Transcript of Chemical Erosion in DIII-D - University of Toronto
Chemical Erosion in DIII-DTHE DIVERTOR OF THE DIII-D TOKAMAK
by
for the degree of Doctor of Philosophy
Graduate Department of Aerospace Engineering
University of Toronto
ii
Quantification of Chemical Erosion in the Divertor of the DIII-D
Tokamak
Doctor of Philosophy, 2009
University of Toronto
The International Thermonuclear Experimental Reactor (ITER) is
currently designed
to use graphite targets in the divertor for power handling and
impurity control. Understanding
and quantifying chemical sputtering is therefore key to the success
of fusion as a clean
energy source. The principal goal of this thesis is to design and
carry out experiments, then
analyze and interpret the results in order to elucidate the role of
chemical sputtering in carbon
sources in the DIII-D tokamak.
A self-contained gas puff system has been designed, constructed,
and employed for
in-situ study of chemical erosion. The porous plug injector (PPI)
releases methane through a
porous graphite surface into the divertor plasma at a precisely
calibrated rate, minimizing
perturbation to local plasma while replicating the immediate
environment of methane
molecules released from a solid graphite surface more accurately
than done previously. For
the first time in a tokamak environment, the methane flow rate used
in a puffing experiment
was the same order of magnitude as that expected from laboratory
experiments for intrinsic
chemical sputtering.
measurement of background and incremental spectroscopic emissions
of both physically and
chemically-released sputtering products and by CI, 910 nm line
profile fitting. Comparison of
background and incremental emissions of chemically-released
products demonstrate a
dramatic drop in production of CH in cold and detached conditions.
Finally, the chemical
erosion yield is calculated in both attached and cold-divertor
conditions and found to be
much closer to that measured ex-situ in ion beam experiments than
previously determined in
DII-D.
These observations represent a positive result for ITER which will
operate at all times
with a detached divertor, i.e., a low chemical sputtering yield.
Results and analysis
techniques presented here point the direction for future
experiments with the PPI for study of
chemical sputtering in the tokamak edge environment.
Truth is sought for its own sake. And those who are engaged upon
the
quest for anything for its own sake are not interested in other
things.
Finding the truth is difficult, and the road to it is rough.
Ibn al-Haytham (Alhazen, 965–1039), one of the key figures in
developing the scientific
method and regarded as the father of modern optics.
iv
Acknowledgements
I would like to thank my supervisor, Peter Stangeby, for his vision
and dedication to
the field of boundary physics and his support of me from early in
my university career and
throughout my BASc, MASc, and PhD research at the University of
Toronto. The early idea
for this research was made by Jim Davis whose assistance and ideas
over these years have
been invaluable. Also at the University of Toronto, I would like to
thank Tony Haasz, Steve
Lisgo (now at MAST), David Elder, Yarong Mu, and Charles Perez for
their help.
I would like to sincerely thank the close collaboration I have had
with Neil Brooks
and the late Phil West at General Atomics. Their knowledge and
support have been
invaluable and each made himself available for my questions and
ideas without hesitation. I
would also like to acknowledge the valuable discussions,
contributions, input, and support I
have received from everyone I had the pleasure of working and
collaborating with at DIII-D,
including Steve Allen, Dana Blatchford, Jose Boedo, Rejean Boivin,
Amy Bozek, Bruce
Bray, Justin Burruss, Greg Campbell, Will Carrig, Lupe Cerda, Ron
Ellis, Max
Fenstermacher, Sean Flanagan, Eduardo Gonzo Gonzales, Matt Green,
Mathias Groth
(now at the Helsinki University of Technology), Matt Hansink,
Richard Harrington, Kurt
Holtrop, Al Hyatt, Gary Jackson, Woodie Jarrett, Kristi Keith,
Arnie Kellman, Jim Kulchar,
Charlie Lasnier, Rick Lee, Tony Leonard, Jim Lynch, Ali Mahdavi,
Kim Moore, Rick
Moyer, Chris Murphy, Alex Nagy, Tom Petrie, Dmitry Rudakov, Mike
Schaffer, Bonnie
Scoville, Tim Scoville, Wayne Solomon, Ted Strait, Derek Sundstrom,
Pete Taylor, Steve
Visser, Mickey Wade, Don Wall, Jon Watkins, Bob Williams, and
Clement Wong.
As well, I would like to acknowledge the support and input I
received from Dennis
Whyte at MIT, Russ Doerner and Eric Hollmann at UCSD, Stuart Van
Deusen and Bill
Wampler at SNL, Don Hillis, Ralph Isler, and Tom Jernigan at ORNL,
and Sebastijan
Brezinsek, Marcin Jakubowski, Andrey Litnovski, and Detlev Reiter
at Juelich.
This thesis was also made possible in part to the drilling
expertise of Jaime Itur riagga
and his staff at Data Circuits (Markham, Ontario), and
contributions from Albert Henning
and his team at the former Redwood Microsystems (Menlo Park,
California).
I would finally like to thank the loving encouragement and support
from my wife,
Sasha, and the longstanding support given to me over the years from
my mom and brothers.
v
1.1 Thesis Objectives
......................................................................................................
3 Chapter 2: Background
.......................................................................................................
4
2.1 The DIII-D Tokamak
................................................................................................
4 2.2 DiMES
......................................................................................................................
7 2.3 Diagnostic Systems
...................................................................................................
8
2.3.1 Langmuir Probes
.............................................................................................
10 2.3.2 Thomson Scattering
........................................................................................
10
2.3.3 DiMES TV
......................................................................................................
11 2.3.4 Thermal
Imaging.............................................................................................
13 2.3.5 High Resolution
Spectroscopy........................................................................
14
2.4 Chemical Erosion
....................................................................................................
18 2.4.1 Spectroscopic Indicators of Chemical Sputtering in
Tokamaks ..................... 20
2.4.2 Effect of Boron on Measurement of Chemical Sputtering
............................. 22 2.5 Hydrocarbon Breakup and Plasma
Interaction .......................................................
26
Chapter 3: Experimental
Design.......................................................................................
27
3.1 The PPI Concept
.....................................................................................................
27 3.2 Probe Design and Construction
..............................................................................
28
3.2.1 The PPI Probe Head
........................................................................................
28 3.2.2 PPI MkI
...........................................................................................................
30 3.2.3 PPI Mk
II.........................................................................................................
32
3.2.4 Common Elements to the PPI
.........................................................................
36 Chapter 4: Experimental Results: Analysis and Discussion
............................................. 39
4.1 Plasma Operation with the PPI
...............................................................................
39 4.2 An Analytic Model for the
Injection.......................................................................
41 4.3 PPI Gas Puff Characteristics
...................................................................................
42
4.4 Diagnostic Observations
.........................................................................................
44 4.5 Post-operations
Observations..................................................................................
50
Chapter 5: Interpretation
...................................................................................................
50 5.1 Spectroscopic Interpretation: The 430 nm Region
................................................. 51
5.1.1 Attached Divertor Plasma Condition
..............................................................
51
5.1.2 Integrated Emissions Versus Time in Attached Divertor
Plasma................... 55 5.1.3 Integrated Emissions Versus PPI
Gas Flow Rate in an Attached Divertor
Plasma 60 5.1.4 Cold Divertor Plasma
Condition.....................................................................
65 5.1.5 Integrated Emissions Versus Time in a Cold Divertor
Plasma....................... 67
5.1.6 Integrated Emissions Versus PPI Gas Flow Rate in a Cold
Divertor Plasma 70 5.1.7 Discussion
.......................................................................................................
73
5.2 Spectroscopic Interpretation: The 514 nm Region
................................................. 75 5.2.1
Integrated Emission Versus Time
...................................................................
78 5.2.2 PPI emission Characteristics: Attached Divertor Plasma
............................... 81
5.2.3 PPI Emission Characteristics: Cold Divertor Plasma
..................................... 84 5.2.4 Discussion
.......................................................................................................
86
5.3 Spectroscopic Interpretation: The 910 nm Region
................................................. 87
vi
5.3.1 Spectral
Analysis.............................................................................................
89 5.3.2 CI Line Profile Analysis
.................................................................................
92
5.3.3 Discussion
.....................................................................................................
103 5.4 Relative Roles of Chemical and Physical Erosion Processes
............................... 104
5.4.1 Procedure
......................................................................................................
105 5.4.2 Results and Observations: The PPI MkI Experiment
................................... 107 5.4.3 Results and
Observations: The PPI MkII Experiment
.................................. 108
5.4.4 The 430 nm Region: Integrated CII Emission Line and CD/CH
Bands ....... 109 5.4.5 The 514 nm Region: Integrated CII Multiplet
and C2 Dimer ....................... 115
5.4.6 The 910 nm Region: Integrated CI Multiplet
............................................... 120 5.4.7
Discussion
.....................................................................................................
123
5.5 Chemical Erosion Yield
........................................................................................
125
5.6 Emission Intensity Accounting
.............................................................................
128 5.6.1 Method
..........................................................................................................
128
5.6.2 Results: Intensity Continuity at CI, 910
nm.................................................. 130 5.6.3
Results: Intensity Continuity at CII, 514 and 427 nm
.................................. 134
Chapter 6: Conclusions and Contributions
.....................................................................
139
6.1 Conclusions
...........................................................................................................
139 6.2 Suggested Future
Work.........................................................................................
143
List of Tables
Table 1: Primary optical filters used in DiMES TV for the PPI
experiment. ......................... 13 Table 2: Spectral features
of the MDS with the Wright Instruments camera (spectra spread
across 770 pixels) for commonly used wavelengths.
.............................................................. 17
Table 3: Spectral features of the MDS with the Princeton
Instruments 1024BFT camera
(spectra spread across 1024 pixels) for commonly used wavelengths.
.................................. 17 Table 4: History of plasma
exposure and time since previous BZN for each PPI run day on
DIII-D......................................................................................................................................
26
Table 5: List of major components used for the PPI. Components
specific to each of the PPI MkI and MKII are listed separately, as
are those common to both models. ...........................
38
Table 6: Summary of DIII-D run days with data collected for the
PPI. Primary data-capture run dates are shown in
bold.....................................................................................................
39 Table 7: Spectral line characteristics of the CII, 514 nm
multiplet. ....................................... 75
Table 8: Spectral line characteristics of the CI, 910 nm multiplet.
......................................... 88 Table 9: Bounds of
integration for spectral features of interest for the PPI.
........................ 106
Table 10: Data required calculating the inferred chemical and PS
contribution to C0 in an attached
divertor....................................................................................................................
122 Table 11: CI, 910 nm emission intensity continuity results for
PPI MkI SAPP shot 122199.
...............................................................................................................................................
130 Table 12: CI, 910 nm emission continuity for PPI MkI shot
122199, assumed yield approach.
...............................................................................................................................................
131 Table 13: CI, 910 nm emission intensity continuity results for
PPI MkII SAPP shot 129692.
...............................................................................................................................................
131
Table 14: CI, 910 nm emission continuity for PPI MkII shot 129692,
assumed yield approach.
...............................................................................................................................
132
Table 15: CI, 910 nm emission intensity continuity results for
SAPPI shot 105500. .......... 133 Table 16: CII, 514 nm emission
intensity continuity results for PPI MkI SAPP shot 122197.
...............................................................................................................................................
134
Table 17: CII, 514 nm emission continuity for PPI MkI shot 122197,
assumed yield approach.
...............................................................................................................................
135
Table 18: CII, 514 nm emission intensity continuity results for PPI
MkII SAPP shot 129691.
...............................................................................................................................................
135 Table 19: CII, 514 nm emission continuity for PPI MkII shot
129691, assumed yield
approach.
...............................................................................................................................
136 Table 20: CII, 514 nm emission continuity for SAPPI shot
105514, assumed yield approach.
...............................................................................................................................................
137 Table 21: CII, 427 nm emission intensity continuity results for
PPI MkII SAPP shot 129693.
...............................................................................................................................................
138
Table 22: CII, 427 nm emission continuity for PPI MkI shot 129693,
assumed yield approach.
...............................................................................................................................
138
viii
List of Figures
Figure 1: The DiMES mechanism beneath DIII-D.
..................................................................
8 Figure 2: Locations and optical views for diagnostics pertinent
to this thesis, projected into . 9
Figure 3: View of the lower divertor shelf near DiMES with
diagnostic views overlaid. ....... 9 Figure 4: The DiMES TV
assembly.
......................................................................................
12 Figure 5: Reciprocal MDS system sensitivity normalized by its
sensitivity in 2007 for each 16
Figure 6: HC erosion yield and species composition versus surface
temperature [4]. ........... 20 Figure 7: Spectral profile in the
430 nm region captured with the MDS. ..............................
21
Figure 8: 430 nm region spectral profile as viewed by the MDS
during plasma shots at ...... 23 Figure 9: BD band emission
contamination ratio vs. time since the last BZN in DIII-D....... 25
Figure 10: HC breakup reactions available [5,6].
...................................................................
27
Figure 11: The graphite porous cap shown before being brazed onto
the DiMES sample. ... 29 Figure 12: a) PPI DiMES sample body and
base separated, and b) assembled. ..................... 30
Figure 13: a) PPI MkI gas chamber shown prior to final assembly,
and b) after ................... 31 Figure 14: a) PPI MkI computer
aided design, and b) implementation inside DiMES. ......... 32
Figure 15: Comparison between experimental canister pressure during
calibration and the . 34
Figure 16: PPI MkII gas chamber, view from bottom (left) and top
(right). .......................... 35 Figure 17: PPI MkII fully
assembled.
.....................................................................................
35
Figure 18: a) PPI MkII design and b) implementation in the DiMES
chamber. .................... 36 Figure 19: a) TC assembly in the
DiMES sample base modified with a flow diffuser, and .. 37 Figure
20: PPI MkI valve control waveform.
.........................................................................
39
Figure 21: SAPP magnetic equilibrium geometry with (2007-onward)
and without the ....... 40 Figure 22: Magnetic and spatial
configuration viewing the inserted PPI sample toroidally. .
40
Figure 23: Angular flux of molecules exiting the PPI cap for short
and long channel cases. 42 Figure 24: Micro valve leak rate and
canister pressure on the dedicated PPI MkI run day. .. 43 Figure
25: Shot-averaged gas flow rate from the PPI MkI.
.................................................... 43
Figure 26: CH4 and He puff rate from the PPI MkII on one
experimental day. Shot............. 44 Figure 27: DiMES TV view of
the PPI, (a) in 2005 with the analog camera, and (b) in .......
45
Figure 28: Sensitivity-normalized spectral features in the 430 nm
region of the MDS ......... 46 Figure 29: Integrated poloidal
full-span CD/CH emission detected by viewing each of the . 47
Figure 30: Integrated poloidal CII, 427 nm emission detected by
viewing each of the on- ... 47
Figure 31: TC data from the porous cap of the PPI MkI. Multiple
curves that peak at.......... 49 Figure 32: Porous cap
plasma-facing surface (a) before and (b) after exposure.
................... 50
Figure 33: (a) 430 nm spectral region measured by the MDS in frame
14 (integration ......... 52 Figure 34: Integrated CD/CH signal
after subtraction of contaminating BD/BH emission ... 56 Figure 35:
Chemical erosion yield for plasma-exposed graphite tiles vs.
surface.................. 57
Figure 36: Integrated BD/BH signal versus time in the attached
divertor condition. ............. 58 Figure 37: Integrated CII, 427
nm signal versus time in the attached divertor condition. .....
59
Figure 38: Integrated D / H signal versus time in the attached
divertor condition. ............ 59
Figure 39: Integrated incremental CD/CH signal measured from the
PPI gas puff versus .... 60 Figure 40: eD/XB value for emission of
CH and CD due to release of CH4 into an attached 61
Figure 41: eD/XBCH4CH+CD values as measured at devices worldwide
(modified from .... 62
Figure 42: Integrated incremental CII, 427 nm signal measured from
the PPI gas puff ........ 64 Figure 43: eD/XB value for emission of
CII due to release of CH4 into an attached divertor 65
ix
Figure 44: (a) The 430 nm spectral region as measured by the MDS in
frame 20 ................. 66 Figure 45: Comparison between spectra
measured during the attached and detached ........... 66
Figure 46: Integrated CD/CH emission versus time in a shot that
transitions from attached . 68 Figure 47: Integrated CII, 427 nm
emission is shown versus time in a shot that transitions . 69
Figure 48: Integrated D / H signal versus time in a plasma shot
which transitions from ... 69
Figure 49: Integrated BD/BH band signal versus time in the same
shot as analyzed for ....... 70 Figure 50: Integrated incremental
CD/CH signal measured from the PPI gas puff versus .... 71
Figure 51: eD/XBCH4CH+CD values as measured at devices worldwide
(modified from .... 72
Figure 52: Integrated incremental CII, 427 nm signal measured from
the PPI gas puff ........ 73 Figure 53: (a) 514 nm spectral region
as measured by the MDS during the PPI MkII .......... 77 Figure 54:
The C2 Swan band as measured during attached (a) and cold divertor
(b) ........... 77
Figure 55: Integrated CII counts are shown for shot 130116 which
transitions from ............ 79 Figure 56: Integrated C2 counts
are shown for shot 130116 which transitions from .............
79
Figure 57: Ratio of integrated C2 over integrated CII emission
measured on all chords ....... 81 Figure 58: Integrated incremental
CII, 514 nm multiplet signal measured from the PPI gas 82 Figure
59: Integrated incremental C2, 517 nm multiplet signal measured
from the PPI gas.. 83
Figure 60: Integrated CII, 514 nm multiplet and C2, 517 nm band
(secondary axis) ............. 83 Figure 61: Major radius of the
OSP versus time in shots with extended attached divertor ....
84
Figure 62: Integrated incremental CII, 514 nm multiplet signal
measured from the PPI ....... 85 Figure 63: Integrated incremental
C2, 514 nm multiplet signal measured from the PPI gas.. 85 Figure
64: Spectral profile measured by the MDS in the 910 nm region. Data
are shown .... 90
Figure 65: Integrated counts in the range of 907 - 912 nm,
including the CI multiplet lines. 91 Figure 66: Integrated counts
in the range of 907 - 912 nm are shown for shot 129064 .........
92
Figure 67: Zoomed view of the 907.82 nm CI line profile. The
wavelength calibrated rest .. 94 Figure 68: Best fit to the
normalized CI line profile viewing the PPI MkI CH4 gas puff ......
96 Figure 69: Best fit of simulated CI line profile to experimental
data viewing the PPI MkI ... 97
Figure 70: Simulated CI line profile replicating the results
measuring background emission98 Figure 71: Best fit of simulated CI
line profile to experimental data viewing off-DiMES .... 99
Figure 72: Best fit of simulated CI line profile to background data
measured off-DiMES.. 100 Figure 73: Best fit of simulated CI line
profile to experimental data viewing off-DiMES .. 101 Figure 74:
Best fit of simulated CI line profile to experimental data viewing
off-DiMES .. 102
Figure 75: (a) Incremental CII vs CD/CH integrated emission counts
for all PPI gas puff . 108 Figure 76: (a) Relationship between
integrated CII emission vs. integrated CD emission .. 110
Figure 77: (a) Spectral emission from the MDS for PPI MkII shot
130117, frame 14 ........ 112 Figure 78: (a) Relationship between
background integrated CII emission vs. integrated .... 114 Figure
79: (a) Relationship between integrated CII emission vs. integrated
C2 emission.... 116
Figure 80: (a) Relationship between integrated CII emission vs.
integrated C2 emission.... 119 Figure 81: Procedure for
determination of the chemical/physical contribution to local ......
121
Figure 82: Relative contribution of chemical and physical
sputtering to local C0 ............... 123 Figure 83: Inferred
chemical erosion yield in attached and cold divertor
............................ 126 Figure 84: Chemical erosion yield
inferred from the PPI MkII vs. laboratory measured .... 127
Figure 85: Semi- log reciprocal sensitivity of the MDS diagnostic
versus time for.............. 133
x
Abbreviations
e Plasma confinement time (equal to energy content/power loss
rate)
Particle flux [/m s]
a Minor radius [m]
A-X The 22 XA electronic system of CD/CH
ADAS Atomic Data and Analysis Structure
ADC Analog to Digital Convertor
ADU Analog to Digital Unit (also referred to as counts)
AIMO Advanced Inverted Mode Operation (also referred to as
MPP)
AMU Atomic Mass Unit, equal to 1.661x10-27 kg (1/12th the mass of
12C)
AVI Audio Video Interleave movie file format
BT Toroidal magnetic field [Tesla]
BD Boron Deuteride molecule
BFT Backlit Frame Transfer
BZN Boronization
CI, CII, CIII Emission from Carbon charge states C0, C+, C2+
CCD Charge-Coupled Device
CS Chemical Sputtering
CW Clockwise
n=3, 4, and 5 to 2
D Dispersion [nm/pixel]
DTS Divertor Thomson Scattering
e- electrons
EDX Energy Dispersive X-ray
EEV English Electric Valve Company Ltd. (presently known as e2v,
known
as Marconi Applied Technologies from 1999-2002)
EFIT Equilibrium Fitting computer code
ELM Edge Localized Mode
f Focal ratio, or relative aperture of an optical system
FFT Fast Fourier Transform
FPS Frames Per Second
G System gain [e-/ADU]
G/mm Grooves per millimeter
n=3, 4, and 5 to 2
HC Hydrocarbon
Isat Ion saturation current (also jsat)
IDL Interactive Data Language
IP Instrumental Profile
IRTV Infrared Television
k Reciprocal sensitivity [(ph/s/m2/sr/nm)/count]
LP Langmuir Probe
MCP Micro channel plate image intensifier
MD Molecular Dynamics
MDS Multichord Divertor Spectrometer
MDSplus Model Data System set of software tools for data
acquisition
MDS-SIM Multichord Divertor Spectrometer Simulator
MEMS Micro Electro-Mechanical System
MFP Mean Free Path
ne Electron density [/m3]
NBI Neutral Beam Injection
NIMO Non Inverted Mode Operation
NIR Near Infrared
NPT National Pipe Thread
OSP Outer Strike Point
PDF Probability Distribution Function
PFZ Private Flux Zone
SNR Signal- to-Noise Ratio
SOL Scrape Off Layer
Te Electron temperature [eV]
Ti Ion temperature [eV]
Tint Integration time [s]
Top Operating temperature [K]
Z Elevation in the tokamak relative to the midplane
Zeff Mean effective charge of a plasma
1
Chapter 1: Introduction and Thesis Objective
Since chemical sputtering (CS) of carbon (C) is a key player in
tritium co-deposition,
target lifetime, and flake and dust formation, understanding its
properties in present day
fusion devices is crucial for the successful operation of an ITER
design incorporating C in
the first wall [1,2,3]. The primary means for studying chemical
erosion and re-erosion in-situ
in operating tokamaks is through analysis and quantification of
spectroscopic emissions from
products of chemical interaction between atomic D+ and D0
bombarding the carbon substrate
and/or layers built up in net deposition zones. Prominent emissions
from these by-products
which are accessible to visible and near infrared (NIR)
spectroscopy include the CD/CH
Gerö system ( 22 XA transition, with bandhead at 430.8/431.3 nm),
the C2 dimer Swan
system ( uad 33 transition, with bandhead at 516 nm), atomic CI
(multiplet at 908.93
nm), and ionized CII (426.7 nm, multiplet at 514.33 nm). Precise
absolute quantification of
these emissions and accurate calibration of those photonic
measurements to particle fluxes
for given plasma conditions and monitoring diagnostics are critical
for understanding CS and
thus the functioning of erosion processes in the plasma-materials
interface.
In the standard view, there are two mechanisms for carbon
erosion—physical
sputtering (PS) leading to the release of energetic C0, and CS
leading to the release of
thermal hydrocarbons. Complicated processes involving the launch of
many CxDy species [4],
numerous possible pathways for decomposition during transport
[5,6], and subjection to
chemical processes, including possible re-composition as
hydrocarbons with 2x and re-
erosion, upon transport to a surface [7,8], all make CS a challenge
to summarize and quantify
in a manner which can be easily applied in modelling codes.
Furthermore, this understanding
has recently been challenged by molecular dynamics (MD) simulations
[9], but this standard
view is still thought to be valid with the caveat that measured
neutral C may not necessarily
be physically sputtered, and that CH, CH2 and other molecular
radicals which were
previously produced chemically then deposited on a plasma facing
surface may be directly
released from the surface by physical means.
To relate measured photon fluxes to particle fluxes at the
measurement location,
inverse photon efficiencies must be derived, usually by gas puffing
experiments [10], and
2
more recently by code calculations [11,12]. For atoms, this
quantity is expressed as an S/XB
value, where S is the ionization rate coefficient, and for
molecules, a D/XB value, where D is
the decomposition rate coefficient. In both cases, X is the
excitation rate for a given breakup
process, and B is the branching ratio for decomposition to a
particular end state. While S/XB
values are relatively well known and catalogued for a wide range of
electron density ne and
temperature Te [13], D/XB values are difficult to accurately
catalogue due to their similar
dependence on ne and Te, and their additional dependence on
interactions which occur at the
plasma-material interface such as prompt deposition, and surface
temperature.
For all hydrocarbons released, transported, decomposed, deposited,
and possibly re-
released, molecular C band system emissions are considered
representative of hydrocarbon
fragment density at the measured location, i.e., CD as a step in
the breakup chain of simple
(CxDy, x=1) hydrocarbons, and C2 for so-called higher‘ hydrocarbons
(x 2). CD is the final
step in most breakup processes before a chemically-released
hydrocarbon becomes a C0 or
C+ (which may also be derived from PS), thus the D/XB for CD is
considered the most
significant measurement to quantify CS in fusion devices. An
additional simplification made
is that, for experimental measurement procedures, CD4 is sometimes
considered a
representative proxy for all chemical and ion- induced hydrocarbon
release mechanisms
because, upon injection, considerable emission from CD is emitted
as a consequence.
Nevertheless it is important to establish experimentally the
probability that a CD is created
from each type of HC molecule entering a plasma. Thus, the
effective‘ D/XB, or eD/XB for
emission of molecular CD (or CH, or C2) from the injection of CD4
(CH4) is a useful way to
link the measured Gerö-band or Swan-band light to an original
chemically-released carbon
influx.
Historically, determination of eD/XB for products of CS has been
made by artificial
injection of stable CH4/CD4, C2H6/C2D6, etc. at known rates into
locations in the fusion
device, typically at or near the targets of the primary divertor
where the impinging
hydrogenic flux is greatest, and emissions from their resulting
molecular fragments, io ns and
atoms are monitored. The injection rate is carefully controlled and
minimized to reduce the
risk of perturbations to the local plasma, but of sufficient flux
so as to provide desired
spectroscopic quality compared to that of the background emission.
Injections are typically
made at single puff locations, i.e., the outlet of a gas tube, or
between tile gaps, but also may
3
be accomplished over larger areas via toroidally symmetric
injection, e.g., using a cryo-pump
chamber without pumping [14, 15]. A comprehensive database of
measurements to date is
found in [16] and the references therein. In most cases, however,
the gas injection geometry
(i.e., the molecular release distribution and density) is
significantly different from that of a
true chemically sputtered source; specifically, released
hydrocarbon molecules may not
experience interaction with the graphite over their mean free path
(MFP) as they would if
released from a PFC tile. Additionally, the injection rate used in
previous experiments is
often considerably greater (10 or more) than that of chemical
erosion due to the challenge
of adequate detection in available steady-state integration periods
and accurate background
subtraction for emissions from injections at intrinsic rates.
As the central element to this thesis, the porous plug injector
(PPI) for DIII-D was
conceived, designed, and built as a self-contained, in-situ
diagnostic to admit methane (or
other hydrocarbons) in a distributed manner through a porous
graphite surface, such that
molecular interaction with the plasma closely approximates a
hydrocarbon molecule released
from a carbon surface by chemical erosion. Injecting methane at a
known and controlled rate
provides direct comparison to background emission levels and
calibration of spectroscopic
signals on monitoring diagnostics (e.g., spectrometers and
spectrally filtered cameras). DIII-
D generally uses D2 fueling; therefore CH4 is used for PPI
injections in order to take
advantage of the 0.5 nm separation of bandheads between CH and CD
near 430 nm. The
porous surface is designed such that size and spacing of the holes
is on the order of the MFP
for breakup of CH4 in divertor plasma. The porosity of the cap is
low so that the probe
closely approximates a solid surface. The injection rate is
selected to correspond to a 2%
DC erosion yield, C
chemY , over the holed area of the PPI cap at the OSP. Emissions
over
background from the region above and around the PPI in the lower
divertor floor are then
studied in typical attached and detached operating regimes in
DIII-D, and D/XB values are
calculated for those conditions.
4
Conceive, design, construct and operate the PPI for DIII-D, then
carry out
experiments using it as a platform to artificially replicate the
release of hydrocarbons
due to CS better than has been done before in a tokamak
environment
Upgrade and/or install new diagnostics on DIII-D with the aim of
achieving the best
possible characterization of the divertor plasma and emissions from
the PPI
Collect, analyze, and interpret data from the PPI and associated
diagnostics using
physical models and computer simulations. Figures of merit to
measure and compute
for given plasma conditions include (a) chemical erosion yield, (b)
effective photon
efficiencies for products of chemical erosion, and (c) the fraction
of C0 and C+ in the
divertor boundary whose source is chemical vs. physical
sputtering
Determine how well our current understanding can explain what is
measured, and to
identify possible missing physics
A detailed description of the PPI is contained in this thesis but
also published in [17].
Results obtained to date with the PPI have been published in [18]
and [19]. Data collected
and analyzed to accomplish these goals will be valuable for
application to predictions of the
performance of ITER and future fusion science and energy devices
using carbon as a first
wall material.
2.1 The DIII-D Tokamak
DIII-D is a medium-sized tokamak (R=1.695 m, a=0.681 m, A=R/a 2.5,
total height
from floor to ceiling = 2.72 m), located in San Diego, USA, and
operated by General
Atomics for the US Department of Energy [20]. Full magnetic BT
field strength is 2.1 T. The
machine was designed for a full plasma current of Ip=3 MA but is
regularly operated at 1.7–
2.1 MA to prevent damage during disruptions. The typical shot
duration is 5.0–7.0 seconds.
The primary armour of the plasma-facing wall in DIII-D is composed
of many
individual tiles with average dimensions ~20 x 20 x 5.5 cm and
average mass of ~5 kg. All
tiles are manufactured from a single material, ATJ graphite
manufactured by Union Carbide
(originally referred to as TS-1792 graphite). Current wall/divertor
coverage is ~89% (67.23
5
m2 of 75.86 m2 total surface area) comprising 3,200 carbon tiles
with a total volume of 4,232
L and a total mass of 7.5 metric tonnes [21]. Coverage by this full
graphite‘ wall is limited
only by the area of diagnostic and operational port openings in the
vessel. Carbon armour
coverage is used in order to reduce metallic impurities into the
plasma, thus reducing its Zeff.
Gaps between armour components range from ~3–5 mm for tiles
installed in
1992/1993, down to 1–3 mm for tiles installed in 2000 (upper
divertor baffles), and 0.5–1.0
mm for tiles installed in 2006 (lower divertor shelf). Tiles on the
shelf are actively cooled
and rated for a peak heat flux of 9.6 MW/m2 for 10 s [22]. Graphite
tiles around the rest of
the vessel are not actively cooled but designed to withstand 5
MW/m2 for 5 s or 3.8 MW/m2
for 10 s. Individual tiles are bolted to studs on the vacuum vessel
using a GRAFOIL
complacent layer to improve thermal conductivity and decrease
stress concentration at the
tile/vessel interface. Tile surface temperature during
low-confinement mode (L-mode)
operation is ~300K (room temperature) for all walls, 350–450K at
the targets. During higher
power high-confinement (H-mode) operation, tile surfaces may reach
400K on the walls, and
up to 1200K at the targets [23].
The DIII-D vacuum vessel has a net volume of 35 m3 and is made of
Inconel 625.
The vessel is regularly pumped down to P<1x10-7 torr between
plasma operations. A total of
180 port penetrations in the vessel allow in-vessel views for
diagnostics and placement of
operational equipment. 90% of the vessel is double layered with
corrugated separators to
allow for water cooling during operations and circulation of hot
air during baking procedures
[24]. Plasma-facing surfaces are not actively cooled during plasma
operation.
Injection of fuel and impurity gases into the vessel is carried out
primarily using
piezoelectric valves [25,26] that include no magnetic components.
Response time for the
valves is 0.5 ms and controlled flow rates are achievable from
0–500 torr L/s (D2). The
primary valve used for injection of fuel is at 300° toroidally and
in the upper ceiling of the
device. Additional valves are available beneath each of the outer
baffles which, upon
injection, have sufficient time to distribute throughout the baffle
region before passing into
the channel between the floor/ceiling and the baffle in a
toroidally symmetric fashion.
Cryo-pumping from three locations is available for density and
recycling control
during plasma shots: beneath the low triangularity lower outer
Advanced Divertor Project‘
(ADP) ring installed in 1992, above the upper outer Radiative
Divertor Program (RDP) 1A‘
6
baffle installed in 1996, and above the upper inner RDP 1B‘ baffle
insta lled in 1999 [27,28].
The lower outer ADP ring was replaced with a new shelf for improved
pumping performance
in high operation from September 2005 to January 2006 [29].
While their effectiveness and necessity is currently a subject of
intense investigation,
DIII-D regularly employs a variety of techniques to maintain
particle control, and for
obtaining reproducible conditions [30]. These techniques are listed
as follows:
Helium glow discharge cleaning (GDC) lasting 5-12 minutes is
carried out after most
plasma shots to desorb hydrogen from the graphite walls and thus
maintain density
control. During a GDC, 1–4 mTorr of He is injected into the vessel
and two anodes
are used at 300-500V and 1–7A to create weakly ionized, long MFP
plasma. After
repeated use, this process was found to lead to faster recovery
after a disruption, and
reliable, high performance operation at low q.
Baking of the vessel to an average temperature of 350°C is done
after each vent to
atmospheric pressure, and every 1-2 months during operations to
desorb hydrogen
and remove impurities in the walls. Baking is accomplished by a
combination of
inductive heating using the ohmic heating coil and circulating hot
air through the
vessel double wall [24]. The bake procedure is (a) an 8–hr warm-up
period to the
target temperature (~380°C on the centrepost, ~320°C on the outer
wall), (b) a steady
4–hr period at bake temperature, and (c) a cool-down period of
~10–12 hr.
Finally, boronization (BZN) of the PFCs has been carried out on
average 3.8 times
per year to reduce impurity release into the plasma. BZN is carried
out during a bake
of the vessel using a glow discharge with injection of a 90% He/10%
B2D6 (diborane)
mixture from eight locations poloidally and toroidally distributed
around the machine
(and fully independent of other locations for fuel/impurity
puffing) [31]. Following a
BZN, He in the core plasma (as measured by the SPRED) is
significantly increased
(>10X for the first ~10 shots), but C and O impurities in the
plasma are significantly
reduced (initially ~75%, reducing to ~30–50% after ~30–50 full
length plasma shots)
and improvements in confinement quality and plasma performance are
observed. B,
Ar, and N concentrations in the plasma are typically unaffected by
BZN. (The lack of
an effect on B in the core after BZN is especially interesting).
Each BZN deposits a
7
layer of B with an average thickness of 82 19 nm. A total of 68
boronization
procedures have taken place in DIII-D in the past 17 years.
2.2 DiMES
The Divertor Materials Evaluation System (DiMES) on DIII-D is a
mechanism to
allow remote insertion of samples to the plasma-facing surface of
the lower divertor followed
by retraction and removal for ex-situ study [32,33]. The system
uses a multi-chamber
hydraulic piston extending from the base of the vacuum vessel to
the floor of the DIII-D pit
(a distance of 4.03 m), centred below R=1.486 m at 150° toroidally.
A 30.5 cm diameter
chamber that is 1.89 m below the surface of the DIII-D floor tiles
encloses a volume where
DiMES samples may be mounted for insertion (see Figure 1). During
sample changes in the
exchange chamber, a torus isolation valve (TIV) is activated to
isolate the DiMES chamber
from the torus vessel. Once the sample is mounted in the chamber,
the exchange port is
sealed, and the DiMES chamber is pumped down to a pressure
set-point of 110-7 torr, at
which time the TIV is allowed to open. Typical pump-down time to
the set-point is ~12 h,
however, if necessary for plasma operations and if the residual gas
analyzer (RGA) is used to
monitor impurities in the torus, the DiMES TIV may be opened by
overriding the set-point
after only ~40–60 min in which time the DiMES chamber may reach
only ~110-6 torr. The
sample may then be inserted by following a guide tube up to a
conical cradle at the bottom of
the lower divertor tiles.
A typical DiMES sample is 5.7 cm in diameter at its base and 9.5 cm
in total height.
2.3 cm of the lowermost portion of the sample makes up the locking
mechanism for the
mount. The upper portion of a DiMES sample which extends all the
way up to be flush with
the divertor surface is 5.3 cm in height and 4.8 cm in diameter.
The penetration into the
graphite tiles of the lower divertor is 5.0 cm in diameter
(poloidal extent of
1.4620 R 1.5098 m), leaving a 1.1 mm circumferential gap between
the sample and
surrounding tiles. Before the long torus opening activities (LTOA)
on DIII-D in 2006, the
DiMES sample penetrated into tiles of the lowermost floor; after
2006, however, the sample
was repositioned up to the newly installed divertor shelf.
Below the DiMES sample mount is an empty space approximately 7.1 cm
high and 5.3
cm in diameter. The height of this space reduces by up to ~0.5 cm
upon insertion of the
8
sample because the lower portion of the assembly is mounted on a
heavy spring intended to
absorb the impact shock at the moment the DiMES sample reaches its
fully extended length.
Beneath the DiMES sample mount is a 2.8 cm diameter, 13-pin
electrical feedthrough which
includes two pins for Type-E thermocouple leads (chromel and
constantan, both of which are
non-magnetic). The remaining pins provide the capability to
transmit electrical power and
data in and out of the DiMES chamber, respectively. Data signals
may be transmitted
optically (for electrical isolation) via five AFL-500 analog
fibre-optically- isolated
transmitter-receiver pairs manufactured by A.A. Lab Systems up to
the primary equipment
annex next to the DIII-D control room, where they may be digitized
and stored for future
reference and analysis.
2.3 Diagnostic Systems
DIII-D is recognized to be one of the best diagnosed tokamaks in
the world. Diagnostic
systems pertinent to the present thesis may be loosely separated
into two groups: those used
primarily for study of background plasma conditions (i.e., upstream
from the divertor
targets), and those used primarily for study of plasma-materials
interactions (PMI). The
former group is composed of Langmuir probes, Thomson scattering
(TS), charge-exchange
recombination (CER), tangential television (TTV) cameras, and
optical filterscope (FS)
monitors. Diagnostics used primarily for studying PMI in the device
include: the multi-chord
9
divertor spectrometer (MDS), Ocean Optics (OO) spectrometers,
infrared television (IRTV),
and DiMES-viewing television (DiMES TV). The position or viewing
location and geometry
for each diagnostic system relevant to the PPI experiment is shown
in Figure 2 and Figure 3
below, projected into the poloidal plane and shown overlaid in real
space in a view from
above the lower divertor.
Figure 2: Locations and optical views for diagnostics pertinent to
this thesis, projected into
the poloidal plane in the DIII-D first wall geometry including the
lower divertor shelf (2006–
onward). A magnified view of the region near the DiMES is shown on
the right.
Figure 3: View of the lower divertor shelf near DiMES with
diagnostic views overlaid.
10
2.3.1 Langmuir Probes
Langmuir probes (LPs) are used in the present thesis to
characterize the background
plasma at the plasma-material interface in DIII-D. LPs function by
insertion of a small
electrode in the plasma boundary, and subsequent measurement of the
current (I)-potential
(V) curve in the Debye sheath by sweeping the external voltage. The
characteristic curve
yields information about Te, plasma floating potential (Vf) and ion
saturation current (Isat),
which then makes it possible to determine ne. DIII-D employs two
arrays of single probes:
the first is in the lower divertor at 172.5° toroidally (19 probes
extending along the divertor
floor previous to the 2007 run campaign, and 20 in the post-LTOA
2007 campaign and
onward), and the second array of 21 LPs at 180° toroidally
installed in 2000 is in the upper
divertor [34]. In the configuration after the LTOA, LP tips on the
floor and shelf are made of
pyrolytic graphite and proud 0.66 mm relative to the surrounding
tile surfaces [35,36].
Analysis of Langmuir probe characteristics in magnetic plasmas of a
tokamak tend to
significantly overestimate Te as measured nearby with TS [37]. In
[38], this observation is
suggested to be caused by the lack of a single Maxwellian electron
temperature distribution
in the edge, which may exist in low-collisionality regimes.
Instead, electrons may be
characterized by a dominant thermal component, with a minor
population of hot electrons
(suprathermal) which skews the measured temperature in a non-
linear manner [39,40].
2.3.2 Thomson Scattering
The multi-pulse Thomson scattering (TS) system on DIII-D provides
highly spatially
and temporally localized measurements of Te and ne [41,42,43,44,45]
used to characterize the
divertor plasma in the present work. This is accomplished by
measuring Nd:YAG (20 Hz, 1
J, 15 ns pulse time, 1064 nm) laser light which has been
elastically scattered in the plasma
(~10-13 of the incident light) and shifted in wavelength, then
comparing the wavelength
profile of the detected light with that of a theoretical Thomson
spectrum [46] using a lookup
table. Seven channel polychromators are used for each view-chord to
separate light of
adjacent wavelengths ranging from 970 to 1070 nm using narrowband (
1.6 nm FWHM),
high throughput (average of 70% transmission) interference filters.
Silicon avalanche
11
photodiodes (RCA—later EG&G, now Perkins Elmer—C30956E) are
utilized due to their
efficient long wavelength response and fast response time (2 ns
rise/fall time) [47].
The range of parameters measured using the Thomson system in DIII-D
are
1<Te<500 eV, and 5x1018<ne<1x1021 m-3. The volume of
plasma measured is approximately
3.5510 mm. TS chords on DIII-D are separated into three systems;
two that intersect laser
beam positions in the core plasma (i.e., radially inside the
separatrix), and one that intersects
a single laser beam in the lower divertor. The first system is
referred to as core TS and is
composed of 32 chords along the R=1.48 m major radius from the
midplane up to an
elevation Z=0.826 m. The second system is referred to as tangential
TS and includes five
additional chords extending approximately along the midplane from
R=1.655 to 1.881 m.
The third system is referred to as the DTS and composed of eight
chords, all located at
R=1.485 m and 120° toroidally. Before the LTOA, the DTS chords were
spatially distributed
from -1.167<Z<-1.344 m (lowermost chord 22 mm above the
floor), and after the LTOA,
DTS chords extended from -1.022<Z<-1.246 m (lowermost chord 4
mm above the floor).
2.3.3 DiMES TV
The DiMES television (TV) is a video camera located at 150°
toroidally at the top of
the torus (adjacent to the fibre bundle for the MDS), viewing the
DiMES port and the
surrounding graphite tiles. DiMES TV is mounted horizontally on an
electrically isolated
bracket made of G-10 fibreglass directed towards a mirror which
directs the view of the
camera downwards into the vessel (Figure 2). Using a 50 mm, f/1.4
lens, the view of the
lower divertor floor extends approximately 60 cm in the toroidal
direction and 50 cm in the
radial direction. The camera views the floor through a 10 cm outer
diameter (OD), 6.35 mm
thick window made of optical glass with a thin sapphire shield on
the vacuum-facing side.
The shield is coated with a single quarter wave thickness MgO2
anti-reflection layer to
improve transmission efficiency and reduce ghosting and halo rings
in the video image.
DiMES TV includes a four-position filter wheel which allows
insertion of a 2.54 cm
diameter spectral bandpass filter in the light path to limit
detection to a single emission line
or band (Figure 4).
12
From its installation in 1996, the standard device used in DiMES TV
is a
commercially available camera manufactured by Marshall Electronics
(model number V-
XA076) utilizing an Omnivision 1/3 (OV7411) 510492 pixel
Complementary Metal-
Oxide Semiconductor (CMOS) imaging sensor. A CMOS chip is used
instead of a CCD
detector because of its improved resistance to radiation damage.
Data from the CMOS chip
are converted to analog composite video output at 30 full frames
per second (FPS) and 310
lines of resolution, transmitted to the DIII-D control room by
isolated 75 Ohm video cable,
and recorded to a video cassette recorder (VCR) tape. For analysis
purposes, the video signal
is digitized into 8-bit (256 level) grayscale images using a
National Instruments NI-1407 PCI
digitizer via a Labview-based video capture program developed by
LLNL [48]. Analog
output from the camera has a signal-to-noise ratio (SNR) of >38
db, indicating a dynamic
range of at least 80 and corresponding to a digital bit depth of at
least 6.3.
A new DiMES TV camera became available for DiMES experiments in
2007 using a
low-cost all-digital 30 FPS FireFly MV CMOS camera manufactured by
Point Grey
Research Inc. and employing a Micron 1/3 752480 pixel detector
[49]. The camera uses a
fibre-optically linked IEEE-1394 Firewire 400 interface from the
DIII-D vessel to the control
room to extend the range of the digital connection and to provide
electrical isolation. The
new camera provides improvements in gain/shutter control, dynamic
range (>55 db), pixel
resolution, and bit depth (10-bit) over the existing analog camera.
Data from the camera are
recorded as audio video interleave (AVI) digital movie files and
processed frame-by-frame
using the Interactive Data Language (IDL).
Figure 4: The DiMES TV assembly.
13
Video from the DiMES TV may be absolutely calibrated using data
from the MDS
over the area which the MDS view chords cover in the camera view.
By backlighting the
DiMES region using a lamp at 150° V+1, DiMES TV may be used for
spatial calibration of
all optical chords used by the MDS.
Designation Centre Bandwidth Blocking Species
CD/CH 22 XA 430.5 nm 3.5 nm UV to 1200 nm >OD4 CD0, CH0
CI 912.0 nm 15.0 nm X-ray to 1200 nm >OD5 C0
CII/C2 514.7 nm 3.0 nm OD4 complete C+, C2 0
HeII 468.6 nm 2.0 nm UV to 1200 nm >OD4 He0
Table 1: Primary optical filters used in DiMES TV for the PPI
experiment.
2.3.4 Thermal Imaging
Infrared emission from the graphite-covered walls of DIII-D provide
a means to
directly measure the tile surface temperature and, using a
numerical solution of a differential
equation for 1-D heat transport into a semi- infinite solid, the
heat flux impinging on the first
wall of the machine. A total of five infrared cameras are used; two
cameras viewing 20°
toroidal sections of the lower and upper divertors, one viewing a
section of the lower divertor
separated by 105° toroidally from the first (165° toroidally), and
two viewing the centrepost
[48,50,51,52]. Data are acquired at 30 FPS and processed using the
digital automated
processing system (DAPS) video tools.
Without locked modes and the presence of a resonant magnetic
perturbation to induce
an ELM-free condition, the measured surface temperature and heat
flux profile is a smoothly
varying function across the divertor region, with obvious peaks at
the target locations.
Localized heating at graphite tile edges caused by gaps between
adjacent tiles is observed
with peaking of surface temperature by ~2X, leading to
significantly more chemical erosion
compared to the rest of the tile plasma-facing surface [53].
In the 2007 DIII-D operations campaign, a substitute IR camera was
installed at 165°
R+2 and centred on R=1.5 m (with the DiMES port at 150° in the
view) while on loan from
TEXTOR. The instrument was a Santa Barbara Focalplane (SBF) model
125 camera
incorporating a 320x256 array of 30 μm indium antimonide (InSb)
pixels [54]. Spatial
14
resolution of the measured data was ~1 cm. The camera is capable of
capturing 14-bit images
at a frame rate of up to 400 Hz with sensitivity at from ~3 to 5.3
μm.
2.3.5 High Resolution Spectroscopy
High resolution spectroscopic study and characterization of emitted
light has been
essential in analyzing the quality and performance of fusion
plasmas. Spectroscopy can be
used to efficiently identify elements present in the plasma (i.e.,
through radiation wavelength
and intensity characteristics), enabling the determination of
impurity influx rates from walls,
limiters, and the divertor target plates. Spectroscopy is also used
in the determination of
impurity concentrations, nimp., investigations of transport
properties and processes, and
determination of plasma parameters including Ti from Doppler
broadening, vrot/vdrift from the
Doppler shift, ne and Te from emission line intensity ratios, and
the strength of the local
magnetic field from splitting of spectral lines.
The MDS diagnostic on DIII-D is a high-resolution McPherson model
209, 1.33 m
focal length, f/11.6 spectrometer using a 20 μm wide entrance slit
and a 1200 G/mm grating
blazed at 750 nm [55]. The MDS views the vessel from four
locations; see Figures 2 and 3.
In the lower-viewing MDS chord array, there is a chord directly
viewing the centre of the
DiMES port (150°, V-1) and two additional chords centred at the
DiMES radius (R=1.486
m), but displaced ~10° in both the CW and CCW directions from the
DiMES. Spatial
calibration of chord positions is verified using an American Optics
(AO) model 11-80, 1.5A
fibre optic illuminator to backlight each optical fibre.
Two separate camera systems have been used on the MDS diagnostic in
the past 12
years, both based on 2-D CCD detectors especially suited for
applications involving high-
speed, low-light level detection. The CCDs used for the MDS are of
a frame-transfer (FT)
design, including both active‘ pixels exposed to light, and an
equivalent number of masked
pixels (the readout/storage area) where light is rapidly shifted
after each period of exposure is
complete. This design allows simultaneous exposure of a frame of
pixels in the active area
and digitization of a stored (previous) frame of pixels in the
readout area. Used extensively in
scientific applications, FT CCDs are valuable when the required
integration time is of the
same magnitude as the onboard analog to digital convertors (ADC)
digitization time. In the
15
case of the MDS, short integration times of 100–200 ms are
typically used, providing ~40-50
frames of data collection during a typical plasma shot on
DIII-D.
From May 1996 to September 2006, the instrument utilized a Wright
Instruments
770576 active pixel camera based on the EEV CCD05-20-0-256 FT CCD.
In November
2007, a Princeton Instruments (PI) VersArray 1024 BFT camera based
on the E2V CCD47-
20 back-lit frame transfer CCD was installed to replace the aging
Wright Instruments camera.
Technical details and analysis of detector noise, pixel binning,
system gain, and bias, dark,
and flat- field frame imaging calibration techniques for both
cameras are given in [56]. The
CCD chip in the new camera has square pixels, 13 μm on a side,
compared with 22.5 μm on
a side in the Wright camera. The Wright camera had ~30% wider
useful spectral range due to
its detectors‘ larger size, but the PI camera has proportionally
finer dispersion.
The new PI camera exhibits significantly improved quantum
efficiency (Qe) (i.e., the
ability with which the CCD converts incoming photons to electrons
within each pixel well),
increased ability for binning of viewed optical tracks, and much
decreased dark current and
system gain (i.e., detected electrons per recorded digital count).
As a result, the MDS with
the new camera has nearly double the resolving power and ~24 less
illumination is required
to produce the same SNR at minimum detectability of the camera
(i.e., SNR=1.0). In
addition, new fibre optics and a new collection lens were installed
to replace those which had
been subject to 9 and 14 years, respectively, of radiation damage
and collection of
contaminants (e.g., dust, dirt, and deposits). As a result of all
improvements made to the
system, its sensitivity when viewing the 430 and 514 nm regions was
increased by a factor of
~100 compared to 2005, allowing study of molecular band emissions
in DIII-D at a level
unprecedented in the history of the MDS diagnostic : Figure 5 shows
the ratio of the
reciprocal sensitivity, k, for the DiMES-viewing chord of the MDS
in each year over that of
2007. The improvement in 2007 is especially prominent in the blue
region (<500 nm) due
mainly to increased Qe in the back- lit PI camera compared to that
of the Wright camera.
A comprehensive, historical, component-based description of the MDS
system,
performance analysis for all its components, and a synthetic model
for the optics that connect
the spectrometer to the vacuum vessel can be found in [57]. Data
are presented indicating
ideal transmission efficiency and temporal transmission losses
inferred from experimental
calibrations which have been incurred in the system as a
consequence of aging and radiation
16
damage over time. Ideal transmission of light directed into the
observable view cone (f/#
11.6) of the system to the CCD detector used in conjunction with
the MDS is estimated to be
0.166% in 2008. Attenuation of light due to accumulation of dust,
dirt, and deposits from in-
vessel activities (e.g., glow discharge cleaning, boronization, and
high temperature baking) is
theorized to be profoundly important for transmission calculations,
as are periods when
cleanings and other procedures are carried out which improve
optical transmission. Temporal
losses without cleaning and/or replacement of components damaged by
radiation are found to
be 20–40% per year of use with dependence on the wavelength of the
spectral region of
interest. Cleaning of the 150° R+2 port window took place after the
1996 and 2001
campaigns, coinciding with significant reductions in k at those
times.
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
Wavelength (nm)
N o
r e c ip
m s
e n
s it
iv it
y
1996
1997
1998
1999
2000
2001
2002
2003
2005
Figure 5: Reciprocal MDS system sensitivity normalized by its
sensitivity in 2007 for each
previous year in which a calibration was performed.
Observational parameters at commonly used wavelength settings,
including spectral
range, dispersion, and resolving power, are shown for the MDS
instrument in Table 2 and
Table 3 for the spectrometer using the previous Wright Instruments
camera and the current
Princeton Instruments VersArray 1024BFT camera, respectively.
17
Line/band CII, CD, BD,
Minimum (nm) 425.564 459.546 508.526 652.397 697.998 904.577
Maximum (nm) 435.750 469.651 518.504 661.940 707.382 913.095
Spectral range (nm) 10.186 10.105 9.978 9.543 9.384 8.518
Dispersion (nm/pixel) 0.01323 0.01312 0.01296 0.01239 0.01219
0.01106
Resolving power 32,555 35,402 39,628 53,025 57,659 82,156
Table 2: Spectral features of the MDS with the Wright Instruments
camera (spectra spread
across 770 pixels) for commonly used wavelengths.
Wavelength setting
Line/band CII, CD,
multiplet
Spectral range (nm) 7.830 7.766 7.669 7.335 7.204 6.547
Dispersion (nm/pixel) 0.007646 0.007584 0.007489 0.007163 0.007035
0.006394
Resolving power 56,256 61,261 68,634 91,744 100,309 142,150
Table 3: Spectral features of the MDS with the Princeton
Instruments 1024BFT camera
(spectra spread across 1024 pixels) for commonly used
wavelengths.
Of particular importance for interpretation of emission line width,
and hence the Doppler
temperature of the emitting species, is the instrumental profile
(IP) of the MDS system, also
known as the point spread function (PSF); that is, the natural
spread of light due to the width
of the entrance slit, diffraction phenomena, aberrations, and the
quality, focus, and alignment
of optical components in the system. In addition, the width of the
IP is a limiting factor for
the spectral resolution of the instrument, i.e., the ability of an
instrument to separate adjacent
spectral lines in the presence of apparent broadening due to
imperfections in the instrument.
The IP for the MDS employing both CCD cameras is described and
determined using
diagnostic data shots with one or more chords viewing an
argon-filled Geissler tube in [58].
To do so, shots with data viewing a variety of central wavelengths
are analyzed together and
fit to both single and double-Gaussian functions. Similar to the
MDS with the Wright
18
camera, the IP of the MDS with the PI camera is found to be best
represented us ing a double
Gaussian function with the form
, (1)
where a1=0.95, a2=0.05, and g= 2ln1 1.2011. For the current PI
camera, the dominant
component of the resulting expression for IP has a FWHM ~20%
thinner than that of the
Wright camera (FWHM1=0.015 nm vs. 0.019 nm), but an equivalent
minor component
(FWHM2=0.045 nm). The IP for the PI camera before the CCD underwent
repair is also
inspected and found to be best represented by a non-symmetric
double Gaussian.
2.4 Chemical Erosion
CS of C is the release of CxHy molecules from the substrate caused
by impacting
energetic D+ and D0 on the exposed plasma facing surface. CS is a
complex, multistep
process with significant dependence on the surface material (type
of graphite, CFC), surface
temperature (typically 300-1200K), impinging particle energy
(1–1000 eV), particle flux, and
surface structure (i.e., annealed graphite vs. amorphous C:H,
a-C:H, layers). For ions,
impinging H energy includes the energy gained by the ion passing
through the sheath and
subsequently striking the solid surface; this is expressed as
kT5kT3kT2E eiimpact (2)
with the standard assumption of ei TT [59]. The impinging H in an
attached divertor plasma
where only ions will strike the surface is inherently more
effective at inducing chemical
erosion than a detached divertor regime where the H-flux is
primarily neutral and therefore
do not gain energy via travel through the sheath.
CS includes a thermal erosion process, as well as a component
caused by effects of
energetic ion or neutral impact in the graphite lattice. Incident
hydrogen (i.e., H or D)
19
becomes implanted in the graphite (>90% in the first ~5 nm for
100 eV H+, >99% within the
first 50 nm) and, upon thermalization, reacts with primarily sp2
graphitic planes that make up
the near-surface to form CH3C complexes in an sp3 configuration
(i.e., hydrogenation to form
lone H-C and C-C covalent bonds). This leads to the creation of
thermal methyl radicals and
methane, forming a hydrogen-rich C:H carbon substrate. CH groups
are then released due to
formation of volatile hydrocarbon species upon their interaction
with recycling thermal H
atoms [60]. The chemical reaction of energetic particles with C
atoms occurs near the peak
range of particles penetrating into the surface of the substrate.
Because of this, volatile
products of chemical reactions must diffuse through the graphite
lattice, na tural porous
structures, or grain boundaries formed by cumulative damage to the
surface [61, 62].
At room temperature, CS of graphite occurs in controlled laboratory
conditions with a
yield of approximately 1-3% at particle energies significantly
below the threshold for PS [4].
While PS of C has a threshold of ~30 eV, CS is thought to have no
kinetic threshold. MD
simulations do, however, suggest that the process of CS – i.e., the
process of creation and
desorption of CxHy from the graphitic substrate – does decrease in
yield toward a threshold in
the range of 2-7 eV [63,64]. Chemical erosion occurring at T<2
eV, then, is due solely to
erosion of a-C:H layers which remain [65]. The yield may be
enhanced significantly due to
the presence of a-C:H which is subject to thermal release
mechanisms. The peak CS yield
occurs at 600-700K, above which annealing effects begin to reduce
the yield significantly,
e.g., [66]. Because of this, operation of future fusion devices
with C armour may benefit
significantly from operation with high temperature walls (up to
T~1000K).
In some studies [67], a decrease in CS yield due to flux dependence
has been
investigated and found to be significant at >~1-3x1021 m-2s-1,
and thus for the conditions
expected in the ITER divertor. Both sputtering of the graphite
lattice and thermal release of
HC radicals are found to have similar flux dependence, indicating
that saturation of radicals
by impinging H is the cause.
The result of the chemical erosion process is the release of
primarily thermal HC
molecules in both simple (CHy) and heavy, or higher‘ HC form (CxHy,
x>1). The eroded
species may be both stable HCs, or unsaturated radicals. This is
demonstrated in data from
[4] presented in Figure 6 which show the significance of simple vs.
higher HCs from CS at
energies indicative of cold (10 eV) and hot (100 eV) divertor
plasma conditions. The
20
presence of higher HCs is found to be more significant in
attachment, and at temperatures
>600K in both conditions.
Figure 6: HC erosion yield and species composition versus surface
temperature [4].
2.4.1 Spectroscopic Indicators of Chemical Sputtering in
Tokamaks
Accurate measurement of faint emission bands requires sufficient
optical
transmission from the source to the detecting spectrometer,
sufficient resolution in the
instrument, and minimal noise sources in the detection process. If
these are available, a great
deal of information may be derived regarding processes giving rise
to the emitting molecule
or radical and the environment through which it moves. As the
simplest neutral molecular
stage before a neutral or ionized carbon atom is produced, CD is a
well studied indicator of
CS from plasma facing components (though it is not useful for study
of sublimated C under
high temperatures). While other hydrocarbon fragments do not emit
strongly in visible
wavelengths, CD is especially convenient for study in a tokamak due
to known emission
bands which extend from the near ultraviolet (137 nm) up to the
mid- infrared (5.2 m ) with
strong contributions in the visible region. Analysis and
interpretation of the 22 XA band
system of CD near 430 nm is often made due to the ease of study
with a visible/NIR region
21
spectrometer with optics optimized for visible wavelengths, and its
proximity to intense CII
and D / H spectral lines.
. ..
Wavelength (nm)
Figure 7: Spectral profile in the 430 nm region captured with the
MDS.
The 22 XA electronic system of CD has a bandhead at 431.0 nm, ~0.5
nm to
the blue of the CH bandhead at 431.5 nm (Figure 7). The (0,0) band
is the most intense
contributor to the emission system, and it corresponds to the
electron transition
ppssppss 22212221 22222
. Early studies by Gerö [68,69],
Fagerholm [70], and Garstang [71] offered spectral data for CD
alongside CH to compare
and contrast their band structure and molecular constants. Observed
electronic states in the
22 XA CD band at 431 nm are of a doublet type, indicating that the
emitter contains
an odd number of electrons (not observed in CD+ and CD2.) The
existence of double P, Q,
and R branches is evident for the ''' JJJ , 0J Q branch, 1J R
branch,
and 1J P branch. Theoretically, twenty branches are possible (ten
pairs, each pair
corresponding to a doubled level); however, eight branches have an
intensity much less
than the main‘ branches lying at nearby wavelengths. Therefore,
these eight satellite‘
branches prove to be very difficult to observe. Fine structure
doubling of the rotational
levels results in splitting of each level into c‘ and d‘ pairs and
splitting of each level
increases with rotational number [72].
22
2.4.2 Effect of Boron on Measurement of Chemical Sputtering
Boron is commonly measured as an impurity in DIII-D plasmas due to
the many BZN
procedures carried out since 1991. The presence of B influences the
measurement of
chemical erosion and also the chemical erosion yield [73,74]. The
p-branch of the BD band
overlaps and thus contaminates the 428-431 nm spectral region in a
manner not readily
distinguishable from the A-X CD/CH emission band [75]. The q-branch
of BD, however, is
relatively isolated in the spectrum at 432.5-432.7 nm. Relating the
intensity of the integrated
p-branch emission to that of the q-branch emission over time, then,
provides an experimental
basis for determining the magnitude of the BD p-branch and
effectively removing the BD q-
branch contamination of the CD/CH band in the range of 428-431
nm.
To determine the significance of BD signal in the CD band
signature, the ratio of the
integrated q-branch emission to that of the p-branch is measured in
every MDS frame for 25
plasma shots spanning 8 months of operations after the BZN that
took place in DIII-D on 16
September 2006. Applicable data are selected from shots with the
minimum number of
plasma-seconds after the BZN was completed, to nearly two
plasma-hours of operation after.
While the MDS detector in use for this period was the less
sensitive Wright camera, it was
equipped with new tangential optical fibres in the lower divertor
which had significantly
improved transmission than all other MDS chords.
Inference of the amount of contamination of the CD/CH A-X band by
BD elucidates
the importance which B has on measuring and understanding chemical
erosion in DIII-D:
The lower limit on the ratio of BD p-branch plus CD/CH emission,
over the BD q-
branch emission gives a measure of the contamination of the CD/CH
band knowing
only the size of the isolated q-branch
The trend in the value of the ratio with plasma exposure time gives
an indication of
the magnitude and duration of the effect which B has on the
chemical erosion yield
The upper limit on the ratio correlates the zero-BD contamination
limit with
measurements of B-contamination on the tile surface as measured
ex-situ
Measurements are made at the MDS tangential chord #1 (T1) viewing
~1 cm above the
lower divertor floor and through the OSP. New optical fibres were
installed for the tangential
MDS chords at 285° R-2 during the LTOA (February 2006).
23
The transition from BD-only, to BD+CD, to CD-dominated spectra in
the region of 427-
432 nm is readily apparent in plasma shots beginning the first day
of plasma operations after
the BZN, 18 September 2006; see Figure 8. In the first plasma shot
to take place, 126709, the
typical indicators of CS and surface recycling – the 430 nm A-X CD
band and 427 nm CII
line, and the 434 nm D line, respectively – are initially nearly
non-existent. Their growth as
integrated plasma exposure to the lower divertor increases is
coupled to a strong decrease in
uncharacteristically large BD q-branch intensity at 433 nm.
Figure 8: 430 nm region spectral profile as viewed by the MDS
during plasma shots at
various times after a boronization on DIII-D.
To analyze temporal trends in the intensity of the combined BD
p-branch and the CD A-
X band, pixels indicating BD p-branch peaks (i.e., those with
intensity greater than a lower
bound equal to a preset noise threshold in the 427-432 nm) are
selected in spectra from the
first plasma shot after BZN (126709, shown in Figure 8). Data from
these peaks are then
tracked over subsequent shots and related to the integrated
intensity of the BD q-branch. In
subsequent shots as the layer of B deposited in the BZN is eroded
away, data on these pixels
includes contribution from both BD p-branch and the CD/CH A-X
band.
MDS data from nearly 1000 frames, each with a 100-125 ms
integration time, are
analyzed. Integrated BD q-branch counts for each frame analyzed are
correlated with the
24
corresponding integrated BQ p-branch signal plus the integrated
signal from the overlapping
CD A-X band for each frame, and the number of plasma-seconds of
exposure since the
previous BZN is tracked. The data from each analyzed shot is found
to be well fit by a linear
trend line. A strong correlation of q-to-p branch + CD/CH emission
with time is found since
the previous BZN. The slope of the shot-specific linear trends
decreases rapidly as plasma
exposure increases from 0 to 600 seconds. In contrast, there is
very little impact on the slope
of the integrated intensity ratio as plasma exposure increased from
3700 to nearly 7000
plasma-seconds, indicating that near steady state has been achieved
in this period, and an
asymptotic effect from erosion/transport of B away from the lower
divertor targets.
The shot-specific trend line with the smallest slope is found to
correspond to the first full-
length shot in DIII-D with the new lower divertor shelf installed.
Installation of the shelf
involved the complete replacement of tiles in the lower divertor
including the three
lowermost rows of centrepost tiles, the 45° facet tile row, two
rows of the floor tiles, and four
rows of tiles on the shelf itself. There is a small but significant
difference in slope between
the linear fit for this shot and that for shots long after
(3700-7000 plasma seconds) the
previous BZN. The slope for the fresh tile‘ case remains non-zero,
however, indicating that
BD is still being measured by the MDS on the tangential chord. The
source of this BD may
be non-local, i.e., due to erosion and transport of B from the main
chamber and upper
divertor to the lower divertor.
The ratio of the integrated BD p-branch plus CD A-X band emission
to the
uncontaminated, integrated BD q-branch may then be plotted versus
the plasma exposure
time since the previous BZN in DIII-D; see Figure 9. A fit to the
data is made in the form
xcexp1baY , where a, the x=0-intercept, is equal to 2, b, the
asymptotic limit
in y, is equal to 17.5, and c, the stiffness of the fit, is equal
to 0.0008.
At the point where the fit meets 0 plasma seconds of exposure, the
function allows
experimental quantification of BD p-branch contamination in the
CD-band – i.e., BD p-
branch emission without an emission contribution from the CD A-X
band – based on the
integrated intensity of the uncontaminated BD q-branch. In the case
of data from the DIII-D
2006 campaign, the number of BD p-branch counts which should be
removed from the
integrated (and contaminated) CD A-X band is equal to 2X that of
the uncontaminated
integrated BD q-branch emission nearby. This correction allows for
greater accuracy in
25
quantification of the CD A-X emission band, and hence measurement
of D/XB for CD/CH
and chemical erosion yield.
Figure 9: BD band emission contamination ratio vs. time since the
last BZN in DIII-D.
For the PPI experiment, the magnitude of the correction for removal
of BD p-branch
contamination in the range of the CD A-X band relative to the
intensity of the CD A-X band
itself is proportional to the time between the PPI run day and the
most recent BZN in DIII-D.
This information is shown in Table 4. For shots in the run day with
minimum exposure time
between it and the previous BZN, ~400 plasma-seconds on 10/14/2004,
Figure 9 indicates
that the ratio of the integrated BD p-branch plus CD A-X emission
over the integrated BD q-
branch intensity is ~6. This leads to the need for a 1/6 * 2 * 100%
= 33% correction to the
magnitude of the measured integrated (apparent CD A-X band) counts
in the 428-431 nm
spectral region. For shots in the run day with maximum exposure
time between it and the
previous BZN, ~6,800 plasma-seconds on 08/07/2007, Figure 9
indicates that the ratio of the
integrated BD p-branch plus CD A-X emission over the integrated BD
q-branch intensity is
~17.5. This leads to the need for a 1/17.5 * 2 * 100% = 11%
correction to the magnitude of
the measured integrated counts in the 428-431 nm spectral
region.
26
First plasma shot after
previous BZN
(year installed)
MkI 20040806 119939-119942 20040626 119134 3,332.5 1992
MkI 20041014 120697-120713 20041002 120608 419.2 1992
MkI 20041022 120823-120839 20041002 120608 877.7 1992
MkI 20050217 122132-122153 20041120 121076 3,043.9 1992
MkI 20050218 122168-122207 20041120 121076 3,192.8 1992
MkI 20050323 122735-122762 20050219 122214 1,826.6 1992
MkI 20050328 122838-122872 20050219 122214 2,176.8 1992
MkI 20050414 123358-123393 20050219 122214 4,600.2 1992
None 20060908 126546-126551 20060610 124889 6,212.5 2006
MkII 20070615 129047-129065 20070317 128160 2,680.2 2006
MkII 20070719 129678-129694 20070317 128160 5,246.0 2006
MkII 20070807 130108-130140 20070317 128160 6,825.7 2006 Table 4:
History of plasma exposure and time since previous BZN for each PPI
run day on DIII-D.
2.5 Hydrocarbon Breakup and Plasma Interaction
Once released into the plasma, HCs interact in a primarily
destructive manner with the
impinging plasma. The energetic ions, neutrals, and electrons in
the tokamak edge region
have the capability to dissociate HC molecules into their H and C
components. Extensive
reaction databases for dissociation and ionization of HCs have been
produced [5,6],
incorporated into edge modelling codes including ERO, and DIVIMP,
and applied to
conditions for both natural CS processes and controlled injections
of CxHy. The complexity
of the breakup chains‘ possible is shown in Figure 10, revealing
many possible channels for
emitted HC species to evolve‘ in the plasma to C0 and C+
states.
Fragments of the breakup process will eventually return to the
walls or divertor
targets of the tokamak. The period required for this to occur may
be very small (~10 -6 s),
especially for HC species which ionize in the breakup process. Once
fragments strike the
surface, they may stick and become part of the a-C:H plasma-facing
layer (before possibly
being re-eroded after some dwell time), or immediately reflect back
toward the plasma with
some surface loss probability. Fragments that stick to the surface
lead to retention of H in the
a-C:H layers. For a device operating with a mixture of D and T,
this process will lead to
significant co-deposition of T on vessel components, leading to T
retention and a potential
radiological hazard. The probability for sticking (or the inverse,
the probability for reflection)
27
is different for each HC species, and is typically >0.1 for HC
radicals, but close to 0 for
saturated species [76,77].
Figure 10: HC breakup reactions available [5,6].
There are many unanswered questions regarding CS, especially with
regard to its
operation and ramifications in a tokamak which this thesis aims to
elucidate, e.g., how to
effectively measure the chemical erosion yield in-situ, and the
efficiency with which CS
leads to impurity contamination of the plasma.
Chapter 3: Experimental Design
3.1 The PPI Concept
The self-contained Porous Plug gas Injection (PPI) system provides
a means to
introduce a controlled, locally-confined influx of gas molecules
(HCs or other species) into
divertor or limiter plasmas in the DIII-D tokamak to replicate the
influx of HCs by chemical
erosion in a well calibrated and controlled manner. The molecular
influx rate was selected to
be comparable to that resulting from hydrogenic impact on graphite
with chemical erosion
yields, C
chemY , as measured in the laboratory, i.e., 1%–3% of the
hydrogenic flux to the surface
[4]. The typical ion flux to the floor, D
recorded in previous Simple-As-Possible-Plasma
28
(SAPP) experiments in DIII-D under attached plasma conditions was
sm/10x3 222
D
[38]. If the total influx of C-atoms in all chemically released HCs
were injected as CH4, a
flux of holedD
injection area provides adequate signal for spectrometer
measurements even at low HC flux
densities.
While the release of chemically sputtered material from a graphite
first-wall is known
to include CxHy species with x=1 to 3 [4], methane (CH4) is
currently used as a proxy for the
mixture of simple and higher HCs by the PPI. This is done because
reaction chains involving
electron and ion impact with HC species are complex and reaction
characteristics, especially
when excited species are involved, are not well known [5,6], and
thus are more difficult to
interpret or model. In future experiments, a mixture of
hydrocarbons or other gases may be
used with the PPI. Similarly, an isotopically enriched 13CH4
methane gas can also be used for
post-exposure study of deposition patterns with surface analysis
techniques. Helium was
included in the injected gas (10-20% by weight) to facilitate
monitoring of the actual gas
flow rates, since the excitation rates of neutral He0 and ionized
He+ (i.e., emission of HeI and
HeII lines) are cataloged in the Atomic Data and Analysis Structure
(ADAS) database [13].
3.2 Probe Design and Construction
3.2.1 The PPI Probe Head
The PPI assembly is composed of two primary in-vessel components: A
hollow
DiMES probe head with a porous graphite face positioned flush with
the DIII-D floor, and
the gas source assembly. The latter consists of a pressurized gas
reservoir, a gauge measuring
absolute pressure in the reservoir, and a means to control gas
input into the vessel. All gas
assembly components are located in the space beneath the DiMES
sample mount (Figure 1).
The porous cap of the PPI is made from a disk of isotropic
fine-grain graphite 4.2 cm
in diameter and 1.5 mm thick. The thin nature of the porous cap was
necessary to make
drilling possible, but means that the PPI is approved only for use
in low power plasmas in
DIII-D (i.e., L-mode or ohmic plasmas). A regular square lattice of
1004 holes of 0.25 mm
diameter and 0.81 mm spacing was mechanically drilled in the cap
over a circular area 3 cm
29
in diameter; see Figure 11. Though other drilling techniques were
attempted, none provided