j n .rm .

,a l o f
n u c l e a r
ELSEVI ER Journal of Nuclear Materials 241-243 (1997) 993-997
Wall conditioning and particle control in Extrap T2
H . B e r g s ~ k e r b , . , D . L a r s s o n b, p . B r u n s e l l a, A . M i S l l e r a, L . T r a m o n t i n c
a Department of Fusion Plasma Physics, Alft?n Laboratory, Royal Institute of Technology, EURATOM/ NFR Fusion Association,
100 44 Stockholm, Sweden
b Department of Physics Frescati, Royal Institute of Technology, EURATOM/ NFR Fusion Association, 104 05 Stockholm, Sweden
c Dipartimento di lngegneria Elettrica, Universith di Padot,a, Padua, Italy
Abs t ract
The Extrap T2 reversed fi el d pinch experi ment is operated with the former OHTE vacuum vessel, of di mensi ons
R = 1.24 m and a = 0.18 m and with a compl et e graphite liner. It is shown that a rudimentary density control can be
achi eved by means of frequent hel i um gl ow discharge condi t i oni ng of the wall. The standard He- GDC is wel l characterized
and reproducible. The trapping and release of hydrogen and impurities at the wal l surfaces have been studied by mass
spectrometry and surface analysis. The shot to shot particle exchange bet ween wall and plasma can be approxi mat el y
accounted for.
Keywords: EXTRAP-T2; Boundary plasma; Reversed field pinch; Wall pumping; Wall conditioning
1. I nt roduct i on
The Extrap T2 experi ment aims at i mprovement s of the
reversed field pinch concept with resistive shell operation
[1]. In its first stage, the experi ment is operated with the
former OHTE vacuum vessel with a compl et e graphite
liner [2,3]. One target has been to reproduce the OHTE
results, though with i mpr oved diagnostic facilities. An-
other maj or obj ect i ve has been to study pl asma surface
interactions and met hods of density control and impurity
control with a graphite wall [4,5].
The di mensi ons of T2 are maj or radius R = 1.24 m and
mi nor radius a = 0.183 m. The gl obal pl asma parameters
so far have been in the ranges of 100- 260 kA plasma
current, 1- 7 10 ~9 m- 3 line averaged density and 100- 200
eV electron temperature. The l oop vol t age has been 70- 130
V, the pulse length 4 - 1 4 ms and the energy confi nement
t i me 50- 80 Ixs. These data are comparabl e to the OHTE
performance [2,3].
The main tools for wall condi t i oni ng have been gl ow
discharges in helium, baking and condi t i oni ng by repeated
RFP discharges. To study the effects of these procedures
* Corresponding author.
on the wall surfaces, quantitative mass spectrometry and
surface analysis of passive probes have been used.
2. Expe r i me nt al
The T2 devi ce has normally been operated at room
temperature. It is routinely baked up to ~ 130C, each
time for about eight hours. The base pressure is around
10 - 5 Pa. The effect i ve pumpi ng speed on the vessel is
limited by the size of the pump ports to roughl y S ~ 0.25
m3 / s at base pressure, S = 0.175 m3 / s at 0.5 Pa. At an
earlier stage the discharge gas was supplied in continuous
flow, to a vessel pressure of 0. 1- 0. 7 Pa. Later, gas puffs of
~ 20 s duration were introduced. For hydrogen removal
and cl eani ng an r f assisted dc glow discharge in hel i um at
r oom temperature with a single recessed anode has been
used [4]. A gl ow which is toroidally uniform within a
factor two can be obtained at a pressure of ~ 0.7 Pa, with
~ 650 V anode potential and an average current density of
80 mA/ m 2 wall area [4]. These parameters will be re-
ferred to bel ow as the standard He- GDC. Occasi onal l y a
gl ow discharge in hydrogen with similar parameters, fol-
l owed by a He- GDC has been made in order to r emove
oxygen more efficiently.
0022-3115/97/$17.00 Copyright 1997 Elsevier Science B.V. All rights reserved.
PI! S0 0 2 2 - 3 1 1 5 ( 9 6 ) 0 0 6 4 6 - 0
9 9 4 H. Bergshker et a l . / Journal of Nucl ear Materials 241- 243 (1997) 993- 997
The edge plasma density and temperature were mea-
sured with a Langmuir triple probe of similar geometry as
in [6]. A quadrupole mass spectrometer mounted directly
on the vessel could be operated up to a total pressure of
~ 10 2 Pa. Another, differentially pumped mass spec-
trometer was calibrated for N 2, H 2 and He in the range of
vessel pressure from 0. I-10 Pa. A probe manipulator
system allows passive probes to be exposed to complete
RFP discharges or to wall conditioning. Following expo-
sure the probes are transferred via a portable UHV cassette
to either of two surface analysis stations, for ion beam
analysis or for X-ray photoelectron spectroscopy and scan-
ning Auger electron spectrometry. In particular, graphite
and silicon probes have been exposed at the wall position
and in slightly retracted positions. The samples have been
transported at 10 -4 Pa or better, and surface analysis
could be performed typically within a few hours after
exposure.
3. Results
Fig. 1 shows the plasma current and density evolution
in three representative examples of RFP discharges. Char-
acteristic is the wide range of different densities that can
occur, depending on the status of the graphite wall. Shot
number 2531 was preceded by a 60 min glow discharge in
helium. The density increase which is observed from
around t = 3 ms in these three discharges is representative
and is often accompanied by an increase in radiation from
carbon and oxygen and an increase in plasma resistance.
Fig. 2 shows the edge temperature and edge density at the
time of maximum current as a function of line averaged
?E 6 0 2 5 4 5 . - ' - \
4 0 [ , ' " 2 5 3 6 ...... ' " " - - '
~ 2 0 ~ " 2 5 5 2
2 0 0
150
o 100
5 0
E 0
0 2 4 6
Time, [ ms]
Fi g. 1. Ex a mp l e s o f c u r r e n t wa v e f o r m a n d de ns i t y e vol ut i on i n
RF P di s c ha r ge s . Shot n u mb e r 2531 wa s p r e c e d e d b y a g l o w
d i s c h a r g e i n He. The de ns i t y r i se at a r o u n d t = 3 ms i s i nt e r pr e t e d
as be i ng due t o a n i n c r e a s e d i mp u r i t y i nf l ux, p r o b a b l y as a r e s ul t
o f l ocal he a t i ng.
E
O. B
o
~C~ 0 . 6
,
"~ 0.4
c:]
0 . 2 '
O. OI
T
~ 20
~ 15 I
F 1
g
~ o
1. 0 , . . . . . . . , . . . . , . . . . . , - -
. . . . . . . . : . . . . . . . . . " " [ 0 ,~'P ~ [ , 9 0 , 2 1 0 ] ~ {1 4 0 ,1 s 5 ] ~ A " A } - - -
.<+
l O 2 0 3 0 4 0
L i n e A v e r a g e d D e n s i t y , [ 1 0 ~8 m - s ]
50
Fi g. 2. The e d g e de ns i t y a n d e dge t e mpe r a t ur e f r o m L a n g mu i r
t r i pl e p r o b e me a s u r e me n t s . The da t a f or t he l o we r c ur r e nt s ar e
t aken at r = 0. 181 m, whi l e t he h i g h c u r r e n t me a s u r e me n t s we r e
ma d e at r = 0. 179 m. At t hes e pos i t i ons t he de ns i t y a n d t e mpe r a -
t ur e d r o p r a di a l l y on t he s cal e l e ngt h o f A ~ 8 r am. The e d g e
p a r a me t e r s d e p e n d onl y we a k l y on t he l i ne a v e r a g e d de ns i t y a n d
p l a s ma cur r ent .
density in high and low current discharges respectively.
The edge parameters depended only weakly on the central
density, and there was no significant dependence on the
plasma current either. The edge temperature was 10-18
e V.
That the graphite wall pumps hydrogen during the
discharge is evident from Fig. 3, where a typical pressure
decrement in connection to an RFP discharge is shown. In
this case the gas supply is set to constant gas flow. During
the plasma discharge the pressure drops by Ap ~ 0.12 Pa,
corresponding to N = 2 A p V/ k T -~ N ~ 5 1019 hydrogen
o. 2o
0 . 1 5
o_ 0 I
c 0 . 1
0 o s :
0 . 2 5 . . . . . . . . . . " . . . . . . . . ' . . . . . . . . . . . . . . . . . . . ' . . . . . . . . . ' . . . . . . . . . ' . . . . . . . . .
0 . 0 0
- 10 0 10 20 50 40 50 60
Time, i s]
Fi g. 3. The par t i al pr e s s ur e t r ans i ent i n c o n n e c t i o n wi t h a n RF P
di s c ha r ge , t he ga s s u p p l y is set t o c o n s t a n t f l ow c o n t r o l A
t r a p p i n g r at e o f 5 . 1 0 r8 H/ m 2 pe r d i s c h a r g e at t he wal l c a n be
i nf er r ed.
H. Bergsfker et al. / Journal of Nuclear Materials 241-243 (1997) 993-997 995
atoms bei ng trapped by the wall, then reverts to its equi l i b-
ri um val ue on the t i me scale of the characteristic pump
down t i me ~'s = V/ S= 3 s. Typi cal l y 5. 1018 H/ m e are
trapped by the wal l in an RFP discharge. Not e that some of
this hydrogen is released within 10- 20 s after the dis-
charge, as evi denced by the slight pressure increase after
the discharge, though it is difficult to estimate this fraction
accurately.
The amount of hydrogen that is r emoved in the hel i um
gl ow discharges has been measured as shown in Fig. 4,
where mass spect romet er data are presented for a set of
standard hel i um gl ow discharges, fol l owi ng similar series
of RFP discharges. There is an i ncrement Apg( t ) in the
hydrogen partial pressure whi l e the gl ow di scharge is on,
correspondi ng to a flux of hydrogen bei ng released from
the wall. The t i me dependence of the release can be fitted
quite wel l to be proportional to t - 0 " 9 . Integrated over the
gl ow discharge the number of hydrogens whi ch have been
released from the wall and pumped out was calculated as
ApeS
AN w=2. f kT dt ( 1)
and for the standard situation shown in the figure A Nw/A
= 1.5 1020 H/ m 2 wall area were removed.
Usual l y sequences of 10- 30 RFP discharges have been
run, each sequence terminated with a standard He- GDC. In
operating this way the standard He- GDC was capable of
restoring the wall surface to the original state as far as
pl asma density is concerned. This is shown in Fig. 5,
whi ch presents the shot to shot density evol ut i on averaged
bet ween gl ow discharges. The discharge duration was 6
ms. Each poi nt in the figure represents an average over
5- 15 discharge sequences. The bars indicate the statistical
standard devi at i on and are drawn only for the medi um
current discharges, whi l e the scatter among the high and
l ow current discharges is similar. The densities at the poi nt
of maxi mum current are plotted, whi ch is prior to the final
density increase (cf. Fig. 1). There is no significant differ-
ence bet ween the 150- 170 kA sequences and the 170- 200
kA case, but in the 100- 140 kA sequences the density
?
# 1 0
eo
G_
c~ 5
o ~ _ / ~ , ~ ! ' ' , i l i
' ~ ' . i ~/! ' '
~ . & L J ~ , ~
2[2
0 . . . . . . . . , . . . . . . . . . , . . . . . . . . . , . . . . . . . . . i . . . . . . . . . , . . . . . . . ,
- 1 0 0 1 0 2 0 5 0 4 0 5 0
T i me , [ T i n . ]
Fig. 4, The partial pressure evolution of hydrogen during the
standard glow discharges in helium, In the 45 Tin standard helium
glow 1.5.10 20 H/ m 2 are removed.
60
50
4 0
E
o
~ , 5 o
g
o 20
1 0
,v , , , , I ' " ' '
0 Lp e [ 1 5 0 , 1 7 0 ] k A
Ip ~ [ 1 7 0 , 2 0 0 ] k A
Ip [ 1 0 0 , 1 5 0 ] k A
' ' ' ' ' I
0 i , i i I i , i i i i i i , L 1
0 5 1 0 1 5
D { s c h a r g e N u m b e r c f [ e r l o t e s t H e - G D C
Fig. 5, The evolution from shot to shot of the line averaged
density at maximum current in a number of discharge sequences
following helium glow discharges The bars which are drawn for
the medium current data indicate the statistical standard deviation.
The scatter is similar for the high and low current discharges.
increase is slower. The low current sequences correspond
largely to the gl ow discharges whi ch are plotted in Fig. 4.
Fig. 6 shows exampl es of impurity accumul at i on on
5. 0
4. 5
4. 0
3.5
3.0
2.5
2.0
1.5
t . 0
0 3
0. 0
o
O j . /
O
5 I 0 15 20 25 30
number of RFP dlscharges
1 4 0
100120 ~
m 80
a ,0
g ~
0 . . . . . : , i r I . , I
0 20 40 60 80 1 O0 120 140
number of RFP discharges
Fig. 6. The trapping of oxygen in a 30 nm thick surface layer of a
graphite surface and the collection of carbon at a silicon surface,
both at wall radius.
996 14. Bergsaker et al./Journal of Nuclear Materials 241-243 (1997) 993-997
5
E
o
s
0
g"
O
T
f o l l o w i n g R F P ~ .[
~
l n g H 2 G D C
0 ' ' ' ' ' . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ' . . . . i , , , ,
50 100 150 200 250 $00 350 400 450
H e G D C t l m e [ m i n u t e s ]
I ~ . . . . i . . . . . . . . . . i i [ . . . . . . . . . T . . . . . . . . . i . . . .
7
- - 14 f o l l o w i n g R F P ~
~ " 8 f o l l o w i n g H ~ G D C
| ,
E a
0
so ~oo iso ~oo 25o soo 3so ,oo 45o
H e G D C t l m e [ m i n u t e s ]
Fig. 7. The trapping of oxygen and stainless steel components on
graphite at wall radius.
initially clean surfaces which have been exposed at the
wall position to l ong sequences of RFP discharges, but
without exposure to the intervening He-GDC. The upper
part shows the areal density of oxygen trapped within a 30
nm thick layer on graphite surface, whereas the l ower part
shows carbon accumul at i on on a silicon surface. Similarly
Fig. 7 shows the accumulation of oxygen and stainless
steel component s in helium gl ow discharges which fol l ow
RFP operation and gl ow discharges in hydrogen, respec-
tively. In all cases the trapping rate is initially linear with
exposure time, and may or may not tend towards a station-
ary equi l i bri um areal density. In RFP discharges represen-
tative initial trapping rates at the wall position have been
2. 5. ] 019 C / m 2, 2. 5. 1018 O/ m 2, and 3. 1016 metal-
a t o ms / m s per discharge. Ot her accumul at i ng impurities
have been nitrogen, chlorine and silicon. The small amounts
of stainless steel which are present on the wall have been
built up mostly during a long period of OHTE operation
and the primary source can not be identified. For compari-
son, the initial trapping rate of hydrogen (or deuterium in
discharges in deuterium) was typically 5. 1019 a t o ms / m 2
per discharge on carbon or silicon surfaces.
4 . D i s c u s s i o n a n d c o n c l u s i o n s
As is well known from ion beam experiments as well
as from other large fusion experiments, the graphite wall
acts in the first approximation as a saturable reservoi r of
hydrogen. In the absence of sufficiently large external
particle sources, the pumpi ng and release by the wall
determines the plasma density. The mass spectrometry and
surface probe data can be used for a more quantitative
account of the particle exchange between wall and plasma.
The initial trapping rate of deuterium in pure graphite,
5. 1019 D/ m 2 per discharge, can be taken to represent
roughly the incident flux density of hydrogen to the wall
[7] and corresponds to a particle confi nement time of
roughly 150/. ts. From Fig. 3 it was seen that the partially
saturated wall traps 5 . 1 0 Is H/ m 2 per discharge.
The effect i veness of He gl ow discharge in r emovi ng
hydrogen from carbon surfaces has been qualitatively
demonstrated before, in several large fusion experiments,
such as OHTE [3], TEXTOR [8], DIII-D [9] and JT-60U
[10]. As shown in Fig. 4, the standard He- GDC in T2
removed 1.5. 10 20 H/ m 2 fol l owi ng 10- 30 low current
discharges, which corresponds reasonably well to the 0. 5-
1. 5. 10 a H/ m 2 that the wall has trapped. The release
mechani sm of hydrogen during helium ion bombardment,
in particular in a gl ow discharge, has not been as exten-
sively discussed in the literature as the trapping, release
and isotopic exchange during hydrogen implantation. The
time dependence of Apg( t ) as shown in Fig. 4 can be
compared to a numerical model which was devel oped for
Tore Supra [11] and which included ion induced detrap-
ping with a fixed cross section, thermal detrapping, diffu-
sion and surface as well as bulk recombination. The pre-
sent time dependence of ~ t -(~9 may be indicative of a
intermediate condition between diffusion limited and re-
combi nat i on limited release. In the present case however,
thermally activated release is clearly negl i gi bl e compared
to the ion induced release,
That the standard He- GDC removes a substantial frac-
tion of the total wall reservoi r of hydrogen which is
interacting with the plasma could be seen from an isotopic
exchange from 1H to D, when mass spectrometry fol l ow-
ing the discharges showed a change of the isotopic ratio at
a rate of ~ 5 % pe? shot [5]. Thus a number N D=
2p~i , V/ kT, ~ 2. 102o deuterium atoms were admitted as
filling gas in every discharge, representing a fraction F~ m
= ND/ ( No + Nw) of the total hydrogen content in the
system, i f N W = NwD + NwH is the wall reservoir. The
deuterium fraction r o = NwD/ Nw in the wall fol l owi ng
the first pulse with deuterium would be rDi = F~ijl and
after the ith pulse rDi = (1 - Ft.m)roo_ t) + Fliu. This
should correspond to the isotope ratio observed in the mass
spectrometry, such that N w= 2 0 N D~ 4 . 1 0 2 j H. or 4 .
10 20 H/ m 2. This areal density corresponds to the satura-
tion l evel of 100 eV deuterium ions implanted into graphite
[12].
Carbon and oxygen are the spectroscopically dominat-
ing impurities in T2 RFP operation. The collection rates of
C and O on initially clean surfaces provi de an estimate of
their absolute concentrations and Fig. 6 may gi ve an idea
H. Bergsfiker et al. / Journal of Nuclear Materials 241-243 (1997) 993-997 997
of t he rat e at whi ch an equi l i br i um condi t i on is appr oached
at t he wal l surfaces. Fig. 7 shows t hat t he He- GDC is abl e
t o r edi st r i but e oxygen. The met al s on t he wal l cl earl y
r each a st at i onar y l evel aft er 250 mi n or so, pr esumabl y
whe n sput t er i ng and r edeposi t i on bal ance, but t hi s behav-
i or is not obs er ved for oxygen.
In concl usi on, i t has been s hown t hat a r udi ment ar y
densi t y cont r ol coul d be achi eved by means of f r equent
hel i um gl ow di schar ge condi t i oni ng of t he wall. The st an-
dar d He- GDC is wel l char act er i zed and r epr oduci bl e. The
hydr ogen and i mpur i t y t r appi ng at t he wal l has been
meas ur ed i n RFP di schar ges and i n gl ow di schar ges so as
t o pr ovi de a basi s for pl as ma per f or mance t o wal l condi -
t i ons. The shot t o shot part i cl e exchange bet ween wal l and
pl as ma can be appr oxi mat el y account ed for. Fl uxes of
maj or i mpur i t i es are gi ven i n t er ms of i ni t i al t r appi ng rat es
on cl ean surfaces at t he wal l posi t i on.
References
[1] P. Brunsell, H. Bergs~,ker, J. Brzozowski et al., Proc. 22nd
Eur. Conf. Controlled Fusion Plasma Phys., Bournemouth,
UK, July (1995) Part III, p. 157.
[2] R.R. Goforth et al., Nucl. Fusion 26 (1986) 515.
[3] G.L. Jackson, T.S. Taylor, T.N, Cadstrom et al., J. Nucl.
Mater. 145-147 (1987) 470.
[4] D. Larsson, H. Bergs~ker, B. Emmoth et al., Proc. 18th
Symp. Fusion Technol. Karlsruhe, August 22-26 (1994) Vol.
1, p. 207.
[5] H. Bergs&ker, P. Brunsell, D. Larsson and L. Tramontin,
Proc. 22nd Eur. Conf. Contr. Fusion Plasma Phys.
Bournemouth, UK, July (1995) Part IV, p. 277.
[6] H. Bergs~aker, A. Mt~ller, G.X. Li et al., J. Nucl. Mater.
220-222 (1995) 712.
[7] W.R. Wampler, D.K. Brice and C.W. Magee, J. Nucl. Mater.
102 (1981) 304.
[8] F. Waelbroeck, T. Banno, H.G. Esser et al., J. Nucl. Mater.
162-164 (1989) 496.
[9] G.L. Jackson, T.S. Taylor, S.L. Allen et al., J. Nucl. Mater.
162-164 (1989) 489.
[10] M. Shimada, A. Sakasai, R. Yoshino et al., J. Nucl. Mater.
196-198 (1992) 164.
[11] C. Grisoglia, L.D. Horton and J. Ehrenberg, J. Nucl. Mater.
220-222 (1995) 516.
[12] G. Staudenmaier, J. Roth, R. Behrisch et al., J. Nucl. Mater.
84 (1979) 149.