In situ and ex situ NO oxidation assisted by sub-microsecond pulsed multi-pin-to-plane corona discharge: the effect of pin density

Abstract

The in situ (direct) and ex situ (indirect) nitric oxide (NO) oxidation experiments were carried out at room temperature in an atmospheric pressure multi-pin-to-plane corona discharge reactor. For in situ oxidation the gas mixture (NO–O₂–H₂O–N₂) was directly treated in the plasma discharge, whereas for the ex situ oxidation the plasma reactor was used to generate ozone and other species using the O₂–N₂ gas mixture and at the outlet of the reactor the NO–N₂ mixture is added. In this study, we have mainly investigated the influence of (i) reactor configuration on discharge morphology and ozone production, (ii) the NO removal efficiency of in situ and ex situ processes, (iii) input pulse energy deposition on NO oxidation and (iv) the influence of humidity on NO oxidation and NO₂ selectivity. It is demonstrated that the increase of pin density (17 and 57 pins) increased the pulse energy (at a given applied voltage), and significantly decreased streamer branching and the corresponding volume occupied by the plasma channels. Therefore, NO conversion decreased. For in situ NO oxidation, at a fixed energy density, the 17 pin reactor was more efficient than the 57 pin reactor. A total NO conversion is obtained by ex situ plasma oxidation, whereas only 42% NO conversion is reached for in situ corona discharge using the 17 pin reactor at 42.5 mJ per pulse. However, in the ex situ plasma oxidation process when the ozone concentration exceeds about 90% of the NO inlet concentration, the NO₂ selectivity dramatically decreases due to the secondary oxidation reaction of NO₂ into NO₃ and further conversion into N₂O₅. It is observed that the addition of 5% humidity does not influence the pulse energy deposition, whereas it has improved the NO conversion by 30% and decreased the NO₂ selectivity by increasing HNO₃ and HNO₂ production.

---

1. Introduction

Nitrogen oxides, NO_x (NO and NO₂) are the major pollutants in the atmosphere, which lead to acid rain and photochemical smog, and have been shown to be detrimental to human health and the environment. Automobiles and other mobile sources contribute to about 50% of the NO_x that is emitted in the atmosphere, and NO is the major portion of it.^1,2 When NO is released into the atmosphere in less than an hour it is oxidized to NO₂. NO₂, under the UV light, produces NO and atomic oxygen. The latter one further reacts with an oxygen molecule (O₂) and significantly increases the atmospheric ozone (O₃) concentration. Consequently, several methods, collectively called DeNO_x process, have been developed to treat the effluents of mobile sources and power plants. Among them Selective Catalytic Reduction (SCR)³ and Selective Non-Catalytic Reduction (SNCR) are currently used to convert NO_x into nitrogen (N₂) molecules.⁴ Granger and Parvulescu⁵ reviewed the recent developments in DeNO_x process and suggested that an increase in the NO₂ content in the exhaust gas greatly improves the efficiency of the DeNO_x catalytic system at low temperature. Koebel et al.⁴ reported about 95% NO_x reduction into N₂ using NH₃ as a reducing agent with 50% NO₂ in the NO_x mixture. Therefore, to meet NO_x regulation and to improve the DeNO_x process efficiency, the pre-oxidation of NO into NO₂ is very much needed.

The atmospheric pressure non-thermal plasma and catalyst coupling systems have emerged as one of the most feasible techniques for DeNO_x processes.^6–11 Non-thermal plasma discharge can oxidize NO into NO₂, then NO₂, which is more active than NO, can increase catalysts performances and lifetime. Several reactors like pulsed corona discharge, dielectric barrier discharge (DBD), surface discharge and ferro-electric pellet packed-bed discharge have been investigated for NO_x removal. Among them, pulsed corona discharge process has proved a competitive technology for NO_x removal owing to the very fast voltage rise time and high active species density in the discharge volume.^12–15 Takaki et al.¹⁶ have investigated the NO_x removal using multi-pin-to-plane (MP), trench and plane-to-plane DBD reactor configurations, and reported that the MP configuration shows more conversion than the other configuration. In addition, they have noticed that the increase in number pins decreases the NO conversion.

In corona discharge, electrical discharges between two conductors occur in the gas phase when the induced electric field exceeds the breakdown voltage required to ionize or excite the gas, however the subjected energy should be insufficient to cause a spark.¹⁷ In atmospheric pressure corona discharge several active species are produced in less than tens of milliseconds. It is widely reported that the O atoms and ozone are main reactive species for NO conversion into NO₂.⁹ In literature, many researchers have proposed the ozone formation¹⁸ and NO removal mechanisms occurring in different gas mixtures by numerical methods correlated with experimental data.¹⁹ Mok and Nam²⁰ investigated the relative importance of O, OH, HO₂ and ozone species on NO oxidation into NO₂ using pulsed corona discharge reactor, and reported that ozone was the most important species involved in NO conversion. Recently, Jõgi et al.²¹ demonstrated the direct (in situ) and indirect (ex situ) plasma oxidation of NO using small and medium-scale test-bench DBD reactors, and observed similar results for NO_x and N₂O₅ formation in both methods. Although the general characteristics of this reaction are well known, still a fundamental work is needed to understand its mechanism and how this relates to the direct conversion of NO into NO₂ under plasma discharge. Indeed, it is necessary to understand the energy efficiencies of the in situ and the ex situ processes under similar operating conditions.

In this study, NO selective conversion into NO₂ is studied using pulsed sub-microsecond multi pin-to-plane corona discharge reactor. The influence of number of pins, i.e. the increase in number of corona discharge pins density under similar operating conditions, on NO conversion is investigated as a function of plasma dissipated energy. A special attention is paid to identify and to quantify the by-products during NO removal experiments. The effect of NO molecules on various active species production, by corona discharge, is examined with and without NO in the gas stream. The in situ and ex situ NO conversion is explored at room temperature under dry and wet conditions.

2. Experimental

The experimental set-up employed for NO removal at room temperature, schematically shown in Fig. 1, includes three main parts: (1) gas and humidity handling system, (2) the non-thermal plasma reactor, and (3) analytical tools.

Fig. 1 General schematic of the experimental setup.

2.1. Gas and humidity handling system

High purity gas mixture containing O₂, N₂, and NO were prepared in a gas handling system and their composition was controlled by calibrated high-precision mass flow controllers (MFC, Brooks). Water vapor was added to the gas stream using controlled evaporation and mixing system (CEM®, Bronkhorst High-Tech). It consists of a liquid (H₂O) flowmeter, a flow controller for the carrier gas (N₂), and a gas–liquid mixing chamber with heat exchanger for total evaporation. The carrier gas, controlled by a MFC is used to stimulate the evaporation process as a mixing component, and furthermore to transport the vapor. The process is highly reproducible and efficient. After mixing in the manifold, the gas stream then passes through a temperature-controlled line that preheats the gas at 90 °C and prevents water condensation. Similarly, the gas line from the reactor downstream to the analytical tools is also preheated at 90 °C to avoid the water and other plasma produced species condensation. For all the experiments the total gas flow was kept constant at 1 L min⁻¹ (STP), the water vapor concentration was fixed at 5%, and the NO and O₂ concentrations was fixed at 500 ppm and 10% respectively, unless otherwise mentioned.

2.2. Non-thermal plasma reactor

The corona discharges are ignited between multi-pin-to a plane electrode configuration in a closed reactor with a total volume of about 40 cm³. Two reactor configurations consisting of 17 and 57 aligned pins (inter-pin distance 13 and 4 mm, respectively) in a length of 230 mm. The discharge gap, i.e. the distance between pins and the plane, is precisely adjusted using a vacuum compatible linear displacement device (MDC BLM-133-1). Results presented in this paper were obtained with a discharge gap of 12 mm at room temperature and atmospheric pressure. The front and rear ends of the reactor was equipped with MgF₂ windows for discharge observation, photography, and optical emission spectroscopy measurements.

Fig. 2 shows the photograph of the 17 pins corona reactor, and the inserted figure represents an example of discharges showing the morphology and branching structure of the streamers developed from each pin in nitrogen atmosphere.

Fig. 2 Photograph of 17 pins-to-plane corona discharge reactor.

The fast discharges were driven by a storage-transfer energy capacitors bank (power supply ALE, 30 kV, 500 J s⁻¹) in a Blumlein-like circuit and electrically switched by a hydrogen thyratron (CV 6022, National Electronics). After commutation, the Blumlein circuit produces pulses with a peak voltage of about 1.2–1.4 times the charging voltage (1–5 kV in this work). Then, this high voltage was feed to the reactor through a pulse-forming network (PFN, Dielectric Sciences 2012 SST coaxial cables), thus a maximum open circuit voltage of eight times the charging voltage of each cable was achieved. Under these conditions, single pulses up to 40 kV peak voltage in 80 ns duration (FWHM) and short rise time (80 ns) were produced at a repetition rate that can be varied from single shot up to 1 kHz. The fast voltage rise time (kilovolts per nanosecond) allows achieving significant over-voltage at breakdown. The operation of the power supply and the repetition rate were adjusted by a remote control through optic fibers. Khacef et al.⁶ demonstrated that for a given reactor, the mode of energy deposition, i.e. by changing the pulse energy or pulse repetition rate, controls the active species production and reaction kinetics. Therefore, in this study, the pulse energy was varied by tuning the charging voltage in the range 1–5 kV, corresponding to applied voltage up to 20 kV and stored energy up to 100 mJ, whilst the frequency was kept constant at 350 Hz.

2.3. Analytical tools

The applied voltage and discharge current were measured using a HV probe (Tektronix P6015A, 1000×) and a current probe (Pearson 4100), respectively. The output signals were transmitted to a digital oscilloscope (Tektronix DPO 3054) and saved for analysis and energy deposition calculation.

An intensified charge-coupled device camera (IStar, 1024 × 256 pixels, Andor Technology) was used to monitor the development of corona discharge and to acquire the discharge images. The ICCD camera is fitted with a 100 mm Canon f/2.8 L Macro lens.

The gaseous species at the outlet of the reactor were analyzed online and quantified using Fourier Transform Infra-Red spectroscopy (FTIR, Nicolet 6700, Thermo Scientific). Additionally, a multi-gas analyzer (X-Stream, Emerson), equipped with IR and UV detectors was used to quantify the NO_x and CO_x (CO and CO₂) species. The ozone concentration was monitored using an ozone analyzer (Ozomat MP, ANSEROS) based on non-dispersive ultra-violet absorption (Lambert–Beer) at wavelength of 254 nm.

For FTIR measurements, a 10 m length path cell coupled with liquid nitrogen cooled Mercury–Cadmium–Tellurium (MCT) detector was used. The spectra were collected using Omnic software with 5 scans per spectrum and a spectral resolution of 0.5 cm⁻¹. The background spectra was collected under O₂ (10%)/N₂ gas stream with 16 scans per spectrum. TQ-Analyst software was used for calibration as well as for data processing. The calibration concentrations were adjusted between 500 ppb and 750 ppm. As reported by Sivachandiran et al.,²² a special attention has been paid to select the vibrational and rotational bands for NO, NO₂, N₂O and ozone to avoid the bands overlapping.

The NO conversion rate is determined using the eqn (1), where, [NO]_in and [NO]_out respectively denote NO inlet and outlet concentrations.

The NO₂ selectivity is calculated using eqn (2), where [NO₂]_out denotes NO₂ concentration quantified at the reactor outlet during corona discharge and [NO₂]_in represents the NO₂ concentration measured before plasma discharge ignition.

The ozone production and NO removal efficiency (η in g kW h⁻¹) have been determined using eqn (3), where [X]_C is the amount of ozone produced or converted NO in g m⁻³, Q is the total flow rate in m³ h⁻¹, and P is the injected power in kW.

As suggested by Kim et al.²³ it is worth to mention that the comparison based on NO removal efficiency does not provide an absolute index regarding which result is better, because it is strongly dependent on SIE and several other operating conditions such as initial concentrations, temperature and a gas composition. However, one can have an overview on the process efficiency. In this study this parameter is used to differentiate the efficiencies of an in situ and ex situ plasma processes for NO removal at room temperature and under similar operating conditions.

2.4. Experimental procedure

In order to study the NO to NO₂ conversion by corona discharge and to understand the reactivity of the various species, produced by plasma discharge, on NO oxidation, the following three kinds of experiments have been performed:

(1) In situ or direct NO to NO₂ oxidation: the plasma discharge was ignited in a gas stream containing NO (500 ppm)–O₂ (10%)–N₂ (balance).

(2) Influence of humidity on direct NO to NO₂ conversion at room temperature: similarly to in situ experiments, the plasma discharge was ignited in a flow of NO (500 ppm)–O₂ (10%)–H₂O (5%)–N₂ (balance).

(3) Ex situ or indirect NO to NO₂ oxidation: the plasma discharge was ignited in O₂ (10%)–N₂ mixture producing ozone and other species, then 50 mL min⁻¹ of NO (1% in N₂) was concurrently injected downstream the plasma reactor as shown in Fig. 3. After mixing NO and O₃ at the reactor downstream, the NO concentration is 500 ppm in 1 L min⁻¹ total flow.

Fig. 3 Schematic representation of ex situ NO conversion under dry condition at room temperature.

3. Results and discussion

3.1. Effect of reactor configuration on the corona discharge

3.1.1. Plasma energy deposition. All measurements were performed for a constant repetition frequency of 350 Hz and the energy deposition in the plasma reactor, the specific input energy (SIE), was evaluated by varying the applied voltage up to 20 kV (charging voltage in the range 1–5 kV). The SIE is determined by:where f is the pulse repetition frequency, Q is the total gas flow rate at standard conditions, and E_p is the energy deposited into the plasma volume per current pulse. E_p was determined through the time-resolved measurements of the applied voltage (U) and discharge current (I) and is given by eqn (5), where T is the pulse duration.

An example of the typical temporal evolution of voltage and current waveforms are reported in Fig. 4 for both reactor configurations (17 and 57 pins). It is worth to mention that the increase in number of pins decreases the distance between the pins. It is noticed that, for all the investigated conditions, even if the applied voltage waveforms appears quite smooth and similar for both reactors, the current waveforms presented two well separated peaks in the case of 17 pins configuration. At the same applied voltage, 57 pins reactor has shown about 1.8 times more discharge current as compared to 17 pins. Therefore, it can be concluded that the increase in number of pins not only alters the current waveform but also increases the discharge current at a given applied voltage.

Fig. 4 Typical voltage and current waveforms at 19 kV set point: (a) 17 pins and (b) 57 pins reactor.

At the working conditions used in these experiments, the measured pulse energy (E_p) did not exceed (42.5 ± 0.5) mJ for the 17 pins and (69.5 ± 0.5) mJ for the 57 pins reactor. It is worth to mention that, the pulse energy could be reduced by optimizing the distance between pins and the plane. As reported in eqn (4), at a given flow rate and frequency, the SIE can be expressed as pulse energy (E_p), therefore, pulse energy is used in the following discussion.

The evolution of pulse energy (mJ) as a function of applied voltage is presented in Fig. 5. For both reactors, pulse energy increases with increasing applied voltage, and 57 pins reactor exhibits about 1.6 times higher pulse energy than 17 pins reactor for all the investigated applied voltage. Lee and Yeom²⁴ observed similar phenomenon in pin-to-plane dielectric barrier discharge reactor under He flow. The authors determined the discharge current as directly proportional to the total area of the powered electrode, i.e. the number of pins. The same authors also evidenced that the increase in number of pins not only increases the discharge current but also increases the emission intensities of He* metastable species and oxygen atoms by increasing the corresponding species concentrations. As can be seen in Fig. 5, the pulse energy is not linearly dependent on the applied voltage under the given operating conditions. The pulse energy (or SIE) profiles of 17 and 57 pins reactors follow the same trend and present a voltage threshold at ≈13 kV, beyond that pulse energy increases two times faster. The increase in pulse energy with increasing the number pins implies the fact that there is a mutual effect between corona discharge generated by each pin.

Fig. 5 Evolution of pulse energy (E_p) as a function of applied voltage for both reactors: Q = 1 L min⁻¹ and f = 350 Hz.

From Fig. 5, the average energy injected per pin can be determined for both reactors. It is evidenced that the injected energy per pin increases with increasing the applied voltage. Moreover, at about 19 kV and 350 Hz, 17 pins reactor has consumed about twice more energy per pin than the 57 pins reactor. This finding emphasis the fact that, at constant operating conditions, the increase in pins density decreases the energy deposited per pin.

3.1.2. Corona discharge morphology. The corona discharge morphology significantly influences the NO conversion at the given operating conditions. The volume occupied by the plasma channels could be varied by increasing the number of pins and the discharge length or by increasing the number pins at a fixed discharge length, i.e. by increasing the pins intensity per cm².

In this study, as reported in Section 2.2, the number pins varied at a constant discharge length. In literature, Takaki et al.¹⁶ have reported that the NO removal decreases in corona DBD reactor with increasing the number of pyramid-to-plane at constant operating conditions. However, authors have not proposed any hypotheses relating number pins and the discharge characteristics. The volume occupied by the plasma channels could be estimated using the discharge photographs and the numerical simulation. The pulsed corona discharge is characterized by a filamentary structure corresponding to complex streamer dynamics that is largely analyzed in the literature.^25,26

Fig. 6 shows the morphology of the streamer developed from each pin in pure N₂ at atmospheric pressure for a given set of operating conditions (inter-pin distance 13 mm (17 pins) and 4 mm (57 pins), applied voltage 16 kV, frequency 350 Hz, pin-to-plane distance 12 mm, and exposure time 5 ms). The illuminating paths, on the photos, indicate the position where the streamer heads have moved. Below a certain applied voltage, the corona discharge filaments are almost invisible to the human eyes, and only at the optimum voltage, which is not sufficient to generate arc discharge, and after a long adaptation in the dark, they can just be observed. In both pictures, many streamers reach the electrode and the branching behavior of the streamers in N₂ is very different in the two reactor geometries studied. The streamers slightly more diffuse in the case of 13 mm inter-pin distance (17 pins reactor). Interestingly, the mutual effect due to the adjacent pins is clearly observed in the case of 4 mm inter-pin distance (57 pins reactor). As can be seen in Fig. 6, when the inter-pin distance is increased to 13 mm or higher, the influence between the pins progressively decreases and the streamlines associated with each pin was similar to that of a single-pin taken separately. The mutual effect due to the adjacent pins was also studied experimentally by Mraihi et al.²⁵ and confirmed by electric field calculations. These authors have shown that the peak current and streamer branching structure are clearly affected by the mutual effect existing between pins. In that case, the streamlines of the electric field show an electrostatic repulsive behavior explaining the reduction of the streamer branches in the inter-pins gap. Meziane et al.²⁷ investigated the reactions kinetic in air–NO mixture in multi-pin to plane atmospheric corona discharge reactor using numerical simulation model. These authors observed that the significant amount of ozone is produced near the pin tip (illuminating region) by the three-body reaction (O + O₂ + M → O₃ + M) and inside the micro discharge streamer branches where the O concentration is high. Therefore, it can be suggested that, it is certainly interesting to have a discharge with an umbrella structure, with more micro discharge branches, similar to a single pin or multi pin separated with an optimum distance as shown in Fig. 6(a) for 17 pins reactor.

Fig. 6 Time integrated pictures of corona discharge in N₂ at atmospheric pressure for two inter-pin distances (a) 13 mm and (b) 4 mm: V_applied = 16 kV, f = 350 Hz, pin-to-plane gap = 12 mm, exposure time = 5 ms.

3.1.3. Ozone and N₂O production. In order to understand the influence of number of pins on ozone and other species production, the plasma was ignited in O₂ (10%)–N₂ mixture, and the gaseous species were continuously analyzed and quantified at the reactor downstream. Fig. 7 shows the typical FTIR spectrum illustrating the plasma processing of O₂–N₂ mixture at room temperature and for an energy pulse deposition of 42.5 mJ.

Fig. 7 FTIR spectrum acquired during the plasma processing of O₂ (10%)–N₂ mixture at room temperature for 17 pins reactor configurations.

Irrespectively of the number of pins, the only species identified by FTIR at the reactor downstream are ozone, N₂O and CO₂ using the absorption bands at 1021–1000 cm⁻¹, 2248–2223 cm⁻¹ and 2388–2383 cm⁻¹; respectively. Notably, the small amount of CO₂ (about 10 ppm) observed at the maximum pulse energy indicated the presence of impurities in the gas handling system. As can be seen in Fig. 7, the absorption bands of NO, NO₂ and HNO₃ are not observed at 1858–1902 cm⁻¹, 1605–1584 cm⁻¹ and 1325–1300 cm⁻¹, respectively. This finding emphasizes the fact that neither NO_x (NO and NO₂) and nor HNO₃ are identified by FTIR spectroscopy. The similar observation is made using multi gas analyzer. However, the production of NO_x species cannot be ruled out in air-plasma chemistry. It can be suggested that, the NO_x and HNO₃ are probably produced at certainly lower level than the detection limits of FTIR (500 ppb) and gas analyzer (1 ppm). In addition, it can also be suggested that the discharge pulse energy is not sufficient to produce NO, which is the main species for NO₂ and HNO₃ formation.¹⁸

The most important step in corona discharge is the excitation and dissociation of O₂ and N₂ molecule by electrons with sufficient energy. In literature, Penetrante et al.²⁸ reported that, when O₂ concentration is 5% or higher, the plasma discharge produce ground-state and metastable excited-state O and N₂ excited species through electron-impact dissociation reactions. However, only the ground-state O(³P) atoms are involved in ozone production reaction as mentioned in eqn (6). Chen and Davidson²⁹ (and reference therein) demonstrated the ozone and NO_x formation mechanisms in DC corona discharge using numerical simulation and suggested that half of the ozone, in total production, is produced by excited metastable N^*₂, i.e. (N^*₂(A³Σ_u⁺)) species (eqn (7) and (8)). In addition, the excited metastable N^*₂ species also involve in N₂O production as mentioned in eqn (9). However, the reaction (7) is two order of magnitude faster than the reaction (9). Therefore, it can be suggested that the exited metastable N^*₂ molecules are involved in O atoms production rather than in N₂O formation.

O(³P) + O₂ + N₂ → O₃ + N₂, k₂₉₈ = 2 × 10⁻¹⁴ cm³ s⁻¹ (ref. 30) (6)

N^*₂ + O₂ → N₂ + 2O, k₂₉₈ = 3 × 10⁻¹² cm³ s⁻¹ (ref. 31) (7)

O + O₂ + N₂ → O₃ + N₂, k₂₉₈ = 1.5 × 10⁻¹⁴ cm³ s⁻¹ (ref. 32) (8)

N^*₂ + O₂ → N₂O + O, k₂₉₈ = 6 × 10⁻¹⁴ cm³ s⁻¹ (ref. 33) (9)

The O₃ and N₂O concentration quantified at the reactor downstream, as a function of pulse energy, is reported in Fig. 8.

Fig. 8 Ozone (a) and N₂O (b) production as a function of energy deposition for the direct plasma processing of O₂ (10%) –N₂ mixture (17 and 57 pins reactor configurations).

As can be seen in Fig. 8(a), for both 17 and 57 pins reactor configurations, ozone concentration gradually increases with increasing the pulse energy (E_p). This can be explained by the increase of O atoms concentration with increasing the applied energy as mentioned in eqn (6). For the 17 pins reactor the maximum of 660 ppm of ozone was measured at 42.5 mJ per pulse energy and the corresponding ozone production energy efficiency is 5.8 g kW h⁻¹. It is noteworthy to mention that the obtained value is around 10 times lower than the corona discharge reactor reported in the literature,³⁴ however, the energy efficiency could be improved by optimizing the reactor operating conditions. Moreover, to produce the same amount of ozone, about 69.5 mJ per pulse energy was required for 57 pins reactor, which is 1.6 times more than the energy deposited in 17 pins reactor. Similar tendency is observed for all investigated energy density. The following two possible explanations could be pointed out for the less ozone production with high energy in 57 pins reactor.

(i) Ozone decomposition by O atoms: as suggested in Section 3.1, the addition of pins increases the O atom formation and, therefore, the ozone can be decomposed to O₂ molecule by the reaction reported in eqn (10). Kogelschatz et al.¹⁸ predicted by the numerical methods that the ozone formation drastically drops down if the micro discharges become “too strong”. As reported in Fig. 5, at 19 kV the 57 pins reactor consumes about 1.6 times more energy, thus the micro discharges become strong enough to decrease the ozone production.

O + O₃ → 2O₂, k₂₉₈ = 8 × 10⁻¹⁵ cm³ s⁻¹ (ref. 32) (10)

(ii) Volume occupied by the plasma channels: as reported in Fig. 6, the decrease in number of pins, i.e. the increase in the inter-pins distance, limited the mutual influence of the pins and then increases the volume occupied by the plasma channels leading to large zone production where the active species are produced and chemistry takes place. As reported by Meziane et al.²⁷ the O atoms are produced inside the micro discharge channel; therefore at given operating conditions, the increase of the number of pins decreases the ozone production. These results emphasis the fact that, the increase of the number of pins only increases the total energy consumption but does not increase the ozone production.

As reported in Fig. 8(b), N₂O profiles for 17 and 57 pins reactor configurations follow the same trend as ozone profiles, and N₂O concentration gradually increases with increasing the pulse energy from 16 to 69.5 mJ. For both reactors, even with maximum pulse energy, about 14 ppm N₂O is quantified. Similarly to ozone production, as reported in Fig. 8(a), to produce the same amount of N₂O, 57 pins reactor requires about 1.6 times more energy as compared to 17 pins reactor. These findings highlight the fact that, irrespectively of the number of pins, the ozone and N₂O production mechanisms are interconnected. Under given operating conditions, 17 pins reactor exhibits higher ozone production with lower pulse energy as compared to 57 pins reactor. Therefore, it can be proposed that, for the better ozone production at low energy deposition, the inter-pin distance should be optimum to minimize the mutual influence of the discharge streamer branches (Fig. 6).

3.2. In situ plasma NO oxidation

3.2.1. O₂ (10%)–NO (500 ppm)–N₂ mixture. In the following we present results of the experiments performed by plasma processing of dry gas mixtures (O₂ (10%)–NO (500 ppm)–N₂ (balance)) at a repetition frequency of 350 Hz in both 17 and 57 pins reactors. The Fig. 9 reports the typical FTIR spectra illustrating the effect of the pulse energy on NO into NO₂ conversion at room temperature for the 17 pins reactor. Similar spectra were observed in the case of 57 pins reactor. Under these oxygen-rich conditions, a significant fraction of input plasma energy is dissipated in the dissociation of O₂ that become the dominant process because O₂ dissociation is much more efficient than N₂ dissociation in the electron energy range studied (6 to 9 eV). This dominant oxidation process induces the NO conversion for pulse energy higher than 7.4 mJ and NO₂, N₂O, and HNO₂ (nitrous acid) are identified at the reactor outlet.

Fig. 9 Typical FTIR spectra acquired during plasma processing of O₂ (10%)–NO (500 ppm)–N₂ (balance) at room temperature for different energy deposition.

It is noticed that without any discharge, about 25 ppm of NO is converted into NO₂ by reacting with O₂ in the gas handling system and inside the reactor. Indeed, the NO₂ concentration in the gas stream, before discharge ignition, is considered for NO₂ selectivity calculation as mentioned in eqn (2). The plasma processing at the highest energy (42.5 mJ per pulse) even for a long period (20 min) reveals significant amount of NO at the reactor downstream, signifying that NO is not completely oxidized by in situ corona discharge.

The weak signals observed in the absorption band in the range 733–900 cm⁻¹ is attributed to the gas phase HNO₂ (ref. 35) indicating the presence of moisture in the system, possibly arising from the air supply or from the moisture adsorbed onto the walls of the gas handling system. The HNO₂ is mainly produced by the reaction between NO and OH radicals produced by plasma discharge.¹⁹ Interestingly, absorption bands at 1325–1300 cm⁻¹ corresponding to HNO₃ is not observed even at the highest energy deposition used in these experiments. However, the amount of HNO₂ and HNO₃ produced can be indirectly determined by calculating the total NO_x balance before and after plasma processing.

As can be seen in Fig. 9, remarkably ozone is not observed (absorption band at 1021–1000 cm⁻¹) whatever the energy deposition, whereas, without NO in the gas stream about 660 ppm of ozone was measured as reported in Fig. 8(a). For the both reactors, about 15 ppm of N₂O are quantified at the reactor downstream, which corresponds to the concentration produced without NO in the gas matrix as reported in Fig. 8(b). This result evidences the fact that the NO added in the gas mixture neither modified the N₂O formation reaction kinetics and nor the added NO converted into N₂O and, indeed, it has significantly modified the ozone production reaction pathways in corona discharge. Even though ozone is not identified at the reactor downstream, the NO conversion and NO₂ production are observed at room temperature and are reported in Fig. 10.

Fig. 10 NO conversion (a) and NO₂ selectivity (b) as a function of pulse energy for 17 and 57 pins reactors: plasma processing of O₂ (10%)–NO (500 ppm)–N₂ (balance) at room temperature.

For both reactors NO conversion rate increases with increasing the pulse energy. Indeed, in the range of energy explored, the 17 pins reactor has shown better NO conversion than the 57 pins reactor. As an example, at 42.5 mJ per pulse, 42% of NO are converted in the 17 pins reactor while only 26% conversion rate was achieved with the 57 pins reactor. These results should be correlated to the previous ozone production data obtained in plasma processing of O₂–N₂ mixture and reported in Fig. 8(a). In these plasma conditions and without NO in the gas stream, about 660 ppm of ozone is produced. However, under the same operating conditions, when 500 ppm of NO are added to the gas stream, only 220 ppm of NO are converted which is only 33% of the expected ozone production. Therefore, it can be proposed that the added NO in the O₂–N₂ mixture remarkably influences the ozone production pathways and decreases the ozone concentration. As reported in electro-hydrodynamic and kinetic modeling studies by Meziane et al.²⁷ the chemical kinetic is no more located inside the discharge channels but is also extended and distributed in the entire discharge volume. Subsequently, the O atoms and ozone molecules are produced near each pins and inside the discharge filaments, and the NO conversion takes place around the pin. Therefore, it can be suggested that owing to a small volume occupied by the plasma channels under each pin, due to the mutual effect between two pins, a low NO conversion is reached. Much work is needed to improve the NO conversion such as optimizing the reactor configuration and the distance between pins and pin-to-plane. In addition, it is established that the hydrocarbon additives such as propene, ethanol could significantly improve the NO conversion rate at room temperature.²³

In literature, the NO conversion into NO₂, have been investigated by several authors using various types of corona discharge reactors and obtained close to the NO conversion value reached in this study. Yoon et al.³⁶ demonstrated NO conversion using micro corona discharge reactor. These authors have achieved about 25% NO conversion in NO (400 ppm)–O₂ (1%)–N₂ gas mixture at 200 mW input power. In an another study, Malik et al.³⁷ compared the NO removal efficiency of the volume and surface discharge corona reactors at room temperature and under dry condition. Interestingly, using volume discharge reactor, authors obtained about 40% NO conversion with SIE of about 100 J L⁻¹ (NO (350 ppm)–O₂ (10%)–N₂) which is comparable to the conversion reached by the 17 pins reactor. Kim et al.³⁸ investigated the NO removal using positive pulsed corona reactor and reported about 26% NO conversion in NO (400 ppm)–O₂ (10%)–CO₂ (10%)–N₂ gas mixture at 4 L min⁻¹ and 50 °C. Therefore it can be proposed that the irrespective of the mode of discharge and energy deposition, without any additive like hydrocarbons, H₂O₂ and SO₂, lower than 50% NO conversion is obtained, however, the NO removal energy efficiency could vary.

In plasma processing of O₂–NO–N₂ mixture, the NO to NO₂ oxidation is directly correlated to O and ozone species produced following the reactions:

NO + O + N₂ → NO₂ + N₂, k₂₉₈ = 2.5 × 10⁻¹² cm³ s⁻¹ (ref. 32) (11)

NO + O₃ → NO₂ + O₂, k₂₉₈ = 1.8 × 10⁻¹⁴ cm³ s⁻¹ (ref. 30) (12)

As reported by Atkinson et al.³² the oxidation reaction (11) is predominant. The consumption of O by NO (eqn (11)) is about two order of magnitude faster than the ozone consumption by NO molecules (eqn (12)). The production of NO₂ and the destruction of NO are also regulated by an oxido-reduction cycle involving reactions (11)–(13).

NO₂ + O → NO + O₂, k₂₉₈ = 9.7 × 10⁻¹² cm³ s⁻¹ (ref. 30) (13)

On the other hand, the consumption of O atoms to form ozone molecules through the reaction (8) is about two order of magnitude slower than the NO to NO₂ oxidation through the reaction (11). Thereby, and in concordance with reaction rates of the reactions (8) and (11), and (13), the ozone formation remain at a very low level because of the limited amount of O atoms available for the reaction.

As can be seen in Fig. 10(b), for both 17 and 57 pins reactors, the NO₂ selectivity gradually increases with increasing the pulse energy until 25 mJ, and thereafter it remains constant at 95% ± 2%. As proposed beginning of this section, the missing NO₂ can be attributed to the amount of HNO₃ and HNO₂ produced during plasma processing. Though in situ corona plasma process exhibits only 42% NO conversion, indeed, more than 95% is converted into NO₂. As demonstrated by Koebel et al.⁴ the NO₂ rich condition enhances the NO_x reduction and N₂ selectivity in DeNO_x process. Therefore, it can be proposed that multi pin-to-plane corona reactor could be incorporated to the upstream of the ammonia DeNO_x reactor to increase the N₂ selectivity and to achieve the complete NO_x removal.

3.2.2. O₂ (10%)–NO (500 ppm)–H₂O (5%)–N₂ mixture. Humidity is inevitably present in the exhaust gas stream. The influence of humidity on NO conversion and NO_x formation has been widely investigated.³⁹ In this study, 5% of humidity was added to the gas mixture of O₂ (10%)–NO (500 ppm)–N₂ (balance) and the NO conversion rate was followed as a function of pulse energy deposition maintaining the other experimental conditions as constant. Complementary experiments have also been performed without NO in the gas stream and evidenced that the humidity does not influence the injected energy density. However, the reactor being maintained at room temperature, water condensation on the reactor wall is observed. Fig. 11 shows typical FTIR spectra illustrating the plasma processing effect on humid gas mixture with and without NO (500 ppm) in the stream for a pulse energy of 42.5 mJ.

Fig. 11 Typical FTIR spectra produced by plasma processing of O₂ (10%)–H₂O (5%)–N₂ (balance) without and with 500 ppm NO at room temperature and for a pulse energy of 42.5 mJ.

It is worth to mention that under humid condition, ozone quantification by gas analyzer with UV detector is not reliable. Therefore, FTIR spectroscopy has been used to compare the ozone production under dry and humid condition using the absorption peaks area between 968 and 1074 cm⁻¹ (results not shown). It is well established that the FTIR peak area does not exhibit linear relationship with high concentration. However, the peak area of the species can be compared for the experiments with and without humidity under the same operating conditions. For all the investigated energy deposition, the addition of 5% humidity decreases the ozone production approximately by a factor 4 as compared to dry condition. As demonstrated in the literature⁴⁰ the O radicals react with H₂O molecules by chain reactions and thus ozone concentration decreases.

For plasma processing of both dry and humid gas mixture, it is evidenced that the ozone is not observed as soon as NO is introduced into the gas stream. It is also observed that the humidity does not affect the N₂O production. Similarly to dry condition, as reported in Fig. 11, without NO in the gas stream and with 5% humidity, the main species identified by FTIR are ozone, N₂O and CO₂ and neither HNO₃ nor N₂O₅ is observed downstream the plasma reactor. However, when NO is introduced into the gas stream, in addition to CO₂ and N₂O, NO₂, HNO₃ and nitrous acid (HNO₂) are identified as the main products in the gas phase. The HNO₂ is produced by the reaction between NO and OH radicals produced by the plasma discharge.

The NO conversion, NO₂ selectivity and HNO₂ estimated concentration, as a function of pulse energy, are shown in Fig. 12. As discussed in Section 3.2.1, in FTIR spectra, the area of absorption peak between 900 and 733 cm⁻¹ has been used to follow the HNO₂ production.

Fig. 12 NO conversion, NO₂ selectivity and HNO₂ formation during plasma processing of NO (500 ppm)–O₂ (10%)–H₂O (5%)–N₂ (balance) gas stream.

In the range of energy deposition explored, the presence of humidity in the gas mixture lead to increase the NO conversion by consumption of ozone. However, the NO₂ selectivity decreased with increasing the pulse energy. For instance, at 42.5 mJ per pulse, about 72% of inlet NO are converted which is 30% more as compared to dry condition. Indeed, at the same energy deposition, the NO₂ selectivity has been decreased by 60%. As can be seen in Fig. 12, the HNO₂ concentration increases with increasing the pulse energy and, remarkably, it follows similar trend as NO conversion profile. It appears that the higher NO conversion exhibited under humid condition is owing to the HNO₂ formation. Notably, the HNO₂ concentration tend to stabilize for pulse energy higher than 30 mJ; indeed, the NO conversion steeply increases while the NO₂ selectivity steeply decreases. This result demonstrates that the reaction between NO and OH species is reached the maximum and the remaining OH species could be involved in NO₂ consumption to produce HNO₃. As reported in the experimental part, all the experiments have been performed at room temperature, therefore, it can be suggested that a major portion of HNO₃ could be adsorbed on the rector walls. Therefore, the missing NO_x can be attributed to the amount of HNO₂ and HNO₃ produced in the gas phase and/or adsorbed on the reactor walls. These findings emphasize the fact that under humid condition, rather than ozone molecules, the O and OH species significantly react with NO and NO₂ molecules and increase NO conversion, however, these species decrease the NO₂ selectivity at room temperature. On conclusion, during the process of selective conversion of NO into NO₂ the presence of humidity shows negative effect.

3.3. Ex situ NO oxidation

In light of the results discussed above, the idea of NO injection downstream the plasma reactor (indirect NO oxidation) was applied as an alternative to the standard application of corona discharge to the gas mixture. Moreover, the effective role of long-lived species such as ozone to oxidize NO into NO₂ could be investigated by overcoming the contribution of the O atoms. In addition, the energy efficiencies of the in situ and ex situ NO removal processes cold be estimated. In these experiments, plasma processing of O₂ (10%)–N₂ was mainly used to produce ozone which in turn will react with NO. As reported in Section 3.1.3, the highest amount of ozone produced by 17 pins corona reactor in O₂ (10%)–N₂ gas mixture at 1 L min⁻¹ and room temperature is about 660 ppm. However, for ex situ NO oxidation, as reported in Fig. 3 the gas flow rate was fixed at 950 mL min⁻¹ for ozone production. In this condition, approximately 645 ppm of ozone is produced. Therefore, to compare the NO removal efficiencies of in situ and ex situ processes, the input power was slightly adjusted to reach the required concentration. The inlet NO concentration was fixed at 500 ppm. The oxidation of NO by ozone is followed by FTIR spectra acquired at different energy deposition (i.e. different ozone concentrations) and reported in Fig. 13.

Fig. 13 Typical FTIR spectra acquired after injection of 500 ppm of NO downstream plasma reactor as a function of ozone concentration (the plasma was ignited in O₂ (10%)–N₂).

After injection of NO the oxidation into NO₂ takes place. As can be seen in Fig. 13, the increasing of ozone inlet concentration lead to the gradual removal of NO from the outlet gas and converted to NO₂, HNO₂, HNO₃ and N₂O₅ species. The ozone concentration up to 243 ppm, only NO₂ is observed as an oxidized product as mentioned in eqn (12). Interestingly, when ozone concentration increased to 660 ppm, neither NO nor NO₂ are observed, indicating that NO are oxidized into NO₂, which in turn could be oxidized to NO₃ and finally into HNO₃ and N₂O₅ as reported in the literature.²¹ The observation of nitric acid (HNO₃) in the gas phase, for the highest ozone concentration, is a strong signature of the reaction of N₂O₅ with water vapor desorbed from the reactor walls. One can note that, with 660 ppm of ozone concentration, a small amount of ozone (≈65 ppm) is measured at the outlet of the system showing that about 90% of ozone are consumed in the reactions.

Fig. 14 shows the NO concentration and NO₂ selectivity measured in ex situ experiments as a function of the ozone inlet concentration. It is evidenced that the NO conversion increased with increasing the ozone inlet concentration. The NO conversion rate is greatly improved compared to those obtained during direct plasma NO oxidation (in situ experiments) as discussed in Section 3.2.1. As an example, for 17 pins reactor at an energy deposition of 42.5 mJ per pulse corresponding to an ozone concentration of about 660 ppm, only 42% of NO conversion is reached in the in situ experiments whereas, in the ex situ process a complete NO conversion is achieved. These results emphasize the importance of the ozone as an oxidizing agent at room temperature as compared to O and OH active species produced by in situ corona discharge. These results are not consistent with the work of Jõgi et al.²¹ who investigated the direct and indirect plasma NO oxidation using DBD reactor at high flow rate (50 L min⁻¹). These authors have reported that, at 25 °C, both direct (in situ) and indirect (ex situ) methods have shown similar results regarding NO conversion and NO_x production.

Fig. 14 Indirect plasma NO oxidation by ozone: inlet NO concentration 500 ppm, 17 pins corona reactor, room temperature.

As can be seen in Fig. 14, with about 243 ppm of ozone around 47% NO conversion is reached. Which is approximately 7% lower than the ozone inlet concentration. It is evidenced that, for all the investigated ozone inlet concentration, the O₃/NO_consumed ratio is about 1.2, which is very close to the value (1.4) reported in the literature.²¹ This finding emphasizes the fact that not all the O atoms produced by ozone decomposition react with NO molecules in the gas phase.

About 100% NO₂ selectivity is obtained for ozone inlet concentration in the range 40–240 ppm. Remarkably, when ozone concentration increased to 440 ppm, the NO₂ selectivity decreased to 84%. These results demonstrate that, when ozone concentration exceeds about 90% of the NO inlet concentration, the secondary oxidation reaction of NO₂ into NO₃ takes place. Subsequently the NO₂ selectivity decreases owing to the production of N₂O₅ by the reaction between NO₂ and NO₃ molecules at room temperature. In addition, when ozone concentration exceeds the NO concentration, NO and NO₂ are completely converted into N₂O₅ and/or HNO₃ at room temperature. From these results, it can be suggested that in ex situ plasma NO oxidation into NO₂ process the ozone inlet concentration must be lower than 50% of the NO inlet concentration.

3.4. Energy efficiency

To highlight the efficiencies of the in situ and ex situ processes under similar operating conditions, the energy efficiencies have been estimated using eqn (3) and compared in Fig. 15.

Fig. 15 Evolution of energy efficiencies of ozone production, in situ and ex situ NO removal under dry condition at room temperature. Ozone was produced without NO in the gas stream under similar operating conditions.

The in situ NO removal efficiency increases with increasing the pulse energy till 30 mJ and then it decreases with increasing the energy. This indicates the negative effect of injected energy on in situ NO removal by corona discharge. The NO removal energy efficiency obtained in this study, for 42% NO conversion, is about 6 times lower than the value reported by Takaki et al.¹⁶ using pin-to-plane DBD reactor with 200 ppm NO inlet concentration with 5 L min⁻¹ flow rate. However, the NO removal efficiency could be improved by optimizing the experimental conditions like flow rate and inter electrode distance. Similarly to in situ process, the ex situ NO removal process increases with increasing the injected energy. Interestingly, when the ozone concentration exceed the NO concentration, as reported in Fig. 14, the energy efficiency decreases. As can be seen in Fig. 15, at 42.5 mJ, the ex situ process shows about 2.2 times more NO removal energy efficiency than the in situ process. This result shows the advantages of ex situ plasma process for NO removal. Moreover, it can be suggested that the added NO to O₂–N₂ mixture, during in situ plasma process, significantly decreases the ozone production efficiency and thus decreases the NO removal energy efficiency at constant operating conditions. More optimization of our reactor and operating conditions is necessary to improve energy efficiency.

4. Conclusions

In this study the oxidation of NO into NO₂, in NO (500 ppm)–O₂ (10%)–N₂ (balance) mixture has been studied at room temperature under dry and humid conditions using multi pin-to-plane sub micro second corona discharge reactors. The influence of corona discharge morphology, i.e. the shape of the streamer discharge and the volume occupied by the plasma channels on NO oxidation is investigated. The decrease of the inter pins distance or increase of the number pins has significantly influenced the corona discharge characteristics, consumed more energy and decreased the NO conversion at given operating conditions.

It is observed that the ozone and N₂O production mechanisms are interconnected. It is evidenced that the added NO into the O₂–N₂ mixture is remarkably influenced the ozone production pathways and decreased the ozone concentration. Nevertheless, it does not affect the N₂O production pathways. The presence of humidity in the gas stream have remarkably improved the NO oxidation, however, the NO₂ selectivity decreased owing to the formation of HNO₂ and/or HNO₃.

For the range of pulse energy studied the indirect (ex situ) plasma oxidation has shown better NO conversion than the direct (in situ) plasma oxidation. Nevertheless, indirect method has shown low NO₂ selectivity owing to the secondary oxidation of NO₂ into NO₃ and further conversion into N₂O₅ and HNO₃. Indeed, the secondary oxidation of NO₂ to NO₃ is dependent on the ozone inlet concentration. In the future, the influence of hydrocarbons, as a reductant, on NO oxidation will be studied.

Acknowledgements

Experiments have been performed in the framework of ANR REMOVAL project. The authors greatly acknowledge the ANR for the financial support and the project collaborators (LAPLACE and SUPELEC) for the fruitful discussions.

References

1. M. Shelef, Chem. Rev., 1995, 95, 209.
2. J. M. Hao, H. Z. Tian and Y. Q. Lu, Environ. Sci. Technol., 2002, 36, 552.
3. D. S. Zhang, L. Zhang, C. Fang, R. H. Gao, Y. L. Qian, L. Y. Shi and J. P. Zhang, RSC Adv., 2013, 3, 8811.
4. M. Koebel, G. Madia and M. Elsener, Catal. Today, 2002, 73, 239.
5. P. Granger and V. I. Parvulescu, Chem. Rev., 2011, 111, 3155.
6. A. Khacef, J. M. Cormier and J. M. Pouvesle, Eur. Phys. J.: Appl. Phys., 2006, 33, 195.
7. G. Djéga-Mariadassou, F. Baudin, A. Khacef and P. Da Costa, NO_x Abatement by Plasma Catalysis, in Plasma Chemistry and Catalysis in Gases and Liquids, WILEY VCH, 2012, vol. 89.
8. F. Wang, X. Tang, H. Yi, K. Li, J. Wang and C. Wang, RSC Adv., 2014, 4, 8502.
9. B. M. Penetrante, R. M. Brusasco, B. T. Merritt and G. E. Vogtlin, Pure Appl. Chem., 1999, 71, 1829.
10. H. Wang, Q. Q. Yu, T. Liu, L. P. Xiao and X. M. Zheng, RSC Adv., 2012, 2, 5094.
11. X. Tang, F. Gao, J. Wan, H. Yi and S. Zhao, RSC Adv., 2014, 4, 5847.
12. E. M. V. Veldhulzen, W. R. Rutgers and V. A. Bityurin, Plasma Chem. Plasma Process., 1996, 16, 227.
13. A. Khacef, J. M. Cormier and J. M. Pouvesle, J. Phys. D: Appl. Phys., 2002, 35, 1491.
14. F. J. C. M. Beckers, W. F. L. M. Hoeben, A. J. M. Pemen and E. J. M. Van Heesch, J. Phys. D: Appl. Phys., 2013, 46, 295201.
15. X. Hu, J. J. Zhang, S. Mukhnahallipatna, J. Hamann, M. J. Biggsa and P. Agarwal, Fuel, 2003, 82, 1675.
16. K. Takaki, M. Shimizu, S. Mukaigaw and T. Fujiwara, IEEE Trans. Plasma Sci., 2004, 32, 1.
17. U. R. Pillai and E. S. Demessie, Chem. Commun., 2005, 17, 2256.
18. U. Kogelschatz, B. Eliasson and M. Hirth, Ozone: Sci. Eng., 1988, 10, 367.
19. I. Orlandini and U. Riedel, J. Phys. D: Appl. Phys., 2000, 33, 2467.
20. Y. S. Mok and I. S. Nam, Chem. Eng. Technol., 1999, 22, 6.
21. I. Jõgi, E. Stamate, C. Irimiea, M. Schmidt, R. Brandenburg, M. Hołub, M. Bonisławski, T. Jakubowski, M. L. Kääriäinen and D. C. Cameron, Fuel, 2015, 144, 137.
22. L. Sivachandiran, F. Thevenet, P. Gravejat and A. Rousseau, Chem. Eng. J., 2013, 214, 17.
23. H. H. Kim, K. Takashima, S. Kastura and A. Mizuno, J. Phys. D: Appl. Phys., 2001, 34, 604.
24. Y. H. Lee and G. Y. Yeom, J. Korean Phys. Soc., 2005, 47, 74.
25. A. Mraihi, N. Merbahi, M. Yousfi, A. Abahazem and O. Eichwald, Plasma Sources Sci. Technol., 2011, 20, 065002.
26. E. M. Veldhuizen and W. R. Rutgers, J. Phys. D: Appl. Phys., 2002, 35, 2169.
27. M. Meziane, O. Eichwald, J. P. Sarrette, O. Ducasse and M. Yousfi, J. Appl. Phys., 2013, 113, 153302.
28. B. M. Penetrante, R. M. Brusasco, B. T. Merritt, W. J. Pitz, G. E. Vogtlin, M. C. Kung, H. H. Kung, C. Z. Wan and K. E. Voss, SAE paper number 982508. In Plasma Exhaust After Treatment, SAE/SP-1395, 1998, p. 57.
29. J. Chen and J. H. Davidson, Plasma Chem. Plasma Process., 2002, 22, 495.
30. NIST: http://www.kinetics.nist.gov/kinetics/index.jsp.
31. J. T. Herron, J. Phys. Chem. Ref. Data, 1999, 28, 1453.
32. R. Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, J. A. Kerr, M. J. Rossi and J. Troe, J. Phys. Chem. Ref. Data, 1997, 26, 521.
33. M. P. Iannuzzi, J. B. Jeffries and F. Kaufman, Chem. Phys. Lett., 1982, 87, 570.
34. M. Simek and M. Clupek, J. Phys. D: Appl. Phys., 2002, 35, 1171.
35. S. W. Sharpe, T. J. Johnson, R. L. Sams, P. M. Chu, G. C. Rhoderick and P. A. Johnson, Appl. Spectrosc., 2004, 58, 1452.
36. S. I. Yoon, S. Heo, S. Song and Y. J. Kim, Analyst, 2010, 135, 1327.
37. M. A. Malik, J. F. Kolb, Y. Sun and K. H. Schoenbach, J. Hazard. Mater., 2011, 197, 220.
38. H. H. Kim, G. Prieto, K. Takashima, S. Katsura and A. Mizuno, J. Electrost., 2002, 55, 25.
39. A. Khacef and J. M. Cormier, J. Phys. D: Appl. Phys., 2006, 39, 1078.
40. G. S. Forbes and L. J. Heidt, J. Am. Chem. Soc., 1934, 56, 1671.

---