Summertime Day-Night Differences of PM2.5 Components (Inorganic Ions, OC, EC, WSOC, WSON, HULIS, and PAHs) in Changzhou, China

This work reports the day-night differences of a suite of chemical species including elemental carbon (EC), organic carbon (OC), water-soluble organic carbon (WSOC), water-soluble organic nitrogen (WSON), selected polycyclic aromatic hydrocarbons (PAHs), and secondary inorganic ions (NO3, SO4, NH4) in ambient fine particles (PM2.5) collected from 23 July to 23 August 2016 in Changzhou, China. Mass concentrations of PM2.5 and SO4 show a 10–20% increase during daytime, while NO3 concentration decreases by a factor of three from nighttime to daytime due to its semi-volatile nature. PAHs, EC, and WSON show higher mass concentration in the night too. Mass ratios of WSOC to OC are high in both day and night, indicating that secondary organic aerosol (SOA) formation could occur throughout the day, while the slightly higher ratio during daytime suggests a more significant contribution from daytime photochemical oxidation. Strong positive correlations between HULIS-C and WSOC, and HULIS-C with O3 both in day and night, imply that HULIS-C, similar to WSOC, is mainly composed of secondary species. HULIS-C accounted for a large fraction of WSOC, with an average of ~60%. Moreover, the average WSON concentrations are 1.08 and 1.46 μg/m3, constituting ~16% and ~18% of water-soluble total nitrogen in day and night, respectively. Correlation analyses suggest that WSON is also predominantly produced from secondary processes. PAHs concentrations are found to be very low in summer aerosols. Overall, our findings highlight the dominant contribution of secondary processes to the major aerosol components in Changzhou, suggesting proper measures to effectively reduce gaseous precursors are also important to improve air quality.


Introduction
Atmospheric fine aerosols (PM 2.5 ), derived from natural and anthropogenic sources, can significantly affect air quality, visibility, human health and the earth's climate. Organic aerosol (OA), constituting a large fraction (~20-90%) of PM 2.5 , contains multiple species that are potentially mutagenic and carcinogenic (such as polycyclic aromatic hydrocarbons (PAHs)) and therefore is particularly harmful to human health [1,2]. As a significant fine aerosol component, OA also contributes to the formation and evolution of atmospheric haze [3]. Despite the increasing number of studies on were collected. Before and after sampling, the filters were conditioned under constant temperature (22 ± 1 • C) and relative humidity (45 ± 5%) for 48 h and weighed by a microbalance (precision of 0.01 mg). The sampled filters were wrapped, sealed in aluminum foil envelopes separately, and stored in a freezer at −20 • C until analysis. Note the samples were analyzed immediately after the campaign, so all samples were analyzed within one month after sampling. Of course, quartz filters can absorb organic gases and hence introduce artifacts, so we used back-to-back blank filters that were treated in the same manner to minimize the influences especially on the carbonaceous species and gravimetric measurements [30].
Meteorological parameters and gaseous O 3 concentrations were recorded at the air quality monitoring station inside the campus, which is about 500 m from the sampling site. The temperature varied from 15.4 to 32.4 • C and the relative humidity (RH) varied from 30 to 100% during the sampling period. The mean temperatures in day and night were 27.5 • C and 18.3 • C, respectively. The average O 3 concentrations were 120.2 and 54.9 µg/m 3 , and the wind speeds (WS) were 2.5 and 1.5 m/s in day and night, respectively.

Water soluble species (WSOC, WSTN/WSON, and inorganic ions):
A quarter of each filter was cut into pieces by using a ceramic scissor, then extracted by 100 mL Milli-Q water (18.2 MΩ cm) under ultra-sonication for 45 min. The extracted solution was filtered through a syringe with 0.2 µm pore size (Whatman, Anotop 25). Then, the extracts were analyzed by a Shimadzu TOC-V CPH analyzer with a TNM-1 unit for the total organic carbon (referred here as WSOC) and total nitrogen (referred here as WSTN) using a thermo-catalytic oxidation approach, as described previously [31]. The major inorganic ions (SO 4 2− , NH 4 + , NO 3 − , NO 2 − , and K + ) of the extracts were analyzed by an ion chromatograph (Aquion, ThermoFisher Scientific, Waltham, MA, USA). Ion chromatograph equipped with a Dionex CS12A column was used for cation analysis (20 mM methanesulfonic acid as eluent), and a Dionex AS23 column was employed for the anions (4.5 mM NaHCO 3 and 0.8 mM Na 2 CO 3 as eluent). The detection limits are determined as three times the standard deviation of blank filters (0.06 mg/L for WSOC, 0.08 mg/L for WSTN, 0.19, 0.20, 0.23, 0.11, and 0.11 mg/L for NH 4 + , K + , NO 3 − , NO 2 − , and SO 4 2− ).
More details regarding the analytical methods can be found in our previous work [29]. The concentration of WSON was determined by subtracting the water soluble inorganic nitrogen (WSIN) from the WSTN as follows: OC and EC: OC and EC were measured using a thermal/optical carbon analyzer (DRI Model 2001A, Desert Research Institute, Las Vegas, NV, USA) following the temperature protocol described by Chow et al. [32]. A punch of each filter was measured stepwise at temperatures of 140 • C (OC 1 ), 280 • C (OC 2 ), 480 • C (OC 3 ), and 580 • C (OC 4 ) in a helium atmosphere, and 580 • C (EC 1 ), 740 • C (EC 2 ), and 840 • C (EC 3 ) in a 2% oxygen/98% helium gas atmosphere. OC is calculated as OC 1 + OC 2 + OC 3 + OC 4 + OP and EC as EC 1 + EC 2 + EC 3 − OP, where OP is the optical pyrolyzed OC [33]. Detection limits of OC and EC were 0.82 µgC/cm 2 and 0.20 µgC/cm 2 , respectively.

HULIS-C:
In this work we focused on measurements of the carbon contained in HULIS (HULIS-C) rather than HULIS. The HULIS fraction was isolated from PM 2.5 sample by solid phase extraction (SPE) technique, as previously described [13,34]. In brief, 1/4 filter was extracted in 50 mL Milli-Q water with ultrasonication for 45 min and then filtered through the syringe with 0.2 µm pore size (Whatman, Maidstone, UK, Anotop 25) to obtain the water extract (WE). The WE was adjusted to pH ≈ 2 using 2 M HCl (Sigma-Aldrich, Saint Louis, MO, USA), and passed through pre-conditioned ENVI-18 SPE cartridges (60 mg/cartridge, CNW, ANPEL laboratory Technologies (Shanghai) Inc., Shanghai, China) with a flow rate of 1 mL/min. Thus, hydrophilic HULIS was retained by SPE cartridge. Note that the ENVI-18 was pre-conditioned with 3 mL methanol and 3 mL water. Then, 2 mL Milli-Q was injected into the SPE cartridge to rinse column in order to remove residues of inorganic constituents, then 5 mL 2% ammonia/methanol mixture (CNW, ANPEL laboratory Technologies (Shanghai) Inc., Shanghai, China) was introduced into the SPE cartridge to elute retained organics (HULIS). Finally, the eluted HULIS was dried under nitrogen gas flow until full dryness; subsequently, it was diluted with 25 mL Milli-Q and then measured by the same TOC analyzer used for WSOC determination (referred here as HULIS-C).
Two standard humic-like substances (Suwannee River fulvic acid, SRFA, 2S101F; Pahokee Peat Humic Acid, PPHA, 1S103H, St. Paul, MN, USA) were purchased from the International Humic Substances Society. We verified the method by adding 20 mL of 10 mgC/L SRFA solutions to blank filters to assess the recovery efficiency and reproducibility. The results showed that our SPE method yielded recovery ratios of 82-90% for SRFA with excellent reproducibility. These results indicate that the ENVI-18 SPE method can efficiently recover HULIS from filters.
The ultraviolet-visible (UV-Vis) spectra of both WSOC and HULIS in each sample were measured by using a UV-Vis spectrometer. We recorded the absorbance at λ = 250 nm (E 250 ) and λ = 365 nm (E 365 ) and E 250 /E 365 ratio.
As mentioned earlier, back-to-back blank filters were treated in the same manner as for the samples, and it is worth to mention that all results reported here were subjected to blank corrections. Table 1 summarizes the concentrations of PM 2.5 , OC, EC, WSOC, HULIS-C, WSTN, WSON, ΣPAHs, the major inorganic ionic species (ammonium, sulfate, and nitrate), as well as the (OC + EC)/PM 2.5 , WSOC/OC, HULIS-C/WSOC, WSON/WSTN ratios in day and night. The PM 2.5 concentrations in day vary from 40.9 to 136.3 µg/m 3 with an average of 77.3 µg/m 3 , whereas they vary from 33.3 to 119.3 µg/m 3 in night with average value of 68.9 µg/m 3 . These values are generally close to 75 µg/m 3 -the second-grade national air quality standard (NAAQS) (GB 3095-2012).

Total Concentration and Compositions of PM 2.5
In Figure 1a, we present the day-to-night mass ratios of some selected chemical species in PM 2.5 . On average, the PM 2.5 mass, OC, WSOC, and HULIS-C are relatively higher (12-36%) in day than those in night, while ΣPAHs, EC, WSTN, and WSON show opposite day-night variation trends. As for the inorganic species, SO 4 2− concentration is~17% higher during daytime, and the difference between day and night samples regarding NO 3 − is very striking, being consistent with day-night variability of nitrate observed in Kanpur, India [35]. The remarkably low levels of nitrate in the day are likely attributed to the evaporation of NH 4 NO 3 due to higher temperature during daytime and/or condensation of nitrate under relatively low temperatures and high RH during nighttime [36].
Overall, the enhancement of secondary aerosol formation is likely a dominant cause for the increase of PM 2.5 concentrations during daytime. In Figure 1a, we present the day-to-night mass ratios of some selected chemical species in PM2.5. On average, the PM2.5 mass, OC, WSOC, and HULIS-C are relatively higher (12-36%) in day than those in night, while ΣPAHs, EC, WSTN, and WSON show opposite day-night variation trends. As for the inorganic species, SO4 2− concentration is ~17% higher during daytime, and the difference between day and night samples regarding NO3 − is very striking, being consistent with day-night variability of nitrate observed in Kanpur, India [35]. The remarkably low levels of nitrate in the day are likely attributed to the evaporation of NH4NO3 due to higher temperature during daytime and/or condensation of nitrate under relatively low temperatures and high RH during nighttime [36]. Overall, the enhancement of secondary aerosol formation is likely a dominant cause for the increase of PM2.5 concentrations during daytime. The average WSOC/OC ratio for samples collected during daytime (60.3%) is higher than that for the nighttime samples (54.8%), similar to results observed in Kanpur, India [35]. The slightly higher WSOC/OC in daytime samples can be attributed to the enhanced SOA formation from atmospheric oxidation of VOCs under strong solar irradiation.
The mass proportions of OM (estimated as 1.6 × OC), EC, and secondary inorganic ions (SO4 2− , NH4 + and NO3 − ) to the PM2.5 mass are illustrated in Figure 1b. On average, OM accounts for ~31% of PM2.5 mass in both daytime and nighttime samples. A large proportion of PM2.5 mass is unidentified, which means other constituents, such as other heteroatoms bound in organic matters (oxygen, hydrogen, and nitrogen), other organic species (in particular the water-insoluble species), mineral dust, heavy metals, and other inorganic species are also important. The water uptake by the aerosol components (especially ammonium nitrate and sulfate) [37] can also influence the mass closure. The relatively high HULIS-C contents (that The average WSOC/OC ratio for samples collected during daytime (60.3%) is higher than that for the nighttime samples (54.8%), similar to results observed in Kanpur, India [35]. The slightly higher WSOC/OC in daytime samples can be attributed to the enhanced SOA formation from atmospheric oxidation of VOCs under strong solar irradiation.
The mass proportions of OM (estimated as 1.6 × OC), EC, and secondary inorganic ions (SO 4 2− , NH 4 + and NO 3 − ) to the PM 2.5 mass are illustrated in Figure 1b. On average, OM accounts for~31% of PM 2.5 mass in both daytime and nighttime samples. A large proportion of PM 2.5 mass is unidentified, which means other constituents, such as other heteroatoms bound in organic matters (oxygen, hydrogen, and nitrogen), other organic species (in particular the water-insoluble species), mineral dust, heavy metals, and other inorganic species are also important. The water uptake by the aerosol components (especially ammonium nitrate and sulfate) [37] can also influence the mass closure. The relatively high HULIS-C contents (that will be shown later) may be in part responsible for the low reconstituted fraction since HULIS usually has a higher OM/OC ratio (>1.9) than 1.6 [38]. In addition, the WSTN concentrations (day: 5.39 µg/m 3 , night: 6.92 µg/m 3 ) can also contribute to the unquantified PM 2.5 mass.

OC/EC Ratios and Correlation between OC and EC
The time series of concentrations of PM 2.5 , EC, OC, and O 3 , as well as OC/EC ratios and WS, are plotted in Figure 2a,b. Table 2 shows Pearson's correlation (r) among various components, and their correlations with O 3 and WS for both day and night samples. As presented in Table 2, all species negatively correlate with WS both in day and night, indicating high WS can dilute and reduce the accumulation of all particle-phase pollutants. PM 2.5 , OC, and EC all show positive correlations with O 3 , suggesting that the secondary formation of aerosols through the atmospheric oxidation processes may be considerable. The average OC and EC mass concentrations are 15.1 and 3.2 µg/m 3 during daytime and 13.5 and 3.3 µg/m 3 during nighttime (Table 1). Scatter plots of OC and EC and their correlation coefficients (r) are shown in Figure 2c. The correlations between OC and EC are relatively weak (r of 0.40 and 0.53 for day and night), suggesting additional secondary sources of OC other than the primary sources as for EC. Average OC/EC ratios are >4, likely supporting the assumption of secondary OC formation. But we need to be aware that the justification of secondary OC based on the OC/EC ratios alone is valid only after careful inspection of local sources of OC and EC [39], as some primary OA (such as biomass burning emissions) may have high OC/EC ratios as well.
Furthermore, as presented in Figure 2d, OC and O 3 show better correlation (r of 0.74) for daytime samples than that of nighttime samples (r of 0.55), likely indicating that more secondary OC is produced during daytime as O 3 is exclusively a secondary photochemical product. Of course, the positive correlation between OC and O 3 in night might also imply that partial OC observed in the night are products of daytime oxidation.    Figure 3 presents concentrations of eight fractions of carbon (i.e., OC 1 to OC 4 , OP and EC 1 to EC 3 ) and their average contributions to the total carbon mass (OC + EC), which can reflect the different features of OA [40]. In this study, OC 1 concentration only accounted for about 1% of total carbon, which was very low compared with other OC types. Since OC 1 and OC 2 evaporate at a temperature of less than 280 • C, they are usually treated as volatile and semi-volatile. The fractions of OC 1 + OC 2 in OC are 22.2% and 25.2% for day and night samples, respectively, indicating low mass percentages of volatile and semi-volatile species in summer OA. On the other hand, OC 3 + OC 4 accounts for near half of the total carbon mass, indicating a large fraction of organic compounds with large molecular weights and low volatility [41]. OP accounts for~11%, further supporting that partial low-volatility species exist in OC. Recently, a combination of the OC/EC results with those of high resolution mass spectrometry data has shown to be useful to identify the sources of OA [41], which will be the subject of our future work.

Characteristics of OC and EC
Overall, the average mass contributions of eight carbon fractions, however, are similar in day and night samples. This is likely due to the complex combined effects from mixed contributions from different sources, secondary oxidation processes, and the meteorological conditions (boundary layer height, wind speed, and direction, etc., which can affect the dispersion of aerosols). of our future work.
Overall, the average mass contributions of eight carbon fractions, however, are similar in day and night samples. This is likely due to the complex combined effects from mixed contributions from different sources, secondary oxidation processes, and the meteorological conditions (boundary layer height, wind speed, and direction, etc., which can affect the dispersion of aerosols). OC, EC, and OP stand for organic carbon, elemental carbon, and pyrolyzed carbon, respectively. Error bar represents the standard deviation.

WSOC versus OC and O3
The WSOC concentration and WSOC/OC ratios can be used to infer SOA formation and significance. As shown in Figure 4a, good correlation is found between WSOC and OC whether in day or night. WSOC has good correlations with O3 in both day and night too ( Figure 4b). It is well known that in summer, the SOA formation during daytime can be attributed to the photochemical reactions under solar irradiation [42], which can be also inferred from the good correlation of OC with O3 in Figure 2d. The WSOC/OC ratios are only slightly lower than those during daytime. On one hand, this result may indicate that the residual SOA formed during daytime survives in night.  The WSOC concentration and WSOC/OC ratios can be used to infer SOA formation and significance. As shown in Figure 4a, good correlation is found between WSOC and OC whether in day or night. WSOC has good correlations with O 3 in both day and night too ( Figure 4b). It is well known that in summer, the SOA formation during daytime can be attributed to the photochemical reactions under solar irradiation [42], which can be also inferred from the good correlation of OC with O 3 in Figure 2d. The WSOC/OC ratios are only slightly lower than those during daytime. On one hand, this result may indicate that the residual SOA formed during daytime survives in night. But on the other hand, we cannot exclude the SOA formed during night as well. In fact, SOA formation under dark conditions via nitrate radical (NO 3 ) oxidation reactions is also possible and efficient [43,44]. Further investigations should be done in the future. But on the other hand, we cannot exclude the SOA formed during night as well. In fact, SOA formation under dark conditions via nitrate radical (NO3) oxidation reactions is also possible and efficient [43,44]. Further investigations should be done in the future. Indeed, NO3 radical is the most important atmospheric oxidant in night, which can contribute to both nitrate and SOA formations, as follows:    (6) and (7)), it may contribute to the enhanced nitrate concentration we observed in night due to higher RH. NO 3 radical (Reaction 9) can also react with VOCs to form SOA [45]. It is very likely that both dark (in night) and photochemical reactions (day) can lead to secondary aerosol formation

HULIS-C and Its Relationships with WSOC, K + , and O 3
Sources and processes leading to formation of HULIS are diverse, including primary sources (marine emission, vehicle exhaust, biomass burning) and secondary sources (oxidation of soot, aqueous reaction in cloud droplet, and formation of SOA). It has been suggested [46] that a major amount of HULIS compounds can associate with biomass burning emissions. Here, the correlation between HULIS-C and the biomass burning emission marker (K + ion) is shown in Figure 5a. The correlations between HULIS-C and K + are not strong in day and night (day: r = 0.58, night: r = 0.28), indicating that biomass burning emission is not a dominant HULIS source. Correlations between HULIS-C and EC, and HULIS-C and PAHs (Table 2) are also poor in both daytime and nighttime samples, indicating that fossil fuel combustion is not an influential source of HULIS in summer either, consistent with results in summer Marseilles [19]. On the other hand, one can find strong positive correlations between HULIS-C and WSOC (day: r = 0.94; night: r = 0.96), and HULIS-C with O 3 (day: r = 0.83; night: r = 0.90) ( Table 2). It is therefore reasonable to infer that HULIS-C, similar to WSOC, closely links with secondary formation. Also, it is also possible that the HULIS is formed secondarily from oxidation of primary emissions (both biomass burning and fossil fuel combustion emissions can be precursors). Recently, Lin et al. [47,48] reported a linear relationship between HULIS-C and WSOC, and also suggested that the increases are caused by the secondary chemical processes. The dominant role of photochemical production of HULIS-C is also found in Shanghai during summer [49]. Kuang et al. [50] show that the secondary formation processes contribute 49-82% of ambient HULIS. Additionally, linear regression slopes in Figure 5b show that HULIS-C/WSOC ratio (0.64) in daytime was slighter higher that in nighttime (0.57), suggesting that the contribution of photochemical production of HULIS-C in day might be more significant than that during night. The HULIS-C/WSOC ratio in this study is comparable to that observed in Shanghai (0.6) [16] and Hongkong (0.57) [50], but higher than that in Lanzhou (0.46) [51].
Here, we show the UV-Vis absorption spectra of HULIS-C and WSOC by using the samples collected on 31 July and 10 August 2016 in Figure 6. Note these two samples have high and low PM 2.5 mass loadings, respectively, and the shapes of their UV-Vis spectra are also representative of the average pattern of all samples. For comparison, the UV-Vis spectra of standard SRFA and PPHA are also presented in Figure 6. Generally, the absorbance of both HULIS-C and WSOC increases with the decreasing wavelength in the range of 200-500 nm, but the WSOC absorbs relatively more light compared to that of HULIS-C at the same wavelength. Compared with the two standard humic substances, the UV-Vis characteristics of both HULIS-C and WSOC are significantly different, with higher absorbance at 365 nm for the latter, indicating they contain different compounds. Such differences are closely relevant with different oxidation reactions of different precursors. For example, our UV-Vis spectra are qualitatively similar to those of pure guaiacol and catechol, but different from those from the guaiacol and catechol-derived SOA [52].
Atmosphere 2017, 8,189 9 of 16 closely links with secondary formation. Also, it is also possible that the HULIS is formed secondarily from oxidation of primary emissions (both biomass burning and fossil fuel combustion emissions can be precursors). Recently, Lin et al. [47,48] reported a linear relationship between HULIS-C and WSOC, and also suggested that the increases are caused by the secondary chemical processes. The dominant role of photochemical production of HULIS-C is also found in Shanghai during summer [49]. Kuang et al. [50] show that the secondary formation processes contribute 49-82% of ambient HULIS. Additionally, linear regression slopes in Figure 5b show that HULIS-C/WSOC ratio (0.64) in daytime was slighter higher that in nighttime (0.57), suggesting that the contribution of photochemical production of HULIS-C in day might be more significant than that during night. The HULIS-C/WSOC ratio in this study is comparable to that observed in Shanghai (0.6) [16] and Hongkong (0.57) [50], but higher than that in Lanzhou (0.46) [51]. Here, we show the UV-Vis absorption spectra of HULIS-C and WSOC by using the samples collected on 31 July and 10 August 2016 in Figure 6. Note these two samples have high and low PM2.5 mass loadings, respectively, and the shapes of their UV-Vis spectra are also representative of the average pattern of all samples. For comparison, the UV-Vis spectra of standard SRFA and PPHA are also presented in Figure 6. Generally, the absorbance of both HULIS-C and WSOC increases with the decreasing wavelength in the range of 200-500 nm, but the WSOC absorbs relatively more light compared to that of HULIS-C at the same wavelength. Compared with the two standard humic substances, the UV-Vis characteristics of both HULIS-C and WSOC are significantly different, with higher absorbance at 365 nm for the latter, indicating they contain different compounds. Such differences are closely relevant with different oxidation reactions of different precursors. For example, our UV-Vis spectra are qualitatively similar to those of pure guaiacol and catechol, but different from those from the guaiacol and catechol-derived SOA [52].
To further investigate the chemical properties of WSOC and HULIS, the E250/E365 ratios are calculated and shown in Figure 7. As suggested by previous studies [34,53], a low E250/E365 ratio indicates the sample contains more aromatic or large molecules. The average E250/E365 values for HULIS-C and WSOC are 7.0 and 5.7 during daytime, and 7.9 and 6.3 during nighttime. Therefore, a slightly smaller E250/E365 ratio in day reveals that the daytime samples include more highly UV absorbing aromatic molecules. The average E250/E365 ratio (either in day and night) of HULIS in our sample is higher than the average value of 5.9 for the HULIS determined in PM2.5 collected in 2010 in Guangzhou [34]. As that study used the same method as us to isolate the HULIS fraction, this difference demonstrates the different chemical characteristics of HULIS in aerosols under different environments [34]. In addition, larger E250/E365 ratio for HULIS fraction than that of WSOC indicates that the HULIS fraction concentrates more aliphatic/less aromatic structures. This finding is opposite to the results from Duarte et al. [53] who found that HULIS had lower E250/E365 ratios than WSOC.  To further investigate the chemical properties of WSOC and HULIS, the E 250 /E 365 ratios are calculated and shown in Figure 7. As suggested by previous studies [34,53], a low E 250 /E 365 ratio indicates the sample contains more aromatic or large molecules. The average E 250 /E 365 values for HULIS-C and WSOC are 7.0 and 5.7 during daytime, and 7.9 and 6.3 during nighttime. Therefore, a slightly smaller E 250 /E 365 ratio in day reveals that the daytime samples include more highly UV absorbing aromatic molecules. The average E 250 /E 365 ratio (either in day and night) of HULIS in our sample is higher than the average value of 5.9 for the HULIS determined in PM 2.5 collected in 2010 in Guangzhou [34]. As that study used the same method as us to isolate the HULIS fraction, this difference demonstrates the different chemical characteristics of HULIS in aerosols under different environments [34]. In addition, larger E 250 /E 365 ratio for HULIS fraction than that of WSOC indicates that the HULIS fraction concentrates more aliphatic/less aromatic structures. This finding is opposite to the results from Duarte et al. [53] who found that HULIS had lower E 250 /E 365 ratios than WSOC.
to the results from Duarte et al. [53] who found that HULIS had lower E250/E365 ratios than WSOC.

Determination and Sources of WSON and WSTN
As discussed in earlier studies [23,54]

Determination and Sources of WSON and WSTN
As discussed in earlier studies [23,54], the determination of WSON concentrations by subtracting WSIN from WSTN in some cases yields negative values due to measurement uncertainties and aggregation of errors. During this study, only one negative WSON concentration (31 July) is found and we set it as zero according to the suggestion of Mace et al. [55]. Figure 8 shows the day/night results of N-containing components. Averaged mass concentrations of WSON, NO 2 − -N, studies have shown the formation of N-containing oligomers in aqueous droplets [56,57]. The WSTN concentrations are similar to those measured in Beijing spring PM2.5 (0.9-6.7 µg/m 3 ) [58]. Accordingly, WSON on average contributes 16.3% and 17.8% of WSTN mass in day and night ( Table 1). The WSON/WSTN ratios in this study are somewhat similar to results at an urban site in Kofu, Japan (16.2%) [27] and in fog water samples collected in Fresno, USA (10-17%) [59]. Differently, the contribution of WSON to WTON was much smaller than results observed by Ho et al. [60] for sampling in summer in downtown background in Xi'an, China, with an average WSON/WTON ratio of 45%. The results demonstrate the large variability of WSON, dependent on source, time, and space. In order to further probe the source of N-species, we need to investigate its correlations with other species. As shown in Table 2, WSON has good correlation with NH4 + in both day and night, especially in night (r = 0.95); it moderately correlates with SO4 2− (r = 0.70 in day and r = 0.69 in night), and correlates poorly with NO3 − in day (r = −0.06) but more tightly in night (r = 0.76). WSON also positively correlates with WSOC (r = 0.68 in day and r = 0.69 in night). These results all suggest that the WSON has significant contribution from secondary production too. The atmospheric reactions of NOx and VOCs would produce secondary organic nitrogen compounds, such as alkyl nitrates and amino compounds, which would form secondary organonitrate aerosols via gas-to-particle conversions during day and night. Matsumoto et al. [27] also suggest that fine-mode WSON is mainly derived from secondary oxidation of corresponding precursor gases. Research in Xi'an [60] suggests that secondary sources and biomass burning can be the major factors influencing the levels of WSON. While in this study, the positive correlation between WSON and HULIS-C (day: r = 0.66, night: r = 0.61), and weak correlation between WSON and K + ions (day: r = 0.35, night: r = 0.46) imply that biomass burning is not a major source of WSON. In order to further probe the source of N-species, we need to investigate its correlations with other species. As shown in Table 2 These results all suggest that the WSON has significant contribution from secondary production too. The atmospheric reactions of NO x and VOCs would produce secondary organic nitrogen compounds, such as alkyl nitrates and amino compounds, which would form secondary organonitrate aerosols via gas-to-particle conversions during day and night. Matsumoto et al. [27] also suggest that fine-mode WSON is mainly derived from secondary oxidation of corresponding precursor gases. Research in Xi'an [60] suggests that secondary sources and biomass burning can be the major factors influencing the levels of WSON. While in this study, the positive correlation between WSON and HULIS-C (day: r = 0.66, night: r = 0.61), and weak correlation between WSON and K + ions (day: r = 0.35, night: r = 0.46) imply that biomass burning is not a major source of WSON.

PAHs Compounds
The mean total PAH concentrations (ΣPAHs) in PM 2.5 are listed in Table 1, and concentrations of individual PAH compounds and their night-to-day mass ratios are also examined and listed in Table 3. Overall, the summer PAHs levels are much lower than those measured during spring and winter in Changzhou [31], probably due to the significantly reduced fossil fuel combustion activities during summer and enhanced partitioning of some PAH species into gas phase owing to warmer temperatures. Table 3 shows that the nighttime ΣPAHs concentrations are higher (3.64 ng/m 3 vs. 2.34 ng/m 3 ) than those during daytime. This day-night difference is similar to that found in Xiamen [61] but opposite to that from Sarajevo (day/night = 1.45) [62]. Specifically, high molecular weight PAHs (5-6 rings) exhibit remarkably higher loadings during nighttime (average night/day ratio of 2.20), while the day-night differences are insignificant for low molecular weight PAHs (2-3 rings) (average night/day ratio of 0.99) and medium molecular weight PAHs (4-rings) (average night/day ratio of 1.21). Such behaviors may link with the volatility of different PAHs, as the lighter PAHs (LMW/MMW-PAHs) can exist in both vapor and particulate phases while the heavier PAHs (≥5-rings) are mainly particle-bound [63]. Warmer temperatures during daytime may enhance the evaporation of lighter PAHs but not heavier PAHs. Overall, the amount of total PAHs in PM 2.5 is very low during the sampling period compared with those during spring and winter at the same site [31]. Future studies are still needed to elucidate the low PAHs levels in details. Total PAHs exhibits good correlations with EC and K + , but weak or negative correlations with WSOC, HULIS-C, and O 3 ( Table 2), suggesting that PAHs are linked to primary sources such as vehicle exhaust and biomass burning.

Conclusions
In this study, we conducted a comprehensive day-night comparison regarding various components in the PM 2.5 samples collected in summer at an urban site in Changzhou, China. Our results show slightly high levels of PM 2.5 , WSOC, HULIS-C, SO 4 2− , and NH 4 + concentrations in the day, while concentrations of PAHs, EC, and WSON are higher in the night. Nitrate concentration is significantly low in the day as it is easy to evaporate at warm temperatures. WSOC contributes, on average,~60% of the OC mass, and it has particularly good correlations with O 3 in day, implying the significant role of secondary photochemical production. Meanwhile the WSOC/OC ratios are only slightly lower in the night, indicating the secondary organic aerosol (SOA) formation likely occurs during nighttime as well. HULIS-C contributes~60% of WSOC mass. Very good relationships between HULIS-C and WSOC (day: r = 0.94, night: r = 0.96), HULIS-C and O 3 (day: r = 0.83; night: r = 0.90) are found, clearly demonstrating the dominant role of the secondary origin of HULIS-C. On the contrary, correlations between HULIS-C and EC, PAHs, and K + , both in day and night, are weak, indicating that primary sources, e.g., fossil fuel combustion and biomass burning, are not major sources of HULIS in summer. Optical properties (E 250 /E 365 ratios) of HULIS indicate that HULIS contains more aliphatic/less aromatic structures than the WSOC in summer. WSON accounts for approximately 17% (daily average) of the WSTN, and correlation analysis shows that it might be dominated by secondary formation as well. However, different from the behaviors of WSOC, it seems like the nighttime formation processes under dark conditions can contribute remarkably to the WSON. PAHs concentrations are very low in summer, and they are emitted by dominant primary sources (e.g., vehicle emissions, biomass burning) based on good relations with EC and K + ion and weak correlations with WSOC, HULIS-C, and O 3 .
Our results, overall, reveal that the secondary formation is a significant contributor to PM 2.5 and its major components such as HULIS, WSON, and WSOC. More sample analysis and seasonal variation should be done to look into the new insights of day-night variation. Future investigations should focus on elucidating the details of precursors and products, the reaction mechanism, and the contribution of different sources in both the day and the night, so as to effectively reduce the fine aerosols and/or ozone pollution in Changzhou, China.