Download Real time, in-situ measurements of secondary electron emission in

Journal of Nuclear Materials 176 & 177 (1990) 877482
North-Holland
Real time, in-situ measurements
in DITE
of secondary electron emission
R.A. Pitts and G.F. Matthews
AEA Furion, Culham Laboratory, UKAEA/Euratom
Ficsion Association, OXON, United Kingdom
Using a new application of the retarding field analyser @WA), it is possible to measure, in-situ, the coefficients of secondary
electron release from chosen, conducting target substrates. Since the rn~~ern~ts
are made in the tokamak, the yields are
due to primary particles characterized by velocity distributions which are close to those experienced by solid surfaces
intercepting the edge plasma. Operating the RFA in the standard mode also permits measurement, under identical discharge
conditions, of Ti, T,, ne and Vshcathat the probe.
Measurements of the coefficients of ion and electron secondary electron emission (Si, 8,) are presented for both graphite
and molybdenum target substrates. For electron energies in the range 0 to - 60 eV, values of 8, near to unity have been
measured for MO and C with the graphite yields being slightly lower than those for molybdenum. Si is found to be 0.3-0.5 for
molybdenum, in good agreement with ion beam measurements at low energy ( - 300-500 ev).
1. Introduction
The retarding field analyser (RFA) is now established as an important probe diagnostic for the scrapeoff layer (SOL) in tokamaks [1,2,3]. By exploiting an
effect normally suppressed during conventional RFA
operation, namely secondary electron emission (s.e.e.),
we have developed a powerful technique for the quantitative study of the coefficients of this release, Si, S,, due
to ions and electrons.
In addition to providing the first measurements of 6,
at low energy (0 to - 60 ev), our new method enables
unique, direct experimental observation of the s.e.e.
yields due to primary particles whose velocity distributions are close to those experienced by walls and
~~ters/divertors
at the plasma edge. Such measurements are important since the s.e.e. yield is an important factor in determining the theoretical sheath
potential difference and hence the sheath heat transmission coefficient [4].
2. The principle
During
device are
electrons,
suppressed
normal RFA operation, potentials within the
arranged such that the release of secondary
in particular, from the collector surface, is
(s.e.e. from the grids is usually negligible.)
Elsevier Science Publishers B.V. (North-Holl~d)
This avoids unwanted offsets on the data since an
escaping secondary electron is equivalent to a collected
positive ion. If, instead, the potentials are changed so
that secondary electrons are allowed to leave the collector surface, it is possible to use the apparent increase in
the collector current as a direct measure of the yield.
This technique is illustrated in figs. la and 2a where the
ion and electron s.e.e. modes of operation are depicted
in schematic form together with the approximate bias
potentials used in this study.
In the ion s.e.e. mode (fig. la), the slit is biassed into
ion saturation to eliminate most of the primary electrons. To remove all electrons (i.e. those able to
surmount the sheath), grid G2 and the collector are held
even more negative than the slit. A square wave oscillation (of peak to peak ~p~tude
- few times T,, i.e.
greater than the maximum energy of the secondary
electrons) is then superimposed on both the collector
and G2 voltages such that the electric field between
them changes direction rapidly (- every 16 ms in this
case). Since the constant negative bias is less than that
at the slit, secondary electrons are able to escape during
one half of the cycle but are suppressed during the other
half. An example of the time dependence of the resulting collector current for a graphite sample is shown in
fig. lb where the effects of s.e.e. are clearly visible. In
the electron s.e.e. mode (fig. 2a), the slit is allowed to
float and the unwanted ions are removed by a large,
R.A. Pius, G. F. Marrhews / Secondal?: elecrron emission in DITE
878
(a
1 Ion S.E.E. Mode
3. Experiment
Sheath
Secondary
The data presented here were obtained using a modified version of a probe described in an earlier paper [6]
and shown in plan view in fig. 3. Samples for which the
s.e.e. yield is required form the collectors of a two grid
RFA. They are held in, and electrically isolated from a
carousel which may be rotated behind two large apertures (area - 0.19 cm2) cut in opposite faces of the
protective
graphite outer housing. Behind one is the
small (width = 5 pm, length = 3 mm, thickness = 30
pm) nickel RFA entrance
slit. The other is unobstructed, providing a means to condition the samples by
exposure to the full plasma flux.
Five modes of operation are possible for the probe,
with each configuration
requiring a separate discharge.
In the LP mode the slit is operated as a Langmuir probe
Electrons
(b)
80
70
;i
a.
(a)
Electron
S.E.E. Mode.
60
E
?!
2
50
40
&
30
g
20
Secondary
S
Electrons
10
0
50
150
250
350
450
Time (m-s)
Fig. 1. The ion s.e.e. mode. a) Schematic arrangement of bias
potentials, b) raw RFA collector current data during the density scan of fig. 4a
constant
positive bias applied to grid Gl. A square
wave voltage is again applied to both Gl and the
collector such that the potential on each oscillates in
antiphase from zero volts to some negative value. Once
again (fig. 2b), the collector current exhibits a step like
change as the electric field direction reverses.
We note here that our measurements
are, in fact, of
the total yield, including true secondary electrons and
reflected particles. In the case of electrons as primary
particles, data for the the reflection coefficient,
r, of
surfaces at low energy are scarce. In the ion s.e.e. mode,
although the primary ions will be reflected with high
probability
[5], the majority will do so as neutrals and
so will not affect the measured current.
(b)
;:
$J
;
u
b
5
2
z
U
-
-20
-30
-40
-50
-60
-70
-10 r
-80 ’
50
I
I
150
I
1
250
I
I
350
I
I
450
i
J
Time(ms)
Fig. 2. The electron s.e.e. mode. a) Schematic arrangement of
bias potentials,
b) raw RFA collector current data during the
density scan of fig. 4a
R.A. Pitts, G.E Matthews
/ Secondary electron emission in DITE
879
Fine
Nickel
Entran
Slit
Sample
Graphite
Exposure
Electrically
Isolated
from
Carousel
Housing
Fig. 3. Section through the probe used in these experiments showing the RFA entrance slit and grids, sample/collector
exposure aperture
to obtain the local T, and ne. Sample exposures are
usually made during this mode. In the electron and ion
RFA modes, the probe is configured as a normal RFA
to measure Ti, V&,,, and also T,. The latter provides a
useful comparison
for that derived from the LP mode.
Finally, in the ion and electron see. modes, the bias
configurations
described in section 2 are used to measure the s.e.e. yields of samples in the carousel.
(4
,
,
I
,
,
I
.A
I
,
holders and
Experiments
have been performed
in a series of
reproducible
helium discharges with I, = 115 kA, B, =
1.8 T with the probe entrance slit at the limiter radius,
a = 24 cm. Since each full set of measurements
requires
many discharges,
our philosophy
was to obtain the
largest possible variation in density during a single shot.
This enabled us to achieve a wide variation in the edge
parameters
but at the expense of steady-state
condi-
I
/
lb)
I&O-, o
160
I
1
I
0
I
q
Ti (RFA)
0
VS,,wt,,
I
I
(W-A)
ne (edge)
50
,
,
I
,
I
I
I
,
100
150
200
250
300
350
400
450
Time (ms)
500
250
350
450
Time (ms.1
Fig. 4. (a) Behaviour of edge and line average central densities during the discharges in this study. (b) Variation of T,, T,(RFA),
W-P) and Lath for the same discharge.
RA. Pius, G.E Matthews
880
/ Secondary electron emission in DITE
tions. The samples used in this study were graphite and
molybdenum. Note that measurements of s.e.e. yields
using this technique are limited to electrically conducting target materials.
4. Results and discussion
Fig. 4a shows how the line averaged central density,
iie and the edge density behave during these discharges.
During the first 350 ms, iI, increases linearly, reaching
a plateau at a value - 8 times its initial value. Although
the edge density also rises reproducibly during this
period, the reason for the decrease at - 300 ms is not
clear. Similar trends are observed in the envelope of the
collected current (figs. lb and 2b). Presumably, the
behaviour is connected with changes in the density
profile. The compilation of signals in fig. 4b shows
rather clearly how T, (calculated assuming Zi = 2), T,
and ) Vsheath
1 decrease with increasing density. Notably,
there is a large variation in the ratio Ti/T,, in common
with previous observations [3], where the density was
increased over a number of discharges. In addition, it is
clear that the RFA electron and LP mode values of T,
are in very good agreement throughout the discharge.
Errors on the data points are not included for clarity
but are generally in the range 5-108. Since the data
acquisition electronics are referenced to the torus, the
value of the local floating potential should be subtracted from our measurements of Vsheath.This has not
1.0
I
1
,
I
been performed in fig. 4b since I$ was separately measured to be only 1 or 2 V different from zero.
In figs. Sa and b we compare the time variation of &
and 6, for unexposed clean graphite and molybdenum
samples (ultrasonic clean only). The yields are computed by averaging the collector current during the
interval corresponding to zero and maximum secondary
emission and subtracting one from the other. The error
bars represent the most probable error calculated from
the standard deviation of each average. Though the data
are more scattered for &, there is a clear increase in
both yields in going from molybdenum to graphite, in
agreement with the general tendency for metal surfaces
to have higher s.e.e. yields than graphite [7]. The apparent increase in S, as time increases and T falls is not
expected. In general, the electron s.e.e. yield is a decreasing function of projectile energy at low energies
[7,8], in contrast to our data. Whilst isolated data points
at the beginning and end of the discharge are associated
with larger errors, the trend with time shown in fig. 5 is
reproducible.
The absolute values of & for molybdenum are reasonably close to those found by Vance [9] for bombardment of atomically clean molybdenum surfaces with
monoenergetic He+ ions at low energy and varying
angle of incidence. His measurements showed 6i to
remain approximately constant at 6i = 0.25-0.3 in the
range Eion = 40-400 eV. Mahadevan et al. [lo] showed
this to remain true at normal incidence up to much
higher energies ( Eion - 1.5 keV). Since in our case the
ions are accelerated to the collector during the ion s.e.e.
I
(b)
(a)
’
’
’
o MO
0.9
0 MO
l C
IJJ 0.6 -
0.01
50
I
’
150
250
Time (ms)
350
I
I
450
I
50
150
250
Time(ms)
Fig. 5. Time variation of see. yields for graphite and molybdenum a) ai b) 6,
350
450
881
R.A. Pitts, G.F. Matthews / Secondary electron emission in DITE
mode, the primary ion energy interval appropriate
to
our dara is in the region Eion = NO-500 eV, taking into
account the ion temperature
and the energy gain after
acceleration
in the sheath. Moreover,
our ion energy
distribution
is Maxwellian and comprises a large number of highly charged impurity ions [ll] in addition to
He’+ ions. Hagstrum [12] measured values of 6i = 0.7
for doubly charged, low energy helium ions incident
normally on clean MO. No analagous
measurements
appear to exist for the ion s.e.e. yield of graphite, but
our data suggests
6i to be similar
to that for
molybdenum.
Our results for the electron s.e.e. yield of graphite
indicate higher values of S, than earlier studies using
monoenergetic
beams [13]. Indeed, calculations
in [13]
for the expected yields due to incident 1D Maxwellian
distributions
(appropriate
to our experiments, where the
electrons
are constrained
to gyrate in tight Larmor
orbits along the magnetic field), suggest S, = 0.2-0.4 at
low primary electron energy. In the context of our data,
low energy means Eelstron = 0 to 60 eV. This follows
from the measured
T, (fig. 4b) which decreases from
- 20 to - 5 eV as the density increases (the electrons
see a retarding field in passing through the sheath so
that their distribution
remains Maxwellian at the same
temperature
[4]). In the case of molybdenum,
we observe S, - 1 + 1.4, in contrast with earlier calculations
[14] which gave 13,= 0.1-0.8 for a Maxwellian distribution of primary electrons at low energy (O-100 ev>. This
discrepancy
(which is also observed for graphite) may
be due simply to the contribution
of reflected electrons
(Bronshtein
[15] measured
r = 0.15 for normally incident, primary electrons with energy 2-10 eV.) Since
reflection is not a problem in the ion s.e.e. mode, the
good agreement between our measurements
of & and
those already published
is further evidence for high
values of r in the electron case. Enhanced reflection at
low energies and fluxes might also explain why the
measured yield appears to increase with time as T, falls.
Following measurements
on the clean surface, each
sample was fully exposed to the plasma during a single
discharge and the experiment
repeated. Within the errors, no difference was detected between the exposed
and unexposed yields, indicating that the surface conditions rapidly or that exposure through the 5 urn hole
during RFA measurements
of T and T, is enough to
condition
the sample. Unfortunately,
insufficient
discharges were available to make prolonged exposures of
the samples. Future experiments
in which this were
possible would be particularly interesting since the s.e.e.
yield is known to be influenced by the deposition
of
metal/carbon
layers and by fuel ion implantation
[8,13].
1
70
0
60
t
I
I
I
1
Experament
-
0
Fit
-1
50
z
t
40-
7
z
30r”
5-”
zolo-
00
0
0
5
10
15
Te
20
25
(eV)
Fig. 6. Experimental and theoretical dependence of 1Gheath
1
on T,.
Returning to fig. 4b, it is instructive to compare
measured
value Vsheath with that obtained
from
simple expression for the sheath potential fall [4]
Vrhra~h=$lOg(
%(
Zi+
the
the
~)(1~8)~2)~(l)
where, for a floating surface, S is the total yield. In our
case, the Geath data has been obtained in the RFA ion
mode with the slit biassed into ion saturation. Since the
majority of electrons are then prevented from reaching
the defining slit, it is more appropriate
to take Si for use
in eq. (1). Moreover, because we observe Si(C) - &(Mo),
it is reasonable
to assume similar values for the ion
s.e.e. yield of the nickel slit plates. Using least squares
cubic spline approximations,
we have fitted the experimental data for T,, T, and S,(Mo) and computed
the
expected Vsheath.Theory and experiment
are compared
in fig. 6, showing that the two are in reasonable agreement except at high values of T, near the start of the
discharge. We have no Vsheath data for the case of a
floating slit where electron s.e.e. also makes a contribution to the total yield.
5. Conclusions
In this paper we have shown that it is possible to
measure secondary electron emission yields due to ion
and electron energy distributions
similar to those experienced by solid surfaces contacting
the tokamak edge
plasma. By using a retarding field analyser inserted into
the plasma boundary,
we have obtained
values of Si
882
R.A. Pins, G.F. Matthews
/ Secondary electron emission m DITE
and 8, for graphite and molybdenum
samples. Our data
for S, is higher than previous values recorded in the
literature, although few other measurements
exist at the
low electron energies characteristic
of this study. In
addition, since our yield includes inelastically reflected
primaries, the true secondary yield may be lower. We
find reasonable agreement between our values of 6, for
molybdenum
and those measured in monoenergetic
ion
beam studies. The ion s.e.e. yields for graphite
are
- lo-30% lower than for molybdenum.
Good agreement is observed between the measured
and that calculated
from standard
theory
Vsheath(Te)
using the experimental
values of T,, T, and 6,. The
latter is most appropriate
since during the measurements of Vsheath,the RFA entrance slit is held at a large
negative potential and most of the primary electrons do
not reach the slit surfaces.
References
[l] G.F. Matthews, J. Phys. D 17 (1984) 2243.
[2] AS. Wan, T.F. Yang, B. Lipschultz and B. Labombard,
Rev. Sci. Instr. 57 (1986) 1542.
[3] R.A. Pitts, G.M. McCracken and G.F. Matthews in: Proc.
16th Eur. Conf. on Controlled Fusion and Plasma Physics,
Venice, 13-17 March 1989, Part III, p. 955.
[41 P.C. Stangeby, in: Physics of Plasma Wall Interactions in
Controlled Fusion, Eds. D.E. Post and R. Behrisch, NATO
AS1 Series (Plenum, New York, 1986) p. 41.
[51 W. Fckstein and H. Verbeek. Nucl. Fusion. Special Issue
(1984) 12.
[61 R.A. Pitts, G.M. McCracken and G.F. Matthews, J. Nucl.
Mater. 162-164 (1989) 568.
171 H. Bruining, Physics and Application of Secondary Electron Emission (Pergamon,
London, 1954).
PI E.W. Thomas, Nucl. Fusion, Special Issue (1984) 94.
[91 D.W. Vance, Phys. Rev. 188 (1968) 252.
1101 P. Mahadevan. J.K. Layton and D.B. Medved, Phys. Rev.
129 (1963) 79.
J.M. Pedgely,
R.A. Pitts and P.C.
Pll G.F. Matthews,
Stangeby,
in these Proceedings
(PSI-9). J. Nucl. Mater.
176 & 177 (1990).
WI H.D. Hagstrum, Phys. Rev. 104 (1956) 672.
P31 M.E. Woods, B.J. Hopkins, G.F. Matthews, G.M. McCracken,
P.M. Sewell and H. Fahrang,
J. Phys. D20
(1987) 1136.
and P.J. Harbour,
Culham Report
[I41 M.C. Saussez-Hublet
CLM-R208
(1980).
[I51 I.M. Bronshtein, Bull. Acad. Sci. USSR, 22 (1958) 442.
P. Sewell, M. Woods
1161 G.F. Matthews, G.M. McCracken,
and B.J. Hopkins, J. Nucl. Mater. 145-147 (1987) 225.