direct measurements of particle flux along gap sides in castellated plasma facing component in...

F

Dp

RARa

b

c

d

h

••••

a

ARR2AA

KPGCTPSC

1

pttptgn

0h

ARTICLE IN PRESSG ModelUSION-7185; No. of Pages 5

Fusion Engineering and Design xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Fusion Engineering and Design

jo ur nal home p age: www.elsev ier .com/ locate / fusengdes

irect measurements of particle flux along gap sides in castellatedlasma facing component in COMPASS

enaud Dejarnaca,∗, Miglena Dimitrovaa, Michael Komma, Bernd Schweerb,lexis Terrab, Aurelien Martinc, Gontran Boizantec, James P. Gunnd,adomir Paneka, the COMPASS teama

Institute of Plasma Physics, AS CR v.v.i., Prague, Czech RepublicInstitute of Energy and Climate Research – Plasma Physics, Forschungszentrum Juelich, GermanyEcole Nationale Superieure des Arts et Metiers, FranceCEA, IRFM, F-13108 Saint-Paul-lez-Durance, France

i g h l i g h t s

We designed a probe to measure plasma deposition into gaps during tokamak discharges.Isat profiles are measured on both side of the gap for different gap orientations.Ion current is measured at the bottom of the gap in the toroidal orientation.Kinetic simulations reproduce well experimental profiles qualitatively.

r t i c l e i n f o

rticle history:eceived 27 August 2013eceived in revised form6 December 2013ccepted 8 January 2014vailable online xxx

eywords:

a b s t r a c t

In this paper, we report results of a dedicated experiment that gives the plasma penetration profiles insidea gap of a tokamak castellated plasma-facing component. A specially designed probe that recreates a gapbetween two tiles has been built for the purpose of this study. It allows to measure ion saturation profilesalong the 2 sides and at the bottom of the gap for both poloidal and toroidal orientations. The noveltyof such experiment is the real time measurement of the plasma flux inside the gap during a tokamakD-shaped discharge compared to previous experimental studies which were mainly post-mortem. Thisexperiment was performed in the COMPASS tokamak and results are compared with particle-in-cell

lasma depositionapastellationokamakrobeimulation

simulations. The plasma deposition is found to be asymmetric in both orientations with a stronger effectin poloidal gaps. The Larmor radius of the incoming ions plays a role in the plasma penetration only inpoloidal gaps but seems to have little impact in toroidal gaps. Profiles are qualitatively well reproducedby simulations. Ion current is recorded at the bottom of a toroidal gap under certain conditions.

© 2014 Elsevier B.V. All rights reserved.

OMPASS

. Introduction

In future fusion devices, as well as in present tokamaks,lasma-facing components (PFCs) of strong plasma–wall interac-ion regions are/will be castellated [1] in order to withstand stronghermo-mechanical stresses [2] due to intense, inhomogeneousarticle and heat fluxes. The direct consequence of this castella-

Please cite this article in press as: R. Dejarnac, et al., Direct measuremcomponent in COMPASS, Fusion Eng. Des. (2014), http://dx.doi.org/10

ion is an increase of the exposed surface of the PFCs due to theaps between tiles or monoblocks. Plasma and charge exchangedeutrals will flow into the gaps where tritium can be trapped.

∗ Corresponding author. Tel.: +420 266052944.E-mail address: [email protected] (R. Dejarnac).

920-3796/$ – see front matter © 2014 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.fusengdes.2014.01.008

Experimental studies in TEXTOR [3–5] have shown significantdeposited layers in the gaps of an ITER-like castellated test-limiterand an enhanced deposition at the bottom of the gap. Enhanced re-deposition of eroded material is not a candidate to explain this lastresult [6]. Numerical studies using particle-in-cell (PIC) techniquehave also been performed in order to understand mechanism thatgovern particle flux deposition in castellated PFCs [6–9]. In orderto assess experimentally the plasma deposition in such a complexgeometry as gaps, we have developed a special gap probe, so-calledSandwich Probe, that recreates a part of a castellated limiter with

ents of particle flux along gap sides in castellated plasma facing.1016/j.fusengdes.2014.01.008

2 tiles separated by a small gap. A dedicated experiment was per-formed on the COMPASS tokamak [10] using the sandwich probethat can measure the ion saturation current profiles along the 2 gapsides during plasma discharges. The experimental set-up as well as

dx.doi.org/10.1016/j.fusengdes.2014.01.008


http://www.sciencedirect.com/science/journal/09203796

http://www.elsevier.com/locate/fusengdes

mailto:[email protected]


IN PRESSG ModelF

2 eering and Design xxx (2014) xxx–xxx

tiponpbTc

2

2

oaapcmtAopIiCadsoflhib1aocssbeboswrotLpwsi

w(abiggo

ARTICLEUSION-7185; No. of Pages 5

R. Dejarnac et al. / Fusion Engin

he probe itself is detailed in Section 2. The results from the ded-cated experiment on COMPASS are presented in Section 3 of thisaper. Plasma deposition profiles are presented for the 2 possiblerientations of the probe, parallel and perpendicular to the mag-etic field lines. In Section 4, a comparison of these experimentalrofiles with results of 2D PIC simulations is presented, as well as arief description of the kinetic model used to achieve those results.he main findings are summarized in Section 5 under the generalonclusion.

. Experimental set-up

.1. Introducing the sandwich probe

The sandwich probe (SP) is a complex probe specially devel-ped for this dedicated experimental study. Its head is made in

TZM (titanium, zirconium, molybdenum) alloy monoblock, with 70 mm in diameter circular shape. The design of the sandwichrobe was based on several complex techniques, which made itsonstruction challenging. One of these was to cut a pocket in theolybdenum monoblock with a complex shape to make the body of

he probe that face the plasma, using the electro-erosion technique.nother very challenging point was to realize a metallic coatingf a thin molybdenum layer on boron nitride (BN). This has beenossible thanks to the Combined Magnetron Sputtering and Ion

mplantation technique [11] developed in Bucharest, Romania. Thiss the same technique that was used to make the W coating of JETFC tiles for the ITER-like wall project [12] but for the first timepplied to boron nitride. The final coating was not straightforwardue to the porosity of the BN but the final result is a successfullyolid Mo layer, which is less than 3 �m thick and with the hardnessf the Mo coating being 800–900 HV 0.025. The layer can surviveuxes in the order of 10 MW/m2. A coated sample was subjected toigh heat flux test using the High Temperature Test Facility (HTTF)

n Bucharest, Romania, where the heating occurs with an electroneam of 1.3 kW. The surface temperature was between 1200 and300 ◦C after a 5 s exposition. A cycle of one hundred pulses waspplied with these parameters. No delamination of the coating wasbserved during this thermal fatigue test. This metallic layer is toreate the base of the conducting segments that measure the ionaturation currents along the gap sides. For this purpose, we madeome parallel cuts in the coating till reaching the insulator locatedelow (i.e. the BN) in order to create measuring segments with anlectric insulation between them. The spatial resolution is limitedy the size of the cutting tool and we achieved the high spatial res-lution of about 0.3 mm. There are 6 measuring segments on bothides of the gap, covering a depth of 4 mm. The SP is also equippedith a conductive segment at the bottom of the gap, −15 mm with

espect to the entrance of the gap, in order to assess the depositionbserved in [4] and not reproduced in [6]. Fig. 1 shows three pho-ographs of the SP during its mounting. The SP is equipped with aangmuir probe (LP), protruding outside the cap, to measure locallasma parameters such as density and temperature. Those valuesill be used as input for the PIC simulations. In order to monitor the

urface temperature during experiments, a thermo-couple is alsomplemented in the body of the probe.

In Fig. 1(c), one can see the conductive segments that are biasedith negative voltage in order to collect the ion saturation current

Isat) during the discharge. The probe head can rotate around itsxis (vertical in Fig. 1) allowing to change the gap orientation toe either parallel or perpendicular to local magnetic field lines to


nvestigate the plasma deposition in toroidal gaps (TG) or poloidalaps (PG), respectively. In order to differentiate the 2 sides of theap, we name side A, the side without the LP and side B, the otherne. The gap is 1 mm wide.

Fig. 1. Detailed photographs of the SP and its inner components.

2.2. Experimental set-up

Measurements were performed on COMPASS with the SPmounted on an horizontal manipulator at the outboard midplane(OMP) location. The probe position was fixed during the dischargeand was 20 mm outside of the separatrix. The plasma configurationwas the standard single-null high triangularity D-shape plasmasused on COMPASS. The experiment was made in ohmic mode withnominal values of Bt = 1.15 T, Ip = 180 kA and the line-averaged elec-tron density varying in the range 2–6 × 1019 m−3 to study the effectof density on the plasma penetration. The segments were biasedwith a constant voltage Vbias = −100 V and the consequent ion satu-ration currents were measured as the drop voltage on independentresistors. The resistance value varies for each segment, increasingwith the depth of each segments in order to compensate the lowsignal on the deepest segments due to the expected decreasingdeposition. The values of each resistance were optimized for eachsegment with the expected currents estimated by PIC simulationsand vary in the range 47.5–330 �. The experimental campaign wasdivided into 2 sessions. On the first one, the SP was in the TG config-uration with the magnetic field lines parallel to the gap having ∼2◦

angle of incidence with respect to top surface of the probe head.4 discharges were performed at nominal magnetic field as above-mentioned and 2 additional discharges were performed at lowermagnetic field (Bt = 0.92 T) in order to investigate the effect of theLarmor radius of incoming ions on the plasma penetration in thegap. Then the probe was turned by 90◦ to position it in the PG con-figuration for the second session. Similar series of discharges (4 + 2)were performed. The probe was monitored by the in situ thermo-couple but during the 250 ms discharges on COMPASS operationswere safe with a recorded temperature always Tthermocouple < 80 ◦C.The Isat profiles were recorded on the 100 ms steady-state plateauof the discharges and averaged over 10 ms.

3. Experimental results

The first series of discharges deal with the TG orientation wherethe magnetic field lines are parallel to the gap. Fig. 2 shows the Isatspatial profiles along the side A (up) and side B (down) of the gapfor the series of discharges with the higher toroidal magnetic field.


Unfortunately, one can see that the two 1st segments are not work-ing, as well as segment #4 of side B. A disruption which happenedduring the tuning of the plasma configuration destroyed them. Theerror bars on the curves are the standard deviation of recorded


ARTICLE IN PRESSG ModelFUSION-7185; No. of Pages 5

R. Dejarnac et al. / Fusion Engineering and Design xxx (2014) xxx–xxx 3

FT

s4s

b(pwstotttstrtgosstSaritsttattiiimtwtvtmiifib

on the negative potential that can deflect ions to the shadowedside [14]. If we compare these profiles with the cases at lowermagnetic field (with ∼30% higher rL), the observed trend is con-served and we find that the decay length is increased by 15%

ig. 2. Spatial profiles of Isat along the side A (up) and side B (down) in the case ofGs with higher Bt (or lower rL).

ignals corresponding to their fluctuations. This curve also averages discharges with the different densities described in the previousection but at fixed, higher magnetic field (Bt = 1.15 T).

The variation between the absolute values of Isat profiles foroth sides and for the different electron densities is rather smallless than 5%) what makes us conclude that the penetration isractically not affected when varying the density. Such a resultas already observed numerically in a previous PIC study for ITER

cenarios [7]. The second important point shown in this graph ishat there is some non-negligible signal recorded at the bottomf the gap. The plasma does penetrate as deep as 15 mm into theoroidal gap under those plasma conditions. The deposits found athe bottom of the gap in TEXTOR [4] are therefore less mysteriouso explain but we have to be careful not to over-interpret this resultince we do not know the exact composition of the plasma reachinghe gap bottom (pure deuterium, hydrocarbon ions?), neither theespective ions energy. Moreover, the gap being open at each side,he plasma can come directly to the bottom without passing by theap entrance (top). However, the fraction of plasma at the bottomf the gap is rather low and it has to be noted that the size of theegment at the bottom of the gap is much larger (1 mm wide) than aingle collecting segment on the side (0.3 mm wide), which wouldend to decrease more its relative contribution in current density.urprisingly, in the series of discharges with lower magnetic fieldnd thus larger Larmor gyration, no signal is found at all. A moreecent dedicated experiment performed in TEXTOR to assess thisssue using a Quartz Micro Balance, to directly measure at the bot-om of a gap the injected CD4 in the far scrape-off layer, did nothow any deposition as well [13]. It is therefore difficult to conclude,he underlying physics being less straightforward than the size ofhe ion gyration. If we now compare both profiles, we observe ansymmetry in the deposition, with side A showing higher valueshan side B. With this orientation of the probe, side A correspondso the side that is favored by the ExB drift, which is consistent withts higher measured values. This feature was also predicted numer-cally in a previous study [14]. We have to note here that the gaps prone to an alignment error of ±2◦ with respect to B-field due to

echanical uncertainty of the probe head. We observe an exponen-ial decay on side A with a decay length �TG

high B = 1.35 ± 0.05 mmhereas the side B seems to have a parabolic distribution. However,

he crucial information lost with the failure of the 1st segment pre-ents us to conclude firmly on this feature for side B. We can notehat for these plasma parameters and for an electron temperature

easured by the LP of Te = 15 eV, the Larmor radius on incomingons is r = 1.36 mm at the probe location (assuming T = 2T ), which


L i e

s slightly larger than the gap width. In the case of lower magneticeld, giving a Larmor radius of rL = 1.72 mm at the probe location,oth sides distributions are symmetric as we can see in Fig. 3 (a

Fig. 3. Spatial profiles of Isat along both sides in the case of TGs with lower Bt (orhigher rL).

zoom has been performed on the first 6 segments discarding thebottom segment since it does not show any signal), with a similardecay length �TG

low B = 1.38 ± 0.05 mm.We can conclude that increasing the Larmor radius (when

already larger than the gap width in both cases) does not affectthe plasma penetration depth into the gap but favors having asymmetric deposition. Indeed, the larger the gyration, the higherprobability to hit the side which is not favored by ExB drift. Thisseems to confirm that the ‘parabolic’ shape seen in Fig. 2 is an arti-fact of the broken 1st segment. Nevertheless, we observe in Fig. 3that segments 3 and 4 show the same value, whilst in Fig. 2 (down)this feature is seen on segments 2 and 3, which corresponds for eachcase approximately to 1 rL. This is the consequence of ions having alarger gyration can reach a deeper distance on the non-favored ExBside

In the second series of discharges, the SP has a PG orienta-tion where the magnetic field lines are perpendicular to the gap.Unfortunately, we lost segment #2 of side B for this series ofshots. No current was recorded on the bottom segment, whichis more understandable since the entrance slit is considerablyreduced from a TG configuration. Fig. 4 shows that the depositionis mainly done on the wetted side (side A) with a decay length of�PG

high B = 0.67 ± 0.03 mm for higher magnetic field cases.We can note that side B collects a significant amount of cur-

rent on segment #3 at 1.35 mm from the entrance of the gap,value very close to the ions rL. This is due to a positive bump


Fig. 4. Spatial profiles of Isat along both sides in the case of PGs with higher Bt (orlower rL).


ARTICLE ING ModelFUSION-7185; No. of Pages 5

4 R. Dejarnac et al. / Fusion Engineering

Fb

gsmdpg

4

4

v[mcotolepcctLmblbsnt

4

elntessTne

ig. 5. Spatial profiles of Isat along side A and side B for a TG with higher Bt giveny PIC calculations.

iving �PGlow B = 0.77 ± 0.03 mm and that the ‘wetted’ area on the

hadowed side B seems to be larger. However, again here the infor-ation that we miss with the failure of segments #1, #2 and #4

oes not allow us to conclude on the latter point. Nevertheless, theenetration depth seems to be directly proportional to the Larmoryration on the wetted side for PGs.

. Numerical model

.1. The 2D particle-in-cell code

The numerical tool used for the comparison is a threeelocity–two-dimensional kinetic code based on PIC technique15]. The code is based on the resolution of the equations of

otion and the integration of Poisson’s equation to obtain the self-onsistent electric field that accelerates the particles. The noveltyf the code is its ability to inject arbitrary velocity distribution func-ions. For the ions, we use a non-Maxwellian distribution given by ane-dimensional quasineutral kinetic calculation of the scrape-offayer [16,17] that satisfies the kinetic Bohm criterion at the sheathntrance. The case considered here is a fully ionized magnetizedlasma with one species of singly charged ions (D+) incident on aompletely absorbing, conducting wall. The uniform magnetic fieldan have an arbitrary orientation. A magnetic sheath [18,19] canhus develop along the surface in the range of 4rL, where rL is thearmor radius. This has been taken into consideration by assuring ainimum distance of 10rL in between the tiles top and the plasma

oundary in order to have no perturbation of the bulk plasma. Theength of the tile tops is also taken large enough to avoid pertur-ations generated by the gap itself due to the periodicity of theystem. More detail can be found in [14]. In our simulations, we sete = 7.5 × 1018 m−3 and Te = 15 eV. Those values were retrieved byhe swept LP located on the cap of the SP.

.2. Results and comparison with experiment

Fig. 5 shows Isat profiles along the two gap sides for the TG ori-ntation by means of PIC calculations for Bt = 0.82 T (at the probeocation, corresponding to Bt = 1.15 T on axis). We can see that theumerical profiles reproduce well the experimental ones quali-atively. The asymmetry is well reproduced and we observe anxponential decay on side A and a more parabolic distribution onide B. However, the absolute values do not match and the profile on

high B,PIC


ide A decays faster than in Fig. 2 (we find �TG = 0.25 mm).he deposition profile in the gap depends strongly on the incli-ation angle [7] and this is where we also have the greatestxperimental uncertainty. Nevertheless, by increasing the angle by

PRESS and Design xxx (2014) xxx–xxx

a factor of 2, we increase the deposition depth by 50%, which is stillfar from the experimental value.

However, it has to be reminded that the SP was measuring atthe OMP of COMPASS in considerably turbulent plasmas seen withthe large error bars of Figs. 2–4. The PIC code used here does nottake into account the cross-field transport and such a discrepancycan be explained by this feature. Indeed, the source of the tur-bulent transport is known to be located at the outer midplane intokamaks [20] and in our ohmic discharges it was clearly dominat-ing the plasma penetration inside the SP during the experiments.Under such conditions it is not straightforward to draw quanti-tative conclusions. However, the relative trends are confirmed bythe calculations as shown in Fig. 5. Simulations of the Isat pro-files in the PG orientation show that the deposition is also stronglyasymmetric with the shadowed side recording almost no plasmaand the wetted side with a shorter decay length with respect tothe TG orientation cases. As in the experimental observations, wefind that �PG

high B,PIC ≈ 0.5 * �TGhigh B,PIC. Moreover, when increas-

ing the Larmor radius by decreasing the magnetic field, we finda +20% increase in the exponential decay length, which is in goodagreement with the +15% enhancement observed experimentally.In both orientations, TG and PG, no plasma is found deep at thebottom of the gap. Here again the turbulent transport must playa large role in this finding. To investigate such effect, we plan toperform in the future the same series of discharges but with the SPlocated in the divertor region of COMPASS. We can also compareOhmic discharges with H-mode discharges where the fluctuationsare strongly reduced.

5. Conclusions

A dedicated probe was specially designed to measure plasmadeposition profiles in a gap between tiles during ohmic D-shapedplasmas in the COMPASS tokamak. The plasma deposition profilesare found to be asymmetric on both sides of the gap and for bothpoloidal and toroidal orientations. In TGs, the side favored by ExBdrift shows an exponential decay and preferentially higher depo-sition. However, by increasing the Larmor radius of the incomingions, the asymmetry tends to diminish but the penetration depthremains unaffected. Density of the upstream plasma also does nothave any major impact on the plasma deposition profiles insidethe gap for both orientations. A surprising result shows that undercertain experimental conditions, ion current is measured at the bot-tom of the TG where simulations fail to reproduce it. However, thismeasurement was not reproduced in a series of discharges at lowermagnetic field. Nevertheless, it is a result in good agreement withdeposits found at the bottom of a castellated experimental limiterin TEXTOR tokamak and which could not be reproduced by sim-ulations. In PGs, the deposition is strongly asymmetric with thedeposition mainly on the wetted side of the PG, following a twicefaster exponential decay than TGs. Larmor radius of the incom-ing ions plays a role in the plasma penetration with a slightlydeeper penetration with larger gyration. Profiles are qualitativelywell reproduced by simulations but the quantitative comparison isstrongly affected by the weakness of our model. Indeed, the cross-field turbulent transport is not taken into account. The probe waslocated at the outer midplane and measurements show large fluc-tuations signature of a strong turbulence. Further experiments areplanned to measure in a reduced fluctuating plasma with the probein the near future.


Acknowledgments

The authors would like to thank D. Sestak and M. Bousek fortheir substantial help in preparing the BN parts to be coated with


ING ModelF

eering

MdtpfT

R

[

[

[

[

[[

[

ARTICLEUSION-7185; No. of Pages 5

R. Dejarnac et al. / Fusion Engin

o, mounting/dismounting the probe and fixing all issues we faceduring the construction phase of this special probe. We would likeo acknowledge FZJ workshop for their great work in making theocket of the probe cap in the TZM monoblock, as well as Dr. C. Rusetor his considerable help and patience in coating the BN samples.his work was supported by the project MSMT LM2011021.

eferences

[1] W. Daenner, M. Merola, P. Lorenzetto, A. Peacock, I. Bobin-Vastra, L. Briottet,et al., Fusion Eng. Des. 61 and 62 (61) (2002).

[2] J. Schlosser, F. Escourbiac, M. Merola, S. Fouquet, P. Bayetti, J.J. Cordier, et al.,Nucl. Fusion 45 (2005).

[3] A. Litnovsky, V. Philipps, P. Wienhold, G. Sergienko, B. Emmoth, M. Rubel, et al.,J. Nucl. Mater. 337–339 (2005) 917–921.

[4] A. Litnovsky, P. Wienhold, V. Philipps, K. Krieger, A. Kirschner, D. Matveev, et al.,


J. Nucl. Mater. 390–391 (2009) 556–559.[5] A. Litnovsky, V. Philipps, P. Wienhold, A. Kreter, A. Kirschner, D. Matveev, et al.,

J. Nucl. Mater. 415 (1) (2011) S289–S292.[6] D. Matveev, A. Kirschner, A. Litnovsky, M. Komm, D. Borodin, V. Philipps, et al.,

Plasma Phys. Control. Fusion 52 (7) (2010) 075007.

[[[[

PRESS and Design xxx (2014) xxx–xxx 5

[7] R. Dejarnac, M. Komm, J.P. Gunn, R. Panek, J. Nucl. Mater. 390–391 (2009).[8] M. Komm, R. Dejarnac, J.P. Gunn, A. Kirschner, A. Litnovsky, D. Matveev, et al.,

Plasma Phys. Control. Fusion 53 (2011) 015005.[9] M. Komm, R. Dejarnac, J.P. Gunn, Z. Pekarek, Plasma Phys. Control. Fusion 55

(2013) 025006.10] R. Panek, O. Bulykova, V. Fuchs, M. Hron, P. Chraska, P. Pavlo, et al., Reinstallation

of the COMPASS-D tokamak in IPP ASCR, Czech. J. Phys. 56 (Suppl. B) (2006)B125–B137, http://dx.doi.org/10.1007/s10582-006-0188-1.

11] C. Ruset, E. Grigore, H. Maier, R. Neu, X. Li, H. Dong, et al., Phys. Scr. T128 (2007)171–174.

12] C. Ruset, E. Grigore, I. Munteanu, H. Maier, H. Greuner, C. Hopf, Fusion Eng. Des.84 (2009) 1662–1665.

13] A. Kirschner, H.G. Esser, D. Matveev, O. Van Hoey, D. Borodin, A. Galonska, et al.,Phys. Scr. T145 (2011) 014005.

14] R. Dejarnac, J.P. Gunn, J. Nucl. Mater. 363–365 (2007) 560–564.15] C.K. Birdsall, A.B. Langdon, Plasma Physics via Computer Simulation, McGraw-

Hill, Inc., New York, 1985.16] V. Fuchs, J.P. Gunn, Proceedings of the 32nd EPS Plasma Physics Conference,

Tarragona, 2005.


17] J.P. Gunn, J. Nucl. Mater. 337–339 (2005) 310.18] R. Chodura, Phys. Fluids 25 (1982) 1628.19] J.P. Gunn, Phys. Plasmas 4 (1997) 4435.20] O.E. Garcia, J. Horacek, R.A. Pitts, A.H. Nielsen, W. Fundamenski, J.P. Graves,

et al., Plasma Phys. Control. Fusion 48 (1) (2006) L1–L10.


http://refhub.elsevier.com/S0920-3796(14)00009-X/sbref0005
























































































































































































dx.doi.org/10.1007/s10582-006-0188-1

























































































































































direct measurements of particle flux along gap sides in castellated plasma facing component in...

Documents