13 CH 4 were introduced. No strong effect on the W behavior was observed. Preliminary studies clearly showed different deposition patterns concerning thickness and shape for W and C limiters under 13 CH 4 puffing. Further analysis of the isotope distribution is in preparation. (3) Utilization of ferritic steel (F82H) as a first wall material A stainless steel limiter a part of which was replaced by ferritic steel (F82H) was exposed to TEXTOR plasmas in order to investigate the influence of local magnetic fields on PMI concerning recycling and impurity behavior. The effect of heat loads onto the limiter was also examined. Owing to its low melting temperature, the limiter was melted and the behavior of the process was analyzed by a TV camera. Analysis of spectroscopic data and postmortem analysis are under preparation. (4) Modeling of erosion and deposition patterns on W and Ta limiters exposed to the TEXTOR edge plasma The erosion and deposition patterns on W and Ta test limiters exposed to the TEXTOR deuterium plasma containing a small amount of C impurities are simulated with the modified EDDY code. At the very top of the W and Ta limiters, there occurs neither erosion nor deposition, instead the erosion proceeds slowly along the surface. When approaching the edge, the surface is covered by a thick C layer, which shows a very sharp boundary similar to the observation in surface measurements. In the erosion zone, the re-deposited C forms a W (Ta)-C mixed layer with small C concentration. Assumptions for chemical erosion yields of 0.01 for W and < for Ta fit the calculated widths of the deposition zone to the experimentally determined values. Possible reasons for the difference between W and Ta are discussed. Work Planned (1) Continuation of high-z limiter tests using different kinds of materials. (2) Artificial hot spot experiments with laser irradiation. (3) Gas puff through a hole in the high-z limiter using methane, oxygen and hydrogen. (4) Investigation of heat load and local erosion and deposition with brush limiters. (5) Simultaneous measurements of several Balmer line emissions using parallel detectors. 85

86 (6) Influence of the DED on impurity production and its local transport. (7) Carbon deposition in plasma shadowed areas and tile gaps. Penning Gauge Spectroscopy on LHD H. Funaba, National Institute for Fusion Science Aims (1) Partial pressure measurement of hydrogen (H) and helium (He) in the vacuum vessel of the Large Helical Device (LHD). (2) Partial pressure measurement of noble gases, such as neon (Ne), argon (Ar), krypton (Kr), xenon (Xe) and so on, which are injected from the gas puff system. Work performed (1) Installation of the optical and spectroscopic system. The penning gauge of LHD is located on an outer port of the vacuum vessel. A window for observation, optical fiber bundles, which transport the observed light from the experimental room to the diagnostics room, and a visible spectrometer have been installed. (2) Measurements during the glow discharge cleaning. The systems were tested by observing the penning discharge during the glow discharge cleaning in LHD. The wavelength was calibrated by several lines of Ne I from 585 nm to 706 nm by using the data which were acquired during the neon glow discharge cleaning. (3) Calibration by hydrogen and helium gases. For the calibration of the partial pressure measurement, the vacuum vessel of LHD was filled with some neutral gases. In the case of H 2 gas, H α and many lines from H 2 were observed. In the He gas case, three lines of He I (587.6, 667.8, nm) ware observed. The intensity of H α and the nm line of He I were measured with varying gas pressure. The intensity of He I is almost proportional to the helium pressure. On the other hand, the intensity of H α tends to saturate at an H 2 pressure region higher than Pa. This may be because of the fraction of the dissociation of H 2 became small at the high pressure region. From these results, it is considered that the partial pressure can be measured by penning gauge spectroscopy in the pressure range below Pa for hydrogen and below Pa for helium. Work Planned (1) Improvement of the system. In order to measure with high temporal resolution and in the wide pressure range of Pa, an improvement of the optical system and the detector is needed. Some other penning gauges which have a more optimal structure for the observation should be tested. (2) Measurement in the plasma experiment. The partial pressure measurement by penning gauge spectroscopy will be carried out during the main plasma discharge in LHD. 86

87 Millimeter-Wave Imaging Diagnostics A. Mase, Kyushu University Aims To develop a 2D-3D millimeter-wave imaging system for measurements of temperature and density profiles and fluctuations. Work performed (1) Dr. Y. Kogi visited the Institut für Plasmaphysik at Forschungszentrum Jülich in order to participate in the millimeter-wave imaging experiment conducted by the UCD/PPPL and FOM groups at TEXTOR. The imaging system is a combined one consisting of ECE imaging (ECEI) and microwave imaging reflectometry (MIR). The imaging array consists of dual-dipole antennas and Schottky barrier diodes bonded between the antennas, which shows excellent performance in both the E-plane and the H-plane. The ECEI system measures 128 spatial points (2 x 8 x 8) and the MIR system measures 64 spatial points (2 x 8 x 4). The sawteeth oscillations and their precursor oscillations have been observed with good signal-to-noise ratio in ECEI system. The calibration experiment of the whole ECEI array has now started by using a hot source. In the MIR system, the method of injecting a probing beam is still a major issue. The curvature of the wave front at the cutoff layer must be matched to the curvature of the cutoff layer. It is planned to use an adaptive array antenna for the injection. (2) A numerical comparison of conventional reflectometry and MIR systems has been performed. The simulations reveal the ability of the two-lens imaging system to obtain a high correlation between the reflectometer signal phase fluctuations and the shape of corrugation for a receiver location far from a target, where the conventional reflectometer signal phase is strongly distorted by interference effects. Work planned (1) One of the group members will visit TEXTOR to participate in the ECEI/MIR experiment again. The physics-oriented experiment is now expected due to the improvement of the system. (2) The simulation study of MIR will be continued. The comparison between the simulation and the experimental results obtained in TEXTOR is planned. Edge Plasma diagnostics A. Tsushima, Yokohama National University Aims Measurement of the edge plasma. Work performed The proposed axisymmetric ion sensitive probe has been improved in order to measure the plasma flow velocity with good spatial resolution in addition to the measurement of ion temperatures. Since 87

88 the spatial resolution of the probe is the distance between two electrodes facing each other, which is considered to be several times the ion gyro radius being in the order of 1 cm it is expected that the probe can be used near a solid wall or a limiter. Work planned To design the axisymmetric ion sensitive probe for measurements of tokamak edge plasmas. 88

89 ANNUAL PROGRESS REPORT 2003 / ASSOCIATION EURATOM FZJ D. PARTNERS OF THE IEA TEXTOR IMPLEMENTING AGREEMENT D.2. CANADA boucher@inrs-emt.uquebec.ca Studies of Plasma Flows with a Gundestrup Probe in the TOMAS toroidal device C. Boucher, A. Litnovski Investigations of plasma flows are of prime importance in physics understanding of edge and divertor plasmas of fusion devices. An electrostatic probe array, called Gundestrup, is being successfully used in several tokamaks and is presently recognized as a powerful tool in flow measurements. However, theoretical models employed to define the flow velocity from Gundestrup data need to be validated. In particular, the ability of models to derive the component of the flow velocity being oriented perpendicular to the B-field needs a validation in dedicated experiments. Such an experiment has been initiated on the TOMAS toroidal device (FZJ, Germany) as a joint project of FZJ/IPP and INRS (Canada) in a framework of the TEXTOR-IEA collaboration programme. The participants from Canadian side were: Prof. C. Boucher, the group leader, Dr. A. Litnovsky, who was leading the experiment, Mr. Ph. Sicard, a master-student, and Ms. I. Uytdenhouven, a summer-student at INRS. A new combined Gundestrup/Langmuir probe system has been designed and built at INRS for use at TOMAS. The system is capable to perform measurements of basic plasma parameters, such as the electron temperature and density, simultaneously with flow measurements. The new probe system is shielded against microwaves ( ECR plasma generation in TOMAS) and protected against sputtering of probe collectors. The system has two easily exchangeable probe heads with flat/curved collectors so that all existing theoretical models can be validated. The complete software package including acquisition and analytic programmes has been developed and is presently in use. Normalized ion saturation current, a.u. B Fig. 1: The Gundestrup/Langmuir system with exchangeable probe heads. Fig. 2: The polar diagram of ion saturation currents collected by the Gundestrup probe. 89