Winters Final

Laser Diagnostics of Reacting Molecular Plasmas for Plasma Assisted
Combustion Applications
DISSERTATION
Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy
in the Graduate School of The Ohio State University
By
Caroline Winters, M.S.
Graduate Program in Aeronautical and Astronautical Engineering
The Ohio State University
2017
Dissertation Committee:
Dr. Igor Adamovich, Advisor
Dr. Jeffrey Sutton
Dr. Seung Hyun Kim

Copyright by
Caroline April Winters
2017
Abstract
This work has produced extensive sets of new data on low-temperature plasma-
assisted fuel oxidation in hydrogen-oxygen-argon and hydrocarbon-oxygen-argon
mixtures. The measurements have been made in two different plasma flow reactors, at an
initial temperature of 500 K and pressures ranging from 300 Torr to 700 Torr. In both
reactors, the plasma is generated by a high peak voltage, ns pulse discharge, operated at
high pulse repetition rates (up to 20 kHz). Metastable Ar atom number density distributions
in the discharge afterglow are measured by Tunable Diode Laser Absorption Spectroscopy
(TDLAS), and used to characterize plasma uniformity. Temperature in the discharge-
excited reacting flow is measured by Rayleigh scattering. Two-photon Absorption Laser
Induced Fluorescence (TALIF) is used to measured absolute H and O atom number
densities. The results are compared with predictions of a kinetic model analyzing reaction
kinetics of excited species and radicals generated by the plasma at low temperatures and
high pressures. The modeling predictions show good agreement with the data, with the
exception of fuel-limited mixtures, when nearly all fuel available in the mixture of reactants
is oxidized in the discharge. Kinetic modeling analysis identified dominant processes of
generation and decay of atomic and radical species in the discharge and in the afterglow.
At the present low-temperature conditions, the effect of chain branching reactions on
plasma-assisted fuel oxidation kinetics is insignificant.

ii
In addition, this work presents results of time-resolved, absolute measurements of
OH number density by Laser Induced Fluorescence (LIF), as well as measurements of
translations-rotational temperature and nitrogen vibrational temperature by ns broadband
Coherent Anti-Stokes Raman Scattering (CARS), in air and lean hydrogen-air mixtures
excited in a diffuse filament ns pulse discharge at high specific energy loading. The main
objective of these measurements is to study kinetics of OH radicals at the conditions of
strong vibrational excitation of nitrogen, below autoignition temperature. The results show
that OH number density and N2 vibrational temperature exhibit transient maxima after the
discharge pulse. Comparison of the experimental data with kinetic modeling predictions
shows that OH kinetics is controlled primarily by reactions of H 2 and O2 with O and H
atom generated during the discharge. At the present conditions, OH number density is not
affected by N2 vibrational excitation directly, e.g. vibrational energy transfer to HO2.
However, as the energy is increased, the model also predicts transient OH number density
overshoot, due to the temperature rise caused by N 2 vibrational relaxation by O atoms. This
may well be the dominant effect on OH number density in discharges with high specific
energy loading.
Finally, the present work provides insight into surface charge dynamics and kinetics
of radical species reactions in nanosecond pulse discharges sustained at a liquid-vapor
interface, above a distilled water surface. LIF and TALIF line imaging are used for in situ
measurements of spatial distributions of absolute OH and H atom number densities in near-
surface, repetitive nanosecond pulse discharge plasmas.
iii
Dedication
To the following people, I dedicate this work:
My father, James Winters, whose hard work and dedication to his family afforded me the
education necessary to obtain this degree, and whose unwavering support and love was
paramount to its realization.
My mother, Carol Winters, whose intelligence, compassion, and friendship has guided
me in becoming the person I am today, and whose achievements have shown me how
much is possible in life.
iv
Acknowledgments
My road to this moment would not be possible without the guidance of the following
people, and for that I am ever thankful:
To my advisor, Dr. Igor Adamovich, whose extraordinary leadership over the last
few years in this laboratory cannot be understated. Thank you. We, your students,
understand and are grateful for all that you have done for us.
To all my teachers throughout the years: you expanded my knowledge, challenged
my reasoning, and offered the very best every day. I will forever be grateful for your
wisdom, encouragement, and challenges. I thank all those from Randolph School and Rose-
Hulman Institute of Technology. I would especially like to thank Wayne Mullins, who
commuted between cities and offered his weekends to teach my classmates and me; you
remain the best physics teacher I have ever had.
To Elliot Schmidt, whose love and friendship I clung to at my worst moments, and
whose humor and intelligence made my best moments, I say thank you. Thank you for the
all the little things that made these five years so fulfilling. No matter where you find
yourself, if the days are too long, the nights too short, and the progress too slow, Don’t
Panic.
To the fellow graduate students who have been along for the ride: Good Luck, and
Godspeed.
v
Vita
November 7th, 1989........................................Born- Los Angeles, CA USA
2011................................................................B.S. Mechanical Engineering, Rose-Hulman
Institute of Technology
2016................................................................M.S. Aerospace Engineering, The Ohio
State University
2015 to present ..............................................Graduate Research Intern, Diagnostic
Science and Engineering, Sandia National
Laboratory, Albuquerque, NM
2012 to present ..............................................Graduate Research Assistant, Department of
Mechanical and Aerospace Engineering,
The Ohio State University
vi
Publications
1. Winters, C., Petrischev, V., Yin, Z., Lempert, W., and Adamovich, I.V., “Surface
charge dynamics and OH and H number density distributions in near-surface
nanosecond pulse discharges at a liquid / vapor interface”, Journal of Physics D:
Applied Physics, vol. 48, 2015, p. 424002
Fields of Study
Major Field: Aeronautical and Astronautical Engineering
vii
Table of Contents
Abstract ............................................................................................................................... ii
Dedication .......................................................................................................................... iv
Acknowledgments............................................................................................................... v
Vita ..................................................................................................................................... vi
Publications ....................................................................................................................... vii
Fields of Study .................................................................................................................. vii
Table of Contents ............................................................................................................. viii
List of Tables ..................................................................................................................... xi
Chapter 1: Introduction ....................................................................................................... 1
1.1. Nonequilibrium plasmas for plasma assisted combustion applications ................... 1
1.2. Review of previous studies of nonequilibrium plasmas in ns pulse discharges..... 12
1.3. Effect of vibrational excitation on plasma chemical reaction kinetics................... 26
1.4. Liquid/Vapor interfaces and plasma medicine ....................................................... 31
1.5. Structure of dissertation ......................................................................................... 34
Chapter 2: Background on Laser Diagnostics .................................................................. 35
viii
2.1. OH Laser Induced Fluorescence (LIF)................................................................... 35
2.2. Two-Photon Absorption Laser Induced Fluorescence ........................................... 45
2.3. Tunable Diode Laser Absorption Spectroscopy..................................................... 56
Chapter 3: Measurements of metastable and radical species in repetitive ns pulse
discharge plasmas in fuel-oxidizer mixtures..................................................................... 62
3.1. Plasma flow reactor design .................................................................................... 63
3.2. Laser Diagnostics Setup ......................................................................................... 69
3.3. Discharge Waveforms and Plasma Images ............................................................ 74
3.4. Plasma assisted combustion kinetic model ............................................................ 85
3.5. Measurements of Ar metastable atoms .................................................................. 86
3.6. TALIF results and discussion................................................................................. 96
Chapter 4: Measurements of OH number density at the conditions of strong vibrational
nonequilibrium ................................................................................................................ 133
4.1. Discharge Test Cell and Electrode Assembly ...................................................... 134
4.2. Pulsed discharge waveforms and plasma images ................................................. 137
4.3. Laser Induced Fluorescence Experimental Setup ................................................ 144
4.4. Coherent Anti-Stokes Raman Spectroscopy Setup .............................................. 152
4.5. CARS Results and Discussion ............................................................................. 157
4.6. LIF Results and Discussion .................................................................................. 167
ix
Chapter 5: Radical species measurements at a liquid/vapor interface ............................ 189
5.1. Electric discharge cell design ............................................................................... 190
5.2. Discharge waveforms and plasma images ........................................................... 194
5.3. OH and H Spatial Distributions ........................................................................... 197
5.4. Plasma assisted combustion kinetic modeling predictions .................................. 206
Chapter 6: Summary and Conclusions............................................................................ 212
Bibliography ................................................................................................................... 220
x
List of Tables
Table 2.1: Two-photon absorption cross section ratios, natural lifetimes, and collisional
quenching rates used in H, Kr, O, and Xe TALIF measurements. ................................... 49
Table 2.2: Bandpass filter transmission factors used in H, Kr, O, and Xe TALIF
measurements. ................................................................................................................... 54
Table 4.1: Quenching rate variation over time after a discharge pulse. ......................... 148
Table 4.2: Quenching rates used in LIF data processing ................................................ 149
xi
List of Figures
Figure 1.1: Energy partition in a discharge in air vs. the reduced electric field, E/N [3]. .. 2
Figure 1.2: Electron temperature, Te, versus reduced electric field, E/N, in (a) nitrogen [3]
and (b) dry air [4] plasmas. ................................................................................................. 3
Figure 1.3: Breakdown voltages in different gases over a wide range of Pd values (Paschen
curves) [3]. .......................................................................................................................... 7
Figure 1.4: Schematic diagram illustrating positive feedback development of the
ionization-heating instability in an electric discharge, resulting in a discharge constriction
and glow-to-arc transition in high pressure plasmas []. .................................................... 10
Figure 1.5: Two-dimensional number density distributions of (a) H atoms in a 1% H 2-Ar
mixture at P= 100 Torr and (b) O atoms in a 1% O 2-He mixture at P= 100 Torr, illustrating
non-uniform distributions of atoms at t= 25 µs after the end of the discharge burst [11]. 14
Figure 1.6: Radial distributions of H and OH concentrations during and after the discharge
burst between two spherical copper electrodes, measured in a mixture of Ar:O 2:H2 =
80:20:2, at P= 40 Torr, T0= 300 K, and a discharge pulse repetition rate of 100 kHz [13].
........................................................................................................................................... 17
Figure 1.7: Time-resolved gas temperature (experimental and predicted) during and after
the main discharge pulse in air and in a H2-air mixture at an equivalence ratio of 0.42 [7].
........................................................................................................................................... 19
xii
Figure 1.8: Comparison of experimental and predicted time-resolved, absolute OH number
densities after ν= 10 kHz, 50-pulse discharge burst in (a) H2-air mixture, (b) C2H4-air
mixtures, and (c) C3H8-air mixtures at P= 100 Torr and T0= 500 K [6]. .......................... 21
Figure 1.9: Predicted OH number density in the atmospheric pressure 3% H 2-air mixture
excited by a ns pulse discharge at different initial temperatures and initial N 2 vibrational
temperature of Tv = 3000 K: (a) “conventional” combustion chemistry only, and (b) with
the inclusion of N2-HO2 V-V transfer of Eqn. 1.18 [2]..................................................... 29
Figure 2.1: Two level energy diagram for single photon laser excitation, showing dominant
energy transfer processes between the ground state, N0, and the excited state, N1. ......... 36
Figure 2.2: Schematic representation of the convolution integral, determined by tuning the
laser lineshape relative to the absorption transition lineshape, and measuring the
convolution signal. ............................................................................................................ 40
Figure 2.3: Energy level diagram for OH. The absorption transition used in the present
work, X2Π  A2Σ R2(2.5), is highlighted in red [65]. ..................................................... 42
Figure 2.4: Three level energy diagram for two-photon laser excitation, showing dominant
energy transfer processes. ................................................................................................. 46
Figure 2.5: Two-photon absorption energy level diagrams for (a) hydrogen and krypton
and (b) oxygen and xenon [8]. .......................................................................................... 51
Figure 2.6: Hamamatsu R3896 photomultiplier tube response curve measured using a
calibrated blackbody source at T=900 K. A spectrometer placed before PMT is used as a
tunable filter. The PMT is used for H, O, Kr, and Xe TALIF measurements. The response
xiii
curve is used to compare TALIF signal in calibration gases (Xe, Kr) with TALIF signal in
O2-Ar and H2-Ar plasmas. ................................................................................................ 52
Figure 2.7: ThorLabs FS650-40 filter transmission curve measured using a broadband light
source and spectrometer. Spectra are measured without (red line) and with (blue line) the
filter. The intensity ratio between the two lines gives the transmission curve, also plotted
in the figure. The same procedure is used for other filters used in TALIF measurements
(see Table 2.2). .................................................................................................................. 53
Figure 2.8: Energy level diagram for atomic oxygen absorption transitions, with relative
transition probabilities indicated [70]. The transitions used in the present work are labeled.
........................................................................................................................................... 55
Figure 2.9: Energy level diagram for Ar atoms showing Ar(3p54s) and Ar(3p54p) states,
using Paschen notations. The transition used in the present work is 1s 52p7, indicated by
a red arrow [74]................................................................................................................. 58
Figure 2.10: Theoretical Voigt absorption line profiles for the Ar(3p 54s) 1s5 Ar(3p54p)
2p7 transition, illustrating the combined effect of Doppler broadening and pressure
broadening at 300 K vs. 500 K at P = 300 Torr and pressure broadening alone at 300 Torr
vs. 700 Torr at T = 500 K. ................................................................................................ 61
Figure 3.1: Schematic (a) and a photograph (b) of a plasma flow reactor / discharge cell
with liquid metal electrodes encapsulated in quartz reservoirs (Cell #1) to generate a
volumetric, double dielectric barrier electric discharge. ................................................... 64
xiv
with copper electrodes encapsulated in a resilient dielectric material (PDMS), to generate
a volumetric, double dielectric barrier electric discharge ................................................. 66
Figure 3.3: Schematic of TDLAS diagnostics. A continuous wave (cw), tunable
narrowband diode laser is mounted on a translation stage. A photodiode is mounted on
another translation stage, after the cell. Micrometric screws are used to make translation
stage adjustments. ............................................................................................................. 69
Figure 3.4: Schematic of H TALIF diagnostics. A nanosecond Nd:YAG laser pumps a
tunable dye laser to generate output at 615 nm , frequency doubled using a Type I BBO
crystal, and mixed with the dye laser output to generate a 205 nm beam. ....................... 70
Figure 3.5: Schematic of O TALIF diagnostics. The 619 nm beam generated by the dye
laser pumped by the second harmonic of the Nd:YAG laser, is frequency mixed with the
third harmonic output of the Nd:YAG laser (355 nm beam) using a Type I BBO crystal.
The three beams are passed through an iris and a UV harmonic separator (denoted as PR
in the diagram) to allow only the 226 nm beam to pass. .................................................. 71
Figure 3.6: Fluorescence signal versus squared laser pulse energy for (a) H TALIF
measured in a 1% H2-Ar mixture excited by a ns pulse discharge burst of 50 pulses with a
pulse repetition rate of 10 kHz, and (b) O TALIF, measured in a 1% O2-Ar mixture excited
by a ns pulse discharge burst of 75 pulses with a pulse repetition rate of 20 kHz at P = 300
Torr and initial temperature of T = 500 K. ....................................................................... 73
Figure 3.7: Voltage and instantaneous power waveforms (a), voltage and coupled energy
(b) measured in Cell #1. 1% H2-Ar mixture excited by a ns pulse discharge burst, pulse
xv
repetition rate 20 kHz, pulse #25 at P = 300 Torr and T 0 = 500 K. Oscillatory behavior is
caused by multiple pulse reflections from the load and the high-voltage pulse generator.
........................................................................................................................................... 75
Figure 3.8: Photograph of a ns pulse discharge in argon (a). Single shot ICCD camera
images of broadband plasma emission during a ns discharge pulse in Cell #1, side view (b)
and end view (c). Ar at T0= 500 K and P= 700 Torr, discharge pulse repetition rate 10 kHz,
pulse #50, camera gate 1 µs. ............................................................................................. 76
Figure 3.9: Single shot ICCD camera images of broadband plasma emission during a ns
discharge pulse in Cell #1, side view (a) and end view (b). 1% H2-Ar at T0= 500 K and P=
300 Torr, discharge pulse repetition rate 10 kHz, pulse #50, camera gate 1 µs. .............. 77
discharge pulse in Cell #1, side view (a) and end view (b). 1% O2-Ar at T0= 500 K and P=
Figure 3.11: Voltage and instantaneous power waveforms (a), voltage and coupled pulse
energy waveforms (b) measured in Cell #2. 1% O2-Ar excited by a ns pulse discharge burst,
pulse repetition rate 20 kHz, pulse #75. Oscillatory behavior is caused by multiple pulse
reflections from the load and high-voltage pulse generator.............................................. 79
Figure 3.12: ICCD camera images of broadband plasma emission during a ns discharge
pulse in Cell #2, single shot side view (a) and end view (b), and 100-shot accumulation,
side view (c) and end view (d). 1% O2-Ar mixture at T0= 500 K and P= 300 Torr, discharge
pulse repetition rate 20 kHz, pulse #75, camera gate 1 µs................................................ 80
xvi
discharge pulse in Cell #2, side view (a) and end view (b) 0.13% H 2-1% O2-Ar mixture at
T0= 500 K and P= 300 Torr, discharge pulse repetition rate 20 kHz, pulse #75, camera gate
1 µs. ................................................................................................................................... 80
discharge pulse in Cell #2, side view (a) and end view (b) 0.25% CH 4-1% O2-Ar mixture
at T0= 500 K and P= 300 Torr, discharge pulse repetition rate 20 kHz, pulse #75, camera
gate 1 µs. ........................................................................................................................... 81
Figure 3.15: Voltage and instantaneous power waveforms measured in Cell #2. Mixture of
75 ppm C2H4 in 1% O2-Ar excited by a ns pulse discharge burst, pulse repetition rate 20
kHz, pulse #75. Oscillatory behavior is caused by multiple reflections from the load and
high-voltage pulse generator. ............................................................................................ 82
waveforms (b) measured in Cell #2. Mixture of 150 ppm C3H8 in 1% O2-Ar excited by a
ns pulse discharge burst, pulse repetition rate 20 kHz, pulse #75. Oscillatory behavior is
caused by multiple pulse reflections from the load and high-voltage pulse generator. .... 82
side view (c) and end view (d) mixture of 150 ppm C2H4 in 1% O2-Ar; discharge pulse
repetition rate 20 kHz, pulse #75, camera gate 1 µs. ........................................................ 84
xvii
side view (c) and end view (d) mixture of 150 ppm C3H8 in 1% O2 -Ar. T0= 500 K and P=
Figure 3.19: Experimental (symbols) and synthetic (line) absorption spectra (Ar 1s 5→2p7
and 1s3→2p2 transitions) in argon excited by a ns pulse discharge burst, used for Ar(1s 5)
and Ar(1s3) number density inference. P=300 Torr, T0=500 K, ν=10 kHz, 50-pulse burst,
time delay after the last pulse 2 μs. In the synthetic spectrum, absorption line shape is
approximated by a Voigt profile. ...................................................................................... 87
Figure 3.20: Time resolved, normalized Ar* TDLAS signal (1s5→2p7 transition) in argon,
1% H2-Ar, and 1% O2-Ar mixtures excited by a ns pulse discharge burst, taken after the
last pulse in the burst. P=700 Torr (argon), P=300 Torr (1% H 2-Ar and 1% O2-Ar
mixtures), T0=500 K, ν=10 kHz, 50-pulse burst, Noise near t=0 is caused by EMI from the
pulser. Note almost complete absorption in argon and 1% H2-Ar mixture. ..................... 88
Figure 3.21: Comparison of time resolved [Ar(3p54s)] decay, in the linear absorption
regime (for I/I0=0.1-0.9), in 1% O2-Ar and 1% H2-Ar mixtures at P= 300 Torr, T0= 500 K,
and pure Ar at P= 700 Torr, T0= 500 K, excited by 50 pulses at a pulse repetition rate of
10 kHz, with kinetic modeling predictions. ...................................................................... 90
Figure 3.22: Schematic of twenty-one TDLAS laser beam locations across the channel
cross section (three different heights and seven spanwise positions)), used for [Ar(3p 54s)]
distribution measurements after the discharge burst. Laser beam locations are overlapped
with the Ar plasma emission image, also shown in Fig. 3.10. [Ar(3p 54s)]is used as a
quantitative measure of plasma uniformity....................................................................... 92
xviii
Figure 3.23: [Ar(3p54s)] distribution across the plasma, measured 1.5 µs after the last pulse
(pulse #50) in the discharge burst. Argon, P=500 Torr, T=500 K, ν=10 kHz, 50-pulse burst.
Data point symbols are the same as in Fig. 3.22. Significant deviation from uniformity, up
to ~ 50%, is detected near the left wall of the channel (at x=2 mm). ............................... 93
Figure 3.24: [Ar(3p54s] distribution across the plasma, measured 0.23 µs after the last pulse
(pulse #50) in the discharge burst. 1% O2-Ar mixture, P=300 Torr, T0=500 K, ν=10 kHz,
50-pulse burst. Nonuniformity across the plasma does not exceed 8%. Note that O 2-Ar
plasma is considerably more uniform compared to Ar plasma (see Fig. 3.23). ............... 94
Figure 3.25: Time-resolved Ar(1s5) number density measured in Cell #1: Argon at P=300
Torr, T0=500 K, discharge pulse repetition rate ν= 98 kHz. Data are taken after the last
pulse in a 25-pulse burst. Experimental data are extrapolated (dashed line) to show peak
[Ar(3p54s)] produced after each pulse. ............................................................................. 95
Figure 3.26: Convolution integral (between the laser lineshape and the absorption
lineshape) for (a) Kr 4p6 1S0  5p’[3/2]2 two-photon absorption transition at 204.13 nm. P
= 47.5 Torr T = 300 K, average laser pulse energy Eavg = 35 µJ. Convolution integral for
(b) H 1s2 S1/2  3d 2D3/2,5/2 two-photon absorption transition at 205.08 nm, P = 300 Torr,
T0 = 500 K, 1% H2-Ar mixture excited by a 50-pulse discharge burst at ν=10 kHz, 100 µs
after the burst, Eavg = 30 µJ. .............................................................................................. 97
lineshape) for (a) Xe 5p6 1S0  6p’[3/2]2 two-photon absorption transition at 225.24 nm.
P = 47.5 Torr T = 300 K, average laser pulse energy Eavg = 35 µJ.Convolution integral for
(b) O 2p4 3P2,1,0  3p 3P1,2,0 two-photon absorption transition at 225.59 nm, P = 300 Torr,
xix
T0 = 500 K, 1% O2-Ar mixture excited by a 50-pulse discharge burst at ν= 20 kHz, 100 µs
after the pulse, Eavg=30 µJ. ............................................................................................... 98
Figure 3.28: Comparison between TALIF signal measured by the PMT and background
noise due to laser scattering: (a) H TALIF, 1% H2- Ar mixture (50-pulse burst, ν=10 kHz),
and (b) O TALIF 1% O2- Ar mixture (75-pulse burst, ν=20 kHz). Both signals are taken
100 µs after the end of the discharge burst and accumulated over 100 laser shots. T 0= 500
K, P= 300 Torr, and PMT. ................................................................................................ 99
Figure 3.29: H TALIF line images (300 laser shot accumulations) taken with an ICCD
camera in 1% H2- Ar mixture in Cell #1 excited by a ns pulse discharge burst, at different
delay times after the burst. P=300 Torr, T0=500 K, 50-pulse burst, ν=10 kHz. Flow
direction from left to right. Both H atom recombination and convection downstream with
flow are evident............................................................................................................... 101
Figure 3.30: Time resolved H atom number density in a 1% H2-Ar mixture excited by 50
discharge pulses at a pulse repetition rate of 10 kHz and in a 1% H 2-0.15% O2-Ar mixture
excited by 25 discharge pulses at a pulse repetition rate of 20 kHz. Coupled discharge pulse
energy is 2.6 mJ/pulse. .................................................................................................... 102
Figure 3.31: H atom number density measured 20 µs after the end of the discharge burst
for different number of pulses in the burst. Cell #1, 1% H2-Ar, P= 300 Torr, T0= 500 K
ν=10 kHz, 2.6 mJ/pulse................................................................................................... 103
Figure 3.32: Predicted radical (H, O, OH, and HO2) and stable (H2O) species number
densities in a 1% H2- 0.15% O2- Ar mixture excited by a ns pulse discharge of 50 pulses at
xx
pulse repetition rate of 10 kHz, in the burst (a) and in the afterglow (b), at P= 300 Torr and
T0= 500 K. Coupled discharge pulse energy is 2.6 mJ/pulse. ........................................ 106
Figure 3.33: Time resolved O atom number density in a 1% O2-Ar mixture, a 0.13% H2-
1% O2-Ar mixture, and a 0.25% CH4-1% O2-Ar mixture excited by 75 discharge pulses at
a pulse repetition rate of 20 kHz. Coupled discharge pulse energy is 4.2 mJ/pulse. ...... 108
Figure 3.34: Predicted O atom and O2 species number densities in a 1% O2-Ar mixture
excited by a ns pulse discharge of 75 pulses at pulse repetition rate of 20 kHz, in the burst,
at P= 300 Torr and T0= 500 K. Coupled discharge pulse energy is 4.2 mJ/pulse. ......... 109
densities in a 0.13% H2- 1% O2- Ar mixture excited by a ns pulse discharge of 75 pulses at
T0= 500 K. Coupled discharge pulse energy is 4.2 mJ/pulse. ........................................ 112
Figure 3.36: Predicted radical (O, H, OH, and HO2) and stable (CH4 and H2O) species
number densities in a 0.25% CH4- 1% O2- Ar mixture excited by a ns pulse discharge of
75 pulses at pulse repetition rate of 20 kHz, in the burst (a) and in the afterglow (b), at P=
300 Torr and T0= 500 K. Coupled discharge pulse energy is 4.2 mJ/pulse.................... 115
Figure 3.37: Time resolved, absolute O atom number density measured in C 2H4 – O2- Ar
mixtures excited by a ns pulse discharge in Cell #2. P= 300 Torr, T 0= 500 K, ν=20 kHz, 75
pulses in a burst, at two coupled pulse energies 2.3 mJ/pulse (a) and 1.4 mJ/pulse (b). 116
Figure 3.38: Predicted radical (O, H, OH, and HO2) and stable (C2H4 and H2O) species
number densities in a mixture of 150 ppm of C2H4 in a 1% O2- Ar excited by a ns pulse
xxi
discharge of 75 pulses at pulse repetition rate of 20 kHz, in the burst (a) and in the afterglow
(b), at P= 300 Torr and T0= 500 K. Coupled discharge pulse energy is 1.4 mJ/pulse.... 118
Figure 3.39: Absolute O atom number density versus initial mole fraction of ethylene in
the mixture excited by a ns pulse discharge. P= 300 Torr, T 0= 500 K, ν=20 kHz, 75 pulses
in a burst, with discharge coupled energy of 1.4 mJ/pulse. Data and modeling predictions
are compared 70 µs after the discharge burst. ................................................................ 120
pulses in a burst, with discharge coupled pulse energy of 1.4 mJ/pulse......................... 121
number densities in a mixture of 75 ppm C3H8 in 1% O2- Ar, in the burst (a) and in the
afterglow (b), and in a mixture of 150 ppm C3H8 in 1% O2- Ar, in the burst (c) and in the
afterglow (d), excited by a ns pulse discharge of 75 pulses at pulse repetition rate of 20
kHz, at P= 300 Torr and T0= 500 K. Coupled discharge pulse energy is 1.4 mJ/pulse.. 123
Figure 3.42: Absolute O atom number density versus initial mole fraction of propane in the
mixture excited by a ns pulse discharge. P= 300 Torr, T0= 500 K, ν=20 kHz, 75 pulses in
a burst, with discharge coupled energy of 1.2 mJ/pulse. Data and modeling predictions are
compared 90 µs after the discharge burst. ...................................................................... 126
number densities in a mixture of 1% C3H8 -1% O2- Ar, in the burst (a) and in the afterglow
(b), and predicted radical (O, H, OH, and HO2) and stable (C3H8 and H2O) species number
densities in a mixture in a mixture of 1% C 3H8 - 1% O2- Ar, in the burst (c) and in the
xxii
kHz, at P= 300 Torr and T0= 500 K. Coupled discharge pulse energy is 1.4 mJ/pulse.. 131
Figure 4.1: (a) Schematic diagram of the experimental apparatus including a glass six-arm
cross test cell with BK-7 glass windows providing optical access to the laser beams and
fused silica windows for fluorescence signal collection at 90° and for plasma imaging.
......................................................................................................................................... 135
Figure 4.2: Left: photograph of the ceramic bracket holding the discharge electrodes, taken
through the 2” diameter optical access window in one of the arms of the cell, with an ICCD
camera image showing a reference image of the electrodes. Right: end view photograph
of the electrode assembly, showing a 1 mm diameter aperture drilled through the electrode
center, taken through the 3” diameter window providing optical access to the laser beams.
......................................................................................................................................... 136
Figure 4.3: Voltage, current, and coupled energy waveforms for Pulser #2. 3% H 2-air
mixture, T0= 300 K, P = 100 Torr, pulser input DC voltage (a) Vin=490 V, positive polarity,
(b) Vin=490 V, negative polarity, (c) Vin=530 V, positive polarity, and (d) Vin=530 V,
negative polarity; pulse repetition rate 10 Hz. ................................................................ 138
Figure 4.4: Single pulse (a) and 100-pulse average (b) ICCD images of broadband plasma
emission in air and H2-air mixtures for Pulser #2. T0= 300 K, P = 100 Torr, Vin=490 V, 10
Hz pulse repetition rate. Camera gate 1 µs. Reference image of the electrodes (c) with
grounded electrode is on the left, high voltage electrode is the right. ............................ 139
xxiii
Hz repetition rate. Camera gate 1 µs. Grounded electrode is on the left, high voltage
electrode is the right. ....................................................................................................... 140
Figure 4.6: Single pulse (a) and 100-pulse average (b) ICCD images ns of broadband
plasma emission in air for Pulser #2. T0= 300 K, P = 100 Torr, Vin=490 V, 60 Hz pulse
repetition rate. Camera gate 100 ns. Sets of images are taken for both positive and negative
polarity pulses. Grounded electrode is on the left, high voltage electrode is the right. .. 141
Figure 4.7: Voltage, current, and coupled energy waveforms for Pulser #1. Taken in a 3%
H2-Air mixture, at T0= 300 K, input DC voltage Vin=420 V, with a discharge pulse
repetition rate of 10 Hz. (a) P= 60 Torr and (b) P= 100 Torr. ........................................ 142
emission in air and H2-air mixtures for Pulser #1 T0= 300 K, P = 60 Torr, Vin=420 V, 10
Hz pulse repetition rate. Grounded electrode is on the left, high voltage electrode is the
right. ................................................................................................................................ 143
right. ................................................................................................................................ 143
Figure 4.10: Schematic diagram of the OH LIF experimental apparatus. The 615 nm dye
laser output is frequency doubled using a Type I BBO crystal. The 308 nm beam is passed
through a half waveplate and polarizer combination to control the laser pulse energy. Both
beams are sent through a set of Pellin-Broca prisms to separate the LIF beam from the dye
laser beam before it is focused in the test section using a f= 500 mm UV lens.............. 145
xxiv
Figure 4.11: OH fluorescence signal versus laser pulse energy in 3% H 2-air mixture excited
by a ns pulse discharge. P=100 Torr, T0=300 K, Pulser #1, coupled discharge pulse energy
6.3 mJ/pulse, and delay time between the discharge pulse and the laser pulse is 100 µs.
Camera gate 200 ns, camera gain 100, 20 laser shot accumulation. All subsequent data are
taken at laser pulse energy of 5 µJ. ................................................................................. 146
Figure 4.12: OH LIF signal vs. time delay after the laser pulse, used for the fluorescence
quenching rate measurements. 3% H2-air mixture, P=100 Torr, T0=300 K, Pulser #2,
Ecoupled=7.5 mJ/pulse. Camera gate 5 ns, every data point is averaged over 300 laser shots.
Equation used for best fit is also shown.......................................................................... 148
Figure 4.13: Experimental data points used for calculation of the convolution integral for
the OH LIF absorption transition. 3% H2-air mixture, a) P=60 Torr, T0=300 K, Pulser #1,
Ec=11 mJ/pulse b) P = 100 Torr, T0= 300 K, Pulser #2, Ec= 7.5 mJ. The data are
approximated by a Gaussian profile, to obtain the convolution integral, with FWHM = 0.5
cm-1 and 0.47 cm-1, respectively. .................................................................................... 150
Figure 4.14: Rayleigh scattering images taken in nitrogen at a) 685 Torr and b) 334 Torr,
at T=300 K. 100 laser shot accumulation, 200 ns camera gate. Rayleigh signal collected
from the central 4 mm in the axial direction and integrated along a transverse region with
a height of ~1 mm, where 95% of the Rayleigh signal is generated. ............................. 151
Figure 4.15: Rayleigh scattering calibration obtained from ICCD camera images such as
those shown in Fig. 4.15, at P= 334-685 Torr in nitrogen. Signal uncertainty is 30%,
controlled by laser scattering noise off the windows. ..................................................... 152
xxv
Figure 4.16: CARS energy level diagram showing excitation of the molecule to a virtual
state and anti-Stokes signal beam generation. ................................................................ 154
Figure 4.17: Experimental diagram of broadband collinear CARS diagnostics. The 532 nm
and 607 nm beams are focused halfway between the discharge electrodes. A set of dichroic
mirrors (R473, T532, T607) after the test cell separates the pump/probe and Stokes beams
from the CARS signal beam. The remaining pump beam light is filtered out by the band
pass filter in front of the spectrometer slit. ..................................................................... 155
Figure 4.18: Axial distribution of the non-resonant background signal from a microscope
slide, illustrating spatial resolution of the collinear CARS measurements (95% of the signal
originates from a region 3.7 mm long). .......................................................................... 156
Figure 4.19: Broadband CARS spectra in a 3% H 2-air mixture at 100 Torr, excited by a ns
pulse discharge sustained by Pulser #2, taken with Nd:YAG laser injection seeding off, for
a) coupled pulse energy of 8.5 mJ, and b) coupled pulse energy of 11 mJ. Time delay after
the discharge pulse is 200 μs. N2 (v=0-2) vibrational bands can be identified in the spectra.
......................................................................................................................................... 158
Figure 4.20: Comparison of experimental and best fit synthetic (CARSFIT) spectra of
N2(v=0) band in room air. Experimental spectra are taken with Nd:YAG laser injection
seeding turned on. Best fit rotational temperature inferred using CARSFIT is 292 ± 12 K.
......................................................................................................................................... 159
N2(v=0) band in air at 100 Torr, excited by a ns pulse discharge sustained by Pulser #2,
with a) coupled pulse energy 8.5 mJ at a time delay after the discharge burst of 400 µs and
xxvi
b) 11 mJ with a time delay of 600 µs. Experimental spectra are taken with Nd:YAG laser
injection seeding turned on. Best fit rotational temperatures inferred using CARSFIT are
528 ± 45 K and 557 ± 50 K, respectively. ...................................................................... 160
Figure 4.22: N2 vibrational temperature and translational-rotational temperature plotted vs.
time delay after the discharge pulse in air and hydrogen-air mixtures excited by a ns pulse
discharge generated by Pulser #2 at 100 Torr, inferred from the CARS spectra such as
shown in Fig. 4.21. Coupled discharge pulse energy is 8.5 mJ (a) and 11 mJ (b). ........ 162
discharge generated by Pulser #1 at 100 Torr, inferred from the CARS spectra. Coupled
discharge pulse energy is 6.3 mJ/pulse. .......................................................................... 164
Figure 4.24: OH LIF signal distribution along the plasma filament at different delay times
after the end of the discharge pulse. ICCD images, 100 laser shot accumulation, 200 ns
camera gate, camera gain 100x. P=100 Torr, T0=300 K, Pulser #2, coupled discharge pulse
energy E= 7.5 mJ/pulse (a) 1% H2-air, (b) 3% H2-air, (c) 5% H2-air. ............................ 167
energy E= 10 mJ/pulse (a) 1% H2-air, (b) 3% H2-air, (c) 5% H2-air. ............................. 168
Figure 4.26: OH LIF signal distribution along the plasma filament in the discharge
afterglow, integrated over a transverse region with a height of 1 mm. 100-shot
accumulation images, with a 200 ns camera gate, taken in a 3% H 2-air mixture with coupled
xxvii
discharge pulse energy 7.5 mJ/pulse (a) and a 5% H2-air mixture with coupled discharge
pulse energy of 10 mJ/pulse (b), excited by a ns pulse discharge sustained by Pulser #2, at
P= 100 Torr and T0= 300 K. ........................................................................................... 169
energy E= 11.6 mJ/pulse (a) 1% H2-air, (b) 3% H2-air, (c) 5% H2-air. .......................... 170
energy E= 6.3 mJ/pulse . (a) 1% H2-air, (b) 3% H2-air, (c) 5% H2-air. ......................... 170
accumulation images, with a 200 ns camera gate, taken in a 3% H 2-air mixture at (a) P= 60
Torr and T0=300 K, with coupled discharge pulse energy of 11.6 mJ/pulse and (b) P= 100
Torr, T0= 300 K, with coupled discharge pulse energy of 6.3 mJ/pulse, excited by a ns
pulse discharge sustained by Pulser #1. .......................................................................... 171
Figure 4.30: Time resolved absolute OH number density integrated over the central 4 mm
region of the discharge filament in the plasma afterglow in H2-air mixtures excited by a ns
pulse discharge (Pulser #2): (a) P= 100 T, T0= 300 K, coupled discharge pulse energy is
7.5 mJ/pulse, and (b) P = 100 Torr, T0 = 300 K, coupled discharge pulse energy is 10
mJ/pulse. ......................................................................................................................... 173
xxviii
pulse discharge (Pulser #1) a) P= 60 T, T0= 300 K, coupled discharge pulse energy is 11.6
mJ/pulse and b) P = 100 Torr, T0 = 300 K, coupled discharge pulse energy is 6.3 mJ/pulse.
......................................................................................................................................... 175
Figure 4.32: Radial distribution of UV/visible plasma emission intensity in hydrogen- air
mixtures at P = 100 Torr, T0 = 300 K. Pulser #2 coupled discharge pulse energies are (a)
7.5 mJ and (b) 10 mJ/pulse, respectively. ....................................................................... 177
Figure 4.33: Comparison of experimental and predicted N 2 vibrational temperature (a) and
gas temperature (b) vs. time delay after the discharge pulse generated by Pulser #2 in air
and H2-air mixtures. Coupled discharge pulse energy is 7.5 mJ/pulse. .......................... 178
Figure 4.34: Comparison of experimental and predicted OH number density vs. time delay
after the discharge pulse generated by Pulser #2 in H2-air mixtures. Coupled discharge
pulse energy is 7.5 mJ/pulse. .......................................................................................... 180
Figure 4.35: Predicted radical species number densities vs. time delay after the discharge
pulse generated by Pulser #2 in a 3% H2-air mixture. Coupled discharge pulse energy is
7.5 mJ/pulse .................................................................................................................... 181
and H2-air mixtures. Coupled discharge pulse energy is 10 mJ/pulse. ........................... 185
xxix
pulse energy is 10 mJ/pulse. ........................................................................................... 186
Figure 5.1: Schematic diagram of the discharge cells: (a) surface ionization wave discharge
cell, and (b) dielectric barrier discharge cell. .................................................................. 191
Figure 5.2: Voltage, current, and coupled energy waveforms, (a) surface ionization wave
discharge cell, last pulse in a burst of 75 pulses, pulse repetition rate 10 kHz; (b) dielectric
barrier discharge cell, last pulse in a burst of 20 pulses, pulse repetition rate 1 kHz. Ar
buffer. P=30 Torr, flow rate 0.1 SLM, FID pulser. ........................................................ 195
Figure 5.3: ICCD camera image of plasma emission in a surface ionization wave discharge
over water in argon buffer at the conditions of Fig. 5.4(a): (a) top view, (b) side view. P=18
Torr, camera gate 1 μs, FID pulser. ................................................................................ 196
Figure 5.4: ICCD camera image of plasma emission in a double dielectric barrier discharge
over water surface in argon buffer at the conditions of Fig. 5.4(b). P=30 Torr, camera gate
100 ns, FID pulser. .......................................................................................................... 197
Figure 5.5: Schematic of experimental setup for OH and H number density measurements.
......................................................................................................................................... 199
Figure 5.6: (a) OH LIF line images and (b) H TALIF line images in surface ionization wave
discharge over liquid water in Ar buffer. P=18 Torr, 4 μs after 75 pulse burst at 10 kHz,
FID pulser. ...................................................................................................................... 201
Figure 5.7: One-dimensional distributions of (a) absolute OH and (b) absolute H atoms
number densities, inferred from OH LIF and H TALIF line images shown in Fig. 5.6. 202
xxx
Figure 5.8: (a) OH LIF line images and (b) H TALIF line images in a double dielectric
barrier discharge over liquid water in Ar buffer. P=30 Torr, 2 μs after 20 pulse burst at 1
kHz, FID pulser. .............................................................................................................. 204
Figure 5.9: Two-dimensional distributions of absolute OH (a) and H atoms number
densities, inferred from OH LIF and H TALIF line images shown in Fig. 5.8. ............. 205
Figure 5.10: [H] and [OH] number densities at the conditions of Fig. 5.8, predicted by a 0-
D kinetic model. 50% - 50% H2O – Ar mixture, P=30 Torr. Discharge pulse repetition rate
1 kHz, coupled pulse energy 3.7 mJ/pulse, estimated .................................................... 208
xxxi
Chapter 1: Introduction
1.1. Nonequilibrium plasmas for plasma assisted combustion applications
Over the last decade, there has been significant progress demonstrating the utility
of non-equilibrium plasmas for augmentation of combustion phenomena, such as reduction
of ignition delay time and ignition temperature, as well as increase in flame stability and
flammability limits [1,2]. High peak voltage, nanosecond pulse duration discharges are of
particular interest for plasma assisted combustion since they can generate diffuse
nonequilibrium plasmas at high pressures (up to ~1 bar) and high pulse repetition rates (up
to ~100 kHz), and are characterized by a high peak reduced electric field, i.e. the ratio of
electric field to the number density, E/N, of several hundred Townsend (1 Td = 10-17
V∙cm2). At these high E/N values, a significant fraction of discharge input energy goes into
population of excited states of molecules (vibrational and electronic), as well as molecular
dissociation and ionization by electron impact. Figure 1.1 illustrates the electric discharge
plasma energy partition in air versus the reduced electric field, predicted by solving the
Boltzmann equation for plasma electrons.
1
Figure 1.1: Energy partition in a discharge in air vs. the reduced electric field, E/N [3].
The reduced electric field is a plasma parameter closely related to the average
electron energy, or electron temperature, 𝑇 , as shown in Figure 1.2 [4]. This relationship
can be illustrated as follows: in a steady-state plasma, the average electron energy gained
over a mean free path is
𝛿𝜀 = 𝑒𝐸𝜆 = 𝑒𝐸 (1.1)
where 𝑒 is the electron charge (C), 𝐸 is the electric field (V cm-1), and 𝜆 is the mean free
path. The mean free path equates to , where 𝜎 is the total electron-neutral collision cross
3
section (cm2), and 𝑁 is the number density (cm-3). The average electron energy is 𝜀~ 𝑇𝑒 ,
2
where 𝑇 is the electron temperature and 𝛿 is a factor such that 0< 𝛿<1. This estimate,
assuming 𝜎 and 𝛿 are independent of the electron energy, gives
𝜀= 𝑇 = (1.2)
2
Equation 1.2 shows that the average electron energy, 𝜀, and the electron
temperature, 𝑇 , are functions of the reduced electric field, E/N.
Figure 1.2: Electron temperature, Te, versus reduced electric field, E/N, in (a) nitrogen [3]
and (b) dry air [4] plasmas.
3
More accurately, solving the Boltzmann equation for plasma electrons, using a set
of energy-dependent collision cross-sections (for both electron-neutral and electron-
electron collisions), and incorporating electron energy loss factors for a number of elastic
and inelastic collision processes, gives the results shown in Figs. 1.1 and 1.2. From Figure
1.1, it can be see that at E/N ≥ 50-100 Td, significant discharge input energy fraction goes
to electronic excitation, molecular dissociation, and ionization. At these conditions,
electron temperature is in the range 𝑇 = 1-2 eV (see Fig. 1.2). At lower electric field values,
E/N ≈ 10-50 Td, a large fraction of discharge input energy goes to vibrational excitation of
N2 (see Fig. 1.1), at electron temperature of 𝑇 ≈ 1 eV (see Fig. 1.2). Electron temperatures
in nonequilibrium plasmas are high, up to several eV (1 eV ≈ 11,600 K), because electrons
gain energy from the electric field between collisions, and lose very little energy in elastic
collisions due to significant mass disparity between the electrons and the neutral species,
2𝑚
𝛿 ~ 𝑀 (1.3)
where 𝑚 is the mass of an electron, and 𝑀 is the mass of the neutral atomic or molecular
collisional partner. Although inelastic collisions result in far more significant energy loss,
up to several eV per collision, the average electron energy per collision remains quite low
[3].
Electron energy partition predicted in Figure 1.1 suggests that air plasmas produce
significant amounts of excited electronic states of nitrogen, such as N2(A3Σ, B3Π, C3Π,
etc.…) and radical species such as O atoms and H atoms (in fuel-air plasmas), by electron
impact processes [2]:
𝑁 (𝑋 𝛴) + 𝑒 → 𝑁 (𝐴 𝛴, 𝐵 𝛱, 𝐶 𝛱 … ) + 𝑒 (1.4)
4
𝑂 + 𝑒 → 𝑂( 𝑃) + 𝑂( 𝑃) + 𝑒 (1.5)
→ 𝑂( 𝑃) + 𝑂( 𝐷) + 𝑒
𝐻 +𝑒 →𝐻+𝐻+𝑒 (1.6)
In addition, collisional quenching of the excited states (including reactive
quenching) and reactions of radical species generated in the discharge considerably expand
the variety of chemical reactions in low-temperature fuel-air mixtures, resulting in fuel
oxidation and ignition at low gas temperatures below ignition temperature [5,6]. This
includes dissociation of molecular oxygen and hydrogen,
𝑁 (𝐴 𝛴, 𝐵 𝛱, 𝐶 𝛱 … ) + 𝑂 → 𝑁 (𝑋 𝛴) + 𝑂 + 𝑂, (1.7)
𝑁 (𝐴 𝛴, 𝐵 𝛱, 𝐶 𝛱 … ) + 𝐻 → 𝑁 (𝑋 𝛴) + 𝐻 + 𝐻, (1.8)
and OH formation in reactions of electronically excited O atoms,
𝑂( 𝐷) + 𝐻 → 𝑂𝐻 + 𝐻 (1.9)
Since H atoms, O atoms, and OH radicals react with fuel species even at relatively
low temperatures, they initiate chain branching reactions of fuel oxidation, such as
𝑂 + 𝐻 → 𝑂𝐻 + 𝐻, (1.10)
and
𝐻 + 𝑂 → 𝑂𝐻 + 𝑂, (1.11)
resulting in a gradual temperature rise and eventually in ignition [5]. Therefore,
measurements of excited electronic states of nitrogen and oxygen, as well as O atoms, H
atoms, and OH radicals, are critical for quantitative insight into kinetics of low-temperature
plasma-assisted fuel oxidation and ignition.
5
Returning to Figure 1.1, it is apparent that at E/N ≈ 10-50 Td, electron impact in air
plasmas can generate significant amounts of vibrationally excited N 2 molecules in the
ground electronic state:
𝑁 (𝑋 𝛴, 𝑣 = 0) + 𝑒 → 𝑁 (𝑋 𝛴, 𝑣 = 1 − 8) + 𝑒 (1.12)
These molecules decay primarily via vibrational-translational (V-T) relaxation by
O atoms in dry air, and with water vapor, carbon dioxide, and hydrocarbons in reacting
fuel-air mixtures. This vibrational excitation of nitrogen, at the very least, results in
significant storage of discharge input energy, and may significantly delay gas mixture
heating in pulsed discharges, which would occur on longer time scales by vibrational
relaxation [7]. Whether energy transfer from vibrationally excited N2 molecules may also
affect chemical reactions of radicals in low-temperature plasmas remains an open question,
and will be discussed in Section 1.3.
Qualitative analysis of modeling predictions plotted in Figures 1.1 and 1.2
illustrates that sustaining electric discharges at high electron temperatures (i.e. at high
reduced electric field values) would result in more efficient generation of electronically
excited species, as well as atomic and molecular radical species. This suggests the use of
high peak voltage, short pulse duration discharges, in which peak electric fields may
significantly exceed DC breakdown fields predicted by the Paschen law [3]. Paschen law
predictions of breakdown voltage in different gases are plotted in Figure 1.3, showing that
breakdown reduced electric field in air is approximately (E/N)BR ≈ 100 Td. Indeed, at P =
760 Torr, and a discharge gap distance of d = 1 cm, breakdown voltage in air is ~25 kV
(see Fig. 1.3), which corresponds to ~ ∗ ~100 𝑇𝑑.
6
Figure 1.3: Breakdown voltages in different gases over a wide range of Pd values (Paschen
curves) [3].
During DC breakdown, the rate of electron production, , is controlled by
electron impact ionization.
=𝛼 𝑛 𝑣 (1.13)
where 𝑣 (cm/s) is the electron drift velocity, 𝑛 (cm-3) is the electron number density,
and 𝛼 is the ionization coefficient (cm-1), also known as the Townsend ionization
coefficient [3]. The Townsend ionization coefficient represents the number of ionization
events caused by a single electron along a 1 cm path in the externally applied field.
Experimental data for the Townsend ionization coefficient are usually approximated by the
following equation,
7
𝛼 = 𝐴𝑝𝑒 (1.14)
where 𝐴 and 𝐵 are constants determined from fitting the experimental data with units of
(cm-1Torr-1) and (Vcm-1Torr-1), respectively, 𝑝 is the pressure (Torr), and is the reduced
electric field in terms of pressure (Vcm-1Torr-1) [3].
Reduced electric field values in quasi-steady-state discharges (DC, RF, and MW)
are usually significantly lower than the breakdown threshold, since at these conditions
ionization occurs not only by electron impact from the ground electronic state of dominant
neutral species,
𝑁 +𝑒 →𝑁 +𝑒+𝑒 (1.15)
𝐴𝑟 + 𝑒 → 𝐴𝑟 + 𝑒 + 𝑒 (1.16)
but also from a variety of other processes, including step-wise ionization of electronically
excited molecules,
𝑁 (𝑋 𝛴) + 𝑒 → 𝑁 (𝐴 𝛴, 𝐵 𝛱, 𝐶 𝛱 … ) + 𝑒 (1.17)
𝑁 (𝐴 𝛴, 𝐵 𝛱, 𝐶 𝛱 … ) + 𝑒 → 𝑁 + 𝑒 + 𝑒 (1.18)
associative ionization,
𝑁 + 𝑁∗ → 𝑁 + 𝑒 (1.19)
and Penning ionization [3],
𝐴𝑟 + 𝐴𝑟 ∗ → 𝐴𝑟 + 𝐴𝑟 + 𝑒 (1.20)
In addition, in quasi-steady-state discharges, ionization may occur predominantly
in near-electrode layers, such as the cathode layer of a DC glow discharge [3], where the
electric field is much higher than in the bulk of the plasma. Therefore, short-pulse duration
8
discharges are more efficient for electronically excited species generation and molecular
dissociation, due to higher peak E/N values and larger energy fraction going into electronic
excitation and dissociation (see Fig. 1.1).
Additionally, short-pulse duration, high peak voltage discharges have been widely
used for plasma assisted combustion applications due to their superior stability at high
pressures and high specific powers [3]. Figure 1.4(a-h) illustrates the mechanism of one
of the most common ionization instabilities occurring in high-pressure discharge plasmas.
Basically, electron density rise in these plasmas (due to random fluctuations) (a) results in
an electric current increase (b) and an increase in Joule heating which accelerates local
temperature rise (c). If this occurs on a sub-acoustic time scale, τa ~ r/a, where r is an
incipient discharge filament radius and a is the speed of sound, the pressure remains
essentially constant (d), resulting in the number density reduction (e). This increases the
reduced electric field (f), increasing the electron temperature (g) and the rate of electron
impact ionization, which depends exponentially on the reduced electric field and the
electron temperature (h), thereby completing the positive feedback loop shown in Fig. 1.4.
9
Figure 1.4: Schematic diagram illustrating positive feedback development of the
ionization-heating instability in an electric discharge, resulting in a discharge constriction
and glow-to-arc transition in high pressure plasmas [8].
At low pressures, the instability feedback shown in Figure 1.4 is limited by
ambipolar diffusion and heat transfer by conduction. Both processes help in dissipating the
incipient discharge filaments. However, at high pressures or high specific powers (i.e. high
Joule heating rates), the characteristic time for ionization stability growth becomes much
shorter, such that a diffuse discharge plasma collapses into a series of high temperature,
high current filaments, or into a single arc filament. The ionization instability can be
delayed by reducing the flow residence time in the discharge (e.g. aerodynamic
stabilization), using individually ballasted cathodes (i.e. reducing the voltage applied to the
cathode, 𝑉 =𝑉 − 𝐼𝑅, where 𝑅 is the ballast resistance), or by repetitive pulsing of
the electric field, such that the pulse duration is much shorter than the characteristic time
for the instability development. In the latter case, discharge pulses need to be repeated at a
10
pulse repetition rate higher than characteristic frequencies for electron recombination and
attachment, which control the decay of electron number density, , and therefore the
plasma decay time in the afterglow.
= −𝛽 𝑛 𝑛 − 𝛼 𝑛 𝑛 𝑁 (1.21)
In Equation 1.21, 𝛽 is the electron-ion recombination coefficient (cm3s-1), 𝛼
is the three-body electron attachment coefficient (cm6s-1), 𝑛 ≈ 𝑛 is the ion number
density, 𝑛 is the number density of the electronegative species, such as O 2 in air, and 𝑁
is the total number density (cm-3).
In summary, efficient generation of excited electronic species and radicals in a
high-pressure plasma requires repetitively pulsed discharges operated at high peak reduced
electric fields (high peak voltages), short discharge pulse durations, and high pulse
repetition rates – an approach used in the present work.
From a fundamental kinetics perspective, the dominant energy transfer and
chemical reaction processes in these plasmas remain not fully understood. The
overwhelming majority of the kinetic data in fuel-air plasma is obtained at near room
temperature and at low pressures [9,10,11,12,13]. This includes rates of collisional
quenching of electronically excited states, rate coefficients of chemical reactions of
electronically exited species [7, 14,15,16, and 17] and state-specific rates of vibrational
energy transfer [18]. Very little data has been obtained in a well characterized plasma at
elevated temperatures. In kinetic modeling predictions of plasma assisted combustion,
these processes are simply combined with the “conventional” chemical reaction
mechanisms, including reactions of species in their ground electronic states. These

11
mechanisms (such as Konnov [19], USC [20], and Popov [14, 21]) were validated in flow
reactors for high temperatures (T > 900 K), and their predictions must be extrapolated to
temperatures below self-ignition (𝑇 ≈ 750 𝐾 − 900 𝐾), down to room temperature.
This creates a significant gap between data obtained in plasmas at low temperatures and
data taken in flames and flow reactors at considerably higher temperatures, leaving the
reacting plasma kinetics over a wide temperature range, 𝑇 = 300 𝐾 < 𝑇 < 𝑇 = 750 𝐾,
largely unexplored. Specifically, direct quantitative measurements of efficiency of primary
radical species generation (such as H and O atoms) in well-characterized ns pulse discharge
plasmas are limited. Additionally, net rates of primary radical consumption in low-
temperature fuel-oxidizer plasmas, which provide insight into the dominant radical chain
propagation, chain branching, and chain termination for hydrocarbon fuels at low
temperatures remains largely uncertain. Without such understanding, predictive kinetic
modeling and analysis of plasma assisted combustion phenomena remains problematic.
This is critical in using nonequilibrium plasmas for applications including gas turbines, jet
engines and afterburners, and scramjet engines, where inlet flow temperatures are up to
several hundred degrees higher than room temperature, due to flow compression through a
compressor or in a shock train, but below ignition temperature.
1.2. Review of previous studies of nonequilibrium plasmas in ns pulse discharges
Measurements of absolute number densities of atomic species and radicals in low-
temperature air and fuel-oxidizer plasmas sustained by ns pulse discharges have been done
previously. Absolute number density measurements of O atoms in a plane-to-plane double
12
dielectric barrier discharge (DBD) by Two-photon Laser Induced Fluorescence (TALIF),
compared to kinetic modeling predictions, determined that up to half of the energy coupled
by a nanosecond pulse discharge in air goes to O2 dissociation, resulting in a significant
initial concentration of O atoms in the afterglow [10]. Comprehensive studies, Ref. [22],
of elevated initial temperature (T0= 1000 K), atmospheric pressure, ns pulse discharge
plasma filaments using TALIF, Cavity Ring Down Spectroscopy (CRDS), and Optical
Emission Spectroscopy (OES) to measure O, N2(A), and N2(B), and N2(C) number
densities, respectively, confirmed that O atoms in the plasma are formed by a two-step
mechanism of electronic excitation of N2 by electron impact, followed by quenching of N2
excited electronic states by O2, with higher number density produced in a constricted
plasma filament (by about an order of magnitude) compared to a diffuse filament plasma
[22]. More recently in Ref. [11], measurements of absolute number density of O and H
atoms by femtosecond TALIF in H2-Ar, H2-He, and O2-He mixtures at a pressure of P=
100 Torr demonstrated that ~60% of the total discharge coupled energy goes to O atom
production, and ~70% of the total discharge coupled energy goes to H atom production,
both by electron impact dissociation and by dissociative quenching of electronically
excited helium and argon [11]. The use of femtosecond laser diagnostics allowed direct O
and H atom quenching rate measurements, and provided two-dimensional distributions of
concentrations of O and H atoms in a diffuse filament discharge, illustrating a non-uniform
spatial distribution of atoms, with more H atoms produced near the cathode and more O
atoms generated near the anode, as shown in Figure 1.5 [11].
13
Figure 1.5: Two-dimensional number density distributions of (a) H atoms in a 1% H 2-Ar
mixture at P= 100 Torr and (b) O atoms in a 1% O 2-He mixture at P= 100 Torr, illustrating
non-uniform distributions of atoms at t= 25 µs after the end of the discharge burst [11].
Experimental studies of radical chemistry have been made in fuel-oxygen-argon
mixtures, used as a substitute for fuel-air mixtures [12,13,23,24]. Adding an argon buffer
simplifies the plasma chemistry, removing kinetics of vibrationally excited nitrogen and
chemical reactions of NOx formation, and allows sustaining diffuse plasmas in dilute fuel-
oxidizer mixtures over a wide range of pressures (from 40 Torr to 760 Torr) [12,13,23,24].
In the first experiment, measurements of gas temperature, made by rotational Coherent
Anti-stokes Raman Spectroscopy (CARS), were used in conjunction with measurements
of absolute atomic oxygen number density, made by TALIF, to identify the relevant
consumption pathways in low temperature (T=300 K), low pressure (P=40 Torr) H 2-O2-Ar
and C2H4-O2-Ar mixtures, in a DBD plasma flow reactor [12]. Comparison of the
experimental results with kinetic modeling predictions using two different chemistry
14
mechanisms for H2-O2 mixtures [14,16] illustrated that the use of the Popov mechanism
[14] resulted in better agreement with the data for low temperature oxidation of hydrogen.
In C2H4-O2-Ar mixtures, the use of the chemistry mechanism for C2H4 [20] resulted only
in fair agreement with the data, underpredicting both the gas temperature and the O atom
number density [12]. Further studies, in plasma flow reactors in dilute H 2-O2-Ar mixtures,
Ref. [23], used ex-situ measurements of stable species (H2 and O2) by gas chromatography
and in-situ measurements of radical species (OH by LIF), to compare fuel consumption
during plasma-assisted-oxidation and during thermal oxidation at atmospheric pressure and
over a temperature range of T= 420-1100 K. During thermal oxidation, with no plasma
present, all hydrogen was oxidized above T= 860 K [23]. During plasma assisted fuel
oxidation, partial oxidation of H2 began at much lower temperatures, T= 470 K, and all
hydrogen was oxidized above T = 800 K [23]. The same authors also studied plasma-
assisted-oxidation and plasma-assisted-pyrolysis of C 2H4-O2-Ar mixtures at atmospheric
pressure, Ref. [24], over a temperature range of T= 520- 1250 K, using only ex-situ gas
chromatography of stable species [24]. The branching ratios inferred for the Ar* + C 2H4
reaction were used in kinetic modeling predictions. Comparison between data and kinetic
modeling predictions showed that ~60% of the C2H4 dissociated in pyrolysis is used in H
atom generation. Stable species number densities during plasma assisted oxidation,
however, were not reproduced well, illustrating deficiencies in the plasma chemistry
mechanism [24].
In a diffuse filament ns pulse discharge in H2-O2-Ar mixtures, measurements of
absolute number density distributions of H atoms by TALIF and absolute number density
15
distributions of OH radicals by LIF were made at low pressure (P= 40 Torr) and low
temperature (T0= 300 K) conditions, to study their time resolved and spatially resolved
evolution during and after a burst of nanosecond discharge pulses [13]. It was found that
peak H atom number density during the burst was higher compared to that of OH radicals.
After the burst, H atoms diffuse rapidly out of the discharge filament, and contribute to OH
production in the afterglow outside of the discharge filament, as shown in Figure 1.6. As a
result, the OH number density radial profile had a main peak on the filament centerline and
two secondary peaks outside of the filament, which remained nearly unchanged up to t= 1
ms after the end of the discharge burst [13].
16
Figure 1.6: Radial distributions of H and OH concentrations during and after the discharge
burst between two spherical copper electrodes, measured in a mixture of Ar:O 2:H2 =
80:20:2, at P= 40 Torr, T0= 300 K, and a discharge pulse repetition rate of 100 kHz [13].
Insight into kinetics of fuel-air mixtures, excited by nanosecond pulse discharges,
requires quantitative understanding of the role of electronically and vibrationally excited
17
nitrogen and oxygen in chemical kinetics [7,25,26]. Previously, vibrational excitation and
relaxation of nitrogen has been studied for two different ns pulse discharge geometries,
repetitively pulsed plane-to-plane discharge and single pulse pin-to-pin discharge, by
measuring N2 vibrational level populations, using picosecond (ps) CARS [25]. At the
conditions of high specific energy loading in air at P = 100 Torr, the fraction of total
coupled discharge energy that went to vibrational excitation of nitrogen by electron impact
was ~50% in a plane-to-plane discharge and ~40% in the pin-to-pin discharge. In the pin-
to-pin discharge, strong N2 vibrational excitation with vibrational levels up to v=9
populated, was detected [25]. On the contrary, in the plane-to-plane discharge, N 2
vibrational nonequilibrium was very weak, near the detection limit of the CARS system,
due to a much larger plasma volume compared to the pin-to-pin discharge and due to a
longer time scale of energy loading using a repetitively pulsed discharge, compared to a
single pulse discharge. In the afterglow, the number density of vibrational quanta per N2
molecule increased, suggesting that energy was added to the N2 vibrational mode from
another “reservoir” [25].
18
Figure 1.7: Time-resolved gas temperature (experimental and predicted) during and after
the main discharge pulse in air and in a H2-air mixture at an equivalence ratio of 0.42 [7].
In another study, two-stage thermalization in air and H2-air mixtures was detected
at low pressure (P = 40 Torr), low temperature (T0= 300 K) conditions, using pure
rotational ps CARS to measure time resolved gas temperature [7], as shown in Figure 1.7.
The first stage is “rapid” (t ≈100 ns -1 µs) heating of the mixture by collisional quenching
of electronically excited states of N2 and O2, and the second stage is “slow” (t ≈200 µs -
500 µs) heating by N2 vibrational relaxation by O atoms (in air), as well as chemical energy
release during partial oxidation of H2 (in hydrogen-air mixtures). The rapid heating stage
was found to strongly affect rates of chemical reactions among radicals generated by
electron impact dissociation and quenching of electronically excited species [7].
19
Subsequent studies of atmospheric pressure, pin-to-plane ns pulse discharges in air, Ref.
[26] confirmed the existence of the two-stage heating mechanism, using spontaneous
Raman scattering (SRS) to measure time-resolved vibrational level populations of N 2 and
O2 during and after a ns pulse discharge, at three different axial locations along the filament
[26]. The authors determined that relaxation time of vibrationally excited N 2 near the pin
electrode (τ ≈ 200 µs) is much faster compared to the relaxation time in the rest of the
discharge filament (τ ≈ 10 ms) [26].
Previous studies of excited species and radicals in fuel-air mixtures have offered
quantitative insight into oxidation and ignition kinetics at low pressure (P ≤ 100 Torr)
conditions [5,6,28]. Plasma assisted oxidation of lean H2-air and hydrocarbon-air mixtures
was studied in a repetitively pulsed ns discharge, at high pulse repetition rates in a plane-
to-plane geometry, at a pressure of P= 100 Torr and elevated initial temperature of T 0= 500
K using measurements of absolute OH radical number density by LIF, and N 2 vibrational
level populations and gas temperature by ps CARS [6]. The OH decay rate in H 2-air and
CH4-air mixtures was sufficiently slow, such that OH accumulated between the pulses. The
OH decay rate in lean ethylene and propane mixtures was much faster, such that nearly all
OH was consumed prior to the next discharge pulse.
20
Figure 1.8: Comparison of experimental and predicted time-resolved, absolute OH number
densities after ν= 10 kHz, 50-pulse discharge burst in (a) H2-air mixture, (b) C2H4-air
mixtures, and (c) C3H8-air mixtures at P= 100 Torr and T0= 500 K [6].
21
Kinetic modeling predictions of several different chemistry mechanisms were
compared with the data, as shown in Figure 1.8, and the use of the kinetic mechanism from
Ref. [27] showed good agreement with experimental data in lean hydrogen, methane, and
ethylene mixtures with air, but only qualitative agreement with the data in lean propane-
air mixtures. This illustrates the need for additional quantitative data on species number
densities, to understand better the production and consumption pathways of radicals at low-
temperatures (below ignition temperature), as well as the need for development of more
accurate kinetic mechanisms for plasma assisted fuel oxidation. The same authors made
further studies of plasma-assisted-ignition of H2-air mixtures, Ref. [5], in the same plane-
to-plane plasma flow reactor, at a pressure of P = 100 Torr and elevated initial temperature
of T0= 470-500 K. During a burst of nanosecond discharge pulses, plasma assisted H 2
oxidation caused a gradual temperature rise, and H atom accumulation leading to ignition,
during which the OH number density increased by a factor of 20-50 [5], and the
temperature peaks at T = 1500-1600 K. The use of nonequilibrium plasma to generate a
pool of primary radicals (O atoms) led to ignition occurring at temperatures up to 200 K
lower than auto ignition, and reduced the ignition delay time. Increasing the number of
nanosecond discharge pulses was found to reduce ignition delay time after the end of the
discharge burst [5]. Closely related studies of vibrational energy transfer and low-
temperature chemical kinetics in air at P= 100 Torr and T0= 300 K, Ref. [28], were made
by measuring radical species number densities (NO by LIF, and N and O by TALIF), gas
temperature, and N2 vibrational level populations (by ps CARS) in the afterglow of a
diffuse filament ns pulse discharge maintained between bare metal, spherical electrodes
22
[28]. In this work, N2 vibrational levels up to v=4 were detected. A strong vibrational
nonequilibrium was observed, with “first level” N 2 vibrational temperature, Tv ≈ 2000 K,
much higher compared to the gas temperature, T≈ 370 K. N2 vibrational temperature was
observed to rise 10-100 µs after the ns discharge pulse, due to the “downward” vibrational-
vibrational (V-V) energy transfer from higher vibrational states to the v=1 state,
𝑁 (𝑣) + 𝑁 (𝑤 = 0) → 𝑁 (𝑣 − 1) + 𝑁 (𝑤 = 1) (1.22)
Plasma chemistry at these high specific energy loading, low-temperature
conditions was dominated by the reverse Zel’dovich reactions and ozone formation
reactions [28], discussed further in Section 1.3.
In a critically important study, absolute number density measurements of HO 2 was
measured at the exit of a laminar flow reactor at 1 atm, using mid-IR Faraday Rotation
Spectroscopy (FRS) [29]. This diagnostic has a sensitivity of < 1ppm over a temperature
range of T = 398-673 K, and is a powerful new technique to measure the HO 2 radical,
which is a critically important species in H atom decay and OH production in low-
temperature plasmas, but is extremely challenging to quantify.
The work discussed above is only a brief overview of significant advances in
studies of plasma assisted combustion kinetics over the last 15 years. A far more
comprehensive review can be found in Refs. [1,2,14]. Summarizing the previous work on
radical species measurements in fuel-air and fuel-oxidizer mixtures excited by
nonequilibrium plasmas, it should be noted that in most of these measurements, specific
discharge energy loading was known with a significant uncertainty. The difficulty in
coupled energy measurements in high-pressure ns pulse discharges is two-fold. First, in
23
discharges sustained between two metal electrodes, such as in Refs. [7,9,11,22], spatial
distribution of coupled energy is strongly affected by (a) cathode layer formation, and (b)
discharge constriction into a relatively small diameter filament, on the order of ~0.1- 1.0
mm [9]. In self-sustained pulsed electric discharges, electric field in the thin cathode layer
exceeds considerably the field in the rest of the discharge plasma, resulting in a significant
fraction of the total input energy to be coupled in the cathode layer. Basically, formation
of the cathode layer is necessary to generate sufficiently high electric field near the cathode,
such that the ions can be accelerated toward the cathode surfaces and produce electrons by
secondary electron emission. The ion current to the cathode balances the electron current
to the anode [30]. Obviously, this requires the electric field to be much higher compared to
that near the anode or in the rest of the discharge, since the ion mobility is much lower than
the electron mobility.
Although a simple theory of the cathode layer in a DC glow discharge predicts the
electric field on the cathode [3], the thickness of the cathode layer, and the voltage drop
across the cathode layer, its predictions cannot be expected to be accurate in high peak
current, ns pulse discharges. Therefore, the amount of energy coupled to the plasma in the
cathode layer is highly uncertain and is difficult to measure directly in high-pressure, short
pulse duration plasmas.
The diameter of the plasma filament, as well as the current density distribution
across the filament, depend on many factors, such as the size of the “cathode spot,” the
electrode geometry, the discharge gap, and the rates of ambipolar diffusion and radial heat
transfer by convection [3]. Therefore, measurements of total energy coupled to the plasma
24
during the discharge pulse are not sufficient to characterize the energy loading in the
discharge filament quantitatively.
Due to this problem, many previous studies of plasma assisted combustion kinetics
used a ns pulse discharge between electrodes separated by one or two dielectric barriers, in
a simple plane-to-plane geometry [5,6,23,25]. It has been shown that at these conditions,
operating the discharge at high pulse repetition rates generates diffuse plasmas,
reproducible pulse-to-pulse [5]. The main role of the dielectric barriers is to prevent
formation of the cathode layer and to limit the current in the incipient filaments, by surface
charge accumulation of the dielectric surfaces.
At these conditions, energy coupled to the plasma can be predicted theoretically
[31], or measured directly [32]. Previous experiments have shown that, as the pressure is
increased, the plane-to-plane, double dielectric barrier discharge (DBD) gradually
develops well-defined, reproducible filaments [31,32]. However, mild preheating of the
gas mixture considerably improves the plasma uniformity, due to enhanced diffusion and
heat transfer by conduction, which helps to dissipate the “hot spots” created by the
filaments [6].
In Refs. [5] and [6], extensive measurements of absolute, time resolved OH number
density have been done in a double DBD, ns pulse discharge, operated at a high pulse
repetition rate in a preheated flow of reactants. However, kinetic modeling analysis of these
results was complicated by generation of the plasma outside of the discharge flow reactor,
such that accurate measurements of energy coupled to the plasma in the reactant flow was
not feasible. The same problem was encountered in a more recent study in a similar
25
discharge geometry, in a preheated flow of reactants at atmospheric pressure [23, 24]. Thus,
one of the main objectives of the present work is to demonstrate operation of a diffuse,
high pressure plasma in a preheated flow, quantify plasma nonuniformity, and to measure
energy coupled to the plasma accurately, by preventing breakdown and/or corona discharge
formation outside of the reactor channel. For this, custom-made, high-bandwidth voltage
and current probes [32] are used, which also considerably reduce the effect of
electromagnetic noise on measurement results.
1.3. Effect of vibrational excitation on plasma chemical reaction kinetics
The role of vibrationally excited nitrogen molecules in the ground electronic state,
𝑁 (𝑋 𝛴, 𝑣), in chemical kinetics of reacting nonequilibrium plasmas remains rather poorly
understood. It may well be significant for two principal reasons, (i) high input energy
fraction into N2 vibrational excitation by electron impact in electric discharges in air and
fuel-air mixtures, and (ii) relatively slow N2 vibrational relaxation rate, even in plasmas
with significant concentrations of hydrogen, hydrocarbon species, and oxygen atoms.
Because of this, considerable amount of energy in transient plasmas is stored in the
vibrational energy mode of molecular nitrogen, which affects the rate of temperature rise
[31,33] and, indirectly, rate coefficients of chemical reactions. A more direct effect of N 2
vibrational excitation on plasma chemistry, either via reactions with 𝑁 (𝑋 𝛴, 𝑣) as one of
the reactants, or via vibrational energy transfer from nitrogen molecules to other reactive
molecular species, has been a matter of some debate.
26
The effect of 𝑁 (𝑋 𝛴, 𝑣) molecules on nitric oxide formation in low-temperature
glow discharges in air has been discussed in the literature extensively (e.g. see
[34,35,36,37,38] and references therein). Direct measurements or high-fidelity theoretical
predictions of state-specific rates of these reactions at low temperatures are not available.
The principal difficulty in quantifying this effect is that it would require measurements of
vibrational populations of the ground electronic state of nitrogen, e.g. by CARS or by
spontaneous Raman scattering, along with measurements of number densities of other
reactants and products, such as NO, O atoms, and N atoms. One recent example of such
measurements is the use of simultaneous N2 CARS, NO LIF, and TALIF (N and O) [28]
to quantify the effect of 𝑁 (𝑋 𝛴, 𝑣) + 𝑂 → 𝑁𝑂 + 𝑁 reaction on the rate of NO formation
in a low-temperature air plasma generated by a ns pulse discharge. At these conditions, the
contribution of this reaction, compared to reactive quenching of excited electronic states
of N2, appears to be minor [9]. However, in quasi-steady state low-pressure plasmas
sustained by DC or RF discharges, at considerably lower reduced electric fields, this
reaction may well be one of the dominant channels of nitric oxide formation [28].
Another reaction pathway, considered in Ref. [2], suggests that vibrationally
excited nitrogen may delay OH recombination due to near-resonance vibrational energy
transfer to relatively “unreactive” HO2 radical,
1 1 1
𝑁 𝑋 𝛴, 𝑣 = 1 + 𝐻𝑂 → 𝑁 𝑋 𝛴, 𝑣 = 0 + 𝐻𝑂 (𝑣 + 𝑣 ) → 𝑁 𝑋 𝛴 + 𝐻 + 𝑂 (1.23)
It was also suggested that this process may sustain chain propagation and chain
branching chemical reactions of H atoms in low-temperature afterglow in fuel-air mixtures,
and thus extend the OH radical lifetime. Recent time-resolved measurements of OH
27
number density in nanosecond pulse discharge afterglow in preheated atmospheric pressure
hydrocarbon-air mixtures [39], which demonstrated anomalously long OH lifetime and its
non-monotonous decay in the afterglow, seem to support this hypothesis. Kinetic modeling
of atmospheric pressure hydrogen-air plasmas with significant initial vibrational excitation
of nitrogen, Tv(N2)=3000 K, incorporating the N2-HO2 V-V process of Equation 1.23,
predicts that this effect would result in transient OH number density rise in the afterglow,
peaking on the time scale of ~ 0.3-4.0 ms at T=600-800 K, as shown in Figure 1.9 [2].
28
Figure 1.9: Predicted OH number density in the atmospheric pressure 3% H 2-air mixture
excited by a ns pulse discharge at different initial temperatures and initial N 2 vibrational
temperature of Tv = 3000 K: (a) “conventional” combustion chemistry only, and (b) with
the inclusion of N2-HO2 V-V transfer of Eqn. 1.23 [2].
However, gas temperature and N2 vibrational temperature in the experiments of
Ref. [39] have not been measured directly, such that a more direct effect of heating the
29
flow by the discharge on chemical reaction rates and OH lifetime in the afterglow cannot
be ruled out. Note that sustaining strong vibrational excitation in the late afterglow in H 2-
air and hydrocarbon-air mixtures, without significant heating of the gas, appears unlikely
due to heating caused by vibrational relaxation. Direct vibrational temperature
measurements at these conditions are critical due to significant uncertainty in state-specific
rates of N2 vibrational relaxation in reacting fuel-air mixtures.
Although it may well be possible that reactions of vibrationally excited N 2
molecules in fuel-air plasmas affect low-temperature fuel oxidation chemistry via reaction
of Eqn. 1.23, as suggested in Ref. [2], additional experimental studies are needed to verify
this hypothesis. In the present work, we use simultaneous N2 vibrational CARS and OH
LIF measurements, (i) to determine time-resolved temperature and N2 vibrational
temperature, (ii) to monitor absolute, time resolved OH number density, and (iii) to detect
a possible effect of N2 vibrational excitation on OH kinetics in lean H2-air afterglow
plasmas sustained by a ns pulse discharge, at elevated gas temperatures and strong
vibrational excitation. The approach used in the present experiments is similar to the one
in our previous work, where a ns pulse, diffuse filament discharge in air was used to
generate strongly nonequilibrium plasmas, up to T=850 K and Tv(N2)= 3000 K [25].
Kinetic modeling of ns pulse discharge and afterglow in H 2-air mixtures using the model
developed in Refs. [6, 40] is used to interpret the experimental results and to provide the
modeling predictions at the conditions where experimental data are not available.
30
1.4. Liquid/Vapor interfaces and plasma medicine
Kinetics of nonequilibrium electric discharges in liquids and at liquid-vapor
interfaces is of great interest from the fundamental viewpoint [41], and is of critical
importance for development of applications such as reactive nitrogen / oxygen species
generation [42], plasma activation of water for sterilization and wound treatment [43,44],
removal of volatile organic compounds from water and aqueous solutions [45,46], and
plasma chemical reforming of liquid hydrocarbons and oxygenates for portable fuel cells
[47,48,49,50], proposed as an alternative to thermo-catalytic fuel reforming. In spite of
extensive experimental (e.g. see [41] and references therein) and modeling (e.g. see
[46,51,52]) studies of nonequilibrium plasmas generated in the presence of liquids, kinetics
of plasma chemical reactions at these conditions remains poorly understood. One of the
main difficulties encountered in quantitative studies of liquid / vapor phase plasma
chemistry is sustaining the plasma at controlled, well-reproducible, ample optical access
conditions, which would lend themselves to in situ, time- and spatially resolved optical
diagnostics. Typically, in the experiments the liquid / vapor interface plasma is generated
away from the liquid surface [44], or sustained as a stochastic manifold of streamers
propagating in the vapor phase [45], in the bulk of the liquid [46,49] (likely via microscopic
bubbles although formation of plasma in the liquid phase has been suggested [53]), in
macroscopic bubbles [54], or over the liquid surface [55,56]. Quantifying plasma
parameters, as well as accurate time-resolved and spatially-resolved species concentrations
measurements at these conditions is extremely difficult. Lack of in situ measurements of
31
plasma parameters at well-characterized conditions also makes difficult development and
validation of kinetic models, and assessing their predictive capability.
In the present work, a novel way of sustaining diffuse, highly reproducible plasmas
at liquid-vapor interfaces is demonstrated, using surface ionization waves generated by
high-voltage, nanosecond pulse duration discharges. The basic concept of this approach is
qualitatively similar to volumetric and surface Fast Ionization Wave (FIW) ns pulse
discharges, which are known to generate diffuse plasmas over large volumes or large
surface areas, in a wide range of pressures [57,58]. Typically, wave amplitude attenuation
rate is fairly low, due to high electrical conductivity of the plasma column (or near-surface
plasma sheet) behind the wave, which acts essentially as an extension of the high-voltage
electrode. The fundamental difference of FIW plasmas from other types of nonequilibrium
plasmas is high peak electric field in the wave front (up to E/N ~ 1000 Td = 10 -14 V∙cm2).
High overvoltage in presence of weak residual ionization results in nearly plane ionization
wave front development, instead of isolated streamer breakdown [57,58]. High electron
energy achieved in the wave front significantly accelerates electron impact processes with
high energy threshold, such as ionization and molecular dissociation, thus generating
reactive radical species in molecular plasmas. Also, at high peak electron energies
involved, the rate of electron impact ionization greatly exceeds the rate of dissociative
attachment, such that diffuse plasmas can be sustained even in strongly electronegative
gases mixtures, such as water vapor. Finally, high wave speed (of the order of ~0.1-1
cm/ns) greatly enhances plasma stability, since ionization instabilities do not have time to
develop during the wave front residence time.
32
Short time scales involved in surface ionization wave discharges launched over a
surface of a dielectric, or weakly conductive, liquid (typically ~10-100 ns) preclude bulk
motion of the liquid that may be caused by charge accumulation on the liquid surface. In
addition, net surface charge accumulation may be controlled, and reduced significantly, by
alternating the high voltage pulse polarity, and by using a weakly conductive liquid. Peak
electric field in the surface ionization wave may be further enhanced if the discharge is
generated over a liquid with high dielectric constant. At these conditions, high energy
electrons generated in the thin near-surface plasma layer initiate electron impact
dissociation and ionization of evaporating reactants, with potentially high yield of radical
species, such as O, H, and OH in surface plasmas sustained over liquid water and aqueous
solutions, and CH3 in surface plasmas over liquid hydrocarbons and oxygenates. The yield
of these reactions may be enhanced considerably by operating the discharge at a high
velocity of a buffer gas flowing above the liquid surface, at the conditions when the net
rate of evaporation is high. Generation of the initial radical pool may result in rich plasma
chemistry and formation of a variety of stable product species, depending on the initial
composition of the liquid and the buffer gas flow. Note that chemical reactions in
repetitively pulsed plasma at a liquid-vapor interface would occur at nearly constant
temperature conditions, since Joule heating would be limited both by high specific heat of
the liquid and by high latent heat of vaporization (~0.4 eV/molecule for water and
alcohols). All this suggests that surface ionization wave discharge plasma sustained at a
liquid-vapor interface can be used as an experimental platform for studies of near-surface
plasma chemical reaction kinetics, at the conditions when the interface acts as a high-yield
33
“plane surface source” of radical species. At this time, kinetics of these reactions, as well
as time evolution of electric field and electron density distributions along the interface
remain largely unknown.
1.5. Structure of dissertation
The following chapters of this dissertation are organized as follows: Chapter 2
presents an overview of the background theory and technical issues for laser diagnostics
used in the present work, LIF, TALIF, and TDLAS. Chapter 3 discusses characterization
of a plane-to-plane double dielectric barrier, repetitively pulsed discharge; time resolved
radical species number densities in this discharge by TALIF; comparison of experimental
results with kinetic modeling predictions; and discussion of implications for low-
temperature plasma-assisted fuel oxidation kinetics. Chapter 4 discusses absolute, time-
resolved OH number density measurements (by LIF), as well as time-resolved gas
temperature and N2 vibrational temperature measurements in a “diffuse filament” ns pulse
discharge; comparison with kinetic modeling predictions; and discussion of the effect of
vibrational excitation on energy thermalization and chemical reaction kinetics in high
nonequilibrium fuel-air plasma. Chapter 5 discusses measurements of OH and H number
densities in a ns pulse, surface ionization wave discharges sustained over a liquid surface,
using the same laser diagnostic techniques.
34
Chapter 2: Background on Laser Diagnostics
2.1. OH Laser Induced Fluorescence (LIF)
In the present work, the number density of OH molecules is measured using single
photon Laser Induced Fluorescence (LIF). LIF is a two-step process: first the molecule of
interest absorbs a laser photon, exciting it from a low energy state (ground electronic and
vibrational state in the present work) to a higher energy state, which subsequently decays
spontaneously by emitting a fluorescence photon and relaxing back either to a lower energy
state or to the original ground state (in resonance LIF) [59]. The fluorescence signal is
detected and calibrated to measure the probe species number density. The LIF signal
(photons/second) is proportional to the number density of excited probe species and the
fluorescence decay rate,
𝑆 =𝑉 𝜂𝑁 𝐴 (2.1)
In this equation, the LIF signal, 𝑆 , is the product of the spontaneous emission
rate, 𝐴 , also known as the Einstein coefficient for spontaneous emission, and optical
parameters including the fluorescence signal collection efficiency, 𝜂, the probe volume, 𝑉
, the fractional solid angle from which the signal is collected, , the effectiveness of
converting fluorescence photons into detector counts (or volts), 𝛽; and the number density
of the excited state, 𝑁 .

35
Only a small fraction of the probe species is excited by the laser radiation. The
relationship between the excited state population and the total number density of the probe
species is obtained as follows. For single photon absorption, consider a two-level system
shown in Figure 2.1 to obtain the state population rate equations.
Figure 2.1: Two level energy diagram for single photon laser excitation, showing dominant
energy transfer processes between the ground state, N0, and the excited state, N1.
Assuming that the laser pulse photon energy is sufficiently low to neglect
photoionization from the excited state, the rate equations are
( )
= −𝑁 (𝑡)𝑓 𝑏 + 𝑁 (𝑡)(𝐴 +𝑄+𝑏 ) (2.2)
( )
= 𝑁 (𝑡)𝑓 𝑏 − 𝑁 (𝑡)(𝐴 +𝑄+𝑏 ) (2.3)
The excited state, 𝑁 , is populated by absorption, 𝑏 , from the ground state, with
the population of 𝑓 𝑁 , where 𝑁 is the total ground state population and 𝑓 is the
Boltzmann fraction which determines the relative population of a specific rotational state
of OH accessible to the laser excitation transition. Depletion of the excited state, 𝑁 , is

36
controlled by spontaneous emission, 𝐴 , collisional quenching, 𝑄, and stimulated
emission, 𝑏 .
A steady-state approximation is valid when the laser pulse duration is longer than
the time required for the population of both states to reach steady state. Since in the present
work, steady state is obtained on a time scale of ~1 ns, which is shorter than the laser pulse
duration of ~10 ns, the use of the steady-state approximation is valid [59]. The relationship
between the ground state and the excited state number densities becomes
[𝑁 ] = (2.4)
[ ]
OH LIF has two limiting signal regimes, the linear regime (weak excitation) and
the saturated regime (strong excitation). If the laser energy is kept sufficiently low (<10
µJ for the present work), the signal increases linearly with the laser energy, and the
depletion mechanisms, fluorescence quenching and spontaneous emission, are much faster
than the stimulated emission rate. As the saturation threshold is approached, the
spontaneous emission can no longer be ignored, resulting in LIF signal nearly independent
of the laser energy. Saturated LIF has been used previously to circumvent the uncertainty
in the quenching rate coefficients. However, high laser energy may also lead to photo-
dissociation, increasing the measurement uncertainty [59]. By maintaining low laser
energy, it can also be assumed that most OH molecules are in the ground electronic state,
and that only a small fraction is excited by the laser photons, leading to the assumption that
𝑁 ≈𝑁 . With these assumptions, Eqn. 2.4 becomes:
[𝑁 ] = (2.5)
( )
37
Substituting Equation 2.5 into Equation 2.1, the relationship between LIF signal,
𝑆 , and OH number density is obtained
𝑆 =𝑁 𝑓 𝑉 𝜂𝐴 (2.6)
( )
Equation 2.6 has both the fluorescence factor, 𝜂𝐴 , and the absorption
( )
factor, 𝑁 𝑓 𝑏 𝑉. First, consider the fluorescence factor. The spontaneous emission
coefficients, 𝐴 , are tabulated for all rotational and vibrational levels of OH in LIFBASE,
a database of spectroscopic constants of diatomic molecules [60]. When the laser excites
OH molecules at a specific ground state rotational level to an excited state, rotational
equilibrium is perturbed [61]. Rotational energy transfer (RET) between the rotational
states in the vibrational level of the excited state re-establishes an equilibrium rotational
distribution. This results in the spontaneous emission rates being weighted by their
rotational Boltzmann fractions in the upper electronic state, 𝐴 = ∑ 𝐴 𝑓 , . The total
spontaneous emission rate depends on the fluorescence quantum efficiency, defined as
𝜑 = . (2.7)
( )
Collisional quenching, 𝑄, depopulates excited OH molecules back to the ground
electronic state before they fluoresce. Quenching is the dominant fluorescence decay
process (compared to spontaneous emission), and it can be either measured directly from
the time-resolved fluorescence decay rate as discussed in Chapter 4, or it can be calculated
based on previously measured quenching cross-sections and gas mixture composition.
Although direct quenching measurement considerably reduces the experimental
uncertainty, it is not always feasible while using nanosecond laser diagnostics, since it may
38
occur on a similar (ns) time scale. Therefore, the total quenching rate, 𝑄, is calculated as a
product of quenching rate coefficients, 𝑘 , and species number densities, 𝑛 , summed
over all quenching species present, as shown in Equation 2.8.
𝑄 = ∑𝑛 ∗𝑘 , (2.8)
The quenching rate coefficients depend on quenching cross sections, 𝜎 , given in
Ref. [62] for stable air and fuel species and H and OH radicals, and gas temperature, 𝑇, as
shown in Equation 2.9.
𝑘 (𝑐𝑚 𝑠 ) = 𝜎 ∗ 𝑣̅ = 𝜎 ∗ =𝜎 ∗ ∗ (2.9)
where 𝑇 is 300 K, and 𝜇 is the molecular weight.
Now considering the absorption factor, 𝑁 𝑓 𝑏 𝑉, the probe volume, 𝑉, can be
assumed to be the product of the laser beam cross sectional area, 𝐴, and the path length, 𝑙.
From Ref. [59], the absorption rate coefficient, 𝑏 , depends on the Einstein coefficient for
absorption, 𝐵, the speed of light, 𝑐, and the spectral intensity of the laser beam 𝐼 , such that
𝑏 = 𝐼 (2.10)
Equation 2.10 is valid only if the laser spectral intensity is independent of the
wavelength and the absorption line is infinitely narrow. This is not the case, since the laser
has a finite lineshape, 𝑔 (𝜈 ), and the absorption transition has a finite lineshape,
𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈), defined by the centerline frequency, 𝜈 , and the line broadening
factors including Doppler broadening, ∆𝜈 , and pressure broadening, ∆𝜈 . Therefore, the
absorption rate coefficient is more accurately given as,
𝑏 = ∫ 𝐼 (𝜈 )𝑔 (𝜈 − 𝜈) ∗ 𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈) 𝑑𝜈 = 𝐼 (𝜈 )𝑔(𝜈 − 𝜈 ) (2.11)
39
where 𝐼 (𝜈 ) is the laser spectral intensity at the line center frequency, 𝜈 , the normalized
laser lineshape is defined as ∫ 𝑔 (𝜈 − 𝜈) 𝑑𝜈 = 1, and the normalized absorption
lineshape is defined as ∫ 𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈) 𝑑𝜈 = 1 [63]. The convolution of the laser
lineshape and the absorption transition lineshape is the convolution integral, 𝑔(𝜈 − 𝜈 ),
such that
𝑔(𝜈 − 𝜈 ) = ∫ 𝑔 (𝜈 − 𝜈) ∗ 𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈) 𝑑𝜈 (2.12)
In the present work, it is measured by scanning the laser lineshape relative to the
absorption transition lineshape, and measuring the fluorescence signal, as shown in Figure
2.2.
Figure 2.2: Schematic representation of the convolution integral, determined by tuning the
laser lineshape relative to the absorption transition lineshape, and measuring the
convolution signal.
40
Thus, the total laser intensity and the convolution integral are measured directly,
and the absorption coefficient, 𝑏′ = , for the excitation transition is tabulated in the
LIFBASE library [60].
The Boltzmann fraction, 𝑓 , is the relative population of the ground electronic and
vibrational state of the probe species molecules, accessible to laser excitation for a given
transition (the OH(X2Π  A2Σ) R2(2.5) transition in the present work). Due to coupling
between the rotation of the nuclei and the electronic motion (i.e. the total angular
momentum of the electrons, representing the sum of the orbital and spin angular momenta),
each rotational level in the ground and excited electronic states of OH(X2Π and A2Σ) is
split into two levels, 𝐹 and 𝐹 , depending on the total angular momentum of the electrons
[64], as shown in Figure 2.3. Additionally, in the ground electronic state, each energy level
is further split into two components due to coupling between the nuclear motion and the
orbital angular momentum of the electrons, known as Λ-type doubling, such that the
𝐹 level has the components 𝐹 and 𝐹 . This results in a total of four energy states for
each rotational level. The energies of all four levels are tabulated in Ref. [65]. The energy
difference between 𝐹 and 𝐹 states is sufficiently large, such that that the Boltzmann
factor needs to be calculated only for one of them. The energy difference between the Λ-
doublet states, however, is small enough such that the contribution from both components
needs to be taken into account.
41
Figure 2.3: Energy level diagram for OH. The absorption transition used in the present
work, X2Π  A2Σ R2(2.5), is highlighted in red [64].
42
The rotational Boltzmann fraction for the OH (X2Π) 𝐹 (2.5) level, is given in
Equation 2.13 below,
( . ) ) ( ( . ) )
( ) ( )
𝑓 = + (2.13)
∑( ) ∑( )
where 𝐽 is the rotational quantum number, 𝐸(2.5) and 𝐸(2.5) are the energies of the
Λ-doubling components of the 𝐹 (2.5) state, 𝐸 are the energies of all rotational quantum
states for the same vibrational level of the ground electronic state of OH, 𝑘 is the Boltzmann
constant, and 𝑇 is the gas temperature. By substituting Eqn. 2.7 and 2.10 into Eqn. 2.6, the
relationship between the LIF signal and the OH number density is given in Equation 2.14,
𝑆 = 𝑁 𝑏′ 𝜑 𝐴𝐼 𝑔(𝜈 − 𝜈 )𝑓 (2.14)
Although the signal collection parameters, 𝑙𝛽𝛺 , in Eqn. 2.14 can be measured in
principle, this is extremely challenging, and instead absolute calibration to determine the
signal collection factors via Rayleigh scattering is employed. Rayleigh scattering as a
calibration technique was discussed in detail in a previous study [13]. The Rayleigh
scattering signal is given as,
𝑆 = 𝑙𝛺𝛽 ∗ (𝑁 𝐸 ) = 𝐷 ∗ (𝑁 𝐸 ), (2.15)
𝐷 = 𝑙𝛺𝛽. (2.16)
For the calibration, the cell is filled with a calibration gas (in the present work
nitrogen with a calculated Rayleigh cross section of = 6.07∙10-27 cm2sr-1 [66], at 𝜈 =
308 nm). By monitoring the laser pulse power, 𝐸 , and varying the pressure in the cell, and
thereby varying the number density, 𝑁 , a linear relationship between Rayleigh signal
43
intensity, 𝑆 , and 𝑁 𝐸 is determined. The slope of the calibration line, 𝐷 , depends on
the Rayleigh cross section, the signal collection efficiency, 𝜖 , which is equal to 𝜂 in Eqn.
2.1 since the same signal collection optics and detector are used for OH LIF and Rayleigh
scattering, the speed of light, 𝑐, Plank constant, ℎ, the laser frequency, 𝜈 , and the optical
collection parameters, 𝑙𝛺𝛽. Dividing Eqn. 2.14 by Eqn. 2.15, a relationship between the
LIF signal and the Rayleigh scattering signal is obtained,
( )
𝑆 = 𝐷 . (2.17)
This procedure implies that if the collection optics are not changed between the LIF
measurements and Rayleigh scattering calibration, the signal collection factors cancel.
Finally, from Eqn. 2.17, the relationship between OH number density and the LIF signal is
obtained,
𝑛 = ( )
, (2.18)
where the OH number density, 𝑛 , depends on previously defined terms including the
LIF signal, 𝑆 , the Rayleigh cross section, , the absorption coefficient, 𝑏′ , the
fluorescence quantum efficiency, 𝜑 , the measured laser power, 𝐴𝐼 , the convolution
integral, 𝑔(𝜈 − 𝜈 ), the Boltzmann fraction, 𝑓 , Planck constant, ℎ, the speed of light, 𝑐,
the frequency of the laser used for Rayleigh scattering, 𝜈 , and the slope of the
calibration line, 𝐷 . This equation has been used in the [OH] data reduction procedure in
the present work.
44
2.2. Two-Photon Absorption Laser Induced Fluorescence
This work presents atomic oxygen and hydrogen concentration measurements
using Two-photon Laser Induced Fluorescence (TALIF). This method is used for
excitation of species where single photon absorption would require laser photons in the
vacuum ultra-violet [59]. It employs an approach similar to LIF: simultaneous absorption
of two laser photons from the atomic ground state to an excited state, followed by single-
photon fluorescence, which is detected and quantified. Due to the two-photon absorption
cross section being much lower compared to that of the single photon absorption, TALIF
is much less efficient than LIF, resulting in a weaker signal.
To process and quantify the TALIF signal, the governing equations must be
analyzed and all assumptions used in the equations validated. Figure 2.4 provides a three-
state energy level diagram, outlining transitions for an excited atom in the N 1 state
populated by the two-photon absorption process.
45
Figure 2.4: Three level energy diagram for two-photon laser excitation, showing dominant
energy transfer processes.
The rate equations for the ground state, 𝑁 , and the excited state, 𝑁 , of the atom,
as well as the atomic ion state, 𝑁 ∗ , populations are provided in Equations 2.19-2.21.
∗( )
= 𝑁 (𝑡)𝑅 ∗ (2.19)
( )
= 𝑊 (𝑡)𝑁 (𝑡) − 𝑊 (𝑡)𝑁 (𝑡) − 𝑁 (𝑡)[ 𝑅 ∗ + 𝑄 + 𝐴] (2.20)
( )
= 𝑊 (𝑡)𝑁 (𝑡) − 𝑊 (𝑡)𝑁 (𝑡) + 𝑁 (𝑡)(𝑄 + 𝐴) (2.21)
In Eqns. 2.19-2.21, the excited state, 𝑁 , is populated by two-photon absorption by
( )
the ground state atoms, 𝑊 ≈ 𝜎 𝐼
ℎ𝜈 , where 𝑊 depends on 𝜎 ( ) , the two-photon
absorption cross section from the ground state to the excited state, 𝐼, the laser intensity, 𝜈,
the transition frequency, and ℎ, the Planck constant. The excited 𝑁 state is depopulated by
ionization following the absorption of a third photon, controlled by the ionization rate,
46
𝜎 𝐼
𝑅∗ ≈ ℎ𝜈 , which depends on the laser intensity and frequency, as well as on the
ionization cross section, 𝜎 . In addition, 𝑁 depopulation is controlled by the stimulated
( )
emission, 𝑊 ≈𝜎 𝐼
ℎ𝜈 , where 𝜎
( )
is to the two-photon stimulated emission cross
section, by collisional quenching, 𝑄, and spontaneous emission, 𝐴, as illustrated in Fig.
2.4. Similar to LIF, the net collisional quenching rate depends on the quenching rate
coefficients, 𝑘 , and number densities of the quenching species, 𝑛 , see Eqn. 2.8.
Using the steady-state approximation, Equations 2.19, 2.20, and 2.21 can be solved
for the concentration of the excited state 𝑁 .
[𝑁 ] = ∗
(2.22)
[ ]
Equation 2.23 shows the explicit dependence of the excited state number density
on the absorption and ionization cross sections, as well as quenching rate coefficients and
spontaneous emission.
( )
[𝑁 ] = ( )
(2.23)
The threshold for photo-ionization and stimulated emission must be determined by
monitoring the fluorescence signal versus the laser intensity. At sufficiently low laser
intensity, the fluorescence signal must vary quadratically with the laser pulse energy, in
which case photoionization and stimulated emission may be neglected, leaving only
quenching and spontaneous emission to depopulate the 𝑁 state. At higher laser pulse
intensity, all depletion terms must be considered, resulting in a non-quadratic dependence
of fluorescence signal on the laser pulse energy. For all TALIF measurements in the
47
present work, the laser pulse energy was kept within the quadratic regime, such that photo-
ionization and stimulated emission were negligible. With this assumption, a simplified
steady state equation for 𝑁 is obtained.

( )
[𝑁 ] = = (2.24)
( )
This equation assumes the relative population of the excited state is very low, such
that 𝑁 is very close to the total number density of the probe atomic species.
From Ref. [59], the expression for the excitation rate, 𝑊 , is given by Equation
2.24, which includes the photon statistical factor, 𝐺 ( )

= 2 [67], and the convolution
integral, 𝑔(𝜈 − 2𝜈 ).
𝑊 = 𝐺 ( ) 𝜎 ( ) 𝑔(𝜈 − 2𝜈 )( ) (2.25)
𝑔(𝜈 − 2𝜈 ) = ∫ 𝑔(𝜈 − 2𝜈) ∗ 𝑔 (𝜈 − 𝜈) ∗ 𝑔 (𝜈 − 𝜈)𝑑𝜈 (2.26)
In Equation 2.26, the convolution integral for TALIF is the convolved lineshape of
the two-photon absorption transition lineshape, 𝑔(𝜈 − 2𝜈), and the laser lineshape
contributed by both photons, 𝑔 (𝜈 − 𝜈). It is determined by scanning the laser frequency
across the transition and measuring the resulting fluorescence signal.
From Ref. [59], the two-photon fluorescence signal is proportional to the number
density of the excited probe species and the fluorescence decay rate, illustrated in Equation
2.27,
𝑆 =𝑉 𝜂𝑁 𝑎 𝑊 . (2.27)
48
In Equation 2.27, 𝑉 𝜂 is the product of the excited volume and optical collection
efficiency parameters, and 𝑎 is the fluorescence quantum yield, 𝑎 = . Taking into

( )
account Eqn. 2.27, the relationship between the fluorescence signal, 𝑆 , and the species
number density is obtained,
𝑆 =𝑉 𝜂𝑁 𝑎 𝑓 (𝑇)𝐺 ( ) 𝜎 ( ) 𝑔(𝜈 − 2𝜈 )( ) . (2.28)
Knowledge of quenching rates is critical for accurate calculation of the fluorescence
yield. Using nanosecond pulse laser diagnostics does not allow resolving of the quenching
rates at high pressures, especially in the presence of fast quenchers, such as oxygen
molecules. A compilation of the ratios of the two-photon absorption cross sections for the
probe and calibration species (Xe and O atoms, and Kr and H atoms, respectively), natural
lifetimes, and quenching rates used in this study is presented in Table 2.1, with references
to the original work. Absolute calibration of TALIF measurements is discussed below.
Table 2.1: Two-photon absorption cross section ratios, natural lifetimes, and collisional
quenching rates used in H, Kr, O, and Xe TALIF measurements.
O (3p 3P1,2,0) Xe(6p’[3/2]2) H (3d 2D 3/2.5/2) Kr(5p’ [3/2]2)

𝜎 ( ) /𝜎 ( ) Xe/O = 1.9 [68] Kr/H = 0.62 [69]
[cm4/cm4]
𝜏 [ns] 35.4±1.4 [11] 40.0±1.6 [68] 16.7±0.7 [11] 33.6±1.1 [11]
O2 8.6±0.2 [11]
9.4±0.5 [69]
𝑘 Xe 3.7±0.14 [11]
-103 -1
[10 cm s ] 3.6±0.4 [68]
Ar 0.14±0.07[69] 4.15±0.03 [11]
H2 19.9±3 [11]
Kr 1.31±0.065
[11]
49
Placing the TALIF signal on an absolute scale requires knowledge of the signal
collection parameters, 𝑉 𝜂, which are obtained from calibration, as discussed below. As
demonstrated in Ref. [69], comparison between the fluorescence signal from atomic
species generated in the plasma to the fluorescence signal from a noble gas cancels out the
signal collection efficiency factors if the same detector and signal collection optics are
used. The noble gas used for calibration must have similar excitation and fluorescence
wavelengths, reducing the variance in the signal collection parameters. Figure 2.5 shows
energy level diagrams for both atomic species (H and O) and the respective noble gases
(krypton and xenon) used for their calibration.
50
Figure 2.5: Two-photon absorption energy level diagrams for (a) hydrogen and krypton
and (b) oxygen and xenon [8].
In the present work, parameters of the collection optics that have wavelength-
dependent responses, such as photomultiplier tubes (PMTs) and bandpass filters, have been
51
quantified. The PMT (Hamamatsu R3896) spectral response is measured using a calibrated
blackbody source (InfraRed Industries Inc. Model 563) at T=900 K, and a spectrometer,
acting as a tunable filter. The spectral response is calculated from the blackbody signal,
recorded by an oscilloscope, divided by the Plank distribution at T = 900 K, and plotted in
Figure 2.6. The signal at the four wavelengths of interest (656, 826, 844, and 834 nm), used
for H, Kr, O and Xe TALIF, respectively, was recorded to compare TALIF signal in
calibration gases (Xe, Kr) with TALIF signal in O2-Ar and H2-Ar plasmas.
Figure 2.6: Hamamatsu R3896 photomultiplier tube response curve measured using a
calibrated blackbody source at T=900 K. A spectrometer placed before PMT is used as a
tunable filter. The PMT is used for H, O, Kr, and Xe TALIF measurements. The response
curve is used to compare TALIF signal in calibration gases (Xe, Kr) with TALIF signal in
O2-Ar and H2-Ar plasmas.
The transmission of the bandpass filters is measured using a broadband xenon lamp
light source and an Ocean Optics Spectrometer (USB6000). To illustrate this, the output
spectrum of the broadband light source measured by the spectrometer, as well as the
52
spectrum transmitted by the FB650-40 filter, are plotted in Figure 2.7. The ratio of the two
spectra yields the filter transmission curve, also plotted in Fig. 2.7. In H TALIF data
analysis, a value of 74% at 656 nm has been used.
Figure 2.7: ThorLabs FS650-40 filter transmission curve measured using a broadband light
source and spectrometer. Spectra are measured without (red line) and with (blue line) the
filter. The intensity ratio between the two lines gives the transmission curve, also plotted
in the figure. The same procedure is used for other filters used in TALIF measurements
(see Table 2.2).
Similar traces were taken for every filter used to collect TALIF signal, and the
resulting transmission coefficients are listed in Table 2.2 below.
53
Table 2.2: Bandpass filter transmission factors used in H, Kr, O, and Xe TALIF
measurements.
Filter Type Wavelength (nm) Transmissivity (%)

FB650-40 656.3 74
830FS20 826.3 63
850-40 844.7 62
850-40 834.7 53
Following Ref. [69], the atomic oxygen concentration is calibrated using the 5p 6
1
S06p’[3/2]2 transition of xenon, as it has two-photon excitation and fluorescence
frequencies similar to that of atomic oxygen [69],
( ) ( ) ( )( )
𝑛 = ( ) ( ) ( )( )
( ) 𝑛 . (2.30)
( )
In Equation 2.30, the solid angle, the optical collection efficiency, and the photon
statistical factor cancel, leaving only the fluorescence signals, 𝑆 , the convolution integrals,
𝑔(𝑖), the fluorescence quantum yields, 𝑎 (𝑖), laser beam frequencies, 𝜈 , the two-photon
absorption cross-sections, 𝜎 ( ) (𝑖), the number density of xenon, 𝑛 , and the Boltzmann
fraction, 𝑓 (𝑇). The Boltzmann fraction is included to account for the triplet splitting of
the ground state of atomic oxygen, 2p4 3PJ, which has sufficiently large energy separation
between the three energy levels J=0, 1, and 2, to allow each state to be excited separately
as shown in Figure 2.8 [70].
54
Figure 2.8: Energy level diagram for atomic oxygen absorption transitions, with relative
transition probabilities indicated [70]. The transitions used in the present work are labeled.
Accurate knowledge of temperature is critical for calculating the Boltzmann
fraction in Eqn. 2.30. In addition, the excited state of atomic oxygen is also a triplet,
however the energy spacing between the triplet components is small, such that fluorescence
from individual states is difficult to isolate.
TALIF measurements of atomic hydrogen concentration rely on similar calibration
using the 4p6 1So  5p’[3/2]2 transition of krypton, as shown in Figure 2.5. Equation 2.31
below, relating H atom number density to the known krypton number density, is similar to
Equation 2.30, except that the Boltzmann fraction for the singlet ground state of H atoms
is equal to unity, since there is no energy level splitting,
55
( ) ( ) ( )( )
𝑛 = ( ) ( ) ( )( )
( ) 𝑛 . (2.31)
Equations 2.30 and 2.31 have been used in the [O] and [H] data reduction
procedures in the present work.
2.3. Tunable Diode Laser Absorption Spectroscopy
Finally, in the present work, metastable argon atom number density is measured
using Tunable Diode Laser Absorption Spectroscopy (TDLAS). This technique is based
on Beer-Lambert’s law of absorption, relating how the light spectral intensity, 𝐼 , changes
along the optical path of an absorbing medium,
( , )
= −𝑘(𝜈, 𝑥)𝐼 (𝜈, 𝑥) (2.32)
In Equation 2.32, 𝜈 is the laser frequency, 𝐼 is the incident light intensity, and
𝑘(𝜈, 𝑥) is the effective spectral absorption coefficient, which is the difference between the
absorption and the stimulated emission coefficients of an absorbing medium. Using the
expressions for the Einstein absorption and stimulated emission coefficients, the absorption
coefficient is determined by Equation 2.33 [71],
𝑘(𝜈) = ∗ [𝜂(𝑙) − 𝜂(𝑢)]𝛿(𝑢)𝐴 𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈) (2.33)
where 𝜆 is the transition wavelength and 𝐴 is the Einstein coefficient for spontaneous
emission, 𝑙 and 𝑢 denote the lower and upper states, 𝜂(𝑙) = is the lower quantum state
()
number density divided by the degeneracy (statistical weight), 𝛿( 𝑙) of the lower state, and
𝜂(𝑢) = is the upper quantum state number density, divided by its degeneracy
( )
(statistical weight), 𝛿( 𝑙). Assuming that only a small fraction of the atoms are excited to
56
the upper state, such that the population of the lower state is much greater than the
( )
population of the upper state, [𝜂(𝑙) − 𝜂(𝑢)]𝛿(𝑢) becomes 𝑛(𝑙) ()
. The absorption
lineshape factor is 𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈), where 𝜈 is the centerline frequency, ∆𝜈 is the
Doppler broadening factor, and ∆𝜈 is the pressure broadening factor. Assuming that
stimulated emission is negligible, the absorption coefficient becomes,
( )
𝑘(𝜈) = ∗ 𝑛(𝑙) ()
𝐴 𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈) (2.34)
The TDLAS measurement is a line of sight measurement, incorporating integration
along the optical path, from x=0 to x=L, where L is the length of the absorbing medium,
such as a plasma. In addition, the transmitted signal is spectrally resolved by scanning the
laser frequency across the absorption transition.
( , ) ( )
ln
( , )
= ()
𝐴 ∫ ∫ 𝑛(𝑙, 𝑥)𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈)𝑔 (𝜈 − 𝜈) 𝑑𝑥𝑑𝜈 (2.35)
where 𝐿 is the path length of the absorbing medium, 𝑔 (𝜈 − 𝜈), is the laser lineshape with
laser frequency 𝜈 , and 𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈) is the absorption lineshape which has a central
frequency, 𝜈 , and is dependent on different broadening factors, ∆𝜈 and ∆𝜈 . The natural
log of the intensity ratio is usually referred to as spectral absorbance, and is denoted here
by 𝑆 (𝜈 ).
The lowest energy argon metastable state Ar(3p54s) has four different energy
levels, while the next excited state, Ar(3p54p), has ten energy levels, as shown in Figure
2.7 [72].
57
Figure 2.9: Energy level diagram for Ar atoms showing Ar(3p54s) and Ar(3p54p) states,
using Paschen notations. The transition used in the present work is 1s 52p7, indicated by
a red arrow [72].
A narrowband continuous wave (cw) diode laser is used to access two transitions
between the 4s and 4p states, 1s52p7 (indicated by the arrow in Fig. 2.7) and 1s32p2.
In the present work, the laser was parked at the center of 1s 52p7 transition in the 3p54s
state. The laser linewidth is much narrower than the absorption transition linewidth (<300
kHz compared to ~3 GHz), such that the convolution integral between the laser lineshape
and the absorption lineshape, 𝑔(𝜈 − 𝜈 ) = ∫ 𝑔 (𝜈 , ∆𝜈 , ∆𝜈 , 𝜈)𝑔 (𝜈 − 𝜈) 𝑑𝜈, is
dominated by absorption line broadening, and is dependent on three main types of line
broadening and their associated lineshapes, taken from Ref. [71].
58
First is collisional (or pressure) broadening caused by the interruption (dephasing)
of a continuous emission process by collisions. It produces a Lorentzian lineshape,
𝑔(𝜈, ∆𝜈 ),
∆
𝑔(𝜈, ∆𝜈 ) = ∆
(2.36)
∗ ( )
centered at a frequency 𝜈 , with ∆𝜈 , the full width half maximum (FWHM) of the
collisional broadening lineshape for the Ar(1s5) state given in Ref. [71].
.
∆𝜈 ( ) [𝐻𝑧] = 𝐾 ∗ 𝑃 ∗ (2.37)
From Ref. [71], 𝐾 = 1.3 ∙ 10 [𝐻𝑧] and for the present experimental conditions (P
= 300 Torr and T0= 500 K), the estimated FWHM collisional broadening is ∆𝜈 ≈ 2.7 GHz.
Second is Doppler broadening, which is a frequency shift due to the result of
random thermal motion of atoms. It produces a Gaussian Lineshape, 𝑔(𝜈, ∆𝜈 ),
( ) ( )
𝑔(𝜈, ∆𝜈 ) = exp − (𝜈 − 𝜈 ) (2.38)
√ ∆ ∆
centered at 𝜈 , with a FWHM, ∆𝜈 , of
( ) . .
∆𝜈 [𝐻𝑧] = 2 ∗ ∗𝜈 (2.39)
where 𝑇 is the gas temperature, 𝑀 is the atomic mass, 𝑘 is the Boltzmann constant, and 𝑐
is the speed of light. At the present experimental conditions, the Doppler FWHM is ∆𝜈 ≈
0.8 GHz.
A third type of broadening is natural broadening, which is the result of a finite
radiative lifetime. It produces a Lorentzian lineshape, given in Eqn. 2.36, with a FWHM
∆𝜈 = (2.40)
59
For the Ar(1s5) state, 𝐴 = 1 𝜏 , τ = 5.18∙106 s-1 [71], leading to a ∆𝜈 = 0.824
MHz. This is much narrower compared to pressure broadening (0.824 MHz << 2.7 GHz)
and is therefore negligible.
Combination of the dominant broadening mechanisms, collisional and Doppler
broadening, result in a Voigt spectral lineshape profile, which is the convolution of
Lorentzian and Gaussian lineshapes:
𝑔(𝜈 − 𝜈 , ∆𝜈 , ∆𝜈 ) = ∫ 𝑔(𝑢, ∆𝜈 ) ∗ 𝑔(𝜈 − 𝑢, ∆𝜈 )𝑑𝑢 (2.41)
Figure 2.10 plots several synthetic Voight profiles calculated using Equations 2.36,
2.38, and 2.41 at the present experimental conditions. It can be seen that pressure increase
from 300 Torr to 700 Torr results in significant broadening of the profile. On the other
hand, temperature increase from 300 K to 500 K results in a reduction in the collisional
broadening and an increase in Doppler broadening. Since the collisional broadening
reduction is greater, the lineshape profile narrows.
60
Figure 2.10: Theoretical Voigt absorption line profiles for the Ar(3p 54s) 1s5 Ar(3p54p)
2p7 transition, illustrating the combined effect of Doppler broadening and pressure
broadening at 300 K vs. 500 K at P = 300 Torr and pressure broadening alone at 300 Torr
vs. 700 Torr at T = 500 K.
Rearranging the terms of Equation 2.35, the equation for the concentration of the
absorbing species, 𝑛(𝑙), is derived,
()
𝑛(𝑙) = 𝑆 (𝜈 = 𝜈 ) ( )
, (2.42)
( ,∆ ,∆ )
where 𝑆 (𝜈 = 𝜈 ) is the absorbance signal taken at the peak of the absorption
transition, when the laser frequency, 𝜈 is equal to the centerline transition frequency, 𝜈 .
()
In addition, 𝜆, is the laser wavelength, ( )
is the ratio of upper to lower state
degeneracies, 𝐴 is the Einstein absorption coefficient, 𝐿 is the path length, and
𝑔(𝜈 − 𝜈 , ∆𝜈 , ∆𝜈 ) is the calculated Voight profile. Equation 2.42 has been used to
measure absolute number population of the Ar(1s 5) state in the present work.
61
Chapter 3: Measurements of metastable and radical species
in repetitive ns pulse discharge plasmas in fuel-oxidizer
mixtures
The focus of this work is to study plasma assisted fuel oxidation kinetics in a flow
reactor excited by a ns pulse electric discharge, operated at elevated temperatures (T= 500-
600 K) and pressures of 0.4-0.9 bar. Plasma uniformity in the reactor at these conditions is
quantified by mapping the spatial distribution of metastable argon atoms produced by the
plasma, using Tunable Diode Laser Absorption Spectroscopy (TDLAS). Temperature rise
in the discharge-excited reacting flow is measured by Rayleigh scattering. The objective
of this work is to obtain a set of experimental data on absolute, time-resolved number
densities of key radical species generated in repetitive nanosecond pulse discharges in lean
fuel / oxidizer mixtures, primarily H and O atoms, by Two-photon Absorption Laser
Induced Fluorescence (TALIF), and to compare the results with kinetic modeling
predictions. These data provide quantitative insight into rates and product distributions of
reactions of fuel species with plasma electrons and excited molecules and atoms, as well
as into plasma-generated radical species reactions at low temperatures (below ignition
temperature) and high pressures (approaching 1 bar).
62
3.1. Plasma flow reactor design
Measurements of H and O atoms, known to be critical in low-temperature plasma
assisted fuel oxidation and ignition kinetics, are conducted in two different plasma flow
reactors designed to sustain a volumetric, near 0-D plasma in reacting fuel-oxidizer
mixtures at pressures up to ~ 1 bar. Reactor Cell #1, shown in Figure 3.1, is based on a
previous plasma flow reactor design for plasma assisted combustion kinetics experiments
at elevated temperatures [6], with changes made to improve plasma uniformity at high
pressures and reduce the uncertainty in the energy coupled to the plasma. The flow reactor
/ electric discharge cell, approximately 50 cm long, is made of quartz, with 25.4 mm
diameter quartz-to-stainless-steel adapters and stainless-steel flanges at the ends. Stainless
steel arms with UV fused silica windows at Brewster’s angle are attached to the flanges at
the ends of the reactor, to provide optical access for laser diagnostics. A premixed flow of
reactants enters the central part of the reactor after passing through two preheating coils
approximately 1 m long, shown in Fig. 3.1(a), to maintain the reactant mixture temperature
at the desired level. A photograph of the central part of the reactor, with the preheating
coils, is shown in Figure 3.1(b).
63
with liquid metal electrodes encapsulated in quartz reservoirs (Cell #1) to generate a
volumetric, double dielectric barrier electric discharge.
The central part of the reactor is a rectangular cross section flow channel 20 cm
long with a cross section of 15 mm by 5 mm, fused at both ends to two circular tubes 25.4
mm in diameter. Two quartz reservoirs, 6 cm long, 1.5 cm wide, and approximately 0.5 cm
height each, are fused to the top and the bottom walls of the channel, as shown in Fig. 3.1.
The reservoirs are filled with liquid metal alloy (Ga-In-Sn) through two quartz tubes 20 cm
long and 1 mm in diameter (inside diameter approximately 0.5 mm), extending along the
flow channel, as shown in Fig. 3.1. The two reservoirs and tubes, filled with liquid metal
64
without air bubbles, are used as electrodes for the electric discharge sustained in the flow
reactor.
Reactor Cell #2 is shown schematically in Figure 3.2(a). It incorporates a
rectangular cross section central channel with preheating coils, all made of quartz, used in
previous time-resolved OH measurements in a ns pulse discharge in preheated fuel-air
mixtures [6]. The inner quartz channel is 20 cm long and has a rectangular cross section of
22 mm by 10 mm, with a wall thickness of 1.8 mm. The central channel is fused on each
side to circular quartz tube sections, 25.4 mm in diameter and 20 cm long, with 1 mm thick
planar quartz windows fused to the channel ends at the same Brewster’s angle as in Cell
#1.
65
with copper electrodes encapsulated in a resilient dielectric material (PDMS), to generate
a volumetric, double dielectric barrier electric discharge
The electrodes are two copper plates machined to be 4 mm thick, 14 mm wide, and
6 cm long, with rounded edges, held by ceramic clamps on the top and bottom of the quartz
channel, as shown in Fig. 3.2(b), and separated by two resilient dielectric pads ~4 mm thick
each. The total distance between the electrodes is approximately 23 mm. The dielectric
pads are made from a high-temperature resilient material (Polydimethylsiloxane or
PDMS). The flow reactor (Cell #1 or Cell #2) is placed in a tube furnace (Thermacraft,
with a 6” diameter, 6″ long heated section, maximum temperature 1200 0 C) and preheated
66
to a temperature of 500 K to improve plasma stability. Preheating coils, shown in Figures
3.1 and 3.2, have a total preheating distance of 10 meters before the reactants enter the test
section.
In our previous work [5], encapsulating copper electrodes in self-hardening
resilient dielectric coating, such as silicone rubber, prevented corona formation outside of
the discharge cell at room temperature, but this cannot be done at temperatures exceeding
~ 2000 C. Encapsulating the electrodes using hard dielectric coatings was found to be
ineffective, due to significant difference between thermal expansion coefficients of copper
and dielectric materials, resulting in cracking of the dielectric or the quartz channel, and
corona discharge formation. To prevent this, a dielectric material that remains resilient to
temperatures of at least 500 K was used. PDMS is an inert, non-toxic, and non-flammable
elastomer material commonly used in microfluidics [73]. The standard mixture of 10:1
elastomer base to curing agent produces a soft, gel-like material after curing for 48 hours,
which does not withstand temperatures above 100°C. At temperatures above 100°C, the
material begins to melt. Varying the mixture ratio improved the material stability at higher
temperatures [73], but excessive reduction of the amount of elastomer curing agent in the
mixture produces a brittle material. An optimum mixture was found to be a 5:1 ratio of
base to curing agent by mass, with an additive of silicon dioxide (quartz) powder to enhance
the dielectric strength of the material and to withstand high voltage. To prevent corona
discharge formation outside the flow reactor, the copper electrodes were coated in PDMS,
and two PDMS sheets cured for 48 hours at room temperature were placed between the
electrodes and the outside channel walls, as shown in Fig. 3.2(a). PDMS tolerance to heat
67
was tested by heating the flow reactor with the electrodes to a temperature of 400 K for
two hours, before cooling and checking for cracks. The process was then repeated at a
temperature of 500 K. No significant change was detected in the structure or flexibility of
the pads, indicating that the material was resilient at this temperature.
The use of encapsulated liquid metal electrodes (in Cell #1) or PDMS dielectric
pads (in Cell #2) prevents corona discharge formation at the electrode surfaces, which
occurred in our previous work [5,6], producing significant EMI noise and introducing
significant uncertainty in the energy coupled to the plasma inside the channel.
H and O atom TALIF measurements were conducted in diluted H2-O2-Ar mixtures.
Additional O TALIF measurements were made in dilute C xHy-O2-Ar mixtures. The flow
rates of the mixture components, premixed 3 meters before entering the reactor, are
metered by MKS mass flow controllers (<0.4% error at the present conditions, according
to the manufacturer’s specifications). The reactant mixture enters the reactor channel 9 cm
upstream of the electrodes, after passing through the preheating coils, and flows between
the electrodes before exiting the channel 4 cm downstream of the electrodes, as shown in
Figs. 3.1(a) and 3.2(a). The flow rate used in the H TALIF experiments (using Cell #1) is
3.6 slm, which corresponds to flow residence time in the plasma of approximately 7 ms at
P=300 Torr and T=500 K. For the O TALIF experiments (using Cell #2), the flow rate is
varied from 2.0 to 3.4 slm, corresponding a flow residence time of 31 to 18 ms in the
plasma. Thus, at these flow rates, the flow residence time in the plasma is typically much
longer compared to the discharge burst duration of 3.7-5 ms (see Section 3) and to the delay
time between the discharge burst and the laser pulse.
68
3.2. Laser Diagnostics Setup
Spatial uniformity of the plasma generated by high voltage repetitive nanosecond
pulses in the reactor channel is monitored by a PI-MAX ICCD camera with a UV lens
(UV-Nikkor 105 mm f/4.5, Nikon). In addition, plasma uniformity is quantified using
spatially resolved, line-of-sight averaged measurements of metastable Ar atom number
density distributions, using TDLAS. For this, a cw diode laser beam (New Focus Model
6312, linewidth < 0.01 cm-1), producing tunable output from 765 nm to 781 nm, was
directed axially through the reactor, as shown in Fig. 3.3. This diode laser is capable of
accessing two absorption transitions of Ar atoms (1s 52p7 at 772.38 nm and 1s32p2 at
772.42 nm), as shown in Fig. 2.9. The laser was mounted on a translation stage, such that
the laser beam could be positioned at multiple locations over the reactor channel cross
section.
Figure 3.3: Schematic of TDLAS diagnostics. A continuous wave (cw), tunable

narrowband diode laser is mounted on a translation stage. A photodiode is mounted on
another translation stage, after the cell. Micrometric screws are used to make translation
stage adjustments.
69
Figure 3.4: Schematic of H TALIF diagnostics. A nanosecond Nd:YAG laser pumps a
tunable dye laser to generate output at 615 nm, frequency doubled using a Type I BBO
crystal, and mixed with the dye laser output to generate a 205 nm beam.
The schematic of H TALIF laser diagnostics is shown in Figure 3.4. For this, a
Nd:YAG (Continuum Model Powerlite 8010) pumped tunable dye laser (Continuum
ND6000) output at 615 nm is frequency-doubled in a type-I BBO crystal. The frequency
doubled output at 308 nm is recombined with the fundamental to be frequency-mixed again
in another type-I BBO crystal to produce a frequency-tripled output of 205 nm. The
linewidth of the 205 nm beam is approximately 0.8 cm-1, which is slightly more than triple
the dye laser linewidth (measured to ~0.2 cm-1 by a High Finesse, Angstrom WS/6
wavemeter). A Pellin-Broca prism separates the three beams, and an identical prism is set
symmetrically, to compensate for alignment changes of 308 nm and 205 nm beams. The
308 nm beam is blocked prior to entering the test cell, and the 205 nm beams is focused
through the flow reactor by a f=500 mm lens, setting the focal point ~3 cm before the test
section, which allows the laser to be at an intermediate focus field as it passes through the
test section. A band pass filter is placed in front of either an ICCD camera lens or a
70
Hamamatsu R3896 photomultiplier tube (PMT), to isolate the H atom fluorescence signal
at ~656 nm from the plasma emission.
Figure 3.5: Schematic of O TALIF diagnostics. The 619 nm beam generated by the dye
laser pumped by the second harmonic of the Nd:YAG laser, is frequency mixed with the
third harmonic output of the Nd:YAG laser (355 nm beam) using a Type I BBO crystal.
The three beams are passed through an iris and a UV harmonic separator (denoted as PR
in the diagram) to allow only the 226 nm beam to pass.
The schematic of O TALIF laser diagnostics, shown in Figure 3.5, illustrates how
the same Nd:YAG and tunable dye lasers are used to generate the 226 nm beam necessary
to excite O atoms by two-photon absorption. The second harmonic output from the
Nd:YAG laser at 532 nm pumps the dye laser to generate a 619 nm output, which is sum
frequency mixed in a Type I BBO crystal with the third harmonic output from the Nd:YAG
at 355 nm. The Type I BBO crystal is held in an Inrad Autotracker III, allowing precise
tuning of the crystal angle to optimize mixing. The three overlapping beams (355, 619,
and 226 nm) are then passed through an iris before entering an Inrad UV Harmonic
71
Separator to disperse the beams and allow only the 226 nm beam to exit. The linewidth of
the 226 nm beam is ~1 cm-1 and is limited by the non-injection seeded third harmonic
output of the Nd:YAG laser (~1 cm-1). This beam is then focused into the test section with
the same f= 500 mm lens. This prevents nonlinear effects caused by the focused laser
beam, such as photo dissociation and photoionization to affect the measurements. The same
PMT was used to take time-resolved O atom fluorescence measurements. A band pass filter
was placed in front of the PMT to collect only the signal from the O atom 3p3s emission
transition at ~844 nm.
The Nd:YAG laser used to pump the dye laser is not injection seeded. Energies of
H and O TALIF interrogation beams were monitored using a photodiode placed before the
test cell optics, as shown in Figs. 3.4 and 3.5. As discussed in Section 2.2, the TALIF signal
depends quadratically on the laser pulse energy, when the laser pulse energy is sufficiently
low. At the present conditions, measurements of fluorescence signal versus laser pulse
energy squared are shown in Figure 3.6, demonstrating that the quadratic dependence
region ranges approximately up to laser pulse energies of 40 µJ (1600 µJ 2). In all present
experiments, laser pulse energy was kept at or below 35 µJ to prevent photo-ionization or
photo-dissociation that would affect the measurement results [74,75].
72
Figure 3.6: Fluorescence signal versus squared laser pulse energy for (a) H TALIF
measured in a 1% H2-Ar mixture excited by a ns pulse discharge burst of 50 pulses with a
pulse repetition rate of 10 kHz, and (b) O TALIF, measured in a 1% O2-Ar mixture excited
by a ns pulse discharge burst of 75 pulses with a pulse repetition rate of 20 kHz at P = 300
Torr and initial temperature of T = 500 K.
Flow temperature in the reactor after the discharge burst was measured by Rayleigh
scattering at 226 nm, using the linear relationship between the scattering signal intensity
and the number density,
𝑇= 𝑇 ∗ (3.1)
In Equation 3.1, the gas mixture temperature, 𝑇, is determined by the ratio of
Rayleigh scattering signal at a known laser pulse energy, , taken at a reference
temperature, (𝑇 = 500 K, for these studies), to the Rayleigh scattering signal at a
measured laser pulse energy, , taken in the plasma. This relationship relies on a
73
constant pressure assumption, which is accurate since the temperature rise is measured
after a burst of 25-50 discharge pulses, 1.3 – 5 ms long. Each discharge pulse is estimated
to heat the flow by only a few degrees K, and the overall gradual temperature rise represents
a cumulative effect of energy coupled to the plasma by multiple pulses. Also, in the present
experiments with dilute fuel-oxidizer mixtures, ignition, which would generate a
significant pressure overshoot, does not occur. Although excitation of the reacting mixture
and radical species generation by the discharge changes its chemical composition, in the
present experiments argon makes up at least 97% of the gas mixture, both at the reference
conditions and the operating conditions. The Rayleigh cross-section for argon also has
weak temperature dependence, and does not need to be accounted for between the two sets
of conditions. Rayleigh scattering measurements are taken 10 µs and 200 μs after the last
pulse in the discharge burst.
3.3. Discharge Waveforms and Plasma Images
The electrodes of both flow reactor cells are connected to an FID GmbH FPG 60-
100MC4 pulse generator (peak voltage up to 30 kV, pulse duration of 5 ns, repetition rate
up to 100 kHz). In the present work, the pulser is operated in repetitive burst mode,
producing bursts of 25-75 pulses at a pulse repetition rate of 10-20 kHz (burst duration of
several ms), and burst repetition rate of 10 Hz. Powering the electrodes generates a double
dielectric barrier, repetitive nanosecond pulse electric discharge in the flow channel, with
the plasma volume of approximately 5 cm3 (Cell #1) or 10 cm3 (Cell #2). Pulse voltage and
current waveforms are measured by custom-made, high-bandwidth probes used in our
74
previous work [32], which are designed to reduce the effect of common mode noise on the
measured pulse waveforms.
(b) measured in Cell #1. 1% H2-Ar mixture excited by a ns pulse discharge burst, pulse
repetition rate 20 kHz, pulse #25 at P = 300 Torr and T 0 = 500 K. Oscillatory behavior is
caused by multiple pulse reflections from the load and the high-voltage pulse generator.
Figure 3.7 shows voltage, instantaneous power, and coupled energy waveforms
measured in a repetitive ns pulse discharge in 1% H2-Ar at initial temperature of T0=500
K in Cell #1 (P=300 Torr, pulse repetition rate ν=10 kHz, pulse #50 in the burst). Pulse
peak voltage at these conditions is 24 kV, and coupled pulse energy is 2.6 mJ/pulse. The
FID pulser generates positive and negative polarity pulse waveforms, sent to the electrodes
through a coaxial transmission line 2.8 meters long. Voltage oscillations, seen in Fig. 3.7,
are cause by multiple reflections of the pulse from the load (i.e. the discharge electrodes)
and the high-voltage pulse generator. This occurs due to the impedance mismatch between
the transmission line and the plasma. As a result, approximately half of the pulse energy is
75
coupled during the incident pulse and the rest during the pulse reflections. Since no corona
discharge was detected at these conditions outside of the reactor, the entire energy is
coupled to the plasma inside the test section.
Figure 3.8: Photograph of a ns pulse discharge in argon (a). Single shot ICCD camera
images of broadband plasma emission during a ns discharge pulse in Cell #1, side view (b)
and end view (c). Ar at T0= 500 K and P= 700 Torr, discharge pulse repetition rate 10 kHz,
pulse #50, camera gate 1 µs.
A photograph of a repetitive nanosecond pulse discharge plasma in Cell #1, taken
in argon at P=300 Torr and initially at room temperature, after a burst of 50 pulses at
discharge pulse repetition rate ν=10 kHz, is shown in Fig. 3.8(a). The liquid metal
reservoirs above and below the flow channel, used as discharge electrodes, are clearly
visible. Note that side view optical access to the plasma near the top and bottom of the
76
discharge, where the reservoirs are fused to the flow channel, is somewhat limited. The
photograph in Fig. 3.8(a) shows that the plasma is diffuse and fills the entire region between
the electrodes, but also shows brighter regions of the discharge filaments. As shown in our
previous work [8,9], preheating the flow in the discharge section considerably improves
plasma stability and precludes formation of the filaments. To illustrate this, Fig. 3.8(b, c)
shows single-shot ICCD camera images of plasmas generated in argon preheated in the
furnace to T0=500 K in Cell #1 (P=700 Torr, ν=10 kHz, pulse #50). These images are
taken at camera gate of 1 μs, and show side view and end view of the plasma, taken through
the side wall of the channel and through one of the optical access windows at the end of
the reactor. From these images, the plasma appears to be diffuse and nearly uniform, with
no sign of discharge filaments.
discharge pulse in Cell #1, side view (a) and end view (b). 1% H2-Ar at T0= 500 K and P=
300 Torr, discharge pulse repetition rate 10 kHz, pulse #50, camera gate 1 µs.
77
Figures 3.9 and 3.10 show single-shot ICCD images of plasmas in 1% H 2-Ar and
1% O2-Ar mixtures, respectively, when Cell #1 is preheated to T0=500 K (P=300 Torr,
ν=10 kHz, pulse #50). In all images, the plasma remains diffuse, with minor variations in
brightness apparent in the end view images (Fig. 3.9(b) and 3.10(b)). Again, no discharge
filaments are detected.
discharge pulse in Cell #1, side view (a) and end view (b). 1% O2-Ar at T0= 500 K and P=
Figure 3.11 shows similar voltage, instantaneous power, and coupled energy
waveforms taken in a 1% O2-Ar mixture in Cell #2 (P=300 Torr, ν=20 kHz, pulse #75 in
the burst). Pulse peak voltage at these conditions is 27 kV, and coupled pulse energy is 4.2
mJ/pulse. These waveforms are similar to those taken for 0.13% H 2- 1% O2- Ar and 0.25%
CH4- 1% O2- Ar mixtures.
78
Figure 3.11: Voltage and instantaneous power waveforms (a), voltage and coupled pulse
energy waveforms (b) measured in Cell #2. 1% O2-Ar excited by a ns pulse discharge burst,
pulse repetition rate 20 kHz, pulse #75. Oscillatory behavior is caused by multiple pulse
reflections from the load and high-voltage pulse generator.
Plasma uniformity in Cell #2 was tested qualitatively, using a series of ICCD
camera images taken in a 1% O2-Ar mixture. Figure 3.12 shows both single pulse images
(a, b), and 100-shot accumulation images (c, d). No discharge filaments are detected, and
the plasma extends over the entire length of the electrodes. Two dark columns in the images
are the ceramic set screws holding the ceramic plate - dielectric pad assembly together.
Similarly, single shot plasma emission images taken in 0.13% H 2- 1% O2-Ar and 0.25%
CH4- 1% O2- Ar mixtures, are shown in Figures 3.13 and 3.14, respectively. It can be seen
that, as fuel is added to the O2-Ar mixture, the plasma volume somewhat decreases,
developing a narrower “waist” approximately halfway between the electrodes. However,
the plasma in the entire reactor channel remains diffuse, with no sign of filaments or other
non-uniformities.
79
side view (c) and end view (d). 1% O2-Ar mixture at T0= 500 K and P= 300 Torr, discharge
pulse repetition rate 20 kHz, pulse #75, camera gate 1 µs.
discharge pulse in Cell #2, side view (a) and end view (b) 0.13% H 2-1% O2-Ar mixture at
T0= 500 K and P= 300 Torr, discharge pulse repetition rate 20 kHz, pulse #75, camera gate
1 µs.
80
discharge pulse in Cell #2, side view (a) and end view (b) 0.25% CH 4-1% O2-Ar mixture
at T0= 500 K and P= 300 Torr, discharge pulse repetition rate 20 kHz, pulse #75, camera
gate 1 µs.
Figures 3.15 and 3.16 plot voltage, instantaneous power, and coupled energy
waveforms taken in Cell #2 in a mixture of 150 ppm of ethylene in 1% O 2-Ar and a mixture
of 150 ppm of propane in 1% O2-Ar. These waveforms were taken using a different set of
dielectric pads than the waveforms shown in Fig. 3.11, and are slightly noisier. The peak
voltages at these conditions are similar, 28 kV, but coupled pulse energies are significantly
lower, 1.4 mJ/pulse and 1.2 mJ/pulse, respectively. It was determined that the difference
between the coupled discharge pulse energy in Fig. 3.11 and Figs. 3.15 and 3.16 is not due
to changes in the gas mixture composition, and is most likely caused by gradual
deterioration of the dielectric pads.
81
Figure 3.15: Voltage and instantaneous power waveforms measured in Cell #2. Mixture of
75 ppm C2H4 in 1% O2-Ar excited by a ns pulse discharge burst, pulse repetition rate 20
kHz, pulse #75. Oscillatory behavior is caused by multiple reflections from the load and
high-voltage pulse generator.
waveforms (b) measured in Cell #2. Mixture of 150 ppm C3H8 in 1% O2-Ar excited by a
ns pulse discharge burst, pulse repetition rate 20 kHz, pulse #75. Oscillatory behavior is
caused by multiple pulse reflections from the load and high-voltage pulse generator.
82
Comparing Figures 3.11, 3.15, and 3.16 illustrates that coupled pulse energy, which
controls the rates of electron impact processes in the plasma, needs to be measured before
every set of TALIF measurements. Figure 3.17 shows single shot (taken with 100x camera
gain) and 100-shot accumulation (taken with no camera gain) plasma emission images
taken in a mixture of 150 ppm C2H4 and 1% O2-Ar, at discharge pulse repetition rate of 20
kHz, for pulse #75, with a camera gate of 1 µs. It can be seen that the plasma is diffuse and
reproducible, although compared to the images taken in Cell #1 (see Figs. 3.8-3.10), it has
a well-defined “waist” seen in the side view (see Fig. 3.17(a, c)) and the end view images
(see Fig. 3.17(b, d)). No discharge filaments are detected, but the plasma appears brighter
near the bottom wall of the reactor channel. Similarly, in Figure 3.18, single shot images
(taken with 100x camera gain) and 100-shot accumulation images (taken with no camera
gain) show diffuse, reproducible plasma in a mixture of 150 ppm C 3H8 and 1% O2 -Ar, at
T0= 500 K and P= 300 Torr, at a discharge pulse repetition rate of 20 kHz, pulse #75, and
a camera gate of 1 µs.
83
side view (c) and end view (d) mixture of 150 ppm C2H4 in 1% O2-Ar; discharge pulse
repetition rate 20 kHz, pulse #75, camera gate 1 µs.
side view (c) and end view (d) mixture of 150 ppm C3H8 in 1% O2 -Ar. T0= 500 K and P=
Summarizing, images taken over the entire range of experimental conditions
demonstrate that the plasma is diffuse and fills nearly the entire volume between the
electrodes. Images take in Cell #2, with a larger distance between the electrodes, indicate
somewhat smaller extent of the plasma halfway between the electrodes, compared to the
84
plasma near the channel walls. Also, at these conditions, formation of near electrode
regions in apparent. However, no discharge filaments are detected at any of the present
experimental conditions.
3.4. Plasma assisted combustion kinetic model
The kinetic model, discussed in detail in Ref. [40], incorporates the two-term
Boltzmann equation for plasma electrons; electron energy equation and heavy species
energy equation; electron impact processes of electronic excitation, dissociation, and
ionization; reactions of excited electronic states of Ar and O atoms; and “conventional”
chemical reactions of species in the ground electronic states (Konnov [19]). Effects of
diffusion, convective heat transfer, and conduction heat transfer are incorporated as quasi-
zero-dimensional corrections [76,77], using calculated species diffusion coefficients.
Time-resolved discharge power, used as one of the model entries, is obtained from
experimental current and voltage waveforms, such as shown in Fig. 3.11. The model is
exercised using a combination of Boltzmann equation solver for plasma electrons, Bolsig+
[78] predicting electron impact rate coefficients versus electron temperature, and
ChemKin-Pro, solving a set of species concentration equations, heavy species energy
equation, and electron energy equation (a moment of the Boltzmann equation). The model
also incorporates an interface pre-processor, merging electron impact processes (with rate
coefficients predicted by the Boltzmann solver), plasma chemical reactions, and
conventional chemistry reactions, and preparing the input reaction kinetics data file in
ChemKin format.
85
Model input parameters include initial temperature, pressure, initial mixture
composition, flow residence time in the discharge, discharge volume, and experimentally
measured power waveform coupled to the plasma. Using the experimental coupled power
waveform (as a source term in the right-hand side of the electron energy equation) make
additional assumptions, such as estimates of the reduced electric field in the plasma,
unnecessary. This considerably reduces the uncertainty of the model predictions. The
model assumes uniform spatial distribution of energy coupled to the plasma. The
parameters predicted by the model include time-resolved gas temperature, electron
temperature, and species number densities during the discharge and the afterglow.
3.5. Measurements of Ar metastable atoms
In the present work, variation of metastable excited Ar atom number density across
the channel cross section is used to characterize plasma non-uniformity in the reactor. For
this, the laser described in Section 3.2 was scanned over several excited Ar atom absorption
transitions to measure the absorption line profiles in argon plasma excited by the ns pulse
discharge. Figure 3.19 shows an experimental TDLAS spectrum in Ar at P= 300 Torr and
T0= 500 K, compared with the synthetic spectrum, illustrating that the absorption lineshape
is reproduced accurately by the Voigt line shape profile.
86
Figure 3.19: Experimental (symbols) and synthetic (line) absorption spectra (Ar 1s 5→2p7
and 1s3→2p2 transitions) in argon excited by a ns pulse discharge burst, used for Ar(1s 5)
and Ar(1s3) number density inference. P=300 Torr, T0=500 K, ν=10 kHz, 50-pulse burst,
time delay after the last pulse 2 μs. In the synthetic spectrum, absorption line shape is
approximated by a Voigt profile.
Although Ar(3p54s) excited electronic state has four different energy levels, as
shown in Figure 2.7, only the 1s32p2 and 1s52p7 transitions are accessible to the diode
laser output. Since the laser output is spectrally narrow (0.01 cm-1), the absorption
lineshapes are controlled by the line broadening mechanisms in the gas mixture, primarily
Doppler and pressure broadening, as well as Van der Waals broadening [71]. From Figure
3.19, it is evident that the absorption transition from the 1s5 state, used for metastable Ar
atom measurements, has higher absorbance. For this, the laser was parked at the center of
1s52p7 transition in the 3p54s state. Absorption signal was measured before and after the
87
last pulse in the ns pulse discharge burst in three different gas mixtures, pure argon, 1%
H2-Ar, and 1% O2-Ar, as shown in Figure 3.20.
Figure 3.20: Time resolved, normalized Ar* TDLAS signal (1s5→2p7 transition) in argon,
1% H2-Ar, and 1% O2-Ar mixtures excited by a ns pulse discharge burst, taken after the
last pulse in the burst. P=700 Torr (argon), P=300 Torr (1% H 2-Ar and 1% O2-Ar
mixtures), T0=500 K, ν=10 kHz, 50-pulse burst, Noise near t=0 is caused by EMI from the
pulser. Note almost complete absorption in argon and 1% H2-Ar mixture.
The noise due to electromagnetic interference (EMI) generated by the nanosecond
pulse discharge (at t= 0) is detected in all absorption spectra. Both in pure argon and in a
1% H2-Ar mixture, the laser beam is absorbed almost completely (see Fig. 3.20), resulting
in low signal-to-noise. Time-resolved absolute number density of Ar(3p 54s) state was
determined from the absorption spectra (from I/I0>0.1 to I/I0>0.9 to prevent any non-linear
88
effects) using Beer’s law, where the absorption lineshape factor was calculated taking into
account Doppler, pressure, and Van der Waals broadening factors [71].
()
𝑛(𝑙) = 𝑆 ( )
(3.2)
(∆ ,∆ )
( , )
where 𝑆 = ln , is the spectral absorbance, 𝜆 is the transition wavelength, 𝛿( 𝑙)
( , )
and 𝛿( 𝑢) are the degeneracy (statistical weights) of the lower and upper state,
respectively, 𝐿, is the path length, 𝐴 is the Einstein coefficient, and 𝑔(𝜈 −
𝜈 , ∆𝜈 , ∆𝜈 )is the spectral lineshape at 𝜈 = 𝜈 . The Einstein coefficient, and 1s5 state
degeneracies were taken from Ref. [71]. The absorption path was assumed to be equal to
the length of the electrodes, based on the plasma emission images shown in Figs. 3.8-3.10.
Equation 3.2 yields time-resolved, line-of-sight averaged absolute population of 1s 5 state
of Ar(3p54s) atoms, plotted in Figure 3.21.
89
Figure 3.21: Comparison of time resolved [Ar(3p54s)] decay, in the linear absorption
regime (for I/I0=0.1-0.9), in 1% O2-Ar and 1% H2-Ar mixtures at P= 300 Torr, T0= 500 K,
and pure Ar at P= 700 Torr, T0= 500 K, excited by 50 pulses at a pulse repetition rate of
10 kHz, with kinetic modeling predictions.
From Figure 3.21, it can be seen that after the last discharge pulse, absorbance on
the line center is gradually reduced, due to collisional quenching of Ar(1s 5) atoms excited
by electron impact in the discharge. In pure argon, the measured number density of Ar*
atoms is the highest, reaching nAr* = 1.4∙1013 cm-3 at t = 1 μs. As expected, Ar* atom
number density in H2-Ar and O2-Ar mixtures decays more rapidly, due to collisional
quenching by H2 and O2, respectively. In a hydrogen-argon mixture, metastable argon
population reaches only nAr* = 7.4∙1012 cm-3 at t= 0.6 µs, when I/I0= 0.1, and in an oxygen-
argon mixture, the highest measured Ar*(1s5) number density is 5.1∙1012 cm-3, at t= 0.1 µs,
when I/I0= 0.25. The characteristic 1 𝑒 quenching time for Ar*(1s5) for these three cases
is τAr= 0.68 μs, τH2=0.31 μs, and τO2= 0.085 μs, respectively. It can be seen that each
90
nanosecond pulse discharge generates peak Ar*(1s5) density in the range of ~ 1012 - 1013
cm-3, and that Ar*(1s5) atoms do not accumulate in the repetitive discharge plasma at these
conditions, due to their rapid quenching, on the time scale much shorter than the time
interval between the discharge pulses (100 µs).
Figure 3.21 also compares the results of TDLAS measurements with the kinetic
modeling predictions. The kinetic model incorporates the total experimental cross section
of excitation of metastable excited electronic states of argon, Ar(1s 2), Ar(1s3), Ar(1s4), and
Ar(1s5), by electron impact [79]. Populations of these four states are lumped together into
one “effective” state, Ar*(3p54s), with number densities of individual states assumed to be
proportional to their statistical weights [71], [Ar(1s 2)]:[Ar(1s3)]:[Ar(1s4)]:[Ar(1s5)] =
3:1:3:5, i.e. [Ar(1s5)]/[Ar*(3p54s)] = 5/12 [71,79]. This assumption is valid only if the rates
of energy transfer among these four states are much faster compared to the rates of their
collisional quenching and radiative decay. This assumption appears to be fairly accurate in
pure argon plasma, but it results in significant overestimation of Ar(1s 5) populations in the
H2-Ar and O2-Ar plasmas, by approximately a factor of ten. However, [Ar(1s 5)] decay rates
in all three cases are reproduced by the model fairly well. In pure argon, the decay is
controlled by three-body quenching,
𝐴𝑟 ∗ + 𝐴𝑟 + 𝐴𝑟 → 𝐴𝑟 + 𝐴𝑟 + 𝐴𝑟 (3.2)
with room-temperature rate coefficient of kAr= 1.4∙10-32 cm6/s [80], while in H2-Ar and O2-
Ar mixtures, the decay is controlled by two-body quenching by hydrogen and oxygen,
𝐴𝑟 ∗ + 𝐻 → 𝐴𝑟 + 𝐻 + 𝐻 (3.3)
and
91
𝐴𝑟 ∗ + 𝑂 → 𝐴𝑟 + 𝑂 + 𝑂, (3.4)
respectively, with room temperature rates kO2= 6.6∙10-11 cm3/s and kH2= 2.1∙10-10 cm3/s
[81].
To quantify plasma uniformity in Cell #1, twenty-one locations were chosen across
the flow reactor cross section to measure Ar*(1s 5) distribution, as shown in Figure 3.22.
This includes seven equally spaced locations across the channel (between X= 2.5 mm and
X= 14 mm), at three different heights: Y= 1.5 mm, Y= 2.5 mm, and Y= 3.5 mm. These
were done by traversing the laser beam vertically and horizontally over the channel cross
section, without clipping of the laser beam by the channel side walls. Ar* number density
distributions in pure argon, at P= 500 Torr and T0 = 500 K, and in a 1% O2-Ar mixture, at
P= 300 Torr and T0= 500 K, are shown in Figures 3.23 and 3.24.
Figure 3.22: Schematic of twenty-one TDLAS laser beam locations across the channel
cross section (three different heights and seven span wise positions)), used for [Ar(3p 54s)]
distribution measurements after the discharge burst. Laser beam locations are overlapped
with the Ar plasma emission image, also shown in Fig. 3.10. [Ar(3p 54s)]is used as a
quantitative measure of plasma uniformity.
92
Figure 3.23 shows the distribution of line-of-sight-averaged Ar*(1s 5) number
density in a repetitive nanosecond pulse discharge for argon at P=500 Torr and T 0=500 K,
1.5 μs after the last pulse, plotted vs. transverse distance across the reactor channel (X=0
corresponds to the left wall of the channel). At these conditions, Ar* number density
increases near both side walls of the channel, by up to 50% on the left side. Comparison
between the data and the ICCD plasma image taken in pure argon at the same conditions
(see Figure 3.8) demonstrates that broadband plasma emission intensity cannot be used as
a quantitative measure of excited species and radical species distribution in the reactor,
illustrating the necessity of quantitative laser diagnostics.
Figure 3.23: [Ar(3p54s)] distribution across the plasma, measured 1.5 µs after the last pulse
(pulse #50) in the discharge burst. Argon, P=500 Torr, T=500 K, ν=10 kHz, 50-pulse burst.
Data point symbols are the same as in Fig. 3.22. Significant deviation from uniformity, up
to ~ 50%, is detected near the left wall of the channel (at x=2 mm).
93
Figure 3.24 shows the distributions of line-of-sight-averaged Ar*(1s 5) number
density in a repetitive nanosecond pulse discharge in a 1% O 2-Ar mixture at P=300 torr
and T0=500 K, taken 0.23 μs after the last discharge pulse in the burst. In this case,
distribution of Ar* number density over the cross section of the reaction channel is nearly
uniform, with variation not exceeding 15%. All fuel-oxygen-argon mixtures are therefore
kept at P= 300 Torr and T0 = 500 K, to remain close to these conditions.
Figure 3.24: [Ar(3p54s] distribution across the plasma, measured 0.23 µs after the last pulse
(pulse #50) in the discharge burst. 1% O2-Ar mixture, P=300 Torr, T0=500 K, ν=10 kHz,
50-pulse burst. Nonuniformity across the plasma does not exceed 15%. Note that O 2-Ar
plasma is considerably more uniform compared to Ar plasma (see Fig. 3.23).
Additional measurements of Ar*(1s5) have been done in a ns pulse discharge
operated at a higher pulse repetition rate of 98 kHz, to determine if nanosecond pulse
plasmas can generate quasi-steady-state concentrations of Ar*(1s 5) ≥ 1∙1013 cm-3, which
94
would make operation of a new metastable laser, operating on the 2p 10  1s5 transition of
Ar(3p54p), feasible [72]. Figure 3.25 plots time-resolved Ar*(1s 5) number density in pure
argon at P= 300 Torr and T0= 500 K, during the burst of discharge pulses at pulse repetition
rate of 98 kHz.
Figure 3.25: Time-resolved Ar(1s5) number density measured in Cell #1: Argon at P=300
Torr, T0=500 K, discharge pulse repetition rate ν= 98 kHz. Data are taken after the last
pulse in a 25-pulse burst. Experimental data are extrapolated (dashed line) to show peak
[Ar(3p54s)] produced after each pulse.
From Fig. 3.25, it is evident that although the highest measured Ar*(1s5) number
density is only ≈ 8.5∙1012 cm-3, peak Ar*(1s5) number density produced during each
discharge pulse (extrapolated from the data) is ≈ 8.7∙1013 cm-3. Assuming single
exponential decay of metastable argon after the pulse, time-averaged Ar*(1s 5) number
95
density during the discharge burst is ≈ 3∙1011 cm-3, due to rapid Ar* self-quenching by the
reaction of Eqn. 3.2. Therefore, the use of a nanosecond pulse discharge would not be
sufficient to generate the quasi-steady-state number density of Ar*(1s 5) necessary for
lasing. However, this does not exclude the possibility of repetitively pulsed lasing due to
high peak estimated Ar*(1s5) number density, which is an order of magnitude higher than
the predicted threshold density [72].
3.6. TALIF results and discussion
Figure 3.26 plots TALIF laser scans obtained when the dye laser was scanned over
the Kr absorption transition used for calibration, taken with 47.5 Torr of krypton in the cell
(Fig. 3.26(a)), and over the H atom two-photon absorption transition (Fig. 3.26(b)) in Cell
#1 (P = 300 Torr, T0 = 500 K, 1% H2-Ar mixture, 100 µs after the discharge pulse). It can
be seen that the convolution lineshape for the H atom is wider compared to the lineshape
for the Kr atom, due to significantly larger Doppler broadening caused by the higher
average velocity of H atoms.
96
lineshape) for (a) Kr 4p6 1S0  5p’[3/2]2 two-photon absorption transition at 204.13 nm. P
= 47.5 Torr T = 300 K, average laser pulse energy Eavg = 35 µJ. Convolution integral for
(b) H 1s2 S1/2  3d 2D3/2,5/2 two-photon absorption transition at 205.08 nm, P = 300 Torr,
T0 = 500 K, 1% H2-Ar mixture excited by a 50-pulse discharge burst at ν=10 kHz, 100 µs
after the burst, Eavg = 30 µJ.
Figure 3.27 plots O atom two-photon absorption transitions in Cell #2 (O TALIF,
P = 300 Torr, T0 = 500 K, 1% O2 -Ar, 100 µs after discharge pulse), and the corresponding
Xe absorption transition taken at P= 47.5 Torr, T= 300 K. In this case, it can be seen that
the convolution lineshape for O and Xe atoms are similar. In Figures 3.26 and 3.27, vertical
scale is proportional to the TALIF signal intensity divided by the laser pulse energy
squared. Red curves indicate best Gaussian fits used for integrating the lineshape. The
integrals of the lineshapes are normalized to 1, such that 𝑔 (𝜈 ), 𝑔 (𝜈 ), 𝑔 (𝜈 ), and
𝑔 (𝜈 ), are the lineshape factors used in Equations 2.30 and 2.31 for H and O atom number
density calculations.
97
lineshape) for (a) Xe 5p6 1S0  6p’[3/2]2 two-photon absorption transition at 225.24 nm.
P = 47.5 Torr T = 300 K, average laser pulse energy Eavg = 35 µJ. Convolution integral for
(b) O 2p4 3P2,1,0  3p 3P1,2,0 two-photon absorption transition at 225.59 nm, P = 300 Torr,
T0 = 500 K, 1% O2-Ar mixture excited by a 50-pulse discharge burst at ν= 20 kHz, 100 µs
after the pulse, Eavg=30 µJ.
It can be seen that, with the exception of H atom TALIF transition scan, the
lineshape convolution, 𝑔(𝜈 ), and therefore the lineshape factor are controlled by the laser
lineshape.
Figure 3.28 plots typical H TALIF and O TALIF signals taken in 1% H2-Ar and
1% O2-Ar mixtures excited by a ns pulse discharge (50 pulses at a 10 kHz repetition rate
and 75 pulses at a 20 kHz repetition rate, respectively) at P= 300 Torr and T 0= 500 K. The
signals are taken 100 µs after the discharge burst. It can be seen that signal-to-noise in O
TALIF measurements is much higher. To reduce the effect of scattered laser light detected
by the PMT, background PMT traces were taken for each set of measurements, with the
discharge turned off, and subtracted before data processing. As can be seen from Figure
98
3.28, background noise is up to ~10% in H TALIF measurements and up to ~3% in O
TALIF measurements.
Figure 3.28: Comparison between TALIF signal measured by the PMT and background
noise due to laser scattering: (a) H TALIF, 1% H2- Ar mixture (50-pulse burst, ν=10 kHz),
and (b) O TALIF 1% O2- Ar mixture (75-pulse burst, ν=20 kHz). Both signals are taken
100 µs after the end of the discharge burst and accumulated over 100 laser shots. T 0= 500
K, P= 300 Torr, and PMT.
The temperature in the plasma flow reactor controls the rates of chemical reactions
among radical species generated in the plasma and also affects fluorescence quenching
rates (see Eqn. 2.9). From Rayleigh scattering measurements in a 1% H2-Ar mixture at P =
300 Torr, excited by 50 discharge pulses at a 10 kHz repetition rate, the temperature
increases from T0 = 500 K to Tf = 590 ± 67 K 10 µs after the discharge burst. A similar
temperature rise occurs in argon at P = 700 Torr, at the same plasma conditions, from T 0=
500 K to Tf = 590 ± 64 K. A more significant temperature rise is measured in 1% O 2-Ar at
P= 300 Torr and T0= 500 K, excited by 75 pulses at a 20 kHz repetition rate. At t = 200 µs
99
after the discharge burst, the temperature is Tf = 756 ± 151 K. The uncertainty of Rayleigh
scattering data is limited by scattering from the walls of the flow reactor. These data are
consistent with the temperature rise predicted by the kinetic model at these conditions, and
the temperatures predicted by the kinetic model are used in the TALIF data reduction.
Hydrogen-Oxygen-Argon mixtures:
Figure 3.29 plots H TALIF line images, taken in a 1% H2-Ar mixture excited by a
ns pulse discharge burst of 50 pulses, at a pulse repetition rate of 10 kHz, on the centerline
of Cell #1, at P= 300 Torr and T0= 500 K. The images are taken at three different delay
times after the discharge burst, 3 µs, 300 µs, and 3 ms. The images are shifted vertically to
compare signal intensity distributions, and are shown on the same intensity scale. It can be
seen that in addition to TALIF signal intensity decay at 3 ms, the entire image is shifted to
the right in the direction of the flow (left to right). This illustrates that on ~1 ms time scale,
the H atom number density between the discharge electrodes is affected by convective
transport. At these conditions, the characteristic H atom diffusion time is 3 ms, and the
estimated flow residence time is 7 ms. However, since in all subsequent TALIF
measurements the fluorescence signal was collected from a region near the center of the
plasma, ~ 1 cm long, the measurement results are not affected by convection transport until
𝑡~ , i.e. at t > 3.5 ms.
100
Figure 3.29: H TALIF line images (300 laser shot accumulations) taken with an ICCD
camera in 1% H2- Ar mixture in Cell #1 excited by a ns pulse discharge burst, at different
delay times after the burst. P=300 Torr, T0=500 K, 50-pulse burst, ν=10 kHz. Flow
direction from left to right. Both H atom recombination and convection downstream with
flow are evident.
Time-resolved absolute H atom number density after the discharge burst in 1% H 2
– Ar and 1% H2 – 0.15% O2 – Ar mixtures measured in Cell #1 is shown in Figure 3.30,
compared with kinetic modeling predictions. The H 2 – Ar mixture is excited by a burst of
50 discharge pulses at pulse repetition rate of 10 kHz, as shown in Fig. 3.30, while the H 2
– O2 – Ar mixture is excited by a burst of 25 pulses at 20 kHz. At these conditions, pulse
discharge energy coupled to the plasma is 2.6 mJ/pulse. Figure 3.31 plots H atom number
density measured in a 1% H2-Ar mixture versus the number of pulses in the discharge burst,
operated at the conditions of Fig. 3.30 (P= 300 Torr, T0= 500 K, pulse repetition rate of 10
kHz, and coupled discharge pulse energy of 2.6 mJ/pulse). All measurements in Fig. 3.31
have been taken 20 µs after the last pulse in the burst. Comparing Fig. 3.30 with Fig. 3.31,
it can be seen that during 10 µs after the end of the burst, H atom number density does not
change significantly, such that these data represent H atom number density “immediately”
after the burst, when H atom decay due to recombination is not significant. The dashed line
in Fig. 3.30 corresponds to one half of the flow residence time between the discharge
101
electrodes, 𝑡~ ~ 3.5 ms, and indicates the approximate time scale when the effect of
convection on the H atom decay becomes significant.
Figure 3.30: Time resolved H atom number density in a 1% H2-Ar mixture excited by 50
discharge pulses at a pulse repetition rate of 10 kHz and in a 1% H 2-0.15% O2-Ar mixture
excited by 25 discharge pulses at a pulse repetition rate of 20 kHz. Coupled discharge pulse
energy is 2.6 mJ/pulse. Dashed line indicates the approximate time scale when the effect
of convection on H atom decay becomes significant.
102
Figure 3.31: H atom number density measured 20 µs after the end of the discharge burst
for different number of pulses in the burst. Cell #1, 1% H2-Ar, P= 300 Torr, T0= 500 K
ν=10 kHz. Coupled discharge pulse energy is 2.6 mJ/pulse.
From Figs. 3.30 and 3.31, it can be seen that the modeling predictions are in good
quantitative agreement with the data, although the model somewhat overpredicts the rate
of H atom decay due to recombination after the burst (see Fig. 3.30), and predicts a more
gradual H atom number density rise during the burst (see Fig. 3.31).
The dominant mechanism of H atom production at these conditions is determined
from comparison of the data with the modeling predictions. In a 1% H 2-Ar mixture, the
model predicts that approximately 80% of the discharge input energy (0.8 ∗ 2.6 ≈
2.08 𝑚𝐽/𝑝𝑢𝑙𝑠𝑒) goes to electronic excitation of argon by electron impact,
𝑒 + 𝐴𝑟 → 𝑒 + 𝐴𝑟 ∗ (3.5)
103
followed by collisional quenching of 𝐴𝑟 ∗ by hydrogen molecules, previously written as
Equation 3.3,
𝐴𝑟 ∗ + 𝐻 → 𝐴𝑟 + 𝐻 + 𝐻 (3.6)
This is the dominant H atom production reaction in the discharge burst, resulting in
dissociation of ~30% of hydrogen initially available in the mixture. At these conditions,
approximately 30% of the energy coupled to the plasma goes to H2 dissociation. In this
strongly diluted mixture, the role of hydrogen dissociation by electron impact is minor.
In the hydrogen-argon mixture, H atom decay is primarily due to three body
recombination,
𝐻+𝐻+𝑀 →𝐻 +𝑀 (3.7)
which accounts for 95% of the total H atom decay rate, although H atom diffusion to the
reactor channel walls and convection with the flow (see Figs. 3.29 and 3.30) also contribute
somewhat.
In the diluted H2-O2-Ar mixture, 20% of the discharge input energy goes to
hydrogen dissociation by the same mechanisms as discussed above, and 16% of the
discharge input energy goes to oxygen dissociation by electronically excited Ar atoms,
previously written as Equation 3.4.
𝐴𝑟 ∗ + 𝑂 → 𝐴𝑟 + 𝑂 + 𝑂 (3.8)
At these conditions, dominant H atom consumption mechanisms are recombination
with O2 molecules, accounting for 36% of the total H atom decay rate,
𝐻 + 𝑂 + 𝐴𝑟 → 𝐻𝑂 + 𝐴𝑟 (3.9)
reactions with HO2, accounting for 32% of the total H atom decay rate,
104
𝐻 + 𝐻𝑂 → 2𝑂𝐻 (3.10)
and
𝐻 + 𝐻𝑂 → 𝐻 + 𝑂 (3.11)
and three-body recombination, accounting for 23% of the total H atom decay rate.
𝐻+𝐻+𝑀 →𝐻 +𝑀 (3.12)
Summarizing analysis of the kinetic modeling prediction at these conditions,
approximately 30% of the discharge power coupled to electronic excitation in 1% H 2-Ar
mixture goes to dissociating ~30% of the molecular hydrogen initially available during in
the discharge burst, with nearly all H atoms recombining back to H 2. In a 1% H2-0.15%
O2- Ar mixture, 36% of the input power goes to generation of primary radicals (H and O
atoms) by reactions of Eqns. 3.6 and 3.8, the majority of which are converted to OH radicals
via an intermediate radical, HO2, by reaction of Eqn. 3.10. Approximately 20% of H atoms
recombine back to H2. Figure 3.32 plots number densities of the dominant species (H, O,
OH, HO2, and H2O) during and after the discharge burst, at the conditions of Fig. 3.30 (H2-
O2-Ar mixture).
105
densities in a 1% H2- 0.15% O2- Ar mixture excited by a ns pulse discharge of 50 pulses at
T0= 500 K. Coupled discharge pulse energy is 2.6 mJ/pulse.
It can be seen that H and O atoms are the two dominant species generated during
the discharge burst (see Fig. 3.32a), and most of them are converted to water vapor by
106
chemical reactions during and after the burst (see Fig. 3.32b). Comparing the predicted
total number density of the primary radicals, H and O atoms, generated during the
discharge burst, with the number density of the stable fuel oxidation product, H 2O, after
[ ]
the discharge burst, the estimated reaction chain length is [ ] [ ]
≈ 1.1, indicating that
every primary radical generated is consumed to form a stable species, which shows that at
the present low-temperature conditions the role of the chain branching reactions, such as
𝐻 + 𝑂 → 𝑂𝐻 + 𝑂 (3.13)
and
𝑂 + 𝐻 → 𝑂𝐻 + 𝐻 (3.14)
is fairly insignificant, and that the amount of fuel oxidized is controlled by the initial
amount of primary radicals. At the present conditions, the fraction of H 2 molecules
dissociated by a single discharge burst is approximately 24% of the H 2 mole fraction
initially available, i.e. 2400 ppm, and the total mole fraction of water vapor generated is
2374 ppm, as shown in Fig. 3.32.
Figure 3.33 plots time-resolved absolute O atom number density measurements in
1% O2-Ar, 0.13% H2- 1% O2- Ar, and 0.25% CH4 – 1% O2 – Ar mixtures excited by a
burst of 75 pulses at a 20 kHz pulse repetition rate, at P=300 Torr, T 0=500 K, and high
coupled discharge energy, 4.2 mJ/pulse. These, as well as all subsequent measurements are
done in Cell #2. It is evident that at these conditions the model reproduces O atom number
density at the end of the burst, in all mixtures, and reproduces O atom number density
during the decay, in the H2-O2-Ar and CH4-O2-Ar mixtures. In the O2-Ar mixture, the
model overpredicts the O atom number density during the decay.

107
Figure 3.33: Time resolved O atom number density in a 1% O2-Ar mixture, a 0.13% H2-
1% O2-Ar mixture, and a 0.25% CH4-1% O2-Ar mixture excited by 75 discharge pulses at
a pulse repetition rate of 20 kHz. Coupled discharge pulse energy is 4.2 mJ/pulse. Dotted
line indicates the approximate time scale when the effect of convection on O atom decay
becomes significant.
Figure 3.34 plots O atom number density and O2 number density during the
discharge burst, at the conditions of Figure 3.33. Based on the modeling predictions, O
atoms at these conditions are produced mainly by quenching of electronically excited Ar
atoms by O2, by reaction of Eqn 3.8. Approximately 48% of the discharge input power
goes to O atom generation, with 67% of the coupled discharge pulse energy going to Ar*
excitation by electron impact, and approximately 48% of O2 initially available is
dissociated, as illustrated in Figure 3.34. In the O2-Ar mixture, O atom recombination is
due to three body recombination,

108
𝑂+𝑂+𝑀 →𝑂 +𝑀 (3.15)
with O atom diffusion to the walls and convection with the flow also contributing at t > 14
ms. Ozone chemistry, also incorporated in the model, is negligible at these relatively high
temperatures.
Figure 3.34: Predicted O atom and O2 species number densities in a 1% O2-Ar mixture
excited by a ns pulse discharge of 75 pulses at pulse repetition rate of 20 kHz, in the burst,
at P= 300 Torr and T0= 500 K. Coupled discharge pulse energy is 4.2 mJ/pulse.
As discussed in Section 3.4, the kinetic model incorporates the effect of diffusion
of O, H, and OH species as a 0D correction. At these experimental conditions, the estimated
time for transverse diffusion of O atoms to the channel wall is τ diff ≈ (h/π)2/D ≈ 37 ms,
where h= 1 cm is the channel height and D = 2.09∙102 cm2/s is the O atom diffusion
coefficient in argon at the present conditions [76]. This is longer compared to the measured
1 3
𝑒 O atom decay time, ≈ 10 ms. The diffusion coefficient of H atoms in argon is D=1∙10
109
cm2/s, and the diffusion coefficient for OH molecules in argon is D= 1.78∙10 2 cm2/s [77],
and the estimated time for transverse diffusion of H atoms and OH molecules to the channel
wall is τdiff ≈ 8 ms, and τdiff ≈ 43 ms, respectively.
Similar to the estimated time for the transverse diffusion of O atoms, centerline O
atom decay time due to convection out of the fluorescence signal collection region is
𝜏 ≈ ≈ 31 𝑚𝑠, where 𝐿~6 𝑐𝑚 is the length of the plasma region and 𝑢~134 𝑐𝑚/𝑠
is the centerline flow velocity estimated from the mass flow rate of 2.0 slm, assuming a
fully developed flow in the channel, in a 1% O2-Ar mixture. Again, this is longer compared
to the measured O atom decay time (see Fig. 3.33). However, the time scale for combined
diffusion and convection losses, 𝜏 =𝜏 + 𝜏 ≈ 10 𝑚𝑠, is comparable with the
recombination decay time, as shown by the dashed line in Figure 3.33. Incorporating the
effect of convection makes the agreement between the data and the modeling predictions
in the O2-Ar mixture noticeably better.
In a 0.13% H2- 1% O2- Ar mixture, the model also reproduces production and decay
rate of O atoms quite well. When hydrogen is added to the mixture, 46% of the discharge
input power goes to generating O atoms (by reaction of Eqn. 3.8), and an additional 1% of
the input power goes to H atom generation (by reaction of Eqn. 3.6). At these conditions,
O atom reaction with OH accounts for 68% of the total decay rate,
𝑂 + 𝑂𝐻 → 𝐻 + 𝑂 (3.16)
O atom reaction with HO2 accounts for 14% of the total decay rate,
𝐻𝑂 + 𝑂 → 𝑂𝐻 + 𝑂 (3.17)
110
and chain branching reaction with H2, Equation 3.14, accounts for 10% of the total O atom
decay rate. Three-body recombination reaction,
𝑂 + 𝐻 + 𝑀 → 𝑂𝐻 + 𝑀 (3.18)
and three-body recombination (by reaction of Eqn. 3.15) play almost no role in O atom
decay in this mixture. Reactions with OH and HO2 result in significant reduction of O atom
number density generated by the end of the discharge burst, compared to O 2-Ar mixture,
as well as its more rapid decay after the burst (see Fig 3.33).
111
densities in a 0.13% H2- 1% O2- Ar mixture excited by a ns pulse discharge of 75 pulses at
T0= 500 K. Coupled discharge pulse energy is 4.2 mJ/pulse.
Figure 3.35 plots number densities of the dominant species (H, O, OH, HO 2, and
H2O) during and after the discharge burst, at the conditions of Figure 3.33 (H2-O2-Ar
mixture). It can be seen that H and O atoms are the dominant species generated during the
discharge burst (see Fig. 3.35(a)), and most of them are converted to water vapor in
chemical reactions during and after the burst (see Fig. 3.35(b)). Comparison of the total
number density of primary radicals, O and H atoms generated during the discharge burst,
112
with H2O number density produced during and after the burst, predicted by the model at
these conditions, again shows that the estimated reaction chain length is fairly low,
[ ] [ ]
[ ]
≈[ ] [ ]
≈ 0.1. The conversion of primary radicals to stable species is
low because at these hydrogen-lean conditions, the conversion is limited by H atom
generation. Indeed, the ratio of peak water vapor number density to the total number
[ ]
density of H atoms generated during the discharge burst is [ ]
≈ 2.7. Again, the role of
major chain branching reactions, Equations 3.13 and 3.14, is insignificant at these
relatively low-temperature conditions. As expected, the mole fraction of water vapor
produced [681 ppm], is close to the mole fraction of hydrogen oxidized [863 ppm] at the
present conditions, both during the discharge burst (see Fig. 3.36(a)) and in the discharge
afterglow (see Fig. 3.35(b)).
Hydrocarbon-Oxygen-Argon mixtures:
Figure 3.33 compares O atom number density measurements in a 0.25% CH 4- 1%
O2- Ar mixture excited by a ns pulse discharge, at the same conditions as 1% O 2-Ar and
0.13% H2- 1% O2-Ar mixtures. It is apparent that, similar to O2-Ar and H2-O2-Ar mixtures,
the modeling predictions are in good agreement with the data, including O atom number
density at the end of the burst and its decay after the discharge burst. Analysis of kinetic
modeling predictions demonstrates that in this case, 31% of the discharge coupled energy
goes to O2 dissociation (by reaction of Eqn. 3.8), similar to O2-Ar and H2-O2-Ar mixtures.
Additionally, 13% of input energy is spent on CH 4 dissociation by excited argon atoms,
𝐴𝑟 ∗ + 𝐶𝐻 → 𝐴𝑟 + 𝐶𝐻 + 𝐻 (3.19)
113
Similar to H2-O2-Ar mixtures, O atom number density after the discharge burst
decays predominately due to the reaction with OH, Equation 3.16, accounting for 66% of
the total decay rate. 70% of OH produced is by H atom reaction with HO 2, Equation 3.10.
The main production channel of HO2 is the recombination reaction between H atoms and
O2, accounting for >90% of the total production rate (Eqn. 3.9). These three reactions are
equivalent to the following net O atom decay process,
𝑂 + 𝐻 → 𝑂𝐻 (3.20)
Therefore, the net rate of O atom decay after the discharge burst is controlled almost
entirely by the total number density of H atoms generated during the burst. As can be seen
in Figure 3.36(a), the H atom number density increases throughout the discharge burst, as
does the OH number density. By the end of the discharge burst, the predicted H number
density is ~ 2•1015 cm-3 (see Fig. 3.36(b)), which is close to the measured O atom number
density, ~1.5•1015 cm-3 (see Fig. 3.33), which accelerates O atom decay significantly.
114
Figure 3.36: Predicted radical (O, H, OH, and HO2) and stable (CH4 and H2O) species
number densities in a 0.25% CH4- 1% O2- Ar mixture excited by a ns pulse discharge of
75 pulses at pulse repetition rate of 20 kHz, in the burst (a) and in the afterglow (b), at P=
300 Torr and T0= 500 K. Coupled discharge pulse energy is 4.2 mJ/pulse.
Figure 3.36 plots number densities of dominant species (O, H, OH, HO 2, H2O and
CH4) during and after the burst. At these conditions, a single discharge burst oxidizes
115
approximately 52% of the fuel initially available in the discharge burst, as shown in Fig.
3.36(a). Similar to H2-O2-Ar mixtures, the reaction chain length is estimated as
[ ] [ ] [ ] [ ] [ ] [ ]
[ ]
≈ [ ] [ ] [
≈ 0.6, is low. Nearly all carbon in the fuel
]
(CH4) oxidized [1614 ppm mole fraction] is converted to three stable species, [𝐶𝐻 𝑂] +
[𝐶𝑂] + [𝐶𝑂 ] = [1590 𝑝𝑝𝑚], and nearly every H atom in the oxidized fuel
[4 × 1614 𝑝𝑝𝑚 = 6456 𝑝𝑝𝑚] is converted to three dominant stable species, 2 ∗
([𝐶𝐻 𝑂] + [𝐻 ] + [𝐻 𝑂]) = [6170 𝑝𝑝𝑚].
pulses in a burst, at two coupled pulse energies 2.3 mJ/pulse (a) and 1.4 mJ/pulse (b).
116
Figure 3.37 compares time resolved O atom number density measurements in lean
C2H4-O2-Ar mixtures excited by a ns pulse discharge, for two different C 2H4 mole
fractions, 75 ppm and 150 ppm, and two different coupled pulse energies, 2.3 mJ/pulse and
1.4 mJ/pulse. Figure 3.38 plots number densities of dominant species (O, H, OH, HO 2,
H2O, and C2H4) in a C2H4-O2-Ar mixture with an initial fuel mole fraction of 150 ppm
during (Fig. 3.38(a)) and after (Fig. 3.38(b)) the burst. Similar to H2-O2-Ar and CH4-O2-
Ar mixtures, at these conditions 45% of the discharge coupled energy goes to O and H
atom generation (predominately O atoms) by quenching of electronically excited Ar atoms.
However, unlike in H2-O2-Ar and CH4-O2-Ar mixtures, at these conditions the agreement
between the data and the modeling predictions is not nearly as good. Specifically, the
model considerably underpredicts the O atom decay rate after the burst. This effect
becomes more significant as discharge pulse energy is reduced (see Fig. 3.37). Note that at
these very fuel-lean conditions, nearly all fuel (C2H4) is oxidized after only ~30 discharge
pulses, as illustrated in Figure 3.38(a). In addition, Fig. 3.38(a) shows that H atom number
density during the burst reaches a plateau as the fuel is depleted, unlike in the 0.25% CH 4-
O2-Ar mixture, where H atoms number density is increasing during the entire burst (see
Fig. 3.36(a)). The leveling off of H atoms occurs in spite of dissociative quenching of Ar*
by intermediate ethylene dissociation and oxidation products, as well as water vapor, which
contributes to H atom generation. As a result, H atom number density in the afterglow is
lower than in the CH4-O2-Ar mixture, by approximately an order of magnitude (compare
Figs. 3.36 and 3.38).
117
number densities in a mixture of 150 ppm of C2H4 in a 1% O2- Ar excited by a ns pulse
discharge of 75 pulses at pulse repetition rate of 20 kHz, in the burst (a) and in the afterglow
(b), at P= 300 Torr and T0= 500 K. Coupled discharge pulse energy is 1.4 mJ/pulse.
At the beginning of the discharge burst, after one discharge pulse, the O atom decay
is controlled by reactions with ethylene,
𝑂 + 𝐶 𝐻 → 𝐶𝐻 + 𝐻𝐶𝑂, (3.21)
118
𝑂 + 𝐶 𝐻 → 𝐶𝐻 + 𝐻𝐶𝑂 + 𝐻 (3.22)
and
𝑂 + 𝐶 𝐻 → 𝑜𝑡ℎ𝑒𝑟 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑠 (3.23)
which accounts for 31%, 18%, and 4% of the total O atom decay rate, respectively.
After five discharge pulses, reactions of O atoms with OH and HO2 become as
significant as O atom reactions with fuel and fuel products. Specifically, at this point in the
discharge burst, 38% of the O atom decay rate is controlled by reaction with OH, Equation
3.16, and 20% of the total decay rate is O atom reaction with HO2. OH is produced by two
mechanisms, H atom reaction with HO2 (Eqn. 3.10), and O atom reaction with HO2 (Eqn.
3.17), accounting for 47% and 43% of the total production rate, respectively. The
remaining O atom decay rate is controlled by oxidation of C 2H4, Eqn. 3.21 (22% of the
total decay rate).
After 75 discharge pulses, all ethylene initially available in the mixture has been
oxidized, and the O atom decay after the burst is controlled by the reaction with OH, Eqn.
3.16 (47% of the total decay rate), reaction with HO2, Eqn. 3.17 (27% of the total decay
rate), as well as three-body recombination, Eqn. 3.15 (12% of the total decay rate). While
the production of OH is controlled by primary radical reactions with HO 2, a significant
difference compared to H2-O2-Ar and CH4-O2-Ar mixtures is that H atom reaction with
HO2 now accounts for only 30% of the total OH production rate, while O atom reaction
with HO2 accounts for 60% of the total OH production rate. Nearly all HO2, 97% of the
total production rate, is produced by the reaction of Eqn. 3.9 (recombination of H and O 2).
At these conditions, the net equivalent O atom decay process becomes,
119
𝑂+𝑂 →𝑂 (3.24)
the rate of which is slower compared to that of Eqn. 3.20, and is no longer controlled by
the amount of hydrogen atoms generated in the discharge.
Similar to the time-resolved measurements, the modeling predictions for O atom
number density versus C2H4 mole fraction in the mixture, at the lower coupled pulse energy
value of 1.4 mJ/pulse, overpredict the O atom number density, as shown in Figure 3.39.
The data and the modeling predictions are compared 70 µs after the discharge burst.
Figure 3.39: Absolute O atom number density versus initial mole fraction of ethylene in
the mixture excited by a ns pulse discharge. P= 300 Torr, T 0= 500 K, ν=20 kHz, 75 pulses
in a burst, with discharge coupled energy of 1.4 mJ/pulse. Data and modeling predictions
are compared 70 µs after the discharge burst.
Finally, Figure 3.40 compares time resolved O atom number density measurements
in lean C3H8-O2-Ar mixtures excited by a ns pulse discharge, for two different mole
120
fractions of C3H8 initially available in the discharge burst, 75 ppm and 150 ppm, with the
modeling predictions, showing that the model overpredicts the O atom decay rate as the
mole fraction of propane is reduced, as evident from Figure 3.40.
pulses in a burst, with discharge coupled pulse energy of 1.4 mJ/pulse.
Figure 3.41 plots number densities of dominant species (O, H, OH, HO 2, H2O and
C3H8) in a C3H8-O2-Ar mixture with a mole fraction of fuel initially available in the
discharge burst of 75 ppm during (Fig. 3.41(a)) and after (Fig. 3.41(b)) the burst, as well
as in a C3H8-O2-Ar mixture with a mole fraction of fuel initially available in the discharge
burst of 150 ppm during (Fig. 3.41(c)) and after (Fig. 3.41(d)) the burst. At these
conditions, 46% of the discharge coupled energy goes to O atom (45%) and H atom (1%)
generation, by quenching of electronically excited Ar atoms. Unlike in fuel-lean C 2H4-O2-
121
Ar mixtures, complete oxidation of 75 ppm of C 3H8 occurs much later in the burst, after
56 discharge pulses, as shown in Figure 3.41(a), and oxidation of 150 ppm of C 3H8 is still
occurring after 75 discharge pulses, as shown in Figure 3.41(c).
In C3H8-O2-Ar mixtures, after one discharge pulse, 60% of decay rate of O atoms
is due to fuel oxidation,
𝑂 + 𝐶 𝐻 → 𝐶 𝐻 + 𝑂𝐻 (3.25)
while 30% of the total decay rate is O atom reaction with OH, Eqn. 3.16. At this time, OH
is produced by H and O atom reactions with HO2, Eqn. 3.10 (46% of total OH production)
and Eqn. 3.17 (40% of total OH production).
122
number densities in a mixture of 75 ppm C3H8 in 1% O2- Ar, in the burst (a) and in the
afterglow (c), and in a mixture of 150 ppm C3H8 in 1% O2- Ar, in the burst (b) and in the
kHz, at P= 300 Torr and T0= 500 K. Coupled discharge pulse energy is 1.4 mJ/pulse.
After a 75 pulse discharge burst, in a mixture of 75 ppm C 3H8 in 1% O2-Ar, 47%
of the total O atom decay rate is due to reaction with OH, Eqn. 3.16, 30% of the total decay
rate is due to reaction with HO2, Eqn. 3.17, and 9% of the total decay rate is due to three-
body recombination of O atoms. In this fuel-limited mixture (i.e. at the conditions when all
fuel initially available in the mixture is completely oxidized during the burst), nearly 60%
of OH production rate is due to the reaction of Equation 3.17 (O atom reaction with HO 2).
123
In addition, 93% of the total HO2 production rate is by H atom recombination with O2,
Eqn. 3.9. Similar to C2H4-O2-Ar mixtures, the production of H atoms during the discharge
burst levels off (see Fig. 3.41(a)), and the predicted net O atom decay process, Eqn. 3.24,
is independent of H atoms number density in the mixture.
After a 75 pulse discharge burst in a mixture of 150 ppm of C 3H8 in 1% O2-Ar, O
atom reaction with OH (Eqn. 3.16) and reaction with HO2 (Eqn. 3.17) accounts for 58%
and 24% of the total O atom decay rate, respectively. Figure 3.41(b) shows that H atom
number density keeps increasing throughout the burst, and the dominant OH production
mechanism is H atom reaction with HO2 (60% of the total productions rate). This
corresponds to the same net O atom decay process as in H2-O2-Ar and CH4-O2-Ar mixtures,
Eqn. 3.20.
[ ]
At these conditions, the reaction chain length, estimated as [ ]
≈
[ ] [ ] [ ] [ ] [ ]
[ ] [ ] [
≈ 0.2, is low, indicating that chain branching mechanisms are
]
insignificant, and every C atom and H atom released by the decomposition of propane is
converted into stable species.
Comparing the results for mixtures with two different mole fractions of propane
initially available in the discharge burst, the main decay mechanisms of O atoms after the
burst are similar, but the production mechanisms of OH differ, depending on the amount
of C3H8 left after the discharge burst. The fact that the model underpredicts O atom decay
rate in the 75 ppm propane mixture, as well as in both C 2H4 mixtures, suggests that the
amount of hydrogen-containing species in the plasma may be higher, either due to an
ambient air (water vapor) leak into the reactor or due to the effect of additional reactants.
124
As is evident from the plasma emission images in Figure 3.18, there are regions near the
side walls of the reactor channel where the plasma is not generated, and where unreacted
mixture is present. Since the reactor leak rate, 57 Torr per hour, is approximately six orders
of magnitude lower compared to the reactant mixture flow rate, the upper bound estimate
of water vapor impurity in the mixture is ~1 ppm, much lower compared to the level
necessary to affect the modeling predictions significantly (estimated to be ~200 ppm), this
indicates the effect of additional fuel species as a possible reason for the discrepancy.
Additional modeling calculations incorporating transverse diffusion of the fuel from the
“unreacted” regions near the side wall toward the “reacted” center of the flow, showed that
this effect is sufficient to increase the H atom number density near the center of the flow,
and to accelerate the predicted O atom decay to the value observed in the experiments,
demonstrating that kinetics no longer plays a significant role in O atom decay at these fuel-
limited conditions.
Comparison of O atom number density measured 90 µs after the discharge burst
with the modeling predictions at different mole fractions of fuel (propane) initially
available in the discharge burst, is shown in Figure 3.42, indicating that the model generally
overpredicts the number density of O atoms. However, the agreement between the
modeling predictions and the data improves as the mole fraction of propane is increased.
125
Figure 3.42: Absolute O atom number density versus initial mole fraction of propane in the
mixture excited by a ns pulse discharge. P= 300 Torr, T0= 500 K, ν=20 kHz, 75 pulses in
a burst, with discharge coupled energy of 1.2 mJ/pulse. Data and modeling predictions are
compared 90 µs after the discharge burst.
Summarizing, this chapter presents the results of temperature measurements and
absolute, time-resolved number density measurements of excited metastable Ar atoms, as
well as H and O atoms, in mixtures of H2, O2, CH4, C2H4, and C3H8 with argon, excited by
a repetitive nanosecond pulse discharge. These measurements have been made in a diffuse,
double dielectric barrier discharge plasma sustained in two different plasma flow reactor
geometries, at the initial temperature of 500 K and pressures ranging from 300 Torr and
700 Torr. Metastable Ar atom number density distributions in the discharge afterglow,
measured by Tunable Diode Laser Absorption Spectroscopy (TDLAS), are used to
quantify the plasma uniformity, showing that Ar*(1s5) number density variation across the
channel cross section does not exceed 10-25%. Temperature in the discharge-excited
reacting flow is measured by Rayleigh scattering. The results are compared with
126
predictions of a kinetic model for analysis of radical reaction kinetics at low temperature,
for nonequilibrium plasma assisted combustion applications.
Two-photon Absorption Laser Induced Fluorescence (TALIF) H and O atom
measurements after the discharge burst demonstrate that at the present conditions, large
fractions of hydrogen and oxygen initially available in the mixture are dissociated in the
plasma, up to 36% in a 1% H2-Ar mixture and up to 48% in a 1% O2-Ar mixture. Analysis
of the modeling predictions indicate that significant fractions of discharge input energy
goes to generation of H and O atoms, primarily by quenching or electronically excited
argon atoms, in good agreement with the data. In H2-O2-Ar mixtures, the model predicts
accurately both H and O atoms number densities at the end of the discharge burst, and their
decay after the burst. The number density of water vapor generated after the burst is
comparable with the total number density of H and O atoms generated during the discharge
burst, indicating that at these low-temperature conditions, the role of chain-branching
reactions is minor, such that the primary radicals either participate in chain propagation
reactions or recombine, terminating reaction chains.
In lean mixtures of hydrogen and hydrocarbon fuels with oxygen-argon, the
modeling predictions show good agreement with the data only when the initial fuel species
is still present at the end of the discharge burst. In H2-O2-Ar, CH4-O2-Ar mixtures, as well
as in a 150 ppm C3H8 -O2-Ar mixture, O atoms after the discharge decay primarily in a
reaction with OH, generated during H atom reaction with HO2, with the latter produced by
three-body H atom recombination with O2. Therefore, the equivalent net process of O atom
decay is 𝑂 + 𝐻 → 𝑂𝐻, such that rate of O atom decay in the afterglow is controlled by the
127
amount of H atoms generated in the discharge. In C 2H4-O2-Ar mixtures and in the 75 ppm
C3H8-O2-Ar mixture, the agreement with the data and modeling predictions is not nearly
as good. Analysis of the modeling predictions indicates that at these conditions, when the
number density of hydrogen atoms is low, OH, which remains the dominant O atom
scavenger, is generated primarily in O atom reaction with HO2. The equivalent net process
of O atom decay becomes 𝑂 + 𝑂 → 𝑂 , such that rate of O atom decay in the afterglow
approaches that in the O2-Ar mixture and is only weakly dependent on H atom number
density. Lack of agreement between the modeling predictions and the data at these
extremely lean conditions suggests the effect of transverse diffusion of the fuel from the
“unreacted” peripheral regions near the reactor channel in the “reacted” core flow, thereby
enhancing the H number density in the reacting mixture. O atom number density predicted
at the end of the discharge burst in C2H4-O2-Ar mixtures and C3H8-O2-Ar mixtures versus
the initial amount of fuel, is in fairly good agreement with the data, especially as the initial
mole fraction of fuel species is increased.
The present results show kinetic modeling predictions to be in good agreement with
the data, with the exception of extremely fuel-lean mixtures. If the fuel mole fraction is
sufficiently high, such that it is not completely oxidized during the discharge burst, the
present results demonstrate that the model predictions would be accurate.
Finally, the modeling predictions show that O atoms generated in the plasma
rapidly oxidize hydrogen and hydrocarbon fuel species. However, at temperatures
significantly below ignition temperature, both H and O atoms generated in the plasma in
significant amounts predominately recombine to produce HO 2 radicals, effectively
128
terminating chain propagation processes and limiting the effect of the plasma on fuel
oxidation.
In hydrocarbon-oxidizer mixtures that are not fuel-limited, radical species produced
in the plasma are consumed rapidly by fuel oxidation reactions. At these conditions,
consumption of the radical species generated in the plasma is so rapid that there is almost
no accumulation over multiple pulses, making measurements of radical species extremely
challenging, due to the detection limit of the diagnostics. Figure 3.43 plots the predicted
number densities of dominant species in a 1% C 2H4 - 1% O2-Ar mixture during (Fig.
3.43(a)) and after (Fig. 3.43(b)) the discharge burst, and in a 1% C 3H8-1% O2-Ar mixture
during (Fig. 3.43(c)) and after (Fig. 3.43(d)) the burst.
In a C2H4-O2-Ar mixture, the O atoms generated in the plasma by the reaction of
Eqn. 3.8, and the H atoms generated by the following reaction,
𝐴𝑟 ∗ + 𝐶 𝐻 → 𝐴𝑟 + 𝐻 + 𝐶 𝐻 (3.27)
are consumed by 10 microseconds after the discharge burst. 58% of O atoms react with the
fuel, by reactions of Eqns. 3.20, 3.22-3.24, producing radical species. Another fuel
oxidation reaction accounts for 42% of the total H atom production rate in the discharge
afterglow,
𝑂 + 𝐶 𝐻 → 𝐻 + 𝐶𝐻 𝐻𝐶𝑂 (3.30)
At these conditions, there is no significant O atom decay reaction between O atoms
and OH molecules. Additional O atom consumption processes include reactions with
hydrocarbon and oxygenated radical species, Eqns. 3.17 (6%), 3.21 (6%), and 3.26 (9%).
129
The production rate of H atoms in the discharge afterglow is approximately half of the
consumption rate, with 36% of the total decay rate by the reaction,
𝐻+𝐶 𝐻 +𝑀 →𝐶 𝐻 +𝑀 (3.31)
and the reaction of H atoms with HO2, Eqn. 3.10, accounting for 18% of the total decay
[ ]
rate. At these conditions, the reaction chain length is estimated as [ ]
≈
[ ] [ ] [ ] [ ] [ ]
[ ] [ ] [ ]
≈ 2.3, which indicates that even when there is a sufficient
amount of fuel available, chain branching remains insignificant at these low temperature
conditions.
130
number densities in a mixture of 1% C3H8 -1% O2- Ar, in the burst (a) and in the afterglow
(c), and predicted radical (O, H, OH, and HO2) and stable (C3H8 and H2O) species number
densities in a mixture in a mixture of 1% C3H8 - 1% O2- Ar, in the burst (b) and in the
kHz, at P= 300 Torr and T0= 500 K. Coupled discharge pulse energy is 1.4 mJ/pulse.
In 1% C3H8- 1% O2-Ar mixtures, O atom decay kinetics are controlled by fuel
oxidation, Eqn. 3.25, reaction with CH2O, Eqn. 3.28, and reaction with CH3, Eqn. 3.21,
accounting for 36%, 18%, and 11% of the total decay rate, respectively. Equations 3.26
and 3.21 are also the main production rate mechanisms for H atoms in the discharge
131
afterglow, 41% and 22%, respectively. At these conditions, H atom production rate is 20%
of the total decay rate. The H atom decay rate is controlled by reactions with hydrocarbon
and oxygenated species. At these conditions, the reaction chain length is estimated as
[ ] [ ] [ ] [ ] [ ] [ ]
[ ]
≈ [ ] [ ] [ ]
≈ 3, indicating that at these conditions,
chain branching remains a fairly minor effect.
Thus, in the entire range of fuel-oxidizer mixtures considered in the present
experiments and modeling calculations, chain branching remains generally ineffective.
Enhancing the chain branching effect would require increasing the initial temperature of
the mixture, which will be the focus of future work.
The present results also demonstrate that radical species using ns TALIF
measurements are feasible only in lean fuel-oxidizer mixtures, where quenching of atomic
species is not prohibitively rapid.
132
Chapter 4: Measurements of OH number density at the
conditions of strong vibrational nonequilibrium
The role of vibrationally excited nitrogen molecules in the ground electronic state,
𝑁 (𝑋 𝛴, 𝑣), in chemical kinetics of reacting nonequilibrium plasmas is not well
understood. A relatively slow N2 vibrational relaxation rate affects the rate of temperature
rise in the plasma afterglow and, indirectly, the rate coefficients of chemical reactions. A
more direct effect of vibrationally excited N2 via vibrational energy transfer from nitrogen
molecules to other reactive molecular species, as discussed in Section 1.3, suggests that
N2(v=1) may sustain chain propagation and chain branching chemical reactions of H atoms
in low-temperature afterglow in fuel-air mixtures, and thus extend the OH radical lifetime.
Although it may well be possible that reactions of vibrationally excited N 2 molecules in
fuel-air plasmas affect low-temperature fuel oxidation chemistry, additional experimental
studies are needed to quantitatively understand whether this is a direct or indirect effect.
The objective of this work is to make time-resolved measurements of absolute OH
number density, nitrogen vibrational temperature, and translational-rotational temperature
in air and lean hydrogen-air plasmas generated in a diffuse filament, nanosecond pulse
discharge, at high specific energy loading. The use of simultaneous N2 vibrational Coherent
Anti-Stokes Raman Spectroscopy (CARS) and OH Laser Induced Fluorescence (LIF) for
133
these measurements provide quantitative insight into the effect of N2 vibrational excitation
on OH kinetics in low-temperature H2-air plasmas.
4.1. Discharge Test Cell and Electrode Assembly
A schematic of the discharge cell used in the present experiments is shown in Figure
4.1. The discharge is sustained between two spherical electrodes, 7 mm in diameter, made
of copper, placed 8 mm apart. The electrodes are attached to copper tubes 1/4″ in diameter
and 16 mm long, with inside diameter of 3 mm, and held by a U-shaped bracket machined
of macor ceramic. A 1 mm diameter hole is drilled through each electrode, as shown in
Figure 4.2, and the electrodes are aligned to allow optical access for the laser diagnostics.
The OH fluorescence signal is collected at a 900 angle, as shown in Figure 4.1.
134
Figure 4.1: (a) Schematic diagram of the experimental apparatus including a glass six-arm
cross test cell with BK-7 glass windows providing optical access to the laser beams and
fused silica windows for fluorescence signal collection at 90° and for plasma imaging.
The electrode assembly is placed inside a six-arm cross glass cell, with two 3″
diameter BK-7 glass windows providing optical access for the laser beams, two 2″ fused
silica windows used to collect the fluorescence signal and to take plasma emission images,
and two 1″ flanges for high-voltage electrode feedthroughs. The cell is mounted on a three-
dimensional translation stage, for the alignment of the electrodes to the laser beams, and is
equipped with two 1/4″ diameter ports for the gas flow inlet and exit. The laser beams are
directed along the centerline of the filament, shown schematically in Figure 4.1. This
approach provides a significant advantage in probing excited environments with relatively
low spatial gradients of parameters in the direction of the laser beams.
135
Figure 4.2: Left: photograph of the ceramic bracket holding the discharge electrodes, taken
through the 2” diameter optical access window in one of the arms of the cell, with an ICCD
camera image showing a reference image of the electrodes. Right: end view photograph
of the electrode assembly, showing a 1 mm diameter aperture drilled through the electrode
center, taken through the 3” diameter window providing optical access to the laser beams.
OH LIF experiments are made in lean hydrogen-air mixtures (1-5% H 2) and CARS
experiments are done in lean hydrogen-air mixtures (1% and 3% H2) and dry air. All
measurements are carried out at a pressure of either 60 or 100 Torr, initial temperature of
300 K, and total flow rate through the cell of 1.3 slm. The flow velocity through the cell,
estimated from the flow rate and the cell arm cross sectional area, is approximately 4 cm/s.
Gases are delivered to the cell from high-pressure cylinders, where hydrogen and air are
mixed approximately 3 meters upstream of the cell. The flow through the cell is maintained
by a vacuum pump, and the flow rates are measured by Omega and MKS mass flow
controllers. The leak rate of the cell is approximately 6 Torr per hour, such that the
estimated water vapor impurity level is ~10 ppm.

136
The discharge electrodes are connected to the feedthroughs and powered by two
different custom-built, externally triggered, high-voltage pulse generators used in our
previous work, referred to as Pulser #1 and Pulser #2 in the present work [82]. In these
experiments, Pulser #2 generates an alternating polarity pulse sequence with peak voltage
of approximately 10 kV, FWHM pulse duration of about 100 ns, and pulse repetition rate
of 10 Hz (for LIF measurements) and 60 Hz (for CARS measurements). Pulser #1 generates
a set of positive polarity pulses; with primary pulse duration of ~100 ns and peak voltage
of ~ 7.5 kV and secondary pulse duration of ~100 ns and peak voltage of 2.5 kV, which
occurs approximately 400 ns after the end of the primary pulse. Pulse voltage and current
are measured by a high voltage probe (Tektronix P6015) and current probe (Pearson 2877),
respectively. At the present conditions, high-voltage pulses generate a diffuse filament
discharge between the electrodes, approximately 3 mm in diameter, both in air and in H 2-
air mixtures. Plasma emission images are taken by an ICCD camera (Princeton
Instruments, PI-MAX III) with a UV lens (UV-Nikkor 105 mm f/4.5, Nikon).
4.2. Pulsed discharge waveforms and plasma images
Figure 4.3 plots positive and negative polarity pulse discharge voltage, current, and
coupled energy waveforms from Pulser #2 in a 3% H2-air mixture at 100 Torr, at two
different pulse energies coupled to the plasma (controlled by DC voltage input of the high-
voltage pulse generator), 7.5 mJ (a, b) and 10.0 mJ (c, d). Waveforms taken in air and other
H2-air mixtures (1% and 5% of H2), as well as coupled pulse energies, are very similar.
137
Pulse peak voltage at these conditions is approximately 10-11 kV, and peak current is
approximately 34 A and 42 A (at 7.5 mJ/pulse and 10.0 mJ/pulse, respectively).
Figure 4.3: Voltage, current, and coupled energy waveforms for Pulser #2. 3% H 2-air
mixture, T0= 300 K, P = 100 Torr, pulser input DC voltage (a) Vin=490 V, positive polarity,
(b) Vin=490 V, negative polarity, (c) Vin=530 V, positive polarity, and (d) Vin=530 V,
negative polarity; pulse repetition rate 10 Hz.
During breakdown, applied voltage drops by a few hundred Volts when the
discharge current begins to rise. Current pulse duration is approximately 50 ns FWHM. As
expected, negative polarity pulse waveforms in all gas mixtures are close to the positive
polarity waveforms, since the electrode assembly is symmetric.

138
Hz pulse repetition rate. Camera gate 1 µs. Reference image of the electrodes (c) with
grounded electrode is on the left, high voltage electrode is the right.
Figures 4.4 and 4.5 show images of broadband plasma emission generated in 1-5%
H2-air mixtures at 100 Torr, at coupled energy of 7.5 mJ/pulse and 10.0 mJ/pulse,
respectively. Both single-pulse (positive polarity) and 100-pulse average images are
shown. In the images, the grounded electrode is on the left, the discharge is operated at a
10 Hz repetition rate, and the camera gate is 1 μs, incorporating the entire discharge current
pulse. A reference image of the electrodes with no plasma is shown in Figure 4.4(c).
139
Hz repetition rate. Camera gate 1 µs. Grounded electrode is on the left, high voltage
electrode is the right.
Plasma images taken in air at a pulse repetition rate of 60 Hz, with a short camera
gate, 100 ns, shown in Figure 4.6, are essentially identical to the ones shown in Fig. 4.4,
since plasma emission decays very rapidly after the discharge pulse, within a few tens of
nanoseconds. For all cases shown, the plasma filament is diffuse, with a diameter of ~3
mm. Plasma emission intensity near the filament centerline is somewhat lower than in the
periphery, likely due to the 1 mm laser beam access holes in the electrodes. Increasing the
pulse energy from 7.5 mJ/pulse to 10.0 mJ/pulse results in the onset of somewhat brighter,
partially constricted filaments, which are “averaged out” in 100-pulse average images.
140
Figure 4.6: Single pulse (a) and 100-pulse average (b) ICCD images ns of broadband
plasma emission in air for Pulser #2. T0= 300 K, P = 100 Torr, Vin=490 V, 60 Hz pulse
repetition rate. Camera gate 100 ns. Sets of images are taken for both positive and negative
polarity pulses. Grounded electrode is on the left, high voltage electrode is the right.
Figure 4.7 plots voltage, current, and coupled energy waveforms from Pulser #1 in
a 3% H2-air mixture at 60 Torr (a) and 100 Torr (b). Pulse energy coupled to the plasma
at these conditions is 11.6 mJ and 6.3 mJ, respectively, is calculated as a sum of energy
coupled by the primary pulse and smaller (secondary) pulse. Discharge waveforms and
coupled energy in other H2-air mixtures (1% H2 for 60 Torr and 1% and 5% H2 for 100
Torr) are similar. At these conditions, Pulser #1 generates single polarity pulses, with pulse
peak voltage of 6-7.5 kV and peak current of approximately 40 A and 15 A (at 11.6
mJ/pulse and 6.3 mJ/pulse, respectively). A few hundred nanoseconds later, a secondary
voltage pulse (pulse peak voltage of 5 and 2.5 kV, respectively) is detected. Current pulse
duration is approximately 100 ns FWHM, for both primary and secondary pulses.
141
Figure 4.7: Voltage, current, and coupled energy waveforms for Pulser #1. Taken in a 3%
H2-Air mixture, at T0= 300 K, input DC voltage Vin=420 V, with a discharge pulse
repetition rate of 10 Hz. (a) P= 60 Torr and (b) P= 100 Torr.
Figures 4.9 and 4.10 show both single pulse and 100-pulse averaged broadband
plasma emission, taken in 1-5% H2-air mixtures at 60 Torr and 100 Torr. The grounded
electrode is again shown on the left, and all images were taken with a 1µs gate to detect
both the primary and the secondary pulses generated by Pulser #1. From single shot images
taken at 60 Torr and 100 Torr, it can be seen that the filament becomes partially constricted,
generating a brighter central region in the plasma. This effect is averaged out in 100-pulse
average images, which also show that the filament has a somewhat lower intensity near the
electrode centerline.
142
right.
right.
143
These images are taken at a pulse repetition rate of 10 Hz, to match the pump laser
repetition rate used for LIF measurements. CARS data taken in the afterglow of the
discharge generated by Pulser #2 are taken at a higher repetition rate of 60 Hz, which
improves plasma stability. However, CARS data in the afterglow of the discharge
generated by Pulser #1 are taken at a repetition rate of 10 Hz, to prevent discharge filament
constriction.
4.3. Laser Induced Fluorescence Experimental Setup
The schematic of the OH LIF apparatus is shown in Figure 4.10. For this, the 532
nm output from a Nd:YAG laser (Continuum Model Powerlite 8010) is used to pump a
tunable dye laser (Continuum ND6000), generating output at 615 nm, which is frequency-
doubled in a Type-I BBO crystal. To obtain the best beam quality and maximum
conversion, the crystal is pumped with ~30 mJ of laser energy, resulting in ~20 µJ of 308
nm second harmonic output. Two dichroic mirrors are used to separate the 308 nm second
harmonic beam from the 615 nm fundamental beam, creating a delay line (see Fig. 4.10).
In the delay line, a half waveplate and polarizer combination is used to reduce the energy
of the 308 nm beam to ~5 µJ, within the linear LIF regime, while maintaining a good beam
profile. A Pellin-Broca prism then separates the second harmonic from the remaining
fundamental, and an identical prism is set symmetrically, to compensate for alignment
changes of the 308 nm beam. The 308 nm beam is then focused into the six-arm cross
discharge cell, through the holes in the electrodes using a f=500 mm lens, as shown in Fig.
4.1. The measurement volume is kept in the intermediate field of the focused pump beam
144
to reduce laser power density and to prevent any non-linear effects or laser induced
breakdown.
Figure 4.10: Schematic diagram of the OH LIF experimental apparatus. The 615 nm dye
laser output is frequency doubled using a Type I BBO crystal. The 308 nm beam is passed
through a half waveplate and polarizer combination to control the laser pulse energy. Both
beams are sent through a set of Pellin-Broca prisms to separate the LIF beam from the dye
laser beam before it is focused in the test section using a f= 500 mm UV lens.
The fluorescence signal is collected at 900, through a 2” UV fused silica window.
A band pass filter (UG11 Colored Glass UV-Passing Filter, 275 - 375 nm), is placed in
front of the camera lens to filter out plasma emission and scattered light from the
fundamental beam from the fluorescence signal of the OH A 2Σ+ → X2Π (v′=0,v′′=0)
emission band at ~ 308 nm. The use of resonant LIF, i.e. collecting fluorescence at the
same wavelength as the that of the pump laser, increases interference due to pump laser
145
beam scattering off the electrodes. This effect is reduced by gating the ICCD camera; for
these measurements, the camera gate was set to open 3 ns after the end of the laser pulse,
significantly reducing laser beam scattering noise, and the gate width was 200 ns. The
wavelength of the laser beam pumping the OH absorption transition, R 2(2.5), λ = 307.8
nm, is monitored by a High Finesse, Angstrom WS/6 wavemeter, and the laser pulse
energy, 3-6 μJ/pulse, is measured by a Scientech P09 power meter. In all LIF
measurements, potential photodissociation was minimized by keeping the laser energy in
the LIF linear regime (≤ 10 μJ/pulse), which was determined by varying laser pulse energy
and collecting fluorescence signal in the plasma afterglow 100 µs after the discharge pulse,
as shown in Figure 4.11.
Figure 4.11: OH fluorescence signal versus laser pulse energy in 3% H 2-air mixture excited
by a ns pulse discharge. P=100 Torr, T0=300 K, Pulser #1, coupled discharge pulse energy
6.3 mJ/pulse, and delay time between the discharge pulse and the laser pulse is 100 µs.
Camera gate 200 ns, camera gain 100, 20 laser shot accumulation. All subsequent data are
taken at laser pulse energy of 5 µJ.
146
The fluorescence signal is put on an absolute scale by measuring the fluorescence
quenching rate, the convolution of the laser lineshape and the absorption transition
lineshape (the convolution integral), and the efficiency of the signal collection optical
system, i.e. the parameters that control the intensity of the measured LIF signal, as specified
by Equation 2.18, repeated below,
𝑛 = ( )
. (2.18)
The total fluorescence quenching rate, (𝐴 + 𝑄), is measured directly for each set
of experimental conditions, at t=100 µs after the discharge pulse, when the fluorescence
signal is near maximum. For this, the ICCD camera is gated for a 5 ns duration. The camera
gate is shifted relative to the laser pulse, using a Stanford Research Systems Digital Delay
Generator (DG654) as the master clock, and 300 laser shots are accumulated at each time
delay. The quenching rate was inferred from the best exponential fit to the LIF signal decay,
as shown in Figure 4.12.
147
Figure 4.12: OH LIF signal vs. time delay after the laser pulse, used for the fluorescence
quenching rate measurements. 3% H2-air mixture, P=100 Torr, T0=300 K, Pulser #2,
Ecoupled=7.5 mJ/pulse. Camera gate 5 ns, every data point is averaged over 300 laser shots.
Equation used for best fit is also shown.
To quantify the quenching rate variation in time in the afterglow, it was measured
at three different time delays after the discharge pulse in a mixture of 3% H 2-air at P = 100
Torr and T0 = 300 K, and the results are summarized in Table 4.1.
Table 4.1: Quenching rate variation over time after a discharge pulse.
Input DC (A+Q) (s-1)

Pulser
Voltage (V) t = 30 µs t = 100 µs t = 300 µs
#2 530 7.9∙107 7.02∙107 6.5∙107
148
As can be seen from Table 4.1, the quenching rate for Pulser #2 after the discharge
pulse varied by about 15%. Varying H2 mole fraction in the mixture from 1% to 5%, at t=
100 µs, resulted in quenching rate variation by 8%. Therefore, in all subsequent
experiments, the quenching rate was measured only once, at t= 100 µs in 3% H2-air
mixture. Table 4.2 lists quenching rates measured for both pulsers, at different pressures
and different coupled energies.
Table 4.2: Quenching rates used in LIF data processing
Pulser Pressure (Torr) Input DC (A+Q) (s-1)

Voltage (V)
#2 100 490 1.1∙108
#2 100 530 7.02∙107
#1 60 420 2.51∙107
#1 100 420 6.94∙107
The convolution integral has been measured in several H 2-air mixtures at t=100 µs
after the discharge pulse, by scanning the laser wavelength across the OH absorption
transition, R2(2.5), and measuring the fluorescence signal. A total of eight data points at P
= 60 Torr, and ten data points at P = 100 Torr, taken during the laser scan are used to
calculate the convolution integral, as shown in Figure 4.13. During the rest of the
measurements, the laser was parked at the center of the absorption transition.
149
Figure 4.13: Experimental data points used for calculation of the convolution integral for
the OH LIF absorption transition. 3% H2-air mixture, a) P=60 Torr, T0=300 K, Pulser #1,
Ec=11 mJ/pulse b) P = 100 Torr, T0= 300 K, Pulser #2, Ec= 7.5 mJ. The data are
approximated by a Gaussian profile, to obtain the convolution integral, with FWHM = 0.5
cm-1 and 0.47 cm-1, respectively.
Finally, the signal collection efficiency is determined via Rayleigh scattering
calibration. For this, the cell is filled with nitrogen, and the laser scattering signal off
nitrogen molecules is collected by the same optical system used to collect the fluorescence
signal. Sample Rayleigh scattering images taken at the highest pressure of nitrogen used
(P = 685 Torr) and at the lowest pressure (P = 334 Torr), are shown in Figure 4.14.
150
Figure 4.14: Rayleigh scattering images taken in nitrogen at a) 685 Torr and b) 334 Torr,
at T=300 K. 100 laser shot accumulation, 200 ns camera gate. Rayleigh signal collected
from the central 4 mm in the axial direction and integrated along a transverse region with
a height of ~1 mm, where 95% of the Rayleigh signal is generated.
To remove laser scattering off the electrodes during Rayleigh calibration
measurements, the test cell was translated away from the camera, such that the laser beam
passed in front of the ceramic bracket holding the electrodes. The Rayleigh scattering
signal was integrated over the same axial distance as the LIF signal, ~4 mm, and over ~ 1
mm in the transverse direction, such that <95% of the total Rayleigh scattering signal was
accounted for in the integration. By keeping the laser pulse energy constant and varying
the nitrogen pressure in the cell, the scattering signal intensity is measured versus the
number density, exhibiting a linear trend shown in Figure 4.15.
151
Figure 4.15: Rayleigh scattering calibration obtained from ICCD camera images such as
those shown in Fig. 4.15, at P= 334-685 Torr in nitrogen. Signal uncertainty is 30%,
controlled by laser scattering noise off the windows.
The slope of this line provides the overall signal collection efficiency used for
absolute calibration of the LIF signal, as specified in Equation 2.16. The error associated
with the calibration is 30%, almost entirely due to signal-to-noise caused by laser scattering
from the cell window surfaces. Finally, the gas temperature necessary to calculate the
Boltzmann factor in the expression for the fluorescence signal intensity, given by Equation
2.13, is taken from the CARS measurements discussed in Section 4.5.
4.4. Coherent Anti-Stokes Raman Spectroscopy Setup
Coherent Anti-Stokes Raman Spectroscopy (CARS) is an inelastic, coherent light-
scattering process used for thermometry and measuring vibrational/rotational level
populations of major species of interest. CARS produces a coherent signal beam, with high
152
signal-to-noise, high resolution in both space and time, and effective rejection of
background optical interference from plasma emission and laser beam scattering.
In CARS, two lasers beams, the pump and Stokes, are tuned such that their
frequency difference (𝜔 −𝜔 ) matches a vibrational and/or rotational transition
frequency of a molecule. The pump/Stokes pair of electromagnetic waves, excites the
molecules by causing the electron clouds to oscillate, at the difference frequency. The
molecules excited by the pump and Stokes beam pair oscillate in phase, creating a Raman
coherence. When a third electromagnetic wave, the probe laser beam, 𝜔 , passes
through the excited medium, the forward scattering from the oscillating molecules causes
constructive interference, producing a fourth, blue shifted electromagnetic wave, the anti-
Stokes signal beam, 𝜔 . The anti-Stoke beam has a frequency of 𝜔 =
𝜔 −𝜔 +𝜔 , and a CARS energy level diagram is shown in Figure 4.16.
153
Figure 4.16: CARS energy level diagram showing excitation of the molecule to a virtual
state and anti-Stokes signal beam generation.
The phase matching condition (i.e. the photon momentum conservation), also
shown in Fig. 4.16, determines the propagation direction of the resulting anti-Stokes signal
beam. For these experiments, a collinear CARS beam geometry was employed, such that
the pump, Stokes, and anti-Stokes beams are parallel and directed along the centerline of
the electrodes, providing maximum signal. Since the pump and Stokes beams are parallel,
the CARS interaction region length may be fairly significant, and is controlled by the laser
beam diameters prior to focusing and by the focal distance of the lens.
154
Figure 4.17: Experimental diagram of broadband collinear CARS diagnostics. The 532 nm
and 607 nm beams are focused halfway between the discharge electrodes. A set of dichroic
mirrors (R473, T532, T607) after the test cell separates the pump/probe and Stokes beams
from the CARS signal beam. The remaining pump beam light is filtered out by the band
pass filter in front of the spectrometer slit.
The CARS apparatus schematic is shown in Figure 4.17. The second harmonic
output from an injection seeded, externally triggered Nd:YAG laser (Surelite, SLIII-10,
pulse energy up to 400 mJ, linewidth with and without injection seeding 0.015 cm -1 and
0.4 cm-1, respectively) is separated with a 90:10 beam splitter, with the first beam pumping
a custom built broadband dye laser [83] and the second acting as both pump beam and
probe beam. The Stokes beam is centered at 605 nm with a full width half of maximum
155
(FWHM) of 3 nm, and is directed through the cell, passing through a dichroic mirror (R532
nm, T607 nm) which combines the pump and probe beams (see Fig. 4.17). At this point,
the laser intensity is not high enough to generate the signal. All three beams are focused
through the cell using a spherical lens (f=150 mm). The length of the interaction region
was measured by translating a glass slide (~1 mm thick) through the focal region of the
beams between the discharge electrodes and measuring the intensity of the non-resonant
signal (NRS). The result, shown in Figure 4.18, demonstrates that 90% of the CARS signal
originates over a distance of approximately 4 mm.
Figure 4.18: Axial distribution of the non-resonant background signal from a microscope
slide, illustrating spatial resolution of the collinear CARS measurements (95% of the signal
originates from a region 3.7 mm long).
Once the signal is generated, the beams are passed through a set of dichroic mirrors
to separate the pump and Stokes from the anti-Stokes beam. Finally, the signal beam,
156
centered at 473 nm, is focused into a spectrometer slit with the width of 300 µm. The band
pass filter (FB470-10 Filter, 460-480 nm) is used to remove the residual pump and Stokes
beams, as well as emission from the plasma. The spectrometer (Action SP300i with Andor
750 with Andor CCD camera) used a 3600 l/mm grating, employed for N 2 vibrational
population measurements, and to resolve rotational structure of N 2(v=0) band to infer the
rotational temperature. Without injection seeding, the spectral resolution of the
measurements is dominated by the spectrometer’s instrument function, approximately 0.9
cm-1. To improve spectral resolution for more accurate rotational-translational temperature
inference, some of the CARS spectra were taken with the injection seeder on, resulting in
a higher resolution of 0.4 cm-1 FWHM.
After plasma emission background subtraction, the experimental spectra are
divided by the non-resonant signal spectrum obtained from 200 Torr of argon in the test
cell, and a square root of the normalized intensity is taken. Nitrogen vibrational level
populations and “first level” N2 vibrational temperature are inferred from the integrated
area under the vibrational band peaks in the square root intensity spectra, which is
proportional to the difference of vibrational level populations. The rotational-translational
temperature is inferred by fitting the experimental square root intensity N 2(v=0) band
spectra with synthetic CARSFIT spectra [84].
4.5. CARS Results and Discussion
Figure 4.19 plots two broadband N2 CARS spectra in a 3% H2-air mixture taken
without Nd:YAG laser injection seeding, at a coupled discharge pulse energy of 8.3 mJ
157
and 10.2 mJ, respectively, 200 µs after the discharge pulse. Nitrogen vibrational bands
v=0-2 are readily identified, with rotational structure of the vibrational bands partially
resolved.
Figure 4.19: Broadband CARS spectra in a 3% H 2-air mixture at 100 Torr, excited by a ns
pulse discharge sustained by Pulser #2, taken with Nd:YAG laser injection seeding off, for
a) coupled pulse energy of 8.3 mJ, and b) coupled pulse energy of 10.2 mJ. Time delay
after the discharge pulse is 200 μs. N2 (v=0-2) vibrational bands can be identified in the
spectra.
158
Figure 4.20 plots a CARS spectrum of N2(v=0) band in room air, taken with
injection seeding turned on. The temperature inferred from the best fit CARSFIT spectrum
is 292 ± 12 K, with the uncertainty defined as the temperature range where the least squares
error calculated by CARSFIT is within 20% of the minimum value, which corresponds to
the best fit temperature.
N2(v=0) band in room air. Experimental spectra are taken with Nd:YAG laser injection
seeding turned on. Best fit rotational temperature inferred using CARSFIT is 292 ± 12 K.
Figure 4.21 plots a CARS spectrum of N2(v=0) band, taken with injection seeding
turned on in a 3% H2-air mixture. The noise in the baseline is due to several factors,
including interference from stray light, interference caused by the dark current from the
camera, the readout noise from the camera chip, and EMI from the pulser. Background was
subtracted from each spectrum before the temperature was inferred, to mitigate these
159
effects. The temperature inferred from the CARS spectra at 400 µs after the discharge pulse
at coupled pulse energy of 8.3 mJ/pulse is T= 528 ± 45 K (see Fig. 4.21(a)), and
temperature inferred 600 µs after the discharge pulse with coupled pulse energy of 10.2
mJ/pulse is T= 557 ± 50 K (see Fig. 4.21(b)).
N2(v=0) band in air at 100 Torr, excited by a ns pulse discharge sustained by Pulser #2,
with a) coupled pulse energy 8.3 mJ at a time delay after the discharge burst of 400 µs and
b) 10.2 mJ with a time delay of 600 µs. Experimental spectra are taken with Nd:YAG laser
injection seeding turned on. Best fit rotational temperatures inferred using CARSFIT are
528 ± 45 K and 557 ± 50 K, respectively.
160
Figure 4.22 plots the rotational-translational temperature inferred from the
rotational structure of N2(v=0) band CARS spectra, such as shown in Figure 4.21, as well
as the “first level” N2 vibrational temperature,
E10
Tv  (4.1)
 f 
ln  0 
 f1 
versus time delay after the discharge pulse in air and hydrogen-air mixtures, for coupled
pulse energies of 8.3 mJ (see Fig. 4.20(a)) and 10.2 mJ (see Fig. 4.20(b)). In Equation 4.1,
E10 = 3353 K is the energy difference between vibrational levels v=0 and v=1, and f0 and
f1 are their relative populations, respectively.
161
discharge generated by Pulser #2 at 100 Torr, inferred from the CARS spectra such as
shown in Fig. 4.21. Coupled discharge pulse energy is 8.3 mJ (a) and 10.2 mJ (b).
In Figure 4.22(a), the N2 vibrational temperature in air increases from Tv0 ≈ 1200
K at t = 10 µs to Tv ≈ 1900 K at t=100 μs. A similar trend is seen when 1% and 3% H2 is
162
added to the air. The vibrational temperature for 1% H2 is initially Tv0 ≈ 1300 K at t = 20
µs and increases to Tv ≈ 1900 K at t=90 μs. Similarly, for 3% H2, the initial vibrational
temperature is Tv0 ≈ 1000 K at t = 10 µs, increasing to Tv ≈ 1700 K at t=100 μs. Figure
4.22(b) shows similar trends. In air, initial vibrational temperature is slightly higher than
at the lower coupled energy, Tv0≈ 1300 K at t = 20 µs; however, vibrational temperature
increases to nearly the same peak value, Tv = 1900 K at t = 100 µs. In a 1% H2-air mixture,
Tv0 = 1300 K at t = 20 µs, increasing to Tv = 1900 K at t = 100 µs. In a 3% H2-air mixture,
the initial vibrational temperature is lower, Tv0 = 1000 K at t = 20 µs, increasing to a
somewhat smaller peak temperature, Tv = 1600 K, at t = 200 μs after the discharge pulse.
Figure 4.23 shows the results of Tv(N2) and temperature measurements in air and
H2-air plasmas sustained by Pulser #1, at a lower coupled energy of 6.3 mJ/pulse. The
trends exhibited at these conditions are similar to the results shown in Fig. 4.22, although
peak vibrational and translational-rotational temperature values are lower.
163
discharge generated by Pulser #1 at 100 Torr, inferred from the CARS spectra. Coupled
discharge pulse energy is 6.3 mJ/pulse.
As shown by kinetic modeling calculations [82,85], the initial vibrational excitation
of nitrogen in air is caused by electron impact,
𝑁 (𝑣 = 0) + 𝑒 → 𝑁 (𝑣 = 8) + 𝑒 (4.2)
The increase in Tv(N2) observed at all sets of conditions tested (see Figs. 4.22 and
4.23) is controlled by the downward N2-N2 vibration-vibration (V-V) energy exchange,
reducing high N2 vibrational level populations and increasing N2(v=1) population [31].
𝑁 (𝑤) + 𝑁 (𝑣 = 0) → 𝑁 (𝑤 − 1) + 𝑁 (𝑣 = 1) (4.3)
After reaching a peak vibrational temperature at t ≈ 200 µs, vibrational relaxation
of N2 by O atoms becomes the dominant process. Vibrational-translational (V-T) relaxation
of N2 by O atoms occurs on timescales of t = 200-700 µs.
𝑁 (𝑣) + 𝑂 → 𝑁 (𝑣 − 1) + 𝑂 + ℎ𝑒𝑎𝑡 (4.4)

164
The rate coefficient of N2 vibrational relaxation by O atoms at room temperature is
𝑘 (𝑣 = 1 → 0) = 3 ∙ 10 (cm3s-1) [25]. In ns pulse discharges in air, O atoms are
produced both by electron impact and by collisional quenching of electronically excited N 2
molecules.
𝑂 +𝑒 → 𝑂+𝑂+𝑒 (4.5)
𝑁 ∗ (𝐴, 𝐵, 𝐶, 𝑎) + 𝑂 → 𝑁 (𝑋) + 𝑂 + 𝑂 (4.6)
At longer time delays, t ≥ 700 µs, N2 vibrational levels decay mainly by radial
diffusion from the plasma filament into the surrounding non-excited gas mixture.
Kinetics controlling N2 vibrational populations in hydrogen-air mixtures are similar
since V-T and V-V relaxation of nitrogen by H2, e.g.
𝑁 (𝑣 = 1) + 𝐻 → 𝑁 (𝑣 = 0) + 𝐻 (4.7)
𝑁 (𝑣 = 1) + 𝐻 (w = 0) → 𝑁 (𝑣 = 0) + 𝐻 (𝑤 = 1) (4.8)
→
is relatively slow, 𝑘 (𝑣 = 1 → 0) = 4 ∙ 10 [86] and 𝑘 ( ) = 1 ∙ 10
→
(cm3s-1) [87]. Although H atoms formed in H2-air plasmas by electron impact and by
quenching of electronically excited N2 molecules,
𝐻 +𝑒 → 𝐻+𝐻+𝑒 (4.9)
𝑁 ∗ (𝐴, 𝐵, 𝐶, 𝑎) + 𝐻 → 𝑁 (𝑋) + 𝐻 + 𝐻 (4.10)
𝑁 (𝑣 = 1) + 𝐻 → 𝑁 (𝑣 = 0) + 𝐻 (4.11)
may also contribute to the net rate of N2 V-T relaxation, the rate coefficient of this process
has not been measured. In addition, the H atom number density predicted by kinetic
modeling calculations (discussed below) is much lower compared to that of O atoms, such
165
that N2 V-T relaxation by O atoms is the dominant process. Therefore peak vibrational
temperature in H2-air mixtures is not very different from that in air (see Figs. 4.23-4.25).
In air for both discharge coupled energies, 8.3 mJ/pulse and 10.2 mJ/pulse, rotational-
translational temperature increases from T0= 300 K to T≈ 400 K, at t= 100 ns - 1 µs after
the discharge pulse. Previous measurements in air at similar conditions [82] show this is
due to rapid energy thermalization on the time scale of t ~ 0.1 – 1 μs [82], and previous
kinetic modeling calculations [85] show that rapid heating in air is dominated by quenching
of electronically excited N2 molecules by molecular oxygen.
𝑁∗ + 𝑂 → 𝑁 + 𝑂 + 𝑂 (4.12)
The temperature remains quasi-steady until t = 200 µs, when it increases again to
T = 650 K. This secondary rise is due to the heat released by V-T relaxation of nitrogen
by O atoms in reaction of Eqn. 4.3. The effect is readily apparent in Fig. 4.22(b), at the
coupled energy of 11 mJ.
In hydrogen-air mixtures, the temperature at t = 20 µs is approximately 400 K,
before a gradual temperature rise at t = 50 - 900 μs increases it to approximately 600 K,
due to both V-T relaxation of nitrogen and additional chemical energy release during
hydrogen oxidation in reactions of radical species generated by the plasma [12]. The
temperature slowly decreases by radial diffusion at t > 1 ms.
Summarizing, CARS measurements of N2 vibrational populations and temperature
demonstrate that strong vibrational nonequilibrium is sustained after the discharge pulse in
air, 1% H2-air, and 3% H2-air mixtures, up to at least Tv ≈ 1900 K and T ≈ 600 K at t=200
166
μs. Time-resolved temperature measured after the discharge pulse in air was used to infer
absolute OH number density from LIF measurements in H 2-air mixtures, discussed below.
4.6. LIF Results and Discussion
Figures 4.24 and 4.25 show OH LIF images taken in 1-5% H2-air mixtures at P=100
Torr, at different delay times after the discharge pulse, and at two different values of
discharge pulse energy generated by Pulser # 2, 7.5 mJ (see Fig. 4.24) and 10.0 mJ (see
Fig. 4.25). Similar to plasma emission images shown in Fig. 4.5, the grounded electrode is
on the left. These images are accumulated over 100 laser shots, at a camera gate of 200 ns.
From the figures, it can be seen that laser scattering off the electrodes is a minor factor.
energy E= 7.5 mJ/pulse (a) 1% H2-air, (b) 3% H2-air, (c) 5% H2-air.
167
energy E= 10 mJ/pulse (a) 1% H2-air, (b) 3% H2-air, (c) 5% H2-air.
Time evolution of the OH LIF signal intensity distribution along the discharge
filament centerline (integrated over the transverse region with the height of 1 mm) at these
conditions is shown in Figure 4.26, for coupled energies of (a) 7.5 mJ and (b) 10 mJ,
respectively. At short delay times after the discharge pulse, t ~ 10 μs, the LIF signal peaks
near the grounded electrode (at X=0 mm), with a lower intensity peak also appearing near
the high-voltage electrode at t ~ 10-100 μs (see Fig. 4.26(a)). This effect becomes more
pronounced as the discharge pulse energy is increased from 7.5 mJ to 10.0 mJ, as can be
seen from comparison of Fig. 4.26(a) and 4.26(b). This behavior is consistent with recent
TALIF H atom measurements in ns pulse discharges in H2-He and H2-Ar mixtures at
similar conditions [11], indicating that the specific energy loading in the plasma near the
electrodes is higher compared to the rest of the discharge gap. At t > 100 μs, these maxima
become less pronounced and gradually merge.
168
accumulation images, with a 200 ns camera gate, taken in a 3% H 2-air mixture with coupled
discharge pulse energy 7.5 mJ/pulse (a) and a 5% H2-air mixture with coupled discharge
pulse energy of 10 mJ/pulse (b), excited by a ns pulse discharge sustained by Pulser #2, at
P= 100 Torr and T0= 300 K.
Figures 4.27 and 4.28 show OH LIF images taken in 1 and 3% H2-air mixtures at
60 Torr and 1-5% H2-air mixtures at 100 Torr, respectively, in plasmas generated by Pulser
#1 at coupled energies of 11.6 mJ and 6.3 mJ. Again, the images show slight asymmetry
in OH LIF signal at the grounded electrode, on the left, which is more significant at higher
pressures, as shown in Figure 4.28.
169
energy E= 11.6 mJ/pulse (a) 1% H2-air, (b) 3% H2-air, (c) 5% H2-air.
energy E= 6.3 mJ/pulse . (a) 1% H2-air, (b) 3% H2-air, (c) 5% H2-air.
Figure 4.29 plots centerline signal distribution spatially integrated in the same way
as in Fig. 4.26, versus time delay after the discharge pulse generated by Pulser #1 at coupled
energies of 11.6 mJ and 6.3 mJ, in 1 and 3% H2-air mixtures at 60 Torr (a) and 1-5% H2-
air mixtures at 100 Torr (b), respectively. OH initially peaks near the grounded electrode.
170
At 60 Torr, this effect is not nearly as pronounced as at 100 Torr, as can be seen by
comparing Fig. 4.29(a) and 4.29(b). At 100 Torr and t= 10 µs, the signal peak near the
grounded electrode exceeds that in the rest of the discharge gap by about an order of
magnitude. At longer time delays, t = 90 – 400 µs, the OH peak near the grounded electrode
decays and the peak near the high-voltage electrode becomes more pronounced, leading to
a less asymmetric distribution. After t = 400 µs, both maxima nearly disappear, and the
signal decreases over a few hundred microseconds.
accumulation images, with a 200 ns camera gate, taken in a 3% H 2-air mixture at (a) P= 60
Torr and T0=300 K, with coupled discharge pulse energy of 11.6 mJ/pulse and (b) P= 100
Torr, T0= 300 K, with coupled discharge pulse energy of 6.3 mJ/pulse, excited by a ns
pulse discharge sustained by Pulser #1.
Figure 4.30 plots the absolute OH number density vs. time delay after the discharge
pulse generated by Pulser #2 in three different H2-air mixtures, at discharge pulse energy
171
of 7.5 mJ (a) and 10.0 mJ (b). Figure 4.31 plots the absolute OH number density vs. time
delay after the discharge pulse generated by Pulser #1 taken in two different H 2-air
mixtures at 60 Torr (a) and three different H2-air mixtures at 100 Torr (b), with coupled
discharge pulse energies of 11.6 mJ and 6.3 mJ, respectively. OH number density is
inferred from the LIF signal distributions, such as those shown in Figs. 4.26 and 4.29,
integrated over the central region of the discharge filament ~4 mm long, and calibrated
using Rayleigh scattering. Gas temperature in the discharge afterglow, used to calculate
absolute OH number density, is taken from CARS measurements in hydrogen-air mixtures
at the same discharge pulse energy, as discussed above (see Figs. 4.22 and 4.23). The
combined uncertainty of the inferred OH number density is estimated to be ±30%, with
uncertainty in the Rayleigh scattering calibration being the dominant factor.
172
pulse discharge (Pulser #2): (a) P= 100 T, T0= 300 K, coupled discharge pulse energy is
7.5 mJ/pulse, and (b) P = 100 Torr, T0 = 300 K, coupled discharge pulse energy is 10
mJ/pulse.
From Fig. 4.30(a), the OH number density in the 1% H2-air mixture for both
coupled pulse energies increases monotonously by about a factor of two at t = 10 – 300 μs,
173
before decaying at t = 300 μs – 1 ms. At these conditions, the estimated characteristic time
for radial diffusion out of the filament is approximately 1 ms, such that [OH] decay is likely
due to OH chemical reactions in the afterglow. No measurable OH fluorescence signal was
detected at t > 1 ms. Similar behavior is observed in 3% and 5% H 2 mixtures, although at
these conditions OH number density is significantly higher and peaks sooner, at t = 150 μs
and t = 100 μs, respectively, followed by a more rapid decay. Increasing the coupled
discharge pulse energy by 25% (from 7.5 mJ/pulse to 10 mJ/pulse) results in an increase
of OH number density produced by a factor of 2.5. Figure 4.30(b) shows that the OH
number density behaves in a similar way as Fig. 4.30(a), although the OH number density
peaks faster, at t= 150 µs in a 1% H2-air mixture and t= 100 µs and t= 90 µs in the 3% and
5% H2-air mixtures, respectively. After the peaks, no secondary transient maxima of OH
number density is detected. Peak OH number density in 1% H2 and 3% H2 mixtures,
achieved at t = 200-300 μs, approximately coincides in time with peak N 2 vibrational
temperatures in these mixtures, Tv ≈ 1700-1900 K at t=200 μs (see Fig. 4.22(b)).
174
pulse discharge (Pulser #1) a) P= 60 T, T0= 300 K, coupled discharge pulse energy is 11.6
mJ/pulse and b) P = 100 Torr, T0 = 300 K, coupled discharge pulse energy is 6.3 mJ/pulse.
From Figure 4.31(a), OH number density produced in both 1% and 3% H 2-air
mixtures at 60 Torr increases by a factor of two at t= 10-100 µs, before decaying at t= 200
µs - 1ms. Peak OH number densities are 3.3∙1014 cm-3 and 1.4∙1014 cm-3, in 3% H2-air and
175
1% H2-air mixtures, respectively. Peak OH number density is higher at 60 Torr than 100
Torr due to a higher coupled discharge pulse energy, 11.6 mJ/pulse and 6.3 mJ/pulse,
respectively. At 100 Torr the OH number density produced in all three H 2-air mixtures
increases by a factor of two before decaying, as can be seen from Fig. 4.31(b). In a 1% H2-
air mixture this increase occurs at t= 10 – 300 µs. The OH number density produced in the
3% and 5% H2-air mixtures peaks more rapidly, at t= 150 µs and t = 100 μs, respectively
(see Fig. 4.31(b)), decaying by t ~ 1ms.
The kinetic model used in the present work, discussed in detail in Ref. [40],
incorporates two-term Boltzmann equation for plasma electrons; electron energy equation
and heavy species energy equation; electron impact processes of vibrational excitation,
electronic excitation, dissociation, and ionization; state-specific vibrational kinetics of
nitrogen and hydrogen, including vibration-vibration (V-V) and vibration-translation (V-
T) processes; reactions of excited electronic states of N 2 and O; and chemical reactions of
species in the ground electronic states. Kinetic model of hydrogen-air plasma is the same
as in Ref. [40]. Plasma filament diameter used in the model, 2.4-2.6 mm FWHM, is
estimated from plasma emission intensity (see Figures 4.4 and 4.5), as shown in Figure
4.32.
176
Figure 4.32: Radial distribution of UV/visible plasma emission intensity in hydrogen- air
mixtures at P = 100 Torr, T0 = 300 K. Pulser #2 coupled discharge pulse energies are (a)
7.5 mJ and (b) 10 mJ/pulse, respectively.
Radial diffusion of species out of the discharge filament is accounted for as a 0-D
correction, using diffusion coefficients calculated as in Ref. [76]. Pressure variation in the
filament due to sub-acoustic time scale heating and subsequent gas dynamic expansion is
incorporated as discussed in Ref. [40]. Time-resolved discharge power, used as one of the
model entries, is obtained from experimental current and voltage waveforms (see Fig. 4),
with power waveform scaled to account for the cathode voltage fall, which significantly
reduces the energy coupled to the plasma outside of the cathode layer [40]. Figures 4.33-
4.38 compare kinetic modeling predictions with the experimental results in air and H 2-air
mixtures, excited by a ns pulse discharge generated by Pulser #2, at coupled pulse energy
of 7.5 mJ (see Figs. 4.33 - 4.35) and 10 mJ (see Figs. 4.36 - 4.38).
177
and H2-air mixtures. Coupled discharge pulse energy is 7.5 mJ/pulse.
Figure 4.33 compares experimental and predicted N2 vibrational temperature (a)
and gas temperature (b) vs. time delay after the discharge pulse in air and in 1% and 3%
178
H2-air mixtures. It can be seen that the model predictions are in good agreement with the
experimental data, indicating that the model correctly reproduces the effect of dominant
energy thermalization processes over a wide range of time scales (t ~ 1 μs – 10 ms). As
discussed in Refs. [7,31], N2 vibrational temperature rise after the discharge pulse (at t ~ 1
– 200 μs) is caused by the “downward” N2-N2 V-V exchange, which depopulates high N2
vibrational levels and increases populations of vibrational levels v=1 and v=2, thereby
increasing vibrational temperature defined by Eqn. 4.3. At t > 100-200 μs, vibrational
temperature decays primarily due to V-T relaxation by O atoms. The model also reproduces
transient gas temperature reduction at t ~ 2 – 10 μs, caused by gas dynamic expansion of
the filament following its heating on sub-acoustic time scale [88], as well as subsequent
temperature rise caused by N2-N2 V-V exchange and N2-O V-T relation (in air) and by
chemical energy release due to oxidation of hydrogen in plasma chemical reactions in H 2-
air. At t > 1 ms, temperature decreases due to radial diffusion.
179
pulse energy is 7.5 mJ/pulse.
180
Figure 4.35: Predicted radical species number densities vs. time delay after the discharge
pulse generated by Pulser #2 in a 3% H2-air mixture compared to experimental data (shown
as symbols). Coupled discharge pulse energy is 7.5 mJ/pulse.
Figure 4.34 compares modeling predictions and experimental data for time-
resolved, absolute OH number density after the discharge pulse in three different H 2-air
mixtures, excited by nanosecond pulse discharge generated by Pulser #2, with coupled
energy of 7.5 mJ/pulse. The model reproduces peak OH number density for a 3% H 2-air
mixture, but underpredicts peak OH number density for 1% and 5% H 2- air mixtures. For
all mixtures, the model overpredicts the OH decay rate, and does not reproduce the OH
number density rise at t ~ 100 µs (see Fig. 4.34). The modeling predictions (in particular,
absence of OH number density overshoot in the afterglow for 1% H 2-air mixtures) are
consistent with our previous modeling calculations [6,40], which are in very good
181
agreement with the experimental data taken at similar experimental conditions (lean H 2-air
mixtures at P=100 Torr and T=500 K) [40]. The difference between the modeling
predictions and the present data, indicating significant [OH] overshoot, is not fully
understood. Adding Equation 4.1, vibrational energy transfer from N 2 to HO2, to the kinetic
model demonstrated that it does not enhance the rate of OH production, since at the present
conditions nearly every HO2 radical formed by a three-body recombination reaction, is
converted to OH by a rapid reaction with O atoms,
𝐻+𝑂 +𝑀 𝐻𝑂 + 𝑀 (4.13)
𝑂 + 𝐻𝑂 𝑂𝐻 + 𝑂 (4.14)
Also, as discussed above, gradual temperature rise caused by N 2 vibrational
relaxation is too slow to explain the [OH] overshoot detected in the experiment. Since at
the present conditions the diffuse discharge filament is generated slightly off the electrode
symmetry axis, it is possible that the overshoot is caused by diffusion of H atoms from the
filament centerline into the line of sight of the LIF laser beam, on the time scale of several
tens of μs. This would significantly accelerate the rate of HO2 formation, with subsequent
OH production by the reaction with O atoms (Eqns. 4.13 and 4.14)
Figure 4.35 plots dominant radical species number densities (ground state O( 3P)
atoms, H atoms, OH, and HO2, vs. time delay after the discharge pulse in a 3% H2-air
mixture). Both O atoms and H atoms are generated by electron impact dissociation of
molecular oxygen and hydrogen, as well as by collisional quenching of electronically
excited nitrogen molecules, N2*, by O2 and H2 (Eqns. 4.5, 4.6, 4.9, and 4.10). A strong
182
overshoot of OH number density predicted on ~100 ns time scale is due to OH formation
in a reaction of electronically excited O(1D) atoms with molecular hydrogen,
𝑂( 𝐷) + 𝐻 → 𝑂𝐻 + 𝐻 (4.15)
On a ~1 μs time scale, OH decays by a reaction
𝑂𝐻 + 𝐻 𝐻 𝑂+𝐻 (4.16)
which is also an additional source of H atoms. The kinetic model also incorporates kinetics
of vibrationally excited H2 molecules, 𝑣 = 1 − 3, generated in the plasma by electron
impact. Populations of H2 vibrational levels in the plasma are controlled by the V-V energy
transfer,
𝐻 + 𝑒 → 𝐻 (𝑣 = 1 − 3) + 𝑒 (4.17)
𝐻 (𝑤) + 𝐻 (𝑣) → 𝐻 (𝑤 − 1) + 𝐻 (𝑣 + 1) (4.18)
and by V-T relaxation by O atoms,
𝐻 (𝑣) + 𝑂 → 𝐻 (𝑣 − 1) + 𝑂, (4.19)
Incorporating H2 vibrational kinetics is necessary to estimate the effect of OH
production on the time scale of t ~3 μs – 100 μs by the reaction of O atoms with H 2(v),
𝐻 (𝑣) + 𝑂 → 𝑂𝐻 + 𝐻, (4.20)
( )
with the ratio of room temperature rate coefficients, ( )
= 2600 [89]. However,
kinetic modeling calculations show that H2 V-T by O atoms, with temperature dependent
rate coefficient in Reference [90], dominates the reaction of Equation 4.20, such that H 2
vibrational kinetics has a negligible effect on OH formation in the afterglow. Kinetic

183
modeling calculations demonstrate that the dominant reactions of OH formation on the
time scale of 3 µs – 1 ms are reactions of Eqns. 4.13, 4.14, and H atom reaction with O 2,
𝐻+𝑂 𝑂𝐻 + 𝑂 (4.21)
At this time scale, OH decays by reaction with O atoms,
𝑂 + 𝑂𝐻 𝑂 +𝐻 (4.22)
and reaction of Equation 4.16. Using a quasi-stationary approximation, OH number density
can be described by the following equation:
[ ]∗[ ]∗( ∗ )
[𝑂𝐻] ≈
[ ]∗ [ ]∗
(4.23)
which is in good agreement with the full kinetic model prediction. From Equation 4.23, it
can be seen that OH number density in the afterglow basically follows the number densities
of O and H atoms (the latter is also strongly affected by radial diffusion of H atoms from
the filament at t > 100-200 μs), as well as gas temperature, which affects the rate
coefficients of chemical reactions. However, kinetic modeling shows that the gradual
temperature rise on ~10 μs – 1 ms, caused by vibrational relaxation of nitrogen by O atoms
and by chemical energy release during hydrogen oxidation (see Fig. 4.33(b)), which
increases the rate coefficient of reactions of Eqns. 4.20 and 4.21, remains a relatively minor
factor. Basically, the effect of the temperature rise on OH number density is strongly
diminished by H atom radial diffusion on the time scale of ~ 100-200 μs, which prevents
[OH] overshoot due to the temperature increase. Thus, the kinetic model does not predict
OH number density rise in the afterglow, at variance with the experimental results.
184
and H2-air mixtures. Coupled discharge pulse energy is 10 mJ/pulse.
185
pulse energy is 10 mJ/pulse.
Comparison of the modeling predictions at a higher pulse energy, 10 mJ pulse,
plotted in Figs. 4.36 and 4.37, exhibits similar trends. The kinetic modeling predictions for
these mixtures predict a rise in OH number density for 3% and 5% H 2- air mixtures on the
timescale of 50µs - 500 µs, caused by the temperature overshoot, shown in Figure 4.36(b).
This demonstrates that temperature rise caused by vibrational relaxation of nitrogen in the
afterglow of ns pulse discharges in lean H2-air mixtures at high specific energy loading,
resulting in strong vibrational nonequilibrium, may be a dominant effect of OH number
density overshoot. Again, additional energy release during hydrogen oxidation in plasma
chemical reactions amplifies this effect.
186
Summarizing, time-resolved measurements of temperature, N2 vibrational
temperature, and absolute OH number density in nitrogen, air, and 1-5% H 2-air mixtures
at P=100 Torr are used to study kinetics of OH formation and decay, and a possible effect
of nitrogen vibrational excitation on low-temperature kinetics of HO 2 and OH radicals, at
high specific energy loading. Broadband plasma emission images show that a discharge
pulse between two spherical electrodes separated by an 8 mm gap generates a diffuse
plasma filament ~2 mm in diameter. Translational-rotational temperature and N 2
vibrational temperature in the discharge and afterglow are measured by ns broadband,
degenerate-pump CARS in collinear phase-matching geometry, with pump and Stokes
laser beams directed along the discharge filament through holes drilled in the discharge
electrodes. Spatial resolution of the present CARS measurements along the discharge
filament is approximately 4 mm. The uncertainty of translational-rotational temperature
and N2 vibrational temperature inference is ±50 K and ±200 K, respectively. OH number
density is measured by LIF, using direct measurements of LIF quenching rate and absolute
calibration by Rayleigh scattering. The uncertainty of absolute OH number density
measurements is ±30%.
CARS measurements demonstrated that the discharge generates strong vibrational
nonequilibrium in air and H2-air mixtures for delay times after the discharge pulse up to ~
1 ms, with peak N2 vibrational temperature of up to Tv = 1700-1900 K at T = 500-580 K.
N2(v=0-2) vibrational levels are detected both in air and H2-air mixtures. The kinetics of
population and decay of N2 vibrational levels at these conditions are well understood, based
both on previous CARS measurements and kinetic modeling predictions. Laser induced
187
fluorescence line images along the discharge filament show that the LIF signal peaks near
the electrodes (primarily near the grounded electrode), indicating that specific energy
loading in these regions is higher compared to the rest of the filament, as expected. LIF
measurements show that OH number density increases gradually after the discharge pulse,
starting at t = 10 μs, before peaking at t ~ 100-300 μs (depending on H 2 fraction in the
mixture), and decaying on a longer time scale, until t ~ 1 ms. Increasing the discharge pulse
energy from 7.5 mJ to 10.0 mJ results in peak OH number density increasing by up to a
factor of 2.5. Peak OH number density in H2-air mixtures is reached at approximately the
same time as peak N2 vibrational temperature. Comparison of the experimental data with
kinetic modeling predictions shows that OH kinetics at the present conditions is controlled
primarily by reactions of H2 and O2 with O and H atoms generated during the discharge.
At the present conditions, OH number density is not affected by N 2 vibrational excitation
directly, via vibrational energy transfer to HO2, or indirectly, due to temperature rise caused
by N2 vibrational relaxation by O atoms. However, the modeling calculations indicate that
at higher discharge specific energy loading vibrational relaxation of nitrogen by O atoms,
producing significant temperature rise in the afterglow, may result in transient overshoot
of OH number density, such as detected in previous OH LIF measurements in ns pulse
discharge in fuel-air mixtures at atmospheric pressures [39].
188
Chapter 5: Radical species measurements at a liquid/vapor
interface
The main focus of this chapter is characterization of low-temperature plasmas
sustained by ns pulse discharges near liquid surfaces, using laser diagnostics. For this work,
diffuse, highly reproducible plasmas were sustained near a planar liquid-vapor interface,
using discharges generated by high-voltage, nanosecond duration pulses [91]. At high
estimated peak electric fields near the surface [91], the rate of electron impact ionization
greatly exceeds the rate of dissociative attachment, such that diffuse plasmas can be
sustained even in strongly electronegative gases mixtures, such as water vapor. Also, high
electron energy achieved in these discharges significantly accelerates electron impact
processes with high energy threshold, such as ionization and molecular dissociation, thus
generating reactive radical species in molecular plasmas. Finally, relatively short discharge
pulse duration, ~10-100 ns, enhances plasma stability and precludes bulk motion of the
liquid which may be caused by charge accumulation on the liquid surface.
Peak electric field in the surface ionization wave may be further enhanced if the
discharge is generated over a liquid with high dielectric constant, such as water (ε~80).
High energy electrons generated in the near-surface plasma layer would initiate electron
impact dissociation and ionization of evaporating reactants, with potentially high yield of
189
radical species, such as O, H, and OH in surface plasmas sustained over liquid water and
aqueous solutions. Chemical reactions in repetitively pulsed plasma at a liquid-vapor
interface are expected to occur at low temperature conditions, since Joule heating is limited
both by high specific heat of the liquid and by high latent heat of vaporization (~0.4
eV/molecule for water). This suggests that plasma sustained near a liquid-vapor interface
can be used as an experimental platform for studies of near-surface plasma chemical
reaction kinetics, at the conditions when the interface acts as a high-yield “plane surface
source” of radical species.
The objectives of the present work are to obtain in situ measurements of absolute
OH and H number density distributions in nanosecond pulse discharges generated near the
liquid-vapor interface in simple geometry test sections, using OH LIF and H TALIF
diagnostics similar to those used for characterization of reacting fuel-oxidizer plasmas in
Chapters 3 and 4. Two simple, “canonical” geometry test sections provided optical access
for in situ optical diagnostics for quantitative studies of plasma dynamics and
nonequilibrium plasma chemistry, with high temporal and spatial resolution.
5.1. Electric discharge cell design
The experiments have been conducted using two different discharge cells of similar
design, shown schematically in Figure 5.1(a, b).
190
Figure 5.1: Schematic diagram of the discharge cells: (a) surface ionization wave discharge
cell, and (b) dielectric barrier discharge cell.
The first cell, shown in Fig. 5.1(a), consists of a 10 cm long rectangular cross
section quartz channel (10 mm width by 22 mm height), fused at both ends to 25 mm
outside diameter quartz tube sections 3 cm long. Two plastic (Delrin) flanges with 25 mm
191
inside diameter circular openings served as both electrode feedthroughs and connectors to
two 25 mm outside diameter, 11 cm long quartz tube end pieces. The centerpiece and the
end pieces are held together by the flanges equipped with rubber O-rings to provide a
vacuum seal, and tightened using plastic screws. At each end, UV fused silica windows
were glued at Brewster’s angle (calculated at 308 nm) to allow optical access to laser beams
used for LIF and TALIF species number density measurements in the plasma afterglow.
Gas flow inlet and exit was through ¼ inch diameter quartz tubes. Between the
experiments, the inlet and exit tubes were used to partially fill the cell with a liquid
(distilled water), to the depth of 6 mm, using a syringe and a small diameter flexible plastic
tube. The flow inlet and exit were connected to gas delivery and vacuum lines by Ultratorr
vacuum fittings. Argon buffer flow (flow rate 0.1 SLM, measured by an MKS mass flow
controller) was maintained in the channel, at the conditions when the vapor pressure in the
cell was close to saturated vapor pressure, 15.4-18.6 Torr for water at room temperature
(291-294 K), and at total pressure in the cell of 18-40 Torr (measured by an MKS Series
910 DualTrans Transducer). The buffer flow rate was kept sufficiently low to prevent
excessive evaporative cooling of the test cell, as well as water vapor condensation on the
test cell walls. The test cell was placed on a translation stage, such that its vertical position
(i.e. the distance between the laser beam and the liquid surface) could be adjusted, to collect
the fluorescence signal at different heights above the liquid surface.
Two cylindrical segment copper foil electrodes (0.1 mm thick) were located near
the bottom of the flanges, aligned along the inner surface of the circular opening of the
flange, and held in place by the flange covers, as shown in Fig. 5.1(a). The discharge
192
electrodes were placed inside the cell, to enable a conduction current path between the
electrodes after the surface ionization wave, initiated near the high voltage electrode and
propagating over the liquid surface, reached the grounded (or opposite polarity) electrode.
Both electrodes were in direct contact with a layer of liquid (distilled water) partially filling
the channel. The electrodes were connected to brass pin feedthroughs, vacuum sealed using
a Dow Corning 3165 adhesive. The buffer gas flow in the cell was directed toward the high
voltage electrode. A section of adhesive copper tape 10.5 cm long and 1 cm wide was
attached to the bottom outside wall of the channel and connected to one of the brass pin
electrode feedthroughs, serving as a waveguide to the surface ionization wave discharge
sustained in the cell.
The second cell, shown in Figure 5.1b is made of a 16 cm long rectangular quartz
channel with the same cross section as in the first cell, fused at both ends to 25 mm outside
diameter quartz tube sections 3 cm long, with ¼” quartz tubes used for flow inlet and exit
and fused silica windows at Brewster’s angle. Between the experiments, the inlet and exit
tubes were used to partially fill the cell with distilled water, to the depth of 6 mm. The
electrodes, made of adhesive copper foil 10 mm wide and 20 mm long, were attached to
the outside bottom wall of the cell 30 mm apart from each other, as shown in Fig. 5.1(b),
to generate an ionization wave plasma in a double dielectric barrier configuration. Kapton
dielectric tape was placed between the electrodes to prevent breakdown outside the cell.
193
5.2. Discharge waveforms and plasma images
Figure 5.2(a) plots pulse voltage and current waveforms in the surface ionization
wave cell, as well as calculated coupled energy waveforms measured by the same custom
made, high bandwidth voltage and current probe outlined earlier in Chapter 3. The
electrodes, located 16 cm apart, were powered by a FID GmbH FPG 60-100MC4 pulse
generator (peak voltage up to 32 kV, pulse duration 5 ns, repetition rate up to 100 kHz), to
sustain a near-surface ionization wave discharge inside the test section. In this case, the
positive polarity terminal of the pulser (+ 16 kV) was connected to the “high voltage
electrode”, and the negative polarity terminal (-16 kV) was connected to the “grounded
electrode” and the waveguide outside the cell (see Fig. 5.1(a)). This “dual polarity”
ionization wave generated by this pulser, similar to the one used in our previous work [91],
was used to significantly reduce pulse voltage rise time and increase peak voltage
difference between the electrodes. Note that the voltage between the electrodes may be
higher than the difference between the incident pulse peak voltages (by up to a factor of
two), due to incident voltage pulse reflection. Multiple peaks in voltage, current, and
coupled energy are representative of the reflections along the transmission line. The
incident pulse peak voltage is approximately 30 kV with a pulse duration of 10 ns, FWHM.
In this case, the pulser was operated in repetitive burst mode, at pulse repetition rate of 10
kHz, 75 pulses in the burst, and burst repetition rate of 10 Hz, to match the laser pulse
repetition rate. A coupled energy of 2.6 mJ per pulse was maintained during the entire burst
duration.
194
Figure 5.2: Voltage, current, and coupled energy waveforms, (a) surface ionization wave
discharge cell, last pulse in a burst of 75 pulses, pulse repetition rate 10 kHz; (b) dielectric
barrier discharge cell, last pulse in a burst of 20 pulses, pulse repetition rate 1 kHz. Ar
buffer. P=30 Torr, flow rate 0.1 SLM, FID pulser.
Figure 5.2b plots similar pulse voltage, current, and coupled energy waveforms for
the dielectric barrier discharge cell. The electrodes were powered by the FID generator,
operated at half peak output voltage (using only two out of four output channels available)
to sustain a near-surface DBD ionization wave. The positive polarity (+ 8 kV) and the
negative polarity (‒ 8 kV) still resulted in a peak incident voltage pulse of approximately
30 kV and a coupled energy of 3.7 mJ per pulse. The pulser was operated in repetitive bust
mode, at a pulse repetition rate of 1 kHz, 20 pulses in the burst, and burst repetition rate of
10 Hz. High temporal resolution capacitive probe measurements have shown that the
discharge is formed by two opposite polarity surface ionization waves propagating from
the electrodes toward the center of the discharge cell. For these measurements, the
capacitive probe, which is sensitive to the charge accumulated on the dielectric surface,
195
was moved along the discharge cell, and its output was monitored on the oscilloscope. The
use of the capacitive probe is discussed in greater detail in our previous work [91].
Figure 5.3: ICCD camera image of plasma emission in a surface ionization wave discharge
over water in argon buffer at the conditions of Fig. 5.4(a): (a) top view, (b) side view. P=18
Torr, camera gate 1 μs, FID pulser.
Broadband plasma emission images in the discharge cells were taken using two
different gated PI-MAX ICCD cameras with a UV lens (UV-Nikkor 105 mm f/4.5, Nikon),
one with minimum gate of 2 ns, 512 x 512 pixels, and the other with minimum gate of 0.5
ns, 1024 x 255 pixels. The latter camera was also used to obtain LIF / TALIF line images.
196
ICCD images of plasma emission generated over the distilled water surface in argon buffer
flow in surface ionization wave discharge cell and in double dielectric barrier discharge
cell at the conditions of Figure 5.2 are shown in Figures 5.3 and 5.4, respectively.
Figure 5.4: ICCD camera image of plasma emission in a double dielectric barrier discharge
over water surface in argon buffer at the conditions of Fig. 5.4(b). P=30 Torr, camera gate
100 ns, FID pulser.
In these images, camera gate is 1 μs (Fig. 5.3) and 100 ns (Fig. 5.4), such that
ionization wave propagation during the discharge pulse is not resolved. The approximate
location of distilled water surface is the bottom edge of both ICCD images. Figure 5.3
shows the diffuse ionization wave plasma, ~1 mm thick, follows the liquid surface. Figure
5.4 shows the ionization wave plasma above the dielectric barrier follows near the surface
above the electrodes (the locations of which are indicated in Fig. 5.4), but lifts 1-2 mm
above the liquid surface between electrodes.
5.3. OH and H Spatial Distributions
The schematic of OH LIF / H TALIF laser diagnostics, shown in Figure 5.5, has
been previously described in Chapters 3 and 4. Changing the position of the laser beam
197
relative to the liquid surface, by moving the discharge cell mounted on the translation stage,
a series of OH LIF and H TALIF line images from the rectangular part of the quartz channel
were collected. The fluorescence signal in the line images is calibrated to obtain absolute
one-dimensional H and OH number density distributions along the laser beam, using Kr
TALIF and Rayleigh scattering, for H and OH respectively. Rayleigh scattering signal from
a dry test cell filled with nitrogen was collected with the ICCD camera. The camera is not
sensitive enough at ~830 nm to detect Kr TALIF fluorescence signal, so H and Kr TALIF
signals were collected by a PMT (Hamamatsu R3896) from a region along the fluorescence
line approximately 6 mm long, 1 mm above the liquid surface. Two-dimensional [H] and
[OH] distributions can be reconstructed from these line image series. Species
concentrations as functions of height above the surface are inferred based on spline-fit of
concentrations measured at several heights. The use of this approach is justified since
pulse-to-pulse reproducibility of the data is very good. Although the present setup can also
be used to obtain two-dimensional planar LIF (PLIF) images, the use of line images
significantly improves signal-to-noise.
198
Figure 5.5: Schematic of experimental setup for OH and H number density measurements.
In this work, partial pressure of water vapor above the liquid surface was assumed
to be 17.5 Torr (close to saturated vapor pressure at room temperature of 293 K). The near-
saturated vapor assumption is justified since at the present fully developed laminar flow
conditions (flow velocity u ≈ 25 cm/s, Reynolds number based on channel hydrodynamic
diameter of Δ = 1.2 cm ReΔ ≈ 6, Sherwood number for convection mass transfer ShΔ ≈ 3.0),
convection mass transfer coefficient is fairly high, h = ShΔ ∙D/Δ ≈ 15 cm/s [91]. Here D ≈
6 cm2/s is the diffusion coefficient of water in argon. In a fully developed, initially dry flow
( )
of an evaporating liquid surface, the vapor density varies as = 1 − exp(−ℎ𝑥 𝑢∆),
( )
such that the estimated distance over which reaches 90% is 𝑥 % = 2.3 ∗ 𝑢∆ ℎ ~ 5𝑐𝑚,
which is close to the distance from the flow inlet to the part of the channel where LIF and
TALIF images were taken. Since the fluorescence quenching rate of OH by Ar is much
lower than quenching by H2O [92], water vapor was assumed to be the only quenching
199
species. Similarly, H atom fluorescence quantum efficiency was calculated considering
water vapor to be the only quenching species present in the cell. Note, however, that the
validity of these assumptions have not been verified experimentally, since fluorescence
quenching rate was too high to be resolved by measuring the fluorescence signal decay.
In all LIF / TALIF measurements, the laser pulse energy was kept below the
saturation thresholds to mitigate any photo-dissociation or photolytic effects, ELIF < 8 µJ
and ETALIF < 35 µJ. In addition, the camera gate was set wide enough, 2 μs, to capture the
entire duration of fluorescence decay after the laser pulse.
Figure 5.6 shows OH LIF and H TALIF line images (6 for OH and 3 for H) taken in
the surface ionization wave discharge cell 4 μs after the last pulse in the discharge burst
over distilled water in Ar buffer at total pressure of 18 Torr, and a flow rate of 0.1 SLM.
Both OH and H are detected in the entire field of view of the camera, approximately 50
mm long, up to several mm above the liquid surface.
200
Figure 5.6: (a) OH LIF line images and (b) H TALIF line images in surface ionization wave
discharge over liquid water in Ar buffer. P=18 Torr, 4 μs after 75 pulses burst at 10 kHz,
FID pulser.
Figure 5.7 shows one-dimensional distributions of absolute OH and H atoms number
densities inferred from OH LIF line images. Comparing the inferred density distributions
with the ICCD image of plasma emission, both OH and H diffuse / convect over several
mm from the liquid surface, significantly exceeding the extent of the near-surface plasma.
Peak [OH] value measured at these conditions is ≈ 3.5∙10 13 cm-3, such that the estimated
OH mole fraction in the gas/vapor phase is ~ 6∙10 -5. Highest OH number density region is
detected near the positive polarity electrode. Peak H atom number density is ≈ 2∙10 15 cm-
3
, with estimated mole fraction of 0.3%. Estimated combined uncertainties of absolute
[OH] and [H] measurements at these conditions are ±30% and ±50%, respectively, with
201
the uncertainty of the fluorescence quenching cross sections being the dominant
contributing factor.
Figure 5.7: One-dimensional distributions of (a) absolute OH and (b) absolute H atoms
number densities, inferred from OH LIF and H TALIF line images shown in Fig. 5.6.
202
Figure 5.8 shows OH LIF and H TALIF line images (13 for OH and 7 for H) taken
in the double dielectric barrier discharge cell, 6 µs after the last pulse in the discharge burst
over distilled water in Ar buffer at total pressure of 30 Torr, and a flow rate of 0.1 SLM.
In these measurements, the laser beam was brought to within 300 μm from the liquid
surface. Figure 5.9(a) shows a two-dimensional distribution of absolute OH number density
inferred from OH LIF line images. Comparing this plot with the ICCD image of plasma
emission shown in Fig. 5.4, it can be seen that OH distribution follows the plasma emission
intensity. Peak [OH] value measured at these conditions is ≈ 3∙10 14 cm-3, such that the
estimated OH mole fraction in the gas/vapor phase is about 3∙10 -4. The highest OH number
density region is located near the positive electrode, and lifted away from the water surface.
Near the top wall of the quartz test cell, [OH] decreases to near the detection limit. At the
low flow rate used in the present experiments (estimated flow residence time between the
electrodes of ~ 0.2 s, much longer compared to burst duration, 20 ms), both the flow and
diffusion do not seem to affect the location of peak OH number density region.
203
Figure 5.8: (a) OH LIF line images and (b) H TALIF line images in a double dielectric
barrier discharge over liquid water in Ar buffer. P=30 Torr, 2 μs after 20 pulses burst at 1
kHz, FID pulser.
Figure 5.9(b) shows a two-dimensional distribution of absolute H number density,
inferred from H TALIF line images. Peak [H] value is ≈ 2∙1016 cm-3, such that H atom mole
fraction in the mixture is about 2%, much higher compared to OH mole fraction. As can
be seen from comparison of the two-dimensional density distributions, H atoms diffuse
further away from the liquid surface / plasma region, and the peak H atom number density
was achieved above the negative electrode, unlike peak OH number density, where the
peak is located closer to the positive electrode at the same location as the highest plasma
emission intensity. H number density decay near the top wall of the test section is also
detected. Estimated uncertainties of absolute [OH] and [H] populations plotted in Figure
5.9 are ±22% and ±50%, respectively.
204
Figure 5.9: Two-dimensional distributions of absolute OH (a) and H atoms number
densities, inferred from OH LIF and H TALIF line images shown in Fig. 5.8.
Both OH and H atom number densities measured in the double dielectric barrier
discharge, plotted in Figure 5.9, are significantly higher compared to the results obtained
in the surface ionization wave discharge, shown in Figure 5.8, by approximately an order
of magnitude. Since pulse peak voltage and coupled pulse energy in both types of
discharges are comparable, as shown in Figure 5.2, such a significant difference is most
likely due to a larger distance between the electrodes in the ionization wave discharge, 16
205
cm, compared to the double dielectric barrier discharge, 3 cm, and therefore lower reduced
electric field in the plasma behind the wave front.
5.4. Plasma assisted combustion kinetic modeling predictions
In the argon-water vapor plasma, above the liquid surface, both OH and H are likely
to be generated at the same rate by two dominant processes, electron impact dissociation
of water vapor,
𝐻 𝑂 + 𝑒 → 𝑂𝐻 + 𝐻 + 𝑒 (5.1)
and reactive quenching of metastable Ar* atoms,
𝐻 𝑂 + 𝐴𝑟 ∗ → 𝑂𝐻 + 𝐻 + 𝐴𝑟 (5.2)
A quasi-zero-dimensional kinetic model which incorporates ionization and
electronic excitation of argon, ionization and dissociation of H 2O by electron impact, H2O
dissociation during quenching of metastable Ar* atoms, and chemical reactions among H,
O, OH, H2, O2, H2O, HO2, and H2O2 was used to obtain qualitative insight into kinetics of
plasma chemical reactions in the plasma afterglow above a liquid/vapor interface in the
DBD cell. Cross sections of electron impact processes were taken from Bolsig+ database
[78], rates of Ar* reactions from Ref. [81], and rates of hydrogen-oxygen chemical
reactions from Ref. [93]. Experimental coupled energy waveform in a repetitively pulsed
ns discharge, from Figure 5.2(b), was used as a model input, and the volume of the plasma
was estimated from the plasma emission image shown in Figure 5.4.
This model is essentially the same as the one used for kinetic modeling of reacting
206
fuel-oxidizer plasmas in Chapter 3. However, due to considerably more complex discharge
geometry in the surface ionization wave, the modeling predictions cannot be expected to
be quantitative, unlike the near 0-D diffuse plasmas sustained in plasma flow reactors,
discussed in Chapter 3.
Quantitative predictions of the chemical composition of the plasma in surface
ionization wave would require using a two-dimensional discharge model, taking into
account time-resolved electric field and electron density distributions in the ionization
wave, as well as coupled kinetics and transport of reacting neutral species, such as the
model developed in Reference [94]. The modeling predictions in the present work can be
considered as semi-qualitative estimates.
207
Figure 5.10: [H] and [OH] number densities at the conditions of Fig. 5.8, predicted by a 0-
D kinetic model. 50% - 50% H2O – Ar mixture, P=30 Torr. Discharge pulse repetition rate
1 kHz, coupled pulse energy 3.7 mJ/pulse, estimated ≈1 cm 3, burst duration of 20 pulses.
Figure 5.10 shows kinetic modeling predictions in 50%-50% H 2O vapor – Ar
plasma afterglow at P=30 torr, excited by a burst of 20 discharge pulses at pulse repetition
rate of 1 kHz, coupled pulse energy of 3.7 mJ/pulse. The estimated plasma volume is ≈ 1
cm3. Figure 5.10(a) plots predicted temperature, [H], and [OH] during and after the
discharge burst, and illustrates rapid [H] and [OH] rise during each discharge pulse,
primarily due to Reactions 5.1 and 5.2. Quenching of electronically excited argon
metastables contributes to the temperature rise during and after the discharge pulses.
Between the pulses, [OH] decays rapidly in bimolecular reactions such as,
𝑂𝐻 + 𝑂𝐻 → 𝐻 𝑂 + 𝑂, (5.3)
208
with the room temperature rate coefficient of k=1.5∙10-12 cm3/s and characteristic time of
𝜏 ~(𝑘 ∗ [𝑂𝐻]) ~(1.5∙10-12 cm3/s ∙ 4∙1015 cm-3)-1 ~ 100 μs. This is much faster compared
to H atom decay by three-body recombination,
𝐻 + 𝑂𝐻 + 𝑀 → 𝐻 𝑂 + 𝑀, (5.4)
with the room temperature rate coefficient of k=3∙10-30 cm3/s and characteristic time of
𝜏 ~(𝑘 ∗ [𝑂𝐻] ∗ [𝑀]) ~(3∙10-30 cm6/s ∙ 2∙1014 cm-3 ∙ 1018 cm-3)-1 ~ 1 ms. This prediction
is qualitatively consistent with the present experimental results shown in Figure 5.9. The
model also predicts the temperature in the near-surface plasma to increase significantly,
exceeding T=500 K, and to decrease between the pulses, primarily due to rapid conduction
heat transfer to the cold liquid surface. Figure 5.10(b), which plots [H], [OH], and
temperature after the discharge burst on a log-log scale, provides a better illustration of the
difference between H atom and OH number density decay after the discharge burst.
Quantitative prediction of [OH], [H], and temperature distributions requires the use of a
kinetic model incorporating diffusion and transport of species, as well as heat transfer in
the near-surface plasma, preferably in two-dimensional geometry. Also, the modeling
results suggest that temperature in the plasma may vary significantly, such that temperature
distribution measurements would be necessary to verify the modeling predictions.
Summarizing, the present experiments provide insight into the kinetics of radical
species reactions in nanosecond pulse discharges sustained at a liquid-vapor interface,
above distilled water surfaces in nitrogen and argon buffer flow, at pressures ranging from
18 Torr to 30 Torr. The near-surface plasma is sustained using two different discharge
209
configurations, a surface ionization wave discharge between two exposed metal electrodes
and a double dielectric barrier discharge.
Laser Induced Fluorescence (LIF) and Two-Photon Absorption LIF (TALIF) line
imaging are used for in situ measurements of spatial distributions of absolute OH and H
atom number density distributions in near-surface, repetitive nanosecond pulse discharge
plasmas over distilled water surface in argon buffer at pressures of 18 Torr and 30 Torr.
Measurement results in a surface ionization wave discharge after a burst of 75 discharge
pulses at a pulse repetition rate of 10 kHz demonstrated that peak H number density, [H] ≈
2∙1015 cm-3, is much higher compared to peak OH number density, [OH] ≈ 3.5∙10 13 cm-3.
Higher OH number density was measured near the regions with higher plasma emission
intensity. Both OH and H atoms diffuse out of the plasma volume, over several mm from
the liquid surface. Measurements in a double dielectric barrier discharge after a burst of 20
discharge pulses at a pulse repetition rate of 1 kHz also show that peak H number density,
[H] ≈ 2∙1016 cm-3, is much higher compared to peak OH number density, [OH] ≈ 3∙10 14 cm-
3
. Highest OH number density was measured in the region where plasma emission peaks,
while H atoms convect downstream with the flow and diffuse further away from the surface
compared to OH. Kinetic modeling calculations using a quasi-zero-dimensional H 2O vapor
/ Ar plasma model are in qualitative agreement with the experimental results. Quantitative
analysis of plasma chemical reactions of radical species in plasma sustained over a liquid-
vapor interface would require using a kinetic model incorporating diffusion and transport
of species, as well as heat transfer from the near-surface plasma to the liquid. The results
demonstrate experimental capability of in situ radical species number density distribution
210
measurements in liquid-vapor interface plasmas, in simple canonical geometry that lends
itself for validation of kinetic models.
211
Chapter 6: Summary and Conclusions
This work has produced extensive sets of new data on low-temperature plasma-
assisted fuel oxidation in hydrogen and hydrocarbon-oxygen-argon mixtures, radical
production and decay in a ns pulse discharge in hydrogen-air mixtures, at high specific
energy loading, and radical production in plasmas sustained at a liquid/vapor interface. The
experimental data have been compared with kinetic modeling predictions, to obtain insight
into kinetic processes initiated in the plasma. A summary of the results and the conclusions
drawn from each experiment is discussed below.
One of the main results of the present work is conception, demonstration, and
testing of two new plasma flow reactor designs to prevent plasma filamentation and corona
discharge formation, and to significantly reduce the uncertainty in measured coupled
discharge energy in ns pulse discharges in high-pressure, preheated fuel-oxidizer mixtures.
The first plasma flow reactor employs two quartz reservoirs filled with liquid metal (Ga-
In-Sn), used as electrodes. The second flow reactor uses high temperature dielectric gel
pads and dielectric gel coating of the electrodes, which remain resilient at elevated
temperatures (up to at least T= 500 K). In both flow reactors, the plasma is generated by a
high peak voltage, ns pulse discharge, operated at high pulse repetition rates (up to 10-20
kHz). Plasma uniformity in the liquid-metal electrode reactor has been quantified using
measurements of metastable argon atoms made by Tunable Diode Laser Absorption
212
Spectroscopy (TDLAS), indicating that their number density across the reactor channel
cross section varies by ~15% at P= 300 Torr. At higher pressures, P= 500 Torr, metastable
Ar atom variation increases to ~50%, primarily near the channel walls.
Plasmas generated in both types of reactors are volumetric and diffuse, with no sign
of isolated discharge filaments. At these conditions, energy coupled to the plasma in a
repetitively pulsed ns discharge was measured using custom-made high-bandwidth, low-
noise voltage and current probes. Accurate measurements of the energy coupled to the
plasma is critical for quantitative understanding of the effect of electron impact processes,
reactions of excited electronic species, and radical reactions in fuel oxidation kinetics.
Second, comparing the results of H atom and O atom number density measurements
in ns pulse discharges in fuel-oxidizer mixtures with kinetic modeling predictions showed
that primary radicals generated efficiently in repetitively pulsed ns discharges at elevated
temperatures oxidize hydrogen and hydrocarbon fuels, but do not initiate chain branching
reactions. Time resolved, absolute number densities of H atoms and O atoms in these
plasmas have been measured using Two-photon Absorption Laser Induced Fluorescence
(TALIF), after a burst of nanosecond discharge pulses. H atom number density after the
discharge burst, as well as its decay rate in H2-Ar and H2-O2-Ar mixtures, are reproduced
well by the kinetic model. Similarly, O atom number density after the discharge burst and
during the decay in the afterglow, are reproduced well by the kinetic model in O 2-Ar, H2-
O2-Ar, and CH4-O2-Ar mixtures. In 1% H2-Ar and 1% O2-Ar mixtures at elevated initial
temperature, nonequilibrium plasmas effectively generate primary radical species, by
dissociating up to 36% of molecular hydrogen initially available in the mixture and up to
213
48% of the molecular oxygen initially available in the mixture during the discharge burst.
The reaction chain length, estimated by comparison of the stable species number density
(H2O) to the number density of primary radicals generated by the plasma (H and O atoms)
is low, on the order of one, indicating that chain branching reactions at the present
conditions remain insignificant.
O atom decay kinetics in lean hydrogen-O2-Ar and hydrocarbon-O2-Ar mixtures is
primarily controlled by the reactions
𝑂 + 𝑂𝐻 → 𝐻 + 𝑂
𝐻 + 𝐻𝑂 → 2𝑂𝐻
𝐻 + 𝑂 + 𝐴𝑟 → 𝐻𝑂 + 𝐴𝑟
which sum up to the equivalent net mechanism,
𝑂 + 𝐻 → 𝑂𝐻
Therefore, H atom generation processes, primarily dissociation of H 2 and
hydrocarbons by electron impact and during quenching of electronically excited Ar atoms,
are critical for understanding the consumption mechanisms of primary radicals generated
in the plasma. In hydrocarbon-O2-Ar mixtures, the model reproduces absolute O atom
number density and its decay rate in the discharge afterglow only when the fuel is not
completely oxidized during the discharge burst, i.e. when the fuel is still present in the
afterglow, as occurs in 0.13% H2-O2-Ar, 0.25% CH4-O2-Ar, and 150 ppm C3H8-O2-Ar
mixtures. In fuel-limited C3H8-O2-Ar and C2H4-O2-Ar mixtures (75 ppm of C2H8 and 75-
150 ppm of C2H4), when the fuel is almost completely oxidized before the end of the
214
discharge burst, the model consistently underpredicts O atom decay rate in the discharge
afterglow.
Comparing the O atom decay kinetics predicted by the model in CH 4-O2-Ar and
C2H4-O2-Ar mixtures shows that in the latter case, O atom decay is controlled by the
following reactions,
𝑂 + 𝑂𝐻 → 𝐻 + 𝑂
𝐻𝑂 + 𝑂 → 𝑂𝐻 + 𝑂
𝐻 + 𝑂 + 𝐴𝑟 → 𝐻𝑂 + 𝐴𝑟
which sum up to the equivalent net mechanism,
𝑂+𝑂 →𝑂
This occurs due to a shift of the dominant OH production channel from the reaction
of H atoms with HO2,
𝐻 + 𝐻𝑂 → 2𝑂𝐻,
to O atom reaction with HO2.
𝑂 + 𝐻𝑂 → 𝑂𝐻 + 𝑂 .
Comparison of the modeling predictions with the data indicate that after the fuel is
completely oxidized, the model is lacking a mechanism of H atom generation, resulting in
a “leveling off” of H atoms produced during the burst. Analysis of plasma images taken at
these conditions, as well as modeling calculations, shows that the difference between the
data and the modeling predictions is most likely due to transverse diffusion of fuel from
the peripheral “unreacted” regions near the side walls to the “reacted” core flow. This
provides the additional source of H atoms in the reacting mixture, thereby accelerating the
215
rate of O atom decay in the afterglow. Summarizing, the kinetic model is in good agreement
with the present data, with the exception of extremely fuel-lean mixtures. These results
illustrate the far-reaching impact of the present measurements on understanding low
temperature plasma fuel oxidation kinetics, and on truly predictive kinetic model
development.
The present work demonstrated that vibrationally excited nitrogen does not affect
radical species kinetics directly, via vibrational energy transfer hypothesized in the
previous work. A mechanism of vibration-vibration transfer from N 2 to HO2 hypothesized
in Ref. [2],
1 1 1
𝑁 𝑋 𝛴, 𝑣 = 1 + 𝐻𝑂 → 𝑁 𝑋 𝛴, 𝑣 = 0 + 𝐻𝑂 (𝑣 + 𝑣 ) → 𝑁 𝑋 𝛴 + 𝐻 + 𝑂
was predicted to enhance the radical production by generating more H atoms in a discharge
afterglow. To test this hypothesis, in the present work, time resolved measurements of
temperature and N2 vibrational excitation are made by ns Coherent Anti-Stokes Raman
Spectroscopy (ns-CARS), and absolute measurements OH are made by Laser Induced
Fluorescence (LIF), on the centerline of a ns pulse discharge diffuse filament in H 2-air
mixtures at the conditions of high specific energy loading. The results demonstrate a rise
in N2(v=1) population on a t~ 100 µs time scale after the discharge pulse, with peak N 2
vibrational temperature of Tv(N2)≈ 1900 K, while the gas temperature increases from T0=
300 K to T≈ 600 K. This indicates strong vibrational nonequilibrium generated by the
discharge. A rise in OH number density on the same time scale, t~ 100 µs, was observed,
216
with the OH number density in the 1% H2-air mixture following the N2 vibrational
temperature.
Comparing the data with kinetic modeling predictions shows that the model
reproduces the temperature and N2 vibrational temperature in all H2-air mixtures tested.
The model also predicts that the OH number density in the afterglow, at time scales t~ 10
µs – 1 ms. While the model reproduces the scale of the OH number density and OH decay
rate, indicating that all major OH formation and decay processes are incorporated, it does
not predict the increase in OH number density at t~ 10 – 100 µs, detected in the
experiments. This effect may be caused by rapid diffusion of H atoms into the line of sight
of the laser beam, resulting in additional OH formation, which is not incorporated into the
present model. Analysis of the modeling predictions also demonstrates that incorporating
the process of V-V energy transfer from N2 to HO2 makes the agreement with the model
worse, since this process dissociates HO2, a critical precursor species for OH formation at
the present conditions. The experimental results obtained at a higher discharge pulse
energy, 10 mJ/pulse, as well as the modeling predictions, indicate that vibrational
nonequilibrium may affect OH number density in the discharge afterglow indirectly, via
the temperature rise caused by N2 vibrational relaxation by O atoms. Also, the modeling
calculations show that the effect of a chemical reaction of vibrational excited hydrogen,
H2(v), with O atoms, resulting in additional generation of OH, is minor at the present
conditions.
Finally, the present measurements provided quantitative insight into radical species
generation in near-surface discharges at a liquid-vapor interface. OH and H radicals were
217
measured in the afterglow of nanosecond pulse discharge plasmas near a liquid/vapor
interface. One-dimensional LIF and TALIF line images yielded spatial distribution of these
species above the liquid surface. Peak H number density is much higher compared to peak
OH number density. Although kinetic modeling predictions indicate that both H and OH
are generated at the same rate, by electron impact dissociation of water vapor,
𝐻 𝑂 + 𝑒 → 𝑂𝐻 + 𝐻 + 𝑒
H atom number density is significantly higher than OH number density due to the
difference in their characteristic decay time, τOH ~ 100 µs by the reaction,
𝑂𝐻 + 𝑂𝐻 → 𝐻 𝑂 + 𝑂
versus τH ~ 1 ms, by the reaction,
𝐻 + 𝑂𝐻 + 𝑀 → 𝐻 𝑂 + 𝑀
respectively. Rapid decay of OH number density predicted by the model between the
discharge pulses in repetitive discharges, on ~100 μs time scale, demonstrates the need for
time-resolved [OH] distribution measurements using a high sampling rate, on the order of
~ 10 kHz. These measurements would yield quantitative insight into kinetic processes of
OH generation and decay in ns pulse discharge plasmas at a liquid-vapor interface, and
would be a significant advance over data taken using a 10 Hz pulse repetition rate laser
used in the present work. High sampling rate [OH], [H], [O], and [CH 2O] diagnostics
capability would also be critical for characterization of liquid-vapor interface discharges
sustained in highly reactive liquid-vapor mixtures, such as hydrocarbon or oxygenate fuel
vapor in air buffer. In this case, time-dependent number densities of atomic species
generated in the plasma by electron impact and reactive quenching of excited species would
218
be strongly affected by their reactions with evaporating reactants (hydrocarbons or
oxygenates). Finally, characterization of liquid-vapor interface plasmas would be
significantly advanced by in situ measurements of the electric field near the dielectric liquid
surface, e.g. using CARS / four-wave mixing diagnostics such as has been used recently
for measurements of electric field in surface ionization wave discharges over a solid
dielectric [95].
219
Bibliography
1) Ju, Y., & Sun, W. (2015). Plasma assisted combustion: dynamics and
chemistry. Progress in Energy and Combustion Science, 48, 21-83.
2) Starikovskiy, A., and Aleksandrov, N. (2013). Plasma-assisted ignition and

combustion. Progress in Energy and Combustion Science, 39(1), 61-110.
3) Raizer, Y. P. (1997). Gas discharge physics. Berlin: Springer.
4) Lim, M., Jayapalan, K.K., Zulazlan, A., and Chin, O. (2017). Characterization of an
atmospheric air non-thermal plasma device in relation to ozone production. International
Journal of Environmental Science and Development, 8(1).
5) Yin, Z., Adamovich, I. V., & Lempert, W. R. (2013). OH radical and temperature
measurements during ignition of H 2-air mixtures excited by a repetitively pulsed
nanosecond discharge. Proceedings of the Combustion Institute, 34(2), 3249-3258.
6) Yin, Z., Montello, A., Carter, C. D., Lempert, W. R., & Adamovich, I. V. (2013).
Measurements of temperature and hydroxyl radical generation/decay in lean fuel–air
mixtures excited by a repetitively pulsed nanosecond discharge. Combustion and
Flame, 160(9), 1594-1608.
7) Lanier, S., Shkurenkov, I., Adamovich, I. V., & Lempert, W. R. (2015). Two-stage
energy thermalization mechanism in nanosecond pulse discharges in air and hydrogen–air
mixtures. Plasma Sources Science and Technology, 24(2), 025005.
8) Schmidt, J. (2015). Ultrashort Two-Photon-Absorption Laser-Induced Fluorescence in

Nanosecond-Duration, Repetitively Pulsed Discharges. PhD Dissertation, Ohio State
University, Ohiolink
9) Shkurenkov, I., Burnette, D., Lempert, W. R., & Adamovich, I. V. (2014). Kinetics of
excited states and radicals in a nanosecond pulse discharge and afterglow in nitrogen and
air. Plasma Sources Science and Technology, 23(6), 065003.
10) Uddi, M., Jiang, N., Mintusov, E., Adamovich, I. V., & Lempert, W. R. (2009). Atomic
oxygen measurements in air and air/fuel nanosecond pulse discharges by two photon laser
induced fluorescence. Proceedings of the Combustion Institute, 32(1), 929-936.
220
11) Schmidt, J.B., Kulatilaka, W.D., Shkurenkov, I., Adamovich, I.V., Lempert, W.R.,
Gord, J.R., and Roy, S., “Femtosecond, two-photon-absorption, laser-induced-
fluorescence (fs-TALIF) imaging of atomic hydrogen and oxygen in non-equilibrium
plasmas”, Journal of Physics D: Applied Physics, vol. 50, 2017, p. 015204.
12) Lanier, S., Bowman, S., Burnette, D., Adamovich, I. V., & Lempert, W. R. (2014).
Time-resolved temperature and O atom measurements in nanosecond pulse discharges in
combustible mixtures. Journal of Physics D: Applied Physics, 47(44), 445204.
13) Yin, Z., Eckert, Z., Adamovich, I. V., & Lempert, W. R. (2015). Time-resolved radical
species and temperature distributions in an Ar–O 2–H2 mixture excited by a nanosecond
pulse discharge. Proceedings of the Combustion Institute, 35(3), 3455-3462.
14) Popov, N. A. (2008). Effect of a pulsed high-current discharge on hydrogen-air

mixtures. Plasma Physics Reports, 34(5), 376-391.
15) Pancheshnyi, S. V., Starikovskaia, S. M., & Starikovskii, A. Y. (2000). Collisional

deactivation of N2 (C3Πu, v= 0, 1, 2, 3) states by N2, O2, H2 and H2O molecules. Chemical
Physics, 262(2), 349-357.
16) Popov, N. A. (2011). Effect of singlet oxygen O2 (a 1Δg) molecules produced in a gas
discharge plasma on the ignition of hydrogen–oxygen mixtures. Plasma Sources Science
and Technology, 20(4), 045002.
17) McNeal, R. J., Whitson, M. E., & Cook, G. R. (1972). Quenching of vibrationally
excited N2 by atomic oxygen. Chemical Physics Letters, 16(3), 507-510.
18) Adamovich, I. V. (2001). Three-dimensional analytic model of vibrational energy

transfer in molecule-molecule collisions. AIAA journal, 39(10), 1916-1925.
19) Konnov, A. A. (2000). Detailed reaction mechanism for small hydrocarbons

combustion. Release 0.5.
20) Wang, H., You, X., Joshi, A. V., Davis, S. G., Laskin, A., Egolfopoulos, F., & Law, C.
K. (2007). USC Mech Version II. High-Temperature Combustion Reaction Model of
H2/CO/C1-C4 Compounds.
21) Popov, N. A. (2011). Fast gas heating in a nitrogen–oxygen discharge plasma: I.

Kinetic mechanism. Journal of Physics D: Applied Physics, 44(28), 285201.
221
22) Stancu, G. D., Kaddouri, F., Lacoste, D. A., & Laux, C. O. (2010). Atmospheric
pressure plasma diagnostics by OES, CRDS and TALIF. Journal of Physics D: Applied
Physics, 43(12), 124002.
23) Tsolas, N., Togai, K., Yin, Z., Frederickson, K., Yetter, R. A., Lempert, W. R., &
Adamovich, I. V. (2016). Plasma flow reactor studies of H 2/O2/Ar kinetics. Combustion
and Flame, 165, 144-153.
24) Tsolas, N., Yetter, R. A., & Adamovich, I. V. (2017). Kinetics of plasma assisted
pyrolysis and oxidation of ethylene. Part 2: Kinetic modeling studies. Combustion and
Flame, 176, 462-478.
25) Montello, A., Yin, Z., Burnette, D., Adamovich, I. V., & Lempert, W. R. (2013).
Picosecond CARS measurements of nitrogen vibrational loading and
rotational/translational temperature in non-equilibrium discharges. Journal of Physics D:
Applied Physics, 46(46), 464002.
26) Lo, A., Cessou, A., Boubert, P., & Vervisch, P. (2014). Space and time analysis of the
nanosecond scale discharges in atmospheric pressure air: I. Gas temperature and
vibrational distribution function of N2 and O2. Journal of Physics D: Applied
Physics, 47(11), 115201.
27) Konnov, A. A. (2008). Remaining uncertainties in the kinetic mechanism of hydrogen

combustion. Combustion and flame, 152(4), 507-528.
28) Burnette, D., Montello, A., Adamovich, I. V., & Lempert, W. R. (2014). Nitric oxide
kinetics in the afterglow of a diffuse plasma filament. Plasma Sources Science and
Technology, 23(4), 045007.
29) Brumfield, B., Sun, W., Ju, Y., & Wysocki, G. (2013). Direct in situ quantification of
HO2 from a flow reactor. The journal of physical chemistry letters, 4(6), 872-876.
30) Adamovich, I. V., Nishihara, M., Choi, I., Uddi, M., & Lempert, W. R. (2009). Energy
coupling to the plasma in repetitive nanosecond pulse discharges. Physics of
Plasmas, 16(11), 113505.
31) Shkurenkov, I., & Adamovich, I. V. (2016). Energy balance in nanosecond pulse
discharges in nitrogen and air. Plasma Sources Science and Technology, 25(1), 015021.
32) Takashima, K., Yin, Z., & Adamovich, I. V. (2012). Measurements and kinetic
modeling of energy coupling in volume and surface nanosecond pulse discharges. Plasma
Sources Science and Technology, 22(1), 015013.
222
33) Pintassilgo, C. D., Guerra, V., Guaitella, O., & Rousseau, A. (2014). Study of gas
heating mechanisms in millisecond pulsed discharges and afterglows in air at low
pressures. Plasma Sources Science and Technology, 23(2), 025006.
34) Macheret, S. O., Rusanov, V. D., Fridman, A. A., & Sholin, G. V. (1980).
Nonequilibrim plasma chemical synthesis of nitric oxides. Sov. Phys.—Tech. Phys, 50,
705-15.
35) Dmitrieva, I. K., & Zenevich, V. A. (1984). Nitrogen vibrational-excitation effect on

N2(v)+O-] NO+N reaction rate constant information theory approximation. Khimicheskaya
fizika, 3(8), 1075-1080.
36) Levitskii, A. A., Fridman, A. A., & Macherer, S. O. (1986). Simulation of

nonequilibrium chemical processes stimulated by vibrational excitation of the reactants in
plasmas. High Energy Chem., 19(5).
37) Guerra, V., & Loureiro, J. M. A. H. (1997). Electron and heavy particle kinetics in a
low-pressure nitrogen glow discharge. Plasma Sources Science and Technology, 6(3), 361.
38) Pintassilgo, C. D., Guaitella, O., & Rousseau, A. (2009). Heavy species kinetics in low-
pressure dc pulsed discharges in air. Plasma Sources Science and Technology, 18(2),
025005.
39) Wu, L., Lane, J., Cernansky, N. P., Miller, D. L., Fridman, A. A., & Starikovskiy, A.
Y. (2011). Plasma-assisted ignition below self-ignition threshold in methane, ethane,
propane and butane-air mixtures. Proceedings of the Combustion Institute, 33(2), 3219-
3224.
40) Adamovich, I. V., Li, T., & Lempert, W. R. (2015). Kinetic mechanism of molecular
energy transfer and chemical reactions in low-temperature air-fuel plasmas. Phil. Trans. R.
Soc. A, 373(2048), 20140336.
41) Bruggeman, P., & Leys, C. (2009). Non-thermal plasmas in and in contact with
liquids. Journal of Physics D: Applied Physics, 42(5), 053001.
42) Graves, D. B. (2012). The emerging role of reactive oxygen and nitrogen species in
redox biology and some implications for plasma applications to medicine and
biology. Journal of Physics D: Applied Physics, 45(26), 263001.
43) Sakiyama, Y., Graves, D. B., Chang, H. W., Shimizu, T., & Morfill, G. E. (2012).
Plasma chemistry model of surface microdischarge in humid air and dynamics of reactive
neutral species. Journal of Physics D: Applied Physics, 45(42), 425201.
223
44) Pavlovich, M. J., Chang, H. W., Sakiyama, Y., Clark, D. S., & Graves, D. B. (2013).
Ozone correlates with antibacterial effects from indirect air dielectric barrier discharge
treatment of water. Journal of Physics D: Applied Physics, 46(14), 145202.
45) Ognier, S., Iya-Sou, D., Fourmond, C., & Cavadias, S. (2009). Analysis of mechanisms
at the plasma–liquid interface in a gas–liquid discharge reactor used for treatment of
polluted water. Plasma Chemistry and Plasma Processing, 29(4), 261-273.
46) Locke, B. R., & Thagard, S. M. (2012). Analysis and review of chemical reactions and
transport processes in pulsed electrical discharge plasma formed directly in liquid
water. Plasma Chemistry and Plasma Processing, 32(5), 875-917.
47) Petitpas, G., Rollier, J. D., Darmon, A., Gonzalez-Aguilar, J., Metkemeijer, R., &
Fulcheri, L. (2007). A comparative study of non-thermal plasma assisted reforming
technologies. International Journal of Hydrogen Energy, 32(14), 2848-2867.
48) Lindner, P. J., & Besser, R. S. (2012). Hydrogen production by methanol reforming in
a non-thermal atmospheric pressure microplasma reactor. International Journal of
Hydrogen Energy, 37(18), 13338-13349.
49) Malik, M. A., Hughes, D., Malik, A., Xiao, S., & Schoenbach, K. H. (2013). Study of
the production of hydrogen and light hydrocarbons by spark discharges in diesel, kerosene,
gasoline, and methane. Plasma Chemistry and Plasma Processing, 33(1), 271-279.
50) Kolb, T., Voigt, J. H., & Gericke, K. H. (2013). Conversion of methane and carbon
dioxide in a DBD reactor: influence of oxygen. Plasma Chemistry and Plasma
Processing, 33(4), 631-646.
51) Babaeva, N. Y., & Kushner, M. J. (2009). Structure of positive streamers inside
gaseous bubbles immersed in liquids. Journal of Physics D: Applied Physics, 42(13),
132003.
52) Tian, W., & Kushner, M. J. (2014). Atmospheric pressure dielectric barrier discharges
interacting with liquid covered tissue. Journal of Physics D: Applied Physics, 47(16),
165201.
53) Starikovskiy, A., Yang, Y., Cho, Y. I., & Fridman, A. (2011). Non-equilibrium plasma
in liquid water: dynamics of generation and quenching. Plasma Sources Science and
Technology, 20(2), 024003.
54) Sommers, B. S., Foster, J. E., Babaeva, N. Y., & Kushner, M. J. (2011). Observations
of electric discharge streamer propagation and capillary oscillations on the surface of air
bubbles in water. Journal of Physics D: Applied Physics, 44(8), 082001.
224
55) Anpilov, A. M., Barkhudarov, É. M., Kop’ev, V. A., Kossyĭ, I. A., & Silakov, V. P.
(2006). Atmospheric electric discharge into water. Plasma Physics Reports, 32(11), 968-
972.
56) Belosheev, V. P. (2000). Discharge leader self-organization on the water

surface. Technical Physics, 45(7), 922-927.
57) Lagarkov, A. N., & Rutkevich, I. M. (2012). Ionization waves in electrical breakdown
of gases. Springer Science & Business Media.
58) Vasilyak, L. M., Kostyuchenko, S. V., Kudryavtsev, N. N., & Filyugin, I. V. (1994).
Fast ionisation waves under electrical breakdown conditions. Physics-Uspekhi, 37(3), 247.
59) Eckbreth, A. C. (1996). Laser diagnostics for combustion temperature and

species (Vol. 3). CRC Press.
60) Luque, J., & Crosley, D. R. (1999). Lifbase (version 1.6). computer program by SRI
International, SRI report No. MP, 99-009.
61) German, K. R. (1975). Direct measurement of the radiative lifetimes of the A 2Σ+(V′=
0) states of OH and OD. The Journal of Chemical Physics, 62(7), 2584-2587.
62) Tamura, M., Berg, P. A., Harrington, J. E., Luque, J., Jeffries, J. B., Smith, G. P., &
Crosley, D. R. (1998). Collisional quenching of CH(A), OH(A), and NO(A) in low pressure
hydrocarbon flames. Combustion and Flame, 114(3), 502-514.
63) Chu, P. K., & Lu, X. (Eds.). (2013). Low temperature plasma technology: methods and
applications. CRC Press.
64) Herzberg, G. (1953). Molecular spectra and molecular structure: vol. I.: spectra of
diatomic molecules. D. Van Nostrand Company.
65) Dieke, G. H., & Crosswhite, H. M. (1962). The ultraviolet bands of OH fundamental
data. Journal of Quantitative Spectroscopy and Radiative Transfer, 2(2), 97-199.
66) Miles, R. B., Lempert, W. R., & Forkey, J. N. (2001). Laser rayleigh
scattering. Measurement Science and Technology, 12(5), R33.
67) Loudon, R. (2000). The quantum theory of light. OUP Oxford.
68) Knake, N., Niemi, K., Reuter, S., Schulz-von der Gathen, V., & Winter, J. (2008).
Absolute atomic oxygen density profiles in the discharge core of a microscale atmospheric
pressure plasma jet. Applied Physics Letters, 93(13), 131503.
225
69) Niemi, K., Schulz-Von Der Gathen, V., & Döbele, H. F. (2001). Absolute calibration
of atomic density measurements by laser-induced fluorescence spectroscopy with two-
photon excitation. Journal of Physics D: Applied Physics, 34(15), 2330.
70) Saxon, R. P., & Eichler, J. (1986). Theoretical calculation of two-photon absorption
cross sections in atomic oxygen. Physical Review A, 34(1), 199.
71) Hübner, S., Sadeghi, N., Carbone, E. A. D., & Van Der Mullen, J. J. A. M. (2013).
Density of atoms in Ar*(3p54s) states and gas temperatures in an argon surfatron plasma
measured by tunable laser spectroscopy. Journal of Applied Physics, 113(14), 143306.
72) Rawlins, W. T., Galbally-Kinney, K. L., Davis, S. J., Hoskinson, A. R., Hopwood, J.
A., & Heaven, M. C. (2015). Optically pumped microplasma rare gas laser. Optics
express, 23(4), 4804-4813.
73) Ren, K. N., Zheng, Y. Z., Dai, W., Ryan, D., Fung, C. Y., & Wu, H. K. (2010, October).
Soft-lithography-based high temperature molding method to fabricate whole teflon
microfluidic chips. 14th Int. Conf. on Miniaturized Systems for Chemistry and Life Science
Micro TAS 3-7 October, 2010, Groningen, The Netherlands.
74) Goldsmith, J. E. (1987). Photochemical effects in two-photon-excited fluorescence

detection of atomic oxygen in flames. Applied optics, 26(17), 3566-3572.
75) Gasnot, L., Desgroux, P., Pauwels, J. F., & Sochet, L. R. (1997). Improvement of two-
photon laser induced fluorescence measurements of H-and O-atoms in premixed
methane/air flames. Applied Physics B: Lasers and Optics, 65(4), 639-646.
76) Morgan, J. E., & Schiff, H. I. (1964). Diffusion coefficients of O and N atoms in inert
gases. Canadian Journal of Chemistry, 42(10), 2300-2306.
77) Tang, M. J., Cox, R. A., & Kalberer, M. (2014). Compilation and evaluation of gas
phase diffusion coefficients of reactive trace gases in the atmosphere: volume 1. Inorganic
compounds. Atmospheric Chemistry and Physics, 14(17), 9233-9247.
78) G.J.M. Hagelaar and L.C. Pitchford, "Solving the Boltzmann equation to obtain
electron transport coefficients and rate coefficients for fluid models", Plasma Sources Sci.
Technol. 14 722-733 (2005)
79) Mason, N. J., & Newell, W. R. (1987). Total cross sections for metastable excitation
in the rare gases. Journal of Physics B: Atomic and Molecular Physics, 20(6), 1357.
226
80) Tachibana, K. (1986). Excitation of the 1s5, 1s4, 1s3, and 1s2 levels of argon by low-
energy electrons. Physical Review A, 34(2), 1007.
81) Velazco, J. E., Kolts, J. H., & Setser, D. W. (1978). Rate constants and quenching
mechanisms for the metastable states of argon, krypton, and xenon. The Journal of
Chemical Physics, 69(10), 4357-4373.
82) Lanier, S., Shkurenkov, I., Adamovich, I. V., & Lempert, W. R. (2015). Two-stage
energy thermalization mechanism in nanosecond pulse discharges in air and hydrogen–air
mixtures. Plasma Sources Science and Technology, 24(2), 025005.
83) Nishihara, M., Frederickson, K., & Lempert, W. R. (2016). Dual-pump CARS
measurements in a vibrationally nonequilibrium supersonic mixing layer. In 54th AIAA
Aerospace Sciences Meeting (p. 1762), January 4-8, 2016, San Diego, CA .
84) Clark, G., Palmer, R., & Farrow, R. (1990). CARSFIT computer code, developed at
Los Alamos and Sandia National Laboratories.
85) Shkurenkov, I., Burnette, D., Lempert, W. R., & Adamovich, I. V. (2014). Kinetics of
excited states and radicals in a nanosecond pulse discharge and afterglow in nitrogen and
air. Plasma Sources Science and Technology, 23(6), 065003.
86) Garscadden, A., & Nagpal, R. (1995). Non-equilibrium electronic and vibrational
kinetics in H2-N2 and H2 discharges. Plasma Sources Science and Technology, 4(2), 268.
87) Allen, D. C., Chandler, E. T., Gregory, E. A., Siddles, R. M., & Simpson, C. J. S. M.
(1980). Vibrational deactivation of N2 (v= 1) by n-H2 and by p-H2 in the temperature range
300-80 K. Chemical Physics Letters, 76(2), 347-353.
88) Frederickson, K., Hung, Y. C., Lempert, W. R., & Adamovich, I. V. (2016). Control
of vibrational distribution functions in nonequilibrium molecular plasmas and high-speed
flows. Plasma Sources Science and Technology, 26(1), 014002.
89) Light, G. C. (1978). The effect of vibrational excitation on the reaction of O( 3P) with
H2 and the distribution of vibrational energy in the product OH. The Journal of Chemical
Physics, 68(6), 2831-2843.
90) Gorse, C., Capitelli, M., Bacal, M., Bretagne, J., & Lagana, A. (1987). Progress in the
non-equilibrium vibrational kinetics of hydrogen in magnetic multicusp H− ion
sources. Chemical physics, 117(2), 177-195.
227
91) Petrishchev, V., Leonov, S., & Adamovich, I. V. (2014). Studies of nanosecond pulse
surface ionization wave discharges over solid and liquid dielectric surfaces. Plasma
Sources Science and Technology, 23(6), 065022.
92) Yagi, I., Ono, R., Oda, T., & Takaki, K. (2014). Two-dimensional LIF measurements
of humidity and OH density resulting from evaporated water from a wet surface in plasma
for medical use. Plasma Sources Science and Technology, 24(1), 015002.
93) Popov, N. A. (2008). Effect of a pulsed high-current discharge on hydrogen-air

mixtures. Plasma Physics Reports, 34(5), 376-391.
94) Lietz, A. M., & Kushner, M. J. Addressing plasma-liquid interactions in a global

model.
95) Goldberg, B. M., Shkurenkov, I., O’Byrne, S., Adamovich, I. V., & Lempert, W. R.
(2015). Electric field measurements in a dielectric barrier nanosecond pulse discharge with
sub-nanosecond time resolution. Plasma Sources Science and Technology, 24(3), 035010.
228

Winters Final

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Winters Final

Uploaded by

Copyright:

Available Formats

Laser Diagnostics of Reacting Molecular Plasmas for Plasma Assisted

Caroline Winters, M.S.

Graduate Program in Aeronautical and Astronautical Engineering

The Ohio State University

Dr. Igor Adamovich, Advisor

Dr. Jeffrey Sutton

Dr. Seung Hyun Kim

Caroline April Winters

assisted fuel oxidation in hydrogen-oxygen-argon and hydrocarbon-oxygen-argon

(TDLAS), and used to characterize plasma uniformity. Temperature in the discharge-

excited reacting flow is measured by Rayleigh scattering. Two-photon Absorption Laser

Induced Fluorescence (TALIF) is used to measured absolute H and O atom number

is oxidized in the discharge. Kinetic modeling analysis identified dominant processes of

At the present low-temperature conditions, the effect of chain branching reactions on

plasma-assisted fuel oxidation kinetics is insignificant.

OH number density by Laser Induced Fluorescence (LIF), as well as measurements of

translations-rotational temperature and nitrogen vibrational temperature by ns broadband

objective of these measurements is to study kinetics of OH radicals at the conditions of

shows that OH kinetics is controlled primarily by reactions of H 2 and O2 with O and H

affected by N2 vibrational excitation directly, e.g. vibrational energy transfer to HO2.

of radical species reactions in nanosecond pulse discharges sustained at a liquid-vapor

measurements of spatial distributions of absolute OH and H atom number densities in near-

surface, repetitive nanosecond pulse discharge plasmas.

To the following people, I dedicate this work:

people, and for that I am ever thankful:

To all my teachers throughout the years: you expanded my knowledge, challenged

remain the best physics teacher I have ever had.

November 7th, 1989........................................Born- Los Angeles, CA USA

2011................................................................B.S. Mechanical Engineering, Rose-Hulman

2016................................................................M.S. Aerospace Engineering, The Ohio

2015 to present ..............................................Graduate Research Intern, Diagnostic

Science and Engineering, Sandia National

2012 to present ..............................................Graduate Research Assistant, Department of

Mechanical and Aerospace Engineering,

The Ohio State University

Major Field: Aeronautical and Astronautical Engineering

Publications ....................................................................................................................... vii

Fields of Study .................................................................................................................. vii

Table of Contents ............................................................................................................. viii

List of Tables ..................................................................................................................... xi

Chapter 1: Introduction ....................................................................................................... 1

1.1. Nonequilibrium plasmas for plasma assisted combustion applications ................... 1

1.2. Review of previous studies of nonequilibrium plasmas in ns pulse discharges..... 12

1.3. Effect of vibrational excitation on plasma chemical reaction kinetics................... 26

1.4. Liquid/Vapor interfaces and plasma medicine ....................................................... 31

1.5. Structure of dissertation ......................................................................................... 34

Chapter 2: Background on Laser Diagnostics .................................................................. 35

2.2. Two-Photon Absorption Laser Induced Fluorescence ........................................... 45

2.3. Tunable Diode Laser Absorption Spectroscopy..................................................... 56

Chapter 3: Measurements of metastable and radical species in repetitive ns pulse

discharge plasmas in fuel-oxidizer mixtures..................................................................... 62

3.1. Plasma flow reactor design .................................................................................... 63

3.2. Laser Diagnostics Setup ......................................................................................... 69

3.3. Discharge Waveforms and Plasma Images ............................................................ 74

3.4. Plasma assisted combustion kinetic model ............................................................ 85

3.5. Measurements of Ar metastable atoms .................................................................. 86

3.6. TALIF results and discussion................................................................................. 96

Chapter 4: Measurements of OH number density at the conditions of strong vibrational

nonequilibrium ................................................................................................................ 133

4.1. Discharge Test Cell and Electrode Assembly ...................................................... 134

4.2. Pulsed discharge waveforms and plasma images ................................................. 137

4.3. Laser Induced Fluorescence Experimental Setup ................................................ 144

4.4. Coherent Anti-Stokes Raman Spectroscopy Setup .............................................. 152