1. Fast Imaging of Visible Phenomena in NSTX R. J. Maqueda Nova Photonics C. E. Bush ORNL L. Roquemore, K. Williams, S. J. Zweben PPPL 47 th Annual APS-DPP Meeting October 24-28, 2005 Denver, Colorado Poster RP1.00014 Movie clips are hyperlinked with this “camera” symbol.

2. 1 Abstract Edge phenomena are important for global plasma confinement as well as power and particle handling and distribution to plasma facing components. High frame rate, 2-D imaging is a powerful tool to access the physics behind these phenomena which include: edge turbulence and "blobs", ELMs, and MARFEs. This diagnostic is also useful in general plasma equilibrium and dynamics measurements, like those during Coaxial Helicity Injection discharges, and in pellet injection experiments. A new Phantom 7 fast-framing digital camera has been installed in NSTX which has been used at frame rates typically ranging between 68000 frames/s and 120000 frames/s and full discharge coverage (frames recorded for over 2 s). Examples will be presented showing the usefulness of this diagnostic for physics studies in the areas mentioned above. Work supported by DoE grant DE-FG02-04ER54520.

3. 2 Outline Two-dimensional imaging at fast frame rates (>10000 frames/s) has many applications in magnetically confined plasmas: Edge turbulence and “blobs”: Gas Puff Imaging (GPI) diagnostic. L-H transitions: Where does the transition start? Edge Localized Modes (ELMs): Heat pulse evolution and interaction with plasma facing components. Multifaceted Asymmetric Radiation From the Edge: Is a MARFE axisymmetric? Plasma positioning and equilibrium: Development of non-inductive current initiation by Coaxial Helicity Injection (CHI). Lithium pellet injection: Ablation dynamics and plume development. A new fast framing camera capable of capturing 120,000 frames/s with full discharge coverage is being used in NSTX.

4. 3 NSTX’s Phantom 7 Camera* Frame rates of: ≤4,800 frames/sat 800 x 600 pixel resolution ≤ 68,000 frames/sat 128 x 128 pixel resolution ≤ 120,000 frames/sat 64 x 64 pixel resolution Minimum frame exposures of 2 µs. Digitization: 12-bit. C-MOS detector with 30%-40% Q.E. and 22 µm x 22 µm pixels. Full discharge coverage with 2 GB of on-board memory. Fast download speeds through Ethernet connection (100 Mbit/s network). Control through LabView, synchronized to MDS+ shot cycle. Coherent fiber bundles used to transmit image to camera. Interference filters used to select visible bands of spectrum. * Manufactured by Vision Research, Wayne, NJ.

5. 4 GPI Diagnostic Camera used to view visible emission from edge just above midplane. Gas puff is injected to increase image contrast and brightness. Gas puff does not perturb local (nor global) plasma. Emission filtered for D  light from gas puff: I  n o n e f(n e,T e ) D  emission only seen in range ~ 5 eV < T e < 50 eV View aligned along B field line to see 2-D structure  B. Typical edge phenomena has a long parallel wavelength, filament structure. For more details: “Gas puff imaging of edge turbulence”, R.J. Maqueda et al., Rev. Sci. Instrum. 74(3), p. 2020, 2003.

6. 5 Summary of GPI Results Ohmic H-mode Edge turbulence observed during Ohmic H-modes in NSTX is similar to that measured in neutral beam heated H-modes. Quiescent H-mode edge is present with the turbulence much reduced respect to the preceding L-mode phase. Only small amplitude poloidal modulations of the emission has been observed during H-modes. The fluctuation level decreases from a typical 10%-40% RMS level in L-mode to an also typical 5% RMS level in a quiescent H-mode. The poloidal autocorrelation lengths appear to be somewhat smaller than those previously reported in H-modes (S.J. Zweben et al., Nucl. Fusion 44, p. 134, 2004). For details see: C. E. Bush, poster RP1.00028

7. 6 GPI: L-H Transition Transition takes place at ~192.1 ms L-mode Separatrix Antenna limiter shadow 24 cm radial 24 cm poloidal Spontaneous transition into quiescent H- mode “Blobs” Ohmic H-mode 0.65 ms mosaic D 2 puff D  filter ~11 MB

8. 7 GPI: Time Evolution Time (s) Image pixel (a.u.) RMS fluctuation level Divertor D  (a.u.) H-mode drsep = -5 cm drsep = 0 cm drsep =+5 cm Shot 115513 L-mode burst 2 kHz “breathing” mode Reduced fluctuation level during H-mode

9. 8 RMS fluctuation level Average image brightness (a.u.) Poloidal auto-corr. length (cm) R mid -R sep (m) average 0.191 s - 0.192 s (L-mode) average 0.1922 s - 0.1932 s (H-mode) average 0.207 s - 0.208 s (H-mode) average 0.209 s - 0.210 s (L-mode) GPI: Radial Profiles Shot 115513 FWHM

10. 9 GPI: Active H-mode H-mode edge with “blobs”...micro-ELMs? Separatrix Antenna limiter shadow 24 cm radial 24 cm poloidal “Blobs” Active Quiet 4.5 MW NBI 0.65 ms mosaic D 2 puff D  filter ~5.5 MB

11. 10 L-H Transition Where does the transition start? The image intensity is consistently reduced first at the midplane near the center stack. This is followed soon after (20-30  s) by the outer divertor strike point. The inner leg of the divertor region is delayed respect to the outer strike point by ~150  s, with a slower decay rate. This, perhaps, introduced by atomic physics of highly radiating MARFE-like region. NOTE: Data available for only LSN H-modes with high field side fuelling and fixed plasma parameters (800 kA, 4.5 kG, 4 MW NBI). Only shots with “clean transitions” (no dithers and low fluctuation levels) were selected. Time traces normalized to “1” before the transition and “0” after the transition

12. 11 DD CII (657.8 nm) Time (s) R.O.I. intensity (a.u.) L-H Transition: Time Sequence Image intensity drop sequence: 1) Center stack near midplane, with  drop ~100  s. 2) Outer divertor strike point, within 20-30  s. 3) Other locations later, with slower decays. Similar sequence in D  light. Fish-eye view

13. 12 Edge Localized Modes (ELMs) Type I, III, and V ELMs are routinely seen in NSTX’s H-mode shots. (Type II ELMs have recently been observed too.) Fast camera imaging shows evolution characteristics of impurity emission layers in divertor region during the different types of ELMs. Type V ELMs show heat pulse propagation characteristics consistent with energy/particles ejection from the closed field line region near the lower strike point, low field side. (For more details see: R. Maingi, invited talk CI1b.003.) Lower divertor tangential view Phantom camera image

14. 13 Type I ELM Shot 117407 CII (657.8 nm) Carbon sputtering Energy/particle dumpRecovery Energy dump into divertor region causes CII emission layer to move to smaller (and larger) major radii. EFIT reconstructions show the X-point moves upward and inward on the order of a few centimeters. 0-600 scale 0-1568 scale 0-4095 scale Unperturbed emission ~9.3 MB

15. 14 Type III ELM Shot 117432 Outer strike point brightens Recovery Energy dump affects inner leg detachment but emission layer persists close to separatrix. There is no measurable movement of X-point. Inner leg re-attaches Modes on inner separatrix Relaxed to unperturbed emission Unperturbed emission CII (657.8 nm) 800 kA 4.2 MW NBI Double null Time (ms) Divertor D  (a.u.) Shot 117432 1 2 3 456 123456 ~4.4 MB

16. 15 Type V ELM Shot 117407 Secondary band on outer strike point...and reaches X-point region Heat pulse propagates on inner separatrix, delayed from outer strike point band. There is no measurable movement of X-point. Heat pulse propagates on inner separatrix Propagation continues Relaxed to unperturbed emission Unperturbed emission CII (657.8 nm) 800 kA 4.2 MW NBI LSN Time (ms) Divertor D  (a.u.) Shot 117407 1 2 3 456 123456 ~4.8 MB

17. 16 Type V ELM Time (s) Image intensity (counts) Image intensity (counts) Outer divertor Inner separatrix From peak in cross-correlation function: In-out delay = 0.32 ms Shot 117407 t = 255.477 ms

18. 17 Type V ELM: Heat Pulse Propagation Time (ms) peak in cross-correlation function Poloidal distance along inner separatrix (cm) 1.1 Km/s Slow down reaching X- point Shot 117407 t = 255.587 ms

19. 18 MARFE evolution Are MARFEs toroidally symmetric? MARFEs are seen on the divertor region and center stack of NSTX. Although the evolution of the MARFE is varied, some new characteristics have been observed. MARFEs born near the lower divertor move upward (against ion grad-B drift direction) as toroidally localized condensation, while rotating toroidally, following the magnetic field pitch. Upward movement stagnates and becomes a “more typical” toroidally symmetric ring. MARFE then moves downward towards lower divertor, while still rotating. Presence of highly radiating MARFE coincides with decrease in divertor recycling. MARFE: “Multifaceted Asymmetric Radiation From the Edge” [B. Lipschultz et al, Nucl. Fusion 24, p. 977, 1984].

20. 19 Center stack Lower divertor Toroidally localized, rotating condensation 1.0 ms mosaic D  filter 9  s exposures contrast enhanced MARFE evolution Upper divertor Stagnation MARFE moves downward Ion grad-B drift Fish-eye view 900 kA 6.4 MW NBI Double null ~3.8 MB

21. 20 Plasma Positioning and Equilibrium Coaxial Helicity Injection (CHI) R. Raman (U. Washington) Solenoid-free plasma startup is important for the spherical torus concept. Coaxial Helicity Injection (CHI) is a promising method to achieve this goal. Fast-framing digital camera gives operators feedback on plasma positioning and equilibrium during CHI experiments. 60 kA of closed flux current generated using only 7 kJ of capacitor bank energy. In some discharges, the current channel shrinks to a small size and persists for more than 200 ms. For details see: R. Raman, contributed oral GO3.00011

22. 21 Coaxial Helicity Injection (CHI) Discharge evolution For details see: R. Raman, contributed oral GO3.00011 Time (ms) Plasma current (kA) Injector current (kA) Shot 118342 Fish-eye view No filter 9  s exposures Detached plasma Decay Fully grown plasma Fast crowbar Current persistence Breakdown and growing plasma ~4.5 MB

23. 22 Lithium Pellet Injection Lithium pellet injection is used in NSTX to modify the conditions of plasma facing components, as well as, for diagnostic purposes. Lithium pellets with masses between 0.43 mg and 5 mg are injected just above the outer midplane at ~150 m/s. Fast-framing digital camera shows pellet penetration, ablation of pellet material and transport along field lines towards divertor regions. Ablated pellet material shows structure of underlying electron density (filamentary structure) and flux surfaces (if deep penetration). For details on lithium pellet injection experiments see: H. Kugel, contributed oral GO3.00008

24. 23 Shot 117909 LiII (548.5 nm) Pellet injection ~1.4 ms ~20 cm penetration 0-255 scale 0-1023 scale 0-4095 scale Lithium Pellet Injection NBI Heated H-mode Fish-eye view Ablation begins in SOL Pellet material “diverted” Pellet material burns-out Filamentary structure Pellet material deposited on divertor surfaces 600 kA - 4.4 MW NBI - double null ~4.5 MB

25. 24 Shot 117094 LiII (548.5 nm) 0-600 scale 0-1568 scale 0-4095 scale Lithium Pellet Injection Ohmic Helium Plasma Full penetration Fish-eye view Outer edge Core flux surfacesInside edge 500 kA - Ohmic – inner wall limited Flux surface Flows? Pellet ablation first seen at 240.47 ms ~7.7 MB

26. 25 Summary of Physics Results Ohmic H-modes in NSTX appear similar to neutral beam heated H- modes. They are categorized among the “quiescent” group of H-modes. During the L-H transition (NBI shots), the recycling (and CII light) is first reduced at the midplane near the center stack and soon followed (20-30  s) by the outer divertor strike point region. Type V ELMs show heat pulse propagation characteristics consistent with energy/particle ejection from the closed field line region near the lower strike point, low field side. MARFEs appear to originate as toroidally localized condensations that later become “more typical” toroidally symmetric rings.

27. 26 Summary and Conclusions Fast-framing visible cameras have multiple uses in magnetically confined plasmas. Examples have been shown pertaining to edge turbulence, ELMs, L-H transitions, MARFEs, solenoid-free startup and pellet injection. Other possible uses include: - ELMs using GPI and fish-eye views. - “GPI” using pellets and/or supersonic gas injector. - Interaction between MARFEs and ELMs. With more than 210 Gbyte of Phantom camera data collected in the 2005 experimental campaign of NSTX the first challenge becomes automated analysis. (Note: Camera was used on only 1/3 of NSTX’s plasma shots.) Nearly every short portion of data contains interesting, valuable information!