Diverse mixing states of amine-containing single particles in Nanjing, China

The mixing states of particulate amines with different chemical components are of great significance in studying the formation and evolution processes of aminecontaining particles. In this work, the mixing states of single particles containing trimethylamine (TMA) and diethylamine (DEA) are investigated using a high-performance single-particle aerosol mass spectrometer located in Nanjing, China, in September 2019. TMAand DEA-containing particles accounted for 22.8 % and 5.5 % of the total detected single particles, respectively. The particle count and abundance of the TMA-containing particles in the total particles notably increased with enhancement of ambient relative humidity (RH), while the DEA-containing particles showed no increase under a high RH. This result suggested the important role of RH in the formation of particulate TMA. Significant enrichments of secondary organic species, including C2H3O, 26CN−, 42CNO−, 73C3H5O−2 , and HC2O4 , were found in DEA-containing particles, indicating that DEA-containing particles were closely associated with the aging of secondary organics. The differential mass spectra of the DEA-containing particles showed a much higher abundance of nitrate and organic nitrogen species during the nighttime than during the daytime, which suggested that the nighttime production of particulate DEA might be associated with reactions of gaseous DEA with HNO3 and/or particulate nitrate. In the daytime, the decrease in DEAcontaining particles was observed with the enrichment of oxalate and glyoxylate, which suggested a substantial impact of photochemistry on the aging process of DEA-containing particles. Furthermore, more than 80 % of TMAand DEAcontaining particles internally mixed with nitrate, while the abundance of sulfate was higher in the DEA-containing particles (79.3 %) than in the TMA-containing particles (55.3 %). This suggested that particulate DEA existed both as nitrate and sulfate aminium salts, while the particulate TMA primarily presented as nitrate aminium salt. The different mixing states of the TMAand DEA-containing particles suggested their different formation processes and various influencing factors, which are difficult to investigate using bulk analysis. These results provide insights into the discriminated fates of organics during the evolution process in aerosols, which helps to illustrate the behavior of secondary organic aerosols. Published by Copernicus Publications on behalf of the European Geosciences Union. 17954 Q. E. Zhong et al.: Diverse mixing states of amine-containing single particles

Abstract. The mixing states of particulate amines with different chemical components are of great significance in studying the formation and evolution processes of aminecontaining particles. In this work, the mixing states of single particles containing trimethylamine (TMA) and diethylamine (DEA) are investigated using a high-performance single-particle aerosol mass spectrometer located in Nanjing, China, in September 2019. TMA-and DEA-containing particles accounted for 22.8 % and 5.5 % of the total detected single particles, respectively. The particle count and abundance of the TMA-containing particles in the total particles notably increased with enhancement of ambient relative humidity (RH), while the DEA-containing particles showed no increase under a high RH. This result suggested the important role of RH in the formation of particulate TMA. Significant enrichments of secondary organic species, including 43 C 2 H 3 O + , 26 CN − , 42 CNO − , 73C 3 H 5 O − 2 , and 89 HC 2 O − 4 , were found in DEA-containing particles, indicating that DEA-containing particles were closely associated with the aging of secondary organics. The differential mass spectra of the DEA-containing particles showed a much higher abundance of nitrate and organic nitrogen species dur-ing the nighttime than during the daytime, which suggested that the nighttime production of particulate DEA might be associated with reactions of gaseous DEA with HNO 3 and/or particulate nitrate. In the daytime, the decrease in DEAcontaining particles was observed with the enrichment of oxalate and glyoxylate, which suggested a substantial impact of photochemistry on the aging process of DEA-containing particles. Furthermore, more than 80 % of TMA-and DEAcontaining particles internally mixed with nitrate, while the abundance of sulfate was higher in the DEA-containing particles (79.3 %) than in the TMA-containing particles (55.3 %). This suggested that particulate DEA existed both as nitrate and sulfate aminium salts, while the particulate TMA primarily presented as nitrate aminium salt. The different mixing states of the TMA-and DEA-containing particles suggested their different formation processes and various influencing factors, which are difficult to investigate using bulk analysis. These results provide insights into the discriminated fates of organics during the evolution process in aerosols, which helps to illustrate the behavior of secondary organic aerosols.

Introduction
Amines are ubiquitous organic components in aerosols and have a wide range of sources, including animal husbandry, industrial emissions, vehicle exhaust, biomass burning, vegetation emissions, and ocean emissions (Ge et al., 2011b;Facchini et al., 2008;Youn et al., 2015). Due to being highly water soluble and having strong alkaline properties, amines play an important role in new particle formation and substantially contribute to the secondary organic aerosol (SOA) mass (Zhao et al., 2011;Tao et al., 2016). The formation processes of particulate amines are commonly associated with the gas-to-particle partitioning of gaseous amines and acidbase reactions in the particles (Ge et al., 2011a;Pratt et al., 2009). Therefore, ambient relative humidity (RH; Rehbein et al., 2011;Zhang et al., 2012), temperature (Huang et al., 2012), particle acidity (Pratt et al., 2009;Rehbein et al., 2011), amine-ammonium exchange (Chan and Chan, 2013;Chu and Chan, 2017;Qiu et al., 2011), and oxidants (Tang et al., 2013;Price et al., 2016) all influence the formation of particulate amines.
Many field observations have been used to investigate the influence of RH on the formation of amines. A high RH is beneficial for the formation of amines in most cases. Zhang et al. (2012) observed a sharp increase in trimethylamine (TMA) during fog events with high RH. Zhou et al. (2019) found that the concentrations of low molecular weight (LMW) amines increased significantly under high RH conditions (> 90 %). According to the seasonal distributions of amines during the summer and winter, low temperature was found to be favorable for the partitioning of gaseous amines into particles. Huang et al. (2012) found that the number fraction (N f ) of amine-containing particles during winter was 4 times higher than that during summer.
Gaseous amines can react with sulfuric acid, nitric acid, and organic acids to form aminium salts, which underscores the important role of sulfate and nitrate information for particulate amines (Berndt et al., 2010;Murphy et al., 2007). Berndt et al. (2010) and Wang et al. (2010) found that the formation of aminium salts via a neutralization reaction can affect the growth of particles and the generation of SOAs. Although the concentrations of amines are generally lower than ammonia, the amine-ammonium exchange still contributes to particulate amine formation due to the stronger alkalinity of amines compared to ammonium (Ge et al., 2011b;Sorooshian et al., 2008). Chan and Chan (2013) found that the exchange reactions between ammonia and amines showed different reaction rates and product ratios with changes in the aerosol phase state. Qiu et al. (2011) also found that amines can exchange with ammonium to release ammonia. The particulate amines produced from the above pathways and reactions constitute a substantial proportion of the SOAs that impact the physical and chemical properties of fine particles. In addition to the direct contribution of the SOA mass, the oxidation of amines by OH radicals, NO 3 radicals, and O 3 is also a substantial source of SOA production (Price et al., 2016;Tong et al., 2020). Different amines exhibit inconsistent behavior under the same oxidation environments (NO 3 radicals, OH radicals, or ozone; Price et al., 2014;Silva et al., 2008;Murphy et al., 2007). In chamber studies, the oxidation of TMA and diethylamine (DEA) by OH vs. NO 3 radicals resulted in different SOA yields, with differences greater than 1 order of magnitude (Tang et al., 2013). Furthermore, even the same amine showed completely different SOA yields due to OH and NO 3 radical oxidation. Also, the same amine showed distinct trends under the different temperature-changing trends. The formation and oxidation processes of particulate amines are not well understood, and these processes require additional comprehensive field observational studies in order to be elucidated.
Most of the field observations did not distinguish between the different behaviors of each type of amine molecule under the same ambient influencing factors. Actually, due to the different mixing states of amines with other chemical components, the amine molecules typically exhibited different behaviors in terms of being oxidized by OH radicals, forming aminium salts, and altering the hygroscopicity of the particles (Healy et al., 2015;Cheng et al., 2018;Chu et al., 2015;Price et al., 2016). Therefore, the formation processes of the different amines are important to reveal the evolution process of organic aerosols (OAs), and these processes are of great significance to comprehensively understand the influencing factors of OA production.
The mixing states of organics in single particles are commonly investigated by electron microscope and mass spectrometry (Yu et al., 2019;Li et al., 2016a;Cheng et al., 2018). The technique of an electron microscope coupled with energy-dispersive X-ray spectrometry (EDX) can provide the morphological, physical, and some chemical information of organics in single particles (Li et al., 2016b. However, an electron microscope coupled with EDX cannot distinguish or identify the specific organic molecules (Li et al., 2016b), making it incapable of analyzing amines in single particles. The technique of single particle mass spectrometry can identify the real-time presence and relative abundance of specific organic ions, which provides substantial data to understand the sources, formation, and evolution processes of selective organic markers, providing a feasible approach for investigating the formation processes of different particulate amines Chen et al., 2019;Angelino et al., 2001). Chen et al. (2019) found that high RH was favorable for the uptake of DEA, leading to a DEA-rich substance in the particle phase both during winter and summer. However, Cheng et al. (2018) and Lian et al. (2020) found that RH was not strongly correlated with the formation of amine-containing particles during winter and summer. Pratt et al. (2009) reported that more acidic particles during summer were favorable for the formation of aminium salts compared with the particles present during autumn, indicating that the particle acidity affected the gas-to-particle partitioning of amines. Rehbein et al. (2011) found more TMA entered the particles as the amount of acidic particles increased. Based on these studies, although the influences of ambient RH and particle acidity on the specific type of amines formed have been reported, comparative studies between different amines under the same atmospheric environment using field observation are lacking.
In the present study, the mixing states of TMA-and DEAcontaining single particles are investigated during autumn using a high-performance single particle aerosol mass spectrometer (HP-SPAMS) located in Nanjing, China. Nanjing is a typical megacity in the Yangtze River Delta (YRD), which is downwind of other megacities including Shanghai, Changzhou, Suzhou, and Wuxi ( Fig. S1 in the Supplement). In addition, Nanjing suffers from heavy loadings of anthropogenic pollutants and the complex impacts of biogenic and ship emissions (Xu et al., 2021;Zhao et al., 2020;Ding et al., 2013a). The investigation of the mixing states of amines in Nanjing helps to explorer the formation and evolution processes of OAs. The two types of amine-containing particles exhibited different mixing states with secondarily produced OA species. The influences of ambient RH, T , and particle acidity on the mixing states of the two amine-containing particles are evaluated. In addition, the potential heterogeneous formation of DEA during the nighttime is also discussed. The results revealed the distinct chemical behaviors of TMA-and DEA-containing particles and implied the potential role of DEA as an indicator of the aging process of OA.
2 Experimental methods

Sampling site
Ambient single particles were sampled using the HP-SPAMS from 2-16 September 2019 in Nanjing, China. The campaign was conducted at the Station for Observing Regional Processes of the Earth System (SORPES) station on Nanjing University Xianlin Campus (Ding et al., , 2013a. The instrument was set up on top of a small hill (40 m above the ground) on the Nanjing University campus. The ambient single particles were introduced into the HP-SPAMS through a copper tube.

Instrumentation of the HP-SPAMS
In this work, HP-SPAMS (Hexin Analytical Instrument Co., Ltd., China) was used to detect single particles. The design and principles of SPAMS has previously been described in detail (Li et al., 2011). In short, particles are introduced into the aerodynamic lens through a critical orifice at a flow rate of 75 mL/min (Gong et al., 2021;Li et al., 2011). Individual particles are focused and accelerated to specific velocities, which are detected by two continuous diode Nd:YAG laser beams (532 nm) and then ionized using a pulsed Nd:YAG laser (266 nm). Finally, the z-shaped bipolar time-of-flight mass spectrometer is used to detect the generated ions. The improvements and modifications from the SPAMS to the HP-SPAMS are comparatively presented below. The improvement in the SPAMS primarily includes the following three parts: the application of a concentration device, a delay extraction technology, and a multichannel acquisition technology Li et al., 2018). First, the addition of the concentrator increases the injection flow rate by 6 times, which allows for an improved separation of the gas and particles. Second, the generated ions from the laser ionization of single particles first enter the zone without an electric field. Then, the pulsed electric field will be added to accelerate the same kind of ions flying to the detector. This pulsed electric field, instead of the constant electric field, will prevent the initial deflection of the same kind of ions, and the pulsed electric field also provides sufficient energy at the appropriate time to improve the resolutions of positive and negative ions. The mass resolutions of the positive (> 1000 at maximum half-width) and negative (> 2000 at maximum halfwidth) ion spectra are then significantly improved. Third, the multichannel acquisition technology is used to divide the signal into two channels, thus detecting the high and low intensity signals simultaneously without signal loss. This new acquisition technology enables a detectable dynamic signal from 5-20 000 mV, which is approximately 40 times higher than that of SPAMS. Generally, the particle size measured by HP-SPAMS ranges from 0.2 to 2.0 µm and was calibrated by polystyrene latex particles before and after the sampling campaign (Li et al., 2011. The field observation result from the scanning mobility particle sizer (SMPS) is necessary to scale the measurements of HP-SPAMS to demonstrate the authentic size distributions of the specific type particles. Because the SMPS was not available in this sampling campaign, the size distributions of amine-containing particles were not discussed here.

Data analysis
The size and chemical compositions of single particles obtained using the HP-SPAMS were analyzed using the Computational Continuation Core (COCO) toolkit in MATLAB software. According to previous studies that have utilized aerosol time-of-flight mass spectrometer (ATOFMS) and SPAMS, the amine-containing particles were identified by querying 59 (CH 3 ) 3 N + , 74 (C 2 H 5 ) 2 NH + 2 , 86 (C 2 H 5 ) 2 NCH + 2 , 101 (C 2 H 5 ) 3 N + , 102 (C 3 H 7 ) 2 NH + 2 , and 143 (C 3 H 7 ) 3 N + (Healy et al., 2015;Angelino et al., 2001;Cheng et al., 2018;Zhang et al., 2012). In this work, the marker ions of 59 (CH 3 ) 3 N + , 74 (C 2 H 5 ) 2 NH + 2 , and 86 (C 2 H 5 ) 2 NCH + 2 were detected as the abundant species, and their particle counts and ratios in the total detected single particles are shown in Table 1. Single particles containing 86 (C 2 H 5 ) 2 NCH + 2 only accounted for 3.7 % of total particles, which was primarily due to the occasional increases on 5 September (Fig. S2), possibly due to the outburst of special emissions such as combustion and in- dustry. Thus, in this work, particles containing 59 (CH 3 ) 3 N + and 74 (C 2 H 5 ) 2 NH + 2 were selected to discuss the mixing states and formation processes of the particulate amines. The marker ions of 62 NO − 3 , 97 HSO − 4 , and 18 NH + 4 were used to identify the nitrate, sulfate, and ammonium in the aminecontaining particles . Based on field and chamber studies using SPAMS and ATOFMS, the 43 C 2 H 3 O + ion was identified as the representative oxygen-containing organic (Healy et al., 2015;Pratt et al., 2009). The particles containing 26 CN − and 42 CNO − were considered to be representative of the organic nitrogen-containing particles (Pratt et al., 2011). In addition, the 73 C 3 H 5 O − 2 and 89 HC 2 O − 4 ions were designated as glyoxylate and oxalate markers, respectively (Cheng et al., 2017;Zhang et al., 2020). It should be noted that HP-SPAMS measurements cannot provide the quantitative mass concentrations of amines and related chemical species due to the size-dependent transmission efficiencies of particles through aerodynamic lens and composition dependent matrix effect Gong et al., 2021). Currently, the observational results were used to illustrate the distinct impacts of same influencing factors on the behaviors of amine-containing particles.
3 Results and discussion

Characteristics of amine-containing particles
In this work, the amine-containing particles accounted for 32.1 % of total detected single particles, which was higher than in previously reported results for the Pearl River Delta (PRD) region (9.4 %-11.1 %) and Chongqing (8.3 %-12.7 %), China. The TMA-containing particles showed a much higher abundance in the total particles (22.8 %) than the DEA-containing particles (5.5 %; Table 1), which could have been due to their differential emissions and atmospheric processing Chen et al., 2019;Liu et al., 2020;Ge et al., 2011b). Temporal variations in meteorological parameters, PM 2.5 concentration, and the count of amine-containing particles are shown in Fig. 1. Although the TMA-and DEA-containing particles exhibited similar temporal trends at a lower particle count, their increasing peaks appeared at different periods, suggesting that the reasons for their increase in the particle count were different. Generally, peaks in the DEA-containing particles frequently appeared during the nighttime, which was possibly due to their enhanced source emissions and/or favorable nighttime production (Tang et al., 2013). The ambient RH was relatively high during the entire sampling period (74 % ± 14 %), especially from 5-7 September, when the count of TMA-containing particles sharply increased. However, no obvious enhancement in DEA-containing particles count was found, which suggested other influencing factors on their formation process in addition to the ambient RH.
Additionally, the periods of high concentration of the amine-containing particles were not consistent with the increase in PM 2.5 concentration, which could have been due to the integrated effects of the emission sources and the secondary formation processes. The gaseous TMA and DEA are mainly from agriculture, industry, vehicle exhaust, biomass combustion, biological, and marine sources (Zhou et al., 2019;Hemmilä et al., 2018;Sintermann et al., 2014;Ge et al., 2011b;Zhang et al., 2017). Their concentrations vary greatly, depending on the influence of source strength near the sampling site. For example, the gaseous concentration of DEA was 14 and 2-5 times higher than TMA in polluted urban areas in China (Yao et al., 2016) and the USA (You et al., 2014), respectively, while a higher concentration of TMA than DEA was observed in the forest site (You et al., 2014). Both the online and offline measurements are difficult to quantitatively resolve their emission sources (You et al., 2014;Yao et al., 2016;Kieloaho et al., 2013;Hellén et al., 2014). Here the backward trajectories of the air masses from sampling site were discussed to explore their possible different sources. The backward trajectories of the air masses (48 h; 500 m) associated with the spatial distributions of the two amine-containing particles during the entire sampling period (Fig. 2). More than 70 % of the air masses (clusters 1 and 4) were from east of the sampling site, which were both connected with anthropogenic emissions in the YRD and marine sources in the East China Sea. TMA-containing particles were primarily from the air masses of cluster 1 and cluster 4, while the DEA-containing particles were associated with the air masses of cluster 3 and cluster 4, which underwent long-range transport. These results suggested potential different emission sources and atmospheric formation processes of TMA-and DEA-containing particles, which was further investigated by examining their mixing states.
Diurnal variations in TMA-and DEA-containing particles are shown in Fig. 3. The particle count of TMA-containing particles and their abundance in the total particles exhibited identical variation patterns. According to the molecular characterization of particles from vehicle exhaust, TMA was detected as one of the directly emitted organics from vehicle exhaust Li et al., 2020), which was in accordance with the field studies conducted during traffic hours Chen et al., 2019). Thus, the significant increase in TMA-containing particles in the morning was possibly associated with direct emissions from vehicle exhaust. The DEA-containing particles showed a completely different diurnal pattern compared with the TMA-containing particles. DEA-containing particles increased during nighttime, but sharply decreased during the afternoon, when the photochemistry was the most active. The nighttime increase could have been due to the high ambient RH and/or enhanced heterogeneous reactions (Zhou et al., 2019;Huang et al., 2012;Zhang et al., 2019). However, since the increase in the DEA-containing particles was not prominent under high RH (Fig. 1), an enhanced heterogeneous production of particulate DEA could be a more reasonable explanation. The decrease in the DEA-containing particles during the afternoon could have been associated with the photodegradation of DEA and/or repartitioning of particulate DEA under high temperatures during the day (Ge et al., 2011a;Pitts et al., 1978;Murphy et al., 2007). In order to investigate the impact of RH on the formation process of amine-containing particles, the particle counts of amine particles and the relative peak areas (RPAs) of amines in the particles with an increase in the RH are presented in Fig. 4. The particle count of TMA-containing particles and the RPA of the TMA showed remarkable increasing trends, with an enhancement in RH during the entire sampling period. This result suggested a significant role of RH in the formation of particulate TMA. This was consistent with a field study conducted in Guangzhou, China, which also found an instant increase in TMA-containing particles after the occurrence of fog events . In contrast, the particle count of DEAcontaining particles only exhibited an increased RH range between 70 %-80 % and decreased with the continuous increase in the RH. Additionally, the RPA of the DEA showed little change, with an increase in the RH, which suggested the minor influence of a change in RH on the formation of particulate DEA. The different responses of TMA and DEA with RH changes signified their different formation processes in the particles.

Different mixing states of amine-containing particles
It is important to understand the chemical compositions of amine-containing particles in order to understand their mixing states and track their formation processes. Hence, the positive and negative mass spectra of the amine-containing particles are shown in Fig. 5. Generally, TMA-and DEAcontaining particles both contained organic fragments, such as 27 C 2 H + 3 , 37 C 3 H + , 43 C 2 H 3 O + , 51 C 4 H + 3 , and 61 C 5 H + , in the positive mass spectra. In addition, their negative mass spectra were both characterized by nitrate, sulfate, and nitric acid ( 125 H(NO 3 ) − 2 ). However, DEA-containing particles contained many more organic fragments and a higher abundance of hydrocarbon clusters than the TMA-containing particles. In the positive mass spectra, the abundance of the hydrocarbon fragments with an m/z below 60 was 2-3 times higher in DEA-containing particles than that in TMA-containing particles. In addition, hydrocarbon fragments with an m/z above 60 were barely detectable in TMAcontaining particles, while abundant hydrocarbon fragments with an m/z ranging from 60-150 were observed in DEAcontaining particles. Furthermore, the DEA-containing particles also contained abundant secondary organic marker ions, including organic nitrogen ( 26 CN − and 42 CNO − ), acetate ( 59 C 2 H 3 O − 2 ), glyoxylate ( 73 C 3 H 5 O − 2 ), and oxalate ( 89 HC 2 O − 4 ) in the negative mass spectra, and these were not found in the TMA-containing particles. This was in accordance with the linear regressions between these secondary organic ions containing particles with two amine-containing particles (Table 2), which showed no correlations in the TMA-containing particles (r 2 < 0.1) but good correlations in the DEA-containing particles (r 2 > 0.57). The differential mass spectral features in the distributions of organics in the two amine-containing particles (Fig. 6) suggested that more secondary organics accumulated in DEA-containing particles than in TMA-containing particles. This result also implied that multiple factors influenced the mixing state of DEA-containing particles in addition to ambient RH.
In order to further characterize the mixing states of DEAcontaining particles with secondary organic ions, temporal variations and diurnal patterns of secondary organic ions in the DEA-containing particles are presented in Fig. 7. As the oxidation products of various organics, the abundances of glyoxylate and oxalate commonly increased between 12:00  . The diurnal variations in particle counts and number fractions of the two amine-containing particles in total particles during the entire sampling period. and 18:00 LT (local time; Fig. 7), when the photochemistry was most active during the daytime. This result suggested the deep photochemical aging state of DEA-containing particles. This might explain the decrease in the particle counts of DEA-containing particles (Fig. 3), which was partially associated with the photo degradation of particulate DEA. Pitts et al. (1978) reported that, under sunlight, particulate DEA decomposed to acetamide, while DEA in the gas phase was oxidized to acetaldehyde, PAN, amide, and imine. Gaseous DEA can be oxidized into carbonyl compounds and other amines by ozone and OH radicals that primarily include acetaldehyde and N-ethylethanimine (Tuazon et al., 2011;   et al., 2020). The organic nitrogen markers of 26 CN − and 42 CNO − showed different temporal trends with glyoxylate and oxalate. Although the abundances of organic nitrogen markers also increased after 12:00 LT, like oxalate, the markers still maintained high abundances during the nighttime when glyoxylate and oxalate sharply decreased. This result suggested that the aging process of organics during the nighttime was slower than that during the afternoon in the DEAcontaining particles. Thus, the increase in DEA-containing particles (Fig. 3) could have been due to the enhanced production of particulate DEA during the nighttime. In addition, the differential mass spectra of DEA-containing par-ticles (Fig. 8) between the nighttime (22:00-02:00 LT) and daytime (14:00-18:00 LT) showed a significant enrichment of nitrate during the nighttime. This result suggested that nighttime production of particulate DEA was associated with gaseous HNO 3 and/or particulate nitrate (Price et al., 2016).
Temporal variations in NO x and the N f of the DEAcontaining particles in the total detected particles are presented in Fig. 9. They showed similar increasing patterns during the nighttime, and a high abundance of nitrate in the DEA-containing particles was also observed. This result suggested the important role of nitrate in the formation of particulate DEA. The particulate DEA during the nighttime could  have been produced from the reaction of gaseous DEA with HNO 3 during the gas phase (Reaction R1), followed by the gas-to-particle partitioning (Reaction R2) and/or the direct heterogeneous formation pathway (Reaction R3; Price et al., 2016;Nielsen et al., 2012). The high ambient concentration of NO x is favorable for the production of NO 3 radicals and the heterogeneous production of nitrate, which might explain the distinct enhancement of the DEA-containing particles. However, the same formation pathways were also applied to TMA, yet there was no significant increase in the N f of the TMA-containing particles in the total particles (Fig. 3). This could have been due to the different particle/gas dissociation constant (K p ) for DEA q HNO 3 and TMA q HNO 3 , which was several orders of magnitude lower than that for DEA q HNO 3 (7.01 × 10 −9 ) compared with TMA q HNO 3 (1.65 × 10 −6 ) at 25 • C (Price et al., 2016;Ge et al., 2011a). During the entire sampling period, the ambient temperature during the nighttime was approximately 24 • C. Thus, the produced DEA q HNO 3 tended to stay in the particles, while a portion of the TMA q HNO 3 repartitioned back into the gas phase. This resulted in an insignificant increase in the TMA-containing particles. Further studies should consider the influence of the different volatilities of DEA q HNO 3 and TMA q HNO 3 on the formation of particulate amines in chamber experiments due to the lack of quantitative results in this study.

Formation of aminium salts
To study the acid-base reactions of TMA and DEA with sulfate and nitrate, the N f s of nitrate-, sulfate-, and ammoniumcontaining particles in total detected particles and aminecontaining particles are shown in Table 3. More than 80 % of TMA-and DEA-containing particles internally mixed with nitrate, which was higher than the N f of nitrate in the total particles (72 %). Interestingly, the N f of sulfate in DEAcontaining particles (79.3 %) was much higher than that in TMA-containing particles (55.3 %) and in the total particles (60.1 %). This was similar to a study performed by Lian et al. (2020), who found a stronger correlation between 86 (C 2 H 5 ) 2 NCH + 2 with sulfate than that between TMA and sulfate. In addition, in this work, robust linear correlations (r 2 > 0.9) between nitrate-containing particles and amine particles were both observed in TMA-and DEAcontaining particles (Table 2). However, a weak linear correlation (r 2 = 0.32) was found between the sulfate-containing particles and the TMA-containing particles, while a better linear correlation (r 2 = 0.86) was observed in the DEAcontaining particles. According to reported studies, the vapor pressure of diethylaminium sulfate (DEAS) (0.2 × 10 −12 -12.8 × 10 −12 Pa) was 3 orders of magnitude lower than that of trimethylaminium sulfate (TMAS) (0.6 × 10 −9 -1.8 × 10 −9 Pa) at 298 k. In addition, the enthalpy of evaporation was higher than that of TMAS (DEAS -168 ± 5 kJ mol −1 ; TMAS -114 ± 2 kJ mol −1 ; Lavi et al., 2013). Therefore, the thermostability of DEAS was stronger than TMAS (Qiu and Zhang, 2012), which led to the higher N f of sulfate in the DEA-containing particles than in the TMA-containing particles.
The N f of ammonium in DEA-containing particles (13.2 %) was lower than in TMA-containing particles (35 %) and total particles (19.4 %). The low abundance of NH + 4 in DEA-containing particles had been observed in our previous studies in the PRD region , which was partially attributed to the ammonium-amine exchange reactions in the particles. The related laboratory experiments primarily involved the preferential uptake of LMW amines in the H 2 SO 4 particles (Sauerwein and Chan, 2017;Chan and Chan, 2013;Chu and Chan, 2017). In this work, the distinct low N f of NH + 4 in the DEA particles suggested the possible displacement of NH + 4 by DEA. Moreover, the higher abundance of sulfate in DEA particles than in TMA particles was more favorable for the occurrence of ammonium-amine exchange reactions in DEA particles. This disparity could imply different roles of DEA and TMA in the new particle formation process (Wang et al., 2010;Yin et al., 2011;Zhao et al., 2011).
The temporal trends of the N f s of nitrate-, sulfate-, and ammonium-containing particles in TMA and DEA particles are shown in Fig. 10. The N f of nitrate-containing amine particles exhibited similar variation patterns with each type of amine particle, while the N f of sulfate-containing amine particles only showed a similar variation pattern with DEAcontaining particles. Although aminium nitrate and sulfate salts were both produced in TMA-and DEA-containing particles, the different temporal trends of sulfate and nitrate in the two amine particles suggested that both sulfate and nitrate DEA salts existed in the DEA-containing particles, while nitrate TMA salt dominated in TMA-containing particles Pratt et al., 2009). This difference in the form of aminium salts could signify the potential different influences in the hygroscopic property of secondarily processed particles internally mixing with different amines (Rovelli et al., 2017;Clegg et al., 2013;Lavi et al., 2013). The relative acidity ratio (Ra), defined as the ratio of the sum of the sulfate and nitrate peak areas to the ammonium peak area, has been proposed in field studies that use single particle mass spectrometry to roughly estimate particle acidity (Huang et al., 2013;Cheng et al., 2018). Although the feasibility of Ra has been supported by the robust linear correlation, with the authentic particle acidity calculated from mass concentrations of inorganic ions (Huang et al., 2013), this semi-quantitative approach should be carefully treated when it comes to the discussion about the actual acidity of atmospheric particles. In this work, temporal variations in Ra in TMA-containing particles (Ra 1 ) and DEAcontaining particles (Ra 2 ) are shown in Fig. 11. The average Ra 1 was 6.3 ± 1.8 in TMA-containing particles, and Ra 2 was 36.1 ± 21.8 in DEA-containing particles. This result could suggest a more acidic nature of DEA particles than TMA particles. However, after including the peak area of amines (TMA/DEA) in the calculation of Ra, the new Ra reduced to 2.1 ± 0.5 in the TMA-containing particles and 6.5 ± 1.2 in the DEA-containing particles. The gap of Ra between the two amine particles significantly decreased after including amines in the calculation. The larger reduction ratio of Ra in DEA-containing particles than in TMA-containing particles  suggested the effective buffering effect of amines under the absence of ammonium in the particles.

Implications of the diverse mixing states of amines particles
The mixing states and formation processes of the two aminecontaining particles were investigated under the same atmospheric environment, and their different atmospheric behaviors against the same influencing factors suggested their differential contributions to SOA mass. The prominent impact of ambient RH on the formation of particulate TMA suggested a significant role for the gas particle partitioning process to the high water soluble species in the SOA. However, the slight influence of RH on the formation of the particulate DEA implied the inconsistent role of high RH on the same group of water soluble organic molecules. In addition, the distinct distribution patterns of secondary organic species in two amine-containing particles also signified that the mixing states of the OA are important to explore their formation processes. Furthermore, the heterogeneous processing of the DEA-containing particles during the nighttime, and the photochemical degradation of the DEA during the daytime, both generated more fractions of nitrogen-and oxygen-containing species in the particles than in the TMA-containing particles. This result suggested different roles of particulate TMA and DEA in the evolution of the hygroscopicity and aging state of the SOA. In summary, understanding mixing states and formation processes of different amines in single particles is of great significance to reveal the unique response of each type of amine to the same atmospheric environment. Singleparticle analysis provided insights into the mixing states of specific organic species to further understand the formation process of the SOA.

Summary and conclusions
TMA-and DEA-containing single particles were collected and analyzed on September 2019, using HP-SPAMS in Nanjing, China, and accounted for 22.8 % and 5.5 % of total detected particles, respectively. The mixing states and formation processes of TMA-and DEA-containing particles were studied. With increased RH, the counts of particulate TMA and the RPA of the TMA displayed an obvious upward trend, while the particle count of the particulate DEA slightly increased when the RH was 70 %-80 %. In addition, the RPA of the DEA showed no difference in reaction to the RH change during the entire sampling period. This suggested a Figure 10. Temporal trends of TMA-and DEA-containing particle counts, and number fractions of nitrate, sulfate, and ammonium in TMAand DEA-containing particles. Figure 11. Temporal trends of the relative acidity ratios (Ra; Ra ) in TMA-and DEA-containing particles.
differential role for ambient RH during the formation processes of particulate TMA and DEA. The possible formation processes were further evaluated by analyzing the mixing states of the amine-containing particles. The mass spectra of the amine-containing particles showed that the secondary organic species were enriched in the DEA-containing particles. The differential distributions of the secondary ions effectively explained the sharp increase in DEA-containing particles during the nighttime, which could have been due to the heterogeneous reactions of gaseous DEA with HNO 3 and/or nitrate particles. The prominent decrease in the DEAcontaining particles during the afternoon was attributed to photo-degradation of particulate DEA. Due to the differences in the thermodynamic properties, the N f of sulfate in the particulate DEA was higher than that in the particulate TMA. The amine-ammonium exchange reaction resulted in particulate DEA containing less NH + 4 . In addition, the particulate DEA was abundant in sulfate, which was more favorable for the exchange of amine and ammonium. The higher relative acidity ratio in DEA-containing particles relative to TMA-containing particles could suggest that DEA particles are more acidic. After including the peak area of amines (TMA/DEA) in the calculation, the larger reduction ratio of the Ra in DEA-containing particles than in TMA-containing particles suggested the effective buffering effect of amines under the absence of ammonium in the particles. These re-sults revealed the distinct mixing states and chemical behaviors of TMA-and DEA-containing single particles and could imply a potential role for DEA as an indicator of the OA aging process.
Data availability. The observational data, including HP-SPAMS and the meteorological parameters, obtained in this study are available from the corresponding author upon request (chengcl@jnu.edu.cn).
Author contributions. QEZ, CC, and ZW developed the methodology and wrote the original draft. DG, LW, YL, WN, XC and AD were responsible for the methodology and sampling. LL, ML, SY, DC, and ZZ engaged in discussions and helped to revise original draft.
Competing interests. At least one of the (co-)authors is a member of the editorial board of Atmospheric Chemistry and Physics. The peer-review process was guided by an independent editor, and the authors also have no other competing interests to declare.
Disclaimer. Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.  Review statement. This paper was edited by Alex Lee and reviewed by Weijun Li and one anonymous referee.