INSTITUTE OF PHYSICS PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS J. Phys. D: Appl. Phys. 38 (2005) 3812–3824 doi:10.1088/0022-3727/38/20/007 Singlet oxygen generation in a high pressure non-self-sustained electric discharge

Adam Hicks, Seth Norberg, Paul Shawcross, Walter R Lempert, J William Rich and Igor V Adamovich

Nonequilibrium Thermodynamics Laboratories, Department of Mechanical Engineering, Ohio State University, Columbus, OH 43210, USA

Received 22 May 2005, in final form 19 August 2005 Published 28 September 2005 Online at stacks.iop.org/JPhysD/38/3812 Abstract This paper presents results of singlet oxygen generation experiments in a high-pressure, non-self-sustained crossed discharge. The discharge consists of a high-voltage, short pulse duration, high repetition rate pulsed discharge, which produces ionization in the flow, and a low-voltage dc discharge which sustains current in a decaying plasma between the pulses. The sustainer voltage can be independently varied to maximize the energy input into electron impact excitation of singlet delta oxygen (SDO). The results demonstrate operation of a stable and diffuse crossed discharge in O2–He mixtures at static pressures of at least up to P0 = 380 Torr and sustainer discharge powers of at least up to 1200 W, achieved at P0 = 120 Torr. The reduced electric field in the positive column of the sustainer discharge varies from E/N = 0.3 × 10−16 to 0.65 × 10−16 Vcm2, which is significantly lower than E/N in self-sustained discharges and close to the theoretically 1 predicted optimum value for O2(a ) excitation. Measurements of visible 1 3 emission spectra O2(b → X ) in the discharge afterglow show the 1 O2(b ) concentration to increase with the sustainer discharge power and to decrease as the O2 fraction in the flow is increased. Rotational temperatures inferred from these spectra in 10% O2–90% He flows at P0 = 120 Torr and mass flow rates of m˙ = 0.73–2.2 g s−1 are 365–465 K. SDO yield at these conditions, 1.7% to 4.4%, was inferred from the integrated intensity of the 1 3 (0,0) band of the O2(a → X ) infrared emission spectra calibrated using a blackbody source. The yield remains nearly constant in the discharge afterglow, up to at least 15 cm distance from the discharge. Kinetic modelling calculations using a quasi-one-dimensional nonequilibrium pulser–sustainer discharge model coupled with the Boltzmann equation for plasma electrons predict gas temperature rise in the discharge in satisfactory agreement with the experimental measurements. 1 However, the model overpredicts the O2(a ) yield by a factor of 2–2.5, which suggests that the model’s description of nonequilibrium O2–He plasma kinetics at high pressures is not quite adequate. (Some figures in this article are in colour only in the electronic version)

1. Introduction motivation for these efforts is to reduce the weight, complexity and operational difficulties associated with a chemical oxygen- Development of an electrically pumped oxygen–iodine laser ion laser (COIL), such as excessive weight, hazardous liquid has recently attracted considerable attention [1–9]. The main chemical storage and the use of two-phase pumping systems.

0022-3727/05/203812+13$30.00 © 2005 IOP Publishing Ltd Printed in the UK 3812 Singlet oxygen generation experiments One of the main goals of this research is to achieve significant stable at much higher pressures and energy inputs compared 1 yields of singlet delta oxygen (SDO) molecules, O2(a ), with self-sustained discharges, which at these conditions are in nonequilibrium gas discharge oxygen plasmas at low prone to instability development [10]. temperatures. Based on the nonequilibrium oxygen–iodine Recent experiments in a non-self-sustained discharge mixture kinetics, the threshold SDO yield required to achieve sustained by a high-energy electron beam [3,5] suggested that positive gain in oxygen–iodine laser systems is given by it can be successfully used for efficient generation of SDO at the expression [8] high pressures, up to P = 100 Torr. However, in this type of discharge, special care should be exercised to prevent foil [O (a 1)] 1 2 = , (1) breakage, which may result in contamination of the electron 3 · [O2(X )] 1+1.5 exp(403/T) beam apparatus with iodine as well as in severe damage of the where T is the flow temperature in the laser cavity. The electron gun cathode. An alternative approach is to use two strong temperature dependence of the threshold yield (14.8% at overlapping discharges, one producing a series of high-voltage, T = 300 K, 8.2% at T = 200 K and only 1.2% at T = 100 K) short pulse duration, high repetition rate ionizing pulses and provides motivation to rapidly reduce the flow temperature the other providing a dc sustainer voltage at E/N values in the before it enters the laser cavity, which can be done in a optimum SDO excitation range. In such discharges, uniform rapid supersonic nozzle expansion downstream of the electric ionization is produced by a high-voltage pulse, which is turned discharge section. For example, after an M = 3 expansion of a off before ionization instabilities develop and a uniform pulsed room temperature 10% O2–90% He flow, the static temperature discharge collapses into an arc filament. On the other hand, will be only about T = 80 K. Indeed, recent encouraging the dc sustainer voltage is far too low to produce ionization, so experiments in an RF discharge in O2–He mixtures at P = that between the pulses the decaying plasma remains stable. 10 Torr, followed by an M = 2 expansion demonstrated both At these conditions, the sustainer voltage, tailored to maximize positive gain on a 1315 nm iodine atom transition [7] and the energy input into the singlet oxygen states, draws the cw lasing with a 220 mW output power at the laser cavity electric current and couples the power to the decaying plasma. temperature of 180 K [8]. Positive gain has also been measured To increase the sustainer discharge energy loading, the pulse in a low-pressure microwave discharge in subsonic near room repetition rate should be sufficiently high to avoid complete temperature O2–Ar mixtures (P = 1.5 Torr, T = 350 K) [9]. plasma decay between the high-voltage pulses. This approach, Using an electric discharge followed by a supersonic flow first suggested and experimentally demonstrated by Hill [11], expansion would also potentially allow the achieving of high has been previously used to develop a high power, fast flow mass flow rates and therefore high laser powers. However, CO2 laser [10, 12]. this approach to reduce the flow temperature and the gain Recent experiments at the Nonequilibrium Thermody- threshold would also require operating at rather high stagnation namics Group using this crossed discharge technique [13, 14] pressures. Indeed, for an M = 3 laser cavity pressure of showed that, indeed, it allows producing stable and diffuse = P 3–5 Torr, the stagnation pressure in a 10% O2–90% He plasmas at higher pressures and much higher energy loadings = flow should be P0 100–160 Torr. Note that the discharge compared with self-sustained discharges (dc and RF). In these excitation section should be located close to the nozzle plenum experiments, the crossed discharge was generated in M = 3 to increase the flow residence time, and therefore the energy and M = 4 supersonic flows of nitrogen and air, at static pres- loading in the discharge, i.e. the pressure in the discharge sures of P = 5–10 Torr. The objective of the present paper region would be rather close to stagnation pressure. Also, to is to study singlet oxygen generation in the crossed discharge 1 optimize the O2(a ) yield in the plasma, the discharge should generated in subsonic flows (M ∼ 0.1–0.2, P ≈ P0) and at operate at reduced electric field (E/N) values where the energy significantly higher pressures. 1 input into the target O2(a ) state is maximum. Finally, to fully utilize the advantage of lowering the gain threshold by reducing the cavity temperature, the flow temperature rise in 2. Experimental the discharge should be rather modest (preferably, of the order of a few tens of degrees). This requires dilution of the feedstock The present measurements have been conducted in a new species, oxygen, in an inert carrier gas which should also have blowdown facility at the Nonequilibrium Thermodynamics 1 low collisional quenching rates of the O2(a ) state, such as Laboratories, which was specifically designed for development helium, argon or nitrogen. of an electrically pumped oxygen–iodine laser. A schematic It is well known that in nonequilibrium plasmas, the of the facility is shown in figure 1. Premixed helium/oxygen energy fraction going into electron impact excitation of and argon/oxygen flows are produced by mixing a carrier gas 1 1 O2(a ) and O2(b ) states in O2–Ar and O2–He mixtures (helium or argon, respectively) with a 50%/50% mixture of reaches a maximum at rather low E/N values, E/N < the same carrier gas with oxygen. The carrier gas cylinders 1 × 10−16 Vcm2 [1,3,4]. These E/N values are considerably are stacked to increase the overall available operation time. lower than those achieved in self-sustained nonequilibrium After mixing, the gas flow is delivered to the test section via electric discharges (dc, RF or microwave), E/N ∼ (1–10) × a 1 inch diameter, 15 ft long supply line. A slit sonic choke − 10 16 Vcm2. Therefore optimization of the energy input into plate, placed downstream of the diagnostics section, allows the singlet oxygen states suggests the use of non-self-sustained inference of both the mass flow rate through the test section electric discharges, with an external ionization source not and mole fractions of oxygen and carrier gas from their partial coupled to the applied electric field. An additional well-known pressures. The flow rate can be varied by changing the slit advantage of non-self-sustained discharges is that they remain cross section area.

3813 A Hicks et al

Figure 1. Schematic of experimental set-up.

The pulsed electrodes are powered by a Chemical Physics Technologies custom designed high-voltage (20–25 kV), short pulse duration (10–20 ns), high pulse repetition rate (up to 50 kHz) pulsed power supply. Therefore, the duty cycle of the power supply is very low; ∼1/1000. The pulse voltage is measured using a Tektronix P6015A high-voltage probe and a Tektronix TDS 3032B digital oscilloscope. The dc sustainer electrodes are powered by a DEL high current, low- voltage (3 A, 2 kV, 6 kW max) dc power supply, operated in Figure 2. Overall view of the discharge section. Flow is from left to the voltage-stabilized mode. Adjustable high-power ballast right. resistors (Powerohm, 0–10 k) are connected in series with the dc power supply to limit the maximum sustainer current. The discharge test section, the overall view of which In the present experiments, the ballast resistance has been is shown in figure 2, is made of acrylic plastic and has a varied between 0.5 and 2.0 k. The dc sustainer current rectangular inner cross section of 1 cm×5 cm. Two rectangular is measured using a Tektronix A6303 current probe with a dc electrodes, each 5 cm long and 1 cm wide, and two square Tektronix AM503B amplifier and a digital oscilloscope. In all shaped pulsed electrodes, 5 cm × 5 cm, are flush-mounted in measurements reported in this paper, the pulser was operated = the side walls and in the top/bottom walls of the discharge at the pulse repetition rate of ν 40 kHz. section, respectively (see figure 3). The dc electrodes are made The discharge test section is followed by a 17 cm long of copper and are exposed to the flow. Each pulsed electrode, optical diagnostic section with the same cross section (5 cm also made of copper, is insulated from the flow by a dielectric width and 1 cm height). Five sets of BK-7 glass windows, ceramic plate approximately 1 mm thick (see figure 3). On flush-mounted with the inside walls and evenly spaced along the opposite side, the pulsed electrodes are covered by a layer the test section provide optical access at various streamwise of plastic to prevent electrode surface exposure to air and/or locations. Visible emission spectroscopy measurements have test section gases and the development of corona discharge been conducted using a Roper Scientific Optical Multichannel − during the operation. Both sets of electrodes are located Analyzer (OMA) with a 0.5 m monochromator, 1200 g mm 1 at the same streamwise location to form a crossed pulser/dc grating blazed at 700 nm and a Roper Scientific liquid sustainer discharge, as shown in figure 3. The overall length nitrogen cooled 2D 512 × 512 pixel CCD array camera. of the test section is about 12 cm, with a crossed discharge Infrared emission measurements have been conducted using length of 5 cm. the same monochromator, 600 g mm−1 grating blazed at 1 µm

3814 Singlet oxygen generation experiments

Figure 3. Schematic of the pulsed electrode block (left) and of the discharge section with pulsed electrode blocks removed (right). and a liquid nitrogen cooled 1D 512 pixel InGaAs CCD In the present experiments, the test section total pressure array camera, on loan from Roper Scientific. To reduce was typically set at P0 = 120 Torr, and it remained constant electromagnetic interference from the pulsed power supply, during the run. At these conditions, the mass flow rate was the emission was collected using a Thor Labs 5 m long varied in the range of 0.5–2.5 g s−1 depending on the gas AFS fibre optic bundle with collimators on each end. The mixture and the choke slit area. Downstream of the slit choke collimators were positioned in front of an optical access plate, a 4 inch diameter shutoff ball valve (see figure 1) allows window in the diagnostics section wall and in front of the for the test section removal while keeping the vacuum tank slit opening of the spectrometer, respectively. To maximize under vacuum. The vacuum tank volume is approximately the signal collected by the 2D array camera, a spot-to-slit 250 ft3. Between the runs, the vacuum tank is pumped down converter has been attached to the end of the fibre optic cable. to below 1 Torr using a Stokes 212-H 150 cfm vacuum pump. The use of the fibre optic link resulted in a nearly complete During the run, the tank pressure increased by a few Torr. 1 electromagnetic noise removal. The O2(a ) concentration in the discharge afterglow and the SDO yield were evaluated 3. Kinetic model by the calibration of the fibre optics/OMA/CCD camera signal collection system using an Infrared Systems calibrated The kinetic model used in the present paper is a quasi- blackbody source IR-564. In particular, during the calibration one-dimensional compressible flow nonequilibrium pulser– the number of blackbody photons captured by the fibre optics sustainer discharge model coupled with the Boltzmann collimator, calculated from the Planck distribution, was related equation for plasma electrons. The model incorporates to the total number of the OMA counts, SBB, registered quasi-one-dimensional compressible flow equations, species 3 − across the wavelength range sampled by the OMA detector, concentrations equations for the neutral species, O2(X g ), = 1 1 + 1 3 1 λ 73 nm. Then, in the actual experiment, using the same O2(a g),O2(b g ),O2(c ),O(P),O(D),O3, Ar and collimator, the number of the OMA counts integrated over the He, and species concentrations equations for the charged 1 → 3 − + + + − − + + + + O2(a X ) band, SSDO, was related to the SDO number species, e ,O,O2 ,O4 ,O ,O2 ,Ar,Ar2 ,He and He2 . 3 + 3 1 − density, nSDO. The equation used for the SDO number density Since the energies of the A u , A u and c u electronic calculations is as follows: states of an oxygen molecule are rather close to each other (4.1–4.4 eV), they were combined into one effective level, SSDO I(λ,T)· λ · (1/ε) · τBB 1 n = · , which is referred to as O2(c ). The model also incorporates SDO · · (2) SBB A (L/4π) τSDO a number of electron impact processes in the nonequilibrium plasma, such as ionization, dissociation, electronic excitation where I(λ,T) is the Planck distribution (blackbody spectral and dissociative attachment, as well as electron–ion and ion– radiation intensity in W/m2/µmatλ=1.268 µm in the direction ion recombination, electron attachment, detachment and ion normal to the source), ε is the blackbody photon energy conversion processes. The list of kinetic processes and rates in joule, A is the Einstein coefficient for spontaneous emission 1 used is given in table 2 in the appendix. Note that the model of the O2(a ) state, L = 0.05 m is the width of the test 1 incorporates a rapid three-body O2(a ) deactivation process section (i.e. the length of the distributed SDO emission source recommended by the authors of [5]: sampled by the collimator), 4π is the full solid angle and τ is the signal collection time (in seconds) for the actual O2(a) +O+M→ O2 +O+M, experiment (SDO) and for the blackbody calibration run (BB). = −32 6 −1 = Note that the cross section of the collimator signal collection k 10 cm s (M O2), (3) region cancels out since it is the same in the actual experiment k = 0.63 · 10−32 cm6 s−1 (M = Ar), and in the blackbody calibration. However, the diameter 1 of the nearly cylindrical region sampled by the collimator which may well contribute to the overall O2(a ) deactivation (d = 2.5±0.5 mm) was also determined during the calibration rate in high pressure flows. The rate of this process with helium by moving the collimator relative to the pinhole blackbody as a third collision partner is assumed to be the same as for aperture (0.3 mm), using a three-dimensional translation stage. argon.

3815 A Hicks et al

10

0

Voltage, kV -10 Figure 5. Photograph of the discharge section in operation. 10% O2 in helium, P = 120 Torr, pulse repetition rate 40 kHz.

-20 Current, A and Voltage, kV -100 -50 0 5 00 2 voltage Time, nsec 1.5 Figure 4. Typical single-pulse discharge oscillogram. Pulse peak 1 voltage 19 kV, pulse FWHM 25 ns 10% O2 in helium, P = 120 Torr. 0.5 current

The model is coupled with the two-term Lorentz 0 expansion Boltzmann equation solver [1] with the set of 0 40 80 120 160 µ experimental cross sections [15] used as inputs. The Time ( sec) Boltzmann solver calculates the electron energy distribution -16 2 function (EEDF) in the plasma (both in the pulsed discharge Power, kW and E/N, 10 Vcm and in the dc sustainer discharge), averages the cross sections 2 over the EEDF and provides the resultant electron impact power process rate coefficients as functions of the reduced electric 1.5 field, E/N, to the main code. In the modelling calculations, 1 we used experimentally measured values of the sustainer E/N ` voltage after subtracting the cathode voltage fall. Since 0.5 the pulse voltage fall across the dielectric plates and across the sheaths is not known, the repetitively pulsed discharge 0 voltage was assumed to be an adjustable parameter, with a 0 40 80 120 160 Gaussian shape pulse duration (FWHM) of 25 ns. The pulse Time (µsec) voltage was adjusted to fit the calculated sustainer discharge Figure 6. Oscillograms of dc sustainer current, voltage, power and power to the experimentally measured value. reduced electric field (E/N) in the crossed discharge. 10% O2 in helium, P = 120 Torr, pulse repetition rate 40 kHz. 4. Results and discussion

4.1. Experiments can be seen that the pulse peak voltage is 19 kV, with the pulse full width at half maximum of about 25 ns. After the flow is started and the test section pressure is Figure 5 shows a photograph of a repetitively pulsed stabilized, the dc sustainer voltage is turned on. At these crossed discharge produced in the test section at these conditions, no breakdown is produced in O2/He or O2/Ar conditions, at a constant dc power supply voltage of UPS = flows since the maximum applied dc voltage is too low (2 kV 2 kV and ballast resistor of R = 0.46 k. For the maximum). The crossed discharge is initiated only after measurements reported in the present paper, the typical run starting the high-voltage, high repetition rate pulse sequence. time with the crossed discharge turned on was from 3 to High-voltage, short duration pulses produce ionization in the 5 s. Figure 6 shows oscillograms of the sustainer current and test section, and the relatively low sustainer voltage draws voltage, as well as of the sustainer discharge power coupled current across the test section between the pulses, while to the flow and the estimated reduced electric field, E/N, electron density is decaying due to electron recombination in the crossed discharge at these conditions. In figure 6, and attachment. In the present experiments, stable repetitively E/N is evaluated based on the flow number density at room pulsed discharge in O2–He flows was maintained at test section temperature and neglecting the cathode voltage fall in the pressures of at least up to P0 = 380 Torr. Figure 4 shows sustainer discharge. It can be seen that the sustainer current a typical oscillogram of a single high-voltage pulse fired between the pulses drops from approximately I = 2A to in a 10% O2–90% He flow, at the test section pressure of about I = 0.5 A, while the sustainer voltage, U = UPS − IR, −1 P0 = 120 Torr and the mass flow rate of m˙ = 2.2gs .It increases from U = 1.2 to 1.8 kV. At these conditions,

3816 Singlet oxygen generation experiments

2.0 0.8

10% oxygen in helium 1.5 P=120 torr, 2.2 g/sec 0.6

1.0 0.4 Current, A 0.5 O2-He

0.2 O2-Ar 0.0 Time-averaged current, A Time-averaged current, 0 50 100 150 200 Time, µsec 0.0 0 400 800 1200 1600 2000 Figure 7. Oscillograms of dc sustainer current in 10% O2–He and 10% O2–Ar mixtures. P = 120 Torr, UPS = 2kV,R = 0.5k Time-averaged voltage, V (helium) R = 1.5k (argon). Pulse repetition rate 40 kHz. Figure 8. Current–voltage characteristic of the sustainer discharge. −1 10% O2 in helium, P = 120 Torr, mass flow rate 2.2 g s . Cathode voltage fall inferred from the linear slope is 360 V. the time-dependent sustainer power varies from above 2 kW to about 0.8 kW, and the reduced electric field E/N varies from E/N = 0.5 × 10−16 to E/N = 0.8 × 10−16 Vcm2. 0.8 Note that these E/N values remain far too low for the dc discharge to become self-sustained. In fact, the discharge 2 0.6 always terminated as soon as the pulser was turned off. At the test section pressure and dc sustainer voltage used in = = V cm the present study (P0 120 Torr and UPS 2 kV), the crossed -16 0.4 discharge appeared diffuse, uniform and stable in a wide range of the ballast resistances (at least down to R = 0.46 k in oxygen–helium flows and down to R = 1.5k in oxygen– E/N, 10 0.2 argon flows). The discharge stability was also verified by checking the sustainer current oscillograms spanning a wider 0.0 time period, up to 2 ms, which exceeds the flow residence 0 200 400 600 80 000 time in the discharge (see figure 7). As the ballast resistance was reduced, the current traces in oxygen–helium mixtures Power, W = remained stable (see figure 7, R 0.46 k, top), while in Figure 9. Time-averaged E/N versus time-averaged sustainer oxygen–argon mixtures this resulted in the appearance of a power into positive column. 10% O2 in helium, P = 120 Torr. low-frequency ‘ripple’ in the sustainer current (see figure 7, R = 1.5k; bottom). At R = 1.0k in a 10% oxygen– 90% argon mixture, the current oscillations amplitude became when the conductivity in the positive column of the discharge very large (of the order of ∼0.5 A) as the sustainer discharge is high. The cathode voltage fall was estimated from the became unstable. On the other hand, as expected, the crossed x-axis intercept of the linear slope of the current voltage discharge in oxygen–helium mixtures remained quite stable in characteristic in figure 8, Uc = 360 ± 50 V, which is rather the entire range of the ballast resistance tested. We conclude close to the normal glow discharge cathode fall, 370 V in that operating the crossed discharge in oxygen–argon mixtures oxygen and 180 V in helium (both for copper cathode) [10]. at high-power loadings would most likely require an additional At dc voltages exceeding the cathode fall, Uc, a linear current stabilization technique, such as blowing helium across the voltage characteristic (see figure 8) is expected, because in the sustainer electrode faces, which would help in dissipating non-self-sustained discharge the rate of ionization produced incipient arc filaments. by high-voltage pulses is independent of the sustainer field. Figure 8 shows a current voltage characteristic of the Figure 9 shows time-averaged values of the reduced sustainer discharge in a 10% O2–90% He flow at P0 = 120 Torr electric field in the sustainer discharge, E/N =(U − −1 and the mass flow rate of m˙ = 2.2gs . It can be seen that at Uc)/Nd, plotted against the time-averaged power added to low-voltages the sustainer current remains very low and nearly the positive column of the sustainer discharge, (U − Uc) · I, independent of the applied voltage, while at high-voltages for the conditions of figure 8 (10% O2–90% He mixture, −1 the current exhibits linear voltage dependence. Basically, P0 = 120 Torr, m˙ = 2.2gs ). Note that both E/N if the applied voltage is low, the voltage across the cathode and sustainer discharge power values are calculated after layer of the discharge, Uc ≈ UPS, is insufficient to accelerate subtracting the cathode voltage fall, Uc, from the voltage ions toward the cathode, release enough secondary electrons between the electrodes. First, it can be seen that the from the cathode surface and multiply them in the cathode sustainer power coupled to the flow in these cases, achieved layer to sustain a significant current [10], even at the conditions at the lowest ballast resistance tested of R = 0.46 k is

3817 A Hicks et al

50 1E+0

40 1E-1 30

20 10% O2-He 1E-2

20% O2-He 10 5.0 O2 1E-3 ) ε Input power percentage 0 00.5 .5 f( 1E-4 0.3 2.0 E/N (10-16 V cm2) 1E-5 Figure 10. Energy fractions into electron impact excitation of 1.0 O (a 1) versus reduced electric field for different O –He mixtures, 2 2 0.5 0.7 predicted by the Bolzmann solver. 1E-6 048121620 ε, eV approximately 900 W (at the total sustainer discharge power of 1200 W). This can be converted into the energy loading of Figure 11. EEDFs in a 10% O2–He mixture at different values of 0.28 eV/O2 molecule in the positive column of the discharge. E/N (labelled in the figure), predicted bythe Bolzmann solver. For comparison, in 5% O2–He and 15% O2–He mixtures (m˙ = 2.0gs−1 and 2.4 g s−1, respectively), the highest energy loadings achieved were 0.48 eV/O2 and 0.17 eV/O2 molecule, respectively. Further increase of the sustainer discharge power would require either increasing the sustainer voltage beyond UPS = 2 kV (limited by the power supply used in the present work) or further reduction of the ballast resistance, i.e. below R = 0.46 k. Total sustainer discharge powers measured in these experiments (i.e. power added to the cathode layer and to the positive column together) allow estimating the upper bound of the flow temperature rise (i.e. the temperature increase if all input discharge energy would thermalize instantaneously),

T = 150 K in a 5% mixture, T = 160 K in a 10% mixture Intensity, (arbitrary units) and T = 140 K in a 15% mixture. The actual temperature rise is somewhat less due to energy storage in the electronic 758 760 762 764 766 768 770 levels of oxygen and O atom generation. From figure 9, one can also see that the reduced electric Wavelength, nm field in the sustainer discharge increases with the discharge − − Figure 12. Experimental (——) and synthetic (----)partially = × 16 × 16 2 1 3 power, from E/N 0.3 10 to 0.65 10 Vcm . Actual resolved O2(b → X ) emission spectra. 10% O2 in helium, E/N values are somewhat smaller due to the flow temperature P = 120 Torr, T = 365 K. rise in the discharge. Comparing these experimental results with the modelling calculations predicting the energy balance in a nonequilibrium plasma as a function of E/N, using a power and to optimize the discharge performance, we first 1 Boltzmann equation solver (see figure 10), one could expect measured O2(b )visible emission spectra at 762 nm. Indeed, approximately 20% of the input discharge power in the 10% both electron impact cross section data and kinetic modelling O2–He mixture to go to the electron impact excitation of the calculations [1] suggest that both these low-energy electronic 1 O2(a ) state. This would result in a yield of [0.28 eV/O2 states are most efficiently excited at rather close E/N values. 1 1 molecule]/[0.98 eV/O2(a)molecule] · 20% ≈ 6%. Similar Therefore, one could expect the O2(a ) and O2(b ) yields estimates for a 15% O2 and for a 5% O2 mixture predict to scale in a similar way. Also, we used rotationally resolved 1 1 3 3.5% and 10% O2(a ) yield, respectively. At these values of O2(b → X ) spectra to infer the flow temperature sustainer discharge power and E/N, higher energy loadings in the discharge afterglow. Figure 12 shows a partially 1 1 3 per O2 molecule and consequently higher O2(a ) yields rotationally resolved (0,0) band of the O2(b → X ) can be achieved by increasing the flow residence time in the emission spectrum with the band centre at 762 nm, in a 10% −1 discharge, i.e. by reducing the mass flow rate through the test O2–He mixture at P0 = 120 Torr and m˙ = 2.2gs . Figure 12 section. Figure 11 shows the exponential part of the EEDF also shows the synthetic spectrum calculated at the rotational at different values of the reduced electric field, including the temperature of T = 365 K. The spectroscopic constants and E/N range in which the fraction of the discharge power into the rotational line strengths for the synthetic spectrum were 1 −16 2 O2(a ) state peaks, E/N = (0.3–0.5) × 10 Vcm . taken from [16,17]. The rotational temperature was evaluated During these measurements, only limited access was from the comparison of the two spectra by the least squares available to an infrared CCD array camera. Therefore, to method, T = 365 ± 15 K at the conditions of figure 12. 1 estimate the dependence of the O2(a ) yield on the discharge At the same time, we also measured the excited O atom

3818 Singlet oxygen generation experiments

16000

5% O2 2.2 g/sec 12000 10% O2 0.73 g/sec 15% O2

8000

4000 Intensity, arbitrary units Intensity, arbitrary units 0 0 200 400 600 800 1000 1230 1240 1250 1260 1270 1280 1290 1300 1310 Power, W Wavelength, nm

1 → 3 1 3 Figure 13. 762 nm O2(b X ) ( ) and 777 nm O atom Figure 14. O2(a → X ) emission spectra at m˙ = 2.2 and  ♦ −1 line ( ) intensities versus time-averaged sustainer power into 0.73 g s . 10% O2 in helium, P = 120 Torr. positive column. 5%, 10% and 15% O2 in helium, P = 120 Torr.

in the discharge. This was done by reducing the slit choke line emission at 777 nm to estimate the dependence of the area, A∗, while keeping the discharge pressure constant, which oxygen dissociation fraction on the discharge power. Earlier resulted in the mass flow rate reduction, experiments [7, 9] suggested that high O atom concentrations 1
could adversely affect O (a ) yield and small signal gain in   − 2 P A∗ γ 2 (γ +1)/(γ 1) nonequilibrium oxygen–iodine mixtures. m˙ = ρuA = √0 . (4) 1 3 R γ +1 Figure 13 plots the O2(b → X ) emission intensity, T0 which is proportional to the O (b 1) concentration, as well as 2 In equation (4), u, ρ, P and T are velocity, density, pressure the O atom line emission intensity versus the discharge power, 0 0 and temperature in the discharge section, respectively, A is for three different O2–He mixtures at P0 = 120 Torr (at mass − the cross section area of the discharge section, γ = 1.625 flow rates of m˙ = 2.0, 2.2 and 2.4 g s 1). It can be seen 1 is the specific heat ratio and R is the gas constant. If the that the O2(b ) population grows approximately linearly as a function of the discharge power, both in 5% O –helium and sustainer discharge power is kept the same while the mass 2 flow rate is reduced, this would increase the discharge energy in 10% O2–helium mixtures. Note that at the same discharge power, the O (b 1) emission intensity in the 5% O –helium loading per oxygen molecule, which would increase both SDO 2 2 yield and the flow temperature. Figure 14 shows typical mixture is nearly the same as in the 10% O2–helium mixture. 1 3 1 O2(a → X ) emission spectra, (0,0) band with the band This means that the O2(b ) concentrations in both mixtures 1 centre at 1.268 µm, at the mass flow rates of m˙ = 2.2 and are close and the O2(b ) yield in the 5% O2 mixture, defined −1 as the fraction of all available oxygen molecules in the b 1 0.73 g s . The spectra in figure 14 are shown after subtracting a nearly flat baseline from the raw infrared signal. It can be state, is nearly double that in the 10% O2 mixture. On 1 (b 1) seen that, indeed, the O2(a ) emission intensity increases as the other hand, the O2 emission intensity in the 15% 1 the mass flow rate is reduced. On the other hand, O2(a ) O2 mixture is only about 60% compared with the other two 1 emission intensity measured at several locations downstream mixtures, which indicates that at these conditions the O2(b ) of the discharge section using five optical access windows (see yield is only about one-third of that in the 10% O2 mixture. Open symbols in figure 13 indicate the O atom line emission figure 5) was found to be nearly independent of the distance intensity, which in the 5% O mixture considerably exceeds from the discharge (up to at least 15 cm). As expected, this 2 1 demonstrates that at the present conditions O2(a ) molecules that in the 10% and 15% O2 mixtures (by a factor of 3–4). This result suggests that the oxygen dissociation fraction in in the discharge afterglow survive for a relatively long time, at least several milliseconds. the low O2 percentage mixtures may be significantly higher than in flows with significant oxygen fractions, which may During these measurements, the sustainer discharge power 1 into the positive column was found to be nearly independent well reduce the O2(a ) yield at these conditions. Visible emission spectroscopy measurements have also of the mass flow rate (see table 1). The SDO yield was inferred from the experimental been conducted in O2–Ar mixtures. However, at these 1 3 infrared emission spectra using blackbody calibration. During conditions we did not detect any O2(b → X ) emission, which indicates the emission intensity to be at least a factor of the calibration, both the signal collection volume sampled 30–50 lower than in O2–He flows. In this case, it is possible by the fibre optics collimator and contributions of different 1 that the O2(b ) state may be rapidly quenched in resonance parts of the collection volume into the overall SDO signal energy transfer collisions with Ar atoms. were determined. Note that the calibration is significantly 1 O2(a ) yields were measured in a 10% O2–90% He simplified by the fact that the flow is optically thin for the 1 → 3 mixture, at the same test section pressure of P0 = 120 Torr and O2(a X ) emission due to an extremely long radiative − 1 at three different mass flow rates, m˙ = 2.2, 1.1 and 0.73 g s 1, lifetime of the O2(a ) state, with the Einstein coefficient for = × −4 −1 i.e. for three different flow velocities and flow residence times spontaneous emission of Arad 2.2 10 s [18, 19].

3819 A Hicks et al Table 1. Time-averaged Time-averaged Energy Time-averaged Mass flow positive column power into positive loading, eV/O2 positive column −1 −16 2 rate, g s voltage, U − Uc(V ) column, W molecule E/N,10 Vcm 2.2 1330 840 0.27 0.65 1.1 1290 920 0.59 0.64 0.73 1140 860 0.82 0.57

Temperature, K 1.0E-5 6.0 600 Temperature Yield 5.0 500 1.0E-6 Threshold yield in M=3 flow 4.0 400 1.0E-7 3.0 300 1.0E-8 Yield, % 2.0 200 e- 1.0 100 + 1.0E-9 O2 Ionized Species Fractions - 0.0 0 O 0.5 1.0 1.5 2.0 2.5 1.0E-10 Mass flow rate, g/sec 0 0.01 0.02 0.03 0.04 0.05

1 Distance, m Figure 15. O2(a ) yields and flow temperatures as functions of = the mass flow rate. 10% O2 in helium, P 120 Torr. Figure 16. Ionized species fractions in the pulser–sustainer discharge in 10% O2–90% He mixture at P = 120 Torr. Mass flow rate is 1.1 g s−1. Figure 15 shows the SDO yields and the flow temperatures at these conditions. In figure 15, the error bars in the SDO yield indicate a combined uncertainty in the infrared spectra 1.0E-2 baseline and in the flow temperature. It can be seen that the SDO yield increases from 1.7% ± 0.2% to 4.3% ± 0.7% as 1.0E-3 the mass flow rate is reduced (see figure 15). However, the O2(a) higher yield is achieved at a somewhat higher flow temperature, O T = 465 ± 15 K at m˙ = 0.73gs−1 versus T = 365 ± 15 K at 1.0E-4 0 0 O (b) m˙ = 2.2gs−1. Figure 15 also plots the SDO threshold yield 2 O calculated from equation (1) at the static temperature in an 1.0E-5 3 M = 3 flow for the stagnation temperatures, T0, measured in the discharge section, 1.0E-6

  Neutral Species Fractions − −1 γ 1 2 T = T0 1+ M . (5) 2 1.0E-7 0.00 0.02 0.04 0.06 0.08 0.10 0.12 In equation (5), γ = 1.625 for a 10% O2–90% He mixture. It can be seen that the actual yield, although rather low, exceeds Distance, m the threshold yield by almost a factor of two (see figure 15). Figure 17. Oxygen species fractions in the pulser–sustainer Note that both the test section pressure and the mass discharge and in the M = 3 supersonic expansion in a 10% O2–90% flow rate through the test section, m˙ , are dictated by the He mixture at P = 120 Torr. Mass flow rate is 1.1 g s−1. target flow parameters in the supersonic laser cavity, such as static pressure and temperature (which determines the Mach number; see section 1) and cavity height (which determines we estimate that in the present setup the distance between the the mass flow rate). For this reason, both density and mass pulser electrodes, and therefore the flow residence time, can flow rate in equation (4) cannot be easily varied within a be increased by at least a factor of two. wide range. Therefore a practical way of reducing the flow velocity and increasing the flow residence time in the discharge 4.2. Modelling calculations would be to increase the cross section area of the discharge, for example by moving the pulsed electrodes farther apart. Figures 16 and 17 show the axial distributions of key Since the repetitively pulsed discharge in the test section was ionized and neutral species concentrations along the crossed successfully generated at pressures of up to P0 = 380 Torr, discharge in the 10% O2–90% He mixture, at P0 = 120 Torr,

3820 Singlet oxygen generation experiments

− m˙ = 1.1gs 1 and the sustainer discharge positive column 600 6 voltage of U − Uc=1.29 kV. These flow conditions correspond to the discharge power into the positive column 500 5 of 920 W and E/N = 0.64 × 10−16 Vcm2 (see table 1). Mach Number The discharge length in the streamwise direction is 5 cm. 400 4 Downstream of the discharge, the mixture flows through a converging–diverging nozzle to reach M = 3 in the laser 300 3 cavity. The nozzle throat is located 3.25 cm downstream of 2.2 g/sec the discharge, i.e. at x = 8.25 cm. From figure 17, it can 200 1.1 g/sec 2 be seen that soon after the flow enters the discharge, the Temperature, K 0.73 g/sec electron density between the ionizing pulses remains very 100 Mach number (2.2 g/sec) 1 low because of the three-body attachment of electrons to O2 − 0 0 molecules, with subsequent rapid ion conversion from O2 to O−. Because of the low electron density, the sustainer current 0.00 0.02 0.04 0.06 0.08 0.10 0.12 density in the upstream half of the discharge also remains Distance, m low, which results in a very slow O (a 1) and O (b 1) 2 2 Figure 18. Gas temperatures in the pulser–sustainer discharge and generation by the sustainer discharge. Because of this effect, in the M = 3 supersonic expansion in 10% O2–90% He mixtures at a significant part of the discharge volume generates very little P = 120 Torr and different mass flow rates. 1 1 singlet oxygen. However, as the O2(a ),O2(b ) and O atom concentrations increase, the electron detachment from − − 1 12 O and O2 ions becomes more rapid since both O2(a ) 1 and O2(b ) molecules as well as O atoms are very efficient 2.2 g/sec electron detachers (see table 2), which contributes to the rapid 10 electron density increase. High peak electron density and high 1.1 g/sec peak E/N during the pulses also result in significant oxygen 8 0.73 g/sec dissociation (see figure 17). Higher electron density, and therefore higher sustainer 6 current density in the downstream half of the discharge 1 0.73 g/sec in turn accelerates the O2(a ) generation (see figure 17). 4 Note that ozone formation in the discharge remains quite SDO Yield, % 1.1 g/sec insignificant, especially after the flow temperature starts 2 increasing because of Joule heating (see figures 17 and 18). 2.2 g/sec (a 1) From figure 17, one can see that the O2 yield, defined as 0 the ratio of the O (a 1) concentration over the initial oxygen 2 0.00 0.02 0.04 0.06 0.08 0.10 0.12 concentration, reaches approximately 6.5% at the discharge exit, which is close to the calculated oxygen dissociation Distance, m 1 fraction, [O]/[O2], 7.5%. At the discharge exit, the calculated Figure 19. Calculated O2(a ) yield as a function for different sustainer discharge power added to the flow is 900 W, which is mass flow rates. 10% O2 in helium, P = 120 Torr. close to the experimentally measured value at these conditions, 920 W (see table 1). Figure 17 also shows that during a supersonic expansion to M = 3 downstream of the discharge, is reduced, are significantly higher than the experimentally 1 the O2(a ) yield remains nearly constant because of its very measured values, by more than a factor of 2. At this time, slow collisional deactivation rate. The predicted freezing of the reason for this difference remains unclear. Conceivably, it 1 1 the O2(a ) concentration is consistent with measurements of might be due to some additional rapid O2(a ) deactivation 1 the O2(a ) emission intensity at five different locations in the process that the present kinetic model does not include, subsonic (i.e. much slower) flow downstream of the discharge although it appears unlikely. Indeed, in the present model 1 section, which was nearly constant (see section 4.1). the process of equation (3) is by far the most rapid O2(a ) Figure 18, which plots the axial temperature distributions quenching process, with a characteristic time of τquench ∼ 2ms in the discharge and in the supersonic nozzle at m˙ = 0.73, (assuming 5% O2 dissociation fraction). At these conditions, −1 1 1.1 and 2.2 g s , shows that higher electron density in the the characteristic time for two-body O2(a ) quenching by downstream part of the discharge results in a more rapid Joule O atoms is much longer; τquench ∼ 0.2 s. For comparison, heating of the flow. Calculated flow temperatures at the end of the flow residence time in the discharge is τres ∼ 0.5–1.5 ms, 1 the discharge are in satisfactory agreement with the rotational depending on the flow rate, with O2(a ) being generated over 1 3 temperatures inferred from the O2(b → X ) spectra such only about half of that time, 0.25–0.75 ms (see figure 19). For 1 as shown in figure 12 (see figure 18). As can be seen from this reason, even rapid three-body quenching of O2(a ) by figure 18, the flow temperature in the supersonic expansion to the process of equation (3) has a relatively modest effect on the M = 3 is reduced to T = 90–140 K. SDO yield (within about 25%), which was found by excluding 1 Figure 19 shows the measured and the calculated O2(a ) the reaction of equation (3) from the model. Basically, the flow yields for the three different mass flow rates. One can see residence time both in the discharge and in the afterglow at the that the predicted yields, which increase as the mass flow rate conditions of the present experiments is so short that collisional

3821 A Hicks et al

1 deactivation of O2(a ) simply does not seem to have enough rotational temperatures in the 10% O2–90% He flows in the time to significantly affect the SDO yield. Again, this is crossed discharge at P0 = 120 Torr and mass flow rates of 1 = −1 = consistent with the O2(a ) emission intensity measurements m˙ 2.2–0.73 g s , T0 365–465 K. The SDO yield at these at several locations downstream of the discharge section, which conditions, 1.7% to 4.4%, was inferred from the integrated 1 1 3 show that the O2(a ) concentration in the discharge afterglow intensity of the (0,0) band of the O2(a → X ) infrared 1 remains unchanged. This fact also suggests that O2(a ) emission spectra calibrated using a blackbody source. The quenching on the test section walls is unlikely to significantly yield remains nearly constant in the discharge afterglow, up to reduce the SDO yield. Wall quenching also appears unlikely at least 15 cm distance from the discharge. because of the short flow residence time and the high pressure Kinetic modelling calculations using a quasi-one- in the test section. dimensional nonequilibrium pulser–sustainer discharge model Since in the modelling calculations the calculated coupled with the Boltzmann equation for plasma electrons 1 discharge power was tailored to the actual sustainer discharge predict O2(a ) yields of 3.5% to 11.5% and a gas temperature power, these arguments suggest that the present kinetic rise of T = 50–200 K at the sustainer discharge power model may significantly overestimate the sustainer discharge going into the positive column of approximately 900 W and energy fraction going into direct electron impact excitation of the mass flow rate of m˙ = 0.73–2.2 g s−1. The predicted 1 O2(a ), as predicted by the Boltzmann solver (see figure 10) temperature rise in the discharge is in satisfactory agreement for the experimentally measured E/N values (see figure 9). with the emission spectroscopy temperature measurements. Note that predictions of the Bolzmann solver with the cross The model also predicts a fairly high oxygen dissociation sections used in this paper are in good agreement with the fraction in the discharge, from 4% to 9%. The modelling results of kinetic modelling calculations by other authors (e.g. calculations show that increasing the flow residence time in see [1]). Further experiments, where the O2 fraction in the the crossed discharge by reducing the mass flow rate results flow will be increased up to 15–20% and the sustainer voltage in an increase of the O (a 1) yield. However, comparison of will be varied within a wider range (up to U = 5 kV versus 2 PS the experimental results and the modelling calculations shows U = 2 kV in this work) using a higher voltage dc power PS that the model significantly overestimates the SDO yield (by supply, are expected to provide additional insight into this a factor of 2–2.5), which suggests that the model’s description issue. of nonequilibrium O2–He plasma kinetics at high pressures is not quite adequate. 5. Summary

This paper discusses the results of recent experiments on Acknowledgments singlet oxygen generation in a high-pressure, non-self- sustained crossed discharge. The discharge consists of This work has been supported by the Phase II MDA STTR completely overlapping high-voltage (15–20 kV), short pulse grant, technically directed by Directed Energy Directorate duration (10–20 ns), high repetition rate (up to 50 kHz) pulsed of AFRL, Gas Laser Laboratories, under a subcontract from discharge, which produces ionization in the flow, and a low- PSI, Inc. We would like to express our sincere gratitude to voltage dc sustainer discharge which sustains current in a Fiodar Pliavaka and Syarhei Harbatau from Chemical Physics decaying plasma between the pulses. The sustainer discharge Technologies, Inc. for continuing support and collaboration voltage can be independently varied to choose the reduced as well as for their help in setting up and operating the pulsed electric field value such that it maximizes the sustainer energy power supply. We would also like to thank Roper Scientific for input into electron impact excitation of SDO. The results loaning us an infrared CCD array camera used in the present demonstrate the operation of a stable and diffuse crossed infrared emission measurements. discharge in O2–He and O2–Ar mixtures at static pressures of at least up to P0 = 380 Torr and sustainer discharge powers of at least up to 1200 W (discharge power and energy loading in Appendix the positive column of up to 920 W and 0.82 eV/O2 molecule, respectively). The reduced electric field in the positive column of the sustainer discharge in O2–He flows varies Table 2. Reactions and rates used in the kinetic model. from E/N = 0.3 × 10−16 to 0.65 × 10−16 Vcm2, which is 3 −1 significantly lower than E/N in self-sustained discharges and Reaction Rate (cm s ) Reference 1 − → + − − a close to the theoretically predicted optimum value for O2(a ) O2 +e O2 +e +e σ [15] − → + − − excitation. O2 +e O+O +e +e σ [15] − → − × −29 · −0.5 Measurements of visible emission spectra O (b 1 → O2 +e +O2 O2 +O2 3.878 10 T [20] 2 − − → − 3 O2 +e O +O σ [15] X ) downstream of the discharge in O2–He flows show − − O2 +e → O2(a) +e σ [15] the O (b 1) yield to increase nearly proportionally to the − − 2 O2 +e → O2(b) +e σ [15] − − sustainer discharge power and to decrease as the O2 fraction O2 +e → O2(c) +e σ [15] − → − in the flow is increased. Excited O atom emission intensity O2 +e O+O+e σ [15] − → − is shown to increase as the O fraction in the flow is reduced, O2 +e O+O(d) +e σ [15] 2 + − → × −6 · −0.5 O2 +e O+O 5.196 10 T [20] which suggests a higher oxygen dissociation fraction at these + − → × −5 · −1.0 O2 +O2 O2 +O2 5.960 10 T [20] conditions. The partially resolved rotational structure of O+ +O− → O +O+O 1.000 × 10−7 [6] 1 3 2 2 2 the O2(b → X ) (0,0) band have been used to infer

3822 Singlet oxygen generation experiments Table 2. (continued) References + − → × −5 · −1.0 O2 +O O+O2 5.960 10 T [20] + − → × 7 [1] Napartovich A P, Deryugin A and Kochetov I 2001 Discharge O2 +O O+O+O 1.000 10 [6] → × −16 production of the singlet delta oxygen for an iodine laser O2(a) +O O2 +O 2.000 10 [6] J. Phys. D: Appl. Phys. 34 1827–33 → × −32 O2(a) +O+O2 O2 +O+O2 1.000 10 [5] [2] Schmiedberger J and Fujii H 2001 Radio-frequency plasma jet → × −33 O2(a) +O+He O2 +O+He 6.300 10 [5] generator of singlet delta oxygen with high yield Appl. → × −18 O2(a) +O2 O2 +O2 2.200 10 [4] Phys. Lett. 78 2649–51 → × −15 O2(a) +O3 O2 +O2 +O 3.800 10 [4] [3] Ionin A A, Klimachev Y M, Kotkov A A, Kochetov I V, → × −18 O2(a) +O2(a) O2(b) +O2 2.200 10 [4] Napartovich A P, Seleznev L V, SinitsynDVand → × −11 O2(b) +O3 O+O2 +O2 1.540 10 [4] Hager G D 2003 Non-self-sustained electric discharge in → × −17 O2(b) +O2 O2(a) +O2 3.200 10 [4] oxygen gas mixtures: singlet delta oxygen production → × −14 O2(b) +O O2 +O 4.000 10 [4] J. Phys. D: Appl. Phys. 36 982–9 −14 O2(b) +O→ O2(a) +O 4.000 × 10 [4] [4] Vagin N P, Ionin A A, Klimachev Yu M, Kochetov I V, → × −12 O2(b) +O3 O2(a) +O2 +O 7.000 10 [4] Napartovich A P, Sinitsyn D V and Yuryshev N N 2003 −13 O2(c) +M→ O2(a) +M 1.000 × 10 [4] Glow discharge in singlet oxygen Plasma Phys. Rep. 29 −13 O2(c) +M→ O2(b) +M 9.000 × 10 [4] 211–19 −34 O+O2 +O2 → O3 +O2 6.900 × 10 [20] [5] Vasiljeva A N, Klopovskiy K S, Kovalev A S, Lopaev D V, −15 O+O3 → O2 +O2 4.200 × 10 [4] Mankelevich Y A, Popov N A, RakhimovATand −33 1 O+O+M→ O2 +M 4.800 × 10 [4] Rakhimova T V 2004 On the possibility of O2(a ) −33 O+O+M→ O2(a) +M 2.400 × 10 [4] production by a non-self-sustained discharge for −33 O+O+M→ O2(b) +M 1.200 × 10 [4] oxygen–iodine laser pumping J. Phys. D: Appl. Phys. −34 O+O2 +M→ O3 +M 6.900 × 10 [4] 37 2455–68 → × −15 [6] Stafford S D and Kushner M 2004 O (1) Production in O+O3 O2(a) +O2 2.800 10 [4] 2 − → × 15 He/O2 mixtures in flowing low pressure plasmas J. Appl. O+O3 O2(b) +O2 1.400 10 [4] − → − × −10 Phys. 96 2451–65 O+O2 O +O2 3.300 10 [4] + → + × −29 [7] Carroll D L, Verdeyen J T, King D M, Zimmerman J W, O+O +M O2 +M 2.000 10 [6] − − −10 Laystrom J K, Woodard B S, Richardson N, Kittell K, O +O→ e +O2 1.900 × 10 [4] O− +O+ → O+O 2.000 × 10−7 [4] Kushner M J and Solomon W C 2004 Measurement of − → − × −10 positive gain on the 1315 nm transition of atomic iodine O +O2(a) e +O3 3.000 10 [4] 1 − − −10 pumped by O2(a ) produced in an electric discharge Appl. O +O2(b) → O+e +O2 7.000 × 10 [4] − − − Phys. Lett. 85 1320–2 O +O → O +e 5.000 × 10 15 [4] 2 3 [8] Carroll D L et al 2005 Continuous-wave laser oscillation on O− +O → O +O +e− 3.000 × 10−10 [4] 3 2 2 the 1315 nm transition of atomic iodine pumped by − + → . × −25 O +O2 +M O+O2 +M 2000 10 [6] O (a 1 ) produced in an electric discharge Appl. Phys. − + → × −25 2 g O +O +M O+O+M 2.000 10 [6] Lett. 86 111104–111104-3 − + → × −7 O2 +O O2 +O 2.000 10 [6] [9] Rawlins W T, Lee S, Kessler W J and Davis S J 2005 − → − × −18 O2 +O2 O2 +e +O2 2.200 10 [20] Observations of gain on the transition I(2P →2 P ) by − → − × −10 1/2 3/2 O2 +O2(a) O2 +O2 +e 2.000 10 [4] energy transfer from O (a 1 ) generated by a microwave − → − × −10 2 g O2 +O2(b) O2 +O2 +e 3.600 10 [4] discharge in a subsonic-flow reactor Appl. Phys. Lett. 86 − → − × −10 O2 +O O3 +e 3.000 10 [4] 051105–051105-3 + → + × −11 O +O2 O2 +O2 1.900 10 [4] [10] Raizer Y P 1991 Gas Discharge Physics (Berlin: Springer) + → + × −30 O2 +O2 +O2 O4 +O2 2.400 10 [4] [11] Hill A E 1973 Continuous uniform excitation of + → + × −13 O4 +O2 O2 +O2 +O2 1.730 10 [4] medium-pressure CO2 laser plasmas by means + → + × −10 O4 +O2(a) O2 +O2 +O2 1.000 10 [4] of controlled avalanche ionization Appl. Phys. Lett. + → + × −10 O4 +O2(b) O2 +O2 +O2 1.000 10 [4] 22 670 + → + × −10 O4 +O O2 +O3 3.000 10 [4] [12] Generalov N A, Zimakov V P, Kosynkin V D, Raizer Yu P and + − −26 O +e +O2 → O+O2 1.000 × 10 [4] Roitenburg D I 1975 Method for significantly increasing the + → + × −11 O +O2 O2 +O 2.000 10 [6] stability limit of the discharge in fast-flow large-volume + → + × −10 O +O3 O2 +O2 1.000 10 [6] lasers Tech. Phys. Lett. 1 201 + − → × −7 O4 +e O2 +O2 2.200 10 [4] [13] Nishihara M, Meyer R, Cundy M, LempertWRand He + e− → He+ +e− +e− σ [21] Adamovich I V 2004 Development and operation of a + − → × −8 · −0.9 supersonic nonequilibrium MHD channel 35th He2 +e He + He 1.040 10 T [22] + → + × −31 Plasmadynamics and Lasers Conf. (Portland, OR, He +He+He He2 +He 0.630 10 [10] − + → × −9 · −0.7 29 June–1 July 2004) AIAA Paper 2004-2441 e +O2 O(d) +O 8.880 10 T [6] e− +O+ → O(d) 5.300 × 10−13 · T−0.5 [6] [14] Nishihara M, Jiang N, Lempert W R, AdamovichIVand e− +e− +O+ → O(d) +e− 5.120 × 10−27 · T−0.45 [6] Gogineni S 2005 MHD supersonic boundary layer control O(d) +O→ O+O 8.000 × 10−12 [6] using pulsed discharge ionization 43rd Aerospace Sciences −11 Meeting and Exhibit (Reno, NV, January 2005) AIAA Paper O(d) +O2 → O+O2(b) 2.047 × 10 [6] −12 2005-1341 O(d) +O2 → O+O2(a) 1.296 × 10 [6] −12 [15] Eliasson B and Kogelschatz U 1986 Basic data for modeling of O(d) +O2 → O+O2 3.839 × 10 [6] −10 electrical discharges in gases: oxygen ABB Corp. Tes. O(d) +O3 → O2 +O 1.200 × 10 [6] − Report KLR 96-11 O(d) +O → O +O 1.200 × 10 10 [6] 3 2 2 [16] Babcock H D and Herzberg L 1948 Fine structure of the red O(d) +He→ O+He 1.00 × 10−10 [6] system of atmospheric oxygen bands Astrophys. J. 108 Ar + e− → Ar+ +e− +e− σ [21] 167–90 + − → . × −5 · −0.6 Ar2 +e Ar + Ar 2 300 10 T [10] [17] WatsonJKG1968 Rotational line intensities in 3–1 + → + × −31 Ar +Ar+Ar Ar2 +Ar 1.460 10 [10] electronic transitions Can. J. Phys. 46 1637–43 a The rate is calculated by the Boltzmann solver using an [18] Newman S M, Lane I C, Orr-Ewing A J, NewnhamDAand Ballard J 1999 Integrated absorption intensity and enistein experimental cross section, σ. 1 → 3 − coefficients for the O2(a g X g ) (0,0) transition: a comparison of cavity ringdown and high resolution fourier

3823 A Hicks et al transform spectroscopy with a long-path absorption cell comparing calculations within the kinetic and fluid models J. Chem. Phys. 110 p 10749 of the positive column plasma of a dc oxygen discharge [19] Lafferty J, Solodov A M, Lugez C L and Fraser G T 1998 Tech. Phys. 48 983–94 Rotational line strength and self-pressure-broadening [21] Rapp D and Englander-Golden P 1965 Total cross sections 1 → 3 − = → coefficients for the 1.27 µm a g X g v 0 0 for ionization and attachment in gases by electron band of O2 Appl. Opt. 37 2264 impact. I. Positive ionization J. Chem. Phys. 43 1464–79 [20] Bogdanov E A, Kudryavtsev A A, Tsendin L D, [22] Carata L, Orel A E and Suzor-Weiner A 1999 Dissociative + Arslanbekov R R, Kolobov V I and Kudryavtsev V V 2003 recombination of He2 molecular ions Phys. Rev. A 59 Substantiation of the two-temperature kinetic model by 2804–12

3824