A study on the density shoulder formation in the SOL of h mode plasmas

A study on the density shoulder formation in the SOL of H mode plasmas ARTICLE IN PRESS JID NME [m5G; December 23, 2016;6 53 ] Nuclear Materials and Energy 0 0 0 (2016) 1–5 Contents lists available at[.]

journalhomepage:www.elsevier.com/locate/nme

plasmas

D Carraleroa,∗, J Madsenb, S.A Artenea,c, M Bernerta, G Birkenmeiera,c, T Eicha,

G Fuchertd, F Laggnera, V Naulinb, P Manzc, N Vianelloe,f, E Wolfruma, the EUROfusion

MST1 team1, the ASDEX Upgrade Teama

a Max Planck Institute for Plasma Physics, Boltzmannstr 2, 85748 Garching, Germany

b Aalto University, Espoo, Finland

c Physik-Department E28, Technische Universität München, Garching, Germany

d Max Planck Institute for Plasma Physics, Greifswald, Germany

e Consorzio RFX, Padova, Italy

f Ecole Polytechnique Fédérale de Lausanne, Swiss Plasma Center (SPC), Lausanne, Switzerland

a r t i c l e i n f o

Article history:

Received 15 July 2016

Revised 20 October 2016

Accepted 18 November 2016

Available online xxx

PSI-20 keywords:

ASDEX-Upgrade

Detachment

Edge plasma

Intermittent transport

H-mode

a b s t r a c t

Theterm“shoulderformation” referstoanincreaseofthedensitydecaylengthinthescrape-off layer (SOL)observedinmanytokamaksduringL-modeoperationwhenadensitythresholdisreached.Recent experimentsinASDEXUpgrade(AUG)andJET haveshownthattheshoulderformswhenthedivertor collisionality inthe divertorelectricallydisconnects filamentsfromthe wall.Thisleadstoatransition fromthesheathlimitedtotheinertialregimeandtoanenhancementofradialparticletransport,ingood agreementwithanalyticalmodels.Inthepresentwork,thevalidityofsuchamechanismisinvestigated

inthemorereactor-relevantH-moderegime.Forthis,acolddivertorH-modescenario isdevelopedin AUGusingdifferent levelsofDpuffing and Nseeding,inwhichinter-ELMfilaments and SOLdensity profilesaremeasured.Thebasicrelationbetweenfilamentsizeanddivertorcollisionalityisstillvalidin H-modeplasmas,albeitanadditionalcondition relatedtothegasfueling ratehasbeen foundforthe formationoftheshoulder

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

1 Introduction

The next generation of magnetic confinement fusion devices

willneedtosolvetheproblemofextremeheatandparticlefluxes

on Plasma-Facing Components (PFC),which will almost certainly

stretchavailablematerialstotheirtechnicallimitsonpowerloads

anderosionlevels[1].Inparticular,thepredictionofparticleand

heat fluxesontothemain vesselcomponentswillrequirethe

de-velopment ofaworkingmodelforperpendicular transportinthe

far Scrape-off Layer (SOL), including the propagation of

filamen-tary structures [2] Recent work carried out in ASDEX Upgrade

(AUG) [3] advanced in this direction by confirming

experimen-tallythepredictionsofanalyticalmodels forfilamentpropagation

[4-6],whichexplainedthebroadeningofL-modeSOLdensity

pro-∗ Corresponding author

E-mail address: daniel.carralero@ipp.mpg.de (D Carralero)

1 See http://www.euro-fusionscipub.org/mst1

filesobservedinmanytokamaks(sometimesreferred toas “den-sityshoulder”)[7,8]intermsofafilamentregimetransition.Itwas shownhowtheparameterregulatingshoulderformationisthe ef-fective divertorcollisionality, div [5] When thisparameter, rep-resentingtheproductofcharacteristicparalleltransport timeand theion-electroncollisionfrequencyinthedivertorregion,becomes

div>1,filamentselectricallydisconnectfromthewallthus tran-sitioningfromthesheathlimited[4]totheinertialregime[6].This greatlyenhancesradialtransport

According to these findings, baseline scenarios for ITER and DEMOwouldfeaturefullydevelopeddensityshoulders,asthe es-timatedvaluesofdivwillgreatlyexceedthat threshold[9].This couldbeofgreatpracticalimportance,sinceitcouldsubstantially increasetheparticlefluxarrivingatthefirstwall,aswell as con-tributeto spreadparticleandpowerfluxesonthedivertortarget Botheffectswouldhaverelevant consequencesforthe sputtering yieldfromthePFCsofbothregions,thuschangingthelifetimesof severalcomponents.However, so farthe regime transitionmodel

http://dx.doi.org/10.1016/j.nme.2016.11.016

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:D.Carraleroetal.,AstudyonthedensityshoulderformationintheSOLofH-modeplasmas,NuclearMaterials

de-vices will operatein H-mode Therefore, it is necessaryto prove

that shoulder formation does also happen in H-mode, and if so,

thatthesamemechanismfoundinL-mode applies.Regardingthe

firstquestion,severalexamplesofan H-modeshoulderformation

canbe foundintheliterature[10,11],butnosystematicstudyhas

beenrealizedsofar.Besides,thisproblemmustbeseparatedfrom

ELM-relatedtransport,whichusuallydominatestheSOLofpresent

daymachinesoperatingin H-mode,asITERandespeciallyDEMO

willmostlikelyfeaturesubstantially reducedlevelsofELM

activ-ity[12].Inthepresentwork,wepresenttheresultsofaseriesof

experimentscarriedoutonAUGwiththeaimofinducinga

shoul-derformation duringinter-ELM H-mode periodsin orderto

vali-datethefilament transitionmodelbymeasuring theevolution of

filamentsanddivduringtheprocess

2 Experiments

In order to evaluate the filament transitionmodel, the

evolu-tion offilament characteristics and SOL densityprofiles required

tobemeasured whilediv wasvariedacross theL-mode

thresh-old (div[0.1, 10]) in an otherwise stationary H-mode plasma

Thus,anewscenariowasdevelopedusingthesamemagnetic

ge-ometry(LSNedgeoptimizedconfiguration)andplasmaparameters

(BT=−2.5T,Ip=800kA,q95=4.85)asinL-modedischargesused

inthepreviouswork[3].SufficientECHandNBIpowerwasadded

toaccesstheH-mode,while keepingthetotalheatingpowerlow

enoughtoallowmidplanemanipulator(MPM)measurementswith

amultipinprobeheadastheoneusedinpreviousL-mode

exper-iments[3].Mainparametersoftheconsidereddischargesare

pre-sentedinTable1.Giventhepowerlimitation, thisscenariois far

fromLFSdivertordetachmenthencenitrogenseedingwasrequired

toobtainacold,collisionaldivertor.Inordertodisentanglethe

ef-fectsofnitrogenanddeuteriumfuelingonthedivvalues,

differ-entfuelingratesforbothgases(NrateandDrate)wereused,roughly

dividingthe data set in four scenarios:A) low power discharges

withlow Nrate andDrate values; B)dischargesincludingboth NBI

andECH heating, strong nitrogen seeding and a low Drate; C) A

dischargeinwhichdiv>1isachievedonlybymeansofastrong

densityfueling with no nitrogen; D) discharges with full power,

andbothhighNrate andDrate values

A typical dischargeis presentedin Fig.1, wherethree phases

can be distinguished: first, only 300kW of ECRH heating power

isused toestablish a referenceL-mode Next,full NBIandECRH

power is injected and H-mode withtype-I ELMs is accessed

Fi-nally,thedivscaniscarriedout byincreasingNrate and/orDrate

Thisendsthe type-I ELMsandreplaces them withsmaller, more

frequent ones Also, the divertor temperature is reduced, bring-ing theLFS divertorto differentlevelsof detachment.Duringthe whole discharge, divertor conditions and midplane density pro-files are measured, respectively, by a set of divertor fixed Lang-muirprobes [13] andthelithium beamdiagnostic [14].Also, the evolution of the main plasmadensity is moniteredusing a line-integrated interferometer measurement covering the region out-sidetheρ>0.875fluxsurface[15].diviscalculatedasexplained

in[3],usingdivertortargetvaluesofne andTeanda 1/5fraction

ofπRq95 asthecharacteristicparallellength(where Risthe ma-jorradiusofAUG,andq95isthesafetyfactorneartheseparatrix) Additionally,theprobeheadintheMPMprovidesthe perpendicu-larsizeandvelocityoffilamentsbycorrelatingionsaturation cur-rentmeasurements inpoloidally andradiallyseparated pins[15]

Ascanbe seeninFig.1,theMPMisplungedmultipletimes dur-ingthedischarge,coveringeachphase,andseveralvaluesofdiv

3 Analysis and results

3.1 ELM conditional averaging

In H-mode plasmas, ELMs introduce a newlevel of complex-ityinthepresentanalysis:inL-mode,fluctuationsdonot substan-tiallyalter thebackgroundconditionsofthe SOL,andboth diver-torconditionsandmidplanedensityprofilescanbemeasured sep-arately However, the ejection of ELMs inH-mode creates a ma-jorandintermittentperturbationacrossthewholeSOL.Therefore, the ELM cycle has to be taken into account when defining the turbulent characteristics oftransport Withthis aim,the H-mode phaseofeachdischargeisdividedin250mswindows(which pro-vide a sufficient numberof events) andELMs are detected using the thermoelectric current to the divertor, Idiv [16] This current

is mainly caused by the temperature difference between the in-nerandouterdivertor.Sincetheinner targetistypicallycold,Idiv

isagoodmeasureofthetemperatureintheouter divertorandit

isthusroutinely usedforthedetectionofELMs inAUG.Then di-vertorLangmuirprobesandlithiumbeammeasurements are syn-chronizedwithIdivinordertoobtainaconditionallyaveraged evo-lutionof divand midplanene profilesasa function oft−tE,max, where tE,max is the time of the maximum Idiv value Finally, for eachwindow, pre-ELM(t−tE,max=−2ms)densityand collisional-ityvaluesare taken,asrepresentativeoftheinter-ELMconditions

[17] The fraction of the ELM cycle corresponding to such inter-ELMstateincreasesasthesizeofELMsisreduced,goingtypically fromaround20% fortype-IELMs toover65% forthesmallELMs found at the endof the discharges This method is not applica-ble forMPM data, forwhich only the limited intervalsin which

Fig 1 The three phases on a typical discharge Top) Edge line integrated density N edge (blue) and heating power (P ECH in green, P NBI in red) are shown along with the radial position of the MPM (black) Middle) Divertor thermoelectric current (indicative of ELM activity), divertor T e and  div in the ρ= 1.02–1.04 area of the target in blue, red and green, respectively Bottom) Deuterium fueling and N seeding rates D rate (blue) and N rate (green) (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig 2 Experimental results a) Several inter-ELM Lithium beam density profiles during the two H-mode phases of the discharge Separatrix and far SOL e-folding lengths

λn,sep and λn,far are indicated b) Parameter diagram, where the collisionality and D rate of each discharge are displayed c) Evolution of inter-ELM λn,far with  div for the whole set of discharges The shaded area represents the range of L-mode values shown in [3] The color code in all plots corresponds to the four scenarios A-D presented in Table 1 (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

the manipulatorisinserted areavailable Inthiscase, Idiv isused

to separate “inter-ELM” probe data from “close to an ELM” data

(i.e., 3ms>t−tE,max>−2ms) Then, filament conditional analysis

isperformedon“inter-ELMdata” usinga2.5σthreshold,asin

pre-viouswork[15]

3.2 Shoulder formation

AccordingtothemodelvalidinL-mode,inthosedischargesin

which theDrate and/or Nrate are highenough to achievediv>1,

a shoulder formationshouldbe observed bythe endofthethird

phase (high collisionality H-mode) In Fig 2a, some inter-ELM

density profiles are presented at the beginning of the H-mode

(t=3.75s,dashedlines)andattheendofthethirdphase(t=6s,

solidlines).Ascanbeseen,discharges#33056and#33059exhibit

a clearflattening at theend ofthe discharge, while inthe other

two caseslittleornoflatteningcanbe seeninthesameinterval

In ordertoquantify thiseffect,two e-foldinglengthsare defined

on theprofiles: First,λn,far,representingthe gradient ofthe

pro-file inthe farSOL,isfitted intheradial rangeρp[1,1.04] This

parameter is equivalentto theλn used inprevious worksto

de-tecttheshoulderformation[3,15].Second,λn,sep,representingthe

evolution around theseparatrix, is fittedin the range ρp[0.98,

1.01] Also, div is calculated using the fixed divertor probes in

therangeρp [1.02,1.04].InFig.2b,theevolutionofdivthrough thedischargeisrepresentedasafunctionofDrateinthefour sce-nariosdescribed inSection 2 The samecolor codeis used asin

Table 1 (respectively, blue/green/black/red for scenarios A-D) As can be seen, independently of the levels of Drate only scenarios withhighNrateclearlysurpassthediv>1threshold.Instead,the low Nrate, lowDrate scenario remains clearlyunderthethreshold, andthe one with no nitrogen surpasses slightly div=1 by the endof thedischarge The differences betweenthe fourscenarios canbeseeninFig.2c,wheretherelationbetweeninter-ELMdiv andλn,farisshownusingthesamecolorcodeoverashaded back-groundrepresentingthetrendfoundinL-modeexperiments Con-sideringthe point cluster as a whole, a behavior similar to that

ofL-mode isfound, with λn,far rising when div>1, even if the increaseisnotassharpandtheλn,farvaluesarenotashigh How-ever,whenthefourscenariosareconsideredindividually,aclearer trend is found: discharges from scenario A display low levels of

div anddonot accessthe highertransport regime Thesame is validessentiallyforscenarioC,albeitslightlyhigherdiv(andthus

λn,far) valuesareachieved.Instead,alldischargesfromscenarioD develop a clearshoulder,achievinga significant increase inλn,far

athighervaluesofdiv.Interestingly,dischargesinscenarioB, dis-playingsimilardiv valuesasthose inscenarioD, fail toachieve highλn,farvalues

Fig 3 Inter-ELM filament size a) As a function of line integrated edge density b)

As a function of div Shaded area corresponds to L-mode values from [3] Color in-

dicates L/H-mode (For interpretation of the references to color in this figure legend,

the reader is referred to the web version of this article.)

3.3 Filament characteristics

The main resultsfromaconditional analysisofMPM dataare

shownin Fig.3:theevolution ofthefilament sizewiththeedge

line-integrateddensityshowstwothresholdsabovewhichthesize

increases sharply: the first one, observed in the L-mode phase,

takes place around Nedge=2× 1019m−3, and coincides with the

L-modeshoulder formation detected previously for300kW

oper-ation [3] The second, observed in H-mode, is observed around

Nedge=3.75× 1019m−3, and would correspond to an equivalent

thresholdforH-mode.Instead,whenthesamedatapointsare

rep-resentedasafunctionofdivinFig.3b,thetwothresholdsmerge

arounddiv=1,ingoodagreementwiththeexpectationsfromthe

filamentmodel,andsuggestingthatthesamedisconnection

mech-anism applies in H-mode Still, as with the shoulder formation,

thetransitionisnot asclearastheonefoundinL-mode.Besides

that,some propertiesofthefilamentschangewhengoing fromL

toH-mode:first,ascanbe seeninFig.3,pre-transitionfilaments

tendto be larger insize in H-mode.Also, the relative amplitude

offilaments(expressed as theratio betweenthe standard

devia-tionandthemeanvaluesofionsaturationpins,σ/μ)isincreased

by30%inH-mode(fromσ/μ∼ 0.35−0.4toσ/μ∼ 0.45−0.55).This

effectis mostly due to the reduction in densityin the SOL, and

disappearswhenσ/Nedgeisusedinstead.Finally,thedetection

fre-quencyisreducedfordiv<1,withtypicalvaluesintherangeof

fd[2000,2500] −1 andfd[1000,1500] −1 forLandH-mode,

re-spectively.Awordofcaution isinorderhere,asfd ismostlikely

not onlyaffected by the SOLconditions butalso by thefilament

generation frequency,which can be expectedto change

substan-tiallywiththeformationoftheH-modepedestal

3.4 Near SOL evolution

Besides theevolution ofλn,far,therelation ofthedensity

gra-dientattheseparatrixwiththeprocessofshoulderformationhas

beenobserved: AsshowninFig.4,bycomparingλn,far andλn,sep

inthe dataset, it could be concluded that they are uncorrelated,

andthusλn,sep is neithercorrelated withthe div parameter

re-sponsibleforshoulderformation.Instead,λn,sepseemstobeclearly

correlatedtoDrate indischargeswithvariablelevelsofgasfueling

(nosuchrelationisfoundbetweenNrate andλn,sep,though).Also,

theλn,sepvalueisproportionaltotheELMfrequency

4 Discussion and conclusions

The firstconclusioncouldbe thataphenomenon analogousto

the well documented L-mode shoulder formation can be found

when a density threshold is surpassed under certain conditions

tor Asthe datainFig.2 reveal,thishappensfordischargeswith low Drate, indicating the existence of a second threshold for the shoulder,relatedtothedeuteriumpuffinglevel.Thisisnotrelated

tothefuelingofthemainplasma:ascanbeseeninFig.5a,ahigh

Nedgedoesnottriggertheshoulderformationeither,sinceScenario

B reachesthesamerangeofedge densitiesasScenarioD There-fore,the mechanism relatingfueling and shoulder formation can probably be found in the SOL One possible explanation forthis wouldbethattheshoulderformationrequiresahighrecyclingrate

atthemainwallontopoftheincreasedconvectivetransport[18] Thiswouldbeingoodagreementwithrecentexperimentswhich show how far-SOL ion temperatures drop after the shoulder for-mation[19,20],suggestingthatalargefractionoftheionsinitdo not come fromthe confinedplasma butfromionization of recy-cledneutrals.In such acase, a minimumlevelofneutraldensity

inthefarSOLcould berequiredtostart theprocess,thus requir-ing a minimumvalue of Drate The precise determination ofthis secondthresholdandtheidentificationoftheunderlyingphysical mechanismwillbethesubjectofforthcomingwork

Asecond conclusionwouldbethat asimilar regimetransition seems totake placeasin L-mode,since thefilament sizeis sub-stantially increased when high collisionality disconnects the SOL from the wall Also, the relation between transition in the fila-mentdynamicsandshoulderformationseemstoholdinH-mode,

as they both sharethe same div and Nedge thresholds A more directcomparisoncan be seeninFig.5b,where thesizesof fila-mentsmeasuredduringsome dischargesfromFig.2areshown.A generalcorrelation betweenfilamentsize andλn,far appears, sug-gestingarelationbetweenincreasedfilamentarytransportandthe flattening of the profiles This correlation would also mean that largefilamentsonly appearwhenashoulder isformed However, filament anddensitydataare not both available inall cases(not all dischargeshadthe MPMequippedwiththe rightprobehead, probedataislimitedtothereciprocations,etc.), sogiventhe lim-ited amountofdata andthelarge errorbars inthefilament size calculation,no conclusivestatementcan bemadeyet.The clarifi-cationofthissubjectwillalsobeaddressedinforthcomingworks Finally, the density gradient length around the separatrix has been found to be largely independent ofthe collisionalityat the divertor, but correlated to Drate and proportional to the ejection frequencyofELMs The evolutionofλn,sep withDrate andNrate is consistent withrecentstudies onthe high-fieldsidehighdensity (HFHSD)regioninAUG[21,22].Thisregion,whichtendstoflatten thedensity gradientatthe separatrix,forms asDrate isincreased (whichwouldcorrespondinFig.4atotheincreaseofλn,sepwhen goingfromScenarioAtoBandfromBtoC)andhasbeenproven

tobestronglyreducedwithnitrogenseeding(whichwould corre-spondinFig.4atothereductionofλn,sepfromScenarioCtoD) Summarizing, the formation of a shoulder has been observed

in inter-ELM H-mode plasmas The general link between shoul-der formationandfilamenttransitionalsoseems toremainvalid Collisionality remains the necessary condition for the shoulder formation, but deuterium fueling seems to play an additional role These results allow the extension of the general shoulder

Fig 4 Inter-ELM separatrix density e-folding length, λn,sep a) as a function of λn,far b) as a function of puffing rate D rate c) As a function of ELM frequency Color code as in Fig 2 (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig 5 a) Relation between filament size and λn,far b) Scaling of λn,far with edge

line integrated density, N edge Same set of discharges and color code as in Fig 2

(For interpretation of the references to color in this figure legend, the reader is

referred to the web version of this article.)

formation mechanism validated in L-mode to H-mode plasmas

withtheintroductionofafewnewelements.Futureworkwill

ad-dress the interplayof filamentary transport andneutral

penetra-tionintheSOLinordertoclarifytheprecisemechanismrelating

Drate withλn,far.Thisis anecessarystep onthewaytoa general

scalingoftheSOLwidth,capableofimprovingcurrentpredictions

forITERandDEMOoperation

Acknowledgments

This work hasbeen carried out within the framework of the

EUROfusion Consortium and has received funding from the

Eu-ratom research and training programme 2014–2018 under grant

agreement No 633053 The views and opinionsexpressed herein

donotnecessarilyreflectthoseoftheEuropeanCommission

References

[1] A Loarte , M Sugihara , M Shimada , et al , 22nd IAEA Fusion Energy Conference,

2008 [2] S.I Krasheninnikov , Phys Lett A 238 (2001) 368–370 [3] D Carralero , P Manz , L Aho-Mantila , et al , Phys Rev Lett 115 (2015) 215002 [4] S.I Krasheninnikov , D.A D’Ippolito , J.R Myra , J Plasma Phys 74 (2008) 679717 [5] J.R Myra , D.A Russell , D.A D ´Ippolito , Phys Plasmas 13 (2006) 112502 [6] O.E Garcia , R.A Pitts , J Horacek , et al , J Nucl Mater 363–365 (2007) 575–580

[7] B LaBombard , R.L Boivin , M Greenwald , et al , Phys Plasmas 8 (2001) 2107 [8] D.L Rudakov , J.A Boedo , R.A Moyer , et al , Nucl Fusion 45 (2005) 1589–1599 [9] D Carralero , H.W Müller , M Groth , et al , J Nucl Mater 463 (2015) 123–127 [10] H.W Müller , M Bernert , D Carralero , et al , J Nucl Mater 463 (0) (2015) 739–743

[11] H.J Sun , E Wolfrum , T Eich , et al , Plasma Phys Control Fusion 57 (12) (2015)

125011 [12] R.P Wenninger , M Bernert , T Eich , et al , Nucl Fusion 54 (2014) 114003 [13] M Weinlich , A Carlson , Contrib Plasma Phys 36 (S) (1996) 53–60 [14] M Willensdorfer , G Birkenmeier , R Fischer , et al , Plasma Phys Control Fusion

56 (2014) 025008 [15] D Carralero , G Birkenmeier , H.W Müller , et al , Nucl Fusion 54 (2014) 123005 [16] A Kallenbach , R Dux , J.C Fuchs , et al , Plasma Phys Control Fusion 52 (2010)

055002 [17] A Burckhart , E Wolfrum , R Fischer , et al , Plasma Phys Control Fusion 52 (2010) 105010

[18] B Lipschultz , D Whyte , B LaBombard ,Plasma Phys Control Fusion 47 (2005) 1559–1578

[19] D Carralero , H.J Sun , S.A Artene , et al , 42th EPS Conference on Plasma Physics, Lisbon, Portugal, 2015

[20] S.Y Allan , S Elmore , G Fishpool , et al , Plasma Phys Control Fusion 58 (2016)

045014 [21] S Potzel , M Wischmeier , M Bernert , J Nucl Mat 463 (2015) 541–545 [22] F Reimold , M Wischmeier , S Potzel , et al , J Nucl Mat Energy (2016)