University of Southern California Dissertations and Theses

The physics and application of compact pulsed power to transient plasma ignition

(USC Thesis Other)

The physics and application of compact pulsed power to transient plasma ignition

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

THE PHYSICS AND APPLICATION OF COMPACT PULSED POWER TO
TRANSIENT PLASMA IGNITION

by

Daniel R. Singleton

________________________________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ELECTRICAL ENGINEERING)

August 2010

Copyright 2010 Daniel R. Singleton
ii
Acknowledgments
I would like to thank my advisor, Prof. Martin Gundersen for his guidance,
encouragement, and wisdom, which have shaped me greatly over the past four years. He
has helped me grow not only as an engineer and a scientist, but also as an entrepreneur, a
leader, and an educator. Whether we discussed physics, history, film, or something
entirely different, I could always count on a good conversation when I stepped into his
office. It has truly been a great pleasure to work with him, and I look forward to
continuing to do so in the future.
I would like to acknowledge those whose help was invaluable to the completion
of my Ph.D.: Dr. Andràs Kuthi, for his wisdom and technical expertise, Prof. José
Sinibaldi, who has been a great friend and colleague ever since my first experiment in
Monterey, and Prof. Hai Wang, who was always willing and able to discuss any kind of
problem, technical or otherwise, always with a calm smile and contagious enthusiasm.
I would like to thank Prof. Chris Brophy and Dr. Campbell Carter. It was highly
educational and enjoyable to work with these great researchers and experimentalists.
Their talent, patience, and kindness made my experiments with them some of the most
enjoyable I have had.
I would like to thank my family and friends, who supported me to pursue what
interests me, and I therefore owe my time at USC to them. I would like to thank my lab
mates, Jason Sanders, Esin Sözer, and Scott Pendleton, for being wonderful friends and
colleagues. Without them my time at USC would have so much less meaning.
iii
Finally, I would like to thank Shannon for her unconditional love and support
throughout my work on my Ph.D.
iv
Table of Contents
Acknowledgments ii
List of Tables vi
List of Figures vii
Abbreviations xiii
Abstract xv
Chapter 1: Introduction to Transient Plasma 1
1.1 Introduction 1
1.2 Non-Equilibrium Plasmas at Atmospheric Pressure 2
1.2.1 Pulsed Corona Discharge 4
1.2.2 Electrode Design 6
1.2.3 Streamer Theory 8
1.3 Transient Plasma Ignition 14
Chapter 2: Pulse Generator Design 21
2.1 Introduction 21
2.2 Switches 22
2.2.1 Gas Switches 23
2.2.2 Solid State Switches 27
2.2.3 Magnetic Switches 32
2.3 Pulse Generator Architectures 35
2.3.1 Marx Generators 36
2.3.2 Blumlein 37
2.3.3 Magnetic Pulse Compression 41
2.4 Development of a 12 ns Pulse Generator 43
Chapter 3: Application of Transient Plasma 50
3.1 Introduction 50
3.2 Transient Plasma Ignition in Pulse Detonation Engines 51
3.2.1 Measurement of Ignition Delay in a PDE at NPS 56
3.2.2 Flame Propagation in a PDE at WPAFB 68
3.3 Transient Plasma Ignition in a Constant Volume Reactor 73
3.3.1 Varying Pulse Width 77
3.3.2 Varying Gap Size 81
3.3.3 Varying Anode Length 85

v
3.3.4 Varying Cathode Porosity 88
3.3.5 Varying Pulse Amplitude 90
3.4 Transient Plasma Ignition in an Internal Combustion Engine 91
Chapter 4: Transient Plasma Physics 99
4.1 Introduction 99
4.2 The Effect of Humidity on Transient Plasma Ignition 100
4.2.1 Measurement of Streamer Intensity 101
4.2.2 Measurement of OH and O
3
104
4.3 Spatially Resolved Transient Plasma Ignition in C
2
H
4
-air 131
Chapter 5: Conclusion & Future Work 139
5.1 Conclusion 139
5.2 Future Work 141
5.2.1 Measurement of Atomic Oxygen in Transient Plasma Ignited CH
4
-Air 141
5.2.2 Pulse Generator Design 144
5.2.3 Electrode Design for an Internal Combustion Engine 145
5.2.4 Soot Reduction 147
5.2.5 Gas Turbine Ignition and Relight 148
5.2.6 Pulsed Electric Field Treatment of Wine Grapes 149
Bibliography 152

vi
List of Tables
Table 1 – A summary of various switch parameters for pulsed power applications
(Schamiloglu, 2004). 23
Table 2 – Characteristics of pulse generators at USC used for TPI. 49

vii
List of Figures
Figure 1 – Left: Typical electrode used for transient plasma ignition; Right:
streamers generated by a single 60 kV, 54 ns pulse in a 15 mm gap.
Photo taken using a 15 second exposure in a dark chamber. 5
Figure 2 – Electric field distribution prior to breakdown in a coaxial cylinder,
where x is the radial distance from the center axis. The gap distance
is d = b – a. 7
Figure 3 – An electric field plot of the ignition geometry used in many transient
plasma ignition experiments when 90 kV is applied to the anode.
Black lines show equipotential lines at 5000 V increments. 8
Figure 4 – Electric fields in a gap, d, containing an electron avalanche. Left:
Lines of force of the space charge induced field, E', and E
0
, shown
separately; Right: Lines of force resulting from E = E
0
+ E' (Raizer,
1998). 11
Figure 5 – Left: Cathode directed streamer propagation with secondary
avalanches moving toward the positive head of the streamer; Right:
Lines of force of the enhanced electric field near streamer head
(Raizer, 1998). 12
Figure 6 – Left: Streamer heads, where the main production of radicals occurs,
propagate across the gap; Right: The streamers have bridged the gap
and the faint plasma channels that trail the streamer head become
highly conductive, seen by an increase in intensity. 13
Figure 7 – Comparison of rates of electron impact and thermal reactions
producing important species for combustion such as O and OH
(Starikovskaia, 2006). 15
Figure 8 – Combustion of stoichiometric ( ϕ=1) CH
4
-air at 1 atm. The transient
plasma was generated using a 100 ns, 75 kV pulse in a 47 mm gap.
The spark was generated using a 10 µs, 10 kV pulse. 19
Figure 9 – A pulsed power system accumulates energy over a relatively long
period (left), and then delivers the energy in a short period with high
peak power (right) (Mankowski, 2000). 22
Figure 10 – Basic thyratron (Burkes, 1979). 25
Figure 11 – Pseudospark switch diagram for the FS2000 (Alstom). 27
viii
Figure 12 – Left: Basic structure of an SCR; Center: Equivalent model using two
bipolar transistors; Right: SCR circuit symbol (Linder, 2006). 28
Figure 13 – Left: Cross-section of an n-channel MOSFET; Right: MOSFET
circuit symbol (Linder, 2006). 30
Figure 14 – Left: Cross section of a vertically-integrated IGBT; Right: IGBT
circuit symbol (Linder, 2006). 31
Figure 15 – BH loop (hysteresis loop) to describe magnetic switch behavior
(Smith, 2002). 34
Figure 16 – Variation of voltage and current in the LC circuit shown left (Smith,
2002). 35
Figure 17 – Basic circuit of a three-stage Marx generator (adapted from Smith,
2002). 37
Figure 18 – Circuit diagram of the basic Blumlein pulse forming line (Smith,
2002). 38
Figure 19 – Blumlein operation after the switch in Figure 18 closes (Smith, 2002). 40
Figure 20 – Pseudospark switched pulse generator based on Blumlein architecture
used for transient plasma ignition experiments. 41
Figure 21 – Basic magnetic pulse compressor circuit (Smith, 2002). 42
Figure 22 – Voltage during the operation of a basic magnetic pulse compression
stage (Smith, 2002). 43
Figure 23 – Schematic for a 65 kV, 12 ns compact pulse generator. 44
Figure 24 – The 90 kV, 85 ns pseudospark pulse generator (left), and the 65 kV,
12 ns solid state pulse generator (right). 45
Figure 25 – Output voltage and current of the 65 kV, 12 ns pulse generator. The
load is 200 Ω. 47
Figure 26 - Rapid Charging Circuit 48
Figure 27 – Typical waveform into a 200 Ω load from pulse generators at USC
used for TPI. 49
Figure 28 – Projected efficiency of a variety of airborne engines (Bussing, 1997). 52
ix
Figure 29 – Schematic showing the detonation front structure consisting of
multiple shock waves and reaction zones (Ciccarelli, 2006). 55
Figure 30 – PDE operating cycle (Hackard, 2007). 56
Figure 31 – Top: A section view of the transient plasma igniter and grounded
screen for the 12 ns TPI system; Bottom: threaded rod and grounded
screen for the 85 ns TPI system. Dimensions are in mm. 58
Figure 32 – Top: The PDE at NPS with the compact igniters installed; Bottom: A
section view of the TPI PDE system. 60
Figure 33 – Results of ignition delay in a PDE using a 12 ns TPI system (80 mJ)
and an 85 ns TPI system (800 mJ). Spark ignition delays are not
shown because they are much greater (around 8 ms). 63
Figure 34 – Ignition delay time vs. reduced electric field in a stoichiometric H
2
-air
mixture at 1 atm. T = 1000 K. Total energy release E = 4 × 10
-3

J/cm
3
(Aleksandrov, 2001). 67
Figure 35 – The electrode used in the PDE at WPAFB. Dimensions are in mm. 69
Figure 36 – Head-on view of the PDE engine block, with four fuel-air valves, and
a transient plasma igniter (left) and a traditional spark plug igniter
(right). 70
Figure 37 – Setup of the PDE at WPAFB for Schlieren photography in a clear
section with a vertically mounted Spark or TPI plug. 71
Figure 38 – Flame propagation of stoichiometric C
2
H
4
-air in a clear section of a
PDE tube. The images were taken 10 ms after ignition. 72
Figure 39 – Section view of the combustion reactor with removable cathode to
allow a change in gap size. Shown with a quartz window where a
flange with a pressure transducer port could be inter-changed. 74
Figure 40 – Electrode for the combustion reactor. A high-voltage cable plugs in
from the right. The gap size shown is 6 mm. Dimensions are in
mm. 75
Figure 41 – Setup of transient plasma ignition experiments in a constant volume
reactor. 76
Figure 42 – Photo of the combustion chamber and mixing manifold. 77
x
Figure 43 – Ignition delay of quiescent stoichiometric C
2
H
4
-air at 1 atm in 1.73 L
stainless steel chamber. The gap is 6 mm and the pulse amplitude
for both TPI systems was 42 kV. 78
Figure 44 – Photographs of ignition of C
2
H
4
-air ( ϕ = 1.1). Time indicates time
after discharge. Top: automobile ignition coil with spark plug;
Bottom; 12 ns TPI system. 79
Figure 45 – Transient plasma produced ignition delay of quiescent stoichiometric
C
2
H
4
-air at 1 atm in 1.73 L stainless steel chamber using a 12 ns
pulse generator. The maximum applied electric field strength is 100
kV/cm for all gap sizes. 82
Figure 46 – Transient plasma produced ignition delay of quiescent stoichiometric
C
2
H
4
-air at 1 atm in 1.73 L stainless steel chamber using am 85 ns
pulse generator. The maximum applied electric field strength is 100
kV/cm for all gap sizes. 84
Figure 47 – Different configurations of the constant volume reactor for testing
different anode lengths. Top: long threaded rod; Bottom left: short
threaded rod; Bottom right: spark plug. 86
Figure 48 – Transient plasma ignition characteristics of ethylene-air with varying
anode lengths, and constant E/n. 87
Figure 49 – Images of flame propagation 6 ms after the discharge of a 12 ns, 42
kV pulse and the corresponding cathodes with varying porosity (hole
number × hole size in inches). The gap was 6 mm. 89
Figure 50 – Ignition delays produced by a 12 ns pulse with varying pulse
amplitude (from 38 kV to 54 kV). 91
Figure 51 – TPI in an ICE at Nissan Research Center. Pressure vs. crank angle,
for the spark, 100 ns pulse, and 20 ns pulse, ф =0.72 (Cathey, 2007). 94
Figure 52 – 800 HP engine used in the experiment at Earnhardt-Childress Racing. 95
Figure 53 – Measurement setup for an internal combustion engine experiment at
Earnhardt-Childress Racing. 96
Figure 54 – Peak pressure in the cylinder in an internal combustion engine using
transient plasma ignition and spark ignition at while the engine is
idle at 3000 rpm. 97
xi
Figure 55 – Detonation wave velocities in the PDE at NPS with the vitiator ON
(moist air) and vitiator OFF (dry air). Humidity was above 5% with
the vitiator ON. 101
Figure 56 – Streamer intensity and energy delivered vs. mole fraction H
2
O. 102
Figure 57 – Intensity of streamer discharge with different levels of humidity. 103
Figure 58 – Effect if humidity on the uniform-field breakdown voltage in
atmospheric air (Blair et al., 1963). 104
Figure 59 – Simulation of OH and O
3
after the application of an electric field in
air, with varying amounts of humidity. Courtesy of J. Luginsland. 106
Figure 60 – Experimental setup for OH-LIF measurement. 108
Figure 61 – Electronic states and energy levels of OH. W indicates the absorption
process, Q
e
the electronic quenching process, Q
V,10
the vibrational
energy transfer process and A the fluorescence process (K. Kohse-
Hoinghaus, 1994). 112
Figure 62 – Left: Cross-section of a 10.1 cm ID, 20.2 cm long combustion
chamber and; Right: a photo of the combustion chamber. 114
Figure 63 – Streamers propagating from a washer anode. 120
Figure 64 – Boltzmann modeling of O
3
concentration as a function of water vapor
100 µs after a discharge. 123
Figure 65 – Simulation of peak OH concentration after a single 70 kV, 100 ns
discharge in air with varying levels of humidity. 124
Figure 66 – Result of OH LIF experiment with varying levels of humidity. 125
Figure 67 – Measured O
3
density after discharge with varying levels of humidity. 127
Figure 68 – Measured O
3
density after discharge with varying electric field
strength. 128
Figure 69 – Left: Streamers generated by a 56 kV, 54 ns pulse (Maximum E/n ≈
400 Td); Right: Flame propagation from multiple ignition sites at the
base of the streamers after a single pulse in ϕ=1.1 C
2
H
4
-air. 133
xii
Figure 70 – Combustion of ϕ=1.1 C
2
H
4
-air 6 ms after ignition. Left: conventional
spark ignition using a standard 105 mJ spark ignition system with a
spark plug. Right: transient plasma ignition using a 12 ns, 52 kV
pulse in a 6 mm gap. 134
Figure 71 – Electrode with field enhancement at four protrusions from the
cathode. Left: Photograph showing ten 12 ns, 50 kV pulses in air;
Right: Ignition occurring at multiple locations after a single pulse in
ϕ=1.1 C
2
H
4
-air. 135
Figure 72 – Schematic of atomic oxygen energy levels showing the two photon
absorption and emission. Two photons at 225.7 nm are absorbed,
exciting the O atom to the 3p
3
P state. The fluorescence at 845 nm is
then measured (Bamford, 1986). 143
Figure 73 – Proposed electrode design that eliminates the undesirable recessed
volume. 146
Figure 74 – Particle size distribution of soot produced during ignition by spark
and transient plasma for a rich ethylene-air mixture. 147
Figure 75 – A typical flight relight envelope for gas turbines (Rolls Royce). 149
Figure 76 – Pulsed (left) and non-pulsed (right) juice extracted from Chardonnay
grapes. The pulsed sample yielded significantly more juice and
better color. 150

xiii
Abbreviations
AFRL Air Force Research Laboratory
air O
2
+ 3.76N
2
CJ Chapman / Jouguet
CV constant volume
CW continuous wave
DA direct absorption
DBD dielectric barrier discharge
DDT deflagration-to-detonation transition
DOS diode opening switch
ECR Earnhardt-Childress Racing
HV high-voltage
ICE internal combustion engine
IGBT insulated gate bipolar transistor
IMEP indicated mean effective pressure
IR infrared
LIF laser-induced fluorescence
MOSFET metal–oxide–semiconductor field-effect transistor
MSD multiple spark discharge, an automobile ignition system
NASCAR National Association for Stock Car Auto Racing
NPS Naval Postgraduate School
PDE pulsed detonation engine
xiv
PLIF planar laser-induced fluorescence
RPM revolutions per minute
SCR silicon controlled rectifier
STP standard temperature and pressure, i.e. P = 1 atm, T = 298 K
TALIF two-photon absorption laser-induced fluorescence
TDC top dead center
TPI transient plasma ignition
USC University of Southern California
VET vibrational energy transfer
WPAFB Wright-Patterson Air Force Base

xv
Abstract
This dissertation presents an experimental study of the application and the
underlying physics of transient plasma ignition. Transient plasma generated by
nanosecond electrical discharges has demonstrated lean-burn capability and reductions in
ignition delay in a variety of engines, resulting in higher combustion efficiencies and
lowered emissions compared to conventional spark ignition. The experiments performed
demonstrate the effects of transient plasma ignition and attempt to understand the basic
physics behind it by examining the production of radicals, conditions for ignition, and
combustion characteristics.
Critical to the study of the application of transient plasma is the enabling
technology: pulsed power systems. To realize this technology in aircraft and automobile
engines, the size, weight, cost, reliability, and electrical energy consumption of the pulsed
power systems must be comparable to that of the current spark ignition systems. To that
end, a compact, solid state 12 ns pulse generator was developed and successfully
implemented in transient plasma ignition experiments.
Transient plasma was applied in both quiescent and flowing fuel-air mixtures in a
pulse detonation engine (PDE), an internal combustion engine (ICE), and a constant
volume reactor. Reduced ignition delays and lean-burn capability were demonstrated. A
12 ns, 70 mJ pulse was used achieve similar gains as those produced by an 85 ns, 800 mJ
pulse, which means that more compact pulse generators may be used for this application.
It was demonstrated that water inhibits the performance of TPI, and optical diagnostic
techniques were used to determine that this was due to a significant decrease in
xvi
production of atomic oxygen, which plays an important role in enhancing combustion
kinetics.
It was demonstrated that after a transient plasma discharge, ignition kernels are
formed at the ends of spatially separated streamer channels, where there is an enhanced
reduced electric field (hundreds of Td) and significant energy transferred into
electronically excited species. Evidence is presented that transient plasma ignition occurs
in two phases; a distinct non-thermal phase of ignition, wherein the efficient production
of electronically excited species accelerates reaction rates, and a subsequent thermal
phase, driven by exothermic fuel oxidation in reactions with free radicals and the decay
of active species. It is concluded that TPI has the potential to improve combustion
efficiency compared to traditional spark ignition for wide-ranging applications to engine
technology.
1
Chapter 1: Introduction to Transient Plasma
1.1 Introduction
The combustion of fossil fuels has driven the development of the U.S. over the
past century. Today, the U.S. consumes almost one million gallons of petroleum every
minute of each day for the transport of people and goods alone, which only accounts for
60% of the country’s total energy consumption (U.S. Department of Energy). As
important as it is, the combustion of fossil fuels results in air pollution, including nitrogen
oxides (NOx) and carbon dioxide (CO
2
). Our staggering energy consumption rates have
significant implications for the environment as well as the economy. The motivation for
this research is to develop transient plasma ignition for improved combustion efficiencies
and lowered emissions in engines of varying types.
Non-equilibrium plasma in the transient, formative phase of an arc, applied to
ignition and combustion (transient plasma ignition or TPI) has several advantages over
conventional spark ignition used in most combustion applications. TPI has consistently
demonstrated reductions in ignition delay, lean-burn capability, and the ability to ignite
higher mass flow rates, resulting in improved efficiency and reduced emissions from a
variety of airborne engines and internal combustion engines.
2
The approach to transient plasma ignition is different from traditional approaches
in that a short (typically <100 ns) high-voltage pulse is used to initiate ignition, rather
than a many microsecond to several millisecond high-voltage pulse for spark ignition
(Liu, 2005; Wang, 2005). The short pulses prevent an arc from occurring and produce
non-equilibrium plasma that initiates different molecular processes. It is hypothesized
that high energy electrons produced during the electrical discharge are responsible for
electron impact dissociation, ionization, and fragmentation of molecules in the gas,
resulting in the formation of radicals that drive the combustion process.
At the core of this research is the development of compact pulsed power systems
to generate high-voltage (roughly 30 - 90 kV) pulses on a nanosecond timescale in order
to produce transient plasma. Additionally, it is critical to understand the science of
highly non-equilibrium energy states that occur between matter and energy to realize
transient plasma ignition technology. The focus of this dissertation is both the
engineering of more compact and reliable transient plasma ignition systems and the
application of these systems in experiments to improve our understanding of the process.
1.2 Non-Equilibrium Plasmas at Atmospheric Pressure
The most familiar ignition system used in both automobile and aircraft engines
consists of an ignition coil and a spark plug. An arc produced in a traditional spark plug
is an equilibrated (thermal) or nearly equilibrated plasma. Ignition is achieved by local
heating of the gas, which increases the dissociation rate and reactions of chain
3
prolongation and development (Samano, 1993; Polak, 1975; Eichenberger, 1999). In this
process, only fractions of the energy supplied to the spark plug are delivered to the
mixture. The energy transfer efficiency is low since most of the energy is dissipated by
the resistances in the transformer, spark plug wires, and spark plugs (Hilliard, 1984;
Rohwein, 1997).
Non-equilibrated (non-thermal) plasma is characterized by a high electron
temperature T
e
and a low gas temperature T
g
(T
e
>> T
g
). In the extreme case, the gas
temperature can be close to room temperature and the electron temperature can be well
above 20,000 K (Becker, 2005). In non-equilibrated plasma, the ionization process is
dominated by field-driven, energetic electrons impacting with non-excited atoms and
molecules. The energy transfer efficiency is high and significant energy goes into
creating highly energetic electrons instead of heating of the gas.
Another difference between equilibrated and non-equilibrated plasmas is that non-
equilibrated plasmas are very selective in terms of the allowed plasma chemical
reactions. An application that makes use of this selectivity is the treatment of exhaust
gasses, where radicals generated by the non-thermal plasma oxidize or decompose
pollutant molecules (Mizuno, 1986; Gallimberti, 1988). In a thermal system for the same
application, the gas would be heated to destroy the pollutants.
In non-equilibrium plasma, the mean electron energy is much greater than the
thermal energy of molecules, and the electron energy spectrum and rate of ionization by
electrons depends on the electric field strength E and the gas density n. Electrons gain
energy from the field and pass it on to atoms and molecules with a collision frequency v
m

4
that is proportional to n. The mean electron energy, ̅, is proportional to E/n, known as
the reduced electric field, which is the electric field strength over the gas density (Raizer,
1997; Bazelyan, 1998):

̅ =

=
√

√ ≈ 0.8

√ ~
eq. (1)

In eq. (1), k is the Boltzmann constant, T
e
is the electron temperature, e is the
electron charge, E is the electric field, l is the path length, n is the gas density and δ is the
elastic loss coefficient denoted as 2m/M. High energy electrons are responsible for the
dissociation and ionization of atoms and molecules by electron impact and are critical to
the transient plasma ignition mechanism, and since the mean electron energy is
proportional to E/n, E/n is an important concept in this work. The units of E/n are often
given in Townsends (Td), where, 1 Td = 10
-17
V·cm
2
. As an example, at E = 100 kV/cm
and p = 1 atm in air, E/n = 398 Td = 3.98× 10
-15
V·cm
2
.
1.2.1 Pulsed Corona Discharge
Non-equilibrium plasmas can be produced using various methodologies,
including glow and RF discharges at low pressures, and short pulses, barrier and hollow
cathode discharges at atmospheric pressures. The primary method used for this research
is a pulsed corona methodology, using short (less than 100 ns), high-voltage (usually 30 –
90 kV) pulses to generate non-equilibrium plasma at atmospheric pressures.
5
In ambient air, the most commonly observed non-equilibrated plasma is a corona
discharge. A corona discharge is typically a low-current, faintly luminous discharge that
appears on sharp edges or points where field enhancement is occurring. It is spatially
non-uniform where a strong electric field, luminosity, and ionization all occur near one
electrode. In a continuous corona discharge, the discharge volume is typically small and
is characterized by a low discharge power. In a pulsed corona discharge, the discharge
volume is increased by overvolting a gap with short pulses. Using these pulses, the
corona discharge is broken up into separate thin, conductive plasma channels, called
streamers (Figure 1) (Raizer, 1998).

Figure 1 – Left: Typical electrode used for transient plasma ignition; Right: streamers
generated by a single 60 kV, 54 ns pulse in a 15 mm gap. Photo taken using a 15
second exposure in a dark chamber.

6
Streamers generated by a pulsed corona methodology are the precursor to an arc.
The short, high-voltage pulses allow the electrical energy to be coupled into a non-
equilibrated plasma without a transition from streamers to an arc. The total time from the
development of an electron avalanche to streamer propagation across a gap is typically
100 – 300 ns. If the pulse is too long, it will cause a spark channel to form and the pulsed
corona discharge will collapse into an arc. The amplitude of the pulse is also important
in determining whether there is a breakdown of the gap. Using a pulsed corona
methodology allows a significant overvolting of the system without causing breakdown,
increasing the maximum electric field strength in the gap. When designing a pulsed
power system for transient plasma ignition, pulse width and amplitude must be matched
to the discharge conditions to ensure that transient plasma, not an arc, is delivered to the
fuel-air mixture.
1.2.2 Electrode Design
There are many common configurations of electrode systems for pulsed corona
discharges, which include pin-to-plane, wire-to-plane, and coaxial wire-cylinder. In this
research, a coaxial geometry is primarily used to generate the non-uniform electric field.
Using this geometry, a positive voltage is applied to the center electrode (anode), and the
outer electrode is grounded (cathode). In this work, the gap between the two electrodes is
typically between 5 – 50 mm. This geometry was largely chosen for ease of installation
in various engine types because of its similarity to a spark plug. Figure 2 shows the
7
equation for the electric field distribution in the coaxial geometry prior to breakdown
(Fridman, 2004).

Using this geometry the maximum electric field occurs when x equals the inner
electrode radius, a. Active particles are produced throughout the discharge volume, but
the highest concentration is generated near the anode where the electric field is the
highest. This is where most of the chemical reactions take place during transient plasma
ignition. This geometry has a field enhancement factor of 1.82 (maximum field/average
field). Figure 3 presents a simulation in Quickfield of the electric field prior to being
disturbed by the plasma for a typical electrode used in this work. For most applications
in this dissertation, the anode is a threaded rod, resulting in significant electric field
enhancement.

Figure 2 – Electric field distribution prior to breakdown in a coaxial cylinder, where
x is the radial distance from the center axis. The gap distance is d = b – a.
=
[ (
)]

8

1.2.3 Streamer Theory
The breakdown mechanism of transient plasma is based on streamer theory,
which is a modified version of Townsend breakdown theory. Townsend breakdown
theory is valid for pd < 1000 Torr cm, where p is the pressure and d is the gap distance
(Razier, 1998). Breakdown by short pulses in larger pressure gaps proceed much faster
than can be explained by the Townsend breakdown mechanism, so streamer breakdown
theory was developed by Loeb and Meek (1941), Raether (1964), and Meek and Craggs
Figure 3 – An electric field plot of the ignition geometry used in many transient
plasma ignition experiments when 90 kV is applied to the anode. Black lines
show equipotential lines at 5000 V increments.

9
(1978). The theory is built around the concept of a streamer, a thin ionized channel that
follows a primary avalanche across a gap.
Streamer properties are difficult to characterize, but a simplified version of
streamer theory starts with an electron avalanche. An avalanche is the primary element
of any breakdown mechanism and is initiated by electrons that leave the cathode due to
an external field. Consider an applied electric field, E
0
= V/d, where V is the voltage
applied to the electrodes and d is the gap. An electron gains enough energy in the electric
field to ionize an atom or molecule, and then loses its energy. There are now two slow
electrons, which gain energy from the field and ionize more atoms or molecules, and so
on, resulting in an electron avalanche. In the electron avalanche, the electrons quickly
drift toward the anode due to the electric field, while heavier positive ions slowly drift
toward the cathode. When an ion reaches the cathode, it knocks out another electron
from the cathode with a certain probability. If the number of electrons from the cathode
increases in each of these cycles, then the number of avalanches increases as well, and
charges fill up the gap (Razier, 1998).
Space charges generated during an electron avalanche produce their own field, E',
that add up with the applied field E
0
to give a field, E, stronger than E
0
(Figure 4)
(Raizer, 1998). For an avalanche to develop into a streamer the space charge induced
field has to increase to a value on the order of the applied external field. This is called
the criterion of streamer formation:

′ =
e
( /)
≈
eq. (2)
10

In eq. (2), r
D
is the avalanche head radius, and α is the Townshend ionization
coefficient, which refers to the electron production per unit length (in air at STP,
breakdown occurs at approximately 30 kV/cm and α ≈ 10 cm
-1
). If it is assumed that x is
the gap length, d, and that the avalanche head radius is the ionization length, 1/ α, then the
Meek breakdown condition can be obtained, which demands that αd ≥ 20, and the
number of electrons in the avalanche N
e
= exp( αd) ~ 10
8
. Assuming this condition is
met, the primary avalanche can transform into a streamer when it reaches the anode.
Since the discharge gap for transient plasma ignition applications is relatively small and
there is significant overvoltage, cathode-directed or positive streamers are formed.
11

The primary electron avalanche creates a positively charged trail that a streamer
will follow back across the gap from the anode toward the cathode. A large number of
secondary avalanches are produced by photons near this trail, which are uniformly
emitted in all directions during the ionization process (Figure 5) (Raizer, 1998). The
electrons that are produced are pulled into the trail by its field and mix with the primary
avalanche electrons to form a streamer channel. The secondary electrons enhance the
positive charge at the tip of the streamer head thus attracting additional electrons and
initiating another wave of secondary avalanches. This is how the streamer propagates

Figure 4 – Electric fields in a gap, d, containing an electron avalanche. Left:
Lines of force of the space charge induced field, E', and E
0
, shown separately;
Right: Lines of force resulting from E = E
0
+ E' (Raizer, 1998).
12
across the gap. When the streamers bridge the gap between the anode and the cathode,
the plasma channel becomes highly conductive and an arc can form. The total time for
avalanche development, avalanche to streamer development and streamer propagation
between the electrodes is typically 100 – 300 ns. The electron drift velocities in the field
are on the order of 10
7
cm/s, while the velocities near the head of the streamer are on the
order of 10
8
cm/s, and the charge density is around 10
12
cm
-3
.

Figure 6 illustrates the different stages of streamer development in two
photographs. A washer was used as the anode to ensure that the streamers were in the
same plane for clearer images. The first phase is the streamer discharge. This phase
happens during the rising edge of the applied high-voltage pulse. Bright streamer heads,

Figure 5 – Left: Cathode directed streamer propagation with secondary avalanches
moving toward the positive head of the streamer; Right: Lines of force of the
enhanced electric field near streamer head (Raizer, 1998).
13
where a high reduced electric field (hundreds of Td) arises, propagate across the gap from
the anode toward the cathode, leaving dark streamer channel characterized by low
electric fields. The mean electron energies in the streamer heads can be 5-10 eV or
greater (Veldhuizen, 2002). The second phase of streamer development occurs after
streamers bridge the gap and terminates when the current falls to zero. The previously
dark streamer channels increase in conductivity and light is emitted from newly created
ions in the sheath of these conductive channels.

It is worth emphasizing two distinct differences between streamers and arcs; the
discharge volume and the time scales. The electrodes can be designed in such a way that
the streamers are distributed throughout a volume (for example, if the washer was
removed from the electrode in Figure 6, the streamers would be distributed in half the
volume of the cylinder). By comparison, an arc follows a single path, regardless of any
Figure 6 – Left: Streamer heads, where the main production of radicals occurs,
propagate across the gap; Right: The streamers have bridged the gap and the faint
plasma channels that trail the streamer head become highly conductive, seen by an
increase in intensity.

14
changes to the electrode configuration. The effective volume excited by the streamers is
around 60-80% of the total volume between the electrodes (Becker, 2005). The
volumetric property of transient plasma suggests the possibility of initiating multi-point
ignition in a volume, rather than single-point ignition as with a spark plug. The
difference in timescales between a streamer discharge and an arc discharge is dramatic;
while streamers occur on the nanosecond time scale, an arc occurs over hundreds of
microseconds to milliseconds, sometimes more than 10,000 times longer. The short time
scale of transient plasma allows the researcher to separate the combustion kinetics from
the discharge and its products.
1.3 Transient Plasma Ignition
Transient plasma ignition has demonstrated substantially shorter ignition delays,
lean-burn capability, improved ignition of high mass flow rates, and reduced NOx
emissions, but there is still little understanding of the fundamental causes for the
observed phenomenon. It is known that the total reaction rate during combustion is
higher when the chain is artificially initiated by free radicals, such as O•, OH•, and H•,
and it is thought that high energy electrons generated during the nanosecond discharges
produce free radicals in the volume through electron impact dissociation and ionization of
molecules in the gas (Starikovskaia, 2006). It has been shown experimentally that
energetic electrons in the streamers produce active radicals throughout the ignition
volume through the dissociation of hydrocarbon chain molecules (Cathey, 2007).
15
Ignition delay time, a critical parameter for the application of TPI to pulse
detonation engines (see Chapter 3), is limited by the rate that active centers (atoms and
radicals) are produced. For spark ignition, the ignition delay time is determined by local
heating of the gas and the subsequent increase in the rate of thermal dissociation. For
transient plasma ignition, it is proposed that the ignition delay time is decreased by
dissociation and excitation of molecules by electron impact. This is both a more efficient
and a faster process than traditional thermal methods, which can be seen by the
significantly higher reaction rate constants produced by dissociation by electron impact
compared to thermal methods (Figure 7).

Figure 7 – Comparison of rates of electron impact and thermal reactions producing
important species for combustion such as O and OH (Starikovskaia, 2006).

16
A simplified model of hydrogen and oxygen can demonstrate the importance of
free radicals in combustion chemistry. Starting with a mixture of H
2
and O
2
, the free
radical chain reaction is initiated through reaction (R1), where an H• atom is produced
(Wang, 2008):

H
2
+ O
2
→ H• + HO
2
• (chain initiation) (R1)

The resultant H• atom then attacks O
2
to produce two more radicals, O• and OH•:

H• + O
2
→ O• + OH• (chain branching) (R2)

The two radicals then attack the fuel:

O• + H
2
→ OH• + H• (chain branching) (R3)
OH• + H
2
→ H
2
O + H• (chain propagation) (R4)

Note that beginning with only one H• atom, a single cycle through the chain
produces three H• atoms. Reactions (R2) and (R3) generate two OH• molecules that
form two H• atoms through reaction (R4), where the third H• is produced. This cycle
repeats, now starting with the 3 H• atoms, resulting in 9 H• atoms, which produce 27 H•
atoms, etc. This leads to exponential growth in the reaction rate, and, due to the
exponential rate of heat release eventually surpassing the rate of heat removal from the
17
water molecule produced each cycle, an explosion. Reactions (R2) and (R3) are called
chain branching reactions because the reactions of one free radical yield two free radicals.
Not all combustion reactions are chain branching, but even a small number of chain
branching reactions can have a significant impact on the time it takes for combustion to
occur (Wang, 2008; Starikovskaia, 2006).
Applying a transient plasma discharge to the mixture in the above example is
thought to increase the number of free radicals initially present in the system through
electron impact. The mean electron energy of the discharge is determined by the
reduced electric field, E/n. For this work, an E/n greater than 100 Td is typically used,
which can produce streamers with electron energies greater than 10 eV in an overvolted,
pulsed gap. Assuming streamer heads with electron energy of 10 eV are produced in the
transient plasma discharge, reaction (R1) above can be replaced with electron impact
dissociation reactions (R5) and (R6).

H
2
+ e → H • + H • + e (chain initiation) (R5)
O
2
+ e → O • + O • + e (chain initiation) (R6)

For H
2
this occurs through b
3
Σ
u
+
, and a
3
Σ
g
+
electronically excited states, and for
O
2
this occurs through A
3
Σ
u
+
and B
3
Σ
u
-
, with thresholds of approximately 7 eV (Trevisan,
2002; Starikovskaia, 2006). Reactions (R5) and (R6) increase the pool of radicals that
initiate chain branching, increasing the reaction rate, theoretically giving TPI a significant
advantage over traditional thermal ignition. Additionally, since the transient plasma
18
discharge produces radicals over a large volume compared to traditional spark ignition,
ignition can occur at multiple locations simultaneously. This volumetric effect, in
combination with the production of active species, is what is believed to lead to a reduced
ignition delay and greater burning rate and allow ignition to occur under conditions not
possible with traditional spark ignition. Figure 8 shows a comparison of traditional spark
ignition and TPI in CH
4
-air at 1 atm in a 1.73 L constant volume reactor. Note the
decrease in ignition delay, the increase in the pressure rise time, and the increase in peak
pressure when using transient plasma to ignite the mixture.
19

Figure 8 – Combustion of stoichiometric ( ϕ=1) CH
4
-air at 1 atm. The transient
plasma was generated using a 100 ns, 75 kV pulse in a 47 mm gap. The spark was
generated using a 10 µs, 10 kV pulse.

20
While it is thought that only a small fraction of the electrical energy goes into
thermal heating of the mixture, it is important to consider the contribution of thermal
effects, if there are any. In theory, the pulses applied are too short to allow the ions to
respond to the field, leaving the gas temperature cold relative to the electron temperature.
However, any thermal effect is important because chemical reaction rates, k, have an
exponential dependence on temperature: =exp(−
) (Arrhenius Law). Here, A is
the pre-exponential factor and E
a
(kJ/mol) is the activation energy. This strong
temperature dependence of the reaction rates has a significant impact on combustion
(Wang, 2008).
21
Chapter 2: Pulse Generator Design
2.1 Introduction
The design of pulsed power systems is critical to the advancement of transient
plasma ignition technology. An efficient pulse generator design is matched to the
impedance of the load, which in the case of TPI, is a gap with a fuel-air mixture in
between. The pressure and gap size determines the breakdown conditions. The gap is
typically less than 5 cm and the pressures are typically less than 10 atm. Streamers
generated at the high-voltage electrode typically bridge the gap in less than 100 ns
(streamer velocities are on the order of 10
8
cm/s), so the pulse generators must produce a
high-voltage pulse with a width less than that to hold-off spark breakdown.
The fundamental purpose of a pulsed power system is to convert a low-power,
long-time input into a high-power, short-time output (Figure 9). Traditionally, research
in pulsed power has been driven primarily by the military, focusing on fusion and high-
energy physics, and has been dominated by large devices producing electric pulses of
microsecond to millisecond duration. However, the development of compact and short-
pulse technology in the nanosecond regime in recent years has enabled new and exciting
applications in combustion, as well as in medicine, agriculture, and environmental
studies.
22

Figure 9 – A pulsed power system accumulates energy over a relatively long period
(left), and then delivers the energy in a short period with high peak power (right)
(Mankowski, 2000).

The application of pulsed power to transient plasma ignition typically requires
pulses with amplitudes in the 10’s of kilovolts and pulse widths in the 10’s of
nanoseconds. To improve pulsed power system performance for transient plasma
ignition, it is critical to increase reliability and decrease system volume and weight since
combustion systems typically require components that can operate in high-temperature
and high-vibration environments where size and weight place further constraints on the
system (Gaudet, 2004).
2.2 Switches
In pulse generator design the switch is one of the most critical considerations. In
particular, the main problem in developing high-voltage (10-100 kV) short pulse (1 ns -
100 ns) and high-repetition rate (up to 1 kHz) pulse generators for transient plasma
23
ignition is to choose a reliable and fast high-voltage switch (Simek, 2006). A switch can
be categorized by function, e.g. a closing or an opening switch, and by type, e.g. solid,
liquid, gas, or plasma. Each switch has different limitations, including opening and
closing times, maximum repetition rates, forward voltage drop, voltage hold-off strength,
and peak current ratings. Typical characteristics for different switch types used for high-
voltage pulsed power applications are shown in Table 1 (Schamiloglu, 2004), and the use
of these switches is discussed in the following sections.

Table 1 – A summary of various switch parameters for pulsed power applications
(Schamiloglu, 2004).
Switch type
Hold-
off
voltage
(kV)
Peak
current
(kA)
Forward
drop (V)
Spark Gap 100
10 to >
1000
20
Thyratron 125 20 150
Pseudospark 100 5 -100 200
Thyristor 1 to 9 1 to 50 2
MOSFET 1 0.1 V
DS

IGBT 7 1 3

2.2.1 Gas Switches
Gas switches are commonly used for high-voltage and high-power applications
because they have very high hold-off voltages and current capabilities. The gas between
the electrodes inside the switch acts as in the insulator while the switch is off, so the
breakdown voltage is determined by the dielectric strength of the gas and the gap
24
between the electrodes. The switch is typically electrically or optically triggered, which
causes a portion of the gas to become ionized, producing a plasma. The plasma allows
very fast switching times and the transport of very high currents, and the power
dissipation is relatively low. The most common gas switches for pulsed power
applications are spark gaps, thyratrons, and pseudosparks.
Spark Gaps
Spark gaps are a gas closing switch that are among the most commonly used
switches in laboratory pulsed power experiments because they are simple, inexpensive,
and have extremely high operating voltage and current (Burkes, 1979). A spark gap
typically consists of two electrodes separated by a gas (or another insulating medium
such as a vacuum, liquid, or solid). The switch is closed when the gap is significantly
overvolted, or by applying a trigger (electrical or optical) to a third electrode. The
disadvantages of a spark gap are that they have limited pulse repetition frequency
because of the necessary recovery of the gap, they can have significant jitter (variation in
delay time from shot to shot), and they have a limited lifetime due to the erosion of
electrodes (Simek, 2006).
Thyratron
A thyratron is a gas closing switch that can have hold-off voltages greater than
100 kV and can handle peak currents greater than 10 kA. Thyratrons provide a switch
with a sufficiently long lifetime at a reasonable cost for high-power applications (Simek,
25
2006). The basic thyratron consists of an anode, cathode, baffle, control grid, and gas
reservoir (Figure 10). The gas reservoir is filled with low-pressure (approximately 0.5
torr) hydrogen (H
2
) or deuterium (D
2
), and the cathode is heated to emit electrons into the
neutral plasma formed by the gas (Burke, 1979). An electrical pulse is applied to the grid
to cause the switch to close by ionizing the fill gas. Compared to the spark gap, the
thyratron has a much longer lifetime (billions of shots), however, a major disadvantage of
this switch is the hot cathode, which adds significant overhead to the pulse generator
design.

Pseudospark
A pseudospark switch is a gas closing switch similar to a thyratron in that a
plasma is formed in a gas that is around 0.5 torr, but the pseudospark differs from the
thyratron in that it uses a cold (hollow) cathode. A schematic of a pseudospark switch
made by Alstom that is used in a pseudospark based pulse generator applied in this work
(see Section 2.3.2) is shown in Figure 11. A heater is still used, but it is used to control

Figure 10 – Basic thyratron (Burkes, 1979).
26
the gas pressure in the reservoir, not to heat the cathode. The breakdown voltage of the
switch can be adjusted by changing the gas pressure in the switch. The limitation in the
peak current capability for both the thyratron and the pseudospark is in the cathode, and
the pseudospark was designed with the goal of increasing the peak current capability of
the thyratron. In the pseudospark, a hollow cathode discharge is formed, which results in
the self-heating of a thin cathode surface layer due to the field and the plasma. The
cathode layer reaches extremely high emission through both field emission and
thermionic emission (Hartmann, 1988). Peak currents can be greater than 100 kA. The
pseudospark switch used in this work is capable of switching 30 kA, with a rise time of 8
kA/ns. Similar to the thyratron, a major disadvantage of the pseudospark is the overhead
associated with heating, as well as the size of the switch itself.
27

2.2.2 Solid State Switches
Solid state switches are attractive for ignition applications because they eliminate
the overhead (e.g. heating) associated with a gas switch, reduce size and weight, and
increase reliability and lifetime. The drawback of solid state switches is lower hold-off
voltages and peak current capacities compared to gas switches, which place strict
requirements on the rest of the system to compensate for those limitations. The most
common solid state switches for pulsed power applications are thyristors, MOSFETs, and
IGBTs.

Figure 11 – Pseudospark switch diagram for the FS2000 (Alstom).
28
Thyristor
The thyristor is a three-terminal solid state closing switch that is commonly used
for high-voltage and high-current applications. By definition it consists of an npnp-
structure; it has four layers and 3 pn-junctions (Figure 12). This work is most concerned
with a particular type of thyristor known as a silicon-controlled rectifier (SCR), which is
a gate controlled thyristor (Blicher, 1976). In particular, this is the switch chosen for the
12 ns pulse generator developed for this work (see Section 2.4). It is able to carry very
high-currents with an extremely low voltage drop because it behaves like a pin-diode
during its on-state. It is inherently stable, and can be scaled easily to withstand higher
voltages and currents (Linder, 2006).

Figure 12 – Left: Basic structure of an SCR; Center: Equivalent model using
two bipolar transistors; Right: SCR circuit symbol (Linder, 2006).
29
The disadvantage of an SCR is that the flow of current cannot be turned off at an
arbitrary instant. Once a current is applied to the gate and the SCR is switched on,
internal feedback keeps the current flowing, and an applied current is no longer required.
This regenerative condition is known as latching. There are only two ways to interrupt
the current flow of a latched SCR; the current must drop below a certain minimum value,
known as the holding current, or the voltage must be reversed to reverse the direction of
the current flow.
MOSFET
The metal-oxide-semiconductor field effect transistor (MOSFET) is a four
terminal solid state closing switch that is commonly used for switching voltages less than
a kilovolt for applications that require fast switching times. MOS refers to the sequence
of materials of the device, where the top layer is the metal, the middle layer is the oxide,
(e.g. SiO
2
), and the bottom layer is the semiconductor (e.g. silicon) (Figure 13). This
work is primarily concerned with power MOSFETs, which are fundamentally the same as
the low-voltage MOSFET found in digital integrated circuits, the only difference is that
device geometries and doping levels in power MOSFETs are optimized for power
electronics applications (Linder, 2006). A power MOSFET is vertically integrated,
which maximizes the utilization of the silicon area and allows for higher hold-off
voltages.
30

Power MOSFETs are attractive for certain pulsed power applications for several
reasons. Power MOSFETs have a high gate input impedance, which allows for small
gate drivers with low power consumption; they have a high current turn-off capability;
they have intrinsic overvoltage protection; and they have fast switching speed and low
turn-off losses due to uni-polar current flow (Linder, 2006). The disadvantage of power
MOSFETs is their low hold-off voltage and current handling capability.
IGBT
The insulated gate bipolar transistor (IGBT) is a three terminal solid state closing
switch that is based on power MOSFET technology. When comparing the cross-sections
of the IGBT (Figure 14) and the power MOSFET (Figure 13), the only difference is that
Figure 13 – Left: Cross-section of an n-channel MOSFET; Right: MOSFET
circuit symbol (Linder, 2006).
31
the n-doped drain region of the power MOSFET is replaced by a p-layer in the IGBT.
IGBTs were originally introduced as a replacement for bipolar transistors (BJT) since
they combined the desirable low on-state voltage drops of a BJT with the high blocking
voltage capabilities and fast switching speeds of MOSFETs. IGBTs are commonly used
in medium- to high-power applications; they can have higher hold-off voltages than
power MOSFETs. While the IGBT is one of the most popular solid state switches
currently used, they have a slower turn-off time so they still cannot compete with the
switching speed of MOSFETs, or the hold-off voltages and peak current capabilities of
gas switches.

Figure 14 – Left: Cross section of a vertically-integrated IGBT; Right: IGBT
circuit symbol (Linder, 2006).
32
2.2.3 Magnetic Switches
Magnetic switches are based on the nonlinear relation between the magnetic field
H applied to a core made from amorphous metals (i.e., metallic glass) and the resulting
flux density B. The inductance of a winding around the magnetic core depends on the
current passing though the winding and can change very quickly. Magnetic switches can
operate in the 10 ns range with a repetition frequency in the kHz range with no moving
parts and no discharge. A disadvantage of using magnetic switches for high-power
applications is that they have an efficiency of 50–70% so the core requires cooling
because the saturation flux density is temperature dependent (Simek, 2006). Magnetic
switches will be discussed in some detail because they are used in for magnetic
compression (see Section 2.3.3) in the 12 ns pulse generator developed for this work (see
Section 2.4).
The basic operating principle of magnetic switching is that sufficient current is
driven through the winding on a magnetic core so that the applied field H produces a flux
density B in the core in excess of the core’s saturation flux, causing the inductance in the
winding to change from a relatively high value to a very low value (Smith, 2002). The
change in µ can be greater than three orders of magnitude, e.g. from around 1000 to
around 2. A typical BH loop, known as a hysteresis curve or hysteresis loop, is shown in
Figure 15. The change in flux density in a magnetic core resulting from a voltage pulse
V
p
is:

33
∆ =

=

eq. (3)

In eq. (3), N is the number of winding turns, A is the core area, and t
p
is the
duration of the voltage pulse. If the amplitude of the applied pulse is great enough, the
core will saturate at time t
s
:

=
∆
=
(
±
)
eq. (4)

In eq. (4), B
sat
is the saturation flux density for the core and B
r
is the residual flux
density in the core at the beginning of the pulse. It is desirable to maximize ΔB in order
to minimize the area of core material required for the switch, which is accomplished if
the core is initially at position a in Figure 15. If the core is initially at position a, the
slope of the loop is steep and the relative permeability µ
r
will be high, but if an applied
magnetic field causes to core to move from position b to position c, the slope is shallow
and µ
r
will be very low. The position of the core on the hysteresis curve can be
controlled by applying a negative bias current through the winding before the operation
of the magnetic switch.
34

The simple LC circuit in Figure 16 can be used to demonstrate how a saturable
magnetic core responds to both a positive and negative field H caused by the discharge of
capacitor C. Starting at position a (see Figure 15), the inductance L is very high so there
is very little current flowing through the circuit. As the current begins to rise the
magnetic flux in the core moves to position b on the hysteresis loop, and the core starts to
saturate. At position c, the core is fully saturated, the inductance L is very low and the
impedance drops significantly, resulting in a significant current rise. In this example
circuit, the current has a sinusoidal dependence so it will start to fall and then reverse.
The process repeats, starting with the magnetic flux at position d (negative flux),
saturating at position e and fully saturated at position f. The current will rise again
Figure 15 – BH loop (hysteresis loop) to describe magnetic switch behavior (Smith,
2002).
35
significantly, although in this condition it will be negative, and the flux returns to position
a. The final voltage, V(T), will be less than the original voltage, V(0), because of the loss
in the circuit resulting from being driven around a cycle of the hysteresis loop (Smith,
2002).

Magnetic switches are excellent for pulsed power applications because they have
a very long lifetime and are very reliable compared to gas switches, they can be operated
at high repetition rates (>1 kHz), and they can allow pulse rise times comparable to gas-
switches. Their biggest disadvantage is that they must be cooled in high-power
applications, and that core reset must be monitored carefully to ensure pulse consistency.

2.3 Pulse Generator Architectures
A typical pulsed power system consists of a power source (e.g. HV DC power
supply), an energy storage device (e.g. capacitor or inductor), a pulse forming network,
Figure 16 – Variation of voltage and current in the LC circuit shown left (Smith,
2002).
36
and the load (e.g. gap). The application of pulsed power to transient plasma ignition
requires pulses with amplitudes around 30-90 kV and pulse widths below 100 ns. The
Marx bank, the Blumlein, and magnetic compression are pulse forming network
architectures that can be used to produce pulses that meet these requirements.
2.3.1 Marx Generators
Solid state switches are preferred to gas switches for the application of pulsed
power to transient plasma ignition, but as discussed in Section 2.2, solid state switches
have much lower hold-off voltages than gas switches. Marx generators can be used to
increase the operating voltage of a pulse generator using switches with lower hold-off
voltages by using multiple capacitors that are charged in parallel and discharged in series
through multiple switches (Marx, 1925). The Marx generator can produce a voltage gain
roughly equal to the number of capacitors used and it can be operated at high-repetition
rates. It is one of the most useful and most common architectures in high-voltage and
high-power pulsed power applications.
The basic circuit for a three-stage Marx generator (or Marx bank) is shown in
Figure 17. The capacitors are initially charged to a potential V through the charging
resistors R, where R is large enough that it can be ignored once the circuit is operated.
Before the switches S
1
– S
3
are triggered, the potential at points A, C, and E are V and the
potential at points B, D, and F are 0. When the switches S
1
– S
3
are triggered, the
potential at point B rises to V and the potential at point C rises to 2V, then the potential at
point D rises to 2V and the potential at point E rises to 3V, and finally the potential of the
37
load resistor R
l
rises to 3V. In practice, the operation is more complex since the triggers
to each switch must be isolated and there is stray capacitance and inductance in the
system. The leading edge of the pulse waveform is primarily controlled by the
inductance of the Marx generator (not shown), as well as the series resistance and the
capacitance of the load C
l
(also not shown). The falling edge of the pulse is controlled by
the capacitance of the Marx C and the load resistor R
l
. A peaking capacitor and switch
can be used to reduce the rise time if needed.

2.3.2 Blumlein
A Blumlein is a type of pulse forming line that solves the major problem of a
single pulse forming line, which is that in a single pulse forming line the voltage
delivered to a matched load is equal to half the voltage to which the transmission line is
charged. A Blumlein is basically two equal length transmission lines that are charged in

Figure 17 – Basic circuit of a three-stage Marx generator (adapted from Smith, 2002).
38
parallel and discharged in series (Figure 18). If the Blumlein is properly terminated it
can produce an output voltage that is equal to the charging voltage. It is popular for high-
power pulse generator applications because of its simplicity and reliability.

Figure 18 – Circuit diagram of the basic Blumlein pulse forming line (Smith, 2002).

The operation of the Blumlein architecture is illustrated in Figure 19. The pulse
forming action is initiated by the closure of the switch and the end the line 1 (see Figure
18) is shorted, which makes the reflection coefficient at that end of line 1 equal to -1.
The reflection coefficient at the end of line 2, which is open, is equal to +1. The
reflection coefficient where line 1 meets the load is loaded by the combination of the load
in series with the impedance of line 2. For the Blumlein to operate correctly, the load
impedance must be equal to twice the characteristic impedance of line 1 and line 2, so
line 1 is loaded by an impedance of 3Z
0
and the reflection coefficient of the load is:

=

=
eq. (5)

39
The potential at the end of line 1 is reduced to zero and a step amplitude -V
propagates along the line toward the load when the switch is closed, discharging the line
as it goes. When the step reaches the load it is partially transmitted and partially reflected
(after a transit time delay δ) with a reflection coefficient of +1/2, which launches a step of
-V/2 back into line 1 propagating toward the switch. The step has an amplitude equal to
the sum of the incident step V
+
= -V and the reflected step V
-
= -V/2, which is -3V/2.
Considering the load and line 2 as a simple voltage divider yields a step of amplitude -V/2
being launched into line 2 with a potential V across the load. This step now moves along
line 2, discharging the line to a potential of V/2, and reflects at the open of line 2 with a
reflection coefficient of +1. This fully discharges line 2 as the -V/2 step moves back to
the load. The step that was launched in line 1 charged the line to -V/2 and then reflected
with a coefficient of -1 to discharge line 1 as it propagated back toward the load. As the
two steps arrive at the load, they eliminate each other and the pulse on the load terminates
with a duration of 2 δ (Smith, 2002).

40

Figure 19 – Blumlein operation after the switch in Figure 18 closes (Smith, 2002).
41

In this work, a 50-100 ns (depending on the load) line type pseudospark switched
pulse generator is used to generate transient plasma in several applications. The pulse
generator is based on the Blumlein architecture, and is capable of delivering up to 90 kV
(Figure 20). Instead of using transmission line as discussed above, capacitors are used to
make the pulse forming network. LC circuits do not combine as cleanly as transmission
lines, so the shortest possible critically damped (non-ringing) output pulse delivers a
maximum output voltage of 64% of the charging voltage (Wang, 2005). For most
electrodes in this work, the pulse width delivered to the gap is 85 ns, so this pulse
generator is often referred to as the 85 ns pseudospark pulse generator in this dissertation.

2.3.3 Magnetic Pulse Compression
A sequence of multiple magnetic switches and capacitors can be used to
successively reduce the rise time and compress the width of a pulse propagating on a line.
This is the type of pulse forming network used in the 12 ns pulse generator developed for
Figure 20 – Pseudospark switched pulse generator based on Blumlein architecture
used for transient plasma ignition experiments.

42
this work (see Section 2.4). Magnetic compressors, also referred to as Melville lines,
increase the peak power of the pulse by using magnetic switches to transfer energy
between capacitors (Melville, 1951). Each capacitor is charged and discharged faster
than the previous capacitor if the inductance of the switches is successively reduced
along the line, compressing the pulse width. The pulse amplitude can also be
significantly increased if the value of capacitance is also successively reduced along the
line (Smith, 2002).

A basic magnetic pulse compression configuration is shown in Figure 21. C
0
is
initially charged to potential V(0), and the switch is closed at t(0), allowing C
0
to
discharge through inductor L
0
into capacitor C
1
. The potential on C
1
rises until L
1

saturates, as discussed in Section 2.2.3. Once L
1
saturates, C
1
discharges rapidly into
capacitor C
2
, and again, the potential rises in C
2
until L
2
saturates. Assuming L
2
has a
lower saturated inductance than L
1
, then C
2
will discharge even more quickly into
capacitor C
3
. This process continues until the last capacitor, C
n
, is discharged into the
load L. Voltage plots are shown for the basic magnetic pulse compressor circuit in
Figure 22. If the capacitors all have the same value, then the impedance falls along the
Figure 21 – Basic magnetic pulse compressor circuit (Smith, 2002).
43
line and significant current gain can be achieved. If the capacitors are reduced in value
sequentially along the line, considerable voltage gain can be achieved. The maximum
gain or compression per compression stage is around 5.

2.4 Development of a 12 ns Pulse Generator
A compact 12 ns magnetic compression based pulse generator was developed
using an SCR as a switch in order to decrease system size and weight, and to decrease the
pulse width compared to the 20 ns IGBT-based pulse generator previously used for this

Figure 22 – Voltage during the operation of a basic magnetic pulse compression
stage (Smith, 2002).
44
work (Tang, 2006). The advantage of decreasing the pulse width was shown in an
internal combustion engine experiment in collaboration with Nissan Research Center
(Cathey, 2007). In that experiment, because of the small gap (6.4 mm), using a 20 ns
pulse compared to an 85 ns pulse allowed a significant increase in the reduced electric
field E/n without causing spark breakdown, which increased the number of high-energy
electrons capable of dissociation and ionization and resulted in an additional 20%
increase in peak pressure in the engine cylinder.
The new pulse generator is capable of delivering a 65 kV, 12 ns (FWHM) pulse
(180 mJ) into a 200 Ω load. The amplitude is scalable while maintaining the same pulse
width down to an output voltage of 20 kV. The rise time of the pulse is approximately 5
ns. The pulse generator consists of three stages: a charging and energy storage stage, a
voltage doubling and magnetic compression stage and a diode opening switch (DOS)
stage (Figure 23).

Figure 23 – Schematic for a 65 kV, 12 ns compact pulse generator.

45
An inexpensive SCR was chosen to replace the relatively expensive IGBT as the
switch. SCRs can handle high peak currents, but their hold-off voltage is typically lower
than that of IGBTs, which subsequently required that the charging voltage be reduced
from 1 kV to 500 V. The size of the system was reduced by over four times to a case
with a volume less than 6500 cm
3
, compared to a case with a volume of 28000 cm
3
for
the 20 ns DOS pulse generator and pseudospark pulse generators (Figure 24).

Figure 24 – The 90 kV, 85 ns pseudospark pulse generator (left), and the 65 kV, 12
ns solid state pulse generator (right).

In the first stage of the pulse generator, a DC power supply or resonant charging
circuit is used to charge the 6.8 µF capacitor (C
1
) to 500 V or less, storing up to 850 mJ.
The SCR triggering circuit consists of a 1:1 gate drive transformer (T
in
) and capacitor
(C
in
) to protect the Trigger IN supply from current reflections. When the SCR is turned
on by applying a 5 V, 50 μs TTL pulse across the gate and the cathode, the stored energy
is transferred across a saturable transformer (T
1
) into the second stage and the voltage is
increased by a factor of 20. To make sure that T1 is initially biased at the negative
46
saturation flux on its BH curve (see Section 2.2.3), 1 A DC current produced by an
inexpensive internal +3.3 V power supply is passed through an auxiliary winding to
reverse the magnetization field. The second stage consists of a voltage doubling and
magnetic compression system.
In the second stage, a capacitive voltage doubling methodology was used to
compensate for the loss in amplitude caused by decreased input voltage compared to the
previous 20 ns IGBT-switched pulse generator (Tang, 2006; Rukin, 1995). Capacitors C
2

and C
3
are charged to 10 kV in parallel. When transformer T
1
saturates, the polarity of
C
3
momentarily reverses and C
2
and C
3
discharge in series through L
1
. The voltage
polarity is reversed and the amplitude is doubled to approximately 20 kV. The diodes
(D
1
– D
3
) in this stage can also be replaced with a saturable magnetic core used as a
magnetic switch to improve the pulse generators’ cooling characteristics (Lyubutin,
1997). A nine turn saturable inductor (L
1
) after the diodes is used to compress the pulse
width.
The final stage of the pulse generator consists of transformer T
2
to double the
voltage, and a DOS to sharpen the pulse and increase the amplitude. Transformer T
2

reverses the polarity of the pulse so that there is a positive output pulse. The charge
current of capacitor C
4
passes through diodes D
4
– D
12
so that they are forward biased.
When T
2
saturates, C
4
discharges through the reverse-biased DOS. When the current
reaches a value close to the maximum, the energy is transferred to the inductance of the
T
2
secondary winding. When the DOS cuts off this current, a 12 nanosecond pulse is
generated at the device output. Typical output voltage and current waveforms are shown
47
in Figure 25. Current and voltage were measured using a fast rise-time Pearson
Electronics current transformer (Model 6223) and a custom 1:1000 high-voltage divider
and were recorded using a 1 GHz Tektronix digital oscilloscope (DPO 4104).

Figure 25 – Output voltage and current of the 65 kV, 12 ns pulse generator. The load
is 200 Ω.

To increase the repetition rate and decrease the input voltage, a rapid charging
circuit (resonant LC charging circuit) was developed, which decreased the input voltage
by more than a factor of two (225 V input required instead of 500 V), and allowed
repetition rates up to 1 kHz (Figure 26). For this circuit, C
1
is 55 uF, HV
in
is 225 V, L
1
is
200 µH, and V
out
is 500 V. When the IGBT switch is closed the LC circuit rings up to
48
the DC charging voltage of 225 V. The time to charge capacitor C
1
to 225 V is
determined by the quarter-period of oscillation √ /2. The diodes prevent the
downward LC oscillation, forcing the inductor to deliver its remaining energy as forward
current into the capacitor, doubling the charge stored and therefore the voltage across the
input capacitor to the pulse forming network (C
1
in Figure 23). This doubles the total
time to charge the capacitor, bringing it to a half-period of oscillation. The capacitance
C
1
is determined by the initial capacitance needs of the pulse forming network, and the
inductance L
1
is chosen to optimize peak current and charging time.

Figure 26 - Rapid Charging Circuit

The new 12 ns pulse generator extended the capabilities of pulse generators
already being used for this application at USC (Table 2). It is the most compact pulse
generator on the list, it delivers the lowest energy per pulse, and it has the shortest pulse
width. It was the primary pulse generator used in this work. Typical waveforms from the
49
four pulse generators used for transient plasma ignition at USC in the past and present are
shown in (Figure 27).

Table 2 – Characteristics of pulse generators at USC used for TPI.
Pulse generator by
switch type (year
implemented)
Peak
voltage
(kV)
Pulse
width
(ns)
Max
energy per
pulse (mJ)
Thyratron (1998) 50 150 1000
Pseudospark (2003) 90 85 1500
IGBT (2006) 60 20 300
SCR (2008) 65 12 200

Figure 27 – Typical waveform into a 200 Ω load from pulse generators at USC used
for TPI.

50
Chapter 3: Application of Transient Plasma
3.1 Introduction
Transient plasma was applied in experiments in pulse detonation engines (PDE),
in an internal combustion engine (ICE), and in a constant volume reactor. Primarily, two
different pulse generators were used; the 12 ns SCR-based pulse generator that was
developed for this work (see Section 2.4), and the 85 ns pseudospark-based pulse
generator (see Section 2.3.2). Here, transient plasma ignition system refers to the pulse
generator, the transmission cable, and the electrode. Significant reduction in ignition
delay and an increase in flame speed under a variety of temperature and flow conditions
for ethylene (C
2
H
4
) fuel were demonstrated in PDEs while reducing the size and weight
of both the pulse generator and the electrodes. It was shown that in both flowing and
quiescent conditions, a shorter pulse with an order of magnitude less energy per pulse
than previously applied could be used to achieve the same ignition delays. It was also
shown that small, spark plug sized electrodes could be used with a 12 ns pulse generator
to initiate ignition in an ICE.
When discussing combustion experiments, the fuel-air mixture is described as
being fuel lean, stoichiometric, or fuel rich. Equivalence ratio, ϕ, is used to define these
conditions:
51
=
( / )

( / )

eq. (6)
where
= < 1 fuel lean
1 stoichiometric
> 1 fuel rich

A complete combustion reaction occurs when the fuel-to-oxygen molar ratio is
stoichiometric. In this condition, there is no excess oxygen; all of the carbon in the fuel
is oxidized to CO
2
and all of the hydrogen in the fuel is converted to H
2
O. The
stoichiometric ratio for any fuel C
m
H
n
can be calculated using the complete
stoichiometric reaction, which yields a stoichiometric ratio of 1/( +
4
) (Wang,
2008). If there is excess oxygen the equivalence ratio is fuel lean, and if there is a lack of
oxygen the equivalence ratio is fuel rich.
3.2 Transient Plasma Ignition in Pulse Detonation Engines
One promising application of transient plasma ignition is pulse detonation
engines. The PDE provides impulse through pulsed operation and through fuel
detonation. Thrust is provided by a supersonic shock wave that is generated by the
exhausting gases. The advantages of this engine methodology include intrinsic
efficiencies, design simplicity, and operation at subsonic and supersonic speeds. PDEs
are the only engine platform that can theoretically efficiently operate from a dead stop to
hypersonic speeds (Mach 4+) (Figure 28) (Bussing, 1997). For hypersonic speeds, the
PDEs primary competitor is the ramjet, which needs a booster to reach speeds where it
52
becomes operational. Transient plasma is attractive as a technology for the ignition of
PDEs because it results in reductions in ignition delay, which allows for higher repetition
rates and therefore greater thrust, and because it has been shown to ignite leaner mixtures,
which offers lower specific fuel consumption, high-altitude operation, and reduced NOx
emissions (Bozhenkov, 2003; Starikovskia, 2004).

Figure 28 – Projected efficiency of a variety of airborne engines (Bussing, 1997).

The detonation of a fuel-air mixture is fundamental to PDE operation and
provides it with a thermodynamic advantage over a ramjet’s deflagration combustion
process. A detonation wave is classically described as a one-dimensional supersonic
53
combustion wave coupled to a supersonic shock wave, followed by a reaction zone
propagating at steady velocity (Ciccarelli, 2006). Chapman and Jouguet were the first to
present a theory describing detonation waves, and created the CJ (Chapman-Jouguet)
theory, which considers a detonation wave as a discontinuity with an infinite reaction rate
(Fickett and Davis, 1979).
Using CJ theory, it is possible to calculate the detonation velocity if the gas
mixture and the initial pressure and temperature are known. Depending on the mixture
energy content, the combustion wave propagates at approximately 1,500 to 2,000 meters
per second in a nearly constant volume condition. In comparison, deflagration produces
a relatively slow, subsonic combustion wave with a small pressure rise. The combustion
wave propagates at approximately 1 to 10 meters per second in a nearly constant pressure
condition, and the transport of thermal energy and reactants govern the flame front
propagation rate (Hackard, 2007).
There are three ways to initiate a detonation; through direct initiation, shock
induced initiation, and deflagration-to-detonation transition (DDT) process. In this work,
only DDT is considered. For direct initiation, a high-energy ignition source is required
(e.g. for stoichiometric propane-air, 300 kJ is required for direct initiation) (Bull, 1978).
A lower energy source can initiate detonation through flame front acceleration in a
sufficiently long tube. Since DDT length and time are critical for PDE applications, a
key issue is the reduction of the transition length and time in detonation tubes under the
minimum energy of initiation (Rakitin, 2008). In addition to other advantages of TPI
discussed in Chapter 1, TPI has been employed in PDE’s as a low-energy source that can
54
produce detonation waves in conditions under which traditional spark ignition cannot
(Rakitin, 2008; Singleton, 2009).
The DDT process begins with ignition and flame propagation. As the gas burns,
it expands and pushes unburned gas ahead of the flame front. A turbulent boundary
layer grows and significantly increases the burning rate. The enhanced burning rate
creates an increase in the flow velocity ahead of the mixture due to the expanding gases,
and unsteady compression waves ahead of the mixture result in an increase in
temperature that accelerates reaction rates. A shock front is formed once the compression
waves coalesce, and detonation begins when there is a sudden appearance of explosion
centers in the shock front (Figure 29) (Kentfield, 2002; Ciccarelli, 2006). Once
detonation begins it will propagate in a relative steady self-sustaining wave at CJ
conditions. Obstacles (e.g. Schelkhin spiral or ramps) in the tube are often needed to
increase turbulence to successfully create a detonation wave.

55

Figure 29 – Schematic showing the detonation front structure consisting of multiple
shock waves and reaction zones (Ciccarelli, 2006).

The operation of a simple PDE is shown in Figure 30 (Hackard, 2007). A cycle
begins with the injection of a fuel-oxidizer mixture upstream of the main tube (1). This
can be done either with valves, such as in the PDE used at Wright Patterson Air Force
Base (WPAFB), or without valves, such as in the PDE used at the Naval Postgraduate
School (NPS). The oxidizer is typically air. The combustion tube fills with the fuel-air
mixture (2). The ignition event, either a spark discharge or transient plasma discharge,
occurs and creates the initial deflagration combustion wave (3). The deflagration wave
accelerates though the unburned fuel and transitions to a detonation wave as it travels
down the tube, possibly with the help of a turbulence creating device such as a Schelkhin
spiral (4). The detonation wave exits the tube (5). A series of rarefaction waves form
and begin to travel back down the combustion tube (6). The rarefaction waves continue
56
to propagate back down the combustion tube, reducing the elevated pressures and
purging the combustion products (7). Once the combustion products are purged, the
cycle is complete and the PDE is ready for the next cycle.

Figure 30 – PDE operating cycle (Hackard, 2007).
3.2.1 Measurement of Ignition Delay in a PDE at NPS
The new 12 ns pulse generator, combined with transient plasma igniters
comparable to traditional spark plugs, were successfully tested in the PDE at the NPS in
Monterey, CA. The new TPI system had smaller electrode geometry and delivered an
57
order of magnitude less energy than in previous experiments (Wang, 2005). The ignition
delays of two transient plasma ignition systems were measured and compared; one
system with the 85 ns pseudospark pulse generator and threaded anode, which discharges
a pulse of 800 mJ (henceforth called the 85 ns TPI system in this section), and the other
system with the 12 ns solid state pulse generator and transient plasma igniter which
discharges a pulse of 80 mJ (henceforth called the 12 ns TPI system in this section).
Electrode Design
The electrode used with the 85 ns TPI system in the PDE at NPS consisted of a
stainless steel rod, threaded for field enhancement, and insulated by a ceramic (Macor)
(Figure 31). A steel coaxial cylinder was grounded and acted as the return path for the
discharge current, and created a large discharge gap (36.6 mm). Previous Macor
insulators have failed due to mechanical vibration and were not easily replaced, so a
smaller, more robust, and easily replaceable solution was desired.
The electrode system used with the 12 ns solid state pulse generator was similar in
size and form to a commercial spark plug, but 20 mm of the grounding shell was
removed, eliminating the grounding arm and leaving exposed 18.6 mm of the insulator
and a 2.6 mm long smooth anode (Figure 31). A 4.5 mm long threaded rod (2.1 mm
diameter) was attached to the end of the anode to increase the length. It was threaded for
electric field enhancement. A perforated steel pipe was used as the cathode. Initial tests
were performed with a porosity of 25% on the cathode. This provided a lower velocity
region inside the discharge volume and enabled sufficient development of a flame front
58
in the tested fuel-air mixtures, which subsequently produced successful DDT in the main
detonation combustor after passing over turbulence generating devices.

Figure 31 – Top: A section view of the transient plasma igniter and grounded screen
for the 12 ns TPI system; Bottom: threaded rod and grounded screen for the 85 ns TPI
system. Dimensions are in mm.

59
Experimental Setup
The tests performed at NPS investigated the variation of ignition delay times as a
function of fuel-air equivalence ratios in a PDE configured to simulate Mach 2.5 cruise
conditions at an altitude of 40,000 ft. At this velocity, stagnation temperatures near 500
K are required. The NPS configuration is a direct-connect test rig, therefore air heating is
required. The airflow was heated using a hydrogen fueled vitiator system, which utilizes
oxygen to rebalance the oxygen content of the heated air up to the standard mole fraction
of 21%. A short distance (less than 1 m) downstream of the vitiator, a stagnation plenum
is used to feed the four mixing arms in which fuel is injected and allowed to mix with the
heated airflow. Either liquid or gaseous fuels or a combination of these can be injected.
In these investigations, gaseous ethylene (C
2
H
4
) was used. Gaseous C
2
H
4
fuel is often
used in PDE experiments because it is easily detonable and because liquid fuels
envisaged for practical PDEs are expected to chemically decompose to C
2
H
4
or C
2
H
4
-like
species before detonation events (Ma, 2002). The ethylene-air mixtures were introduced
into the PDE as shown in Figure 32, where a portion of it passes through the porous
cathode and fills the TPI discharge volume with much lower velocities than the
surrounding flow. It is the lower velocities that enable a fully developed flame-front to
evolve, and it evolves rapidly in conjunction with the volumetric discharge of the TPI
system. The inlet Reynolds number was Re ~ 200,000 and the temperature of the fuel air
mixture at the entrance to the PDE was 490 K.
60

Figure 32 – Top: The PDE at NPS with the compact igniters installed; Bottom:
A section view of the TPI PDE system.

61
Once evolved, the flame-front must still go through DDT within the main
detonation combustor, where Schelkhin-type obstacles strategically placed along the
main detonation combustor enable DDT to occur within 1 m, in a 75 mm diameter
combustor. After detonation, a 2 ms or longer purge process is allowed prior to the start
of the next cycle in order to prevent the hot products from pre-igniting fresh ethylene-air
mixtures. Although the flow rates can be varied from 0.15 kg/s to greater than 0.5 kg/s,
all tests here were performed at 0.35 kg/s, which allowed PDE operating frequencies of
40 Hz without any adverse behavior.
Instrumentation was composed of high-speed Kistler Type 603B1 dynamic
pressure transducers with matching charge amplifiers. These respond within 2 µs and
their signals were digitized at a 1 MHz sampling rate using 14-bit National Instruments
Analog-Input DAQ boards mounted on a PXI system. In addition, laser-based absorption
spectroscopy was used to measure the ethylene content in the mixture, which was then
used to calibrate set equivalence ratios. A CW HeNe laser tuned to the infra-red 3.39 µm
wavelength was used with a PbSe-based diode detector. As the ethylene-air mixture
flowed through the engine, the laser beam intensity decreases and this intensity decrease
is proportional to the concentration of ethylene in the mixture from Beer’s Law. These
measurements were performed only during non-ignition tests as blast waves in the test
cell formed from exiting detonation waves could damage the HeNe. Previous
investigations in collaboration with Stanford University have demonstrated fuel
concentration (among others, such as temperature, water concentration, etc.)
measurements during hot-tests using fiber launched tunable diode lasers (Klingbeil, 2007;
62
Hanson, 2006; Klingbeil, 2006). Lastly, type-K thermocouples were used throughout to
determine airflow inlet temperatures and to control vitiator operating temperatures.

Results
The results presented are ignition delays for a varying equivalence ratio and a
fixed mass flow rate of 0.35 kg/s (Figure 33). The error bars represent the standard
deviation of a minimum of 10 data points (with usually around 40 data points) for each
equivalence ratio tested. Both transient plasma ignition systems produced similar
ignition delay times near stoichiometric conditions despite the large differences in the
energy delivered and discharge volume. The 12 ns TPI system produced slightly longer
average ignition delay times than the 85 ns TPI system when the mixtures were far away
from stoichiometric conditions, and the differences in average ignition delays were most
dramatic at very lean conditions ( ϕ < 0.6). Both TPI systems were able to produce
successful DDT in the main combustor. In comparison, a traditional spark plug ignition
system will not produce DDT in this setup without additional oxygen.
63

Figure 33 – Results of ignition delay in a PDE using a 12 ns TPI system (80 mJ) and
an 85 ns TPI system (800 mJ). Spark ignition delays are not shown because they are
much greater (around 8 ms).
Discussion
Ignition delay is defined here as the time from the ignition pulse to a pressure rise
due to heat release (Busby, 2006). Pressure transducers were located on the main
detonation combustor at three locations. Location 1 was directly at the TPI discharge
location, which was used to measure pressure rise from the combustion event – this was
used to determine ignition delay times. The response time of the pressure transducer (2
µs) was much faster than the ignition delay times (ms). The ignition delay time was
taken between the rise of the high-voltage discharge event to the time at which the high-
64
speed pressure transducer at Location 1 registered a 1 bar pressure increase. This criteria
possesses a local dp/dt value that exceeds that of (Busby, 2006), thus making the ignition
delay definition for the setup used in this experiment more conservative. It is difficult to
compare the two rigs since the PDE in the cited work is valved, while the PDE used in
this experiment has both forward and aft venting, thus making the pressure rise more
difficult to achieve and reinforcing the statement that our definition is more conservative.
MATLAB was used to analyze the high-frequency transducer signals and TPI discharge
signals recorded in LabView.
Pressure transducer Locations 2 and 3 were set near the detonation combustor exit
and were placed 100 mm apart from each other in order to measure detonation wave
pressures and times of arrival so that detonation velocities could be verified. This served
two purposes; indication of successful DDT transition and quantification of detonation
velocities for the given equivalence ratios. However, the latter is more difficult to
analyze because under- and over-driven detonation states occur near detonation transition
location, which occurred within the 100 mm spacing between transducers for equivalence
ratios away from 1.0. Therefore, detonation velocities vs. equivalence ratios are not
reported here. At lean and rich equivalence ratios, a longer tube would be required to
obtain steady-state detonation velocities.
Since the geometries of the electrodes in the two ignition systems used in this
experiment were different, it is relevant to address the possible influence of
hydrodynamic characteristics on the ignition delay time measured values. Although the
shrouded volume around the electrode was larger in the 85 ns TPI system and likely
65
resulted in a greater volume of excited reactants than in the 12 ns TPI system, convection
of the flow over such time difference of approximately 60 ns would only correspond to a
spatial variation of 10's of microns. The improvements in the ignition delay times for the
rich and lean conditions were more likely due to the increased size/amount of the initial
flame kernels for the 85 ns TPI system.
When comparing the measured ignition delay times between the two TPI systems,
it is important to consider the difference in total energy deposited (~800 mJ and ~80 mJ
for the 85 ns and 12 ns TPI systems, respectively), as well as the differences in the
discharge volume and the reduced electric field. The discharge volume of each system
was calculated using the length and diameter of the anode and the cathode. Streamers
were present throughout this volume. The 85 ns TPI system had a discharge volume of
58.4 cm
3
and the 12 ns TPI system had a discharge volume of 0.3 cm
3
. It is therefore
interesting to note that there were very small differences in ignition delay times between
both systems at near stoichiometric conditions ( ϕ near 1.0).
While the discharge volume with the 12 ns TPI system was about 200 times less,
the energy density delivered was about 20 times more. The energy density deposited in
the 85 ns TPI system was approximately 15 mJ/cm
3
, while the energy density deposited
in the 12 ns TPI system was approximately 270 mJ/cm
3
. The increased maximum
electric field for the 12 ns TPI system (approximately 250 kV/cm compared to 150
kV/cm) resulted in an increase in the reduced electric field E/n (estimated 1100 Td
compared to 650 Td. This theoretically increases the number of energetic electrons
capable of disassociating hydrocarbon molecules, producing more active free radicals in
66
the ignition volume. These approximations do not take into account the voltage fall
across the plasma sheaths, and the anode was assumed to be smooth. The experiments
described used threaded anodes which further enhanced the electric field near the anode.
It has been reported that there is a threshold dependence of ignition delay times on
the reduced electric field, and that an increase in E/n beyond approximately 400 Td will
not result in any further decrease in ignition delay (Figure 34) (Aleksandrov, 2001;
Bozhenkov, 2003). While ignition delays were similar for the two systems at
stoichiometric conditions, the 12 ns TPI system produced longer ignition delays in very
lean conditions. Therefore, when considering the differences in ignition delay using the
two systems in lean mixtures, the effective area of the igniters (the length of the anodes)
must also be considered.

67

Previous study has shown that ignition occurs near the anode, despite the
volumetric distribution of the streamers (Cathey, 2008). The electric field is the highest
near the anode and therefore the highest densities of radicals are found there. In the
geometry used in this experiment, the anode was seven times shorter (7.1 mm compared
to 50.7 mm) than the previous ignition system, resulting in fewer ignition kernels and
contributing to the slower ignition delay times further away from stoichiometric
conditions. The ethylene-air mixtures become less sensitive (less energetic) when the
equivalence ratio is far away from 1.0, even more so at lean conditions, thus the

Figure 34 – Ignition delay time vs. reduced electric field in a stoichiometric H
2
-air
mixture at 1 atm. T = 1000 K. Total energy release E = 4 × 10
-3
J/cm
3

(Aleksandrov, 2001).
68
differences in ignition delay times become more significant, as expected from
conventional flame propagation theory (Kuo, 2005).
Conclusions
Transient plasma generated by short, high-voltage pulses, has consistently shown
reductions in ignition delay times and broader lean flammability capability relative to
traditional spark ignition. A new 12 ns compact solid state pulse generator for ignition
applications was tested using ignition geometry comparable to a spark plug. This system
was compared to the 85 ns pseudospark pulse generator and threaded rod electrode
configuration in a pulse detonation engine. The ignition delay times achieved using the
solid state pulse generator and small igniter were comparable with those using the
pseudospark pulse generator and threaded rod throughout, except at very lean conditions
( ϕ < 0.6). The pulse generator is attractive for mobile applications because of its small
size, simple operation, and reliability.

3.2.2 Flame Propagation in a PDE at WPAFB
An investigation of flame propagation initiated by transient plasma was
performed in the PDE at WPAFB with the 12 ns TPI system in collaboration with the Air
Force Research Laboratory (AFRL) and the University at Buffalo. Schlieren
photography was used to compare the flame speeds of spark and transient plasma ignited
69
ethylene-air in a clear section of a valved PDE, using a new small gap electrode based on
a standard 10 mm, ¾ inch reach spark plug (Figure 35).

Figure 35 – The electrode used in the PDE at WPAFB. Dimensions are in mm.

Schlieren photography is commonly used to look at the flow of fluids of varying
density. In concept, it is simple; it relies on the change of refractive index with density.
A light source is collimated with a lens coming from behind the target object, and a knife
edge is placed at the focal point blocking half the light. In flow with a uniform density,
the photograph will just be half as bright, but any change in air density causes part of the
beam to be refracted and to fall above or below the knife edge. That part of the image
will appear brighter or darker than the background (Settles, 2001).

Experimental Setup
The PDE at WPAFB is different from the PDE at NPS because it uses valves to
control the injection of fuel into the main combustor. It is based on a General Motors
Quad 4 engine head and normally uses four 5.2 cm internal diameter, 1.85 m long
70
detonation tubes (Figure 36). In this experiment, only one tube, with a clear section
made of two Plexiglas windows, and two steel cross-bars was used. The electrode was
installed into the steel cross-bar vertically in the clear section of the tube so it could be
seen clearly (Figure 37). The fuel-air manifold pressure and the camshaft/valve timing
in the engine head were used to regulate the fill volume of the fuel-air mixture (Busby,
2006). The fuel-air mixture (C
2
H
4
-air) approximately filled the entire tube before
ignition.

Figure 36 – Head-on view of the PDE engine block, with four fuel-air valves, and a
transient plasma igniter (left) and a traditional spark plug igniter (right).

For the Schlieren setup, a compact white light source was enclosed in a box with
an aperture to control the intensity of the light. To collimate the light source, a concave
mirror was placed about 1 m away from the light source. The collimated light was then
directed to a plane mirror placed at a 45º angle to the PDE tube. The light then passed
through the clear test section of the PDE and onto a second plane mirror on the other side,
71
which directed the light toward a second concave mirror. This mirror focused the light to
a point where the knife edge was placed. Light diverging from the knife edge passed
through a convex lens which collimated the light and formed an image on the sensor of
the high-speed camera.

The 12 ns pulse generator was used to generate transient plasma in the 4.35 mm
gap of the custom electrode and a custom shielded coaxial cable was used to transmit the
pulse. The amplitude of the pulse was varied from 20 kV to 50 kV, delivering 30 mJ and
80 mJ pulses, respectively. A standard MSD ignition system was used to deliver a 105
mJ pulse to the spark gap of a standard spark plug for comparison.
Results
High-speed Schlieren photography was used to create video of flame propagation
comparing a spark ignition system to a transient plasma ignition system with varying
pulse amplitudes (Figure 38). It was found that the initial flame speeds in the 80 mJ
transient plasma ignited case were two times faster than that of the spark ignited case.
This is a promising result, since this is the smallest transient plasma igniter tested with

Figure 37 – Setup of the PDE at WPAFB for Schlieren photography in a clear
section with a vertically mounted Spark or TPI plug.
72
this pulse generator to date, with a gap size of 4.35 mm. The compact transient plasma
igniter fits in a standard automobile internal combustion engine (ICE) without any
modification, thus the potential utility of the system is apparent.

Figure 38 – Flame propagation of stoichiometric C
2
H
4
-air in a clear section of a PDE
tube. The images were taken 10 ms after ignition.

73
3.3 Transient Plasma Ignition in a Constant Volume Reactor
A stainless steel constant volume reactor was used to investigate different
configurations of transient plasma ignition systems in quiescent fuel-air mixtures. The
purpose of these experiments was to increase the understanding of transient plasma
ignition and to produce guidelines to design the most effective transient plasma ignition
system for different applications. The 1.73 L cylindrical stainless steel static combustion
reactor had a length of 229 mm, an internal diameter of 98 mm, and a wall thickness of
2.6 mm (Figure 39). Flanges at the end of the cylinder provide gas inlet and exit lines
and a port for a high-speed dynamic pressure transducer (Omega DPX101) and the
electrode. A pressure transducer port was located directly opposite the electrode in the
center of the flange. This flange could also be replaced with a flange with a quartz
window so that high-speed photography could be used to observe the flame propagation
from the electrode. The high-speed camera used was an IDT X-Stream VISION XS3
with a 55 mm f/1.2 lens. The voltage and current of each discharge were measured using
a 1:1000 high-voltage divider (Northstar PVM-5) and a fast current transformer (Pearson
6223), respectively. The pressure, voltage, and current signals were recorded on a 1 GHz
Tektronix digital oscilloscope (DPO 4104).
74

The electrode consisted of a stainless steel anode and cathode in a coaxial
geometry (Figure 40). A custom shielded twisted-pair high-impedance (200 Ω) cable
connects both the 12 ns and 85 ns pulse generators to the electrode. This is necessary to
transmit the pulse with minimum attenuation and distortion. The anode was a stainless
steel rod, threaded (8-32) for field enhancement, and insulated by Teflon. A 1.25 mm
thick stainless steel coaxial cylinder, perforated to allow the flame to propagate more
quickly through the volume, was grounded and acted as the return path for the discharge
current. The electrode system was designed so that the anode and cathode could be
Figure 39 – Section view of the combustion reactor with removable cathode
to allow a change in gap size. Shown with a quartz window where a flange
with a pressure transducer port could be inter-changed.
75
changed, allowing the testing of different gap sizes, anode lengths, electrode materials,
and cathode perforation patterns. Unless otherwise specified, a 32.6 mm long anode was
used. The pulse width was varied by using two different pulse generators.

Figure 40 – Electrode for the combustion reactor. A high-voltage cable plugs in
from the right. The gap size shown is 6 mm. Dimensions are in mm.

The typical configuration of the chamber is shown in Figure 41. A 1 L stainless
steel cylinder was used as a mixing manifold (Figure 42). A vacuum pump was used to
evacuate the cylinder, which was then filled with the amount of fuel for the desired fuel-
air ratio, and then compressed air was added. Equivalence ratios were calculated based
on partial pressures, taking the current atmospheric pressure into consideration. The
volume of fuel-air mixture in the mixing manifold was enough for two runs in the main
combustion chamber. Once the fresh fuel-air mixture was prepared, it was used to fill the
main combustion chamber to 1 atm. A trigger signal generated by a delay generator
(SRS DG535) was used to trigger the high-voltage pulse generators, initiating
combustion. The time delay from the rise of the trigger pulse to the rise of the high-
76
voltage discharge in the combustion chamber was 18 ± 1 µs for the 85 ns pulse generator,
and 9 ± 0.5 µs for the 12 ns pulse generator.

Figure 41 – Setup of transient plasma ignition experiments in a constant volume
reactor.

77

Figure 42 – Photo of the combustion chamber and mixing manifold.
3.3.1 Varying Pulse Width
The pulse width was varied by using two different pulse generators, while the
pulse amplitude was kept constant at 42 kV, resulting in an estimated maximum reduced
electric field strength (E/n) of 400 Td (E
max
=100 kV/cm) at the anode surface. The
energy delivered per pulse was 365 mJ for the 85 ns TPI system, and 75 mJ for the 12 ns
TPI system.
The ignition delays produced by both the 85 ns TPI system and the 12 ns TPI
system across a broad range of equivalence ratios are plotted against the ignition delays
produced by a spark ignition system in Figure 43. Each run was repeated at least six
78
times, and the mean value was plotted. The error bars represent the standard deviation
from the mean, which was typically less than 1%.

Figure 43 – Ignition delay of quiescent stoichiometric C
2
H
4
-air at 1 atm in 1.73 L
stainless steel chamber. The gap is 6 mm and the pulse amplitude for both TPI
systems was 42 kV.

As seen in previous experiments in flowing fuel-air mixtures in a PDE, both of
the transient plasma ignition systems produced similar ignition delay times near
stoichiometric conditions in the quiescent mixture, despite the large difference in the
energy delivered. The important difference here is that the same electrode was used for
both pulse generators. The 85 ns TPI system produced slightly shorter average ignition
79
delay times near ϕ = 1.0. Note that the spark ignition system could not ignite C
2
H
4
-air at
an equivalence ratio less than ϕ = 0.7. Both TPI ignition systems could ignite a mixture
down to ϕ = 0.5 (not shown because the ignition delay is much longer).
Flame propagation was recorded with the high-speed camera through a window
on the flange opposite the electrode. The images were recorded at 670 fps with a 300 µs
exposure. Figure 44 compares the spark ignition system with the 12 ns TPI system. The
calculated flame speeds for the transient plasma ignited runs at ϕ = 1.1 were an average
of 10% faster than for the spark ignited runs. No significant difference in the flame
propagation was seen between the 12 ns TPI system and the 85 ns TPI system.

Figure 44 – Photographs of ignition of C
2
H
4
-air ( ϕ = 1.1). Time indicates time after
discharge. Top: automobile ignition coil with spark plug; Bottom; 12 ns TPI system.

The discharge volume of the transient plasma ignition systems was approximately
25 cm
3
. Using either pulse generator, approximately 50 streamers were present during
each discharge and were spread throughout this volume. Combustion enhancing radicals,
80
including excited species of oxygen, are produced in the streamers during the discharge,
and the highest density of radicals is found along the anode, where the electric field is
highest (Ono, 2003; Cathey, 2007). At 1 atm, ignition occurs along the anode, not along
the streamers.
A previous experiment in a pulse detonation engine showed that using a flowing
fuel-air mixture, similar ignition performance could be achieved while using smaller
electrode geometry and lower pulse energies than what had been used. This experiment
showed that in a quiescent fuel-air mixture, with constant electrode geometry (gap size)
and constant pulse amplitude, similar ignition performance could be achieved using a
pulse with less energy. The significance of this result is that it shows that pulse
generators that deliver pulses with energy on the order of 10’s of mJ, and perhaps even
less, may be used for this application.
In future work, the transient plasma ignition system including the pulse generator
and electrode will be optimized for pulse width, amplitude, rise time, gap size (anode and
cathode diameter), anode material, and anode length. Understanding how these
parameters affect combustion will help the investigation of the mechanism behind
transient plasma ignition. The results presented show that with identical electrode
geometries, similar ignition delays can be achieved across a broad range of equivalence
ratios using much lower energy per pulse than previously used (5 times reduction in
energy delivered), and comparable energy per pulse to traditional spark ignition. This
means that more compact, less expensive (due to the reduction in voltage and current
requirements for the switch) pulse generators may be used for this application.
81
3.3.2 Varying Gap Size
The relationship between gap size and combustion performance (ignition delay
and pressure rise time) was investigated. Previous work has shown that for a transient
plasma discharge, as the gap length increases, the effective length of the region where
active particles are produced decreases if the electrode voltage is held constant
(Pancheshnyi, 2000). It was found that for small gaps, the linear density of active
particles did not change much across the gap, but for longer gaps (> 24 mm in their case),
the linear density [N*] beyond the area where the streamer is formed decreased rapidly,
and could be represented by:

[ ∗
]( ) = [ ∗
]

−
( )
eq. (7)

In eq. (7), [N*]
max
is the peak density of active particles, which is reached at the
distance x
0
, and η is the damping length of the emission intensity, which characterizes the
uniformity of the production of active particles along the streamers. The particles being
looked at were N
2
(C
3
Π
u
, v = 0), N
(B
2
Σ

, v = 0), and NO(A
2
Σ
+

, v = 0).
The increase in the production of active particles in a shorter gap is due to
enchanced electronic states at the streamer head in a higher external field. For smaller
gaps, the streamers bridge the gap in the presence of a high average electric field,
forming conductive channels. For larger gaps, the same voltage applied results in a lower
average electric field, and the streamers do not bridge the gap, but rather freely propagate
82
across the gap in a weaker field. The density of active particles decreases because of the
lower average field in the gap. The region where the streamer head is characterized by
the maximum density of active particles is independent of the gap size, and is determined
by the strength of the electric field near the high-voltage anode. This means that in a
combustion application, the ideal configuration would apply the maximum electric field
strength in a gap for that pulse width without transitioning to an arc.

Figure 45 – Transient plasma produced ignition delay of quiescent stoichiometric
C
2
H
4
-air at 1 atm in 1.73 L stainless steel chamber using a 12 ns pulse generator. The
maximum applied electric field strength is 100 kV/cm for all gap sizes.

83
The gap size was varied from 4 – 15 mm. The gap is determined by the inner
diameter of the cylindrical cathode that is bolted onto the flange of the chamber. Figure
45 shows the ignition delays with varying equivalence ratio using the 12 ns pulse
generator. The difference between the ignition delays in the gap that achieved the
shortest delays (6 mm) and the gap that produced the longest delays (10 mm) was around
15%. It is interesting to note that the 6 mm gap slightly outperformed the 4 mm gap,
since the density of active particles decreases as the gap sized increases. However, in the
4 mm gap, the pulses resulted in an arc, so it is thought that less of the energy went into
the production of active species.

84

Figure 46 – Transient plasma produced ignition delay of quiescent stoichiometric
C
2
H
4
-air at 1 atm in 1.73 L stainless steel chamber using am 85 ns pulse generator.
The maximum applied electric field strength is 100 kV/cm for all gap sizes.

Figure 46 shows the ignition delays with varying equivalence ratio using the 85
ns pulse generator. These results are interesting because using the applied electric field
strength of 100 kV/cm, every pulse in the 6 and 8 mm gaps transitioned to an arc,
however, for this higher energy pulse generator, this does not seem to have had much of
an effect on the ignition delays.
These results are important because they suggest that the combustion process is
less sensitive to the transient plasma discharge than was once thought. When the
discharge transitioned to an arc, there was only a slight, if any, decrease in combustion
85
performance. This suggests that the rise time of the pulse is what is critical for producing
the advantages that transient plasma ignition offers. As a result of these experiments, for
all other experiments in the constant volume reactor the 6 mm gap was chosen for use
with the 12 ns pulse generator because it produced slightly faster ignition delays, and the
15 mm gap was chosen for all other experiments using the 85 ns pulse generator because
it provided the most visibility of the discharge for experiments where high-speed
photography was used.
3.3.3 Varying Anode Length
The anode length was varied in order to examine the impact of the volumetric
effect of TPI. A spark initiates ignition at a single point, whereas transient plasma
initiates ignition at multiple points along the anode. With the anode reduced to 3 mm,
ignition kernels could only form along a small surface area, making this condition more
similar to a spark plug. With the volumetric advantage of TPI reduced, the ignition delay
times and combustion pressure waveforms were compared to that of spark ignition. It
was expected that the ignition delay times for transient plasma ignited C
2
H
4
-air would
still be faster than spark ignition.

86

Figure 47 – Different configurations of the constant volume reactor for testing
different anode lengths. Top: long threaded rod; Bottom left: short threaded
rod; Bottom right: spark plug.

87
Five different length anodes were used with the pseudospark pulse generator to
ignite stoichiometric ethylene-air with lengths from 0.3 cm to 8.9 cm. Here, the length of
the anode refers to the amount of anode that is not insulated, which is the exposed length
from which streamers propagate to the chamber wall. The reduced electric field (E/n)
was kept constant at 475 Td. The gap wall of the combustion chamber was used as the
cathode, resulting in a 47 mm gap. This longer gap changes the load conditions for the
pulse generator, so the pseudospark pulse generator delivered a 100 ns, 75 kV pulse.

Figure 48 – Transient plasma ignition characteristics of ethylene-air with varying
anode lengths, and constant E/n.

88
As the anode length was decreased, so was the volumetric effect of TPI, since
ignition has been shown to occur along the anode, where the electric field is highest. The
results are shown in Figure 48. As the anode length was decreased from 8.9 cm to 0.3
cm, ignition delay increased from 5.8 ms to 8.3 ms (compared to 18.3 ms for spark
ignition), and peak pressure decreased from 178 psi to 154 psi, presumably since fewer
ignition kernels were present. For the shortest anode, the peak pressure was 5% less than
that produced by spark ignition, but the ignition delay was more than 30% faster than
spark ignition. The “hump” on the rise of the spark ignited case is thought to be caused
by the competition of heat loss to the wall and the heat release due to continuing flame
propagation. The reduced appearance of this effect on the rise in TPI compared to spark
ignition indicates faster combustion, despite similar effective areas.
These results suggest that the gains observed in peak pressure while using TPI are
caused by the volumetric advantage of the system, but that the gains in ignition delay are
caused by the non-thermal mechanism resulting from the transient plasma discharge. In
order to achieve increased peak pressure using TPI, the discharge should fill as much of
the combustion volume as possible to distribute the radicals throughout the volume.
However, to achieve gains in ignition alone, the discharge volume can be much smaller.
3.3.4 Varying Cathode Porosity
The purpose of using a porous cathode for these experiments is two-fold; 1) For
flowing fuel-air mixtures, a perforated cathode provides a lower velocity region inside
the discharge volume to enable sufficient development of a flame front, and 2) for
89
quiescent and flowing fuel-air mixtures, the holes or slots allow the flame initiated inside
the cathode to propagate through it to reach unburned fuel around it without having to
exit at the end. A series of experiments were performed to verify this effect and to
optimize hole size and number to achieve the fastest pressure rise times.
For this experiment, the 12 ns pulse generator was used with five 70 mm long
porous cathodes (6 mm gap) and a 50 mm long anode. Each of the five cathodes had
different hole sizes and a different number of holes. High-speed images were taken of
the transient plasma initiated combustion event at 700 fps using a 300 µs exposure
(Figure 49). From these images the effect of the porosity of the cathode is clear: A
higher number of smaller holes allows for the best flame propagation.

As discussed in section 3.1, the burning velocity of a fuel-air mixture can be
enhanced by turbulence. As the flame exits the holes, the burning velocity is increased
Figure 49 – Images of flame propagation 6 ms after the discharge of a 12 ns, 42 kV
pulse and the corresponding cathodes with varying porosity (hole number × hole size
in inches). The gap was 6 mm.
90
due to the effect of turbulence in wrinkling the flame and therefore extending its surface
area (Ballal, 1975). A higher number of holes spaced closely together increased the
turbulence of the flame, and thereby the burning velocity. Future work will use mesh
with different grid sizes as the cathode.
3.3.5 Varying Pulse Amplitude
The amplitude of the 12 ns pulse was varied in order to examine the impact of a
change in E/n on ignition delay. The amplitude of the pulse was increased from 38 kV to
54 kV by increasing the charging voltage of the pulse generator. This results in an E/n
increase from 360 Td to 520 Td. When a 12 ns pulse with an amplitude below this
threshold was applied, streamer breakdown could not be generated reliability in the 6 mm
gap at atmospheric pressure. As shown in Figure 34, an increase in ignition delay is not
expected over this range of E/n. The results confirmed that an increase of E/n beyond
360 Td does not have a significant impact on ignition delay (Figure 50).
91

3.4 Transient Plasma Ignition in an Internal Combustion Engine
The application of transient plasma ignition to internal combustion engines is less
developed compared to other applications, but preliminary work in this area is promising.
TPI is attractive for ICE’s for a many reasons, including its demonstrated ability to ignite
leaner mixtures and its potential for volumetric ignition. Combusting under a lean
condition results in higher thermal efficiency and reduced NOx production (Glassman,
1987). In an ideal ICE, constant volume combustion occurs at the minimum cylinder

Figure 50 – Ignition delays produced by a 12 ns pulse with varying pulse
amplitude (from 38 kV to 54 kV).
92
volume under a high compression ratio, but with spark ignition, lean combustion does not
occur quickly enough to approach this ideal CV cycle. A solution to this problem is to
use a multi-point ignition source so that ignition occurs over a volume. Transient plasma
ignition has the potential to fill this role due its volumetric capabilities. Both the lean
burn capability and the volumetric effect of the transient plasma gives it an inherent
advantage over spark ignition, in particular, in the reduction of NOx production.
There are two important differences to consider when comparing the operating
conditions of an ICE and a PDE; (1) the operating pressure in the combustion chamber,
and (2) the size of the discharge gap. (1) In a PDE, the pressure in the main combustor is
generally between 1 and 1.5 atm, while in an ICE, the pressure in the cylinder typically
peaks around 10 atm. This increase in pressure results in a reduction in the mean free
path of the electrons, which increases the number of collisions and decreases the effective
electron energy relative to the same discharge at lower pressures. In order to maintain
effective electron energies in the transient plasma so that it can dissociate the molecules
in the mixture, a higher electric field needs to be applied without causing a transition to
an arc. This means that shorter pulses are necessary. (2) The discharge gap in an ICE is
smaller than in a PDE. There is less freedom to design large gap electrodes for use in an
ICE, due to constraints such as valve positions and a moving piston. The gaps used in
PDE’s for TPI can be bigger than 30 mm. In an ICE, the maximum gap size when
placing the electrode in the same configuration as a spark plug is on the order of 5 mm.
The differences in the pressure and the discharge gap place strict requirements on the
93
pulse generators for this application, since discharges in high pressures and small gaps
have a tendency to arc.
Previously, an experiment applying TPI in an ICE was performed with Nissan
Research Center in Japan (Cathey, 2007). It was found that by using transient plasma
ignition, the ignition delay period, which is an index of performance in a conventional
gasoline engine, can be improved relative to a spark plug. Stable lean combustion was
shown to occur at levels not realizable by spark ignition. Transient plasma produced by a
20 ns pulse was found to increase peak pressure relative to spark ignition by more than
20% (Figure 51). Recently, road tests with a transient plasma ignition system have
shown an increase in efficiency (increase in mpg) of up to 15% (Tropina, 2009).

94

Figure 51 – TPI in an ICE at Nissan Research Center. Pressure vs. crank angle, for
the spark, 100 ns pulse, and 20 ns pulse, ф =0.72 (Cathey, 2007).

To follow up on this work, an experiment was conducted in collaboration with
Earnhardt-Childress Racing in Welcome, NC, applying the new 12 ns TPI system in an
800 HP ICE (Figure 52). The spark plug in a single cylinder was replaced with the same
transient plasma electrode used in the experiment in the PDE at WPAFB (See Section
3.2.2). For this experiment the pressure in the engine at top dead center (TDC) was
around 10 atm. The pressure in each cylinder was measured with respect to crank angle.
The experimental setup is shown in Figure 53. The fuel-air conditions in this experiment
were different from previous work in that the fuel-air mixture was rich, as it is run that
way for races.
95

Figure 52 – 800 HP engine used in the experiment at Earnhardt-Childress Racing.

For spark based ignition systems, the electrical systems are designed to dampen
noise from a relatively long pulse (µs to ms), not from the high-frequency noise caused
by a high-voltage discharge with rise times of a few nanoseconds. There were both
electromagnetic noise problems and problems with the electrode used in this experiment.
Initially, the engine was run under 100% load conditions at 4,500 RPM. Under these
conditions, one of the custom electrodes was destroyed due to the high-temperatures in
the cylinder. With time limitations, some system grounding issues were solved, a custom
shielded HV cable was fashioned to transmit the 12 ns pulse to the cylinder with minimal
noise, and the engine was run at 3,000 RPM with no load to protect the electrode.
96

Figure 53 – Measurement setup for an internal combustion engine experiment at
Earnhardt-Childress Racing.

Only one successful run was captured in the brief time allotted for the experiment.
The data was taken while the engine was idling at 3,000 RPM (Figure 54). Of particular
interest is the number of misfiring engine cycles at idle with both ignition systems
(misfires are cycles with below zero Indicated Mean Effective Pressure or IMEP). Data
with misfires cannot be analyzed for burn rate because the burn rate algorithm
intrinsically assumes a 100% combustion efficiency, where in reality, a misfiring engine
cycles has very near zero combustion efficiency. Heat release algorithms will produce
burn rate information for misfiring cycles, but the data is not useful because a negligible
97
amount of combustion occurred. It is difficult to extract comparable cycles with such a
high misfire frequency.

In the experiment, the engine fired after several cycles and produced extremely
high IMEP because the chamber was fully scavenged (zero residual gases). The
following cycle had some residual burned gas content and produced lower IMEP. The
next cycle had more residual gases and produced still lower IMEP, and the next cycle
plunged back into misfire until the residual burned gas was once again fully replaced by

Figure 54 – Peak pressure in the cylinder in an internal combustion engine using
transient plasma ignition and spark ignition at while the engine is idle at 3000 rpm.

98
fresh fuel-air mixture. There were no consecutive cycles where the engine reached any
kind of steady state operation. It appeared that the pressures in the transient plasma
ignition case were higher, but for the reasons stated above, nothing conclusive can be
drawn from this data. While this was not an ideal result, the experiment did show that the
small electrode and 12 ns pulse generator could be used to ignite gasoline in an ICE, and
provided insight as to how to successfully integrate a transient plasma ignition system
into a high-performance engine.
The spark ignition system for automobiles has been used for over 100 years and is
very well developed. If transient plasma ignition is to be accepted as a replacement to the
spark ignition system in the automotive industry, size, efficiency, cost and reliability of
nanosecond pulsed power systems must be improved. Recent pulsed power
developments and transient plasma ignition experiments suggest that this technology can
be realized in the near future.
99
Chapter 4: Transient Plasma Physics
4.1 Introduction
The application of transient plasma to ignition and combustion reduces ignition
delays and improves combustion efficiency in a range of hydrocarbon-air mixtures in a
wide variety of engine types, relative to conventional spark ignition. The fundamental
understanding of this ignition enhancement is not yet well developed, however, critical to
the understanding is the identification of the relative roles of non-thermal vs. thermal
processes. The experiments discussed in this chapter attempt to improve the
understanding of the transient plasma ignition mechanism by examining two aspects of
TPI; how TPI is impacted by the presence of humidity and how the flame is initiated after
a transient plasma discharge.
A clue toward understanding TPI was provided when it was observed that
humidity impacts the combustion performance of TPI, and a series of experiments was
performed to determine the cause of this effect. It was shown that humidity impacts the
production of ozone (O
3
), providing evidence that the decrease in combustion
performance is due to a reduction in the production of atomic oxygen (O). A follow up
experiment is discussed in Chapter 5.
100
Using high-speed photography, it was shown that while active species are
produced in the streamer channels that branch across the volume, ignition kernels are
formed at the ends of the spatially separated streamer channels. At these locations, the
reduced electric field, E/n, is on the order of hundreds of Td and a significant fraction of
the energy is transferred into electronic processes. As a result, the transient plasma
produces two phases of ignition; a distinct non-thermal phase of ignition, wherein the
efficient production of electronically excited species accelerates reaction rates and
reduces ignition delays, and a subsequent thermal phase, driven by exothermic fuel
oxidation in reactions with free radicals and the decay of active species.
4.2 The Effect of Humidity on Transient Plasma Ignition
To simulate conditions of supersonic cruising at altitude in a PDE experiment, the
airflow is heated using a hydrogen fueled vitiator system, which utilizes oxygen to
rebalance the oxygen content of the heated air up to the standard mole fraction of 21%.
While running a transient plasma ignition system during an experiment (Singleton, 2009),
it was noted that TPI produced the best detonation wave velocities in the PDE a few
seconds after the vitiator was turned off, when the air was dry (Figure 55). Water is a
byproduct of the vitiator system, and is believed to be responsible for the decrease in
performance. The detonation wave velocities were not affected by the vitiator when
using spark ignition. This discovery lead to further investigation, because it can be used
101
to help determine which radicals produced in a transient plasma discharge are most
important for ignition and combustion.

4.2.1 Measurement of Streamer Intensity
In an experiment in collaboration with Stanford University and NPS, the electrical
energy delivered and the relative intensity of the streamers were measured while applying
transient plasma discharges to a coaxial electrode in air with varying levels of humidity in
the constant volume reactor described in Chapter 3 (Figure 56). The energy delivered

Figure 55 – Detonation wave velocities in the PDE at NPS with the vitiator ON
(moist air) and vitiator OFF (dry air). Humidity was above 5% with the vitiator
ON.
102
did not vary significantly (<10%) with an increase in water content in the air, indicating
that most of the energy is still being delivered to the gas. The streamer intensity,
however, was heavily impacted by the increase in moisture content, with a 70% reduction
of intensity measured when the mol fraciton of H
2
O was 18% compared to dry air
(Figure 57).

Figure 56 – Streamer intensity and energy delivered vs. mole fraction H
2
O.
103

Figure 57 – Intensity of streamer discharge with different levels of humidity.

Several authors have found an increase in breakdown voltage with increase of
humidity (Waters, 1978; Aleksandrov, 1998). Increased moisture content in the air
results in an increase in hydrated ions, for which the electron recombination rate is higher
than of ions in dry air. The conductivity of the streamer channel decreases more rapidly,
and the field required for the streamer to cover the gap increases (Figure 58). While this
experiment verified that humidity impacts the transient plasma discharge, more detailed
measurements were necessary to understand the nature of this effect.
104

Figure 58 – Effect if humidity on the uniform-field breakdown voltage in
atmospheric air (Blair et al., 1963).
4.2.2 Measurement of OH and O
3

During transient plasma ignition, the initial pool of free radicals seed many of the
chain branching and chain propagation reactions in the fuel, and are key to initiating and
maintaining combustion. Relatively small amounts of free radicals can shift the system
equilibrium and initiate combustion (Starikovskaia, 2006). A previous experiment
determined the spatial distribution and concentrations of OH created by TPI using laser
105
induced fluorescence (LIF) (Cathey, 2007). While there are a number of radicals that are
interesting to the combustion process (e.g. H, O, OH, and CH
3
), OH was chosen because
it can be measured relatively easily. In that experiment it was shown that during a
transient plasma discharge, OH was distributed throughout the volume, mostly in the
streamers, and then decayed rapidly. OH concentration fell below the detectible limit
within 100 μs after the discharge while ignition did not occur until 1 ms after the
discharge, suggesting additional chemical processes.
Simulations were performed to help develop links between the short timescale
plasma chemistry and the long timescale combustion chemistry, with an emphasis on the
dependence of the performance of TPI on H
2
O. Zero-D Boltzmann modeling provided
spatial information from an E/n input (Figure 59). The long timescale evolution
suggested that O
3
, as opposed to OH, persists long enough to impact combustion
chemistry. A three order of magnitude difference in O
3
production occurs in the presence
of 20% mol fraction of water. O
3
is a good oxidizer and has be shown to impact flame
speed (1% O
3
resulted in a 6% increase in flame speed in C
2
H
4
-air in one study) (Kelly,
1957; Yagyu, 2008).

106

Previous experiments by Ono and Oda found significant reductions in O
3

production using a small gap (15 mm), point-to-plane electrode geometry, and a 27 kV,
200 ns pulse (Ono, 2003). The TPI systems here use a coaxial geometry with shorter
pulses and larger gaps. In this work, a line-type pseudospark switched pulse generator
was used to deliver a 60 kV, 100 ns pulse across a 47 mm gap. The reactive species were
produced throughout a 730 cm
3
volume in a stainless steel combustion chamber.
An experiment was performed at WPAFB to measure both OH and O
3
in a
flowing, humid air system, with varying water concentrations. OH density was measured
by planar laser-induced fluorescence (PLIF), and O
3
density was measured by ultraviolet
(UV) absorption. The Q
1
(3) transition at 282.2 nm was chosen for excitation, since its
Figure 59 – Simulation of OH and O
3
after the application of an electric field in air,
with varying amounts of humidity. Courtesy of J. Luginsland.
107
ground state offers a reasonable compromise between low-temperature population and
insensitivity to temperature changes at low temperatures.
The experiment presented consisted of two separate measurements; 1) a planar
laser-induced fluorescence (PLIF) measurement to probe OH development and
production induced by the transient plasma in the presence of varying levels of humidity
and, 2) a UV LED absorption measurement to characterize the effect of varying levels of
humidity on O
3
production. The air was delivered from a facility compressor; it was
initially dried and filtered for particles. Water was then added to the air by directing a
portion of the air flow through a bubbler, which consisted of a Teflon jar with inlet and
outlet ports. The bubbler was also immersed in a heated water bath, but in general
humidity levels in the chamber were limited to room-temperature saturation values. This
chamber used is the same chamber described in Chapter 3. The anode was a steel 8-32
threaded rod 8.9 cm long. A line type pseudospark switched pulse generator charged
with a Glassman high-voltage DC supply was used to create a 100 ns, 60 kV pulse
responsible for generating the transient plasma (see Section 2.3.2). A diagram of the
setup is shown in Figure 60.
108

Figure 60 – Experimental setup for OH-LIF measurement.
Measurement of H
2
O Concentration
The water density measurements were made via a line-of-sight absorption
technique using tunable diode laser absorption spectroscopy (TDLAS) which yields a
path-averaged value for the local density at the exit plane of the test chamber. Three
temperature-stabilized diode lasers with central output wavelengths near 1.394 μm (7172
cm
-1
) are repetitively tuned across 5 strong water transitions in this spectral region. The
tuning is achieved by sequentially ramping the current to the diodes (time multiplexing)
at a repetition rate of 1 kHz. Most of the output of the diode lasers (~90%) was
combined into a single fiber that delivered the light to edge of the exit plane of the test
109
chamber. This single-mode fiber was fitted with a collimator that produced a 1-mm
beam that traversed the exit plane along a diameter about 1 cm above the exit plane. The
transmitted light was collected by a similar collimator/fiber combination that in turn
delivered the light to an InGaAs photodetector. The voltage output of the photodetector
was digitized using a commercial (National Instruments) data acquisition system. The
pitch-and-catch collimator/fiber assemblies were positioned such that transmitted beam
passed only above a region defined by the diameter of the test chamber exit plane. This
prevented the need to make any signal corrections due to ambient water vapor in the lab
area surrounding the test chamber.
The recorded transmission signals were processed using a Beer’s Law
interpretation of the signal yielding absorption spectra. A Boltzmann analysis of these
spectra was then used to extract the water density. For an incident beam intensity of I
o
,
the transmitted intensity, I, is expressed as a function of the water density, N, and total
pathlength, L via Beer’s Law.

− =
= eq. (8)

In eq. (8), S denotes the tabulated line strength for the individual transition and g
ν

the lineshape function which is normalized such that the integral of g
ν
over all
frequencies is equal to unity. Values for the line strengths are found in the literature for
specific temperatures – typically room temperature. A
ν
denotes the frequency dependent
110
absorbance and each target transition has a unique value for A
ν
that depends only on the
exit plane temperature and water density for the fixed pathlength L. To extract numerical
values for the water density, the acquired transmission signal, − , was integrated
for each transition over frequency to obtain measured integrated values of the absorbance
(A
meas
) for multiple transitions. The set of measured absorbances was then analyzed in
the form of a Boltzmann plot to obtain the path-integrated temperature and water density.
Specifically, for each transition, the measured integrated absorbance was normalized by
the corresponding line strength and the natural logarithm was taken of this ratio. This
quantity was then plotted versus the ground state energy for each of the accessed
transitions. The resulting graph is linear; the reciprocal of the slope yields the
temperature and the y-axis intercept yields the water density.
The raw transmission signals were recorded at 1 kHz and averaged in real time
such that post-processing yielded water density measurements at 1 second time intervals.
The 10% of the diode laser outputs not directed toward the sample chamber was split
50/50 between a photodetector measuring the incident light levels (I
o
) and a solid etalon
providing an accurate frequency tuning reference.
LIF for OH Measurement
A. OH PLIF Experimental Setup
OH was excited by pumping a transition in the A
2
Σ
+
← X
2
Π (v′=1,v ″=0) band
(Figure 61) at 282.209 nm. To generate this wavelength, the output of a Q-switched,
111
frequency-doubled Nd:YAG laser pumped a Lumonics Hyperdye HD-300 dye laser
operating at 564 nm; the dye beam was frequency-doubled within an Inrad AT III, to
generate the 282-nm OH probe beam. The UV beam was then isolated from the dye
beam within an Inrad Prism Harmonic Separator. Pump energy at 282 nm was kept
reasonably low, about 4 mJ at the output of the prism separator, to reduce transition
saturation effects. The laser probe sheet was formed using a combination of cylindrical (-
100-mm focal length) and spherical lenses (1-m focal length); the cylindrical lens was
oriented such that the beam was expanded in the horizontal direction; the spherical lens
collimates the sheet in the horizontal direction while focusing it in the “thickness”
direction. The sheet thickness is estimated to be 300 μm in the probe region. A small
portion of the beam was reflected by a fused-silica flat, placed before the cylindrical lens,
and directed over a reference flame and then to a photodiode to monitor the respective
LIF from the nascent OH, using a photomultiplier tube. Both photodiode and
photomultiplier signals were recorded on a LeCroy WaveRunner 44Xi digital
oscilloscope, and the signals were integrated over the pulse width within the oscilloscope
and then saved for later analysis. The purpose of monitoring the OH LIF signal was to
ensure that the laser remained tuned to the OH Q
1
(3) transition throughout the
experiment.
112

Figure 61 – Electronic states and energy levels of OH. W indicates the absorption
process, Q
e
the electronic quenching process, Q
V,10
the vibrational energy transfer
process and A the fluorescence process (K. Kohse-Hoinghaus, 1994).

The laser was operated at its normal pulse repetition frequency of 10 Hz;
however, it was necessary to operate the HV pulser at a much lower frequency, and 1 Hz
was selected as a reasonable compromise (it was also necessary to remove all discharge
products, e.g., O
3
, between laser measurements and this was facilitated by the 1-Hz
sample frequency). To accomplish synchronization and control of time delay between
the laser and the HV pulser, a Quantum Composer 9328 Delay Generator was employed.
This device operated at a 10-Hz fundamental frequency to provide triggers to the
Nd:YAG laser system; a Stanford Research Systems DG535 provided a 1-Hz “cueing”
pulse to the Quantum Composer that allowed synchronization of the HV pulser (and the
PLIF camera described below) with the laser at a 1-Hz sub-harmonic frequency.
113
To achieve a uniform flow of air, the chamber was mounted vertically, and glass
beads were placed in the bottom of the chamber (Figure 62). A 45-degree mirror
reflected the fluorescence to a Princeton Instruments PIMAX UV sensitive (Superblue)
intensified CCD camera (512×512 pixel array. The camera was positioned to view a
portion of the chamber from the outer wall (where the laser sheet entered the chamber) to
just past the central electrode. The camera was fitted with a fast Cerco 45 mm focal
length f/1.8 UV lens. The camera gate was set to 100 ns, to block emission from the
discharge; furthermore, the camera’s micro-channel plate (MCP) was also gated, in a
process called bracket pulsing, which further improves the on-off ratio of the intensifier.
UG-5 and WG-305 Schott-glass filters were used to block the (v′=1,v″=0) fluorescence
and 282-nm scattering and allow passage of the strong (v′=1,v″=1) and (v′=0,v″=0) bands
at ~312-322 nm and ~306-316 nm, respectively (Figure 61). Note that the v′=0 (prime
designating an electronically excited state) vibrational state is populated through
vibrational energy transfer, VET, from v′ = 1. It should be noted that the WG-305 filter
preferentially attenuates the (0,0) band fluorescence, which is blue-shifted relative to that
from the (1,1) band. The Q
1
(3) transition was chosen for excitation, as its ground state
offers a reasonable compromise between low-temperature population and insensitivity to
temperature changes at low temperatures.

114

Figure 62 – Left: Cross-section of a 10.1 cm ID, 20.2 cm long combustion chamber
and; Right: a photo of the combustion chamber.

B. OH PLIF Image Processing
To estimate the OH number densities within the discharge, a 25.4-mm-square
Hencken burner was used as a calibration source (Barlow, 2001). The flame was
generated from a near-stoichiometric H
2
-N
2
-air mixture for which adiabatic equilibrium
calculations give a temperature of about 2150 K; for the calculations below, major
species product concentrations are assumed to be the adiabatic equilibrium values while
the temperature is assumed to be 2100 K. It was expected that the OH number density
would be close to the adiabatic equilibrium value, though it was not clear that the OH
would be equilibrated at the probe height used (about 25 mm above the burner). Note
115
that for the calibration measurement, the camera was looking down on the flame and the
laser sheet probed a horizontal plane across the burner.
OH absorption measurements were performed with this same laser system (but
different laser dye for the different operation wavelength); for this purpose, the laser was
scanned in wavelength across the Q
2
(7) transition in the A
2
Σ
+
← X
2
Π (v′=0,v ″=0) band,
and the total integrated absorbance was derived from the average of several such scans.
To then derive an accurate map of the OH number density across the Hencken burner, an
accurate relative profile of number density must be determined; however, main-branch
transition that are used for absorption cannot easily be used, as the strong absorption
distorts the OH profile. Thus, the OH LIF line profile was recorded (in conjunction with
the absorption measurements) with the laser tuned to the Q
12
(7) transition (adjacent to the
Q
2
(7) transition); satellite transitions such as these are sufficiently weak to mitigate the
absorption bias. Increased laser energy (above the 10 μJ/pulse used for absorption) was
used to compensate for the smaller Einstein B coefficient (and associated smaller LIF
signal strength).
The time-integrated fluorescence signal (counts), S
f
, from collection volume V
c
is
proportional to the detected number of fluorescence photons:

=

eq. (9)

116
Here, β accounts for net optical transmission and detector gain; Ω
c
is the detection solid
angle; N
OH
is the OH number density while f
B
is the Boltzmann fraction of OH molecules
in the absorbing state. , the specific fluorescence, is defined as

≡
(
)∙

∆

eq. (10)

which for linear fluorescence conditions becomes

= ∙
,
/
,

,

,

∆
eq. (11)

when VET from v′=0 to 1 or from v′=1 to 2, is negligibly small. Here, N
i
, ε
i
, A
i
and Q
e,i

are the respective number density, fluorescence collection efficiency, fluorescence rate,
and electronic quenching rate for the A-state v′= i (0 or 1) level; Q
V,10
is the rate for VET
from v′=0 to 1. W is the laser excitation rate, which here equals approximately 2×10
9
s
-1
,
for the flame, and 1.8×10
9
s
-1
,

for room temperature air, based on the estimated irradiance
at the probe region; the time integral of W is proportional to laser pulse energy. The
collection efficiencies were derived from a measurement of the transmission (T vs. λ) of
the WG-305 filter and the modeled fluorescence spectrum.
117
With these rates one can calculate the specific fluorescence using eq. (11) within
the reactants and the burnt gas of the calibration flame and use this to estimate the
streamer number density:

,
=
,
∙ , ,
∙

∙ ,
,
eq. (12)

However, to better understand the excitation and collisional processes, it is also
useful to employ a rate equation model. Thus, a 5-level model was formulated
representing the populations of the directly pumped rovibronic levels, X(v ″=0, N=3) and
A(v′=1, N=3), the remaining rotational levels of the A(v′=l) and X(v ″=0) levels, and the
A(v′=0) level. It is assumed that the electronically quenched molecules are not returned
to X(v ″=0) but rather are “lost” to the system in the time-frame of laser excitation (i.e., 10
ns). Collision rates Q
e,1
, Q
e,0
, and Q
V,10
are from Steffens and Crosley, for room
temperature air. For the Hencken flame, N
2
cross sections for VET and electronic
quenching are from (Paul, 1995) (in each case the average of the 1900- and 2300-K
values). The H
2
O cross section for VET was from (Hartlieb, 1997), though this value is
too small to make a significant difference. The H
2
O electronic quenching cross sections
were chosen to be 26×10
−16
cm
2
for both v′=1 and 0; these cross sections are consistent
with measurements of Beaud et al. and other literature values (e.g., see Hartleib et al. and
Paul). Sources for rotational energy transfer rates for the rate-equation model include
Kliner and Farrow, Burris et al. and Zizak et al.
118
At low excitation rates (small W), the 5-level model yields the same specific
fluorescence (eq. (10)) as given by eq. (11), as expected. As the excitation rate increases,
from low (linear fluorescence) values to 2×10
9
s
-1
, the ratio /

increases only
modestly from about 1.5 to about 1.6, due to the combination of ground-state bleaching at
room temperatures and transition saturation at high temperatures. Thus, a value of
/

= 1.5 (±0.5) is assumed to be a reasonable approximation; here, the
uncertainty in /

is based on uncertainty in collision rates coupled with that
produced by transition saturation effects. Furthermore, because transient plasma deposits
a small amount of energy (500 mJ) that is distributed throughout a substantial volume
within the chamber, there is negligible heating of the gas, an assumption verified with
Filtered Rayleigh Scattering (FRS) measurements. Consequently, T
air
= 293 K within the
streamers (and T
Hencken
= 2100 K), and the ratio of Boltzmann fractions is , /
, ≈ 0.27.
Finally, in processing the PLIF images, an average background image was
subtracted from each PLIF image to account for dark charge, camera offset, and laser
scattering (principally from the anode). A flatfield image, accounting for the laser
intensity distribution and other imaging artifacts, was acquired by doping the inflow air
with a small amount of acetone and imaging the LIF; for this purpose, it was necessary to
remove the UG-5 filter (which otherwise would block most of the fluorescence). This
flatfield was then used to “correct” all images, including the OH calibration image (from
the Hencken burner).
119
C. UV LED Absorption for O
3
Measurement
O
3
density was measured using a Sensor Electronic Technology UV LED
(equipped with a hemispheric lens) and an amplified photodiode detector (Thor Labs
PDA25K). The UV LED provided output with a peak at 260 nm and a width (FWHM) of
about 25 nm. A Semrock bandpass filter centered at 250 nm was placed in front of the
photodiode to prevent saturation of the photodiode from the strong emission generated by
HV pulser; use of the filter effectively blue-shifted the probe wavelength by about 5 nm.
The LED was placed at the one optical access slits while the photodiode assembly was
placed at the other. A 1-ms-duration TTL pulse was provided to the LED, again at the 1-
Hz frequency of the HV pulser, and the photodiode output was recorded on the digital
oscilloscope. The oscilloscope recorded the baseline (no LED output) part of the
photodiode signal along with a portion of the output before the pulser was triggered; with
these two values, I
0
was accurately determined. Absolute timing of the pulser trigger was
known by the appearance of a noise-spike on the photodiode oscilloscope channel.
Rather than average traces within the oscilloscope, individual traces were recorded and
stored, and the concentrations for each scope trace were then derived using Beer’s Law.
For this purpose, 200 individual sample traces were recorded. For the purposes of
display, it is assumed that the concentration is uniform across the chamber, but
presumably, concentration close to the anode will be highest. With the threaded rod
electrode, O
3
is presumably generated throughout the chamber volume, and no O
3
could
be detected under these conditions; for this configuration, the detection limit for O
3

appears to be around a few ppm (averaged across the chamber). Consequently, the anode
120
was changed to a washer aligned with the detection region (in line with the window slits
near the chamber’s central chord) to ensure production of O
3
near the detection path
(Figure 63).

Figure 63 – Streamers propagating from a washer anode.
Results, Simulations, and Discussion
To better understand the various plasma mechanisms that can potentially
influence the combustion process, a set of simulations were developed to study the
coupling of electrical energy into the reacting gas. The energy associated with the
electrical discharge is strongly coupled into the electric field, and because of the non-
equilibrium nature of the plasma, the electric field couples energy directly into the
121
electrons in the plasma, rather than in heating the fuel-air mixture as would occur during
a normal spark discharge.
In previous work involving simulations of transient plasma ignition, the particle-
in-cell method was used to provide both spatial resolution and species dependent
evolution (Birsall, 2004). This technique self-consistently solves for the evolution of the
electric fields, and the plasma represented by a collection of Lagrangian markers that are
allowed to collide via Monte-Carlo collisions. This method is quite successful at
modeling the breakdown and formation of plasma streamers for time scales less than
hundreds of nanoseconds. For the challenges posed by this work, however, the evolution
needs to be tracked for much longer time scales to understand what chemical species are
present during combustion. For this purpose, the transport equations were solved
directly, namely solving a Boltzmann equation for every species of interest via Kinema
software (Kinema, 2010):

f(,, t)/t + ●f(,,t)/ + ●f(,,t)/ = C(f,x,v,t)
Df(,, t)/Dt = C(f,,, t) eq. (6)

Here, f is the distribution function of a given species with charge q; E is the electric
field from discharge, and C represents the collision operator between the species. In
these simulations, there are 63 species in the air model, starting with N
2
, O
2
, H
2
O, and
CO
2
. The electric field derives from both the applied pulse and the neutralizing plasma
streamers.
122
After verifying that the energy deposition observed in the simulation agreed with the
experimental results, a matrix of simulations was performed with an emphasis on the
long-time scale evolution of the chemistry. Most important for this work was the
resultant neutral species that would (a) be present to influence the combustion process,
and (b) have long-time scale density that exhibits a strong correlation to the initial
presence of water vapor. In particular, the electron energy distribution function (EEDF)
is the source of free-energy which allows the formation of novel chemical species during
a transient plasma ignition event. In a conventional spark ignition, a significant fraction
of the energy goes into raising the temperature of the gas. The non-equilibrium nature of
a transient plasma discharge, however, results in a significant fraction of the energy being
used to generate high-energy electrons. During the discharge there is a significant
population of electrons with energy in excess of 10 eV. These electrons open new
chemical pathways that are inaccessible in neutral (non-plasma) kinetics.
The long time scale evolution simulations of OH and O
3
are shown in Figure 59.
The simulations indicate that O
3
, which has the potential to influence combustion
chemistry, is created in significant quantities by the transient plasma, and that the peak
density of the O
3
is a strong function of the initial concentration of water. This strong
dependence is shown in Figure 64, which gives the density of O
3
at 100 µs after the
plasma discharge, as the initial conditions of the simulation were varied. The O
3
density
drops by nearly 63% as the initial percentage of water increases from 0.2% water to
1.0%. At five percent water vapor, the O
3
production drops precipitously, with the peak
O
3
density decreased by nearly three orders of magnitude relative to dry air.
123

O
3
is a good oxidizer and has be shown to impact flame speed (1% O
3
resulted in
a 6% increase in flame speed in C
2
H
4
-air in one study) (Kelley, 1957). In a transient
plasma discharge, O
3
and OH are primarily formed by:

O
2
+ e → O(
1
D) + O(
3
P) + e (R7)
e + H
2
O → H • + OH • + e (R8)
O(
1
D ) + H
2
O → 2OH • (R9)

Figure 64 – Boltzmann modeling of O
3
concentration as a function of water vapor
100 µs after a discharge.
124

The range of humidity tested in this experiment was limited to a maximum around
1% by the bubbling method used to create the humid air. In the range of humidity from
dry air to 1% water vapor, an increase in OH production is expected, rather than a
decrease. There is little OH produced in nearly dry (0.09% humidity) air, and there is a
peak in OH production around 1% absolute humidity, and then a decrease thereafter
(Figure 65).

Figure 65 – Simulation of peak OH concentration after a single 70 kV, 100 ns
discharge in air with varying levels of humidity.
125
The calibrated measurements of OH are in Figure 66. A peak concentration of
4.6 × 10
14
/cm
3
was measured. It is difficult to characterize OH production induced by the
transient plasma because of the spatial non-uniformity of the streamers. As seen
previously, the streamer-induced OH seems to be exhausted around 100 μs. In air, OH
decays by recombination reactions:

OH + OH → H
2
O + O (R10)
OH + OH + M → H
2
O
2
+ M (R11)

Humidity had a relatively small impact on the formation of OH, which was
expected at such low concentrations of water. The presence of water did, however, have

Figure 66 – Result of OH LIF experiment with varying levels of humidity.

126
a significant impact on the formation of O
3
. The increase of humidity from 0.18% to
1.02% resulted in a decrease of O
3
concentration by 67% (Figure 67). This correlates
very well with the simulation, where the O
3
density dropped by 63% as the humidity
increased from 0.2% water to 1.0%. Water vapor reduces the amount of O, which leads
to reduced O
3
production by R9. There are several reactions that could increases the O
atom decay rate with humidity, including (Atkinson 1992; Matzing, 1991, Oda, 2005):

O + H
2
O → OH + OH (R12)
O + OH → O
2
+ H (R13)
O + HO
2
→ OH + O
2
(R14)
O + H
2
O
2
→ OH + HO
2
(R15)
127

Figure 67 – Measured O
3
density after discharge with varying levels of humidity.

It should be noted that the geometry of the experiment limited the accuracy of an
absolute measurement of the concentration of OH and O
3
. OH and O
3
is expected to be
produced in streamers, and while the washer in the O
3
measurement helped to promote
production of O
3
near the detection area of the photodiode, much of the streamers from
the discharge are out of its path for both the O
3
and OH measurements. O
3
density in
streamers by a single pulse is expected to be larger than 2.5×10
15
cm
-3
(Ono, 2003).
Concentrations of O
3
were measured with varying electric field strength. The results are
shown in Figure 68, where a significant impact of increasing electric field on O
3

concentration was observed. While the concentration may not be enough to
128
significantly impact flame speed, the decrease in concentration of O
3
due to an increase
in humidity is an indicator of the impact of H
2
O on atomic oxygen, which is has been
shown to impact flame speed (Bourig, 2009). A significant decrease in O density could
significantly affect ignition delay and flame speed.

Figure 68 – Measured O
3
density after discharge with varying electric field strength.

A transient plasma discharge generates streamers with electrons with energies >10
eV. Through electron impact, these streamers can dissociate O
2
, which produces
strongly oxidizing species including O and O
3
(Zhao, 2005). Several excited states of
atomic oxygen can be accessed by electron impact, but the most energetically accessible
in experimental conditions is R7 above, which has a 7.3 eV threshold (Kossyi, 1992;
129
Klopovskiy, 2004). As an example, consider a discharge in a fuel-air mixture with a
simple fuel, such as methane (CH
4
): at room temperature O(
1
D) reacts with CH
4
near the
collision limit (k = 1.4x10
-10
cm
3
/s), and O(
3
P) reacts with CH
4
at high temperature (k =
6x10
-19
cm
3
/s at 293 K).

CH
4
+ O(
1
D) → CH
3
• + OH• (R16)

The process in equation R7 does not occur in a thermal discharge (electron energy
<1 eV), so the presence of high-concentrations of O(
1
D) would be evidence of
dissociation by electron impact. The CH
3
and OH radicals produced in equation R16 can
initiate combustion chain reactions through the following;

CH
3
• + O
2
→ CH
3
O
2
(chain propagation) (R17)
CH
3
O
2
• + CH
4
→ CH
3
O
2
H + CH
3
•

(R18)

CH
3
O
2
H → CH
3
O• + OH• (chain branching) (R19)
CH
3
O• + O
2
→ CH
2
O + HO
2
•

(R20)
CH
2
O + OH• → H
2
O + HCO• (R21)
CH
3
O
2
H + OH•→ CH
3
O
2
• + H
2
O (R22)

HCO• + O
2
→ CO + HO
2
•

(chain propagation) (R23)
HO
2
• + CH
4
→ H
2
O
2
+ CH
3
•

(R24)
HO
2
• + CH
2
O → H
2
O
2
+ HCO• (R25)

OH• → wall (chain termination) (R26)
130
CH
3
•

→ wall (R27)
HO
2
• → wall (R28)

In addition to O
3
and OH, these results suggest that atomic oxygen plays an
important role in transient plasma ignition. In future work, the production of atomic
oxygen in a transient plasma discharge will be characterized (see Chapter 5).
Conclusion
It was shown that an increase of humidity impacts the detonation wave velocities
produced in a pulse detonation engine while using transient plasma ignition. Simulations
performed indicated that O
3
produced in high-concentrations in the discharge lasted long
enough to impact combustion chemistry. The production of OH and O
3
in a transient
plasma discharge in air with varying levels of humidity was measured. Peak OH
concentrations of approximately 5x10
14
/cm
3
were measured, which decayed below 1.3 ×
10
14
/cm
3
at 100 µs. In this experiment, increasing the humidity from 0.18% to 1.02%
resulted in a 67% decrease in O
3
concentration in a single discharge, which was
confirmed by simulation. While the measured concentration of O
3
cannot be said to be
an accurate absolute measurement due to the spatial non-uniformities of the transient
plasma discharge, the results show that relatively low levels of humidity have a
significant impact on its production. The decrease in production of O
3
can be explained
by a reduction of the lifetime of atomic oxygen as a result of the increase in water vapor.
The density of oxygen atoms is known to be related to ignition delays and flame speed
131
(Bourig, 2009; Aleksandrov, 2009). Atomic oxygen plays an important role in transient
plasma ignition, and its production in the discharge will be characterized in future work.
4.3 Spatially Resolved Transient Plasma Ignition in C
2
H
4
-air
Plasma, in a transient state generated by short (ns), high-voltage (kV) pulses,
favorably alter pre-combustion chemistry and physics to reduce ignition delays, but
mechanisms are not well understood. It is known that a transient plasma discharge with a
reduced electric field, E/n, on the order of hundreds of Td results in the production of
active particles in streamer channels through electron impact dissociation, excitation, and
ionization of atoms and molecules, and that these active species significantly impact
chain branching reactions, reducing ignition delay times and allowing for lean burn
combustion (equivalence ratio, ϕ < 0.7) (Starikovskaia, 2006; Wang, 2005; Cathey,
2007). An experiment was performed that showed that ignition occurs where the density
of active particles produced in the discharge is highest, at the base of the streamer, rather
than in the streamer channel between the electrodes, resulting in multiple spatially
separated ignition sites.
The experimental apparatus for these studies was described in detail in Section
3.3. It incorporates a stainless steel cylindrical combustion chamber, two pulse
generators used independently, and a gas manifold with controller gauges. Ethylene-air
(C
2
H
4
-air) with ϕ=1.1 was used for this study. A number of electrode configurations
were used; the typical geometry consisted of an 8-32 stainless steel rod of 3.2 cm in
132
length as the anode with a varying diameter stainless steel tube as the outer electrode (10
- 36 mm) to make gap sizes from 4 - 15 mm. The anode surface was threaded for local
electric field enhancement, although smooth anodes also tested were found to produce
similar results to those reported here. A porous cathode in close proximity to the anode
allowed the flame to propagate through the electrode and reach unburned gases in the
larger combustion volume more quickly. A high-speed camera (IDT X-Stream VISION
XS3) with a 55 mm f/1.2 lens was used to image the combustion process.
Two different pulse generators supplied high-voltage pulses to the electrode with
amplitudes up to 60 kV. The pseudospark pulse generator (Section 2.3.2) delivered a
pulse width of 54 ns, and the solid-state pulse generator (Section 2.4) delivered a pulse
width of 12 ns. By measuring the voltage and current signals of the discharge, it was
possible to determine the energy deposition into the gas, which for the 54 ns pulse
generator was typically 370 mJ/pulse and for the 12 ns pulse generator was typically 70
mJ/pulse. The 54 ns pulse generator was used to achieve streamer breakdown in larger
gaps so that the streamers and the anode could be seen more clearly. Streamers were
imaged using near UV and visible emission from the N
2
(C3Πu, v') →N
2
(B3Πg, v'')
electronic transition. The intensity of emission is nearly directly proportional of the
population of N
2
(C3Πu, v'), which provides an analog to the magnitude of the local
electric field and the concentration of active species in the gas (Pancheshnyi, 2000).
The 54 ns and 12 ns pulses produced similar results while delivering pulses with
the same amplitude but different pulse widths, despite the difference in energy delivered,
as demonstrated previously (Singleton, 2009). The ignition delays measured were
133
10.9±1% ms, 5.5±2% ms, and 6.1±1% ms for spark ignition, the 54 ns pulse, and 12 ns
the pulse, respectively.

At atmospheric pressure using a coaxial electrode geometry with constant gap
distance, multiple flame kernels were formed along the anode at the bases of the streamer
channels. The spatially separated flame kernels propagated outward resulting in a
cylindrical, pointed flame front (Figure 69). The flame continued to propagate outward
from the anode unobstructed along the streamer channels until it reached the porous
cathode. The flame passed through the holes, forming a cylindrical clover shaped flame
front in contrast to the hemispherical smooth flame front in the spark ignited case (Figure
70). Using a sequence of images taken at 700 fps, an average increase in flame speed of
15±5% was measured in the transient plasma ignited cases relative to the spark ignited
case.

Figure 69 – Left: Streamers generated by a 56 kV, 54 ns pulse (Maximum E/n ≈
400 Td); Right: Flame propagation from multiple ignition sites at the base of the
streamers after a single pulse in ϕ=1.1 C
2
H
4
-air.
134

In order to investigate the cause of the location of ignition, screws were added to
the cathode to create four points where local electric field enhancement would occur.
During a transient plasma discharge, the intensity of optical emission from the streamers
was observed to be highest at the tip of the screws, as well as along the anode. This
supports the hypothesis that this configuration produces increased densities of active
species at the cathode compared to the traditional constant gap coaxial geometry. A
single 12 ns, 50 kV pulse was applied to a ϕ=1.1 C
2
H
4
-air mixture and ignition was
shown to occur simultaneously at the tip of each screw, as well as at the bases of the
streamer channels along the anode (Figure 71). These results confirmed that ignition
occurs where the electric field is enhanced in the streamer channels, whether that is near
the anode or the cathode.

Figure 70 – Combustion of ϕ=1.1 C
2
H
4
-air 6 ms after ignition. Left: conventional
spark ignition using a standard 105 mJ spark ignition system with a spark plug.
Right: transient plasma ignition using a 12 ns, 52 kV pulse in a 6 mm gap.
135

This result is significant because it implies that multiple spatially separated
ignition sites can be generated efficiently and employed to improve combustion
efficiency (Ronney, 1994; Phuoc, 2000). It is worth noting a useful benefit in that the
energy distributed to each site is small—the overall energy requirement is approximately
the same or less than that required for traditional spark ignition, and some reduction in
electrode erosion may be anticipated (Cathey, 2007).
In these studies, the 12 ns ignition pulse delivered approximately 70 mJ to the
mixture, distributed across many streamer channels (typically ≈50). In that discharge,
≈1.5 mJ is applied to the fuel-air mixture in each streamer. Although the distribution of
energy to specific states is complex, studies at high E/n (hundreds of Td) in N
2
concluded
that a significant fraction of energy is deposited in the excitation of electronic states and

Figure 71 – Electrode with field enhancement at four protrusions from the
cathode. Left: Photograph showing ten 12 ns, 50 kV pulses in air; Right: Ignition
occurring at multiple locations after a single pulse in ϕ=1.1 C
2
H
4
-air.
136
dissociation of molecules relative to vibrational processes (Nagpal, 1996). In air it is
known that about 30% of the energy spent on the production of active species is
transformed into heat over ~1 µs (Popov, 2001). Therefore, when considering the
mechanism of ignition via transient plasma discharge, it is important to address the
thermal phase of the mechanism after the active species have been produced, and the
minimum energy required for ignition.
The minimum thermal ignition energy of the mixture, E
min
, is estimated as the
amount of energy needed to raise a spherical volume of fuel-air mixture, known as the
quenching distance, d
q
, up to flame temperature (Ballal, 1977). For C
2
H
4
-air at standard
conditions it is on the order of 100’s of µJ (Fenn, 1951). For ignition to occur, it is
critical that the principal dimensions of the thermal kernel are equal to or greater than d
q

everywhere to ensure that heat is not dissipated faster than it is produced from chemical
reactions. For quiescent, gaseous mixtures, d
q
is determined by two factors: 1) the
thermal diffusivity – the faster the heat escapes the larger the quenching distance; and 2)
the laminar flame speed – the faster the flame propagates the smaller the quenching
distance. For C
2
H
4
-air at standard conditions, d
q
is on the order of 1 mm (Calcote, 1952).
In addition to the results presented here, previous experiments have shown that
TPI results in faster flame propagation than traditional spark ignition due to increased
reaction rates as a result of the non-thermal production of active species (Cathey, 2007;
Cathey, 2008). Recently, Bourig et al. performed simulations using CHEMKIN, and
reported that laminar flame speed increases by 15% when adding 5% excited oxygen into
a gaseous H
2
-O
2
mixture (Bourig, 2009). Considering higher concentrations, the authors
137
reported an exponential increase in laminar flame speed with an increase in O
2
(a
1
Δg) and
O(
1
D) percentage. The excitation of oxygen molecules to O
2
(a
1
Δg) results in a
significant enhancement of the formation of highly reactive atoms such as O, H and OH
radicals.
The production of active species is also related to the shorter ignition delays
reported here. In particular, recent work by Aleksandrov et. al. investigating the
mechanism of ignition by non-equilibrium plasma showed that the decrease in ignition
delay is related to the density of oxygen atoms produced in the discharge via electron
impact dissociation (Aleksandrov, 2009). The authors also reported that the high-voltage
nanosecond discharge led to a significant decrease in the gas temperatures at which the
mixtures would ignite.
In a fuel-air mixture at atmospheric pressure and room temperature, evidence
suggests that flame initiation results from chemical energy release resulting from
exothermic fuel oxidation in reactions with low temperature radicals, and from fast gas
heating due to the decay of active species (Popov, 2001). Initially, radical pools are
generated in the high E/n conditions of a transient plasma discharge, which aids
significantly in starting chain branching reactions (Starikovskaia, 2006). However, the
lifetimes of these radicals are short in comparison to the ignition delay time (ns to µs
compared with ms). As the initial flame kernels grow from the bases of the streamers,
radicals further away from the anode (or cathode) recombine before the flame arrives,
resulting in an increase in local temperature along the streamer channels. The
temperature increase is not enough to ignite the mixture (tens of K), but may nevertheless
138
accelerate chemical reactions (Gaivoronskii, 1986). These processes are complex and are
the subject of ongoing investigation. Additionally, the recombined species can have
substantial effects on the combustion chemistry (e.g. O
3
, which is produced in the
discharge and has been shown to increase flame speed (Kelley, 1957)).
The results from this work show that ignition sites resulting from transient plasma
occur at multiple spatially separated locations, and in particular, at the base of the
streamers, where they aid in processes supported by high local electric fields and E/n. As
a result, a significant fraction of the power is deposited in electronic excitation of radical
ions and dissociative species. Among the active species produced, excited species of
oxygen are known to reduce ignition delays, increase flame speed, and reduce the energy
required to ignite the fuel-air mixture. A thermal phase of TPI that occurs on a longer
time scale after the active species are produced increases reaction rates and results in
ignition. Based on these results, TPI has potential to improve combustion efficiency
compared to traditional spark ignition for wide-ranging applications to engine
technology, and also through improved electrode life.

139
Chapter 5: Conclusion & Future Work
5.1 Conclusion
A compact, solid state 12 ns pulse generator was developed and successfully
implemented in transient plasma ignition experiments. Transient plasma was applied in
both quiescent and flowing fuel-air mixtures in a pulse detonation engine (PDE), an
internal combustion engine (ICE), and a constant volume reactor. In the PDE at NPS and
in the constant volume reactor, it was demonstrated that a 12 ns, 70 mJ pulse could be
used to achieve the same ignition delays and peak pressures as those produced by an 85
ns, 800 mJ pulse. Additionally, in the PDE at WPAFB, it was shown that a 12 ns, 10 mJ
pulse could be used to produce faster flame propagation than a traditional spark plug.
These results demonstrate that more compact, lower-energy pulse generators may be used
for transient plasma ignition.
In the constant volume reactor, it was shown that even if a short (<100 ns) fast
rise time (<50 ns) pulse transitions to an arc, significant gains in ignition delay can be
achieved compared to traditional thermal ignition. It was also demonstrated that
increasing the length of the anode can result in an increase in peak pressure in the
combustion volume, and therefore, an increase in combustion efficiency by increasing the
number of ignition kernels after the discharge and distributing radicals in the volume.
140
When a short anode and a discharge volume similar to a spark plug were used, an
improvement in ignition delay was still achieved through the TPI mechanism. The
optimum gap size for the 12 ns pulse generator was determined to be 6 mm, and the pulse
amplitude should result in E/n >360 Td for maximum reductions in ignition delay.
It was demonstrated that water inhibits the performance of TPI and optical
diagnostic techniques were used show that an increase in humidity of from 0.01% to 1%
resulted in a 67% decrease in O
3
concentration, due to a reduction in atomic oxygen
production. The peak density of OH measured in the discharge was 4.6 × 10
14
/cm
3
,
which decayed below 1.3 × 10
14
/cm
3
at 100 µs.
It was shown that ignition sites resulting from transient plasma occur at multiple
spatially separated locations, and in particular, at the base of the streamers, where they
aid in processes supported by high local electric fields and E/n. Evidence was presented
that there are two distinct phases occurring during ignition; an initial non-thermal phase,
and a subsequent volume-distributed thermal phase. It was concluded that TPI has
potential to improve combustion efficiency compared to traditional spark ignition for
wide-ranging applications to engine technology.
Although the ignition mechanism of TPI is not fully understood, transient plasma
ignition has been shown to have significant advantages over spark ignition. An increased
understanding of TPI will lead to improvements of pulsed power technology critical to
this work. Improved transient plasma ignition technology can result in better fuel
efficiency and better control of the burning processes in various types of engines. The
141
ability to ignite under the ultra-lean condition in an automobile engine, for example, can
have a great impact on pollutant emissions.
5.2 Future Work
At the core of this research are the applications of pulsed power, which are
copious and diverse. In the following sections, future work in a variety of applications is
briefly described, including an experiment to continue the investigation of the mechanism
behind transient plasma ignition by conducting a time resolved atomic oxygen
concentration measurement, and experiments to continue my work involving the pulsed
electric field treatment of wine grapes.
5.2.1 Measurement of Atomic Oxygen in Transient Plasma Ignited CH
4
-Air
To follow up on my studies examining the impact of water on transient plasma
ignition, time resolved atomic oxygen concentration should be measured in transient
plasma ignited methane-air (CH
4
-air). Previous results showed that the presence of
humidity had a significant impact on the production of O
3
in a transient plasma discharge,
and evidence suggested that this was due to the reduction of atomic oxygen in the
presence of H
2
O.
In a CH
4
-air mixture, O(
1
D) reacts with CH
4
to produce CH
3
•, an important
radical in the combustion process. As previously discussed, free radicals are critical to
initiating and maintaining combustion as they seed many of the chain branching and
142
chain propagation reactions. Most of the work done looking at active species produced
by non-equilibrated plasmas is at much lower pressures (~60 Torr). This experiment
would determine the time-resolved concentrations of atomic oxygen created by transient
plasma in a quiescent CH
4
-air mixture at atmospheric pressure. The two-photon
absorption laser induced fluorescence (TALIF) technique could be used to measure
concentrations of O(
3
P) in order to indirectly monitor O(
1
D). This experiment would
help determine to what extent the transient plasma is seeding the discharge volume with
these radicals, as well as whether the decrease in performance using TPI in humidity is
due to a decrease in of the formation of atomic oxygen.
Two-Photon Absorption Laser-Induced Florescence (TALIF)
The most common method used to investigate minor species in combustion is
laser-induced florescence (LIF). In this technique, a laser is tuned to a wavelength to
match an absorption line of the atom or molecule of interest, causing that species to be
elevated to an excited state and fluoresce. A few minor species, including O, H, N and
CO, require tuning the laser to a wavelength so that two photons are absorbed, because
the species cannot be accessed by the absorption of a single photon (Jeffries, 2002). The
technique has been used in a variety of combustion and plasma experiments and is well
documented (Goldsmith, 1986; Tserepi, 1995; Uddi, 2009).
For atomic oxygen, the transition between the ground 2p
3
P state and the excited
3p
3
P state is two-photon, one-color allowed with a photon wavelength of 225.7 nm
(Figure 72). From this excited state the atom can make a single photon relaxation to the
143
3s
3
S state, with emission at 844.6 nm. The detection threshold for TALIF for atomic
oxygen is in the range 10
11
–10
12
cm
−3
(Amorim, 2000).

Figure 72 – Schematic of atomic oxygen energy levels showing the two photon
absorption and emission. Two photons at 225.7 nm are absorbed, exciting the O atom
to the 3p
3
P state. The fluorescence at 845 nm is then measured (Bamford, 1986).
Additional Considerations
The TALIF measurements can be performed in the cylindrical stainless steel static
combustion reactor used for previous combustion experiments, however, the coaxial
electrode configuration is not ideal since the exact location of the streamers cannot be
controlled, so a point to plane or wire to plane geometry could be used.
144
An important consideration in looking at atomic oxygen with TALIF is the
photolysis of O
2
, which is when the UV laser produces the measured atomic oxygen
atoms, resulting in an inflated fluorescence signal. This problem is especially significant
at high temperatures, for example after ignition has occurred. Since this experiment will
be looking at atomic oxygen concentrations prior to ignition, the temperatures will be
relatively low. However, one way to mitigate this problem is to collimate the laser beam
to a diameter of approximately 1 mm using a Galilean telescope (Oostendorp, 1993).
Using the collimated beam reduces the intensity, and therefore the fluorescence. The
beam has a lower photon flux density so the weaker signal is somewhat mitigated. The
desired beam intensity is ~0.1 GW/cm
2
or less.
5.2.2 Pulse Generator Design
A nanosecond pulsed power system comparable in size to an automobile ignition
coil would be enabling technology for more efficient combustion in a variety of engines.
This requires careful design of pulsed power circuits, as well as the development of
compact capacitors, magnetic cores, and high-voltage insulators. As a first step, the
current 12 ns solid state pulse generator could be modified to be made more compact, and
could be re-designed to interface with an automobile ignition system (12 V battery,
connection to an electronic ignition control system). The majority of the volume in the
12 ns pulse generator is the HV capacitors and the magnetic cores, so to decrease system
size, smaller replacements must be used.
145
The 12 ns solid state power generator developed for this research utilizes
saturable magnetic cores for a magnetic compression stage. Saturable cores are useful
because they can switch high currents, but they also make for a significant amount of the
weight and size of the system in its current configuration. Removing the saturable cores
would significantly reduce the size and weight of the system. A similar design to an air-
core based pulse generator used at USC for biomedical applications could potentially be
applied for the generation of transient plasma (Sanders, 2009). Since the amplitude of the
pulse produced by this architecture is much lower (around 5 kV), the gap size would have
to be decreased significantly to achieve a high maximum electric field strength. Since
ignition occurs only along the anode and not in the volume, a smaller gap size may not
impact combustion significantly. However, the volume in which radicals are produced
will be significantly less, which could negatively affect combustion performance.
Another issue that needs to be addressed is cable design. A cable for the 12 ns
pulse generator for an ignition application needs to be high-impedance, shielded, and
must be able to withstand high temperatures. With this and the other described
developments, the application of pulsed power to ignition and combustion will be at the
point where it could be commercialized.
5.2.3 Electrode Design for an Internal Combustion Engine
A modified spark plug was used in an ICE experiment with Earnhardt-Childress
Racing and in a PDE experiment with the AFRL. While the igniter worked under certain
conditions, there were problems with the design. A recessed volume in the plug was not
146
easily purged of gas each cycle, leaving pre-ionized gas in the volume at faster repetition
rates, which resulted in arcing (Figure 73). Additionally, the ceramic in the modified
plug is not rated for the voltages being applied. While the ceramic will work for a while
since it is being operated in pulsed mode, it will eventually break down. The ceramic
boron nitride insulator could be used on the new electrode. The ratio of the diameter of
the insulator to the diameter of the cathode will also need to be considered because it
determines the temperature range for the plug. The new electrode could include a
cathode so that it is one complete unit, similar to a spark plug. The gap size of the
transient plasma igniter is limited by the geometry of the ignition cylinder, so the
maximum gap size should be used.

Figure 73 – Proposed electrode design that eliminates the undesirable recessed
volume.
147
5.2.4 Soot Reduction
Soot emission from practical combustion devices such as automobile engines
causes serious problems to human health and the environment, and is an undesirable
effect with regard to combustion efficiency (Cignoli, 1994; Hayashida, 2006). The
combustion community recognizes soot reduction as a significant challenge and there is
great interest in reducing soot formation (Santoro, 2002). In a preliminary experiment
conducted at USC comparing soot formation using transient plasma ignition (TPI) and
spark ignition (SI), TPI has shown the potential to produce less soot (Figure 74).

Figure 74 – Particle size distribution of soot produced during ignition by spark and
transient plasma for a rich ethylene-air mixture.

148
Further work is necessary to verify and explain these results. It is postulated that
TPI may produce less soot because it takes less time to combust the mixture, essentially
shortening the window in which soot is produced. More complex chemical processes,
such as the O
3
produced by TPI consuming soot precursors, may also be a factor.
5.2.5 Gas Turbine Ignition and Relight
Gas turbine powered aircraft are certified to certain altitudes depending on their
ability to relight the engines in the event of engine flame out (Figure 75). Relighting the
engine at high altitudes is difficult since the cold conditions prevent effective fuel
vaporization. The emergency procedure for an engine flameout is essentially to descend
and attempt to relight. For military aircraft, if the aircraft must descend to a lower
altitude in order to relight the engine, the mission of that aircraft can be compromised or
aborted. Transient plasma ignition may be able to extend the in-flight relight envelope
due to demonstrated lower lean ignitability limit, ability to ignite high mass flow rates,
and its improved performance with liquid fuels compared to spark ignition.

149

5.2.6 Pulsed Electric Field Treatment of Wine Grapes
The application of pulsed electric fields (PEF) has been demonstrated to increase
juice yield and quality in white wine grapes in my experiments and in published literature
(Praporscic, 2007). An electric field of moderate intensity (500-1000 V/cm), applied to
the grape tissue causes an irreversible electroporation of the cell membranes, allowing
selective extraction of cell components without thermally affecting the grapes (Bazhal,
2001). The high-voltage electrical pulses can be applied to a stationary volume, or to a
flowing volume, suggesting its application to grape processing in batch or on a
continuous basis.
Figure 75 – A typical flight relight envelope for gas turbines (Rolls Royce).
150
A 2 µs, 20 kV pulse generator designed for a 10 Ω load was developed for this
application, and an experiment was performed in collaboration with Dr. Waterhouse’s
group at UC Davis. Electroporation of the cell membrane in two white wine grape
varietals, Chardonnay and Sauvignon Blanc was shown to positively increase juice yield
as much as 30%, as well as reduce the effect of browning, shown in Figure 76.

Figure 76 – Pulsed (left) and non-pulsed (right) juice extracted from Chardonnay
grapes. The pulsed sample yielded significantly more juice and better color.

The next experiment to be performed on this work should examine red wine
grapes. It is theorized that cell permeability will be positively affected in red grapes,
releasing more anthocyanins thereby improving grape and wine quality. A considerable
benefit to industry is envisioned if electroporation increases anthocyanin diffusion, and
reduces losses from enzymatic oxidation, thus avoiding lengthy macerations and/or
higher temperatures sometimes needed for maximum extraction. Rapid color extraction
151
without browning losses would well be suited to the wine producer using grapes that are
have difficulties with color due to the environment in which the grapes are grown. There
is also a growing trend among wine producers concerned with the practice of
sustainability and environmental consciousness. This new technology could reduce the
need for synthetic chemicals such as Etherel or additions like color enzymes, in addition
to increasing both the yield and quality of white juice.

152
Bibliography
Abramoff, M.D., P.J. Magelhaes, and S.J. Ram, "Image Processing with ImageJ".
Biophotonics International, volume 11, issue 7, pp. 36-42, 2004.
Aleksandrov, N. L., E . M. Bazelyan, and D. A. Novitski, Influence of moisture on the
properties of long streamers in air. Technical Physics Letter, Volume 24, Number 5,
May 1998.
Aleksandrov, N. L., N. B. Anikin, E. M. Bazelyan, D. V. Zatsepin, S. M. Starikovskaia,
A. Yu. Starikovskii “Chemical Reactions and Ignition Initiation in Hydrocarbon-Air
Mixtures by High-Voltage Nanosecond Gas Discharge,” 32nd AIAA Plasma
dynamics and Lasers Conference and 4th Weakly Inized Gases Workshop, June 2001
Anaheim, CA, AIAA paper 2001-2949.
Aleksandrov, N. L., S. V. Kindysheva, I. N. Kosarev, S. M. Starikovskaia, A. Y.
Starikovskii, “Mechanism of ignition by non-equilibrium plasma,” Proceedings of the
Combustion Institute 32 (2009) 205-212.
Alstom FS2000 pseudospark switch datasheet.
Amorim, J., G. Baravian,J. Jolly, “Laser-induced resonance fluorescence as a diagnostic
technique in non-thermal equilibrium plasmas”, Journal of Physics D: Applied
Physics, Volume 33, Issue 9, pp. R51-R65 (2000).
Anderson, M., M. Gundersen, J. Sanders, D. Singleton, and A. Waterhouse, “Effects of
Pulsed Energy Field Treatments on White Wine Grapes.” American Society for
Enology and Viticulture, June 2009.
Atkinson, R., D. L. Baulch, R. A. Cox, R. F. Hampson, Jr, J. A. Kerr and J. Troe,
“Evaluated kinetic and photochemical data for atmospheric chemistry supplement”
IV J. Phys. Chem. Ref. Data 21 1125, 1992.
Ballal, D. R. and A. H. Lefebvre (1977). "Ignition and Flame Quenching in Flowing
Gaseous Mixtures." Proceedings of the Royal Society of London. Series A,
Mathematical and Physical Sciences 357 (1689): 163-181.
Ballal, D. R., A. H. Lefebvre, “The Structure and Propagation of Turbulent Flames,”
Proceedings of the Royal Society of London. Series A, Mathematical and Physical
Sciences, Vol. 344, No. 1637 (Jun. 24, 1975), pp. 217-234.
Bamford, J. D., L. E. Jusinski, W. K. Bischel, “Absolute two photon absorption and three
photon ionization cross sections for atomic oxygen”, 1986, Vol 34, No 1, Physical
Review A, pp 185- 198.
153
Barlow, R. S., A. N. Karpetis, J. H. Frank, and J.-Y. Chen, Combust. Flame 127 (2001)
2102–2118.
Barrett, D. M. “Core reset considerations in magnetic pulse compression networks”
Pulsed Power Conference, 1995. Digest of Technical Papers. Tenth IEEE
International, Vol. 2, Iss., 3-6, Pages: 1160-1165, Jul 1995.
Bazhal, M. I., N. I. Lebovka, E. Vorobiev, “Pulsed electric field treatment of apple tissue
during compression for juice extraction”, J. Food Eng. 50 (2001) 129–139.
Becker, K. H., U. Kogelschatz, K. H. Schoenbach, R. J. Barker, “Non-Equilibrium Air
Plasmas at Atmospheric Pressure”, IOP Publishing LTD, Philadelphia, 2005.
Birsall, C. K., and A. B. Langdon, Plasma Physics via Computer Simulation. Taylor and
Francis, London, 2004.
Bischel, W. K., B. E. Perry, D. R. Crosley, “Detection of fluorescence from 0 and N
atoms induced by two-photon absorption”, Applied Optics 21 (8) (1982) 1419–1429.
Blair, D. T. A., F. M. Bruce and D. J. Tedford, 1963 Proc. 6th Int. Conf. on Phenomena
in Ionised Gases, Paris Vol II (Paris: Association) p 321.
Blicher, A., “Thyristor Physics,” Springer-Verlag New York Inc., New York, 1976.
Blumlein, A. D. UK Patent 589127 (1941)
Bourig, A., D. Thévenin, et al. (2009). "Numerical modeling of H2-O2 flames involving
electronically-excited species O2(a1[Delta]g),O(1D)O(1D) and OH(2[Sigma]+)."
Proceedings of the Combustion Institute 32(2): 3171-3179.
Bozhenkov, S. A., S. M. Starikovskaya, and A. Yu. Starikovskii, “Nanosecond Gas
Discharge Ignition of H2- and CH4- Containing Mixtures,” Combustion and Flame,
133 (2003) 133-146, 2003.
Bull, D., C., Elsworth, J. E., and Hooper, G., “Initiation of Spherical Detonation in
Hydrocarbon/Air Mixtures," Acta Astron., Vol. 5, No. 11, 1978, pp. 997-1008.
Burkes, T. R., J. P. Craig, M. O. Hagler, M. Kristiansen, W. M. Portnoy, “A review of
high-power switch technology,” IEEE Transactions on Electron Devices, Vol 26,
Issue 10, Oct 1979 Page(s):1401 – 1411.
Burris, J., J. J. Butler, T. J. McGee, and W.S. Heaps, “Collisional deactivation rates for
A 2+(v’=1) state of OH,” Chem. Phys. 124 (1988) 251-258.
154
Busby, K., J. Corrigan, S. Yu, S. Williams, C. Carter, F. Schauer, J. Hoke, C. Cathey, and
M.A. Gundersen, “Effects of Corona, Spark and Surface Discharges on Ignition
Delay and Deflagration-to-Detonation Times in Pulsed Detonation Engines” AIAA-
2007-1028 at 45th Aerospace Sciences Meeting , Reno, NV, Dec. 2006.
Bussing, T. R. A., T.E. Bratkovich, and J. B. Hinkley Jr, “Practical Implementation of
Pulse Detonation Engines,” AIAA 2748-1997.
Calcote, H. F., C. A. Gregory, C. M. Barnett, R. B. Gilmer, “Spark Ignition. Effect of
Molecular Structure,” Industrial & Engineering Chemistry 1952 44 (11), 2656-2662.
Cathey, C., “Spatial and Time Resolved Study of Transient Plasma Induced OH
Production in Quiescent CH4-Air Mixtures”, PhD Dissertation, University of
Southern California, August 2007.
Cathey, C., J. Cain, H. Wang, M. A. Gundersen, C. Carter, and M. Ryan, “OH production
by transient plasma and mechanism of flame ignition and propagation in quiescent
methane-air mixtures,” Combustion and Flame 154 (2008) 715–727, 2007.
Cathey, C., T. Tang, T. Shiraishi, T. Urushihara, A. Kuthi, and M. A. Gundersen,
“Nanosecond Plasma Ignition for Improved Performance of an Internal Combustion
engine”, IEEE Trans on Plasma Sci., Dec. 2007.
Ciccarelli, G., Card, J., “Detonation in Mixtures of JP-10 Vapor and Air,” AIAA Journal,
Vol. 44, No. 2, February 2006.
Cignoli, F., S. Benecchi, and G. Zizak, “Time-Delayed Detection of Laser-Induced
Incandescence for the Two-Dimensional Visualization of Soot in Flames”, Applied
Optics . Vol. 33, No. 24. 20 August 1994, pp. 5778-5783.
Eichenberger, D. A., W. L. Roberts. Combust. Flame 118, 469 (1999).
Fenn, J. B., “Lean Flammability Limit and Minimum Spark Ignition Energy”, Industrial
& Engineering Chemistry 1951 43 (12), 2865-2869.
Fickett, W., and W. C.Davis, "Detonation". University of California Press, Berkeley,
1979.
Frank, J. H., T. B. Settersten, “Two-photon LIF imaging of atomic oxygen in flames with
picoseconds excitation”, Proceedings of the Combustion Institute 30 (2005) 1527-
1534.
Fridman, A., and L. A. Kennedy, Plasma Physics and Engineering, Taylor and Francis,
New York, 2004.
155
Gaivoronskii, A. S., and I. M. Razhanskii, (1986) “Numerical model of the development
of cathode-directed streamer in air gaps with non-uniform field” Zh. Tekh. Fiz. 56
1110–6.
Gallimberti, I., Pure Appl. Chem. 60, 663-674 (1988).
Gaudet, J., R. J. Barker, M. A. Gundersen, A. Kuthi, A. Neuber, J. Dickens, W. Nunnally,
E. Schamiloglu, C. Christodoulou, S. Tyo, K. H. Schoenbach, R. P. Joshi, M.
Laroussi, J. Kolb, R. Vidmar, “Research Issues in Developing Compact Pulsed Power
for High Peak Power Applications on Mobile Platforms.” Proc IEEE 92:1144–1165,
2004.
Glassman, I., Combustion, 2nd Ed., Academic Press, 1996.
Goldsmith, J. E. M., N. M. Laurendeau, “Two photon excited fluorescence measurements
of OH concentration in a hydrogen in a hydrogen-oxygen flame”, 1986, Vol 25, No 2,
Applied Optics, pp. 276-283.
Hackard, C., “Ignition characteristics for transient plasma ignition of ethylene/air and jp-
10/air mixtures in a pulse detonation engine.” Master’s Dissertation, Naval
Postgraduate School, Monterey, California, December 2007.
Hanson, R. K.., D. W. Mattison, L. Ma, D. F. Davidson, and S. T. Sanders, "Laser
Diagnostics and Combustion Chemistry for Pulse Detonation Engines" Chap. 10, pp
365-376 in Combustion Processes in Propulsion, ed. G. Roy, Elsevier Press, 2006.
Hartlieb, A.T., D. Markus, W. Kreutner, K. Kohse_Hoinghaus, Appl. Phys. B. vol. 65
(1997), 81-91.
Hartmann, W., and M. A. Gundersen, “Origin of anomalous emission in superdense glow
discharge,” Phys. Rev. Lett., vol. 60, pp. 2371–2374, 1988.
Hayashida, K., K. Amagai, K. Satoh, A. Masataka, “Experimental Analysis of Soot
Formation in Sooting Diffusion Flame by Using Laser-Induced Emissions”, Journal
of Engineering for Gas Turbines and Power. Vol. 128, April 2006, pp. 241-246.
Hilliard, J. C., G. S. Springer, Fuel Economy in Road Vehicles Powered by Spark Ignition
Engines, Plenum Press, New York, 1984.
Hutcheson, P.D., “Design, Modeling and Performance of a Split Path JP-10/air Pulse
Detonation Engine,” Master’s Dissertation, Naval Postgraduate School, Monterey,
California, December 2006.
Kelley, R. J., “Effect of ozone upon flame speeds of some combustible mixtures”,
University of Michigan, College of Engineering Technical Reports, UMR2392, 1957.
156
Kentfield, J. A. C., “Fundamentals of Idealized Airbreathing Pulse-Detonation Engines,”
Journal of Propulsion and Power, Vol. 18, No. 1, January-February 2002.
Kinema Boltzmann solver from Kinema Research Software (http://www.kinema.com),
and references therein.
Kliner, D. A., and R. L. Farrow, J. Chem. Phys. Vol. 110 (1999) 412-422.
Klingbeil, A. E., J. B. Jeffries, and R. K. Hanson, "3.39 micron Laser Absorption Sensor
for Ethylene and Propane Measurements in a Pulse Detonation Engine," AIAA-2006-
0925 at 44th Aerospace Sciences Meeting, Reno, NV, Jan. 2006.
Klingbeil, A. E., J. B. Jeffries, and R. K. Hanson, "Design of a Fiber-Coupled Mid-IR
Fuel Sensor for Pulse Detonation Engines,” AIAA J. 45, 772-778 (2007).
Klopovskiy, K. S., D. V. Lopaev, O. V. Proshina, A. T. Rakhimov, and T. V. Rakhimova,
“Effect of the Nonlocal Nature of the Electron Energy Spectrum on the Dissociation
of Oxygen Molecules in a Discharge”, Plasma Physics Reports, Vol. 30, No. 6, 2004,
pp. 542 548. Translated from Fizika Plazmy, Vol. 30, No. 6, 2004, pp. 586–592.
Kohse-Hoinghaus, K., “Laser Techniques for the Quantitative Detection of Reactive
Intermediates in Combustion Systems,” Prog. Energy Combust. Sci. Vol 20, 1994.
Kossyi, A., A. Yu. Kostinsky, A. A. Matveyev, and V.P. Silakov, “Kinetic Scheme of the
Nonequilibrium Discharge in Nitrogen-Oxygen Mixtures, Plasma Sources Sci.
Technol., vol. 1, 1992, pp. 207-220.
Kuo, K., Principles of combustion, 2nd Ed., John Wiley & Sons, 2005.
Liu, J. B., F. Wang, G. Q. Li, , A. Kuthi, E. J. Gutmark, P. D. Ronney, and M. A.
Gundersen, “Transient plasma ignition,” IEEE Transactions on Plasma Science,
33:326-327, 2005.
Loeb, L. B., J. M. Meek, The Mechanism of the Electric Spark, Stanford: University
Press, 1941.
Luginsland, J. W., D. Singleton, M. Gundersen, C. Brophy, J. Sinibaldi, E. Barbour, D.
Waxman, R. Hanson, “The Role of Water in Transient Plasma Ignition for
Combustion,” IEEE 36th International Conference on Plasma Science, 2009
Lyubutin, S. K., G.A. Mesyats, S.N. Rukin; B.G. Slovikovskii, "Repetitive Nanosecond
All-Solid state Pulsers Based on SOS diodes," Pulsed Power Conference, 1997.
Digest of Technical NPapers. 1997 11th IEEE International , vol.2, no., pp.992-998
vol.2, 29 Jun-2 Jul 1997.
157
Ma, L., S. T. Sanders, J. B. Jeffries, and R. K. Hanson, “Monitoring and control of a
pulse detonation engine using a diode-laser fuel concentration and temperature
sensor,” Proceedings of the Combustion Institute, Volume 29, pp. 161-166, 2002.
Mankowski, J., M. Kristiansen, "A Review of Short Pulse Generator Technology," IEEE
Transactions on Plasma Science , vol.28, no.1, pp.102-108, February 2000.
Marx, E. “Versuche uber die Prufing von Isolatoren mit Spannugsstossen”. Elektro-
technishnishe Zeitschrift 45 (1925) 625
Matzing, H. “Chemical kinetics of flue gas cleaning by irradiation with electrons” Adv.
Chem. Phys. 80 315, 1991.
Meek, J. M., J. D. Craggs, Electrical Breakdown of Gases, Wiley: New York (1978).
Melville, W. S., “The Use of Saturable Reactors as Discharge Devices for Pulse
Generators,” Proc. IEE 98 (1951) 185-207.
Mintusov, E., A. Serdyuchenko, I. Choi, W. R. Lempert, I. V. Adamovich, “Mechanism
of plasma assisted oxidation and ignition of ethylene-air flows by a repetitively
pulsed nanosecond discharge”, Proceedings of the Combustion Institute, Volume 32,
Issue 2, 2009, Pages 3181-3188.
Mizuno, A., J. S. Clements, R. H. Davis, IEEE Trans. Ind. Appl. 22, 58-64 (1986).
Mori, K., N. Goto-Yamamoto, et al. (2007). "Loss of anthocyanins in red-wine grape
under high temperature." J. Exp. Bot. 58(8): 1935-1945.
N.A. Popov, Plasma Physics Reports 27 (10), 886 (2001).
Nagpal, R., B. N. Ganguly, P. Bletzinger, A. Garscadden, “Power deposition in H2 and
H2-N2 glow discharges,” Chemical Physics Letters 257 (1996) 386-392.
Niemi, K., V. Schulz-von der Gathen, and H. F. Döbele, “Absolute atomic oxygen
density measurements by two-photon absorption laser-induced fluorescence
spectroscopy in an RF-excited atmospheric pressure plasma jet”, Plasma Sources Sci.
Technol. 14 375-386, 2005.
Oda, T., Y. Yamashita, K. Takezawa, R. Ono, “Oxygen atom behavior in the nonthermal
plasma, Thin Solid Films”, Volumes 506-507, The Joint Meeting of 7th APCPST
(Asia Pacific Conference on Plasma Science and Technology) and 17th SPSM
(Symposium on Plasma Science for Materials) - 7th APCPST/17th SPSM, 26 May
2006, Pages 669-673.
158
Oda, T., Y. Yamashita, K. Takezawa, R. Ono, “Oxygen atom behavior in the nonthermal
plasma”, Thin Solid Films, Volumes 506-507, The Joint Meeting of 7th APCPST
(Asia Pacific Conference on Plasma Science and Technology) and 17th SPSM
(Symposium on Plasma Science for Materials) - 7th APCPST/17th SPSM, 26 May
2006, Pages 669-673.
Ombrello, T., X. Qin, Y. Ju, A. Gutsol, A. Fridman, and C. Carter, AIAA J. 44 (2006)
142-150.
Ono, R., and T. Oda, J. Appl. Phys. 93 (2003) 5876-5882.
Ono, R., and T. Oda, J. Phys. D: Appl. Phys. 35 (2002) 2133-2138.
Ono, R., Y.Yamashita, K. Takezawa, and T. Oda, “Behaviour of atomic oxygen in a
pulsed dielectric barrier discharge measured by laser-induced fluorescence”, J. Phys.
D: Appl. Phys. 38 (2005) 2812–2816.
Oostendorp, D. L., H. B. Levinsky, C. E. van der Meij, R. A. A. M. Jacobs, and W. T. A.
Borghols, “Avoidance of the photochemical production of oxygen atoms in one-
dimensional, two-photon laser-induced fluorescence imaging,” App. Opt., 32, 24
(1993).
Pancheshnyi, S., S. Sobakin, et al. (2000). "Discharge dynamics and the production of
active particles in a cathode-directed streamer." Plasma Physics Reports 26 (12):
1054-1065.
Pancheshnyi, S., S. Sobakin, S. Starikovskaya et al., Plasma Physics Reports 26 (12),
1054 (2000).
Paul, P. H., J. of Quantitative Spectroscopy and Radiative Transfer, vol. 51 (1994), 511-
524.
Paul, P. H., J. Phys. Chem. 99 (1995) 8472-8476.
Pendleton, S. J., D. Singleton, A. Kuthi, M. A. Gundersen, “Compact Solid State
Variable Amplitude High Repetition Rate Pulse Generator,” IEEE Pulsed Power
Conference, 2009
Phuoc, T. X., C.M. White, Opt. Commun. 181 (2000) 353
Pierret, R., Semiconductor Device Fundamentals, Addison-Wesley Publishing Company:
Massachusetts, 1996, pg 465.
Polak, L. S., A. A. Ovsiannikov, D. I. Slovetskii, F. B. M. Vurzel. Theoretical and
Applied Plasmochemistry, p. 222, Nauka, Moscow (1975).
159
Popov, N.A., Plasma Physics Reports 27 (10), 886 (2001).
Praporscic, I., N. Lebovka, E. Vorobiev, M. Mietton-Peuchot, “Pulsed electric field
enhanced expression and juice quality of white grapes”, Separation and Purification
Technology, Volume 52, Issue 3, January 2007, Pages 520-526.
Raether, H. Electron Avalanches and Breakdown in Gases, Butterworth & Co: London
(1964).
Rohwein, G. J. “An Efficient, Power-Enhanced Ignition System”, IEEE Transactions on
Plasma Science, vol. 25, bo. 2, April 1997.
Rohwein, G. J., S.R. Babcock, M. T. Buttram, L. S. Camilli, "Advanced automotive
ignition systems," Pulsed Power Conference, 1995. Digest of Technical Papers, Tenth
IEEE International, vol.1, pp.40-45, 3-6 Jul 1995.
Ronney, P. D., Opt. Eng. 33:510 (1994)
Rukin, S. M., S. K. Lyubutin, V. V. Kostirev, V. A. Telnov, "Repetitive 200 kV
Nanosecond All Solid State Pulser with a Semiconductor Opening Switch," Pulsed
Power Conference, 1995. Digest of Technical Papers., Tenth IEEE International , vol.
2, pp.1211-1214 vol.2, 3-6 Jul. 1995.
Samano, E. C., W. E. Carr, M. Seidl, B. S. Lee. Rev. Sci. Instrum. 64, 2746 (1993).
Santoro, R., C. Shaddix, Applied Combustion Diagnostics, Katharina Kohse-Hoinghaus
and Jay B. Jeffries, Taylor and Francis, London 2002, pp. 252-286.
Saxon, R. P., and J. Eichler, “Theoretical calculation of two photon absorption cross
sections in atomic oxygen”, Phys. Rev. A 34, 199 - 206 (1986).
Schamiloglu, E. et al, Proceeding of the IEEE, Vol 92(7) 1014, July 2004.
Settles, G. S., Schlieren and shadowgraph techniques: Visualizing phenomena in
transparent media, Berlin: Springer-Verlag, 2001.
Shiraishi, T., A. Kakuho, T. Urushihara, C. Cathey, T. Tang, M. A. Gundersen, “A Study
of Volumetric Ignition Using High-Speed Plasma for Improving Lean Combustion
Performance in Internal Combustion Engines”, SAE International Journal of Engines
April 2009 1:399-408.
Simek, M., M. Clupek, V. Babicky´, and P. Sunka, “Production of reactive species by
atmospheric pressure streamers in N2–O2 mixtures,” Pure Appl. Chem., Vol. 78, No.
6, pp. 1213–1225, 2006.
160
Singleton, D, C. Cathey, A. Kuthi, and M. A. Gundersen, “Applications of Power
Modulator Technology to Ignition and Combustion”, IEEE International Power
Modulators and High Voltage Conference, Proceedings of the 2008, vol., no., pp.174-
177, 27-31 May 2008.
Singleton, D., A. Simone, and M. A. Gundersen, “Optimization of Compact Power
Modulators for Transient Plasma Ignition”, IEEE International Power Modulators and
High Voltage Conference, May 2010.
Singleton, D., J. Sinibaldi, C. Brophy, A. Kuthi and M. A. Gundersen, “Compact Pulsed
Power System for Transient Plasma Ignition”, IEEE Trans. of Plasma Sci., vol. 37,
no. 12, pp. 2275- 2279, Dec. 2009.
Singleton, D., M. A. Gundersen “Study of the Volumetric Effect of Transient Plasma
Ignition in Quiescent Ethylene-Air,” IEEE 36th International Conference on Plasma
Science, 2009
Singleton, D., M. A. Gundersen, C. Brophy, J. Sinibaldi, J. W. Luginsland,
"Development of High Rep-Rate Pulse Detonation Engines Based on Transient
Plasma Ignition Technology," IEEE 35th International Conference on Plasma
Science, 2008, pp.1-1, 15-19 June 2008.
Starikovskaia, S. M., “Plasma assisted ignition and combustion,” J. Phys. D: Appl. Phys.,
vol. 39, no. 16, pp. R265–R299, Aug. 2006.
Starikovskaia, S. M., A. Y. Starikovskii, D. V. Zatsepin, “Hydrogen oxidation in a
stoichiometric hydrogen-air mixture in the fast ionization wave,” Combustion Theory
and Modelling, 5:97-129, 2001.
Tang, T., A. Kuthi, F. Wang, C. Cathey, and M. A. Gundersen, “Design of 60kV 20 ns
Solid State Pulse Generator Based on Magnetic Reactor Driven Diode Opening
Switch,” 27th International Power Modulator Conference 2006, Washington D. C.,
District of Columbia, May 14-18th, 2006.
Trevisan, C. S., J. Tennyson , “Calculated rates for the electron impact dissociation of
molecular hydrogen, deuterium and tritium,” Plasma Phys. Control. Fusion 44 (2002).
Tropina, A., L. Lenarduzzi, S. Marasov, A. Kuzmenko, “Comparative analysis of engine
ignition systems”, IEEE Transactions on Plasma Science, Special Issue on Plasma
Assisted Combustion, Dec. 2009.
Tserepi, A. D., T. A. Miller, “Spatially and temporally resolved absolute O-atom
concentrations in etching plasmas”, Journal of Applied Physics 77 (2) (1995) 505–
511.
161
Turro, N., Modern molecular photochemistry, University Science Books, April 1991.
Uddi, M., J. Naibo, E. Mintusov, I. V. Adamovich, W. R. Lempert, “Atomic oxygen
measurements in air and air/fuel nanosecond pulse discharges by two photon laser
induced fluorescence”, Proceedings of the Combustion Institute, Volume 32, Issue 1,
2009, Pages 929-936.
Veldhuizen, E. M., and W. R. Rutgers, “Pulsed positive corona streamer propagation and
branching,” J. Phys. D: Appl. Phys. 35 (2002).
Wang, F., A. Kuthi, M. A. Gundersen, “Compact high repetition rate pseudospark pulse
generator,” IEEE Transactions on Plasma Science, 33:1177-1181, 2005.
Wang, F., J. B. Liu, J. Sinibaldi, C. Brophy, A. Kuthi, C.Jiang, P. Ronney, M. A.
Gundersen, “Transient plasma ignition of quiescent and flowing air/fuel mixtures,”
IEEE Transactions on Plasma Science, 33:844-849, 2005.
Wang, H., Course Notes AME 579: Combustion Chemistry and Physics, Fall 2008.
Waters, R. T., “Electrical Breakdown in Gases”, edited by J. M. Meek and J. D. Craggs.
Wiley, New York, 1978 , pp. 385–532.
Yagyu, Y., N. Hayashi, et al. (2008). "Fundamental studies on effect of ozone injection to
the internal-combustion engine - FTIR spectrum of hydrocarbon compound
reformulated by ozone." Journal of Physics: Conference, 2008.
Zhao, G. B., S. V. B. J. Garikipati, X. Hu, M. D. Argyle, and M. Radosz, “Effect of
Oxygen on Nonthermal Plasma Reactions of Nitrogen Oxides in Nitrogen”,
Environmental and Energy Engineering, AIChE Journal Vol. 51, No. 6, June 2005.
Zizak, G., G. A. Petrucci, C. L. Stevenson, and J. D. Winefordner, “Ground state
saturated population distribution of OH in an acetylene-air flame measured by two
optical double resonance pump-probe approaches,” Appl. Opt 30 (1991) 5270-5275.

Abstract (if available)

Abstract This dissertation presents an experimental study of the application and the underlying physics of transient plasma ignition. Transient plasma generated by nanosecond electrical discharges has demonstrated lean-burn capability and reductions in ignition delay in a variety of engines, resulting in higher combustion efficiencies and lowered emissions compared to conventional spark ignition. The experiments performed demonstrate the effects of transient plasma ignition and attempt to understand the basic physics behind it by examining the production of radicals, conditions for ignition, and combustion characteristics.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Spatial and time resolved study of transient plasma induced oh production in quiscent CH4-air mixtures

PDF

Physics and applications of energetic electrons resulting from nanosecond-scale transient plasma

PDF

An experimental investigation by optical methods of the physics and chemistry of transient plasma ignition

PDF

Topics in high voltage pulsed power plasma devices and applications

PDF

Design and implementation of full solid state high voltage nanosecond pulse generators

PDF

Studies of nanosecond pulsed power for modifications of biomaterials and nanomaterials (SWCNT)

PDF

The physics of pulsed streamer discharge in high pressure air and applications to engine technologies

PDF

Development of the two-stage micro pulsed plasma thruster

PDF

A study of ignition effects on thruster performance of a multi-electrode capillary discharge using visible emission spectroscopy diagnostics

PDF

A study of junction effect transistors and their roles in carbon nanotube field emission cathodes in compact pulsed power applications

PDF

Mitigation of ion motion in future plasma Wakefield accelerators

PDF

Transient pulsed plasma for pollution remediation and energy conversion applications

PDF

Biophysical studies of nanosecond pulsed electric field induced cell membrane permeabilization

PDF

Power density modes for determining electromagnetic source fields

PDF

Physics of particle trapping in ultrarelativistic plasma wakes

PDF

Applications of optical microring and microdisk resonators as physical, chemical, and biological sensors

PDF

The cathode plasma simulation

PDF

Kinetic studies of collisionless mesothermal plasma flow dynamics

PDF

Power efficient multimedia applications on embedded systems

PDF

Development of impulse measurement techniques for the investigation of transient forces due to laser-induced ablation

Asset Metadata

Creator Singleton, Daniel R. (author)

Core Title The physics and application of compact pulsed power to transient plasma ignition

Contributor Electronically uploaded by the author (provenance)

School Andrew and Erna Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Electrical Engineering

Publication Date 08/05/2010

Defense Date 06/28/2010

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag combustion efficiency,ignition,nanosecond discharge,non-equilibrium plasma,OAI-PMH Harvest,pulsed power,transient plasma,transient plasma ignition

Language English

Advisor Gundersen, Martin A. (committee chair), Steier, William H. (committee member), Wang, Hai (committee member)

Creator Email daniel.r.singleton@gmail.com,dsinglet@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m3289

Unique identifier UC1313015

Identifier etd-Singleton-3848 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-369247 (legacy record id),usctheses-m3289 (legacy record id)

Legacy Identifier etd-Singleton-3848.pdf

Dmrecord 369247

Document Type Dissertation

Rights Singleton, Daniel R.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu