Effects of Electrode Structure and Electron Energy on Abatement of NO in Dielectric Barrier Discharge Reactor

Featured Application: In this special issue, we only investigated the compare of two type DBD reactors (CC-DBD and TM-DBD). However, further studies are required about the mathematic relation between electrode structure parameters (electrode length, discharge gap and so on) and electron energy. It is important for industrial application of DBD to understand the effect of O 2 content for NO x removal because O 2 usually exist in industrial waste gas. The research for electron energy of DBD by using VSim is an effective way, while relevant studies are rare. DBD technologies are paid considerable attention by many scientists and I think the research results in the paper would provide helpful information for the future investigation in the relevant research ﬁled. Abstract: Electrode structure and electron energy effects on NO abatement were studied in two different structure DBD reactors. Final product analysis of NO abatement in coaxial cylinder dielectric barrier discharge (CC-DBD) and tubular multilayer dielectric barrier discharge (TM-DBD) reactors indicated that the electrode structure of TM-DBD was better under low O 2 concentration conditions, but the result was opposite because the new NO x was produced in TM-DBD when O 2 concentration was increasing. In addition, results of particle-in-cell with Monte Carlo collisions (PIC-MCC) simulation manifested that the largest and the average electron energy were 12.09 eV and 3.35 eV in TM-DBD reactor, respectively, while they were 5.25 eV and 2.96 eV in CC-DBD reactor, respectively. CC-DBD electrode structures are preferable for better NO abatement and no new NO x under oxygen-containing condition.


Introduction
NO x , one of the primary air pollutants, can cause various effects such as photochemical smog, secondary aerosols, and tropospheric ozone [1][2][3][4][5], although NO x plays a positive role in biology [6]. Therefore, abatement of NO x is a subject of great concern around the world. Nonthermal plasma (NTP) can destroy NO x in the gas stream with relatively low energy consumption, ease of operation, and compact system. Dielectric barrier discharge (DBD), one of the most promising NTP generation technologies at an atmospheric pressure with highly efficient and eco-friendly advantages, has even paid attention by 'plasma medicine' [7] and particularly distinguished performance in the decomposition of NO x via conventional methods [8][9][10].
Given that NO is the main component of NO x in exhaust gas, the removal of NO is the core problem involved in the denitration field [11][12][13]. Recently, substantial research work using DBD reactors were developed for NO abatement [14][15][16][17][18][19]. Moreover, in most studies, high electron energy was considered the key factor in controlling plasma processes for NO removal [20,21]. Therefore, many researchers have paid attention to the investigation of improving the electron energy of DBD reactors to achieve higher destruction and removal efficiency (DRE) of NO [22,23]. However, high electron energy can lead to the production of new NO x by the oxidation of N generated from N 2 dissociation in the DBD reactor. Therefore, preventing new NO x production and improving the better by-product selectivity gradually becomes a great concern on the NO abatement by DBD technique [24,25]. Many researchers investigated the role of electron energy in NO abatement and studied several influencing factors such as input power, electrode gaps and dielectric relative permittivity [26][27][28][29][30][31][32][33][34][35][36][37][38][39][40]. Nevertheless, limited attention has been given to the relation between electron energy and electrode structure although the latter is considered as one of the most important factors in controlling electron energy [41,42].
Numerical simulation, an effective tool in developing the NTP theories and determining the operating parameters, has been developed in recent years [43,44]. Numerical study of the DBD operation is conventionally focused on plasma modeling approaches, namely, fluid and kinetic plasma description. Particle-in-cell with Monte Carlo collisions (PIC-MCC), a new plasma modeling approach, has been verified to predict different types of discharge and to describe the breakdown stage of discharge or plasma decay stage [45]. These modeling results can provide important information about the qualitative description of the electron energy and the force generation by DBD plasma [46]. These results have illustrated the distribution of electron energy and active particles in the atmospheric air. However, electron energy and its distribution in DBD reactors with different electrode structures for NO abatement were not discussed.
This study aims to investigate the effect of electron energy and electrode structure on the abatement of NO x and preferable by-product selectivity in two different DBD reactors. Reaction products and chemical reaction mechanism of NO abatement were performed by Fourier transform infrared (FTIR) to provide further analysis information on NO destruction and new NO x production. Then, electron energy, electron energy distribution and partial discharge of two different plasma reactors were investigated in a NO-N 2 -O 2 system using PIC-MCC simulation method. The electrical signals in DBD reactors were also analyzed to explore the discharge effect of electrode structures. Based on experimental and numerical research, the mechanism involved and preferable means of controlling the new NO x production in DBD reactor are also suggested. The results can provide useful information on DBD application for NO abatement.

Experiment
The schematic of the experimental system shown in Figure 1 was composed of a gas distribution system, a reaction system and an analysis system. Figure 2 shows the two electrode structures of plasma reactors that were designated as coaxial cylinder dielectric barrier discharge (CC-DBD) and tubular multilayer dielectric barrier discharge (TM-DBD). The CC-DBD reactor consisted of an inner high-voltage electrode (stainless steel sheet), two quartz tubes (outer tube with 20 mm diameter and 200 mm length, inner tube with 12 mm diameter and 200 mm length), and an outer electrode (aluminum foil). The quartz tubes were coaxial cylinder in shape with 4 mm gap.
The TM-DBD reactor consisted of three quartz tubes arranged in a row with 6.0 mm diameter inner electrode and 31 mm length. The stainless-steel electrodes were inserted into the quartz tubes. NO was prepared in cylinder containing very few N 2 O (1.98 × 10 −3 ppm). The NO concentration was 345 ppm and gas flow rate was fixed at 6 L/min. Mass flow controller (MFC) (SEC-4400, Horbia, Kyoto, Japan) was used to adjust the flow of NO, O 2 , N 2 , and reactant gases.  The final products of NO degradation were measured by FTIR spectroscopy (Nicolet Nexus 470, Thermo Fisher, Waltham, MA, USA) and the resolution of the FTIR is 1 cm −1 . The plasma reactor was driven by AC power supply (homemade power). The voltage was maintained at 5 kV (driven frequency of 20 kHz) and input power was 128 W measured by power meter (PF9800, Everfine, Hangzhou, China).
Destruction and removal efficiency (DRE) of NO and NO x was calculated from the equations

Numerical Simulation
The simulation is performed by PIC-MCC method using the software VSim8.0 (Boulder, CO, America, 2016) by Tech-X Corporation. The two types of DBD reactors with different electrode structures are shown in Figure 2. The geometric structure and discharge modeling as well as geometry parameters used in the simulation are also shown in Figure 2.
VSim is an efficient parallel PIC model, but few DBD simulations using VSim have been reported. Therefore, the key setting of parameters is presented. In the simulation, the electromagnetic field was calculated by 2D modeling based on Maxwell's equations, and particle collision was handled by Monte Carlo collisions. Only the collisions and movement process of electron have been taken into account in model to simplify simulation. This simulation is 2D CFDTD (conformal finite-difference time-domain) PIC code, which can run parallel with CPU cores. Those parameters used in PIC-MCC model are shown in Table 1. We mainly used the electron movement data to examine the electron energy and the effect on NO abatement by the numerical simulation in the TM-DBD and CC-DBD reactors. Other data were VISM built-in data. The entire model was based on 2D Cartesian coordinates. Table 1. Model parameters of DBD in VSim and Geometry parameters of CC-DBD and TM-DBD in VSim.

Model Parameters of DBD in VSim
Grid numbers X * Y * Z = 100 * 100 * 200 Numbers of gyrating circles 50 Numbers of microparticle 50,000 Geometry Parameters of CC-DBD and TM-DBD in VSim TM-DBD R T /mm r T /mm D T /mm 3 1.5 4 CC-DBD R C /mm r C /mm D C /mm R oC /mm 3 1.5 4 1 Figure 3 showed the DRE of NO and NO x under different O 2 concentration in the TM-DBD and CC-DBD reactors. O 2 not only competitively shares the input power, but contributes to the oxidation of NO. Therefore, the DRE of NO x decreases from 75% to 35% in CC-DBD reactor and from 87% to 20% in the TM-DBD reactor with an increase of O 2 from 0 to 15.9%. However, DRE of NO first decreases and then sharply increases to 99% in CC-DBD reactor, while the DRE of NO first decreases and then keeps little change at about 69% in the TM-DBD reactor. In addition, the results also indicated that the DRE of NO and NO x in the TM-DBD reactor was better than in the CC-DBD reactor when O 2 concentration was low, and the opposite result was seen under higher O 2 concentration. Thus, it can be seen that the electrode structure has important impact on the DRE of NO and NO x .

Final Product Analysis by FTIR
In order to get further information about NO abatement, the final products in the two reactors were identified by FTIR. Figure 4 showed the FTIR spectra of the final products for NO abatement in TM-DBD and CC-DBD reactors when N 2 was taken as the buffer gas. The results showed a decrease behavior for NO abatement in two reactors except for the peak intensity. Compared with the untreated gas, the intensity of NO absorption decreased after discharge in all reactors. The peak of N 2 O increased in CC-DBD, but it decreased in TM-DBD. Based on the facts, the final products of NO abatement included N 2 O in CC-DBD. In addition, the peak intensity of NO and N 2 O in TM-DBD correspondingly became weaker than them in CC-DBD. The results indicated that the electrode structure of TM-DBD was better for abatement of NO and NO x under low O 2 concentration. Figure 5 showed the FTIR spectra for NO abatement with 345 ppm NO under the buffer gas of 8% O 2 and 92% N 2 in the TM-DBD and CC-DBD reactors. Compared with the untreated gas shown in Figure 4, the peak intensity of NO 2 and N 2 O with the peaks at 2200 cm -1 and 1600 cm -1 all increased obviously in the two reactors. The results indicated that the final products in the TM-DBD and CC-DBD reactor included few N 2 O (about 9.9 × 10 −3 ppm) and NO 2 , respectively. According to Figure 5, there was still quite a peak intensity of NO in TM-DBD reactor while NO was almost not identified in the CC-DBD reactor; moreover, the intensity of NO 2 and N 2 O absorption in TM-DBD was higher than that in CC-DBD. The difference between TM-DBD and CC-DBD was ascribed to the conclusion that the DRE of NO as well as NO x was better in CC-DBD reactor under high O 2 concentration. To explore the difference of above experiment results, the mixture of 92% N 2 and 8% O 2 without NO was induced into the TM-DBD and CC-DBD reactors. Figure 6 showed the FTIR spectra of the final products produced in the TM-DBD and CC-DBD reactors, respectively. It can be seen that NO, NO 2 and N 2 O were identified with the peaks at around 1900, 1600 and 2200 cm -1 in TM-DBD reactor, while the final product was only O 3 with the peak at around 1000 cm -1 in the CC-DBD reactor. The different results indicated that N 2 was dissociated by electron impact dissociation reactions in TM-DBD reactor, namely, e+N 2 →e+2N and N would recombine with O and O 2 to produce new NO x in the TM-DBD reactor, which also verified the results of the final products shown in Figure 5. O 2 dissociation could occur in TM-DBD but O 3 was not detectable, because it is selectively consumed by NO. However, in the CC-DBD reactor, only the identification of O 3 production was determined and no evidence of N 2 dissociation reactions could be seen. Above all, NO abatement was performed mainly through the reduction channel in the TM-DBD and CC-DBD reactors under low O 2 concentration, and moreover, the electrode structure of TM-DBD was preferable. However, When O 2 concentration increased, O· produced by the dissociation of O 2 in reaction system rapidly enhanced and the channel of oxidation reaction was predominant. The new NO x , generated from the reaction of N+O→NO, N+O 2 →NO+O, and NO+O→NO 2 [21] cannot be ignored in the TM-DBD. The electrode structure of the CC-DBD reactor was advantageous under high oxygen conditions. The results also revealed that the electrode structure was the key role in the new NO x produced in the DBD reactors.

Numerical Simulation Result
Numerical simulation was a method of investigating the relationship between discharge parameters and electron energy. Figure 7 demonstrated the electric potential distribution of two types of electrode structures in the TM-DBD and CC-DBD reactors. Similarly, the voltage was linearly decreasing from anode to cathode in the two reactors, and the electromagnetic field was uniform distribution in the CC-DBD reactor, while that was stronger in the central discharge zone in the TM-DBD reactor with nonuniform voltage distribution. The distribution of electron velocity and electron drift in the space of the CC-DBD and TM-DBD reactors are shown in Figure 8. Once the plasma was generated and the number of micro-discharges inside the cell exceeded the threshold limit (10 macro-particles per particle), the particles were combined into larger micro-discharge [47]. Electron velocity distribution presented periodic drift as a variation of discharge voltage. According to Figure 8, electron velocity distribution was more balanced in the CC-DBD reactor. Summing up the electron velocity data of simulation, the largest electronic velocity and average electron velocity were 1.345 × 10 6 m/s and 1.014 × 10 6 m/s in the CC-DBD reactor, respectively. On the other hand, electron velocity distribution in the TM-DBD reactor was typical Poisson distribution [48] and stronger in the center area between the two electrodes. The maximum and average of electron velocity were 2.052 × 10 6 m/s and 1.081 × 10 6 m/s in the TM-DBD reactor, respectively. Considering that the motion energy of the electron is predominant in a plasma reactor, we determined the electron energy based on electron velocity distribution. The largest electron energy and average electron energy in the TM-DBD reactor were 12.09 eV and 3.35 eV, respectively. While they were 5.25 eV and 2.96 eV in the CC-DBD reactor, respectively. The bond energies for N 2 , NO and O 2 were 9.76, 6.5 and 5.12 eV, respectively [49]. Thus, O 2 could be dissociated all in TM-DBD and CC-DBD reactors in view of the largest electron energy of 12.09 eV and 5.25 eV respectively in the two reactors, but N 2 could be decomposed into N· by the electron impact dissociation reaction only in the TM-DBD reactor.

Discussion
Based on the above facts, the possible reaction channels [28] involved in two reactors under different O 2 concentration were shown in Table 2. In the TM-DBD and CC-DBD reactors, the reduction reactions were the main channels for NO abatement under low O 2 concentration, while the channel of oxidation reaction was predominant when O 2 concentration increased.
As the result, the electrode structure of TM-DBD was preferable when O 2 concentration was low, but the electrode structure of CC-DBD was better under high O 2 concentration condition. In other words, at low O 2 concentration, a higher electron energy yielded a better DRE of NO and NO x , while the opposite result was seen when O 2 concentration increased. These results verified that the electrode structure influenced the electron energy and further decided the production of new NO x in a DBD reactor.

Conclusions
The DRE of NO and NO x were obviously influenced by electrode structure. Under deficient oxygen, better DRE of NO and NO x is realized in the TM-DBD, and the electron impact dissociation mainly determined the destruction rate of NO with the conversion products of N 2 as well as N 2 O. When O 2 concentration increased, the DRE of NO x decreased, whereas the DRE of NO decreased and then increased in the two DBD reactors. Under O 2 abundant conditions, NO was actively turned into NO 2 by the oxidation reaction. Higher DRE of NO and NO x is realized in the CC-DBD reactor because new NO x was produced by the oxidation of N generated from N 2 dissociation in the TM-DBD reactor.
The simulation results showed that the electron energy distribution in the TM-DBD reactor was typical Poisson distribution differing from the CC-DBD reactor. The largest electron energy and average electron energy were 12.09 eV and 3.35 eV in the TM-DBD reactor, respectively, while they were 5.25 eV and 2.96 eV in the CC-DBD reactor, respectively.
These experimental and simulation results verified that enough energy could initiate the reaction of N 2 →2N and then produce new NO x in the presence of O 2 in the TM-DBD reactor. Therefore, the electrode structure of the CC-DBD reactor is preferable for better NO abatement and fewer new NO x generation in the DBD reactor under higher O 2 concentration condition.