close

Вход

Забыли?

вход по аккаунту

?

Optical emission spectroscopy and effects of plasma in high power microwave pulse shortening experiments

код для вставкиСкачать
INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI films
the text directly from the original or copy submitted. Thus, some thesis and
dissertation copies are in typewriter face, while others may be from any type of
computer printer.
The quality of th is reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality illustrations
and photographs, print bleedthrough, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and continuing
from left to right in equal sections with small overlaps.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations appearing
in this copy for an additional charge. Contact UMI directly to order.
Bell & Howell Information and Learning
300 North Zeeb Road, Ann Arbor, Ml 48106-1346 USA
800-521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
OPTICAL EMISSION SPECTROSCOPY AND EFFECTS OF PLASMA IN HIGH
POWER MICROWAVE PULSE SHORTENING EXPERIMENTS
by
William Erwin Cohen
A dissertation subm itted in partial fulfillment
o f the requirem ents for the degree o f
D octor o f Philosophy
(N uclear Engineering)
in The University o f Michigan
2000
Doctoral Committee:
Professor Ronald M. Gilgenbach, Chair
Associate Professor M ary L. Brake
Associate Professor Brian E. Gilchrist
Professor Yue Y. Lau
Dr. Thomas A. Spencer, Nuclear Engineer, USAF
A ir Force Research Lab
Phillips Research Site
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number. 9977136
Copyright 2000 by
Cohen, William Erwin
All rights reserved.
UMI
UMI Microform9977136
Copyright 2000 by Bell & Howell Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
Bell & Howell Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
William Erwin Cohen
All Rights Reserved
2000
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In dedication to m y parents, wife, family, and friends,
for all their support and unconditional love.
ii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ACKNOWLEDGMENTS
First o f all, I would like to thank my advisor, Professor Ronald Gilgenbach. I am
thankful for his support and encouragement throughout my tenure as a graduate student at
the U niversity o f Michigan. I especially appreciate his trust and confidence in me. I am
thankful for the opportunities that he has made available to m e for professional growth
and developm ent. I know that I am a much better researcher and experimentalist because
o f my interaction with him. This dissertation would not have been completed without his
constant encouragement.
As well, I’m thankful for having an advisor that always
acknowledges the hard work o f his students to colleagues. I’ve enjoyed the interaction
that this has yielded for me in the high power microwave com m unity.
I thank Professor Y. Y. Lau for teaching me a great deal about Plasma Physics.
I’m thankful for the opportunity to have interacted with such an extraordinary theorist. I
would also like to thank him for helping to fund my graduate education by allocating
AASERT funding for me through the Air Force Office o f Science Research.
I thank Dr. Tom Spencer for the data acquisition software used for taking the data
in this dissertation. I appreciate his many helpful comments concerning this dissertation.
I thank Professor M ary Brake for help with the spectroscopy o f the RF plasma discharge
used for the RF cleaning experiments presented in this dissertation.
I appreciate her
optimism and encouragement through the years. I thank Professor Brian Gilchrist for
taking tim e out o f his busy schedule to be on my committee and his helpful comments on
this dissertation.
I thank M ark Perreault for fixing problems for me during crises, helping me run
experiments w hen I was in a bind for additional assistance, and all the interesting
iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
conversations. I appreciate help from Ron Spears on electrical devices for the experiment
and all his w ork on MELBA.
I am thankful for all the help that I’ve received over the years from Pam Derry,
Helen Lum, Ann Bell, Wendy Derby, Rhonda Sweet, Liz Tompkins, Zonda Ketola,
Diana Corey, M ary Ann Marshall, and Sue Greenwood. I ’m glad to have known Karen
Balson and always appreciated her help.
I thank Ed Birdsall for his help with the
computers in the Plasma Bay.
I thank Dr. Josh Rintamaki for his help and encouragement during my
undergraduate and graduate years at the University o f Michigan.
I thank Dr. Scott
Kovaleski for his encouragement and sense o f humor. I thank Dr. Jonathan Hochman for
his work in setting up the magnetic cusp used in this research. I appreciate his assistance
on his experim ent such that I could m easure optical emission spectroscopy from the RCS
gyrotron. I thank Dr. Reggie Jaynes for his work in setting up the magnetic cusp and the
initial setup o f the coaxial gyrotron used in this research. I am thankful for his help in
running experim ents and the interesting conversations.
To Chris Peters, I appreciate his help running experiments and work with TimeFrequency Analysis. I wish you the best o f luck finishing up your graduate education. I
thank Mike Lopez for his help running experiments and the interesting conversations.
Thanks for your sense o f humor and good luck with the relativistic magnetron. To Scott
Anderson, I appreciate all his help running experiments and his encouragement on the
more difficult days. Thanks for the tim e you got me lunch so that I could keep running
my experiment. I wish you luck with everything. I thank Rex Anderson for all his help,
interesting conversations, making us coffee to stay awake, and ju st being such a
delightful person. I’m glad that you moved your experiment up to the Plasma Bay, so
that we had the opportunity to work together and get to know each other better. To Mark
Johnston, thanks for your help and good luck with your project. To Bo Qi, good luck
with your project and thanks for your daily humor.
iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I thank Antwan Edson for his help doing both the fun jobs and the more dull
tasks. I appreciate the times you stayed late to help me get things done when you knew I
would be doing it by m yself if you didn’t. To Nick Eidietis, thanks for your help running
experiments and enthusiasm on some o f the more thankless tasks around the lab. I thank
Mark Porter for his help running experiments.
I enjoyed our conversations.
I thank
Justin Benoit for his help and his ability to quickly learn some o f the tasks involved with
running M ELBA which enabled me to continue running experiments late into the day in
order to com plete several o f the data sets presented in this dissertation.
To Paul and Megan Grekowicz, thanks for your friendship and support over the
years. I appreciated having a place to stay in town on the late nights o f working on this
research. I am thankful for the great times w e’ve had together since we met.
To Cemal Sozener and Erin Booth, I am thankful for two great friends. The days
that I needed to work late on some o f the more difficult problems I encountered during
this research were made tolerable by our dinner conversations. Cem al, I appreciate the
more than several times you let me stay at your place when it was too late to drive home
to Battle Creek.
Thank you to all who I have had the pleasure o f interacting with during my
involvement with the Michigan Marching Band. I know that I am a better person because
o f it. I thank Mr. Don Shepherd for his friendship and support.
I appreciate all that my past teachers and professors have provided me with
throughout my life. My ability to achieve is a direct reflection o f their efforts.
I appreciate all that my w ife’s family has provided for me over the years. With
my family being far away, they have all been a positive and important part o f my life. I
appreciate the support and encouragement o f my wife’s parents, Bill and Karen Thumer.
I appreciate the love and support o f my family throughout m y life. They have
been the foundation for me. I am thankful for all the encouragement from m y sister and
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
her husband over the year. I thank m y parents, Herb and Kularb Cohen, for their endless
love and support. I appreciate them letting me find my own path in life and being there
to help and encourage me along the way.
All the thanks in m y heart go to my wife, Laura. Her love and support mean the
world to me.
I know that I’m truly thankful and blessed to have her in my life.
I
especially appreciate her understanding o f the late hours doing research, performing data
analysis, and writing this thesis.
This work has been supported by the Air Force Office o f Scientific Research
(AFOSR) H igh Power M icrowave M ultidisciplinary University Research Initiative
(MURJ) program contracted through Texas Tech University and by Air Force Research
Lab - Phillips Research Site and Northrop Grumman Corp.
I am thankful for the
additional funding provided for my graduate education through an AFOSR-AASERT.
vi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
TABLE OF CONTENTS
DEDICATION........................................................................................................................... ii
ACKNOWLEDGMENTS......................................................................................................iii
L IST O F F IG U R E S ....................................................................................................................... ix
LIST OF TABLES.................................................................................................................xiii
L IS T O F A P P E N D IC E S ............................................................................................................xiv
CHAPTER
1.
INTRODUCTION.................................................................................................. 1
1.1 P ream ble.............................................................................................. 1
1.2 High-Power Microwave Pulse Shortening R esearch.................. 2
1.3 Current Experimental W ork............................................................. 3
2.
B A CK G R O U N D T H E O R Y ..................................................................................... 5
2.1 Gyrotron Dispersion Relation.......................................................... 5
2.2 Cause o f Plasma in High Power M icrowave Sources................. 6
2.2.1 Background Gas Ionization............................................. 6
2.2.2 Electron Beam Impact and Scraping.............................. 8
2.3 Effect o f Plasma on the Dispersion R elation ................................9
2.4 M echanisms for Plasma G row th....................................................12
2.4.1 Plasma Critical Density...................................................12
2.4.2 Electron Cyclotron Resonance (ECR) Heating
12
2.4.3 Upper Hybrid Resonance H eating ............................... 14
2.5 RF Plasm a Sputtering .....................................................................14
2.6 Adsorption o f H 2O M olecules........................................................ 15
2.7 Paschen Curve for High Frequency D ischarges........................ 16
3.
EXPERIMENTAL CONFIGURATION AND DIAGNOSTICS...............17
3.1 MELBA Accelerator and D io d e....................................................17
3.2 Electron Beam Voltage and Current D iagnostics...................... 19
3.2.1 MELBA Voltage M easurem ent....................................19
3.2.2 Diode Current................................................................... 21
3.2.3 Aperture Current, Cavity Entrance Current, and
Cavity Exit C urrent......................................................... 22
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3
3.4
3.5
3.6
3.7
Diode and Electron Beam Transport M agnetic Fields..............23
MELBA Triggering Sequence...................................................... 24
Electron Beam Extraction and Glass Witness P la te .................26
Microwave Cavity Structure and Cold T e sts............................. 28
Microwave Extraction and D etection.......................................... 30
3.7.1 M icrowave Diode Detector........................................... 31
3.7.2 Heterodyne Mixing and Time-FrequencyAnalysis.............................................................................32
3.8 Optical Emission Spectroscopy D iagnostics..............................32
3.9 RF Plasma Cleaning Equipm ent................................................... 34
3.10 Residual Gas Analyzer (RGA)................................................... 36
3.11 RF Tuned Langmuir P robe.........................................................38
3.12 RF Cleaning Plasma Characterization..................................... 41
4.
EXPERIMENTAL METHODS AND RESULTS........................................ 45
4.1 Optical Emission Spectroscopy o f P lasm a.................................45
4.2 Correlation o f M icrowave Emission and Plasm a.......................48
4.3 Improvements in Total Microwave Energy E m ission..............52
4.3.1 RF Plasma Cleaning........................................................54
4.3.2 Gas B ackfilling............................................................... 72
4.4 Microwave Pulse Shape and Voltage Fluctuations................... 78
5.
CONCLUSIONS................................................................................................. 80
APPENDICES.........................................................................................................................82
BIBLIOGRAPHY.................................................................................................................. 94
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF FIGURES
Figure
2.1
W aveguide dispersion relation showing the beam and structure m o d e s..........6
2.2
E-beam impact ionization o f argon, nitrogen, and water vapor for t = 0
to 700 n s........................................................................................................................9
2.3
Plot o f ei/e 0 (X) and zi/z0 (•) versus plasm a density.......................................... 11
2.4
Paschen curves for high frequency breakdown o f air and a rg o n ......................16
3.1
Large-orbit, coaxial gyrotron experimental configuration with axial
m agnetic field profile................................................................................................18
3.2
(a) V oltage pulse and (b) diode current produced by the M ELBA
Marx B a n k ................................................................................................................. 20
3.3
Calibration setup for the voltage m o n ito r............................................................ 21
3.4
Experim ent setup for Rogowski coil calib ratio n ................................................ 23
3.5
M agnetic field profile for diode magnetic field coil charging o f 1.08
kV and solenoid magnetic field coil charging o f 200V .................................... 25
3.6
Glass W itness Plate o f the e-beam used to drive the coaxial gyrotron
used in this experim ent............................................................................................27
3.7
Diagram o f the coaxial cavity and waveguide setup with dim ensions.......... 29
3.8
Fiber optic bundle setup (center line and acceptance angle o f fiber
optic bundle show n)................................................................................................. 33
3.9
Schem atic o f the RF cleaning plasm a................................................................... 34
3.10 RF plasm a cleaning equipment s e tu p ................................................................... 36
3.11 Typical RGA trace o f the background gas in the HPM coaxial gyrotron......37
3.12 Setup for RF tuned Langmuir probe experim ent................................................39
3.13 I-V characteristics for RF tuned Langm uir probe position one and tw o .......41
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.14 Plot o f squared probe current versus probe voltage for (a) probe
position one and (b) probe position t w o ..............................................................42
3.15 Ion density o f the RF cleaning plasm a versus radial position in the
coaxial w aveguide from RF tuned Langm uir probe m easurem ents.............. 43
3.16 Spectra o f N 2 1st and 2nd positive system s for the RF cleaning plasm a........ 44
3.17 Spectra o f the N 2 2nd positive system for the RF cleaning p lasm a................ 44
4.1
Optical em ission o f H-alpha (0.75 m Spectrograph with ICCD gated
for 1 p s to capture entire voltage pulse, ICCD gain=10.0)..............................46
4.2
Optical em ission o f H-beta (0.75 m Spectrograph with ICCD gated for
1 ps to capture entire voltage pulse, ICCD gain= 10.0)....................................47
4.3
Tem porally resolved optical em ission spectroscopy o f H-alpha shown
with M ELB A voltage and microwave pow er signal from the
m icrow ave diode detector.......................................................................................48
4.4
Plot o f M ELB A voltage, microwave pow er diode signal, PMT signal
(H-alpha), and heterodyne m ixer signal versus tim e ........................................ 50
4.5
Plot o f H -alpha optical emission start tim e versus microwave power
cutoff tim e (average MELBA voltage start tim e subtracted from x- and
y- axis) Base vacuum pressure was ~ 8 x l O'6 Torr on March 6 , 2000
and -7 x 1 O'6 Torr on March 8 , 2000...................................................................... 51
4.6 Plots o f shot-averaged microwave pow er and H-alpha optical emission
signal traces versus time for (a) 30 shots from March 6 , 2000 and (b)
18 shots from M arch 8 , 2 0 0 0 ................................................................................. 53
4.7
a) Peak m icrow ave power and b) m icrowave energy plotted versus
m icrow ave pulselength for data taken on M arch 8 , 2000. Base
vacuum case pressure was —7 x l0 '6 Torr. RF cleaning shots were taken
at - l x l O ' 5 Torr. Post-RF cleaning case pressure was ~ 7 x l0 ‘6 T orr.............. 58
4.8 Shot-averaged (a) microwave power and (b) H-alpha optical emission
for M arch 8 , 2000. 1) Base vacuum case (18 shots, -7x1 O'6 Torr), 2)
RF plasm a cleaning (20 shots, - lx l O '5 Torr during shot, alternating
between regular shot and cleaning), and 3) Post-RF cleaned (6 shots,
—7x1 O’6 T orr)..............................................................................................................59
4.9
a) Peak m icrow ave power and b) microwave energy plotted versus
m icrow ave pulselength for data taken on M arch 15, 2000. Base
vacuum case pressure was —5x1 O'6 Torr. RF cleaning shots taken at
- lx l O ’5 Torr. Post-RF cleaning case pressure was - 5 x 1 c 6 T o r r ................62
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.10 Shot averaged (a) m icrow ave power and (b) H-alpha optical emission
for March 15, 2000. 1) Initial base vacuum case (15 shots, -5x1 O'6
Torr), 2) RF plasm a cleaning (14 shots, - lx lO ’5 Torr during shot, each
shot), 3) Post-RF cleaned (15 shots, -5 x 1 0 ^ Torr)...........................................63
4.11 a) Peak microwave pow er and b) microwave energy plotted versus
microwave pulselength for data taken on M arch 28, 2000. Initial base
vacuum case pressure was ~ lx lO '5 Torr. Base vacuum case pressure
(after breaking vacuum ) was - 1 .7xl0 ‘5 Torr. RF cleaning shots taken
at 1.3xl0 '5 to 2 .3 x l0 ' 5 Torr. Post-RF cleaning case pressure was
-9x1 O'6 T o rr............................................................................................................ 66
4 . 12 Shot-averaged (a) m icrowave power and (b) H-alpha optical emission
for March 28, 2000. 1) Initial base vacuum case (16 shots, - lx lO ' 3
Torr), 2) base vacuum case (after system purge to lab air, 4 shots,
1.7xl0 '5 Torr), 3) RF plasma cleaning (17 shots, 1.3xl0 '5 to 2.3xl0 *5
Torr, each shot), 4) Post-RF cleaned (15 shots, -9X 10-6 T o rr)......................67
4.13 a) Peak microwave pow er and b) microwave energy plotted versus
microwave pulselength for data taken on M arch 29, 2000. Base
vacuum case pressure was ~ 1.3xl0 ‘5 Torr. Post-RF cleaning case
pressure was lxlO *5 to 7.4x1 O'6 T o rr..................................................................70
4.14 Shot-averaged (a) microwave power and (b) H-alpha optical emission
for March 29, 2000. 1) Initial base vacuum case (15 shots, -1.3x10 ‘5
Torr) and 2) Post-RF cleaned case after 1.5 hours o f RF cleaning (12
shots, lxlO '5 to 7.4x1 O'6 T o rr).............................................................................71
4.15 Peak microwave pow er versus microwave pulse length summary o f
RF plasma cleaning results. Data plotted are the averages o f the
individual case averages examined. Pulse shortening curves
(Power=constant/time) are shown for each case................................................72
4.16 Summary o f m icrow ave energy production for each backfill case
examined. The standard deviation is included for each average.
Arrows below graph indicate purge o f experiment to atmospheric
pressure with a ir....................................................................................................... 74
4.17 Summary o f peak m icrow ave power for each backfill case examined.
The standard deviation is included for each average. Arrows below
graph indicate purge o f experiment to atmospheric pressure with a ir.........75
4.18 Shot-averaged microwave power signals for the a) Base vacuum case
(5 shots, -4 x 1 0 ^ Torr), b) Argon backfill case (28 shots, 1.6xl0 "5
Torr), and c) SF 6 backfill case (28 shots, 1.6xl0 '5 Torr) from February
16, 2000...................................................................................................................... 76
xi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4.19 Shot-averaged microwave power signals for the a) Base vacuum case
(30 shots, -4x1 O'6 Torr) and b) SF 6 backfill case (30 shots, 1.6x1 O' 5
Torr) from February 18, 2000 ................................................................................77
4.20 MELBA voltage and microwave power signal averaged over 30 shots...... 79
A .l
Heterodyne m ixer schematic for time-frequency analysis..............................84
A.2
Optical emission spectroscopy schematic for Intensified CCD data............84
A.3
Optical emission spectroscopy schematic for photom ultiplier tube data.....85
B. 1
MELBA triggering schem atic.............................................................................. 86
D. 1
Picture o f the University o f Michigan experiment for argon RF plasma
discharge cleaning o f SLAC nose pieces.............................................................89
D.2 University o f M ichigan SEM pictures o f SLAC nose piece area alpha
for base condition, 50 W o f argon RF plasma cleaning, and then 100
W o f argon RF plasm a cleaning. Cleaning was perform ed for 20
minutes at a discharge pressure o f -5 m Torr......................................................91
D.3
University o f M ichigan SEM pictures o f SLAC nose piece area beta
for base condition, 50 W o f argon RF plasma cleaning, and then 100
W o f argon RF plasm a cleaning. Cleaning was perform ed for 20
minutes at a discharge pressure o f ~5 m Torr......................................................92
xii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF TABLES
Table
3.1
Calculated TE Cavity M odes................................................................................. 30
4.1
Sum m ary o f RF plasm a cleaning study data sets................................................57
xiii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
LIST OF APPENDICES
Appendix
A.
D A TA ACQUISITION SETUP (FIN A L )........................................................... 83
B.
M ELBA TRIGGERING SCH EM A TIC............................................................... 86
C.
M A TLA B SUBROUTINE FOR ANALYSIS OF ROGOWSKI
CO IL S IG N A L S ......................................................................................................87
D.
SCA N N IN G ELECTRON M ICROSCOPE ANALYSIS OF RF
PLA SM A CLEANED COPPER FROM UNIVERSITY OF
M ICHIGAN/SLAC CO LLABORATION.......................................................... 88
E.
AN A LY SIS OF VARIANCE (A N O V A )............................................................ 93
x iv
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 1
INTRODUCTION
1.1 Preamble
The immense improvement in achievable pow er levels in high power microwave
(HPM) devices over the past decade has been met with a reduction in the achievable
microwave pulse length. This phenomenon is known as microwave pulse shortening and
generally holds the radiated energy at more or less a constant as power is increased
[AGE98]. The physical causes o f pulse shortening were grouped by Benford into four
basic types [BEN97]. They are plasma generation, electron streaming, high-electric-field
breakdown, and beam disruption.
This thesis is an experimental study o f the relation o f microwave pulse shortening
with the plasma generated (first physical case cited above) inside an operating high
power microwave device.
These HPM experiments were conducted in the Intense
Energy Beam Interaction Laboratory at the University o f Michigan. The multi-megawatt,
large-orbit, coaxial gyrotron used in these experiments was driven by the Michigan
Electron Long Beam Accelerator (MELBA) at parameters: V= -800 kV, Icaihode= 6 kA,
Itubc= 0.8 kA, and pulselengths o f 0.5-1.5 ps. Pulse shortening effects on some HPM
devices are not apparent because most intense electron beam accelerators produce a
pulselength o f a few hundred nanoseconds or less. M ELBA incorporates an Abramyan
circuit [ABR77] to flatten the output voltage for up to 1.5 ps, which makes the
accelerator ideal for studying microwave pulse shortening effects. The subject o f this
thesis includes the use o f RF plasma cleaning o f the cavity/waveguide structure to reduce
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
the quantity o f w ater vapor on the structure surface in order to mitigate observed
microwave pulse shortening.
1.2 High-Power Microwave Pulse Shortening Research
The issue o f m icrowave pulse shortening has becom e an immense area o f research
during the past five years. The focus o f research on m icrow ave pulse shortening in the
United States was a result o f the Air Force Office o f Scientific Research sponsored
M ultidisciplinary University Research Initiative (M URI) program on High Power
Microwaves involving nine universities and the Air Force Research Laboratory [AGE98].
The 1998 IEEE Transactions on Plasma Science Special Issue on High Power M icrowave
Generation (June 1998, Vol. 26, Number 3) includes a section devoted to microwave
pulse shortening papers.
M icrowave pulse shortening has been studied in all m ajor types o f HPM devices,
including Gyrotrons [GIL98], Backward-Wave Oscillators (BW Os) [GRA98,GOE98b],
Relativistic M agnetrons [PRI98], Reltrons [MIL98], PASOTRON [GOE98], Travelling
Wave Tubes (TW Ts) [GOE98b], and a M agnetically Insulated Transmission Line
Oscillator (M ILO) [AGE96, HAW98], as well as work related to a variety o f HPM
devices [PRJ98b].
Diagnostics to investigate the behavior o f plasma inside operating
HPM devices have included plasm a characterization with optical emission spectroscopy
[GIL98], electron density measurements with laser interferometry [HEG98], and plasma
monitoring in a slow wave structure (SWS) with optical photodetectors [GOE98]. Papers
which survey m icrow ave pulse shortening research have also been written [BEN97],
[AGE98]. State o f the art HPM devices (e.g. Air Force - M ILO) can produce microwave
energies on the order o f 250 J [AGE98].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
Most m icrow ave pulse shortening research regards plasm a generated in the
operating HPM device as deleterious and undesired. However, plasm a has been found to
be beneficial in som e devices. These include plasma-filled backward-wave oscillators
(BWOs) [CAR89], [GOE99], [GOE98], [GRA98], [NUS98], [SHK98] and plasma-filled
travelling wave tubes (TW Ts) [NUS98], [KOB98], [SHLOO], [POI 8 8 ]. Optical emission
spectroscopy has been performed inside a plasma-filled backward-wave oscillator and
used to measure the electric field inside the slow wave structure [ZHA94].
The
references listed represent a sample o f the current and past research on plasma-filled
devices. They contain extensive references on this subject.
1.3 Current Experimental Work
In this dissertation plasm a H-alpha line radiation is measured inside the
microwave cavity and electron beam collector via fiber optics with a 0.275 m
m onochrom ator/photom ultiplier tube and correlated with output microwave power. The
temporal correlation between the reduction o f output microwave pow er and growing
H-alpha optical em ission is measured. A strong correlation suggests that the plasma is
reducing the output microwave pow er as the plasma approaches critical density
(~ 8 xl O10 cm '3). Initial results o f this research performed on a rectangular-cross-section
(RCS) gyrotron were published in [GIL98], the first optical emission spectroscopy
experiments perform ed in an operating gyrotron.
RF plasm a cleaning is examined on the coaxial cavity and e-beam collector to
determine its effect on the microwave pulse shortening characteristics o f this gyrotron
device. The RF plasm a discharge is expected to produce ions that will sputter excess
water vapor from the cavity/waveguide structure.
The liberated w ater vapor is then
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
removed from the system by vacuum pumps. No previous work on RF plasma cleaning
o f HPM devices is known.
Following this introduction, this dissertation contains four more chapters.
Chapter 2 is devoted to background theory used in this research.
Chapter 3 explains
configuration and diagnostics for the experiments. Chapter 4 deals with experimental
methods and results. Chapter 5 provides a summary o f the experiments and conclusions.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER 2
BACKGROUND THEORY
2.1 Gyrotron Dispersion Relation
A detailed explanation o f the cyclotron resonance interaction which drives the
gyrotron m echanism is given in [HOC98] which utilizes m any excellent references. The
dispersion relation for the operating conditions used for the coaxial gyrotron in this
experiment is given in Figure 2.1.
The dispersion relation graphically shows what
conditions are required for the e-beam to couple to the w aveguide modes.
The
waveguide cutoff frequency for the T E t i mode is 2.286 GHz for this coaxial gyrotron.
The grazing condition is when the e-beam cyclotron wave line is tangential to the
waveguide part o f the dispersion relation and therefore has a single intersection. This
represents the condition o f strongest interaction between the e-beam and the waveguide
mode [LAU97]. The expected frequency o f operation is ~2.6 GHz. The main mode o f
oscillation for this coaxial gyrotron is the TEn? mode [JAYOO]. The operating frequency
as measured by heterodyne mixing is -2 .5 5 GHz.
The modes shown are derived in Chapter 3 o f this dissertation.
The axial
magnetic field in the interaction region is 0.15 Tesla. The e-beam cyclotron wave line for
an axial m agnetic field o f 0.16 Tesla is shown for comparison. The magnetic field used
in the experim ent places the beam line relatively close to a grazing intersection with the
waveguide mode.
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6
F [GHz]
CO c u to ff
t e
115
TEi
14
TE„3
t e II2
TEm
CObcam
V =
k z V oz
COcyc
C
0 . 1 5 T e s la !
-20
40
60
Figure 2.1. W aveguide dispersion relation showing the beam and structure modes.
2.2 Cause of Plasma in High Power Microwave Sources
Plasm a can cause shortening o f the expected microwave output power anywhere
it appears [BEN97].
The plasma can change the resonance condition o f the e-
beam /oscillator interaction due to its presence inside the m icrowave cavity. A plasma
approaching cutoff density will reflect or absorb energy from the microw aves and further
the growth o f the plasm a density (Section 2.4.1). The plasma generation mechanisms
investigated in this work are e-beam im pact scraping and background gas ionization.
2.2.1 Background Gas Ionization
Ionization collisions occur in the beam tubes o f accelerators as a source o f plasma
since it is not possible to obtain a perfect vacuum [REI94]. Assum ing that ionization
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
occurs only from e-beam impact, the change in the gas molecular density is given by
Equation 2.1,
dn„
- ± = - n bnga iV
( 2 . 1)
where ng is the gas density, nb is the beam density, a; is the ionization cross section, and v
is the velocity o f the beam [REI94].
The ionization cross sections o f atoms and
molecules for high-energy electron impact can be found in [RIE72]. Equation 2.1 can be
integrated and an equation written for the ion density, where the ion density n /t) is the
difference betw een ng(t=0) and ng(t). Effects o f recombination can be ignored for low
percent ionization (< 0 . 1% for this experiment) and short times o f interest (< 1 0 0 u s for
densities sim ilar to this experiment [BR094]). The ion density is given by Equation 2.2.
= ngo (l - e x p (-n 6v<T,r))
( 2 .2 )
The ionization density as a function o f tim e from Equation 2.2 is plotted in Figure
2.2 for 0 to 700 ns for argon, nitrogen, and w ater vapor using a beam voltage o f -8 0 0 kV
and an e-beam current o f 500 A. Nitrogen and water vapor together compose the bulk o f
the background gas present in the experiment.
Argon is included in this plot for
discussion contained in Chapter 3. From Figure 2.2, the ion density after only 100 ns is
greater than l x l O 7 cm*3. The importance o f this plasma density will be discussed in
Section 2.4.1.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
8
2.2.2 Electron Beam Impact and Scraping
E-beam impact and scraping occur through the cavity/waveguide structure along
both the inner and outer conductor surfaces. This will be clearly shown from a glass
witness plate presented in Chapter 3.
E-beam im pact o f the outer conductor surface
occurs mainly when the e-beam is dumped into the wall o f the waveguide just after the
exit o f the microwave cavity. As well, the support structure at the front o f the microwave
cavity provides an additional source o f plasma from e-beam impact.
Water vapor is the principal obstacle in vacuum systems for fast pumpdown times
and low base pressures, as water is a highly charged polar molecule which tenaciously
attaches itself to any surface [HARSH]. There may be as many as 10-20 monolayers o f
water on a fairly smooth and normal surface with a roughness factor o f 10, which would
yield a surface density reaching 1017 H 2O m olecules/cm 2 [HAR91].
This large
concentration o f H 2O molecules becomes a natural source for the hydrogen line emission
presented in this dissertation. The area o f the e-beam collector of the coaxial gyrotron
used in this dissertation is approximately 20 cm 2 and would generate 2 x l 0 18 H 2O
molecules if the e-beam completely liberated all the H 2O monolayers. As will be shown
in Section 2.4.1, the maximum plasma density allowable for propagation o f the output o f
microwaves o f this coaxial gyrotron is 8 x l0 10 cm '3. Only a small fraction o f the total
H :0 molecule inventory at the e-beam collector is required to exceed this density. Given
that ablated wall-neutrals/plasma can have an expansion velocity o f 1-10 cm /ps [GIL91],
the H 2O molecules can expand rapidly enough to fill the volume. These estimates are
meant to illustrate that water liberation and expansion is capable o f creating plasma that
is dense enough to reduce the output microwaves o f this coaxial gyrotron.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
Argon
Nitrogen
W ater Vapor
l.E+07
0
100
200
300
400
500
600
700
Time [nanoseconds]
Figure 2.2. E-beam impact ionization o f argon, nitrogen, and water vapor for t = 0 to 700
ns.
2.3 Effect of Plasma on the Dispersion Relation
Due to the production o f plasm a by the e-beam in this experiment as shown in
Section 2.2, the solutions to the vacuum waveguide modes need to be modified to
account for the presence o f plasma. In the limit o f an infinite magnetic field, the plasm a
dielectric tensor reduces to a form where only transverse magnetic (TM) modes are
affected by the presence o f plasm a; transverse electric (TE) m odes are not affected by the
presence o f plasm a due to the confinem ent o f plasma electrons to motion along the
m agnetic field [KRA73].
Trivelpiece and Gould derived the dispersion relation for a
plasm a filled cylindrical waveguide with an infinite axial magnetic field [TRJ59], Due to
the infinite magnetic field, this dispersion relation is for TM modes and consists o f an
upper band pass for the electrom agnetic waveguide modes and a lower band pass for
space charge waves. For the upper band pass, the dispersion relation for the plasma filled
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
waveguide m odes for TM modes are given by Equation 2.3, where ©wo is the frequency
o f the propagating wave, k~ is the w ave number, c is the speed o f light, coculofr is the cutoff
frequency o f the vacuum waveguide mode, and cop is the plasma frequency.
o>lc = k W + oKutoff + <»\
(2.3)
The device used in this research operates with TE modes. To examine the effect
o f the plasm a on TE modes, the infinite m agnetic field assumption used in [TRI59] must
be relaxed. The effect o f plasm a on waveguide modes when the axial magnetic field is
finite is discussed in [KRA73] and [ALL63].
However, the solutions are very
com plicated and hinder the interpretation o f the features o f these waves.
From the plasma dielectric tensor, the effect o f plasma for this device can be
examined.
The plasma dielectric tensor for an infinite, cold, collisionless, and
hom ogenous plasm a is given in Equation 2.4, where ei, 62, and 83 are defined by
Equation 2.5, Equation 2.6 and Equation 2.7, respectively.
is ,
£ =
-ie -,
(2.4)
0
0
£, =£•„ 1 +
coz. -co '
( 2 .6)
CO C O Z ,-C O '
ce
c
*3 = £ o
(2.5)
CO'
co:
CO
\
£ 37
J
coz->\
1+ - ?
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(2.7)
11
For an infinite axial magnetic field, £t—> z0 and £2—> 0. For a finite axial magnetic
field, the range o f plasm a densities that will approximate this condition may be found by
plotting S|/ 8o and E2/e 0 versus plasm a density for a given axial magnetic field.
If this
condition is approximately satisfied, the plasm a will not have a significant effect on the
TE waveguide modes. In Figure 2.3,
Z \ / zq
an axial magnetic field o f 0.15 Tesla.
and £2/60 are plotted versus plasma density for
For plasma densities less than l xl O 9 cm \ the
dielectric constant is approximately that o f the infinite axial magnetic field case.
Therefore, the plasm a should not significantly affect the waveguide modes. For plasma
densities above l xl O 9 cm '3, the plasm a will affect the waveguide modes and could
change the perform ance o f the device due to the complexity o f the solutions for this case.
2
1.5
0.5
0
l.E+06
l.E+07
l.E+08
l.E+09
l.E+10
Plasma Density [cm-3|
l.E+11
Figure 2.3. Plot o f Ei/e0 (X) and £2/e 0 (•) versus plasma density.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
l.E+12
12
2.4 Mechanisms for Plasma Growth
2.4.1 Plasma Critical Density
For electrom agnetic waves propagating in a plasma, the dispersion relation is
given by Equation 2.8, where ro is the frequency o f the propagating wave, cop is the
plasma frequency (Equation 2.9), c is the speed o f light, and k is the wave propagation
vector. For the w aves to propagate, k must be a real number. Therefore, o)p must be less
than co.
The m axim um plasma density through which an electrom agnetic wave o f
frequency (co) can propagate is the found by solving Equation 2.8 for the plasm a density
(n) with k:=0. This is given by Equation 2.10. The operating frequency as measured by
heterodyne m ixing is -2 .5 5 GHz. For co=2.55 GHz, the maximum plasm a density that
the wave can propagate through is ~ 8 x l0 10 cm '3. If this maximum plasm a density is
exceeded, the wave will be reflected or absorbed.
1
7
' I
~
(o' = co'p + c'k~
r n e '•> V
<°P =
«<
/2
( 2 .8 )
(2.9)
\ £om j
c o 'c m
( 2 . 10)
2.4.2 Electron Cyclotron Resonance (ECR) Heating
The magnetic field profile for this device (given in Chapter 3) yields the
necessary m agnetic field for electron cyclotron resonance (ECR) heating o f the plasm a to
occur in the output waveguide region o f the experiment. ECR heating is a mechanism for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
wave/plasma interaction w hich would further increase the plasm a density beyond that o f
e-beam impact and ionization. In this region, the wave propagation vector (k) is parallel
to the axial magnetic field. The main operating mode o f the device \s f m~2.6 GHz. The
equation for the electron cyclotron frequency is given by Equation 2.11 [CHE84], where
B is the magnetic field, e is the electron charge, and m is the electron mass. Therefore, an
axial magnetic field o f 0.10 Tesla is necessary for the microwaves to interact and give
energy to the plasm a electrons. This magnetic field exists at the center o f the waveguide
region, where
( 2 . 11)
The effective resonance zone width for ECR heating in this experiment can be
calculated from Equation 2.12 [LIE94], where vres is the axial electron velocity at the
resonance position, co is the frequency o f the electromagnetic wave, and a is the magnetic
field gradient at the resonance position.
(2 -12)
For this experim ent, co and a are known, but the axial electron velocity at the
resonance position (vres) is unknown. An electron tem perature o f leV is reasonable for
common plasmas.
This can be used to calculate an electron velocity and obtain an
approximation for the effective resonance zone width.
For (o/2n = 2.55 GHz, a=2
Tesla/m (from Figure 3.5), and vrei= 5.9xl05 m/s (for Te= l eV), Azres is equal to 1.1 cm.
Therefore, the effective length that the propagating microwaves give energy to the
plasma through ECR heating is on the order o f a centimeter.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
14
2.4.3 Upper Hybrid Resonance Heating
When the propagating wave vector is perpendicular to the applied magnetic field,
upper hybrid resonance heating o f the plasma by the propagating wave is possible when
Equation 2.13 is satisfied [CHE84]. This condition m ay exist at the e-beam dumping
magnets. This interaction w as not investigated, but is mentioned here for completeness.
(2.13)
2.5 RF Plasma Sputtering
As mentioned previously in this chapter, water vapor is the principal contaminant
in vacuum systems. M oreover, it has been shown how this w ater vapor could produce a
plasma that will be deleterious to the radiation o f microwaves produced in the coaxial
gyrotron that is examined in this dissertation. The ability o f RF plasma sputtering to
remove H 2O contam inant is examined and presented in Chapter 4.
When an RF
discharge is established inside the vacuum, ions oscillating in the RF field will strike the
water layers and cause desorption o f the water [BER95]. The liberated water vapor can
then be removed from the system by the vacuum pumps.
The maximum heat o f physical adsorption o f HiO is 58.6 kJ/mol [HAY64], This
is equivalent to 0.61 eV/FhO molecule. For RF sputtering to reduce the amount o f water
vapor attached to the surface, ions must have an energy greater that 0.61 eV. Due to the
fact that RF sputtering o f the copper structure was observed in the experiment for RF
powers greater than 50 W, an approximation for the ion energy can be made.
The
standard heat o f formation (AHf°) is the energy required to change one mole o f a
substance from its reference form. To form Cu(g) from Cu(s), 341.1 kJ/mol is needed
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
15
[EBB90]. This is equivalent to 3.5 eV/Cu atom. The fact that it is possible to sputter
copper in this coaxial gyrotron for RF powers greater than 50 W implies that it is possible
to sputter HjO molecules o ff the surface at an RF power o f 50 W.
2.6 Adsorption of H2O Molecules
The rate o f adsorption o f water vapor can be calculated from Equation 2.14
[CUN99], w here dn/dt is the adsorption flux in molecules/(cm2s), s is the sticking factor,
P is the gas pressure in Torr, M is the molecular weight in amu, T is the gas temperature,
n/na is the ratio o f occupied versus available sites for adsorption, and x is the order o f the
desorption reaction,
dn
dt
3.5x10" sP 1 - ^ V
4mt
"a/
(2.14)
.
Some simplifications can be made in order to determine how many monolayers o f
H 2O impact the surface each second for typical operating parameters o f the coaxial
gyrotron exam ined in this dissertation (n/na can be set to zero since all sites are
unoccupied for this calculation and s is set to 1 to include all H 2O molecules impacting to
the surface).
For a gas pressure o f lxlO '5 Torr and gas temperature o f 300K, dn/di is
equal to 4 .7 6 x l0 15 H 2O molecules/(cm2s). This can be divided by the number o f H 2O
molecules in a monolayer/cm2 to see how many monolayers are striking the surface per
second.
From [DUS62], there are 5.27x10 14 H 2O molecules in a 1 cm2 monolayer
(assumes a flat surface).
Therefore, approximately 9 monolayers o f H 2O strike the
surface o f the experiment every second at lxlO*5 Torr.
For a background pressure o f
lx l 0 ° Torr, the amount decreases to 0.9 monolayers.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
2.7 Paschen Curve for High Frequency Discharges
From the Paschen curves, the voltage required to initiate the plasma discharge can
be estim ated from the gas pressure and distance between the electrodes. The Paschen
curves for high frequency breakdown o f argon and air are plotted in Figure 2.4 [B R 094],
The Paschen curve for nitrogen should be approximately that o f air. For the RF plasma
cleaning experiments presented in this dissertation, the param eters to initiate the RF
discharge are not satisfied. In order to get the plasma to breakdown, the pressure and/or
RF pow er m ust be initially set higher. After the plasma is stable, the pressure and power
are set to the parameters required for the experiment.
250
200
5 /3
"o
150
75
100
>
<y
«DC
Argon
50
Air
0.1
1
10
Pd [Torr x cm)
Figure 2.4. Paschen curves for high frequency breakdown o f air and argon.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
100
CHAPTER 3
EXPERIMENTAL CONFIGURATION AND DIAGNOSTICS
This experiment utilizes a large-orbit, coaxial gyrotron.
The main operating
characteristics o f the device are presented in [JAYOO]. Further adaptations have been
made to im prove the operating conditions and diagnostic capabilities necessary for this
experiment. The finai configuration is shown in Figure 3.1 and will be explained further
in the following sections.
Experim ents performed in this w ork are concerned with
plasma production inside the m icrow ave cavity and e-beam collector. Optical emission
spectroscopy is used to characterize the plasma during the operation o f this gyrotron.
M icrowave pulse shortening is exam ined in this experiment by using optical emission
spectrum to correlate the plasm a w ith the microwave output o f the device.
The large-orbit cusp setup for this experiment was designed in [HOC98] when
work on long-pulse, large-orbit (RCS) gyrotrons was started in the Intense Energy Beam
Interaction Laboratory at the U niversity o f Michigan.
3.1 MELBA Accelerator and Diode
M echanism s o f m icrow ave pulse shortening were investigated in a multi­
megawatt, large-orbit, coaxial gyrotron driven by the M ichigan Electron Long Beam
Accelerator (MELBA) at param eters: V= -800 kV, Icathodc= 6 kA, ItUbc= 0.8 kA, and
pulselengths o f 0.5-1.5 ps. M E L B A ’s voltage and current profiles have been archived
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
B7 [Tesla]
Magnetic
Cusp Plate
“0 .1
S-Band Rectangular
Waveguide
Solenoid Coils
RGA Head
G raphite Anode,
Aluminum
Cathode Tip
RF Power to
Coaxial Center
Conductor
5
Cathode Stalk
T o Heterodyne Mixer
2
j]
~3
Fiber Optics
I
Microwave Absorber
T o Cryogenic
Pump
Vertical Polarization
M icrowave Power Signal to
Attenuators and Diode Detector
To Cryogenic
Pump
Diode Magnetic
Field Coils
Coaxial
Microwave
Cavity
Coaxial
W aveguide
Rocowski Coils
1. Aperture CutTent Coil
2. Entrance CutTent Coil
3. ExitCuiTcntCoil
T o Spectrograph
Equipment
Figure 3.1. Large-orbit, coaxial gyrotron experim ental configuration with axial m agnetic field profile.
oo
19
since its initial experim ents [GIL85] and throughout its years o f operation. MELBA is a
modified Marx generator and consists o f sixteen Aerovox Type No. PX400D27 1.02
microfarad, 100 kilovolt (rated) capacitors arranged in pairs.
Spark gaps using sulfur
hexafluoride (SF6> insulating gas connect seven o f the capacitor pairs forming a standard
Marx circuit (M ain Marx). The remaining pair has a resistor and SF6 insulated spark gap
to form an Abram yan circuit. The Abramyan circuit is charged with reverse voltage with
respect to the M ain M arx and provides voltage compensation to the Main Marx to
ameliorate the im pedance collapse o f the diode load and RC decay o f the Main Marx
[ABR77]. This m odification allows M ELBA to achieve a flatter (± 7%) output voltage
pulse for long (>1 microsecond) pulselengths. In a Marx generator, the spark gaps allow
for the capacitors to be charged in parallel and discharged in series. For this experiment,
alternate capacitors are charged to +57 kV and -5 7 kV to keep the midplane o f the
switches near ground potential. With these charging voltages, M ELBA yields a flattop
output voltage o f approximately -800 kV.
Typical MELBA voltage pulse and diode
current traces are show n in Figure 3.2a and Figure 3.2b, respectively. A detailed design
description o f M ELBA can be found in the original design proposal [PSI83],
An
equivalent RC circuit model for MELBA can be found in [PSI83] and [CUN89].
3.2 Electron Beam Voltage and Current Diagnostics
3.2.1 MELBA Voltage Measurement
M ELBA’s voltage is measured across the insulating stack in the diode region by
means o f a balanced resistive divider (voltage monitor).
The voltage monitor is
constructed o f clear acrylic tube (length = 60cm , I.D. = 2.54 cm, O.D. = 3.81cm) and is
filled with copper sulfate solution. A small copper pickoff electrode is used near the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
grounded end o f the voltage monitor to provide the second part o f the resistive divider.
The signal from the voltage monitor is attenuated through a solid-state resistive divider, a
Tektronix lOx attenuator, and a 50 Q splitter. It is displayed on a Tektronix DSA602A
with 50 Q internal termination.
-200
-400
Voltage
[kV]
-600
-800
-1000
Tim e (lOOns/div)
4
Diode
Current 3
(kA|
2
0
Time (lOOns/div)
Figure 3.2. (a) Voltage pulse and (b) diode current produced by the M ELBA Marx Bank.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
21
The total calibration o f the voltage m onitor is currently 169 kV/V.
The
calibration setup is shown in Figure 3.3. A Velonex High Power Pulse Generator (Model
350 with a 1:10 voltage step up plug-in) provides a flat voltage pulse with a voltage range
o f 1 to 9 kV for the calibration. A Tektronix high voltage probe is used to accurately
measure the Velonex pulser output voltage. Prior to August 1999 (M ELBA Shot# 8517),
the calibration was 310 kV/V. This change in calibration was the result o f disassembling
the voltage m onitor for maintenance. A full description o f the operation and previous
calibration method o f the voltage m onitor is contained in [LUC88].
Voltage
Monitor
HV
V elonex H igh Pow er
Pulse G enerator (M odel 350)
w ith 1:10 V oltage Step Up Plug-in
Resistive
Divider
GND
Signal
Out
Tektronix
Tektronix TD S3052
O scilloscope
C h. 1
HV in
HV
P ro b e
C h. 2
T ektronix HV Probe
Com pensation Box
Figure 3.3. Calibration setup for the voltage monitor.
3.2.2 Diode Current
Diode current is measured with a small magnetic probe, a B-dot loop. The probe
is located in the M ELBA tank anode plate and extends approximately 1 cm into the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
22
transformer oil.
The probe is constructed o f a 1.1 cm core with five windings o f
magnetic wire and coated with Glyptal insulating enamel. Current travelling along the
cathode stalk induces a time-varying azimuthal magnetic field, which is detected by the
B-dot loop probe.
The detected signal is integrated with an RC integrator circuit
(t = 20 ps) to provide a voltage signal which is proportional to the diode current. The
calibration is 9.62 ± 0.48 mV/kA. More information on the calibration and operation o f
the B-dot loop probe can be found in [LUC88] and [MIL89].
3.2.3 Aperture Current, Cavity Entrance Current, and Cavity Exit Current
Once the electrons are emitted at the cathode tip, they travel through the anode
and into the interaction region. Rogowski coils are used to measure the e-beam current at
three places in the interaction region.
The aperture Rogowski coil is placed just
downstream o f the anode plate. The entrance Rogowski coil and exit Rogowski coil are
positioned at the entrance and exit o f the coaxial microwave cavity, respectively. The
voltage signal from each Rogowski coil is proportional to the derivative o f the current
(dl/dt) travelling through the coil. The signal is digitally integrated during data analysis.
The setup for calibrating the Rogowski coils is shown in Figure 3.4. The calibration for
the aperture Rogowski coil is 17 GA/integrated unit. The entrance and exit Rogowski
coil calibrations are 2.9 GA/integrated unit.
Any DC offset in the original signal will cause the integrated signal to have an
additional artificial positive or negative slope added to the base signal. Using the average
o f the initial part o f each signal as a method for removing the DC offset proved to be
unreliable. Analyzed data signals would still have some additional artificial slope after
using the described process.
A more reliable way was to identify the slope after the
integration and have it removed as a part o f the digital process. This is accomplished
with a short Matlab code that would first digitally integrate the signal using the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
trapezoidal method o f integration. In M atlab, the integrated data is plotted and the cursor
is manually used to identify the amount o f artificial slope contained in the initial part o f
the signal prior to the accelerator firing. The slope is then rem oved and the final signal
saved for later analysis. A sample code is shown in Appendix C.
Pearson C oil
(0.1 V/A)
Aperture
Rogowski
Coil
Entrance
Rogowski
Coil
Exit
Rogowski
Coil
Febetron Pulser
Module
To Digital Oscilloscope
for Analysis
Figure 3.4. Experiment setup for Rogowski coil calibration.
3.3 Diode and Electron Beam Transport Magnetic Fields
Two sets o f magnetic field coils are used to focus and control the e-beam. Five
pulsed “pancake” electromagnetic coils comprise the first set. They control the magnetic
field in the diode region and were designed to produce up to 1.0 Tesla. A full description
o f their design and operation can be found in [CUN89]. A solenoidal winding made up
o f two layers o f 12 gauge enameled copper wire form the second set. This set controls
the magnetic fields in the interaction region and has the capability to produce a magnetic
field o f up to 0.4 Tesla. Details o f the solenoidal winding construction as well as setup
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
details o f the m agnetic cusp plate used to produce the large-orbit, axis-encircling e-beam
can be found in [HOC98].
Because reproducibility o f data is especially im portant when attempting to
com pare averages o f different experimental cases, the magnetic field settings used for
this experiment were set to where this gyrotron would yield consistent shot to shot results
and were not further optim ized. A charging voltage o f 1.08 kV for the diode magnets is
used which creates an axial magnetic field o f -0 .0 6 Tesla in the diode region. A charging
voltage o f 200 V for the solenoidal magnets is used which creates an axial magnetic field
in the interaction region o f -0 .1 5 Tesla pointing opposite o f the diode magnetic field.
These settings w ere found to provide a stable regime for this study and were used for the
m ajority o f this experiment.
The m agnetic field was measured axially with a Bell Gaussmeter Model 610
equipped with a Hall effect axial probe. The microw ave cavity and waveguide were
installed during the m agnetic field profile measurements, so the probe was not centered
on-axis due to the center conductor. This magnetic field profile is shown in Figure 3.5.
The magnetic field in the cavity region is -0.15 Tesla. The measurement o f the magnetic
field is appreciably different when measured without the cavity and waveguide installed.
In the cavity region, the m agnetic field is -0.19 Tesla when measured without the cavity
and waveguide installed. This is due to suppression o f the pulsed magnetic field by the
cavity and waveguide structure.
3.4 MELBA Triggering Sequence
The triggering equipm ent for MELBA is shown in Appendix B (Figure B .l) and
is installed with the data acquisition equipment in a Faraday cage. Both the diode and
solenoidal magnet capacitor banks are discharged on a m illisecond time scale, whereas
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
25
the M ELBA capacitor bank is discharged on a nanosecond tim e scale. This requires the
triggering to occur as follows: diode magnet, solenoidal magnet, and then MELBA. For
each shot o f M ELBA, the MELBA capacitor bank, diode magnet capacitor bank, and
solenoidal m agnet capacitor bank are first charged. Once they are all charged, a single
trigger pulse is given from a Systron Donner 101 Pulse Generator. The initial trigger
pulse is used to start the discharge o f the diode magnet capacitor bank and trigger two
BNC Model 7050 Digital Delay Generators. The first delay generator is used to trigger
the solenoidal magnet capacitor bank to start discharging 108.3 ms after the initial trigger
pulse. The second delay generator is used to trigger MELBA to fire 116.0 ms after the
initial trigger pulse. A third delay generator is used to fire a trigger pulse to the MELBA
crowbar to end the voltage pulse by shorting out the diode.
o.io
^ “C av ity Region
lb
•
..
0.05----
"
W aveguide Region
3%---------------------------------------------------------------------------------------------uu
o
‘SXO
**
p
•
J2
^
Cl.
j
o .o o --------------------------------- eu
• ---------------C1
3 -----------------------------------------------------------------.
g
.
2
•
••
2
•
•
£ - 0 . 0 5 ---------------------------------------------------------------------------------------------------------------------------------------
u
3
m
•
••
- o . i o -----------------------------------------------------------------------------------------------------------------------
i
«
1
•
runju
•
•
•
•
___________________
i
Ic
<
-0.25 J----------------0
20
1---------------------------40
L
u
-
Ja
U
Anode
0.20 —3 -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------Cathodi
-
60
80
100
12C
D istance from Cathode [cm]
Figure 3.5. M agnetic field profile for diode magnetic field coil charging o f 1.08 kV and
solenoid m agnetic field coil charging o f 200 V.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
For the m ajority o f these experiments, the crow bar delay was set to fire 960 ns after the
MELBA trigger pulse. Both MELBA trigger pulses (initial and crowbar) are stepped up
from 5 V to 300 V before leaving the screen room.
approximately 260 ns to erect.
The M ELBA Marx bank takes
For a crowbar setting o f 960 ns, the average voltage
pulselength including the rise time is 700 ns.
3.5 Electron Beam Extraction and Glass Witness Plate
The coaxial gyrotron used in this experiment is driven by a large-orbit, axisencircling e-beam . The cathode tip is made o f aluminum and is covered with Glyptal
insulating enam el except for an annular portion where it is bare aluminum and scratched
with a razor blade to provide electric field enhancement for electron emission.
The
emitting annular ring has an inner radius o f 2.0 cm and outer radius o f 2.5 cm. The ebeam is em itted at the cathode and travels through a 7.6 cm diam eter opening in the
anode plate. In this region, the electrons are confined by the diode magnetic field. The ebeam then enters the magnetic cusp region where a radial magnetic field acts on the
electrons.
The electrons experience an azimuthal Lorentz force and rotate about the
center axis.
During the construction o f the magnetic cusp used in this experiment, glass
witness plates were used to determine the e-beam velocity ratio, a ( y j v,,) by measuring
the e-beam ’s Larm or orbit radius (R) [HOC98], [JAY99], [JAY00]. The off-centering o f
the Larmor orbits (Ar) can be measured from the glass plates.
A graphite anode with
eight 1 mm pinhole apertures placed at a radial distance o f 2.25 cm from the centerline o f
the experiment was used. The values measured for similar conditions to this experiment
are a = 1.0, R s 1.1 cm, and Ar = 1.1 cm.
Details o f those measurements, related
calculations, and experimental setup can be found in [HOC98] and [JAY00].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Prior
27
measurements in [HOC98] and [JAYOO] were performed w ithout the microwave cavity
and waveguide installed.
For this experiment, a glass witness plate was created to examine beam position
and distribution.
This measurem ent was performed with the cavity and waveguide
installed to give the m ost accurate picture possible o f the e-beam. The glass witness plate
from this measurem ent is shown in Figure 3.6. The setup o f the experiment is the same
as in Figure 3.1, with the glass plate placed between the m icrowave cavity and waveguide
but before the cavity term ination washer.
The glass witness plate shows that the e-beam is concentrated within a 2.5 cm
radius o f the center o f the coaxial cavity. The hole in the center o f the glass plate is for
the center conductor. E-beam scraping is highly probable along the center conductor o f
the coaxial cavity from the darkening which is visible near the center o f the glass witness
plate. E-beam scraping is also evident from comparisons o f the current measured at the
cavity entrance and exit for this coaxial gyrotron [JAYOO].
Figure 3.6. Glass W itness Plate o f the e-beam used to drive the coaxial gyrotron used in
this experiment.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
28
3.6 Microwave Cavity Structure and Cold Tests
The coaxial cavity used in this experiment is shown in Figure 3.7. The outer wall
is constructed o f an “L” type copper tube with an internal radius o f 3.79 cm. The inner
conductor is made from soft drawn copper tube with an outer radius o f 3.18 mm. The
interaction region is 26 cm long and is defined by a set o f stainless steel 4-40 threaded
rods at the front o f the cavity and two washers at the cavity exit. The stainless steel rods
provide a front support for the center conductor and keep the center conductor at ground
potential.
They also short the front o f the cavity such that backward traveling
microwaves are reflected. The two end washers are each 1.9 cm in diameter and 1.5 mm
thick. They are used to raise the cavity (Q) by reflecting a portion o f the microwave
power produced back into the cavity. This is enhanced by a gap o f 1.8 cm between the
cavity and waveguide. A fter the e-beam exits the coaxial cavity, it is dumped into the
outer wall o f the waveguide region several centimeters past the cavity exit by a set o f
permanent magnets.
The waveguide modes for this coaxial gyrotron can be calculated from
Equation 3.1,
j
: (kra)y;(krb ) =
(krb )r;(k,a )
<3.1 )
where J l n is the derivative o f the nth order Bessel Function o f the first kind, Y ~ n isthe
derivative o f the nth order Bessel Function o f the second kind, a is the radius o f the center
conductor o f the coaxial gyrotron (3.18 mm), b is the radius o f the outer conductor o f the
coaxial gyrotron (3.79 cm), and kc is the propagation constant [POZ98]. Equation 3.1 is a
transcendental equation and m ust be solved numerically for the roots o f the equation.
Once the mth root o f kc is found for the nlh order Bessel Functions, the cutoff frequency
for the TEnm waveguide m ode can be calculated from kc using Equation 3.2,
=
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3 2 )
29
where c is the speed o f light. The azimuthal part o f the cavity mode is contained in the
lhcorder o f the Bessel Function and the radial part o f the cavity mode is contained in the
nlh
mth root o f the solution o f Equation 3.1.
The axial part o f the cavity m ode can be modeled as a transmission line [JAYOO].
The stainless steel rods at the entrance o f the coaxial cavity approximate a short circuit in
the transmission line model whereas the exit o f the coaxial cavity approximates an open
circuit. The frequency o f the axial resonant condition is given by:
c f
1
’2
(3.3)
where L is the length o f the coaxial cavity (26.0 cm), / is an integer greater or equal to
one, and c is the speed o f light.
The resonant cavity modes are the combination o f the radial, azimuthal, and axial
resonant conditions found in this section. The frequency o f these modes is calculated
using Equation 3.4.
/ w = <JfL + f 2
29.6 cm
(3-4)
47.9 cm
1.8 cm
Cavity
W aveguide
Center Conductor
2 6 .0 cn>
Center Conductor
Supporting Rods
■102.9 cm
Cavity Termination
W asher
Figure 3.7. Diagram o f the coaxial cavity and waveguide setup with dimensions.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
All TEnmi cavity modes below 4 GHz are shown in Table 3.1. In this experim ent,
microwave signals above 4 GHz are cutoff by 4 GHz low pass filters.
Because the
primary operating mode (T E 112) oscillates at ~2.6 GHz, microwave signals cutoff by the
low pass filters are not o f interest in this experiment.
M icrowave cavity cold tests o f this coaxial gyrotron were performed in [JAYOO]
using a Hewlett Packard 8722D 50 MHz-40 G H z 2-port Network Analyzer.
Both the
TE ui and T E 112 m ode calculated frequencies were verified with the cold tests.
The
loaded cavity (Q) w as m easured to be 94 and 505, respectively for each mode.
Table 3.1. Calculated TE Cavity Modes.
TEnml
Fnml [GHzj
TE,,,
2.304
TEm
2.444
TE„3
2.703
TE„4
3.050
TE„5
3.459
t e 211
3.858
TEi 16
3.911
t e 2I2
3.943
3.7 Microwave Extraction and Detection
The interaction region o f this gyrotron consists o f a coaxial cavity. High power
microwaves produced in the cavity are directed out by a coaxial waveguide section with
the same transverse dim ensions as the coaxial cavity. The microwaves leave the vacuum
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
through a Lucite output window.
The output window is placed ju st inside a large
cham ber that is lined with m icrow ave absorber. At the other end o f the large chamber are
two S-band waveguides oriented in a vertical and horizontal polarization.
The
waveguides have thin copper w ires to prevent cross-polarized m icrowaves from entering
the w aveguide.
For this experim ent, only the S-band waveguide in the vertical
polarization is used for extraction o f microwave power. M icrowave signals are extracted
from this S-band waveguide by a 30 dB S-band M icroline directional coupler and a
0-20 dB HP-S375A variable attenuator.
Both are attached to HP-S281A adapters for
converting the signal from S-band to N-type coax.
Both signals travel in RG-214/U
coaxial cables shielded in copper pipe to the Faraday cage for analysis.
3.7.1 Microwave Diode Detector
T he microwave signal from the 30 dB directional coupler is further attenuated by
HP coaxial attenuators, filtered by a 4 GHz low pass filter, and measured with a
calibrated Narda 4503 diode detector.
This diode detector is used to measure power
production o f the coaxial gyrotron. Peak microwave pow er radiated and pulse duration
give the pulse shortening characteristics o f high power m icrowave devices.
The total
energy produced by the m icrow ave device is measured by digitally integrating the power
signal.
Analyses o f peak m icrow ave pow er versus m easured pulselength and o f total
m icrow ave energy produced are used as a figure o f m erit for different variations o f this
experim ent. The calibration for the microwave diode detector used in this experiment is
given in Equation 3.5.
v[mW] = 4.8987x10 ”2 + 0 .1 1 2 8 7 -z + 3.0644x10 ‘3 - z 2;z = [m V ]
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3.5)
32
3.7.2 Heterodyne Mixing and Time-Frequency-Analysis
A heterodyne mixer system is used to “beat-down” the operating frequency o f this
coaxial gyrotron.
This allows for digital sampling o f the oscillating frequencies and
determ ination o f the operating modes o f the coaxial gyrotron.
Traditionally, this has
been done w ith a Fast Fourier Transform (FFT) o f the spectrum to see which frequencies
are most dom inant. This allows for determination o f the operating modes o f the device.
However, the FFT lacks the resolution o f time, which becomes more important when two
or more m odes are present in the spectrum. For this experiment, time-frequency-analysis
is perform ed on the heterodyne m ixer data by means o f a reduced interference
distribution (RID) program developed by W. Williams [JE 092].
Because of the high
resolution o f both time and frequency, a better understanding o f the operating
characteristics o f the coaxial gyrotron is attainable.
The microwave signal from the variable attenuator is further attenuated by coaxial
attenuators and then mixed with the local oscillator signal. The heterodyne mixer system
setup is shown in Appendix A, Figure A .I. The frequency o f the intermediate signal is
equal to the local oscillator frequency plus or minus the unknown frequency. For this
experiment, the local oscillator frequency is set below the fundamental operating
frequency o f the coaxial gyrotron such that the unknown frequency is always the sum o f
the interm ediate signal frequency and the local oscillator frequency.
More detailed
information o f heterodyne mixers can be found in [POZ98].
3.8 Optical Emission Spectroscopy Diagnostics
This experiment utilizes both a 0.75 m Acton Research Corp (ARC) spectrograph
with two output ports and a 0.275 m ARC monochromator. They are both installed in the
Faraday cage to protect the electronics and reduce noise from the spark gap triggers; lead
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
33
bricks shield x-rays produced by the e-beam. The 0.275 m m onochrom ator is installed
with a Ham am atsu R928 photomultiplier tube. The 0.75 m spectrograph output can be
measured with either a Hamamatsu R928 photomultiplier tube or a Princeton Instruments
intensified charged-coupled device (ICCD) camera. The electrical and tim ing setup for
both are shown in Appendix A (Figure A.2 and Figure A.3). Three acrylic fiber optic
lines (1000 p m dia. each) are used to transm it optical emission from the experiment to
the spectroscopy equipment. The fiber optic lines are bundled together and view along
the experiment axis. The final setup used for this experiment is shown in Figure 3.8.
Output Window
Waveguide
30.5000
30.5000
Fiber Optic
Bundle
Figure 3.8.
Fiber optic bundle setup (center line and acceptance angle o f fiber optic
bundle shown).
Use o f internal fiber optic probes was examined for spatial resolution within the
coaxial cavity and waveguide.
The fiber probes were difficult to position inside the
vacuum beam tube due to the thickness o f the cavity/waveguide m aterial (2.2 mm) and
the space between the cavity/waveguide and the vacuum beam tube (~1 cm). The fiber
probes must be outside o f the inner cavity/waveguide wall to protect them against
electron impact, yet close enough to gather sufficient light em ission for detection.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
34
Moreover, the RF cleaning plasma would dam age the fiber probes after several
experiment sessions.
This led to the decision to locate the fiber optics outside the
vacuum.
3.9 RF Plasma Cleaning Equipment
RP Plasm a cleaning was examined in this experiment as a m ethod for removing
contaminants from the cavity and waveguide walls and increasing the energy output o f
this coaxial gyrotron. The center/outer conductor configuration o f this coaxial gyrotron
is ideal for a capacitive RF discharge.
A schematic o f the capacitively coupled RF
plasma discharge is shown in Figure 3.9. The RF cleaning equipment setup is shown in
Figure 3.10.
Microwave
Output
W indow
Diode
End
Nitrogen RF Plasma
c 'b c a m
direction
M icrow ave Cavity
M icrow ave W aveguide
if
Solenoid
Switch
RF Pow er
S u p p ly
System
Figure 3.9. Schem atic o f the RF cleaning plasm a.
The RF pow er source that was available for this research oscillates at 13.56 MHz.
This frequency is com m only used for RF discharges as it is an ISM (Industrial, Scientific,
Medical) frequency allocated by the FCC. The discharge gas was flow controlled by a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
35
leak valve, filtered by a Millipore (WFRG02 M ini XL) filter, and then leaked into the
system near the m icrow ave output window.
In order to maximize the RF sputtering
process o f water from the cavity and waveguide structure, the maximum RF power
possible should be used. However, for RF powers above 50 W, sputtering o f the copper
cavity and waveguide structure was observed. Therefore, an RF discharge pow er o f 50
W was used for the experim ents presented in this dissertation.
At a base pressure o f lxlO '5 Torr, approximately nine monolayers o f H 2O
molecules collide w ith the surface each second (Section 2.6). Therefore, it is important
to minimize the tim e between ending the RF cleaning and firing the accelerator in order
to maximize the effect o f discharge cleaning. However, the main limitation o f how short
this time can be is the amount o f time to pump down the system low enough to fire the
accelerator. The tim e needed to pump down is approximately thirty seconds. Because
the charging time o f M ELBA is approximately two minutes, remote switches are used to
shutoff the RF equipm ent such that the RF cleaning can occur during the charging cycle
o f the MELBA accelerator. During a typical M ELBA shot cycle, the gate valve for the
beam tube cryogenic pump is closed and the RF cleaning plasma is started before
beginning the charging cycle o f MELBA. Approximately thirty seconds before firing the
accelerator, the RF switch is toggled to send a 5 V TTL signal to the ENI RF Power
Generator as an external o ff signal. Then, the AC switch is toggled to switch o ff all AC
power to the equipment. Finally the cryo-switch is toggled to open the gate valve to the
beam tube cryogenic pump.
The RF pow er is delivered to the center conductor o f the coaxial gyrotron via a
piece o f RG-8/U coaxial cable stripped o f the outer jacket and braid. An alligator clip is
used to attach the m odified RG-8/U cable to the center conductor support, such that a
lead weight used with the solenoid switch shown in Figure 3.9 can remove the connection
remotely after the cleaning is complete and before firing the accelerator.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
36
Pneumatic Gate Valve
to Beam Tube
Cryogenic Pump
RF Remote Switch
Microwave
Output Window
120 VAC
5V TTL
Power Supply
AC Power
Relay Box
120 VAC
ENI ACG-3 XL
RF Power Generator
Solenoid
Switch
- 1 ............................
Gas Valve
for RF Processing
ENI MW-5D
Impedance Matching
Network
24V Power
Supply
Figure 3.10. RF plasm a cleaning equipment setup.
Initial experiments used argon as the breakdown gas for the RF discharge
cleaning. Argon is typically used for RF discharge cleaning. However, data collected
after argon was introduced into this coaxial gyrotron tended to have degraded mode
selection.
The discharge gas was switched to nitrogen (rated: extra-dry) for the
experiments presented here since nitrogen is always present in the experiment (RGA
data). Nitrogen has a lower electron ionization cross section compared to argon (Figure
2.2) and the cryogenic pumps used on this experiment have a pumping speed o f -1500
L/s for nitrogen and -1 2 0 0 L/s for argon.
3.10 Residual Gas Analyzer (RGA)
The composition o f the background gas for the experiment was examined with a
Stanford Research Systems RGA200 Residual Gas Analyzer (RGA). The detector head
o f the RGA was mounted to a 6-way, 6-inch Conflat adapter located near the output
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
37
window o f the experiment. The RGA is a mass spectrom eter that utilizes a hot filament
to analyze residual gases by ionizing some o f the gas molecules and separating the
resulting ions with a quadrupole mass filter.
The current from the collected ions is
measured with a Faraday cup and analyzed on a PC with software included with the
system. It was not possible to use the RGA during the MELBA pulse or during the RF
cleaning as the electronics and filament cannot survive operation in either environment.
A typical background trace for the RGA is shown in Figure 3.11. In general, the
background gas present in the experiment is -7 0 % w ater vapor and -2 5 % nitrogen. The
remaining 5% is com prised o f oxygen, hydrogen, carbon dioxide, molecular pump oil,
argon, and nitrous oxide.
4. E-08
Water Vapor
3. E-OS
ok.
o
—
Nitrogen
3
to
to
o 2.E-08
Hydrogen
Oxygen
I E-08
O.E+OO
Atomic Mass Units
Figure 3.11. Typical RGA trace o f the background gas in the HPM coaxial gyrotron.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
38
An attempt was made to com pare the RGA trace before and after a MELBA shot.
Little to no change was observed. M easurement o f changes in the system is complicated
by the placement o f the RGA detector in relation to the interaction region and cryogenic
pumps. Changes in the background gas caused by the operation o f the HPM gyrotron are
most likely undetectable due to the tim e that is required to start the detector after taking a
shot and the path molecules m ust take to reach the RGA detector versus the likelihood o f
reaching one o f the cryogenic pumps.
3.11 RF Tuned Langmuir Probe
An RF tuned Langmuir probe is used in this experim ent to measure the ion
density o f the RF cleaning plasm a. Details o f its construction can be found in [RIN99],
A special output window was built to perform the Langmuir probe measurements. Two
slip seals were installed in the output window to allow for the Langmuir probe to be
inserted into the RF cleaning plasm a oriented along the axis at two different radial
positions. Both measurements are taken with the tip o f the probe at approximately the
axial center o f the waveguide region. The first position is at a radius o f 1.4 cm from the
center conductor surface. The second position is at a radius o f 2.7 cm away from the
center conductor surface, or 1.1 cm away from the outer waveguide wall surface. The
experimental setup is shown in Figure 3.12.
A conventional Langm uir probe would have a distorted current-voltage (I-V)
characteristic due to the plasm a potential oscillating due to the 13.56 MHz RF power
supply [LIE94].
The RF-induced distortion is minimized by using a set o f tuned
inductors within the probe to reduce the RF voltage across the plasm a sheath o f the
probe. The tuned inductors ensure that the impedance o f the sheath is small compared to
the impedance between the probe and ground [PAR90].
Interpretation o f the I-V
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
39
characteristic from the RF tuned Langm uir probe is given in [PAR90].
The I-V
characteristics for probe position one and two are given in Figure 3.13.
1.7 cm (Position 1)
3.0 cm (Position 2)
A ttach
to Probe
13.56MHz Low Pass Filter
Micronta 9A4 Multimeter
(measure probe current)
ooooao
____
To experiment ground
Fluke 412B High
Voltage Power Supply
Triplett 2202 Multimeter
(measure probe voltage)
Figure 3.12. Setup for RF tuned Langmuir probe experiment.
The ion density is obtained in this analysis. Because electron saturation is never
reached, the electron density can only be inferred from the ion density. The RF plasma
discharge is non-Maxwellian. Because no single electron temperature is representative of
non-M axwellian electron energy distribution functions, calculation o f the electron
temperature was performed, but is not reported in this work.
This calculation is
performed w ith a plot o f the natural logarithm o f the probe current versus the probe
voltage, w here the electron temperature is the inverse o f the slope [PAR90].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
40
The ion density is obtained by plotting the square o f the probe current versus the
applied probe voltage for the ion saturation region o f the I-V characteristic, since the
positive ion current is related to the square root o f the probe voltage by Equation 3.6 (Ip =
ion current, e = electron charge, Ap = area o f probe, np = ion density, Vp = probe voltage,
Vs = plasma potential, and mp = ion mass). This equation is derived using standard orbit
motion limit theory (OML) and presented in [PAR90]. In order to solve Equation 3.6 for
the ion density, a least-squares fit is performed to the data points. The coefficients o f the
linear fit are represented by the constants in Equation 3.7 and are used in Equation 3.8 to
obtain the ion density.
I p =eA„nn
p p
' - 2 e ( V p - V s ) V '2
(3-6)
\
p
j
I 2 = KVp + B
z' IAAIp r 2m „ y /;
np =
(3.7)
(3-8)
2 e*A;
The squared probe current versus probe voltage plots used to obtain the linear fits
for probe positions one and two are given in Figure 3.14(a) and Figure 3.14(b),
respectively. The calculated ion density was l.8 5 x l0 9 cm'3 for position one and 5 .5 1 x l0 8
cm '3 for position two. They are plotted in Figure 3.15. The ion density is larger toward
the center. This is expected since the RF power for the plasm a is applied to the center
conductor o f the experiment. The larger ion density towards the center o f the experiment
implies that the water vapor attached to the surface o f the center conductor will be
removed more efficiently than w ater vapor attached to the outer wall.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
41
200
150
Position #1 (near Center Conductor)
j
Position #2 (near Outer Wall)
I
-100
eu
X X X X X X X X X X X X X X X X X X X X X X X X X X X
-50 4
-250
-200
-150
-100
0
50
Probe Voltage [V]
-50
100
150
200
250
Figure 3.13. I-V characteristics for RF tuned Langmuir probe position one and two.
3.12 RF Cleaning Plasma Characterization
For this experiment, an RF cleaning plasma (non-reactive) is produced in a
nitrogen backfill (N 2 supply rated extra dry). An RF power o f 50 W is used to drive the
cleaning plasma. Background pressures o f 15 mTorr and 25 m Torr w ere examined for
the RF plasm a discharge. The nitrogen backfill pressure is controlled w ith an analog gas
flow valve and was calibrated for the setting used during the experim ent with a Pirani
gauge from Kurt J Lesker (P/N KJL902006). The calibration was perform ed by replacing
the m icrowave output window with the Pirani gauge. Pressures m easured in the diode
region during cleaning were approxim ately 1 mTorr.
The RF pow er used for the
experim ent is a maximum based on observation o f sputtering o f copper above this power
level.
The low RF cleaning pressures used in the experiment w ere based on the
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
42
requirement o f pum ping out the cleaning gas to less than 8x1 O'5 Torr before firing the
accelerator in a reasonable amount o f tim e (—30 seconds).
I 6E -09
jy = -8.364E -I2x + 5.219E-I0[
1.4E-09
R: = 9.973 E-01
1.2E-09
2X
1.0E-09
8.0E -IO
6 .0E -10
4 .0E -10
2 .0E -10
O.OE+OO
Position #1
-150
-100
0
-50
100
50
150
Probe Voltage [V]
1.2E-10
iy = -7.463E-13x - 3.846E-111
1.0E-10
i
<
R” = 9.970E-01
8 .0 E -1 1
0
1
6.0E-11
ej
"c
t—
a.
4.0E-11
2.0E-1
Position #2
-250
-200
-150
-100
-50
0
50
100
Probe Voltage [V]
Figure 3.14. Plot o f squared probe current versus probe voltage for (a) probe position
one and (b) probe position two.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
43
2.0E+09
1.5E +09
s
(j
u
.a
•a
s
1.0E+09
V
o
s
e
0
0.5
1.5
2
2.5
3
3.5
4
Radial Position [cm|
Figure 3.15. Ion density o f the RF cleaning plasma versus radial position in the coaxial
waveguide from RF tuned Langmuir probe measurements.
For this analysis, the 0.75 m spectrograph with the ICCD cam era is used. Its
configuration is described in this chapter. A low resolution scan o f the nitrogen spectra
obtained during RF plasm a cleaning is shown in Figure 3.16. A higher resolution spectra
o f a portion o f the second positive system is shown in Figure 3.17. The band systems and
rotational lines are identified in Figure 3.16 and 3.17 from [PEA84], The rotational lines
o f the (0-3) and (1-4) vibrational bands can in principle be used to determine the
rotational temperature o f the RF cleaning plasma.
The rotational temperature is
indicative o f the gas temperature o f the plasma [PAS91]. By comparing the data o f this
experiment with that o f N 2 spectra produced by a microwave discharge, it can be
estimated that the rotational temperature is around room temperature [BRA00].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
44
7.E+0S
N; - 2nd Positive System
N: - 1st Positive System
CJrL.-B3n t
B3n r AII*,
5 E -05 - 2 8 1 43nm to 497.64nm
“
S03.08nmto 1051.00nm
W
J
I
3.E+05
2.E+0S
O.E-OO
375
425
475
525
575
625
675
725
775
Wavelength [nm]
Figure 3.16. Spectra o f N 2 1st and 2nd positive systems for the RF cleaning plasma.
5.E-OS
j
4.E-05
^
u
4.E+05
t
3 E -05
—
3 E -05
i
2.E -05
u
(0.3)
N - S e c o n d P o s itiv e S y s te m
j ~ c 3n u-B3n g
|
(1.4)
i
!
(2.6)
(3.6)
I
i
V 2.E-OS y1
1
1
5 E -04 i
0 E -00
380
(3.7)
I
I E+05 1
.
—J i
385
390
T jJ
J
1 J
___
395
400
405
410
415
Jt
420
425
Wavelength [nm]
Figure 3.17. Spectra o f the N 2 2>nd positive system for the RF cleaning plasma.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CHAPTER4
EXPERIMENTAL METHODS AND RESULTS
This chapter reports the results o f high power m icrow ave (HPM) pulse shortening
experiments on a large-orbit coaxial gyrotron. The principal diagnostics utilized for
analysis o f these experim ents are optical emission spectroscopy (both time-integrated and
temporally resolved) and diode detector m easurem ents o f the radiated microwave power.
The effects o f plasm a produced in the operating coaxial gyrotron are examined.
RF
plasma cleaning o f the cavity and waveguide region, as well as gas backfilling are
examined to determ ine their effect on the pulse shortening characteristics o f this gyrotron
device. The diagnostics used in these experiments are described in Chapter 3.
4.1 Optical Emission Spectroscopy of Plasma
The plasm a investigated in this experiment is generated in two ways. The first
mechanism is desorption o f contaminants by e-beam im pact on structures. This includes
e-beam interception o f the cavity/waveguide structure and e-beam collection where a set
o f permanent m agnets dumps the e-beam into the inner surface o f the outer wall o f the
coaxial waveguide. The second mechanism is volum e ionization as the e-beam travels
through the background gas o f the experiment.
Both mechanisms are discussed in
Chapter 2.
The first obstacle o f this research was to configure optical emission spectroscopy
diagnostics to observe and characterize the plasm a produced in the operating HPM
gyrotron. This setup was performed on a rectangular-cross-section (RCS) gyrotron and
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
46
initial results published in [GIL98]. The configuration was then modified and transferred
to the HPM coaxial gyrotron that is reported in this work. This diagnostic system was
used to link plasm a to the observation o f pulse shortening seen in this HPM coaxial
gyrotron. This system is described in Chapter 3.
In initial survey spectra o f the plasma in the gyrotron, the strongest optical
emission peaks were from H-alpha (656.28 nm) and H-beta (486.13 nm).
Time-
integrated spectra using the 0.75 m spectrograph with the intensified-CCD (ICCD)
cam era o f H-alpha and H-beta are shown in Figure 4.1 and Figure 4.2, respectively. Halpha is the em ission o f hydrogen from the energy level transition n=3 to n=2. H-beta is
the em ission o f hydrogen from the energy level transition o f n=4 to n=2. H-alpha is the
most intense optical emission line o f hydrogen in the visible spectrum [HER44],
6000
H-alpha (656.28 nm)
5000 -
S ' 4000
3000
ec
2000
1000
■
654
J jlJ
j
i_jlL
654.5
655
655.5
656
656.5
657
Wavelength [nm]
Figure 4.1. Optical emission o f H-alpha (0.75 m Spectrograph with ICCD gated for 1 ps
to capture entire voltage pulse, ICCD gain=10.0).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
47
14000
e
1 2000
£
10000
•
--
8000
6000
-
4000
-
2000
-
483
484
485
486
487
488
489
Wavelength [nm]
Figure 4.2. Optical emission o f H-beta (0.75 m Spectrograph with ICCD gated for I ps
to capture entire voltage pulse, ICCD gain=10.0).
H -alpha emission is always detected by optical emission spectroscopy during
operation o f the HPM coaxial gyrotron. H-beta radiation is detected, but is sometimes
comparable to the background noise o f the collected ICCD spectra. Therefore, H-alpha
line em ission is used in this experiment for temporally resolved optical emission
spectroscopy. The source o f H-alpha line emission is believed to be ionization o f the
background w ater vapor in the experiment as well as water vapor liberated from the
coaxial cavity and waveguide structure. The response time (~2 ns) o f the photomultiplier
tube allows for measurement o f fast changes in the plasma optical emission.
Visible
nitrogen lines/bands were expected in the plasm a spectroscopy since nitrogen is present
in the vacuum system (shown in next section). However, nitrogen lines/bands were not
detected during the e-beam pulse. The equipment configuration for tem porally resolved
optical em ission spectroscopy is discussed in Chapter 3.
A typical photomultiplier tube signal from H-alpha line em ission is given in
Figure 4.3 with the MELBA voltage and microwave diode detector.
The microwave
pulse ends before the end o f the voltage pulse. During the decay o f the microwaves, H-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
48
alpha line em ission increases.
This observation is discussed further in the following
section.
Microwave Power
[2 MW/div]
MELBA Voltage
- [-169 kV/div]
H-alpha
PMT Signal
[Rel. Intensity]
Time [100 ns/div]
Figure 4.3. Tem porally resolved optical em ission spectroscopy o f H -alpha shown with
M ELBA voltage and microwave power signal from the microwave diode detector.
4.2 Correlation of Microwave Emission and Plasma
In this experiment, a 0.275 m m onochromator equipped w ith a photomultiplier
tube is used to measure temporal optical em ission o f H-alpha during operation o f the
HPM coaxial gyrotron for comparison to microwave power radiated. B ecause this device
operates in the S-band, it is especially susceptible to microwave c u to ff due to an overdense plasm a. The main operating m ode (T E 112) o f this gyrotron is ~2.6 GHz. Therefore
m icrow aves will be cut o ffb y plasm a densities exceeding 8 x l0 10 cm '3 (Section 2.4.1).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
49
As discussed in Section 2.2.1, a plasma density o f lx lO 7 cm '3 is achievable by
ionization o f background gas in 100 ns (V= -800 kV, Ibeam= 500 A).
This density is
approaching that o f cutoff. As shown in Section 2.2.2, a plasm a density exceeding cutoff
density is achievable by plasm a generation at the e-beam collector alone for S-band
devices. M oreover, e-beam scraping is highly probable along the center conductor o f the
coaxial cavity from the darkening which is visible near the center o f the glass witness
plate (Section 3.5).
In addition to the previously mentioned mechanisms for plasma
production, the magnetic field profile for this device (given in Chapter 3) yields the
necessary m agnetic field for electron cyclotron resonance (ECR) heating o f the plasma to
occur in the output waveguide region o f the experiment (Section 2.4.2).
In Figure 4.4, pulse shortening o f the microwave signal is demonstrated, by
noting the e-beam voltage continues beyond the end o f the m icrowave signal. Early in
this research, the observation was made that the optical emission from H-alpha would
begin at approxim ately the same time as the end o f the high pow er microwave diode
signal. The heterodyne mixer signal is included in Figure 4.4 to show that after the end
of the m icrow ave signal the device is still oscillating, however, at lower power levels.
This behavior is suggestive o f microwave cutoff due to the plasm a density exceeding the
necessary cu to ff density.
The low-level oscillations are detectable on the heterodyne m ixer because it is
configured such that the high pow er output o f the HPM device is above the 1 dB
compression point o f the mixer. The 1 dB compression point is a quantitative measure o f
the onset o f saturation and is defined as the input power for which the output signal is 1
dB below that o f the ideal linear response regime o f the mixer [POZ98]. Below the 1 dB
compression point, the mixer responds linearly to the input microwave signal power. As
input power is increased beyond the 1 dB compression point, the m ixer’s output signal
amplitude becom es constant.
However, its frequency response is conserved.
At the
sacrifice o f pow er resolution at the peak power o f the coaxial gyrotron, the behavior o f
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
50
the device at low pow er levels can be observed and analyzed.
In general, low level
oscillations are detected from the heterodyne mixer signal for most microwave pulses
produced by this experim ent due to the long pulse capability (up to 1.5 ps) o f MELBA.
Heterodyne Mixer Signal
Microwave Power
[1 MW/di v]
MELBA Voltage
[-169 kV/div]
PMT Signal (H-alpha)
[R el. Intensity]
Time [100 ns/Div]
Figure 4.4.
Plot o f M ELBA voltage, microwave power diode signal, PMT signal (H-
alpha), and heterodyne m ixer signal versus time.
Figure 4.5 is a plot o f the time at which optical emission from H-alpha is detected
versus the tim e at w hich the microwave signal falls below a 1 M W threshold. A 1 MW
threshold is used in this dissertation because this research is concerned with multi-
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
51
megawatt m icrow ave output. A roughly linear relation is shown to exist between the
time at w hich the microwaves end and the start o f the optical em ission from H-alpha.
The linear correlation coefficient (r) value for the linear fit to this set o f data is 0.78
(r2=0.60). T he closer r is to unity, the m ore the data exhibit a linear correlation. The
significance o f the rv alu e for this linear fit can be checked quantitatively by examining
the probability that a set o f uncorrelated data points could produce a linear relation with a
r value o f 0.78. From [TAY82], there is less than a 0.1% probability that this set o f data
is actually uncorrelated and only appears to be linearly correlated, even though the exact
correlation m ay not be known.
500
00MAR06-Basc Vacuum ■
450
* 0OMAR08-Basc Vacuum !
-
400
sc
•3; 350
“ •300
250
y = 0.729x + 64.507|
S 150
R2 = 0.6012
j
50
0
100
200
300
400
500
600
Microwave Power CutofTTime [ns|
Figure 4.5. Plot o f H-alpha optical em ission start time versus microwave power cutoff
time (average M ELBA voltage start tim e subtracted from x- and y- axis) Base vacuum
pressure w as -8X10*6 T orron M arch 6, 20 0 0 and -7X10"6 Torr on M arch 8, 2000.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
52
A second m ethod to examine the correlation o f the optical emission o f the plasma
to the apparent m icrow ave cutoff is to average the traces o f a data set to obtain a
qualitative sum m ary o f the set o f traces.
The microwave pow er signal and
photom ultiplier signal have been averaged from two different data sets and are shown in
Figure 4.6(a) and Figure 4.6(b). The timing between the microwave cutoff and start time
o f optical em ission from H-alpha is seen here in the average sense, and confirm s that the
correlation o f the two signals is seen over m any shots o f the experiment. The reduction
o f H-alpha optical em ission between the two data cases in Figure 4.6 is due to a longer
pumpdown time used for the data in Figure 4.6(b). Even after the microwave pulses are
averaged over m any shots, they still exhibit distinct, deterministic features. This is seen
for all shot-averaged microwave power signals presented in this dissertation and is
discussed in Section 4.4.
4.3 Improvements in Total Microwave Energy Emission
Optical em ission spectroscopy identifies w ater vapor as the main contaminant in
the undesired plasm a through detection o f H -alpha optical emission during the operation
o f this coaxial H PM gyrotron. The plasm a is found to be deleterious to the microwave
production o f the device from correlations o f H-alpha optical emission to the microwave
pulse shortening characteristics o f the device.
Therefore, techniques for reducing the
effect o f pulse shortening on the device should concentrate on reducing the water vapor
inventory in the system. RF plasma cleaning and the use o f a backfill gas are examined
in this work.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
53
M ic ro w a v e P o w e r [2 M W /div]
P M T S ig n a l (H -alp h a)
[R e l. In te n sity ]
/
- *V'v,
i
M icrow ave Pow er [2 M W /div]
P M T Signal (H-alpha)
[Rel. Intensity]
i
l
’
!
I
Time [100 ns/div]
Figure 4.6. Plots o f shot-averaged microwave pow er and H-alpha optical emission signal
traces versus time for (a) 30 shots from March 6, 2000 and (b) 18 shots from M arch 8,
2000 .
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
54
4.3.1 RF Plasma Cleaning
Methods used for RF plasm a cleaning o f high voltage diodes [RIN99],[CUN99]
are m odified and adapted in this experim ent to examine the effect o f RF plasma cleaning
o f the cavity/waveguide structure on microwave pulse shortening characteristics. In this
experiment,
RF
plasma
cleaning
is
used
to
rem ove
contaminants
from
the
cavity/waveguide structure by m eans o f sputtering with nitrogen ions created by the RF
plasma discharge. Before installing the cavity and waveguide in the experiment, it was
chem ically cleaned with Citranox liquid acid detergent (manufactured by Alconox, Inc.)
to remove copper oxide and contamination. The base vacuum was less than IxlO '5 Tonafter approximately 5 hours o f pumping on the system with the cryogenic pump.
Background pressures o f 15 m Torr and 25 mTorr were used for the RF discharge and the
plasma was driven with 50 W o f RF power. The RF plasm a cleaning system setup is
given in Section 3.9.
RF plasm a cleaning o f the cavity/waveguide structure is examined here in four
separate data sets. The data sets are summarized in Table 4.1. Information is included in
Table 4.1 for data taken on M arch 6, 2000, which was used in Figure 4.5 and 4.6(a). RF
cleaning data are compared to the base vacuum case for each day by plotting both peak
microwave pow er and microwave energy versus the length o f the microwave pulse using
a 1 MW threshold. These are given in Figures 4.7(a,b), 4.9(a,b), 4.11 (a,b), and 4.13(a,b).
The averages for the peak microwave power, microwave energy, and microwave
pulselength are included on each graph for each case with error bars indicating the
standard deviation for the averages.
Pulse shortening curves are included on plots o f
peak m icrowave power versus pulse length for com parison o f the averages o f the
different cases in terms o f m itigation o f pulse shortening (Power = constant/time). An
increase in the value o f the constant is the desired result. The average peak microwave
power, average microwave energy, and average microwave pulselength are compared for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
55
each case and statistically tested with Analysis o f Variance (ANOVA). ANOVA allows
for a quantitative measure o f the confidence that two averages are statistically different
and are not from the sam e distribution [OTT93], [FRE97], [CRC90].
A statistical
confidence o f 95% is typically used as the cutoff o f significance for ANOVA.
A pplication o f this technique is given in Appendix E. In order to examine the effect o f
RF plasm a cleaning on the m icrow ave power signal and photom ultiplier tube signal, both
signals have been averaged for all shots in each particular case to obtain a qualitative
sum m ary o f each set o f signals.
They are presented in Figures 4.8(a,b), 4.l0(a,b),
4.12(a,b), and 4.14(a,b).
The first RF plasm a cleaning procedure examined in this experiment was to
alternate betw een taking a normal shot and RF cleaning before the shot.
The shot
sequence w as to RF clean for —5-3/4 minutes prior to the shot, wait for -3 0 seconds for
the system to pump down, take the shot, wait another —6 m inutes for the system to
recontam inate, take a shot on the accelerator, and then repeat the process. Therefore, the
data points o f this RF cleaning case are mixed between shots that were RF cleaned
im m ediately preceding the shot and shots that have had tim e to recontaminate. The gas
pressure used for the RF discharge was -15 mTorr.
minutes at 50 W o f RF power.
The cleaning time was —5-3/4
Additional shots were taken without RF cleaning to
examine recontam ination on a longer time scale.
These are referred to as post-RF
cleaning shots.
Figures 4.7(a and b) show the microwave power and microwave energy plotted
against the microwave pulselength.
For the mixed RF cleaning case compared to the
base vacuum case, the average microwave energy increased by 39% and the average
m icrow ave pulselength increased by 80%. A small loss o f 5% was seen in the average
peak m icrow ave power; this is believed to be due to the higher N 2 gas pressure at the
time o f the microwave pulse from the pumpdown lim itation o f the cryogenic pumps.
From A N O V A , the increases in average microwave energy and average microwave
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
56
pulselength have a statistical confidence o f 99.39% and 99.85%, respectively.
The
ANOVA analysis yields a statistical confidence o f only 45% for the change in average
peak m icrow ave power and is not considered significant.
From Figures 4.7(a and b), the post-RF cleaning case average peak microwave
power was increased by 1% and the average microwave energy w as increased by 5%
compared to the base vacuum case. These changes are not considered significant, with a
statistical confidence o f 6% and 51 %, respectively. The average m icrowave pulselength
was 33% higher for the post-RF cleaning case compared to the base vacuum case. From
ANOVA, the change in average microwave pulselength is significant with a statistical
confidence o f 99.24%.
The post-RF cleaned case exhibits an average microwave
pulselength in-between the average microwave pulselength for the base vacuum case and
the RF cleaned case. This is expected since the cavity/waveguide structure should be less
contaminated due to the cleaning effects o f the RF cleaning, yet m ore contaminated after
the RF cleaning has ceased due to the fast recontamination times mentioned in Chapter 2.
The shot-averaged power signals shown in Figure 4.8(a) highlight the additional
microwave pulselength seen for the mixed RF cleaning case (trace 2). The shape o f the
shot-averaged microwave power signals for all three cases are fairly consistent with each
other for —175 ns. After that point in time, the RF cleaning case (trace 2) produces an
additional —300 ns o f power compared to the base vacuum case (trace 1) and post-RF
cleaned case (trace 3). In examining the data for these cases shot-by-shot, the increase in
microwave pulselength was not seen until after 10 shots o f the m ixed RF cleaning case.
This implies that the RF cleaning required a threshold o f cleaning tim e be exceeded to
affect the m icrow ave pulse shortening characteristics of the device.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
57
Table 4.1. Sum m ary o f RF plasma cleaning study data sets.
Date
March 6, 2000
Remarks
N ew ly refurbished cathode stalk an d tip
F irst data set run on chem ically cleaned cavity/w aveguide structure.
S ystem roughed dow n starting at 9:00 am, cryogenic p u m p on system at 12:10 pm.
F irst shot taken at 1:51 pm at P sS x lO -6 Torr.
M arch 8. 2000
B ase vacuum case
S ystem roughed dow n starting at 5:17 pm (M ar07), cry o g en ic pum p on system at
8:23 am. First shot taken at 10:18 am at a P ^ x lO " 45T orr.
March 15, 2000
B ase vacuum case exam ined. RF cleaning exam ined by alternating betw een taking a
sh o t and perform ing cleaning for -5 -3 /4 m inutes before th e shot w ith N 2 at 50 W RF
Pow er. Elapsed tim e betw een end o f cleaning and tak in g shot w as - 3 0 seconds.
Pressure for cleaning w as - 1 5 m T orr within interaction region. T he shot was taken
a t P = lx l 0 '5 Torr.
Post-R F cleaning case exam ined after cleaning portion at
P = 7 x lO * T o rr.
S ystem roughed dow n starting at 8:35 pm (M a rl4 ), cry o g en ic pum p on system at
9:0 0 pm (M arl4 ). First shot taken at 10:40 am at P=5.x 1O'* Torr.
B ase Case exam ined. RF cleaning exam ined by perform ing cleaning for
—10-1/2 m inutes before every shot with N 2 at 50 W RF Pow er. Elapsed time
betw een end o f cleaning and taking shot w as - 3 0 seconds. Pressure for cleaning
w as -1 5 m T orr w ithin interaction region. T he shot w as taken at P= 1x10 s Torr.
P ost-R F cleaning case afterw ards at P sS xlO '6 Torr.
N ew ly refurbished cathode stalk and tip
March 28, 2000
S ystem roughed dow n starting at 4:35 pm (M ar27), cry o g en ic pum p on system at
6:55 am. First shot taken at 9:16 am at P = lx l0 '5 T orr (b ase vacuum case).
March 29, 2000
B ase vacuum case exam ined. System purged with a ir to fix problem inside
w aveguide. Roughed dow n for 18 m inutes and cry o g en ic pum p on line for
—48 m inutes before starting the second base case (P = 1 .7 x l0 's T orr). RF cleaning
exam ined by perform ing cleaning for -1 0 -1 /2 m inutes b efo re every shot with N 2 at
50 W RF Power. E lapsed tim e betw een end o f clean in g and taking shot was
- 3 0 seconds. Pressure for cleaning was - 2 5 m T orr w ith in interaction region. The
sh o t w as taken at P = 1 .3 x l0 ‘5 to 2.3x10 s T o n \ Post-R F clean in g case afterwards at
P = 9 x lO ^T o rr.
S ystem roughed overnight and purged to atm ospheric p re ssu re w ith lab air at 9:14
am fo r five m inutes, roughed dow n starting at 9:19 am , cry o g en ic pum p on system at
9 :45 am . First shot taken at 10:40 am at P = lx l0 '5 T orr (b a se vacuum case).
S ystem was RF cleaned w ith N 2 at 50 W RF P ow er fo r - 1 .5 hours. RF cleaning
p ressure was - 2 5 m T orr w ithin interaction region, P s l x l O '5 to 7.4x1 O’* Torr for
P ost-R F cleaning case
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
58
March 8, 2000
-'4200/tim e [ns]
3350/time [ns]
25
= r
D Base Vacuum (~7e-6 Ton-)
• Mixed RF Cleaning (~le-5 Torr)
X Post RF-Cleaned (~7c-6 Torr)
2500/time [ns]
0
100
200
300
400
500
600
Pulse Length (Full Width 1 MW) [nanoseconds|
2.5
“S
s
U
0.5
D Base Vacuum (-7e-6 Torr)
• Mixed RF Cleaning (—1c-5 Torr)
x Post RF-Cleaned (~7e-6 Torr)
March 8. 2000
0
100
200
300
400
500
600
Pulse Length (Full Width 1 MW) [nanoseconds|
Figure 4.7.
a) Peak microwave pow er and b) microwave energy plotted versus
microwave pulselength for data taken on M arch 8, 2000. Base vacuum case pressure was
~ 7 x l0 '6 Torr.
RF cleaning shots were taken at ~ l x l 0 '5 Torr.
Post-RF cleaning case
pressure was ~7xl0"6 Torr.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
Microwave Power
[2 MW/div]
PMT Signal (H-alpha)
’ [rel. intensity]
'-a
1
«
*
100 ns/div
Figure 4.8.
Shot-averaged (a) microwave pow er and (b) H-alpha optical em ission for
March 8, 2000. 1) Base vacuum case (18 shots, ~ 7 x l O'6 Torr), 2) RF plasm a cleaning (20
shots, - l x l 0 '5 Torr during shot, alternating betw een regular shot and cleaning), and 3)
Post-RF cleaned (6 shots, -7x1 O'6 Torr).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
60
From Figure 4.8(b), the shot-averaged H-alpha optical emission during the
operation o f the device was delayed in time for the mixed RF cleaning case (trace 2)
compared to the base vacuum case (trace 1).
retarded the grow th o f the plasma.
This implies that the RF cleaning has
This supports the theory that the microwaves are
being reduced by a plasma since the microwave pulselength was extended (Figure 4.8(a),
trace 2) w hen the H-alpha optical em ission was delayed (Figure 4.8(b), trace 2). The
s'not-averaged optical emission o f H-alpha for the post-RF cleaning case (trace 3)
increases to m ore than the original base vacuum condition (trace 1). The shot-averaged
optical em ission o f H-alpha should increase after the RF cleaning has ceased due to the
recontam ination rate. However, the reason the optical emission for the post-RF cleaning
case is larger than the original base vacuum case is not understood.
One possible
explanation is that the liberated water vapor has not redeposited on the walls and forms a
larger percentage o f the bulk gas com position than the water vapor did for the base
vacuum case.
The second technique examined in this experiment was to RF clean before every
shot o f the RF cleaning case for -1 0 minutes. The gas pressure used for the RF discharge
was -1 5 m Torr with 50 W o f RF power. Post-RF cleaning shots were taken after the RF
cleaning shots to examine recontamination on a longer time scale.
The results were
similar to the first technique examined.
From Figure 4.9(a), for the RF cleaning case compared to the base vacuum case,
the average peak microwave power was reduced by 4% and the average microwave
pulselength w as increased by 21%.
Figure 4.9(b) shows that the RF cleaning case
compared to the base vacuum case increased the average m icrowave energy by 15% .
From AN O VA , the increase in average microwave energy and average microwave
pulselength
are significant with
statistical confidences o f 96.8%
and 99.59%,
respectively. However, the change in average peak microwave power is statistically not
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
61
significant with an ANOVA statistical confidence o f 44%. For this set o f cases, the postRF cleaning case compared to the base vacuum case exhibits a slightly larger average
peak m icrow ave power and average microwave energy (2% and 6% , respectively), yet a
lower average microwave pulselength (12%) is measured (Figures 4.9(a and b)). From
ANOVA, none o f these changes are significant w ith a statistical confidence o f 22% for
the change in average peak microwave power, 68% for the change in average microwave
energy, and 85% for the change in average m icrow ave pulselength. M oreover, the data
exhibit som e signs o f conditioning during the base vacuum case, shown by the large
variance o f the microwave pulselengths o f the base vacuum case (Figures 4.9(a and b)).
From Figure 4.10(a), the increase in average microwave pulselength for the RF
cleaning case (trace 2) compared to the base vacuum case (trace 1) is noticeable in the
shot-averaged microwave power signal plotted. This was also seen in the first technique
examined in this experiment. In Figure 4.10(b), the shot-averaged photom ultiplier tube
signals exhibit the same behavior as in the first technique examined.
The optical
em ission from H-alpha is reduced/delayed in tim e during the RF cleaning case (trace 2)
com pared to the base vacuum case (trace 1). The optical emission from H-alpha is larger
for the RF-post cleaning case (trace 3) than the base vacuum case (trace 1). The second
technique examined has not yielded more inform ation about the observation that the Halpha em ission is stronger for the post-RF cleaning case when com pared to the base
vacuum case.
Before the next set o f experiments was performed, the system was disassembled
for m aintenance on the cathode o f the accelerator.
The system w as at atmospheric
pressure for over a week. The second technique was then examined again in another set
o f data. The same parameters were used for the discharge plasma, except the discharge
pressure w as increased to 25 m Torr to keep the discharge plasma stable. The reason the
plasma w as not stable at 15 m Torr as in the previous two data sets is unclear. Two base
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
62
6200/time [ns]
— 25
March 15.2000
4800/time [ns]'
XI
5300/time [ns]
20
2 15
n Base Vacuum (~5e-6 Torr)
• RP Cleaning (~ le -5 T orr)
* Post R F-Cleaned (~5e-6 Torr)
0
100
200
300
400
500
600
Pulse Length (Full Width 1 MW) [nanoseconds]
2.5
b)
2
&
« 1.5
VHflL r - *
J fc j!'
s
U
>
I
x
%
■
0.5
n Base Vacuum (-5 c -6 Torr)
• RP Cleaning (~1 e-5 Torr)
x Post RF-Cleaned (~5e-6 Torr)
March 15.2000
100
200
300
400
500
600
Pulse Length (Full Width I MW) [nanoseconds|
Figure 4.9.
a) Peak microwave power and b) microwave energy plotted versus
m icrowave pulselength for data taken on M arch 15, 2000. Base vacuum case pressure
was ~5xl0*6 Torr.
RP cleaning shots taken at - lx lO '5 Torr.
pressure w a s -5 x 1 0
Torr.
Post-RF cleaning case
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
63
Microwave Power
[2 MW/div]
. PMT Signal (H-alpha)
[rel. intensity]
.J
100 ns/div
Figure 4.10. Shot averaged (a) microwave pow er and (b) H-alpha optical emission for
March 15, 2000.
1) Initial base vacuum case (15 shots, -5X10*6 Torr), 2) RF plasma
cleaning (14 shots, ~ l x l 0 '5 Torr during shot, each shot), 3) Post-RF cleaned (15 shots,
-5 x 1 0 ^ Torr).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
64
vacuum cases w ere run for this set. After 18 shots on the experiment, a degradation o f
power and m ode selection was observed. The cavity/waveguide structure was checked
for damage since past observation o f this behavior has been caused by the e-beam
damaging the center conductor and/or copper particulate in the cavity region
due to e-beam impact. Indeed, small copper flakes were found in the cavity region. The
vacuum
system
w as
purged
to
atmospheric
cavity/waveguide cleaned o f all particulate.
pressure
with
lab
air
and
the
The system was then pum ped down and
experiments resum ed approximately two hours later.
Four shots were taken to establish this new base vacuum before proceeding to the
RF cleaning case.
As seen in Figure 4.11 (a and b), the average microwave energy
increased by 111% and the average microwave pulselength increased by 124% between
the RF cleaning case and the new base vacuum case.
From ANOVA, the statistical
confidences for the increase in the average microwave energy and average microwave
pulselength are 99.18% and 97.7%, respectively, The average peak microwave power
was increased by 10%, but is not significant from ANOVA, with a 44% statistical
confidence.
As seen in Figures 4.11 (a and b), the post-RF cleaning case yields an increase o f
63% for the average microwave energy, an increase o f 23% for the average peak
microwave power, and an increase o f 27% for the average microwave pulselength
compared to the RF cleaning case. Only the change in average microwave energy is
statistically significant from ANOVA with a statistical confidence o f 99.43%.
The
statistical confidences for the change in average peak microwave power and average
microwave pulselength are 92.7% and 89.5%, respectively.
For the com parison o f the post-RF cleaning case to the second base vacuum case
(Figures 4.11 (a and b)), the average microwave energy increased by 245%, the average
peak microwave pow er by 34%, and the average microwave pulselength by 186%. From
ANOVA, the increase in average microwave energy and average m icrowave pulselength
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
65
are found to be statistically significant, with a 98.9% and 99.51% statistical confidence.
The increase in average peak microwave power was not significant, with a statistical
confidence o f 83.6%.
As shown in Figure 4.12(a), the post-RF cleaning case (trace 4) exhibits longer
microwave pulselength characteristics than the RF cleaning case (trace 3), as well as both
base vacuum cases (traces 1 and 2).
The additional area under the shot-averaged
microwave pow er signal for the RF cleaning case signal (trace 3) and the post-RF
cleaning case signal (trace 4) compared to both base vacuum case signals (traces 1 and 2)
shows how much additional energy is produced during the RF cleaning case and post-RF
cleaning case.
From Figure 4.12(a and b), the second base vacuum showed a shorter microwave
pulse length and higher H-alpha optical emission than the first base vacuum case. This is
expected as the vacuum conditions for the second base case were not as good as for the
first base case. The first base vacuum case was roughed down overnight, where as the
second base vacuum case was roughed down for only 18 minutes. Also, the first base
vacuum case was pum ped on with the cryogenic pum p for two hours and 21 minutes
before starting the experim ent whereas the second base vacuum was only pum ped on
with the cryogenic pum p for 48 minutes before starting. From Figure 4.12(b), the optical
emission for the post-R F cleaning case (trace 4) is in-between the base vacuum case
(trace 1) and the RF cleaning case (trace 3). This was expected for the first two data sets
examined but was not observed. Moreover, the H-alpha optical emission is significantly
retarded (-1 0 0 ns).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
66
4 Base Vacuum (-le -5 Torr)
D Base Vacuum after purge (- 1 .7e-5 Torr)
• RF Cleaning (1.3e-5 to 2.3e-5 Torr)
* Post RF-Cleaned (~9e-6 Torrl
3240/time [ns]
50K)/time [ns]
1300/time [ns]
50
100
1450/time [ns]
150
200
March 28. 2000
250
300
350
400
450
500
Pulse Length (Full Width 1 MW) [nanoseconds]
4 Base Vacuum (~lc-5 Torr)
D Base Vacuum after purge (~!.7c-5 Torr)
• RF Cleaning (1.3e-5 to 2.3e-5 Torr)
x Post RF-Cleaned (-9e-6 Torr)_________
3
—8 2.5
*
b)
X
^
X
X
X
X
X
w
-
•
a
r
1
•
J
•
e2
w
c 2
U
«
ei
i 1.5
,
X
0.5
•
March 28, 2000
50
100
150
200
250
300
350
400
450
j
500
Pulse Length (Full Width 1 MW) [nanoseconds|
Figure 4.11.
a) Peak m icrow ave power and b) m icrow ave energy plotted versus
m icrow ave pulselength for data taken on March 28, 2000.
pressure was ~ l x l 0 '5 Torr.
Initial base vacuum case
Base vacuum case pressure (after breaking vacuum) was
~ 1 .7 x l0 '5 Torr. RF cleaning shots taken at 1.3xlO'5 to 2 .3 x l0 '5 Torr. Post-RF cleaning
case pressure was ~9x 10 Torr.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
67
PMT Signal (H-alpha)
- [rel. intensity]
X"v
100 ns/div
Figure 4.12. Shot-averaged (a) microwave power and (b) H-alpha optical emission for
March 28, 2000. 1) Initial base vacuum case (16 shots, -lxlO *5 Torr), 2) base vacuum
case (after system purge to lab air, 4 shots, 1.7x10‘5 Torr), 3) RF plasm a cleaning (17
shots, 1 .3 x l0 '5 to 2 .3 x l0 '5 Torr, each shot), 4) Post-RF cleaned (15 shots, -9x10^ Torr).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
68
The main differences between this data set and the two previous data sets are the
short pum ping time for the second base vacuum case and the gas pressure o f 25 mTorr
used for the RP discharge. The im proved results suggest that this pressure is better for
the cleaning mechanisms. If the discharge pressure is too high, the ability o f the ions to
clean the surface through sputtering is hindered by collisions.
However, at lower
discharge pressures, less ions are available to sputter the surface [CUN99].
The recontamination o f the coaxial structure surfaces has been controlled, which
is evident from the better perform ance o f the post-RF cleaning case in com parison to both
the RF cleaning case and base vacuum case. The most likely explanation is that the RF
cleaning has succeed in removing the majority o f the water vapor from the structure
surfaces by the end o f the RF plasm a cleaning case. After which, the e-beam continues to
keep the w ater vapor from significantly building up on surfaces due to e-beam impact and
scraping.
Given that the post-RF cleaning o f the previous data set yielded the best results,
the third technique examined was to use one long, continuous RF cleaning cycle. Data
were taken for the base vacuum case, the system was cleaned with an RF plasma
continuously for 1.5 hours, and then data were taken for the post-RF cleaning case. The
pressure for the discharge was ~25 mTorr. From Figure 4.13(a and b), the average peak
microwave pow er increased by 30% and the average microwave energy increased by
63% for the post-RF cleaning case compared to the base vacuum case.
The average
microwave pulselength was increased by 36%. Using ANOVA, the increases in average
peak m icrow ave power, average m icrow ave energy, and average m icrowave pulselength
are all statistically significant with high levels o f confidence. The statistical confidence
o f the change in average peak m icrow ave power was 99.55%. The statistical confidences
o f the change in average m icrow ave energy and average m icrow ave pulselength are
greater than 99.99% for both. O f the four data sets examined with RF plasm a cleaning,
these data had the highest statistical confidence for the com pared averages.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This can
69
m ost likely be attributed to the simplicity o f the case with on ly one long RP plasma
cleaning session and data collection before and after the cleaning.
From Figure 4.14(a), the increase in microwave pulse length for the post-RP
cleaning case (trace 2) com pared to the base vacuum case (trace 1) is unmistakable.
From Figure 4.14(b), the optical emission o f H-alpha has been delayed for the post-RP
cleaning case (trace 2) in com parison to the base vacuum case (trace 1).
A summary plot o f peak microwave power versus the m icrow ave pulse length for
each case averaged over all days examined is shown in Figure 4.15. The averages are
performed from each case average to prevent the plot from being skewed by differences
in the number o f shots for a given case on a given day. It is seen in Figure 4.15 that the
shots taken during RF plasm a cleaning and in the post-RP plasm a cleaning cases
improved the pulse shortening characteristics o f this device in com parison to the base
vacuum cases examined.
This is shown quantitatively by the increase in the pulse
shortening curve constant (Pow er = constant/time) for the different cases.
The RF plasm a discharge pressure o f 15 m Torr show ed mitigation o f pulse
shortening characteristics o f the device during the RF cleaning case in comparison to the
base vacuum case and post-RP cleaning case. For the RF plasm a discharge pressure o f
25 mTorr, the post-RF cleaning case showed im provem ent o f pulse shortening
characteristics for the device in comparison to the base vacuum case and RF cleaning
case. These improvements clearly show that the RF plasm a cleaning has succeeded in
m odifying the pulse shortening curve for this device.
Appendix D discusses surface modification that was exam ined in a collaborative
effort between the University o f Michigan and Stanford Linear A ccelerator Center within
the Air Force Office o f Scientific Research HPM M ultidisciplinary University Research
Initiative.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
70
10810/time [ns]
40
C. 35
6070/time [ns]
&
i 20
a Base Vacuum (~1.3e-5 Torr)
March 29, 2000
* Post RF-Cleaned ( le -5 to 7.4e-6 Torr)
0
50
100
150
200
250
300
350
400
450
Pulse Length (Full Width 1 MW) [nanoseconds|
a Base Vacuum ( - 1 .3c-5 Torr)
x Post RF-Cleaned (1 e-5 to 7.4e-6 Torr)
b)
> .4
e®
X X
>
□
©
S
2
x
□
March 29. 2000
50
100
150
200
250
300
350
400
450
Pulse Length (Full Width I MW) [nanoseconds]
Figure 4.13.
a) Peak m icrowave power and b) microwave energy plotted versus
microwave pulselength for data taken on March 29, 2000. Base vacuum case pressure
was ~1.3x10 '5 Torr. Post-RF cleaning case pressure was l x l 0 '5 to 7.4x1 O'6 Torr.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
71
Microwave Power
[2 MW/div]
PMT Si
100 ns/div
Figure 4.14. Shot-averaged (a) microwave power and (b) H -alpha optical emission for
March 29, 2000. 1) Initial base vacuum case (15 shots, ~ 1 .3 x l0 '5 Torr) and 2) Post-RF
cleaned case after 1.5 hours o f RF cleaning (12 shots, lxlO*5 to 7.4x1 O'6 Torr).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
72
35
3070/T[ns] — ►Vx
3200/t[ns]
30
4030/t[ns]
w
125
5150/x[ns]
7650/t[ns]
£ .20
>
«
015
u
-*10
C
&S
a.
d B a s e V acuum
° R P C le a n in g (15 m T o rr)
• R P C le a n in g (2 5 m T o rr)
x P o st-R F C lea n in g (15 m T o r r ) !
& P o st-R F C le a n in g (2 5 m T o rr)_________________________________;
50
Figure 4.15.
100
150
200
25 0
300
Pulse Length (Full Width 1 MW) [nanoseconds]
350
400
Peak m icrow ave power versus microwave pulse length summary o f RF
plasma cleaning results.
Data plotted are the averages o f the individual case averages
examined. Pulse shortening curves (Power = constant/time) are shown for each case.
4.3.2 Gas Backfilling
The concept o f using a backfill gas in the experiment evolved from the ionization
calculations presented in C hapter 2. By filling the experiment w ith a gas that would be
ionized less by the e-beam or would suppress plasma growth, the microwave pulselength
could potentially be increased and therefore the amount o f energy radiated would be
increased. Sulfur hexaflouride (SF6) was chosen due to its use as an insulating gas and its
ability to scavenge electrons from a discharge [KLI79], [CR.083].
This concept was
further tested by also com paring the pulse shortening characteristics o f argon backfilling
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
73
to the base vacuum case and SF6 backfilling, since its e-beam ionization cross section is
larger than nitrogen and w ater vapor (Section 2.2.1).
A sum m ary o f average microwave energy production and average peak
microwave pow er for the backfill cases examined in this experim ent is given in Figure
4.16 and Figure 4.17, respectively.
After a base vacuum case was performed for
comparison, an argon backfill was examined. The base vacuum case was at 4x1 O'6 Torr,
while the argon backfill w as at 1.6xl0'5 Torr. From Figure 4.16 and Figure 4.17, the
average m icrowave energy increased by 31%, while the average peak microwave power
was reduced by 5%. From ANOVA, the increase in the average microwave energy is
significant and has a statistical confidence o f 99.94%.
T he change in average peak
microwave pow er is found to be not significant as the statistical confidence is 81%. It
was expected that the average microwave energy produced by the device would decrease
for the argon backfill due to the higher electron ionization cross section and higher
background pressure, but this was not the case. This implies that the mechanism is not
sim ply ionization o f the background gas by the e-beam, but is m ore complex.
The effect o f SF6 backfilling on the pulse shortening characteristics o f this coaxial
gyrotron was then exam ined after purging the experiment w ith air to displace the argon.
The same backfill pressure o f 1.6xlCT5 Torr was used for com parison. Figure 4.16 and
Figure 4.17 show that the average microwave energy increased by 52% and the average
peak microwave pow er increased by 12% for the SF6 case in comparison to the initial
base vacuum case.
From ANOVA, the increase is average microwave energy is
significant with a statistical confidence o f greater than 99.99% .
The change in the
average peak m icrowave pow er is not significant with a statistical confidence o f 54%.
This improvement in average microwave energy is noteworthy because the increase
occurs ju st after the experim ent was purged with air to atm ospheric pressure.
Any
conditioning effects such as removal o f water vapor from the w alls o f the experiment by
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
74
the e-beam during the base vacuum case and argon backfill case were most likely lost
during the purge.
0.9
0.8
0 .7
0.6
SO
§ 0 .5
u
j
fc
o
0.4
2
0.3
0.2
0.1
fc
ca
uj —
£ E 'O
= uj
oo =
2
-> > is
5 8£
O
S3
5
ca
5-
U
uj
LuJ O
®—
? 5 “>
^« O
| JM
i «CS
O<
O
f“I
U*
2
C
Q
LU
Li. —
o
o
•
r=
^(J
\0
•-
V)
©
m ffl
o J=
C/3
«
O
tfc oo
C/3 <N
o
H
ca
tu
u.
i
|
3 ^
a
S
5
<
o2
^
®
£
VI
aUJ
UJ
0o =
**- 5
• y C/J
iS £©
n O
-“ tc o^
O c/3 m
oo
0
Figure 4.16. Summary o f microwave energy production for each backfill case examined.
The standard deviation is included for each average. Arrows below graph indicate purge
o f experim ent to atmospheric pressure with air.
Shot-averaged plots o f the m icrowave power signal are shown in Figure 4.18 for
the base vacuum case, argon backfill case, and SF6 backfill case. The base vacuum case
(trace a) has two thin peaks o f m icrowave power.
The shot-averaged plot o f the
m icrow ave pow er signal for the argon backfill (trace b) shows that the microwave
pulselength o f each peak has been increased. The SF6 case (trace c) shows even larger
increases than the argon case. It appears from the data that the argon and SF6 backfills
are possibly suppressing water vapor that reduced the microwave pulselength o f the base
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
75
vacuum case. The properties o f the SF6 as an insulating gas and its ability to scavenge
electrons support the increases seen in the experiment. However, the result for the argon
backfill is not understood.
14
12
ZJ
o>
C3
/■*
CS
>
a
£
- Y
tw C
-
O
3
m
*
>
i
f
Z «J
S3 !/3
o fl ^
Figure 4.17.
t:
o
a
V
U_
0
UJ
e2
UJ
?
i;
-2
-c
flQ c/3
^O•2< 0<N
0
^
fc
£
f2
CQ *0
U
. —0
§ £ • *3 in
“*5
o 2 r0
en o
C/5
£N
OW
U J
UJ
00
OO
aUJ
m
u.m —
t
£
co
t:
f2
w*»
0
u j
_U J
frl
U.
0
8 1 “
■
«
i
f
> 2
ao
oo
.c:
^o CS
I no
§ C -
. -s «
^cm C
"u 0o
—. ^
C/3
©
U.
o
C/3 r - i
Summary o f peak microwave power for each backfill case
examined. The standard deviation is included for each average.
Arrows below graph
indicate purge o f experiment to atm ospheric pressure with air.
The second set o f data for the backfill experiment consists o f a base vacuum case
and SF6 backfill case. From Figure 4.16 and Figure 4.17, the average microwave energy
increased by 7% and the average peak microwave power increased by 2% for the SF6
case in com parison to the initial base vacuum case.
Only the change in average
microwave energy is significant w ith a statistical confidence o f 97.6% from ANON A.
The change in average peak m icrow ave pow er is not significant from ANOVA with a
statistical confidence o f 13%. From Figure 4.19, it is seen in a plot o f shot-averaged
m icrowave pow er signals for each case, that the power is divided into two peaks. The
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
76
additional energy o f the SF6 case (trace b) is a result o f the extension o f the length o f the
second pulse’s peak power region in com parison to the base vacuum case (trace a).
>
'-3
£
2
<o
>
2
Time [100 ns/div]
Figure 4.18.
Shot-averaged m icrow ave pow er signals for the a) Base vacuum case (5
shots, -4X1CT6 Torr), b) Argon backfill case (28 shots, 1.6xl0'5 Torr), and c) SF6 backfill
case (28 shots, 1.6x1 O'5 Torr) from February 16, 2000.
The increase in energy production o f the base vacuum case on February 18, 2000, in
com parison to February 16, 2000 shown in Figure 4.16 appears to be due to conditioning
o f the device caused by the SF6- The pumpdown time was longer for the lower energy
case, w hich is generally not observed. The reason for this is not understood.
The SF6 backfilling did show improvement to the microwave energy production
o f the experim ent, but is still inconclusive. The use o f SF6 backfill gas to suppress pulse
shortening w ould be better studied on a repetitive pulse device, as the SF6 appears to
condition the device. The num ber o f shots required for statistical purposes to understand
this m echanism are too cumbersome for a high power accelerator, given the amount o f
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
77
time required to charge the accelerator as well as the maintenance required after several
hundred shots.
A better plan o f attack would be to use a repetitive device to test the
characteristics o f the SF6 backfill and to understand the mechanisms involved. If the SF6
backfill were shown conclusively to indeed improve pulse shortening characteristics, the
SF6 backfill could be examined on a single-shot HPM device driven by a high power
accelerator, such as MELBA.
a.
Time [100 ns/div]
Figure 4.19. Shot-averaged microwave power signals for the a) Base vacuum case (30
shots, ~4xl O'6 Torr) and b) SF6 backfill case (30 shots, 1.6xl0"5 Torr) from February 18,
2000 .
A fter several experiments using the SF6 backfill, yellowish pow der contaminant
was visible inside the cavity/waveguide structure. It is most likely sulfur, or a sulfur
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
78
compound. Analysis o f this powder was not performed, but it appears to be sim ilar to the
residue found in high voltage switches that use SF6 insulating gas, as in the MELBA
capacitor bank. Obviously, SF6 backfilling would be unsuitable for thermionic cathode
devices.
4.4 Microwave Pulse Shape and Voltage Fluctuations
Shot-averaged microwave power signals presented in this dissertation exhibit
distinct, determ inistic features. More precisely, they are not smooth pulses that simply
turn on and then turn off, but rather fluctuate in a somewhat reproducible shape near the
beginning o f the microwave pulse. The likely cause is fluctuations o f the e-beam diode
voltage [PET98].
The consistent fluctuations o f the diode voltage and the distinct,
correlated features o f the shot-averaged microwave power signal are shown in Figure
4.20, where both signals are averaged for 30 shots. The two microwave peaks occur at
approximately the same voltage, at which the coupling is better for the gyrotron. As the
e-beam diode voltage fluctuates, the effective coupling between the e-beam and
waveguide modes changes.
The microwave power increases and decreases as the
effective coupling improves and worsens, respectively.
The consistency o f the MELBA voltage fluctuations yields the consistent features
of the m icrowave pow er signal as seen when averaged over many shots.
Due to this
strong coupling o f the voltage to microwave emission, the fluctuations o f the e-beam
diode voltage m ust be minimized to permit the longest microwave pulses.
The loaded cavity quality factor, Q, may be used to compare with the observed
time scale. Using the analysis for the transient response o f passive circuits [SCH93] and
several equivalent definitions for Q [POZ98], Equation 4.1 is obtained. The decay time
o f the circuit can be examined from the exponential term in Equation 4.1.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
For Q =505 (from Section 3.6) and (oJ2n=2.55 GHz, the time for the m icrowave
energy stored in the cavity to decay is - 6 0 ns from the exponential decay term o f
Equation 4.1. Therefore, the observed time scale o f the changes in microwave pow er is
reasonable, once the diode voltage departs from the optimal value.
Microwave Power
[1 MW/div]
MELBA Voltage
[-169 kV/div]
V\
Time [100 ns/div]
Figure 4.20. M ELBA voltage and microwave pow er signal averaged over 30 shots.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
CH APTER 5
C O N C L U SIO N S
Optical em ission spectroscopy has been used to conclusively link the plasm a to
pulse shortening in the high power microwave (HPM ) coaxial gyrotron studied in this
dissertation. Experim ental results show a roughly linear relationship between premature
microwave pow er cu to ff and growth o f H-alpha optical emission. Heterodyne m ixer data
show that the m icrow ave cavity is still oscillating and producing microwaves beyond
apparent cutoff, yet only low levels o f m icrow ave power are radiated from the
experiment. These observations support the theory that microwaves are being attenuated
as the plasma approaches critical density (-8 x 1 0 10 cm '3 for 2.55 GHz microwaves) for
this HPM coaxial gyrotron.
RF plasm a cleaning has been used to m itigate the pulse shortening characteristic
o f this HPM device. The plasm a discharge was driven with a 50 W RF (13.56 M Hz)
pow er supply. Gas pressures o f 15 mTorr and 25 m T orr were examined. The 25 m Torr
discharge pressure yielded better results.
Improvements in the average microwave
energy output o f this device ranged from 15% to 245%.
The average microwave
pulselength increase ranged from 21% to 186%. The average peak microwave pow er for
the different cases show ed varied results and was not significant for all but one case
comparison. Analysis o f Variance (ANOVA) was used to show that RF plasm a cleaning
statistically improved the microwave energy and m icrow ave pulse length in the four data
sets examined.
The m echanism for this improvement is believed to be sputtering o f the
contaminants from the cavity/waveguide structure b y ions produced in the nitrogen RF
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
81
plasm a discharge. The contam inant is subsequently removed by the vacuum pumps, thus
reducing the amount o f H 2O molecules available to contribute to the plasma.
This is
supported by the reduction o f H-alpha optical emission measured during the RP plasma
cleaning cases examined in this experiment.
Gas backfilling experiments had limited results. The SF6 backfilling did show
improvement to the energy production o f the experiment, but is still inconclusive. The
use o f SF6 backfilling to suppress pulse shortening would be better studied first on a
repetitive pulse device, since the SF6 appears to condition the device. If positive results
were demonstrated, the technique could then be applied to a single-shot HPM device/high
power accelerator.
A
yellowish
powder contaminant
was
visible
cavity/waveguide structure after use o f SF6 as a backfill gas.
inside
the
The effect o f this
contam inant would need further investigation.
Future work o f this research would entail optim izing the gas pressure used for the
discharge as well as the cleaning time.
Optical emission spectroscopy could be
performed on other HPM devices (e.g., Air Force-MILO) in order to examine the effects
o f plasm a on the microwave output o f the device. If harmful plasma effects are observed,
the RF plasm a cleaning techniques demonstrated here could be adapted and applied,
where a 50% improvement in energy would be a significant contribution. This process
would be best suited for devices that are not bakable “hard-tubes” and operate under
moderate vacuum pressures ^ l O -6 Torr). This process would especially be useful for
devices in L-band and S-band frequency ranges where the critical plasma density is easily
achievable due to the low operating frequency o f the device.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A PPE N D IC E S
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
83
APPENDIX A
DATA ACQUISITION SETUP (FINAL)
I)
M ELBA Voltage: voltage m onitor —> resistive divider —> RG -58/U cable to
screen room —> lOx attenuator (RM G B l) —> 50 Q Splitter —> Tekronix DSA602A
oscilloscope internally term inated at 50 Q (400 MHz)
II)
Diode current: B-Dot loop —> RG-58/U cable to screen room —> 50 Q terminator
(A 3) —» RC integrator (#5) w ith t = 20 ps, Tekronix DSA 602 oscilloscope
term inated at 1 MQ (20 M Hz)
III)
A perture current: Rogowski coil —> RG-58/U cable to screen room —> lOx
attenuator ( D 1) -> lOx attenuator (C2) —> 50 Q term inator (D3) —> Tekronix
DSA602 oscilloscope terminated at 1 M Q (100 MHz)
IV)
C avity entrance current: Rogow ski coil —> RG-58/U cable to screen room —> lOx
attenuator ( A l) -> Tekronix DSA 602 oscilloscope internally terminated at 50 Q
(100 MHz)
V)
Cavity exit current: Rogowski coil —> RG-58/U cable to screen room —> lOx
attenuator (6/22/97) —> Tekronix DSA602 oscilloscope internally terminated at 50
Q (100 M Hz)
VI)
Diode m agnet current: shunt resistor (0.0477 Q) —> RG -58/U cable to screen
room —►Tekronix TDS210 oscilloscope terminated at 1 M Q (20 M Hz)
VII)
Solenoid current: Pearson Coil (0 .0 1 V/A) -> RG-58/U cable to screen room —>
Tekronix TDS210 oscilloscope terminated at 1 M Q (20 M Hz)
VIII) V ertically polarized S-Band m icrow ave signal: S-Band w aveguide at white tank
—>■30dB Directional Coupler —*■S-Band Waveguide to N -type A dapter —>
R G -214/U cable to screen room —►0-20dB HP coaxial attenuators —>
M icrolab/FX R LA-40N 4 G H z low pass filter —> Narda 4503 diode crystal
detector —> RG-58/U cable -> Tekronix DSA602A oscilloscope internally
term inated at 50 Q (100 M Hz)
IX)
Heterodyne mixer schematic shown in Figure A. 1
X)
O ptical Emission Spectroscopy PM Tube schematic show n in Figure A.2
XI)
O ptical Emission Spectroscopy Intensified CCD schem atic shown in Figure A.3
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
84
S-Band W aveguide
in W hite T ank
HP Variable A ttenuator
Model S375A (0-20dB)
HP Adapter
Model S28IA
R G 2I4/U Cable
to Screen Room
HP 8350B Sw eep O scillator
with 83590A RF Plug-in
2-20 GHz
M icrolab /FX R LA-tON
4 GHz Low Pass Filter
UTE M icrow ave Inc.
2-4GH z Isolator
C T -3040-O T (S/N R8638)
M idwest M icrow ave
3 dB Splitter M odel 2533
O m nispcctra M od 20600-10
10 dB SM A A ttenuator
Tek TDS3052 Oscilloscope
Term inated at 50 Cl
Full Band W idth (500 M H z)
M iniCircuits ZEM -4300M H
M ixer (15542/0-9704)
O m nispcctra M od 20600-10
10 dB SMA A ttenuator
Figure A. 1. Heterodyne mixer schematic for time-frequency analysis.
COM2
DELL GXMT5133
Computer
COM I
Princeton Instruments
A/D Convertor/Buffer
Computer Card
P rinceton Instrum ents
D e tec to r C o n tro lle r
S T -I3 8 S
Trig In I i
ARC Motor Controller
Unit -750
Input Ports
Acton Research Corp (ARC)
SpectraPro-750
0.75 m eter Spectrograph
'HjO Supply Line
•N, Supply Line
Princeton Instruments
IC C D C am era
Detector
* i ICCD
HV
H ,0 Return Line
Not Scan
Trigger Pulse —^
Not Inhibit
RS-232
DSA602A terminated
at 1 MO (20 MHz)
Monitor Gate 1
TDS220
(I MC2.20.MHz)
Princeton Instrum ents
Pulse G enerator
PG-200
Figure A.2. Optical emission spectroscopy schematic for Intensified CCD data.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Gate
Out I
85
DELL GXMT 5133
Computer
COM2
ARC Motor Controller
Unit #750
Input Ports
Acton Research Corp (ARC)
SpectraPro-750
0.75 meter M onochrom ator
Filter Wheel Assembly
ARC Filter Wheel
Controller FA-448
PM T u b e ----------- »
HV
Tektronix
602(A ) DSA
s
o
n
Terminators
ARC Motor Controller
Unit #275
Acton Research Corp (A RC)
SpectraPro-275
0.275 m eter M onochrom ator
PM Tube Current Signal
ARC High Voltage
Supply PHV-400
Input Port
Flulcc 412B High Voltage
Power Supply
PM Tube
Figure A.3. O ptical emission spectroscopy schematic for photom ultiplier tube data.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
86
APPENDIX B
MELBA TRIGGERING SCHEMATIC
S v ttr o o D o r a e r - 101 P u k e G c t « r a » r
S y iir a i D p w c f - 101 P u i^ G tn g ,
MELBA Trigger G enerator
Crowbar
3 0 0 V 'O U T P U T
B N C M odel 7050
D ig ita l D e la y G c n o a io r
MARX Trigger
^
M A R X T rig h ( P T -S f|
^
C r o w tm r T n g h ( F T - 55
B N C M odel 7050
D ig ita l D e la y G e n e a t e r
“ **“1
B N C M odel 7050
D ig ita l D e l a y G e n O a te s '
30cv o u t p u t
000.0 ms
00.96 (is j
116.0 ms j
| 00.00 fas |
108.3 ms 1
00.00 (is
-0
Msmx”
r csntA i ^
|
M q ^n ct O ia lla K x p e T n g g a*
Figure B. 1. M ELBA triggering schematic.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
«x»
C a lm ~ i d \
j
r .u i
j
87
APPENDIX C
M ATLAB SUBRO U TINE FOR ANALYSIS OF ROGOWSKI COIL SIGNALS
% I n t e g r a t e t h e ENTC s i g n a l t o c u r r e n t (kA )
e n tc in t = z e ro s ( s i z e ( e n t c (1:e n d ,2 ) ) ) ;
e n t c i n t = c u m t r a p z ( e n t c ( 1 : e n d , 2) ) * d t * 2 . 9 E 6 ,e n t c ( 1 : e n d , 2 ) = e n t c i n t ,-
% get slope information for ENTC and take out offset
figure(1)
plot(entc(1:end,2))
[xxl,yyl] = ginput(l);
s lope=yy 1/xxl
for p = l :2048
entc(p,2)=entc(p,2 ) -p*slope;
end
plot(entc(1:end,2))
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
88
APPENDIX D
SCANNING ELECTRON M ICROSCOPE ANALYSIS OF RF PLASMA CLEANED
COPPER FROM UNIVERSITY OF MICHIGAN/S LAC COLLABORATION
Surface modification was examined in a collaborative effort between the
University o f M ichigan and Stanford Linear Accelerator Center (SLAC) within the Air
Force Office o f Scientific Research HPM M ultidisciplinary University Research
Initiative. G.P. Scheitrum and L.L. Laurent were the two collaborators that interacted
from SLAC.
The goal was to examine RF plasma cleaning effects on high voltage
breakdown in X-Band microwave cavities.
Some insight and understanding can be
gained o f RF plasm a cleaning o f the cavity/waveguide structure presented in this
dissertation through use o f initial plasma cleaning results studied with the University o f
Michigan scanning electron microscope (SEM, Philips XL30 FEG).
The collaboration was set up to examine possible improvements o f the microwave
breakdown holdoff o f SLAC X-band microwave breakdown experiments through the use
o f RF plasma cleaning techniques being investigated at the University o f Michigan,
Intense Energy Beam Interaction Laboratory.
Data from this collaboration are presented here to show how an RF plasma can
modify the surface o f a piece o f OFE101 copper as examined with an SEM. This assists
in understanding the effect that the nitrogen RF plasma discharge can have on the surface
o f the high power m icrowave coaxial gyrotron studied in this work.
A picture o f the experimental configuration is shown in Figure D .l.
The RF
power is fed into the SLAC OFE101 nose piece through the tube seen at the top o f the
picture. The system is pumped with a cryogenic pump located o ff to the left side o f the
picture. The nose piece is suspended from the top plate o f the vacuum chamber with a
stainless steel rod. The ground electrode is made from a solid 2-3/4” OFE copper gasket
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
89
purchased from MDC. The spacing between the SLAC nose piece and ground electrode
was ~5 cm . The argon gas used for the R F plasm a discharge is filtered w ith a Millipore
(W FRG02 M ini XL) filter.
T o V acuum System
(Cryogenic Pump)
Hot Lead for
RF Discharge
SLAC
Nose Piece
Ground for
RF Discharge
Figure D .l. Picture o f the University o f M ichigan experiment for argon RF plasma
discharge cleaning o f SLAC nose pieces.
The data presented here were from RF plasma cleaning tests perform ed at the
University o f M ichigan on a previously dam aged nose piece to check and practice the
technique that was going to be used for new SLAC nose pieces.
The following
University o f M ichigan SEM pictures (Figures D.2 and D.3) show two different regions
on the surface o f the nose piece before RF plasma cleaning, after 20 m inutes of RF
plasma cleaning at 50 W, and after an additional 20 minutes o f RF plasm a cleaning at
100 W. Both cleaning plasmas were with argon at a discharge pressure o f ~5 mTorr.
In Figure D.2 is shown a region o f the nose piece where there is a large amount o f
damage for the different stages o f RF plasm a cleaning. The SEM picture showing the
region before RF cleaning has smooth features with several large particles and shadowed
regions believed to be hydrocarbon contam inants. After 50 W o f RF plasm a cleaning
with argon for 20 minutes, the hydrocarbon contam inants are removed and small particles
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
90
(<1 p m ) are visible on the surface. After 100 W o f RF plasma cleaning with argon for 20
m inutes after the 50 W RF plasm a cleaning, the surface features appear reduced and a
large num ber o f small particles (~1 pm ) are present. It appears that the contaminants are
rem oved and surface features are reduced at the cost o f adding small particulate to the
surface.
In Figure D.3, a region with a trench in the surface is shown before RF cleaning.
A fter RF cleaning at 50 W with argon for 20 m inutes, the surface features o f the trench
appear cleaned and smoothed. Small particles (<1 pm ) are visible in small num bers on
the surface.
After 100 W o f RF plasma cleaning with argon for 20 minutes after the
50 W RF plasm a cleaning, the trench in this region appears further cleaned and
smoothed. A large num ber o f small particles (<1 pm ) are visible on the surface.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
91
Area Alpha
Before RF plasma cleaning
Area Alpha
After 50 W argon
RF plasma cleaning
for 20 minutes (~5 mTorr)
Area Alpha
After 100 W argon RF
plasma cleaning
for 20 minutes (~5 mTorr)
Figure D.2. University o f M ichigan SEM pictures o f SLAC nose piece area alpha for
base condition, 50 W o f argon RF plasm a cleaning, and then 100 W o f argon RF plasma
cleaning. Cleaning was performed for 20 minutes at a discharge pressure o f ~5 mTorr.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
92
Area Beta
Before RF plasma cleaning
Area Beta
After 50 W argon
RF plasma cleaning
for 20 minutes (~5 mTorr)
Area Beta
After 100 W argon RF
plasma cleaning
for 20 minutes (~5 mTorr)
Figure D.3. University o f M ichigan SEM pictures o f SLAC nose piece area beta for base
condition, 50 W o f argon RF plasma cleaning, and then 100 W o f argon RF plasma
cleaning. Cleaning was performed for 20 minutes at a discharge pressure o f ~5 mTorr.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
93
APPENDIX E
ANALYSIS OF VARIANCE (ANOVA)
Analysis o f Variance (ANOVA) is a statistical tool for testing the equality o f a set
o f means by m easuring the variation within each data set and the variation o f the entire
population. The technique yields a confidence level for assuming that the means are not
equal.
The calculations can be performed with a spreadsheet program such as
Microsoft® Excel as well as statistical analysis packages.
For convenience, ANOVA
calculations were performed in Microsoft® Excel for this dissertation. The output from
the analysis tool appears below.
A nova: S in g le F acto r
SUM M ARY
Groups
Count
Average
Sum
4
15
E-Base v acu u m 2
E-Post RF
1.854
23.979
0.4635
1.5986
Variance
0.019812
0.610667
ANOVA
Source o f Variation
SS
df
B etw een G roups
W ithin G roups
4.068796
8.608779
1
17
Total
12.67757
18
MS
F
P-value
4 .0 6 8 7 9 6
0 .5 0 6 3 9 9
8.034767
0.011441
F crit
4.451323
From the P-value listed above, the statistical confidence level can be calculated as
(1 - {P-value)) x 100%. For this example, the A N O V A statistical confidence that the two
means exam ined (E-Base vacuum 2 and E-Post RF) are different is 98.86%.
The references used for ANOVA in this dissertation included [OTT93], [FRE97],
and [CRC90],
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
BIBLIOGRAPHY
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
95
BIBLIOGRAPHY
ABR77
E.A. Abramyan, B.A. Altercorp, and G.D. Kuleshov, “Microsecond
Intensive Electron Beams,” in Proc o f the 2nd Int. Tropical Conference on
High Power Electron and Ion Beam Research and Technology, vol. 2, pp.
755-760, 1977.
AGE96
F.J. Agee, S.E. Calico, K. Hendricks, M. Haworth, and T. Spencer, “Pulse
Shortening in the magnetically insulated line oscillator (MILO),”
Processings o f the SPIE, vol. 2843, pp. 144-152, 1996.
AGE98
F.J. Agee, “Evolution o f Pulse Shortening Research in Narrow Band, High
Pow er Microwave Sources,” IEEE Transactions on Plasma Science
Special Issue on High Power M icrowave Generation, vol. 26, no. 3, pp.
235-245, June 1998.
ALL63
A.P. Allis, S.J. Buchsbaum, and A. Bers, Waves in Anisotropic Plasmas,
M.I.T. Press, 1963.
BEN97
J. Benford and G. Benford, “Survey o f Pulse Shortening in High-Power
M icrowave Sources,” IEEE Transactions on Plasma Science Special Issue
on Pulsed Power Science and Technology, vol. 25, no. 2, pp. 311-317,
Apr. 1997.
BER95
A. Berman, “Water vapor in vacuum systems,” Vacuum, vol. 47, no. 4,
pp. 327-332, 1996.
BRAOO
M.L. Brake, University o f M ichigan, personal communication, 2000.
B R 094
S.C. Brown, Basic Data o f Plasma Physics, AIP Press, New York, 1994
CAR89
Y. Carmel, K. Minami, R.A. Kehs, W.W. Destler, V.L. Granatstein, D.
Abe, and W.L. Lou, “Dem onstration o f Efficiency Enhancements in a
High-Power Backward-Wave O scillator by Plasma Injection,” Phys. Rev.
Letters vol. 62, pp. 2389-2392, 1989.
CHE84
F.F. Chen, Introduction to Plasma Physics and Controlled Fusion-Volume
I: Plasma Physics, second edition, Plenum Press, New York, 1984.
CRC90
W. H. Beyer (ed.), CRC Standard Probability and Statistics Tables and
Formulae, CRC Press, Ann Arbor, 1990.
CR 083
R.W. Crompton and G.N. Haddad, “Thermal Electron Attachment to SF6
and CFCI3 ,” Australian Journal o f Physics, vol. 36, pp 15-25, 1983.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
96
CUN89
M.E. Cuneo, “Characterization o f the Time-Evolution o f a M icrosecond
Electron Beam Diode with Anode Effects,” Ph. D. Dissertation, University
o f M ichigan, 1989.
CUN99
M.E. Cuneo, “Effect o f electrode contamination, cleaning and
conditioning on high-energy pulsed-power device perform ance,” IEEE
Transactions on Dielectics and Electrical Insulation, vol. 6, no. 4, pp. 469485, 1999.
DUS62
S. Dushm an and J.M. Lafferty (ed.), Scientific Foundations o f Vacuum
Technique, second edition, John W iley & Sons, New York, 1962.
EBB90
D.D. Ebbing and M.S. W righton, General Chemistry, third edition,
Houghton M ifflin Co., Boston, 1990.
FRE97
R.J. Freund and W.J. Wilson, Statistical Methods, revised edition,
Academic Press, Inc., New York, 1997.
GIL85
R.M. Gilgenbach, L.D. Horton, R.F. Lucey, Jr., S. Bidwell, M.E. Cuneo, J.
Miller, and L. Smutek, “M icrosecond Electron Beam Diode Closure
Experim ents,” invited paper in Digest o f Technical Papers o f the Fifth
IEEE Pulsed Power Conference, pp. 126-132, June 1985.
GIL91
R.M. Gilgenbach and P.L.G. Ventzek, “Dynamics o f excim er laserA blated aluminum-neutral-atom plum e measured by dye-laser-resonanceabsorption photography,” Appl. Phys. Lett., vol. 58, p. 1597, 1991.
GIL98
R.M. Gilgenbach, J.M. Hochman, R.L. Jaynes, W.E. Cohen, J.I.
Rintamaki, C.W. Peters, D.E. Vollers, Y.Y. Lau, and T.A. Spencer,
“Optical Spectroscopy o f Plasma in High Power M icrowave Pulse
Shortening Experiments Driven by a p s e-Beam,” IEEE Transactions on
Plasm a Science Special Issue on High Power Microwave Generation, vol.
26, no. 3, pp. 282-289, June 1998.
GOE98
D.M. Goebel, R.W. Schumacher, and R.L. Eisenhart, “Performance and
Pulse Shortening Effects in a 200-kV PASOTRON HPM Source,” IEEE
Transactions on Plasma Science Special Issue on High Power M icrowave
Generation, vol. 26, no. 3, pp. 354-365, June 1998.
GOE98b
D.M. Goebel, “Pulse Shortening C auses in High Power BWO and TW T
M icrowave Sources,” IEEE Transactions on Plasma Science Special Issue
on H igh Power Microwave Generation, vol. 26, no. 3, pp. 263-274, June
1998.
GOE99
D.M. Goebel, Y. Carmel, and G.S. Nusinovich, “Advances in plasmafilled microwave sources,” Physics o f Plasmas, vol. 6, no. 5, pp. 22252232, M ay 1999.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
97
GRA98
C. Grabowski, J.M . Gahl, and E. Scham iloglu, “Initial Plasma-Filled
Backward-W ave Oscillator Experim ents Using a Cathode-M ounted
Plasma Prefill Source,” IEEE Transactions on Plasma Science Special
Issue on High Pow er M icrowave Generation, vol. 26, no. 3, pp. 653-668,
June 1998.
HAR91
J. Harvell and P. Lessard, “M anaging w ater vapor in a vacuum process,”
Semicond Int, vol. 14, no. 7, pp. 88-90, June 1991.
HAW 98
M.D. Haworth, G. Baca, J.N. Benford, T. Englert, K. Hackett, K..J.
Hendricks, D. Henley, M. LaCour, R.W. Lemke, D. Price, D. Ralph, M.
Sena, D. Shiflfler, and T. A. Spencer, “Significant Pulse-Lengthening in a
Multigigwatt M agnetically Insulated Transm ission Line Oscillator,” IEEE
Transactions on Plasm a Science Special Issue on High Power M icrowave
Generation, vol. 26, no. 3, pp. 312-319, June 1998.
HAY64
D.O. Hayward and B.M.W. Trapnell, Chemisorption, second edition,
Butterworths, 1964.
HEG98
F. Hegeler, C. Grabowski, and E. Scham iloglu, “Electron Density
Measurements During M icrowave Generation in a High Power BackwardWave O scillator,” IEEE Transactions on Plasm a Science Special Issue on
High Pow er M icrowave Generation, vol. 26, no. 3, pp. 275-281, June
1998.
HER44
G. Herzberg, A tom ic Spectra an d Atom ic Structure, Dover Publications,
New York, 1944.
HOC98
J.M. Hochman, “M icrowave Em ission o f Large and Small Orbit
Rectangular Gyrotron Devices,”
Ph. D. Dissertation, University o f
Michigan, 1998.
JAY99
R.L. Jaynes, R.M . Gilgenbach, J.M . Hochm an, N. Eidietis, J.I. Rintamaki,
W.E. Cohen, C.W . Peters, Y.Y. Lau, and T.A. Spencer, “Velocity Ratio
M easurement Diagnostics and Sim ulations o f a Relativistic Electron Beam
in an Axis Encircling Gyrotron,” IEEE Transactions on Plasma Science
Special Issue on Images in Plasm a Science, vol. 27, no. 1, pp. 136-137,
Feb. 1999.
JAYOO
R.L. Jaynes, “Generation o f High Power M icrowaves in a Large Orbit
Coaxial G yrotron,” Ph. D. Dissertation, University o f Michigan, 2000.
JE 0 9 2
J. Jeong, W.J. W illiams, “Alias-Free Generalized Discrete-Time TimeFrequency Distributions,” IEEE Transactions on Signal Processing, vol.
40, no. 11, 2757-2765, Nov. 1992.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
98
KLI79
L.E. Kline, D.K. Davies, C.L. Chen, and P.J. Chantry, “Dielectric
properties o f SF6 and SF6 mixtures predicted from basic data,” Journal o f
Applied Physics, vol. 50, no. 11, 6789-6796, Nov. 1979.
K.OB98
S. Kobayashi, T.M. Antonsen, Jr., and G. S. Nusinovich, “Linear Theory
o f a Plasma Loaded, Helix Type, Slow Wave A m plifier,” ,” IEEE
Transactions on Plasma Science Special Issue on High Power Microwave
Generation, vol. 26, no. 3, pp. 669-679, June 1998.
KRA73
N.A. Krall and A.W. Trivelpiece, Principles o f Plasma Physics, McGrawHill Book Co., New York, 1973.
LAU97
Y.Y. Lau, Electron-Coherent Radiation course notes (NERS673), Winter
Term 1997.
LIE94
M.A. Lieberman and A.J. Lichtenberg, Principles o f Plasma Discharges
and Materials Processing, John Wiley & Sons, New York, 1994.
LUC88
R.F. Lucey, Jr., “Long-Pulse Relativistic Electron Beam Generation and
Propagation in Gases and in UV Laser Ionized Channels,” Ph. D.
Dissertation, University o f M ichigan, 1988.
MIL89
J.D. Miller, “Transport o f Long-Pulse Relativistic Electron Beams in
Preformed Plasma Channels in the Ion Focus Regim e,” Ph. D.
Dissertation, University o f M ichigan, 1989.
MIL98
R.B. Miller, “Pulse Shortening in High-Peak-Power Reltron Tubes,” IEEE
Transactions on Plasma Science Special Issue on High Power Microwave
Generation, vol. 26, no. 3, pp. 340-347, June 1998.
NUS98
G.S. Nusinovich, Y. Carmel, T.M. Antonsen, Jr., D.M. Goebel, and J.
Santoru, “Recent Progress in the Development o f Plasma-Filled
Traveling-Wave Tubes and Backward-Wave O scillators,” ,” IEEE
Transactions on Plasma Science Special Issue on High Power Microwave
Generation, vol. 26, no. 3, pp. 628-645, June 1998.
OTT93
R.L. Ott, An Introduction to Statistical Methods and Data Analysis, fourth
edition, Duxbury Press, Belmont, 1993.
PAR90
A.P. Paranipe, J.P. M cVittie, and S.A. Self, “A tuned Langm uir probe for
measurements in r f glow discharges,” Journal o f Applied Physics, vol 67,
no. 11, June 1990.
PAS91
M.L. Passow, “M icrowave Resonant-Cavity-Produced Air Discharges,”
IEEE Transactions on Plasm a Science, vol. 19, no. 2, pp. 219-228, Apr.
1991.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
99
PEA84
R.W.B. Pearse and A.G. Gaydon, The Identification o f M olecular
Spectroscopy, fourth edition, Chapman and Hall, New York, 1984.
PET98
C.W. Peters, R.L. Jaynes, Y.Y. Lau, R.M. Gilgenbach, W.J. Williams,
J.M. Hochman, W.E. Cohen, J.I. Rintamaki, D.E. Vollers, and T.A.
Spencer, “Time-frequency analysis o f modulation o f high-power
microwaves by electron-beam voltage fluctuations,” Physical Review E,
vol. 58, no. 5, pp. 6880-6883, Nov. 1998.
POI88
T.D. Pointon and J.S. De Groot, “Particle simulations o f plasma and
dielectric Cerenkov m asers,” Phys. Fluids, vol. 31, pp. 908-915, Apr.
1988.
POZ98
D.M. Pozar, Microwave Engineering, second edition, John W iley & Sons,
New York, 1998.
PRI98
D. Price, J.S. Levine, and J.N. Benford, “Diode Plasma Effects on the
Microwave Pulse Length from Relativistic M agnetrons,” IEEE
Transactions on Plasma Science Special Issue on High Power Microwave
Generation, vol. 26, no. 3, pp. 348-353, June 1998.
PRI98b
D. Price and J.N. Benford, “General Scaling o f Pulse Shortening in
Explosive-Emission-Driven Microwave Sources,” IEEE Transactions on
Plasma Science Special Issue on High Power Microwave Generation, vol.
26, no. 3, pp. 256-262, June 1998.
PSI83
Pulse Sciences Inc., “Electron Beam Accelerator,” PSI-P-83-105, Aug.
1983.
RE 194
M. Reiser, Theory and Design o f Charged Particle Beams, John Wiley &
Sons, New York, 1994.
RIE72
F.F. Rieke and W. Prepejchal, “Ionization Cross Sections o f Gaseous
Atoms and Molecules for High-Energy Electrons and Positrons,” Phys.
Rev. Letters vol. 6, no. 4, pp. 1507-1519, 1972.
RIN99
J.I. Rintamaki, “Effects o f RF Plasma Processing on the Impedance and
Electron Emission Characteristics o f a MV Beam Diode,” Ph. D.
Dissertation, University o f Michigan, 1999.
SCH93
S.E. Schwarz and W.G. Oldham, Electrical Engineering: An Introduction,
second edition, Saunders College Publishing, New York, 1993.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
100
SHK98
A.G. Shkvarunets, S. Kobayashi, Y. Camel, J. Rodgers, T.M. Antonsen,
Jr., L. Duan, V.L. Granatstein, “Operation o f a Relativistic BackwardWave Oscillator Filled with a Preionized High-Density Radially
Inhomogeneous Plasm a,” EEEE Transactions on Plasma Science Special
Issue on High Power M icrowave Generation, vol. 26, no. 3, pp. 646-652,
June 1998.
SHL00
A. Shlapakovski, G. M elnikov, and A. Petrov, “Influence o f near-surface
plasm a layer on the wide-bandwidth dielectric Cherenkov m aser
operation,” Proceedings o f SPIE (Intense M icrowave Pulses VII) vol.
4031,2000.
TAY82
J.R. Taylor, An Introduction to Error Analysis, University Science Books,
M ill Valley (CA), 1982.
TRI59
A.W. Trivelpiece and R.W . Gould, “Space Charge Waves in Cylindrical
Plasma Columns,” Journal o f Applied Physics, vol. 30, no. 11, Nov. 1959.
ZHA94
X. Zhai, E. Garate, R. Prohaska, A. Fisher, and G. Benford, “Electric field
m easurement in a plasm a-filled X-band backw ard wave oscillator,”
Physics Letters A, vol. 186, pp. 330-334, Mar. 1994.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
4 371 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа