Multi-Scale Modeling and Study of Aerosol Growth in an Amine-based CO 2 Capture Absorber

: A monoethanolamine (MEA) aerosol growth model was developed to quantify the aerosol growth factor in an amine-based CO 2 capture absorber that considers the gas-liquid interactions, and it is empirically validated by measuring the aerosol particle size and concentration. The aerosol growth model, using sucrose as the aerosol nuclei instead of sulfuric acid to prevent the corrosion of the test equipment, accurately predicted that the outlet aerosol size increased to the same level regardless of the sucrose concentration. It also found that particle concentration was the primary factor a ﬀ ecting aerosol growth and amine emissions. We found an inverse relationship between aerosol particle concentration and the aerosol size, while the MEA emissions were proportional to particle concentration.


Introduction
The capture of post-combustion CO 2 through chemical capture and absorption by chemical solvents, and absorption by amine solutions, are commercialized technologies. However, the scale-up of these technologies has experienced challenges associated with the vaporization of chemical solvents and organic solutions. One example is the amine emissions, which are known to cause harm to humans and the environment. In one example of the capture process, the CO 2 in a flue gas is chemically absorbed by the amine solvents. The enriched amine solvent is regenerated in a stripper and is recycled to the absorber [1]. Monoethanolamine (MEA) is an organic solvent commonly used for CO 2 capturing in coal-fired power plants to absorb CO 2 . The absorbent mechanism was based on MEA that reacted with CO 2 in the solution to become amine carbamate. However, MEA partially volatilizes and is absorbed into the nuclei of the complex flue gas effluents to form aerosol particles. The particle sizes of these aerosols are too small to be effectively removed through absorbers, water washes, and demisters, resulting in amine solvent loss through emissions.
Thitakamol et al. [2] reported the emission (loss) of monoethanolamine (MEA) as vapor to be 0.11-0.72 kg/tonne CO 2 from an absorber with a removal capacity of 1000 tonne CO 2 /day, using 30 wt% MEA. This significant amount, 110-720 kg/day, of amine emissions increases operating costs. Furthermore, the lost amine reacts with NO x in the flue gas to form nitrosamines [3,4]. Kamijo et al. [5] compared amine emission between MEA and KS−1 TM (Mitsubishi Heavy Industries, Ltd., Japan) solvents, reporting that emissions were 0.8-67.5 and 0.4-23.2 ppmv, respectively, based on the fuel gas containing 0-3 ppmv of SO 3. Their study also found that the amine was entrained by the aerosols produced from SO 3 . A pilot plant at Karlsruhe Institute of Technology (Germany) conducted a series the multi-scale model. Aspen Plus was used to simulate the absorber profiles, and gPROMS was used to build an Amine Aerosol growth model. Figure 1 shows the method adopted for connecting Aspen Plus and gPROMS with Microsoft Excel used as the connecting interface. On the Aspen Plus side, Visual Basic for Applications (VBA) was used to transfer the absorber data from Aspen Plus to Excel; the extracted data were gas and liquid temperatures, gas and liquid mole fractions, liquid holdup, gas-liquid interfacial area, and mass transfer coefficients. On the gPROMS side, foreign objects, which are Excel macros provided by gPROMS, exchanged the parameters between Excel and gPROMS. In this study, the Amine Aerosol growth model sent the position of the aerosol in the absorber to receive the corresponding data from the absorber.
Environments 2020, 7, x FOR PEER REVIEW 3 of 13 coupling was used to construct the multi-scale model. Aspen Plus was used to simulate the absorber profiles, and gPROMS was used to build an Amine Aerosol growth model. Figure 1 shows the method adopted for connecting Aspen Plus and gPROMS with Microsoft Excel used as the connecting interface. On the Aspen Plus side, Visual Basic for Applications (VBA) was used to transfer the absorber data from Aspen Plus to Excel; the extracted data were gas and liquid temperatures, gas and liquid mole fractions, liquid holdup, gas-liquid interfacial area, and mass transfer coefficients. On the gPROMS side, foreign objects, which are Excel macros provided by gPROMS, exchanged the parameters between Excel and gPROMS. In this study, the Amine Aerosol growth model sent the position of the aerosol in the absorber to receive the corresponding data from the absorber.

Amine Aerosol Growth Modeling
The Amine Aerosol growth model was modified from Kang et al. [13] to make it applicable for CO2 absorption using MEA. Changes in the aerosol growth and amine composition during the gas phase were considered. Mass transfer was established, based on the two-film theory, to predict the aerosol growth, assuming that mass transfer between the aerosol and the gas phases was similar to the mass transfer between the liquid phase and the gas phase.
Four components, H2O, CO2, MEA, and N2, were employed in the model. For aerosol growth, the material and energy balances are expressed as follows: where i represents the index of the component, ℎ represents the position of the aerosol in the absorber, , represents the mole number of species i, represents the average velocity of the droplets, which was assumed to be the same as the gas velocity, represents the height of the column, , represents the gas-aerosol mass transfer flux of species i, and represents the surface area of a droplet. Furthermore, is the temperature of the aerosol phase, is the heat transfer flux between the gas and aerosol phases, represents the total number of moles of a droplet, and , is the heat capacity of the droplet. Since MEA and H2O are the solvent components, the mass transfer flux , for H2O and MEA is expressed as follows: For the hardly soluble gas CO2 in the amine solution, the mass transfer flux , should be modified as

Amine Aerosol Growth Modeling
The Amine Aerosol growth model was modified from Kang et al. [13] to make it applicable for CO 2 absorption using MEA. Changes in the aerosol growth and amine composition during the gas phase were considered. Mass transfer was established, based on the two-film theory, to predict the aerosol growth, assuming that mass transfer between the aerosol and the gas phases was similar to the mass transfer between the liquid phase and the gas phase.
Four components, H 2 O, CO 2 , MEA, and N 2 , were employed in the model. For aerosol growth, the material and energy balances are expressed as follows: where i represents the index of the component, h represents the position of the aerosol in the absorber, n d,i represents the mole number of species i, v drop represents the average velocity of the droplets, which was assumed to be the same as the gas velocity, H col represents the height of the column, N gd,i represents the gas-aerosol mass transfer flux of species i, and A gd represents the surface area of a droplet. Furthermore, T d is the temperature of the aerosol phase, Q gd is the heat transfer flux between the gas and aerosol phases, n d represents the total number of moles of a droplet, and C p,d is the heat capacity of the droplet. Since MEA and H 2 O are the solvent components, the mass transfer flux N gd,i for H 2 O and MEA is expressed as follows: Environments 2020, 7, 58 4 of 13 For the hardly soluble gas CO 2 in the amine solution, the mass transfer flux N gd,CO 2 should be modified as N gd,CO 2 = k g,CO 2 P g,CO 2 − P * d,CO 2 ,corr , where k g,i represents the mass transfer coefficients for H 2 O and MEA; ϕ is the Fuchs-Sutugin corrective factor; P g,i represents the partial pressure of species i obtained from the absorber profile of Aspen Plus, P * d,i,corr represents the corrected equilibrium partial pressure of species I, T g represents the temperature of the gas phase, and R represents the ideal gas constant. Furthermore, k g,CO 2 is the enhanced mass transfer coefficient of CO 2 , evaluated using the chemical reaction rates, which are defined as follows: where k 2 represents the reaction rate constant of CO 2 , C f ree MEA represents the concentration of free MEA, D d,CO 2 ,mx represents the CO 2 diffusivity of the mixture solution during the aerosol phase, and H CO 2 represents the Henry law constant of CO 2 .
The Fuchs-Sutugin corrective factor ϕ was used for the mass transfer rate of the micron-scale aerosol droplet and is expressed as follows: The Knudsen number Kn is defined as follows: where α is the accommodation coefficient, λ is the mean free path of the surrounding fluid molecules, and d drop is the diameter of an aerosol droplet. The correction of vapor pressures for the droplets was also considered in the aerosol growth model because the vapor pressure of the curved surface was higher than the normal vapor pressure. Hence, in the aerosol model, the Kelvin effect was used to correct the equilibrium partial pressure in a droplet. The corrected equilibrium partial pressure P * i,corr is expressed as follows: where P * d,i is the equilibrium partial pressure of component i, σ is the surface tension between the gas and aerosol phases, v l,i is the partial molar volume of component i in the aerosol phase, k B is the Boltzmann constant, T drop is the temperature of the aerosol phase, and d drop is the diameter of the aerosol.
The heat transfer flux Q gd is expressed as follows: where ∆H H 2 O , ∆H MEA , and ∆H CO 2 are the heat of vaporization for H 2 O and MEA and the heat of absorption for CO 2 , respectively; h g is the heat transfer coefficient; and T g is the temperature of the gas phase.

of 13
The mass transfer coefficients k g,i for H 2 O and MEA and the heat transfer coefficient h g were calculated using the Sherwood number (Sh) and Nusselt number (Nu), assuming that the aerosols were spherical. The Sherwood and Nusselt numbers are expressed as follows: where Re g is the gas Reynolds number, Sc g,i is the gas Schmidt number for species i, and Pr g is the Prandtl number. However, in the absence of convection, the Sherwood and Nusselt numbers can be considered to be 2. Hence, the mass transfer coefficients k g,i for H 2 O and MEA are calculated as follows: The heat transfer coefficient was calculated as follows: where D g,i is the gas diffusion coefficient of species i, obtained from the absorber profile of Aspen Plus; d drop is the diameter of an aerosol droplet; and λ g is the gas thermal conductivity. The diameter of the droplets was calculated as follows: where ρ drop is the density of the aerosol droplets and MW drop is the molecular weight of the aerosol droplets. The equilibrium partial pressures during the aerosol phase P * d,i are estimated based on the modified Raoult's law as follows: where P * d,i and P sat d,i are the equilibrium and saturated partial pressures of H 2 O and MEA during the aerosol phase, respectively; γ d,i is the activity coefficient; and x d,i is the mole fraction. The equilibrium partial pressures of CO 2 during the aerosol phase were obtained from the correlation [17].
With regard to the change in the gas phase, the change in MEA concentration in the gas phase was considered in the Amine Aerosol growth model because the MEA concentration was significantly lower than that of H 2 O and CO 2 . This resulted in the mass transfer of MEA across the gas-aerosol interface, and significantly affected the gas-phase MEA mole fraction. Compared to the other two components, the change in the MEA composition caused by the gas-aerosol phase interaction may have been lower than that caused by the gas-liquid interaction. As a result, the MEA composition was recalculated by considering both the gas-aerosol and gas-liquid interactions. Furthermore, the mole fractions of N 2 , H 2 O, and CO 2 were obtained from process modeling using Aspen Plus. The MEA material balance can be simplified as a differential equation of the MEA partial pressure, dP MEA , which is expressed as follows: v where N gl,MEA is the mass transfer flux of MEA between the gas and liquid phases, a gl is the gas-liquid contact area, N p is the number of particle concentration, and R is the ideal gas constant. This equation considers the mass transfers of gas-aerosol and gas-liquid interfaces. In the gas-aerosol mass transfer, a new term affecting PNC was introduced to estimate the total mass transfer Environments 2020, 7, 58 6 of 13 flux from gas to aerosol in a column. The gas-liquid mass transfer flux of MEA N gl,MEA is expressed as follows: where k g,MEA is the gas mass transfer coefficient of the MEA for the gas-liquid interface, obtained from the absorber profile of Aspen Plus, P g,MEA is the partial pressure of the MEA, and P * l,MEA is the equilibrium partial pressure of the MEA for the liquid phase, obtained from the absorber profile of Aspen Plus.
The reactions in the aerosol phase were assumed as the same as the reaction in the liquid phase [18].

Physical Properties
The thermodynamic properties and mass transfer parameters of the gas and liquid phases were obtained from the absorber model using the Excel interface. The thermodynamic properties of the aerosol phase, such as the activity coefficients, heat capacity, surface tension, and true composition, were obtained from the Aspen Properties database, using the electrolyte non-random-two-liquid (e-NRTL) model for MEA through the CAPE-OPEN Thermo-interface to connect with gPROMS. A few missing parameters were included based on the literature, such as vapor pressure and diffusion coefficient for the aerosol phase. The origins of these properties are listed in Table 1. Table 1. Summaries of the sources of thermodynamic properties and mass transfer parameters.

Amine Absorber Modeling Using Aspen Plus
The rate-based absorber was modified by [19], employing a rigorous e-NRTL thermodynamic framework and reaction kinetics. Models for the mass transfer and interfacial area were based on Onda et al. [5] and the liquid holdup was based on Stichlmair et al. [20]. The number of stages for the absorber was set to 100 to export smooth profiles. Figure 2 shows the experimental flow employed in this study. A lab-scale packed-bed absorber [21] was used to carry out the experiment for validating aerosol growth. The equipment contained four removable glass packed sections with their inner diameter and height being 25.4 and 300 mm, respectively, however, in this study only one packed section was used to investigate CO 2 removal and aerosol size. The packed section was fully filled with plastic Raschig rings, as described in Table 2, with the porosity and packing surface area being 0.86 and 610 m 2 /m 3 , respectively. The liquid solvent flow was set to 0.2 L/min, and 1.5 and 3 L/min of inlet gas flows were evaluated. The inlet flow containing 20% CO 2 and 80% N 2 passed through a solution and an aerosol generator (Model 3076, TSI) to carry the aerosol into the absorber. The solution was prepared with sucrose instead of sulfuric acid Environments 2020, 7, 58 7 of 13 to the nuclei of the aerosol to prevent erosion of the detection equipment. While sulfuric acid reacts with the amine, the amount of sulfuric acid in the aerosol was too few so the reaction between sulfuric acid and MEA can be ignored. Sucrose does not react with MEA solvent. Thus, sucrose can be used to simulate sulfuric acid in order to react with MEA solvent.

Experiments
Environments 2020, 7, x FOR PEER REVIEW 7 of 13 exceeding the PNC count limitation, an aerosol diluter (Model 3332, TSI) was installed between the particle size and gas flows with a dilution ratio of 1/100. The detailed operating conditions are listed in Table 2.   Figure 3 shows the inlet aerosol PSD at different sucrose concentrations. The PNCs were found to be 73,094, 133,114, and 225,076 #/cm 3 for sucrose concentrations of 1000, 3000, and 10,000 ppm, respectively. As shown in Figure 3, most particles were found to be less than 0.3 um in size. Figure 4 shows the distribution of the outlet aerosol particle sizes. Compared to the inlet aerosol particle sizes, the average sizes of the outlet aerosol were found to be 1.43, 1.36, and 1.33, μm and the PNCs were observed to be approximately 3.2 × 10 , 3.3 × 10 , and 3.6 × 10 #/ for sucrose concentrations of 1000, 3000, and 10,000 ppm, respectively. The results showed that the sucrose concentration was not affected by the outlet aerosol size.  Sucrose is a nonvolatile solute that provides a condensed nucleus during the atomization process and produces aerosols with stable size distributions because of its fixed velocity when passing through an atomizer [22,23]. An optical particle sizer (Model 3330, TSI) was used to measure the aerosol particle size distribution (PSD) and PNC at the inlet and outlet of the gas. The particle sizer sizes in the range of 0.3-10 µm. The PNC range of the particle sizer was 0-3000 #/cm 3 . To prevent exceeding the PNC count limitation, an aerosol diluter (Model 3332, TSI) was installed between the particle size and gas flows with a dilution ratio of 1/100. The detailed operating conditions are listed in Table 2. Figure 3 shows the inlet aerosol PSD at different sucrose concentrations. The PNCs were found to be 73,094, 133,114, and 225,076 #/cm 3 for sucrose concentrations of 1000, 3000, and 10,000 ppm, respectively. As shown in Figure 3, most particles were found to be less than 0.3 um in size. Figure 4 shows the distribution of the outlet aerosol particle sizes. Compared to the inlet aerosol particle sizes, Environments 2020, 7, 58 8 of 13 the average sizes of the outlet aerosol were found to be 1.43, 1.36, and 1.33, µm and the PNCs were observed to be approximately 3.2 × 10 5 , 3.3 × 10 5 , and 3.6 × 10 5 #/cm 3 for sucrose concentrations of 1000, 3000, and 10,000 ppm, respectively. The results showed that the sucrose concentration was not affected by the outlet aerosol size.   Figure 5 shows that the absorber model could predict the gas outlet CO2 mole fraction of the experimental data at 1.5 and 3 L/min gas flow rates. Results from the absorber model were found to be in good agreement with the experimental data, with an average error of 3.2%. Figure 6 shows that the aerosol growth model could predict the average outlet aerosol size with an average error of 2.6%, after adjusting the accommodation coefficient, , to 0.025.     Figure 5 shows that the absorber model could predict the gas outlet CO2 mole fraction of the experimental data at 1.5 and 3 L/min gas flow rates. Results from the absorber model were found to be in good agreement with the experimental data, with an average error of 3.2%. Figure 6 shows that the aerosol growth model could predict the average outlet aerosol size with an average error of 2.6%, after adjusting the accommodation coefficient, , to 0.025.   Figure 5 shows that the absorber model could predict the gas outlet CO 2 mole fraction of the experimental data at 1.5 and 3 L/min gas flow rates. Results from the absorber model were found to be in good agreement with the experimental data, with an average error of 3.2%. Figure 6 shows that the aerosol growth model could predict the average outlet aerosol size with an average error of 2.6%, after adjusting the accommodation coefficient, α, to 0.025.   Figure 7 shows the predicted aerosol size growth in the experimental absorber using four select aerosol input PNCs and = 0.025. The Z/Ztot represents the normalized height of the absorber. 0 means the bottom of the column and 1 indicates the top column. Aerosols will continue to grow in the absorber because the MEA solvent is transferred into the aerosol to disrupt the partial pressure equilibrium. CO2 is also absorbed because the concentration of amine increases in the aerosol phase.    Figure 7 shows the predicted aerosol size growth in the experimental absorber using four select aerosol input PNCs and = 0.025. The Z/Ztot represents the normalized height of the absorber. 0 means the bottom of the column and 1 indicates the top column. Aerosols will continue to grow in the absorber because the MEA solvent is transferred into the aerosol to disrupt the partial pressure equilibrium. CO2 is also absorbed because the concentration of amine increases in the aerosol phase.

Aerosol size of Prediction (μm)
Aerosol size of experoment (μm) Figure 6. Agreement of the gas outlet aerosol size between the aerosol growth model and experimental data of the aerosol with different accommodation coefficients α. Figure 7 shows the predicted aerosol size growth in the experimental absorber using four select aerosol input PNCs and α = 0.025. The Z/Z tot represents the normalized height of the absorber. 0 means the bottom of the column and 1 indicates the top column. Aerosols will continue to grow in the absorber because the MEA solvent is transferred into the aerosol to disrupt the partial pressure equilibrium. CO 2 is also absorbed because the concentration of amine increases in the aerosol phase.  Figure 8 shows the contour lines of the outlet aerosol PSD. PNC was observed to be the main factor affecting aerosol growth because the outlet aerosol size decreased as the PNC increased, whereas the initial aerosol size only slightly affected the growth as the PNC was lower than 10 6 #/cm 3 . When the PNC was 10 5 #/cm 3 , the outlet aerosol size with 0.25 µm of the initial aerosol size was smaller than other larger initial aerosol sizes because vapor pressures were high due to the high surface tension when the drop diameter was 0.25 µm and the aerosol might easily evaporate to limit growth in the absorber. The outlet aerosol size was only related to PNC because the aerosol growth rate was related to 1/D3. As the aerosol particle size increased, the growth rate decreased. Hence, regardless of the initial sizes used, the aerosol size would grow to a similar size if the resident time of the aerosol in the absorber was sufficiently long.   Figure 8 shows the contour lines of the outlet aerosol PSD. PNC was observed to be the main factor affecting aerosol growth because the outlet aerosol size decreased as the PNC increased, whereas the initial aerosol size only slightly affected the growth as the PNC was lower than 10 6 #/cm 3 . When the PNC was 10 5 #/cm 3 , the outlet aerosol size with 0.25 µm of the initial aerosol size was smaller than other larger initial aerosol sizes because vapor pressures were high due to the high surface tension when the drop diameter was 0.25 µm and the aerosol might easily evaporate to limit growth in the absorber. The outlet aerosol size was only related to PNC because the aerosol growth rate was related to 1/D3. As the aerosol particle size increased, the growth rate decreased. Hence, regardless of the initial sizes used, the aerosol size would grow to a similar size if the resident time of the aerosol in the absorber was sufficiently long.  Figure 8 shows the contour lines of the outlet aerosol PSD. PNC was observed to be the main factor affecting aerosol growth because the outlet aerosol size decreased as the PNC increased, whereas the initial aerosol size only slightly affected the growth as the PNC was lower than 10 6 #/cm 3 . When the PNC was 10 5 #/cm 3 , the outlet aerosol size with 0.25 µm of the initial aerosol size was smaller than other larger initial aerosol sizes because vapor pressures were high due to the high surface tension when the drop diameter was 0.25 µm and the aerosol might easily evaporate to limit growth in the absorber. The outlet aerosol size was only related to PNC because the aerosol growth rate was related to 1/D3. As the aerosol particle size increased, the growth rate decreased. Hence, regardless of the initial sizes used, the aerosol size would grow to a similar size if the resident time of the aerosol in the absorber was sufficiently long.   Figure 9 shows the MEA emissions by computing the MEA concentrations of the gas and aerosol phases. The results showed that the MEA emission increased as the PNC increased. An emission of 331 mg/m 3 was observed at a PNC of 10 8 #/cm 3 , which is tripled in comparison to 155 mg/m 3 at a PNC of 10 5 #/cm 3 . Figure 10 shows the contours of the MEA emissions, and from them, it can be observed that PNC was the main factor in removing MEA from the absorbers. The initial size of the aerosol was also observed to have an effect on finial aerosol size when the PNC was below 10 7 #/cm 3 . No matter what initial aerosol size was used, the outlet aerosol size was similar. With the same aerosol size, the MEA concentration for a single aerosol droplet was also similar. Thus, the initial aerosol size has only a slight effect on MEA emissions. Owing to the fixed MEA concentration for a single aerosol, the amount of MEA emissions increased when the PNC increased.

Effect of PNC on MEA Emissions
Environments 2020, 7, x FOR PEER REVIEW 11 of 13 331 mg/m 3 was observed at a PNC of 10 8 #/cm 3 , which is tripled in comparison to 155 mg/m 3 at a PNC of 10 5 #/cm 3 . Figure 10 shows the contours of the MEA emissions, and from them, it can be observed that PNC was the main factor in removing MEA from the absorbers. The initial size of the aerosol was also observed to have an effect on finial aerosol size when the PNC was below 10 7 #/cm 3 . No matter what initial aerosol size was used, the outlet aerosol size was similar. With the same aerosol size, the MEA concentration for a single aerosol droplet was also similar. Thus, the initial aerosol size has only a slight effect on MEA emissions. Owing to the fixed MEA concentration for a single aerosol, the amount of MEA emissions increased when the PNC increased. .

Conclusions
In this study, we built an MEA aerosol growth model that incorporated the effect of PNC, between the aerosol and gas phases, on aerosol growth. We carried out an experiment to measure the Environments 2020, 7, x FOR PEER REVIEW 11 of 13 331 mg/m 3 was observed at a PNC of 10 8 #/cm 3 , which is tripled in comparison to 155 mg/m 3 at a PNC of 10 5 #/cm 3 . Figure 10 shows the contours of the MEA emissions, and from them, it can be observed that PNC was the main factor in removing MEA from the absorbers. The initial size of the aerosol was also observed to have an effect on finial aerosol size when the PNC was below 10 7 #/cm 3 . No matter what initial aerosol size was used, the outlet aerosol size was similar. With the same aerosol size, the MEA concentration for a single aerosol droplet was also similar. Thus, the initial aerosol size has only a slight effect on MEA emissions. Owing to the fixed MEA concentration for a single aerosol, the amount of MEA emissions increased when the PNC increased.
. Figure 9. Total MEA concentration profile of the gas phase in the absorber with different PNCs using the aerosol growth model.

Conclusions
In this study, we built an MEA aerosol growth model that incorporated the effect of PNC, between the aerosol and gas phases, on aerosol growth. We carried out an experiment to measure the

Conclusions
In this study, we built an MEA aerosol growth model that incorporated the effect of PNC, between the aerosol and gas phases, on aerosol growth. We carried out an experiment to measure the outlet aerosol particle size. In the experiment, sulfuric acid was replaced with sucrose as the nuclei of the aerosol to prevent corrosion of the equipment. Experimental results showed that the inlet aerosol sizes Environments 2020, 7, 58 12 of 13 were smaller than 0.3 µm and the outlet aerosol size increased to 1.38 µm with a PNC of 3.3 × 10 5 #/cm 3 . The aerosol growth model was validated using in-house experimental data. After adjusting the accommodation coefficient to 0.025, the aerosol growth model was able to predict the aerosol size with an accuracy of 2.6%. PNC was observed to be the main factor affecting aerosol growth and MEA emission, because when the PNC increased, the aerosol size was observed to decrease, while the MEA emission was observed to increase.