A comprehensive study on impurity behavior in LHD long pulse discharges

A comprehensive study on impurity behavior in LHD long pulse discharges ARTICLE IN PRESS JID NME [m5G; November 26, 2016;13 14 ] Nuclear Materials and Energy 0 0 0 (2016) 1–9 Contents lists available[.]

journalhomepage:www.elsevier.com/locate/nme

discharges

Y Nakamuraa,b,∗, N Tamuraa, M Kobayashia,b, S Yoshimuraa, C Suzukia, M Yoshinumaa,b,

M Gotoa,b, G Motojimaa,b, K Nagaokaa, K Tanakaa, R Sakamotoa,b, B.J Petersona,b,

K Idaa,b, M Osakabea,b, T Morisakia,b, the LHD Experiment Group

a National Institute for Fusion Science, National Institutes of Natural Sciences, Toki 509-5292, Japan

b SOKENDAI (Graduate School for Advanced Studies), Toki 509–5292, Japan

a r t i c l e i n f o

Article history:

Received 28 June 2016

Revised 14 October 2016

Accepted 12 November 2016

Available online xxx

Keywords:

Long pulse discharge

Impurity accumulation window

Radial electric filed

Impurity screening

Turbulent impurity transport

a b s t r a c t

Impuritybehavior isstudiedinavarietyofLHD(LargeHelicalDevice)longpulsedischarges,i.e stan-dardhydrogenplasmas,superdensecore plasmas,heliumplasmaswith ICH(IonCyclotronFrequency Heating),multi-speciesplasmasmixedwithHandHe.Densityscanexperimentsshowaspecificdensity rangeofimpurityaccumulationforonlyhydrogendischarges.Strongsuppressionofimpurity accumula-tivebehaviorisobservedinhightemperatureplasmaswithhighpowerheating.Themaincontributions

toimpuritytransportareextractedbyacomprehensivestudyonimpuritybehavior,i.e.investigatingthe criticalconditionsforimpurityaccumulationandtheparameterdependences.Itisfoundthatthe impu-ritybehaviorisdeterminedbythreedominantcontributions,i.e.neoclassicaltransportmainlydepending

onradialelectricfield,turbulenttransportincreasingwithheatingpowerandimpurityscreeningathigh edgecollisionalityintheergodiclayer.Themappingofimpuritybehavior onn-T(electrondensity and temperature)spaceattheplasmaedgeshowsaclearindicationofthedomainwithoutimpurity accu-mulationandprovidesoperationscenariostobuildupfusion-relevantplasmas

© 2016TheAuthors.PublishedbyElsevierLtd ThisisanopenaccessarticleundertheCCBY-NC-NDlicense (http://creativecommons.org/licenses/by-nc-nd/4.0/ )

1 Introduction

Understanding impurity transportis one ofthemajor tasksin

presentfusionresearch.Impuritiesintheplasmacoreenhance

ra-diation lossesandplasmadilutionwithdeleteriousconsequences

for fusion reactivity, and an uncontrolled impurity accumulation

mayeventerminate thedischarge.Recently, thechallengesof

us-ing the high-Z plasma facingmaterial W havebeen encountered

in fusion devices such as ASDEX and JET [1] The feasibility of

theuseoftungstenindivertortokamakshasbeenintensively

dis-cussed fordesigning next generation fusion devicessuch as ITER

andDEMO.Thesameissueisofcourseofrelevancetohelical

de-vices.Inparticular,impurityhandlinginlargesuperconducting

he-liotron/stellarator devices such as LHD [2] and W7-X [3], which

should demonstratequasi-stationary plasma operation, is an

im-portantsubject[4,5]

LHD long pulse dischargeswithNBI heatingshowed impurity

accumulative behavior on a long time scale (several seconds) for

∗ Corresponding author

E-mail address: nakamura.yukio@LHD.nifs.ac.jp (Y Nakamura)

high-Z impurities such as Fe, which come from the plasma fac-ing components The intrinsic impurities were accumulated in a specific range of impurity collision frequency (impurity accumu-lationwindow),whichwasaroundthetransitionfromtheplateau regimetothePfirsch–Schlüter(PS)regime[6,7].Impuritytransport studybyusingactiveimpuritypelletinjectionindicatedalong im-purityconfinementtime [8].Such an impurity behavior wasalso observed in other helical devices and a better understanding of impurity transportwasobtained fromtheinter-machine compar-ison [9] Theoretical predictions based on neoclassical transport theory fornon-axisymmetric configurations underline the impor-tanceofradialelectricfield.Inthestandardcasewithnegative ra-dialelectricfield,the so-calledion-root regime,high-Z impurities aredrawntowardthecenter.Onlyinthelow-densityregime,itis possibletoestablishtheelectronrootwithpositive radialelectric field, which flushes out impurities On the other hand, impurity transportstudiesinthescrape-off-layer(SOL)regiondemonstrated

afavorableimpactontheimpurityscreening,i.e.thescreeningof impurityinfluxfromthedivertorplates[10,11].Impuritytransport simulations indicated aclear physicalpicture ofimpurity screen-ing in the SOL [12] The cross-field heat conduction governs the http://dx.doi.org/10.1016/j.nme.2016.11.005

2352-1791/© 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ) Pleasecitethisarticleas:Y.Nakamuraetal.,AcomprehensivestudyonimpuritybehaviorinLHDlongpulsedischarges,Nuclear

Mate-ion energy transport across the islands under high density, low

temperatureconditions.The frictionforce dominatesover theion

thermalforce, dragging impurities outwards Turbulent transport

isanotherimportantcontributiontoimpuritytransportandmight

gainincreasingimportanceinfusion-relevantplasmasathigh

tem-peratures,butthereisnoclearevidenceinexperimental

observa-tion andalso in theoretical analysis including simulation studies

fornon-axisymmetricconfigurations

In order to predict impurity behavior in fusion-relevant

plas-mas,itisveryimportanttoextendtheoperationalregimetohigh

temperaturesandhighdensities withhigh powerheatingandto

explore favorable scenarios capable of preventing impurity

accu-mulation.Recently, highNBIheatingpowermorethan 10 MWis

availabletolong pulseoperation anda varietyofdischargeswith

hydrogen plasmas, super dense core (SDC) plasmas and helium

plasmashavebeencarriedoutinLHD.Inthispaper,impurity

be-haviorisinvestigatedinthesedischargesandacomprehensive

un-derstandingofimpurity behavior inLHD longpulsedischarges is

presented.Firstofall,experimental observationsaredescribedfor

each dischargeand some kind of parameter dependences of

im-puritybehaviorare indicated.Forunderstandingthemain

contri-butionstoimpuritytransport,theimpurity screeningeffectinthe

ergodiclayer,theroleofradialelectricfieldandtheimpactof

tur-bulenttransportinthecoreplasmaarediscussed.Finally,the

map-pingofimpurity behavioronplasmaparameter(n-T)spaceatthe

plasmaedgeispresentedandoperation scenariosforsteady-state

dischargesinLHDarediscussed

2 Experimental observation on impurity behavior

LHD is a large superconducting magnetic confinement device,

whichemploysaheliotronconfigurationwithapairofl/m=2/10

helicalwindings[13].Aconfiningmagneticfieldofup to2.9Tis

providedinsteadystatebymeansofthefullsetof

superconduct-ingcoils.Themagneticaxiswithoutplasmacanbeadjustedinthe

rangeofmajor radius R=3.6–3.9m andit isone ofkey

parame-tersforproducingSDCplasmas.TheLHDplasmahasanelongated

shapewithadouble-nullstructure andtheaverageplasmaradius

isabout0.6m.Longpulsedischargescanbeproducedbyonly

in-jecting the heating power of ECH, ICH and NBI and the plasma

densityiscontrolledbygaspuffingandpelletinjection

2.1 Standard hydrogen plasma with gas puffing

Firstofall,thedistinctivefeaturesofimpuritybehaviorin

stan-dardhydrogendischargeswiththe magneticaxisofR=3.6mare

described here Fig 1 shows a typical long pulse hydrogen

dis-charge with impurity accumulation The line averaged electron

density is controlled with hydrogen gas puffing during the

dis-chargetokeep aconstant densityof4.5×1019m−3.The radiation

power density in the plasma center (ρ=0) increases with time

though there is no bigchange in the radiation in theperipheral

region (ρ=0.945) The impurity line emission (Fe XXIII)

bright-ensintheplasmacoreandthecentralcarbondensitynC also

in-creasesremarkably withtime Fig 2 showsthe radial profiles of

electrontemperature,electrondensity,radiatedpowerdensityand

carbondensity.Theelectrontemperatureanddensityprofileshave

apeakedshapeandaflat one,respectively,andthey keepnearly

the same shapesduring the discharge In contrast,a remarkable

peakingisobserved inthe radiationandthe carbondensity

pro-files,whicharemeasuredby foilbolometerarraysandcharge

ex-changespectroscopy(CXS),respectively

Densityscanexperimentsinstandardhydrogendischargeswith

gaspuffingrevealtheimpurityaccumulationwindowasshownin

Fig.3.Inthiscase,thehydrogendischargesareconductedwiththe

NBIheatingpowerof 6.7MWatvarious plasmadensities in the

Fig 1 Typical long pulse discharge with impurity accumulation Time evolutions

of (a) heating power of neutral beam injection and averaged electron density, (b) central electron temperature and density, (c) radiated power densities in the center and the peripheral region, (d) impurity line intensity and central carbon density are indicated

magneticconfigurationwithR=3.6m.Therelativestrengthof im-purityaccumulationisestimatedbytheincreasingrateofradiated powerinthe coreplasma(dSrad /dt(ρ=0)).As foundpreviously,

no impurity accumulation occurs in the low-density operational regimeandtheremarkableincreaseofcoreradiationappearswith increasing density The increasing rate of coreradiation abruptly goesdownandnoimpurityaccumulationisobservedinthe high-densityoperationalregime.The densityrangeoftheimpurity ac-cumulation window depends on the plasma heating power and shiftstothehighdensitysidewithincreasingheatingpower.High temperature hydrogen plasmas with the heating power of more than10MWrevealanewaspectofimpuritybehaviorandthe ac-cumulationwindowvanishes,asdescribedinSection2.5

2.2 Super dense core (SDC) plasma with pellet injection

LHD can produce SDC plasmas (ne0=3∼ 5×1020m−3) [14, 15] and recentlya quasi-steady state dischargewith SDCplasma [16] in the magnetic configuration R=3.75m has been demon-strated over 4 by repetitivepellet injectionsasshown inFig.4 The sequential pellet injectiondrives the formation ofa strongly peakeddensityprofileasshowninFig.5(a).Thedensityprofile be-comesremarkablypeakeduntil4.7s.Thenthecentraldensity(ne

(ρ=0))decreasesandtheedgedensity(ne(ρ=1))increaseswith time.Thisisduetoedgeparticlerecycling,whichresultsinthe de-creaseofedge densitybystrong wallpumpingin theearly stage

of thedischarge andthe gradual increase withtime by reducing the capability of wall pumping The radiation intensity is mea-suredwithbolometerarraysofabsoluteextremeultraviolet photo-diodes(AXUVD)[17] andtheradiationsignalsareintegratedalong each chordline.A remarkableincrease ofradiationatthe central chord is observed in the early stage of the discharge The radia-tionprofiles areestimatedbyreconstructingthe two-dimensional distributionoflocalradiationandbysuperimposingeachmagnetic fluxsurface.Theremarkablepeakingoftheradiationprofileis

0.5

1.0

1.5

2.0

2.5

0 5 10 15 20

# 114887

Te

e (10

t = 5 s

Te

ne (a)

0 20 40 60 80

# 114887

t = 4 s

t = 7 s

Srad

3) (b)

0 2 4 6 8 10 12

# 114887

t = 3.54 s

t = 5.24 s

nC

-3) (c)

Fig 2 Radial profiles of (a) electron temperature and density, (b) radiated power density and (c) carbon density along normalized minor radius ρin the long pulse discharge (#114887)

0

5

10

15

3/s)

n e_bar (1019m-3)

Standard hydrogen plasma (P

nbi = 6.7 MW)

Fig 3 Impurity accumulation window in density scan experiment with standard

hydrogen plasmas The observable indicator for impurity accumulation is the esti-

mated increasing rate of central radiation

served as shown in Fig 5(c) The central radiation increases by

more than three timesat 4.7 incomparison with that at 4.3s,

despite keepingtheprofiles ofdensityandtemperature(Fig.5(a)

and(b)).Theincreaseofimpuritylineintensity(FeXIX)inthecore

plasma is alsoobserved and the globaltemporal behavior is the

sameasthatofthecentralradiation.Inthiscase,impurity

behav-ior isdominatedby thecoreimpurity transportinthePSregime,

wheretheimpurityfluxstronglydependsonthedensitygradient

asdiscussed intokamakplasmasandtheimpurities are

accumu-lated into the central region Anotherimportant feature in these

discharges is that the peaking of radiation profiles slowly stops

duringthedischargeandnoradiationcollapseduetoimpurity

ac-cumulationisobservedafterthesaturation(t>4.7s).This

satura-tioniscorrelatedtothereduction ofimpurityinfluxintothecore

plasmaasdescribedlater(Section3.1)

2.3 Helium plasma with ICH minority heating

InLHD,agreatdealofefforthassofarbeendevotedto

achiev-ing a steady state operation with high performance plasma and

great progresshasbeenmadeinterms ofdischargeduration and

injected energy[4,5] Thehighperformance plasmawithionand

electron temperatures of 2keV and an averaged electron density

of 1.2× 1019m−3 hasbeen sustained for 48min Long pulse

dis-charges with ICH were usually conducted in the scheme of

hy-drogen minorityheatingandheliumdominantplasmaswere

sus-tainedbycontrollingtheminorityratio(nH/nHe<0.1).Mostoflong

Fig 4 Long pulse operation with super dense core (SDC) plasma by multiple pellet

injection in the magnetic configuration R = 3.75 m Time evolutions of (a) averaged electron density and H αsignal, (b) central and edge electron densities, (c) central and edge electron temperature, (d) line-integrated radiation in the central and pe- ripheral chord, (e) impurity line intensities of Fe XIX and C III are indicated

pulsedischarges withICH were terminated by radiation collapse dueto the increase of plasmadensity or the penetration of im-purityflakesintotheplasma[18].However,therehasneverbeen

aneventoflong-termimpurityaccumulation,whichisobservedin hydrogendischarges.Thisnotableresultmightbeattributedtothe differencebetweenhelium andhydrogen plasmas,i.e.the change

ofimpuritytransportduetobackgroundions,asdescribedlater

2.4 Multi-species plasma mixed with H and He

Asdescribedin theprevious subsection,no impurity accumu-lation phenomenon was observed in long pulse discharges with

1

2

3

4

5

# 104258

t = 5.25 s

t = 4.3 s

t = 4.7 s

ne

-3 )

(a)

0.0 0.5 1.0 1.5

# 104258

t = 5.25 s

t = 4.3 s

t = 4.7 s

T e

(b)

0 2 4 6 8 10

0.0 0.2 0.4 0.6 0.8 1.0

# 104258

t = 5.25.s

t = 4.3 s

t = 4.7 s

(c)

Fig 5 Radial profiles of (a) electron density, (b) electron temperature and (c) local radiation intensity in the long pulse SDC discharge (#104258)

0.0

0.5

1.0

1.5

2.0

2.5

H rich plasma

He rich plasma

S rad

n

e_bar (1019m-3)

(a)

0.0

0.5

1.0

1.5

2.0

2.5

S rad

n

H / (n

H+n

He)

n

H/n

He~ 2.7

PF rad~ 0.5

(b)

H rich

He rich

n e_bar= 3.5 ~ 5 x 1019m-3

Fig 6 Dependences of radiation peaking factor on (a) average electron density and

(b) density ratio n H /(n H + n He ) for multi-species plasmas mixed with H and He The

peaking factor (PF rad ) is estimated by the ratio of core radiation (S rad (0.54)) to edge

one (S rad (0.945))

heliumdominantplasmas.Therefore, impurity behaviorin

multi-speciesplasmasmixedwithHandHeisinvestigatedtoclarifythe

effectof background plasma ions on impurity transport Density

scan in long pulse discharges shows a clear difference of

impu-rity behavior between H and He rich plasmas as shown in Fig

6(a), where the peaking factor of the radiation profile is plotted

asa function of average electron density The impurity

accumu-lationwindow isobserved forH rich plasmas,while thereexists

-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

P

nbi ~ 13 MW P

-3 /s)

n

e_bar (1019m-3)

Fig 7 Dependence of the increasing rate of core carbon density on averaged elec-

tron density for two different heating power cases (P nbi =7.5 MW and 13 MW)

no impurity accumulative behavior forHe rich plasmas Fig.6(b) showsthedependenceoftheradiationpeakingfactorontheratio

of hydrogeniondensity nH to total iondensity (nH+nHe) inthe dischargeswithne_bar=3.5∼ 5× 1019m−3.Spectroscopic measure-mentswiththelineemissionofneutralhydrogen(HI)andionized helium(HeII)providetheaverageiondensityratioofhydrogento helium (nH/nHe) [19], which is available to estimate the density ratio inthe plasma corefor steadystate plasmas [20].The radi-ationpeakingfactorPFradremains constantuntilthedensityratio

nH/nHe of 2.7and then abruptlyincreases ina transitional man-ner.Thisdrasticchangedependsontheparticletransportof back-groundions,whichinfluences theradial electricfield determined

byambipolardiffusionandalsothefrictionforce bythecollisions betweenimpurities andbackgroundions, asdescribedin Section

3

2.5 High temperature plasma with high power NBI heating

Recently, high NBI heatingpowers (Pnbi>10MW, td∼ 5s) are availableforsteady-statedischargesanditenablesustostudy im-puritybehavior inhightemperatureplasmas.Inthesedischarges,

itisfoundthattheimpurityaccumulativebehaviorisdramatically suppressed withhigher NBI heatingpower Fig 7 shows the de-pendence of the increasing rate of core carbon density on line-averagedelectrondensityfortwodifferentheatingpowercases.In thecaseoflowpowerheating(Pnbi=7.5MW),thedensitywindow for impurity accumulation is found in the carbon density

mea-surement asobserved before inradiation measurements (Fig 3).

However, the impurity accumulation window almost vanishes in

the discharges withhighpower heating(Pnbi=13MW) The high

temperature plasmasin the impurity accumulation windowhave

a large negativeradial electric field in the edge region andit is

expectedthat theimpuritiesaredriven intothecoreplasma.The

impurity screening in the ergodic layer dependson the

temper-ature anddensity inthe SOL, butthe plasmaparameters do not

satisfy the criterion for impurity screening in the ergodic layer

A new contribution to impurity transport is required for

under-standingtheresultinhightemperatureplasmasanddiscussedin

Section3.2.3

3 Comprehensive understanding of impurity behavior

The impurity evolution inplasmas dependson transport

pro-cessesinthethreeplasmaregions:(a)theplasma-surface

interac-tion zone, (b)the scrape-off layer (SOL)and(c) theplasma

con-finementregion.Insteadystatedischarges,impurityinfluxdueto

the sputteringofplasma facingcomponents remains almost

con-stantbecausethereisnobigchangeintheplasmaparametersin

the SOL during the discharge Therefore, the screening processes

intheSOLandtheconfinementpropertiesofthecoreplasmaare

comprehensivelystudiedusingexperimentalobservationsina

va-rietyoflongpulsedischarges

3.1 Impurity influx screening in the ergodic layer

Particleandenergytransport intheSOLhavebeenextensively

investigatedinhelicalsystemsaswellasintokamaks.Recent

stud-ies of edge impurity transport with the EMC3/EIRENE code

indi-cated a clear physical picture of impurity screening in the SOL,

whichis auniquefeature inhelicaldevices[12].The balance

be-tween friction and ionthermal forcesgoverns the edge impurity

transportandtheimpurityinfluxintothecoreisstronglyreduced

in the friction-dominated regime This reduction is caused by a

strong suppression ofthe thermalforce dueto the enhancement

ofcross-fieldheatconductionthroughthemagneticislands

There-fore,theratiooftheparallel(Q//D)totheperpendicular heatflux

(Q⊥D)canberegardedasakeyparameterfortheimpurity

screen-ingeffect[21]:

The field-line pitch  plays an important role in the

trans-port process Thecontribution ofcross-field transportremarkably

increases onlyfor helicalsystems with=0.0001∼ 0.001,which

is extremely different from ∼ 0.1 for tokamaks The impurity

screeningintheergodiclayercanbeexpectedwhenthecross-field

heatconductionbecomesevenmoredominantunderthecondition

ξi<1 [22].Thisgeometrical 3Deffectisunderinvestigation with

themulti-machinecomparison[23]

In the standard hydrogen discharges with gas puffing

(R=3.6m), the impurity accumulation is strongly suppressed

in the highdensityregion (Fig.3) The experimental database of

impurity behavior has been constructed by scanning the plasma

densityinlongpulsedischargeswithNBIheatingpowerbetween

1MWand9.5MW.The experimental study ofthe critical

condi-tion on impurity screening indicated an empirical scaling based

on the above impurity transport theory, i.e ξi∼ 0.11 [21] Here,

it is important to investigate whether ornot the scaling can be

appliedtootherdischargessuchastotheSDCplasmaortotheHe

rich plasma Fig.8 showsamapping oftheSDCplasmasandthe

standard hydrogen plasmas onthe two-dimensional space ofthe

electron density and temperature at the last closed flux surface

(LCFS) In this case, both SDC and standard hydrogen plasmas

0 100 200 300 400

R = 3.75 m

T e

n e LCFS (1019m-3)

SDC discharge

Standard H discharge

i = 0.09

dS rad/dt ~ 0

dS rad/dt > 0

Fig 8 Mapping of SDC plasmas and standard hydrogen plasmas on n-T space at

the plasma edge Both SDC and standard hydrogen plasmas are produced in the magnetic configuration R = 3.75 m The closed and open symbols indicate the plas- mas with and without impurity accumulation at each time in the discharges (e.g #

104258 in Fig 4 ), respectively The dashed line is fitted by the curve of constant ξi parameter

0 100 200 300 400 500

T e

n e LCFS (1019m-3)

H

* = 0.014

H (

i = 0.11)

E

r shielding

impurity screening

impurity accumulation

Fig 9 Mapping of He rich plasmas and standard hydrogen plasmas on n-T space

at the plasma edge The light gray symbols and open triangles indicate hydrogen plasmas and He rich plasmas, respectively The solid and dashed lines of light gray indicate the empirical scaling on impurity screening for hydrogen plasmas [21] The bold solid line is estimated for pure helium plasmas by using the same ξi parame- ter for hydrogen plasmas

are produced in the magnetic configuration with R=3.75m The closedsymbols indicate theplasmas withimpurity accumulation (dSrad/dt>0)andtheopen symbolstheplasmaswithoutimpurity accumulation(dSrad/dt∼ 0).Thecriticalconditioncanbeestimated

by thecurve ofconstant parameter ξi (seeEq (1)), which is ex-pressed asT=CH n0.4 witha specific constant value of CH given

by(χi ⊥ξi/2κi0)0.4.The curve withtheconstant CH∼ 125makes

a clear separation between impurity accumulation and impurity screeningregions.Assumingthattheperpendicularthermal diffu-sivityχi ⊥∼ 1 m2/sforLHDwith∼ 10−4,theconstantCH(= 125) corresponds totheratio oftheparallel tothe perpendicular heat fluxξi∼ 0.09,which isalmost thesameasthat (ξi∼ 0.11) inthe magneticconfigurationwithR=3.6m

ForHe plasmas, there is a possibilityof reducing theparallel heat conduction in comparison with the H plasmas because the coefficientofionheat conductivitydependsonthe massandthe chargeofbackgroundplasmaion(κi0∝mi−1/2Zi−2).Fig.9 showsa

mappingoftheHerichplasmasandtheprevious standard

hydro-genplasmasonthen-TspaceattheLCFS Thelightgraysymbols

(solidoropen)indicatethehydrogenplasmasandtheopen

trian-glestheHe richplasmas,inwhichthereexistsnoimpurity

accu-mulation over all densityrange The solid line oflight gray

rep-resentstheempirical scaling forimpurity screeningfor hydrogen

plasmas andthe solid line for pure helium plasmas, taking into

accountthedifferencesinthemassandthechargeincomparison

withpure hydrogenplasmas.Thoughtheimpurity screeningarea

expandstothelowcollisionalitysidefortheHeplasmas,itisnot

enoughtoexplainthesuppressionofimpurityaccumulationinthe

lowcollisionalityregion.Evenwiththat,theimpurityscreeningat

highedgecollisionalityiseffectiveforalllongpulsedischarges

3.2 Impurity transport in the core plasma

Impurity transport studies in the coreplasma havebeen

per-formedinvarioushelicaldevices[8,24,25],analyzingthetemporal

andspatialevolutionofimpurityradiationinresponsetotransient

impuritysources.Forthispurpose,impurityionsdifferentfromthe

intrinsicimpurities wereinjected intotheplasmaeitherby using

a tracer-encapsulated solid pellet (TESPEL) [26] or a laser

blow-off technique.The decaytimesofthecorresponding lineemission

ofthe highestionization statesafter the injectionwere taken as

a measure of the impurity confinement time The inter-machine

comparisonstudies show a clear density dependence of the

im-purityconfinementtime,whichincreaseswiththeplasmadensity

andyieldslongerconfinementtimesathigherdensity[9].This

re-sultmay be connected with neoclassical andturbulent impurity

transportandtheunfavorable densitydependenceisqualitatively

ingood agreementwithneoclassicaltheory,wheretheambipolar

radial electric field dominates impurity transport for high Z

im-purities In the tracer approximation without the interaction

be-tween impurities and backgroundions, the neoclassical impurity

fluxdensityinthebulkplasmacanbeexpressedby

z = −n z D z

11



z

12

11



(2)

forimpurityionswiththeionicchargeeZ,densitynZandthe

ra-dialelectricfieldErdeterminedbytheambipolarcondition[9,27]

Withincreasing Z,the second term inside the brackets becomes

dominantintheconvectiontermsanddependsstronglyonthe

ra-dial electric field Therefore, high-Z impurities are drawn toward

thecenterbynegativeErandpushedoutwardbypositiveEr.Asa

result,theimpuritybehaviorisstronglyinfluencedbythepolarity

oftheradialelectricfield.Ontheotherhand,thereisnoclear

evi-denceofturbulentimpuritytransportinexperimentalobservation

andtheoretical analysis includingsimulations In this subsection,

therole oftheradialelectricfield inimpuritybehavioris

investi-gatedfora variety oflongpulsedischarges andsome convincing

indications pointing to the essential role of anomalous impurity

transport are discussed for high temperature plasmas with high

powerNBIheating

3.2.1 Impurity core confinement time

Firstofall,theimpurityconfinementtimeisinvestigatedby

us-ingTESPELtoconfirmthesimilarityofimpuritybehaviorbetween

intrinsicimpuritiesandexternallyinjectedimpurities.Inthe

stan-dardhydrogendischarges,theimpurityconfinementtimeincreases

withthecollisionalityandtheextrinsicimpuritieshaveaverylong

confinementtime inimpurity accumulation window The similar

longconfinementbehaviorisseenforSDChydrogendischarges.On

theotherhand,inheliumdischargeswithoutimpurity

accumula-tion, the injected impurities are flushed out instantaneously For

hightemperaturehydrogendischarges,thestrongpumpingout of

-15 -10 -5 0 5 10 15

H (P

nbi = 3~5 MW)

H (Pnbi = 6~8 MW)

H (P

nbi = 13 MW)

E r

ion

*( = 1.0)

He plasma

H plasma

High temperature plasma

Fig 10 Radial electric field in the plasma edge region as a function of background

ion collisionality for H and He plasmas with different heating powers

thetracerimpurityisobservedasinheliumdischarges.Thus simi-larconfinementbehaviorisobservedbetweenintrinsicand extrin-sicimpurities.Anotherimportantpointisthattheextrinsic impu-ritieshavealongconfinementtimeeveninthehighdensityrange, wheretheSOLimpurity screeningmechanismbecomesactive,for standardhydrogenplasmasandSDCplasmas

3.2.2 Role of radial electric field

ErmeasurementshavebeencarriedoutinavarietyofLHD hy-drogenplasmaaswithawiderangeofdensityandmagneticaxis ThedependencesofEron densityandmagneticconfigurationare qualitatively consistent with neoclassical theory [28] In particu-lar,a significant changeofEr appears intheedge plasmaregion

Er in the edge regionchanges its signfrom positive inthe elec-tron roottonegative intheion rootwithincreasing density.Fig

10 shows the dependence of Er on background ioncollisionality (νion =ν/(vth/qR0))attheLCFS(ρ=1.0)forhydrogenandhelium plasmaswitha wide rangeofheatingpower anddensity.Inthis figure,onecanseethattheradialelectricfieldattheplasmaedge (ρ∼ 0.9)doesnotdependontheheatingpowerbutdependsonly

ontheioncollisionality.Therefore,thebackgroundion collisional-itycan be regardedasa goodpredictorof Er.Anotherimportant pointisthedifferencebetweenHandHeplasmasintheErvalue Forhelium plasmas,itis hardto enterthe ionroot(negativeEr) deeplyeveninthehighcollisionalityregion.Thisisduetothe dif-ference oftheparticleflux betweenHandHeionspecies,which stronglyinfluences theErdetermined bythe ambipolarcondition ( Zaana=0).TheheliumionfluxinpureHeplasmasis remark-ablyreducedincomparisonwiththehydrogenionfluxinH plas-mas andthe radial electric field does not reach a large negative valueeveninthehighdensityregion[29]

LHD has a large flexibility in changing the ambipolar radial electricfieldErbycontrollingtheeffectiverippleandthemagnetic topologyandtheimpactofEronimpuritytransportwasstudiedin differentmagneticconfigurations [21].Boththeimpurity accumu-lationwindowandErpointswereshiftedtothehighcollisionality sidewitha shiftingofthemagneticaxisoutwardandthecritical conditionforimpurity accumulationwasingoodagreementwith thechangeofEr.Moreover,recentsimultaneousmeasurementsof both key parameters (Er and coreradiation) indicate a direct ca-sualrelationshipbetweenErandimpuritybehaviorinthelow col-lisonality region asshown inFig 11.In this figure, no discharge existswithadensityofmorethan5× 1019m−3,thereby eliminat-ing theimpurity screeningeffectinthehighcollisionalityregion

0

2

4

6

8

10

12

H (standard)

He (standard)

H (high T)

3 /s) (

E

r (kV/m) ( ~ 0.9) High temperature plasma

He plasma

H plasma

Fig 11 Relationship between the radial electric filed at the plasma edge and the

increasing rate of core radiation in standard H, He and high temperature plasmas

In thestandard hydrogen discharges,when Er decreases with

in-creasingdensity,theobservableindicatorofimpurityaccumulation

(dSrad/dt) abruptly increases at a specific value of Er∼ −3kV/m

Theimpurity accumulativebehaviorisalwaysobservedintheion

rootregime withthelargenegativeErvalue (<−3kV/m).For

he-liumdischarges,Ercannotreachthespecificvalueevenatthe

den-sity ofaround 5× 1019m−3 and noimpurity accumulation is

ob-served (dSrad/dt∼ 0) From both results,it seems that the

impu-rity transport in the plasmacoreis attributedto theradial

elec-tricfield forHandHe standarddischarges.In SDChydrogen

dis-charges(R=3.75m),itishardtodiscusstherelationshipbetween

Erandimpurityaccumulationbecausethereisnodatabaseonthe

radialelectricfield.However,theEr canbeexpectedtobealarge

negative value fromthe plasmaparameters (collisionality) inthe

edge region and therefore theintrinsic impurities can enter into

theplasmacore.Thecoreplasmawithveryhighdensityisinthe

PSregimeandtheimpuritiesmaybedrawntowardthecenterdue

toasteepdensitygradientandaflattemperatureprofile,whichis

predictedbyneoclassicaltheoryintokamakplasmas[30].Atlast,

anyclearindicationofrelevance betweenErandimpurity

behav-iorcannotbefoundforhightemperatureplasmaswithhighpower

heating There isno impurity accumulationeven inhigh

temper-atureplasmas withlargenegativeEr values,whichis very

differ-entfromtheimpuritybehaviorinstandardhydrogenplasmas.This

suggeststhatthereexistssomekindofoutwardconvectiondriven

byturbulenceasdiscussedinthenextsubsection

3.2.3 Impact of turbulent transport

Intokamaks,impuritytransportstudieshavebeeninvestigated

with neoclassicaltransport theory andtemperature-screening

ef-fectinimprovedplasmaswithhighiontemperatureiswellknown

[31].Ontheotherhand,recentstudiesofturbulentimpurity

trans-port are starting todemonstrate promising qualitative and

quan-titative agreement between impurity measurements and

gyroki-netic turbulencesimulations.However, thescreening effectisnot

expected in helical system except for high collisionality regime

[9] andturbulentimpuritytransportisverylimitedinexperiments

and theoreticalanalysis forhelical devices Inour case, it isalso

difficulttoobtaindirectexperimentalevidenceofturbulent

trans-port.Therefore,someconvincingindicationspointingtothe

essen-tialroleofanomalousimpuritytransportaredescribed

First of all, in the experiment with impurity pellet injection,

anomalous coefficients originating from turbulent transport

pro-cesses haveto beused tofit theexperimental results.The

trans-port simulations reveal diffusion coefficients being one order of

-1 0 1 2 3 4 5

-3 /s)

R

ax/ L

Ti

= 0.5 ion* ~ 0.01

(E

r~ - 5 kV/m

at = 0.9)

High power heating

Fig 12 Correlation between the increasing rate of carbon density and ion tempera-

ture gradient in hydrogen plasmas with the constant ion collisionality ( νion ∗ ∼ 0.01)

at ρ=0.5 The ion collisionality corresponds to that of 0.025 at ρ=1.0 and

E r ∼ −5 kV/m at the plasma edge ( ρ∼ 0.9)

magnitudeenhanced overneoclassicallevelsforhightemperature plasmas[32].Secondly,asdescribedintheprevioussubsection,for standard hydrogen discharges withPnbi<10MW,impurity trans-port inthe coreplasmais dominated bythe radial electricfield, which is predicted by neoclassical theory for high Z impurities

TokeeptheErcontributionconstant,specificdischargeswith con-stantioncollisionalityare selectedfromthosewithvarious heat-ingpowers,becausetheioncollisionalityisagoodpredictorofEr Theturbulentcontributiontoimpuritytransportisextractedby in-vestigatingthedependenceoftheimpurity pinchontheion tem-peraturegradient,whichcandrive turbulence(ITG mode).Fig.12 showsthe correlationbetweentheincreasing rateofcarbon den-sityandtheiontemperaturegradient forhydrogenplasmaswith theconstantioncollisionality(νion ∼ 0.01)atρ=0.5.Theion col-lisionality corresponds to that of around 0.025 at ρ=1.0, where impurity accumulative behavior remarkablyappears asshown in Fig 9 In this case, the strength of the impurity pinch is calcu-lated by the time derivative of the carbon density at the mid-radius (ρ=0.5) At the sameradial position,the normalized log-arithmic ion temperature gradient Rax/LTi=−(Rax/Ti)dTi/dr is es-timated fromthe ion temperatureprofile measured by CXS One canseea significantdecreasingtrendoftheimpuritypinchwhen the temperature gradient is increased with the heating power Consequently, there is no observation of impurity accumulative behavior in high temperature plasmas with high power heating (Pnbi=13MW) In this database, as a matter of course, there is

no sensitivityto the radial electricfield, which depends on only ioncollisionality(Er∼ −5kV/m atρ=0.9 forνion ∼ 0.01) There-fore,thistrendcannotbeexplainedbyneoclassicalimpurity trans-port.Anotherimportantfeatureisseeninthecarbondensity pro-fileforhightemperatureplasmasasshowninFig.13.Thecarbon densityprofile becomeshollow withdecreasing the plasma den-sity, thereby observinga strong hollow profileof carbon,the so-called impurity hole in high ion temperature mode [33–35] Al-thoughthe impurityhole isobservedasatransientphenomenon

in high Ti mode, such a strong hollow carbon profile is sus-tainedforlongtimeduringthestandardhydrogendischargewith highpowerheating.Thehollownessofthecarbonprofilebecomes strongerwithdecreasingbackgroundioncollisionalityasshownin Fig.14,wherethenormalizedlogarithmiccarbondensitygradient

Rax/Lnc=−(Rax/nc)dnc/drisplottedasafunctionofion collisional-ity.Ingeneral,theradialelectricfieldincreaseswithdecreasingion collisionalityandbecomespositivetodrivetheimpuritiesoutward

1

2

3

4

P nbi = 13 MW

n C

-3 )

n e_bar = 4.34

n e_bar = 2.86

ne_bar = 1.52

Fig 13 Carbon density profiles in high temperature plasmas with different average

densities Each carbon density profile is maintained during the discharge

-25

-20

-15

-10

-5

0

5

10

R ax

ion

* = 0.5

peaked

hollow

Fig 14 Correlation between carbon density profile and ion collisionality The ver-

tical axis indicates the normalized logarithmic carbon density gradient (R ax /L nc ) at

the mid-radius

However,theradialelectricfieldsinthecoreregioncanbe

evalu-atedtobenegativeforhightemperatureplasmas,becausetheEris

amonotonicfunctionofminorradius inLHD [36] andhasa

neg-ativevalueattheplasmaedge(Fig.10).Therefore,thestrong

out-wardconvectionataroundthemid-radiuscannotbeexplainedby

neoclassicalimpurity transport alone Oneof mostprobable

can-didates for anomalous impurity transport is turbulence such as

theITGmode, whichcan be driven inhightemperature plasmas

(Rax/LTi>5) in LHD [37] Although turbulent impurity transport

simulation is under investigation in LHD, there are some results

onturbulentimpuritytransportfortokamakplasmas.Inareversed

magneticshearconfiguration,whichhasanegativemagneticshear

aswell asLHD, the curvature pinchprovides the main

contribu-tiontothetotal convectivevelocity andbecomesoutward in

tur-bulenttransport simulation [38] Comparison between measured

boronprofilesandgyrokineticsimulationsindicatesthat

thermod-iffusion(iontemperaturegradient)androtodiffusion(toroidal

rota-tiongradient)termscontribute totheoutward convection in

tur-bulentimpurity transport [39,40] Theseresults indicate that ITG

turbulence drives the impurities outward in experimental

obser-vationsandsimulationstudies.InLHD,recentstudiesonimpurity

transportinthedischargewithNBItorqueinputshowthattoroidal

rotationplaysanimportantroleintheimpuritytransport[41].The

impuritybehaviorinLHDmaybedominatedbyacombination

ef-R = 3.6 m

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

T LCFS

n

strong outward convection due to E

r and turbulence

impurity screening

in the ergodic layer

suppression of impurity accumulation due to turbulence impurity

accumulation due to Er

i= const

ion

*= const

Fig 15 Classification of impurity behavior on n-T space at the plasma edge in the

operational regime on LHD with R = 3.6 m The n-T diagram is made for standard hydrogen plasmas with flat density and peaked temperature profiles The dominant contribution to impurity transport is indicated in each domain The solid lines are based on the empirical scaling determined by the mapping of various discharges

fectofiontemperaturegradientandrotationgradienttermsin tur-bulentimpurity transport The anomalous impurity transport ob-servedinhightemperatureplasmashasagoodsimilaritywiththe recentresultsintokamakplasmas

3.3 Operational regime without impurity accumulation

Alargenumberoflongpulsedischargeshavesofar been per-formed toexplore theoperationalregime withoutimpurity accu-mulation Theexperimental database enablesusto make a map-ping ofimpurity behavior inthe steadystate operational regime Fig.15 showsadiagram forimpuritybehavior ontheplasma pa-rameters (n, T) at the plasma edge, which is made for standard hydrogenplasmas withgaspuffing inthe magneticconfiguration withR=3.6m.Inthesedischarges,theprofilestructuresindensity and temperature (flat density and peaked temperature) are kept forscanningthedensityandheatingpower.Thedegree of turbu-lenceincreaseswithincreasingheatingpower(temperature gradi-ent).Theimpuritybehaviorisbasicallydeterminedbythree domi-nantcontributionstoimpuritytransport:(a)neoclassicaltransport duetotheradialelectricfield,(b)anomaloustransportdueto tur-bulenceand(c)impurityscreeningintheergodiclayer.Inthelow collisionalityregionwithlownandhighT,theplasmahasalarge positive Er which prevents theimpurities from entering into the plasmacore.ThesteepiontemperaturegradientdrivesITG turbu-lence,inducing theoutwardconvection ofimpurityions.As a re-sult, extremely hollowimpurity profiles areobserved bya strong outwardconvectionduetothesynergeticeffectofpositiveErand turbulence On the other hand, in the high collisionality region withhighnandlow T,though theplasmahasalarge negativeEr drawingtheimpuritiesintotheplasmacore,theimpurity screen-ing bythefriction force iseffectiveintheSOL.Consequently,the intrinsicimpuritiesareretainedintheSOLregionandnoimpurity accumulationisobserved.Theboundaryconditionsaredetermined

bytheconstantioncollisionalityandξiparameter,whichare em-piricallyderivedfromtheexperimentaldatabase.Inthe intermedi-atecollisionalityregion,theplasmaisintheionrootregimewith negativeEr,whileturbulenceincreasesinintensitywithincreasing heatingpower As the Er does not dependon the heatingpower andtheneoclassicalpinchtermEr/Tidecreaseswithincreasingion temperature[42],turbulenttransportbecomesdominantinhigher

temperatureplasmas, thereby resultingin thestrong suppression

of accumulation.Here, the boundary conditionis empirically

de-termined by the heatingpower, Pnbi=10MW, which corresponds

toapowerdensityofapproximately1MW/m3 intheplasmacore

Fromthisfigure,itisfoundthatthedomainofimpurity

accumula-tionisrestrictedtothespecificcollisionalityrangeandvanishesin

thehighertemperatureregion,whichisabigadvantagefor

achiev-ingsteadystateoperationinfusion-relevantplasmas

4 Summary

A systematic study of impurity behavior has been performed

withavarietyoflongpulsedischarges.Somekindsofimpurity

ac-cumulativebehaviorareobservedandtheintrinsicimpuritiessuch

as iron and carbon are accumulated into the plasma core

Den-sity scans instandard hydrogen dischargesreveal a specific

den-sityrangeofimpurityaccumulation,i.e.animpurityaccumulation

window.InSDCdischarges,transientimpurityaccumulationis

ob-servedintheinitialstageofthedischargeandthereafterthe

long-termaccumulativebehaviordoesnotappear.Thisiscausedbythe

change ofedge densitydueto particle recycling Thereis no

ob-servationofimpurityaccumulationinsteadystatedischargeswith

ICHminorityheating,whichrequiresaverysmalldensityratioof

HtoHeions.Multi-speciesplasmasmixedwithHandHeshowa

distinctivedifferenceinimpuritytransportduetobackgroundions

Although there exists no impurity accumulation window for He

rich plasmas, the accumulationwindow appears in a transitional

mannerwithadecreasingdensityratioofHetoH.High

tempera-tureplasmaswithhighpowerheatingindicatethestrong

suppres-sionof impurityaccumulative behaviorandnoimpurity

accumu-lationisobservedovertheentiredensityrange

Theimpuritybehavioriscomprehensivelyinvestigatedfromthe

threepointsofviewofimpurity transport,i.e.impurityscreening

intheSOL,therole ofEr inneoclassicaltransportandthe

contri-bution ofturbulence.Impurityscreeningdueto thefriction force

at high edge collisionality is studied for standard hydrogen

dis-chargesandanempiricalscalingonthecriticalconditionisfound

This scaling can be applied to the SDC plasmas and the

multi-species plasmasmixed withHand He,and the impurity

screen-ing iseffectiveforall discharges.In theplasmacore, neoclassical

impuritytransportisbasicallydominatedbyEr,whichdependson

thebackgroundioncollisionality.SinceHeplasmasdonotenterin

a deepionrootregime witha large negativeEr eveninthe high

densityregion, the Er contribution issignificantly reduced inthe

impurity accumulationwindow.ThedirectcorrelationbetweenEr

andcoreradiationalso showsthe importanceof theEr

contribu-tion tocoreimpuritytransport forHandHeplasmas.Some

con-vincingcontributionsofturbulenceareobservedinhigher

temper-atureplasmas.Thesuppressionofimpurityaccumulativebehavior

increaseswithincreasingiontemperaturegradientunderthe

con-stantErcontribution.Carbondensitymeasurementsrevealastrong

hollowprofileinhightemperatureplasmasatlowdensityandthe

carbondensitygradientbecomesstrongerwithdecreasingion

col-lisionality.One ofmostprobablecandidates foranomalous

impu-ritytransportisturbulencesuchastheITGmode,whichdrivesthe

impuritiesoutwardfortokamakplasmas.However,further

investi-gationintoturbulenttransportisrequired

The impurity behavior in long pulsedischarges can be classi-fiedonthen-Tdiagramattheplasmaedgeanddividedintofour categories:(a)impurityscreeningathighedgecollisionalityinthe ergodiclayer,(b)strongoutwardconvectionduetopositiveErand turbulence inthe low collisionality region,(c) impurity accumu-lationby the dominant contributionof negativeEr, (d)turbulent transport exceeding neoclassicaltransport (Er/Ti pinch term) and suppressingimpurity accumulation From thisdiagram, operation scenariosofsteadystateplasmasinLHDcanbedevelopedtowards therealizationoffusion-relevantplasmas

Acknowledgments

TheauthorsaregratefultotheLHDTeamfortheirexcellent co-operation andto the device engineeringgroup of LHD for main-taininggoodoperatingconditions.Oneoftheauthors(YN)thanks therefereesfortheirimportantcommentsandfruitfuldiscussions This work is supported by the budget for the LHD project (NIF-SULRR702)

References

[1] M.N.A Beurskens , et al , Plasma Phys Control Fusion 55 (2013) 124043 [2] O Kaneko , et al , Nucl Fusion 53 (2013) 104015

[3] T.S Pedersen , et al , Nucl Fusion 55 (2015) 126001 [4] Y Nakamura , et al , Nucl Fusion 46 (2006) 714 [5] T Mutoh , et al , Nucl Fusion 53 (2013) 063017 [6] Y Nakamura , et al , Plasma Phys Control Fusion 44 (2002) 2121 [7] Y Nakamura , et al , Nucl Fusion 43 (2003) 219

[8] N Tamura , et al , Plasma Phys Control Fusion 45 (2003) 27 [9] R Burhenn , et al , Nucl Fusion 49 (2009) 065005 [10] M Kobayashi , et al , J Nucl Mater 390-391 (2009) 325 [11] M Kobayashi , et al , Nucl Fusion 53 (2013) 033011 [12] Y Feng , et al , Plasma Phys Control Fusion 53 (2011) 024009 [13] A Iiyoshi , et al , Nucl Fusion 39 (1999) 1245

[14] N Ohyabu , et al , Phys Rev Lett 97 (2006) 055002 [15] R Sakamoto , et al , Nucl Fusion 49 (2009) 085002 [16] H Yamada , et al , Nucl Fusion 51 (2011) 094021 [17] B.J Peterson , et al , Plasma Phys Control Fusion 45 (2003) 1167 [18] H Kasahara , et al , Phys Plasma 21 (2014) 061505

[19] M Goto , et al , Phys Plasma 10 (2003) 1402 [20] K Ida , et al , Rev Sci Instrum 86 (2015) 123514 [21] Y Nakamura , et al , Plasma Phys Control Fusion 56 (2014) 075014 [22] Y Feng , et al , Nucl Fusion 46 (2006) 807

[23] M Kobayashi , et al , Nucl Fusion 55 (2015) 104021 [24] R Burhenn , et al , Fusion Sci Technol 46 (2004) 115 [25] C Hidalgo , et al , Nucl Fusion 45 (2005) S266 [26] S Sudo , et al , Nucl Fusion 52 (2012) 063012 [27] H Maassberg , et al , Plasma Phys Control Fusion 41 (1999) 1135 [28] M Yoshinuma , et al , Plasma Phys Control Fusion 46 (2004) 1021 [29] H Sasao , et al , J Plasma Fusion Res Ser 3 (20 0 0) 431

[30] S.P Hirshman , D.J Sigmar , Nucl Fusion 21 (1981) 1079 [31] M.R Wade , et al , Phys Rev Lett 84 (20 0 0) 282 [32] S Sudo , et al , Plasma Phys Control Fusion 55 (2013) 095014 [33] M Yoshinuma , et al , Nucl Fusion 49 (2009) 062002 [34] K Ida , et al , Phys Plasma 16 (2009) 056111 [35] M Yoshinuma , et al , Nucl Fusion 55 (2015) 083017 [36] K Ida , et al , Nucl Fusion 45 (2005) 391

[37] M Nakata , et al , Plasma Phys Control Fusion 58 (2016) 074008 [38] S Futatani , Phys Rev Lett 104 (2010) 015003

[39] C Angioni , et al , Nucl Fusion 51 (2011) 023006 [40] E.J Casson , et al , Nucl Fusion 53 (2013) 063026 [41] Y Nakamura , in: 26th IAEA Fusion Energy Conference (Kyoto, Japan), 2016,

p EX/P8-4 [42] J.L Velasco , et al , Nucl Fusion 57 (2017) 016016