Rui FAN(樊瑞),Yaogong WANG(王耀功),∗,Xiaoning ZHANG(张小宁),∗,Zhentao TU (屠震涛) and Jun ZHANG (张军)
1 Key Laboratory of Physical Electronics and Devices of the Ministry of Education, Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China
2 School of Electronic Science and Engineering, Xi’an Jiaotong University, Xi’an 710049, People’s Republic of China
Abstract The regulation of memory effect that the residual charges generated during and after discharge act on the initiation and development of subsequent discharge is explored by adjusting the pulse parameters, which have an influence on the discharge characteristics.The memory effect is quantified by the measurement of ‘wall voltage’ through a series of reference capacitors.The influences of memory effect on the discharge properties corresponding to rising/falling time 50–500 ns,pulse width 0.5–1.5 μs,and frequency 200–600 Hz are analyzed.It is found that the‘wall voltage’ increases from 1.4 kV to 2.4 kV with rising/falling time from 50 ns to 500 ns, it varies in the range of 0.18 kV with frequency of 200–600 Hz, and 0.17 kV with pulse width of 0.5–1.5 μs.The propagation velocity of wavelike ionization under the negative pulse slows down from 2184 km s−1 to 1026 km s−1 as the rising/falling time increases from 50 ns to 500 ns due to the weakening of the electric field by the surface memory effect.More intense and uniform emission can be achieved through faster rising/falling time and higher frequency based on the volume memory effect, while pulse width has less influence on the emission uniformity.Furthermore, similar laws are obtained for spectral and discharge intensity.Therefore, the memory effect is most effectively regulated by rising/falling time, and the discharge properties are affected by the surface and volume memory effect.
Keywords: memory effect, discharge properties, wall voltage, pulse parameters
Low-temperature plasma has been proven to be useful for pesticide elimination, surface treatment, gas purification, and laser technology [1–9].Dielectric barrier discharges (DBDs)with simple scalability [10] excited by a nanosecond pulse can generate low-temperature plasma with high electron density, high energy efficiency, and better uniformity [11].The collective behavior of filamentary and patterned DBDs can be influenced by residual species [12].The residual species and neutral active species (vibrationally excited molecules,metastable molecules,positive ions,etc)generated during the previous discharge and after discharge cease have an influence on subsequent discharge that is called the memory effect[13–16].The investigation of regulation of the memory effect and its influence on discharge characteristics is valuable for both an improvement of discharge efficiency and a reduction of sustain voltage [17].
Memory effect agents include residual charges, neutral active particles, the detachment of negative ions, and the photon-assisted detachment of weakly bonded electrons on the dielectric surface[18,19].The memory effect universally facilitated the subsequent discharge initiation.Specifically,the reproducibility of the micro-discharge channel position is governed by the volume memory effect, and the surface memory effect is responsible for micro-discharge jittering in time[20].The electric field offset before the discharge pulse,streamer initiation, and propagation are also attributed to the memory effect [21–24].In addition, these discharge properties related to memory effects, such as electron avalanche development and pre-ionization distribution, can be modulated by voltage waveform parameters and dielectric material[25–29].High-energy electrons, discharge current density,electron density,and streamer propagation velocities,with the applied voltage,pulse repetition rate,pulse width,and rising/falling time were investigated[30–35].The propagation speed of ionization waves decreases as the rising time slows down in a micro-DBD at low gas pressure [36].In addition, the pulse polarity also has an impact on the properties of discharge in asymmetrical arrangements, while the polarity dependence of volume DBDs in symmetrical arrangement is not known [37–40].
In summary, dominant memory effect agents are significantly affected by many factors, such as the voltage pulse waveform, gas pressure, gas composition, discharge history/mode, electric field inhomogeneity, and ‘pulse off’ period[13].Major memory effect agents, their influential mechanisms, and decay pathways have been summarized [41].Moreover, the discharge peculiarities associated with the memory effect at the effects of waveform parameters were studied [42–44], while the modulation of memory effect by pulse parameters needs to be explored.
In this work, the memory effect is quantified by measuring the‘wall voltage’—a term introduced from the plasma display panel, which has the same structure as DBD—by using a reference capacitor.The ‘wall voltage’ with pulse parameters (rising/falling time, pulse width, and frequency)between the positive and negative half cycle of a bipolar pulse is investigated.Moreover, the influences of memory effect(surface and volume) on the discharge properties, such as ionization propagation, discharge intensity, and uniformity emission, with pulse parameters in a DBD device at atmospheric-pressure N2are preliminarily explored.
The rest of the paper is organized as follows.Section 2 describes the experimental setup and the principle of the memory effect.Section 3 elaborates on the influence of pulse parameters on wall voltage,presents the experimental results,and discussions on the memory effect acting on discharge characteristics.Section 4 gives the conclusions.
A schematic of the memory effect in the DBD device used in our experiment is indicated in figure 1.The memory effect discussed in this work mainly refers to the effect of the residual charge on the negative pulse after the discharge of positive pulse when the DBD is driven by a bipolar pulse.The bipolar high-voltage pulse is input to the left electrode of the DBD and the right electrode is grounded.After a positive pulseVswith amplitude higher than the ignition voltage acts on DBD,micro-discharge channels are formed in the gas gap.Positive charges and electrons are accumulated on the surface of the dielectric layer close to the cathode and anode,respectively.The formed voltage of accumulated charge is called the wall voltageVw.Then, in stage 1 (the interval between positive and negative pulses), the residual charge mainly exists in the former discharge channel and the surface of the dielectric layer.In stage 2, the negative pulse −Vsis applied, and the remaining wall voltageVwis superimposed on the negative pulseVs, which affects the discharge in stage 2.
In order to measure the wall voltage and investigate the influence of the memory effect on discharge properties with pulse parameters,the experimental system is set up as shown in figure 2.The plasma reaction chamber is pumped to 10−1kPa by a mechanical pump, and then the reaction chamber is filled with nitrogen to atmospheric pressure in order to analyze the discharge properties in a static nitrogen environment.The closed chamber is made of plexiglass.Two pieces of 1 mm thick indium tin oxide(ITO)glass(glass area 10×10 cm2, ITO area 5×5 cm2) are fixed together,forming a DBD device with a 3 mm gap between the two dielectric layers.The DBD device is excited by a parameterized nanosecond pulse power supply that can be programmed to output a bipolar pulse with adjustable pulse waveform[40,45].The reference capacitorCcis connected in series with the DBD to measure the wall voltage.The pulse voltages and discharge currents are measured with an oscilloscope (DSO90254A), a 1000:1 high-voltage probe (Tektronix P6015A) and a current probe (Magnelab FCT).In addition, optical properties are measured by a spectrometer(Ocean optics 2000), discharge photos are taken by a camera(Canon EOS 550D),and the morphologies of DBD discharge propagation are captured by an intensified charge-coupled device (ICCD) (Andor DH334).The electric fields of the discharge gap parallel to the electrode surface are denotedExandEy, and that perpendicular to the electrode surface isEz.
The applied voltage, current waveform, and voltage waveform on the reference capacitorCcfor typical parameters(rising/falling time 50 ns,pulse width 1 μs,frequency 500 Hz)of the bipolar pulse used in the experiment are shown in figure 3.Without specific declaration,the default time interval between the positive and negative pulse is 8 μs, and the amplitude is 14 kV, while the rising/falling time can vary in the range of 50–500 ns,pulse width in the range of 0.5–1.5 μs,and frequency in the range of 200–600 Hz.The voltage waveform on the reference capacitorCcis measured as shown in figure 3(b) when the high-voltage pulse in figure 3(a) is applied to the DBD.According to the measured voltageVcwonCcduring the time interval between the positive and negative pulse, the wall voltageVwcan be calculated by using equation (1).The specific principle has been described in our previous work [45].
whereCm=ε0εrS/dis assumed as an equivalent capacitor of the DBD,ε0is the absolute permittivity of a vacuum,and εris the relative permittivity of glass.
The memory effect agents may mainly include the positive particles(N2+and N4+)and metastable particlesandin atmospheric-pressure N2;some typical reactions[14] are listed:
By applying the measured equivalent capacitorCmof the DBD and reference capacitorCcwhich are 15.38 pF and 633 pF, respectively, the wall voltage can be calculated.The calculated wall voltages and the measured minimum sustain voltages change with the rising/falling time,pulse width,and frequency as shown in figure 4.According to figure 4(a), the wall voltage increases from 1.4 kV to 2.4 kV when the rising and falling time increases from 50 ns to 500 ns, as indicated by the purple circle curve.The ionization rates will cause changes in electron density and ion density, which leads to different time evolutions of the gap voltage and electric field for different rising and falling times.At the falling time of the positive pulse, the space charges will be recombined under the action of the electric field.The faster the falling time, the faster the electric field changes, resulting in a greater probability of recombination and lower residual wall charges,and vice versa.Therefore,as the falling edge slows down,there is more residual charge after the positive pulse, thus the wall voltage is greater.Furthermore, the minimum sustain voltage to maintain discharge varying with the rising and falling time is reflected by the gray square curve.It can be seen that the sustain voltage decreases from 11.7 kV to 9.1 kV when the rising and falling time increases from 50 ns to 500 ns.It also verifies that the wall voltage gradually increases with the decrease of the rising and falling time.
Table 1.Maximum transferred charge qmax,residual charge qres,and energy with rising and falling times of 50 ns, 100 ns, 200 ns, and 500 ns.
In addition,Q–VLissajous curves at rising/falling times of 50 ns, 100 ns, 200 ns, and 500 ns are indicated in figure 5.The maximum transferred chargeqmaxafter the positive pulse is calculated by theQ–VLissajous curves [46] and residual chargeqres, corresponding to rising/falling times of 50 ns,100 ns,200 ns,and 500 ns,as listed in table 1;moreover, the energyEof each pulse can be calculated by voltage–current–time integration[47],as shown in equation(2).It can be seen that theqmaxincreases from 842 nC to 922 nC,qresincreases from 24 nC to 255 nC when the rising and falling time increases from 50 ns to 500 ns, whileEdecreases.
whereMis the record length of data, Δtis the sample interval, andVn,Inare the values of measured voltage and current at the recording pointn.
In figure 4(b), the wall voltage remains nearly unchanged with pulse width within a certain range.When the pulse width changed from 0.5 μs to 1.5 μs, the wall voltage varied from 1.26 kV to 1.43 kV, and the amplitude of the sustain voltage changed from 10.8 kV to 10.4 kV.The pulse width refers to the pulse flat top period for the convenience of independent regulation of pulse parameters.Since the process of the accelerated electron collision, excitation ionization, and recombination mainly occurs during the rising and falling time, and the metastable particles e.g.during the flat top have little effect due to the small range of pulse width[48].In addition,the wall voltages and sustain voltages are almost unaffected by frequency of 200 Hz to 600 Hz; the wall voltages and sustain voltages vary in the range of 0.18–0.2 kV, respectively, as indicated in figure 4(c).The frequency has an enhanced effect on the discharge times, i.e.the cumulative discharge intensity;however,it has little effect on the wall charge accumulation in a single pulse period.Therefore, the memory effect of DBD discharge,namely the wall voltage,can be effectively regulated by the rising and falling time of the pulse in a certain range.
Due to the recombination of positive ions, the larger the interval between adjacent discharges, the more the influence of the memory effect may be reduced.In this work, the influence of residual charges and metastable particles generated by previous discharge on the rising edge of negative pulse is mainly studied.In nitrogen, the time-resolved, timeintegrated ICCD images of the DBD from the front and side view subjected to a bipolar pulse during the rising time of the negative pulse when the rising/falling time is 50 ns, 100 ns,200 ns,and 500 ns are studied in figures 6(a)–(d).The left part of figure 6 shows the voltage and current waveforms of the rising edge of the negative pulse.The middle part is the timeresolved images from the front view.The gate width of the ICCD detector is fixed at 3 ns, and a series of false color intensity images are captured in a continuous time range.The timet=0 ns is defined as the zero-crossing preceding the positive pulse of the bipolar pulse ofV(t).The four ICCD images correspond to the four moments of the rising edge of the negative pulse.For a fixed value oft, the obtained timeresolved images are the average of the corresponding image of 100 cycles of the pulse waveforms.The right part of figure 6 is the time-integral image from the front and side view for 100 cycles, i.e.0.2 s.
A wavelike ionization wave is observed in the timeresolved images at the negative pulse phase of the bipolar pulse.The discharge starts from the upper left corner and propagates along the diagonal to the bottom right corner because of the lead position of the electrode, as shown in figure 2, and the discharge propagation presents a smooth periphery of the packet.As the rising and falling time slows down,the wavelike ionization waves become more dispersed until 500 ns.This mechanism of wavelike ionization wave propagation can be mainly referred to the combined action of electric field distortion, surface memory effect, and photoemission.The discharge is weakened which is caused by the reduction of the electric field distortion, as the rising and falling time slows down.On the other hand, the surface memory effect is enhanced with the increase of residual charge.The residual charge particles, combined with the applied electric field, produce the horizontal electric fieldEy(the electric field parallel to the electrode), as shown in figure 2, to move the ionization wave forward.TheEydecreases with the increase of the residual charges distributed on the surface of the dielectric layer, resulting in weaker propagation characteristics [49].
The false color image (in 3D view) of DBD discharge propagation in nitrogen during the rising time of the negative pulse is indicated in figure 7(a).The propagation velocities of the first packet changing with the rising and falling time of the negative pulse along theYaxis marked in figure 7(a) are shown in figure 7(b).The propagation velocities along theYdirection refer to the propagation velocities of the peak discharge intensity in theYdirection calculated by MATLAB.The general trend is that the propagation speed decreases with the increase of the rising and falling time during the negative pulse, since in the case of fast rising time, the electric field distortion is more significant and ionization velocity is faster,while in the case of slow rising time, the electric field distortion alleviates and the ionization velocity is relatively lower.In addition, as the rising/falling time slows down, the surface memory effect is enhanced and the electric fieldEyalong the electrode plane weakens, which leads to the decrease of propagation velocity [49].Moreover, the propagation speed at the falling time of the negative pulse is lower than that at the rising time of the negative pulse, because the current of the falling time of the negative pulse is mainly affected by the residual charge particles generated by the applied electric field during the rising time.
DBD discharge photos are taken by a camera (Canon EOS 550D)and the emission uniformity of DBD discharge photos is evaluated by using the nine-point display uniformity method, which is usually used to evaluate the uniformity of plasma display panel displays [45], and the rules of emission uniformity, discharge intensity with rising and falling time,pulse width, and frequency in nitrogen are studied, as represented in figure 8.The emission uniformity is reflected by the average deviation of the brightness of each test area and the average brightness of the nine areas; therefore, the larger the value of nonuniformity, the worse the uniformity.With the increase of rising/falling time, the emission uniformities and discharge intensities decrease gradually as shown in figure 8(a).The discharge intensity and emission uniformity are both affected by the electric field distortion and volume memory effect.They are calculated from the discharge photos, which are the cumulative process of the discharge.Since the micro-discharges have a spatial ‘memory,’ every subsequent micro-discharge appears at exactly the same location occupied by the micro-discharge at the preceding half-period, which is caused by the volume memory effect[20].The higher the wall voltage, the stronger the discharge intensity at the position of the previous discharge due to the volume memory effect, which may result in increased discharge intensity differences between regions, and worse uniformity.In addition, as the discharge intensity is the average value of the whole discharge area, it is mainly affected by the electric field.
The influence of pulse width on the emission uniformity and discharge intensity in nitrogen is indicated in figure 8(b).With the increase of pulse width, the emission uniformity does not vary obviously when the pulse width changes from 0.5 μs to 1.5 μs.According to the above results, the wall voltage hardly changes with the pulse width in a certain range, as does the electric field.However, the discharge intensity decreases slightly with the increase of pulse width.The discharge at the rising time of the pulse is excited by the impact of an applied electric field,while at the falling time the discharge is the recombination of residual charged particles.Since the discharge intensity is a cumulative process, the increase of the pulse width will cause the decrease of residual particles during the falling time due to the recombination of positive ions, resulting in a small drop in the overall cumulative discharge intensity.In addition, figure 8(c) shows the influence of the pulse frequency on emission uniformity and intensity in nitrogen.With the increase of pulse frequency,the emission uniformity and discharge intensity gradually increase.The discharge photo is a process of integration.The higher the discharge frequency, the more discharge times per unit time, and the greater the discharge intensity.Furthermore, more discharge times will make the discharge in each area more uniform per unit time.Although the wall voltage does not change much with the frequency of 200–800 Hz, the micro-discharge will slightly shift between each discharge, so that the emission uniformity per unit time is improved.
For further analysis, the discharge intensity gradually decreases with the increase of the rising and falling time because of weaker electric field, although the wall voltage gradually increases.Figure 9 shows the schematic diagram of electric field distribution considering the memory effect.When the subsequent pulse is applied to the DBD, the actual electric field of the DBD is the superposition of the applied electric field and the electric field generated by the wall charge; therefore, the intensity of the subsequent discharge is determined by the applied electric field and the wall voltage.Since the amplitude of the wall voltage is 14.3% of the applied voltage, the discharge intensity is mainly affected by the applied voltage.The shorter rising time means the plasma is driven under significant overvoltage breakdown, plasma should be more reactive and have higher electron temperature,and more atoms should be excited to higher energy states [33].
Figure 1.Schematic of the DBD device and memory effect.
Figure 2.Schematic of the experimental setup.
Figure 3.(a)Applied voltage and current waveform of rising/falling time 50 ns and (b)voltage waveform on the reference capacitor under the action of the waveform in (a).
Figure 4.Wall voltage and sustain voltage change with (a) rising and falling time, (b) pulse width, and (c) frequency.
Figure 5. Q–V Lissajous curves with rising/falling times of 50 ns,100 ns, 200 ns, and 500 ns.
Figure 6.Voltage and current waveform, time-resolved and time-integrated ICCD images during the rising edge of the negative pulse half cycle at slopes of (a) 50 ns, (b) 100 ns, (c) 200 ns, and (d) 500 ns.
Figure 7.(a)3D process of DBD discharge and(b)the propagation velocity of the first packet change with the rising/falling time during the negative pulse along the Y direction.
Figure 8.Uniformity and discharge intensity changes with (a) rising/falling time, (b) pulse width, and (c) frequency in nitrogen.
Figure 9.Schematic of electric field distribution considering the memory effect.
Figure 10.Spectra change with (a) rising/falling time, (b) pulse width, and (c) frequency in N2.
The variations of spectrum (N2) with rising and falling time, pulse width, and frequency are shown in figures 10(a)–(c).The wavelength corresponding to the peak of the spectrum almost does not change with the rising and falling time,pulse width, and frequency, while the peak intensity of the spectrum gradually decreases with the increase of the rising and falling time.A faster rising and falling time means higher dv/dt, higher electron energy, more violent collisions between the particles, and more intense discharge.Although the wall voltage increases as the rising and falling time slows down, the wall voltage 1.4–2.4 kV is expected to be smaller than the applied voltage 14 kV, and the applied electric field plays the main role.As shown in figure 10(b), the peak intensity of the spectrum basically does not change with the increase of the pulse width.Since the accelerated collision and excitation processes of particles mainly occur in the rising and falling time, the pulse width has little effect on spectral intensity in a certain range.In addition, with the increase of frequency,the spectral peak intensity increases gradually.The measurement of the emission spectrum is also a cumulative process.With the increase of pulse frequency, the number of active particles excited per unit time will increase; therefore,the discharge intensity and spectral intensity will increase.At present, the wavelength corresponding to N2+(B2∏u+) greater than 350 nm is mainly measured.Although the intensity ratio of N2+(B2∏u+)/N2(C2∏u)varies with the pulse parameters,the effect of the pulse parameters on the spectral intensity of N2(C2∏u) qualitatively should be similar to that of N2+(B2∏u+); specifically, the shortening of the rising edge not only increases the spectral intensity of N2+(B2∏u+), but also the spectral intensity of N2(C2∏u).The discharge device will be improved to measure the influence of parameters on the spectral intensity during the ultraviolet band further.
By adding the reference capacitance,the quantification of the partial memory effect (i.e.wall voltage) between the positive and negative half cycle of a bipolar pulse has been indirectly measured, and the influence of pulse parameters including rising/falling time, pulse width, and frequency on memory effect has been analyzed.It has been shown that the rising/falling time plays a dominant role in wall voltage.The steeper the falling time of the positive half cycle, the more likely leads to the recombination and deexcitation of relatively more positive ions and neutral active particles,and the wall voltage is reduced, while the memory effect is almost unaffected by pulse width and frequency within a certain range(0.5–1.5 μs,and 200–600 Hz).Furthermore, the influence of memory effect on propagation properties, discharge intensity, and emission uniformity has been investigated at atmosphericpressure N2.The discharge properties are affected by the complex interaction of the applied electric field and memory effect.The mechanism of wavelike ionization wave in the rising edge of the negative half cycle is not affected by rising/falling time,while the propagation speed decreases due to the reduction of the electric field distortion and the effect of surface residual charge on the electric field in the direction of parallel electrodes.When the rising/falling time ranges from 50 ns to 500 ns, the applied electric field plays a dominant role in the discharge intensity; that is, the steeper the rising/falling time, the stronger the discharge intensity.Moreover,the discharge intensity decreases with the increase of pulse width and the decrease of frequency because of the memory effect aging.Due to the enhancement of the memory effect and attenuation of frequency, the difference of micro-discharge intensity in different positions increases, which leads to the decrease of the emission uniformity.In addition, the spectral intensity of N2+(B2∏u+)decreases with the increase of rising/falling time and the decrease of frequency, while the pulse width hardly affects the spectral intensity.Further, we will investigate the regulation of a wider range of pulse parameters on the memory effect and mechanism of action.
Acknowledgments
The authors would like to acknowledge the financial support provided by National Natural Science Foundation of China(Nos.51807156 and 61771382), Projects of International Cooperation and Exchanges Shaanxi Province(No.2018KW-034), China Postdoctoral Science Foundation (No.2017M623174) and Central University Basic Scientific Research Operating Expenses (No.xpt012019041).
Plasma Science and Technology2021年10期