2009 Nguyen Thesis

Hydrogen Production in a Radio-Frequency
Plasma Source Operating on Water Vapor
by
Son-Ca Viet Thi Nguyen
A dissertation submitted in partial fulfillment

of the requirements for the degree of
Doctor of Philosophy
(Aerospace Engineering)
in The University of Michigan
2009
Doctoral Committee:
Professor Alec D. Gallimore, Co-Chair

Associate Professor John E. Foster, Co-Chair
Professor Iain D. Boyd
Professor Ronald M. Gilgenbach
Son-Ca Viet Thi Nguyen 2009
c
° All Rights Reserved
To my Mom and Dad,
Lam-Ngoc Thi Le and Truong-Son Viet Nguyen.
And in loving memory of my friend,
Mark de la Cruz Sarsoza.
ii
Acknowledgements
This thesis would not have been possible without the support of my advisors, col-
leagues, friends, and family. I am humbled to have this opportunity to express my deepest
gratitude for the guidance and encouragement that I have received over the years.
Foremost, I would like to thank my advisor, Dr. Alec Gallimore, for giving me the
opportunity to fulfill my educational goal and the freedom to pursue my research interest.
Dr. Gallimore allowed me to make my own mistakes and to learn from them, and he guided
me in the right direction whenever I wandered off path. I value his advices, including those
extended far beyond the subject of research. Without the support of Dr. Gallimore, this
thesis simply would not have been possible. I owe my deepest gratitude to my mentor,
Dr. John Foster, for many reasons: for introducing me to the field of plasma physics,
for introducing me to Dr. Gallimore and encouraging me to pursue my doctoral work
at the University of Michigan, and for our countless meetings where I always left learning
something new. Our official mentorship ended many years ago after my internship at NASA
concluded, but he has continued to serve as my mentor and role model throughout the years.
I am forever indebted to him for his role in my education, and I thank him for showing me
the humor in research, science, and life. I also wish to extend my warmest thanks to my
other committee members, Dr. Iain Boyd and Dr. Ronald Gilgenbach, for reviewing my
thesis.
iii
From the aerospace engineering department, I wish to thank Cindy Enoch and Denise
Phelps for sorting out my paperwork, and Thomas Griffin, David Mclean, and Eric Kirk
for their help with my setup and in other numerous occasions. I thank Professor Bram van
Leer for welcoming me to the department five years ago and continuing to be my confidant.
Research and collaboration are synonyms, and I would like to thank everyone at Elec-
troDynamic Applications, Inc. and the Hydrogen Energy Technology Laboratory for their
collaboration and for loaning their equipment. I am also indebted to a number of agencies
who have financially supported me: the National Science Foundation, Zonta International
through the Amelia Earhart Fellowship, and the Gates Millenium Scholars Program. In ad-
dition, I am thankful for the many teaching opportunities that the Multicultural Engineering
Programs Office offered me and in particular, I would like to thank Darryl Koch.
PEPL was a stimulating learning environment for me, mostly because I was surrounded
by highly motivated colleagues who were always willing to help each other. In particu-
lar, I wish to acknowledge the following individuals for training me, for showing me the
proper lab techniques and some short-cuts, and for their valuable input: Tim Smith, Joshua
Rovey, Allen Victor, Jesse Linnell, Robert Lobbia, David Kirtley, Daniel Brown, Bryan
Reid, and Kristina Lemmer. I am also very grateful to be in good company with Thomas
Liu, Bailo Ngom, Ricky Tang, Rohit Shastry, Michael McDonald, Laura Spencer, Adam
Shabshelowitz, Raymond Liang, David Huang, and Roland Florenz.
On a personal note, I would like to thank my friends, especially those I met in Ann
Arbor, who have given me the necessary encouragement to finish my thesis and helped
me achieve a balance between work and life through out my five years in graduate school.
In particular, I would like to thank David, Caro, Serena, Cyril, Cecile, Jeff, Sebastien,
Tosh, Tina, Kiran, Bryan, Arlyne, Nicolas, Andreja, Lisa, Mauricio, and Charleen. Above
iv
all, I would like to thank Pierre-Yves for his patience, humor, and kindness. Because of
him, I am a much better person, and with him, life is more interesting and meaningful. I am
fortunate to have him as a travel partner as we discover the world together. I would also like
to extend my sincere thanks to Pilar, Mathieu, Jean-Philippe, and Marie-Christine for their
encouragement. More importantly, I thank Pilar and Mathieu for bringing into this world
the person with whom I am lucky to share my life. (En particulier, je souhaite remercier
Pilar et Mathieu pour avoir mis au monde la personne avec qui j’ai la chance de partager
ma vie.)
Lastly, I would like to thank my family for their continuous support: my parents, brother
Vinh-Thinh, and sister Thu-Huong. Everything that I have accomplished was possible
because of the opportunities that my parents have created for me through their sacrifices.
The best way for me to express my gratitude to my parents is by repeating a Vietnamese
proverb that I learned when I was a child, but only now understand its full meaning. It is
loosely translated as: The work of a father is as significant as Mt. Thai Son, and the love
of a mother is as bountiful as the spring water flowing from its source.
v
Table of Contents
Dedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvi
List of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Chapter
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Aim of Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
II. Background–Energy . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 Energy and Climate Challenges . . . . . . . . . . . . . . . . . . . 7

2.1.1 Energy Challenge . . . . . . . . . . . . . . . . . . . . . 7
2.1.2 Climate Challenge . . . . . . . . . . . . . . . . . . . . 10
2.1.3 Case Study: United States . . . . . . . . . . . . . . . . 16
2.2 Hydrogen: Benefits and Drawbacks . . . . . . . . . . . . . . . . . 21
2.3 Hydrogen Production Methods . . . . . . . . . . . . . . . . . . . 23
2.3.1 Natural Gas Reforming . . . . . . . . . . . . . . . . . . 24
vi
2.3.2 Partial Oxidation of Methane . . . . . . . . . . . . . . . 25
2.3.3 Gasification . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.4 Conventional Electrolysis . . . . . . . . . . . . . . . . 26
2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
III. Background–Plasma Overview . . . . . . . . . . . . . . . . . . . . . . . 31
3.1 Overview of Plasma Physical Processes in Water Vapor Discharge 31

3.1.1 Collisional Phenomena in Plasmas . . . . . . . . . . . . 31
3.1.2 Mechanisms of Electron and Ion Creation . . . . . . . . 33
3.1.3 Mechanisms of Electron and Ion Destruction . . . . . . 37
3.1.4 Dissociation of Molecules . . . . . . . . . . . . . . . . 46
3.2 Methods of Plasma Production . . . . . . . . . . . . . . . . . . . 47
3.2.1 Joule Heating . . . . . . . . . . . . . . . . . . . . . . . 48
3.2.2 Secondary Electron Emission Heating . . . . . . . . . . 50
3.2.3 Stochastic Heating . . . . . . . . . . . . . . . . . . . . 51
3.2.4 Resonant Wave-Particle Interaction Heating . . . . . . . 53
3.3 Plasma Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.3.1 Direct Current Plasma Sources . . . . . . . . . . . . . . 54
3.3.2 Capacitively-Coupled Plasma Source . . . . . . . . . . 55
3.3.3 Inductively-Coupled Plasma Source . . . . . . . . . . . 55
3.3.4 Helicon Plasma Source . . . . . . . . . . . . . . . . . . 58
3.3.5 Benefits of Radio-Frequency Plasma Sources . . . . . . 60
3.4 Water Plasma Studies . . . . . . . . . . . . . . . . . . . . . . . . 61
3.4.1 General Applications of Water Plasma . . . . . . . . . . 61
3.4.2 Plasma Electrolysis . . . . . . . . . . . . . . . . . . . . 64
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
IV. Experimental Setup and Diagnostics . . . . . . . . . . . . . . . . . . . 67
4.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . 67

4.1.1 Vacuum Chamber . . . . . . . . . . . . . . . . . . . . . 67
4.1.2 Plasma Source . . . . . . . . . . . . . . . . . . . . . . 69
vii
4.1.3 Water Feed System . . . . . . . . . . . . . . . . . . . . 72
4.2 Diagnostics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.2.1 Residual Gas Analysis System . . . . . . . . . . . . . . 74
4.2.2 Langmuir Probe . . . . . . . . . . . . . . . . . . . . . 80
4.2.3 Spectrometer . . . . . . . . . . . . . . . . . . . . . . . 88
V. Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.1 Argon and Water Vapor Plasma Comparison . . . . . . . . . . . . 89

5.1.1 Optical Emission and Residual Gas Analyzer Spectra . . 90
5.1.2 Plasma Densities . . . . . . . . . . . . . . . . . . . . . 93
5.1.3 Electron Temperature . . . . . . . . . . . . . . . . . . . 100
5.1.4 Floating and Plasma Potentials . . . . . . . . . . . . . . 102
5.1.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.2 Effects of Applied Axial Magnetic Field . . . . . . . . . . . . . . 108
5.2.1 Hydrogen Production . . . . . . . . . . . . . . . . . . . 108
5.2.2 Electron and Ion Number Density . . . . . . . . . . . . 111
5.2.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.3 Effect of Water Input Flow Rate . . . . . . . . . . . . . . . . . . . 113
5.3.1 Hydrogen Production . . . . . . . . . . . . . . . . . . . 113
5.3.2 Production of Other Gases . . . . . . . . . . . . . . . . 116
5.3.3 Electron and Ion Number Density . . . . . . . . . . . . 119
5.3.4 Optical Emission Spectra . . . . . . . . . . . . . . . . . 121
5.3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 123
5.4 Conversion and Energy Efficiencies . . . . . . . . . . . . . . . . . 124
5.5 Summary of Experimental Results . . . . . . . . . . . . . . . . . 126
VI. Kinetic Simulation Setup and Results . . . . . . . . . . . . . . . . . . . 129
6.1 Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

6.2 Description of Global Kinetic Model . . . . . . . . . . . . . . . . 130
6.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . 135
6.3.1 Effects of Flow Rate on Electron Number Density . . . 136
viii
6.3.2 Effects of Flow Rate on Electron Temperature . . . . . . 137
6.3.3 Effects of Flow Rate on Energy and Conversion Effi-
ciencies . . . . . . . . . . . . . . . . . . . . . . . . . . 138
6.3.4 Effects of Background Pressure on Energy Efficiency . . 139
6.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
VII. Conclusions and Future Work . . . . . . . . . . . . . . . . . . . . . . . 142
7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

7.1.1 Water Plasma Source Development . . . . . . . . . . . 143
7.1.2 Electronegative and Electropositive Plasmas Comparison 143
7.1.3 Water Plasma Properties and Hydrogen Production Char-
acterization . . . . . . . . . . . . . . . . . . . . . . . . 144
7.1.4 Water Plasma Behaviors . . . . . . . . . . . . . . . . . 145
7.1.5 Computational Work . . . . . . . . . . . . . . . . . . . 147
7.2 Suggestions for Future Work . . . . . . . . . . . . . . . . . . . . 148
7.2.1 Current Experimental Setup . . . . . . . . . . . . . . . 148
7.2.2 New System Development . . . . . . . . . . . . . . . . 149
7.2.3 Other Applications . . . . . . . . . . . . . . . . . . . . 150
Appendix
A. GlobalKin Data File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
ix
List of Figures
Figure
2.1 Total world energy consumption, 1980-2030 . . . . . . . . . . . . . . . . 8
2.2 Total non-OECD energy consumption, 1980-2030 . . . . . . . . . . . . . 8
2.3 Distribution of proven reserves of oil at the end of 2005 in billion barrels . 10
2.4 Distribution of proven reserves of natural gas at the end of 2005 in trillion
cubic meters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.5 Distribution of proven reserves of coal at the end of 2005 in billion tons . 11
2.6 Annual global energy balance . . . . . . . . . . . . . . . . . . . . . . . . 13
2.7 United States and total world’s CO2 emissions from fossil fuels . . . . . . 14
2.8 Vibration modes of CO2 . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.9 Temperature anomaly relative to the mean temperature between 1961 and
1990 [1]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.10 Rise in global average temperature and sea level and decrease in area of
snow cover in Northern Hemisphere . . . . . . . . . . . . . . . . . . . . 17
2.11 Distribution of energy consumption in the US in 2007 . . . . . . . . . . . 18
2.12 World maps reflecting land distribution, energy export, population, and
energy import . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
x
2.13 Schematic diagram of an electrolysis system . . . . . . . . . . . . . . . . 27
2.14 Schematic diagram of the PEM electrolyte . . . . . . . . . . . . . . . . . 28
3.1 Potential curve for molecule AB . . . . . . . . . . . . . . . . . . . . . . 33
3.6 Plasma sheath in a DC plasma discharge . . . . . . . . . . . . . . . . . . 50
3.7 Plasma sheath in a RF plasma discharge . . . . . . . . . . . . . . . . . . 52
3.8 Experimental setup to study decomposition of phenol solutions . . . . . . 61
3.9 Experimental setup to study diamond growth in a water plasma source . . 62
3.10 Experimental setup to study CO2 dissociation . . . . . . . . . . . . . . . 64
3.11 Experimental setup to study hydrogen production in a water plasma source 65
4.1 Drawing of vacuum facility . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.2 Photograph of setup including the magnetic coils, matching network, and
the RF power supply . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.3 Photograph of plasma source at PEPL . . . . . . . . . . . . . . . . . . . 70
4.4 Mapping of magnetic field at 60-A magnet current with an outline of the
quartz tube and an adapter flange. . . . . . . . . . . . . . . . . . . . . . . 71
4.5 Schematic diagram of plasma source . . . . . . . . . . . . . . . . . . . . 71
xi
4.6 Schematic diagram of water delivery system. . . . . . . . . . . . . . . . 73
4.7 Components of the RGA . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.8 Differential pump system for the RGA . . . . . . . . . . . . . . . . . . . 76
4.9 Calibration curves for hydrogen, oxygen, and water . . . . . . . . . . . . 77
4.10 Effect of argon flow rate on hydrogen production . . . . . . . . . . . . . 79
4.11 Effect of argon flow rate on ion and electron densities . . . . . . . . . . . 79
4.12 Effect of argon flow rate on floating and plasma potential . . . . . . . . . 80
4.13 A typical Langmuir probe curve for water vapor plasma . . . . . . . . . . 81
4.14 Semi-log plot of Langmuir probe trace . . . . . . . . . . . . . . . . . . . 85
4.15 Flow chart for Langmuir probe analysis . . . . . . . . . . . . . . . . . . 86
4.16 Diagram of the spectrometer setup . . . . . . . . . . . . . . . . . . . . . 88
5.1 Optical emission spectrum of an argon discharge . . . . . . . . . . . . . 90
5.2 Optical emission spectrum of a water plasma discharge . . . . . . . . . . 91
5.3 Cross-sections of electron impact with water molecule . . . . . . . . . . 92
5.4 RGA spectrum of the water plasma . . . . . . . . . . . . . . . . . . . . . 93
5.5 Ion plasma density of an argon discharge as a function of RF power for

0-A, 30-A, and 60-A applied magnet current . . . . . . . . . . . . . . . . 94
5.6 Ion density of the water plasma as a function of RF power for 0-A, 30-A,
and 60-A applied magnet current . . . . . . . . . . . . . . . . . . . . . . 95
5.7 Electron density of the water plasma as a function of RF power for 0-A,
30-A, and 60-A applied magnet current . . . . . . . . . . . . . . . . . . 96
xii
5.8 Photographs of the water plasma . . . . . . . . . . . . . . . . . . . . . . 98
5.9 Electron temperature of the argon plasma as a function of RF power for

5.10 Electron temperature of the water plasma as a function of RF power for

5.11 Floating and plasma potential of the argon plasma as a function of RF

power for 0-A, 30-A, and 60-A applied magnet current . . . . . . . . . . 103
5.12 Floating and plasma potential of the water plasma as a function of RF

power for 0-A, 30-A, and 60-A applied magnet current . . . . . . . . . . 103
5.13 Hydrogen production rate as a function of RF power for 0-A, 30-A, and
60-A applied magnet current and 75-sccm water input flow rate . . . . . . 109
5.14 Hydrogen production rate as a function of RF power for 0-A, 30-A, and
60-A applied magnet current and 125-sccm water input flow rate . . . . . 110
5.15 The rate of hydrogen production as a function of RF power for 25, 50, 75,
100, and 125-sccm water input flow rates, operating without an applied
magnet current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.16 The rate of hydrogen production as a function of RF power for 25, 50, 75,
100, and 125-sccm water input flow rates, operating with 30-A applied
magnet current . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
5.17 The rate of hydrogen production as a function of water input flow rate for
250 – 1000 W . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
5.18 Production of hydrogen, oxygen, and hydroxyl in a plasma discharge op-

erating on water vapor with 25-sccm water input flow rate and 30-A ap-
plied magnet current . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

erating on water vapor with 75-sccm water input flow rate and 30-A ap-
plied magnet current . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
xiii
erating on water vapor with 125-sccm water input flow rate and 30-A
applied magnet current . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
5.21 Electron density as a function of RF power for 25, 75, and 125-sccm
water input flow rates without an applied magnet current . . . . . . . . . 120
water input flow rates with 30-A applied magnet current . . . . . . . . . . 120
5.23 Ion density as a function of RF power for 25, 75, and 125-sccm water
input flow rates without an applied magnet currentt . . . . . . . . . . . . 121
water input flow rates with 30-A applied magnet current . . . . . . . . . . 122
5.25 Optical emission spectra for 25, 75, and 125-sccm water input flow rates
operating at 500-W RF power without an applied magnet current . . . . . 122
5.26 Optical emission spectra for 25, 75, and 125-sccm water input flow rates
operating at 500-W RF power and 30-A applied magnet current . . . . . . 123
5.27 Conversion efficiency for the case without an applied magnetic field . . . 125
5.28 Conversion efficiency for the case with an applied magnetic field . . . . . 125
5.29 Energy efficiency for the case without an applied magnetic field . . . . . 126
5.30 Energy efficiency for the case with an applied magnetic field . . . . . . . 127
6.1 Schematic diagram of plasma discharge corresponding to computational

model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
6.2 Simulation scheme of Global Kin . . . . . . . . . . . . . . . . . . . . . 132
6.3 Computational and experimental results for electron density as a function

of RF power for 25, 50, 75, 100, and 125-sccm water input flow rates
without an applied magnet current . . . . . . . . . . . . . . . . . . . . . 136
xiv
6.4 Computational and experimental results for electron temperature as a
function of RF power for 25, 50, 75, 100, and 125-sccm water input flow
rates without an applied magnet current . . . . . . . . . . . . . . . . . . 137
6.5 Computational and experimental results for energy efficiency as a func-

tion of RF power for 25, 50, 75, 100, and 125-sccm water input flow rates
without an applied magnet current . . . . . . . . . . . . . . . . . . . . . 138
6.6 Energy efficiency for operation with 250-W RF power as a function of

water vapor input flow rate for 50, 100, 250, and 500 mtorr . . . . . . . . 140
xv
List of Tables
Table
3.1 Ionization potential energy . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2 Electron affinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.3 Dissociative electron attachment . . . . . . . . . . . . . . . . . . . . . . 42
3.4 Three-body electron attachment . . . . . . . . . . . . . . . . . . . . . . 42
3.5 Associative detachment . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.6 Ion-ion recombination . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.1 Electron impact reactions with H2 O . . . . . . . . . . . . . . . . . . . . 92
xvi
List of Symbols
~a Acceleration vector
aij Stoichiometric coefficient of species i in reaction j
Ap Surface area of the probe
As Surface area of the sheath
Di Diffusivity of species i
Da,i Ambipolar diffusivity of species i
E Electric field
EA Activation energy
F~ Force vector
fij Returned fraction of species j as species i from the wall
Hj Enthalpy of species j
kj Reaction rate coefficient for reaction j
L Length of plasma discharge
Mi Mass of species i
ṁ Mass flow rate
N Total number density
Ni Number density of species i
ne Electron number density
ni Ion number density
P Total pressure
pi Partial pressure of species i
Pd Power deposited
R Radius of plasma discharge
~r Position vector
Ru Universal gas constant
Si Reaction source term
T Temperature
Tg Gas temperature
Te Electron temperature
xvii
Tw Wall temperature
t Time
V~ Velocity vector
γj Wall sticking coefficient of species j
γs e Secondary electron emission coefficient
² Electron energy
κ Thermal diffusivity
Λ Diffusion length
λD Debye length
νmi Momentum transfer collision frequency between electrons
and species i
σ Electron impact cross section
τ Mean time between collisions with neutral atoms
ω Frequency of plasma oscillation
Physical Constants
kB Boltzmann constant, 1.3807 × 10−23 J/K

e, q Charge of an electron, 1.6022 × 10−19 C
me Mass of an electron, 9.1095 × 10−31 kg
²o Permittivity of free space, 8.8542 × 10−12 F/m
µo Permeability of free space, 4π × 10−7 F/m
πao 2 Atomic cross section, 8.7974 × 10−21 m2
NA Avogadro number (molecules/mole), 6.0220 × 1023
R Gas constant, 8.3144 J/(K mol)
AM U Atomic mass unit, 1.6606 × 10−27 kg
To Standard temperature, 298.15 K
Po Standard pressure, 1.0133 × 105 Pa
Acronyms
Btu British thermal unit (1 Btu = 0.29 watthour)

CCP Capacitively-coupled plasma
CF Calibration factor
col Collision
DC Direct current
EEDF Electron energy distribution function
ICP Inductively-coupled plasma
HHV Higher heating value
xviii
OECD Organization for Economic Co-operation and Development
ppm Part-per-million
RF Radio-frequency
xix
Chapter I
Introduction
1.1 Problem Statement
“Each day brings further evidence that the ways we use energy strengthen our adver-
saries and threaten our planet,” acknowledged President Barack Obama on the urgency and
relevancy of the energy and climate challenges in his 2008 inaugural speech. All fossil fuels
(natural gas, liquid petroleum, or solid coal) require hundreds of millions of years to form;
hence, they are considered non-renewable natural resources when taking into account the
relatively short time scale of the human race. Currently, the world relies on non-renewable
fossil fuels to meet 85% of its energy demand [2] and the rate of world fuel consumption is
expected to increase rapidly due to population and economic growth [3]. This summarizes
the energy challenge.
At the same time, burning of fossil fuels introduces carbon dioxide gas ( CO2 ) into the
atmosphere, raising the concentration of greenhouse gases. CO2 is an essential by-product
from combustion of fossil fuels, and its increasing concentration level has contributed to
abnormal climate changes [1]. Consequently, the use of fossil fuels challenges the sustain-
ability of our planet and the climate challenge is to reduce CO2 emissions.
1
The energy and climate challenges are therefore deeply intertwined and they must be
solved simultaneously. The increasing global energy demand and CO2 emissions have
motivated innovative techniques to satisfy demand while minimizing emissions. Many
experts believe that the solutions to address the energy and climate challenges require a
diversified portfolio of different methods, ranging from sequestration of CO2 emissions
to a variety of methods of alternative energy production. In the immediate future, fossil
fuels will remain the main source of energy because of their availability and their high
energy content. Consequently, additional emission of CO2 is inevitable and some types
of sequestration of CO2 or other methods of CO2 removal are necessary in addressing the
climate challenge.
Ultimately, to address both the energy and climate challenges, humanity must be com-
pletely independent of fossil fuels. To date, there is not a perfect formula to address these
intricate energy and climate issues. Nevertheless, the need to reduce consumption of fossil
fuels and a transition into an economy less dependent on hydrocarbon is urgently necessary.
On this quest to find alternative energy supplies, one solution is to use another secondary
energy source other than electricity–hydrogen. As an energy carrier, hydrogen has the po-
tential to address many aspects of the energy problem, particularly in the transportation
sector. Hydrogen can be used in fuel cells or in combustion reactions.
One main attractive feature of hydrogen as an energy carrier is its potential to satisfy
the need to fulfill some energy demand without further producing CO2 emissions. But
these energy and environmental benefits of hydrogen depend on how the gas is produced.
If hydrogen is produced via renewable methods, those which neither emit CO2 nor use cur-
2
rent non-renewable resources (e.g. water), the potential benefits of hydrogen are realized.
However, due to their high energy and conversion efficiencies, currently 95% of hydrogen
produced in the United States is through steam-methane reforming, a non-renewable pro-
cess [4]. Most of the other 5% of the hydrogen is produced primarily through conventional
electrolysis. One main drawback with electrolysis is its requirement of a catalyst, which is
often an expensive, rare resource such as platinum. Research is underway to find other and
cheaper catalyst options to reduce the total cost, but it still remains an expensive option.
It is also argued that the widespread adoption of conventional electrolysis systems has
a physical limitation. In these systems, hydrogen must diffuse in liquids, but their diffusion
rate in a liquid is slower than their diffusion rate in a gas. As an alternative, hydrogen
production via plasma electrolysis is investigated. However, many works in this area in-
vestigate the dissociation of liquid water [5–7], but the plasma processing of the liquid is
localized in the region within the gas bubbles around the electrodes.
This work investigates a method of hydrogen production through the use of a high-
density radio-frequency (RF) plasma source that operates in both capacitive and inductive
modes to dissociate water vapor (as opposed to liquid water). Both experimental and com-
putational works are carried out to assess the feasibility of this method. In addition to the
main motivation, there are other needs for hydrogen gas and water plasma in general. For
example, hydrogen is also needed in hydro-cracking, a process in which crude oil is mixed
with hydrogen to produce many of the end-products that we use today, including gasoline
[8]. Hydrogen is also used to produce chemicals such as ammonium and methanol [9] and
to make hydrogenated oil [10].
3
In addition, the presence of energetic hydroxyl and oxygen in a water plasma offers
many advantages in materials processing applications. Water plasma is being studied for a
number of applications, including diamond film growth, ultraviolet light sources, and even
medical applications [11–21]. A physical understanding of how a plasma source operates
on water vapor offers insights into many areas of plasma chemistry research; yet, much
of the plasma properties of a water vapor plasma have not been characterized. This work
characterizes the properties of the water plasma, including electron density, ion density,
electron temperature, floating potential, and plasma potential. The production of hydrogen
is also quantified in this work. From these results, the main reaction mechanisms, especially
dissociation mechanisms, in the water plasma source are identified. The experimental work
in this dissertation is accompanied by a global kinetic simulation of the plasma discharge
operating on water.
1.2 Aim of Project
The aim of this work is to evaluate the feasibility of hydrogen production through a
method of dissociating water vapor in a RF plasma source. To reach this aim, the following
steps are followed:
1. Generate a stable RF plasma discharge operating on water vapor. The main design
of the plasma source–geometry of discharge, antenna configuration, and matching
network design–follows closely with a similar setup from Professor Boswell’s group
at the Australian National University [22–24]. This RF plasma source was used
4
previously to simulate re-entry plasma conditions [25]. A water delivery system was
designed to meter the amount of water injected into the chamber.
2. Develop diagnostic tools to characterize the water vapor plasma. The following di-
agnostics were used in this work: residual gas analyzer (RGA), Langmuir probe, and
optical emission spectrometer. The RGA was used to identify the gas species in the
plasma and to quantify hydrogen production. However, RGAs are designed to op-
erate below 10−5 torr, much lower than the operating pressure of up to 500 mtorr in
this experiment. A design of a differential pump system was required to operate the
RGA. The Langmuir probe was used to characterize plasma properties. Finally, an
optical emission spectrometer was used to identify species in excited states.
3. Apply a global kinetic model to simulate reaction kinetics and to study how disso-
ciation of water molecules is affected by operating conditions including background
pressure, RF power level, and water input flow rate.
4. Finally, with both experimental and theoretical results, infer the energy and conver-
sion efficiencies of this method of hydrogen production, and provide directions for
future work.
1.3 Thesis Outline
To achieve the aim discussed in the previous section, Chapter 2 begins with a descrip-
tion of the intertwined global energy and climate challenges. In this chapter, hydrogen is
discussed as having a potential to satisfy some of the alternative energy needs. A survey of
5
current conventional hydrogen production methods is listed. This survey shows that while
the current methods of hydrogen production address some parts of the energy problems,
they do not address the climate challenge. This leads to an overview of plasma. Chapter 3
first gives an overview of plasma physical processes in the water vapor discharge, and then
lists the methods of plasma production and plasma sources. Finally, a literature review of
water plasma studies and in particular hydrogen production methods from a plasma source
is presented.
Chapter 4 shows the experimental setup and the diagnostics. The experimental setup
section describes the main chamber, the water delivery system, and the radio-frequency
plasma source. The diagnostics section describes the RGA, the Langmuir probe, and the
spectrometer. Chapter 5 presents the experimental results. First, characterization of the
plasma on argon is given and a comparison of plasma properties between an argon plasma
and a water plasma is made. Electron density, ion density, electron temperature, plasma
and floating potentials are characterized as functions of RF power, applied DC magnet
current, and water vapor input flow rate. Lastly, hydrogen production is examined. In
addition to experimental work, a global kinetic model, Global Kin, is used to determine
the theoretical energy efficiency as a function of water input mass flow rate and operating
pressure in the regimes that are not currently experimentally accessible. Chapter 6 gives
a description of this model and presents the computational results. Electron density and
electron temperature are characterized, and the effects of water flow rate and background
pressure on these two parameters are examined. Finally, conclusions on the feasibility of
the proposed method are drawn in Chapter 7.
6
Chapter II
Background–Energy
The previous chapter summarizes the energy and climate challenges and motivates al-
ternative energy research to address these problems. The aim of this chapter is to substanti-
ate the energy and climate challenges by providing supporting evidence. In Section 2.1, a
history and predictions of energy consumption and CO2 emissions are presented. This sec-
tion also explains the role of CO2 in causing climate change. Section 2.2 outlines benefits
and drawbacks of the proposed use of hydrogen to replace fossil fuels in the transportation
sector. Section 2.3 lists current conventional hydrogen production methods.
2.1 Energy and Climate Challenges
2.1.1 Energy Challenge
The main cause of the energy problem is the world’s dependency on fossil fuels as the
primary energy source. All fossil fuels used today are formed from plants and animals that
lived up to hundreds of millions of years ago, during the same time period when dinosaurs
lived. The time span required to make fossil fuels therefore exceeds the relatively short
time scale of the human race. As a result, fossil fuels are considered non-renewable natural
7
Figure 2.1: World energy consumption, Figure 2.2: Total non-OECD energy con-
1980-2030 [2]. sumption, 1980-2030 [2].
resources. On the other hand, economic growth and population increase necessitate an even
higher energy demand in the future. The energy challenge is to address the future energy
demand with limited supplies of fossil fuels.
Despite a limit in fossil fuel supplies, its demand is expected to grow. Even with a
decrease in population growth rate in terms of percentages from 2% in the early 1960’s to
1.2% in 2007, the world’s population is estimated to reach over 9 billion in 2040 compared
to 6.7 billion today [3]. Currently, 85% of the world’s energy consumption comes from
fossil fuels, and this dependency is expected to continue in the next several decades [2].
Figure 2.1 shows the total energy consumption from 1985 to 2005 and predictions up to
2030 in quadrillion (1015 ) Btu per year (1 Btu = 1054 J) [2]. In 1980, the energy con-
sumption was below 300 quadrillion Btu, but it is expected to reach 700 quadrillion Btu
in 2030. These data substantiate the energy problem–the world continues to depend on
non-renewable fossil fuel supplies.
8
In addition to population growth, economic growth also contributes to the increasing
energy demand. In Figure 2.1, the histogram is divided into two groups: countries in the
Organization for Economic Co-operation and Development (OECD) and those outside of
OECD. The rate of energy consumption increase for non-OECD countries is much higher
than that of OECD countries for the past several decades and this trend is expected to
continue in the next several decades. Among those countries in the non-OECD group,
India and China’s predicted energy needs account for most of the energy demand increase.
As shown in Figure 2.2, there is almost an indiscernible change in energy increase for
non-OECD countries in the Middle East, Africa, Central America, South America, Europe
and Eurasia. However, non-OECD Asian countries are predicted to double their energy
consumption between now and the year 2030 [2]. As countries such as China are building
more nuclear power plants, the future dependency on fossil fuels may reduce.
These predictions of significant increases in energy demand as shown in Figure 2.1
and Figure 2.2 will likely occur unless countries around the world drastically alter their
energy consumption habits. However, primary energy sources still remain finite. Figures
2.3– 2.5 show proven reserves of oil, natural gas, and coal as of 2005 [26]. Even though
finding more fossil fuel reserves is a possibility, the fact remains that fossil fuels are limited
to finite quantities. This calls for urgent actions to reduce global energy consumption, to
become less dependent on fossil fuels, and to find alternative sources of energy. These
figures of proven reserves further illustrate an uneven distribution of current global energy
supplies. The Middle East has most of the oil reserves and a large part of the natural gas
reserves. Europe has a large share of the natural gas and coal. North America and Asia
9
Figure 2.3: Distribution of proven reserves of oil at the end of 2005 in billion barrels [26].
Pacific has a large share of coal. For those countries that have direct access to reserves of
fossil fuels from their own land mass or within their water boundary, finding alternative
energy sources will ensure future energy security. For countries without a direct access to
reserves of fossil fuels and therefore have to import them, finding alternative sources of
energy has immediate economic and political benefits.
2.1.2 Climate Challenge
In addition to the energy challenge, the world also currently faces an equivalently grave
problem–the climate challenge. There are still a few skeptics who attribute the recent ob-
served global warming to the climate’s natural cycle, but ample evidence has linked climate
change to increased anthropogenic (man-made) CO2 emissions. To understand this link, the
greenhouse gas effect is examined.
10
Figure 2.4: Distribution of proven reserves of natural gas at the end of 2005 in trillion cubic
meters [26].
Figure 2.5: Distribution of proven reserves of coal at the end of 2005 in billion tons [26].
11
Earth can harbor human life because its temperature is between freezing and boiling
point for water (its mean surface temperature is 287 K) [27], unlike Mars (186-268 K)
[28] or Venus (729 K) [29]. The properties and the composition of Earth’s atmosphere are
different from its neighboring planets that allow human life to sustain. Earth’s atmosphere
uniquely acts as a blanket to retain some of the solar radiation through a natural greenhouse
gas effect.
Figure 2.6 shows Kiehl and Trenberth’s estimate of the energy balance [30]. Approxi-
mately 31% of the incoming solar radiation–mostly long wavelength in the infrared range–
is reflected by clouds, aerosol, atmospheric gases, and the surface. The atmosphere absorbs
19% of the radiation and Earth’s surface absorbs the rest of the 49% of the solar radiation,
mostly short wavelength in the visible spectrum. Earth maintains an energy equilibrium
by radiating infrared radiation. Without greenhouse gases, this radiation would be lost to
space and Earth’s temperature would be ∼ 30 K colder [31]. Therefore, greenhouse gases
have a vital role in nature to maintain a comfortable range of temperature on this planet,
making it a habitable place for human beings.
Since the industrial revolution, an increased use of fossil fuels has resulted in a sig-
nificant increase in CO2 emissions in the atmosphere. In fact, CO2 makes up the largest
portion of anthropogenic greenhouse gas. The sharp increase in the world’s total CO2
emissions from fossil fuels is shown in Figure 2.7. Even though greenhouse gases exist in
nature and are needed to maintain a comfortable temperature range on Earth, an increased
concentration creates an energy imbalance and causes climate change.
In addition to CO2 , other greenhouse gases include H2 O, O3 , CH4 , NO, and a trace
12
Figure 2.6: Annual global energy balance [30].
amount of other gases [1]. In any combustion process of fossil fuels (hydrocarbon), the
following reaction takes place [33]:
m m
Cn Hm + (n + )O2 → nCO2 + H2 O + heat (2.1)
4 2
The burning of fossil fuels therefore necessarily emits CO2 , which explains why CO2
is the largest anthropogenic greenhouse gas emissions. Similar to other greenhouse gases,
CO2 has a complex structure, which allows for excitation of asymmetric vibration modes.
CO2 is linear and it has four fundamental vibrations. Figure 2.8 shows three of the four
modes of vibration: symmetric stretching, asymmetric stretching, and one bending. The
fourth mode is another bending mode that acts exactly like case (c) but in the opposite
direction. In symmetrical stretching of CO2 (case (a)), the center of mass and center of
charge do not change. Therefore, energy is conserved in symmetrical stretching and the
band in the infrared spectra cannot be absorbed [34]. On the other hand, in asymmetrical
13
Figure 2.7: United States and total world’s CO2 emissions from fossil fuels [32].
stretching and bending vibrations, case (b) and case (c), the center of mass and the center
of charge change, which creates a dipole moment. This dipole moment can be created by
the absorption of infrared radiation by CO2 . The energy and momentum of the absorbed
infrared photons are transferred to CO2 , resulting in asymmetric stretching and bending.
As the Earth’s surface radiates long wavelength radiation back to space, CO2 and other
greenhouse gases absorb the infrared radiation and become vibrationally excited. When
they relax, the total energy absorbed is released as infrared radiation in all directions, some
is lost to space and some is directed back to Earth, thus warming it.
Evidence of climate change is widely observed. Figure 2.9 shows average global tem-
perature normalized to the average temperature from 1961 to 1990. The figure illustrates
two sharp temperature rise periods, one in the early 1900s and the most recent rise in the
14
Figure 2.8: Vibration modes of CO2 : (a) symmetric stretching, (b) asymmetric stretching,
(c) bending.
last several decades. The world has experienced drastic effects due to the climate change.
According to the International Panel on Climate Change [1], CO2 concentration had risen
from 280 ppm (part-per-million) in the pre-industrial period to 379 ppm in 2005. This value
is the highest concentration in 650,000 years. This panel blames the rise in atmospheric
CO2 concentration for the increase in global average air and ocean temperature, melting of
snow and ice, and the rise in the global average sea level, as shown in Figure 2.10.
In conclusion, the world’s dependence on fossil fuels as the primary energy source
presents a problem that is twofold. First, fossil fuels are available in finite supplies, and
they cannot address increasing future energy demand. Second, burning of fossil fuels has an
indirect consequence of causing climate change through the emission of CO2 . Therefore, to
solve both the energy and climate challenges, the world must alter our energy consumption
habits and must get rid of our dependency on fossil fuels.
15
Figure 2.9: Temperature anomaly relative to the mean temperature between 1961 and 1990
[1].
2.1.3 Case Study: United States
In the previous sections, the energy and climate challenges are presented from a global
perspective. These challenges have different levels of impact on individual countries. This
section examines the impact on one specific country–the United States (U.S.). In the case
of the U.S., finding solutions to the energy and climate challenges is also important for the
country’s foreign relations and thereby improving its national security.
Similar to other nations, the U.S. heavily relies on fossil fuels. Figure 2.11 shows the
distribution of energy consumption in the U.S. in 2007: 84% fossil fuels (petroleum, nat-
ural gas, and coal), 8% nuclear electric power, and only 7% renewable energy [35]. Note
that 34% of the renewable energy comes from hydroelectric power. Alarmingly, the U.S.
consumes 25% of the world’s energy. Figure 2.7 shows the U.S. emission of CO2 com-
16
Figure 2.10: Rise in global average temperature and sea level and decrease in area of snow
cover in Northern Hemisphere [1].
17
Figure 2.11: Distribution of energy consumption in the US in 2007 [37].
pared with the world’s emission. In 2005, the United States emitted 25% of the world’s
CO2 emission, 87% of which was related to energy consumption [36]. Among emissions
from fossil fuels consumption in the U.S., 44% of emission comes from petroleum, 20%
from natural gas, and 36% from coal [36]. For a country with only 4.5% of the world’s
population, the U.S. is a major consumer of non-renewable energy and contributor of an-
thropogenic CO2 emissions.
The need to reduce fossil fuel consumption, and therefore CO2 emission, to address the
energy and climate challenges is clear. In addition, the health of the American economy
depends on its access to energy. From production of domestic goods to all technological
innovations, the U.S. economy cannot thrive if the country does not have access to energy,
and this is true for all nations. Figure 2.12 shows current maps of the world based on
land distribution along with those based on population distribution, fuel import, and fuel
export. As shown, the size of the U.S. based on land distribution is similar to that based on
population distribution. However, the size of the US based on fuel import is significantly
18
Figure 2.12: World maps reflecting (a) land distribution, (b) energy export, (c) population,
and (d) energy import [38].
larger than that of fuel export.
In the last two decades, the consumption of petroleum (processed from crude oil) in the
U.S. has increased while its production has decreased. This imbalance in fuel import and
export requires a continuous increase in petroleum import to satisfy the country’s energy
demand. As an example, according to the Department of Energy’s Annual Energy Review
in 2008 [39], the U.S. produced 4.96 million barrels (0.789 million m3 ) of crude oil per
day, but it imported 9.76 million barrels (1.55 million m3 ) per day mostly from Canada,
Saudi Arabia, Mexico, Venezuela, Nigeria, Iraq, and Russia. With regard to natural gas,
the U.S. withdrew 26.05 trillion ft3 (0.737 trillion m3 ) while importing 3.96 trillion ft3
(0.112 trillion m3 ) and 90% of natural gas imports were from Canada. Coal is the only type
of fossil fuel where the U.S. has a surplus. In 2008. it produced 1062 million metric tons,
imported 31.0 million metric tons, and exported 73.9 million metric tons to mostly Canada,
19
Netherlands, Brazil, United Kingdom, France, Italy, and Belgium. From these figures, it is
clear that the U.S. strongly depends on import of foreign oil.
As illustrated in Figure 2.3, most of the oil reserves are located in the least politically
stable part of the world; the Middle East has 67% of the world’s oil reserve. The ten
countries that have the most oil in reserves as of January 2006 in descending order are:
Saudi Arabia, Canada, Iran, Iraq, Kuwait, United Arab Emirates, Venezuela, Russia, Libya,
and Nigeria [40]. According to a report by the Council on Foreign Relations [41], there
are national security consequences of U.S. oil dependency. First, countries with a control
over large oil revenues have the flexibility to adopt policies that oppose U.S. interests and
values; Iran’s program for nuclear weapons is an example. Second, oil dependency has
forced current allies of the U.S. to realign their political partnerships with other oil-rich
countries, and these realignments affect the U.S. in forming its partnerships. For example,
many countries within the European Union are forming relationships with Russia and Iran
because of their dependency on imported oil and gas, and these new realignments reduce
the political influence that the U.S. can have in those regions. Third, due to high oil prices,
some countries invest in oil-rich regions, where new relationships may pose a problem for
the U.S. One example is China’s investment in the Middle East and Africa. Fourth, the
revenues from oil in some countries can undermine local governance, favoring totalitarian
governments, such as those in Sudan. And finally, because the U.S. is dependent on foreign
supplies of oil, any significant interruption in oil supply can cause adverse political and
economic consequences in the U.S. For example, this had happened recently to several
European countries during a dispute between Russia and Ukraine on oil prices, and as a
20
result, Ukraine did not allow natural gas to flow from Russia into Europe.
For these reasons, reducing dependency on foreign sources of non-renewable energy
is important for U.S. foreign relations. But the U.S. foreign relations are also affected by
other countries’ dependency, and therefore, the U.S. must also encourage others to reduce
consumption of non-renewable natural resources. The U.S. should lead by example, but it
is currently the highest consumer of non-renewable energy and contributor of emissions. In
summary, finding alternative sources of energy to reduce dependency on limited supplies
of fossil fuels and to reduce emissions is a paramount task that the world faces. To maintain
continued global economic growth and to sustain this planet, the world has to find solutions
to the energy and climate challenges.
2.2 Hydrogen: Benefits and Drawbacks
To address the energy and climate challenges, several alternative energy sources are
investigated. One proposed solution is to use a secondary energy source–hydrogen–as an
energy carrier to replace some of the fossil fuel consumption, especially in the transporta-
tion sector.
Hydrogen is the most abundant element in the universe, but it is not available in its
natural form on Earth. Hydrogen must be produced from a feedstock that contains its
molecules, such as natural gas, petroleum, and water. Therefore, hydrogen is considered a
secondary energy source, similar to the case of electricity. There are a number of hydrogen
production methods. As discussed previously and shown in Figures 2.3 - 2.5, oil, natural
gas, and coal are not available as natural resources in all regions. Further, in many cases,
21
only a few countries have ownership of much of the world’s non-renewable resources.
On the other hand, hydrogen can be more widely accessible because hydrogen can be
produced anywhere water and power are available. With the ability and flexibility to be
a decentralized source of energy, hydrogen is an attractive option.
When hydrogen is used with oxygen in fuel cells, the only by-product is water. That is,
the use of hydrogen as an energy carrier in the transportation sector is completely emission-
free. Recently, the Energy Information Administration (EIA) estimates that petroleum con-
sumption in the transportation sector can be reduced between 37.1 and 84.1% of the cur-
rent consumption rate and CO2 emission can be decreased between 8.8 and 83.8% from the
current emission rates when hydrogen-powered fuel cell vehicles enter the market in 2050
[42]. The actual petroleum and CO2 emission reduction will depend on the actual level of
fuel cell vehicle penetration rate and fuel economy. Nevertheless, hydrogen-powered fuel
cell vehicles are a good example of the potential benefit of hydrogen to replace fossil fuels.
Hydrogen shows a great promise to address the energy and climate challenges; still,
there are some drawbacks. Mainly, hydrogen is not a primary energy source. As such, it
is only as clean of an energy carrier as its method of production. To produce hydrogen by
a clean method has many technical challenges, which explains why 96% of hydrogen is
still produced from non-renewable resources such as methane and coal worldwide [43]. In
these processes, CO2 emission presents a problem as well. Further, hydrogen production
from non-renewable resources does not solve the original problem, which is to reduce our
dependency on fossil fuels. Most of the remaining 4% of hydrogen is produced from con-
ventional electrolysis [43]. In electrolysis, the energy efficiency is still lower compared to
22
other non-renewable methods [44]. The maximum energy efficiency of the steam methane
reformation method is 70-85% and for electrolysis is 40-70% [45].
Beyond the technical challenges in hydrogen production, the costs of transportation
and storage are two major drawbacks. Hydrogen is the lightest gas, with a density ap-
proximately 7% that of air [46]. In addition, the condensation temperature of hydrogen
at atmospheric pressure is 20 K [46]. These two characteristics pose some difficulties in
transportation and storage of hydrogen. These are the current challenges that need to be
addressed before hydrogen can be applied as intended. The following section outlines the
main methods of hydrogen production.
2.3 Hydrogen Production Methods
Steam methane reformation (SMR), partial oxidation of heavy hydrocarbons, and coal
gasification are current methods employed to produce approximately 96% of hydrogen
worldwide and most of the other 4% is produced via conventional electrolysis [43]. Hy-
drogen when used with oxygen addresses the energy and climate challenges by reducing
the consumption of fossil fuels and by reducing CO2 emission. However, the first three
methods listed below do not fully address these challenges. Non-renewable resources are
used to produce hydrogen, and in many cases, CO2 emission is a by-product in these pro-
cesses.
23
2.3.1 Natural Gas Reforming
Methane gas makes up 95% of natural gas. Globally, 80% of hydrogen is produced
from SMR, and this process accounts for 95% of the hydrogen produced in the U.S.
Steam methane reformation:
Cn Hm + nH 2 O(+heat) =⇒ nCO + (n + 0.5m)H2 (2.2)
In this endothermic process, heat is added to initiate the reaction. Methane is subjected
to a high-temperature steam (700-1000 C) under a pressure between 300 kPa to 2500 kPa
to produce a syngas, which contains a varying amount of CO and H2 [4]. In the SMR
process, a catalyst (typically nickel because of its low cost) is introduced to the system
[47]. The CO product from the SMR process is often used to further produce hydrogen in
what is called the water gas shift (WGS) reaction:
H2 O + CO =⇒ H2 + CO2 (+heat) (2.3)
In WGS reaction, H2 O and CO with the help of a catalyst react to produce H2 and CO2
and a small amount of heat. SMR is currently the most cost-effective and energy-efficient
method of hydrogen production, with an energy efficiency up to 85% [45, 48]. In addition,
the natural gas pipeline already has an established infrastructure, and this makes SMR the
method of choice for hydrogen production. However, the U.S. currently imports 15% of
its natural gas, and therefore, this method of hydrogen production does not address the
24
desire of the U.S. to reduce gas imports from foreign sources. CO2 emission is also another
concern.
2.3.2 Partial Oxidation of Methane
Similar to the SMR method, partial oxidation of methane is also another hydrogen
production method where the product mixture contains a syngas, which undergoes a WGS
reaction to further produce H2 from CO.
Partial Oxidation of Methane:
Cn Hm + nO2 =⇒ nCO + 0.5mH2 + heat (2.4)
This is an exothermic process, but its energy efficiency (70-80%) is less than what can
be achieved with SMR. Therefore, this method is not pursued commercially.
2.3.3 Gasification
In the process of gasification, solid coal is first gasified and then it reacts with water
to produce a syngas, similar to the SMR method. Hydrogen is extracted from this syngas.
This method of hydrogen production is more costly than the SMR method, and has a lower
energy efficiency (63%) [48]. Despite the higher cost and lower energy efficiency, this
method of hydrogen production is under consideration by the U.S. Department of Energy
because the U.S. currently has a supply of coal that will last for more than 250 years, the
only type of fossil fuel for which the U.S. has a surplus. The reactions for this method are:
25
2C + O2 =⇒ 2CO (2.5a)
C + H2 O =⇒ CO + H2 (2.5b)
C + CO2 =⇒ 2CO (2.5c)
2.3.4 Conventional Electrolysis
Methods of hydrogen production previously discussed dominate the world market or are
under consideration, but they can only serve as a transition stage for hydrogen production.
In order to realize the full benefits of hydrogen as a “clean” energy carrier, the process in
which it is produced must have minimal CO2 emissions and must not rely on non-renewable
natural resources. The method of electrolysis is one such example–it uses an electric current
to split water into hydrogen and oxygen in what is known as an electrolyzer [49]. This
method is considered renewable if the electricity comes from a clean source. However,
currently 43% of electricity in the United States is generated from coal [39]. That, however,
does not limit the usage of renewable electricity generated from wind, solar, biomass, etc.
When electricity is obtained from renewable or nuclear sources, then the energy cycle is
sustainable. The total reaction for electrolysis is as follows:
2H2 O + electricity =⇒ O2 + 2H2 (2.6)
There are several types of electrolyzer: alkaline electrolyzer, polymer electrolyte mem-
26
Figure 2.13: Schematic diagram of an electrolysis system, courtesy of Sandia National
Laboratory.
brane (PEM), and solid oxide electrolyzer. The alkaline electrolyzer is widely available
commercially. A diagram of the alkaline electrolyzer is shown in Figure 2.13.
In an alkaline electrolyzer, the electrolyte is an alkaline solution (sodium or potassium
hydroxide). The reactions at the anode and cathode are:
At the anode:
1
2OH − =⇒ O2 + H2 O + 2e− (2.7)
2
At the cathode:
2H2 O + 2e− =⇒ H2 + 2OH − (2.8)
The PEM electrolyzer is similar to the alkaline electrolyzer, except that the electrolyte
is a solid specialty plastic material [4]. A schematic diagram of a PEM electrolyzer is
shown in Figure 2.14
At the anode, the water reacts with the material of the anode to form oxygen and posi-
tively charged hydrogen ions. When an external source is connected to the electrolyzer, the
electrons flow through the external circuit and hydrogen ions move to the cathode through
27
Figure 2.14: Schematic diagram of the PEM electrolyte, courtesy of F. Barbir [50].
the PEM. Hydrogen is created when hydrogen ions and electrons combine. The reactions
are summarized as:
At the anode:
1
H2 O =⇒ O2 + 2H + + 2e− (2.9)
2
At the cathode:
2H + + 2e− =⇒ H2 (2.10)
Alkaline and PEM electrolytes function very similarly, and they operate in relatively
low temperature ranges: 100-150 C for alkaline electrolyzers and 80-100 C for PEM elec-
trolyzers [4]. Solid oxide electrolyzers are different from these two types, mainly because
they operate at a much higher temperature range, between 500 and 800 C. The electrolyte
for these systems is a solid ceramic material.
In addition to the need of an electrolyte, catalysts are also required. The most commonly
used catalysts are platinum or platinum alloys–a rare and expensive metal [50]. This is one
28
disadvantage of the electrolysis method of hydrogen production. The electrolyzer energy
efficiency ranges from 40-70% [45], but the overall efficiency is from 13 to 24% assuming
that the efficiency of electricity generation is 33%. As a comparison, SMR method has an
energy efficiency up to 85%. Currently, hydrogen can be produced through the method of
electrolysis in small amounts of 0.45 kg/hr to as much as 434 kg/hr [44].
2.4 Conclusion
Hydrogen has an enormous potential to address the challenges that we face today. Be-
cause of its low density, hydrogen has an energy density of 120 MJ/kg compared to 42.7
MJ/kg and 41.9 MJ/kg for gasoline and diesel. Even when produced from SMR, gasifica-
tion, or oxidation, hydrogen used in fuel-cell powered vehicles is a good transition stage as
explained in a 2002 technical report on guidance for transportation technologies.
“Fuel cell vehicles running on hydrogen produced from natural gas would use
less energy and emit less carbon than a conventional internal combustion en-
gine, even when considering the full fuel cycle from energy source to produc-
tion to end-use.” [51]
Nevertheless, there is still a need to reduce our dependency on non-renewable natural
resources, and therefore non-renewable methods of hydrogen production are not accept-
able in the future. Thus, the renewable electrolysis method is possibly the best candidate
but the cost of hydrogen production through electrolysis is high and this process also re-
quires a rare resource, primarily platinum metal. In addition to the methods listed above,
29
a number of other approaches are under investigation. For example, hydrogen is produced
from water using sunlight and specialized microorganisms in a photobiological water split-
ting process, and hydrogen is also produced from water using sunlight in conjunction with
photoelectrochemical materials in what is known as photoelectrochemical water splitting
[4]. The list of hydrogen production methods under investigation further includes direct
biomass gasification, sulfur iodine cycle, and many other derivatives of the methods afore-
mentioned [48]. Among those novel approaches is plasma electrolysis, the use of a plasma
source to directly dissociate water molecules into hydrogen and oxygen. This is the topic
of this dissertation and the subject of plasma is introduced in the next chapter along with a
literature review on current water plasma research.
30
Chapter III
Background–Plasma Overview
A review of existing hydrogen production methods in Chapter II reveals that current
methods employed to produce more than 96% of hydrogen do not directly address the en-
ergy and climate challenges. Hydrogen production from electrolysis with the use of “clean”
electricity is the primary renewable hydrogen production method. Conventional electroly-
sis, however, has some drawbacks, and plasma electrolysis–the subject of this dissertation–
is argued to have the potential to address some of these drawbacks. First, Section 3.1
provides an overview of plasma physical processes in a water vapor discharge. Then Sec-
tion 3.2 explains the methods of plasma generation. Section 3.3 lists some examples of
plasma sources, and finally Section 3.4 provides a literature review of the water plasma
studies.
3.1 Overview of Plasma Physical Processes in Water Vapor Discharge
3.1.1 Collisional Phenomena in Plasmas
Before a discussion on how a plasma is generated, the structures of an atom and a
molecule need to be examined. A thorough treatment is given in References 52 and 53.
31
Only a brief description is given in this section to serve as background information for the
discussions in the upcoming sections.
For atoms, collisions with electrons can result in elastic scattering in which the total
kinetic energy is conserved but the momentum of the electron is changed, or in inelastic
processes including electronic excitation and ionization in which the total kinetic energy is
not conserved. A collision between an atom and an ion can result in an elastic scattering,
where the momentum and energy of the two species are exchanged [52]. However, the
structure of a molecule is more complex due to a higher number of degrees of freedom
it possesses. Unlike an atom, a molecule has vibrational and rotational motions. Within
each electronic states, there are discrete vibration and rotation energy levels. The Born-
Oppenheimer Approximation assumes that the electronic motion and the nuclear motion in
molecules can be separated. The nuclear motions (vibration and rotation) are slow com-
pared to the electronic motions [52], and therefore, we can draw potential energy curves
for the electronic states of a diatomic molecule as shown in Figure 3.1. The diagram shows
the potential energy versus the internuclear distance, each representing quantized electronic
level for a frozen set of nuclear positions [52].
In the ground electronic state, the minimum potential energy must have a value that
represents the intermolecular distance R. For higher electronic states, the minimum poten-
tial energy can shift [52]. This observation is important in future discussions on different
dissociation mechanisms.
32
Figure 3.1: Potential curve for a molecule AB [52].
3.1.2 Mechanisms of Electron and Ion Creation
Generating and sustaining a plasma require production of electrons and ions, which is
accomplished through ionization processes [52–54]. Ionization is a result of an inelastic
collision. If most of the products of such collisions are positive ion and electron pairs, then
the plasma is an electropositive plasma. All noble gases, such as argon and xenon, produce
electropositive plasmas. However, a neutral gas particle can also become a negative ion,
instead of a positive ion, if the gas has a high electron affinity. Electron affinity is the
difference in energy between a neutral atom and its negative ion; it is the energy released
when an electron combines with a neutral to become a negative ion [55, 56]. Note that the
electron affinity is positive if the negative ion has a lower energy than the neutral atom (e.g.
O2 , Cl2 , and F), and is negative if the negative ion has a higher energy than the neutral atom
(e.g. noble gases such as Ar and Xe) [55]. A plasma is then an electronegative plasma if
33
the majority of the ions are negative ions. In this section, three main ionization processes
are discussed: direct ionization by electron impact, stepwise ionization by electron impact,
and ionization by collision with heavy particles [54]. All of these processes result in a
production of positive ion and electron pairs.
Direct Ionization by Electron Impact
Direct ionization by electron impact is among the most common ionization processes in
many plasma sources–including the water plasma source investigated in this work. In the
direct ionization by electron impact for an atom, a positive ion and electron pair is created.
Ionization occurs if the electron’s energy exceeds the ionization energy of the state of the
atom–the energy required to remove an outermost electron from an atom or molecule. The
reaction equation for this process is shown in Equation 3.1(a). For the molecules, the direct
ionization by electron impact can result in two outcomes. First, non-dissociative ionization
produces a molecular ion and electron pair as shown in Equation 3.1(b), similar to the
case of an atom. However, for molecules, there is also another possibility; i.e. dissociative
ionization, as shown in Equation 3.1(c).
e + A → A+ + e + e (3.1a)
e + AB → AB + + e + e (3.1b)
e + AB → A + B + + e + e (3.1c)
34
Figure 3.2: Potential curve for a molecule AB, showing dissociative ionization of molecule
AB [52, 53].
The Franck-Condon Principle is introduced to explain the dissociative ionization pro-
cess. Inside a molecule, the fastest internal motion of atoms is molecular vibration, but the
time scale for molecular vibrations is still much longer than the time scale for the interac-
tion between the electrons and the molecules [53]. Therefore, the atoms inside a molecule
are considered being frozen when a molecule, stimulated by electron impact, becomes
electronically excited [52, 53], and this is known as the Franck-Condon Principle. This
principle allows for the potential energy curves representing the potential energy versus
internuclear distance of the molecule to be drawn. Figure 3.2 shows the potential energy
curves for molecule AB. Vertical line a shows the case for non-dissociative ionization, in
which the molecule is excited into a high quantum number bound vibration state. Vertical
line b shows the case of dissociative ionization, where the dissociation of the molecule is
energetically permitted because the molecule is excited to a vibrationally unbound state
[52, 53].
35
Table 3.1: Ionization potential energy [52, 54]
Atom or Molecule Energy (eV)
Ar 15.8
H 13.6
O 13.6
OH 13.2
O2 12.2
H2 O 12.6
Table 3.1 gives a list of ionization energies for atoms and molecules in a water plasma
and for argon gas. As noted, ionizing noble gases such as argon is harder than ionizing gas
species present in a water plasma. References 54, 57–59 and 52 provide detailed physical
insights on this topic.
Stepwise Ionization by Electron Impact
Stepwise ionization by electron impact is similar to direct ionization, but this process
occurs particularly when the plasma density is high. In a high density plasma, the col-
lisional frequency between electrons and neutrals is increased, but the energy transferred
from the electrons to the neutrals is not necessarily high enough to cause ionization. In
many cases, the neutral particles become excited neutrals through an inelastic collision
with electrons; however, this means the density of excited neutral is also high. Therefore,
ionization is possible when an electron of sufficient energy collides with an excited neutral.
Effectively, the neutrals are ionized in two steps: excitation and then ionization. The net
ionization energy in this process is equal to that in the direct ionization process, but step-
wise ionization rate is higher than direct ionization [54] because collisional frequency is
higher in dense plasma.
36
Ionization by Collision with Heavy Particles
In general, heavy particles by nature are massive and usually do not possess enough
energy to create ionization. However, ionization is possible if the heavy particles are in
excited states. When a metastable atom A∗ collides with another atom B, if the excitation
energy of A∗ exceeds ionization potential of B, ionization occurs. This is known as Penning
ionization [54].
Penning ionization:
A∗ + B → A + B + + e (3.2)
In addition, a collision between a pair of heavy particles can also result in a charge-
exchange process. When an energetic ion hits a cold neutral, the products of this collision
can be a cold ion and an energetic neutral [60]. Take argon as an example, the charge-
exchange process is:
Ar+ (f ast) + Ar(slow) → Ar(f ast) + Ar+ (slow) (3.3)
The list of ionization processes discussed above is not complete, but these are the main
processes in most plasmas.
3.1.3 Mechanisms of Electron and Ion Destruction
The previous section outlines important ionization processes in which electrons and
positive ions are created. In this section, the main mechanisms of electron and ion losses are
37
examined. The main loss mechanisms for positive ions and electrons are through recombi-
nation processes [52, 54]. But electrons can also be removed through several attachment
processes in a plasma when species with high electron affinity are present [52, 54]. The
processes to be discussed are: electron-ion recombination, electron attachment, detachment
processes, ion-ion recombination, and surface recombination .
Electron-Ion Recombination
In electron-ion recombination, an electron and positive ion pair recombines to produce
a neutral atom for the case of an atomic ion. For the case of a molecular ion, the recombi-
nation can result in either a production of a neutral molecule or dissociation into multiple
neutral atoms. For both cases, either electron-atomic ion or electron-molecular ion recom-
bination, the process is exothermic.
For atomic ions, the first electron-ion neutralization process is a three-body electron-ion
recombination, which is important in a high-density plasma:
Three-body electron-ion recombination:
e + e + A+ → A∗ + e (3.4)
Here one electron acts as a third-body partner and the atomic positive ion becomes
an excited neutral after the recombination. Any excess energy from this recombination
goes into the kinetic energy of the free electron [53]. This three-body electron-ion re-
combination process dominates in most density regimes. But in the case of a relatively
38
low-density plasma, another process can compete with this three-body electron-ion recom-
bination process, and that is the radiative electron-ion recombination. Here, the excess
energy is converted into radiation [53].
Radiative electron-ion recombination:
e + A+ → A∗ → A + h̄ω (3.5)
The above two processes are described for a recombination between an electron and
an atomic ion. For an electron-molecular ion pair, the most common recombination is
dissociative electron-ion recombination.
Dissociative electron-ion recombination:
e + AB + → (AB)∗ → A + B ∗ (3.6)
This process can be best described using the potential curves shown in Figure 3.3.
When an electron and molecular ion (AB+ ) recombine, the electron is captured, and there-
fore the electron does not have a means of carrying away a part of the reaction energy [52].
The molecular ion AB+ after the recombination with an electron has acquired an additional
energy equal to ²d or ²d0 as shown in Figure 3.3, which results in a repulsive state of A and
B∗ . Because the potential curve for AB+ is always above the repulsive state of AB, the
electron-molecular ion recombination in this case can never result in ground states of A
and B [52].
39
Figure 3.3: Potential curve for a molecule AB, showing dissociative recombination of
molecule AB [52].
Note that for an electron-molecular ion pair, radiative electron-ion recombination is also
possible. However, the rate constant for radiative recombination for an electron-molecular
ion pair is between three and five orders of magnitude lower than that for dissociative
recombination [52].
Electron Attachment
Through a collision between an electron and a neutral, one possible outcome is ioniza-
tion as discussed in Section 3.1.2. But when the neutral (or one atom within the molecule)
has a positive electron affinity, which can be thought of as an ability to form stable negative
ions, then the outcome of such collision is electron attachment–not ionization. Electron
attachment processes include dissociative electron attachment and three-body electron at-
tachment to molecules. Similar to the case of dissociative electron-ion recombination, in
40
Table 3.2: Electron affinity [54]
Atom or Molecule Energy (eV)
H− = H + e 0.75
O3− = O3 + e 2.0
OH − = OH + e 1.8
O− = O + e 1.5
O2− = O2 + e 0.44
HO2− = HO2 + e 3.0
dissociative electron attachment the molecule AB acquires energy when it recombines with
an electron. In the case of dissociative recombination, the products are A and B∗ , but in the
case of dissociative electron attachment, the electron is attached to one of the atoms (A or
B) that has a positive electron affinity.
Dissociative electron attachment:
∗
e + AB → (AB − ) → A + B − (3.7)
This electron attachment can only occur if the atom has a positive electron affinity. For
example, electron dissociative attachment is possible for O2 or H2 O because both H and O
have positive electron affinities. But electron dissociative attachment is not possible for N2
because the electron affinity for N is negative (-0.07 eV) [55]. Note again that the electron
affinity is positive if the negative ion has a lower energy and is negative if the negative ion
has a higher energy than the neutral atom.
Dissociative electron attachment is an important process in water plasmas because its
molecular fragments have positive electron affinities, shown in Table 3.2. Table 3.3 shows
the relevant dissociative electron attachment processes in a water plasma.
41
Table 3.3: Dissociative electron attachment [54].
e + O2 → O − + O
e + H2 → H − + H
e + H2 O → H − + OH
e + H2 O → O− + H2
e + H2 O → H + OH −
Table 3.4: Three-body electron attachment [54].

e + O2 + O2 → O2− + O2
e + O2 + H2 O → O2− + H2 O
e + O2 + H2 → O2− + H2
In addition to dissociative electron attachment, three-body electron attachment is an-
other process in which negative ions are formed. This three-body reaction is only dominant
at high pressures (p > 0.1 atm) [54] and similar to the process discussed above, one of the
particles must have a high positive electron affinity.
Three-body electron attachment to molecules:
e + A + B → A− + B (3.8)
Here A and B can be a molecule, but the energy of the electron is not high enough to
cause dissociative electron attachment as in the previous case. An excellent example of this
process takes place in an oxygen plasma at high pressure (atmospheric): e + O2 + M →
O2 − +M . The third body (M) is needed in this process to carry away the excess energy from
the colliding two-body system (e + O2 ) in order to satisfy the conservation of momentum
and energy [61]. Table 3.4 shows examples of electron attachment to oxygen in three-body
electron attachment processes.
42
Figure 3.4: Potential curve for a molecule AB, showing dissociative detachment of
molecule AB [52].
Detachment
In addition to the loss of electrons through attachment, there are recombination mecha-
nisms that result in the loss of negative ions. Some of those include associative detachment
and electron impact detachment.
The first process, associative detachment, is the reverse process of dissociative attach-
ment. Figure 3.4 illustrates the potential energy diagram and is shown to accompany the
explanation of this process.
Associative detachment:
∗
A− + B → (AB − ) → AB + e (3.9)
When A− and B interact, they can follow the path a-b-c labeled in Figure 3.4 to become
an unstable AB− and after which, the electron detaches carrying excess energy so that the
unstable AB− relaxes into the AB ground state 1 at c [52]. Table 3.5 shows possible
43
Table 3.5: Associative detachment [54].
H − + H → H2 + e
H − + O2 → HO2 + e
O − + O → O2 + e
O− + O2 → O3 + e
O − + H2 → H2 O + e
OH − + O → O2 + e
OH − + H → H2 O + e
associative detachment processes in water plasmas.
Similarly, electron impact detachment is also an important process and it is analogous
to the process of direct impact ionization of neutrals.
Electron impact detachment:
e + A− → A + e + e (3.10a)
e + AB − → AB + e + e (3.10b)
This process occurs in both negative atomic or molecular ions. In the case of direct
impact electron-neutral ionization, the neutral is ionized if the electron colliding into the
neutral has a potential greater than the ionization potential energy of the neutral. Here,
impact detachment occurs if the electron colliding into the negative ion (either atomic or
molecular) has an energy higher than the electron affinity of A or AB [52].
44
Ion-Ion Recombination
Ion-ion recombination processes remove negative and positive ions in both binary and
three-body collisions. These processes are usually dominant at high pressures, in tens of
torr. Because the work of this dissertation investigates only a low-pressure plasma (p <
1 torr), these processes are not dominant, but are listed here for future works that may
consider operating at higher pressures.
First, it is important to note that these processes require the presence of both negative
and positive ions. Therefore, in the absence of negative ions, these processes do not take
place. Binary ion-ion recombination occurs at lower pressure (p < 20 -30 torr) and three-
body ion-ion recombination requires a higher pressure (p > 20-30 torr) [53].
Binary ion-ion recombination:
A− + B + → A + B ∗ (3.11)
Three-body ion-ion recombination:
A− + B + + M → A + B + M ∗ (3.12)
In a binary collision, the excess energy goes into exciting one of the neutrals. For the
three-body ion-ion recombination, the excess energy is absorbed by the third-body (M)
[52]. Table 3.6 lists some examples of ion-ion recombination processes.
45
Table 3.6: Ion-ion recombination [54].
Reactions Released Energy
H− + H+ → H + H 12.8
O− + O+ → O + O 12.1
O− + O2+ → O + O2 11.6
O2− + O+ → O2 + O 13.2
O2− + O2+ → O2 + O2 11.6
Surface Recombination
Similar to Section 3.1.2 on plasma generation mechanisms, this section on loss mech-
anisms has not exhausted the list of all possible recombination processes, but has described
the relevant ones in water plasmas. The last process to be discussed is most dominant in
low-pressure plasmas, which is the regime of this investigation. In low-pressure plasmas,
the main loss mechanism is not necessarily through the interactions between neutrals, elec-
trons, and ions, but is due to wall-surface recombination. Electrons or charged particles
when colliding onto the wall surface either recombine or react with the wall material. This
is the main concern for the water plasma because hydrogen, being a very low-mass particle,
can easily be lost to wall-surface recombination.
3.1.4 Dissociation of Molecules
The previous two sections describe some electron and ion generation and destruction
processes. There is another important process that does not fall into the above categories but
is an important process in a plasma with a molecular gas; i.e. electron impact dissociation.
For direct dissociation, the reaction is:
46
Figure 3.5: Potential curve for a molecule AB, showing dissociation processes of molecule
AB [52].
e + AB → A + B + e (3.13)
Figure 3.5 shows the energy potential curves that can result in electron impact dissoci-
ation. The dissociation of molecule AB is possible if through a collision, the ground state
of AB is excited to a repulsive state of AB, shown as reaction a and a’ in Figure 3.5 [52].
3.2 Methods of Plasma Production
Plasma as an ionized medium has competing processes for the loss of electrons and
ions. The rate of electron and ion losses, usually via recombination, is dependent on a
number of factors including operating pressure, electron number density, gas properties,
and electron temperature. In ionization processes, the presence of electrons and collisions
between those electrons and other species in the discharge are necessary. This section
describes how the electrons gain their energy in a plasma source; i.e. electron heating
47
mechanisms. These electron heating mechanisms are: Ohmic or Joule heating, secondary
electron emission heating, stochastic heating, and wave-particle interaction heating. Note
that for any particular discharge, a combination of several heating mechanisms is possible
in order to generate and sustain the discharge.
3.2.1 Joule Heating
Joule heating, also known as Ohmic heating, is the most common electron heating
mechanism in most plasma sources. Electrons, in the presence of an electric field (DC or
RF), experience an electric force and accelerate in the direction opposite of the electric
field:
F~ = m~a = eE
~ (3.14)
where F~ is the force acting on the electron, m is the electron mass, ~a is the acceleration
~ is the electric field.

of the electron, e is the electron charge, and E
Independent of the presence of the magnetic field and whether or not the electric and
magnetic fields are uniform, in the absence of collisions, electrons will accelerate and gain
kinetic energy in response to an electric field. When an accelerating electron collides into
another particle, possible outcomes include both elastic and inelastic collisions. In elas-
tic collisions, the sum of kinetic energies are conserved, but particles can exchange their
kinetic energies. In this case, some of the kinetic energy of the electron is transferred to
the other particles and the electron loses its direct motion and travels in a random motion.
48
In inelastic collisions, the kinetic energy of the particles can convert into other forms of
energy such as excitation or ionization of a particle. Joule heating, therefore, refers to the
process in which electrons gain kinetic energy in an electric field and transfer some of their
kinetic energy to other species in the plasma through collisions. The power transferred to
the electron from the electric field in the presence of collisions in a weakly ionized gas is
[53, 54]:
ne e2 E 2
P = σE 2 = (3.15)
mνen
where P is the power density transferred from the electric field to plasma electrons (in
W/m3 ), σ is the electron conductivity, νen is the electron-neutral collision frequency. Again
E, ne , and m are the electric field, electron density and mass, respectively.
The relationship established in Equation 3.15 is derived from the fact that power dissi-
pated is a dot product of current and electric field:
P = J~ · E
~ (3.16)
and a substitution of Ohm’s law (J~ = σ E)

~ is made and assuming that the electron-
neutral collision frequency is constant. The electron conductivity is:
ne e2
σ= (3.17)
mνen
49
Figure 3.6: Plasma sheath in a DC plasma discharge [52].
3.2.2 Secondary Electron Emission Heating
Plasmas are quasineutral (ni ≈ ne ); however, this is only true in the bulk of the dis-
charge. Eventually, the plasma comes into contact with the wall of the discharge. In a layer
that typically has a length (known as a sheath thickness) equal to a few Debye lengths,
quasineutrality is not maintained. This layer is referred to as the sheath. Quasineutrality
cannot be maintained in the sheath because electrons, being low-mass particles, are much
more mobile than the ions. The electron’s thermal velocity is larger than the ion’s by a fac-
p
tor of (Te M )/(Ti m), where Te and Ti are the electron and ion temperature, respectively,
and m and M are the electron and ion mass, respectively. Typically, this value is on the
order of 1000 [53]. An illustration of the plasma with two sheaths are shown in Figure
3.6.
Secondary electron emission heating occurs in the cathode sheath, where there is a
50
sharp potential drop as seen in Figure 3.6. As a result of this potential hill, an ion at the
edge of the plasma sheath is accelerated toward the cathode or the wall at ground potential
as is the case shown in Figure 3.6. Without collisions, the ion, when it strikes the cathode,
has an energy equal to the initial sheath potential energy. If this energy is high enough, a
secondary electron is released from the cathode. In the presence of a strong electric field
in the sheath, the secondary electron gains kinetic energy as it travels up the potential hill.
If the secondary electron collides with a gas particle, ionization can occur if the electron’s
energy is sufficiently large [52].
3.2.3 Stochastic Heating
Another heating mechanism is stochastic heating, also known as collisionless heating;
it is unique to RF plasma sources. In RF discharges, the sheath edge oscillates in response
to the oscillating applied electric field as shown in Figure 3.7. The ion density is taken to
be constant across the sheath assuming that the ion mass is large and therefore the ions do
not respond to the instantaneous potentials. On the other hand, electrons are assumed to re-
spond to the instantaneous potentials and that the electron density is zero in the sheath [52].
This is only a simplified model, but is sufficient to apply to describe stochastic heating.
When electrons in the bulk plasma impinge on the oscillating sheath edge, they suffer
a change of velocity. If the sheath moves into the bulk plasma, the electrons impinging
the sheath are reflected and their velocities are increased by approximately twice that of
the sheath edge velocity [52, 54, 59]. This process is analogous to a low-mass ping pong
ball hitting a moving wall. If the sheath moves away from the bulk plasma, the electrons
51
Figure 3.7: Plasma sheath in a RF plasma discharge.
impinging the sheath lose some of their energy in order to follow the sheath. On average,
if the net electron flux to the boundary moving from the electrode equal that moving to the
electrode, the electron would have a net zero energy gain over one cycle. However, the
electron flux to the boundary moving from the electrode is higher than that moving to the
electrode; therefore, energy is transferred to the electrons [54].
A simplified equation similar to Equation 3.15 can be derived for stochastic heating.
Assuming that the ion density is uniform in the sheath and that the electrons at the sheath
edge have a Maxwellian distribution function, the stochastic electron power for each sheath
is [52]:
1 mv̄e 2
S̄stoc = J1 (3.18)
e e2 n
where S̄stoc is the stochastic electron power per unit area and J1 is the current density,
which is entirely due to conduction inside the plasma.
52
3.2.4 Resonant Wave-Particle Interaction Heating
Despite the high densities that can be achieved by an inductive plasma source–1017 to
1018 m−3 –there are applications in which an even higher density is desired. For example,
water molecule dissociation in the discharge relies on electron-impact reactions. The higher
electron density increases the rate of these reactions. However, inductively-coupled plasma
sources have a physical density limitation [52]. The desire for creating higher plasma
density discharges motivates the work in the design of plasma sources where the electron
heating mechanism involves the interaction between waves and particles. Such heating
mechanisms occur in electron-cyclotron resonant (ECR) and helicon plasma sources. In
ECR sources, a right-hand circularly polarized wave are resonantly absorbed by electrons
in the presence of the magnetic field if the RF frequency is approximately equal to the
electron cyclotron frequency (ωRF ≈ ωce ).
3.3 Plasma Sources
The previous section outlines the main methods of plasma production by discussing
different types of heating mechanisms. Several common types of plasma discharge are
presented in this section: direct current (DC), capacitively-coupled, inductively-coupled,
and helicon plasma sources. This section provides basic descriptions of these sources and
explains the benefits of using the helicon plasma source for the work of this dissertation for
dissociating water molecules through direct electron-impact reactions.
53
3.3.1 Direct Current Plasma Sources
In a direct current (DC) plasma source, the main electron heating mechanisms are de-
scribed previously: Ohmic heating and secondary electron heating. Ohmic heating occurs
in the bulk of the plasma and secondary electron heating occurs in the cathode sheath. A
typical DC glow discharge plasma source consists of two electrodes connected to a DC
power supply. In a positive column discharge, the potential decreases continuously at a
constant rate in the direction from the anode to the cathode [52], establishing a constant
electric field. An electron in the positive column accelerates in the presence of the electric
field, and Ohmic heating is achieved through collisions between electrons and other par-
ticles in the bulk discharge. In the cathode sheath, secondary electrons are emitted from
the cathode surface or from the wall via ion bombardments and secondary electron heating
occurs due to the presence of a strong electric field within the sheath.
In both cases–Ohmic heating and secondary electron heating–gas pressure strongly af-
fects the ionization rate. In Ohmic heating, if the pressure is too high, the mean free path of
the electrons are too short for them to gain enough energy from the electric field to cause
ionization. However, when the pressure is too low, not enough collisions occur to gener-
ate substantial ionization. Likewise, there is an optimum pressure in the cathode sheath to
achieve secondary electron emission. If the pressure is too high, the ions will lose energy
through collisions before they reach the cathode.
54
3.3.2 Capacitively-Coupled Plasma Source
The capacitively-coupled plasma source is the first of the three RF plasma sources to
be discussed in this section. A capacitively-coupled plasma is sustained by RF currents
and voltages directly applied to the electrodes and the electrodes are placed either directly
inside the plasma or outside the chamber. Typically, RF plasma systems are driven by a 50-
Ω RF power source that operates at 13.56 MHz frequency. A matching network is needed
to match the impedance of the RF power supply output to the impedance of the plasma
load. Unlike a DC discharge, a capacitive plasma discharge has stochastic or collisionless
heating that results from the interaction of the RF sheath and bulk plasma. Essentially, as
the electrons oscillate with the time-varying electric field, they impart some of their energy
through collisions with other particles in the bulk discharge. At the same time, depending
on the gas pressure, Ohmic heating is also taken place in the bulk discharge if the pressure
is high enough such that there are sufficient electron-neutral collisions.
3.3.3 Inductively-Coupled Plasma Source
In the capacitively-coupled plasma sources, in general, the electromagnetic fields are
applied directly to the electrodes. For inductively-coupled plasma source, the electromag-
netic field is induced by an inductive coil [54], which can be planar or cylindrical. Com-
pared to capacitively-coupled plasma, the inductively-coupled plasma discharges can reach
high antenna currents, high electric conductivity, and high electron density at relatively
low values of electric field and voltage [54]. The inductive coil can be considered as a
transformer, where the coil represents the primary windings and the plasma the secondary
55
winding. Therefore, the coil can effectively increase the current going into the plasma dis-
charge. It is this effective coupling between the inductive coil and the plasma that allows
the inductively-coupled plasma to achieve as much as 10 times higher in electron density
compared to the capacitively-coupled plasma [54]. Further, an inductively-coupled plasma
source is electrodeless and does not have any issues relating to contamination of the elec-
trode.
In an inductive source, RF current and voltage are usually applied to a planar or cylin-
drical coil outside the discharge. The RF power is coupled to the plasma through a dielectric
window or wall. Due to the time-varying current applied to the coil, it generates an induc-
tive magnetic field and electric field inside the discharge. The electric and magnetic fields
can only penetrate into the plasma within a distance from the surface of the discharge in
what is known as the skin layer. The thickness of this layer is known as the skin depth. The
skin depth δ is the inverse of the spatial decay constant α within a plasma for an electro-
magnetic wave normally incident on the boundary of a uniform density plasma [52]:
1 ω
α= = − ImKp 1/2 (3.19)
δ c
p
In Equation 3.19, ω is the applied RF frequency, c = 1/(µo ²o ) is the speed of light,
and Kp is the relative plasma dielectric constant, which is:
2
ωpe ωpe 2
κp = 1 − ≈− 2 (3.20)
ω(ω − jνm ) ω (1 − jνen /ω)
where νen is the electron-neutral collision frequency. For the collisionless regime,
56
where νen ¿ ω, and substituting the expression for electron plasma frequency, the skin
depth is:
r
m
δ= (3.21)
e2 µ o ns
The skin depth thickness is therefore inversely proportional to the square root of the
electron number density. For example, if the density is 1016 m−3 , the skin depth thickness
is 5 cm, but if the density is one order of magnitude higher at 1017 m−3 , the skin depth
thickness is 1.7 cm.
The collisionless assumption made in deriving Equation 3.21 is valid only for the low
pressure regime in this experiment. For higher pressures where νen ¿ ω condition no
longer holds, the collisional skin depth thickness is:
r r
2νen 2
δc = δ = (3.22)
ω ωµo σdc
where σ is the electron conductivity, which is proportional to the electron number den-
sity. For 250 mtorr and 13.56 MHz, the collisional skin depth is greater than the value of
the collisionless skin depth by a factor of 2-3. In both collisionless or collisional regimes,
power from the electric field can only be transferred to the plasma electrons within this
skin depth layer near the plasma source. The plasma density, therefore, is concentrated in
this layer. The layer thickness can be controlled by the frequency, but at the same time, is
affected by the electron number density. Higher electron number density reduces the skin
depth thickness.
57
Lastly, it is important to note that even if an inductive mode is excited, there is always
some capacitive coupling [52]. At high plasma densities, the skin depth thickness is small,
and inductive coupling dominates. But at lower densities, the skin depth thickness can
be large (for example, 5-10 cm) and power deposition through capacitive coupling can be
significant.
3.3.4 Helicon Plasma Source
The plasma source employed for the experimental part of this dissertation was a helicon-
type plasma source. Helicon waves were first researched in the early 1960s but it was not
until 1970 that Boswell’s dissertation work led to a discovery that a high, dense plasma
source could be generated by exciting helicon waves [62,63]. Since then, numerous exper-
imental and computational investigations of the properties and physics of helicon plasma
sources have been studied [22, 25, 64–70]. A helicon plasma source has advantages sim-
ilar to an inductive plasma source (electrodeless system and high density plasma), but its
density does not have the same limitation as an inductive plasma source (1017 − 1018 m−3 ).
In addition, in contrast to ECR sources, a helicon plasma source does not require a high
magnetic field.
Chen [64] derived the dispersion relation for helicon waves as follows. From linearized
Maxwell’s equations, in particular:
~
~ = − ∂B
∇×E (3.23)
∂t
58
~ = µo J~
∇×B (3.24)
and
~
~ = J~ × Bo
E (3.25)
eno
~ B,
where all the values without a subscript (E, ~ and J)
~ are perturbed values, and no and
~ o are equilibrium values. Chen arrived at the dispersion relation for the helicon waves
B
after solving the Besel’s function for m=1 mode as:
Bo eµo R ³ ω ´
= (3.26)
no 3.83 k
where R is the radius of the discharge, and k is the wave number. In arriving at this
dispersion relation, it was assumed that ωLH ¿ ω ¿ ωce , where ωLH is the lower hybrid
frequency. By assuming that ωLH ¿ ω, the ion motion is neglected. And by assuming
ω ¿ ωce , the electron cyclotron motion is also neglected because it is too fast compared
to the operating frequency. Equation 3.26 reveals an important relationship between the
dimensions of the discharge and the magnetic field required and the wave number (therefore
wavelength). This relationship can be used to design a helicon discharge.
In a helicon plasma discharge, low-frequency whistler waves are confined to an insulat-
ing cylinder with an applied DC magnetic field. In addition to collisional heating, Landau
damping and Trivelpiece-Gould modes coupling have been attributed to the efficient trans-
59
fer of the wave energy to the electrons [65, 67]. Although the exact mechanism of energy
transfer is still a subject of debate, it is believed that electrons gain energy from the helicon
mode as it propagates along the column by collisional or collisionless (Landau) damping
[52]. Collisional heating takes a more significant role only at higher pressures. In Landau
damping, the energy of a wave is transferred to electrons which have velocities near the
phase velocity of the wave [52, 66].
3.3.5 Benefits of Radio-Frequency Plasma Sources
Sections 3.2 and 3.3 describe the methods of plasma production and some main types
of plasma discharges. The RF plasma sources (capacitive, inductive, and helicon) in gen-
eral can produce higher plasma density, and at a higher efficiency than DC plasma sources.
Within the choices of RF plasma sources, inductive sources can produce higher plasma den-
sity than capacitive sources, but helicon plasma sources can produce an even higher density
and more efficiently. The selection of the discharge for this dissertation work depends on
several criteria and a helicon-type discharge was chosen due to its versatility. A helicon-
type discharge can operate in capacitive, inductive and/or helicon modes and furthermore,
it was chosen for the following reasons:
Electrodeless: A radio-frequency plasma source as used in this work is an electrodeless
system. In a water plasma source, presence of highly reactive radicals and water molecules
can destroy the electrode. Thus, an electrode-less system is necessary for this application.
High ne : Radio-frequency plasma sources are known to produce among the highest
electron number densities. As this method of hydrogen production relies on electron impact
60
Figure 3.8: Experimental setup to study decomposition of phenol solutions [12].
collisions, higher electron density is favorable.
Efficiency: Radio-frequency plasma sources, particularly those operated in inductively-
coupled or helicon modes can generate high plasma density very efficiently.
3.4 Water Plasma Studies
Water plasmas have gained widespread utility in many fields. This section reviews
applications of water plasmas in a variety of areas including for hydrogen production.
3.4.1 General Applications of Water Plasma
Plasma discharges operated on water–in either liquid or vapor form–can be used in a
wide range of applications [11–21]. For example, the decomposition of organic contami-
nants in waste water, such as phenol, into more benign byproducts first started by seminal
work by Clements [11]. For example, Figure 3.8 shows a setup of an experiment to study
the decomposition of phenol solutions in a water plasma.
61
Figure 3.9: Experimental setup to study diamond growth in a water plasma source [17].
In addition to directly generating a discharge in liquid water, one can also generate
a plasma from water vapor. In the intermediate pressure regime between tens of mtorr
to one torr, water vapor plasmas are now widely used in plasma-assisted chemical vapor
deposition for diamond film formation. Currently, CH4 and H2 are injected into a plasma
chamber to fabricate diamond films [16, 17]. When O2 is introduced, the presence of O2
and OH radicals is stipulated to increase the rate of diamond growth. Instead of injecting
O2 and H2 separately, other researchers have shown that the addition of H2 O vapor in the
main gas feed favorably contributes to diamond growth rate. Furthermore, water is easier
to handle and is essentially cost-free. Figure 3.9 shows an experimental setup of a work
led by Singh [17] in which water with the addition of methanol is used to study diamond
growth.
Ultraviolet (UV) light source plasma is another example of an application of water
vapor plasma. UV light sources have widespread applications. As an example, Hg vapor
62
is the current working gas in fluorescent lamps to produce UV radiation. UV lamps are
also used to sterlize tools in laboratories and hospitals. However, the current method uses
Hg vapor to generate the UV light source. Water vapor plasma as UV light sources can
replace current Hg lamps, mainly because water vapor is easier to handle than Hg. Also,
water vapor does not have the harmful effects that Hg vapor has on human. Recently, Oh
et al. investigated the potential application of water vapor plasmas excited by microwaves
as a UV light source [20].
Lastly, water plasma is observed to have played an important role in an emerging field
of plasma medicine where an interest in applications of non-thermal (non-equilibrium)
plasma is growing [21, 71]. In plasma medicine’s nascent stage, reactive oxygen species
(O and OH) are shown to have the ability to inactivate various organisms. This can be
applied in sterilization of medical instruments [71]. For example, one can envision that
after medical instruments are washed, they can be placed directly into a plasma discharge.
The working gas of this discharge would be the water that remained on the wet instruments.
For this application to be utilized, the cost of the system must be reduced. In particular,
the plasma source must be able to operate at atmospheric pressure to reduce any pumping
systems, hence additional cost. In addition, biological systems are composed of water, and
therefore a plasma can be produced within these environs. Water plasma has many potential
applications and is emerging rapidly in the medical plasma research community because
it contains an abundance of reactive species such as OH− , O2 − and H. A good review of
current research in plasma medicine is given in References 53 and 21.
63
Figure 3.10: Experimental setup to study CO2 dissociation [53].
3.4.2 Plasma Electrolysis
There are other plasma methods for hydrogen production, such as plasma catalytic re-
forming of natural gas [72,73]. However, this literature review only focuses on applications
of plasma to directly dissociate water molecules for hydrogen production.
For direct dissociation of water molecules for hydrogen production, the first known
published work is by V.K. Givotov from I.V. Kurchatov Institute of Atomic Energy in 1981
from Moscow [74, 75]. Initially, this group investigated the direct dissociation of CO2 in
a microwave plasma discharge as shown in Figure 3.10. The plasma discharge operated
typically at 2.4 GHz at 1.5kW, and the diameter of the quartz tube was 28 mm [53].
The typical pressure was 50-100 torr. This plasma source was later used to study water
dissociation with a flow rate of 0.05–0.6 g/s (4–48 slm).
The energy efficiency achieved from this work was reported to be 30–40% [74]. This
group also investigated dissociation using other discharge methods. The glow discharge
was used and the energy efficiency reached 12–20%, lower than the microwave discharge.
64
Figure 3.11: Experimental setup to study hydrogen production in a water plasma source
[5].
In addition, a glow discharge with a hollow cathode was also used, with an energy efficiency
of 2%. The power level was 1–300 W and the pressure range was 0.4–4 torr.
In addition, Chaffin et al. dissociated water by placing electrodes above a liquid water
surface, as shown in Figure 3.11 [5]. Chaffin’s method mimics electrolysis, but uses
electrodes that are conventionally used in a plasma discharge instead of using a membrane
separation method.
Chaffin argues that the widespread adoption of conventional electrolysis systems has
a physical limitation. In these systems, hydrogen must diffuse in liquids, but its diffusion
rate in a liquid is slower compared to its diffusion rate in a gas [5]. Chaffin suggested
that in using plasmas, the cost of electrolytes and catalysts required in conventional elec-
trolysis can be removed and attributed the presence of high-energy electrons as the key in
electrolyzing water.
65
3.5 Summary
In summary, this chapter begins with an overview of the plasma physical processes in
water vapor discharge. Then, the method of plasma production and several types of plasma
sources are discussed. Lastly, a survey of water plasma studies are provided, first with the
general applications followed by electrolysis applications of water plasma.
66
Chapter IV
Experimental Setup and Diagnostics
This chapter describes the experimental set-up including the vacuum chamber, the water
feed system, and the plasma discharge. It also describes the diagnostics employed in this
work: residual gas analyzer (RGA), Langmuir probe, and optical emission spectrometer.
4.1 Experimental Setup
4.1.1 Vacuum Chamber
Testing of the water vapor RF plasma source was performed at the Plasmadynamics
and Electric Propulsion Laboratory (PEPL) in the Cathode Test Facility (CTF). The vacuum
facility was a cylindrical aluminum chamber that is 2.44 m in length and 0.61 m in diameter.
An Edwards XDS 35i dry scroll pump was used to evacuate the chamber. Unlike oil-based
vacuum pumps, dry scroll pumps are essentially oil-free and can therefore be operated in
water vapor-rich environments. With a maximum pumping speed of 35 m3 /hr on N2 [76],
the base pressure achieved was below 3 mtorr. The plasma source was attached to a 15-cm-
67
Figure 4.1: Drawing of vacuum facility
diameter side port on the CTF. A CAD drawing of the vacuum facility is shown in Figure
4.1.
In addition, the vacuum facility is also fitted with a CVI model CGR 411-LS cryogenic
pump that can achieve a base pressure below 10−6 torr. For gases such as argon, within a
reasonable flow rate these cold surfaces can maintain a temperature approximately 14-20
K and the cryopump can run for hours. When water vapor is introduced into the chamber,
water vapor contaminates the charcoal surface of the cryogenic pump and light gases such
as hydrogen can no longer be pumped.
68
4.1.2 Plasma Source
Figure 4.2 shows a photograph of the plasma source assembly and Figure 4.3 shows
a photograph of the plasma source. The source consisted of a 15-cm-diameter quartz tube
that was 50 cm in length. Three magnetic coils connected in series to a Lambda EMI-EMS
40-V, 60-A DC power supply produced a peak axial magnetic field on axis of approximately
200 G and 400 G at 30-A and 60-A magnet current, respectively. The magnetic field was
measured by a 3-axis Hall probe that was connected to a 3-channel Gauss meter from
Lakeshore (model 460). The mapping was achieved by placing the Hall probe on three-
axis motion control system. Figure 4.4 shows the mapping of the magnetic field for the
60-A magnet current setting in the horizontal plane of the quartz tube at the centerline, with
an outline of the quartz tube and the flange that is used to connect the quartz tube to the
CTF vacuum chamber. The magnetic field mapping for a magnet current of 30 A is exactly
identical to that at 60 A, with the exception that the magnetic field strength is reduced in
half.
A double helical (m = 1 mode) antenna circumscribed the quartz tube, and was sand-
wiched between the quartz tube and the magnetic coils, as shown in Figure 4.5. A Plas-
maTherm 13.56-MHz RF power supply was used to excite the antenna up to 1.5 kW. A
π-style, water-cooled, 5-kW matching network purchased from Manitou Systems Inc. was
employed to match the impedance of the RF power supply output with the impedance of
the plasma, reducing the reflected power to less than 5% of input power for all conditions
investigated. More details on the plasma discharge can be found in Reference 25.
The assembly, as shown in Figure 4.2, was housed in a Faraday cage including the en-
69
Figure 4.2: Photograph of setup including the magnetic coils, matching network, and the
RF power supply
Figure 4.3: Photograph of plasma source at PEPL. Plasma source shown attached to the
CTF vacuum chamber at PEPL with one magnet pulled away to show antenna
and quartz tube.
70
Figure 4.4: Mapping of magnetic field at 60-A magnet current with an outline of the quartz
tube and an adapter flange.
Figure 4.5: Schematic diagram of plasma source.
71
tire RF system: the quartz tube, the magnetic coils, the matching network, and importantly
the RF power supply. The Faraday cage was built from very fine copper mesh with a hole
size of a few mm to prevent RF interference on diagnostic sensors and computer systems
throughout the lab.
4.1.3 Water Feed System
An Eldex water pump–a positive displacement, reciprocating piston pump–with a flow
rate range of 12.5 sccm - 25,000 sccm H2 O was used to meter liquid water into the plasma
source. A back pressure regulator from Upchurch Scientific was placed in the gas line be-
tween the water pump and the vacuum chamber to maintain a pressure of 250 psi. In this
work, the vacuum chamber background pressure ranged between 50 and 500 mtorr. In this
pressure range and at room temperature, water is in the vapor phase. Therefore, a small
amount of liquid water (up to 125 sccm) injected into the plasma discharge was immedi-
ately vaporized. Nonetheless, the line between the water pump and the quartz tube was
heated with heating tape up to 400 K to ensure that water molecules were not condensed
onto the metal surface of the feed line and that water remained in vapor state upon entry
into the plasma source. De-ionized water was used. The water passed through a 10-µm fil-
ter before reaching the pump. The filter was used as a precaution to prevent any particulate
from entering the gas line. Figure 4.6 depicts the layout of the water delivery system. As
shown, water delivery system was placed outside of the Faraday cage, and was connected
to the quartz tube inside the cage via a ceramic-to-metal adapter to electrically isolate the
gas feed lines from the RF system.
72
Figure 4.6: Schematic diagram of water delivery system.
73
In addition to the water pump, a low-flow 10 sccm Celerity (UNIT 7300 model) and
a high-flow 2 slm Celerity (also UNIT 7300 model) mass flow controllers were used to
regulate flow of argon, hydrogen, and oxygen gases when flow of those gases were needed.
The need to flow argon and the process of calibrating the RGA using hydrogen and oxygen
gases are discussed in Section 4.2.1.
4.2 Diagnostics
In this investigation, Langmuir probe data were used to determine plasma properties
for both argon and water vapor discharges. The optical emission spectrometer was used to
identify the presence of excited hydrogen and hydroxyl in the discharge to confirm water
dissociation. The RGA was also used to identify, as well as, to make quantitative estimates
on the concentration of hydrogen, oxygen, hydroxyl, and water in the discharge.
4.2.1 Residual Gas Analysis System
A commercial residual gas analyzer–Stanford Research Systems RGA100–was used
in this work to identify gas species inside the chamber. The residual gas analyzer (RGA)
system as shown in Figure 4.7 is a mass spectrometer consisting of an ionizer, quadrupole
probe, and an electronic control unit. When gas molecules enter the RGA, some are ionized
by the ionizer. A combination of RF and DC voltages are applied to the quadrupole probe
to select ions with a certain mass-to-charge ratio to pass through the probe system. The
RGA varies the RF and DC applied to the quadrupole rods to sweep through the entire
spectrum of mass-to-charge ratio which ranges between one and 100 AMU (Atomic Mass
74
Figure 4.7: Components of the RGA, courtesy of SRS [77].
Unit). After passing through the probe filter, the ions are focused on an ion detector, and
an electrometer is used to measure the analog current.
The RGA was designed to operate with a maximum allowable pressure of 10−5 torr,
significantly lower than background pressure of up to 500 mtorr in this experiment. There-
fore, a differential pump system was utilized to reduce the pressure to allow for RGA
operation. Figure 4.8 depicts the layout of this system. A variable leak valve isolated the
RGA chamber from the plasma chamber. A 70-l/s Varian turbopump was used to evacuate
the RGA chamber to achieve a base pressure below 10−7 torr. The turbopump was backed
by a mechanical pump from Edwards, E2M30 [78].
Without a proper calibration, the RGA only yields qualitative information, such as the
presence of certain gas species in the plasma. To quantitatively estimate the amount of
hydrogen, oxygen, and hydroxyl produced in the plasma, the RGA was calibrated for each
gas individually by injecting a known amount (10 sccm) of argon and a varying amount of
the gas of interest; i.e. hydrogen or oxygen. For example, to calibrate hydrogen, a mass
flow controller was used to meter the flow rate of hydrogen into the chamber while a dif-
ferent mass flow control was used to meter a fixed (10 sccm) flow rate of argon. The partial
pressures of argon and hydrogen were recorded for each flow rate setting of hydrogen.
75
Figure 4.8: Differential pump system for the RGA including a variable leak valve, 70 l/s
turbo pump, and pressure transducer.
A calibration factor (CF) (Equation 4.1) relating the ratio of the partial pressure of the
gas of interest to the partial pressure of argon could therefore be obtained from the ratio of
the flow rates of the two gases.
Pi ṁi
= CF (4.1)
PAr ṁAr
where PAr is the partial pressure of argon, Pi is the partial pressure of the gas of interest,
ṁAr is the known mass flow rate of Ar, and ṁi is the rate of production of the gas of interest.
The same procedure was followed to calibrate the oxygen and water flow rates detected
by the RGA. Figure 4.9 shows the calibration curves of hydrogen, oxygen, and water. The
pressure ratio varies linearly with mass flow rate ratio as expected. Note that even though
the flow rate of water vapor injected into the system was known, this calibration was later
76
Figure 4.9: Calibration curves for hydrogen, oxygen, and water relating the RGA partial
pressure ratio to the flow rate ratio.
used to determine the quantity of undissociated water molecules in the discharge. When
two gases were being injected into the system, the water pump as shown in Figure 4.6 was
replaced by a mass flow controller.
With CF known for each gas, the mass flow rate of a gas of interest produced in the
water vapor plasma source is estimated following Equation 4.2.
ṁAr Pi
ṁi = (4.2)
CF PAr
As shown, the mass flow rate of the gas of interest can be determined if a known amount
of argon is injected into a multiple species plasma. The mass flow rate of a gas is deter-
mined when the CF and the mass flow rate of argon are known and the partial pressures of
the gas of interest and of argon are obtained from the RGA.
Unlike the case of hydrogen or oxygen, hydroxyl (OH) is highly reactive and short-
77
lived, and therefore it is not a commercially available gas that can be used to calibrate
the production of hydroxyl in the plasma discharge. To estimate the flow rate of OH pro-
duced in the plasma, the calibration factor for O2 was used. Between the three choices of
calibration factor that are available (hydrogen, oxygen, and water), hydroxyl most closely
resembles oxygen. The hydrogen gas is too low-mass and the water vapor is not as reac-
tive as the hydroxyl. Therefore, a choice was made to use the calibration factor of oxygen
because they both are reactive and have the same order of magnitude in mass. As a conse-
quence, the OH production rate reported in this work is only used to study the trend in OH
production, not necessarily to assess its precise production rate.
Effects of the addition of argon to the plasma properties and hydrogen production was
also studied. The addition of 10-sccm Ar to calibrate the RGA is shown to have a negligible
effect on the plasma. As can be seen in Figure 4.10, the rates of hydrogen production for
the addition of 5-sccm and 10-sccm Ar are almost identical at all RF power levels except
for 750 W. Even at this power level, however, the two results are within the uncertainty of
the measurement technique. Furthermore, Figures 4.11 and 4.12 illustrate the invariance
of electron density, ion density, plasma potential, and floating potential with argon flow
rate.
For the RGA results, the sample size is four sets of three spectra, and the error bars
also represent one standard deviation of spread in each set of measurement. Based on
the calibration of the RGA, the error for determining hydrogen and oxygen production is
5–10%, and for determining water vapor flow rate is 10–20%.
78
Figure 4.10: Effect of argon flow rate on hydrogen production for 75-sccm H2 O with 5-
sccm and 10-sccm Ar operating with 30-A magnet current.
Figure 4.11: Effect of argon flow rate on ion and electron densities for 75-sccm H2 O with
5-sccm and 10-sccm Ar operating with 30-A magnet current.
79
Figure 4.12: Effect of argon flow rate on floating and plasma potential for 75-sccm H2 O
with 5-sccm and 10-sccm Ar operating with 30-A magnet current.
4.2.2 Langmuir Probe
Langmuir probe is used to determine plasma properties such as density, electron tem-
perature, and floating and plasma potentials. It was first coined by Irving Langmuir [79,80]
in 1926 and has been used widely by the plasma research community. The Langmuir probe
in its basic form is a conducting tip that is electrically isolated from its probe holder. The
Langmuir probe system together includes the conducting probe tip, an external power sup-
ply, and an ammeter. The ammeter measures the current to the probe and the power supply
sets the bias voltage on the probe. Consequently, a current-voltage (IV) curve is obtained
for each sweep.
A typical IV curve is shown in Figure 4.13. When the applied voltage is set at a
value that results in a zero net current to the probe (I+ = I− ), then the value of the applied
voltage is known as the floating potential. The region where the applied voltage is below
the floating potential is the ion collection region. In this region, positive ions are attracted
80
Figure 4.13: (¦) A typical Langmuir probe curve for water vapor plasma operating with
75-sccm H2 O and 10-sccm Ar at 500-W RF power and 30-A magnet current.
(◦) Probe current squared versus voltage for the same plasma, indicating the
slopes in the ion and electron saturation regions taken to calculate ion and
electron densities, respectively.
to the probe while electrons and negative ions are repelled. At a sufficiently large negative
applied potential, all the electrons are repelled and the probe only collects the available ion
current as established by sheath conditions. This current is known as the ion saturation
current. On the other hand, if the applied voltage is higher than the floating potential, the
electrons and negative ions increase in number that can be collected at the probe. Similarly
at a sufficiently large positive applied potential, only electrons are collected by the probe,
and the collected current is referred to as the electron saturation current.
A commercial RF-compensated single Langmuir probe (LP) system from Hiden Cor-
poration is used in this experiment to determine plasma density, floating potential, plasma
potential, and electron temperature. This commercial system includes a data acquisition
software package, RF compensation circuitry, a driver power supply and an ammeter. The
LP collector is 0.12 mm in diameter and 3 mm in length. The probe was placed at the axis
81
of symmetry and just downstream of the plasma source axially.
Langmuir probe analysis falls into two main categories (collisionless or collisional),
depending on the ratio of the local electron and ion mean free paths and the Debye thickness
(λD ) [81]. This ratio represents the dimensionless Knudsen number:
λi,e
Kn = (4.3)
λD
where Kn is the Knudsen number, λi,e is the local ion and electron mean free path, and
λD is the Debye thickness.
For a background gas pressure between 50 and 500 mtorr, the neutral number density
ranges between 1021 and 1022 m−3 . The maximum value for electron and ion density is
below 1017 m−3 ; therefore, collisions are likely to occur between a neutral and an electron
or ion instead of between charged particles. The local mean free path of the electron and
ion is:
1
λi,e = √ (4.4)
2σn
where σ is the collisional cross section (≈ 8.8×10−21 m−2 ) and n is the neutral density.
The Debye length is calculated from:
r
k B Te ² o
λD = , (4.5)
ne2
Although the operating pressure for this investigation can be up to 500 mtorr, the local
82
electron and ion mean free paths are found to range between 3.2 and 32 mm, which is
sufficiently higher than the Debye length, which is approximately 0.4 mm. Therefore, the
collisionless probe model is applied for LP analysis in this experiment.
Due to the presence of an applied magnetic field in the plasma, magnetic effects were
examined for the Langmuir probe analysis. Even though the maximum peak of the mag-
netic field strength is 200 G and 400 G for an applied magnet current of 30 A and 60 A,
the LP data are collected in the region downstream of the quartz tube. There, the magnetic
field strength is reduced significantly to only 40 G and 80 G for 30-A and 60-A applied
magnet current, respectively. In the presence of the magnetic field, both electrons and ions
gyrate along the field lines. To determine if the magnetic field affects the current collected
by the LP probe tip, the cyclotron radius of the electrons:
v⊥
rc = , (4.6)
|ωc |
are compared with the probe radius (0.06 mm) and the Debye length (Equation 4.5).
To calculate the electron cyclotron radius, the magnitude of the perpendicular velocity (v⊥ )
is approximated as the average electron speed:
r
8eTe
v̄e = (4.7)
πm
For electron temperature between 1 and 3 eV, electron density between 5 × 1015 and
1016 m−3 , and magnetic field 40 and 80 G, the electron cyclotron radius is between 0.4 mm
and 1.5 mm. The Debye length is between 0.02 and 0.18 mm.
83
In general, the electron cyclotron radius is always greater than the probe radius and the
Debye length. Therefore, the magnetic effects on the LP can be neglected.
Figure 4.13 shows a typical current versus voltage (IV) Langmuir probe curve and
Figure 4.14 shows the same data in semi-log plot. The LP date files were imported and
were analyzed in MATLAB. As shown, the floating potential is obtained directly from the
IV trace, the value of the bias voltage where the net current to the probe is zero. The
electron current is obtained from the IV characteristic data by subtracting the ion current
from the measured current in the electron collection region. The effect is shown in Figure
4.14, where the linear region is further extended when the ion current is subtracted. The
electron temperature is derived from the slope of the semi-log plot, and the plasma potential
is taken to be the bias voltage where the two lines intersect in this semilog plot, as shown
in Figure 4.14.
Within the collisionless model of the LP analysis, the ion density is calculated in two
ways based on the ratio of the probe radius and the Debye length, rP /λD . Figure 4.15
shows a flow chart of the LP analysis routine. First, a thin sheath (rP /λD > 10) is assumed
and the ion density is calculated as follows [81].
r
Ii,sat Mi
ni,sat = , (4.8)
0.61As e eTe
where ni is the ion density, e is the electron charge, As is the sheath surface area, Ii,sat
is the ion saturation current, and Te is the electron temperature. In this work, an effective
ion mass, Mi , is calculated from a weighted average of the product of mass (mi ) and partial
84
Figure 4.14: Semi-log plot of Langmuir probe trace showing the linear region where elec-
tron temperature is calculated and the point at which the plasma potential is
derived.
pressure (pi ) obtained from RGA data.
P
i mi pi
Mi = P (4.9)
i pi
After an estimated ion density is obtained, the Debye length is calculated, and the ef-
fective probe area is calculated following the derivation in Hutchinson [82]. The new ion
density is calculated based on the effective probe area and the Debye length is recalculated
to account for the new ion density. As such, the effective probe area, ion density, and De-
bye length are interdependent. In Figure 4.15 the steps in calculating the ion density are
outlined. A loop is applied for steps 7 through 9. After a converged solution is obtained
for the ion density, the ratio of probe radius to Debye length, rP /λD , is calculated. If the
ratio is greater than 10, the previously assumed thin-sheath analysis is used and ion density
85
Figure 4.15: Flow chart for Langmuir probe analysis – ratio = rp /λD
is the ion density obtained in step 9.
However, if the ratio is less than three, the ion number density is calculated using the or-
bital motion limited (OML) analysis model [81, 83–85]. In a thick sheath, orbital motions
limit some of the particles that enter the sheath from hitting the probe. The current to the
probe in this case is independent of the size of the sheath surrounding the probe. Lafram-
boise derived a method to determine the ion number density in this regime assuming that
the plasma is collisionless, stationary, and isotropic [83, 84]
s µ ¶
π Mi dIi2
ni = (4.10)
Ap 2e3 dV
In Equation 4.10, Ap is the surface area of the probe, and dIi2 /dV is the slope of current
86
squared versus the bias voltage for ions as shown in Figure 4.13.
If the ratio is between 3 and 10, a weighted average density between the two methods
is used. Again, the Debye length and ion number density are interdependent, therefore a
loop is applied for steps 9 through 11.
Despite this laborious effort to ensure that the ion density was calculated appropriately
within the right regime, it was determined that only OML analysis was needed because the
probe radius is sufficiently small that the ratio rP /λD is always less than three.
For argon plasma, quasineutrality is assumed, hence ni ' ne . However, because of
the presence of negative ions in the water plasma, quasineutrality in this case requires that
ni+ ' ne + ni− , where ni+ and ni− are positive and negative ions. The electron density is
calculated separately, using Equation 4.11:
r
π me dIe2
ne = (4.11)
Ap 2e3 dV
In Equation 4.11, dIe2 /dV is the slope of current squared versus the bias voltage for
the electrons as shown in Figure 4.13.
Results given in this dissertation are the mean values of a sample of measurements.
Error bars represent one standard deviation among the spread in the measurements sampled.
For the Langmuir probe data, the sample size includes four sets of 5 to 10 Langmuir probe
traces. The measurement is taken in a random order of RF power. While the absolute error
in the Langmuir probe diagnostics is 20–30% for electron temperature and 50–60% for
plasma densities [82], the relative error from point to point is small, which allows for a
87
Figure 4.16: Diagram of the spectrometer setup, showing light is collected by the collimat-
ing lens and is transferred to the spectrometer via a fiber optic cable.
study of trends in plasma properties.
4.2.3 Spectrometer
In addition to the RGA, two spectrometers from Ocean Optics are used to obtain optical
emission spectra from the plasma: USB2000 and HR4000. The USB2000 model spectrom-
eter has a 25 µm entrance slit and a detector wavelength range of 200-850 nm. The HR4000
model spectrometer has a 5 µm entrance slit and a detector wavelength range of 400-850
nm. A collimating lens is used to focus light into a fiber optic cable that is connected to the
spectrometer via a SMA connector. Figure 4.16 shows this setup in the vacuum chamber.
88
Chapter V
Experimental Results
A water vapor plasma as discussed in Chapter 3 is electronegative. To characterize the
differences between an electronegative and an electropositive plasma, plasma properties of
an argon plasma (electropositive) and a water plasma are presented in Section 5.1. Section
5.2 examines the effects of RF power and magnetic field strength on hydrogen production
and plasma properties of the water plasma. The effects of water input flow rates on hy-
drogen production and plasma properties are presented in Section 5.3. Finally, Section 5.4
summarizes the experimental results in this chapter.
5.1 Argon and Water Vapor Plasma Comparison
The aim of this section is to introduce some key characteristics of a water plasma dis-
charge by comparing its plasma properties with those of an argon discharge. The operating
pressure for the comparisons presented in this section is approximately 300 mtorr. Plasma
properties–plasma density, floating and plasma potential, and electron temperature–are pre-
sented for matching conditions of RF power and axial magnetic field strength.
89
Figure 5.1: Optical emission spectrum of an argon discharge showing most of the argon
neutral lines in the 400-850 nm wavelength range.
5.1.1 Optical Emission and Residual Gas Analyzer Spectra
Optical emission and residual gas analyzer spectra are used to identify the gas species
inside the plasma discharge. Figures 5.1 and 5.2 show an optical emission spectrum
from the plasma source operating on argon and on water vapor, respectively. An HR4000
spectrometer is used to obtain the spectrum shown in Figure 5.1. In order to detect the UV
OH bands, an USB2000 spectrometer equipped with a detector that has a wider wavelength
range was used and the spectrum is shown in Figure 5.2.
As illustrated, the optical emission spectrum of the argon plasma contains all of the
expected prominent argon neutral lines as reported in the literature from NIST [86]: 696.5,
706.7, 738.4, 750.4, 751.5, 763.5, 772, 794, 800, 801, 810, 811, 826, 840, and 842 nm. A
similar spectrum for a plasma discharge operating on water vapor shows a strong presence
of hydrogen and hydroxyl. In particular, the first two lines in the Balmer series [86]–Hα at
90
Figure 5.2: Optical emission spectrum of a water plasma discharge showing the hydrogen
lines and the two OH bands.
656.3 nm and Hβ at 486.1 nm–and two OH bands in the UV range [87]–281-303 nm and
305-330 nm–are clearly present.
The production of hydrogen and hydroxyl as shown in Figure 5.2 indicates that water
molecules are being dissociated in the plasma discharge. Table 5.1 lists the major electron-
water impact reactions. The cross-sections for these electron impact reactions with water
are shown in Figure 5.3 as reported in a literature review by Itikawa and Mason in Ref-
erence 88. As shown, the dissociative mechanisms that dominate at the lower electron
energies are dissociative attachments. Direct dissociation and dissociative ionization only
occur at high electron energies greater than 10 eV.
The presence of excited atomic hydrogen lines and hydroxyl bands confirm that water
molecules are being dissociated in the main plasma discharge. In addition to using optical
spectrometers, an RGA is also employed to confirm the dissociation of water molecules in
91
Table 5.1: Electron impact reactions with H2 O
Momentum transfer H2 O + e ⇒ H2 O + e
Vibrational Excitation H2 O + e ⇒ H2 O + e
Dissociative Attachment H2 O + e ⇒ H − + OH
H2 O + e ⇒ O − + H2
H2 O + e ⇒ OH − + H
Dissociation H2 O + e ⇒ H + OH + e
H2 O + e ⇒ O + 2H + e
Ionization H2 O + e ⇒ H2 O+ + 2e
H2 O + e ⇒ OH + + H + 2e
H2 O + e ⇒ O+ + H2 + 2e
H2 O + e ⇒ H2+ + O + 2e
H2 O + e ⇒ H + + OH + 2e
Figure 5.3: Cross-sections of electron impact with water molecules.
92
Figure 5.4: RGA spectrum of the water plasma indicating the presence of molecular hy-
drogen and oxygen, hydroxyl, and water. An argon peak is observed because a
small amount of argon was added to the gas line for calibration purposes.
the discharge. For a discharge operating on only argon gas, there are two peaks: one dom-
inant peak for singly-ionized argon at 40 AMU/e and one small peak for doubly-ionized
argon at 20 AMU/e. For a plasma discharge operating on water vapor, the RGA spectrum
contains several lines indicating the presence of molecular hydrogen (2 AMU/e) and oxy-
gen (32 AMU/e), hydroxyl (17 AMU/e), and water (18 AMU/e), as shown in Figure 5.4.
Further, when a small amount of argon is added to the gas line, a peak at 40 AMU/e is also
observed. Therefore, RGA spectra further validate the dissociation of water molecules in
the plasma discharge.
5.1.2 Plasma Densities
All plasma properties presented in this dissertation are obtained from Langmuir probe
data collected downstream of the quartz tube and at the axis of symmetry for the source (r
93
Figure 5.5: Plasma density of an argon discharge as a function of RF power for 0-A, 30-A,
and 60-A applied magnet current.
= 0 location) as explained in Section 4.2.1. For an electropositive plasma such as argon,
quasineutrality requires that the positive ion density equal to the electron density (ni+ =
ne ). Figure 5.5 shows the ion density of the argon plasma as a function of RF power for
0-A, 30-A, and 60-A applied magnet current settings. The density is shown ranging from
1.8 × 1017 to 2.8 × 1017 m−3 . The error bars are shown, and they range from 0.04 ×
1017 to 0.54 × 1017 m−3 , which correspond to 1% to 19% of the maximum value of the
ion density. As shown, the error bars can be large at high RF power levels. However, the
general trend shows that the ion density increases with RF power as expected for all applied
magnet current settings. The result at 1000-W RF power level for the case of 60-A is an
outlier to this trend, which is probably due to a mode change in the plasma.
Quasineutrality in the water discharge requires that the total positive ion density is
equal to the sum of the negative ion density and the electron density (ni+ = ne + ni− ).
94
Figure 5.6: Ion density of the water plasma as a function of RF power for 0-A, 30-A, and
60-A applied magnet current.
In this discharge, dissociative electron attachment (Equation 3.7) and three-body electron
attachment (Equation 3.8) are expected to be nontrivial due to the presence of high electron
affinity species such as oxygen and hydroxyl that form negative ions. Thus, the ion density
and the electron density are presented separately in Figures 5.6 and 5.7 for the water
plasma. Under matching operating pressure, the ion density for the water vapor plasma
is approximately one order of magnitude less than that of the argon plasma, ranging from
0.85 × 1016 to 4.6 × 1016 m−3 . The error bars range from 0.03 × 1016 to 0.52 × 1016 m−3 ,
which correspond to 0.1% to 11% of the maximum value of the ion density. The electron
density of the water plasma is up to two orders of magnitude lower than the electron density
of the argon plasma, ranging between 0.45 × 1015 m−3 and 7.9 × 1015 m−3 . The error bars
range from 0.03 × 1015 to 2.37 × 1015 m−3 , which correspond to 0.3% to 30% of the
maximum value of the electron density.
95
Figure 5.7: Electron density of the water plasma as a function of RF power for 0-A, 30-A,
and 60-A applied magnet current.
As expected, the electron density of the water plasma is lower than that of the argon
plasma because the water plasma has more collisional energy loss pathways for electrons.
In the argon plasma, the collisional energy losses are electronic excitation and ionization.
But for the water plasma, additional collisional energy losses include excitation of vibra-
tional and rotational energy levels, and molecular dissociation. Therefore, if an equal
amount of energy is deposited into the two discharges, there is less energy available for
electrons to create ion-electron pairs in the water plasma than in the argon plasma. Addi-
tionally, in the water plasma, electrons are also consumed in dissociative electron attach-
ment and three-body electron attachment processes to produce negative ions. Thus, the low
electron density in the water plasma compared to that in the argon plasma are expected
results.
In the water plasma, the magnetic field strength and RF power appreciably affect both
96
ion and electron densities. For the ion density, there is an almost linear increase with RF
power. Similarly, the electron density also increases with RF power in the water plasma.
The observed increase in densities with RF power is expected for the argon plasma and
the water plasma. As more energy is deposited into the discharge, the total ionization rate
increases with density (Kiz ∝ ne ng < σv >).
On the other hand, the electron and ion densities decrease with applied axial magnetic
field. For the water plasma, there is a negligible difference in ion densities for 0-A and
30-A applied magnet current settings, but a significant decrease from 30-A to 60-A applied
magnet current. The electron density clearly decreases with applied magnet current. These
trends can be explained by examining how RF power is coupled to the plasma source and
the effects of the additional axial magnetic field on the mobility of the electrons and ions.
As discussed in Section 3.3.3, the electric and magnetic fields can only penetrate into
the plasma within the skin depth layer in an inductive plasma discharge. The thickness
of this skin depth layer is inversely proportional to the square root of the electron density
(Equations 3.21 and 3.22). For a low electron density (e.g. 1015 m−3 ), the skin depth layer
thickness can be as large as 16 cm, which is the diameter of the discharge. On the other
hand, for a higher electron density (e.g. 1017 m−3 ), the skin depth thickness is only 1.7
cm. In a different previous experiment, at a lower pressure of a few mtorr and with argon
as the operating gas, the ion number density could be as high as 2.5 × 1019 m−3 , where
the helicon mode was detected [25]. It was also shown that the ion density ranges from
1 × 1018 m−3 to 5 × 1018 m−3 for the capacitive mode and from 1 × 1019 m−3 to 2 ×
1019 m−3 for the inductive mode. Even though these high densities are not observed in the
97
Figure 5.8: Photographs of water plasma operating at (a) 0-A and (b) 30-A applied magnet
current.
water plasma, the discharge is still expected to operate in either capactive or inductive mode
based on the behavior of the discharge and the calculated skin depth thickness. Figure 5.8
shows two photographs of the discharge, without and with an applied magnetic field.
As shown in the photographs in Figure 5.8, the tip of the LP probe is at a fixed r = 0 po-
sition (plasma source centerline) downstream of the discharge. The density measurements
presented here are local measurements–not line-integrated or volume-averaged densities.
Without the magnetic field (Figure 5.8a), the discharge appears diffused making the local
measurement more representative of a volume average. But if the electrons are confined
to the annular region near the wall of the discharge (Figure 5.8b), this density does not
represent the volume-averaged density.
The applied axial magnetic field affects the electrons in two ways. First, the magnetic
field enhances electron confinement in the discharge because the electrons gyrate along the
field lines. Second, the magnetic field reduces the mobility of the electron in the direc-
tion that is perpendicular to the magnetic field lines. The diffusion coefficient across the
magnetic field is [81]:
98
D
D⊥ = (5.1)
1 + ω2τ 2
where D is the diffusion coefficient in the absence of a magnetic field, ω is the cyclotron
frequency, and τ is the mean collision time. As an example, for a gas pressure of 300 mtorr,
electron temperature of 2 eV, and ion temperature of 0.05 eV, and a magnetic field of 400 G,
the diffusion coefficient across the magnetic field is reduced by 16000 times for electrons,
but only by a factor of 23 for ions.
The reduction in the ion and the electron number densities in the water plasma in the
presence of the magnetic field as shown in Figures 5.6 and 5.7 are then the results of a
combination of the following: First, the applied axial magnetic field enhances the electron
confinement, and the electrons created near the wall of the discharge cannot easily cross the
field lines. Therefore, the measurements made outside confinement zones may not reflect
the density increase. Additionally, because the electrons cannot easily diffuse into the bulk
of the discharge due to its reduced mobility in the perpendicular direction, the electron
density in the bulk discharge can actually be reduced. Second, the ions and neutrals are not
affected by the presence of the applied magnetic field. Further, some of the electrons that
diffuse into the bulk discharge are consumed through either three-body electron attachment
or dissociative attachment processes. The combination of these effects is the likely cause
of the trend shown in Figure 5.7, where electron density decreases with applied magnet
current for any RF power level. As a result of this reduction in electrons in the core of the
discharge, ion density is also reduced. Note that electrons are confined in the argon plasma
in the presence of a magnetic field, but they are not consumed in attachment processes in
99
the core of the discharge.
Bohm time gives an estimate of the electron diffusion time, and it is another parameter
used to determine the effect of the magnetic field to the mobility of the electrons. Equation
5.2 shows how Bohm time is calculated:
R2
τB = (5.2)
2DB
where τB , R, and DB are the Bohm time, radius of the cylinder, and Bohm diffusion
coefficient, respectively. Unlike the kinetic collisional diffusion coefficient, the Bohm dif-
fusion coefficient is independent of the density:
1 Te
DB = (5.3)
16 B
For Te of 2 eV and B of 400 G, the Bohm time was calculated to be 0.1 second. There-
fore, while the magnetic field confines the electrons, some of them do cross the field lines
and enter the core of the discharge, with the diffusion time of 0.1 second.
5.1.3 Electron Temperature
The electron temperature of the water and argon plasmas are compared. Figures 5.9
and 5.10 show the electron temperature for argon and water plasmas, respectively. For
the argon plasma, the electron temperature varies between 1.2 and 1.4 eV with error bars
between 0.2 and 0.3 eV (14% to 21% of the maximum value). The electron temperature is
100
Figure 5.9: Electron temperature of the argon plasma as a function of RF power for 0-A,
30-A, and 60-A applied magnet current.
not affected by RF power or applied magnet current as expected. In general, the electron
temperature is determined by particle conservation alone, and is independent of the plasma
density and thus the input power [52]. Therefore, electron temperature is independent of
RF power.
For the water plasma, the electron temperature has significantly larger error bars; and
it ranges from 2.7 to 3.9 eV with error bars between 0.7 and 2.2 eV (18% to 56% of the
maximum value). The large error bars are introduced in the process of smoothing the noisy
IV traces. Further, the water plasma in general was less stable than the argon plasma.
Due to these large error bars, no clear trend for electron temperature with RF power or
applied magnet current is observed. It has been reported that the collisional energy loss per
electron-ion pair created for a molecular gas is 2-10 times higher than an atomic gas when
the electron temperature is below 7 eV [52]. Thus, in order to sustain a discharge operating
101
Figure 5.10: Electron temperature of the water plasma as a function of RF power for 0-A,
30-A, and 60-A applied magnet current.
on a molecular gas, the electron temperature must necessarily be higher in order to achieve
adequate ionization, which may explain why the electron temperature for the water plasma
is higher than that of the argon plasma. In addition, it is also possible that for a similar
input power, the water plasma has a lower density, which results in a high capacitive field
and corresponding high electron temperature.
5.1.4 Floating and Plasma Potentials
The floating potential is the potential applied to the Langmuir probe to achieve equal
total positive and total negative currents to the probe (I + = I − ), where I + represents the
positive ion current and I − represents the electron current in an electropositive plasma or
the sum of electron and negative ion currents in the electronegative plasma. Figures 5.11
and 5.12 show floating and plasma potentials for the argon and water plasma, respectively.
102
Figure 5.11: Floating and plasma potential of the argon plasma as a function of RF power
for 0-A, 30-A, and 60-A applied magnet current.
Figure 5.12: Floating and plasma potential of the water plasma as a function of RF power
for 0-A, 30-A, and 60-A applied magnet current.
103
Floating and plasma potentials for the water plasma are higher than those of the argon
plasma. For the argon plasma, the floating potential ranges from 17 to 25 V with error bars
between 0.08 and 0.63 V. The floating potential increases with RF power, but is not strongly
affected by the applied magnet current except at high RF power (1000 W). The plasma
potential is consistently 4-5 V higher than the floating potential. For an electropositive
plasma, the electron temperature is directly related to the difference between the plasma
and floating potentials as shown in Equation 5.4. Therefore, the nearly constant difference
between the plasma and floating potentials indicates that the electron temperature for the
argon plasma is constant with RF power and magnetic field. This conclusion agrees with
the constant electron temperature.
µr ¶
Te πm
Vp − Vf = ln (5.4)
2 2M
Equation 5.4 is derived for an electropositive plasma by setting the electron current
equal to the ion current [52, 81]:
Ie = Ii (5.5)
In Equation 5.5, the electron and ion currents are:
µ ¶
1 VB − V p
Ie = ene v̄e Aexp (5.6)
4 Te
104
Ii = ens uB A (5.7)
where ns is the density at the sheath (ns ≈ 0.61no ), uB is the Bohm velocity, A is the
area of the probe, and VP is the plasma potential. VB is the bias voltage of the probe and
is equal to VF when the condition in Equation 5.5 is satisfied. The relationship between
plasma potential, floating potential, and electron temperature derived in Equation 5.4 is
only applicable for an electropositive plasma. For an electronegative plasma, Equation 5.5
becomes:
Ii+ = Ie + Ii− (5.8)
where Ie , Ii+ , Ii− are the electron current, the positive ion current, and the negative
ion current. Following a similar derivation for the electropositive plasma, Equation 5.9 is
established for the electronegative plasma [89].
µ ¶ µ ¶
1 VF − Vp 1 VF − Vp
uB (1 + αs ) = v̄e exp + αs ui− exp (5.9)
4 Te 4 Ti−
where αs = ni− /ne is the electronegativity. Without electronegative species (ni− = 0),
Equation 5.9 reduces to Equation 5.4. The first term on the right-hand side of Equation
5.9 represents the electron current and the second term represents the negative ion currents.
For a relatively low ion temperature, the second term on the right-hand side reduces to zero
due to the exponential function of a large negative number. The equivalence of Equation
105
5.4 for an electronegative plasma is approximated as:
Ã r !
1 Mi
Vp − Vf ≈ Te ln (5.10)
1 + αs 2πme
Equation 5.10 shows that the difference between the plasma and floating potential for
an electronegative plasma is only a weak function of the electronegativity.
In the water plasma, the difference between the plasma and the floating potential is up
to 10 V compared to 5 V in the case of the argon plasma. Higher electron temperature
offers an explanation for the larger difference between the plasma and floating potential
in the water plasma. For the water plasma, the effective ion mass is between 9 and 11
AMU. Note that the difference between the plasma and the floating potential is not a strong
function of electronegativity or the relative ion mass.
The results for the water plasma show that the floating potential ranges from 21 to 53
V with error bars ranging from 0.2 to 3 V (0.4% to 6% of the maximum value). With
the 60-A applied magnet current setting, the water plasma’s floating potential is twice that
of the argon plasma for RF power greater than 500 W. The exact physical phenomenon
that causes the floating potential to increase in the water plasma has not been identified.
However, one plausible explanation is as follows.
In addition to quasineutrality:
ni+ = ni− + ne , (5.11)
106
the water plasma must also maintain ambipolar diffusion:
Γi+ = Γi− + Γe (5.12)
where Γ is the flux of particles. At high magnetic field, the electrons are confined to
the production region near the wall of the discharge, but the negative and positive ions
are moving freely into the core of the discharge. In order to maintain quasineutrality, the
plasma potential then needs to increase to allow the electrons to flow to the core of the
discharge. Due to the nature of the electronegative plasma, many of the electrons that
make it to the core of the plasma are being consumed in attachment processes. The plasma
potential increases with magnetic field because the electrons are more confined to the wall
at higher magnetic field. Therefore, a higher plasma potential is needed in order to maintain
quasineutrality in the core. The floating potential increases as a result of Equation 5.10.
5.1.5 Summary
OES and RGA spectra are presented and they validate the presence of the species from
dissociated water molecules in the water vapor plasma discharge. An electropositive argon
plasma was characterized and compared with the electronegative water plasma. In an argon
plasma, the only collisional energy losses are ionization and excitation. On the other hand,
the water plasma’s collisional energy losses include excitation of vibrational and rotational
energy levels, and molecular dissociation. Further, electrons are also lost in electron attach-
ment and dissociative attachment processes. Therefore, the argon ion density (ni = ne ) is
higher than the electron density of the water plasma by up to two orders of magnitude. The
107
electron temperature of the water plasma contains large error bars (up to 50%) that result
from the error introduced when smoothing the IV traces. The average electron temperature
is shown to be higher than that of the argon plasma by 1.5 eV.
5.2 Effects of Applied Axial Magnetic Field
Section 5.1 compares the differences in plasma properties of the plasma discharge
operating on argon gas and on water vapor. This section investigates the effect of RF power
and magnetic field strength on hydrogen production in the water plasma discharge. In
addition, the electron and ion densities are also examined to offer physical explanation of
the observed trends in the hydrogen production rate.
5.2.1 Hydrogen Production
The hydrogen production rate is reported for 0-A, 30-A, and 60-A applied magnet cur-
rent settings as a function of RF power in Figures 5.13 and 5.14 for 75-sccm and 125-sccm
water vapor input flow rate, respectively. The hydrogen production rate is determined from
the method outlined in Section 4.2.1 via the RGA. Without an applied magnet current (0-A
setting), the hydrogen production rate increases linearly with RF power. For the 30-A and
60-A applied magnet current settings, the hydrogen production rate increases from 250-W
to 500-W RF power, but it saturates starting at 500 W. Despite this observed saturation as a
function of RF power, the hydrogen production rate is clearly higher with an applied axial
magnetic field than without it at all RF power levels. The effects of RF power and applied
magnet current on the hydrogen production rate are similar for both 75-sccm and 125-sccm
108
Figure 5.13: Hydrogen production rate as a function of RF power for 0-A, 30-A, and 60-A
applied magnet current and 75-sccm water input flow rate.
water input flow rate. In fact, these effects are also observed in all other water input flow
rates in this investigation.
The observed saturation of the hydrogen production rate with RF power in the pres-
ence of an applied axial magnetic field could be explained by examining the electron and
ion densities of the discharge. As shown in Figure 5.8 and discussed in the previous
section, the axial magnetic field confines electrons to the plasma production zones, which
include the annular volume adjacent to the wall of the quartz tube, closest to the antenna. In
this inductively-coupled plasma source, power is transferred from the electric fields to the
plasma electrons within a skin depth layer near the plasma surface by Ohmic and stochas-
tic heating in this annular region [52]. Therefore, much of the production of electrons is
within this layer, whose thickness varies inversely with the square root of the electron den-
sity (Equations 3.21 or 3.22). As a function of RF power, the electron density increases
109
Figure 5.14: Hydrogen production rate as a function of RF power for 0-A, 30-A, and 60-A
applied magnet current and 125-sccm water input flow rate.
with the skin depth thickness decreasing in response.
The magnetic field inhibits electrons from diffusing into the core of the discharge to
initiate water-electron impact reactions. Despite an increase in electron density with RF
power, electrons cannot easily cross the field lines. However, those electrons that are in
the core of the discharge have a better confinement in the presence of the magnetic field.
This improvement in electron confinement also leads to more electron attachment and dis-
sociative attachment. This explains the increase in hydrogen production when the magnetic
fields are present. However, with the magnetic field, hydrogen production saturates with
RF power. Again, electron density is proportional to RF power, but the skin depth thickness
is inversely proportional to the square root of the density. Therefore, even though the mag-
netic field enhances the electron production in the discharge, it also reduces the electrons’
rate of diffusion into the discharge to initiate electron-water impact reactions.
110
Consequently, for a constant applied magnet current, the magnetic field generates two
competing factors that affect the rate of hydrogen production. On the one hand, the mag-
netic field inhibits electrons that are primarily produced near the edge of the source from
entering the core of the discharge. On the other hand, the magnetic field enhances electron
confinement, thus increasing the probability of electron collisions. These two effects cancel
each other, and this explains why hydrogen production saturates with RF power when there
is a magnetic field. The fact that the rate of hydrogen production increases with applied
magnet current suggests that the second effect–enhanced electron confinement–is stronger
than the first effect at higher magnetic field strength. This behavior could also be due to
enhanced ionization in the heating zone, the annular region closest to the antenna.
5.2.2 Electron and Ion Number Density
The ion and electron densities for the water plasma with 75-sccm water input flow rate
are shown in Figures 5.6 and 5.7, respectively. The densities increase with RF power,
but decrease with applied magnet current. As more energy is deposited into the plasma,
electron production necessarily increases through various electron heating mechanisms in-
cluding Ohmic and stochastic heatings. However, the densities in the core decrease as a
function of magnet current. Note again that the results presented here are local measure-
ments at r = 0 location downstream of the discharge. These results of densities decreasing
with applied magnet current again suggest that the electrons are produced near the wall of
the quartz tube and radial diffusion rate is reduced due to the presence of the axial magnetic
field. This saturation effect is not seen as strongly with ions since they are not magnetized.
111
Rather they grow linearly with RF power. Presumably negative ions assist in maintaining
quasineutrality. Negative ions production is suggested by the saturation of electron density
but a linear growth in ion density. As shown previously, hydrogen production increases
with applied magnet current. Dissociative attachment is speculated to be the main dissoci-
ation mechanism in the water plasma discharge:
H2 O + e → OH − + H (5.13a)
H2 O + e → OH + H − (5.13b)
H2 O + e → O − + H2 (5.13c)
The first two mechanisms (Equations 5.1a and 5.1b) are favored over the last mecha-
nism (Equation 5.1c) because they require less energy [52].
5.2.3 Summary
In summary, the RGA results show that the hydrogen production rate increases with ap-
plied magnet current, but saturates with RF power after 500 W in most cases. Electron and
ion densities, however, decrease with applied magnet current, but increase with RF power.
In the presence of the magnetic field, electron confinement in the core of the discharge is
enhanced. At the same time, diffusion of electrons produced near the antenna to the bulk
discharge is limited because electrons cannot easily cross the field lines.
112
5.3 Effect of Water Input Flow Rate
Section 5.2 discusses the effect of the magnetic field on hydrogen production in the
water plasma discharge. This section investigates the effect of the water input flow rate on
the hydrogen production rate and the plasma properties of the water plasma discharge.
5.3.1 Hydrogen Production
To investigate the effects of water vapor input flow rate on the hydrogen production rate
and the plasma properties, the water flow rate into the system is set at 25, 50, 75, 100, and
125 sccm. Figures 5.15 and 5.16 show the hydrogen production rate for these input water
flow rates as a function of RF power for 0-A and 30-A magnet current setting, respectively.
As noted in Section 5.2, the rate of hydrogen production increases linearly with RF power
without an applied magnet current but saturates with RF power with an applied magnet
current albeit at higher RF power levels. Figures 5.15 and 5.16 illustrate the consistency
of these trends for all water input flow rates.
At 0-A applied magnet current and 250-W RF power, an abrupt increase in hydrogen
production is observed between 25-sccm and 50-sccm water vapor input flow rates and
between 75-sccm and 100-sccm water vapor input flow rates. For the other RF power
levels–500, 750, and 1000 W–there is a clear jump in hydrogen production from 25-sccm
to 50-sccm water vapor input flow rate, but no further increase is observed within the un-
certainty of the measurement. At 30-A applied magnet current, adding more water vapor
increases the rate of hydrogen production but this trend stops at 100-sccm water vapor
flow rate. Hydrogen production for the case of 125-sccm water vapor is lower than that of
113
Figure 5.15: The rate of hydrogen production as a function of RF power for 25, 50, 75, 100,
and 125-sccm water input flow rates, operating without an applied magnet
current.
Figure 5.16: The rate of hydrogen production as a function of RF power for 25, 50, 75,
100, and 125-sccm water input flow rates, operating with 30-A applied magnet
current.
114
100-sccm water vapor.
This may be explained as follows. For a given RF power level, a higher water vapor
input flow rate means the energy available per molecule is reduced. There is an optimum
energy per molecule to dissociate water molecules at a given flow rate (pressure). There-
fore, hydrogen production rate initially increases with water input flow rate up to the op-
timum energy per molecule. Further addition of water vapor reduces the energy available
per molecule to a value below the optimum value. Therefore, beyond 100 sccm, the energy
available per molecule is not consistent with a higher production rate, hence hydrogen pro-
duction rate decreases. Figure 5.17 shows the rate of hydrogen production as a function
of water input flow rate. As illustrated, at 100 sccm water input flow rate, the hydrogen
production is at a maximum for all RF power levels. It was expected that the maximum
peak would shift to the right of the plot (towards a higher flow rate) for a higher RF power.
However, this is not clearly demonstrated in Figure 5.17. The hydrogen production rate for
500, 750, and 1000-W RF power are nearly equal. For 500 W RF power and a hydrogen
production rate of 28 sccm, the cost of energy to produce one H2 is 266 eV. As a reference,
the first theoretical dissociation energy (H2 O → H + OH) for the water molecule is 5.17
eV per molecule and for the second (OH → H + O) is 4.52 eV per molecule. This suggests
that input energy is not efficiently going towards the dissociation of the water molecules,
and also any further production of hydrogen is lost through recombination processes as the
flow rate is increased.
115
Figure 5.17: The rate of hydrogen production as a function of water input flow rate for 250
– 1000 W
5.3.2 Production of Other Gases
In this section, the production of other gases in the plasma discharge operating on water
vapor is investigated. Figures 5.18, 5.19, and 5.20 show the rate of the un-dissociated
water vapor and the production rate of hydrogen, oxygen, and hydroxyl for 25, 75, and
125-sccm water vapor input flow rate obtained from the RGA, respectively. Note that one
sccm of H2 O carries one sccm of H2 but only one-half sccm of O2 .
At 25-sccm water vapor input flow rate, the dominant species in the plasma source is in
fact hydrogen, followed by water vapor, oxygen, and hydroxyl. However, at 75-sccm water
vapor input flow rate, there is a slight increase in hydrogen production, but the dominant
species in the plasma discharge is un-dissociated water molecule. This trend also continues
in the case of 125-sccm water vapor. In all these three cases, hydrogen production increases
116
Figure 5.18: Production of hydrogen, oxygen, and hydroxyl in a plasma discharge oper-
ating on water vapor with 25-sccm water input flow rate and 30-A applied
magnet current.
magnet current.
117
magnet current.
by 50%, from a maximum value of 20 sccm to 30 sccm for 25-sccm and 125-sccm water
vapor, respectively. Similarly, there is also a small increase in the production of oxygen as
a function of water input flow rate, with a maximum value of 8 sccm to 12 sccm. However,
there is a significant percent increase in hydroxyl production, from 3 sccm to 25 sccm for
25-sccm and 125-sccm water flow rates, respectively.
This suggests that the first two dissociative mechanisms (Equations 5.13a and 5.13b)
are favored over the last mechanism (Equation 5.13c). This result is consistent with the fact
that the first two mechanisms require less energy than the last mechanism. As the water
vapor flow rate increases, initially only one O-H bond in the water molecule is broken first
via dissociative attachment. Therefore, as more water is added, there is more OH available
in the plasma source. On the other hand, the atomic hydrogen from OH-H dissociation is
118
very mobile due to its low mass, and therefore may have very little opportunity to combine
with another atomic hydrogen to become a molecular hydrogen. But the atomic hydrogen
produced in the OH-H dissociation is very likely combine to the surfaces of the chamber
wall. Further, the fact that oxygen does not have a large percent increase with water flow
rate suggests that any possible further dissociation of OH into O and H is not favored. The
cross sections for these reactions are shown in Figure 5.3. The electron temperature is
observed to decrease with increasing water input flow rate. This trend would tend to reduce
reaction rates.
5.3.3 Electron and Ion Number Density
The electron density for varying water input flow rates is shown in Figures 5.21 and
5.22 as a function of RF power for 0-A and 30-A applied magnet current setting, respec-
tively. In general, the electron density increases with RF power for a fixed water flow rate
and decreases with water vapor input flow rate for a fixed RF power level. The increase in
electron density with RF power is expected because as more energy is deposited into the
discharge, the total ionization rate (∝ ng ne < σV >) increases through the increase in
electron density. As discussed previously, as more water vapor is introduced into the dis-
charge at a fixed RF power, the energy per molecule is reduced. This explains the reduction
in electron density with water input flow rate.
The ion density is shown in Figures 5.23 and 5.24 as a function of RF power for 0-A
and 30-A applied magnet current setting, respectively. Similar to electron density, the ion
density increases with RF power. This again is due to the increased total ionization rate
119
Figure 5.21: Electron density as a function of RF power for 25, 75, and 125-sccm water
input flow rates without an applied magnet current.
input flow rates with 30-A applied magnet current.
120
Figure 5.23: Ion density as a function of RF power for 25, 75, and 125-sccm water input
flow rates without an applied magnet current.
resulting from an increase in deposited energy. Also similar to electron density, the ion
density decreases with water input flow rates.
5.3.4 Optical Emission Spectra
Finally, optical emission spectra for variation in water vapor input flow rate are pre-
sented in Figures 5.25 and 5.26. The OH band is clearly identified in the UV range–281-
303 nm and 305-330 nm. Hα at 656.3 nm and Hβ at 486.1 nm are also observed. OH and
H lines decrease as a function of water vapor flow rate. In particular, the hydrogen line
at 656.3 nm significantly reduces from 25-sccm to 75-sccm water input flow rate. Both
the OH band and the H lines decrease as water vapor input flow rate increases, indicating
classic electron cooling. The trends showing a decrease in optical emission lines agree with
the observed reduction in electron temperature with the water input flow rate.
121
input flow rates with 30-A applied magnet current.
Figure 5.25: Optical emission spectra for 25, 75, and 125-sccm water input flow rates op-
erating at 500-W RF power without an applied magnet current.
122
Figure 5.26: Optical emission spectra for 25, 75, and 125-sccm water input flow rates op-
erating at 500-W RF power and 30-A applied magnet current.
5.3.5 Summary
It has been shown that at a fixed power, the water input flow rate affects hydrogen
production rate. There is an optimum energy per molecule for dissociation. At a fixed
RF power level, the rate of hydrogen production increases with water input flow rate up
to 100 sccm and begins to reduce at 125 sccm. Although hydrogen and oxygen increase
(by 50%) with water flow rate, the percent increase in hydroxyl is much higher in terms
of percentage (by 800%). The absolute value for hydroxyl is influenced by the lack of an
accurate calibration factor because hydroxyl was not used (see Section 4.2.1). Nevertheless,
the trend shown here suggests that dissociative attachment processes are responsible for
creating hydroxyl.
123
5.4 Conversion and Energy Efficiencies
The conversion and energy efficiencies to be discussed in this section are defined as:
ṁH2,produced
ηcon = (5.14)
ṁH2O,input
ṁH2 HHV H2
ηe = (5.15)
Pin
where ηcon is the conversion efficiency, ηe is the energy efficiency, HHV is the higher
heating value of hydrogen (141.1 MJ/kg), and Pin is the power deposited in W.
Figures 5.27 and 5.28 show the conversion efficiency for the case with and without
the applied magnetic field. The findings of this study show that the highest conversion
efficiency is obtained at a lower water input flow rate. For 25-sccm water input flow rate,
the conversion efficiency is 80%. The conversion efficiency decreases for a higher water
input flow rate. For 75-sccm water input flow rate, the conversion efficiency is 30%, and
for 125-sccm water input flow rate, the conversion efficiency reduces further to 22%.
While the conversion efficiency can be relatively high at a low water input flow rate, the
energy efficiency is relatively low. Figures 5.29 and 5.30 show the energy efficiency for
the case with and without the applied magnetic field. To convert sccm into kg/s, 1 sccm =
1.6 × 10−8 m3 /s, and for hydrogen at standard pressure (1 atm) and standard temperature
(293 K), the density is 0.083 kg/m3 . Applying Equation 5.15, the energy efficiency for the
case without and with magnetic field is shown in Figures 5.29 and 5.30. The maximum
124
Figure 5.27: Conversion efficiency for the case without an applied magnetic field
Figure 5.28: Conversion efficiency for the case with an applied magnetic field
125
Figure 5.29: Energy efficiency for the case without an applied magnetic field
energy efficiency is obtained for the case with the higher water flow rate (125 sccm) and
lowest RF power (250 W), and the efficiency is 1.7%. This is a very low value compared
to the energy efficiency of electrolysis, 40-70% [45].
5.5 Summary of Experimental Results
Argon and water plasmas are compared and the effects of magnetic field strength and
water vapor input flow rate on the hydrogen production rate are investigated in Sections
5.1–5.3. In this section, a summary of the results is given.
From the optical emission and residual gas analyzer spectra, it is shown that water
molecules are dissociated in the plasma discharge operating on water vapor. With matching
background pressure and operating conditions (RF power level and axial applied magnetic
field strength), plasma properties of the discharge operating on argon are quite different
126
Figure 5.30: Energy efficiency for the case with an applied magnetic field
from those operating on water vapor. In particular, the argon plasma density is higher than
that of the water plasma at similar RF power levels and background pressure. The ion
density of the argon plasma (ni ≈ ne by quasineutrality) is up to two orders of magnitude
higher than the electron density in the water discharge. The difference between the ion and
electron densities in the water plasma is due to the presence of species with high electron
affinity in the discharge.
The presence of the magnetic field is shown to affect the electron’s diffusion from the
production zones (in the annular region near the antenna) to the core of the discharge. As a
function of magnetic field, the ion and electron densities at r = 0 location of the discharge
reduce for both the argon and the water plasmas. However, for the water plasma, there
are additional electron loss mechanisms in the core of the discharge, mainly via three-body
collision attachment or dissociative attachment with species that have high positive electron
127
affinities.
A relationship is obtained for both the electronegative and the electropositive plasma
between the plasma potential, floating potential, and electron temperature. These relation-
ships are derived by setting the total positive current equal to the total negative current,
and they show that the difference between the plasma and the floating potential is only a
weak function of the electronegativity and the relative ion mass. Therefore, the difference
between the plasma and floating potential is more likely due to the increase in electron tem-
perature. The plasma potential (and therefore the floating potential) of the water plasma is
higher than that of the argon plasma. It is discussed that in order for the water plasma to
maintain quasineutrality when the electron density is low (on the order of 1015 m−3 com-
pared to the argon plasma’s ion density at 1017 m−3 ), the plasma potential of the water
plasma must be higher than that of the argon plasma in order to set up the electric field
needed to achieve ambipolar diffusion.
In general, when water flow rate increases, the hydroxyl production rate increases at
a higher rate in terms of percent increase than hydrogen or oxygen production. Electron
temperature decreases with increasing water flow rate, and OES spectra concurrently show
the reduction in H emission lines and OH bands, suggesting electron cooling.
Finally, the conversion efficiency as high as 80% is measured. Conversion efficiency
decreases as the water input flow rate increases. The maximum energy efficiency is rela-
tively low (1.7%), which is lower than what has been achieved in conventional electrolysis,
40-70%.
128
Chapter VI
Kinetic Simulation Setup and Results
This chapter describes the zero-dimensional kinetic model Global Kin that is used to
estimate the upper limit theoretical conversion and energy efficiencies. Following the de-
scription of the model, the reactions considered in this model are presented in Section
6.2. In Section 6.3, the computational results are given. The electron number density and
electron temperature are examined as a function of water input flow rate. A comparison
between the theoretical and experimental efficiencies is presented.
6.1 Reactions
This model considers 28 species and 283 gas phase and electron impact reactions (see
Appendix A). For the gas phase reactions, the production rate coefficients are given in
Arrhenius form as shown in Equation 6.1.
³ ´
−E A
b
k(Tg ) = AT e Ru Tg
(6.1)
where A, b, and EA are empirical parameters obtained from experiments for each gas
phase reaction. EA is the activation energy and Ru is the universal gas constant.
129
For the electron impact reactions, the reaction rate coefficients are obtained as shown
in Equation 6.2:
Z ∞ µ ¶0.5
2²
k(Te ) = f (²) σ(²)d² (6.2)
0 m
where ² is the electron energy and σ is the electron impact cross section.
In Global Kin, a collection of Arrhenius parameters and cross sections are available in
a database. Table 5.1 shows the reactions added to the data file to account for electron
impact reactions with water.
The cross sections for elastic and inelastic electron collisions with water are compiled
in a literature review by Itikawa and Mason in Reference 88. An interpolation tool is used
to obtain a continuous curve for the cross sections in order to perform the integration as
shown in Equation 6.2. The cross sections are shown in Figure 5.3.
6.2 Description of Global Kinetic Model
A global, zero-dimensional kinetic model called Global Kin is used to study the disso-
ciation of water molecules in a plasma source [90–93]. The plasma discharge is modeled
as a cylinder with radius R and length L as shown in Figure 6.1. In this simulation, the
water flow rate into the chamber and the gas pressure are kept constant for each simulation.
Power is deposited into the discharge as a function of time. It is set to ramp up from zero to
a pre-set level in 5 µs, and then it remains constant for the duration of the simulation that
is typically of a few seconds. Therefore, the power level is essentially constant with time.
130
Figure 6.1: Schematic diagram of plasma discharge corresponding to computational model.
The code outputs gas temperature, electron temperature, and species density as a function
of time.
The number density inside the chamber is dependent on both the gas temperature and
pressure, following the ideal gas law shown in Equation 6.3.
P = N kB Tgas (6.3)
Thus, when the gas temperature increases or when gas dissociation occurs, the gas must
expand to maintain a constant pressure. A description of the model provided here follows
the example of Stafford in Ref. 90.
In the zero-dimensional (0-D) kinetic model, Global Kin assumes a homogeneous plasma
where the concentration of species is spatially independent. Global Kin has three modules:
an offline Boltzmann solver module, a plasma chemistry and transport module, and an ordi-
nary differential equation (ODE) solver module. The scheme of the simulation is shown in
Figure 6.2. The code works in the following manner: The chemistry and transport module
constructs continuity equations for neutral and charged species. The time rate of change
of a neutral species as shown in Equation 6.4 is accounted for by diffusion to the wall,
diffusion from the wall, reaction sources and gas expansion. The time rate of change of
131
Figure 6.2: Simulation scheme of Global Kin.
charged species as shown in Equation 6.5 is a result of ambipolar diffusion and reaction
sources.
Ã !
dN i X Ni dT g
= −∇ · −∇ (Di Ni ) + ∇ · (Di Nj ) γj fij + Si − (6.4)
dt j
Tg dt
dN ±
i
= −∇ · [−∇ (Da,i Ni )] + Si (6.5)
dt
In Equations 6.4 and 6.5, Ni is the density of neutral species i, N± is the density for
charged species, γj is the wall reactive sticking coefficient of species j, Di and Da,i are
the diffusivity and the ambipolar diffusivity of species i in the mixture, fji is the returned
fraction of species j as species i from the wall, Si is the reaction source term for species
i, and Tg is the gas temperature. The ambipolar diffusion coefficients for charged species
are based on instantaneous ion and electron mobilities and diffusivities. The details of this
model are further described in References 90, 91.
One simplification can be made in a 0-D model by substituting the diffusion length Λ
for the partial derivatives is shown in Equation 6.6:
132
Ã !
dN i 1 X Ni dT g
= 2 −Di Ni + Di Nj γj fij + Si − (6.6)
dt Λ j
Tg dt
In Equation 6.6, ∇2 is approximated as 1/Λ2 .
The source term for the gas phase reactions and the electron impact reactions are de-
scribed as follows:
X¡ ¢ Y aLHS
Si = aRHS
ij − aLHS
ij kj Nl ij (6.7)
j l
In Equation 6.7, aij is the stoichiometric coefficients of species i in reaction j on the
right hand side and left hand side, and kj is the reaction rate coefficient for reaction j.
In the chemistry and transport module, the differential equations are set up to solve for
the gas and electron temperature. The energy conservation equation (Equation 6.8) for
the heavy species includes contribution to gas heating from elastic and inelastic collisions
with electrons, gas phase reaction sources, and conduction to the walls, and the transfer of
internal energy to kinetic energy as the gas expands and flow velocity increases.
X3 µ ¶ X
d 2me
(N cp Tg ) = ne νmi kB (Te − Tg ) + ne kj Nj ∆²j
dt i
2 Mi j
µ ¶ (6.8)
X κ d 1
− ∆Hj + 2 (Tw − Tg ) − Mw N vx2
j
Λ dt 2
In Equation 6.8, N is the total gas density, cp is the mixture averaged heat capacity, ne
is the electron number density, νmi is the momentum transfer collision frequency between
133
electrons and species i, me is the electron mass, Mi is the mass of species i, kB is the
Boltzmann constant, Te is the average electron temperature, kj and ∆²j are the rate constant
and the energy contribution from inelastic process j, ∆H j is the heat of reaction of process
j, κ is the thermal diffusivity, and Mw is the mixture averaged molecular weight.
For electron heating, the energy conservation (Equation 6.9) accounts for Joule heating
and energy transfer in elastic and inelastic collisions with heavy species:
µ ¶ X3 µ ¶ X
d 3 2me
n e k B Te = Pd − ne νmi kB (Te − Tg ) + ne kl Nl ∆²l (6.9)
dt 2 i
2 Mi l
In Equation 6.9, Pd is the time-averaged power deposition into the electrons over many
rf cycles. This model assumes that the power deposited into the system all goes into ac-
celerating the electrons. When the electrons collide into other particles, they impart some
of the energy into thermal energy–also known as Joule heating. Note that this model does
not account for other types of heating such as stochastic heating or any wave-particle inter-
action. However, there is not a global kinetic model of plasma chemistry that can account
for other modes of heating. Results from Global Kin should still be a good first order of
magnitude estimate when accounting for global kinetics.
As discussed in Section 6.1, the reaction rate coefficients for electron impact reactions
require an integral of the EEDF and the cross sections (Equation 6.2). The cross sections
are obtained from the literature. For the EEDF, it is obtained from the offline Boltzmann
solver. The Boltzmann module solves the Boltzmann equation (Equation 6.10) for the
134
EEDF.
µ ¶
∂f ~ F~ ∂f
+ V · ∇~r f + · ∇~v f = (6.10)
∂t m ∂t col
A table is generated for the average electron energy, transport coefficients, and the rate
coefficient constants, k, for a range of electric field/density (E/N ). The EEDF changes
with the composition of the gas mixture, and therefore the Boltzmann solver is invoked
at specified intervals during the simulation to account for this change. After the transport
coefficient and rate coefficients are calculated and tabulated for a range of E/N, the ODE
equations (Equations 6.4 - 6.9) are integrated using the stiff ODE solver developed at
Lawrence Livermore National Laboratory [94].
6.3 Simulation Results
This section presents the simulation results. First, the model conditions are matched
with those in the experiment: RF power is varied between 500 and 1000 W for a range of
water vapor flow rates (25, 50, 75, 100, and 125 sccm). The operating pressure is set to
match those measured in the experiment. A study of energy efficiency is then carried out
independently as a function of background pressure and water vapor input flow rate. Note
that this simulation does not account for any magnetic field effect; the plasma is assumed
to be isotropic.
135
(a) Computational results. (b) Experimental results.
Figure 6.3: (a) Computational and (b) experimental results for electron density as a func-
without an applied magnet current.
6.3.1 Effects of Flow Rate on Electron Number Density
Figure 6.3(a) shows the calculated electron density as a function of RF power for 25,
50, 75, 100, and 125-sccm water input flow rate. The electron density is shown to range
from 1 × 1015 to 35 × 1015 m−3 , within the same order of magnitude as that determined
in the experiment, reproduced in Figure 6.3(b). The results clearly show that the electron
density increases with RF power and decreases with water input flow rate. Again, at a fixed
power level, a decrease in flow rate increases the energy available per molecule, therefore
causing more dissociation and ionization. This agrees qualitatively with the experimental
results.
136
Figure 6.4: (a) Computational and (b) experimental results for electron temperature as a
function of RF power for 25, 50, 75, 100, and 125-sccm water input flow rates
6.3.2 Effects of Flow Rate on Electron Temperature
Figure 6.4(a) shows the electron temperature for the same set of operating conditions as
described in Section 6.3.1. The electron temperature ranges from 2 to 3 eV, which closely
matches the experimental results, reproduced from Chapter 5 in Figure 6.4(b). Similar to
the experimental results, the electron temperature is not affected by RF power. The result
for 125 sccm at 500-W RF power is high relative to the other conditions because at low
RF power and high water flow rate, the electron temperature needs to be relatively high in
order to have adequate ionization to sustain the discharge.
137
Figure 6.5: (a) Computational and (b) experimental results for energy efficiency as a func-
6.3.3 Effects of Flow Rate on Energy and Conversion Efficiencies
Figure 6.5(a) shows the energy efficiency as defined in Equation 5.15. It illustrates
that the energy efficiency increases with water vapor input flow rate and decreases with RF
power. Again, the background pressure for each corresponding flow rate is set to match the
experimental value. In general, the energy efficiency decreases with increasing fractional
dissociation as power goes into producing fragments. Therefore, the trend observed here
is physical. It also suggests that in order to increase the energy efficiency of hydrogen
production in a plasma discharge, throughput is an important parameter. For 125 sccm
water flow rate, the maximum energy efficiency is 7.5% if 250-W RF power is applied.
Figure 6.5(b) shows the corresponding energy efficiency obtained experimentally. The
trend is similar to that observed in the simulation results: the energy efficiency increases
138
with flow rate and decreases with RF power. The maximum experimental energy efficiency
for 125 sccm flow rate at 250 W is 1.2%, which is relatively lower than what is predicted
by the simulation. The experimental energy efficiency is lower than the computational en-
ergy efficiency mainly because the experimental efficiency is determined using the results
from the RGA, and there are loss mechanisms in the hydrogen gas as it traveled from the
production zone (in the plasma discharge) to the detector (RGA chamber). For example,
hydrogen can easily be absorped to the metal surfaces of the chamber. Therefore, it is
expected that the rate of hydrogen production is higher than what the RGA measured. Fur-
ther, the simulation did not account for all the possible inherit loss mechanisms (e.g. wall
loss), therefore, the simulation result only gives an upper limit value. In reality, the energy
efficiency is a value between what was measured experimentally and what was calculated
computationally.
The theoretical conversion efficiency for 125 sccm is 80%. The conversion efficiency
obtained from the experiment is 24% for 125 sccm water flow rate and 80% for 25 sccm
water flow rate.
6.3.4 Effects of Background Pressure on Energy Efficiency
The computational results presented thus far are for conditions that match those avail-
able in the experiment. And they show that the maximum energy efficiency is 7.5%. Next,
the operating pressure and the water input flow rate are independently studied to determine
their effects on the energy efficiency.
Figures 6.6 shows the energy efficiency as a function of water input flow rate for 50-500
139
Figure 6.6: Energy efficiency for operation with 250-W RF power as a function of water
vapor input flow rate for 50, 100, 250, and 500 mtorr.
mtorr pressure range and for 250-W RF power. The energy efficiency is shown to increase
with pressure. At higher pressure, diffusion losses for electrons and ions are lowered,
and therefore, the plasma is produced more efficiently. It is also shown that the energy
efficiency increases with water input flow rate, similar to Figure 6.5(a). Finally, these
results show that the energy efficiency is optimized if the plasma discharge can operate at a
significantly higher water input flow rate and a higher pressure. The theoretical conversion
efficiency at 1000 sccm water input flow rate is 25%. A dielectric barrier discharge or
plasma torch may provide higher energy efficiencies than the current system employed in
the experimental investigation.
140
6.4 Summary
In summary, when the operating conditions match those in the experimental investi-
gation, the simulated electron temperature closely matches the experimental results. The
simulated electron density is greater than experimental results by a factor between two and
four. However, it is shown that in order to increase the energy efficiency, the ability to
operate and sustain the plasma discharge at higher flow rates is needed. In a large range of
background pressure (50-500 mtorr), the energy efficiency increases with water input flow
rate. Thus, the water input flow rate is a critical parameter to optimize the energy efficiency
in this method of hydrogen production.
141
Chapter VII
Conclusions and Future Work
7.1 Conclusions
The aim of this dissertation research project was to evaluate the feasibility of hydro-
gen production through a method of dissociating water vapor in a radio-frequency plasma
source and to study the basic behaviors in a water plasma discharge. The expected shortage
of future energy resources and climate change have motivated development of innovative
techniques to satisfy energy demand while minimizing emissions. To this end, hydrogen as
an alternative non-carbon transportation energy source is an attractive option. The technical
challenge, however, is to extract hydrogen from water at a low cost and a high efficiency.
This work investigated the feasibility of producing hydrogen in a radio-frequency plasma
discharge operating on water vapor. Additionally, water plasma has many applications in
other research areas. The information on basic behaviors of a water plasma discharge will
provide fruitful insights for other researchers. The following summarizes works that have
be done and outlines the main conclusions.
142
7.1.1 Water Plasma Source Development
To carry out this investigation, a radio-frequency plasma discharge was designed to
operate on water vapor. A water delivery system was constructed to allow for an accurate
measurement and control of the water input flow rate. To characterize the discharge, a
Langmuir probe and optical emission spectrometers were used. In addition, a RGA was
calibrated to estimate the rate of hydrogen production. In order to operate the RGA, a
differential pump system was designed and constructed.
7.1.2 Electronegative and Electropositive Plasmas Comparison
The water plasma has been identified as an electronegative plasma due to the presence
of species that have high positive electron affinities such as O2 and OH. To understand
how the water plasma behaves differently from an electropositive plasma, a comparison of
plasma properties of the water and argon plasmas is made. The argon and water plasmas
operated at the same background pressures and operating conditions and the following
summarizes the comparison:
• Based on the LP results obtained at the r = 0 location downstream of the discharge,
the argon plasma’s ion density is higher than the ion density (by up to one order of
magnitude) and the electron density (up to two orders of magnitude) of the water
plasma.
• The electron temperature of the argon plasma is one eV higher than that of the water
plasma.
143
• The floating and plasma potential of the argon plasma is lower than those of the water
plasma.
7.1.3 Water Plasma Properties and Hydrogen Production Characterization
The plasma discharge was operated on water vapor and through the optical emission
and RGA spectra, it was confirmed that dissociation of water molecules in the discharge
was achieved. The effects of RF power, magnetic field strength, and water input flow rate
on the rate of hydrogen production and plasma properties were investigated.
• In the absence of the applied magnetic field, the hydrogen production rate increases
linearly with RF power.
• In the presence of the axial magnetic field, the hydrogen production rate increases
with RF power from 250 W to 500 W, but begins to saturate in most cases after 500
W.
• At a fixed RF power level, the rate of hydrogen production initially increases with
water vapor input flow rate up to 100 sccm, but begins to decrease after that.
• As the water vapor input flow rate increases, the rate of hydrogen and oxygen pro-
duction increases (up to 50%), and the rate of hydroxyl production increases more
significantly in terms of percentage (up to 800%).
• Local measurements at the axis of symmetry for the plasma discharge (r = 0 location)
downstream of the discharge show that electron density increases with RF power but
144
decreases with applied axial magnetic field for a fixed water input flow rate. For a
fixed power level, the electron density decreases with water input flow rate.
• The floating and plasma potential increases with applied magnetic field and with
water input flow rate.
• The electron temperature is not affected by the RF power or the applied axial mag-
netic field.
• The difference between the plasma potential and the floating potential is a weak
function of the electronegativity and the relative ion mass in the water plasma.
7.1.4 Water Plasma Behaviors
For the main results summarized in Sections 7.1.2 and 7.1.3, the following offers some
explanations for the observed trends and describes the behavior of the water plasma.
In the argon plasma, collisional energy losses include electronic excitation and ion-
ization. For the water plasma, these also include excitation of rotational and vibrational
energy levels, and molecular dissociation. Therefore, the available energy-per-molecule in
a water plasma is lower than that of the argon plasma. This explains why the water plasma
has lower plasma densities. In addition, in the presence of the applied axial magnetic field,
the positive and negative ions are free to diffuse from the production zone in the annular
region near the antenna into the core of the discharge. However, electrons are confined to
the production region because their mobility in the direction perpendicular to the field lines
is significantly reduced. For those electrons that are able to diffuse into the discharge, they
145
are likely to be consumed through three-body electron attachment or dissociative attach-
ment processes in the core, due to the presence of species with high electron affinities in
that region.
The plasma potential of the water plasma is higher than that of the argon plasma, and
this is possibly due to the water plasma having a lower electron density. Quasineutrality
is maintained even in the water plasma. The positive ion density is balanced with the sum
of the negative ion density and electron density. Because the electron density is low in the
water plasma, the plasma potential must increase to set up the electric field needed to bring
the electrons to the core to maintain quasineutrality.
In the absence of the applied axial magnetic field, for a fixed water input flow rate, the
increase in RF power effectively increases the energy available per molecule. Therefore,
it is seen that the rate of hydrogen production increases linearly with RF power. However,
in the presence of the applied axial magnetic field, the electrons cannot easily diffuse into
the core of the discharge because their radial mobility coefficient is significantly reduced.
Therefore, electron-impact reactions with water molecules are limited to the annular region
near the antenna. As RF power increases, the electron density increases, but in the presence
of the magnetic field, the electrons are confined to the small annular region. Consequently,
electron-impact reactions with water molecules in the core of the discharge are reduced,
which may explain the saturation of the hydrogen production rate with RF power observed
when there is an applied axial magnetic field.
146
7.1.5 Computational Work
The experimental work offered insights on the properties and behaviors of a plasma
discharge operating on water. Based on the current setup, the energy efficiency for hy-
drogen production is less than 1.7%. However, the experimental setup is not optimized
for achieving high efficiency. In particular, in the current experimental setup, one can-
not independently control the water input flow rate and the background pressure. This
and other limitations, along with the desire to estimate the theoretical efficiency of hydro-
gen production using a plasma discharge–where the dissociation is mostly dominant by
electron-impact reactions–motivated the computational investigation.
Using a zero-dimensional global kinetics model Global Kin, it was shown that the
trends in electron density and temperature from the computational results closely match
those in the experimental work when operating conditions are similar. The computational
investigation shows that energy efficiency is optimized by increasing water input flow rate.
Hence, the study of the effect of water input flow rate–in the ranges outside those available
in the experiment–along with operating pressures was performed. It was determined that
the energy efficiency is highest when water input flow rate and pressure are the highest.
The maximum theoretical energy efficiency achieved was almost 20% with a flow rate of
1000 sccm, and the maximum theoretical conversion efficiency achieved was 80% with a
flow rate of 125 sccm.
In conclusion, this experimental work shows that although it is feasible to dissociate wa-
ter molecules for hydrogen production via electron impact reactions, the energy efficiency
is too low to make this an economic choice compared with the current technologies through
147
electrolysis. However, through this work, properties and behaviors of a water plasma dis-
charge can provide helpful insights for future research which may take advantage of some
of the properties of the water plasma.
7.2 Suggestions for Future Work
This dissertation work has shown that hydrogen production in an RF plasma source is
feasible, but the current system is not optimized for energy efficiency. Both experimental
and computational results point to the need to increase the water input flow rate. The
following outlines suggestions for future work, for both the current plasma source setup
and ideas for the development of a different system.
7.2.1 Current Experimental Setup
To further this current work to understand the behavior of the water plasma in a low-
pressure regime, a radial profile of electron density inside the discharge is valuable. How-
ever, because the electron density is too high in the annular region closest to the antenna, a
Langmuir probe is not recommended. In addition, the presence of the physical probe would
affect the measurement. Therefore, alternative methods of measuring electron density
through the method of interferometry or other non-intrusive methods are recommended.
Further, to increase the energy efficiency, one can also study the effect of a seed gas on
the dissociation rate. A metastable atom can help in the dissociation process of the water
molecules in the discharge. If a helicon-type plasma source is used, a concentric geometry
with the gap distance equal the skin depth layer thickness is recommended. This geometry
148
would maximize the heating zone.
The existing plasma discharge setup can only operate at relatively low pressures, be-
low 500 mtorr. However, the current vacuum pump is limited to only 35 m3 /hr on N2 .
Therefore, a different pumping system with a higher pumping speed will allow one to ex-
perimentally investigate the regime where the flow rates are high but the pressures are low.
Finally, the hydrogen and oxygen gas produced in the existing system are being pumped
out of the system through a vacuum pump. However, a design of the hydrogen-oxygen
separation system is needed.
7.2.2 New System Development
The previous section suggests some future work that can be done to the current system
to provide more information on the plasma properties or to allow the existing system to op-
erate in the different regimes. However, as both experimental and theoretical results have
shown, in order to increase the energy efficiency of this method, the water input flow rates
must drastically increase. As such, operating the water plasma source at higher pressures
are desirable. Therefore, the current design of the low-pressure plasma source is not recom-
mended. Therefore, an entirely new system is suggested. For example, dielectric barrier
discharges or some types of plasma torch are recommended because these are known to
behave very well at high pressures and high flow rates. In addition, for reasons that are
to be discussed in the next section, a removal of the entire pumping system (atmospheric
operation) is desired.
149
7.2.3 Other Applications
This work began by examining the feasibility of breaking up the O-H bonds in the
water molecules. The motivation is to produce hydrogen to address the energy and climate
challenges, and to provide basic characterization of the water plasma discharges. However,
in its current stage, the conversion and energy efficiencies are relatively low compared to
what can be obtained from electrolysis that the proposed method cannot be competitive
with electrolysis, despite some disadvantages in electrolysis.
Even though the energy efficiencies are still an issue for hydrogen production, this work
reveals that dissociation of other molecules via electron-impact is possible and can be of
benefits in other applications. In particular, work has begun at PEPL to investigate the
feasibility of dissociating carbon dioxide gases in the plasma discharge [95]. Similar to
the application of hydrogen production, the ability to operate with high flow rate and high
pressure is recommended for this application of CO2 dissociation.
Moreover, this experimental work has demonstrated that the RF plasma source can
break up water molecules to produce highly useful radicals such as O and OH. Therefore,
if the cost of the overall system can be reduced (e.g. removing the vacuum pump system),
the water plasma discharge can be used in medical applications: i.e. sterilization of medical
instruments.
150
Appendix A
GlobalKin Data File
151
152
153
154
155
156
157
Bibliography
[1] IPCC, Climate Change 2007: The Physical Science Basis. Contribution of Working
Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate
Change. Cambridge University Press, 2007. IPCC-2007: WG1-AR4.
[2] Energy Information Administration, “International energy outlook 2007,” Tech.
Rep. DOE/EIA-0484(2007), Energy Information Administration, Office of Integrated
Analysis and Forecasting, U.S. Department of Energy, 2007.
[3] US Bureau of the Census, “Total midyear population of the world: 1950-2050,” tech.
rep., US Bureau of the Census, 2003.
[4] U. S. Department of Energy, “Hydrogen production.” http://www1.eere.
energy.gov/hydrogenandfuelcells/production/natural_gas.
html, July 2009.
[5] J. H. Chaffin, S. M. Bobbio, H. I. Inyang, and L. Kaanagbara, “Hydrogen production
by plasma electrolysis,” Journal of Energy Engineering, pp. 104–108, 2006.
[6] Z. C. Yan, C. Li, and W. H. Lin, “Hydrogen generation by glow discharge plasma
electrolysis of methanol solution,” International Journal of Hydrogen Energy, vol. 34,
2008.
[7] H. Kabashima, H. Einaga, and S. Futamura, “Hydrogen generation from water,
methane, and methanol with nonthermal plasma,” IEEE Transactions on Industry Ap-
plications, vol. 39, no. 2, 2003.
[8] J. Scherzer and A. J. Gruia, Hydrocracking Science and Technology. Marcel Dekker,
Inc., 1996.
[9] E. Worrell, D. Phylipsen, D. Einstein, and N. Martin, “Energy use and energy in-
tensity of the u.s. chemical industry,” Tech. Rep. LBNL-44314, Energy Analysis De-
partment, Environmental Energy Technologies Division, Lawrence Berkeley National
Laboratory, 2000.
[10] E. S. Jang, M. Y. Jung, and D. B. Min, “Hydrogenation of low trans and high con-
jugated fatty acids,” Comprehensive Revies in Food Science and Food Safety, vol. 1,
pp. 22–30, 2005.
[11] J. S. Clements, M. Sato, and R. H. Davis, “Preliminary investigation of prebreakdown
phenomena and chemical reactions using a pulsed high-voltage discharge in water,”
IEEE Transactions on Industry Applications, vol. IA-23, no. 2, pp. 224 – 35, 1987.
158
[12] A. T. Sugiarto and M. Sato, “Pulsed plasma processing of organic compounds in
aqueous solution,” Thin Solid Films, vol. 386, no. 2, pp. 295–299, 2001.
[13] T. Maehara, H. Toyota, M. Kuramoto, A. Iwamae, A. Tadokoro, S. Mukasa, H. Ya-

mashita, A. Kawashima, and S. Nomura, “Radio frequency plasma in water,”
Japanese Journal of Applied Physics, vol. 45, no. 11, pp. 8864–8868, 2006.
[14] A. A. Ivannikov, V. M. Lelevkin, A. V. Tokarev, and V. A. Yudanov, “Atmospheric-

presure glow discharge treatment of water,” High Energy Chemistry, vol. 37, no. 2,
pp. 115–118, 2003.
[15] A. V. Samokhin, N. V. Alekseev, T. F. Korovkina, E. V. Troitskaya, and Y. V. Tsvetkov,

“Oxidizing purification of water using thermal plasma,” Theoretical Foundations of
Chemical Engineering, vol. 41, no. 5, pp. 613–619, 2007.
[16] R. A. Rudder, G. C. Hudson, J. B. Posthill, R. E. Thomas, R. C. Hendry, and D. P.

Malta, “Chemical vapor deposition of diamond films from water vapor rf-plasma dis-
charges,” Appl. Phys. Lett., vol. 60, no. 3, pp. 329–331, 1992.
[17] R. K. Singh, D. Gilbert, R. Tellshow, P. H. Holloway, R. Ochoa, J. H. Simmons, and

R. Koba, “Low-pressure, low-temperature, and remote-plasma deposition of diamond
thin films from water-methanol mixtures,” Appl. Phys. Lett., vol. 61, no. 24, pp. 2863–
2865, 1992.
[18] B. Sun, M. Sato, and J. S. Clements, “Oxidative processes occurring when pulsed
high voltage discharges degrade phenol in aqueous solution,” Environmental Science
and Technology, vol. 34, no. 3, pp. 509 – 513, 2000.
[19] B. Sun, M. Sato, and J. S. Clements, “Use of a pulsed high-voltage discharge for
removal of organic compounds in aqueous solution,” Journal of Physics D: Applied
Physics, vol. 32, no. 15, pp. 1908 – 1915, 1999.
[20] J. S. Oh, K. Kawamura, B. K. Pramanik, and A. Hatta, “Investigation of water-vapor

plasma excited by microwaves as ultraviolet light source,” IEEE Transactions on
Plamsa Science, vol. 37, no. 1, pp. 107–112, 2009.
[21] G. Fridman, G. Friedman, A. Gutsol, A. Shekhter, V. N. Vasilets, and A. Fridman,

“Applied plasma medicine,” Plasma Processes and Polymers, vol. 5, pp. 503–533,
2008.
[22] M. M. Balkey, Optimization of a Helicon Plasma Source for Maximum Density with
Minimal Ion Heating. PhD thesis, West Virginia University, 1999.
[23] R. W. Boswell and D. Vender, “An experimental study of breakdown in a pulsed heli-
con plasma,” Plasma Sources Science and Technology, vol. 25, pp. 534–540, 1995.
[24] O. Sutherland, C. Charles, N. Plihon, and R. Boswell, “Experimental evidence of

a double layer in a large volume helicon reactor,” Physical Review Letters, vol. 95,
2005.
159
[25] K. M. Lemmer, Use of a Helicon Source for Development of a Re-Entry Blackout
Amelioration System. PhD thesis, University of Michigan, 2009.
[26] British Petroleum, “Quantifying energy - bp statistical review of world energy june
2006,” tech. rep., British Petroleum, 2006.
[27] University Corporation for Atmospheric Research, “Windows to the universe -
earth statistics.” http://www.windows.ucar.edu/tour/link=/earth/
statistics.html, July 2009.
[28] NASA Jet Propulsion Laboratory, “Solar system - venus.” http://www.jpl.
nasa.gov/solar_system/planets/mars_index.html, July 2009.
[29] NASA Jet Propulsion Laboratory, “Solar system - venus.” http://www.jpl.
nasa.gov/solar_system/planets/venus_index.html, July 2009.
[30] J. T. Kiehl and K. E. Trenberth, “Earth’s annual global mean energy budget,” Bulletin
of the American Meteorological Society, vol. 78, no. 2, pp. 197 – 208, 1997.
[31] T. Dowling and A. P. Showman, “Earth as a planet: Atmosphere and oceans,” in En-
cyclopedia of the solar system (L.-A. McFadden, P. R. Weissman, and T. V. Johnson,
eds.), pp. 169–187, Academic Press, second ed., 2007.
[32] G. Marland, T. A. Boden, and R. J. Andres, “Global, regional, and national co2 emis-
sions. in trends: A compendium of data on global change,” tech. rep., Carbon Diox-
ide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department
of Energy, Oak Ridge, Tenn., U.S.A., 2008.
[33] S. R. Turns, An Introduction to Combustion. McGraw Hill, second ed., 2000.
[34] L. D. S. Yadav, Organic Spectroscopy. Springer, 2005.
[35] Energy Information Adminsitration, “Annual energy review 2007,” Tech. Rep.
DOE/EIA-0384(2007), Energy Information Administration, Office Energy Statistics
from the US Government, June 2008.
[36] Energy Information Administration, “Emissions of greenhouse gases report,” Tech.
Rep. DOE/EIA-0573(2006), Energy Information Administration, Office Energy
Statistics from the US Government, 2007.
[37] Energy Information Administration, “Renewable energy consumption and electricity
preliminary 2007 statistics,” tech. rep., International Information Adminstration, Of-
fice of Coal, Nuclear, Electric and Alternate Fuels, U.S. Department of Energy, 2008.
[38] D. Dorling, M. Newman, and A. Barford, The Atlas of the Real World. Thames and
Hudson, 2008.
[39] Energy Information Administration, “Annual energy review 2008,” Tech. Rep.
DOE/EIA-0384(2008), Energy Information Administration, Office of Integrated
Analysis and Forecasting, U.S. Department of Energy, 2009.
160
[40] “Worldwide look at reserves and production,” Oil and Gas Journal, vol. 103, no. 47,
pp. 24–25, 2005.
[41] J. Deutch, J. R. Schlesinger, D. G. Victor, G. T. Allison, N. R. Augustine, R. A. Belfer,

S. W. Bosworth, H. L. Croft, C. J. DiBona, and J. P. Einhorn, “National security
consequences of u.s. oil dependency,” tech. rep., Council on Foreign Relations, 2006.
[42] Energy Information Adminstration, “The impact of increased use of hydrogen

on petroleum consumption and carbon dioxide emissions,” Tech. Rep. SR/OIAF-
CNEAF/2008-04, Energy Information Administration, Office Energy Statistics from
the US Government, August 2008.
[43] B. Ewan and R. Allen, “A figure of merit assessment of the routes to hydrogen,”
International Journal of Hydrogen Energy, vol. 30, pp. 809–819, 2005.
[44] J. Ivy, “Summary of electrolytic hydrogen production,” Tech. Rep. NREL/MP-560-

36734, National Renewable Energy Laboratory, September 2004.
[45] J. Holladay, J. Hu, D. King, and Y. Wang, “An overview of hydrogen production,”
Catalysis Today, vol. 139, pp. 244–260, 2009.
[46] K. Pant and R. B. Gupta, “Fundamentals and use of hydrogen as a fuel,” in Hydrogen
Fuel: Production, Transport, and Storage (R. B. Gupta, ed.), pp. 4–29, CRC Press,
2009.
[47] N. Muradov, “Production of hydrogen from hydrocarbons,” in Hydrogen Fuel: Pro-

duction, Transport, and Storage (R. B. Gupta, ed.), pp. 34–86, CRC Press, 2009.
[48] K. McHugh, S. Eisele, and J. Nestell, “Hydrogen production methods,” Tech. Rep.
MPR-WP-0001, MPR Associates, Inc., 2005.
[49] K. Cox and J. Wiliamson, Hydrogen: its technology and implications. CRC Press,
1977.
[50] F. Barbir, “Pem electrolysis for production of hydrogen from renewable energy
sources,” Solar Energy, vol. 78, pp. 661–669, 2005.
[51] A. D. Little, “Guidance for transportation technologies,” Tech. Rep. 35340-00, Arthur
D. Little, Inc., February 2002.
[52] M. A. Lieberman and A. J. Lichtenberg, Principles of Plasma Discharges and Mate-

rials Processing. John Wiley and Sons, Inc., 2005.
[53] A. Fridman, Plasma Chemistry. Cambridge University Press, 2008.
[54] A. Fridman and L. A. Kennedy, Plasma Physics and Engineering. Taylor and Francis
Books, Inc., 2004.
[55] P. Atkins and J. de Paula, Elements of Physical Chemistry. W. H. Freeman, fifth ed.,
2009.
161
[56] M. Silberberg, Chemistry: The Molecular Nature of Matter and Change. McGraw
Hill, second ed., 2000.
[57] A. Engel, Ionized Gases. Oxford University Press, 1955.
[58] T. D. Mark and G. H. Dunn, Electron Impact Ionization. Springer-Verlag, 1985.
[59] Y. P. Raizer, Gas Discharge Physics. Springer-Verlag, 1991.
[60] R. Goldston and P. Rutherford, Introduction to plasma physics. CRC Press, 1995.
[61] Y. Itikawa, Molecular Processes in Plasmas. Springer, 2007.
[62] R. W. Boswell, A Study of Waves in Gaseous Plasmas. PhD thesis, Flinders University
of South Australia, 1974.
[63] R. W. Boswell, “Plasma production using a standing helicon wave,” Physics Letters
A, vol. 33A, no. 7, pp. 457–8, 1970.
[64] R. W. Boswell, “Plasma ionization by helicon waves,” Plasma Physics and Controlled
Fusion, vol. 33, no. 4, pp. 339–364, 1991.
[65] R. W. Boswell and F. F. Chen, “Helicons - the early years,” IEEE Transactions on
Plasma Science, vol. 25, 1997.
[66] F. F. Chen and R. W. Boswell, “Helicons - the past decade,” IEEE Transactions on
Plasma Science, vol. 25, 1997.
[67] D. D. Blackwell, T. G. Madziwa, D. Arnush, and F. F. Chen, “Evidence for trivelpiece-

gould modes in a helicon discharge,” Physical Review Letters, vol. 88, 2002.
[68] F. F. Chen, “Experiments on helicon plasma sources,” Journal of Vacuum Science and
Technology A, vol. 10, pp. 1389–1401, 1992.
[69] A. W. Degeling, C. O. Jung, R. W. Boswell, and A. R. Ellingboe, “Plasma production

from helicon waves,” Physics of Plasmas, vol. 3, pp. 2788–2796, 1996.
[70] L. Porte, S. M. Yun, D. Arnush, and F. F. Chen, “Superiority of half-wavelength

helicon antenna,” Plasma Sources Science and Technology, vol. 12, pp. 287–293,
2003.
[71] M. Kong, M. Laroussi, and A. Fridman, “Plasma medicine - opportunities and chal-
lenges for science and healthcare.” Conference presentation at 36th International Con-
ference on Plasma Science in San Diego, CA, June 2009.
[72] L. Bromberg, D. Cohn, A. Rabinovich, and N. Alexeev, “Plasma catalytic reform-

ing of methane,” International Journal of Hydrogen Energy, vol. 24, pp. 1131–1137,
1999.
162
[73] G. Petitpas, J. D. Rollier, A. Darmon, J. Gonzalez-Aguilar, R. Metkemeijer, and
L. Fulcheri, “A comparative study of non-thermal plasma assisted reforming tech-
nologies,” International Journal of Hydrogen Energy, 2007.
[74] V. K. Givotov, A. A. Fridman, M. F. Krotov, E. G. Krasheninnikov, and B. I. Patru-

shev, “Plasmochemical methods of hydrogen production,” Int. J. Hydrogen Energy,
vol. 6, no. 5, pp. 441–449, 1981.
[75] R. I. Asisov, A. K. Vakar, A. F. Gutsol, V. K. Givotov, E. G. Krasheninnikov, and M. F.

Krotov, “Plasmachemical methods of energy carrier production,” Int. J. Hydrogen
Energy, vol. 10, no. 7, pp. 475–477, 1985.
[76] BOC Edwards, XDS Dry Pump, 2007.
[77] Stanford Research Systems, Models RGA100, RGA200, and RGA300 Residual Gas
Analyzer, 2005.
[78] BOC Edwards, E2M28 and E2M30 Rotary Vacuum Pumps, 2003.
[79] I. Langmuir, “The interaction of electron and positive ion space charges in cathode
sheaths,” Physical Review, vol. 33, no. 6, pp. 954–989, 1929.
[80] I. L. H. M. Mott-Smith, “The theory of collectors in gaseous discharges,” Physical

Review, vol. 28, no. 4, pp. 727–763, 1926.
[81] F. F. Chen, “Electric probes,” in Plasma Diagnostic Techniques (R. H. Huddlestone

and S. L. Leonard, eds.), pp. 113–200, Academic Press, 1965.
[82] I. H. Hutchinson, Principles of Plasma Diagnostics. Cambridge University Press,

2002.
[83] J. Laframboise and L. Parker, “Probe design for the orbit-limited current collection,”
Journal of Fluids, vol. 6, pp. 629–636, 1973.
[84] J. Laframboise and J. Rubinstein, “Theory of a cylindrical probe in a collisionless

magnetoplasma,” Journal of Fluids, vol. 19, pp. 1900–1908, 1976.
[85] I. D. Sudit and R. C. Woods, “A study of the accuracy of various langmuir probe
theories,” J. Appl. Phys., vol. 76, no. 8, pp. 4487–4497, 1994.
[86] National Institute of Standards and Technology - Physics Laboratory, “Nist atomic
spectra database.” http://physics.nist.gov/PhysRefData/ASD/
lines_form.html, July 2009.
[87] R. Stocker, “Excitation-emission-spectra.” http://www.

combustion-database.com/AES/aes_home.htm, July 2009. Insti-
tute of Thermal Power Systems.
[88] Y. Itikawa and N. Mason, “Cross sections for electron collisions with water
molecules,” J. Phys. Chem. Ref. Data, vol. 34, no. 1, pp. 1–22, 2005.
163
[89] Y. Lee, J. Oh, and J. Shin, “Dynamic model and electrical characteristics for rf-biased
electronegative plasma sheath,” IEEE Transaction on Plasma Science, vol. 30, no. 3,
pp. 1320–1330, 2002.
[90] D. S. Stafford, “O2(1delta) production in he/o2 mixtures in flowing low pressure plas-
mas,” Journal of Applied Physics, vol. 96, no. 5, pp. 2451–2465, 2004.
[91] D. S. Stafford, “Modeling of singlet-delta oxygen yields in flowing electric dis-

charges,” Master’s thesis, University of Illinois at Urbana-Champaign, 2004.
[92] R. Dorai and M. J. Kushner, “Consequences of propene and propane on plasma reme-
diation of nox,” Journal of Applied Physics, vol. 88, no. 6, pp. 3739–3747, 2000.
[93] R. Dorai, “Modeling of plasma remediation of nox using global kinetic models
accounting for hydrocarbons,” Master’s thesis, University of Illinois at Urbana-
Champaign, 2000.
[94] P. N. Brown, G. D. Byrne, and A. C. Hindmarsh, “Vode, a variable-coefficient

ode solver,” SIAM Journal on Scientific and Statistical Computing, vol. 10, no. 5,
pp. 1038–1051, 1989.
[95] L. Spencer, S. V. T. Nguyen, and A. D. Gallimore, “Dissociation of co2 in a radio-

frequency plasma source,” in CAPPSA 2009 - 4th International Congress on Cold
Atmospheric Pressure Plasmas: Sources and Applications, (Ghent, Belgium), June
2009.
164

2009 Nguyen Thesis

Uploaded by

Copyright:

Available Formats

2009 Nguyen Thesis

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2009 Nguyen Thesis

Uploaded by

Copyright:

Available Formats

Hydrogen Production in a Radio-Frequency

Plasma Source Operating on Water Vapor

A dissertation submitted in partial fulfillment

Professor Alec D. Gallimore, Co-Chair

for introducing me to Dr. Gallimore and encouraging me to pursue my doctoral work

Shabshelowitz, Raymond Liang, David Huang, and Roland Florenz.

The best way for me to express my gratitude to my parents is by repeating a Vietnamese

of a mother is as bountiful as the spring water flowing from its source.

List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvi

List of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii

1.1 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2.1 Energy and Climate Challenges . . . . . . . . . . . . . . . . . . . 7

III. Background–Plasma Overview . . . . . . . . . . . . . . . . . . . . . . . 31

3.1 Overview of Plasma Physical Processes in Water Vapor Discharge 31

IV. Experimental Setup and Diagnostics . . . . . . . . . . . . . . . . . . . 67

4.1 Experimental Setup . . . . . . . . . . . . . . . . . . . . . . . . . 67

5.1 Argon and Water Vapor Plasma Comparison . . . . . . . . . . . . 89

VI. Kinetic Simulation Setup and Results . . . . . . . . . . . . . . . . . . . 129

6.1 Reactions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

VII. Conclusions and Future Work . . . . . . . . . . . . . . . . . . . . . . . 142

7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142

A. GlobalKin Data File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

2.1 Total world energy consumption, 1980-2030 . . . . . . . . . . . . . . . . 8

2.2 Total non-OECD energy consumption, 1980-2030 . . . . . . . . . . . . . 8

2.6 Annual global energy balance . . . . . . . . . . . . . . . . . . . . . . . . 13

2.8 Vibration modes of CO2 . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.11 Distribution of energy consumption in the US in 2007 . . . . . . . . . . . 18

2.14 Schematic diagram of the PEM electrolyte . . . . . . . . . . . . . . . . . 28

3.1 Potential curve for molecule AB . . . . . . . . . . . . . . . . . . . . . . 33

3.2 Potential curve for molecule AB . . . . . . . . . . . . . . . . . . . . . . 35

3.3 Potential curve for molecule AB . . . . . . . . . . . . . . . . . . . . . . 40

3.4 Potential curve for molecule AB . . . . . . . . . . . . . . . . . . . . . . 43

3.5 Potential curve for molecule AB . . . . . . . . . . . . . . . . . . . . . . 47

3.6 Plasma sheath in a DC plasma discharge . . . . . . . . . . . . . . . . . . 50

3.7 Plasma sheath in a RF plasma discharge . . . . . . . . . . . . . . . . . . 52

3.8 Experimental setup to study decomposition of phenol solutions . . . . . . 61

3.9 Experimental setup to study diamond growth in a water plasma source . . 62

3.10 Experimental setup to study CO2 dissociation . . . . . . . . . . . . . . . 64

3.11 Experimental setup to study hydrogen production in a water plasma source 65

4.1 Drawing of vacuum facility . . . . . . . . . . . . . . . . . . . . . . . . . 68

4.3 Photograph of plasma source at PEPL . . . . . . . . . . . . . . . . . . . 70

4.5 Schematic diagram of plasma source . . . . . . . . . . . . . . . . . . . . 71

4.7 Components of the RGA . . . . . . . . . . . . . . . . . . . . . . . . . . 75

4.8 Differential pump system for the RGA . . . . . . . . . . . . . . . . . . . 76

4.9 Calibration curves for hydrogen, oxygen, and water . . . . . . . . . . . . 77

4.10 Effect of argon flow rate on hydrogen production . . . . . . . . . . . . . 79

4.11 Effect of argon flow rate on ion and electron densities . . . . . . . . . . . 79

4.12 Effect of argon flow rate on floating and plasma potential . . . . . . . . . 80

4.13 A typical Langmuir probe curve for water vapor plasma . . . . . . . . . . 81

4.14 Semi-log plot of Langmuir probe trace . . . . . . . . . . . . . . . . . . . 85

4.15 Flow chart for Langmuir probe analysis . . . . . . . . . . . . . . . . . . 86

4.16 Diagram of the spectrometer setup . . . . . . . . . . . . . . . . . . . . . 88

5.1 Optical emission spectrum of an argon discharge . . . . . . . . . . . . . 90

5.2 Optical emission spectrum of a water plasma discharge . . . . . . . . . . 91

5.3 Cross-sections of electron impact with water molecule . . . . . . . . . . 92

5.4 RGA spectrum of the water plasma . . . . . . . . . . . . . . . . . . . . . 93

5.5 Ion plasma density of an argon discharge as a function of RF power for

5.9 Electron temperature of the argon plasma as a function of RF power for