1 © 2016 IOP Publishing Ltd Printed in the UK

1. Introduction

Over the past several decades the continuously evolving field of low-temperature plasma physics has taken a number of leaps forward with regard to new plasma technologies. One area that has recently experienced significant growth is non-equilibrium plasmas generated by nanosecond (ns)-duration, pulsed electrical discharges. These plasmas are being explored in various geometries for a number of different applications

that range from combustion (Ju et al 2015) and high-speed flow control (Leonov et al 2016) to biology and medicine (Graves 2012). All of these applications share an interest in understanding temporally resolved spatial distributions of key atomic species in molecular plasmas. For example, gen-eration of O and H atoms in fuel-air plasmas initiates fuel-oxidation pathways and chain-branching processes at low temperatures (Ju et al 2015). Additionally, O-atom genera-tion in air plasmas controls the kinetics of ‘rapid heating’ and

Journal of Physics D: Applied Physics

Femtosecond, two-photon-absorption, laser-induced-fluorescence (fs-TALIF) imaging of atomic hydrogen and oxygen in non-equilibrium plasmas

Jacob B Schmidt1, Sukesh Roy1, Waruna D Kulatilaka1, Ivan Shkurenkov2, Igor V Adamovich2, Walter R Lempert2 and James R Gord3

1 Spectral Energies LLC, Dayton, OH 45431, USA2 The Ohio State University, Columbus, OH 43210, USA3 Air Force Research Laboratory, Wright-Patterson Air Force Base, OH 45433, USA

E-mail: roy.sukesh@gmail.com

Received 10 September 2016, revised 20 October 2016Accepted for publication 2 November 2016Published 23 November 2016

AbstractFemtosecond, two-photon-absorption laser-induced fluorescence (fs-TALIF) is employed to measure space- and time-resolved distributions of atomic hydrogen and oxygen in moderate-pressure, non-equilibrium, nanosecond-duration pulsed-discharge plasmas. Temporally and spatially resolved hydrogen and oxygen TALIF images are obtained over a range of low-temperature plasmas in mixtures of helium and argon at 100 Torr total pressure. The high-peak-intensity, low-average-energy fs pulses combined with the increased spectral bandwidth compared to traditional ns-duration laser pulses provide a large number of photon pairs that are responsible for the two-photon excitation, which results in an enhanced TALIF signal. Krypton and xenon TALIF are used for quantitative calibration of the hydrogen and oxygen concentrations, respectively, with similar excitation schemes being employed. This enables 2D collection of atomic-hydrogen and -oxygen TALIF signals with absolute number densities ranging from 2 × 1012 cm−3 to 6 × 1015 cm−3 and 1 × 1013 cm−3 to 3 × 1016 cm−3, respectively. These 2D images are the first application of TALIF imaging in moderate-pressure plasma discharges. 1D self-consistent modeling predictions show agreement with experimental results within the estimated experimental error of 25%. The present results can be used to further the development of higher fidelity kinetic models while quantifying plasma-source characteristics.

Keywords: fs-TALIF, femtosecond, plasmas, hydrogen atom, oxygen atom

(Some figures may appear in colour only in the online journal)

J B Schmidt et al

Printed in the UK

015204

JPAPBE

© 2016 IOP Publishing Ltd

50

J. Phys. D: Appl. Phys.

JPD

1361-6463

10.1088/1361-6463/50/1/015204

Paper

1

Journal of Physics D: Applied Physics

IOP

2017

1361-6463/17/015204+17$33.00

doi:10.1088/1361-6463/50/1/015204J. Phys. D: Appl. Phys. 50 (2017) 015204 (17pp)

J B Schmidt et al

2

vibrational relaxation (Rusterholtz 2013, Shkurenkov 2016), which is critical for plasma flow control. Finally, reactive-oxygen species (ROS) generated in the plasma volume have proven highly effective for inducing apoptosis in cancerous cells (Ishaq et al 2014). One of the most critical challenges in studies of low-temperature plasmas for these applications is the lack of accurate in situ characterization of the plasma source, which limits considerably our understanding of the effect of plasmas on the kinetics of reacting gas mixtures and biochemical processes. For example, many previous studies of plasmas in biology and medicine relied on an empirical approach, i.e. generation of unspecified amounts of reac-tive oxygen and/or nitrogen species by the plasma, in an attempt to detect and isolate their effect on bacteria, cells, and tissue. Although this approach may have been justified in the past, further development of plasma technologies for biomedical applications requires high-fidelity in situ charac-terization of chemical compositions in plasmas through the use of non-invasive laser-based approaches. Since quenching of excited species in high-pressure plasmas is the main lim-iting factor with regard to the accuracy of conventional ns-laser diag nostics, the use of ultra-short-pulse (picosecond (ps) and femtosecond (fs)) lasers are necessary for devising a quenching-free detection scheme. The quenching-free detec-tion of OH and NO employing various ultrafast laser-based spectroscopic approaches has already been demonstrated in reacting flows and gas cells (Reichardt et al 2000, Roy et al 2002, Wrzesinski et al 2016). However, further development of signal-detection schemes will be required to exploit the unique features afforded by the ultrafast lasers for quantitative detection of atomic-species concentrations.

While a large number of diagnostic techniques have been applied to various plasma sources, validation and verification of more comprehensive plasma-kinetic models require these diagnostics to be non-intrusive, in situ, species-selective, and highly sensitive to permit detection of low densities of reac-tive (short-lived) species. Traditional diode-laser and ns-laser-based spectroscopic techniques exhibit certain shortcomings, despite providing a non-intrusive, in situ, species-selective measurement platform. For example, absorption spectro-scopic techniques are line-of-sight methods that lack sufficient spatial resolution. Laser-induced fluorescence (LIF) is gener-ally based on single-photon absorption and offers high spatial resolution; however, because of relatively strong absorption cross-sections, large concentrations can prove to be optically thick, resulting in significant probe-beam attenuation or stim-ulated-emission effects. In addition, many key intermediates such as atomic hydrogen, oxygen, and nitrogen have large energy spacings between the initial and excited electronic states. These spacings require single-photon energies with wavelengths in the vacuum-ultraviolet (VUV) region, which are generally difficult to generate and pose significant diffi-culty during propagation through air.

To address these complications, multi-photon excitation has been developed and employed. The multi-photon approach offers two significant advantages: (1) red-shifted excitation wavelengths from the VUV region allow beam propagation with minimal absorption in the air and (2) smaller absorption

cross sections enable species measurements with high con-centrations. Two-photon-absorption laser-induced fluores-cence (TALIF) was first demonstrated for atomic hydrogen and deuterium by Bokor et al (1981) and has been dramati-cally expanded to detect many other ground-state atomic species, including oxygen (Bischel et al 1981, Aldén et al 1982, DiMauro et al 1984) and nitrogen (Bischel et al 1981). Studies conducted by the Miller group (Preppernau et al 1989, 1995, Tserepi et al 1992), the Döbele group (Czarnetzki et al 1994, Niemi et al 2001, Boogaarts et al 2002, Döbele et al 2005) and others (Amorim et al 1994, 1995, Miyazaki et al 1996) have significantly expanded atomic-hydrogen TALIF as a diagnostic method for low-temperature-plasma research.

Traditionally, ns-laser systems have been employed to probe TALIF transitions through the use of numerous exci-tation schemes. Specifically, the n = 3 or n = 4 level of atomic hydrogen is directly excited from the ground state using two or three photons, and the fluorescence signal resulting from the transition to the n = 2 state is collected. However, many other schemes such as resonant ionization spectr oscopy (Hurst et al 1988), stimulated emission (Aldén et al 1982, Goldsmith et al 1990), and the (2 + 1)-photon excitation scheme (von der Gathen et al 1991, Sasaki et al 2001), which employs two single-color photons to excite to the n = 2 level and a third photon of a longer wavelength to excite the n = 3 or n = 4 level simultaneously, have been used and compared (Czarnetzki et al 1994). The single-color direct-excitation method has the distinct advantage that it requires only photons of a single wavelength (typically in the UV) for excitation; these photons are spectrally shifted from the fluorescence signal, which simplifies the rejection of scattered light. In most cases quantitative fluorescence measurements are performed in the unsaturated regime so that the fluorescence intensity scales quadratically with the pump-laser fluence and is proportional to the atomic-hydrogen ground-state density, which enables concentration measurements.

A limitation of multi-photon excitation is the high laser flu-ence required to overcome the reduced absorption cross sec-tion. For typical ns-laser pulse durations, high laser fluence may cause significant photo-dissociation within the medium. These photolytic interferences can be significant, making quantitative measurements very difficult. To circumvent this problem, ultrafast (ps) lasers have been used in place of ns systems. High-peak-intensity, ultrafast excitation schemes are capable of producing signals that are similar to those of com-parable ns systems with significantly lower average energies (Settersten et al 2002, Frank et al 2005, Kulatilaka et al 2007, 2009). The low average energy of the ultrafast system limits photo-dissociation—an interference often found in ns excita-tion schemes (Kulatilaka et al 2008, 2009). These efforts have been extended into the fs regime, which essentially elimi-nated the presence of photolytic interference and allowed 2D imaging of atomic species afforded by the higher intensity fs-laser pulses (Kulatilaka et al 2013, 2014). 2D imaging of atomic species with conventional ns lasers is quite unimagi-nable because of the requirement of unrealistic laser energies at UV frequencies.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

3

In the current work, we demonstrate a fs-laser-based, two-photon-absorption laser-induced-fluorescence (fs-TALIF) scheme for 2D imaging of absolute concentrations of atomic hydrogen and oxygen in non-equilibrium plasmas. Proof-of-principle, fs-TALIF planar imaging of the atomic species is demonstrated in a low-temperature ns-pulse discharge in a ‘canonical’ pin-to-pin geometry. This technique can also be used to provide quantitative insight into the mechanism of atomic-species generation in ns-pulse ionization-wave dis-charges propagating over dielectric capillary tubes and in atmospheric-pressure plasma jets (Robert et al 2009) as well as their transport-to-target areas that may be covered with liquids. Thus, this approach will aid the determination of the chemical composition of ns-pulse plasmas used for bio-medical applications, removing one of the major uncertainties associated with the prevalent empirical approach.

2. Experimental

The test system selected for demonstrating fs-TALIF imaging of atomic hydrogen and oxygen is a small-volume, ns-dura-tion, high-voltage pulsed discharge that employs a modified, spherical, pin-to-pin electrode geometry. This system was chosen because it exhibits reproducible atomic-hydrogen dis-tributions that are readily captured by fs-TALIF imaging while simultaneously offering cylindrical symmetry and moderate property gradients that facilitate high-fidelity modeling of the plasma environment. The resulting images demonstrate the advantages of fs-TALIF excitation in a discharge relevant to the verification and validation of the plasma kinetic-modeling effort. Additional benefits of employing this system are low-electrical-energy requirements for initiating a small-volume plasma and ample optical access for implementing laser-based imaging. The modified, spherical pin-to-pin geometry was selected to maintain a small-volume plasma while reducing strong gradients that are normally present in a standard pin-to-pin arrangement. Each copper electrode consists of a 7.5 mm diameter solid-sphere end and a 6 mm diameter hollow stem with an overall length of 50 mm. The electrodes are affixed to a 0.75 mm diameter nickel wire that passes through a ceramic bulkhead to allow adjustability of the electrode position as well as electrical isolation from the remainder of the cell. The electrode gap was fixed at 8 mm for the measurements pre-sented here. The cell (shown in figure 1) consists of a six-way glass cross to allow optical access of the laser probe orthog-onal to the imaging system as well as the electrodes, with each arm of the cell being 50.8 mm in diameter. The optical win-dows are UV-grade fused silica of 3.175 mm thickness, which minimizes temporal broadening of the fs-duration laser pulse, and are attached to vacuum flanges at the end of each arm. The cell is mounted on a three-axis translation stage with linear resolution of ±0.01 mm.

The pulser system that generates the electrical discharge is based on a MOSFET (metal-oxide-semiconductor field-effect-transistor) switch system, with high voltage sup-plied from a high-voltage DC power supply (Glassman). In the present configuration, +5.0 kV peak voltage pulses

of 500 ns duration are supplied to the top electrode at a pulse repetition rate of 100 Hz, and the bottom electrode is grounded. The pulser, DC power supply, discharge cell, and the table are grounded to a common building ground with low-inductance high-voltage cables. The length of the cables was kept very short to minimize negative contributions from ground loops or other inductive effects. Applied-voltage and discharge-current waveforms were recorded for each test with a Northstar 1000:1 high-voltage probe that is con-nected at the anode and a Pearson inductive probe between the cathode and the ground, respectively. Typical voltage and current waveforms for discharges in 1% H2/He, 1% H2/Ar, and 1% O2/He mixtures over an 8 mm gap are shown in figures 2(a)–(c), respectively. For shot-to-shot reproduc-ibility, a non-inductive power resistor with a value of 1 kΩ was inserted between the pulser output and the anode to limit the current and prevent DC-glow-to-arc transition of the discharge. The total pressure was set at 100 Torr for all measurements performed. The estimated flow velocity of the gas mixture through the cell and the discharge repetition rate were set to ensure single discharge during the gas residence time between the electrodes.

Different gas mixtures were used to produce atomic species for each discharge condition. For example, mole fractions of H2 were varied in either helium- or argon-based discharges to produce atomic hydrogen. Similar mixtures on O2 were added to helium-based discharges to produce atomic oxygen. All of the gases employed were ultra-high purity (UHP 99.999%). The total gas flow through the cell was controlled with MKS mass flow controllers. The buffer-gas flow rate was fixed at 8 SLPM. The small amount of either H2 or O2 gas was mixed with the buffer gas 1 m before entering the cell to ensure homogeneity. By measuring the individual flow rates, a mole fraction of 0.01% of H2 or O2 relative to the buffer-gas flow rate could be accurately supplied. The total pressure in the test cell was regulated by a single-stage pump with a needle valve and a bypass. Unfortunately, argon emission within the plasma discharge from three excited states decaying back to the 3s23p5( /P2

1 20 )4s state near 845 nm spectrally overlaps with

the oxygen TALIF signal. For this reason, argon was omitted as a background gas, and only helium was used for atomic-oxygen fs-TALIF imaging. This was not an issue for hydrogen TALIF, and both argon and helium were used as background gasses.

The laser system employed for atomic-hydrogen excita-tion near 205 nm consists of an amplified Ti:sapphire laser system and a fourth-harmonic generator (FHG) operated at a 10 kHz repetition rate with a pulse duration of ~100 fs; the system is described elsewhere in greater detail (Kulatilaka et al 2014). A regenerative and a single-pass amplifier were used to produce ~1 mJ/pulse at the fundamental wavelength, which was then delivered to a home-built FHG. This FHG is equipped with very thin beta-barium-borate (BBO) crys-tals for frequency conversion and mixing and produces ~10 µJ/pulse at 10 kHz near the required 205.1 nm wavelength. For excitation, a second Ti:Sapphire laser system with an optical parameteric amplifier (OPA) was used. This system

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

4

is capable of producing ~15 µJ/pulse at 1 kHz repetition rate near 225 nm. These excitation schemes are shown in figure 3. A 500 mm spherical lens and a −450 mm cylindrical lens were used to form a 2 mm tall, 0.09 mm thick laser sheet within the plasma volume.

An Andor NewtonEM charge-coupled device (CCD) was externally intensified with a LaVision (IRO) gated intensifier for image collection. The external intensifier had an exposure duration of 100 ns, operated at a repetition rate of 100 Hz, and acted as a shutter for the CCD camera that had an exposure

Figure 1. Six-way glass cross used for low-pressure TALIF experiments. Broadband emission is shown from ns-pulsed discharge in 1% hydrogen in helium at 100 Torr.

Figure 2. Typical voltage and current waveforms for discharges in (a) 1% H2/He, (b) 1% H2/Ar, and (c) 1% O2/He mixtures across 8 mm gap at 100 Torr.

Figure 3. Two-photon excitation schemes for atomic-hydrogen and -oxygen with their noble calibration gases of krypton and xenon, respectively.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

5

time of 1 s. The repetition rate was selected to match that of the pulsed discharge. Each data point presented here represents an average of more than 100 laser shots. A band-pass filter with an in-band transmission of more than 95% was used to block pho-tons from laser scatter and plasma emission. Light was focused on the intensifier photocathode with an f/1.8 85 mm lens with 35 mm of lens-tube extensions. The spatial resolution of the image with this imaging system was 16 µm × 16 µm per pixel. The CCD was electronically cooled to −80 °C to reduce on-chip noise. The signal-to-noise ratio ranged from a ‘best-case’ level of 80:1 to typical values of 20:1. For performing initial plasma-emission studies, an intensified CCD (ICCD) camera was used with the same 85 mm lens. The camera gate was set at 200 ns, and signals were averaged over 100 laser shots.

Quantification of the atomic-species number density is nec-essary for the applications where TALIF is used and requires accurate calibration of the detection system. A number of calibration techniques have been used, ranging from known-concentration reference sources (Clyne et al 1979) to single-photon-absorption methods (Amorim et al 1994) to manual evaluation from first principles, based on known parameters involving the exciting radiation, interaction volumes, and cross section. These methods have proved to be difficult because of their rigorous nature, small absorption cross sections, or a lack of comparable absorbers. Two more common calibration tech-niques are NO2 titration and noble-gas calibration. In titration a known quantity of NO2 is introduced into the optical-detec-tion region where it quenches the atomic hydrogen through a fast single-step reaction (Meier et al 1990). The addition of NO2 can be accomplished at various pressures, temperatures, and volumes to determine the linear relationship with the total atomic hydrogen present before titration. While very accurate, this method can be difficult to implement, especially within a short-timescale discharge without negatively affecting the dis-charge parameters. A noble-gas calibration is more straight-forward to implement and relies on excitation and emission characteristics that are similar to those of the atomic spe-cies being measured (Niemi et al 2001, Döbele et al 2005). In addition, this method can be used for calibration of both atomic-hydrogen and -oxygen species. For calibration of atomic-hydrogen populations, krypton was selected since the 4p6 1S0 → 5p′ [3/2]2 transition of krypton is very close to the 1s 2S1/2 → 3d 2D3/2,5/2 transition of hydrogen; for calibration of atomic-oxygen concentrations, xenon was selected since its 5p6 1S0 → 6p′ [3/2]2 transition of Xe lies very close to the 2p4

3P2,1,0 → 3p 3P1,2,0 transition of oxygen. These transitions are shown in figure 3. This calibration was performed in a non-flowing condition, where the cell was evacuated to 0.01 Torr total pressure and then sealed. A sequence of TALIF images of the calibration gas was acquired, ranging from 0.3 to 25 Torr. With the entire calibration requiring <15 min to perform and a total leak-up rate of <5 Torr h−1 at 1 Torr, the cell was assumed to be filled with pure calibration gas for the duration of the calibration. From this calibration a relationship can be obtained between the unknown number density of the atomic species (nX) and the known number density of the calibration gas (nCal). This relationship, which is shown below (equa-tion (1)), is derived from the collision-free case and includes

the Boltzmann-distribution correction factor CB, the detector sensitivities, the attenuation factor if used for the calibration measurements η, the incident laser fluence / υΦ = TE hAi i i i (composed of optical attenuation factor, Ti, measured laser pulse energy Ei, area of incident beam Ai, and laser frequency υi), the fluorescence quantum yields /= +a A A Qi23 23 2 (where A23 and Q are the spontaneous emission and quenching rates, respectively), the two-photon absorption cross sections ( )σi

2 , and the observed fluorescence levels S. This equation assumes that collection parameters such as solid angle, gain, and expo-sure durations are held constant during the calibration and fluorescence-collection events. It should be noted that the reported limits are only as accurate as the calibration per-formed and its corrections. In this effort the accuracy of these measurements was ~25% and was primarily determined by laser performance and quenching corrections.

ηη

σ

σχ=

Φ

Φ=n C

a

an C

S

Sn .X

X X X X

XB

Cal Cal2

2Cal Cal

2

2 Cal BCal

Cal

( )

( ) (1)

Collisional quenching is an important depopulation mech-anism of the excited state. Calibrated signals were corrected based on known collisional cross sections between the atomic species and other major constituents, specifically hydrogen or oxygen and helium or argon (Bittner et al 1988, Niemi et al 2001, 2005). It was assumed that the mixtures were homoge-neous at 300 K. It is important to note that self-quenching of the atomic species was neglected because of their relatively low number densities. These assumptions will contribute to overall error estimation. Multi-photon ionization occurs when a third photon is absorbed by the excited oxygen atom, which produces an oxygen ion, and must be accounted for, espe-cially with high-peak laser-fluence-based measurements. This process affects the overall relationship between the oxygen ground-state population and the expected TALIF signal and can be assumed to be negligible if a squared dependence of the laser intensity on the TALIF signal is observed. To ensure that photoionization did not affect the measurement, a second calibration was conducted by incrementally attenuating the incident laser fluence and recording the TALIF signal. This calibration assures that photoionization is insignificant as long as the incident laser fluence is kept below levels where the squared-laser-intensity relationship with the TALIF signal is maintained. In addition to quenching and photodissociation, fluorescence images were corrected for background-light levels, camera noise, and laser scatter, if present, as well as in-band plasma emission.

A complication exists for quantification of atomic-hydrogen fluorescence using the scheme shown in equation. In addition to the 1s 2S1/2 → 3d 2D3/2,5/2 transition in atomic hydrogen, the 1s 2S1/2 → 3s 2S1/2 transition is also excited with the same two 205 nm photons since these two transitions are only ~30 cm−1 apart. Each transition is allowed, and both 3d 2D3/2,5/2 and 3s 2S1/2 excited states radiate to the 2p 2P1/2,3/2 state, emit-ting a photon at ~656.3 nm. The very similar characteristics make it very hard to discriminate between the two transitions. However, a significant uncertainty can result from inaccurately implementing corrections for these two transitions.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

6

Because of the selection rules (∆ = ±l 0, 2), the l = 0 (s-state) and l = 2 (d-state) are the only states that are directly populated during excitation—not the l = 1 (p-state). Additionally, the s-state and d-state have different absorption cross sections. The ratio of the d-state cross-section to the s-state cross section is 7.56 (Tung et al 1986). Since, even for ns-duration laser pulses, the incident laser linewidth is larger than the energy separation between the s- and d-states, both states are directly excited with the incident laser radiation. Because of the differences in absorption cross section, ~88% of the n = 3 population resides in the d-state after excitation. This can be significant since the d-state has a natural lifetime of 15.6 ns, whereas the natural lifetime of the s-state is 159 ns.

Because of the simultaneous s- and d-state pumping, two forms of non-radiative decay must be considered, i.e. col-lisional quenching and angular momentum mixing between all three n = 3 states due to collisions after laser excitation. The model developed by Preppernau et al (1995) was employed to account for these corrections. This model predicts the rate of change for the population of each individual state, including both collisional quenching and mixing. These relationships are shown in equation. These time-dependent populations can then be used to determine the effective radiative rate for the two channels of interest that produce the observed fluores-cence from the s- and d-state, as shown in equation, where Ns, Np, and Nd are the population of the s, p, and d angular momentum states, respectively, NQ is the number density of the quenching species, τs, τp, and τd are the natural lifetimes of the s, p, and d angular momentum states, respectively, τeff is the effective natural lifetime from the s- and d-states, and kmix and kQ are rate constants associated with collisional popula-tion mixing of the angular momentum l-states and non-radia-tive decay of the entire n = 3 state, respectively.

( ) ( )

( ) ( )

( ) ( )

⎛⎝⎜

⎞⎠⎟

⎛

⎝⎜

⎞

⎠⎟

⎛⎝⎜

⎞⎠⎟

τ

τ

τ

= − + + + +

= − + + + +

= − + + + +

N

tN N k k N k N N

N

tN N k k N k N N

N

tN N k k N k N N

d

d

18

d

d

16 3

d

d

14 5

ss

sp d

pp

ps d

dd

ds p

Q mix Q Q mix

Q mix Q Q mix

Q mix Q Q mix

(2)

( ) ( )τ

τ τ= +

N N0 0.s

s

d

deff (3)

The lower O 2p4 3PJ and upper O 3p 3PJ′ states in atomic oxygen are divided into three levels with orbital angular momentum quantum numbers J = 2, 1, 0 and J′ = 1, 2, 0. While the upper states are very closely spaced and cannot be distinguished during laser excitation or fluorescence, the spacing between the lower states is much larger. Because the population of each of these levels follows a Boltzmann distri-bution, which is dependent on local gas temperature, conven-tional ns-duration laser-based diagnostics require knowledge of either the relative population distribution between the three sublevels or the local temperature. Usually, the popu-lation distribution for each of the sublevels is determined

by scanning over each of the lower levels individually at 225.685 nm, 225.988 nm, and 226.164 nm. The total atomic-oxygen ground-state density is then determined by summing all of these J-level number densities. The large bandwidth of the fs-laser pulse (>1.2 nm FWHM) allows simultaneous excitation from all three sublevels. Each of the individual J-sublevel excitations must be normalized against the incident laser intensity at that wavelength, but neither scanning of the laser wavelength nor knowledge of local gas temperature is required if an equilibrium distribution can be assumed between the sublevels. Once normalized, the collected fs-TALIF signal does not require additional correction since all J sublevels are simultaneously excited and fluoresce during the single col-lection sequence. Relative two-photon absorption cross sec-tions for each of these specific transitions (Saxon and Eichler 1986) are used when making a relative comparison between different excitation schemes during the TALIF measurements.

In the present work quenching rates were measured over a wide range of pressures to obtain the natural or radiative lifetime of the excited-state atomic species and collisional-quenching coefficients for specific quenchers. The results of these measurements are summarized in the plots shown in figure 4 for atomic hydrogen in helium (top left), atomic hydrogen in argon (top right), and krypton in krypton (bottom). To minimize the effects of temperature on the measured quenching rates, the discharge was operated at low-discharge-energy conditions, and long integration times were used to improve the signal-to-noise ratios.

Traditionally, these plots and the constants obtained from them have been collected over a significantly lower pressure range (Adams and Miller 1998, Niemi et al 2005, van Gessel et al 2013) because of the discharge instabilities at higher pressures constraining production of atomic species and the inability to accurately resolve the quenching lifetimes limited by the laser-pulse duration. These constraints are shown as the ns-TALIF limit in each of the plots in figure 4 for an 8 ns duration laser pulse. The fs-TALIF system, along with the fast photomultiplier tube (PMT), allows accurate collection down to a time resolution of 500 ps and can greatly extend the range of pressures where quenching rates can be accurately obtained. Each plot in figure 4 shows data collected both above and below these limits. Collisional quenching of atomic hydrogen in argon is significantly higher, resulting in faster fluorescence decay. Data were collected at the 100 Torr oper-ating conditions to ensure the accuracy of the measurements, but the upper pressure limit for these measurements was near 120 Torr. At higher pressures the finite-impulse-response fea-ture of the PMT is responsible for the same issues that limit the lifetime measurements with ns-TALIF.

The rates obtained from each of these measurements are summarized in table 1 and are compared with data found in the literature. The effective natural lifetime listed for atomic hydrogen, 16.7 ns, is longer than that for the s-state but signifi-cantly shorter than that for the d-state. This suggests that the effective natural lifetime is dominated by the d-state, which is expected because of the significantly higher population of this state. An observed trend in the quenching-coefficient data is that higher efficiency quenchers exhibit larger variation in

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

7

the reported data. This can be attributed to the difficulty in measuring short decay times, which leads to large uncertain-ties. In addition, these measurements are more susceptible to small impurities and interference caused by other quenching species. In general, the quenching-rate constants determined from the present experiments are not in agreement with those in the literature, and it is assumed that the present experimental data are more accurate because of the wider pressure range over which they were obtained. This further demonstrates the strength of the fs-based measurements.

In a manner similar to the atomic-hydrogen measure-ments, quenching rates were collected over a wide range of pressures to obtain the natural lifetime of the excited-state atomic species and the collisional-quenching coefficients for specific quenchers. These plots are shown in figure 5 for atomic oxygen in helium (left) and xenon in xenon (right). To minimize the effect of temperature on the quenching rates obtained, the discharge was operated at a low-pulse-energy condition, and long integration times were used to improve the signal-to-noise ratios. The rates obtained from each of these measurements are summarized in table 2 and are compared with data found in the literature. As with the data in figure 4, the fs-based measurements are close to those in the literature but differ by a sufficiently significant amount that the fs-based measurements are assumed to be more accurate.

3. fs-TALIF results and discussion

For initial plasma characterization a Princeton Instruments ICCD was used to image the unfiltered spatial distribution of the plasma emission, which not only revealed the spatial structure of the plasma but also indicated the lifetime of the plasma emission. It was necessary to determine this timescale since this emission interferes with the fluorescence signal, necessitating correction for number-density calculations. An example of these images in false-color scale is presented in figure 6(a) for 1% hydrogen in argon at 100 Torr. The lifetime of the atomic-hydrogen emission was varied by changing the mixture ratio of hydrogen and the buffer gas and was observed to last nearly 30 µs, with most of the emission being generated in the near-cathode region.

3.1. Spatial-reconstruction results

The Andor camera, when equipped with an external intensi-fier, proved to have higher collection efficiency and signal-to-noise ratio than the Princeton Instruments ICCD and was used for the fluorescence measurements. Each fluorescence image was corrected for plasma emission, background levels, and camera noise as well as incident laser power and quenching. Initially, a line was imaged to optimize the image-correction

Figure 4. Stern–Volmer plots generated to determine quenching-rate constant and natural lifetime for atomic hydrogen in helium (top left), atomic hydrogen in argon (top right), and krypton in krypton (bottom).

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

8

technique because its signal-to-noise ratio was higher than that of the subsequently used sheet imaging. Figures 6(b) and (c) show the typical signal collected from single-shot, line fs-TALIF images from a 1% hydrogen-in-helium mixture before and after correction schemes were applied, respectively. Figure 6(d) shows a corrected, single-shot, planar fs-TALIF image collected near the cathode. Because of the reduced

signal-to-noise ratio when the sheet is used for imaging, 100 shots were averaged to generate a single image for each data point. To obtain the atomic-species distributions across the entire electrode gap, the 2 mm tall sheet was scanned vertically between the electrodes along the centerline of the plasma. To accomplish this, the entire cell was vertically translated in increments of 0.5 mm, resulting in a 75% overlap between subsequent image pairs. These 100-shot-averaged images were then assembled to generate a composite 2D image of the atomic-species distribution.

Results obtained in 1%-hydrogen-in-helium and 1%- hydrogen-in-argon mixtures at 100 Torr are shown in figures 7 and 8, respectively. These images were collected 25 µs after the onset of the applied voltage. This delay was chosen to reduce background plasma emission and, thereby, enhance signal-to-noise ratio. Comparison of the images in figures 7 and 8 shows a marked increase in the spatial extent of the atomic-hydrogen distribution in helium, compared to that measured in argon for the same hydrogen mole fraction and total pressure. A cross-sectional plot depicting the atomic-hydrogen number-density distribution along the centerline of the plasma is shown for both cases. While the absolute value of the atomic-hydrogen number density differs by ~10% for the two cases, which is within the measurement uncertainty, the qualitative behavior is quite similar. The peak concentration is located within 1.5 mm of the cathode in each composite image. The con-centration decreases by about a factor of two as it reaches a local minimum ~3–4 mm from the anode before increasing to a local maximum again. Each cross-sectional profile exhibits a maximum number density of ~5.0 × 1015 cm−3 about 1 mm from the cathode and a minimum of ~1.5 × 1015 cm−3 about 3.5 mm from the anode.

The difference in the 2D distributions of the atomic-hydrogen populations observed in the helium- and argon-based discharges is caused by (1) a shallow dependence of the ionization coefficient on the reduced electric field, E/N, for He compared to Ar (Raizer 1991), (2) rapid diffusion of metastable He* atoms compared to metastable Ar* atoms, and (3) rapid ambipolar diffusion in He compared to Ar. These

Table 1. Two-photon absorption cross sections, 2( )σ , natural lifetimes, τ, quenching coefficients, ki, and fluorescence quantum yields, a21, for atomic-hydrogen and -krypton excited states at 300 K.

H (3d 2D3/2.5/2) Kr (5p′ [3/2]2)

2( )σ (cm4) Kr/H = 0.60 (g)Kr/H = 0.62 (f)

τ (ns) τeff = 16.7 ± 0.7a 33.6 ± 1.1a

15.7 ± 1.5 (d) (a) 26.9 (e)

20.9 ± 0.8 (b) 35.4 ± 2.7 (g)17.6 (f) 34.1 (f) (h)10.0 ± 0.5 (g)

ki (10−10 cm3 s−1) H2 19.9 ± 3 (b) 8.44 (f)25 (c)17.8 ± 0.14 (d)

He kQ = 0.317 ± 0.002a 0.78 (f)

kmix = 0.07 ± 0.01a

0.099 ± 0.05 (b)0.53 ± 0.005 (d)

Ar kQ = 4.15 ± 0.03a 1.29 (f)

kmix = 0.9 ± 0.1a

4.6 ± 0.5 (b)3.8 ± 0.3 (d)

Kr 11.0 ± 1 (b) 1.31 ± 0.065a

1.46 (f) (h)a21 He 0.2583–0.037a

Ar 0.0522a

Kr 0.131a

aThis study, (a) Wiese et al (1969), (b) Bittner et al (1988), (c) Bletzinger and Ganguly (1995), (d) Preppernau et al (1995), (e) Mazouffre et al (2001), (f) Niemi et al (2001), (g) Boogaarts et al (2002), (h) Es-Sebbar et al (2009).

Figure 5. Stern–Volmer plots generated to determine quenching-rate constant and natural lifetime for atomic oxygen in helium (left) and xenon in xenon (right).

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

9

effects result in a larger plasma-filament diameter and associ-ated distribution of atomic hydrogen.

Atomic oxygen was generated in the test cell filled with different mixtures of oxygen (mole fractions ranging from 0.5% to 5%) in helium at 100 Torr. Argon could not be used as a buffer gas because of the strong emission near 845 nm over-lapping with the atomic-oxygen fluorescence that was ema-nating from the three excited states back to the 3s23p5(2

/P1 2o )4s

state, as shown in figure 3. Multiple planar images were col-lected and reassembled to generate the 2D, atomic-oxygen number-density distribution between the electrodes. The discharge shown in figure 9 was sustained in a 1%-O2/He

mixture at 100 Torr, and the fluorescence was collected 25 µs after the discharge. The reconstruction shows that atomic-oxygen fluorescence is significantly stronger near the anode and spreads in the radial direction near the cathode. This overall distribution is observed for all the O2 mole fractions tested. The main parameter that changed with O2 mole frac-tion was the peak number density. For the 1%-O2/He mixture shown in figure 9, the peak atomic-oxygen number density of 3.0 × 1016 cm−3 occurred ~1 mm from the anode and the minimum of 4.1 × 1015 cm−3 occurred near the cathode.

When the 1%-oxygen-in-helium data shown in figure 9 are compared with the single-pulse 1%-H2/He data shown in

Table 2. Two-photon-absorption cross-sections, 2( )σ , natural lifetimes, τ, quenching coefficients, ki, and fluorescence quantum yields, a21, for atomic-oxygen and -xenon excited states at 300 K.

O (3p 3P1,2,0) Xe (6p′[3/2]2)

( )σ 2 (cm4) 2.66 ± 0.80 × 10−34 (a) 4.94 ± 0.98 × 10−34 (b)τ (ns) 35.4 ± 1.4a 40.0 ± 1.6a

34.7 ± 1.7 (b) 40.8 ± 2.0 (b)35.1 ± 3.0 (e) 40 ± 6 (c)36.2 ± 0.7 (f) 37 ± 2 (g)

30.7 ± 2.2 (h)

ki (10−10 cm3 s−1) O2 8.6 ± 0.2 (a)9.4 ± 0.5 (b)9.3 ± 0.4 (e)6.3 ± 0.1 (f)

He 0.016 ± 0.002a 5.7 ± 0.3 (h)0.017 ± 0.002 (b) 5.7 ± 0.6 (c)0.07 ± 0.02 (e) 9.3 ± 2.0 (i)0.15 ± 0.05 (f)

N2 5.9 ± 0.2 (b) 5.1 ± 0.45a

4.3 (f)Xe 3.7 ± 0.14a

3.6 ± 0.4 (b)4.2 ± 0.5 (c)4.3 ± 0.1 (d)

a21 He 0.22–0.054a

Xe 0.27a

aThis study, (a) Bamford et al (1986), (b) Niemi et al (2005), (c) Alekseev et al (1996), (d) Bruce et al (1989), (e) Niemi et al (2001), (f) Bittner et al (1988), (g) Inoue et al (1984), (h) van Gessel et al (2013), (i) Zikratov and Setser (1996).

Figure 6. (a) Example image of broadband emission of non-equilibrium discharge in 1%-hydrogen/99%-argon mixture at 100 Torr collected 10 µs after initial discharge. (b) and (c) Typical single-shot, fs-TALIF line image (b) raw and (c) after corrections applied. (d) Typical planar, single-shot fs-TALIF image after corrections applied. (a) and (b) are linearly scaled between 100 and 30 000 counts, (c) is linearly scaled between 0 and 20 000 counts, and (d) is linearly scaled between 0 and 1000 counts using the colorbar shown on the right.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

10

figure 7, it is observed that the location of the peak atomic-species production is inverted in the two data sets. However, it is also observed that the relative widths of these discharge conditions are approximately the same. At a location 1 mm from the anode, where the peak atomic-oxygen number den-sity is observed, the fluorescence FWHM of the discharge in O2/He is ~3.5 mm. At the same location in the H2/He dis-charge, the FWHM is ~1.9 mm. At a location 1 mm from the cathode, where the peak atomic-hydrogen number density is observed, the fluorescence FWHM of the discharge in H2/

He is ~8.2 mm. At the same location in the discharge in O2/He, the fluorescence FWHM is ~7.8 mm. The reasons for the observed radial extent of the fluorescence are discussed above and are consistent in both H2/He- and O2/He-mixture discharges because of the large mole fraction of helium (99%) in each mixture. The inverted atomic-species distributions are somewhat surprising. One of the kinetic processes that may result in the ‘inverted’ O atom number density distribution detected in the experiment (with a well-pronounced maximum near the anode) is dissociation of metastable excited electronic

Figure 7. 2D composite image of atomic hydrogen in a 1%-hydrogen-in-helium mixture at 100 Torr total pressure. Image is generated from fluorescence collected 25 µs after voltage onset. Broadband image obtained at the same condition is shown on the right.

Figure 8. 2D composite image of atomic hydrogen from 1%-hydrogen-in-argon mixture at 100 Torr total pressure. Signal is collected 25 µs after voltage onset.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

11

states of O2, a1Δ (0.96 eV) and b1Σ (1.6 eV), both by electron impact and during quenching of excited helium atoms. These low-energy excited states are generated more efficiently near the anode, where the reduced electric field is not as high as near the cathode, which may enhance net O2 dissociation yield near the anode. This effect is not present in hydrogen, since H2 does not have bound electronic states below disso-ciation limit. However, this qualitative conjecture needs to be verified using systematic kinetic sensitivity studies, which would determine the effect of different kinetic processes on O and H atom spatial distributions in the discharge. It is impor-tant to note that the differences in fluorescence FWHM are within several hundred microns. The spatial resolution, along with the high rate of data collection, allows this measurement to be made much more efficiently and accurately than with conventional ns-based laser-diagnostic systems since large volumes of data can be collected before significant heating of the electrodes can occur.

One of the first issues to be addressed with regard to the present results is the capability of the ns-duration, pulsed electrical discharge to produce the observed concentration of hydrogen atoms, based on energy conservation. Figures 2(a) and (b) show typical voltage and current traces in 1%-H2/He and 1%-H2/Ar discharges, respectively, across an 8 mm gap at 100 Torr in a modified pin-to-pin geometry. The coupled pulse energy is ~270 µJ and 210 µJ in the helium- and argon-based discharges, respectively. The per-pulse coupled energy must be compared with the energy required to break sufficient hydrogen–hydrogen bonds to produce the atomic-hydrogen number densities shown in figures 7 and 8. Based on spatially integrating the atomic-hydrogen number density in each of the images, the estimated energy required to produce this amount

of atomic hydrogen is ~187 ± 32 µJ and ~149 ± 25 µJ for the helium- and argon-based discharges, respectively, which is ~69 ± 12% and ~71 ± 12%, of the total coupled energy in the discharge, respectively. These unique measurements have determined the net discharge-coupled fractional energy going to H-atom production (both by electron-impact dissociation and dissociative quenching of Ar* and He* by H2), which could not have been accomplished without the 2D imaging scheme afforded by the fs laser.

As with atomic hydrogen, the estimated energy required to produce the observed amount of atomic oxygen is compared with coupled-discharge pulse energy. Figure 2(c) shows cur-rent and voltage waveforms for the conditions in figure 9. At these conditions energy coupled per pulse is ~370 µJ, which is compared to the energy required to produce the measured amount of atomic oxygen, ~210 ± 33 µJ. Thus, the fraction of energy stored in the oxygen atoms is ~57 ± 9% of the total coupled energy. Again, these unique measurements have determined the net discharge-coupled energy fraction going to O-atom production (both by electron-impact dissociation and dissociative quenching of He* by O2).

3.2. Kinetic modeling

The experimental composite images are used to verify and validate the predictions from a quasi-1D model that couples the Poisson, Boltzmann, and non-equilibrium plasma-chem-istry kinetic equations to determine the spatial distribution of atomic hydrogen at a given temporal delay after a pulsed discharge. The kinetic model includes time-dependent spe-cies-concentration equations for electrons, ions (He+, +He2 ,

Figure 9. 2D reconstruction of atomic-oxygen number density in a 1%-O2/He mixture at 100 Torr total pressure. Fluorescence signal is generated in single ns-duration discharge pulse and collected 25 µs after discharge.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

12

H+, +H2 , and +H3 ; O−, −O2 , −O3 , O+, +O2 , +O4 —for oxygen), and neutral species (He, He*, ∗He2, H2(X1Σ,v = 0–3), H2(d3Πu),

H2(a3∑+g ), H2(b3Σg), H2(c3Πu), H2(B1∑+u ), H2(C1Πu), ∗H2,

and H(n = 1,3); O2, O2(1Δg), O2(b1Σg), ∗∗O2 , O(3P), O(1D),

O(1S), O3—for oxygen). Here ∗H2 is a sum of Rydberg-excited electronic states; O2(1Δg), O2(b1Σg) are two lower electron-excited singlet states; and ∗∗O2 is a sum of Herzberg states (A3Σu + C3Δu + c1Σ−u ) of the O2 molecule; O(3P), O(1D), O(1S) are the ground and electron-excited states of atomic oxygen. These equations are coupled to the Poisson equation for the electric field, the two-term expansion Boltzmann equation for the electron-energy-distribution func-tion (EEDF), the equation for the electron temperature, and the equation for the gas temperature. All excited states are produced by electron-impact excitation of He and H2 or O2 (depending on the gas mixture). The EEDF and the rates of electron-impact reactions are predicted by the Boltzmann-equation solver. Electron-impact cross sections are taken from (Engelhardt et al 1963, Braginskiy et al 2005), and the rate coefficients for neutral-species reactions and ion-neutral reac-tions are taken from (Kiefer et al 1966,] Hassouni et al 1999, Stone et al 2002, Braginskiy et al 2005). The system of equa-tions is solved self-consistently, and the detailed description of the model can be found in (Shkurenkov et al 2014).

For comparison purposes the model-predicted and exper-imentally observed atomic-hydrogen, number-density axial profiles at the centerline for the case of 1%-hydrogen-in-helium and 1%-oxygen-in-helium at 100 Torr are displayed in figures 10 and 11, respectively. The qualitative trends of the hydrogen-in-helium data are similar, showing a bimodal dis-tribution within the electrode gap where peak concentrations are located near the electrodes. Both model predictions and experimental data show higher atomic concentrations near the electrodes, specifically near the cathode, which are due to the higher electric field in these regions. This results in more available energy for molecular dissociation and higher atomic populations in these areas. Quantitatively, the agreement between the (H) fs-TALIF experimental results and the mod-eling predictions is generally quite good and generally only

differs at the local maximum locations occurring near both the anode and the cathode and the local minimum occurring in the middle of the discharge. The modeling predictions for the oxygen-in-helium data are also in fairly good agreement with the data. Although the model does not reproduce a trend of (O) reduction at the anode and cathode, the measured and predicted (O) number densities agree within a factor of two. One source for the difference between the model predictions and the experimental results is evident in the current resulting from each of these processes, as shown in figure 12. The blue and red curves in the figure are the voltage and current recorded from the experimental results, respectively. The 1D plasma-kinetic model utilizes only the exper imentally meas-ured voltage as an input parameter and predicts the current, shown in green in figure 12. While the current at breakdown predicted by the model is slightly higher than the exper-imentally measured current, the model over-predicts the rate of cur rent decay during the discharge pulse—the predicted coupled pulse energy of 193 µJ is 29% lower than the meas-ured value of 271 µJ for the hydrogen-in-helium mixture and 21% lower than the measured value of 329 µJ for the oxygen-in-helium mixture. The difference between the experimental results and predictions may indicate the presence of an addi-tional ioniz ation mechanism that is not taken into account by the model. Remaining discrepancies can be attributed to experimental error in the calibration process or the collisional-quenching rates used.

The largest contribution to the experimental error may arise from the temperature-sensitive correction factors since local temperatures were not measured. Additionally, col-lisional-quenching rates were not determined experimentally but taken from the literature, assuming a constant and uni-form 300 K temperature distribution within the discharge. Figure 13 shows the model-predicted temperature along the centerline for the H2–He and O2–He discharge conditions. Most of the discharge volume is predicted to be just above the assumed 300 K. Under the assumption that quenching rates are scaled according to a T0.5 relationship, the predicted small temperature increase results in an under-prediction of the

Figure 10. Axial profiles of atomic-hydrogen number density collected along centerline of non-equilibrium plasma in 1%-hydrogen-in-helium mixture at 100 Torr.

Figure 11. Axial profiles of atomic-oxygen number density collected along centerline of non-equilibrium plasma in 1%-oxygen-in-helium mixture at 100 Torr.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

13

population by ~5%. However, the temperature profiles in the near-cathode locations show temperatures exceeding 450 K. Assuming the same temperature-based scaling, the previously shown results under-predict the atomic-species concentrations by ~22%. This is consistent with the previously given acc-uracy estimations but relies on the model-predicted temper-atures. Should actual heating from the discharge exceed the predictions because of additional current being passed, as discussed previously, these measurements will be negatively affected and will result in further under-prediction. Future work will include in situ measurements of these quenching rates to enhance the accuracy.

3.3. Temporal-decay results

For each set of discharge conditions investigated, the signal from a 400 µm × 400 µm region in the middle of the plasma volume, shown in figure 14(a), was spatially integrated to yield total local signal intensity. This integration was con-ducted to increase the signal-to-noise ratio while isolating local fluctuations in the spatial distribution of atoms from

their temporal behavior. When the time delay after the dis-charge pulse is varied, an atomic-hydrogen number-density temporal decay is obtained for this middle-of-the-discharge location. For most of the atomic-imaging-data conditions, emission data could not be accurately collected for time delays shorter than 10 µs from the onset of the applied voltage pulse since the strong plasma emission caused sig-nificant image-correction issues. Figure 14(b) displays a temporal comparison of the results in 1%-H2/He to 1%-H2/Ar mixtures, both at 100 Torr total pressure, where the helium buffer-gas condition shows higher initial number densities of atomic hydrogen. The 1%-H2/He-mixture number density decays with a /e1 decay time constant of 195 ± 20 µs, while the 1%-H2/Ar-mixture number density decays with a /e1 decay time constant of 91 ± 8 µs, showing that the atomic-hydrogen decay rate in the H2/He mixture is slower than that in the H2/Ar mixture.

Figures 15(a) and (b) show data sets collected at different mole fractions of hydrogen from 0.5% to 5% in argon and helium, respectively. In the signal-integration region, atomic-hydrogen number densities in both types of mixtures peak at a H2-admixture percentage of 1%. When the H2-admixture percentage is increased to 2% or 5% or decreased to 0.5%, atomic-hydrogen number densities decrease both in argon- and helium-based mixtures. The atomic-hydrogen decay rates in these data sets are quite similar. The /e1 decay time constant is 208 ± 22 µs, 195 ± 20 µs, 257 ± 15 µs, and 283 ± 26 µs for the 0.5%, 1%, 2%, and 5% H2/He mixtures, respectively, and 136 ± 14 µs, 91 ± 8 µs, 128 ± 15 µs, and 152 ± 17 µs for the 0.5%, 1%, 2%, and 5% H2/Ar mixtures, respectively. Examination of the discharge current and voltage recorded for each of these conditions shows that an increased amount of H2 in the mixture causes a reduction in the current and, in turn, the efficiency of atomic-hydrogen generation in the plasma when the H2 mole fraction is increased to more than 1%. A peak current of 2.93 A is recorded for the 1%-H2-mole-fraction condition in helium and decreases by 8% to 2.70 A in the 2%-H2/He condition. In the 0.5%-H2/He the current remains near the peak value, but a reduction in atomic hydrogen is observed.

Figure 12. Typical voltage and current (shown in blue and red, respectively) in (a) 1%-hydrogen-in-helium mixture and (b) 1%-oxygen-in-helium mixture at conditions of figures 10 and 11, respectively. Current predicted by model is shown in green and under-predicts per-pulse energy by 29% and 21%, respectively.

Figure 13. 1D model prediction of filament temperature on centerline for 1%-hydrogen- in-helium and 1%-oxygen-in-helium discharge conditions at 100 Torr pressure.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

14

The temporal decay of the atomic hydrogen is almost certainly dominated by diffusion. At standard temperature (273 K) and pressure (1 bar), the coefficient of binary dif-fusion, Dab for H2 diffusing into helium is 1.64 cm2 s−1 (Mostinsky 2011). From figure 13 it can be observed that the predicted temperature at the midway point between the elec-trodes is ~350 K. Utilizing standard diffusion rates, an esti-mate for the characteristic time scale for diffusion is found to be ~400 µs, assuming a filament diameter of ~0.2 cm (see figure 7, which is in reasonable agreement with the observed /e1 decay time constant in figure 15(b)).

Another decay process considered is the recombination of atomic hydrogen, given in equation, where M is a third body (helium or argon). The characteristic timescale for this reaction is given in equation. The room-temperature rate constant for this reaction in helium is 5.79 × 10−33 cm6 molecule−2 s−1 (Mitchell and Le Roy 1977), which indi-cates that the characteristic timescale for this recombination at the present conditions is on the order of 20 ms. The rate constant is similar for argon, 8.0 × 10−33 cm6 molecule−2 s−1 (Lynch et al 1976), suggesting a similar characteristic timescale.

→+ + +H H M H M2 (4)

τ =k

1

H M.

[ ] [ ] (5)

Based on the decay times in figure 15 for the H2/He and H2/Ar mixtures, the estimated recombination times are too long to be significant for the atomic-hydrogen decay observed, fur-ther suggesting diffusion of hydrogen atoms as the dominant decay mechanism.

Figure 16 shows time-resolved atomic-oxygen number density measured for different O2 mole fractions in the helium-based mixture. Almost independent of the O2 mole fraction, the decay time of the discharge-produced atomic oxygen is on the millisecond scale, with full decay occurring over sev-eral tens of milliseconds. The atomic-oxygen decay rates in these data sets are quite similar. The /e1 decay time constant is 1.27 ± 0.10 ms, 1.21 ± 0.08 ms, and 1.21 ± 0.07 ms for the 0.5%, 1%, and 5% O2/He mixtures, respectively.

Recombination of atomic oxygen in helium has been con-sidered to explain the long decay times observed in figure 16. For these conditions the room-temperature rate constant for

Figure 14. (a) Location of the 400 µm × 400 µm region for temporal-decay studies shown by a small square in the middle of the discharge. (b) Comparison of temporal decays of atomic-hydrogen number density from the location shown in (a) generated by a ns-pulse discharge in different buffer gases (He and Ar) at 100 Torr total pressure.

Figure 15. Comparison of atomic-hydrogen number-density temporal decays after a ns discharge pulse for different mole fractions of H2 in (a) argon and (b) helium at 100 Torr total pressure. All decays are measured in the middle of the discharge.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

15

the recombination reaction in helium is 1.27 × 10−33 cm6 molecule−2 s−1 (Campbell and Thrush 1967), which yields a characteristic recombination time of ~10 ms. This is consistent with the observed decay times, suggesting that atomic-oxygen recombination, as given in equation, is probably a significant process that affects the decay timescale.

+ + +O O He O He.2→ (6)

Results of the atomic-oxygen temporal evolutions show that the discharge-produced oxygen atoms have /e1 decay time constants on the order of 1 ms, nearly independent of the mole fraction of O2.

Figure 17 shows the model-predicted temporal decay of both atomic hydrogen and atomic oxygen in the 1%-H2/He-mixture and 1%-O2/He-mixture discharges, respectively. It can be seen that the model predictions for delay times after the discharge pulse that exceed a few tens of microseconds are clearly at variance with the experimental results. Specifically, the model predicts much faster H-atom and O-atom decay than detected in the present experiments, by about three orders of magnitude over 1 ms, compared to factor of approximately ten for H atoms and a factor of approximately three for O atoms in the experiment.

Possible reasons for the difference observed in the decay times of atomic oxygen include their accumulation or temper-ature rise from previous pulses. At these conditions the flow residence time in the cell is estimated to be on the order of 100 ms. A discharge-pulse repetition rate of 200 Hz implies the occurrence of at least 20 discharge pulses during the flow residence time in the cell, which may result in atomic-spe-cies accumulation over multiple pulses. In addition, temper-ature increases after each pulse may also accumulate over multiple pulses for the same reason. In O2/He mixtures the temperature increase may explain the difference between the measured and predicted decay times. The model utilizes the two-step atomic-oxygen recombination mechanism given in equation and equation, where M is a third body.

→+ + +O O M O M2 3 (7)

→+ +O O O O .3 2 2 (8)

Ozone, thermally stable at low temperatures, may be dis-sociated if a significant temperature rise due to multiple dis-charge pulses occurs. This would reduce the net rate of the reaction given in equation, which is one of the main atomic-oxygen recombination mechanisms, while producing more atomic oxygen. In this case the decay of atomic oxygen would be controlled by the three-body recombination mechanism given in equation.

+ + +O O M O M.2→ (9)The time constant for this reaction is on the order of 1 ms,

which is consistent with the experimentally observed atomic-oxygen decay rates in O2/He mixtures.

While this is the first work of its kind to produce highly accurate, spatially resolved, 2D atomic-species concentrations that are directly compared with a model, certainly more work needs to be performed on model development to incorporate a 2D geometry and the effects of multiple discharge pulses as the model struggles to predict these experimental results. In gen-eral, any atomic-species accumulation from previous pulses is unlikely since the model predictions are in fairly good agree-ment with the experimental data for short delay times after the pulse. Additionally, although a temperature rise accumulated over multiple sequential pulses is possible, it is fairly unlikely because of the overall gas-refresh rate through the cell; further-more, it would not explain the observed slow decay rate of the H atom. Finally, the diffusion of atoms in the axial direction (e.g. along the line connecting the electrodes) could be a pos-sible explanation for the discrepancy in these results since this is a major difference between a 1D model and a 2D diagnostic.

4. Conclusion

In conclusion, proof-of-principle 2D imaging of atomic hydrogen and oxygen has been demonstrated inside a modified low-temperature, ns-pulsed, sphere-to-sphere non-equilibrium

Figure 17. Model-predicted decay rates of atomic-hydrogen and atomic-oxygen number densities in 1%-H2/He- and 1%-O2/He-mixture discharges at 100 Torr. The results are predicted on the discharge-filament centerline, halfway between the electrodes.

Figure 16. Temporal decay curves of atomic oxygen measured for different O2 mole fractions in helium at 100 Torr total pressure. Fluorescence signal is generated by single ns-duration, pulsed-plasma discharge pulse.

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

16

discharge using fs-TALIF. Again, this is the first work of its kind to produce highly accurate, spatially resolved, 2D con-centration maps of different atomic species that are compared with a plasma-kinetics model. Multiple images of atomic-species concentrations were assembled to reconstruct an 8 mm discharge that provides a composite 2D image of atomic-spe-cies number density at a specified time after the pulsed-plasma discharge. These images depict local number densities ranging from 2 × 1012 cm−3 to 6 × 1015 cm−3 for atomic hydrogen and 1 × 1013 cm−3 to 3 × 1016 cm−3 for atomic oxygen with far greater spatial resolution and acquisition rates than those of previous laser-based diagnostics systems. The present results also yield a discharge-energy fraction going to H- and O-atom generation in ns-pulse discharges at the present conditions—about 70% for H2 dissociation in H2/He and H2/Ar mix-tures and nearly 60% for O2 dissociation in O2/He mixtures. Atomic- hydrogen and -oxygen decay rates were also mea-sured over a wide parameter space, including inert background gas and admixture composition. Atomic-hydrogen populations were shown to decay with a characteristic time scale of ~100 µs after the pulsed discharge, independent of the buffer gas. The results of atomic-oxygen populations were shown to be very different than those of atomic hydrogen, with character-istic decay timescales in excess of 1 ms. 1D kinetic-modeling results are in fairly good agreement with spatial distributions of the produced atomic species but failed to reproduce the measured decay times for both atomic hydrogen and oxygen. An additional modeling effort is required to interpret these results. This diagnostic approach is quite unique in its ability to provide spatially resolved quantitative measurements of key atomic species with limited interferences. The accuracy and speed of this diagnostic lays the initial groundwork for the full characterization of ns-pulse plasmas as well as the develop-ment of accurate, predictive modeling capabilities.

Future efforts will be focused on resolving the modeling discrepancies described above. Additionally, these ns-plasma results will be expanded to include other key atomic species such as nitrogen. These efforts will include in situ collisional-quenching measurements and subsequent corrections. To sup-port these measurements and previous modeling predictions, temperature distributions within the plasma volume will also be measured with alternative non-linear diagnostic approaches. Finally, this fs-TALIF scheme will be extended to atmospheric pressure for various combustion and plasma sources.

Acknowledgments

Funding for this research was provided by the United States Department of Energy under Contract No. DE-SC0009637, by the Air Force Research laboratory under Contract No. FA8650-15-D-2518, and by the Air Force Office of Scien-tific Research (LRIR: 15RQCOR202). Approved for public release: distribution unlimited (#88ABW-2016–4437).

The authors gratefully acknowledge the Department of Energy—University of Michigan Low Temperature Plasma Science Center—for their contributions toward model development.

References

Adams S F and Miller T A 1998 Chem. Phys. Lett. 295 305–11Aldén M, Edner H, Grafström P and Svanberg S 1982 Opt.

Commun. 42 244Amorim J, Baravian G and Ricard A 1995 Plasma Chem. Plasma

Proc. 15 721Amorim J, Baravian G, Touzeau M and Jolly J 1994 J. Appl. Phys.

76 1487Bamford J D, Jusinski L E and Bischel W K 1986 Phys. Rev. A

34 185–98Bischel W K, Perry B E and Crosley D R 1981 Chem. Phys. Lett.

82 85Bittner J, Kohse-Hoinghaus K, Meier U and Just T 1988 Chem.

Phys. Lett. 143 571Bletzinger P and Ganguly B N 1995 Chem. Phys. Lett. 247 584Bokor J, Freeman R R, White J C and Storz R H 1981 Phys. Rev. A

24 612Boogaarts M G H, Mazouffre S, Brinkman G J, van der

Heijden H W P, Vankan P, van der Mullen J A M, Schram D C and Döbele H F 2002 Rev. Sci. Instrum. 73 73

Braginskiy O V, Vasilieva A N, Klopovskiy K S, Kovalev A S, Lopaev D V, Proshina O V, Rakhimova T V and Rakhimov A V 2005 J. Phys. D: Appl. Phys. 38 3609–25

Bruce M R, Layne W B, Whitehead C A and Keto J W 1989 J. Chem. Phys. 92 2917

Campbell I M and Thrush B A 1967 Proc. R. Soc. 296 222–32Clyne M A and Nip W S 1979 Reactive Intermediates in the Gas

Phase ed D W Setser (New York: Academic)Czarnetzki U, Miyazaki K, Kajiwara T and Muraoka K 1994 J. Opt.

Soc. Am. B 11 2155DiMauro L F, Gottscho R A and Miller T A 1984 J. Appl. Phys.

56 2007Döbele H F, Mosbach T, Niemi K and Schulz-von der Gathen V

2005 Plasma Sources Sci. Technol. 14 S31Engelhardt A G and Phelps A V 1963 Phys. Rev. 131 2115Es-Sebbar E, Sarra-Bournet C, Naude N, Massines F and

Gherardi N 2009 J. Appl. Phys. 106 073302Frank J H and Settersten T B 2005 Proc. Combust. Inst. 30 1527Goldsmith J E M and Rahn L A 1990 Opt. Lett. 15 814Graves D B 2012 J. Phys. D: Appl. Phys. 45 263001Hassouni K, Gicquel A, Capitelli M and Loureiro J 1999 Plasma

Sources Sci. Technol. 8 494Hurst G S and Payne M G 1988 Principles and Applications of

Resonance Ionization Spectroscopy (Bristol: Hilger)Inoue G, Ku J K and Setser D W 1984 J. Chem. Phys. 81 5760–74Ishaq M, Evans M and Ostrikov K 2014 Int. J. Cancer 134 1517Ju Y and Sun W 2015 Prog. Energy Combust. Sci. 48 21Kiefer J H and Lutz R W 1966 J. Chem. Phys. 44 668Kulatilaka W D, Frank J H, Patterson B D and Settersten T B 2009

Appl. Phys. B 97 227Kulatilaka W D, Gord J R, Katta V R and Roy S 2013 Opt. Lett.

37 3051Kulatilaka W D, Gord J R and Roy S 2014 Appl. Phys. B 116 7Kulatilaka W D, Lucht R P, Roy S, Gord J R and Settersten T B

2007 Appl. Opt. 46 3921Kulatilaka W D, Patterson B D, Frank J H and Settersten T B 2008

Appl. Opt. 47 4672Leonov S B, Adamovich I V and Soloviev V R 2016 Plasma

Sources Sci. Technol. 25 063001Lynch K P, Schwab T C and Michael J V 1976 Int. J. Chem. Kinet.

8 651Mazouffre S, Foissac C, Supiot P, Vankan P, Engeln R, Schram D C

and Sadeghi N 2001 Plasma Sources Sci. Technol. 10 168Meier U, Kohse-Hoinghaus K, Schafer L and Klages C-P 1990

Appl. Opt. 29 4993Mitchell D N and Le Roy D J 1977 J. Chem. Phys. 67 1042Miyazaki K, Kajiwara T, Uchino K, Muraoka K, Okada T and

Maeda M 1996 J. Vac. Sci. Technol. A 14 125

J. Phys. D: Appl. Phys. 50 (2017) 015204

J B Schmidt et al

17

Mostinsky I L 2011 Diffusion Coefficient (Danbury, CT: Begell House) (doi:10.1615/AtoZ.d.diffusion_coefficient)

Niemi K, Schulz-von der Gathen V and Döbele H F 2001 J. Phys. D. Appl. Phys. 34 2330

Niemi K, Schulz-von der Gathen V and Döbele H F 2005 Plasma Sources Sci. Technol. 14 375

Preppernau B L, Dolson D A, Gottscho R A and Miller T A 1989 Plasma Chem. Plasma Process 9 157

Preppernau B L, Pearce K, Tserepi A, Wurzberg E and Miller T A 1995 Chem. Phys. 196 371

Raizer Y P 1991 Gas Discharge Physics (New York: Springer)Reichardt T A, Teodoro F, Farrow R L, Roy S and Lucht R P 2000

J. Chem. Phys. 113 2263Robert E, Barbosa E, Dozias S, Vandamme S, Cachoncinlle C,

Viladrosa R and Pouvesle J M 2009 Plasma Process Polym. 6 795

Roy S, Lucht R P and Reichardt T A 2002 J. Chem. Phys. 116 571Rusterholtz D L 2013 J. Phys. D: Appl. Phys. 46 464010Sasaki K, Nakamoto M and Kadota K 2001 Rev. Sci. Instrum.

72 2298Saxon R and Eichler J 1986 Phys. Rev. A 34 199

Settersten T B, Dreizler A and Farrow R L 2002 J. Chem. Phys. 117 3173

Shkurenkov I 2016 Plasma Sources Sci. Technol. 25 015021Shkurenkov I, Burnette D, Lempert W R and Adamovich I V 2014

Plasma Sources Sci. Technol. 23 065003Stone P M, Kim Y K and Desclaux J P 2002 J. Res. Natl Inst.

Stand. Technol. 107 327Tserepi A D, Dunlop J R, Preppernau B L and Miller T A 1992 J.

Appl. Phys. 72 2638Tung J, Tang A, Salamo G and Chan F 1986 J. Opt. Soc. Am. B

3 837van Gessel A F H, van Grootel S C and Bruggeman P J 2013

Plasma Sources Sci. Technol. 22 055010von der Gathen V S, Bornrmann T, Wagner D and Döbele H F

1991 Proc. 10th Int. Symp. on Plasma Chemistry (Bochum, Germany) vol 2, pp 2.1–2

Wiese W, Smith M and Miles B 1969 Atomic Transition Probabilities (Washington, DC: National Bureau Standards)

Wrzesinski P J, Stauffer H U, Schmidt J B, Roy S and Gord J R 2016 Opt. Lett. 41 2021

Zikratov G and Setser D W 1996 J. Chem. Phys. 104 2243

J. Phys. D: Appl. Phys. 50 (2017) 015204