Transient impurity events in JET with the new
ITER-like wall
M.Sertoli
1∗
, J.C. Flannegan
2
, A. Cackett
2
, E. Hodille
3
, P.
de Vries
4
, I.H. Coffey
5
, B. Sieglin
1
, S. Marsen
6
, S.
Brezinsek
7
, G.F. Matthews
2
, J.W. Coenen
7
and
JET-EFDA Contributors
#
JET-EFDA Culham Science Centre, Abington, OX14 3DB, UK
1
Max-Planck-Institut fr Plasmaphysik, EURATOM Association, 85748
Garching, Germany
2
Euratom/CCFE Fusion Association, Culham Science Centre, Abingdon, OX14
3DB, UK
3
Universite Claude Bernard Lyon I, 69622 Villeurbanne, France
4
FOM institute DIFFER, Association EURATOM-FOM, P.O. Box 1207,
3440BE, Nieuwegein, Netherlands
5
Queen’s University, Belfast, BT7 1NN, UK
6
Max-Planck-Institut fr Plasmaphysik, EURATOM Association, 7491
Greifswald, Germany
7
Forschungszentrum Juelich GmbH, EURATOM Association, 52425 Juelich,
Germany
#
See the Appendix of F. Romanelli et al., Proceedings of the 24th IAEA
Fusion Energy Conference 2012, San Diego, US
E-mail: marco.sertoli@ipp.mpg.de
Abstract. Transient impurity events leading to unexpected increase in radiated
power have been studied in JET from the installation of the ITER-like wall. A
total of 1800 events over 2800 plasma discharges have been detected. None have
led to permanent changes in plasma conditions. Of all events 60% show traces of
W, 25% of either Ni, Fe or Cr from either Inconel or steel structures. They occur
mainly in diverted magnetic configuration, independently of strike-point position.
The effect of disruptions on dust redistribution has been investigated using the
Thomson scattering diagnostic and correlated with TIE occurrence. The number
of dust events detected increases with disruption force and, in comparison to
the full-C wall, the amount of dust mobilized is found to be about an order of
magnitude lower. Their time evolution correlates well with that of the transient
impurity events.
PACS numbers:
Keywords: tokamaks, dust, impurity, disruption, VUV spectroscopy
1. Introduction
The understanding of dust production and control in future tokamaks such as ITER
is a major concern not only for safety reasons and for diagnostic interpretation (see
e.g. [1, 2, 3]), but also for reliable plasma operation. Depending on their composition
and size, if such particles reach the plasma core they can perturb it with intense
radiation spikes which have been seen to lead to MARFE (stranding for Multifaceted
Transient impurity events in JET with the new ITER-like wall 2
Asymmetric Radiation From the Edge) instabilities or disruptions, especially in long
pulse operation [4].
This contribution reports on the study of the occurrence of such transient impurity
events (TIEs, also called UFOs) in JET from the installation of the new ITER-Like
wall [5]. Such events are visible as sharp increases in the total radiated power similar to
what caused by laser ablation experiments. All plasma phases with I
p
> 1 (MA) from
pulse number JPN 81000 have been considered. The impurities responsible have been
studied using vacuum-ultra-violet (VUV) spectroscopy. Correlations with auxiliary
heating power, plasma geometry (limiter or diverted magnetic configurations, as well
as positions of the strike-points) external causes such as arcs from the lower-hybrid
antennas or reciprocating probe plunges and disruptions have been investigated. This
contribution reports on the time evolution of these events during the first campaigns
with the ILW, on the correlation with plasma geometry and on the elements responsible
(section 2). Moreover, studies of the dust mobilized by disruptions examined using
the high resolution Thomson scattering (HRTS) diagnostic and their correlations with
TIEs are reported (section 3).
2. Transient impurity event analysis
The occurrence of transient impurity events has been analysed in all phases of plasma
discharges with current I
p
> 1 M A. A total of 2792 discharges have been analysed
for a total of 1781 events detected starting from JPN 81000. The events have been
selected by searching for a sharp increase in the total radiated power similar to what
commonly observed in impurity injection experiments. None of these events have
been seen to lead to permanent changes in plasma conditions (e.g. permanent loss in
confinement). TIE-induced disruptions are not reported in this contribution since the
analysis of cause-and-effect requires more in-depth investigations than what can be
reported in here.
For each event, the values and variations of the ohmic and auxiliary heating power,
plasma current, plasma geometry (high- or low-field-side limiters, strike point positions
for diverted plasmas), excursion in radiated power and total injected deuterium rate
have been tabulated. In order to correctly normalize the event distributions, a
database containing the total plasma operation time (in steps of 10 ms) for each
geometrical configuration and each correlation quantity listed above has been created.
A new database containing the total plasma operation time (in steps of 10 ms)
for each geometrical configuration and each correlation quantity listed above provides
the normalization parameters for the event distributions. The probability of TIE
occurrence is therefore given in # of events / operation time, i.e. as a frequency in
units of Hz. Its error bars are evaluated through propagation of the statistical error
of the number of events, equal to its square root.
2.1. Evolution during the campaign
The evolution in time of the probability of TIE-occurrence is shown in figure 1. The
discharges have been binned according to pulse number (bin size = 200), but each bin
is independently normalized to the total plasma time in that bin. The distribution
shows two main jumps in probability at JP N ∼ 81501 and JP N ∼ 83301, sitting on
an almost constant background value ∼ 0.03 Hz, i.e. 3 events every 100 seconds of
plasma operation.
Transient impurity events in JET with the new ITER-like wall 3
AUX
Figure 1. Time evolution of the TIE probability (black with white error
bars) and of the maximum heating power (M ax P
AU X
in green) and total
injected energy (
R
P
AU X
dt in orange).
The first increase (JP N = 81501 ± 100) in the campaign sets in when additional
power by neutral-beam injection (NBI) was available and routinely used in the
discharges. In the phases before, almost pure ohmic or ion cyclotron resonance
frequency (ICRF) heated discharges had been performed. The maximum heating
power (in green in figure 1) undergoes a sharp increase from < 5 to ∼ 13 M W , more
than 80% of which is due to NBI. This suggests a correlation with input power with
threshold at > 4 M W which is the typical power threshold for type I ELMy (edge
localized modes) H-mode. Since most TIEs occur in diverted configuration (see section
2.2), this is most probably a correlation with power delivered to the divertor. This
statement requires in-depth analysis and cross-correlation with other quantities as well
as space for sufficient discussion, so it will be left as a speculation in this contribution
and discussed elsewhere.
The second increase in event probability (JP N = 83301 ± 100) occurs instead
when the total injected energy is increased almost of a factor 3. Since the normalization
of the number of events has been performed over total plasma-time, this increase is
probably caused by simple cumulative effects due to higher fraction of total time
operated at high input power. More in this regard is discussed in section 3.
2.2. Geometrical dependencies and impurity identification
The probability of TIEs occurring in limited plasmas is very low (≤ 10 mHz) and
only ∼ 7% of all events occur in this configuration. Most are generated in divertor
configuration, the probability increasing to ∼ 40 − 50 mHz, almost independently
of strike point configuration (figure 2). Configurations with outer strike point on
stack A of tile 5 account for only 10 events so the probability is in this case subject
to large errors (black circles on coloured blobs whose size is proportional to the TIE-
probability). Similarly, configurations with outer strike point on tile 7 account instead
for only 6 events, while the outer strike point has never been positioned on tile 8, so
Transient impurity events in JET with the new ITER-like wall 4
T1
T3
T4
T5
A
B
C
D
T6
T7
T8
Figure 2. TIE probability (in mHz) depending on strike point position
for the inner (blue) and for the outer (red) strike points. The size of the
coloured blobs are proportional to the TIE-probability written close to
them with the same colour. Tile names are labeled in black (T #), stacks
of tile 5 with letters.
no statistics is available.
Configuration with [inner, outer] strike points on tiles [3, 5] has been the most
commonly used with the ILW, followed by [4, 6], [1, 5], [3, 7], [3, 6] and [4, 5] used for
respectively 25%, 17%, 15%, 7%, 4% of the operation time spent in configuration
[3, 5]. Since most high power scenarios were performed in configuration [3, 5], its
∼ 30% higher probability with respect to [4, 6] is most probably due the power
dependence of TIEs. Due to the fact that the normalization has been performed
over total plasma-time in each specific configuration, the TIE-probability shown in
these plots will increase with the fraction of total time spent at higher heating powers.
Configurations with the outer strike point on vertical tile 7 seem to be less prone
to TIE-occurrence than those with the outer strike point on the horizontal tiles 5
or 6 (1.9 mHz vs. 40 − 50 mHz). Since the horizontal part of the divertor is a
deposition zone for material coming from the main chamber, dust which accumulates
in this region could partly be the cause of the occurrence of TIEs. On the other
hand, these results are biased by the low externally injected heating power and energy
in this configuration, never exceeding ∼ 5 M W and ∼ 0.05 GJ respectively, versus
∼ 25 M W and ∼ 12 GJ delivered in configuration [3, 5]. More study is therefore
required to better understand these issues.
For each event, the elements responsible have been analysed using two VUV
spectrometers, measuring respectively in range [10, 100] nm and close to 5 nm. The
first one measures spectral lines of Ni, Fe, Cr, Al and Cu, and can provide an indication
of tungsten. The presence of this element has been further cross-checked with the
second spectrometer which measures the quasi-continuum spectral features of tungsten
ions W
27+
− W
35+
[6]. This spectrometer is not subject to spectral contamination
from other elements within the probed spectral range, so provides a trustworthy
measurement of the presence W. An attempt has also been done to detect Be or
C through visible spectroscopy [7].
Since both stainless steel (material of the support structures) and Inconel (support
structure of the divertor W mono-block tiles) are made up of Ni, Fe and Cr (although
Transient impurity events in JET with the new ITER-like wall 5
in different percentages), dust particles from these parent structures could lead to
events of either of these elements. The elements seen have thus been sub-divided into
three groups: W - Ni/Fe/Cr - Others. The latter group accounts for all those events
for which neither of the first four elements have been detected. These events could
be due to light impurities not distinguishable through VUV or visible spectroscopy
(e.g. Be or C) or to the elements listed above but in conditions unfavorable for their
detection by spectroscopic means (e.g. too low photon flux for detection through
the VUV spectrometers). Data from the visible divertor spectroscopy, sensitive to C
and Be, show only few of the events present in the database. Identification of the
composition and origin of the source have so far been unsuccessful [7].
All divertor configurations show approximately the same element dependency:
∼ 60% of all events show traces of W , ∼ 20% of either Ni, F e or Cr and ∼ 20%
are of unknown origin. About 90% of the W events do not show any traces of Ni, Fe
or Cr. Limiter TIEs show instead a lower probability of W (∼ 50%), and increase of
Ni/Fe/Cr (∼ 35%) and of unknown elements (∼ 25%). In limited plasmas, W could
originate from W-coated neutral beam shine-through protection tiles as well as from
few outer poloidal limiters, recessed inner limiter tiles and restraint ring protection
tiles. At the moment, there is no clear indication of which of these sources could be
the major contributor to W limiter-TIEs.
Reciprocating probe (RCP) plunges have shown to be the cause of 15 events
(< 1%), while arcs at the lower hybrid (LH) antennas are responsible for 7 events
(< 0.5%). All LH-arc events and 80% of the RCP ones show traces of F e, consistent
with the material of the LH launcher (stainless steel) and an indication that material
from the probe shaft (Inconel and steel) could have been scratched away or eroded
during the plunges.
3. Disruptions and HRTS dust analysis
Since dust could be one of the possible causes of TIEs, it is important to try and
correlate the results obtained in the present analysis with studies performed with other
JET-C
JET-ILW x 30
Figure 3. HRTS number of dust events vs. disruption force, normalized
to the number of disruptions in each bin for the JET-C wall (blue) and for
the JET-ILW (red), the latter multiplied by a factor 30.
