UC Irvine
UC Irvine Previously Published Works
Title
Fast-ion D-alpha measurements at ASDEX Upgrade
Permalink
https://escholarship.org/uc/item/8t82v54k
Journal
Plasma Physics and Controlled Fusion, 53(6)
ISSN
0741-3335
Authors
Geiger, B
Garcia-Munoz, M
Heidbrink, WW
et al.
Publication Date
2011-06-01
DOI
10.1088/0741-3335/53/6/065010
Copyright Information
This work is made available under the terms of a Creative Commons Attribution License,
availalbe at https://creativecommons.org/licenses/by/4.0/
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
IOP PUBLISHING PLASMA PHYSICS AND CONTROLLED FUSION
Plasma Phys. Control. Fusion 53 (2011) 065010 (19pp) doi:10.1088/0741-3335/53/6/065010
Fast-ion D-alpha measurements at ASDEX Upgrade
B Geiger
1
, M Garcia-Munoz
1
, W W Heidbrink
2
, R M McDermott
1
,
G Tardini
1
, R Dux
1
, R Fischer
1
, V Igochine
1
and the ASDEX Upgrade
Team
1
Max-Planck Institut f
¨
ur Plasma Physik, Garching, Munich, Germany
2
Department of Physics and Astronomy, University of California, Irvine, CA 92697, USA
E-mail: benedikt.geiger@ipp.mpg.de
Received 14 January 2011, in final form 14 March 2011
Published 7 April 2011
Online at
stacks.iop.org/PPCF/53/065010
Abstract
A fast-ion D-alpha (FIDA) diagnostic has been developed for the fully tungsten
coated ASDEX Upgrade (AUG) tokamak using 25 toroidally viewing lines
of sight and featuring a temporal resolution of 10 ms. The diagnostic’s
toroidal geometry determines a well-defined region in velocity space which
significantly overlaps with the typical fast-ion distribution in AUG plasmas.
Background subtraction without beam modulation is possible because relevant
parts of the FIDA spectra are free from impurity line contamination. Thus,
the temporal evolution of the confined fast-ion distribution function can be
monitored continuously. FIDA profiles during on- and off-axis neutral beam
injection (NBI) heating are presented which show changes in the radial
fast-ion distribution with the different NBI geometries. Good agreement
has been obtained between measured and simulated FIDA radial profiles
in MHD-quiescent plasmas using fast-ion distribution functions provided by
TRANSP. In addition, a large fast-ion redistribution with a drop of about 50%
in the central fast-ion population has been observed in the presence of a q = 2
sawtooth-like crash, demonstrating the capabilities of the diagnostic.
(Some figures in this article are in colour only in the electronic version)
1. Introduction
In present fusion devices, fast ions (ions with energies significantly above the thermal energy)
are produced by external heating systems such as neutral beam injection (NBI) and ion cyclotron
resonance heating (ICRH). Their confinement is essential because fast ions contribute to plasma
heating and current drive and can, if poorly confined, even lead to damage of the first wall [1].
Several techniques exist to study the fast-ion confinement, such as fast-ion loss detectors
(FILD) [2], neutral particle analyzers (NPA) [3], gamma ray tomography [4], collective
0741-3335/11/065010+19$33.00 © 2011 IOP Publishing Ltd Printed in the UK & the USA 1
Plasma Phys. Control. Fusion 53 (2011) 065010 B Geiger et al
Thomson scattering (CTS) [5], neutron spectroscopy [6] and fast-ion D-alpha (FIDA) [7]
measurements. The FIDA technique, which is similar to spectroscopic measurements of
fast alpha particles [8], was first implemented at DIII-D [9] and has become a widely used
method due to its relatively good spatial and temporal resolution. It is now used as well in
TEXTOR [10], LHD [11] and NSTX [12] and has yielded results on the fast-ion distribution
due to micro-turbulence [13], Alfven waves [14] and different injection geometries [15].
The FIDA technique is charge-exchange recombination spectroscopy (CXRS) [16])
applied to fast D-ions. Through charge-exchange reactions with neutrals, the fast ions
can capture an electron into an excited quantum n-level, become neutral themselves, and
subsequently emit line radiation. The radiation is typically emitted before the neutralized
fast ions (fast neutrals) move more than a few centimeters as the de-excitation back to the
equilibrium happens very quickly. Several spectral lines from the fast neutrals exist in
fusion plasmas; however, FIDA measurements tend to use the Balmer alpha emission line
(λ
0
= 656.1 nm, n = 3ton = 2). Balmer Alpha radiation permits the use of standard lenses
and spectrometers as it is in the visible range. The FIDA emission is localized mainly along the
path of the NBI as there is no significant overlap of fast ions and neutrals elsewhere. Hence,
multiple lines of sight (LOS) that intersect neutral beams at different radii can be used to obtain
profile information.
Information on the velocity distribution of fast ions is contained in the spectral shape of
FIDA radiation. The shape is dominated by the Doppler effect which depends on the velocity
vector of the fast neutrals, v, or, in other words, on the fast neutrals energy parallel to the
viewing direction, E
:
λ
Doppler
= λ
0
·
cos(α) ·|v|
c
∝
E
. (1)
Here, c is the speed of light, λ
0
is the unshifted wavelength and α is the angle between v
and a given LOS. The Doppler effect shifts the D-alpha emission considerably in wavelength.
The spectral shape of the FIDA emission has consequently broad spectral wings. It deviates
significantly from a Gaussian curve because fast ions do not follow a Maxwell distribution.
It should be noted that the nature of the Doppler effect makes an energy and pitch (v
/v
tot
,
i.e. the projection of v on the magnetic field lines) resolved measurement virtually impossible.
That is to say, one cannot relate a specific shift in wavelength to a unique energy as fast ions
with a higher energy and larger angles α can result in similar shifts.
Information on fast-ion density profiles is contained in the measured spectral radiances.
The FIDA radiance is proportional to the density of fast neutrals in the n = 3 level present
along a LOS. This density depends not only on the fast-ion density, but also on the probability
for charge exchange and on the mechanisms that populate the n = 3 level. The probability for
a fast ion to undergo charge exchange in which the electron ends up in the n-level, i,is
prob
→i
=
j
k
n
NBI(j,k)
· σ
cx(j→i)
(v
rel(k)
) · v
rel(k)
(2)
where n
NBI(j,k)
is the density of injected and halo neutrals present along the NBI with different
n-levels, j , and energies, k, v
rel(k)
is the relative velocity between fast ions and neutrals and
σ
cx(j→i)
is the energy-dependent charge-exchange cross section. All of the various neutral
beam energy components and n-levels as well as the halo population have to be accounted for
separately because the charge-exchange cross-sections depend on n-levels and energies. The
halo neutrals originate from charge-exchange reactions between the injected beam neutrals and
the thermal D-ions (the bulk plasma ions). Due to these reactions as well as subsequent charge-
exchange reactions of the halo neutrals with thermal D-ions, a cloud of neutrals surrounding
the NBI is built whose density is typically comparable to or even larger than the beam neutral
2
Plasma Phys. Control. Fusion 53 (2011) 065010 B Geiger et al
density. The NBI halo density is important for FIDA measurements because the high collision
energy between the fast ions and the halo neutrals provides a significant probability for charge-
exchange reactions.
A considerable fraction of fast neutrals are in the n = 3 level directly after charge exchange.
Additionally, the n = 3 level can be populated by the radiative decay from higher n-levels that
have been populated by charge exchange or other mechanisms. Furthermore, fast neutrals in
lower n-levels can be excited into the n = 3 level by collisions along their path through the
plasma. This effect yields non-localized FIDA radiation but, fortunately, does not strongly
limit the spatial resolution of the technique. The FIDA radiation emitted in the equilibrium
(through collisions) is comparably small. The radiation which is emitted directly after charge
exchange is more intense as the charge-exchange reaction strongly overpopulates the excited
n-levels.
The interpretation of FIDA spectra is difficult. In particular, the moderate resolution
in velocity space (one cannot resolve information on the energy and pitch of fast ions
independently) and the energy-dependent charge exchange cross sections make a direct
de-convolution of FIDA measurements to fast-ion distribution functions virtually impossible.
However, two different methods can be applied when analyzing FIDA measurements to
circumvent this difficulty: first, a forward model can be used to quantitatively interpret
FIDA measurements. By simulating the neutralization and photon emission of fast ions that
correspond to a given distribution function it is possible to calculate synthetic FIDA spectra.
These can then be compared with the measurement. Second, after accounting for changes in
the density profiles of injected and halo neutrals, changes in the FIDA data can be interpreted
as variations in the fast-ion distribution function. Hence, relative changes of the fast-ion
distribution function can be studied, e.g. in the presence of MHD activity.
This paper is structured as follows: section 2 is dedicated to the newly commissioned
toroidal FIDA diagnostic. The diagnostic’s setup is presented and the observed FIDA spectra
and radial intensity profiles are discussed. Furthermore, the diagnostic’s velocity space
weighting function is discussed. The subsequent sections are devoted to experimental results.
Section 3 discusses measurements of radial FIDA intensity profiles during on- and off-axis
NBI. Section 4 gives results and a short introduction to the FIDASIM [17] code. The code
enables the comparison between the measurements and the simulations, which use as inputs
the fast-ion distribution functions provided by TRANSP [18]. Section 5 discusses the fast-
ion redistribution due to a q = 2 sawtooth-like crash that ejects from the plasma center
approximately 50% of the fast-ion population. Finally, conclusions are given in section 6.
2. FIDA diagnostic setup at ASDEX Upgrade
Figure 1 shows an overview of the FIDA diagnostic’s LOS and of the NBI heating system used
at ASDEX Upgrade (AUG). Two NBI injectors, with four sources each, can heat discharges
with up to 20 MW of power. Box 1 (NBI 1-4) injects neutrals with a maximum energy of 60 keV
and box 2 (NBI 5-8) with 93 keV. The FIDA diagnostic’s 25 toroidal (roughly tangential to the
magnetic field lines) LOS are focused close to the midplane on NBI 3. Fibers with a diameter
of 400 µm guide the collected light outside the torus hall to a 0.28 m Czerny Turner-like
spectrometer that uses lenses instead of mirrors. After passing though a tunable entrance slit,
the light is dispersed by a 2400 lines mm
−1
grating and then focused on an EM-CCD camera.
The camera has a 16 bit dynamic range, an array of 512 × 512 16 µm pixels, and images a
spectral range of about 9 nm. Typically, the camera is operated in frame transfer mode with
10 ms exposure time and the entrance slit is opened to 200 µm yielding a rectangular-shaped
instrument function with a spectral width of about 0.2 nm on the CCD.
3
Plasma Phys. Control. Fusion 53 (2011) 065010 B Geiger et al
NBI 8
NBI 3
NBI 6
B
t
I
p
A1
A13
FIDA LOS
FIDA LOS
NBI 3
NBI 8
NBI 6
Figure 1. Toroidal and poloidal cross section of the AUG tokamak. The LOS of the FIDA
diagnostic are focused on neutral beam source 3 (60 kV). A1 and A13 indicate the positions of two
gas valves that, when turned on, influence the FIDA measurements with polluting line radiation.
652 654 656 658 660 662 664
Wavelength [nm]
Bremsstrahlung
FIDA
Beam emission
Halo
edge D-Alpha
Impurities
Intenstiy [a.u.]
Bremsstrahlung
Figure 2. Schematic illustration of active and passive contributions present in FIDA spectra when
using a toroidal diagnostic setup. The active contributions are the FIDA emission, the beam
emission and the halo emission. Passive contributions are bremsstrahlung radiation, the edge
D-alpha emission and line radiation from impurity ions.
The geometry of the NBI system and of the LOS, together with the nature of the FIDA
radiation, provide a radial resolution of the FIDA diagnostic of about ±3.5 cm (see the
appendix). Furthermore, the setup is such that most of the FIDA radiation is observed in
the red-shifted wavelength range (up to 662.6 nm for E
= 93 keV). This enables FIDA
measurements with good signal to noise ratios because a significant part of this spectral range
(from 659.5 nm and above) does not contain superimposed bright spectral contributions.
As illustrated in figure 2, the bright active contributions (present only with NBI 3) to FIDA
spectra are the halo and beam emission. The halo emission is D-alpha radiation that is emitted
from NBI halo neutrals. Its spectral shape can be approximated with a Gaussian curve since
the halo neutrals are thermally distributed. The spectral position of the halo emission depends
4