I. INTRODUCTION

Non-thermal plasma reactors as a possible solution to clean the gas exhaust of future vehicle have generated significant interest. In labs experiments already demonstrated that non-thermal plasma discharges can overcame some practical problems such as NOx removal [1, 2]. NOx remediation of gas exhaust via non-thermal plasma reactor is divided into dry and wet way. Several studies have shown that the wet type reactor allows obtain significant results for the gas exhaust cleaning compared to the dry reactor [3-6]. The underlying process is the NO2 dissolution into water and its conversion to nitrate (NO3

-) and nitrite (NO2-) ions.

A large variety of plasma reactors employed for NOx remediation, including dielectric barrier discharge (DBD), corona discharge and others, is based on a volume discharge. Alternatively to the volume discharge, one way to treat the gas exhaust consists of using a surface discharge. Still, this type of discharge is under employed in the field of the pollution control [7, 8], whereas it knows an important development in other areas of research, e.g. the aerodynamic field [9]. This paper presents an experimental study on NOx removal using a non-thermal surface plasma discharge powered by a modulated voltage. The underlying idea is the use of an unsteady actuation, obtained by a switching on and off the plasma alternately, in order to reduce the electrical power consumption of the cleaning device. In a first part, the effect of the amplitude modulation (AM) on NOx removal is analyzed. Then, the burst modulation (BM) with a fixed duty cycle is studied, as well as the variation of the duty cycle of the burst modulation. All experiments are characterized by measuring the power injected into the gas, and the evolution of removal rate of NO and NOx gases via FTIR measurements. In

addition, they are carried out at low flow rate (1 L/min), with 100 ppm of initial content of NO.

II. EXPERIMENTAL PART

A. Non-thermal surface plasma reactor Fig. 1 displays a schematic side-view of the non-

thermal surface plasma reactor used in this study. It is composed of two parts. The first part consists of a test chamber with two openings that allow the inlet and the exhaust of gas flow. This test chamber made of PP plate has a rectangular cross-section of 100 mm × 30 mm for 150 mm long (inner dimensions). Moreover, each plate has a thickness of 2 mm. The second part of the plasma reactor corresponds to the surface discharge placed on the top plate of the test chamber. The whole forms the non-thermal surface plasma reactors.

The surface discharge consists of two electrodes flush mounted on each side of a dielectric barrier, plus two counter-electrodes placed on the topside of the insulating wall. These counter-electrodes are separated relatively to the AC electrode by an air gap of 25 mm, named the SD gap. Each electrode is made of 50-μm-thick aluminum tape strip whose ends are oval in order to reduce the edge effects. Air exposed electrodes are 60 mm long (in spanwise direction) for 10 mm in width. The upper one is 30 mm wide for 60 mm long. The insulating barrier used is a PMMA) plate of 140 mm × 100 mm and 3 mm thick.

The plasma discharge is powered with AC high-voltage (see Fig. 1). The AC high-voltage is obtained with the help of a transformer supplied by a power amplifier (NF Corporation, model 4510). The transformer may supply a maximum peak voltage of 20 kV at driving frequencies up to 5 kHz. The three other electrodes are grounded. In this case, the plasma device operates in the DBD mode [10].

The wet condition is obtained by adding a sodium sulfite (Na2SO3) solution, of concentration 1 mol/L, inside the reactor (see Fig. 1). This solution is used as

NOx Removal Using a Non-thermal Surface Plasma Discharge Powered by a Modulated Voltage

J. Jolibois1, K. Takashima1, G. Touchard1,2, and A. Mizuno1

1Department of Environmental and Life Sciences, Toyohashi University of Technology, Japan 2Pprime Institute, CNRS, University of Poitiers-ENSMA, France

Abstract—This paper deals with the NOx removal with the help of a non-thermal surface plasma discharge powered by a modulated high voltage waveform. The underlying idea is the use of an unsteady actuation in order to reduce the electrical power consumption of the cleaning device. To achieve this, the effect of amplitude and burst modulation on NOx remediation is analyzed. These experiments show an important decrease of the electric power in both cases. In addition, NOx removal is enhanced with the burst modulation (BM) with a duty cycle ranging between 0.3-0.6. So the energy efficiency is 2.5 time higher than the standard case and reaches 4.5 g/kWh. In the case of amplitude modulation (AM), the modulated voltage has no effect on the gas exhaust treatment.

Keywords—Non-thermal plasma, surface discharge, gas treatment, pollution control, NOx removal

Corresponding author: Jerome Jolibois e-mail address: Jerome.jolibois@gmail.com

74 International Journal of Plasma Environmental Science & Technology, Vol. 6, No. 1, MARCH 2012

NOx gas absorbent. Indeed as reported in literature [3-6], a plasma reactor operating in wet conditions can improve the efficiency of NOx removal by dissolving NO2 into the liquid as NO2

- and NO3- ions. However, the

continuous absorption of nitrogen oxides induces saturation and acidification of the liquid, resulting an inhibition of further absorption. The adding of sodium sulfite allows reduce the nitrite and nitrate ion to N2, thus the gas absorption is facilitated. Moreover, the Na2SO3 solution was replaced after each set of experiments. For this study, 150 mL of Na2SO3 solution is used. Thus, due to the size of the reactor, the height of 150 mL solution is 10 mm. In addition, taking into account the height of the reactor (30 mm) and the thickness of the PMMA plate (3 mm), the distance between surface plasma discharge and water is 17 mm. B. Electrical measurements One of key parameters to evaluate a pollution control system is the energy consumption. It is determined from the simultaneous measurements of the discharge current and the applied voltage. The current is deduced from the voltage across a shunt resistor (100 Ω) connected in series between the upper electrode and the ground. The applied voltage is measured by using an HV probe (Tektronix, model P6015A). Each electrical signal is monitored and recorded via a fast digital oscilloscope (Tektronix, model DPO 2024). From both voltage and current measurements, the electrical power consumption can be calculated as follows:

P 1

nTv(t)i(t)dt

t0

nT

(1)

where v(t) and i(t) are the measured voltage and current versus time, respectively, T is the waveform period, and n is the number of periods. In order to compare the efficiency of the gas exhaust treatment with the different configurations and power consumed, the energy density Ed is generally used:

Ed P 60

Q (2)

where Ed, P and Q are the energy density (J/L), power consumed (W) and gas flow rate (L/min), respectively. C. Gas concentration measurements Fig. 2 shows the schematic view of the experimental setup. The entire device consists of the non-thermal surface plasma reactor, the gas feeding system, the electrical part, and the gas analyzing system. The polluted gas is simulated. It consists of a mixture of nitrogen oxide (NO, N2 base) and dry air, kept in room temperature and atmospheric pressure. The composition of the gas test is performed through the mass flow controller. It also helps regulate the flow rate of polluted gas, whose the value is controlled through the flow meter. The treated gas is sampled at the exit of the plasma reactor, and concentrations of NO, NO2, and NOx are measured by FTIR gas analyzer (BEST SOKKI, model

Fig. 1. Schematic side-view of the wet type plasma reactor used.

Fig. 2. Schematic view of the experimental setup.

Jolibois et al. 75

SESAM 3-N). This model of FTIR is intended to be used for the monitoring of the diesel gas with a minimum accuracy of few ppm. Cleaning experiments are carried out as follows: The evolution in concentrations of the gas exhaust is recorded for 15 min. During this recording, the discharge is ignited on twice for 3 min, as shown in Fig. 3. Each ignition is spaced about 3 min, allowing the initial concentration to be reached. From these results, we can determine the variation of mean concentrations of NO, NOx and NO2, so to calculate the removal efficiency of the plasma reactor. The removal efficiency is given as:

R NOx i

NOx NOx i

100 (3)

where [NOx]i and [NOx] are the initial and final concentrations of NOx (ppm), respectively. The mean concentration is determined by taking into account the rate of pollutants when the plasma is switched on. So, this value is underestimated, resulting in a reduction of the value of the effectiveness of treatment. The energy efficiency, η (g/kWh), is defined as the removal amount of NO per unit power of 1 kWh, and is calculated as follows:

Q 60 NO 106

R

100

30

22.4

1

W (4)

where Q (L/min) is flow rate of the gas containing NO, [NO] is NO concentration (ppm). Molecular weight of NO is 30 g/22.4 L for 1 mol, and ΔR (%) is NO removal rate, W (kWh) is the energy consumed by the plasma.

III. EXPETIMENTAL RESULTS

To highlight the effect of amplitude and burst modulations on the NOx removal process, the AC high-voltage is modulated by a modulating sine wave (fAM) in the case of AM mode (see Fig. 4a). For the BM mode, the AC high-voltage is modulated by a Rectangular function. This modulation results in an AC sine wave alternatively switched on and off: the discharge is assured by the carrier signal whereas the modulation frequency fBM drives its ignition (see Fig. 4b). In both cases, the present carrier frequency (fAC) corresponds to the frequency driving the plasma in steady actuation.

In addition, all experiments are carried out at low flow rate (1 L/min), with 100 ppm of initial content of NO and 150 mL of Na2SO3 solution (1 mol/L). With these conditions, the gas residence time is about 15 sec. A. Effect of amplitude modulation on NOx removal

In this section, we focus on the effect of amplitude

modulation on the gas exhaust treatment. To achieve this, we applied a voltage modulated by a modulating sine wave (fAM = 100 Hz) with a modulation depth fixed at 80%, across terminals of the plasma discharge. First, we study the evolution of the electrical power consumption as a function of the peak voltage value (see Fig. 5). Here, the amplitude of carrier and the peak voltage of the standard actuation (i.e. steady actuation) are included between 6 to 13 kV with fAC = 1 kHz.

It appears that the power consumed by the plasma discharge is significantly reduced when we apply the modulated voltage compared with the standard case. The

Fig. 3. Time evolution of the NO concentration as a function of the

applied voltage.

Fig. 4. Modulated voltage in AM (a) and BM (b) mode.

76 International Journal of Plasma Environmental Science & Technology, Vol. 6, No. 1, MARCH 2012

power gap reaches nearly 60% with V = 13 kV peak. This explains by the fact that the applied voltage across terminals of the discharge is not constant. The voltage passes through a minimum and a maximum depending on the variation of the modulating sine wave. When the modulating sine wave tends toward its minimum, the value of applied voltage decreases, resulting a decrease of the average discharge current. In addition, the modulated voltage becomes lower than the onset voltage so the plasma is quenched during a fraction of actuation time. Therefore, the power consumption is reduced.

Fig. 6 presents the removal efficiencies of NO and NOx as a function of the energy density with a standard and modulated actuation. It appears clearly that the amplitude modulation used as type of actuation does not improve the treatment of NOx. B. Burst modulation with a fixed duty cycle In this second set of experiments, we investigate the effect of the burst modulation with a duty cycle of 50%. The modulation frequency equal to 100 Hz and the carrier frequency is set at 1 kHz. Fig. 7 displays the power consumed with the peak voltage value (6 ≤ V ≤ 13 kV) for the sinusoidal and modulated waveforms. As

expected, the burst modulation with a 50% duty cycle allows reduce the power consumption of the discharge. This decrease is not exactly 50%, but is between 45 and 55% because the electrical power varies from one cycle to another. Moreover, we can also see that the burst modulation has no significant effect on the evolution of the power consumption. It appears that its variation is correctly fitted by the expression (i.e. a quadratic function, where P ≈ ΔV2) used by Pons et al. [11]. The evolution of the NOx removal efficiency as a function of the energy density is shown in Fig. 8. When the discharge operates with the burst modulation mode, the injected energy into the polluted gas is lower than the steady actuation, resulting in a reduction of the operating range of the cleaning device. With the sinusoidal waveform, the operating range is between 10 and 135 J/L, against about 5 and 70 J/L with the burst modulation. However, we can remark that at a given energy density, stressing the discharge with the modulated voltage (VBM) seems given higher values of removal efficiency as compared with the sinusoidal AC voltage. The efficiency gain reaches 10% on the treatment of NO, whereas it is only 5% on NOx removal. This result could be explained by improved contact between the polluted gas and plasma but also by an increase of active species produced.

Fig. 5. Electrical power variation versus the applied voltage across

terminals of discharge (: sine wave, : AM wave).

Fig. 6. Removal efficiency versus the energy density (: NO, : NOx with sine wave and : NO, : NOx with AM wave).

Fig. 7. Electrical power variation versus the applied voltage across terminals of discharge (: sine wave, : BM wave).

Fig. 8. Removal efficiency versus the energy density (: NO, : NOx with sine wave and : NO, : NOx with BM wave).

Jolibois et al. 77

On one hand, it appears that two factors contribute to enhance the contact. The first one is due to the voltage difference applied across the terminals of cleaning device for the same energy density injected. For example, at 30 J/L, the maximum peak voltage is 8 kV with the sine wave whereas the maximum of the modulated voltage is 10 kV. However, Benard et al. [12] have shown that the expansion of the plasma is not affected by the burst modulation. Consequently, the size of plasma "seen" by the polluted gas is higher with the modulated voltage, because the extension of plasma over the dielectric surface depends on the value of applied voltage [13, 14]. The last factor that helps to improve the contact between the polluted gas and the plasma seems to be related to eddies generated through the electric wind induced by the discharge. These vortices cover a larger region above the insulating layer [15], resulting in an increase of contact efficiency. In addition, the mixing allows a contribution of gas to be treated evacuate the clean gas by the means of eddies generated downstream the AC electrode [15]. On the other hand, the pulse repetition rate can affect the removal process via an increase of active species production including O3. However, this effect was demonstrated with a fast-rising pulsed voltage or with a high-frequency discharge (see in [16, 17]) where the rise time is ≤ 10 μs and with a high repetition rate (≈ 10 kHz). Here, the sinusoidal waveform is at low frequency (1 kHz), so the rise time is much slower (≈ 250 μs). And the repetition rate is 100 Hz. Consequently, the effect of these two parameters maybe not significant. Only the measure of ozone rate (not carried out) could allow us to highlight their contributions on the removal process. Fig. 9 shows the energy efficiency as a function of the energy density for NO treatment. For energy density ranging between 5 to 60 J/L, it appears that the energy efficiency is higher by using a modulated voltage. With a low energy density (Ed < 20 J/L), the energy efficiency shows a fast rising and it is 2.5 time higher than the steady actuation. The highest energy efficiency reaches 4.5 g/kWh and is obtained at about 17.5 J/L. Then, the energy efficiency follows a gradual decrease. This result is interesting because it confirms that the use of a

modulated voltage allows enhance the effectiveness of the treatment of polluted gas. Beyond 60 J/L, we can see that the energy efficiency is similar with the steady and unsteady actuations. In this case, the energy efficiency is equal about to 3 g/kWh. Moreover, this value corresponds to the maximum of energy efficiency obtained with the stationary case. C. Variation of the duty cycle of the burst modulation In this section, we study the variation of the duty cycle of the burst modulation. The modulation frequency equal to 100 Hz, the carrier frequency is set at 1 kHz and the duty cycle is between 0.1 and 0.9. The influence of the variation of the duty cycle on the electrical power consumption of the plasma discharge is highlighted in Fig. 10. We can observe that the electric power increases linearly with the duty cycle of the applied high voltage (here, V = 11 kV). This can be explained by the fact that the power consumption seems to increase as:

PBM P dc (5)

with P = k(V-Von)

n (2 ≤ n ≤ 3) and dc is the duty cycle (note: k, Von and n do not depend on dc). When the applied voltage is fixed, the present expression gives a linear evolution of the power in this case. In order to verify this result, Table 1 presents the determined power from the voltage and discharge current measurements (see (1)), and the computed power from the previous expression (5) with the duty cycle. It clearly appears that the power values in both cases are nearly

Fig. 9. NO removal energy efficiency versus the energy density.

Fig. 10. Electrical power variation versus duty cycle.

TABLE I POWER CONSUMPTION DETERMINED FROM ELECTRICAL

MEASUREMENTS (P1) AND COMPUTED (PBM) WITH DIFFERENT DUTY CYCLE (dc).

dc 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

P1

(W)0.12 0.3 0.47 0.63 0.85 0.93 1.16 1.22 1.41 1.56

PBM

(W)0.16 0.31 0.47 0.62 0.78 0.94 1.09 1.25 1.4 1.56

78 International Journal of Plasma Environmental Science & Technology, Vol. 6, No. 1, MARCH 2012

similar. In addition, the computed values of the electrical power follow a linear regression. If we compare the removal efficiency of NOx versus the energy density for the steady and the unsteady actuation (see Fig. 11), one can see that the treatment of NO seems slightly enhanced. The maximum gain reaches about 5% for a duty cycle ranging between 0.3 and 0.6. This means that the appropriate energy density in order to obtain the best cleaning efficiency is deposited in the plasma discharge for these duty cycles. A similar trend has been obtained in [23, 24] with the variation of the duty cycle of one cycle sinusoidal power source on the removal rate of NO. However, comparisons between are difficult because the parameters that change the removal efficiency of NO such as voltage, frequency and acting type (direct or indirect) are not the same. For the remediation of NOx, the modulated voltage has no significant effect up to 50 J/L. Beyond 50 J/L, the NOx removal seems less effective than in the normal case. Fig. 12 displays the energy efficiency as a function of the energy density for NO treatment. This result is interesting because it confirms the previous one, i.e. the conversion of NO to NO2 is slightly improved and for a limited range of duty cycle. From this figure, it clearly appears that the treatment of NO is increased for duty cycle values included between 0.3 and 0.6, corresponding

to an energy density ranging from 30 to 60 J/L. Indeed, energy efficiency increased from 2.8 g/kWh on average with the steady actuation to 3.25 g/kWh on average with the unsteady case. The weak gain in energy efficiency can be explained by the value of the applied high-voltage (here, V = 11 kV). This value of voltage corresponds to an energy density about 55 J/L and approaches the limit of efficiency of the modulated voltage (i.e. 60 J/L). So the use of lower voltage value, typically 8 or 9 kV (see Fig. 10), could alter the efficiency obtained with the variation of duty cycle.

IV. CONCLUSION

In this paper, we investigated the NOx removal

using a modulated voltage. First, the amplitude modulation (AM) was studied to characterize its effects on the gas exhaust cleaning. Then, influence of the burst modulation (BM) was investigated, especially the BM with a fixed duty cycle and the variation of the duty cycle. The main results are as follows:

(1) The use of a modulated voltage has a significant effect on the power consumed, i.e. a decrease of it.

(2) With the burst modulation at a fixed duty cycle, the evolution of the power consumption versus applied voltage follows the behavioral law of Pons et al. [18].

(3) In the case of the variation of the duty cycle, the power consumed follows a linear regression. In this way, the electrical power can be controlled by the duty cycle.

(4) The amplitude modulation has no effect on the NOx remediation.

(5) On the contrary, the burst modulation allows enhance the NO conversion with a duty cycle ranging between 0.3 and 0.6. The maximum gain reaches up to 10% with a duty cycle of 50%.

(6) For a low value of energy density, the energy efficiency is 2.5 time higher with the modulated voltage. The maximum energy efficiency reaches 4.5 g/kWh.

The main effect of the modulated voltage seems to

be due to a combination of two effects. The first one is an increased production of active species. And the second one is an improvement of the contact between the plasma and the polluted gas. However, the interaction between the plasma and the polluted gas is not yet totally distinguished. So, further investigations using optical measurements as well as O3 measurements are necessary in order to complete the present study.

REFERENCES [1] C. R. McLarnon and V. K. Mathur, "Nitrogen oxide

decomposition by barrier discharge," Industrial & Engineering Chemistry Research, vol. 39, pp. 2779-2787, 2000.

[2] A. Mizuno, "Industrial applications of atmospheric non-thermal plasma in environmental remediation," Plasma Physics and Controlled Fusion, vol. 49, pp. A1-A15, 2007.

Fig. 11. Electrical power variation versus duty cycle.

Fig. 12. NO removal energy efficiency versus the energy density.

Jolibois et al. 79

[3] A. Mizuno, K. Shimizu, T. Matsuoka, and S. Furuta, "Reactive absorption of NOx using wet discharge plasma reactor," IEEE Transactions on Industry Applications, vol. 31, pp. 1463-1468, 1995.

[4] T. Fujii, Y. Aoki, N. Yoshioka, and M. Rea, "Removal of NOx by DC corona reactor with water," Journal of Electrostatics, vol. 51, pp. 8-14, 2001..

[5] H. Lin, X. Gao, Z. Y. Luo, K. F. Cen, M. X. Pei, and Z. Huang, "Removal of NOx from wet flue gas by corona discharge," Fuel, vol. 83, pp. 1251-1255, 2004..

[6] S. M. Thagard, Y. Kinoshita, H. Ikeda, K. Takashima, S. Katsura, and A. Mizuno, "NO3

- Reduction for NOx Removal Using Wet-Type Plasma Reactor," IEEE Transactions on Industry Applications, vol. 46, pp. 2165-2171, 2010.

[7] N. Harada, T. Moriya, T. Matsuyama, H. Yamamoto, and S. Hosokawa, "A novel design of electrodes system for gas treatment integrating ceramic filter and SPCP (surface corona discharge induced plasma chemical process) method," Journal of Electrostatics, vol. 65, pp. 37-42, 2007.

[8] M. S. Cha, Y. H. Song, J.-O. Lee, and S. J. Kim, "NOx and Soot Reduction Using Dielectric Barrier Discharge and NH3 Selective Catalytic Reduction in Diesel Exhaust," International Journal of Plasma Environmental Science & Technology, vol. 1, pp. 28-33, 2007.

[9] E. Moreau, "Airflow control by non-thermal plasma actuators," Journal of Physics D-Applied Physics, vol. 40, pp. 605-636, 2007.

[10] K. Takashima, N. Zouzou, E. Moreau, A. Mizuno, and G. Touchard, "Generation of Extended Surface Barrier Discharge on Dielectric Surface – Electrical properties," International Journal of Plasma Environmental Science & Technology, vol. 1, pp. 14-20, 2007.

[11] J. Pons, E. Moreau, and G. Touchard, "Asymmetric surface dielectric barrier discharge in air at atmospheric pressure: electrical properties and induced airflow characteristics," Journal of Physics D-Applied Physics, vol. 38, pp. 3635-3642, 2005.

[12] N. Benard and E. Moreau, "Electric wind produced by a surface plasma discharge energized by a burst modulated high voltage," in Proc. 29th International Conference on Phenomena in Ionized Gases, Cancun, Mexico, PA16-11, 2009.

[13] V. I. Gibalov and G. J. Pietsch, "The development of dielectric barrier discharges in gas gaps and on surfaces," Journal of Physics D-Applied Physics, vol. 33, pp. 2618-2636, 2000.

[14] K. Allegraud, O. Guaitella, and A. Rousseau, "Spatio-temporal breakdown in surface DBDs: evidence of collective effect," Journal of Physics D-Applied Physics, vol. 40, pp. 7698-7706, 2007.

[15] J. Little, M. Nishihara, I. Adamovich, and M. Samimy, "High-lift airfoil trailing edge separation control using a single dielectric barrier discharge plasma actuator," Experiments in Fluids, vol. 48, pp. 521-537, 2010.

[16] M. Nishida, K. Yukimura, S. Kambara, and T. Maruyama, "NOx removal using ammonia radicals prepared by intermittent dielectric barrier discharge at atmospheric pressure," Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers, vol. 40, pp. 1114-1117, 2001.

[17] K. Yukimura, K. Kawamura, T. Hiramatsu, H. Murakami, S. Kambara, H. Moritomi, and T. Yamashita, "Efficient decomposition of NO by ammonia radical-injection method using an intermittent dielectric barrier discharge," Thin Solid Films, vol. 515, pp. 4278-4282, 2007.

80 International Journal of Plasma Environmental Science & Technology, Vol. 6, No. 1, MARCH 2012