An atmospheric pressure plasma afterglow to charge ultrafine aerosol particles

A novel flowing plasma system aimed at increasing charging efficiency of particulate matter and effective removal through electrostatic precipitation is studied. Nanoparticles are passed through the spatial afterglow of an atmospheric pressure radio-frequency glow discharge plasma. Particle charging efficiencies and polarities are measured at different plasma-aerosol gaps, aerosol and plasma flow rates, plasma powers, and afterglow DC bias. Various timescales are calculated to explain the transport of charge carriers that facilitate particle charging processes. The experimental results showed increased charging efficiency and net positive charging at longer gaps between the afterglow and aerosol stream and lower aerosol flow rates. Timescale analysis indicates that when ample residence time is provided, transport of charge carriers shifts from ambi-polar diffusion to free diffusion, and electrons are rapidly lost from the afterglow, resulting in highly efficient, net positive charging of particles. The charging efficiency of particles in optimized operating conditions was comparable or higher than reported collection efficiencies of electrostatic precipitators. The findings overall demonstrate that glow discharges are capable of charging particles not immersed in the plasma bulk, and such systems show promise for improving performance of particle mitigation technology.


Introduction
Fine particulate matter, described by the total mass concentration of airborne particles smaller than 2.5 μm in aerodynamic diameter (PM 2.5), can penetrate into human lungs and cause negative health outcomes including asthma and lung and heart diseases [1,2].PM 2.5 is produced from both anthropogenic sources such as fuel combustion in automobiles and power plants [3], and natural sources such as respiratory droplets [4][5][6] and forest fires [7].Various mitigation strategies have been developed and implemented based on the source of particulate matter.High-efficiency particulate air (HEPA) filters are a popular particle control technology, primarily used for cleaning indoor air; they are approved by the Centers for Disease Control and Prevention for mitigating viral-laden aerosols [8] and have been utilized in various commercial indoor air purifiers [9].However, the use of HEPA filters is limited to small-scale processes due to frequency replacement of filters along with high pressure drops and corresponding high operational cost.Electrostatic precipitators (ESPs) do not suffer from these limitations and are therefore a more economical particle control technology, implemented in several large-scale industrial processes.In an ESP, particle-laden industrial exhaust gas flows through corona discharges, which are a type of plasma formed from partial ionization of background gas in a strong, nonuniform electric field.Particles are charged through collisions with these ions and are electrostatically deposited on a collecting plate.While ESPs have a relatively low operational cost and are easily implemented into industrial exhaust streams, they exhibit poor collection efficiency for sub-micron sized particles due to charging inefficiency [10].Hence, ESP design modifications are crucial for improving collection efficiency of sub-micron particles.
Various methods have been explored to improve particle collection efficiency in ESPs by improving particle charging.Such methods include adding an upstream bipolar charging and agglomeration stage [11], utilizing various electrode materials and geometries [12,13], boosting ion concentrations by use of an electrode liner [14], and photoionizing particles with soft x-rays [15].Another promising strategy is to enhance charging using a glow discharge plasma [16,17].Glow discharge plasmas, also known as dielectric barrier discharges, are formed when a gas flowing through a gap between two electrodes, of which at least one is covered by a dielectric surface, is partially ionized by a strong electric potential is applied to the electrodes [18].Glow discharge plasmas have higher ion concentrations than corona discharges by orders of magnitude [19], showing potential to improve charging efficiency.Unlike other plasmas, glow discharges can be readily generated at atmospheric pressure, making them promising candidates for improving charging and subsequent collection of nanoparticles in an ESP.However, glow discharges cannot be easily generated out of the air that carries pollutant particles, limiting the application of previous studies examining aerosol charging inside a glow discharge plasma [16,17].A workaround for this issue is to decouple the aerosol and plasma streams.In such a system, a glow discharge plasma would incorporate into ESPs through a pre-charging stage in which the particle-laden waste stream is mixed with the afterglow of noble gas-containing glow discharge streams.Through collisions with afterglow charge carriers, particles would acquire an initial charge before passing into the main ESP stage, where they are further charged by the corona plasma and collected.Ideally, the pre-charger and corona will charge particles to the same polarity to avoid neutralization and decreased collection efficiency.Introducing the plasma through noble gas side-streams is more feasible than directly ionizing the air around the particles because air has a very high breakdown field at atmospheric pressure and would therefore require excessive energy input.While previous work [16,17,20] extended knowledge of particle charging phenomena inside flowing atmospheric pressure glow discharges, particles were not decoupled from the plasma stream, leaving several questions on how charging phenomena may differ in a more realistic, scalable, decoupled system.Hence, there remains a poor understanding of how particles passing through a remote plasma (rather than directly through the plasma) acquire charge-specifically with regard to factors such a flow rates, geometry, and gaseous composition of the aerosol stream.A thorough understanding of nanoparticle charging in glow discharge plasma afterglows is required in order to modify an ESP to improve collection efficiencies.
Several previous studies have examined the charging of nanoparticles immersed in a flowing plasma [16,17,20].In such systems, a nanoparticle-noble gas mixture is flowed through a radio-frequency pulsed electric field, and a plasma is generated from the background gas.The key measurement is the size-dependent charge distribution, describing the fraction of particles with a specific quantity of charge (e.g., 0, ±1, ±2 elementary charges).Charge distributions of nanoparticles exiting low pressure glow discharge plasmas have been studied and generally exhibit both positively and negatively charged nanoparticles (bipolar charging) [21,22].At atmospheric pressures, nanoparticle charge distributions have been found to contain one charge polarity (unipolar charging) [23] or both polarities depending on selection of plasma geometry and operating parameters such as flow velocity and input power [20].In a particular study [17], bipolar charging of magnesium sulfate particles was observed in an atmospheric pressure, capacitively-coupled, radio frequency discharge.The bipolar charging was explained in a two-stage model supported by a characteristic timescale analysis; particles first charge negatively in the bulk plasma due to higher mobility of hot electrons, then are neutralized and positively-charged in the ion-rich spatial afterglow, which is the post-plasma region where the driving force for the plasma vanishes and the plasma relaxes back to an ambient gas.In a follow-up study [16] a DC bias was applied in the spatial afterglow to remove positive ions and preserve the negative particle charge from the plasma.Particle charge was successfully enhanced and unipolarly negative when a negative DC bias was applied, demonstrating that simple manipulations to the spatial afterglow can influence charge carrier transport.However, adapting these well-studied systems, in which particles are entrained in the plasma, to an ESP precharging stage is not possible because particles in typical waste streams are not immersed in a readily ionizable gas.Therefore, a scalable alternative must be developed that utilizes the high ion concentrations of a flowing glow plasma and charges particles unipolarly, while keeping particles in a separate stream.
In this paper, a system is devised and characterized in which submicrometer sized aerosol particles are flowed through the spatial afterglow of an atmospheric pressure, radio frequency glow discharge plasma.The charging efficiency and polarity of particles is measured as a function of various system operating parameters.A theoretical timescale analysis is performed to explain the ion and electron transport in the afterglow that governs particle charging.

Experimental design
The experimental setup is shown in figure 1.A solution of magnesium sulfate in deionized water (0.17 mg ml −1 ) was atomized (TSI model 3076) in either argon, air, or nitrogen gas at 3 L m −1 , to produce salt droplets.Droplets were dried in silica-containing diffusion driers to yield a salt aerosol.Magnesium sulfate was selected due to its high melting and boiling points, preventing particles from shrinking or growing when exposed to hot plasma afterglow species.Constant particle size is crucial to isolate the size-dependent effects of charging.While composition plays a role in charging processes (e.g., photoionization and thermoionization [24]) we opted to use magnesium sulfate across all experiments.Further, a previous study [20], found that particle charging in a flowthrough plasma was most weakly dependent on composition, whereas other factors such as particle size and plasma parameters dominated charging.Next, a controlled portion of the flow was removed into a vacuum system to control the aerosol flow through the system.The aerosol was then passed through a Kr-85 neutralizer (TSI model 3077A), reducing the number of charged particles.Remaining charged particles were then removed using a charged particle remover (CPR), consisting of a grounded cylindrical tube with an inner electrode fixed at −2 kV.The neutral aerosol was then passed through one branch of the cross-flow plasma reactor.Pure argon gas passed through the other branch, and a glow discharge plasma was generated using a series of three ring electrodes.The first electrode was connected to an RF power supply (13.56 MHz, RF VII Inc., Model RF-3 XII) and a homemade L-type matching network.The second electrode was grounded.The bulk plasma was therefore sustained between the first two electrodes.Because the RF voltage was insufficient to generate a breakdown field for argon at atmospheric pressure, the third electrode downstream of the ground electrode was connected to a high voltage AC source to ignite the plasma.This electrode was also connected to a high voltage DC supply for DC bias experiments.Control of aerosol charging by applying a DC bias in the spatial afterglow was demonstrated in a previous study [16], in which a DC bias applied to the spatial afterglow removed ions and enhanced particle charging.However, this system contained aerosol particles within the bulk plasma instead of passing the aerosol through the afterglow, hence additional experiments were performed to gauge the effect of a DC bias on a particle-free afterglow.An additional ring electrode was placed in the spatial afterglow 1.2 cm downstream of the ground electrode and was connected to a high voltage DC source.The distance between the aerosol and the plasma afterglow was changed by moving the three electrodes along this branch.
At the intersection of the two branches, the afterglow plasma and aerosol streams mixed, and charge carriers from the plasma afterglow collided with and charged particles.The flow rates, distance between the plasma afterglow and the aerosol stream, plasma power, and presence of a DC bias may affect the transport of charge carriers and charging rates, leading to changes in afterglow composition and the resulting particle charging.
The aerosol exiting the cross-flow reactor contained particles charged by the plasma afterglow species.To determine the charging efficiency, or the size-dependent fraction of charged particles, the aerosol passed through another CPR which removed any charged particles.A scanning mobility particle sizer (SMPS: TSI model 3938) sampled the outlet of the CPR and measured the particle size distribution of the aerosol which spanned 17 nm to 250 nm.The size-dependent charge fraction (CF) was calculated by simply calculating the ratio of size distributions with and without a voltage applied to the downstream CPR.
Particle charge polarity is the second measurement objective.The overall fraction of positive and negative particles was measured by bypassing the neutralizer of the SMPS and changing the polarity of the differential mobility analyzer (DMA) within the SMPS.In typical SMPS operation, the aerosol is pre-charged by the neutralizer to a Fuchs charge distribution [25,26].Particles of fixed electrical mobility diameter transmit through the DMA are counted in the condensation particle counter (CPC).A robust inversion algorithm adjusts this count based on the known fraction of un-counted particles that did not transmit through the DMA due to different mobility.The size distribution of the aerosol is measured by scanning through different DMA voltages and repeating the inversion algorithm.If the overall polarity of particles is desired, one can bypass the neutralizer, allowing preservation of the original charge distribution, and send particles directly into the DMA.The resulting particle size distribution with a positive and negative voltage to the DMA will indicate overall concentrations of negatively and positively charged particles, respectively.Ratios of the measured particle size distributions with this method will indicate the polarity fraction of the charged particles.

*
Where n -(+) is the number of negative (positive) particles.While this method helps elucidate the relative fractions of positive and negative particles, the magnitude of charge on particles is still unknown.To further characterize particle charging and understand the extent of multiple charging (q p ± 2), a tandem DMA (TDMA) setup was used [16,17,27].In TDMA analysis, the charged aerosol from the cross-flow plasma reactor first passed through a single DMA column of fixed voltage, allowing only particles of a fixed electrical mobility to exit.While electrical mobility is known, the sizes and charges of the particles exiting the first DMA are unknown.The particles were therefore sent into an SMPS to measure the particle size distribution.The voltage of the first DMA column was adjusted to target same-sized particles of different charges and polarities.The relative integrated peak areas from the particle size distributions were used to obtain the charge distribution of two ultrafine particle sizes, 45 and 90 nm.

Test plan
The experimental plan is summarized in table 1.Operating parameters are systematically varied to understand charge carrier transport and particle charging phenomena in the cross-flow system.Conditions that maximize particle charging efficiency and unipolar charging are identified, and results are presented comparing charging efficiencies between the optimized present system and larger-scale ESPs.

Theoretical analysis 2.3.1. Relevant timescales
To further understand the behavior of charge carriers in the spatial afterglow of the glow discharge, different timescales were calculated and compared.Critical timescales considered in the system were residence time, electron relaxation time, and diffusion regime transition time.Residence time describes the transit time between the start of the spatial afterglow and the aerosol stream, which depends on the geometry of the plasma branch of the reactor and the flow rate of argon gas fueling the plasma.In this study, the length (L) between the afterglow and the aerosol, and the plasma flow rate (V  plasma ) were varied.The next timescale calculated was the electron relaxation time.
In the nonthermal environment of a glow discharge, electrons rapidly drift due to a fluctuating electric field, providing them with orders of magnitude higher energy than ions and neutral species [19].When electrons exit the bulk plasma, the driving force for the nonthermal environment vanishes, and electrons relax to the neutral gas temperature via collisions with gas molecules.
Where δ is twice the electron mass divided by the argon atom mass, n Ar is the neutral gas concentration, β th is the electron thermal velocity, σ mt (ε) is the energy-dependent electron-neutral momentum transfer cross section, and f(ε) is the Maxwellian electron energy distribution function [28].
The third characteristic time is the time required for the afterglow to enrich with ions.Immediately after ions and electrons exit the bulk plasma, they begin to diffuse away and are lost to the reactor walls.If the Debye length (characteristic length scale of the plasma inversely proportional to plasma density) is sufficiently small relative to the system size, the charged species will undergo ambipolar diffusion.In ambipolar diffusion, electrons and ions diffuse together due to a strong space charge and resulting electric field that forms from charge separation [28].As electrons and ions travel through the afterglow and are lost to the walls of the reactor at the same rate, the Debye length continues to increase and approaches the system size.When the ratio of system size to Debye length reaches 100 [29], electrons and ions diffuse independently.Highly light and mobile electrons are rapidly lost from the afterglow stream, resulting in an ion-rich afterglow.The third characteristic time refers to the point at which the transition occurs between ambipolar and free diffusion.
Where τ ambipolar refers to ambipolar diffusion time, D a /Λ 2 , D a is the ambipolar diffusion coefficient, Λ is the system size (R tube /2.4) [30], n e,0 is the electron concentration at the start of the afterglow, and n e,crit is the electron concentration at which the diffusion regime transition occurs.
When the plasma afterglow mixes with the aerosol stream, mixing and charging of nanoparticles takes place.Depending on how much time is available for mixing and charging to take place, particles may or may not reach their equilibrium charge state.Because charging events are expected to occur over micro/millisecond timescales [17], and our setup permits more than one second between mixing and charge characterization, we assume that ample time is provided for mixing to take place.

Kinetic model to estimate plasma parameters
Electron temperature and concentration are required to compute characteristic times.The glow discharge plasma used in this study was characterized in a previous study [17] using measurements of plasma resistance coupled with a plasma fluid model, at plasma powers 60-105 W. Plasma parameters at lower powers were outside of the scope of that study and are therefore not available.Because lower plasma powers were used in this study for the interest of scalability into an industrial system, and input parameters for calculating characteristic times are not available, a kinetic model was used to estimate plasma parameters.
To constrain the electron concentration, the diffusion-reaction equation for electrons in the plasma was solved to obtain the radial profile of electron concentration within the plasma reactor.A symmetry condition was imposed at the radial centerline, and a sink condition was imposed at the reactor wall.MATLAB bvp4c boundary value problem solver was used to resolve the electron concentration profile.
Steady state is assumed, allowing the left-hand side to equal zero.On the right-hand side, the first term is the rate of electron loss to the reactor wall via ambipolar diffusion.The second term is electron generation by ionization of background gas.Rate constant k 1 is a function of electron temperature and the energy-dependent argon ionization cross-section [31], and was calculated assuming a Maxwellian electron energy distribution function.
The third term is the electron loss rate due to dissociative recombination (Ar 2 + + e − → 2Ar).Rate constant k 3 is a function of neutral gas and electron temperature [32,33].The fourth term is the electron loss rate due to threebody associative recombination (Ar + + 2e − → Ar + e − ).Rate constant k 4 is set to a fixed value of 7 ×10 −27 cm 6 s −1 [34].
After resolving the electron concentration profile, the input power was balanced by the sum of power dissipations within the plasma.
The input power used was 60 and 6 W. The first term of the right-hand side is power dissipation via electronneutral collisions, calculated using the rate constant k 1 and the integrated electron concentration profile.The second term is the power dissipated by volumetric recombination reactions in the bulk of the plasma, calculated using the associative and dissociative recombination rate constants k 3 and k 4 along with the integrated electron concentration profile.Equations for each power dissipation term are provided in the supplementary data.Equations ( 6) and (7) were simultaneously solved to obtain the electron concentration for each plasma power, which were then used to compute the diffusion regime transition times.Increasing the distance between the afterglow and the aerosol stream increases the charged fraction and leads to a higher fraction of positively charged particles.Not shown in figure 2, the same trends were observed across all aerosol flow rates and for both air and argon aerosol carrier gases.The shift to positive polarity suggests that the afterglow has a higher ion concentration than electron concentration after longer distances (figure 2(b)).A potential explanation is that, when the plasma is placed further away from the aerosol stream, sufficient residence time is provided for the diffusion regime to shift from ambipolar to free, resulting in increased losses of electrons.Without electrons to compete with and counteract positive charging, charging efficiency is also enhanced.When the plasma is placed closer to the aerosol stream, only ambipolar diffusion occurs and hence the aerosol is exposed to similar concentrations of ions and electrons, resulting in bipolar charging and lower charging efficiency due to competing charging polarity.

Aerosol flow rate
Aerosol charge fractions and polarities at different aerosol flow rates are shown in figure 3. Lower aerosol flow rates enhanced charging and shifted polarity to positive.This can be explained by the aerosol flow rate affecting the dilution of the plasma afterglow.At high aerosol flow rates, the ions and electrons were diluted by the aerosol carrier gas to a greater extent than at lower flow rates.Because aerosol diffusion charging rates depend on the concentration of charge carriers [35], charged fraction decreased with dilution.The effect of aerosol flow rate was observed only at 10 cm.This may be explained by the presence of both ions and electrons at closer electrode distances, resulting in similar concentrations in the aerosol-afterglow mixture regardless of aerosol flow rate.
Because electrons and ions were diluted to the same extent, particle neutralization rates were unchanged, and no charge fraction or polarity shift was observed.

Plasma flow rate
Plasma flow rate had no effect on charging efficiency nor polarity (figure S1).The effect of plasma flow rate on plasma parameters may explain this discrepancy.Previous plasma characterization studies [36,37] have shown evidence of electron temperature and concentration variations with plasma flow rate.Additionally, the length of the afterglow increased when plasma flow rate was increased, indicating higher plasma concentrations.Therefore, electron and ion transport properties (e.g., diffusion coefficients) at different plasma flows may be playing a role and working to counteract the effects of dilution seen when varying aerosol flow rate.The extended afterglow length also indicates that the transition from ambipolar to free diffusion may not be occurring before reaching the aerosol cross-stream, resulting in less efficient charging despite a higher concentration of plasma species in the plasma-aerosol mixture.

Plasma power
Aerosol charge fractions and polarity for 6 W and 60 W plasmas are shown in figure 4. At higher plasma power, charging efficiency was higher, and all particles were bipolarly charged.Additionally, at high plasma power, there was no effect of electrode distance nor aerosol flow rate on charging and polarity.The notable trends with varying electrode distance and aerosol flow were only observed at low plasma power.This may be explained by the effect of plasma power on electron concentration.At higher plasma powers, the fluctuating electric field driving the plasma is strengthened, providing electrons with more energy.Increased electron temperature speeds up argon ionization rates and therefore boosts electron concentration.The initial electron of the spatial afterglow is higher than at lower plasma powers.The effect of plasma power on electron concentration was also experimentally verified for this plasma system [17].The initial Debye length is smaller at higher plasma power, and longer times are required for the ambipolar-to-free diffusion transition to occur.Visual observations of lengthening of the plasma afterglow further support this explanation.Hence, even when the electrode was placed far away from the aerosol stream, insufficient residence time was provided for electron loss to take place due to high initial electron concentrations, resulting in bipolar charging.While increasing plasma power improved charge fraction, it should be noted that charge fraction was more sensitive to the electrode distance and aerosol flow rate.This can be seen by the higher charge fractions for lowerpower optimized conditions (upper curve in figures 2 and 3) than that of the high-power non-optimized condition (curve from figure 4).Thus, when scalability is concerned, operating parameters can be carefully selected to improve charging instead of simply increasing input power, thereby saving on operating costs and energy consumption of the improved ESP system.

DC bias strength
The third ring electrode downstream of the plasma afterglow was connected to a high voltage DC source to test the effect of a DC field on the afterglow and subsequent particle charging.For this set of experiments, the plasma electrodes were kept close to the aerosol stream (4 cm), and medium aerosol flow was used (1 L m −1 ).Relatively inefficient charging and bipolar distributions were observed at these conditions, providing sufficient room for charging improvement to be observed.
There was no observed effect of DC bias on charge fraction for both air and argon aerosol carrier gas.Aerosol polarities at −0.9, 0, and +0.9 kV DC bias is shown in figure 5. DC bias affected aerosol polarity, exhibiting strong dependence on the aerosol carrier gas.When argon was the aerosol carrier gas, only a positive DC bias affected polarity, resulting in more positive particle charging.When air was the aerosol carrier gas, neither DC bias had an effect, resulting in all bipolar distributions.To further understand the dependency on aerosol carrier gas and explain the result with air, pure nitrogen was used as a carrier gas.With nitrogen as the aerosol carrier gas, both positive and negative DC biases affected particle polarity; when a positive bias was applied, more positive particles were detected, and when a negative bias was applied, more negative particles were detected.All trends were the same at low and high plasma powers.
The magnitudes of electric fields in the afterglow are relevant in understanding the results.In the absence of a DC bias, there exists a polarizing electric field [28] that drives ambipolar diffusion.The polarizing field is driven by the charge separation due to differential mobility of electrons and ions, pointing radially towards the walls of the reactor.The field strength is directly proportional to electron temperature and inversely proportional to reactor scale, (k b T e /eR), approximately 500 V m −1 .When a DC bias is applied, an electric field with both radial and axial components is generated between the DC ring electrode and the middle ground electrode.The radial electric field strength depends on both the axial and radial positions within the tube and is approximately ±10 4 V  m −1 at its highest.Hence, the electric field imposed from the DC bias is significantly higher than the polarizing field driving ambipolar diffusion and is expected to dominate the transport of charge carriers in the plasma afterglow.
When a positive DC bias is applied, electrons are forced towards the edge of the reactor and are lost to the walls.Ions concentrate towards the center of the reactor and flow to the aerosol cross stream with minimal wall losses due to low mobility.The aerosol is therefore exposed to higher concentrations of positive ions, despite the close electrode distance.With argon and nitrogen as the aerosol carrier gases, these ions are consumed though collisions with aerosol particles.The peculiar bipolar result with air suggests that other ion consumption pathways are made available by oxygen gas.Oxygen molecules may serve as targets for argon ions to form positive diatomic oxygen ions (O 2 + ) [38,39].The axial electric field, pointing away from the afterglow, may also accelerate ions towards the cross-stream, providing them with sufficient kinetic energy to participate in otherwise inaccessible reaction pathways.
When a negative DC bias was applied, the electric field lines point towards the reactor walls and removed ions from the afterglow stream.Between the DC bias electrode and the aerosol stream, the afterglow is electronrich and free diffusion takes place, resulting in relatively higher electron wall losses than that of ions in the case of positive DC bias.Aerosol particles are then exposed to remaining electrons.In the case of argon aerosol carrier gas, some of these electrons may have had sufficient energy to ionize background argon gas, leading to formation of positive ions and electrons, resulting in bipolar charging.When air was the aerosol carrier gas, electrons reacted with electronegative oxygen molecules in various pathways, including impact ionization, dissociation, and excitation processes.Electrons may have also picked up significant kinetic energy from the axial electric field, providing them with sufficient energy to participate in such reactions.These processes served to reduce the electron current to the aerosol particles, resulting in bipolar charging.In the case of inert nitrogen, no reaction pathways were energetically favorable for electrons, hence electron current to the particle surface was maximized, resulting in more negative charging.

Charging at optimized conditions
Aerosol charge fraction and charge distributions at the optimal operating conditions for maximizing particle charging and unipolar charging, identified in 3.1.1-3.1.5(10 cm electrode distance, 0.25 L m −1 aerosol flow, 6 W plasma power), are shown in figure 6.
Charge fractions of particles passing through the afterglow at optimized conditions were compared to collection efficiencies of lab scale [10] and pilot scale [40] ESPs (figure 6(a)).In the ultrafine (<100 nm) size range, charging efficiency of particles passing through the glow discharge afterglow is comparable to that of the lab scale ESP and higher than that of the pilot scale ESP.For particles larger than 100 nm, charging efficiencies from the present study far exceed that of both the lab and pilot scale ESPs.It should be noted that several studies have examined ESP performance at various scales and operating parameters, hence several collection efficiency curves exist beyond those presented.Nonetheless, the glow discharge afterglow system shows strong potential to improve charging efficiencies in ESP systems.Particle charge distributions and average particle charges of 45 and 90 nm particles at optimized operating conditions are shown in figure 6(b).The larger particle exhibits a higher degree of multiple charging, attributed to the typical size-dependency of ultrafine particle diffusion charging [26,41] and proportionality between particle diameter and capacitance.
Charge distributions of particles passing through a plasma afterglow are different than that of particles directly entrained in the plasma [16,17].Previous studies have reported neutral fractions between 0.45-0.5 for 45 nm particles and 0.19 to 0.26 for 90 nm particles directly flowing through a plasma, notably lower than that observed in the cross-flow geometry.Charged fractions are expected to be higher for particles passing directly through a plasma due to exposure to significantly higher plasma densities.Despite less efficient charging of particles passing through an afterglow, the ion or electron-mediated neutralization of particles flowing through a plasma, that results in bipolar distributions and reduced average particle charge [17], can be mitigated in a cross-flow charging system by simple manipulation of system geometry (adjusting electrode distance).Hence, unipolar charging is more feasible in a cross-flow system.Particles in a flow-through plasma with a DC bias do not undergo neutralization and hence have a very low neutral fraction (0.15 for a 45 nm particle) and nearly unipolar negative charge [16].While the cross-flow system, at least in its laboratory-scale form, did not charge particles as efficiently, it demonstrated charging of particles in a separate stream and control over polarity, important characteristics for incorporation into ESP systems.

Theoretical analysis 3.2.1. Timescales
A theoretical timescale analysis was performed to explain findings from section 3.1 and to support proposed explanations based on plasma and charging mechanisms.Electron temperature and concentration were estimated using a kinetic model (section 2.3) to determine the electron relaxation time and diffusion regime transition time.To validate the model, predictions were compared to experimentally determined parameters for a 60 W plasma [17].Because electron temperature remained constant at 0.51 eV at different plasma powers, the same value was assumed for the lower plasma power, leaving electron concentration as the sole free parameter.Other studies have also reported similar trends with electron temperature and applied power [36].
The electron concentration measured by Sharma et al for a 60 W plasma was 1.2 * 10 20 m −3 .The predicted electron concentration was found to be 1.8 * 10 19 m −3 .While there is a discrepancy between the experimental and calculated values, the kinetic model provided a value within an order of magnitude of accuracy.Therefore, the kinetic model was used to calculate the electron concentration at the lower plasma power.This value was found to be 4.8 * 10 18 m −3 .
The critical Debye length beyond which free diffusion occurs, determined by the ratio Λ/λ De = 100, is 4.2 μm, corresponding to an electron concentration of approximately 3.16 * 10 17 m −3 .This critical value, along with the initial electron concentrations calculated from the kinetic model, were used to determine the diffusion regime transition times for each plasma power (equation ( 5)).
Characteristic times relevant to the plasma system are shown in table 2. Two residence times (τ res ) are provided, corresponding to the two plasma-aerosol spacings.Two diffusion regime transition times (t transition ) are provided, corresponding to the two plasma powers.The electron relaxation time is significantly smaller than other timescales and therefore one can assume a thermal afterglow environment.The residence time and diffusion regime transition times are all on millisecond scales and can therefore be closely compared.Both t transition times are longer than the shorter residence time, corresponding to the closer electrode distance.The t transition time for the 6 W plasma lies in between the two residence times, while the 60 W plasma transition time is nearly equal to the longer residence time.

Proposed mechanism
The proposed mechanism is illustrated in figure 7. The relative timescales support the claim that ion and electron transport phenomena in the afterglow underlie aerosol charging in the cross stream.At the longer afterglow distance and lower plasma power, there is ample residence time for free diffusion to begin, leading to electron losses and the observed positive charging, whereas in the high-power condition, the transition barely occurs before contacting aerosol particles, leading to the observed bipolar charging.At short distances and both plasma powers, there is insufficient residence time for the ambipolar-to-free diffusion transition to occur, resulting in the bipolar and inefficient charging observed.The diffusion regime transition time can be significantly decreased by applying a strong external DC field to the afterglow to accelerate electrons to the wall.However,

Adapting a glow discharge plasma into electrostatic precipitators
Differences in scale between a large-scale ESP system and the small-scale laboratory setup evaluated in this study necessitate further research.Compared to the parameters used in the present study, particle-laden waste streams flow through industrial ESPs at significantly higher rates and through much wider tubing components.Findings from this study suggest that such high aerosol flow rates would lead to reduced charging efficiency.However, flow velocity may be lower or comparable to that of the scaled-down setup due to larger tube diameters.Therefore, future studies must determine if the effect of aerosol flow rate on charging stems from gas velocity or simply the dilution driven by higher flows.Appropriate tube geometries (e.g., one wide tube versus several narrow tubes) and flow rates must then be selected.Scaling the flowing glow plasma is also not straightforward.Some aspects of the lab-scale plasma system may be held relatively constant if adapted into a scaled-up system to ensure a stable plasma can be formed.For example, the inter-electrode spacing (distance between ground and RF electrode) and tube diameter should remain relatively unchanged because igniting a plasma at atmospheric pressures requires a high breakdown field.Other plasma operating parameters should be modified to increase ion concentrations and compensate for higher aerosol throughputs.The afterglow of a scaled-up system should be sufficiently distanced from the aerosol stream to ensure electrons are lost, and DC fields can accelerate this loss if needed.The effect of plasma flow rate on plasma density and temperatures should be investigated as these parameters are tied to charge carrier transport in the afterglow and subsequent aerosol charging.Regardless, because maintaining a plasma flow equal to or higher than the aerosol flow is unfeasible for a scaled-up system, multiple plasma streams can be incorporated along the length of the ESP to boost charging.Other considerations not captured in this study may also play a role in scale-up, such as the effect of other combustion gases in the aerosol stream, temperature of aerosol stream, and initial charge on particles.Incorporation of glow discharges into ESP systems would increase operating costs from powering the plasma, pumping gas through the plasma system, and the consumption of argon gas.While increasing plasma power slightly increased charging efficiencies, optimization of other operating parameters such as electrode distance was able to boost charging efficiencies to a greater extent.Therefore, plasma power should be considered as only a secondary measure to increase charging efficiency.Operating costs from consumption of argon gas can be reduced by using a plasma composed of both argon and a more abundant gas such as carbon dioxide or oxygen.However, more energy may be necessary to sustain the plasma due to higher ionization energies and stronger breakdown fields.Future studies must therefore address the power requirements of argon composite plasmas and the effects of plasma composition on aerosol charging.

Conclusion
Electrostatic precipitators, a widely used particle removal technology, suffer from inefficient collection of ultrafine (<100 nm) particles due to poor charging efficiency.Meanwhile, nonthermal plasmas have gained recent attention for efficiently charging ultrafine particles due to high ion concentrations, and therefore show potential to incorporate into an ESP to improve performance.However, existing studies [16,17,20] have only examined the charging of particles immersed inside a plasma, as a single stream, which is not possible in an ESP due to an unrealistically high breakdown field for typical waste stream gases.Alternatively, a glow discharge afterglow may still be utilized by passing particles through its spatial afterglow, as two separate streams (pure plasma stream and waste particulate stream) rather than one.To develop such a system, the plasma and particle charging physics underlying this modified system, and how operating parameters affect these physics, must be carefully examined.To address this knowledge gap, in this study, nano-sized salt particles were passed through the spatial afterglow of a flowing nonthermal plasma as means to enhance particle charging in a scalable manner.The dependency of particle charging efficiency and polarity on system operating parameters was systematically studied.It was found that particle charging efficiency was enhanced and preferentially positive when the plasmaaerosol gap was increased, and aerosol flow rate was decreased.The flow rate of the plasma stream had no effect on particle charging.Applying a DC bias to the afterglow region affected charging when particles were entrained in nitrogen and argon gas but not air, preferentially positively charging particles when a positive bias was applied, and vice-versa for negative bias.Notably, these particle polarity shifts were observed at 6 W plasma power and not 60 W. A characteristic timescale analysis was performed to explain the ion and electron transport phenomena underlying the charging processes.Taken together, the experimental findings and timescale analyses suggest that the transition from ambipolar diffusion to free diffusion governs the composition of the afterglow and how aerosol particles are subsequently charged.When sufficient residence time is provided for the transition to occur, electrons are lost from the afterglow and aerosol particles undergo positive ion-mediated charging.
This paper demonstrates that particles can be charged in a plasma afterglow, overcoming practical limitations of previous glow discharge systems [16,17] in which particles are entrained in a readily ionizable gas.Charging efficiencies in the cross-flow system were lower than that of the single-stream systems due to lower plasma densities in afterglows, but the majority of particles over 50 nm carried a charge, making the optimal charging efficiency in our system comparable or higher than that of lab-scale and pilot-scale ESPs.Further, precise control over charging polarity was demonstrated here, which is crucial for incorporation into ESPs as a pre-charging stage and a distinct advancement of previous glow discharge work [16,17].These findings provide a starting point for utilizing glow discharge afterglows to boost collection efficiencies of larger-scale electrostatic precipitators.

Figure 1 .
Figure 1.Experimental setup for (A) particle generation via atomization of a salt solution, followed by neutralization and removal of remaining charged particles; (B) particle charging in the afterglow of an argon RF discharge; (C) measurement of particle charging by aerosol polarity (top path) and specific charging state via TDMA (bottom path).

Figure 2 .
Figure 2. Charge fractions (a) and polarity (b) of particles passing through spatial afterglow at different plasma-aerosol spacings (electrode distance of 4 and 10 cm).Polarities shown in (b) are for a 60 nm particle.Data displayed for a fixed aerosol flow rate of 0.25 L m −1 and plasma power of 6 W.

Figure 3 .
Figure 3. Charge fractions (a) and polarity (b) of particles passing through spatial afterglow at two different aerosol flow rates.Polarities shown in (b) are for a 60 nm particle.Data displayed for a fixed electrode distance of 10 cm and plasma power of 6 W.

Figure 4 .
Figure 4. Charge fractions (a) and polarity (b) of particles passing through spatial afterglow at different plasma powers.Polarities shown in (b) are for a 60 nm particle.Data displayed for a fixed electrode distance of 10 cm and aerosol flow rate of 1 L m −1 .

Figure 5 .
Figure 5. Particle polarities of 60 nm particles as a function of DC bias for argon, air, and nitrogen.Operating parameters set at 4 cm electrode distance, 1 L m −1 aerosol and plasma flow, and 6 W plasma power.

Figure 6 .
Figure 6.Charging of particles passing through glow plasma afterglow at optimized operating conditions (10 cm plasma-aerosol distance, 0.25 L m −1 aerosol flow, 6 W plasma power).(A) Charge fraction at optimized conditions compared to lab scale [10] and a pilot scale [40] ESP collection efficiencies.(B) TDMA measurements of particle charge distributions of 45 and 90 nm particles at optimized conditions.Average charge 〈q〉 shown in inset.

Table 1 .
Test Plan.An SMPS was used to measure the charging efficiency and polarity in each condition.Base conditions are 1 L m −1 aerosol and plasma flow rate, 4 cm electrode distance, and 6 W plasma power.

Table 2 .
Characteristic times of plasma cross-flow system.