1 EX/P6-22

Fast Ion Physics Enabled by Off-Axis Neutral Beam Injection

W.W. Heidbrink 1), M.A. Van Zeeland 2), M.E. Austin 3), E.M. Bass 4), X. Chen 1),B.A. Grierson 5), C.M. Muscatello 1), G. Matsunaga 6), G.R. McKee 7), D.C. Pace 2),J.M. Park 8), C.C. Petty 2), R. Prater 2), D.A. Spong 8), B.J. Tobias 5), Y.B. Zhu 1)and the DIII-D Team

1) University of California Irvine, University Dr., Irvine, CA 92697, USA.2) General Atomics, PO Box 85608, San Diego, CA 92186-5608, USA.3) University of Texas at Austin, 2100 San Jacinto Blvd, Austin, TX 78712-1047, USA.4) University of California San Diego, 9500 Gilman Dr., La Jolla, CA 92093-0417, USA.5) Princeton Plasma Physics Laboratory, PO Box 451, Princeton, NJ 08543-0451, USA.6) Japan Atomic Energy Agency, Naka City, Ibaraki, Japan.7) University of Wisconsin at Madison, 1500 Engineering Dr., Madison, WI 53706, USA.8) Oak Ridge National Laboratory, PO Box 2008, Oak Ridge, TN 37831, USA.

e-mail contact of main author: heidbrink@fusion.gat.com

Abstract. Off-axis injection of neutral beams into DIII-D has provided new insights into fast-ioninstabilities that may impact alpha-particle and neutral-beam confinement in ITER. The performance ofthe off-axis beams is validated by injecting individual sources sequentially into low-power plasmas thatare optimized for accurate diagnostic measurements. Off-axis injection can alter the fast-ion gradient∇βf that drives Alfven eigenmodes (AE) unstable. Two-dimensional measurements of reversed shear(RSAE), toroidal (TAE), and beta-induced Alfven-acoustic eigenmodes (BAAE) show that the phaseof the eigenfunction varies with radius. The phase variation is insensitive to ∇βf . Off-axis injectionreduces the amplitude of RSAE activity an order of magnitude. The enhanced stability for off-axisinjection is attributed to flattening of ∇βf near the minimum of the safety factor qmin. Core TAEs arestrongly stabilized. In contrast, at larger minor radius, the fast-ion gradient is similar for on- and off-axisinjection. As a result, switching the angle of injection has little effect on the stability of TAEs thatappear in the outer portion of the plasma. Surprisingly, even though the gradient is flattened and themode frequencies change, switching the angle of injection also has little effect on the stability of BAAEsthat are unstable near the magnetic axis. Off-axis fishbones are modes that resonate with the precessionalmotion of trapped fast ions. In experiments to date, these modes remain unstable when off-axis sourcesare substituted for some of the on-axis sources.

1. Introduction

Controlling instabilities that are driven by fast ions is crucial for the practical realizationof fusion. Manipulation of the distribution function that drives these modes tests ourtheoretical understanding of mode stability and structure and may suggest new avenuesof control. The DIII-D tokamak is equipped with eight neutral-beam sources housed infour beamlines. Previously, all eight sources injected into the tokamak horizontally atthe midplane of the vacuum vessel. Prior to the 2011 experimental campaign, a massivehydraulic lift was installed beneath one of the beamlines to allow vertical steering [1].When the beamline is tilted, the pair of sources in this beamline inject neutrals throughthe midplane port and below the midplane in the plasma. For the maximum elevationof the beamline, the off-axis beams deposit large numbers of ions near the mid-radius(normalized minor radius of ρ ' 0.5).

In general, fast-ion instabilities can be driven by gradients in either velocity or config-uration space. For example, the fast-ion drive term for toroidal Alfven eigenmodes (TAE)

2 EX/P6-22

is linearly proportional to the product of the fast-ion beta βf and the fast-ion diamagneticfrequency ω∗,f [2] (which can also be written ∇βf ). In the experiments reported here, thebeam-ion pitch angles are similar for the on-axis and off-axis beams, so the changes invelocity-space gradients are modest when switching between sources. In contrast, switch-ing between on-axis and off-axis sources makes large changes in ∇βf in the inner half ofthe plasma (ρ<∼0.5).

-1.0

-0.5

0.0

0.5

1.0

1.0 2.0

1.0 1.5R (m)

-1.0

-0.5

0.0

0.5

1.0

z (m

)

Beam Tilt = 0 Deg. Source Tilt = 0 min.

1.0 2.0

1.0 1.5R (m)

Beam Tilt = 16.4 Deg. Source Tilt = 35 min.

(a) (b)

FIG. 1. Dα images of a neutral beam source during in-jection into helium neutral gas. The steerable sourceinjects in the midplane (a) or at the maximum per-missible angle (b). Typical flux contours are overlaidon the images.

This paper begins by summariz-ing initial experiments that confirmthat the tilted sources inject off-axis(Sec. 2). Section 3 discusses the ef-fect of changes in ∇βf on Alfven andoff-axis fishbone instabilities. A briefconclusion follows (Sec. 4).

2. Measurements inMHD-Quiescent Plasmas

Initial work with the tilted sourcessought to confirm beam perfor-mance. Installation of a mirror ina re-entrant port provided a viewof the entire steerable beam fromthe port box to the inner wall.In “beam-into-gas” shots, the deu-terium sources inject into neutralhelium gas. Collisions with thehelium excite Dα emission that isfiltered and measured by a fast

On-axis

Axis

170

3

FIDA

Den

sity

(au)

2

1

0180 190 200

Major Radius (cm)210 220

#144231

Off-axis

FIG. 2. FIDA profiles at the end of 100-msbeam pulses by near-tangential on-axis andoff-axis sources. The ordinate is the FIDAlight obtained by fitting the FIDA feature inthe full Dα spectra of a tangentially viewingdiagnostic [3] divided by the injected neutraldensity.

framing camera. Figure 1 shows measuredimages of the beam at the maximum andminimum possible elevations [3]. The imagesshow that the beam is successfully steeredoff-axis in the chamber but that some clip-ping of the beam occurs on the lower portbox at maximum elevation. To model theneutral deposition, input parameters in theNUBEAM code [4] are adjusted to match themeasured profiles.

Next, the tilted sources injected into low-power plasmas. Extensive comparisons ofon-axis and off-axis injection were performedby injecting individual sources sequentially.Since this work was recently published [5],only a brief summary is provided.

Figure 2 shows measurements of fast-ionDα (FIDA) light obtained at the end of a 100 ms beam pulse. These data confirm thatthe co-passing fast-ion population shifts off-axis with off-axis injection. Qualitatively,

3 EX/P6-22

the off-axis beams have the expected effect on neutron, FIDA, neutral-particle analyzer(NPA), toroidal rotation, fast-ion pressure, and sawtooth signals; however, quantitatively,the off-axis sources produce ∼20% fewer fast ions than expected. For this experiment,the NPA diagnostic measures active signals from charge-exchange reactions with trappedfast ions near the magnetic axis. As expected [6], tiny NPA signals are observed for themagnetic field helicity that is favorable for neutral-beam current drive but larger signalsare observed for the unfavorable helicity.

3. Effect of Off-Axis Injection on Instabilities

A Conditions for Alfven Eigenmode Experiments

Previous experiments have shown that early on-axis beam injection into plasmas with areversed safety factor profile produces a regime with many unstable reversed shear Alfveneigenmodes (RSAEs) and TAEs. (In spectrograms, RSAEs are readily distinguished fromTAEs by their characteristic up-sweeping frequency [7]. TAE frequencies vary more grad-ually in time.) Numerous papers on these plasmas have been published in recent years,including a comparison with MHD mode structure [8], evidence for flattening of the fast-ion gradient [9], direct measurements of fast-ion losses [10], electron cyclotron emissionimaging (ECEI) measurements of radial phase variations [11], and comparisons with thepredictions of gyrokinetic codes [12]. Similar discharges have also been created in ASDEX-Upgrade [13]. Later in these discharges, as the current profile diffuses, RSAE and TAEactivity ceases but beta-induced Alfven-acoustic eigenmode (BAAE) instabilities are ob-served [14].

0.00.5

1.01.5

500 1000 1500Time (ms)

I p (MA)

0

2

4

P inj (M

W)

0

2

4

P inj

(MW

)

On-axis NBI (#146102)

Off-axis NBI (#146101)

RSAEs & TAEs

BAAEsμTurbulence(a)

(b)

(c)

FIG. 3. Time evolution of (a) plasma currentand beam power waveforms during (b) off-axisand (c) on-axis neutral beam injection. Thelabels at the top of the figure indicate the ap-proximate time period for the study of differ-ent instabilities.

This extensively studied condition is se-lected for the new experiments. To inves-tigate the dependence of Alfven eigenmodestability and mode structure on the fast-ionprofile, different combinations of on-axis andoff-axis beams are injected during the cur-rent ramp [Fig. 3(a)]. For the off-axis case,both tilted sources steadily inject ∼4 MW ofpower [Fig. 3(b)]. (In an “off-axis” case, ifthey inject at all, the on-axis beams thatare needed for diagnostics only inject brief5-10 ms pulses.) In the best on-axis com-parison case, the duty cycle of the on-axisbeams is adjusted to match the available off-axis power [Fig. 3(c)]. Several different in-stabilities are studied in a single discharge:RSAEs and TAEs initially, then BAAEs nearthe end of the current ramp, then the effect of microturbulence on fast-ion transport inthe MHD-quiescent phase prior to the onset of sawtooth oscillations.

Figure 4 illustrates the idea of the experiment. The q profile is reversed early in thedischarge (during the RSAE and TAE phase). Later, in a time period with very weakmagnetic shear, BAAEs are excited. Changing between on-axis and off-axis sources altersthe fast-ion gradient in the central half of the plasma. Localized fluctuation diagnosticssuch as electron cyclotron emission (ECE) and beam-emission spectroscopy (BES) mea-

4 EX/P6-22

sure the amplitude and structure of any excited instabilities. In addition, the cross-powerof two interferometer channels supplies a global monitor of mode activity.

2

4

6

8

q (400 ms)

q (800 ms)

On-axisOn-axis

Off-axis

Beam

Den

sity

(1012

cm

-3) RSAE edge TAE

5

4

3

2

1

00.0 0.2 0.4 0.6 0.8 1.0

Minor Radius

BAAEcore TAE

FIG. 4. Illustration of beam-ion and q pro-files for the AE stability experiment. Theq profiles are representative of reversed-shearprofiles early in the discharge (400 ms), whilethe profile with weak shear (800 ms) is rep-resentative of profiles during the BAAEphase. The beam-ion profiles are calculatedby NUBEAM. The labels at the top of thefigure indicate the approximate spatial loca-tions of the spectra shown in Figs. 7–9.

To confirm that variations in beam pro-file are responsible for the observed effects,numerous other parameters were varied inthe experiment. These variations includechanges in elongation (κ = 1.2 and 1.6) andin the helicity of the magnetic field, differentcombinations of on-axis and off-axis beams,and variations in q profile achieved by addingelectron cyclotron heating (ECH) power orby shifting the time of beam injection. Noneof these variations qualitatively alter the re-sults reported here.

B Alfven Mode Structure

(a) Shot #146077

Lines of

pha se

RSAE TAE

BT IP

(c) 142253

BT IP

(b) 142111

0.125

)m(

z

-0.125

0.125

)m(

z

-0.125

-0.250

0.250

R (m)

R (m)

1.85

1.85 2.00

2.00

constant

FIG. 5. 2D phase plots of RSAEs and TAEsobtained from ECEI data. (a) Mode struc-tures are imaged simultaneously using a dual-array capability. (b,c) RSAE mode structureon the low-field side and direction of propaga-tion (arrows) during (b) standard and (c) re-versed toroidal field operation. From Ref. [15].

Examples of ECEI measurements of RSAEand TAE mode structures are shown inFig. 5(a). Although many features of themode structure are consistent with idealMHD theory [8], the observed radial shear-ing of the phase is not predicted in idealMHD but is predicted by gyrofluid [11] andgyrokinetic [15] models. Initially, the shear-ing was attributed to the fast-ion gradient[11]; however, recent experiments with theoff-axis beams show that the radial phasevariations are insensitive to changes in ∇βf[15]. New gyrofluid and gyrokinetic calcula-tions find that only a weak dependence on∇βf is expected and a quantitative descrip-tion of the shearing remains elusive [15]. Irre-spective of the sign of ∇βf (and the fast-iondiamagnetic direction), the observed RSAEsand TAEs always rotate in the thermal iondiagmagnetic direction [Fig. 5(b,c)], which isalso the direction of the applied beam torque.Counter propagating modes exist theoreti-cally but these modes are not observed in the experiments to date. Detailed stabil-ity analysis with accurate fast-ion models is needed to determine if counter-propagatingmodes should be unstable.

The ECEI diagnostic has also obtained high-quality measurements of BAAE modestructure (Fig. 6). Like RSAEs, the mode structure is dominated by a single poloidalharmonic and rotates in the thermal ion diamagnetic direction irrespective of the ∇βfsign.

5 EX/P6-22

C Alfven Mode Stability

ECEI dataReconstructed

BAAE

SimultaneousHFS/LFS

FIG. 6. 2D phase plots of a BAAE ob-tained from ECEI data (right). Thereconstructed mode (left) is domi-nated by a single poloidal harmonic.

RSAEs are strongly excited during on-axis injec-tion but only weakly excited during off-axis injec-tion (Fig. 7). Typically, the average amplitude ofthe RSAE activity on the ECE diagnostic is an or-der of magnitude stronger during on-axis injectionthan during off-axis injection. Vertical FIDA mea-surements indicate that, for the pair of dischargesshown in Fig. 7, the fast-ion profile is nearly flat nearqmin in the off-axis shot but is peaked on axis for thedischarge with on-axis injection. Thus, the large re-duction in mode amplitude is probably caused by areduction in the negative fast-ion gradient ∇βf thatdrives the instability.

The response of TAEs to changes in injection an-gle is more complicated (Fig. 8). TAEs that are ex-cited in the core during on-axis injection [Fig. 8(c)]are stable during off-axis injection [Fig. 8(d)]. In thecore, the fast-ion pressure is flat or hollow during off-axis injection, so the drive is veryweak. In contrast, radially extended TAEs in the outer half of the plasma have similaramplitudes for on-axis and off-axis injection [Fig. 8(a,b)]. This is expected: in the outerhalf of the plasma, swapping sources barely alters ∇βf (Fig. 4).

350 400 450 500 550Time (ms)

020

40

60

80

100

120

140

Freq

uenc

y (k

Hz)

350 400 450 500 550Time (ms)

#146076 #146077(a) On-axis NBI (b) Off-axis NBI

RSAEs

FIG. 7. Cross power of adjacent ECE channels that are located near qmin (at R ' 195 cm)during (a) on-axis and (b) off-axis injection. The same logarithmic color scale is used in bothfigures. In the off-axis case, the mode activity at 400 and 500 ms coincides with brief diagnosticblips of the on-axis beams.

Later in the discharge, BAAEs are observed for both on-axis and off-axis injection. Inspectrograms (Fig. 9), the BAAEs appear as patterns of regularly spaced discrete modesin the plasma interior (ρ ' 0.2). Large shifts in frequency between on-axis and off-axisinjection are consistently observed but the mode amplitude is comparable in the two cases,despite the large changes in ∇βf in the plasma interior.

6 EX/P6-22

350 400 450 500 550Time (ms)

020

40

60

80

100

120

140

350 400 450 500 550Time (ms)

40

60

80

100

120

140

Freq

uenc

y (k

Hz)

Freq

uenc

y (k

Hz)

#146076

#146076 #146077

#146077

(a) On-axis NBI, R=205 cm (b) Off-axis NBI, R=205 cm

(d) Off-axis NBI, R=187 cm(c) On-axis NBI, R=187 cm

FIG. 8. Cross power of adjacent ECE channels during on-axis (a,c) and off-axis (b,d) injectionfor a probe pair that is (a,b) outside and (c,d) inside of qmin. The same logarithmic color scaleis used for each probe pair (a & b, c & d).

800 1000 1200 1400 16000

20

40

60

80

100

120

800 1000 1200 1400 1600

#146080

Time (ms)

Freq

uenc

y (k

Hz)

#146083

Time (ms)

(a) On-axis NBI (b) Off-axis NBI

FIG. 9. Cross power of adjacent ECE channels that are located near the magnetic axis (atR ' 187 cm) during (a) on-axis and (b) off-axis injection. The same logarithmic color scale isused in both figures.

D Off-Axis Fishbones

Off-axis fishbones (also called “Energetic-particle-driven wall modes” [16]) are observedin high-beta plasmas on both JT-60U and DIII-D [17]. The fishbones can trigger resistivewall modes, compromising high-beta operation. Trapped fast ions drive the instabilityand are expelled in bursts by the mode [18].

In the initial DIII-D studies of off-axis fishbones [17, 18], all of the neutral beamsinjected on-axis. In new experiments, two sources inject off-axis. Unfortunately, sincesubstantial beam power is required to access the high beta regime where off-axis fish-bones are unstable, only 29% of the beam power is injected off-axis. Qualitatively, theeffect of this change on off-axis fishbone stability is modest. (The dependence of sta-bility on the q profile and on normalized beta are stronger than the accessible changes

7 EX/P6-22

-400-200

0200

0.960.981.001.02

3036.6 3037.0 3037.4 3037.8-0.50.00.51.01.5

Time (ms)

(a) MIRNOV (T/s)

(b) Neutrons (au)

(c) BES (au)

NPA (au)

#150331

FIG. 10. Time evolution of (a) magneticprobe, (b) neutrons, and (c) NPA and BESsignals during an off-axis fishbone burst. On-axis (off-axis) sources injected an averagepower of 7.0 MW (2.9 MW) at the time of thisburst; all sources inject in the co-current di-rection in this discharge. The active beamsviewed by the BES diagnostic are temporar-ily turned off at the time of this burst, sobeam emission makes no contribution to theBES signal. (The negative dip in the BESsignal following the burst is an artifact of AC-coupling.)

in distribution function.) Figure 10 showsan off-axis fishbone burst in one of the newexperiments. The new data are similarto the previous observations [18]. As themode grows, the magnetic waveform distortsand becomes increasingly non-sinusoidal nearmaximum amplitude [Fig. 10(a)]. The fish-bone burst causes a rapid drop in the neu-tron rate [Fig. 10(b)] that is consistent withthe loss of a portion of the trapped fast-ionpopulation. The modes expel fast ions to theplasma edge [Fig. 10(c)]. In this example,the losses are detected by a perpendicularly-viewing passive NPA and by a novel applica-tion of the BES diagnostic. At the time ofthis particular burst, no active beams are in-jecting in the sightlines of the NPA or BESdiagnostics. The NPA burst is caused by pas-sive charge-exchange reactions between ex-pelled fast ions and edge neutrals. The BESburst is caused by passive FIDA light that oc-curs when expelled fast ions react with edgeneutrals and emit Doppler shifted Dα emission in the passband of the BES optical filter[19].

4. Summary and Future Work

The main results and status of several studies with the new off-axis neutral beams aresummarized here.

1. Comparisons of injection by different neutral beam sources into MHD-quiescentplasmas have been completed [5]. All available measurements indicate that the fast-ion profile is broader with off-axis injection than with on-axis injection as expected;however, the off-axis sources produce ∼20% fewer fast ions than expected, probablybecause the beam power is lower than reported.

2. A study of the effect of the beam gradient on RSAE and TAE mode structure hasbeen completed [15]. The radial variation of the mode phase is insensitive to changesin ∇βf , consistent with recent calculations that include kinetic effects.

3. An experiment on the effect of∇βf on RSAE and TAE stability has been completed.The data are in qualitative agreement with expectations: Reduction in the negativegradient stabilizes and/or reduces the amplitude of both RSAEs and TAEs. Detailedcomparisons with theoretical predictions are planned [20].

4. Much new data on BAAEs have been acquired. The modes are cylindrical insta-bilities dominated by a single poloidal harmonic. Their stability is insensitive tochanges in ∇βf [21].

8 EX/P6-22

5. Off-axis fishbones are unstable in plasmas with mixed on-axis and off-axis beams[22].

This work was funded by the US Department of Energy under SC-G903402, DE-FC02-04ER54698, DE-FG03-97ER54415, DE-FC02-08ER54977, DE-AC02-09CH11466,DE-FG02-89ER53296, DE-FG02-08ER54999, and DE-AC05-0000R22725. Our debt tothe neutral beam group for reorienting and operating the tilted beam is gratefully ac-knowledged, as well as the valuable contributions of other members of the DIII-D team.

References

[1] MURPHY, C., ABRAHIM, M., ANDERSON, P.M., et al., Overview of DIII-D Off-Axis Neutral Beam Project, Proc. 24th Symposium on Fusion Engineering 2011.

[2] FU, G.Y. and VAN DAM, J.W., Phys. Fluids B (1989)1949.[3] GRIERSON, B.A., et al., Rev. Sci. Instrum. 83 (2012) in press.[4] PANKIN, A., McCUNE, D., ANDRE, R., BATEMAN, G., and KRITZ, A., Comp.

Phys. Comm. 159 (2004) 157.[5] HEIDBRINK, W.W., VAN ZEELAND, M.A., GRIERSON, B.A., et al., Nucl. Fusion

52 (2012) in press.[6] MURAKAMI, M., PARK, J.M., PETTY, C.C., et al., Nucl. Fusion 49 (2009) 065031.[7] SHARAPOV, S.E., ALPER, B., BERK, H.L., et al., Phys. Plasma 9 (2002) 2027.[8] VAN ZEELAND, M.A., KRAMER, G.J., AUSTIN, M.E., et al., Phys. Rev. Lett. 97

(2006) 135001.[9] HEIDBRINK, W.W., GORELENKOV, N.N., LUO, Y., et al., Phys. Rev. Lett. 99

(2007) 245002.[10] PACE, D.C., FISHER, R.K., GARCIAMUNOZ, M., HEIDBRINK, W.W., and

VAN ZEELAND, M.A., Plasma Phys. Controlled Fusion 53 (2011) 062001.[11] TOBIAS, B., CLASSEN, I.G.J., DOMIER, C.W., et al., Phys. Rev. Lett. 106 (2011)

075003.[12] SPONG, D.A., BASS, E.M., DENG, W., et al., Phys. Plasma 19 (2012) 082511.[13] VAN ZEELAND, M.A., HEIDBRINK, W.W., FISHER, R.K., et al., Phys. Plasma

18 (2011) 056114.[14] GORELENKOV, N.N., VAN ZEELAND, M.A., BERK, H.L., et al., Phys. Plasma

16 (2009) 056107.[15] TOBIAS, B.J., BASS, E.M., CLASSEN, I.G.J., et al., Nucl. Fusion 52 (2012) in

press.[16] MATSUNAGA, G., AIBA, N., SHINOHARA, K., et al., Phys. Rev. Lett. 103 (2009)

045001.[17] OKABAYASHI, M., MATSUNAGA, G., TAKECHI, M., et al., Phys. Plasma 18

(2011) 056112.[18] HEIDBRINK, W.W., AUSTIN, M.E., FISHER, R.K., et al., Plasma Phys. Controlled

Fusion 53 (2011) 085028.[19] HEIDBRINK, W.W., McKEE, G.R., SMITH, D.R., and BORTOLON, A., Plasma

Phys. Controlled Fusion 53 (2011) 085007.[20] HEIDBRINK, W.W., et al., Nucl. Fusion 53 (2013) to be submitted.[21] VAN ZEELAND, M.A., et al., Plasma Phys. Cont. Fusion 55 (2013) to be submitted.[22] MATSUNAGA, G., et al., EX/5-1.