Key Factors Controlling Number Concentrations of Non-charged Particles: An Experimental Study

The charging state of submicron particles strongly affects the number of particles deposited on the inner surface of human airways during inhalation, and an increase in this number leads to adverse effects on human health. However, the key factors controlling the charging state of particles are poorly understood because of the limited measurements of the number concentrations of charged submicron particles. In order to experimentally investigate the key factors controlling the number concentration of charged particles, the technique for accurately measuring the number concentration of non-charged particles ( N non-charge ) was improved based on a system consisting of optical particle counters and a parallel-plate particle separator to which a high voltage (1.0–1.5 kV) was applied. The ratio of the N non-charge to the total number concentration of particles ( N total ), with diameters 0.3–0.5 µ m, was used as a metric of the number fraction of non-charged particles. A decrease in the N non-charge / N total ratio with an increase in the negative ion concentration ( n ion– ) and the n ion– / N total ratio under stable meteorological conditions showed an increase in the number of charged particles by collisions between aerosols and ions. Under fluctuating humidity conditions, the N non-charge / N total ratio also decreased with an increase in the n ion– and the n ion– / N total ratio. Particularly, the decrease rate in the N non-charge / N total ratio was greater at higher n ion– and n ion– / N total ratio values than that under stable meteorological conditions. This suggests that the generation of water vapor drives an increase in ion concentration, leading to frequent collisions between excessive negative ions and aerosols. A quantitative understanding of the changes in the number fraction of charged particles with increasing humidity and n ion– can assist in estimating the number of particles deposited on the inner surface of human airways


INTRODUCTION
Particulate matter (PM) deposited on the inner surface of human airways causes respiratory and cardiovascular diseases, preterm birth, and an increase in morbidity and mortality rates (Apte et al., 2015;Ching and Kajino, 2018;Jia et al., 2020;Manojkumar et al., 2018;Russell and Brunekreef, 2009).Particles with diameters less than 2.5 µm (hereafter, PM2.5) can penetrate deeply into the lungs and cause serious damage to the human respiratory system (Xing et al., 2016).Therefore, a quantitative understanding of the aerosol particle deposition process in human airways is required to evaluate their health effects.
In general, aerosol particles are deposited in the respiratory tract through several processes, including Brownian diffusion, interception, inertial impaction, and electrostatic attraction (Azhdarzadeh et al., 2014;Bowes and Swift, 1989;Kulkarni et al., 2011).Cohen et al. (1998) demonstrated that the number of singly charged 125-nm-diameter particles deposited onto the surfaces of human airways was approximately six times that of non-charged particles.Experimental and theoretical studies have shown that the deposition efficiency of charged particles in the oral-extrathoracic and nasal-laryngeal airways depends on the particle size, amount of charge, and flow rate of the particles (Azhdarzadeh et al., 2014;Xi et al., 2014).Simulations by Majid et al. (2012) indicated that the deposition efficiency of charged particles in the tracheobronchial region increased with the particle charge in the submicron size range.Therefore, a quantitative understanding of the charging state of aerosol particles is required to accurately estimate the number of particles deposited in human airways.
Aerosol particles are charged by collision with atmospheric ions (Harrison and Carslaw, 2003;Hoppel and Frick, 1986), which are primarily generated by radon decay, gamma radiation, and cosmic radiation (Hirsikko et al., 2011).The charge distribution of the particles is known to reach a stationary distribution in a short time (Fuchs, 1963;Gunn, 1967).The bipolar charging state of atmospheric particles under stationary conditions is typically asymmetric toward negative values (Johnson et al., 2020;Wiedensohler, 1988).The asymmetry of the charge distribution toward negative polarity is due to negative ions having greater mobility than positive ions and, thus, a higher likelihood of attachment to particles (Hoppel and Frick, 1990).According to observations of the number of positively and negatively charged particles in Korea, Antarctica, and the Arctic during spring (Lee et al., 2021), the variation in the ratio of positively charged particles to negatively charged particles is primarily attributed to cosmic rays, precipitation, and solar activity.This suggests that the charging state of the particles in the atmosphere differs from the well-known Boltzmann change distribution.However, techniques for measuring the charge distributions of submicron particles and the number concentrations of charged or non-charged particles are very limited.
In recent studies, a technique has been developed to estimate the charge distribution of particles and the number concentrations of charged and non-charged particles (Ncharge and Nnon-charge, respectively) at 15-min time intervals in the 0.3-0.5 µm optical diameter (D) range, based on a system consisting of a parallel-plate particle separator (Keio-Measurement system of Aerosol Charging State; K-MACS) and optical particle counters (OPCs) (Iwata et al., 2019;Okuda et al., 2015;Yonemichi et al., 2019).The particles pass through a uniform electric field in K-MACS, which consists of two electrode plates that produce a voltage; the charged particles move in the direction of the electrode plates according to their electrical mobility, which is controlled by the magnitude of the voltage applied to the K-MACS (which is proportional to that of the electric field).Therefore, charged and non-charged particles are separated in K-MACS.However, previous studies have not applied a sufficient voltage to clearly separate charged and non-charged particles; thus, uncertainties exist in the estimates of Ncharge and Nnon-charge in the K-MACS/OPC system.
Despite these measurement uncertainties, previous studies have used the K-MACS/OPC system to investigate the key factors controlling the charging state of atmospheric submicron particles (He et al., 2020;Iwata et al., 2019).Iwata et al. (2019) (Zhang et al., 2016), making the humidity ratio an important parameter for controlling the charging state of the particles.These results are important for understanding the behavior of charged particles under atmospheric conditions with high RH.However, identifying the key factors controlling the charging state of particles is difficult because the charging state of particles is controlled by multiple parameters in ambient air (i.e., RH, ion concentration, and PM2.5).Therefore, a detailed experiment investigating the number of charged particles is required based on a technique for directly measuring the number concentration of non-charged or charged particles.
In this study, the technique for accurately measuring Nnon-charge was improved by increasing the applied voltage of the K-MACS and extracting only non-charged particles from the K-MACS with a high time resolution.This technique facilitated a decrease in previous measurement uncertainties, as shown by Iwata et al. (2019).Using this technique, we experimentally investigated the relationships between the number concentration of non-charged submicron particles, atmospheric ion concentrations, and meteorological parameters (e.g., RH, volumetric humidity, and humidity ratio) under conditions of stable or fluctuating humidity in the sampling line to elucidate the key factors controlling the number fraction of non-charged submicron particles.

Previous K-MACS/OPC system
Previous studies used a system described elsewhere consisting of K-MACS and OPCs to estimate the charge distribution of particles and derive the Ncharge and Nnon-charge (Iwata et al., 2019;Okuda et al., 2015;Yonemichi et al., 2019).Briefly, aerosol particles were introduced into the K-MACS at a sample flow rate of 0.3 liters per minute (L min -1 ), together with a sheath flow of 1.2 L min -1 .The K-MACS consisted of two parallel electrode plates (length: 200 mm; width: 100 mm) separated by a distance of 5 mm, and a constant voltage (200 V or 400 V) was applied between the two electrode plates using a Millikan Elementary Charger (MLD-5, Shimadzu).The positively and negatively charged particles moved in the direction of the electrode plate with negative and positive voltages, respectively, whereas the non-charged particles moved only in the middle layer.Charged and non-charged particles were extracted from the side layers of the electrode plate (with an applied positive or negative voltage) and from the middle layer of the K-MACS, respectively.After the extraction, the total number concentration (Ntotal) of the particles for each size range (0.3-0.5 µm, 0.5-1.0µm, 1.0-2.0µm, 2.0-5.0 µm, and > 5.0 µm) in each layer was measured by OPCs.Iwata et al. (2019) summarized the procedure for estimating the Nnon-charge (D = 0.3-0.5 µm) and Ncharge (D = 0.3-0.5 µm).Briefly, when charged particles with electrical mobility pass through a uniform electric field in K-MACS, they move toward the electrode plate at a constant speed.The speed in the direction of the electrode plates was determined by multiplying the magnitude of the uniform electric field with the electrical mobility of the charged particles.In addition, the residence time of the charged particles between the electrode plates is also determined by the length of the electrode plate (200 mm), the flow rate (1.5 L min -1 ), and the cross-sectional area between the electrode plates (5.0 × 10 -4 m 2 ).Therefore, the distance moved in the direction of the electrode plate can be derived from the product of the moving speed and the residence time.Iwata et al. (2019) assumed the particle diameter to be 0.387 µm (the geometric mean value of 0.3-0.5 µm D particles, hereafter D = 0.378 µm) and the shape of the particle to be spherical to calculate the moving distance of the particle with a charge number (p).Iwata et al. (2019) also assumed that the space in the region of the electrode plate of the middle layer was divided into five and the position to which the particles with each p were transported at the residence time was determined by forward trajectory analysis; thus particles with a p starting from the edge of the electrode plate finally reached one of the three layers (i.e., the side layers of the electrode plate with an applied positive or negative voltage and the middle layer) or were deposited on the walls of the electrode plates in the K-MACS, depending on the electrical mobility of particles with a p.By calculating the number of particles with a p transported to each layer, they estimated the probability distribution of charged particles for each layer.The probability distribution is represented as the ratio of number concentration of particles with p (Np) (D = 0.378 µm) to Ntotal.The Np was calculated from the product of the measured Ntotal and calculated Np/Ntotal ratio (D = 0.378 µm) for each layer.Consequently, the K-MACS/OPC system can estimate the charge distribution of the particles.However, because of the lower applied voltages of 200 V and 400 V, the charged and non-charged particles were not completely separated in this system, leading to uncertainties of approximately 20-30% in the estimate of Np on average.This also caused uncertainties in the estimation of the particle charge distribution.To minimize these uncertainties in the estimate of Nnon-charge, the development of the K-MACS is required, as this removes all charged particles within the 0.3-0.5 µm D range from the middle layer.This is described in detail in Section 2.1.2.

Modified K-MACS/OPC system
The magnitude of the electric field in the K-MACS increased with increasing in voltage applied to the electrode plates.This indicates that the charged particles can move significantly toward the electrode plates because the distance moved by the charged particles is strongly controlled by the magnitude of the electric field (Section 2.1.1).This suggests that charged and non-charged particles should be clearly separated in K-MACS at a higher applied voltage.In order to estimate the applied voltage that can remove all charged particles (D = 0.3-0.5 µm) from the particles in the middle layer of the K-MACS (hereafter, K-MACS I) used in Section 2.1.1,the Np/Ntotal ratio was calculated in the middle layer for particles with D = 0.1-3.0µm by using a flow rate of 1.5 L min -1 and several applied voltages (0.0-1.5 kV).When the Np/Ntotal ratio for particles with p = 0 reached unity, all the particles extracted from the middle layer were non-charged.This calculation result indicated that the non-charged particles with D < 0.5 µm can be extracted from the middle layer of the K-MACS I by applying voltages larger than ~852 V (Figs. 1(c) and 1(e)).
In this study, a K-MACS similar to the one mentioned above (hereafter, K-MACS II), consisting of one classification layer within a distance interval of 5 mm between the two parallel plates (length 800 mm; width 225 mm), was developed to simultaneously measure the Nnon-charge (D = 0.3-0.5 µm) in ambient and sample air (Fig. 1(b)).Using the same method, the Np/Ntotal ratio was calculated for the middle layer within the 0.1-3.0µm D range and at a flow rate of 3.0 L min -1 for the several applied voltages (0-1.0 kV).The calculation result indicated that all particles (D < 0.5 µm) extracted from the middle layer of the K-MACS II are non-charged particles at applied voltages larger than ~568 V (Fig. 1(d)).
Considering the above calculations, a voltage of 1.5 kV was applied to the K-MACS I and 1.0 kV was applied to the K-MACS II (Figs.1(e) and 1(f)) by using a high-voltage power supply (HJPM-2N3, MATSUSADA Precision Inc.) to extract only the non-charged particles with D = 0.3-0.5 µm from the two K-MACSs.Immediately after the classification, the Nnon-charge (D = 0.3-0.5 µm) was measured using OPC (KC-01D or KC-01E, RION) at 2-min time intervals (Figs.1(a) and 1(b)).We simultaneously measured Ntotal (D = 0.3-0.5 µm) in front of each K-MACS to derive the Nnon-charge/Ntotal ratio.
The particles with D > 0.5 µm range (i.e., D = 0.5-1.0µm, 1.0-2.0µm, 2.0-5.0 µm, and > 5.0 µm) cannot be completely separated into charged and non-charged particles at the higher applied voltages (Figs.1(e-f)), because the electrical mobility of the larger particles is very low.To minimize the measurement uncertainties of the Nnon-charge/Ntotal ratio, we used the Nnon-charge/Ntotal ratio for the 0.3-0.5 µm D range in this study.

Measurement uncertainties of size-resolved number concentration by OPCs
When the particles passed through the laser beam in the OPC, a pulse of scattered light was detected at certain scattering angles.The pulse number means the particle count and the pulse height is related to the particle size.In general, using a polystyrene latex sphere particle with a refractive index of 1.59, the threshold D (i.e., 0.3 µm, 0.5 µm, 1.0 µm, 2.0 µm, and 5.0 µm) are determined.Thus, the OPC can derive the size-resolved number concentration of particles by measuring both number and height of the pulse.Note that the size-resolved number concentration of atmospheric particles has some uncertainties because atmospheric aerosols have several optical properties (e.g., refractive index and shape) and the threshold D may change (Miura et al., 1981).In this study, we used the D ranges (i.e., 0.3-0.5 µm, 0.5-1.0µm, 1.0-2.0µm 2.0-5.0 µm, and > 5.0 µm) by assuming that the spherical particles with a refractive index of 1.59 are detected.
Coincidence-counting errors occur when two or more particles pass through the sensing volume and multiple scattered lights are simultaneously detected by a photomultiplier tube at the same time (Miura et al., 1981), leading to an underestimate of the number concentration of the particles.Previous studies evaluated the coincidence counting errors by estimating the true number concentration of particles, which is calculated from the measured number concentration of particles and the sensing volume of the Rion KC-01E OPC (5.0 × 10 -4 cm -3 ) using a Poisson distribution (Iwamoto et al., 2018).The coincidence errors were within 5% when the particle number concentration was 100 cm -3 (Iwamoto et al., 2018).In our study, the maximum value of the number concentration of atmospheric particles was 133 cm -3 for the 0.3-0.5 µm D range, corresponding to coincidence errors of approximately 7%.The coincidence errors were corrected using the measured Ntotal values.
To evaluate the performance of the four OPCs we used, we simultaneously measured the Ntotal in the atmosphere for each D range (0.3-0.5 µm, 0.5-1.0µm, 1.0-2.0µm, 2.0-5.0 µm, and > 5.0 µm) to be compared.The average Ntotal values for each D range agreed to within approximately 10%, 20%, 20%, 30%, and 20%, respectively.The instrumental error correction was performed on the experimental dataset.
Gravitational deposition, Brownian diffusion, and inertial deposition are the main causes of the loss of aerosol particles inside K-MACS I, K-MACS II, and the sampling line (Kulkarni et al., 2011).
To estimate the particle loss rate during transmission through the sampling line, the Ntotal (D = 0.3-0.5 µm) before and after passage through the two K-MACSs were compared, with no applied voltage for each experimental condition (Figs.1(a) and 1(b)).Considering the error correction between both instruments, the particle loss rates for D = 0.3-0.5 µm were within 1% in the K-MACS I/OPC system and 7% in the K-MACS II/OPC system.
Uncertainties were observed in the measurement of the number concentration of particles in the OPCs, as mentioned above.Therefore, in our data analysis, we focused on the trend in Ntotal (0.3-0.5 µm), non-charged particle fraction, humidity, and ion concentration to interpret the controlling factors of the Nnon-charge/Ntotal ratio.

Experiment 1
The first experiment was conducted to elucidate the relationship between the charging state of the particles and ion concentrations using K-MACS II and OPC under stable meteorological conditions (e.g., temperature and RH).This study was conducted at the Yagami campus of Keio University, Japan (35.6°N, 139.7°E, 25 m above sea level), located in the Tokyo Metropolitan Area (Iwata et al., 2019), from 10:30 to 14:00 JST on July 18, 2019.Fig. 2(a) shows the experimental setup.Ions were generated in a sampling line using a bipolar diffusion charge neutralizer (Americium-241; 241 Am) and thoroughly mixed with the ambient aerosol particles in the chamber.Ion concentrations were controlled, as were the ambient aerosol number concentrations by changing the flow rates of introduced ions and ambient air at about 30-min intervals, which ranged from 0.0 to 5.0 L min -1 and from 2.0 to 7.0 L min -1 , respectively.The total flow rate was maintained at 7.0 L min -1 .The negative-ion concentration (nion-) in the chamber was measured using an ion counter (COM-3800V; COM System, Inc.).In addition, meteorological instruments (TR-72wb and TR-73U) were used to Fig. 2. Experimental setups for investigating the relationships between (a) charging state of particles and ion concentrations under constant temperature and relative humidity conditions and (b) charging state of particles, ion concentrations, and volumetric humidity under variable relative humidity conditions.measure the RH and temperature in ambient air and sample air at 10-s time intervals.The precision of the above-mentioned measurements (± standard deviation) was ± 5.0% and ± 0.5°C, respectively, and the resolutions of those measurements were 1.0% and 0.1°C, respectively.

Experiment 2
Further experiments were conducted to elucidate the relationships between the charging state of the particles, ion concentration, and meteorological parameters when the RH values were significantly changed in the same manner as in Experiment 1, from 12:04 to 17:34 JST on January 22, 2020.The experimental setup is shown in Fig. 2(b).
To control the RH, the sample air was dried via passage through a diffusion dryer and wet by passage through a bubbler at a total flow rate of 2.0 L min -1 .The flow rates were manually controlled at 0.5 L min -1 increments, within the range of 0.0-2.0L min -1 at 30-min intervals.The 241 Am neutralizer was used to stably generate ions at a constant flow rate of 5.0 L min -1 and mix the ions with ambient aerosols.In the air samples, the nionand meteorological parameters were measured using an ion counter and meteorological instruments, respectively, as discussed in Section 2.2.The Ntotal for each D range (0.3-0.5 µm, 0.5-1.0µm, 1.0-2.0µm, 2.0-5.0 µm, and > 5.0 µm) and Nnon-charge for the 0.3-0.5 µm D range in sample air were measured by a system consisting of the K-MACS I and OPC (Fig. 1(a)), and those in ambient air were simultaneously measured as references using the K-MACS II/OPC system (Fig. 1(b)).
The sample flow rate was manually controlled in order to change the ratio of the nionto Ntotal (D = 0.3-0.5 µm).The Ntotal in sample air decreased with an increase in the nion-(Figs.3(h) and 3(i)).The increase in the nion-/Ntotal ratio could result in frequent collisions between aerosols and ions.Note that a fraction of the negative ions may have disappeared because of collisions between the negative and positive ions (Harrison and Carslaw, 2003;Takahashi, 2003).In this study, the clear decrease of the Nnon-charge/Ntotal ratio (D = 0.3-0.5 µm) with an increase in the nionand nion-/Ntotal ratio (Figs.3(i) and 3(j)) was demonstrated, indicating that the number of charged particles increased because of collisions between aerosols and ions in the chamber.However, the degree of the increase in the number of charged particles may change because of the mixing of ions and particles, which is controlled by the flow rate.In fact, the average values of Nnon-charge, Ntotal, Nnon-charge/Ntotal ratio, and nion-/Ntotal ratio varied within 38%, 31%, 7.0%, and 27%, respectively, over each approximately 30-min interval (Table S1(a)), partly arising from the unstable flow rates and insufficient mixing between ions and particles.Approximately 15 min after switching the flow rate, the variabilities of Nnon-charge, Ntotal, Nnon-charge/Ntotal ratio, and nion-/Ntotal ratio decreased to approximately 5% of their average values (Table S1(b)).The small variabilities may indicate a steady flow rate and sufficient mixing between the ions and particles.Therefore, we used the dataset to investigate the relationships between nionand Nnon-charge/Ntotal ratio and between nion-/Ntotal ratio and Nnon-charge/Ntotal ratio.The rate of decrease in the Nnon-charge/Ntotal ratio was much larger at the lower values of the nionand the nion-/Ntotal ratio, indicating efficient increases in the charged particles (Fig. 4).
After the particles pass through a neutralizer, their charge distribution is known to shift toward negative particles (Flagan, 1998;Wiedensohler, 1988).This is because the electron mobility of negative ions is larger than that of positive ions.Iwata et al. (2019) used a system consisting of a K-MACS and three OPCs to estimate the charge distribution of particles in the atmosphere and that of particles neutralized by 241 Am at the Yagami campus, Japan.They showed that both the distributions of the charge numbers of the particles shifted toward the negative region.In our experiment, when a large number of ions were introduced into the chamber by passing through the 241 Am neutralizer, collisions between the aerosols and positive or negative ions occurred.Considering these results, the charge distribution of the particles may change to one of the following charge distributions, affecting the non-charged particle fraction: 1) the charge distribution has exactly the same broadness, yet shifts toward a more negative charge; 2) the charge distribution has exactly the same mode charge number, yet the width of the charge distribution  b) between the nion−/Ntotal (0.3-0.5 µm) ratio and the non-charged particle fraction for particles within the 0.3-0.5 µm diameter range for all data and selected data.These data represent 2-min averages.The regression curves were presented as the power functions, determined by the least squares method for the 2-min average selected data.The experimental setup corresponds to Fig. 2(a).expands; and 3) the charge distribution shifts toward the negative and the spread of the charge distribution of the particles is large.However, it was difficult to quantitatively elucidate the changes in the charge distribution of the particles in this experiment.
Our experiment suggests that the Nnon-charge/Ntotal ratio changes with the frequency of collisions between the aerosols and ions.The rate of decrease in the Nnon-charge/Ntotal ratio may be controlled by changes in the average charge number and/or width of the charge distribution of the particles.
The variation of the nionin the sample air was similar to that of the RH, volumetric humidity, and humidity ratio, and the nionwas correlated with the RH (Fig. 6(a); r 2 = 0.79), the volumetric humidity (Fig. 6(b); r 2 = 0.83), and the humidity ratio (Fig. 6(c); r 2 = 0.83).In this study, water droplets were generated from the water surface by bubbling water through a bubbler; these droplets broke near the water surface.Previous studies showed that after the droplets break up, smaller fragments carry negatively charged OH -, while the remaining bigger fragments carry larger proton clusters (e.g., H3O + (H2O)20), and the larger positively charged droplets are deposited on the water surface (Kolarž et al., 2012;Lenard, 1915;Luts et al., 2009;Parts et al., 2007).Therefore, these excellent correlations indicate the generation of a large number of ions, including OH -, by vigorously bubbling water on the water surface using a bubbler.Zheng et al. (2019) demonstrated that the charge of particles > 0.1 µm in size increased by more than 50% when the RH increased from 30% to 80%, due to the enhancement in the concentration of ions.This implies that RH is an important determinant of the generation of charged particles in ambient air.However, Iwata et al. (2019) measured the Nnon-charge/Ntotal ratio for the atmospheric particles with a D = 0.3-0.5 µm range at Yokohama during 2017-2018, to elucidate that the volumetric humidity is an important determinant of the charging state of atmospheric particles.In this study, the Nnon-charge/Ntotal ratio for the particles with D = 0.3-0.5 µm correlated well with the nion-(Fig.6(d); r 2 = 0.69), the nion-/Ntotal ratio (D = 0.3-0.5 µm) (Fig. 6(e); r 2 = 0.49), and the temperature (Fig. 6(f); r 2 = 0.52).The Nnon-charge/Ntotal ratio for the particles with D = 0.3-0.5 µm showed an even better correlation with the RH (Fig. 6(g); r 2 = 0.86), volumetric humidity (Fig. 6(h); r 2 = 0.84), and humidity ratio (Fig. 6(i); r 2 = 0.84).Even in an environment where the humidity in the sample air was considerably changed, the nionwas a key factor in generating the charged particles via collisions with ions and aerosols, which was consistent with the result in experiment 1.
In experiment 1, the decrease in the Nnon-charge/Ntotal ratio for the particles with D = 0.3-0.5 µm, slowed down when increasing the ratio of nionto Ntotal (D = 0.3-0.5 µm).In order to compare the degree of decrease in the Nnon-charge/Ntotal ratios of experiments 1 and 2, the Nnon-charge/Ntotal ratio over the range of nion-(1000-3000 cm -3 ) and over that of nion-/Ntotal (where D = 0.3-0.5 µm, 80-300) was calculated by using the least-squares fitting function (Figs. 4(a), 4(b), 6(d), and 6(e)) and normalized at the nionof 1000 cm -3 and nion-/Ntotal ratio of 80 (Fig. S2).As the nionand nion-/Ntotal The derived non-charged particle fractions against (d) nion−, (e) ratio (extracted from the middle layer of the K-MACS) of nion− to the total number concentration of particles within the 0.3-0.5 µm range (Ntotal), (f) temperature, (g) RH, (h) volumetric humidity, and (k) humidity ratio in sample air for all data and selected data.The regression lines were determined by the least squares method for the 2-min average selected data.The experimental setup corresponds to Fig. 2(b).ratio increased, the charged particles were more efficiently produced under the fluctuating humidity condition (Experiment 2) than the stable meteorological condition (Experiment 1).This calculation suggests that an increase in water vapor drives the generation of ions, leading to frequent collisions between the ions and aerosols.He et al. (2019) reported that an increase in the humidity ratio led to the redistribution of dissolved ions in the water film on the surface of the particles by the liquid bridging force, and the charged particles were generated by the separation of particles.This process also drives the generation of charged particles in a high-humidity environment; therefore, it may be important for interpreting the differences in the non-charged particle fraction under stable meteorological conditions and fluctuating humidity conditions.

CONCLUSIONS
The concentration of charged particles is a key parameter that controls the deposition quantity in human airways, which may affect human health.However, only a few experimental studies have investigated the key factors controlling the number of charged particles based on measurements of charged particles with high time resolution.In this study, the K-MACS were first improved by applying voltages of 1.5 and 1.0 kV between the two parallel plates at flow rates of 1.5 and 3.0 L min -1 , respectively, to extract only the non-charged particles with an optical particle diameter (D) of 0.3-0.5 µm.Then, the K-MACS and OPCs were combined to measure the number concentrations of non-charged particles (Nnon-charge) with D = 0.3-0.5 µm at 2-min time intervals under conditions of constant and significantly fluctuating relative humidity.
The increase in the ratio of the number concentration of negative ions (nion-) to Ntotal (D = 0.3-0.5 µm) led to a decrease in the Nnon-charge/Ntotal ratio under conditions of stable meteorological parameters.The results demonstrated that charged submicron particles were efficiently generated, partly due to the increases in the nion-, hence resulting in the high collision frequency between aerosols and ions.The degree of increase in the number of charged particles may be affected by changes in the charge distribution of the particles under stable meteorological parameters (e.g., humidity).
The nionincreased with humidity because of the efficient formation of negative ions from the water surface by bubbling water.Frequent collisions between negative ions and submicron-sized aerosol particles increased the number of negatively charged particles, leading to a decrease in the non-charged particle fraction.We therefore demonstrated that the charging state of particles was strongly influenced by the increase of the nionand was strongly controlled by the RH, volumetric humidity, and humidity ratio.The rate of decrease in the number of non-charged particles was larger under fluctuating humidity conditions than under stable humidity conditions, suggesting the efficient generation of charged particles by collisions between mainly negative ions and aerosols via water vapor.Iwata et al. (2019) found that the atmospheric charging state of particles was strongly controlled by volumetric humidity, whereas the two experiments in this study demonstrated that the noncharged particle fraction was strongly related to ion concentrations and meteorological characteristics, such as RH, volumetric humidity, and humidity ratio.To further understand the process that changes the charge distribution of particles, measurement of the charge distribution of particles and an experimental investigation of the relationship between the charging state of particles and ion concentrations are required.
To accurately estimate the number of charged particles deposited in the respiratory tract, further studies of the deposition mechanisms of charged particles are required.However, considering the high volumetric humidity in human airways, experiments investigating the behavior of water vapor and charged particles are useful to estimate the number of particles deposited on the inner surface of human airways.The findings of this study may not only explain the variation in the number concentrations of non-charged particles in ambient air but also provide an understanding of the changes in the non-charged particle fraction in human airways.

NOMENCLATURE
derived the Ncharge (D = 0.3-0.5 µm) and Nnon-charge (D = 0.3-0.5 µm) at the Yagami campus of Keio University (35.56°N, 139.65°E),Yokohama, Japan, from April 2017 to February 2018.They used the ratio of Nnon-charge (D = 0.3-0.5 µm) to the total number concentration (Ntotal) of particles, with diameters of 0.3-0.5 µm, as an indicator of the charging states of submicron particles, indicating that the Nnon-charge/Ntotal ratio is influenced more by the volumetric humidity than the relative humidity (RH) and temperature in ambient air.He et al. (2020) also derived the Ncharge (D = 0.3-10 µm) and Nnon-charge (D = 0.3-10 µm) at Xi'an Jiaotong University in China from June 2018 to March 2019, showing that the estimated charge amount of an atmospheric particle is strongly related to PM2.5 concentrations and RH.Simulations by He et al. (2019) suggested that agglomeration velocity of particles increases with the atmospheric humidity because of the increased liquid bridging force at the contact interface.The liquid bridge redistributes the dissolved ions in a water film onto the surface of the particles

Fig. 1 .
Fig. 1.Experimental setups for measuring non-charged aerosol number concentrations and total aerosol number concentrations using (a) K-MACS I and (b) K-MACS II, to which voltages of 1.5 kV at 1.5 liters per minute (L min −1 ) and 1.0 kV at 3.0 L min −1 were applied, respectively.Distribution of optical diameters and charge numbers of particles passing through the K-MACS I at the applied voltages of (c) ~852 V and (e) 1.5 kV in Fig. 1(a).Color plots indicate the fraction of the number concentration of particles with charge number p (Np) normalized by the total number concentration for each size, extracted from the middle layer of the K-MACS.(d, f) Is the same as (c) and (e), but represents the K-MACS II at the applied voltages of (d) ~568 V and (f) 1.0 kV in Fig. 1(b).

Fig. 4 .
Fig. 4. (a) Correlation between negative ion concentration (nion−) and non-charged particle fraction (i.e., Nnon-charge/Ntotal ratio) and (b) between the nion−/Ntotal (0.3-0.5 µm) ratio and the non-charged particle fraction for particles within the 0.3-0.5 µm diameter range for all data and selected data.These data represent 2-min averages.The regression curves were presented as the power functions, determined by the least squares method for the 2-min average selected data.The experimental setup corresponds to Fig.2(a).

Fig. 6 .
Fig. 6.Relationships of negative ion concentration nion− for (a) relative humidity RH, (b) volumetric humidity, and (c) humidity ratio.The derived non-charged particle fractions against (d) nion−, (e) ratio (extracted from the middle layer of the K-MACS) of nion− to the total number concentration of particles within the 0.3-0.5 µm range (Ntotal), (f) temperature, (g) RH, (h) volumetric humidity, and (k) humidity ratio in sample air for all data and selected data.The regression lines were determined by the least squares method for the 2-min average selected data.The experimental setup corresponds to Fig. 2(b).