Organic compounds are abundant in ambient particle, most of which are secondary organic aerosol (SOA) in nature, formed from the oxidation of volatile organic compounds (VOCs) through reactions with O₃, •OH, and •NO₃ radicals and nucleation processes or condensation onto pre-existing particles (Kanakidou et al., 2005; Wong et al., 2015). SOA is continually aging and experiencing further chemical processing in the ambient environment, yet their formation and transformation mechanisms in the atmosphere remain not fully characterized (Heald and Spracklen, 2009; Wong et al., 2015). Because SOA production in the atmosphere is very complex, there are large gaps in the current understanding of SOA (Ervens et al., 2011). Moreover, current models often have large uncertainties in predicting the mass concentration of ambient SOA (Cheng et al., 2015). In addition to emissions, meteorological parameters, i.e., relative humidity (RH), temperature, and solar radiation, also have important influences on ambient SOA concentration and production (Sun et al., 2014; Cheng et al., 2015).

RH is an important factor influencing the mechanisms of SOA formation and related chemical and physical properties (Hallquist et al., 2009; Sun et al., 2013). Water is ubiquitous in the atmosphere, which can be easily taken up by soluble chemical components in ambient aerosols. Liquid water accounts for nearly 30% by volume in biogenic SOA under 85% RH (Varutbangkul et al., 2006). Therefore, chemical reactions or physical processes of aging organic aerosol (OA) involve water as a reactant or solvent, and the composition of OA is affected by RH in the atmosphere (Ervens and Volkamer, 2010; Zhang et al., 2012). Recently, the effects of RH on SOA formation have been investigated through both chamber studies (Zhao et al., 2006; Zhang et al., 2011; Zhou et al., 2011) and field observations (Sun et al., 2013; Cheng et al., 2015; Zheng et al., 2015; Wang et al., 2016). Most of the lab studies indicated that heterogeneous reactions under high RH might play important roles in SOA formation, including various precursors, e.g., aromatic compounds (Zhou et al., 2011), ethylene (Jia and Xu, 2016), β-pinene (Boyd et al., 2015), isoprene (Nguyen et al., 2011; Zhang et al., 2011), etc. Based on aromatic systems, Zhou et al. (2011) suggested that aromatic SOA yields under humid conditions can be a factor of 2 to 5 higher than that under dry conditions. Water uptake of glyoxal and methylglyoxal onto pre-existing aerosols contributing to larger SOA mass at higher RH was also demonstrated by Ervens and Volkamer (2010).

Furthermore, plenty of field studies also observed higher SOA yields under humid conditions in ambient air (Sun et al., 2013; Cheng et al., 2015). Ge et al. (2012) found that SOA formation and the oxidation degree of OA were enhanced by aqueous-phase processing during foggy days. Similarly, enhanced SOA concentrations under high RH conditions were observed by Kaul et al. (2011) as well. Based on online non-refractory submicron aerosol (NR-PM₁) observation, the mass concentrations of most aerosol species, including oxygenated organic aerosol (OOA) and coal combustion OA (CCOA) increase linearly as a function of RH during winter in Beijing (Sun et al., 2013). Aqueous-phase production of SOA is an important pathway for SOA formation (Lim et al., 2010; Ervens et al., 2011, 2014). Moreover, different RH conditions have different impacts on organic oxidation product formation (Nguyen et al., 2011; Zhang et al., 2011). A significant change in isoprene SOA composition between humid (90% RH) and dry (< 2% RH) conditions was observed byNguyen et al. (2011). They found that isoprene SOA formed under high RH conditions contains a significantly shorter chain length of oligomer esters (5–7 monomer residues), while that under dry conditions is dominated by long oligomer esters (8–10 monomer residues). However, based on field measurements using an online high-resolution soot-particle aerosol mass spectrometer (SP-AMS) during humid (RH > 60%) late winter conditions in Nanjing,Wu et al. (2018) recently found that ${\rm{SO}}_4^{2-} $ , ${\rm{NO}}_3^{-} $ , and low-volatility OA (LVOOA) responded positively to RH but not to photochemical activities, implying that they were driven by aqueous-phase processing; meanwhile, the other two less oxygenated SOA factors, i.e., local secondary OA (LSOA) and semi-volatile oxygenated OA (SVOOA), behaved in a complete opposite manner, indicating that they were governed by photochemical processing.

In fact, as ambient conditions are much more variable than the controlled chamber conditions, elucidating SOA and its influencing factors is more complex and difficult. Based on the field observations, the aqueous-phase chemistry on different categories of SOA may be dominated by different mechanisms in the ambient environment. Thus, the knowledge of SOA formation and transformation mechanisms under different RH conditions in the ambient atmosphere is rather limited. In this work, we examine the RH dependence of aerosol chemistry, intending to illuminate the impacts of high RH on SOA abundance in the ambient atmosphere. A three-month measurement campaign of aerosol particle composition was conducted during summer 2015 at a regional background location in East China. The ambient daily average RH throughout the study was generally high, ranging from 57% to 99%, and the overall campaign average RH was high at 84%, providing an ideal opportunity to investigate high RH effects on secondary aerosol production.

Samples were collected at Lin’an Regional Atmosphere Background Station (30.3°N, 119.73°E, 138 m a.s.l.) (Fig. 1). The details about the station description can be found in Liang et al. (2017). A total of 86 daily 24 h-PM_2.5 samples were collected from 7 June to 31 August 2015 by a mini-volume sampler operated at an ambient airflow rate of 5 L min^–1.

Figure 1 Location of Lin’an station (star) and the surrounding provinces

Figure options

Carbonaceous components, i.e., organic carbon (OC) and element carbon (EC), were measured on a punch (0.495 cm²) of each quartz filter sample by a thermal/optical carbon analytical method (Liang et al., 2017). The detection limit of OC was 0.82 µg cm⁻² (Liang et al., 2017). Water-soluble inorganic ions (K⁺, Na⁺, Ca²⁺, Mg²⁺, ${\rm{NH}}_4^{+} $ , ${\rm{NO}}_3^{-} $ , ${\rm{SO}}_4^{2-} $ , and Cl^–) were detected by a Dionex ICS-3000 ion chromatograph; detailed description of water-soluble inorganic ions can be found in Liang et al. (2016). The water-soluble inorganic ion data in this paper were used to calculate aerosol pH.

The thermodynamic model ISORROPIA-II was applied in this study to predict aerosol pH. The forward mode in ISORROPIA-II was chosen, which is superior to the reverse mode, as assessed by prior studies (Hennigan et al., 2015; Bougiatioti et al., 2016). Considering the high RH in our case (mean ± standard deviation of 84 ± 10%), aerosol solutions were assumed to be metastable (only a liquid phase) when running the model, which often had better performance than the stable state solution (solid + liquid) and was commonly applied in previous pH predictions (Guo et al., 2015; Bougiatioti et al., 2016; Liu et al., 2017). ISORROPIA-II input data included water-soluble ions (K⁺, Na⁺, Ca²⁺, Mg²⁺, ${\rm{NH}}_4^{+} $ , ${\rm{SO}}_4^{2-} $ , ${\rm{NO}}_3^{-} $ , and Cl^–) in PM_2.5 as well as RH and temperature.

During this campaign, surface ozone at Lin’an was observed by a Model 49C ozone analyzer and CO was observed by a Model 48C carbon monoxide (CO) analyzer (Thermo Fisher Scientific Co., Ltd.). Daily mean concentrations of surface O₃ and CO were calculated from the raw data recorded every 1 min.

Statistical results of carbonaceous components, i.e., OC and EC, and the ratios of SOC to OC in PM_2.5 collected at Lin’an site during summer 2015 are listed in Table 1, and their temporal variation patterns are shown in Fig. 2. In this study, the average ratio of OC/EC was 4.75, varying from 1.80 to 12.0, of which most values were between 2.0 and 6.0 (84%) (Liang et al., 2017). If the OC/EC ratio exceeds 2.0, the presence of SOA is implied (Chow et al., 1996); and this indicates an obvious contribution of SOA to the ambient aerosol at the background site. In addition, secondary organic carbon (SOC) was estimated by the EC-tracer method (Lim and Turpin, 2002). Briefly, SOC was estimated by the following formulae:

Figure 2 Temporal variations of carbonaceous components (POC, SOC, and EC) and the ratios of SOC to OC in PM2.5 collected at Lin’an site during summer 2015

Figure options

Primary OC (POC) = EC × (OC/EC)min,

SOC = OC − POC,

where (OC/EC)min indicates the minimum ratio of OC/EC. The mean concentrations of POC and SOC in PM_2.5 at Lin’an during the sampling time were 6.00 ± 2.64 and 8.26 ± 3.31 µgC m^–3, respectively (Table 1). The ratios of SOC to OC varied from 20.8% to 85.0% with an average of 58.5 ± 12.6%. The frequency distribution of the SOC/OC ratios is shown in Fig. 3. The daily SOC/OC ratios had a primarily skewed distribution, and more than 60% of SOC/OC ratios were distributed in the range of 50%−70%. The SOC/OC ratios in our study were similar to those reported at different sites around the world (Zhang et al., 2007; Ge et al., 2017; Chen et al., 2018; Wu et al., 2018), illustrating that secondary organic aerosol is ubiquitous and presents a dominance of oxygenated species in organic aerosols in the atmosphere.

Figure 3 The mass frequency distribution for the ratios of SOC/OC in PM2.5 collected at Lin’an site during summer 2015

Figure options

During the sampling time, almost all RH values were higher than 60%, except on one day, i.e., 4 August 2016 (RH = 57%), which was excluded in the following discussion. The average RH value was 84%. Based on meteorological data analysis, increases in clouds and precipitation induced higher RH and decreased total solar radiation (Fig. 4). Statistical analysis of RH and total solar radiation showed an obvious negative correlation (r = 0.89, p < 0.01, Fig. 5). In the case of low solar radiation and high RH, the surface temperature fluctuation was inhibited, causing the boundary layer height to decrease; as a consequence, aerosols accumulated, enhancing light scattering and thus decreasing visibility (Fig. 5). In the meantime, photochemical activity was suppressed under low solar radiation, and the formation of secondary aerosol from this pathway became less significant.

Figure 4 Correlations between RH and precipitation and total cloud cover during summer 2015 at Lin’an site. The k indicates regression slope

Figure options

Figure 5 Correlations between RH and visibility, temperature, and total solar radiation in summer 2015 at Lin’an site. The k indicates regression slope

Figure options

In order to explore the impact of higher RH on SOA concentration, the relationships of POC and SOC with RH are shown in Fig. 6. The concentration of POC showed no relationship with RH (p > 0.05, Fig. 6a). On the other hand, SOC exhibited an obvious negative relationship when RH was higher than 60% (r = 0.56, p < 0.01, Fig. 6b). This indicates that chemical processing of SOC appeared to be inhibited under high RH (> 60%) at Lin’an. This observation was similar to that reported byWu et al. (2018), i.e., the two less oxygenated SOA factors (LSOA and SVOOA) responded negatively to RH. These authors concluded that the two less oxygenated SOA products were mainly driven by photochemical process.

Figure 6 Relationships between (a) POC and RH, and (b) SOC and RH. Also shown is color-coded temperature.

Figure options

In addition, high SOC values were observed at high temperatures (red color, Fig. 6), demonstrating that higher temperatures can elevate SOA concentrations in the ambient environment, especially on days of more than 25°C. High temperatures in summer are typically associated with stronger solar radiation, and the concentrations of photochemical oxidants, i.e., O₃ and OH, will increase and the photolysis rates will be larger accordingly. Since SOC is, to a large extent, a product of photochemical reactions, enhanced SOC production rates are expected on days with high temperatures and intense photochemistry. Furthermore, numerous studies have reported that high temperatures can increase the emission of biogenic SOA precursors, e.g., isoprene and monoterpenes (Monson et al., 1994; Dominguez-Taylor et al., 2007). Thus, under high temperatures, SOA formation rates will be higher due to increased biogenic precursor emission and enhanced photolysis rates in the ambient environment.

The variations of average concentrations of POC, SOC, EC, OC, CO, and aerosol pH as a function of RH throughout the study are shown in Fig. 7. SOC exhibited a decreasing trend when RH was higher than 60%. SOC showed the largest mass decreasing rate when RH increased from 60% to 80% (−2.98 μgC m⁻³/10% RH), whereas the decreasing rate of those species was reduced to −0.46 μgC m ⁻³/10% RH at RH between 70% and 90% (Fig. 7, Table 2). Our explanation for this phenomenon is that the rate of SOA production via the photochemical oxidation reaction pathway was strongly decreased at RH between 60% and 80%; if the rate of SOA production via the heterogeneous reactions was not increased enough, the net value of SOA production rate was significantly decreased. However, with the increasing RH, heterogeneous reactions under the higher RH condition enhanced SOA yields, and compensated for the influence from the photochemical oxidation pathway at RH between 70% and 90%. Thus, the net decreasing rate at RH between 70% and 90% was lower than that at RH between 60% and 80%. It should be noted that POC and EC were expected to be non-hygroscopic, while they were observed still slightly elevated as RH increased from 60% to 80%, probably because of their continuous yet weak accumulation. Moreover, likely due to wet deposition, all carbonaceous components appeared to have similar decreasing rates when RH increased from the range between 80% and 90% to the range higher than 90%. Because of both concentrations of POC and SOC decreasing at the range of RH higher than 90%, OC concentrations exhibited a significant decreasing rate of −3.63 μgC m ⁻³/10% RH (Fig. 7, Table 2). Compared to primary aerosol components, i.e., POC and EC, the primary gaseous species CO exhibited much more stable behavior with increasing RH (Fig. 7).

Figure 7 Variations of mass concentrations of POC, SOC, EC, OC (µgC m−3), gas CO (× 10 ppm), and aerosol pH as a function of RH (each point represents the average concentration of the parameter between the former and latter RH conditions).

Figure options

Previous chamber studies have found that heterogeneous reactions under the high RH condition can significantly enhance SOA yields. Taking isoprene SOA chamber studies as an example, earlier work has demonstrated that under lower NO_x conditions, the isoprene SOA yield was enhanced in the presence of wet and acidic sulfate aerosol, especially resulting in higher concentrations of isoprene-derived epoxydiols (IEPOX) SOA (Surratt et al., 2010; Zhang et al., 2011, 2014; Lin et al., 2012). To evaluate the effect of chemical production, SOC concentrations were normalized to EC (derived from only primary emissions and being rather inert to chemical reactions), in order to exclude accumulation and/or dilution effects. The EC-scaled method will largely eliminate the variations due to physical reactions, i.e., mixing and dilution, and better represent the contribution from chemical reactions (Zheng et al., 2015).

Compared to previous chamber studies, the results obtained in our ambient study show contrary patterns: SOC appeared to be inhibited under high RH (> 60%), as indicated by an obvious downward trend of SOC/EC with increasing RH (Figs. 8, 9). Apparently, SOA at our location had no heterogeneous formation pathway under the high RH condition compared to those reported in other studies (Kaul et al., 2011; Ge et al., 2012). On the contrary, there was a negative effect on the formation of SOA observed in our study. This result was similar to that reported by Zheng et al. (2015), i.e., SOC/EC ratios exhibiting constant behavior during winter in Beijing when RH increased from 50% to 80%, although the SOA concentrations were increasing rapidly with rising RH. By using HOA (hydrocarbon-like organic aerosol) instead of EC, a similar trend was also observed by Sun et al. (2013): although the OOA concentration was observed obviously elevated when RH rose from 50% to 90%, the HOA-normalized OOA was fairly constant across the high RH levels.

Figure 8 Change of SOC/EC with RH variation. In the box-and-whisker plots, the boxes and whiskers indicate the maximum, 75th percentiles, 50th percentiles (median), 25th percentiles, and minimum, respectively.

Figure options

Figure 9 Variations of concentrations of SOC, POC, and O3; total solar radiation (TSR); and the ratio of SOC/EC, as a function of RH (each point indicates the mean of the parameter between the former and latter RH conditions).

Figure options

There might be several reasons for the results observed in our study, different from chamber simulations. First, our study was conducted at high RH conditions, which was typically accompanied by increased clouds and precipitation, resulting in reduced solar radiation at the ground level, while photochemical activity is weakened under low solar radiation. Accordingly, the total solar radiation was observed to sharply decrease with RH rise, and O₃ was also detected to decrease when RH > 70%, indicating secondary aerosol formation via the photochemical oxidation reaction pathway became less important ( Fig. 9). On the other hand, Zheng et al. (2015) suggested that aerosol may be accumulated at the high RH stable synoptic condition and increased to support aqueous-phase production, which is not sufficient to compensate for the weakened photochemical activity influence and may lead to a net decrease in the formation of ambient SOA.

Second, based on concentrations of aerosol chemical species, temperature, and RH, the ISORROPIA-II model is able to offer a more rigorous approach to calculate particle pH (Guo et al., 2015). The average pH in this study was 1.86 ± 0.54, ranging from 0.38 to 4.17, predicted by the thermodynamic ISORROPIA-II model. These systematically low pH levels observed in this study may be significantly influencing acid-catalyzed reactions as part of SOA production. Prior studies have found that acidic sulfate particles can significantly enhance the isoprene SOA yields in the atmosphere (Gaston et al., 2014; Riedel et al., 2015). However, due to more water uptake under the higher RH condition, the aerosol droplet acidity decreased and the enhancement of SOA formation by acidity was accordingly reduced. Therefore, the variation of average aerosol pH in our study was observed to be elevated with the enhanced RH as well, increasing from 1.63 to 2.20 when RH changed from the range of 60%–70% to > 90%, indicating aerosol acidity was obviously weakened under the high RH condition (Fig. 7). It should be noted that the predicted aerosol pH in this study was underestimated by lack of the NH₃ data, while it had no impact on the elevated pH trend with increasing RH, which was previously reported (Guo et al., 2015). Moreover, the simulated sulfate was in good agreement with observations (R² = 0.98, slope 0.95), implying that the ISORROPIA-II model performed well.

In addition, it should be mentioned that recent studies have demonstrated that compared to dry conditions, where SOA is semisolid, high RH conditions can lower the particle’s microscopic viscosity, which affects the particle chemical reactivity, allowing for more rapid diffusion of reactants (Renbaum-Wolff et al., 2013; Grayson et al., 2016). Zhang et al. (2018) found that aerosol viscosity also affects the reactive uptake yield of SOA. The adverse effect of aerosol viscosity on reactive uptake is also a function of RH, which is due to RH affecting the viscosity of pre-existing SOA coatings. Moreover, different portions of SOA may have different predominant mechanisms in the ambient environment. Wu et al. (2018) found highly oxygenated SOA (LVOOA) was more associated with aqueous-phase processes, while the less oxygenated components (LSOA and SVOOA) were governed by photochemical processing in ambient aerosols.

High RH conditions are typically associated with weaker photochemistry, leading to a low production rate of SOA via this pathway in ambient aerosols. Although the aerosol aqueous-phase production will be increased under the high RH condition, it is not sufficient to compensate for the weakened photochemical activity influence, leading to a net decrease in the formation of SOA when RH > 60%. Moreover, due to more water uptake, higher RH was more likely to induce relatively lower aerosol acidity, thus the enhancement of SOA formation by acidity was reduced. Due to measurement limitations, we were unable to identify the precursors and couldn’t separate RH effects from other meteorological factors, e.g., temperature, wind direction, solar radiation, etc., in the ambient environment. It may be true that in the highly controlled chamber studies, RH solely causes the increase of SOA tracers, but not so in all ambient environments. In fact, oxidant concentrations, acidity, viscosity, reaction rates, and chemical species in the atmosphere are highly variable, and these factors affect SOA chemical reactions under high RH conditions and render the processes to be very complex. Although our study results represent more likely a specific and localized event, they indicate that some organic oxidation products may behave differently at higher RH compared to low RH conditions in the ambient atmosphere, as some organic oxidation products, i.e., those produced under high-NO _x and low-NO_x conditions under actual ambient conditions in China, are different from the results obtained from chamber studies and ambient observations in the US (Ding et al., 2008; Paulot et al., 2009; He et al., 2018).

Overall, our investigations do show a negative response of SOA to RH. These findings, at least qualitatively, reveal that in moist ambient environments, SOA formation may be more associated with photochemical processes, while aqueous-phase chemistry is less important for some SOA production.

Acknowledgment. Financial support was also provided partly by the Ministry of Science and Technology (MOST) of Taiwan, China (MOST 103-2113-M-007-005). We are grateful to Professor Xiaobin Xu for providing the O₃ and CO data.