Plasma diagnostics

in basic plasma physics devices and tokamaks:

from principles to practice

Fast imaging techniques

Ivo Furno
in collaboration with

H. Reimerdes and B. Labit

with contributions from

R.A. Pitts, B.P. Duval

Version – 03.02.2012
 Content of the unit
•  Radiation emission from plasmas
•  Detecting radiation from plasmas
§  Bolometers
§  Photomultipliers
§  Photodiodes
§  Xray and bolometric
§  Fast cameras
§  Fast framing cameras
§  Fast gated and intensified cameras
§  Infra-red (IR) cameras
 Theory of plasma emission
Cyclotron radiation Bremsstrahlung
It originates from the gyration Free-free radiation arises from
of charged particles around electron-ion collisions, which
the magnetic field. Main consist of small-angle
contribution is from the scattering of electrons in the
electrons ωe=eB/me. (GHz). ion electric field

Line radiation
A discrete spectrum arises
from the transition of an
Recombination
electron from a bound state A continuum spectrum arises
(Ep) to another bound state from electron–ion collisions in
(Eq) with the emission of a which electrons are captured
photon hν= Ep- Eq. in a bound state.
 Bremsstrahlung
In fusion plasmas, Bremsstrahlung mainly arises from e-i collisions. The
contribution from e-e collisions is significant for energies larger than 0.5 MeV.

The power spectral density of the radiated emission can be expressed as:

Maxwell-averaged Gaunt factor

In the case the average photon energy is ~ the

electron temperature.
 Recombination radiation
Another important contribution to the continuum spectrum arises from e-i
collisions in which the final electron state is bound.
Dielectronic recombination
Radiative recombination

Electron final bound energy Conservation of energy

No recombination can occur

Contribute to recombination radiation at hν

 Recombination radiation
For a Maxwellian electron distribution, the power spectral density emitted
during the free-bound transition to an n state is

recombination
=
Bremsstrahlung

Negligible recombination

recombination dominates
all states can contribute
Bremssstrahlung
dominates
 Plasma effective charge - Zeff
In any practical cases, many different ion species j with charge Zj contaminate the
plasma. With Te[eV] and ne[m-3], the total continuum emission is then:

When recombination becomes negligible, i.e. when gff,jZ2 >> fj then the total
continuum emission can be expressed as, defining
Line radiation – an example from TCV

Ep-Eq
Ep

Eq
 Broadband detection: bolometry
T dependent resistive coating
Photons (2eV-10keV)
+ neutrals
T0

Low heat capacity substrate Electrical leads

⎛ dT T − T0 ⎞
P = C⎜ + ⎟
Incident power ⎝ dt τ ⎠
Heat capacity
Au-kapton-Au ~0.4mJ/K cooling time constant
Stainless steel-Mg02 -Ge ~2mJ/K depends on exchange
with surroundings
τ~10ms – 2s
Bolometry: measurements technique

R = R 0 [1 + α (T − T0 )]
Detector resistance
Au: 2.6x10-3 K-1

Blind detectors

U Wheastone bridge Differential amplifier

 Foil bolometric tomography on TCV

The TCV foil bolometer system comprises 5 cameras with a total of 64 lines of
sight. The detectors are Au-foils.
 Photomultiplier tubes

TCV Mach probe head

Photocathode materials: Pros:

§ Excellent time response
Ag-O-Cs: 300-1200 nm § High gain
GaAs:Cs: 300-850 nm § High sensitivity
InGaAs:Cs: higher infrared
sensitivity than GaAs:Cs Cons:
§ No 2D array
§ High voltage required
Photodiodes and photodiode arrays

p+ dead layer
0-1µm depletion zone
n-type bulk Si

Photon Eph
Lp= (Dpτ)1/2 ~ 100-1000µm
electron/hole pair Epn=3.63eV

passivation layer
Si3N4, SiO2
50nm
V ~ µm D ~ 100µm - mm
 Photodiode efficiency
p+ dead layer depletion zone
n-type bulk Si

Photon Eph
Lp=(Dpτ)1/2
electron/hole pair Epn=3.63eV

passivation layer
V D
TCV Mach probe head

probability of being detected

Si absorption coefficient
is a function of energy
 Photodiode efficiency

CENTRONIC LD20-5T:
passivation layer 55nm
dead layer 0.8µm
diffusion length Lp=200µm

Increasing D, Lp increases η at
high energy
Reducing passivation / dead layer
increases η at low energy

Solid state bolometer – AXUV

Tomography using photodiodes: bolometry

The TCV
AXUV
bolometric
system
comprises 7
cameras for a
total of 140
lines of sight
 Tomography using photodiodes: soft X-ray

The TCV tomographic system has 10 cameras with a total of 200 lines of sight.
 Charged coupled devices - CCD
Photon Eph
gate
Insulating layer
Epitaxial layer e- accumulated charge
Si substrate

Sensitivity in visible and IR and recently even

in soft X-ray range
TCV Mach probe head
Excellent quantum efficiency (50%)

Dark current (thermal noise) can be removed

by cooling (needed for IR detection)

Single or multiple 8-12 bits A/D converters

1D and 2D array of 106 - 10µm pixels

 Charged coupled devices - CCD
Photon Eph
gate
Insulating layer
Epitaxial layer e- e-

Si substrate
parallel

A/D converter (CMOS)

serial

TCV Mach probe head

Fast read out possible on
consumer cameras
(CMOS -1kHz)

A/D converter (CCD)

 Fast framing cameras
Recent advances in commercial CCD/CMOS technology mean that routine 2D
imaging at extremely fast speeds in now possible even on “moderate” budgets.

http://www.visionresearch.com http://www.photron.com http://www.prinsci.com

TCV Mach probe head

TCV Mach probe head
Phantom V640 Photron Ultima APX-RS PSI-5 camera used to
1.4 kHz @2560x1600pix 3 kHz @1024x1024pix record 300 frames,
300 kHz @256x16pix 250 kHz @128x16pix 64 x 64 pix/frame
1+ µs electronic shutter 1+ µs electronic shutter Record: up to 1MHz
Photron APX-RS on TORPEX and TCV
FireWire, Gigabit Ethernet and Optical interfaces
available for maximum flexibility and fastest image
download and reliable camera control, regardless of
distance

Now available with 16GB memory option for longer

recording times such as 12 seconds at 1,000 full
resolution frames per second

TCV Mach probe head Full 1024 x 1024-pixels resolution up to 3,000 fps
10,000 fps at 512 x 512-pixels

250,000 fps at reduced resolution – an industry

record

Global electronic shutter to 2µs

User selectable ‘Region Of Interest’ function

 Typical frequencies in tokamaks

drift turbulence EC frequenc

blob-filament >10GHz
energy confinement propagation
equilibrium of time τE
wall particle TCV: 200Hz , ITER~0.25Hz ~100kHz
content
0.01 Hz IC frequency
plasma rotation >10MHz
current diffusion time frequency ~1kHz
MHD instabilities τI= (µ0/η)L2
0 - ~1MHz ~1Hz for TCV
Fast framing cameras: pros and cons

Pros Cons
Plug-and-play Still expensive (~100kCHF)
Passive : non perturbative - Careful with optics (aberrations,
invasive etc…)

High temporal resolution Sensitive to magnetic fields,

especially if intensified
High spatial resolution
Non-local measurements requires
Statistical analysis, Fourier, CS,
inversion or field line tracking
etc… in time and space
Difficult to extract direct plasma
quantities (ne, Te, etc…)
They may require gas puffing
systems
 Gas puffing imaging systems
•  Fast cameras used to view visible emission
with gas puffing imaging
•  Gas puff is injected to increase image
contrast and brightness. Gas puff does
not perturb local (nor global) plasma.
•  Emission filtered for Dα (or He I) light
from gas puff:
I ∝ none f(ne,Te)
•  Dα emission only seen in range
~ 5 eV < Te < 50 eV
•  View aligned along B field line to see 2-D
structure ⊥ B. Typical edge phenomena
has a long parallel wavelength, filament
structure.
•  For more details: “Gas puff imaging of
edge turbulence” R.J. Maqueda et al.,
Rev. Sci. Instrum. (2003)
Imaging of blobs and filaments on NSTX
This field was pioneered by S. Zweben, R. Maqueda, J. Terry on C-Mod and NSTX

TCV Mach probe head

S.J. Zweben et al, Nucl. Fusion 44,134 (2004)

R.J. Maqueda, Rev. Sci. Inst. 74, 2020 (2003)
 Filaments during ELMs

Extracting meaningful
quantities is challenging:

TCV Mach probe head •  filament speed, size

•  birth place
TCV Mach probe head
requires precise field line
mapping

S. Lisgo et al., PSI17, China

 Comparisons with codes are just starting
GEMR
• ”Delta-f" formulation of electromagnetic gyrofluid
equations
• Six variables (n, vII, TII, T⊥, qII, q⊥) for ions and
electrons
• Fields ϕ and AII solved at each time step
• I ∝ none α Te β
http://w3.pppl.gov/degas2/

• Several turbulence quantities are reproduced by

GEMR
- correlation lengths to well within a factor of x2
- timescales to within a factor of roughly x2
- velocities to within a factor of roughly x2
• But relative fluctuation levels x5-10 too low in
code
• Code can not capture large amplitude ‘blob’
events
A lot of work needs to be done here!
Optical boundary reconstruction in MAST

t=125ms t=200ms t=348ms

1.5 1.5 1.5

1 1 1

0.5 0.5 0.5

Z [m]
0 0 0

-0.5 -0.5 -0.5

-1 -1 -1

-1.5 -1.5 -1.5

0.5 1 0.5 1 0.5 1
R [m] R [m] R [m]

EFIT and OFIT boundaries reconstruction

Courtesy of Gillis Hommen [TU/e & Differ (Dutch Institute for Fusion Energy)]
 Fast gated, intensified cameras
www.cookecorp.com Flux rope
Plasma gun

fast shutter down to 3 ns

excellent sensitivity of the system
allows
TCV Machsingle photon
probe head detection
12 bit dynamic range
high resolution MCP-image
intensifier & CCD (1280 x 1024pix)
exposure times from 3 ns – 1000 s
spectral sensitivity from UV to NIR

Only a few time points per shot

E. Hemsing and I. Furno, IEEE TRANS. PLASMA SCI., (2005)

 Phase
Phase resolved
resolved optical
optical emission
emission spectroscopy
spectroscopy (PROES)
Standard optic (C-mount) Fast gated camera: PI-Max Gen II
without optical filter

Specifications:

Fast gated camera: PI-Max Gen II

Programmable timing generator:

ST-133

CCD: 1024x1024, 16bit,

temperature: -20°C

Minimum gate speed: 2ns

Max repetition rate: 50kHz

(max frame rate <1Hz)

Optic (C-mount): Fujinon

TV-Zoom 1:1.8/18-108
 PROES: Experimental setup
RF

RF signal RF signal Frequency divisor

Voltage probe and
Plasma trigger generator

Trigger out
Fast gated Camera
Delay : Pi-Max
29 ns
Trigger + Camera controller
controlled delay ST - 133

2 ns
•  The camera is triggered on the RF voltage signal at a given phase.
•  Since the acquisition time is 2ns, more than 106 repetitions are necessary.
 Example: RF plasma at 13.56MHz

Electrons accelerated
in the electric fields
excite the neutral gas

Conditions:
•  hydrogen, 1mbar
•  Hα (656nm)
•  The lifetime of the
excited state
(including
quenching) is ~2ns

Phase
Example: RF plasma at 13.56MHz

1 3
The emission pattern are due to:
• Electron ejection from the RF (1)
and ground (2) sheaths

• Field reversal in the RF (3) and

ground (4) sheaths

2 Pattern 1-2 permit to estimate the

maximal sheath length
4 [U. Czarnetzki et al., Plasma Sources Sci.
Technol. 8, 1999 ]
 IR thermography (1)
Rapid technology advances in recent

LWIR
years have brought to the market
(from military development) new IR
cameras in both the MWIR range (3-5
µm) and the LWIR range (8-12 µm).
Large Focal Plane Arrays (FPA)
typically with > 70,000 pixels are now
available thanks to semiconductor
and MEMS (micro-electromechanical
MWIR

systems) technology. These are ideal

for IR thermography diagnostics in
tokamaks. This is “passive”
thermography using the natural IR
Some basics: Planck Law – spectral emission of heated surfaces.
distribution of emission from a black-
body: 2 Emissivity: ε = Mreal/MBB
2πhc ⎡ 1 ⎤
M(T, λ) = ⎢ ⎥ Ratio of radiation emitted from a body
λ5 ⎣ e (hc/kλT ) − 1⎦ to that from a Black Body à if ε(λ)
Wien displacement law: λmax = 2.9 x10-3/T known, T can be determined
Stefan-Boltzmann law: Ib = σT4 à total No real object is a true BB radiator
radiant emittance of a Black Body Grey bodies have ε < 1 but const in λ
σ = 5.7 x 10-8 Wm-2K-4 Real objects have ε < 1 varying with λ

 IR thermography (2)
MWIR: Photovoltaic quantum LWIR: micro-bolometer systems à
systems à direct conversion of indirect conversion of incident radiation
incident radiation into an electronic into an electronic signal à IR photons
signal (like in visible CCD camera) à change heat up a conductive layer on
electrons excited from valence to each detector and change resistance (or
conduction band even physical shape – cantilever)

Typical detector dimension: 30x30 µm

FPA size typically 320 x 240, 256 x 256
pixels but larger arrays now becoming
available (eg. 640 x 512)
 IR thermography (3)
M(λ,T) (photons cm-2µm-1)

Spectral response, RD
×

MWIR

LWIR
Wavelength (µm) Wavelength (µm)

M(λ,T)RD(λ) (photons cm-2µm-1)

Product of photon radiance and spectral
response gives total number of photons
detected for any given wavelength range à
MWIR systems generally do better at the
higher temperatures of interest to fusion
experiments.
But LWIR (microbolometers) can be of
interest for routine surveillance when speed
is less of an object, but cost is.

Wavelength (µm)
 IR thermography (4)
MWIR: Photovoltaic quantum systems: LWIR: microbolometer systems
PtSi, CdHgTe (CMT), InSb
InGaAs (SWIR) Advantages:
Room temperature – no cooling à
Advantages: Low cost (~50 kSFr), good portability
High frame rates possible (eg. 880 Hz Disadvantages:
on 256x256) Not snapshot – sequential readout
“Snapshot” mode Because of heat dissipation within
Low NEdT (> 12 mK) absorber material:
Low integration times (> 1 µs) High NEdT (> 120 mK)
Disadvantages: Cross-talk between neighbouring pixels
Cryogenic cooling required – down to Low time resolution (typically ~ 10 ms)
LN2 (77 K) à Stirling coolers with à of no practical use for observing
limited lifetime before servicing tokamak transients (ELMs, disruptions)
Expensive (typically ~ 200 kSFr)

Noise Equivalent Temperature
ΔT = Ttarget - TBackground Difference (NEdT):
NEdT = ΔT/(vsig/vn)
vsig
Minimum detectable T difference.
vn – rms
noise level
An NEdT = 50 mK means that a T
difference of 50 mK produces a
signal equal to the detector’s own
noise level
 IR thermography (5)

LWIR microbolometer systems now getting very

compact – new products on the market are
comparable in size with visible CCD cameras.

FPA is silicon based 388 x 248 pixels in 7 – 14 µm

range with NEdT < 50 mK (they claim)

Yet another military driven

development – this picture
right off the Thermoteknix
website ….
 IR thermography (6)
Materials for MWIR/LWIR - some examples Ge
Imaging first wall surfaces in research
tokamaks always requires a vacuum window
(though not possible in ITER!).
Ordinary glass does not transmit above 2.5 µm.
Sapphire (Al2O3): expensive, but can be bonded

Transmission (%)
to metals for vacuum seals – strong, requires
low thickness.
Al203
Germanium (Ge): Very popular IR material – low
+ flat transmission in MWIR/LWIR – expensive,
high n – requires anti-reflection coatings –
used mostly for lenses
Zinc Selenide (ZnSe): very popular for IR
vacuum windows and lenses – good
transmission over broad wavelength (used on ZnSe
TCV). Highish n – needs AR coatings. Oxidises
above ~250 C in atmosphere
Calcium fluoride (CaF2): popular for vacuum
windows, low n, low cost
Silicon (Si): becoming increasingly popular –
cheaper than Ge – will be used on TCV
Wavelength (µm)
 IR thermography (7)
IR systems now widely employed on almost all tokamaks, including TCV
Microbolometer
IR Camera in
magnetic
shielding

Ge relay
optics

Shutter

NEdT < 120 mK

7-12 µm
320x240 FPA
51x51 µm pixel
12 Bit, 50 Hz
See J. Marki et al., PSI 17 for more
 IR thermography (8)
Trend for wide-angle IR now being exploited also on MAST – good for disruptions

Also Santa Barbara Focal

Plane (SBFP 125)
FPA CMT
320x256 pixels
315 Hz full frame rate

These images in a
disruption sequence
separated by the standard
3.2 ms

Images from MAST:

S. Lisgo et al.,
E. Delchambre et al.,
PSI17, China
 IR thermography (9)
If surface being viewed can be
assumed “semi-infinite”:
1/ 2 q = Wm , T in Kelvin
-2
ΔΤ ⎛ ρCp κπ ⎞
q= ⎜ ⎟
2 ⎜⎝ t ⎟⎠
If material thickness, d, such that:
Aside from general surveillance d2 > 4κt/ρC and no external heat sink
(machine protection), most important Time dependent, semi-infinite
use of IR is to derive power flux from problems can also be treated (eg.
surface temperature. Duhamel’s method), but not if D = D(T)
This requires an “inverse solution” of
the heat conduction equation:
When D à D(T), need numerical
∂T solutions.
ρCp = −∇ ⋅ q = −∇ ⋅ (κ∇T)
∂t One method is to use the “heat
D(T) = κ/ρCp – heat diffusivity potential” and solve with finite
differences: T ' '
U(T ) = ∫T κ(T )dT
0

q = -κ∇T = - ∇U à ∂U/ ∂t = D∇2U

This is the method used by the
THEODOR code (A. Herrmann, IPP)
now used on many tokamaks
 IR thermography (10)
Example from TCV – J. Marki et al., PSI17
Tokamak surfaces are almost always
covered with re-deposited layers.
This poses a problem for transient IR

Power flux (Wm-2)

analysis because the “surface
layer” (with zero heat capacity and
reduced thermal conductivity) behaves
differently to the bulk substrate (tile).
Result is that after transient events (eg.
ELMs), cool down is too rapid for zero
heat flux and codes like THEODOR yield TCV Outer divertor
negative heat fluxes and overestimates
the deposited power Time (s)

To deal with this, assume heat

transfer across a thin layer:
ΔT = q d/κlayer; d = layer thickness à
q = α(Tmeas – Tbulk) and α = κlayer/d is a
∂T ∂T “heat transfer coefficient.
=0 =0 Procedure is to adjust α until negative
∂y ∂y heat fluxes are eliminated. Careful
calibration also required
(thermometric or calorimetric)
 IR thermography (11)
Limiter HEATED Limiter OHMIC DISRUPTION

600

400

200

0°C

JET Pulse 65670

P. Andrew et al, PSI 17
Effect of deposited layers can produce
serious errors if care not taken
Temperature °C

Here the layer on one side of the inner

poloidal limiter produced extremely fast
transient response during a disruption
à incorrect energy accounting
Heated Ohmic
Problem spotted when comparing
ohmic and heated phases of a limiter
Distance along one tile (m) plasma