Fast lithium-beam spectroscopy of tokamak edge plasmas
E. Wolfrum,a) F. Aumayr, D. Wutte, and HP. Winter
Institut fu’r Allgemeine Physik Technkhe Universitiit Wien, Austria
E. Hintz, D. RusbUldt, and FL P. Schorn
Institut fiir Plasmaphysik, KFA Jiilich GmbH, Ass. Euratom-KFA, Germany
(Received 30 December 1992; accepted for publication 26 April 1993)
magnetically confined hot plasmas (the so-called plasma edge) must be well understood for
successful development of future thermonuclear
detailed edge plasma diagnostics are in great demand. By injecting a fast Li beam into the edge
plasma region, a great number of information can be obtained with excellent space and time
resolution. This so-called Li-beam plasma spectroscopy gives access not only to edge plasma
density profiles from the collisionally excited Li atoms, but also to the impurity concentration
and temperature profiles via line emission induced by electron capture from the injected Li
atoms by the impurity ions. Full utilization of all capabilities requires a reliable data base for the
atomic collision processes involving injected Li atoms and plasma constituents (i.e., electrons,
hydrogen ions, and relevant impurities in their various charge states), since a precise modeling
of Li beam attenuation and excited-state composition has to be made for evaluating desired
plasma properties from the related spectroscopical measurements. The most recent methodical
improvement permits a fully consistent determination of absolute edge plasma density profiles
by measuring only relative LiI line emission profiles. This is of special interest for investigating
rapid edge plasma density fluctuations in connection with, e.g., ELMS, L-H mode transition,
turbulence or edge cooling by impurity injection. This paper describes the capabilities of
Li-beam edge plasma spectroscopy by way of illustrative examples from measurements at the
tokamak experiment TEXTOR.
interaction and impurity transport processes in the outermost region of
fusion reactors. To this goal, sufficiently
1. U-BEAM PLASMA SPECTROSCOPY-PRINCIPLES
plasmas for thermonuclear fusion studies relies to a great
deal on the understanding of reaction kinetics and trans-
port phenomena in the plasma boundary (the so-called
edge region). Due to plasma-wall interaction
plasma will become contaminated, which gives rise to en-
ergy and particle losses as well as local and global plasma
instabilities. All these processes influence again the plasma-
wall interaction, and so on. Due to these interrelations a
quite detailed understanding of the plasma edge behavior is
made difficult and can only be gleaned from accurate mea-
surements of all relevant edge plasma parameters.
The so far common plasma diagnostic techniques, in
particular “passive” photon
throughout the electromagnetic spectrum, are not ideally
suited for investigations in the plasma edge because of its
relatively small extension and strong inhomogeneity. Con-
ditions in the plasma edge are determined by a large num-
ber of interrelated atomic collision processes, which neces-
sitates extensive plasma modeling for extraction of any
information of interest from the diagnostic raw data. Such
modeling processes give rise to considerable ambiguities.
Better edge plasma diagnostics are thus needed and should
comply with the following requirements: Excellent spatial
of hot, dense magnetically confined
“Also at Institut fiir Plasmaphysik, KFA Julich GmbH, Ass. Euratom-
rameters should not strongly depend on the modeling pro-
All these goals can be met by fast Li-beam plasma
spectroscopy, i.e., the injection of keV to multi-keV neutral
lithium beams into the plasma edge, to induce collisional
processes which deliver the desired diagnostic informa-
tions. As an example, the plasma electron density can be
determined by observing the LiI resonance line emission
along the injected beam.i4 In close relation to this, mag-
netic field distributions inside a tokamak discharge can be
determined5P6 by observing Zeeman splitting of the electron
impact-induced LiI resonance line. By combining Li beam
injection with laser-induced fluorescence and injection of
other atomic species, the electron temperature can be
measured.‘** Furthermore, electric field strengths can be
determined by observing optically forbidden LiI transitions
as excited by dye laser radiation.“”
All of the so far mentioned diagnostic applications rely
on photon emission from electron impact-excited Li atoms
[Li impact excitation spectroscopy (LGIXS)].
the fact, however, that also collisions with protons and
impurity ions lead to LiI excitation,”
for quantitative evaluation of Li-IXS data (see below).
A second class of atomic collisions relevant for fast
Li-beam plasma spectroscopy involves electron capture
from the injected Li atoms by plasma protons and/or im-
purity ions, which’ produces excited states of the latter. In
a conceptually equivalent way, fast H atom beams are al-
as well as temporal ((ms)
of plasma disturbance, evaluated plasma pa-
We point to
which is of relevance
Rev. Sci. Instrum. 64 (8), August 1993
@ 1993 American Institute of Physics
Li - ion source
+ beam transportlfocussing
neutralizer - cell
cleaning field to edge plasma
FIG. 1. Principal components of a fast neutral Li beam injector for application in IA-beam edge plasma spectroscopy.
ready utilized for detection of fully stripped light impurity
ions in tokamak plasmas.‘2-‘5 However, as first shown in
Ref. 16, impurity ion diagnostics in the edge plasma may
better be carried out with fast neutral Li beams, since
much more favorable signal-to-background conditions for
the detection of fully and, in particular,
stripped impurity ions can be achieved. Furthermore, the
corresponding dominant line emission appears at compa-
rably longer wavelengths, since electrons from Li are cap-
tured into higher excited states, which offers considerable
The first successful application of Li charge exchange
spectroscopy (Li-CXS) in a tokamak edge plasma con-
cerned C5+ ion detection in TEXTOR,”
progress in this field has been made since.” Li-CXS may as
well be utilized for hydrogen ion detection in the edge
plasma region.” Since electron capture from fast Li atoms
into multicharged ions is accompanied by negligibly small
transfer of kinetic energy, profiles of such produced impu-
rity emission lines can well serve for spatially resolved ion
demonstrated.” In summary, fast Li beam spectroscopy
provides a very versatile instrumentary for investigation of
practically all relevant edge plasma properties, from which
the physically relevant information can be derived without
excessive plasma modeling. However, one has to keep in
mind that the injected Li beam is rapidly attenuated and its
excited-state composition is strongly changed along the
atomic beam path through the plasma edge.t8 This is not
only of concern for absolute electron density determination
via Li-IXS, but also for Li-CXS, since the relevant electron
capture processes strongly depend on the ionization energy
of the “active electron” initially bound to the injected Li
atoms. Consequently, a full utilization of Li-beam plasma
spectroscopy needs careful modeling of the Li beam atten-
uation and excited-state composition4 as will be further
explained in Sec. III.
II. NEUTRAL Ll BEAM PRODUCTION AND LINE
A. Production of fast Ll beams
As sketched in Fig. 1, fast neutral Li beam production
essentially involves the following elements:
(a) A high current Li+ ion source with ion extraction
and beam formation components,
Rev. Sci. Instrum., Vol. 64, No. 8, August 1993
(b) a neutralizer cell for producing the fast Li atom
(6) a deflector magnet for removal of remaining Li
ions behind the neutralizer cell,
(d) neutral beam transport components toward the
edge plasma to be diagnosed.
As will be shown below, the Li ion source is quite
critical, while all other components pose no particular
problems. For high-resolution
ments, neutral beam equivalent current densities of O,l
mA/cm* might already be sufficient, but for Li-CXS at
least 1 mA/cm2 is necessary. Assuming an ion neutraliza-
tion efficiency of typically > 50% (cf. below), primary Li’
current densities have to be a factor of 2 higher than the
above stated neutral current densities. For achieving a spa-
tial resolution of typically better than 1 cm, total Li+ ion
currents in the l-10 mA range are needed, and a suffi-
ciently small emittance of the ion source and acceptance of
the subsequent ion-optical systems have to be secured.
Li+ ion beams from &eucryptite
applied for Li-beam plasma spectroscopy by a number of
For 60 keV Li+ ion beam energy, currents of
3-5 mA from a 1-in.-diam emitter surface have been
reported.22 A detailed technical report is available on the
ASDEX f3-eucryptite type Li ion gun.23 While such ion
sources are rather convenient because of their comparably
long lifetime and high ion beam purity, they are severely
limited in extractable total ion current with sufficiently
small emittance. Consequently, the achievable ion current
densities have so far not been higher than about 1
mA/cm’. Moreover, there is no simple way for rapid chop-
ping of the such produced ion beams, as would be neces-
sary for Li-CXS applications.
Another way for Li+ ion production utilizes plasma
ion sources based on various types of low-pressure arc dis-
charges fed by Li vapor together with auxiliary gases.2k26
Extracted ion current densities of 15 mA/cm2 have been
reported for a DUOPICATRON
details on the related ion beam quality.27 At TEXTOR,
20-30 keV Li+ ion beams with typically 4 cm diameter and
currents of well above 10 mA are produced from a reflex
discharge ion source followed by a 55” stigmatically focus-
ing magnet.18 The mass-separated ‘Li ion beam is subse-
quently passed through a Li vapor-filled cell, and the re-
suiting neutral Li atom beam travels another 2 m toward
the TEXTOR plasma edge, with the remaining Li ions
being magnetically deflected into a beam dump. There is
plasma density measure-
emitters2’ have been
source, however, without
Lithium beam spectroscopy
still considerable potential for much better ion beam qual-
ity than so far achieved, if multiple-aperture extraction is
applied instead of the more common single hole ex&action
geometry. However, all kinds of low-pressure arc sources
suffer from limited cathode filament lifetime, Li vapor con-
densation and other technical problems, which concern
both the ion source reliability and its operational lifetime
(typically not more than 5-10 h). To minimize such diffi-
culties, the very attractive features of ECR plasma ion
sources such as their high ion currents, low emittance, and
rather long lifetime’* can probably be utilized for Li+ ion
beam production, as well. A recently developed compact
2.45 GHz ECR ion sourcez9 is now adapted for Li+ ion
beam production, by applying different techniques of Li
vapor feeding into the ECR plasma region.
Conversion of the fast Lif ions into neutral beams is
routinely made via charge exchange in either Li’*‘* or Na3’
filled vapor cells. Given favorable conditions, with both
alkali species neutralization efficiencies of well above 50%
have been reported. Na is technically more convenient be-
cause of its sufficiently high-vapor pressure already at mod-
erate neutralizer cell temperature ( - 220 vs - 500 “C for
Li), but special care is needed to avoid diffusion of the Na
vapor from the exchange cell into the plasma region. On
the other hand, if Li vapor is used for neutralization, the
resulting Li diagnostic beam contains a thermal compo-
nent made up from Li vapor effusing from the charge ex-
change cell. Excitation of this thermal Li atoms inside the
plasma leads to LiI 670.8 nm background radiation which
makes pulsed operation of the fast neutral Li beam man-
B. Spectroscopical techniques
The essential quantities to be measured are the inten-
sities and/or spectral shapes of line emission induced due
to the Li atom beam interaction with the plasma constitu-
ents. To this purpose, light emitted from the interaction
region has to be collected by a lens system and transported
via fiber or lens optics to a spectroscopical detection sys-
tem, which can involve
(TEXTOR), interference filter-photomultiplier
tions (ASDEX, TEXTOR)
CCD camera combinations (TEXTOR).
Li beam induced line emission can be separated from
plasma background radiation by fast chopping of the diag-
nostic beam. Chopper frequencies of up to 500 Hz have
been utilized with the setup at TEXTOR. Spatial profiling
(i.e., measuring line emission intensity along the injected
Li beam axis) can be achieved by an electrically driven
mirror, to scan the line of observation along the beam
direction (TEXTOR), by one- or two-dimensional obser-
vation systems like a CCD camera (TEXTOR),
number of photomultipliers monitoring the beam at differ-
ent locations (ASDEX).
While the CCD setup provides excellent spatial reso-
lution (typically < 1 mm), the so far best time resolution
( < 1 ,w) has been obtained with sets of photomulipliers. If
it is necessary to subtract the background signal, the Li
beam modulating frequency itself limits the achievable
If necessary, the
or by a
Rev. !Sci. Instrum., Vol. 64, No. 6, August 1993
time resolution. The such obtained profiles have still to be
corrected for effects due to the observation geometry. This
is achieved by scanning along a Li diagnostic beam which
is injected into the plasma vessel filled with low-pressure
(typically 10-5-10-4 mbar) Hz or He gas. Under the then
given single collision conditions no appreciable attenuation
of the beam will take place, which can be checked by a
linear signal intensity dependence on the filling gas pres-
sure. The such obtained intensity profiles include the radial
dependences of all calibration-relevant
solid observation angle, transmission of filters (taking also
into account viewing angle-dependent Doppler shifts),
vignetting due to apertures in the (scanning) observation
system, and quantum efficiencies of the applied photon de-
III. Li BEAM MODELING AND RELATED ATOMIC
In order to obtain reliable quantitative diagnostic in-
formation from Li beam spectroscopy, the Li beam inter-
action with the plasma constituents has to be modeled. For
collisions with fast Li atoms one has to regard not only the
plasma electrons (as in the case of thermal Li beams), but
also hydrogen and impurity ions, since the Li atoms can
become excited as well as ionized in all of these collisional
processes (impact excitation, ionization, and charge ex-
change). Consequently, these excited LiI states undergo
both radiative and collisional deexcitation (the faster the
beams, the more important becomes collisional deexcita-
tion in plasma regions with higher density). The effective
lifetimes of these excited states tend to smear out structures
in, the electron density.
As has been shown in Ref. 18, an appropriate model
must assume that the occupation number for the Li(2p)
state [which is proportional
photon intensity at 670.8 nm), the plasma density n, and
the occupation numbers of other Li(nZ) states Ni [i= 1
denotes Li(&), i=2 Li(2p),
following system of coupled differential equations:
to the measured LiI(2.+2p)
etc.] are interrelated by the
T(Z) 1 +bii)Nj(z),
Here a local plasma coordinate z along the injected Li
beam (z=O at the entrance of the Li beam into the
plasma) has been introduced as the independent variable.
For the practical examples presented in Sec. IV, this coor-
dinate z along the neutral Li beam coincides with the small
tokamak radius r. The density of each plasma component
is expressed as the fraction of the electron density n, (for
impurity-free plasmas nP= n,) . Taking also impurity ions
into account requires additional assumptions, e.g., an av-
erage q(z) profile in combination with a model Z,,(z)
profile.4 Coefficients aij (i#j)
excitation and deexcitation processes for the Li atoms from
state j to state i, due to collisions with both the plasma
electrons and hydrogen ions.
in the rate equations refer to
Lithium beam spectroscopy
FIG. 2. (a) Typical radial LiI 670.8 nm emission profile (crosses) measured by 20 keV Li beam spectroscopy during TEXTOR discharge No, 52516.
(b) Plasma electron density as reconstructed from the LiI emission profiie in (a). (c) Li(2.s), Li(2~), and Li(3,1=0,1,2) components of the 20 keV Li
beam, calculated via Eq. (1) for the electron density given in (b). The accuracy of the reconstruction method can be judged from a direct comparison
of the Nzp population in (c) with the LiI(2.s-2p) emission intensity in (a), where Nzp has been added as a solid line.
Attenuation of particular Li states i due to charge ex-
change and ionization in collisions with plasma particles
and excitation as well as deexcitation to other bound states
j are included in coefficients a,. Finally, the spontaneous
emission processes are described by coefficients b,. Since
Eq. ( 1) takes also into account the finite lifetime of differ-
ent Li levels as well as population of these levels via cas-
cading, the calculated occupation number N,(z)
Lif 2p) state is directly related to the measured LiI (2.9-2~)
emission profile, with the proportionality factor depending
on the detection geometry. Of course, knowledge of abso-
lute cross sections or related rate coefficients is required for
all collisional processes between the different plasma con-
stituents and all Li atomic states of relevance. Available
experimental data on excitation, ionization, and charge ex-
change almost exclusively involve the Li(2s) ground state
only. Therefore, a critical evaluation of semiempirical and
theoretical cross-section formulas for collisional processes
also involving excited Li states has been carried out. The
such produced “atomic data base for Li-beam plasma spec-
troscopy” has been compiled at the IAEA.3’ It contains
multiparameter fit formulas for each cross section. The rate
coefficients calculated from these cross section data depend
of course on the Li beam velocity, but (fortunately) only
rather weakly on the plasma temperature T(z),‘~ because
of which in most cases “standard T(z) profiles” can be
used for the evaluation.
If now a particular electron density n,(z) along the
injected beam direction z is introduced, the system of cou-
pled differential equations ( 1) can be solved, resulting in
the local population densities Nj(Z) of the different Li(j)
states. However, the inverse task, i.e., reconstruction of
Rev. Sci. Instrum., Vol. 64, No. 6, August 1993
absolute electron densities n, from only relatively measured
LiI 670.8 nm diagnostic data is considerably more difficult.
Emission profiles L&(z) as delivered from the Li beam
diagnostic setup are directly proportional to the occupa-
tion number l\rZ(z) of the Li(2p) state. A suitable recon-
struction algorithm has recently been developed.4 It recov-
ers the plasma (electron) density exclusively from scanned
profiles of the relative LiI 670.8 nm emission intensity
along the diagnostic beam direction. In principle, in this
way any spatial variations in the electron density can be
reconstructed, despite of the smearing-out
tioned above.32 More details on this novel algorithmus may
be found in Ref. 4.
As an educative illustration, Fig. 2(a) shows a typical
radial LiI 670.8 nm emission profile obtained with the 20
keV Li injector at TEXTOR (for details of this experimen-
tal setup cf. Ref. 18). The corresponding electron density
as evaluated with the above described methods is given in
Fig. 2(b). It shows a pronounced radial structure in the
vicinity of the ALT II limiter position at z=46 cm. The
accuracy of such determined density profiles is subject to
errors in the applied atomic data base (estimated to about
f 20% in absolute n, magnitude, but only a few percent in
the relative radial density profile), and direct experimental
errors as signavnoise etc., for measuring the LiI emission
line intensity along the injected Li beam (dominant error
contribution to the relative radia1 n, dependence * 10%).
Numerical errors of our reconstruction method amount to
less than 3%. In Fig. 2(c), the population of Li(n,/)
cited states calculated from the density profile in Fig. 2 (b)
is shown. The high quality of our reconstruction method
may be judged from a direct comparison of the iyZP popu-
Lithium beam spectroscopy
FIG. 3. Plasma electron densities as observed for a series of TEXTOR
discharges. Comparison is made between data obtained from L&beam
plasma spectroscopy (-) and HCN-laser
(-. .-. .).
interferometry (Ref. 33)
lation in Fig. 2(c) with the LiI( 2s-2p) emission intensity
in Fig. 2(a).
IV. RECENT APPLICATIONS FOR TOKAMAK EDGE
A. Electron density measurements
Reconstruction of plasma edge density profiles from
the spatial variation of Li(2p-2s)
been carried out with the methods described in Sec. III.
Corresponding measurements have already been carried
out at TEXTOR and ASDEX. For example, edge density
profiles have been compared for OH-, L-, and H-discharge
phases at ASDEX. 4P32 In the same context, the remarkable
time resolution capability of Li-IXS could be demonstrated
by following the decay phase of an edge-localized mode
Regarding Li-IXS measurements at TEXTOR,
3 we have compared a series of edge plasma electron den-
sities (“density scan”) as obtained by means of 20 keV
Li-beam plasma spectroscopy with independent density
measurements by means of the TEXTOR
interferometer.33 Despite the large variation in plasma den-
sities between different shots, the agreement between re-
sults of both kinds of measurements is rather good. Re-
maining discrepancies should be attributed to the different
toroidal positions of both diagnostic systems, and espe-
cially to the fact that the spatial resolution of the HCN
diagnostics is rather limited (the related data had to be
obtained from Abel inversion of line-averaged electron
densities derived at nine discrete radial positions, the two
outermost ones being located at r=40 and 47.5 cm, respec-
emission intensity has
8. Impurity ion density measurements at TEXTOR
Li-beam charge-exchange plasma spectroscopy (Li-
CXS) makes use of electron capture processes according to
Eq. (2) from neutral Li atoms into excited states of impu-
rity ions, which then decay under emission of their char-
acteristic lines. From the respective intensities the concen-
tration of any particular impurity
ions Aq+ can be
Since charge exchange with Li takes place with rela-
tively large cross sections and ends up in comparably
highly excited states of the impurity ions, which only to a
small extent are populated by collisional processes with the
plasma constituents, a favorable signal-to-noise ratio can
be achieved. In the course of charge exchange with Li the
kinetic energy transferred to the impurity ion is negligible.
Therefore, the such induced impurity lines give a rather
direct information on the spatial ion distribution
edge plasma. If population of the different Li atomic states
in the plasma along the complete beam path is known, the
impurity ion density n,(z) can be calculated from the sig-
nal of the characteristic line radiation Sn(z):
SA(z)=k,pLinq(z) x np;.
Here z is the distance along the beam axis and vLi the
velocity of the Li atoms. The ni are the densities of the
Li( nl) states to be derived from model calculations (see
Sec. III) with known electron density and temperature and
the measured absolute current density of the Li beam.
The o i are emission cross sections for the characteris-
tic impurity lines produced as the result of electron capture
from a Li (n*,l* = i) atom into impurity ions Aqf , ending
up in atomic states A(q-‘)+ (n,Z). For electron capture
from the Li( 2.r) ground state these emission cross sections
have been determined experimentally
whereas for capture from excited Li states the related cross
sections have been calculated according to the “classical
over the barrier mode1.“35 The parameter kA includes the
wavelength-dependent sensitivity and the aperture of the
optical detection system, and has to be determined exper-
imentally. A detailed description of the experimental con-
ditions at TEXTOR can be found elsewhere.18
The determination of both electron and impurity den-
sities with a 20 keV Li beam has been employed at
TEXTOR to contribute to the study of active edge cooling.
During these experiments, the deliberate and well-adjusted
puffing of neon gas into the plasma boundary was used to
lower the electron temperature considerably in the outmost
10 cm of the plasma, while T, stays more or less constant
in the hot center.36’37 As an example, we report on an
ohmically heated discharge with a line-averaged central
electron density of 2.4X 1013 cmm3 and central electron
temperature of 1.0 keV, the plasma current being 350 kA
and the limiters placed at a minor radius of 46 cm.
Neon has been injected during the time period of 0.8-
2.4 s, where a steady state could be reached by adjusting
the gas influx to a constant intensity of a Ne’+ spectral
emission line. Figure 4(a) shows electron temperatures
taken at 0.7 s (ohmic prephase, without neon) and at 1.5 s
(edge cooling with neon injection) employing optical spec-
troscopy involving the interaction of a thermal helium
beam with the plasma.3*Y39 It can clearly be seen that T, is
(cf. e.g., Ref. 34),
Rev. Sci. Instrum., Vol. 64, No. 8, August 1993
Lithium beam spectroscopy
FIG. 4. (a) Radial electron temperature profiles during ohmic prephase (0.375-0.875 s) and flat top phase (1.125-2.125 s), respectively, of TEXTOR
discharge No. 51337. (b) Electron density profiles for TEXTOR discharge No. 51337 as defined under (a). (c) Radial C6+ profiles obtained by means
of Li-beam plasma spectroscopy for TEXTOR discharge No. 5 1337 as defined under (a). (d) Relative concentration of c6+. Neon gas puffing during
the time interval of 0.9-2.2 s causes lower C6+ concentration and a flatter slope in the investigated region.
reduced considerably by nearly a factor of 3 in the vicinity
of the limiter radius, while its value at r=O (not depicted,
different method) even slightly rises from 1.0 to 1.2 keV.
This somewhat contradictory
creased ohmic heating power
detachmenta and, as a consequence, better particle con-
finement during the injection phase. In parallel, Z,, rises
from 1.8 to 3.0 on the center axis. Neon cooling seems to
be a method which selectively modifies the properties of
the boundary layer while the influence on the overall per-
formance of the center plasma is less pronounced. The
physical mechanism responsible for this is electromagnetic
line radiation especially of the edge-located lower ioniza-
tion states of neon: For the shot under discussion, the total
radiated electromagnetic power is 360 kW during gas in-
jection (measured with a bolometer), while it is only 100
kW in the prephase. Compared to the total power of 400
kW launched into the discharge, nearly 90% of it is radi-
ated isotropically into space and is thus prevented from
being loaded onto limiter surfaces. Li-IXS shows that elec-
tron density profiles are not altered significantly by neon
cooling [see Fig. 4(b)], in contrast to density profiles of
C6+ impurities measured by means of LGCXS. Their ab-
solute [Fig. 4(c)] as well as relative concentrations [Fig.
4(d)] are diminished by roughly a factor of 2. The addi-
tionally measured reduction of the carbon release on the
limiter surfaces by 35% may give an explanation: Lower
electron temperatures lead to lower sheath potentials above
the limiter surfaces and thus to reduced physical sputtering
yields, being especially relevant to self-sputtering by highly
ionized carbon states,20 where, e.g., a potential of 6 x 3kT,
would be given for C6+ -C sputtering.
behavior is due to an in-
induced by plasma
2290 Rev. Sci. Instrum., Vol. 64, No. 8, August 1993
C. Impurity ion temperature measurements
Application of Li-CXS for temperature measurements
of fully stripped C6+ ions in the boundary layer of the
TEXTOR tokamak has recently been demonstrated and
described in detail,20 by utilizing the same atomic collision
processes [cf. Eq. (211 as for impurity ion density determi-
nation. The impurity ion temperature can be deduced from
spectral broadening of the resulting emission line radiation
due to the velocity-dependent Doppler effect, assuming a
Maxwellian velocity distribution of the emitting particles,
However, other mechanisms leading to line broadening
have to be taken into account, as outlined below.
From laboratory measurements34 it is known that after
C6+-Li charge exchange the electrons are captured mainly
into the n = 7 and n = 8 principal shells with high I values,
giving rise to optical radiation at 343 nm (7-5) and 529 nm
(8-7), respectively. Because of more experimental conve-
nience the latter transition has been used at TEXTOR and
its spectral profile has been analyzed. For achieving a good
spatial resolution across the boundary layer, at the moment
still several reproducible discharges are necessary. By as-
suming a Maxwellian velocity distribution, the spectral line
shape of a single transition (n,t) + (n- 1,E’) can be ex-
pressed by a Gaussian profile
S(A) aexp[ -r+)‘],
where A0 is the central wavelength of this profile and AA. its
width, The ion temperature Timp can be obtained from a
least-squares fit of the experimental data to relation (4):
Lithium beam spectroscopy
Plasma Radius [cm3
FIG. 5. Cs+ temperature profiles as obtained for various conditions of a
neutral beam-heated TEXTOR discharge with line-averaged centrai elec-
tron density of 3.5x lOI m-“, which has been heated by two 1.8 MW
neutral deuterium burns in “co” and “counter” injection. The pro8les
have been measured during time intervals with a single and both injectors
in operation, respectively.
with m the mass of the impurity ion and c the velocity of
light. Resides the Doppler effect, two other processes have
to be taken into account, namely, the Zeeman effect and
colhsional Z-level mixing, which both cause additional
broadening of the spectral profile. In tokamak edge plas-
fields of more than 2 T cause considerable
Zeeman splitting of electronic energy levels, and the such
further broadened spectral profiles would lead to system-
atically too high temperatures.
Therefore, so-called derating factors qzeeman have been
calculated dependent on the magnetic field strength and
the measured temperatures.*’
filters could have been used in order to observe only those
Zeeman components which experience the smallest Zee-
man shift. Collisions of the C5+ particles resulting from
charge exchange of injected Li atoms with C6+ with other
plasma constituents can cause transfer between the differ-
ent electronic I states within the same principal n shell. In
tokamak edge plasmas this Z-level mixing occurs on a faster
time scale than the radiative lifetime of the respective state.
Consequently, all allowed transitions between two n shells
will be observed simultaneously. The corresponding lines
have to be superimposed to a sum profile by using the
correct central wavelengths and relative intensities, from
which a temperature-dependent corrective derating factor
flLmir can be derived (for details cf. Ref. 20). The two
nonthermal broadening mechanisms together with the ap-
paratus profile of the applied spectrometer determine a
lower limit of about 20-30 eV for such temperature mea-
surements. As an example, C6+ temperature profiles for
beam-heated TEXTOR plasmas are drawn in Fig. 5. A
discharge with line-averaged central electron density of
3.5~ lOI mm3 has been heated with two 1.8 MW neutral
deuterium beams directed in “co-” and “counter”
tion, and impurity ion temperature profiles for time inter-
vals with respectively one and both injectors in operation
have been shown. Two main features are obvious. ( 1) The
second (“counter”) beam heats the plasma further up and
(2) the gradient of TbP(r)
injection, in parallel with the development of a more or less
constant pedestal at rz4-0 cm. These features may be in-
terpreted as being due to improved plasma confinement in
the hot center.41 In Ref. 20 it has been explained that the
impurity ion temperatures as measured by Li-CXS in the
here described way should more or less agree with the
hydrogen ion temperatures at the same radial positions,
whereas a considerable deviation from the corresponding
electron temperatures may be given. This is caused by the
relatively short equipartition
hydrogen ions (typically several tens of ps), as compared
with the much longer equipartition times between impuri-
ties and plasma electrons (typically several ms) . The latter
times are too long to permit achievement of equilibrium
between electrons and impurities during the edge conflne-
ment times (typically 1 ms) in TEXTOR.
It can be stated that Li-beam plasma spectroscopy has
now reached a certain degree of maturity.
make it a more widely applicable (standard)
tool, further operational experience is desirable and tech-
nical efforts have to be devoted mainly to the construction
of reliable, long-lived Lif ion sources for easy-to-operate
neutral Li beam injectors.
gets steeper during counter
time between impurity and
The Austrian authors have been supported by Kom-
mission zur Koordination der Kernfusionsforschung
Austrian Academy of Sciences.
i K. Kadota, K. Tsuchida, Y. Kawasumi, and J. Fuji& Plasma Phys. 20,
*K. McCormick, M. Murmann, and M. El Shaer, J. Nucl. Mater. 121,48
‘K. McCormick and the ASDEX team, Rev. Sci. Instrum. 56, 1063
4J. Schweinxer, E. Wolfrum, F. Aumayr, M. Pijckl, HP. Winter, R. P.
Schom, E. Hintz, and A. Unterreiter, Plasma Phys. Controlled Fusion
34, 1173 (1992).
5 J. Fujita and K. McCormick, Proceedmgs of the Sixth European Con-
ference on Controlled Fusion and Plasma Physics Moscow, 1973, p.
6K. McCormick, M. Kick, and J. Olivain, Proceedings of the Eighth
European Conference on Controlled Fusion Plasma Physics Prague,
1977, p. 140.
‘E. Hintz and P. Bogen, J. Nucl. Mater. l28/129, 229 (1984).
*K. Kadota, H. Matsuoka, H. J. Ramos, S. Miyake, K. Tsuchida, and J.
Fujita, J. Nucl. Mater. 12W129, 960 (1984).
9U. Rebhan, N. J. Wiegart, and H. J. Kunxe, Phys. Lett. 85A, 228
“T. Gda and K. Takiyama, Proceedings of IAEA TCM on “Research
using small tokamaks” Nagoya, Japan, 1987.
“F. Aumayr, M. Fehringer, and HP. Winter, J. Phys. B 17,4185 (1984).
i2V. V. Afrosimov, Y. S. Gordeev, and A. N. Zinov’ev, Sov. Tech. Phys.
L&t. 3, 39 (1977).
13V V Afrosimov, Y. Gordcev, A. N. Zinov’ev, and A. A. Korotkov, . .
JETP Lett. 28, 500 (1978).
14R. C. Isler, Phys. Rev. Lett. 38, 1359 (1977).
15R. J. Fonck, M. Finkenthal, R. J. Goldstone, D. L. Hemdon, R. A.
Hulse, R. Kaita, and D. D. Meyerhofer, Phys. Rev. Lett. 49, 737
16HP. Winter, Comments At. Mol. Phys. 12, 165 (1982).
Rev. Sci. Instrum., Vol. 64, No. 8, August 1993
Lithium beam spectroscopy
“A. Pospieszczyk, F. Aumayr, H. L. Bay, E. Iiiitz,
Lie, G. G. Ross, D. Rusbiildt, R. P. Schom, B. Schweer, and HP.
Winter, J. Nucl. Mater. 162-164, 574 (1989).
‘*R. P. Schom, E. Hintz, D. Rusbiildt, F. Aumayr, M. Schneider, E.
Unterreiter, and HP. Winter, Appl. Phys. B 52, 71 (1991).
i9F. Aumavr and HP. Winter. Ann. Phvs. (Leiozin) 42. 228 (1985).
20R. ,P. Schom, E. Wolfrum, F. Aumayr:E. Hi&,
Winter, Nucl. Fusion 32, 351 (1992).
*I J P. Blewett and E. 3. Jones, Phys. Rev. 50, 464 (1936).
“W. P. West, D. M. Thomas, E. S. Ensberg, J. S. deGrassie, and J. F.
Baur, Rev. Sci. Instrum. 57, 1552 (1986).
23K. McCormick and M. Kick, Report IPP 111/85, Max Planck Institut
fur Plasmaphysik Garching, Germany, 1985.
24P. G. Weber, Rev. Sci. Instrum 54, 1506 (1983).
2sH. Yamaoka, K. Kadota, K. Okuno, S. Masamune, S. Miyake, and J.
Fujita, Proceedings of the Tenth Symposium on Ion Sources and Ion-
Assisted Techniques Kyoto, Japan, 1986, p. 65.
26H. L. Bay, E. Duilni, and P. Lcismann, Report J&2062 KFA Jiilich,
“P. G. Weber and S. Masamune, Rev. Sci. Instrum. 57, 2032 (1986).
*‘Y Jongcn and C. M. Lyneis, in The Physics and Technology of Ion
Sources, edited by I. G. Brown (Wiley, New York, 1989), Chap. 10.
29S. Fiedler and HP. Winter, Rev. Sci. Instrum. 63, 2532 (1992).
%K. McCormick, Report IPP III/82
physik Garching, Germany, 1983.
P. Leismann, Y. T.
“F. Aumayr, R. K. Janev, M. Schneider, J. J. Smith, D. Wutte,
J. Schweinzer, and HP. Winter, “Atomic data base for Lithium beam
edge plasma spectroscopy,” Report IAEA-INDC-NDS
Wien, Austria, 1993.
32F. Aurnayr, R. P. Schom, M. Piickl, 3. Schweinzer, E. Wolfrum, K.
McCormick, E. Hintz, and HP. Winter, J. Nucl. Mater. 196-198, 928
33H. Sol&v&h, Phys. B 48, 225 (1989).
%E. Wolfium, R. Hoehtra, F. J. de Hecr, R. Morgenstem, and HP.
Winter, J. Phys. B 25, 2597 (1992).
-“A. Niehaus, J. Phys. B 14, 1161 (1986).
36U. Sam, J. Bocdo, G. Bertschinger, K. H. Dippel, H. Euringer, K. H.
Finken, D. Gray, D. Hillis, A. Pospieszcyk, D. Reiter, M. Tokar, and
B. Unterberg, J. Nucl. Mater. 196-198, 633 ( 1992).
37U. Sarnm et aI., Proceedings of the 14th International Conference on
Plasma Physics and Controlled Nuc1ca.r Fusion Research, Wiirzburg,
‘*B. Schweer, A. Pospieszczyk, G. Mank, and U. Samm, Proceedings of
the 18th European Conference on Controlled Fusion and Plasma Phys-
ics, Berlin, 3-7 June 1991. and Enrophys. Conf. Abstracts 15C Part IV,
39B Schwcer, G. Mank, A. Pospieszczyk, B. Brosda, and B. Pohlmeyer,
J.‘Nucl. Mater. 196-198, 174 (1992).
‘%. Samm et aL, KFA-Report Ji&2123, March 1987.
“H. Conrads, H. Euringer, and D. Rusbiildt, J. Sci. Instrum. (in press).
6. Rusbiildt, andHP.
Max Planck Institut fur Plasma-
Rev. Sci. Instrum., Vol. 64, No. 8, August 1993
Lithium beam spectroscopy