Plasma diagnostic tools for optimizing negative hydrogen ion sources
U. Fantz, H. D. Falter, P. Franzen, E. Speth, R. Hemsworth et al.
Citation: Rev. Sci. Instrum. 77, 03A516 (2006); doi: 10.1063/1.2165769
View online: http://dx.doi.org/10.1063/1.2165769
View Table of Contents: http://rsi.aip.org/resource/1/RSINAK/v77/i3
Published by the American Institute of Physics.
Additional information on Rev. Sci. Instrum.
Journal Homepage: http://rsi.aip.org
Journal Information: http://rsi.aip.org/about/about_the_journal
Top downloads: http://rsi.aip.org/features/most_downloaded
Information for Authors: http://rsi.aip.org/authors
Downloaded 15 Mar 2012 to 218.199.90.49. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
g
soci
ul-le
blin
2318
21
spec
n th
ydro
nd a
ITE
utra
sity
how
ourc
e th
tes
5769
Development of negative hydrogen ion sources for neu-
tral beam heating systems of fusion devices such as ITER
stripping and mutual neutralization. This means optimization
demands knowledge of these quantities, and therefore
simple, robust, and noninvasive diagnostic techniques.
REVIEW OF SCIENTIFIC INSTRUMENTS 77, 03A516 �2006�
requires an optimization of negative ion formation in low-
pressure plasmas. Currently two types of discharges, an in-
ductively coupled rf source and an arc source, are envisaged
to meet the ITER requirements: extracted current densities of
20 mA/cm2 in D �28 mA/cm2 in H� over an extraction area
of 2000 cm2 for a pulse length of 3600 s at a source pressure
of 0.3 Pa. Under an extensive development program, the rf
source at the IPP Garching achieved, and even exceeded,
recently the extracted current densities in short-pulse
operation,1,2 whereas the KAMABOKO-III arc source at
CEA, Cadarache focuses on long-pulse operation.3,4 Both
sources operate with cesium evaporation due to the fact that
the negative ions are very efficiently produced by the surface
effect, i.e., the interaction of hydrogen particles �atoms and
ions� with a material of low work function: H, Hx
+
+surface e−→H−. The enhancement of negative ions is
about one order of magnitude in comparison with a cesium-
free operation when negative ions are formed by the volume
effect, i.e., the dissociative attachment, a process which is
therefore not discussed further in this article. Together with
the boundary condition that the survival length of negative
ions is in the centimeter range, optimization of negative ions
is focused on optimizing the cesium layer on the plasma grid,
reflected by the cesium balance in the source, atomic hydro-
gen density, and positive ion density. The electron density ne
II. DIAGNOSTIC METHOD
The technique of optical emission spectroscopy �OES� is
widely used as a noninvasive and powerful diagnostic tool
which provides information on particle densities and
temperatures.5 In general, line emission gives direct access to
the population of the particle in its excited state, which de-
pends on the plasma parameters. The coupling to the ground
state or its ion state is provided by collisional radiative mod-
els. The combination of such models with absolutely mea-
sured line radiation is used to determine the following
plasma parameters: ne, Te, Tgas, atomic and molecular hydro-
gen densities, neutral and singly ionized cesium densities,
and even densities of negative ions.6,7
III. EXPERIMENTAL RESULTS
OES is applied to the rf sources at the IPP and
KAMABOKO-III arc source at CEA. Collimator optics pro-
vide parallel light �diameter �1 cm� in 40 �rf source� and
17 mm �arc source� distances to the grid, respectively, result-
ing in line-of-sight averaged results. The optical windows are
mounted at the source body, the line of sight has a length of
�30 cm �width of the rf source body, length of the arc
source�; the grid contributes �40% of the length of the line
Plasma diagnostic tools for optimizin
U. Fantz, H. D. Falter, P. Franzen, and E. Speth
Max-Planck-Institut fuer Plasmaphysik, EURATOM As
Germany
R. Hemsworth
CEA-Cadarache, EURATOM Association, F-13108 St. Pa
D. Boilson
Association EURATOM-DCU, PRL/NCST, Glasnevin, Du
A. Krylov
RRC Kurchatov Institute, 1 Kurchatov Square, Moscow 1
�Presented on 13 September 2005; published online
The powerful diagnostic tool of optical emission
parameters in negative hydrogen ion sources based o
temperature, electron density, atomic-to-molecular h
presented for two types of sources, a rf source a
development for a neutral beam heating system of
volume is obtained from cesium radiation: the Cs ne
lower than the hydrogen density and the Cs ion den
than the electron density in front of the grid. It is s
useful for monitoring the cesium balance in the s
densities are determined. In a well-conditioned sourc
of magnitude as the electron density and correla
American Institute of Physics. �DOI: 10.1063/1.216
I. INTRODUCTION
0034-6748/2006/77�3�/03A516/4/$23.00 77, 03A51
Downloaded 15 Mar 2012 to 218.199.90.49. Redistribution subject to AIP lic
negative hydrogen ion sources
ation, Boltzmannstrasse 2, 85748 Garching,
z-Durance, France
13, Ireland
2, Russia
March 2006�
troscopy is used to measure the plasma
e surface mechanism. Results for electron
gen density ratio, and gas temperature are
n arc source, which are currently under
R. The amount of cesium in the plasma
l density is five to ten orders of magnitude
is two to three orders of magnitude lower
n that monitoring of cesium lines is very
e. From a line-ratio method negative ion
e negative ion density is of the same order
with extracted current densities. © 2006
�
and electron temperature Te determine the losses by electron
© 2006 American Institute of Physics6-1
ense or copyright; see http://rsi.aip.org/about/rights_and_permissions
of sight for both sources. Typical magnetic filter field
strength in this area is 0.4 mT m �rf source2� and 0.8 mT m
�arc source4�. In the case of the arc source the same diagnos-
tic flange is used for Langmuir probe measurements, measur-
ing electron densities 17 mm in front of the grid center.
Spectroscopic measurements were taken with two abso-
lutely calibrated survey spectrometers �high and low spectral
resolution, ��FWHM=20–35 pm and 1–1.8 nm, respec-
tively� in hydrogen discharges at a pressure of 0.3 Pa with
and without cesium evaporation, with a characteristic pulse
length and power range: 6 s and 60–140 kW for the rf
source and 30–400 s and 25–55 kW for the arc source. Cur-
rent densities of 15–20 mA/cm2 are achieved typically at
110 and 45 kW, respectively. Cesium-free discharges result
typically in 2 mA/cm2 at 0.3 Pa. Figure 1 shows examples
of spectra recorded with a low-resolution survey spectrom-
eter, which is also capable of recording time traces. The im-
portant diagnostic lines of atomic hydrogen �Balmer lines�,
molecular hydrogen, neutral Cs, and Cs+ are identified. The
wavelength range of impurities �Cu, OH, N2, and CN� is
indicated. In principle, the discharges are free of these impu-
rities, except small amounts of oxygen �occasionally� as can
be seen in the spectrum of the arc source.
A. Plasma parameters
Using a diagnostic gas, for example, argon �admixture of
roughly 20%�, the plasma parameters ne and Te can be ob-
tained from a line-ratio method and absolute intensity,
respectively.5 The gas temperature �Tgas� is determined from
the rotational structure of molecular radiation.5 The atomic-
to-molecular density ratio is correlated with the intensity ra-
tio H� line to molecular radiation �Fulcher transition, 600
-630 nm� via the ratio of the corresponding emission rate
coefficients.5,6 As shown in Fig. 2, ne is much higher �almost
by a factor of 10� in the arc source than in the rf source,
whereas Te is roughly 1.5 eV lower. Probe measurements in
the arc source result in reasonable agreement in densities, but
give a 1 eV lower Te. This might indicate a non-Maxwellian
electron energy distribution function. Both sources show
FIG. 1. Examples of spectra with identification of important emission lines
�H� and H� are saturated�.
03A516-2 Fantz et al.
high gas temperatures �arc: 2000±300 K, rf: 1200±300 K�,
almost independent of power. Due to the higher ne and Tgas,
Downloaded 15 Mar 2012 to 218.199.90.49. Redistribution subject to AIP lic
the degree of dissociation is much higher in the arc source.
As expected for low-pressure plasmas, ne and H densities
increase with power, whereas Te remains almost constant.
B. Cesium and negative ions
Analysis of cesium emission lines and results for the rf
source are described in detail in an earlier article.7 In order to
demonstrate the development of extracted current densities
�jH− and je� and a correlation with the cesium line at 852 nm
�neutral cesium�, Fig. 3 shows a sequence of discharges �rf
source� starting with cesium evaporation. The Cs852 line
increases steadily, whereas the plasma parameters �ne, Te,
and H density� remain constant, indicated by a constant H�
line. After 1 h of evaporation jH− increases, whereas the ex-
tracted electron current decreases, clearly indicating the so-
called cesium effect. This means the level of Cs intensity at
constant power and pressure can be used to adjust the evapo-
ration. Time traces of the Cs852 line and H� line during
cesium evaporation in the arc source are shown in the upper
part of Fig. 4. Again a steady increase from discharge to
discharge is observed in the Cs signal, whereas H� shows the
FIG. 2. Plasma parameters in the arc source �left part� and rf source �right
part� determined by OES. The open symbols refer to Langmuir probe mea-
surements.
Rev. Sci. Instrum. 77, 03A516 �2006�
FIG. 3. Temporal development of extracted current densities and spectro-
scopic signals of the rf source after the start of Cs evaporation �in h�.
ense or copyright; see http://rsi.aip.org/about/rights_and_permissions
stable plasma conditions. With increasing pulse length the Cs
signal seems to saturate, however, this behavior depends
strongly on the evaporation rate which influences also the
formation of negative ions. In general, it is observed that the
evaporation rates needed for arc sources are much higher
than those for the rf sources �roughly by a factor of 10�. This
is partly due to the long-pulse operation and partly due to the
evaporation of tungsten from the hot tungsten filaments
��2600 K�. Thus, a fresh layer of cesium will be buried
under tungsten.8 The continuum emission of tungsten is
clearly visible in the lower part of Fig. 1. However, tungsten
emission lines are very weak, just allowing to get an upper
estimate. In contrast to rf sources cesium ion lines are intense
in the arc source �Fig. 1�, resulting typically in an ion-to-
neutral ratio of 1000 �rf sources typically show values of 30�,
which means more cesium is ionized because of the high ne.
Since H� emission is influenced by the mutual neutralization
process at density ratios H−/H�10−3, the line ratio H� /H�
offers a method to obtain the negative ion density.6,9 For
constant plasma parameters the line ratio correlates directly
with the extracted current density as shown in the rf source.6
The lower part of Fig. 4 shows this direct correlation for a
time trace of the arc source. The two experiments yield very
similar negative ion densities in the observed plasma volume
at comparable extracted current densities.
IV. OPTIMIZATION OF NEGATIVE IONS
The whole set of particle densities obtained in the two
discharges close to the plasma grid are summarized in Fig. 5
for hydrogen discharges �0.3 Pa� at 110 �rf� and 45 kW �arc�
which represent similar performance. Due to the higher gas
temperature in the arc source �2000 K� as in the rf source
�1200 K�, the neutral particle density differs for the same
pressure. As already mentioned, ne and thus Hx
+ densities as
FIG. 4. Time traces of spectroscopic signals �H� ,Cs� with Cs evaporation
�top� and correlation of the line ratio H� /H� with the extracted current
density in the arc source �bottom�. The pulse length varies from 50 to 140 s.
03A516-3 Plasma diagnostic tools for H sources
well as H density are much higher in the arc than in the rf
source, whereas Te is lower. Since the surface effect is the
Downloaded 15 Mar 2012 to 218.199.90.49. Redistribution subject to AIP lic
dominant formation mechanism for H−, low ne and Te are
favorable for reducing losses via electron stripping and mu-
tual neutralization. Lowering Te from 3.5 to 2 eV in the rf
source would reduce the losses by a factor of �3, which are
then determined by the mutual neutralization only.6 Losses
are higher in the arc source than in the rf source due to the
much higher ne. On the other hand, the higher Hx
+ and H
densities produce more H− by the surface mechanism. Sys-
tematic measurements show that enhancement of H− corre-
late with the H density in the arc source.4 In balance, H−
densities are similar for the two sources ��1017 m−3� as it is
indeed measured at comparable conditions �Fig. 5�. Operat-
ing in an almost Cs-free source results in a factor of 10 lower
H− densities �light gray bars in Fig. 5�, reproducing exactly
the difference in extracted current densities.
In both sources cesium densities are orders of magnitude
below hydrogen densities �Fig. 5�. Due to this small amount,
changes in plasma parameters caused by cesium evaporation
are not observed. In the arc source much more cesium is
ionized than in the rf source and the total amount of cesium
�neutrals plus ions� is roughly an order of magnitude higher.
This matches with the higher evaporation rate used in the
arc, which is needed because of the simultaneous evapora-
tion of tungsten, as discussed in Sec. III B. The measured
neutral tungsten density is ten times higher than the neutral
cesium density. Since the ionization energy is also low
�7.86 eV� and since Te in the plasma volume close to the
tungsten filaments is around 5 eV, it is expected that most of
the tungsten is also ionized. Comparable densities indicate
that W and Cs evaporation rates are quite similar as sup-
ported by examination of surface layers.8 It can be expected
that cesium consumption can be considerably reduced avoid-
ing poisoning by tungsten. Further reduction can be expected
by optimizing the cesium distribution, monitored with OES.
In summary, it was shown that OES is a very valuable tool
for understanding and monitoring plasma dynamics, in par-
ticular, the cesium balance, which is most important for re-
liable and reproducible operation in long pulses.
1 P. Franzen, H. D. Falter, E. Speth, W. Kraus, M. Bandyopadhyay, A.
Encheva, U. Fantz, Th. Franke, B. Heinemann, D. Holtum, C. Martens, P.
McNeely, R. Riedl, A. Tanga, R. Wilhelm, Fusion Eng. Des. 74, 351
FIG. 5. Particle densities in the two types of negative ion sources �similar
performance� measured by spectroscopic techniques.
Rev. Sci. Instrum. 77, 03A516 �2006�
�2005�
2 E. Speth, H. D. Falter, P. Franzen, U. Fantz, M. Bandyopadhyay, S. Christ,
ense or copyright; see http://rsi.aip.org/about/rights_and_permissions
A. Encheva, M. Fröschle, D. Holtum, B. Heinemann, W. Kraus, A.
Lorenz, Ch. Martens, P. McNee-ly, S. Obermayer, R. Riedl, R. Süss,
A. Tanga, R. Wilhelm, and D. Wünderlich, Nucl. Fusion �submitted�.
3 D. Boilson, H. P. L. de Esch, R. S. Hemsworth, M. Kashiwagi, P.
Massmann, and L. Svensson, Rev. Sci. Instrum. 72, 1093 �2002�.
4 R. S. Hemsworth, D. Boilson, U. Fantz, L. Svensson, H. P. L. deEsch, A.
Krylov, P. Massmann, and B. Zaniol, AIP Conference Proceedings 763
Subseries: Accelerators and Beams, edited by J. D. Sherman and Y. J.
Belchenko, AIP, 2005.
5 U. Fantz, Contrib. Plasma Phys. 44, 508 �2004�.
6 U. Fantz, H. Falter, P. Franzen, D. Wünderlich, M. Berger, A. Lorenz, W.
Kraus, P. McNeely, R. Riedl, and E. Speth, Nucl. Fusion �submitted�.
7 U. Fantz, M. Bandyopadhyay, H. D. Falter, P. Franzen, B. Heinemann, W.
Kruas, P. McNeely, R. Riedl, E. Speth, A. Tanga, and R. Wilhelm, Fusion
Eng. Des. 74, 299 �2005�.
8 A. Krylov, D. Boilson, U. Fantz, R. S. Hemsworth, O. Provitina, S.
Pontremoli, and B. Zaniol, Nucl. Fusion �accepted�.
9 U. Fantz and D. Wünderlich, J. Phys. D: Appl. Phys. �to be submitted�.
03A516-4 Fantz et al. Rev. Sci. Instrum. 77, 03A516 �2006�
Downloaded 15 Mar 2012 to 218.199.90.49. Redistribution subject to AIP lic
ense or copyright; see http://rsi.aip.org/about/rights_and_permissions