Design, optimization, and evaluation of a wet electrostatic precipitator (ESP) for aerosol collection

Abstract

In this study, we developed, optimized, and evaluated in lab and field experiments a wet electrostatic precipitator (ESP) for the collection of ambient PM_2.5 (particulate matter with aerodynamic diameter < 2.5 μm) into ultrapure water by applying an electrostatic charge to the particles. We operated the wet ESP at different flow rates and voltages to identify the optimal operating conditions. According to our experimental measurements, a flow rate of 125 lpm and an applied positive voltage of 11 kV resulted in a lower ozone generation of 133 ppb and a particle collection efficiency exceeding 80–90% in all size ranges. For the field tests, the wet ESP was compared with the versatile aerosol concentration enrichment system (VACES) connected to a BioSampler, a PTFE filter sampler, and an OC/EC analyzer (Sunset Laboratory Inc., USA) as a reference. The chemical analysis results indicated the wet ESP concentrations of metal and trace elements were in very good agreement with those measured by the VACES/BioSampler and PTFE filter sampler. Moreover, our results showed comparable total organic carbon (TOC) concentrations measured by the wet ESP, BioSampler, and OC/EC analyzer, while somewhat lower TOC concentrations were measured by the PTFE filter sampler, possibly due to the limitations of extracting water-insoluble organic carbon (WIOC) from a dry substrate in the latter sampler. The comparable TOC content in the wet ESP and BioSampler samples differs from previous findings that showed higher TOC content in BioSampler samples compared to those collected by dry ESP. The results of the Dithiothreitol (DTT) assay showed comparable DTT activity in the VACES/BioSampler and wet ESP PM samples while slightly lower in the PTFE filter samples. Overall, our results suggest that the wet ESP could be a promising alternative to other conventional sampling methods.

1. Introduction

Particulate matter (PM) has been associated with many toxic health outcomes, including respiratory and cardiovascular diseases (Karimi et al., 2019; Klompmaker et al., 2022), lung cancer (L. Yang et al., 2020), and neurodegenerative disorders (Wang et al., 2017). To understand the impact of PM on human health, aerosols that accurately represent ambient PM concentrations must be used in controlled toxicological in vivo and in vitro studies. Nevertheless, the complex properties of airborne PM pose a formidable challenge to producing samples in the laboratory that represent ambient PM concentrations precisely (Anderson et al., 2012; Filep et al., 2016). Therefore, utilizing methods that can capture and represent ambient PM concentrations with high efficiency is crucial to evaluate the potential health impacts of PM exposure.

Assessing the health effects of exposure to PM in controlled human or animal exposure studies may be difficult because ambient levels of PM are often too low to cause noticeable health effects (Jung et al., 2010; Liu et al., 2014a). PM concentrators and samplers are both widely used in air quality research to measure and collect PM from the ambient environment. Ambient PM concentrators have been developed to overcome such issues by increasing the concentration of ambient PM while preserving the physio-chemical properties of the aerosols, enabling researchers to assess potential health effects at higher and more realistic concentrations (Demokritou et al., 2003; Geller et al., 2005b; Han et al., 2009b; Khlystov et al., 2005; Romay et al., 2010). The versatile aerosol concentration enrichment system (VACES) is an example of an ambient PM concentrator that has been widely used in the literature. As described by Kim et al. (2001a and 2001b), particles are concentrated by 10–30 times using a saturation-condensation system that grows particles to 3–4 μm droplets, followed by a diffusion dryer to remove excess moisture and return the concentrated particles to their original size before supplying them for in vivo exposures or in vitro toxicity studies. The VACES system can be connected to a BioSampler to make highly concentrated liquid suspensions for in vitro toxicity studies. This combination allows the efficient collection of sufficient concentrations of ambient particles for toxicity studies (Kim et al., 2001a; Kim et al., 2001b; Pakbin et al., 2011; Sioutas et al., 1999). However, ambient PM concentrators can be expensive and cumbersome due to the multiple processes involved in their operation.

PM samplers, on the other hand, collect particles on filters or other substrates for further analysis. Examples of PM samplers include filter-based high-volume samplers, which operate at high flow rates, collecting PM on a filter after removing a specific size range of PM through a pre-selective impactor inlet (Misra et al., 2002; Patel et al., 2021; Sugita et al., 2019). Other PM samplers, such as the Andersen cascade impactor sampler, PM_2.5 Very Sharp Cut Cyclone (VSCC), and PM_2.5 Well Impactor Ninety-Six (WINS) employ impactors and cyclones to segregate particles based on their aerodynamic diameter before collecting them on filters or substrates for further analysis (Patel & Aggarwal, 2022). While these samplers facilitate the collection of substantial PM in relatively short time periods, they may also produce both positive and negative artifacts for organic and inorganic species. Positive artifacts may arise from the absorption of gaseous organic carbon (OC) or inorganic gases by the filter substrate and collected particles (Cheng et al., 2012; Maimone et al., 2011). On the other hand, the occurrence of negative artifacts during sampling may be attributed to evaporative loss of semi-volatile material (SVM) (Liu et al., 2014b; Mader et al., 2003).

Electrostatic precipitators (ESPs), which use electrical charges to collect aerosol particles (Hinds & Zhu, 2022), may offer a solution to the issues associated with biases in the abovementioned methods for PM collection due to evaporative and adsorptive artifacts (Cardello et al., 2002; Volckens & Leith, 2003). According to Volckens & Leith (2002a, 2002b), ESP samplers may be a more accurate method for collecting and measuring semi-volatile aerosols, such as alkanes and polycyclic aromatic hydrocarbons (PAHs), compared to the filter samplers. In their experiments, ESP was found to have significantly lower levels of adsorption of gas-phase semi-volatiles, with factors ranging from 5–100, compared to filter-based samplers. This is because ESP minimizes the surface area of collected particles in comparison to the surface area of conventional filters, which reduces the disruption of the gas-particle equilibrium that can lead to adsorption (Volckens & Leith, 2002b). The ESP also showed low levels of evaporation of particles which were approximately 2.3 times lower than TFE-coated glass fiber filters. This was again attributed to the reduced surface area of the collected particles in the ESP, which reduces the rate of evaporation that can occur. Additionally, wet ESPs may offer several advantages over dry ESPs for PM collection (Lin et al., 2010). Wet ESPs are particularly appropriate for collecting adhesive or corrosive particles and are known to be effective at capturing fine aerosols (Chen et al., 2014; Ruttanachot et al., 2011). Furthermore, because PM is collected in a slurry in wet ESPs, they may be more effective for the collection of water-insoluble species than dry ESPs or filter-based samplers, which require the collected samples to be extracted in an aqueous solution (Wang et al., 2013). Previous research has primarily focused on ESPs with considerably low sampling flow rates (e.g., < 5 lpm), which makes it challenging to collect sufficient PM material for toxicological studies (Ning et al., 2008; Sillanpää et al., 2008; Volckens & Leith, 2002a). Utilizing ESPs operating at a high flow rate, higher mass collections can be achieved in the same amount of time compared to low-flow-rate ESPs (Pirhadi et al., 2020). To maintain high collection efficiency at these high flow rates, design modifications such as increasing the collection surface area, increasing the number of needle electrodes, or increasing the applied voltage are required (Hinds & Zhu, 2022).

In this work, we developed and evaluated a wet ESP for the collection of ambient PM_2.5 (i.e., particles with aerodynamic diameters <2.5 μm) working at a high sampling flow rate. The optimization of the wet ESP configuration for maximum collection efficiency and lower ozone generation was achieved by adjusting the flow rate and applied voltage. To assess the performance of the wet ESP, we conducted concurrent field experiments using three different sampling methods: the wet ESP, the VACES system with a PTFE filter, and the VACES with a BioSampler. The collected samples’ chemical composition and oxidative potential were then analyzed and compared.

2. Methodology

2.1. Wet ESP design and configuration

The semispherical wet ESP designed for aerosol sampling is illustrated in Figure 1. The wet ESP consisted of a dome with an inner diameter of 10 cm and a length of 65 cm, fabricated from antistatic slippery polyethylene (McMaster-Carr, LA, CA, USA) with a thickness of 3.175 mm. Inlet and outlet ports with an inner diameter of 2.54 cm were also incorporated. Twelve tungsten electrodes in the wet ESP were positioned on a 70 cm long stainless-steel rod with a 3.175 mm inner diameter, placed 1.25 cm from the top of the dome. Each electrode, with a diameter of 0.76 mm and spaced 5 cm apart, was positioned 2.5 cm away from the ground electrode, with the first and last electrodes placed 3 cm from the inlet and outlet. The ground electrode, made from stainless steel, was press-fitted into a trough inside the dome and left a 0.2 cm gap from the bottom of the dome. The trough was filled with ultrapure Milli-Q water (resistivity 18.2 MΩ cm at 25 °C, total organic carbon (TOC) ≤ 5 ppb), which served as a collection surface for the charged particles. The electric field created by a high-voltage supplier (Model Bertran Series 915, Spellman High Voltage Electronics Corp., Hauppauge, NY, USA) in the vicinity of the corona discharge electrodes causes a continuous cascade of electron collisions, resulting in the accumulation of free electrons and charged ions in the corona region which enables the charging of aerosol particles. The charged particles were subjected to an electric field oriented perpendicular to the flow direction and the collection surface, resulting in the movement of the particles towards the water trough due to their electrostatic migration velocity resulting from the charges imparted on them (Hinds & Zhu, 2022). To maintain the water level in the trough and prevent water loss due to evaporation, a peristaltic pump (Mityflex mode 907, Anko Products Inc., Bradenton, FL) with a flow rate of 5 mL/min was connected to the wet ESP to supply Milli-Q water continuously, which started every 20 minutes and operated for 1 minute.

Figure 1 –

Wet electrostatic precipitator (ESP)

2.2. Laboratory characterization of the wet ESP

The objective of the laboratory experiments was to identify the optimal operating conditions of the wet ESP by adjusting the sampling flow rate and applied voltage to achieve both high collection efficiency with the lowest possible ozone generation. For the experimental setup, as illustrated in Figure 2, a nebulizer (Model 11310 HOPE^™ nebulizer, B&B Medical Technologies, USA) was utilized to aerosolize sodium chloride (NaCl) and ammonium sulfate ((NH₄)₂SO₄) suspensions, both at a concentration of 600 μg/mL. The aqueous solutions of NaCl and (NH₄)₂SO₄ were prepared by adding 60 mg of these species to 50 mL of Milli-Q water, and the solution was homogenized for 30 min using an ultrasonic bath sonicator (3510R-MT, Branson Ultrasonics Corp., Danbury, CT, USA) prior to being aerosolized to ensure a uniform concentration of these species in the aqueous solution. The aerosolized particles were then combined with fresh particle-free air, supplied by a HEPA filter (Model 12144, Pall Corporation, USA), and entered a diffusion dryer (Model 3062, TSI Inc., USA) to remove excess moisture by decreasing the relative humidity to about 50%. The dried particles were then passed through a group of 12 Po-210 neutralizers (Model 2U500, NRD Inc., USA) to remove the electric charge on the particles before being introduced into the wet ESP. In order to measure the particle collection efficiency of the wet ESP, a scanning mobility particle sizer (SMPS, Model 3936, TSI Inc., Shoreview, MN, USA), working in tandem with a condensation particle counter (CPC, Model 3022A, TSI Inc., Shoreview, MN, USA), and an optical particle sizer (OPS) (Model 3330, TSI Inc., USA) were positioned downstream and upstream of the wet ESP to measure the particle number concentration in size range of 0.01–0.7 μm and 0.3–10 μm, respectively. Using the Multi-Instrument Manager (MIM^™) software, the concentration data from both instruments were merged to measure the particle number concentration, and the details of the merging process and its credibility can be found in the work of Sowlat et al. (2016). The wet ESP’s collection efficiency (η) was determined experimentally by using equation (1):

$$\eta =\left(1-\frac{C_D}{C_U}\right)\times 100$$ (1)

where C_D and C_U are particle number concentrations downstream and upstream of the wet ESP, respectively. Also, a detailed theoretical calculation of the wet ESP’s collection efficiency is presented in section S1. Moreover, to determine the evaporation rate in the wet ESP, the amount of water lost over a 2-hour period was measured for various flow rates (i.e., 100, 125, 150, and 200 lpm) without applying high voltage by adding 50 mL of Milli-Q water to the wet ESP and measuring the remaining amount of water after 2 hours, while maintaining room temperature at 20°C and relative humidity at 67%.

Figure 2 –

Schematic representation of the wet ESP experimental setup for laboratory tests

To determine the optimal operating conditions of the wet ESP, the ozone generation at different flow rates (i.e., 100, 125, 150, and 200 lpm) with various applied voltages (i.e., 8–15 kV) was measured using a Thermo Environmental Instrument (Model 49C, Franklin, MA, USA) O3 analyzer. The background ozone concentration was determined by feeding HEPA filter air into the instrument and measuring the observed ozone values. The measured values for each case were subtracted from the background level to accurately determine the ozone produced by the wet ESP. The main concern with ozone generation in ESPs is chemical artifacts caused by the high ozone concentration since the aerosol particles could, in principle, react with the generated ozone and radicals by corona discharge, leading to chemical alteration of the PM samples (Ning et al., 2008). The experimental voltage-current characteristic curve of the wet ESP was plotted (Figure S1) and showed an increasing trend as the applied positive corona voltage increases, which is consistent with previous studies in the literature (Niewulis et al., 2013; Pirhadi et al., 2020).

The subsequent step in the laboratory experimentation was to determine the flow rates and applied voltages resulting in the highest collection efficiency. The particle collection efficiency of the wet ESP was measured for volumetric flow rates of 125 and 150 lpm and an applied positive voltage of 11 kV (determined as the optimal setting for low ozone generation, as discussed in the results section) by introducing the aerosolized particles of both NaCl and (NH4)2SO4 to the wet ESP, to verify the wet ESP’s performance is independent of the type of aerosol particles supplied.

2.3. Field experiments

Field experiments were performed at the University of Southern California particle instrumentation unit (PIU), located 3 km south of downtown Los Angeles and in the vicinity of the I-110 freeway. This location is directly impacted by both primary sources (e.g., vehicular emissions) and secondary photochemical aerosol formation (Farahani et al., 2021; Hasheminassab et al., 2014; Tohidi et al., 2022a). Concurrent field experiments were performed using the wet ESP and VACES (using PTFE filters and BioSampler) to collect PM_2.5 samples. A model-4 semi-continuous thermo-optical OC/EC field analyzer (Sunset Laboratory Inc., USA) was also employed to measure PM_2.5-bound total organic carbon (TOC) levels of the ambient air and compare the results with those obtained from other instruments. Four concurrent sampling tests were conducted over four days in December 2022, each lasting for 7 hours.

The schematic of the experimental setups for the field test of the wet ESP, OC/EC analyzer, and VACES are shown in Figure 3 (a), (b), and (c), respectively. As shown in Figure 3 (a), an L-shaped PM_2.5 impactor inlet (made of aluminum) was utilized upstream of the wet ESP to introduce fine aerosols to the ESP, the design details of which are provided in section S3.

Figure 3 –

Schematic representation of the (a) wet ESP and PM_2.5 impactor design, (b) OC/EC analyzer, and (c) VACES operating conditions and experimental setup for concurrent field sampling.

A 50 mL volume of Milli-Q water was used inside the wet ESP in each set, and the peristaltic pump was set to a flow rate of 5 mL/min. This flow rate was determined based on our tests, and the pump was set to operate for 1 minute every 20 minutes, supplying 5 mL of Milli-Q water into the ESP trough for a total of 15 mL of Milli-Q water per hour. In addition, as shown in Figure 3 (b), a carbon denuder was placed between the PM_2.5 cut-off inlet (operating at a flow rate of 5 lpm) and the OC/EC analyzer to remove gas phase organics and reduce the positive artifact caused by the adsorption of organic vapors on the quartz filter, which can artificially increase the total OC readings (Sardar et al., 2005; Taghvaee et al., 2019). The OC/EC analyzer measures ambient organic carbon levels through a series of temperature plateaus. PM samples collected on quartz filters are incrementally heated to specific temperatures to determine the OC fractions (OC1, OC2, OC3, and OC4) and then to higher temperatures to determine the EC levels, according to the established operational protocol (Birch & Cary, 2007; Chow et al., 2012). As illustrated in 3 (c), in the VACES system, fine ambient particles were first drawn into an ultrapure water tank using a PM_2.5 impactor with a flow rate of 210 lpm and saturated with water vapor at a temperature of 30°C. The airflow was then divided into two parallel lines, each with a flow rate of 105 lpm. The air in these lines entered a cooling tube, connected to a chiller (VWR model 89202–998, Pennsylvania, USA) operating at −4 °C, where the temperature was lowered to 21°C, causing the particles to grow into droplets of 3–4 μm. These droplets were then concentrated by virtual impaction, resulting in a concentration enrichment factor of 20. In-depth descriptions of the VACES system can be found in previous studies (Kim et al., 2001a and 2001b). One line of the VACES system directed the aerosols into a diffusion dryer (Model 3062, TSI Inc., USA) to eliminate excess moisture and restore the concentrated particles to their original size, which were then collected on a 37-mm PTFE (Teflon) filter (Pall Corp., Life Sciences, 2-μm pore, Ann Arbor, MI, USA), and each PTFE filter was extracted in 50 mL of Milli-Q water. The other line directed the grown particles into a BioSampler (BioSampler, SKC West, Inc., Fullerton, CA), which collects a slurry of PM samples through impaction and centrifugal forces (Daher et al., 2011).

2.4. Chemical and toxicological analyses

For chemical and toxicological analysis, four daily sets of PM slurries were collected using the wet ESP and were drained into a sterile vial. Samples collected on PTFE filters by the VACES system were extracted in Milli-Q water through sonication for 30 minutes at room temperature. The Wisconsin State Lab of Hygiene (WSLH) performed chemical analysis for metal and trace elements, as well as inorganic ions, on all sets of samples using inductively coupled plasma mass spectroscopy (ICP-MS) and ion chromatography (IC), respectively (Karthikeyan & Balasubramanian, 2006; Lough et al., 2005). The TOC concentration of the collected samples by the wet ESP, BioSampler, and PTFE filter was determined using a Sievers 900 total organic carbon analyzer (Stone et al., 2008, 2009) and compared to the TOC concentration measured by the OC/EC analyzer.

The oxidative potential of the collected PM samples was determined using the dithiothreitol consumption (DTT) assay. The oxidative potential is a measure of the PM’s ability to generate reactive oxygen species and cause oxidative stress, which can induce negative health effects (Cho et al., 2005; Verma et al., 2014). The rate at which DTT is consumed in a DTT assay indicates the oxidative potential of PM samples, which evaluates the reduction of cellular antioxidant levels during the change to disulfide form (Shafer et al., 2016; Verma et al., 2015). In this assay, the PM samples were incubated with DTT, a reducing agent (NADH and NDPH) that can be easily oxidized by ROS (Charrier et al., 2015; Verma et al., 2014). The oxidative potential of PM samples was represented by normalizing the DTT activity by the volume of sampled air (units of nmol/min. m³ of sampled air) to represent the extrinsic oxidative potential of the PM samples, which reflects the PM dilution, emission rates, and exposure to PM (Fang et al., 2015; Hakimzadeh et al., 2020; Tuet et al., 2016).

3. Results and Discussion

3.1. Laboratory experimentation results

3.1.1. Ozone generation by the wet ESP

As discussed in section 2.2, ozone generation by the wet ESP was measured at different flow rates and applied voltages to identify the optimum condition. The background ozone concentration was determined (30.7 ± 1.1 ppb) and subtracted from the measured values in each case. The ozone generation at different flow rates (i.e., 100, 125, 150, and 200 lpm) with various applied voltages (i.e., 8–15 kV in 1 kV increments) was measured and plotted in Figure 4. For a better illustration of the obtained experimental results, the ozone production plot is magnified for voltages between 8 and 11 kV. As shown in the figure, as the corona voltage increases, particularly above 12 kV, the ozone production significantly increases due to the transfer of more energy to the air molecules, resulting in the dissociation of O₂ molecules by the ions produced at higher applied corona voltages (Viner et al., 1992; Yasumoto et al., 2010). The wet ESP exhibited lower ozone generation at high flow rates, leading to higher dilution of the ozone produced by the wet ESP and a decrease in its concentration. (Ning et al., 2008; Yehia, 2007). For instance, at the voltage of 11 kV, the wet ESP generates 53% more (194 ppb) ozone at 100 lpm compared with the produced ozone at 125 lpm (133 ppb).

Figure 4 –

Comparison of ozone generation by the ESP at different flow rates (100, 125, 150, and 200 lpm) and applied voltages (8–15 kV). The error bars depict the standard deviation.

Minimizing ozone generation is critical as excessive production of ozone in the wet ESP could impact the chemical properties of the aerosol particles within the corona discharge region (Cardello et al., 2002; Kulkarni et al., 2002; Poppendieck et al., 2014; Zhu et al., 2020). Therefore, to mitigate ozone production in the wet ESP, only positive voltage was applied by the high-voltage supplier since negative voltages have been shown to lead to higher ozone concentrations produced by corona discharge (Chen & Davidson, 2003; Zhu et al., 2020). The ozone concentration values produced by the wet ESP in this study are similar to that reported by Volckens & Leith (2002a) (125 ppb). Their results revealed that their ESP is suitable for measuring semi-volatile compounds (SVOCs) such as alkanes and polycyclic aromatic hydrocarbons (PAHs). Furthermore, the ozone concentration of these operating conditions is lower than the value reported by Kaupp & Umlauf (1992) (200 – 300 ppb). The aerosol residence time in the wet ESP was approximately 1 second, which is comparable to the residence time of 0.25 s reported by Volckens & Leith’s (2002a) ESP. This residence time makes it unlikely for any chemical reactions between ozone and aerosols to occur and eliminates the possibility of ozone deposition by diffusion into the wet ESP water layer. Based on the comparable ozone levels and ESP residence times between our study and that of Volckens & Leith (2002a), we believe that semi-volatile compounds (SVOCs) such as alkanes and polycyclic aromatic hydrocarbons (PAHs) can be sampled without significant chemical alteration. In addition, the theoretical ozone production in a small room or laboratory by the wet ESP was calculated, the details of which are presented in section S2. The ozone concentration produced by the wet ESP in a small room (e.g., with a volume of 75 m³) and with a reasonably low air exchange rate (AER) of 1 hour⁻¹ will reach approximately a maximum of 13.3 ppb as the worst-case scenario. This concentration is lower than the background ozone concentration of 30.7 ± 1.1 ppb and cannot pose a health risk according to the safe limit set by the US Environmental Protection Agency (EPA), which is 70 ppb for an eight-hour exposure time (McCarthy et al., 2018). It is important to note that higher AER values in an indoor environment would result in even lower indoor ozone concentrations.

In addition to the precautions taken to minimize ozone production in our wet ESP, we also investigated the literature on the heterogeneous ozonation reactions of polar organic compounds and PAHs to quantify the possible decay of these compounds in the wet ESP. Numerous studies have reported reaction rates for these compounds with ozone at concentrations comparable to or much higher than that in our wet ESP system (e.g., 400 ppb) (Kahan et al., 2006; Kasumba & Holmén, 2018; Miet et al., 2009; Perraudin et al., 2007). Jenkin et al. (2020) studied ozone reactions with a diverse array of unsaturated volatile organic compounds ((a total of 221 VOCs)), and found rate coefficients never exceeding 10⁻¹⁵ cm³/molecule/s. Assumptions of pseudo first-order reactions with O₃ (concentration of 133 ppb or 3.3 × 10¹² molecules/cm³), a reaction rate of 10⁻¹⁵ cm³/molecule/s (as a worst case scenario), and for a reaction time of 1 second, which corresponds to the residence time in our ESP, indicate the maximum ratio of the reactant VOC concentrations exiting and entering the ESP, C/C0 = exp(−k × t), is greater than 0.99 for most cases. This finding suggests that the chemical degradation of these compounds in our wet ESP system is likely minimal, at least based on these theoretical calculations.

3.1.2. Particle collection efficiency of the wet ESP

The collection efficiency of the wet ESP for different particle size ranges of sodium chloride (NaCl) is presented in Figure 5. The collection efficiency curves were obtained experimentally for flow rates of 125 and 150 lpm, with an applied voltage of 11 kV. The theoretical collection efficiency curves were also plotted in the figure based on the equations provided in the S1, considering the electrical charges induced by field and diffusion charging. It is important to note that the theoretical efficiency curves are based on an average electrical field strength value since the actual point-to-plate electrical field produced by the 12-wire electrodes is spatially non-uniform.

Figure 5 –

Comparison of experimental and theoretical collection efficiency of the wet ESP for NaCl particles for different size ranges at flow rates of 125 and 150 lpm and applied voltage of 11 kV. The error bars depict the standard deviation.

As can be seen in Figure 5, the particle collection efficiency decreases with the increase of PM diameter for particles smaller than 0.07 μm, where diffusion is the dominant mechanism for particle charging. We observe higher particle collection efficiency for larger particles (i.e., d_p>0.1μm), where the field charging is dominant, while a minimum collection efficiency is observed for particles with a diameter around d_p≈0.07−0.1 μm, which can be attributed to the lower electrical mobility in this size range (Hinds & Zhu, 2022; Sillanpää et al., 2008). The particle collection efficiency dependence on particle size is similar to those observed in previous ESP studies (Hu et al., 2022; Li et al., 2019; Pirhadi et al., 2020; Sillanpää et al., 2008; Zhu et al., 2012a; Zhu et al., 2012b). Furthermore, it can be observed that there is a decline of 18% in the overall collection efficiency of the wet ESP when the flow rate increases from 125 lpm to 150 lpm, which is consistent with the trends reported in wet ESP in previous studies (Sobczyk et al., 2017; Xu et al., 2015). This is because the increase in the flow rate results in a shorter particle residence time and fewer charges acquired by the incoming aerosol, which decreases the overall particle collection efficiency of the wet ESP (Hinds & Zhu, 2022).

It is expected that the collection efficiency of the wet ESP will increase with the increase in the applied voltage due to the enhancement of particle charging. However, as previously discussed, ozone generation also increased with the applied voltage, and for voltages higher than 11, the ozone generation became excessively high and could affect the chemical properties of the collected samples. Thus, based on the ozone production curves and the collection efficiency data, a flow rate of 125 lpm and an applied voltage of 11 kV were considered the optimal operating conditions for the wet ESP in this study.

To further evaluate the performance of the wet ESP under the optimal operating conditions identified, an additional artificially lab-generated aerosol was used. As previously discussed in the methodology, a nebulizer was used to aerosolize dissolved particles of ammonium sulfate, which is also a major constituent of PM_2.5 (Martin et al., 2004; Toon et al., 1976).

The experimental collection efficiency curves of NaCl and (NH₄)₂SO₄, as well as the corresponding theoretical collection efficiency curves, were plotted in Figure 6 for the optimal operating conditions of 125 lpm and 11 kV. As demonstrated in the figure, the wet ESP performed similarly, with less than a 1% difference, in the collection of aerosolized NaCl and (NH₄)₂SO₄ particles. This indicates that the wet ESP was able to effectively collect particles of different chemical compositions under the optimal operating conditions established in the previous section. As seen in Figure 6, the average wet ESP particle collection efficiency for NaCl and (NH₄)₂SO₄ is more than 80% across all particle sizes, demonstrating the wet ESP’s effectiveness in collecting different types of particles.

Figure 6 –

Comparison of experimental and theoretical collection efficiency curves of NaCl and (NH₄)₂SO₄ particles at the optimal operating conditions of the wet ESP (flow rate of 125 lpm and voltage of 11 kV). The error bars depict the standard deviation.

3.2. Field experiments

3.2.1. Chemical analysis

The PM_2.5-bound TOC levels of the collected samples during four-day concurrent sampling using the wet ESP, BioSampler, and PTFE filter sampler were compared with the OC/EC analyzer results, as demonstrated in Figure 7. The results indicate that the TOC content of the PM collected by the wet ESP and BioSampler was comparable, with percentage deviations of 7.86% and 1.62% from the OC/EC analyzer, respectively, whereas the PTFE filter sampler had 27.36% lower TOC content than OC/EC analyzer. This is in line with the findings of Pirhadi et al. (2020) who found a 27% lower collection efficiency of TOC for the VACES/PTFE sampler compared to the VACES/BioSampler. In a study by Daher et al. (2011), the levels of ammonium and nitrate collected using the same setup on the PTFE filter were 14%−22% lower than those for the BioSampler slurry, suggesting a lower collection efficiency for the VACES/PTFE filter sampler. It was also expected that the wet ESP might exhibit a lower TOC content than the BioSampler since the wet ESP has a collection efficiency of 80–85% across all size ranges. Furthermore, the results showed comparable TOC content in collected samples using the wet ESP and BioSampler (p-value<0.005), which differs from previous findings that showed higher TOC content in BioSampler samples compared to those collected by dry ESPs, likely due to the dry ESP’s limitations in extracting water-insoluble organic carbon (WIOC) from a dry substrate (Pirhadi et al., 2020; Wang et al., 2013). Collecting PM samples directly into a slurry in the wet ESP and BioSampler allows for the SVOCs to be dissolved in the liquid phase, preventing their loss or degradation due to adsorption and evaporation processes that can occur during the collection process on a PTFE filter (Chow et al., 2010; Kumar & Gupta, 2015).

Figure 7 –

Comparison of TOC content in in PM_2.5 samples collected using the wet ESP, VACES BioSampler, PTFE filter sampler, and OC/EC analyzer. The error bars depict the standard deviation.

The comparable TOC levels between the OC/EC analyzer and the wet ESP indicate that the ozone generation in the wet ESP probably does not have a significant impact on the chemical alteration of the collected samples, at least with regard to their bulk organic carbon content. While it is conceivable that some degradation of highly reactive organic compounds with a limited lifespan in the atmosphere might occur in the wet ESP (as also argued by Volckens & Leith, 2002a)), the overall agreement in TOC concentrations between the wet ESP and the presumably artifact-free concentrations OC/EC analyzer suggests that any sampling artifacts are likely to be small.

The concentration of metals and trace elements in PM_2.5 samples collected by the wet ESP, BioSampler, and PTFE filter sampler is illustrated in Figure 8(a). The PM samples collected through the BioSampler and wet ESP (Figure 8 (b)), as well as the PTFE filter sampler and wet ESP (Figure 8 (c)), exhibited a strong correlation with R² values of 0.94 and 0.95, respectively (p-value <0.001 for both cases). The regression lines derived from standard regression analysis for both data sets have slopes (1.09 and 1.12, respectively) that are nearly equal to one. The results show that the concentrations of metals and trace elements are comparable almost for all species collected by the three instruments, especially for Fe, Cu, Mn, V, and Cr, which have been documented as redox-active species and can contribute to PM toxicity (Cheung et al., 2010; Shafer et al., 2010). The strong agreement between TOC measured by our wet ESP and the OC/EC analyzer, particularly during winter when water-insoluble organic carbon (WSOC) levels are low, demonstrates its efficiency in capturing equally water-soluble and -insoluble PM species. Moreover, the strong agreement also observed for trace element and metal concentrations, i.e., species that are known to have varying solubilities, between our wet ESP and the reference sampler highlights its capability to capture a broad range of ambient aerosol constituents, regardless of their solubility properties (Heal et al., 2005; Jiang et al., 2014; Yang et al., 2023).

Figure 8 –

a) Comparing concentrations of metals and trace elements in PM_2.5 samples collected by wet ESP, VACES/BioSampler, and PTFE filter sampler. b) Linear regression analysis between the levels of metals and trace elements collected by the wet ESP vs VACES/BioSampler, and c) wet ESP vs VACES/PTFE filter sampler. The error bars depict the standard deviation.

3.2.2. Toxicological analysis

To evaluate the toxicity of the collected PM_2.5 samples using the wet ESP and VACES, we conducted a DTT assay on the samples obtained by the wet ESP, VACES/BioSampler, and VACES/PTFE filter. The DTT activity, which shows the oxidative potential of the collected PM, was normalized to the volume of the sampled air to express the values in the unit of nmoles/min/m³. The air volume-based DTT activity of the samples collected by different instruments is shown in Figure 9. As can be seen in the figure, the VACES/BioSampler had the highest DTT activity among the three instruments, with an average value of 0.25 nmoles/min/m³, followed by the wet ESP with a DTT activity of 0.22 nmoles/min/m³, and the VACES/PTFE filter with a DTT activity of 0.21 nmoles/min/m³. The results of the DTT assay suggest that all three sampling methods displayed comparable DTT activity, which is in concert with the similar levels of metals, trace elements, and TOC collected by these instruments. The somewhat lower DTT activity observed in the samples collected by the wet ESP compared to the VACES/BioSampler can be attributed to the wet ESP’s average collection efficiency of 80%−90% within all size ranges. Previous studies have shown that the water-insoluble fraction of the collected PM is associated with the oxidative potential of the particles (Daher et al., 2011; Gao et al., 2020). Therefore, the VACES/BioSampler and wet ESP may be more suitable for PM collection in toxicological studies due to the challenges of extracting the water-insoluble fraction from PTFE filters.

Figure 9 –

Comparison of extrinsic oxidative potential of PM_2.5 samples collected by the wet ESP, VACES BioSampler, and PTFE filter sampler. The error bars depict the standard deviation.

4. Conclusions

The purpose of this study was to design, optimize, and assess the performance of a high-flow-rate wet ESP for collecting PM_2.5 into ultrapure water for toxicological studies. Through laboratory experiments focused on evaluating the particle collection efficiency of the wet ESP and ozone generation by applying different voltages, the optimal operating conditions were found to be an applied voltage of 11 kV and a flow rate of 125 lpm. These conditions were associated with a particle collection efficiency of more than 80% across all size ranges and ozone generation of 133 ppb. The results of the chemical analysis showed that the wet ESP had comparable performance in the collection of metal and trace elements to the VACES/BioSampler and VACES/PTFE filter sampler. The TOC levels of the PM collected by the wet ESP, VACES/BioSampler, and OC/EC analyzer were comparable, while the PTFE filter sampler had lower TOC content. The comparable TOC levels between the OC/EC analyzer and the wet ESP, as well as the comparable levels of trace elements and metals for all three instruments used in this study, indicate that ozone generation in the wet ESP probably does not have a significant impact on the bulk chemical composition of the collected samples. The results of the toxicological analysis, conducted through a DTT assay, also demonstrated comparable DTT activity for the three sampling methods. The comparable DTT activity between the sampling methods is in line with the similar levels of metals, trace elements, and TOC collected by these instruments. Given the high collection efficiency of the wet ESP and its performance in capturing different constituents of ambient PM compared to other instruments, as well as the comparable oxidative potential exhibited by the collected samples, the wet ESP could be a practical substitute for PM collection in toxicological studies.

• A wet ESP was developed for the PM collection in toxicological studies.

• Optimal operating conditions of the wet ESP were found through laboratory tests.

• The Wet ESP exhibited a comparable oxidative potential to the BioSampler.

• The wet ESP, like the BioSampler, can preserve the water-insoluble species.