Fusion Plasma Physics Annual Report Division of Fusion Plasma Physics School of Electrical Engineering KTH

Transcription

1 Fusion Plasma Physics Annual Report 2007 Division of Fusion Plasma Physics School of Electrical Engineering KTH Stockholm, May 2008

2 Division of Fusion Plasma Physics School of Electrical Engineering KTH Stockholm, Sweden 2

3 Contents 1. Introduction EXTRAP T2R experiment Plasma diagnostics Active MHD mode control system Fusion research projects Active MHD mode control Improved steady-state error and transient response Reference-spectrum RWM feedback control simulation and design Active in-vessel coils and a conducting wall for AUG Non-linear MHD dynamics Spontaneous QSH regimes Large periodic fluctuations PPCD and active MHD control Plasma-wall interactions Fuel removal by ICRF-assisted discharges in H 2 -N 2 plasma Material mixing and compound formation on tungsten limiters Collector probe: Ion flux measurements in SOL Material Migration Studies by Tracer Techniques First Mirror Test ITER-Like Wall Project Plasma diagnostics Theoretical fusion plasma physics RF-induced rotation Ripple experiments on JET Assessing effects of ripple on ITER Assessing ITER diagnostics for fast ions Minority Ion cyclotron current drive for sawtooth control Fast particle excitation of global Alfvén eigenmodes Anisotropy drive for geodesic acoustic and tornado modes Production of fast deuteron tail by ICRH accelerated 3 He ions Computational methods for fusion plasmas Chaos and self-organization Plasma application projects Plasma based waste treatment International collaborations Education and research training Undergraduate education Graduate education Publications Personnel List of staff and students Membership, honours, responsibilities Academic and expert activity Income & Expenditures

4 4 Fusion Plasma Physics Annual Report 2007

5 1. Introduction The ultimate goal of fusion research is to construct and operate a power-generating system which is being developed as an energy source for future generations. The Fusion Division is a part of the EU fusion program carried out under an agreement between EURATOM and the Swedish Research Council (VR). There are two main areas (1) providing support and R&D needed to construct ITER during the 10-year construction period and (2) preparation for exploitation of the device when the operational period starts in The Alfvén Laboratory group is involved in both areas. The projects involve work carried out on the in-house EXTRAP T2R device as well as activity at JET (Joint European Torus), the European fusion experiment located in England and in collaborations with other major European fusion laboratories. The research carried out at the Fusion Division is focused on the following areas: Magnetohydrodynamics, plasma turbulence and confinement Plasma wall interaction and development of wall materials and diagnostics. Radio frequency wave interaction with the plasma for heating and current drive and plasma transport modelling. MHD, Turbulence and Confinement (EXTRAP T2R) The experimental research programme on MHD and turbulence is mainly carried out on EXTRAP T2R, and in participation in joint projects at large European experimental facilities such as JET and ASDEX Upgrade. The theoretical programme aims at understanding of improved confinement through the development and commission of new mathematical and numerical tools for studies of both MHD and kinetic effects. The MHD mode control system on EXTRAP T2R (installed in stages ) enables world-leading research in active control of unstable modes. The measurements of turbulence in the plasma edge are internationally recognised and a multi-point measurement project is planned for The experimental programme on T2R has lead to collaboration with experiments including JET (UK), ASDEX Upgrade (Germany) and RFX (Italy). The employment and improvement of existing numerical confinement codes in collaboration with Univ. of Madison as well as the development of new, semi-analytical tools for initial-value problems provide understanding of enhanced confinement scenarios. Plasma Wall Interaction & First Wall Materials The major issues to be tackled in the field of Plasma-Wall Interactions (PWI) are: (i) lifetime of plasma-facing materials (PFM) and components (PFC), (ii) accumulation of hydrogen isotopes in PFC, i.e. tritium inventory; (iii) carbon and metal (beryllium, tungsten) dust formation. The work realized at Division in PWI and fusion-related material science has been fully integrated with the international fusion programme. It is demonstrated by the continuous participation in the: (i) European Task Force on PWI and EFDA Technology Programme, (ii) EFDA-JET Programme including the ambitious ITER-Like Wall (ILW) Project, i.e. operation with full metal wall from 2010, (iii) work within International Tokamak Physics Activity (ITPA) and 5

6 Implementing Agreements of International Energy Agency (IEA); (iv) training of young scientists in PWI: lecturing at summer schools and tutorial articles. The main stream of research has been fully concentrated on issues relevant for a reactor-class device (e.g. ITER) with the central goal to provide the best data for predictions of material behaviour in such a machine. This activity comprises: (i) determination of material erosion, migration and re-deposition; (ii) assessment of fuel retention in tokamak and development of fuel removal techniques; (iii) broad characterisation of PFC from present-day tokamaks; (iv) testing of materials (beryllium, carbon-based composites and high-z metals) including also the development of wall materials for ILW at JET. Design, manufacture and application of diagnostic tools for in-vessel use. Experiments have been carried out at JET and TEXTOR tokamaks, whereas the study of plasma-exposed materials are done in a very well equipped and steadily upgraded accelerator laboratory in Sweden (home laboratory). RF-Heating, Modelling and Theory The research area of RF-heating includes modelling and development of RF-heating as a tool to control the plasma. The Group is strongly involved in the exploitation of JET, the European Topical Group on Heating and Current Drive and the Task Force on Integrated Modelling. The group has proposed and carried out experiments on JET to control the plasma by inducing plasma rotation, driving currents and heating. Participation in the experiments as scientific coordinator and RF-modellers. First successful experiment on driving electron current with the fast wave has been carried out by our group. Codes developed to model RF-heating are validated against JET; two unique codes for self-consistent modelling of heating and current drive have been developed in collaboration with JET: PION and SELFO. The PION code is now the standard code for routine analysis of ion cyclotron heating on JET. SELFO is a code for advanced modelling. Leading roles in the European Topical Group of Heating (board member) and Current Drive and the Task Force on Integrated Modelling in the area of RF-heating, current drive and fast particle effects (project leader for this part of the project). The work involves coordinating the development of codes that are to be the standard in Europe and for modelling ITER and proposing experiments at the European facilities. The group is two new codes for RF-heating a more advanced code than PION for routine analysis and an upgraded version of the SELFO code to include a more realistic plasma cross section, better numerical algorithms in order to better study the effect on the confinement and degradation of heating efficiency due to interaction with Alfvén waves. Development of experimentally validated codes for prediction of the plasma performance has a high priority in the EFDA Work Plan for preparation of ITER operation. 6

7 2. EXTRAP T2R experiment The Division of Fusion Plasma Physics is host for the EXTRAP T2R reversed-field pinch device. The research program at the EXTRAP T2R device includes the following areas: MHD instability control. Non-linear MHD dynamics. Plasma turbulence. Plasma-wall interaction. Figure 2.1. The EXTRAP T2R experiment at the Alfvén Laboratory The front-end system of the EXTRAP T2 experiment (including the vacuum vessel and the conducting wall) was replaced during a rebuild in order to have reproducible high quality plasma pulses substantially longer than the magnetic flux penetration time of the conducting wall. The goal was to provide a platform for studies of Resistive Wall Modes (RWMs), a type of MHD instability that appears both in the RFP and the advanced tokamak configuration with a resistive wall. EXTRAP T2R was then upgraded during in collaboration with Consorzio RFX, Padova, Italy. A comprehensive system for active direct magnetic feedback control of non-axisymmetric MHD modes was developed and installed at EXTRAP T2R. The Active MHD Mode Control System installed has capabilities that in conjunction with other attractive features of the EXTRAP T2R experiment provides excellent capabilities for research on active control of MHD instabilities. 7

8 One of the special features of the EXTRAP T2R device is the close-fitting thin shell used for MHD stabilization, Fig The shell time constant for magnetic flux penetration is short compared to the plasma life time, which enables study of resistive wall mode stability and control methods. The shell is manufactures from thin copper plate in two layers of 0.5 mm each. Figure 2.2. Assembly of the EXTRAP T2R front-end system at the Alfvén Laboratory. The upper half of the copper shell is lowered into place on the toroidal vessel. Table 2.1. EXTRAP T2R machine and plasma parameters Parameter Notation Value Unit Major radius R 1.24 m Minor radius a m Wall diffusion time τ v 6.3 ms Plasma pulse length τ d <100 ms Plasma current (typical) I p 100 ka Plasma electron temperature (typical) T e 300 ev Plasma electron density (typical) n e 1x10 19 m -3 8

9 2.1. Plasma diagnostics A number of plasma diagnostics systems are installed on EXTRAP T2R. The main emphasis is on magnetic diagnostics. A total of 900 magnetic sensors have been installed on the vessel surface inside the conducting shell. These sensors are part of a comprehensive diagnostic system of studies of MHD instabilities and active MHD mode control. The list of diagnostics installed include Magnetic coil arrays for study of MHD activity Electric and magnetic probe array for turbulence studies. Collector probes for plasma wall interaction studies. VUV and visible spectroscopy. Thomson scattering. Interferometer. Neutral particle time-of-flight diagnostic. Bolometer array. SXR camera. Magnetic coil arrays for study of MHD activity MHD activity is studied with an extensive set of magnetic sensors, placed between the vacuum vessel and the conducting shell, consisting of pick-up coils for the measurement of the poloidal, toroidal and radial components of the magnetic field (integrated or not) at 4 poloidal and 64 toroidal positions for a total of 768 sensors. The data acquisition system limits the number of signal that can be collected in each pulse to one component of the magnetic field in the 2 64 or 4 32 configuration, resulting in mode resolution m = 1, -31 n 32 and m = 1, m = 0, -15 n 16 respectively. Thomson scattering A single-point single-time Thomson scattering system provides the absolute value of the electron temperature in the ev range and the uncalibrated electron density in the plasma centre. 2.5 n e (0) ( m -3 ) t (ms) 9

10 Neutral particle time-of-flight diagnostic The system, based on the time-of-flight technique, provides the ion central temperature as well as the charge-exchange neutral fluxes Γ CX ( m -2 s -1 ) T i (0) (ev) t (ms) Bolometer array An 8-chord bolometric system, based on gold thin film bolometers, is used to measure the radial profile of plasma radiation losses in the UV-SXR wavelength range with a time resolution of 1 ms. From the emitted radial profile the plasma emissivity is calculated via a tomographic inversion method f (MW m -2 ) p/a 10

11 Interferometer A single line-of-sight two-color interferometer (CO 2 and HeNe) is used to measure the line integrated electron density along the plasma diameter. The two-color system is required for operations with active MHD mode control in order to compensate for mechanical vibration that becomes dominant in long pulses operation. 100 # I p (ka) n e ( m -3 ) t (ms) SXR camera The system consists of 12 line-of-sights that look at the plasma from the outboard side towards the inboard side covering 80 % of the plasma poloidal cross section. The SXR camera is used to study fast MHD phenomena (QSH, PPCD and dynamo) with a maximum sampling frequency of 100 MHz. 11

12 2.2. Active MHD mode control system The active MHD mode control system installed on EXTRAP T2R is based on extensive arrays of active coils and sensors distributed over the toroidal surface as shown in Fig The system has been developed in collaboration with Consorzio RFX. Sensor flux loop inside shell (blue): norm. radius: r c /a=1.08 coil span: 90 o poloidal, 5.6 o toroidal Active saddle coil outside shell (red): norm. radius: r c /a=1.3 coil span: 90 o poloidal, 11.2 o toroidal Figure 2.3. Two-dimensional arrays of sensor flux loops and active saddle coils installed at EXTRAP T2R. The main features of the active control system are: Sensor array A two-dimensional array of magnetic sensors at 4 poloidal and 64 toroidal positions, measuring all three magnetic field components is available - a total of 4x64x3=768 magnetic sensors. Typically, radial field sensors at 4x32=128 positions are used. Active coil array A two-dimensional array of active coils at 4 poloidal and 32 toroidal positions is installed outside the resistive wall - a total of 128 coils, providing a full cover of the wall.. Saddle coils and sensor flux loops are pair-connected at each toroidal position to form 64 independent m=1 coils and sensors. Power amplifiers A total of 32 power amplifier units with 2 channels each are installed, providing 64 amplifier channels in total. Professional audio amplifiers are used with output power of Watt and bandwidth 1 Hz to 25 khz. Amplifier output currents are up to 12

13 20 A, providing 800 At in the power coils and a maximum radial magnetic field at the coil centre of about 3 mt. Integrated digital controller module The system is contained in a VME bus crate and includes ADCs for analog input of 64 magnetic sensor signals and 64 coil current signals, Board with PPC CPU, 500 MHz and 512 MB RAM, DACs for analog output of 64 amplifier control voltages. Controller algorithms are implemented in software. Several control schemes are used at present, e. g.: Intelligent shell : 64 independent sensor-coil systems with PID- controllers for active suppression of the total m=1 radial flux at the wall. Mode control : Spatial FFT of 64 sensor signals and independent P- controllers with complex gains for active control of both in-phase and quadrature components for 32 individual m=1 Fourier modes. "Open loop" mode for various studies involving application of pre-programmed stationary or rotating external controlled fields. Analog feedback controllers An optional set of analog PID-type feedback controllers is available for use with the intelligent shell feedback control scheme. 13

14 3. Fusion research projects 3.1. Active MHD mode control P. Brunsell, L. Frassinetti, M. Cecconello, J. Drake, E. Olofsson (PhD student) In collaboration with: E. Rachlew, S. Menmuir (PhD student), Dept. of Physics, SCI, KTH W. Suttrop, D. Yadikin, Max-Planck-Institut für Plasmaphysik, Garching, Germany R. Paccagnella, G. Manduchi, RFX Team, Consorzio RFX, Padova, Italy Control of MHD instabilities will likely be a requirement for ITER and a fusion reactor to achieve high beta and tolerable wall erosion and co-deposition of tritium. The use of dedicated magnetic field coils is envisaged. The control fields would be utilized in various ways to avoid Edge Localised Modes (ELM) by application of edge resonant external magnetic fields, rotation control of neo-classical tearing modes (NTM) by resonant field interaction, and active feedback suppression of resistive wall modes (RWM). In tokamaks, ideal kink modes may be destabilized by current density or pressure gradients. Close conducting walls can be used for passive stabilization of these modes on a short time scale, but with a real resistive wall, new slowly growing resistive wall modes (RWM) appear. Passive rotation or active feedback stabilization are possible ways to control these modes. For future large tokamak devices, active control using external magnetic fields is a very interesting option. The reversed field pinch (RFP) is dependent on a conducting wall for stabilization of current driven kink modes at all beta. For RFP devices using a thin shell with a time constant shorter than the plasma pulse length a range of unstable RWM are expected. The modes are experimentally observed growing slowly with an exponential growth time of the order of the shell radial magnetic flux penetration time. Since the growth time typically is of the order of milliseconds, active feedback stabilization of the modes using external magnetic field coils is feasible. The theory for active control of RWM has been developed and the intelligent shell concept has been introduced. Feedback controlled saddle coils are distributed over the shell acting to minimize the radial magnetic flux at the shell imitating a perfectly conducting shell. Suppression of multiple independent RWMs was successfully shown for the first time in the EXTRAP T2R device. Experiments on active control of RWM in RFP have been carried out during recent years on the EXTRAP T2R and RFX-mod devices. The results from these experimental studies are very positive, demonstrating the feasibility of RWM feedback stabilization in RFP Improved steady-state error and transient response Recently, new plasma experiments have been performed in EXTRAP T2R with the purpose of studying different digital controllers for intelligent shell feedback. [P. R. 14

15 Brunsell, et al., Phys. Plasmas 14, (2007)] The present study demonstrates that improvement of transient response and steady-state error with a controller having derivative and integral compensation is useful for effective suppression of RWMs, which is required in high current, long discharges in order to avoid excessive local wall heating. A digital proportional-plus-integral-derivative (PID) controller is used in the experiments. Using a P-controller at high proportional gain large oscillation is present (Fig ) indicating that the system is close to the stability limit. Factors contributing to the stability limit are phase-lags in the control system due to latency of the digital controller, finite bandwidth of the audio amplifier, and L/R time constant of the active coil. Feedback using a PD-controller resulted in a reasonable quiet controlled signal without large oscillations. Figure Experiments with intelligent shell feedback. Discharges with P and PD control. Shot P control (dotted), shot PD control (full). From top: a) Plasma current, b) n=+2 control output voltage, c= n=+2 coil current, and d) n=+2 sensor radial field. A series of discharges using a PI-controller with varying integral gain showed improved steady-state error suppression with increasing gain (Fig ). Then the transient response improvement of the PD-controller was combined with the steady-state error improvement of the PI-controller in a proportional-plus-integralplus-derivative (PID) controller. Empirically found good controller gain settings were compared to the well known rules for PID-controller settings knows as Ziegler- Nichols rules. 15

16 Figure Discharges with PI control and varying integral gain. The suppression of the n=+2 mode continuously improves with increasing integral gain Reference-spectrum RWM feedback control simulation and design In line with experimental results from the EXTRAP T2R reversed field pinch device, a general control-oriented formulation of resistive wall mode (RWM) dynamics has been compiled [K. E. J. Olofsson and P. R. Brunsell, 34 th EPS Conf. on Plasma Phys., Warsaw, 2-6 July 2007, ECA Vol. 31F, P (2007)]. A state-space form is considered, where the control quantities directly correspond to cylindrical Fourier modes. A feedback controller strategy capable of steering Fourier modes along essentially arbitrarily preset reference spectrum trajectoria is proposed. Initial simulations are performed hinting interesting possibilities. General mode control is a natural development of the PID-based local field annihilation and Fourier mode suppression techniques tested so far, where stabilization is the main objective. Model parameters have been checked and calibrated against experiments. It is demonstrated, assuming proper applicability of the single-pole ideal cylindrical RFP model that successful control is straightforward and principially possible. Highperforming monitoring and control of RWMs in EXTRAP T2R may require modelbased filtering and feedback. By tracking higher toroidal and poloidal modes (effectively subtracting them from the sensor signals) it is possible to accurately 16

17 follow Fourier modes during active coil transients. EXTRAP T2R spectra amplifies a broad range of modes, but most of them are still highly stable Active in-vessel coils and a conducting wall for AUG It is proposed to extend ASDEX Upgrade with a set of 24 (3x8) saddle coils. [W. Suttrop, et al, incl D. Yadikin, P. Brunsell, J. Drake, 34 th EPS Conf. on Plasma Phys., Warsaw, 2-6 July 2007, ECA Vol. 31F, P (2007)] The coils are mounted inside the vacuum vessel at the low field side and close to the plasma for fast response and flexible field structure. The enhancement will be carried out in several stages. Later a conducting shell will be added on the low field side, in between the existing upper and lower passive stabilizing loops. A feedback system consisting of distributed poloidal field sensors and a digital controller for n=1 RWM control is envisaged at this stage. Finally, with 24 independent AC power supplies, odd and even n perturbations can be made simultaneously, which allows for multi-mode RWM control (n=1, 2,...) Non-linear MHD dynamics L. Frassinetti, M. Cecconello, P. Brunsell, J. R. Drake, W. Khan (PhD student) In collaboration with: A. Hedqvist, E. Rachlew, S. Menmuir (PhD student) Dept. of Physics, SCI, KTH RFX Team, Consorzio RFX, Padova, Italy In recent years, good progress toward a better understanding and control of the plasma performance in the reversed-field pinch has been made. These improvements consist both of the discovery of spontaneous plasma regimes, termed quasi single helicity (QSH) regime, in which part of the plasma core is no longer stochastic, and in the development of techniques for active control of plasma instabilities. In the QSH regime of a RFP, the dynamo mechanism is produced mainly by one tearing mode, which is in contrast to the the standard RFP scenario, where the magnetic configuration is sustained through the interaction among several tearing modes. In the theoretically predicted single helicity (SH) regime the dynamo is produced by the interaction of a single tearing mode with the velocity field produced by an electrostatic drift due to a small charge separation. The SH plasma core is characterized by a magnetic island generated by a single mode and has no magnetic chaos. On the experimental side, regimes in which the dynamo is produced mainly by a single (dominant) mode, but in which other (secondary) modes have a small contribution have been obtained, and are referred to as quasi single helicity (QSH) regimes. In these regimes the plasma core is still characterized by a magnetic island, but part of the core becomes chaotic due to the presence of the secondary modes. Transitions to QSH regimes are characterized by the decrease of the secondary modes. As a result, the magnetic stochasticity in the core decreases, improving the particle confinement. For the thermal transport, conclusive claims on the improvement of global energy confinement is not yet possible, but nonetheless, the magnetic island is 17

18 hotter than the bulk plasma, since the island heat diffusivity seems to be lower that that of the surrounding stochastic region Spontaneous QSH regimes Experiments have been carried out in EXTRAP T2R [L. Frassinetti, et al., Phys. Plasmas 14, (2007)]. QSH regimes can be detected in a simple and direct way by analyzing the toroidal distribution of m=1 magnetic fluctuations. The most common way to study quantitatively this regime is to analyze the toroidal m=1 mode spectrum. In Fig the multi-helicity (MH) and the quasi single helicity states (QSH) are compared. In the QSH plasma, the dominant mode has high amplitude, approximately one order of magnitude higher than in the MH plasma. The comparison of the two regimes shows that the secondary modes are lower in the QSH plasma than in the MH plasma. In fact the total magnetic energy is approximately constant in both regimes, in the transition from MH to QSH, the secondary modes are reduced and the dominant mode is increased. Figure Mode spectrum of tearing modes (TM) and resistive wall modes (RWM) for a MH plasma (blue) and a QSH plasma (red). The dominant mode is n=-11, the innermost resonant TM. A figure of merit that quantifies the presence of the QSH is the spectral index Ns. It measures the spread of the magnetic energy over the toroidal modes. An ideal SH plasma has Ns=1, while a MH plasma with N toroidal modes of similar amplitudes have Ns=N. The time evolution of the spectral index is shown in Fig Note that values of Ns close to one can be reached. This shows that pure QSH states can be obtained in EXTRAP T2R. It is also interesting that when the dominant mode starts to increase, the mode rotation decreases. This can be ascribed to the increase in the electromagnetic braking torque. The duration of the QSH state in EXTRAP T2R is relatively short compared to other devices. The QSH plasmas in EXTRAP T2R can be classified into two groups. In the first group the QSH terminates when the dominant mode stop to rotate, in the second group the QSH duration is limited by the occurrence of sawtooth crashes. 18

19 Figure Examples of QSH plasmas. Mode evolution is shown in frames a) and d), spectral index evolution in frames b) and e) and toroidal rotation frequency of magnetic modes is shown in frames c) and f). QSH periods are characterized by Ns< Large periodic fluctuations A novel type of large periodic fluctuations in the main global plasma parameters and in spectroscopic measurements of emission from the whole plasma, both centre and edge, are seen in EXTRAP T2R [S. Menmuir, et al incl P. R. Brunsell, L. Frassinetti, J R Drake, 34 th EPS Conf. on Plasma Phys., Warsaw, 2-6 July 2007, ECA Vol. 31F, P (2007)]. A fluctuation cycle begins with a steady increase of the plasma current and at the same time the toroidal flux increases. After a few ms there is a very rapid event in the soft x-ray emission, an almost instant and strong increase, followed by an equally fast drop. Shortly afterwards, the emission from neutral molydenum increases and after this, the resistance goes up and the current, theta, and toroidal flux drop rapidly. The behavior is almost reversed to that of the conventional sawtooth crash or discrete dynamo event, where there is a burst of toroidal flux and a slow decay until the next event PPCD and active MHD control Different operating scenarios have been tested in recent years to improve the overall plasma performance of reversed field pinches, namely: quasi-single helicity (QSH), 19

20 pulsed poloidal current drive (PPCD) and active feedback stabilization (AFS). In particular, in the EXTRAP T2R reversed field pinch, AFS and PPCD techniques have been tested simultaneously showing encouraging results as it can be seen in figure 3.2.3, indicating that it is possible to operate an RFP in a regime of suppressed RWMs amplitude in improved confinement scenarios. E // (a) (Vm -1 ) SXR (a.u.) V res (V) Θ I P (ka) a) b) c) d) e) t (ms) Figure Example of PPCD application in combination with active MHD control: (a) plasma current; (b) Θ, (c) resistive loop voltage, (d) SXR signal and (e) parallel electric field Figure Example of a (m,n) = (1,-12) QSH in a discharge with selective feedback in which the mode (m,n) = (1,-13) is left uncontrolled. A preliminary study, in preparation for future studies on active control of the QSH, with selective AFS and PPCD in EXTRAP T2R has been carried out, leaving one or more internally resonant tearing modes uncontrolled and then applying a single PPCD pulse (with an edge poloidal electric field of 1.5 Vm -1 ) fired when the untargeted mode begins to grow. The results indicate that selective AFS does not affect directly the formation of QSH since QSH states were observed not only for the uncontrolled mode but also for the controlled modes as shown in figure Application of PPCD during selective AFS results in the changes in the growth rate of the uncontrolled mode due to the equilibrium change with a more significant increase in the SXR signal compared to the case with intelligent shell (see figure 3.2.5). An interesting feature of SFB discharges is that before and after PPCD the dominant tearing mode for the given equilibrium exhibits a new form of QSH, in which the mode amplitude is much larger then the other TMs but whose amplitude is rapidly varying in time as shown in figure This new QSH state lasts much longer than standard QSH (up to 5 ms compared to 0.2 ms). In addition the non-dominant modes still have significant amplitude contrary to the situation of a pure QSH. Similarly, during this new QSH state, the dominant mode has much larger toroidal rotation compared to all the other non-dominant modes. 20

21 b r -12 (mt) F b'θ (a.u.) SXR (a.u) # between Ω -12 (krads -1 ) t (ms) Figure Example of PPCD application in combination with selective FB control: (a) reversal parameter; (b) radial field amplitude for (m,n)= (1,-12), (c) SXR signal, (d) poloidal field amplitude for (m,n)= (1,-12) and (e) helical phase velocity for (m,n)= (1,-12) n Figure Example of PPCD application in combination with selective FB control: (a) amplitude of the time derivative of the poloidal component of the internally resonant TM; (b) mode spectrum for the top panel. 21

22 3.3. Plasma-wall interactions M. Rubel, H. Bergsåker, P. Sundelin (PhD student) In collaboration with B. Emmoth, IMIT, KTH Plasma-wall interactions (PWI) comprise all processes involved in the exchange of mass and energy between the plasma and the surrounding wall. Two inter-related aspects of fusion reactor operation - economy and safety - are the driving forces for studies of PWI. The major issues to be tackled are: (i) lifetime of plasma-facing materials (PFM) and components (PFC), (ii) accumulation of hydrogen isotopes in PFC, i.e. tritium inventory; (iii) carbon and metal (Be, W) dust formation. PWI is one of the primary areas where integration of the Physics and Technology programmes is being achieved. The work at KTH in the field of PWI and fusionrelated material physics has been fully integrated with the international fusion programme: (i) EU Fusion Programme, (ii) International Tokamak Physics Activity (ITPA), (iii) Implementing Agreements of International Energy Agency (IEA). It is demonstrated by the participation in: European Task Force on Plasma Wall Interactions (EU-TF-PWI). EFDA Technology Programme. EFDA-JET Work Programme: Task Forces E (Divertor Physics), FT (Fusion Technology), D (Diagnostics) and JET Enhancements (Phase 1 and Phase 2) including the ambitious ITER-Like Wall (ILW) Project, i.e. full metal wall at JET. ITPA and IEA activities. Experimental work is carried out at home laboratory, JET and TEXTOR. The research programme is concentrated on: Material erosion, migration and re-deposition. Fuel retention studies and fuel removal techniques. Characterization of plasma-facing materials and components including testing of high-z metals. Development and characterization of wall materials for ILW at JET. Development of diagnostic tools for PWI studies Fuel removal by ICRF-assisted discharges in H 2 -N 2 plasma The aim of this work was a preliminary evaluation of a nitrogen-based fuel removal where different layers were exposed to discharges in the TEXTOR tokamak; its vacuum vessel is shown in Figure 3.3.1a. The study was carried out with three types of probes: silicon substrates coated with amorphous deutereted carbon films (a-c:d), silicon with pre-boronised layers and bulk Inconel. The a-c:d layers were deposited under laboratory conditions, whereas the boronised films were prepared in TEXTOR during a regular boronisation process carried out with hydrogenated diborane (B 2 H 6 ). All three types of probes were mounted on holders, as shown in Figure 3.3.1b, and inserted into the tokamak using the limiter lock system. One set of probes were inserted using limiter lock system 1 in the bottom of the vessel, and one set using lock 3 at the top. The probes were exposed 22

23 to discharges in hydrogen-nitrogen (H 2 -N 2 ) glow discharge plasma assisted by ion cyclotron radiation heating (ICRH). Figure 3.3.1a. View of the TEXTOR interior Figure 3.3.1b. Holder used for exposure. Following the exposure, surface studies were performed by means of scanning electron microscopy (SEM), wavelength dispersive X-ray spectroscopy (WDS) and accelerator-based ion beam analysis (IBA) methods: nuclear reaction analysis (NRA) with a 2MeV 3 He + beam and enhanced proton scattering (EPS) with a 2MeV H + beam. The objective of NRA was mainly to quantify deuterium, while EPS was used to study the boron and carbon content. Surface morphology Images of the Si probes taken with electron microscopy show smooth surfaces both before and after exposure. As shown in Figure 3.3.2a, no change in morphology can be seen. A slight contrast may be due to the small amount of deposited carbon. On the Inconel samples, there is a thin layer formed on the exposed part, as indicated in Figure 3.3.2b. WDS analysis of the layer formed on the Inconel samples show a small increase in carbon, which was deposited during the experiment. The amount of boron in the exposed and unexposed parts is very similar. No nitrogen was detected, despite exposure in the nitrogen environment. Ion beam analysis IBA results are collected in Table The data present the deuterium, carbon and boron contents in the initial and exposed probes. Previous results for oxygen-assisted cleaning are given for comparison. During this experiment, the probes were positioned at limiter lock 3. After the H 2 -N 2 treatment, there is no significant decrease in deuterium content due to the exposure. In the a-c:d probes, all differences are within the variations of the non exposed layer. In the pre-boronised probes, no change is observed at all. While % of the deuterium was removed using oxygenassisted cleaning, almost no effect is achieved with nitrogen. The carbon layer is also effectively removed by He-O 2 glow, while the amount of carbon increases slightly during the nitrogen-assisted cleaning, indicating deposition of carbon eroded from plasma-facing components by the H 2 -N 2 treatment. The significant difference in deposition between the pre-boronised probed and the a-c:d probes is possibly explained by the difference in position on the holder, i.e. some probes may be shadowed by the holder. 23

24 For one of the pre-boronised samples, one can see an increase in boron of around 30 %. This change is small enough to possibly be explained by variations in the original layer thickness, but it might also be a real effect because boron is present in TEXTOR and could be deposited. In general, the analysis show no certain decrease of boron, carbon and deuterium when initial and H 2 -N 2 treated samples are compared. In some cases, there are indications of some decrease, but these changes are quite small. Figure (a) Pre-boronised silicon probe. Original surface (top) compared to exposed surface (bottom). (b) Inconel probe. On the exposed part (right), the carbon layer formed during the experiment is visible. Table 1. Content of D, B and C before (in) and after exposure (ex). Oxygen cleaning data from M. Rubel et al., J. Nucl. Mater (2007). All results in [10 15 cm -2 ]. (1) and (3) denote limiter locks in TEXTOR. Cleaning method Sample D in D ex B in B ex C in C ex H 2 -N 2 Inconel ICRFassisted Inconel Boron plasma Boron a-c:d a-c:d He-O 2 a-c:d glow Boron The results of surface studies show that this type of treatment has little influence on the surface morphology. The amount of carbon on pre-carbonized probes is even somewhat (5-30 %) increased during the exposure, thus indicating carbon deposition on probes in the tokamak. It is also important to note that the deuterium content in the probes is not affected by the H 2 -N 2 discharges. Much stronger effects in the surface structure and composition have been found after oxygen-assisted cleaning. The comparison of both approaches (oxygen versus nitrogen) indicates that oxidation is much more efficient in removing fuel than discharges in hydrogen-nitrogen plasma. However, definitive conclusions should not be drawn from these preliminary results, since the effects of plasma cleaning may be strongly dependant on the position of the probes. New series of experiments will have been performed in

25 Summary of most important results The ICRF-assisted H 2 -N 2 glow in TEXTOR at wall temperature ~200 o C, has not led to a reduction of the fuel and boron content at the location of the sample exposure. The carbon content is increased by approx. 30%. This is attributed probably due to deposition of carbon eroded from the tokamak wall by the plasma treatment No embedded nitrogen has been detected in the samples. The layers do not flake or peel-off as a result of exposure to nitrogen discharges Material mixing and compound formation on tungsten limiters Work done in cooperation VR-FZJ-IPPLM The study was carried out for two tungsten limiters exposed at TEXTOR: (i) a castellated block of mushroom shape and (ii) a vacuum plasma sprayed (VPS) thick coating on graphite block. The castellated block was composed of twelve individual segments separated by 0.5 mm wide gaps. This construction of a test limiter allowed the segments to be detached after the exposure in order to study deposition inside gaps oriented in poloidal direction. The exposure and plasma operation conditions with that limiter have been reported earlier. The calculation of temperature profile in the limiter was based on surface (pyrometer) and bulk (thermocouples) measurements. The second limiter was a graphite block coated with a μm thick layer of VPS tungsten. This block was a part of the array of the main poloidal limiters at TEXTOR. Macro-brush tungsten limiter Images in Fig a show the entire macro-brush limiter (see insert) and deposits formed inside the castellation during the exposure to high-heat loads. There is a melt zone on the top surface (see insert) because during some shots the temperature in that region exceed the W melting point, T m(w) = 3410 o C. The heat propagated to the bulk of tungsten. One observes five distinct regions on the surface located in the gap between segments: shiny metallic areas at the top and bottom and two brownish zones separated by a red region indicating the presence of copper. The most probable origin of copper has been discussed previously: residual metal remaining after spark-erosion cutting of toroidal gaps with a brass wire. The results of X-ray diffraction (XRD) phase analysis performed in the upper brownish zone are plotted in Fig b revealing the presence of elemental tungsten, tungsten dioxide of distorted rutile structure (WO 2 ) as major phases. There are also some traces of copper. The same composition was detected in the second (lower) brownish zone, whereas only metallic tungsten was found in shiny regions at the top and bottom of the limiter. A micrograph in Fig c shows the oxide structure on tungsten surface. The presence of oxygen was confirmed by local EDS analysis. Microscopy studies in all areas containing WO 2 indicate that the oxide was deposited on the W surface by condensation from the gas phase. To explain the presence of a volatile oxide one has to consider the source of oxygen, temperature distribution inside the limiter and thermodynamic data. Plots in Fig a and b show: (a) the temporal evolution of local spectroscopy signals (WI, OII, CII and CuI) and (b) oxygen evolution versus surface temperature during the limiter exposure. There is a significant increase of the OII signal which is most probably related to outgasing of water vapour traces from tungsten at high temperature. 25

26 c W WO 2 W+ Cu W + WO 2 W Cu base W 10 μm W O 2 b WO WO W O 2 W WO W O 2 W Cu Cu Cu Θ scale Cu WO WO2 Figure Overview images of the castellated tungsten limiter and area inside the gap (a), diffractorgram (b) and topography of the oxide deposit (c). line intensity [a.u.] OII(447nm) CII(283.7nm) 80 H a δ (434nm) BII(412.1nm) WI(400.8nm) CuI(324.8nm) SiII(456.8nm) CaII(315.9nm) time [s] OII line intensity [a.u.] surface temperature [K] Figure Temporal evolution of spectroscopy signals (a) and oxygen signal versus surface temperature (b) recorded during the exposure of the castellated limiter. b

27 The most stable oxides are: WO 3 and WO 2. At temperatures above 400º C, tungsten oxidizes forming tungsten oxide (WO 3 ) which may be reduced to WO 2 with CO, H 2 or W. The formation of WO 2 by reduction of WO 3 with CO begins at 450 º C. At temperatures over º C, volatile oxides are formed by evaporation of mainly WO 3. According to Yih and Mrowec, at high temperatures the evaporation rate can be equivalent to the rate of formation. For low pressures the gaseous WO 2 is the dominating form of oxide. The loss of material due to oxidation is increasing linearly up to around 2250º C, after which it decreases due to thermal decomposition of the oxides. It has also been reported that WO 2 remains stable in a hydrogen-rich atmosphere even at temperatures of o C, as it is the case of a tokamak environment. Copper melting inside the castellation (T m(cu) = 1083 o C) proves that the temperature exceeded 1000 o C thus making the volatilization of oxides possible. Finally, taking into account all experimental findings and thermodynamic data one may assume that the most probable pathway leading to the WO 2 deposition was initiated by the reaction of W with water vapour forming WO 3 or H 2 WO 4 followed by the reduction to WO 2 which condensed on the surface. Tungsten-coated graphite limiter Figure 3.3.5a shows the VPS-W coated graphite limiter. One perceives a distinct deposition pattern formed on the side surface and the coating damaged by power loads exceeding 20 MW m -2. The damage to the tungsten layer and material mixing in the W layer have been reported previously. This work is concerned with chemical composition of deposits on side surfaces. Face A (see Fig a) was located in the gap between two limiter stones, whereas the other side (Face B) was the outer surface of the limiter array. The XRD analyses were done in three points on each side: 8, 25 and 60 mm deep in the scrape-off layer (SOL). Figure 3.3.5b shows a diffractogram detailing features of the main and trace phases present on the surface (in Point 1), whereas plots in Figure 3.3.5c show data recorded in three points. These results, fully representative for both faces (A and B), demonstrate that tungsten carbide (WC) is the main tungsten phase on the graphite background. In addition, only traces of tungsten subcarbide (W 2 C) and elemental tungsten are detected. It is noticeable that the graphitic parameter is distorted by the presence of C-D species in the co-deposit. In fact, the deuterium content changes from 14x10 17 cm -2 to 2x10 17 cm -2 with a distance from the limiter top: 8 to 65 mm, respectively, as measured with NRA. The formation of tungsten carbides (WC and W 2 C) begins at o C depending on the form of reactants. Other carbides reported in literature (W 5 C 2, W 3 C 2 ) are unstable. In case of a thin carbon layer on W plate the formation starts already at 800 o C (predominantly subcarbide, W 2 C) and at around 1000 o C W 2 C transforms into WC. One may expect similar temperature range for reactions occurring in highly dispersed systems containing sputtered W and C species. Two pathways may be considered. (i) direct reaction of sputtered atoms in gas phase (near the surface) followed by the WC deposition, (ii) the reaction of sputtered W atoms on hot graphite surface. The XRD analysis identified WC as the main tungsten phase in all points on both sides of the limiter block. While the temperature of side surfaces near the limiter top could exceed 1000 o C during high power deposition, the temperature of the bottom part was in the range of several hundred degrees. This leads to a tentative conclusion that two types of reactions (surface and in gas phase) took place. As a result, tungsten eroded from 27

28 the coating and transported to the gap was nearly completely transformed into carbides. Carbides (WC and W 2 C) were also detected on the plasma-facing surface. a x WC C WC b Intensity [sqrt scale] Sqrt (Counts) Sqrt (Counts) e4 5x10 4 C C W 2 C W 2 C W 2 C W C WC WC C 2-Theta - Scale 2 Θ scale A1 A2 C A3 c Θ scale Figure Tungsten-coated limiter after high power loads at TEXTOR (a) and diffractograms showing major and trace phases in Point 1 near the limiter top (b) and overview of composition in three points located on the side surface of the limiter (c)

29 Concluding remarks Tungsten mixed with carbon, silicon and boron (Si and B from wall conditioning of TEXTOR) has been observed on side surfaces of the VPS tungsten-coated limiter. To our knowledge, the presence of oxide on components retrieved from a tokamak is reported for the first time. The oxidative mechanism of W transport should be taken into account if the application of oxygen-assisted fuel removal techniques would be applied in next-step devices. Such methods of cleaning might create an in-vessel source of oxygen which, as a consequence, would enhance material migration and mixing. The results obtained for a W-coated limiter clearly show that nearly all tungsten eroded from the coated limiter was re-deposited as carbide. Very intense carbon-tungsten mixing was identified earlier by IBA but the XRD result has provided the final proof Collector probe: Ion flux measurements in SOL Ion fluxes in the SOL plasma were measured in TEXTOR using a collector probe installed on the fast transfer system operated in the equatorial plane. These were the first measurements of this kind performed during the Dynamic Ergodic Divertor (DED) operation. The aim was to assess the impact of DED on the deposition of deuterium and plasma impurity species on the silicon collectors. The samples were exposed typically for 200 ms in the flat-top phase of the discharge. The exposure conditions were varied on shot-to-shot basis by changing the neutral beam power (up to ~1.3 MW) and the DED induced magnetic field structure, including the so-called pump-out conditions. After exposure, it was observed that the samples were still shiny except those exposed in pulses with high input power. Nuclear reaction analysis (NRA) and backscattering spectroscopy were applied for detailed studies of the surfaces. Carbon and deuterium were the main species deposited on the probes. For samples exposed in low input power conditions only low carbon concentrations of ~10 15 C/cm 2 were detected. A specific problem was addressed concerning the contributions to the deuterium retention due to co-deposition with carbon and due to the deuterium implantation Material Migration Studies by Tracer Techniques Studies of plasma facing components (PFC: limiter and divertor tiles). On the last operation day with the MkII-SRP divertor 13 C-labelled methane was introduced to JET from the outer divertor. The aim was to study the carbon transport. The analysis of the C-13 distribution and quantity on various PFC was finished. The analysis of Tile 5 (septum replacement plate) has shown a hollow deposition profile of the C-13 marker. The results obtained with ion beam analysis allowed the completion of material migration modeling in operation with the MkII-SRP divertor. The deposition profile is shown in Figure