Transcript 2D MSE imaging on the KSTAR tokamak and future prospects

2D MSE imaging on the KSTAR tokamak
and future prospects
John Howard
J Chung, O Ford, R Wolf, J Svennson, R Konig
1
Outline
• Measurement principles
– Doppler imaging on DIII-D
• Optical system and calibration
• KSTAR measurements
• Modeling results (using full QM
treatment)
An alternative approach to spectroscopy
a simple polarization interferometer gives contrast and phase at a single optical delay
Simple polarization interferometer:
Input
Waveplate at 45 degrees (delay f=2pLB/l)
Interferogram
S = I(1+z cosf)
To recover the fringe properties, measurements are required at
Polarizers
multiple interferometric delays
Interferogram
Spectral
Lines
Fourier transform
3
Spatial heterodyne interferometer
Savart plate introduces angular phase shear
 generates straight parallel fringes
imprinted on image.
Demodulate for local fringe brightness,
contrast and phase.
Phase shift tracks Doppler colour changes
 flow fields
DIII-D Divertor raw image
Tomographically inverted DIII-D divertor
brightness and flow images
Demodulated brightness (top) and
phase (bottom) projections at
representative times during the
divertor detachment for DIII-D
discharge #141170: (a) 500 ms, (b)
2000 ms and (c) 4000 ms.
Corresponding tomographic
inversions of brightness (top) and
phase (bottom)
Reasonable agreement with
UEDGE modeling
With Diallo, Allen, Ellis, Porter, Meyer, Fenstermacher, Brooks, Boivin
5
Motional Stark effect polarimetry senses the
internal magnetic field
Top view KSTAR MSE viewing geometry
Beam
A typical Doppler shifted Stark effect spectrum
Edge
Centre
View range
Edge
Modelled interferometric image of beam
Centre
Courtesy, Oliver Ford, IPP
Motional Stark effect (MSE) polarimetry measures the polarization orientation of Stark-split Da
656 nm emission from an injected neutral heating beam.
The splitting and polarization is produced by the induced E-field (E = v x B ) in the reference
frame of the injected neutral atom. MSE can deliver information about the internal magnetic
field inside a current-carrying plasma
Angle-varying Doppler shift  every observation position requires its own colour filter.
Interferometric approach – periodic filter allows 2-D spatial imaging  Bz(r,z)
6
Imaging spectro-polarimetry for MSE
Recall simple polarization interferometer:
Quarter waveplate
Input
Waveplate (delay f)
Output signal
S = I(1+z cosf)
Polarizers
If input is polarized already (angle y), remove the first polarizer
Resulting interferogram fringe contrast depends on polarization orientation:
S = I(1+z cos2y cosf)
Add a quarter wave plate. Fringe phase depends on polarization orientation:
S = I[1+z cos(f + 2y)]
The p and s components interfere constructively (no need to spectrally isolate)
7
How do we image the multiplet?
For one of the multiplet components (e.g. p), the interferometer output is:
Sp = Ip [1+zp cos(fp+2y)]
For the orthogonal component (y+p/2, slightly different wavelength), it is
Ss = Is [1-zs cos(fs+2y)]
For MSE triplet, after adding the interferograms, the effective signal contrast
depends on the component contrast difference zp – zs. Choose optical delay t to
maximize the contrast difference zp – zs
Model of KSTAR isolated full energy Stark
multiplet and associated nett contrast
Edge
s
Edge
p
Centre
Centre
Good contrast (~80%) across full field of view (i.e. Stark splitting doesn’t
change significantly). But significant phase variation due to large Doppler shift
Optical delay 1000 waves  a-BBO plate thickness ~5 mm
2nm bandpass filter tilted to track Doppler shift across FOV
KSTAR parameters: Bt = 2.0T on axis, Ip = 600 kA
D beam, 85keV/amu, 1.0 degrees divergence
9
Final imaging MSE instrument
A first quarter wave plate and shearing Savart plates results in a phase encoded
double spatial-heterodyne system for y and amplitude encoding for e
S = I0 [1 + z cos(kxx + f+2y) + z cos(kyy - f+2y) ]
Instrument produces orthogonal phase modulated spatial carriers
Demodulate fringe pattern to obtain Doppler shift f and polarization y
Power spectrum of interference pattern
Calibration image using Neon lamp at 660nm
All information is encoded on distinct spatial heterodyne carriers:
Polarimetric angles: (y, e) (orientation and ellipticity)
Interferometer contrast and phase: (z, f) (splitting and Doppler shift)
Power spectrum of interference pattern
-f+2y
f+2y
2e
All information is encoded on distinct spatial heterodyne carriers:
Polarimetric angles: (y, e) (orientation and ellipticity)
Interferometer contrast and phase: (z, f) (splitting and Doppler shift)
Typical calibration data
(a) Central horizontal slices across a sequence of demodulated polarization angle images y. The
Doppler phase image f is insensitive to the calibration polarizer angle.
(a) Deviation from linearity of the measured polarization angle at the centre of the calibration image
versus polarizer angle. Cell size for averaging is ~1.5-2 carrier wavelengths (10-14 pixels). There is
a small systematic variation. Random noise ~0.1 degrees (calibration image).
Optical system layout
Mirror
Telescope
Camera
Cell
Filter
From plasma
Typical MSE double heterodyne image
Plasma
Boundary/
port
opening
Radiation noise
Orthogonal spatial carriers
Pixelfly
1300x1000
100ms exp
Frame rate
10Hz
Internal reflection
and sparks
(not an issue for
imaging MSE)
Beam direction
Day 1
Typical MSE double heterodyne image
Pixelfly
1300x1000
100ms exp
Frame rate
10Hz
Day 2
Typical MSE double heterodyne image
Pixelfly
1300x1000
100ms exp
Frame rate
10Hz
Day 3
Typical MSE double heterodyne image
This is our beam-into-gas
calibration image
Pixelfly
1300x1000
100ms exp
Frame rate
10Hz
Conclusion: Need new camera
Solution: CID camera + remote + shield
Measured and modelled Doppler phase images are in
good agreement
Measurement
Centre
Model
Edge
Line-of sight integration effects may account for the small
discrepancies.
Viewing from above mid-plane accounts for tilt of phase contours
System tolerant of large beam energy changes (70-90 keV)
Measured and model “nett polarization” images
Low brightness
regions
Simple circular plasma model 2.0T, 600kA
Reflection artifact
A typical measured nett polarization image
- 2.0T, 600kA
Note: A fixed constant shift of 16 degrees
has been subtracted
Window Faraday effect? thermal drift?
misalignment?
Nett polarization angle = plasma MSE angle - Gas MSE reference angle
Typical KSTAR midplane radial profile evolution
during RMP ELM suppression xpts
System should be self calibrating – edge polarization
angle is determined by toroidal current and PF coils –
other angles are referred to the edge.
(alleviates issues with thermal drifts, window Faraday
rotation, in situ calibration problems etc.)
Ramp up
Edge
Axis
Axis
Common mode noise structures from
beam-into-gas calibration have been partly
removed
Edge
Tolerant of polarized background reflections
Reflections from
internal structures
have little effect
on inferred
polarization angle.
True for
broadband
emission,
polarized or
unpolarized
22
QM modeling of system polarization response
• Apply QM model developed by Yuh, Scott,
Hutchinson, Isler etal to estimate importance of
Zeeman effect on MSE nett polarization (three
directions with corresponding polarized components
E, v and B)
• No line of sight integration effects
• Statistical populations
• Uniform brightness beam (no CRM modeling)
• KSTAR viewing geometry
• Simple circular flux surfaces with Shafranov shift
• Spectro-polarimeter – sum over 36 Stark-Zeeman
component lines
23
Modeled E,v and B components
Polarization
orientation
Centre
Edge
Ellipticity
E (p)
V (s)
Nett ellipticity angles are comparable in magnitude to polarization tilt
B
(5% of
intensity)
Comparison with ideal Stark effect model
Stark-Zeeman
Difference orientation angle
Geometric model (no Zeeman)
Difference orientation angle
variation across MSE image less
than ~0.1o
 Standard geometric models
for interpretation are OK
Plasma images show strong ellipticity
Orthogonal carriers
(mostly linear)
Elliptically
polarized
light
Beam direction
Typical raw image of beam emission
26
Ellipticity images
Beam emission images show larger than expected ellipticity
Beam into gas
Beam into plasma
85 keV
80 keV
Beam into gas
Ellipticity unlike QM
model. Window linear
birefringence?
Dependence on beam
energy indicates other
than some B field
dependent optical
effect  ZeemanStark coupling
Attributes of MSE imaging approach
 Analyse full multiplet so no need for multiple discrete narrowband filters
 Simple inexpensive instrument  ~103 more channels at <10% total cost
 No filter tuning issues or incidence angle sensitivities
 Tolerant of beam energy changes (10-20%), other beam energy components,
overlapping beams.
 Multiple heterodyne options (spatial/temporal), single channel or imaging
 Insensitive to “broadband” polarized background contamination
 Insensitive to non-statistical populations
 Full Stokes polarimetry  Possibility of self calibration based on unpolarized
plasma radiation (Voslamber 1995) mirror/window degradation
 Can be applied to spectrally complex elliptically polarized multiplets (Zeeman
effect)
 New opportunities. For example: 2D toroidal current imaging (in principle) and
possibility of synchronous imaging of sawteeth, MHD, ELMs, Er etc.
28
Future directions/possibilities
• Replace crystal quartz window to eliminate Faraday rotation
and anomalous ellipticity and install radiation hard camera
for next KSTAR campaign.
• Fast system for real-time equilibrium
• Time-multiplex system for high spatial resolution imaging
(RMP effects)
• Use gated intensified camera to synchronously study
magnetic reconnection for comparison with ECE imaging.
• MSE/Zeeman imaging at ASDEX and DIII-D (for pedestal
and ELM suppression studies)
Conclusion
 Modeling indicates that imaging MSE should be a
reliable tool for obtaining 2d maps of the internal
magnetic field in tokamaks.
 IMSE significantly increases the information available to
infer the current profile.
30