See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/233813127

Optical emission spectroscopy study in the VUV–VIS regimes of a developing

low- temperature plasma in nitrogen gas

Article  in  Journal of Physics D Applied Physics · November 2012

DOI: 10.1088/0022-3727/45/49/495202

CITATIONS READS

34 2,796

3 authors, including:

Andrew Fierro Andreas A Neuber

University of New Mexico Texas Tech University
58 PUBLICATIONS   296 CITATIONS    485 PUBLICATIONS   3,544 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Design Optimization of MFCG View project

Investigation of Plasma in Pulsed High-Current Electron Sources and High Power Microwave Devices View project

All content following this page was uploaded by Andreas A Neuber on 21 May 2014.

The user has requested enhancement of the downloaded file.

Home Search Collections Journals About Contact us My IOPscience

Optical emission spectroscopy study in the VUV–VIS regimes of a developing low-

temperature plasma in nitrogen gas

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2012 J. Phys. D: Appl. Phys. 45 495202

(http://iopscience.iop.org/0022-3727/45/49/495202)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 129.118.19.71
The article was downloaded on 03/12/2012 at 15:00

Please note that terms and conditions apply.

IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS
J. Phys. D: Appl. Phys. 45 (2012) 495202 (11pp) doi:10.1088/0022-3727/45/49/495202

Optical emission spectroscopy study in

the VUV–VIS regimes of a developing
low-temperature plasma in nitrogen gas
A Fierro, G Laity and A Neuber
Center for Pulsed Power and Power Electronics, Department of Electrical and Computer Engineering and
Department of Physics, Texas Tech University, Lubbock, TX 79409, USA
E-mail: andrew.fierro@ttu.edu

Received 31 August 2012, in final form 8 October 2012

Published 14 November 2012
Online at stacks.iop.org/JPhysD/45/495202

Abstract
The mechanisms leading to the development of an atmospheric low temperature plasma along
a surface under pulsed conditions is of current interest. In the early plasma phase, high energy
photons are a contributing factor to the process of generating electron avalanches resulting in
surface flashover. Since only photons in the vacuum ultraviolet (VUV) regime are energetic
enough to cause step-ionization or direct ionization of atmospheric gases, an experiment has
been set up to enable observations of photons with wavelengths shorter than 200 nm while still
allowing observation up to 800 nm. A spectrum simulation software package has been
developed to allow for temperature analysis on the developing plasma in the VUV region.
Observations below 200 nm revealed a Boltzmann distributed excited state population
corresponding to a temperature of 3.1 eV. Time-resolved emission spectroscopy measurements
of the entire electrode region during the streamer phase of breakdown demonstrate the
presence of molecular nitrogen emission lines from the second positive system. Further
photomultiplier tube measurements of the spark phase showed a rapid decrease in intensity of
the second positive system compared to that of a representative atomic emission line in the
VUV regime. This emission dominates the ultraviolet–visible (UV–VIS) spectrum during the
initial phases of breakdown with little detection of other sources of emission during this phase.
(Some figures may appear in colour only in the online journal)

1. Introduction a developing plasma limits the peak power output of high

The role of photons contributing to ionization of gas has
been discussed in texts such as [1]. Recent works [2–4]
have attempted to further quantify the role of photoionization
in atmospheric discharges. Due to the relatively high
ionization energies of molecular oxygen and nitrogen (13.6 eV
and 14.5 eV [5], respectively), only light in the vacuum
ultraviolet (VUV) regime is energetic enough to promote
direct or two-step ionization in the gas volume. Owing
to the strong absorption of VUV radiation by molecular
oxygen [6], observation of such energetic photons is difficult.
Nevertheless, it has been shown previously [7, 8] that VUV
radiation is prominently emitted during the initial stages of a
pulsed atmospheric discharge.
The effect of high energy radiation on breakdown is of
interest in fields such as microwave-induced flashover where
power systems [9, 10]. Medical applications for the generation
of plasmas include the sterilization of bacteria [11]. The
large number of ionized and reactive species generated by
pulsed discharges provide efficient sterilization of the sample.
Furthermore, corona discharges provide ionized species for the
purpose of silicon processing [12].
In a pulsed discharge, as initial free electrons in the
volume travel to the anode due to the applied macroscopic
electric field, some gain enough energy to cause excitations
and ionizations in the high-field region. Such excitation is
virtually always accompanied by spontaneous emission of
photons from the high-field region, thus appearing as streamer
like objects in high-speed photographs. The velocity of these
cathode-directed streamers is on the order of 107 cm s−1 [13],
roughly three orders of magnitude below the speed of light.

0022-3727/12/495202+11$33.00 1 © 2012 IOP Publishing Ltd Printed in the UK & the USA
J. Phys. D: Appl. Phys. 45 (2012) 495202 A Fierro et al

Hence, if the involved spontaneous emission occurs on a sub-

microsecond timescale, light propagation can affect streamer
propagation in centimetre sized or larger gaps, and may
provide additional electrons to enhance streamer formation and
dominate breakdown characteristics.
A symmetrical point-to-point geometry was chosen for
basic study of optical emission in controlled conditions.
This differs from more typical configurations such as point-
to-wire [14] or point-to-plane [15, 16] often utilized for
corona and streamer studies. These more common electrode
configurations display large positive streamer branching due
to the more uniform field profile near the cathode. In the
early stages of streamer formation, the point-to-point geometry
produces visible (VIS) emission that appears to be dominated
by the field-enhanced anode similar to the discharges created
with the asymmetrical geometries [17]. Although branching
still occurs during the later stages of breakdown, as the
discharge progresses towards the spark phase a combination
of cathode and anode-directed streamers is observed in the Figure 1. Energy level diagram for molecular and atomic species of
symmetrical geometry. That is, during the later stages, a main nitrogen showing processes of interest.
streamer begins to develop between anode and cathode known
as the leader. Eventually, this elongated streamer connects
the anode to cathode and the discharge transitions from the
streamer phase into the spark phase.
Previous spectral analysis was conducted on a long,
microsecond timescale [18] investigating the entire discharge.
Furthermore, experimental apparatus limits prevented attempts
to conduct spectra analysis in other regimes such as
the ultraviolet (UV) or VIS regions. As a result
of experimental improvements, these limitations have
been mitigated and subsequent time-resolved emission
spectroscopy measurements have been conducted from the
VUV through VIS regimes. From observations in the
VUV region, the developing plasma produces a Boltzmann
distributed population density of excited states in the early
spark phase. That is, after voltage collapse an approximated
electronic temperature may be calculated using a locally Figure 2. Experimental setup overview. A: location of flashover
developed software package. Emission in the UV regime chamber. B: off-axis parabolic mirrors to correctly focus light from
is dominated by the nitrogen second positive system which 115 to 800 nm. C: diffraction grating turret. D: optical emission
detectors including a vacuum mounted ICCD and PMT.
has been observed to have a shorter lifetime than measured
atomic emission and is attributed to quenching from collisions
with ground state molecular nitrogen. Lastly, the VIS region such, it is likely some emission is self-absorbed in the plasma
displays no significant emission during the first several hundred and is currently under investigation in more detail [20].
nanoseconds of the developing plasma. This is primarily due
to the absence of strong radiative transitions of both molecular 2. Experimental setup
and atomic species of nitrogen with energies in this region.
Figure 1 shows some of the processes of interest for a pure To observe the spectral regime from VUV to VIS (115 to
nitrogen discharge involving molecular and atomic species. 800 nm), an experiment was constructed (figure 2) to correctly
For molecular nitrogen, the radiative transitions from the focus all wavelengths of light in the spectrum of interest and
second positive system (C 3
u → B 3
g ) are apparent in the alleviate the chromatic aberration encountered when using a
discharge. Quenching of the C3 state by ground state molecular focusing lens [21]. The sharpened point-to-point electrodes
nitrogen has been observed. Vibrational levels with v  > 4 in resting on a dielectric surface are excited using a solid-state
the C3 state are known to be strongly predissociated [19] and high voltage pulser with a peak voltage of 40 kV and a rise
are thus not observed during the discharge. time of a few hundred nanoseconds.
If the diatomic nitrogen becomes dissociated (minimum Light emitted from the flashover event (placed at
energy of 9.87 eV), emission from the atomic species may be figure 2A) is transmitted into vacuum (10−7 Torr) through a
observed. One prevalent emission line from atomic nitrogen MgF2 dielectric interface and focused onto the entrance slit
has been identified at 120 nm and is a ground transition. As of the VM-504 spectrograph by two off-axis parabolic mirrors

2
J. Phys. D: Appl. Phys. 45 (2012) 495202
(locations of figure 2B) with greater than 80% efficiency in the
VUV regime. Light is resolved by the diffraction grating turret
(figure 2C) and can be detected by either an ICCD camera or
photomultiplier tube (PMT) instrument (figure 2D). A second
ICCD may be mounted in atmosphere for fast-gated imaging.
For spectroscopy measurements, an Andor 740 series
ICCD is utilized. The camera provides a minimum gate time
of 50 ns, pixel matrix size of 2048 × 512, and peak quantum
efficiency of 18%. Nanosecond gated external imaging of
the developing plasma is captured with an Andor 734 series
ICCD capable of imaging with gate times less than 5 ns.
Supplemental optical diagnostics include a Hamamatsu R8486
PMT sensitive to wavelengths below 300 nm. This VUV-
sensitive PMT provides a rise time of 2 ns and a gain of 107 at a
supply voltage of 1100 V. For wavelengths exceeding 300 nm,
a Hamamatsu PMT model H9305-01 is used for measurements
in the UV through VIS regimes.
Electrical diagnostics include an optically isolated high
voltage probe and a Rogowski coil for current measurement.
The entire system is triggered via a low-voltage TTL pulse
and electrical delays are studied prior to experimentation for
nanosecond resolution data analysis. The system pulse to pulse
jitter is about 1 ns.
3. Spectrum simulation
Simulated spectra for atomic emission of various gases
are calculated using a locally developed software package.
Specifically, the modelling of atomic emission with
wavelength shorter than 180 nm is of interest, although the
software is capable of calculations through every regime where
the necessary input data is available. It is assumed that
the discharge is in local thermodynamic equilibrium (LTE).
First, the population of excited states and line intensity ratios
assuming a Boltzmann distribution are calculated via
ni Aij gi e−Ei /kb T
= , (1)
nj Q(T )
where Aij , gi and Ei are the Einstein coefficient for
spontaneous emission, degeneracy and energy of the excited
state, all available from [22] and are delivered to the program in
a specific file format generated by an additional computer code.
The Boltzmann constant is kb and the electronic temperature
is T . The partition function, Q(T ), is computed automatically
by the software by summing over all possible states.
Measurements of emission spectra typically observe
broadening of lines due to the limited experimental apparatus
resolution as well as physical processes such as Doppler
broadening or pressure broadening. As such, to further
accurately model the measured spectrum, several line
broadening parameters are also included in the program in
the form of generic Gaussian and Lorentzian profiles. The
software also accepts input for a square profile to represent an
optical slit or a square detector pixel profile. Convolution of
the Gaussian and Lorentzian broadening profiles results in a
Voigt profile. A second convolution incorporates the square
parameter into the simulated spectra. Details of simulating
atomic spectra are discussed in the appendix.
3
A Fierro et al
Figure 3. Emission spectrum in the VUV region before and after
voltage collapse. Dashed lines indicate the camera gate of 50 ns.
Thus, if the electronic temperature and apparatus spectral
sensitivity are known, then a simulation of atomic emission
characteristics assuming a Boltzmann distributed discharge
may be computed that accurately predicts what will be
experimentally observed. Furthermore, parametric analysis
to fit a simulated spectrum to a measured spectrum will yield
broadening parameters as well as electronic temperature.
4. Experimental results
4.1. VUV emission spectrum
Several atomic nitrogen emission lines in the VUV region
have been observed (figure 3) with a 50 ns camera exposure.
The spectra prior to voltage collapse (figure 3(a)) shows the
identifiable emission lines at 131.0, 131.9, 141.2 and 149.2 nm.
The emission measured near 131 nm appears to be more intense
compared to the emission line at 141.2 nm and 149.2 nm
with upper energy levels of approximately 13 eV, 12.4 eV and
10.7 eV respectively. This observed line intensity implies that
population of excited states during this stage of the discharge
is not Boltzmann distributed and the configuration of the
upper energy level for 131 nm (2s2 2p2 (3 P)3d) is preferentially
populated.
However, after voltage collapse (figure 3(b)), the intensity
ratio of 149.2 nm to the other three observable emission lines
more closely follows a Boltzmann distributed population.
Although the emission near 131 nm appears stronger than
141.2 nm, one must consider the effect of their respective
transition probabilities (approximately 7 × 107 s−1 compared
to 5 × 107 s−1 ) in addition to the energy of the upper level.
These two parameters largely determine the intensity ratios
observed.
Since the spectrum after voltage collapse (figure 3(b)) is
close to a Boltzmann distributed population, the previously
discussed software package may be used to determine the
electronic temperature of the plasma during this time frame.
By fitting an appropriate instrument function with σ =
0.12 nm for the Gaussian function, γ = 0.40 nm for the
Lorentzian function, and  = 0.05 nm for the square profile,
an approximate electronic temperature was obtained (figure 4)
J. Phys. D: Appl. Phys. 45 (2012) 495202
Figure 4. Experimental and simulated spectrum of atomic nitrogen
emission between 130 and 150 nm.
Figure 5. Root mean square error for the difference between
simulated spectra at different temperatures and the measured
spectrum. The dashed line represents the minimum at 3.1 eV.
to be 3.1 eV. The program accurately models the line intensity
ratios as well as the observed line broadening, see the appendix.
To increase the confidence of the temperature measure-
ment, multiple temperatures were computed assuming a con-
stant instrument profile and compared against the measured
spectrum. A root mean square error is used to measure the
difference between values of the calculated spectrum and the
measured spectrum (figure 5). It is clear that around 3 eV
electronic temperature, a minimum error is achieved. Elec-
tronic temperatures below 3 eV have the largest error and tem-
peratures above 3 eV experience a gradual increase in error
towards 30 eV.
The time required for the excited state population to
respond to a sudden temperature change and be described by
the Boltzmann distribution is defined as the relaxation time
(τrel ). The relaxation time is, using a first order approximation,
dependent upon the electron density and given by the time of
the slowest process in the plasma. As a result,
1 measurement.
τrel ≈ , (2)
ne νexc
with ne representing the electron density and νexc the collisional
excitation rate. Assuming an electron density of 1017 cm−3
and an excitation rate of 2.20 × 10−9 cm3 s−1 , τrel is found
4
A Fierro et al
Figure 6. VUV–UV emission spectrum taken at different times
referenced to voltage collapse. Dashed lines indicate the camera
gate of 50 ns.
to be 4.5 ns. Because the recorded population is observed
to become Boltzmann distributed within the camera gate of
50 ns, a relaxation time on the order of nanoseconds to tens
of nanoseconds would be expected which the relaxation time
calculated agrees with.
4.2. UV emission spectrum
The VUV–UV spectrum for pure nitrogen with a 50 ns camera
gate was taken at different times throughout the discharge
(figure 6). To capture the spectrum from 100 to 350 nm in
a single measurement, a 150 g mm−1 diffraction grating was
used. However, the measured spectrum intensity suffers in
the VUV regime due to the much decreased efficiency of this
diffraction grating below 200 nm.
The spectrum taken approximately 50 ns prior to voltage
collapse (figure 6(a)) contains several UV lines identified as
the second positive system from molecular nitrogen. This
spectrum has been magnified by a factor of five relative
to figure 6(b). Using a simulation developed for the
second positive system [23], the vibrational and rotational
temperatures were determined to be 2700 K and 325 K,
respectively. The insert on this graph contains a background
corrected accumulation of twenty measurements with a higher
efficiency diffraction grating (1200 g mm−1 ) between 125 and
155 nm to capture the VUV content in this range. This
measured spectrum contains the previously identified atomic
nitrogen emission lines (see figure 4).
In the subsequent spectrum taken approximately at
voltage collapse (figure 6(b)), the molecular nitrogen emission
lines reach their peak intensity and are clearly identifiable.
Furthermore, the VUV emission lines from atomic nitrogen
(174 and 149.5 nm) are captured without the need for a separate
The spectrum for 50 ns after breakdown (figure 6(c))
demonstrates a similar intensity of atomic nitrogen VUV
intensity recorded whereas the molecular nitrogen emission
intensity is significantly reduced. The remaining emission
lines at 298 and 348 nm are from second order interference
J. Phys. D: Appl. Phys. 45 (2012) 495202
Figure 7. Photons per second comparison between 120 nm atomic
nitrogen emission and the 337 nm molecular nitrogen emission.
Time t = 0 represents initial current rise.
produced by the observed atomic emission lines in the VUV
range.
By calculating the spectral efficiencies of the experimental
apparatus and applying the PMT’s specified anode radiant
sensitivity (calculated from gain and cathode radiant
sensitivity), the total intensity of light in watts is computed.
The total light intensity may be divided by the energy of
the photon of interest to yield photons per second which is
a comparable quantity. This calculation was completed for
representative lines from molecular nitrogen in the second
positive system regime and atomic emission in the VUV regime
(figure 7). The time t = 0 represents the initial current rise.
The second positive system (337 nm) appears first with
high initial intensity. At this time, small amounts of atomic
emission (120 nm) are also detected. This evidence suggests
that energetically, the electron temperature before breakdown
in the plasma is inadequate to cause atomic excitation through a
one-step process (dissociative excitation) as this would require
energy on the order of 20 eV. The electric field, however, is still
large enough to produce excited molecular nitrogen as well as
some atomic species through the process described by
e + N2 → e + N + N (3)
representing dissociation by electron collision. Electron
collisions with vibrationally excited molecular nitrogen may
also contribute to the overall atomic species density. After the
production of atomic N, a second electron collision must occur
to generate excited N given by the reaction
e + N → e + N∗ , (4)
where N∗ represents an excited atomic nitrogen. Because
the current is low before breakdown, there is insufficient
electron density to excite the newly generated atomic nitrogen
species. After breakdown the current, and thus electron
5
A Fierro et al
Figure 8. Electron collision cross sections for the excitation of the
second positive system, the dissociation of N2 , the excitation of the
atomic 120 nm transition, and the excitation of the 337 nm transition
of the second positive system [26, 27].
density, increases making the excitation of the atomic species
more probable and additionally not as efficiently exciting the
molecular species. Recent experiments have shown similar
atomic nitrogen behaviour at longer wavelengths [24].
The radiative lifetime of the 337 nm second positive
transition is estimated to be 42 ns [25] and for the 120 nm
atomic transition 2.5 ns [22]. Hence, one would expect, for
an identical upper energy level population, that the intensity
of emission from the atomic species to be roughly 15 times
stronger than the molecular emission and decay at a much faster
rate. However, it would be expected that the population of the
excited state atomic nitrogen is much less than that of excited
state molecular nitrogen. Regardless, the atomic nitrogen
intensity is observed to be stronger than the intensity from
molecular nitrogen. The excitation cross section via electron
collision for 120 nm emission line is larger than that for the
337 nm emission line (figure 8). Calculating the excitation
rate for these two transitions assuming a Maxwell–Boltzmann
velocity distribution with a temperature of 3.1 eV yields an
excitation rate of 2.20 × 10−9 cm3 s−1 for the 120 nm emission
line and 7.03 × 10−11 cm3 s−1 for the 337 nm transition.
Neither observed emission decays on a fast, few nanoseconds
timescale, but both oscillate somewhat proportionally to the
current magnitude. It appears that the electron energy
distribution function is still hot enough during this spark phase
such that molecular and atomic species are still excited through
electron collisions.
It is noted, however, that the initial peak intensity of
the molecular emission is much stronger than the subsequent
oscillation. As shown for the first 150 ns in figure 9, the initial
peak of the second positive system decreases much faster than
the radiative transition probability alone indicates. The rate
equation given by
dn2
= −A21 n2 − kq n2 n0 (5)
dt
defines the reduction in population of an excited state n2 as a
function of time due to the rate of photon emission (A21 ) and
J. Phys. D: Appl. Phys. 45 (2012) 495202
Figure 9. PMT Signal for the 337 molecular nitrogen emission line.
Also shown is the calculated intensity of the excited state assuming
only radiative transitions, A = 2.3 × 107 s−1 (dashed), radiative
transitions and quenching rate, 1.3 × 1011 cm3 s−1 , from the
literature [25] (dotted), and radiative transitions and a best fit
quenching rate, 1.0 × 1011 cm3 s−1 , (dashed–dotted).
the rate of quenching (kq ) due to excited state particle collisions
with ground state molecules (n0 ).
Assuming atmospheric pressure and room temperature,
n0 is determined from ideal gas law, and equation (5) is
solved using the appropriate quenching rate analytically as
a simple exponential function. A best fit match was found
for a quenching rate of 1.0 × 10−11 cm3 s−1 , which is a bit
slower than the quenching rate reported elsewhere, 1.3 ×
10−11 cm3 s−1 [27]. This is not too surprising considering the
finite voltage and current fall times as well as the non-zero
current after the initial peak still driving exciting collisions.
4.3. UV–VIS emission spectrum
Analysis of the spectra emitted during the early stages of
surface flashover was conducted from the UV through VIS
regimes (250–750 nm). During these phases, streamers
propagate from anode to cathode resulting from the positive-
ion space charge left in the wake of fast moving electrons
towards the anode. This build-up of space charge results
in a field-enhanced region capable of producing high energy
photons from molecular and atomic species via electron
collision.
The emission spectrum was recorded for the streamer
phase using a 50 ns camera gate and the 150 g mm−1 grating
(figure 10). Figure 10(a) shows the spectra taken 88 ns prior
to voltage collapse and displays emission between 300 nm and
400 nm as well as weaker emission between 600 and 725 nm.
This emission has been identified as the second positive system
with the emission between 600 and 725 nm resulting from
second order interference. During this time, the 0–0 transition
(v  , v  = 0) at 337 nm is the strongest emission detected.
The subsequent emission spectrum taken at 58 ns prior
to voltage collapse (figure 10(b)) shows identical emission
lines except with stronger intensity. The increasing voltage
6
A Fierro et al
Figure 10. Time-resolved spectroscopy between 250 and 750 nm
taken with a 50 ns camera gate for (a) −88 ns prior to voltage
collapse and (b) −58 ns prior to voltage collapse.
Figure 11. External imaging taken with 10 ns camera gate showing
the similar developing streamer characteristics for (a) no optical
filter and (b) UG11 filter. Times are referenced to voltage collapse.
and electron density during this time leads to an increased
electronic excitation rate in the developing plasma.
In addition, streamers were captured using the externally
mounted ICCD with a 10 ns camera gate (figure 11). Two
sets of images were captured and compared to supplement the
observed emission spectrum. The first set represents imaging
taken with no optical filter and the camera is allowed to capture
light emission from 300 nm to 800 nm (figure 11(a)). The
second set of images was taken by impeding light emission to
the camera with a UG11 filter in the light path (figure 11(b)).
This filter has a pass band between 250 and 400 nm thus
allowing mainly second positive emission to be captured by
the ICCD. The times shown on each image are referenced
to voltage collapse. At approximately 50 ns before voltage
collapse, both sets of images show emission from the cathode
as well as developing streamers from the anode. Later in the
discharge, at 30 ns prior to voltage collapse, the first streamer
reaches the cathode producing strong emission in this region
observable on both images.
There is no evidence of alternate emission sources
in the UV–VIS regimes during the streamer phase of
surface flashover. Consequently, the results of time-resolved
spectroscopy and nanosecond gated imaging reveal that most
J. Phys. D: Appl. Phys. 45 (2012) 495202
light emission from 250 to 750 nm is from the nitrogen second
positive system.
5. Atomic ground state density
It was shown that the intensity of emission from atomic
species contributes largely to the observed spectrum in a
pure nitrogen discharge. As such, it is desired to calculate
the total number of ground state atomic nitrogen that is
present after voltage collapse. The photons per second for an
excited transition (in this case the emission line at 149.2 nm)
may be calculated for a measured PMT signal (analogous
to figure 7). Under the assumption of complete LTE, the
Boltzmann distribution for excited states multiplied by the
Einstein coefficient, equation (1), gives the ratio of photons per
second for the excited state and ground state. Consequently,
substituting the measured photons per second into equation (1)
and solving for the ground state yields the total number of
atoms in the ground state. Defining the plasma size to be
5 mm in length and 100 µm in width and height results in an
atomic ground state density of 1.1 × 1016 cm−3 .
However, it is unlikely that our plasma is under complete
LTE due to the transient and non-homogeneous behaviour.
Considerable effort has been made towards determining the
conditions required for complete LTE or partial LTE (pLTE),
first by [28] and later by others [29, 30] for different plasma
conditions. It is possible for some excited states to be
thermalized and follow a Boltzmann distribution defined by
a common temperature [31]. As a result, an alternate method
for determining the atomic ground state density was used. The
rate equation for the population of an excited atomic state is
given by
dnN∗
= nN νexc ne − (A + Q)nN∗ , (6)
dt
where nN∗ is the excited state density, nN is the atomic ground
state density, ne is the electron density, A is the coefficient for
spontaneous emission, and Q = kq n0 representing the product
of the quenching rate of the atomic species and ground state
molecular nitrogen. A second rate equation for the population
of the excited molecular species is as follows:
dnCstate
= n0 ν̃exc ne − (Ã + Q̃)nCstate , (7)
dt
with nCstate representing the excited molecular state density
of the C3(v =0) level, n0 being the molecular ground state
density, and the other variables representing values similar
to equation (6) except for the excited molecular nitrogen.
Equation (7) represents a simplified model for the population
of the C3(v =0) state due to the complexity of molecular energy
levels and numerous possible transitions among excited states.
First, it is assumed that the excitation rate for the C3(v =0)
state includes all vibrational levels of the ground state X1 .
Excitation paths through the lower B3 and metastable A3
electronic states could also contribute to the C3(v =0) state
population. However, excitation into the C3 state due to
collisions between metastable states has a reported reaction
rate of an order of magnitude slower than the excitation
7
A Fierro et al
rate from the ground state [32] and is thus neglected.
The faster radiative decay rate and small population under
Boltzmann conditions compared to the ground population of
the B3 state produce a negligible effect on the C3(v =0) state
population. Additionally, cascade processes from higher lying
energy states could affect the C3(v =0) state population under
consideration. However, evidence suggests that these higher
electronic levels are metastable and/or predissociated [19].
Using these simplified equations, it is assumed that the
rate equations are in equilibrium when the current reaches the
peak (see figure 7) and thus the left-hand side of equations (6)
and (7) are zero. Dividing equations (6) by (7) results in
nn ν̃exc A + Q nN∗
= ( ) , (8)
n0 νexc à + Q̃ nCstate
which represents the ratio of atomic ground state to molecular
ground state densities given the transition probabilities,
excitation rates, quenching rates, and excited state densities.
Using the equation
AnN∗ = β ÃnCstate , (9)
with β representing the ratio between the measured atomic
and molecular emission lines. Solving equation (9) for nN∗
and substituting into equation (8) yields


ν̃exc A+Q Ã
nN = β · n0 , (10)
νexc à + Q̃ A
which has been solved for the atomic ground state density, nN .
Due to the dependence of the atomic ground state density on the
excited atomic state quenching rate, a lower bound of atomic
density may be found assuming the quenching rate, Q, is
negligible. Using the ratio of 120 nm and 337 nm from figure 7
(β = 2), and the corresponding excitation rates and quenching
rates for the atomic transition and molecular transition, a
lower bound is found to be 1.1 × 1017 cm−3 and results in
a dissociation percentage of 0.43% at atmospheric pressure.
The ratio of the excitation rates depends on the electronic
temperature and could change rapidly due to the evolution
of the electron energy distribution function. However, it
was found (using the cross sections shown in figure 8 and
a Maxwell–Boltzmann velocity distribution) that this ratio
only varies at most by a factor of two for temperatures
between 1 and 8 eV. If an approximated quenching rate is
assumed (for the 859 nm atomic nitrogen transition [33]),
the dissociation percentage is several percent. Consequently,
equation (10) requires accurate knowledge of the excitation and
quenching rates along with precise calibration of the measuring
instrument to provide accurate ratios. As such, it is estimated
that the dissociation percentage is likely within an order of
magnitude.
6. Conclusions
The optical emission from a developing low-temperature
plasma in the VUV–VIS regimes has been characterized. A
locally developed software package assists in temperature
J. Phys. D: Appl. Phys. 45 (2012) 495202 A Fierro et al

analysis of the developing plasma assuming the excited which serves as normalizing factor for the population densities.
state population is Boltzmann distributed. Time-resolved The transition frequencies, νk = c/λk , and corresponding
emission spectroscopy in the VUV region revealed a non- photon emission rates, Ik , are stored for later use in the
Boltzmann distributed population before voltage collapse program.
which transitioned into a Boltzmann distributed population The emission lines are broadened depending on the
with an electronic temperature of 3.1 eV. environment. Typically, a Lorentz profile is assumed for the
Further investigation of the UV regime indicated the natural line width
dominant presence of the molecular nitrogen second positive  
1 δνL
system. Since emission from atomic nitrogen appears later Lk (ν) = , (A2)
than the emission from nitrogen molecules, a two-step process π (ν − νk )2 + δνL2
is believed to be the leading mechanism for the generation
where νk represents the frequency of the line centre and δνL
of excited atomic nitrogen. First, dissociation of nitrogen
the half-width at half-maximum (HWHM). The factor 1/π is
molecules occurs due to electron impact, which is then
chosen such that the integral of the profile is normalized to
followed by a second electron collision that results in excitation
one. If there are few collisions with surrounding atoms or
of the dissociated nitrogen, followed by spontaneous emission
molecules the full-width at half-maximum (FWHM) of this
of VUV radiation. After voltage collapse, the increase in
profile is simply the sum of the reciprocal radiative lifetimes
current, and thus electron density, leads to the continuous
of the two involved energy levels divided by 2π .
emission of atomic and molecular emission even though the
Depending on temperature and the mass of the atom, mA ,
applied field is collapsed across the gap.
the spectral lines are also Doppler broadened, which leads to
Emission spectroscopy observations in the VIS regime
a Gauss shaped profile
showed only second order interference from the second
positive system. Subsequent streamer imaging using a   2
ln(2) 1 − ln(2) ν−ν k
nanosecond camera gate and UG11 optical filter confirmed that G(ν) = e δνD
, (A3)
virtually all emission between 200 and 800 nm results from the π δνD
nitrogen second positive system. with the Doppler HWHM
Calculation of the number of atomic species after


breakdown has been calculated under the assumption of νk 2kB Tk 1/2
complete LTE using the Boltzmann distribution which yielded δνD = ln(2) ,
c mA
a density of 1.1 × 1016 cm−3 . However, due the transient
and non-homogeneous nature of the developing plasma, an and the speed of light, c.
alternate method to calculate the ground state atomic density Convolving the Lorentz and Gauss profile results in a Voigt
through rate equations yielded a lower bound density of profile which is usually given by the integral
1.1 × 1017 cm−3 .
∞
e−y
2
δνL
V (x) = dy, (A4)
Appendix. Spectral simulation δνD2 π 3/2 −∞ (x − y)2 + a 2

with
The simulation software [34] was developed utilizing well ν0 − ν
known optical emission spectroscopy concepts as they are x = 2 ln(2) ,
δνD
found in [35] and elsewhere. This supplemental appendix
and
serves to describe in further detail the process of calculating δνL
atomic spectra accounting for radiation as well as apparatus a= ln(2) ,
δνD
profiles.
A specific atom or ionic state of an atom is chosen, where ν0 is the line centre in frequency, and νL and νD
and the following parameters are read for each line tabulated represent the HWHM’s of the Lorentzian and Gaussian profile
from [22]: line positions, λk ; energy of upper energy respectively.
levels, Ek , degeneracy of the upper energy levels, gk ; Since an analytical solution for (A4) does not
Einstein coefficient for spontaneous emission, Ak . Assuming exist, historically several algorithms have been developed
Boltzmann distributed energy levels at a temperature T and a to quickly provide a solution involving error function
total number of atoms, N0 , the number of photons emitted per approximations [36].
second for each tabulated transition, λk , is given by Alternatively, modern computers enable discrete convolu-
tion of the Lorentz, (A2), and Gauss, (A3), profiles if the data
Aj gj e−Ej /kB T e are stored in vectors given by the equation
Ik = N0 , (A0)
Q(T )
∞

with the Boltzmann constant, kB , and the partition function (L × G)[i] = L[m]G[i − m]. (A5)
m=−∞
∞

Q(T ) = gj e−Ej /kT , (A1) To provide even faster computation, this point-by-point
j =0 convolution was done using a parallel algorithm on the GPU

8
J. Phys. D: Appl. Phys. 45 (2012) 495202 A Fierro et al

allowing convolution of vectors with a size in the hundreds of Table A1. Sample data used to produce the emission spectrum
thousands in just a few seconds. produced in figure A1.
After a symmetric Voigt profile centred at zero frequency λ1 λ2 λ3
is calculated, an array containing the desired frequency range
λ (nm) 131.0950 131.9676 132.7924
at an interval of 100 times smaller than the specified FWHM’s
Ej (eV) 13.0332 12.9706 12.9123
is constructed. For any frequency point in this new array, gj 4 4 2
referred to as xi , the distance, i,k , away from the line centre, Aj (106 s−1 ) 17.5 57.6 14.3
νk , is calculated. This distance is translated to an offset Te (eV) 5 σ a (nm) 0.2
from the centre of the Voigt profile array either by solving Tk (K) 5000 γ b (nm) 0.2
(A4) or (A5) and represents a partial contribution of emission mA (amu) 16 c (nm) 0.05
δνL (cm−1 ) 1 Stepd (nm) 0.022 nm
rate to the frequency point xi . The total photon emission
rate per frequency interval at frequency xi is the sum of all a
FWHM of the Gauss apparatus profile.
contributions b
FWHM of the Lorentz apparatus profile.

n c
FWHM of the square apparatus profile.
I˜i = V (i,k )Ik , (A6) d
Wavelength covered by one pixel in array.
k=1
with n the total number of frequency and emission rate pairs spectrum is reduced to a larger step width
from (1); V () is the Voigt profile evaluated at distance 
from the line centres νk .
The spectrum calculated from (A6) is experimentally
observed only for the case that the spectral apparatus resolution
is much finer than the FWHM of the Voigt profile. In the
more general case, the spectrum of (A6) needs to be convolved
with the apparatus profile. Considering a grating spectrograph
in Czerny–Turner configuration, the dispersion is reasonably
constant in the wavelength domain; for instance, about
2 nm mm−1 for a 1/4 m spectrograph with a 1200 lines mm−1
grating. The photon emission rate per wavelength interval is
given by
c
Iˆi = 2 I˜i , (A7)
λi
which ensures that the total photon emission rate is conserved.
That is, the following integrals need to be equal
calculated spectrum from equation (A10).
Iˆ(λ) dλ = I˜(ν) dν. (A8)
The discretized version of (A8) is utilized as a check that the
conversion from the frequency to wavelength domain is done
correctly.
It has been found that the apparatus profile for a Czerny–
Turner spectrograph with entrance slit and intensified CCD
camera as readout device can be conveniently expressed with
a combination of Lorentz, Gauss, and square profile. Note that
the associated profile widths are a property of the apparatus
and not to be confused with the earlier utilized spectral line
widths describing properties of the radiating atoms. Typically,
one determines the apparatus profile, S(λ), by measuring an
isolated line and then fitting the profile parameters such that
a good match is observed. The apparatus convolved spectrum
is then calculated by simply convolving the area normalized
S(λ) with the array given in (A7)
∞


I i= S(i,k )Iˆk , (A9)
k=1
where S() is the apparatus profile evaluated at a distance 
from line centre. Since the size of the individual pixels of the
CCD in the wavelength regime is typically much larger than
the wavelength spacing of the spectrum of equation (A9), the
9
 
m
I i= I k, (A10)
k=1
with m representing the size of the larger step width and
incrementing k by m for every larger interval i. Noting that
the ICCD detects photons rather than intensity, the output
is typically listed in counts, equation (A10) can be directly
compared to the experimental spectrum provided that one
accounts for the apparatus specific wavelength sensitivity.
Note that for the temperature determination only the shape
of the spectra is important. That is, the temperature is varied
until a best match is achieved. However, if the wavelength
sensitivity as well as the measured solid angle of the apparatus
is accurately known, the number of emitting atoms, N0 , can be
as well determined from comparing the experimental with the
Three spectral lines and their emission properties are pre-
sented in table A1. The HWHM for the natural broadening
(δνL ) was chosen to be a constant for all lines regardless of
Einstein coefficient for spontaneous emission due to its neg-
ligible influence on the final profile. In addition, the kinetic
temperature of the radiating species (Tk ) and mass (mA ) for
the Doppler width are given. Lastly, the generic Gauss (σ ),
Lorentz (γ ), and square () profile FWHMs values determine
the apparatus profile. This data was used as input to the soft-
ware to demonstrate the process of calculating atomic emission
spectra (figure A1). Figure A1(a) represents the photons per
second calculated from equations (A1) and (A2). Figure A1(b)
shows the line spectrum due to only natural and Doppler line
widths. Note that Tk has been raised to a somewhat excessive
5000 K just to be able to observe the linewidth at the depicted
scale at all. Figure A1(c) shows the final spectrum convolved
with the apparatus profile usually used to determine tempera-
ture. Note that the apparatus profile frequently dominates any
Doppler or natural line broadening as it does here.
Acknowledgments
This work was supported by the Air Force Office of Sci-
entific Research (AFOSR) through the grant ‘Basic Physics
of Distributed Plasma Discharge,’ with additional student
J. Phys. D: Appl. Phys. 45 (2012) 495202
Figure A1. Steps taken to produce the final spectrum of
equation (A10). (a) Photons per second calculated from
equations (A0)–(A1). (b) Application of natural and Doppler line
widths from equations (A2)–(A6). (c) Spectrum convolved with
apparatus profile from equations (A7)–(A10).
fellowship support from the National Physical Science Consor-
tium (NPSC) in partnership with Sandia National Laboratories.
References
[1] Nasser 1971 Fundamentals of Gaseous Ionization and Plasma
Electronics (New York: Wiley-Interscience)
[2] Pancheshnyi S, Nudnova M and Starikovskii A 2005
Development of a cathode-directed streamer discharge in air
at different pressures: experiment and comparison with
direct numerical simulation Phys. Rev. E 71 016407
[3] Kulikovsky A 2000 The role of photoionization in positive
streamer dynamics J. Phys. D: Appl. Phys. 33 1514
[4] Liu N and Pasko V 2006 Effects of photoionization on
similarity properties of streamers at various pressures in air
J. Phys. D: Appl. Phys. 39 327
[5] Fridman A 2004 Plasma Physics and Engineering (Boca
Raton, FL: CRC Press)
[6] Watanabe K, Inn E and Zelikoff M 1953 Absorption
coefficients of oxygen in the vacuum ultraviolet J. Chem.
Phys. 21 1026–30
[7] Fierro A, Laity G, Hatfield L, Dickens J, and Neuber A 2011
Advanced imaging of pulsed atmospheric surface flashover
Proc. 18th IEEE Int. Pulsed Power Conf. (Chicago, IL,
19–23 June 2011) pp 199–202
[8] Laity G, Fierro A, Hatfield L, Dickens J and Neuber A 2011
Spatially resolved vacuum UV spectral imaging of pulsed
atmospheric flashover IEEE Trans. Plasma Sci. 39 2122–3
[9] Ford P, Beeson S, Krompholz H and Neuber A 2012 A
finite-difference time-domain simulation of high power
microwave generated plasma at atmospheric pressure Phys.
Plasmas 19 073503
[10] Beeson S, Ford P, Foster J, Krompholz H and Neuber A 2011
Imaging of pressure-dependent high-power microwave
surface flashover IEEE Trans. Plasma Sci. 39 2600–1
[11] Choi J, Han I, Baik H, Lee M, Han D, Park J, Lee I, Song K
and Lim Y 2006 Analysis of sterilization effect by pulsed
dielectric barrier discharge J. Electrostat. 64 17–22
10
A Fierro et al
[12] Ikeuchi T, Takahashi K, Ohkubo T and Fujiwara T 2011
Neutralization by a corona discharge ionizer in nitrogen
atmosphere Electr. Eng. Japan 177 1–8
[13] Allen N and Ghaffar A 1995 The conditions required for the
propagation of a cathode-directed positive streamer in air J.
Phys. D: Appl. Phys. 28 331
[14] van Veldhuizen E, Kemps P and Rutgers W 2002 Streamer
branching in a short gap: the influence of the power supply
IEEE Trans. Plasma Sci. 30 162–3
[15] Briels T, Kos J, Winands G, van Veldhuizen E and Ebert U
2008 Positive and negative streamers in ambient air:
measuring diameter, velocity, and dissipated energy J. Phys.
D: Appl. Phys. 41 234004
[16] Ono R and Oda T 2003 Formation and structure of primary
and secondary streamers in positive pulsed corona
discharge—effect of oxygen concentration and applied
voltage J. Phys. D: Appl. Phys. 36 1952–8
[17] Rogers T, Neuber A, Frank K, Laity G and Dickens J 2010
VUV Emission and streamer formation in pulsed dielectric
surface flashover at atmospheric pressure IEEE Trans.
Plasma Sci. 38 2764–70
[18] Laity G, Neuber A, Fierro A, Dickens J and Hatfield L 2011
Phenomenology of streamer propagation during pulsed
dielectric surface flashover IEEE Trans. Dielectr. Electr.
Insul. 18 946–53
[19] Lofthus A and Krupenie P 1977 The spectrum of molecular
nitrogen J. Phys. Chem. Ref. Data 6 113–307
[20] Laity G, Fierro A, Hatfield L, Neuber A, Dickens J and Frank
K 2012 Investigation of vacuum UV absorption during
low-temperature plasma formation in N2 /H2 mixtures at
atmospheric pressure 2012 IEEE IPMHVC Conf. (San
Diego, CA, 3–7 June 2012) at press
[21] Laity G, Neuber A, Rogers G and Frank K 2010 System for
time resolved spectral studies of pulsed atmospheric
discharges in the visible to vacuum ultraviolet range Rev.
Sci. Instrum. 81 083103
[22] Kramida A, Ralchenko Y, Reader J and NIST ASD Team 2012
NIST Atomic Spectra Database (ver. 5.0), [Online].
http://physics.nist.gov/asd
[23] Edmiston G, Krile J, Neuber A, Dickens J and Krompholz H
2006 High-power microwave surface flashover of a
gas–dielectric interface at 90–760 torr IEEE Trans. Plasma
Sci. 34 1782–8
[24] van der Horst R, Verreycken T, van Veldhuizen E and
Bruggeman P 2012 Time-resolved optical emission
spectroscopy of nanosecond pulsed discharges in
atmospheric-pressure N2 and N2 /H2 O mixtures J. Phys. D:
Appl. Phys. 45 345201
[25] Pancheshnyi S, Starikovskaia S and Starikovskii A 2000
Collisional deactivation of N2 states by N2 , O2 , H2 , and
H2 O molecules Chem. Phys. 262 349–57
[26] Stone E and Zipf E 1973 Excitation of atomic nitrogen by
electron impact J. Chem. Phys. 58 4278–84
[27] Itikawa Y 2006 Cross sections for electron collisions
with nitrogen molecules J. Phys. Chem. Ref. Data
35 31–53
[28] Griem H 1963 Validity of local thermal equilibrium in plasma
spectroscopy Phys. Rev. 131 1170–6
[29] Fujimoto T and McWirter R 1990 Validity criteria for local
thermodynamic equilibrium in plasma spectroscopy Phys.
Rev. A 42 6588–601
[30] Engelhardt W 1973 Establishment of partial local
thermodynamic equilibrium in transient and
inhomogeneous plasma Phys. Fluids 16 217–20
[31] Cristoforetti G, Giacomo A, Dell’Aglio M, Legnaioli S,
Tognoni E, Palleschi V and Omenetto N 2010 Local
thermodynamic equilibrium in laser-induced breakdown
spectroscopy: beyond the McWhirter criterion Spectrochim.
Acta B 65 86–95
J. Phys. D: Appl. Phys. 45 (2012) 495202
[32] Kossyi I, Kostinksy A, Matveyev A and Silakov V 1992
Kinetic scheme of the non-equilibrium discharge in
nitrogen–oxygen mixtures Plasma Sources Sci. Technol.
1 207–20
[33] Bischel W, Perry B and Crosley D 1981 Two-photon
laser-induced fluorescence in oxygen and nitrogen atoms
Chem. Phys. Lett. 82 85–8
11
View publication stats
A Fierro et al
[34] SpectraPlot.exe, accessed 8-20-2012,
http://www.p3e.ttu.edu/FTP Files/SpectraPlot.zip
[35] Thorne A, Litzen U and Johansson S 1999 Spectrophysics
Principles and Applications (Berlin: Springer)
[36] Pierluissi J, Vanderwood P and Gomez R 1977 Fast calculation
algorithm for the Voigt profile J. Quant. Spectrosc. Radiat.
Transfer 18 555–8