Mutual promotion effect between aerosol particle liquid water and nitrate formation lead to severe nitrate-dominated particulate matter pollution and low visibility

Abstract. As has been the case in North America and Western Europe, the SO2 emissions substantially reduced in North China Plain (NCP) in recent years. A dichotomy of reductions in SO2 and NOx concentrations result in the frequent occurrences of nitrate (pNO3−)-dominated particulate matter pollution over NCP. In this study, we observed a polluted episode with the nitrate mass fraction in non-refractory PM1 (NR-PM1) up to 44 % during wintertime in Beijing. Based on this typical pNO3−-dominated haze event, the linkage between aerosol water uptake and pNO3− formation, further impacting on visibility degradation, have been investigated based on field observations and theoretical calculations. During haze development, as ambient relative humidity (RH) increased from ~ 10 % up to 70 %, the aerosol particle liquid water increased from ~ 1 μg/m3 at the beginning to ~ 75 μg/m3 at the fully-developed haze period. Without considering the water uptake, the particle surface area and the volume concentrations increased by a factor of 4.1 and 4.8, respectively, during the development of haze event. Taking water uptake into account, the wet particle surface area and volume concentrations enhanced by a factor of 4.7 and 5.8, respectively. As a consequence, the hygroscopic growth of particles facilitated the condensational loss of dinitrogen pentoxide (N2O5) and nitric acid (HNO3) to particles contributing pNO3−. From the beginning to the fully-developed haze, the condensational loss of N2O5 increased by a factor of 20 when only considering aerosol surface area and volume of dry particles, while increasing by a factor of 25 considering extra surface area and volume due to water uptake. Similarly, the condensational loss of HNO3 increased by a factor of 2.7~2.9 and 3.1~3.5 for dry and wet aerosol surface area and volume from the beginning to the fully-developed haze period. Above results demonstrated that the pNO3− formation is further enhanced by aerosol water uptake with elevated ambient RH during haze development, in turn, facilitating the aerosol taking up water due to the hygroscopicity of nitrate salt. Such mutual promotion effect between aerosol particle liquid water and nitrate formation can rapidly degrade air quality and halve visibility within one day. Reduction of nitrogen-containing gaseous precursors, e.g., by control of traffic emissions, is essential in mitigating severe haze events in NCP.


Abstract. As has been the case in North America and Western Europe, the SO2 emissions substantially reduced in North China Plain (NCP) in recent years. A dichotomy of reductions in SO2 and NOx 40 concentrations result in the frequent occurrences of nitrate (pNO3 -)-dominated particulate matter pollution over NCP. In this study, we observed a polluted episode with the nitrate mass fraction in nonrefractory PM1 (NR-PM1) up to 44% during wintertime in Beijing. Based on this typical pNO3 -dominated haze event, the linkage between aerosol water uptake and pNO3formation, further impacting on visibility degradation, have been investigated based on field observations and theoretical 45 calculations. During haze development, as ambient relative humidity (RH) increased from ~10% up to 70%, the aerosol particle liquid water increased from ~1 µg/m 3 at the beginning to ~75 µg/m 3 at the fully-developed haze period. Without considering the water uptake, the particle surface area and the volume concentrations increased by a factor of 4.1 and 4.8, respectively, during the development of haze event. Taking water uptake into account, the wet particle surface area and volume concentrations 50 enhanced by a factor of 4.7 and 5.8, respectively. As a consequence, the hygroscopic growth of particles facilitated the condensational loss of dinitrogen pentoxide (N2O5) and nitric acid (HNO3) to particles contributing pNO3 -. From the beginning to the fully-developed haze, the condensational loss of N2O5 increased by a factor of 20 when only considering aerosol surface area and volume of dry particles, while increasing by a factor of 25 considering extra surface area and volume due to water uptake. 55 Similarly, the condensational loss of HNO3 increased by a factor of 2.7~2.9 and 3.1~3.5 for dry and wet aerosol surface area and volume from the beginning to the fully-developed haze period. Above results demonstrated that the pNO3formation is further enhanced by aerosol water uptake with elevated ambient RH during haze development, in turn, facilitating the aerosol taking up water due to the

Introduction
Aerosol particle hygroscopicity plays an important role in air quality deterioration and cloud formation 80 (Yu, 2009;Fitzgerald, 1973;Kreidenweis and Asa-Awuku, 2014;Wang and Chen, 2019;McFiggans et al., 2006) and can also directly influence aerosol measurements (Chen et al., 2018a). In atmospheric environments influenced by anthropogenic activities, particulate secondary inorganic compounds are often dominated by ammonium sulfate ((NH4)2SO4) and ammonium nitrate (NH4NO3) (Heintzenberg, 1989), which originate from the oxidation of sulfur dioxide (SO2) and nitrogen oxides (NOx) via well-85 established chemical pathways (Calvert et al., 1985). The abundance of secondary inorganic components is one of the most important factors determining particle hygroscopicity (Swietlicki et al., 2008), thereby governing the aerosol liquid water content under ambient moist conditions. Increased aerosol particle liquid water could accelerate secondary inorganic and organic aerosol formation by decreasing the kinetic limitation of mass transfer of gaseous precursors and providing more medium for 90 multiphase reactions (Mozurkewich and Calvert, 1988;Cheng et al., 2016;Ervens et al., 2011;Kolb et al., 2010).
Sulfuric acid (H2SO4) is formed from the oxidation of SO2 via gaseous and multiphase reactions. H2SO4 is subsequently fully or partly neutralized by gaseous NH3 taken up on particles, resulting in the formation of (NH4)2SO4 and / or NH4HSO4. Any remaining NH3 is available to neutralize HNO3 to 95 form particulate NH4NO3 (Seinfeld. and Pandis., 2006) (and further excess NH3 can neutralize any available HCl to form particulate NH4Cl). Over the past several decades, substantial efforts have reduced emissions of both SO2 and NOx improving the local and regional air quality all over the world. For example, SO2 and NOx emissions were reduced by 82% and 54% in the majority of European Environment Agency member countries between 1990 and 2016 (https://www.eea.europa.eu/data-and-100 maps/indicators/main-anthropogenic-air-pollutant-emission s/assessment-4). In consequence, an increasing trend of NO3 -/SO4 2molar ratio was observed in long-term measurements at Leipzig, Germany  and at some other European sites from the European Monitoring and Evaluation Programme (EMEP) (Putaud et al., 2004). In recent years, China has also managed to reduce SO2 emissions by 75% since 2007 (Li et al., 2017a), whereas NOx emissions declined only by ~10% 105 between 2011 and 2015 (de Foy et al., 2016). Similar with European countries, the dominant inorganic component in fine aerosol particles has switched from sulfate to nitrate in the recent years (Sun et al., 2015;Hu et al., 2017;Hu et al., 2016;Wu et al., 2018;Guo et al., 2014;Huang et al., 2014;Huang et al., 2010;Ge et al., 2017;Xu et al., 2019a;Xie et al., 2019;Li et al., 2018). Field measurements show that annually averaged NO3 -/SO4 2molar ratio of NR-PM1 (non-refractory PM1) in 2012 (1.3~1.8) (Sun et 110 al., 2015) has significantly increased compared to that in 2008 (0.9~1.5) .
Comparably, the NO3 -/SO4 2molar ratio of PM2.5 increased substantially, from 1.5 before 2013 to 3.33 in 2017 (Xu et al., 2019a). Model simulations have also shown that the simulated annual mass concentration of nitrate and its mass fraction in secondary inorganic components over North China increased by 17~19% and 7% respectively, while the sulfate mass and fraction decreased by 10~19% 115 and 6% between 2006 and 2015 under the assumption of constant NH3 emissions .
However, NH3 emissions have been observed by satellites to increase by ~30% from 2008 to 2016 over the North China Plain (NCP) (Liu et al., 2018), further increasing the potential for nitrate formation . Over the NCP region, heavy haze events are typically associated with enhanced ambient RH levels. 120 This leads to an increased aerosol liquid water content , which will influence the particulate nitrate formation by changing the reactive uptake of precursors and the thermodynamic equilibrium of ammonium nitrate (Cheng et al., 2016;Yun et al., 2018;Yue et al., 2019). To date, few studies reported aerosol liquid water content over NCP region Bian et al., 2014;Cheng et al., 2016;Wu et al., 2018). However, the observational 125 and theoretical analysis of the relationship between particulate nitrate formation and associated liquid water during haze events in China has been infrequently reported . In this study, a self-amplification effect between particulate nitrate and liquid water is demonstrated by examining a nitrate-dominated fine particle Beijing pollution episode. The facilitation of particulate nitrate formation by abundant liquid water is subsequently theoretically explored through the impacts of 130 liquid water on thermodynamic equilibrium and heterogeneous reactions. Finally, the corresponding impacts on light extinction coefficient, and visibility degradation are estimated. These results improve our quantitative understanding of the development of haze events over the NCP and on formulating emission reduction strategies, as well as may also provide insights for other polluted regions.

Location and instrumentation
Measurements were conducted within the framework of the BEST-ONE (Beijing winter finE particle STudy-Oxidation, Nucleation, and light Extinctions) field campaign from January 1 to March 5, 2016, Detailed information about the sampling site was described in Tan et al. (2018). A weather station (Met 140 one Instrument Inc., USA) was performed to measure meteorological parameters (ambient RH, temperature, wind speed, wind direction) and detailed aerosol particle physical and chemical properties were recorded using a suite of state-of-the-science instrumentation. Hygroscopic growth factor (HGF) of sub-micrometer aerosol particles was measured using a Hygroscopicity-Tandem Differential Mobility Analyzer (H-TDMA, TROPOS, Germany) (Wu et al., 2011;Massling et al., 2011;Wang et al., 145 2018; Wu et al., 2016;Liu et al., 1978) and data retrieval followed the TDMAinv method in Gysel et al. (2009). The hygroscopicity parameter (κ) was estimated using by the κ-Kӧhler approach (Petters and Kreidenweis, 2007;Köhler, 1936). Size-resolved NR-PM1 was recorded by an Aerodyne High-Resolution Time-of-Flight Aerosol Mass Spectrometry (HR-ToF-AMS, Aerodyne Research, Inc., USA) (DeCarlo et al., 2006). Regular calibration procedures followed as reported in Jayne et al. (2000) and 150 Jimenez et al. (2003) and composition dependent correction followed as in Middlebrook et al. (2012).
Gaseous HNO3 and NH3 were measured using Gas-Aerosol Collector (GAC) coupled with Ion Chromatography (IC) (Dong et al., 2012). Mass concentration of equivalent black carbon in aerosol particles (Petzold et al., 2013)  Detailed description on H-TDMA, HR-ToF-AMS and GAC-IC can be found in the supporting 160 information.

Estimation of aerosol particle liquid water
Given the absence of direct liquid water measurement, size-resolved liquid water was calculated using the corresponding HGFs measured at RH=90% (50, 100, 150, 250, 350 nm in stokes diameter), PNSD data (3 nm~10 μm) and meteorological parameters (RH, T), following the method proposed in Bian et 165 al. (2014), referred to below as H-TDMA-derived liquid water. Briefly, the measured PNSD with 57 size bins were fitted using a four-mode lognormal distribution. The classification of four modes and the fitting results are shown in Table S1 and Figure S4. Good agreement between measured values and fitted PNSD was achieved, which indicates the reliability of the four-mode lognormal fitting method.
Based on four-mode lognormal fitting results, the particle number size distribution and number fractions 170 of each mode can be obtained. It has been assumed that particles from the same mode have constant particle hygroscopicity (κ). Under the assumption of constant particle hygroscopicity in each mode (shown in Table S1), the κ values for each mode (κ1, κ2, κ3) can be calculated by equation [1] from the known number fraction of fitted four modes and the κ values of measured particle size from H-TDMA measurement.
Here, κi and fi represent the κ value and the particle number fraction of the i mode. Then, the calculated κ values for each mode and the derived number fraction of each size bin were used to obtain the κ distribution for each size bin. Figure S5 shows the comparison of calculated sized-resolved κ distribution and the κ measured by H-TDMA, the good agreement showed the reliability of the method. 180 Then, based on κ-Kӧlher theory (Petters and Kreidenweis, 2007;Köhler, 1936), the size-resolved HGFs at ambient RH were calculated. Finally, liquid water of size-resolved particles can be derived by calculating the differentials between the dry and wet PNSD of aerosol particles in equation [2]: where j represents the bin number of measured PNSD, Nj and Dp,j represent the number concentration 185 and the diameter of dry particles of the j th bin, respectively, while, HGF and ρw, are the hygroscopic growth factor of aerosol particles and water density (1 g/cm 3 ), respectively.

Condensation rate of trace gases
The condensation rate (k) of trace gases (dinitrogen pentoxide, N2O5 and nitric acid, HNO3 in the constrained conditions, referred as k_N2O5 and k_HNO3 below) was calculated by the method of 190 Schwartz (1986), shown in equation [3]. In order to illustrate the influences of the dry and wet PNSD due to water uptake on condensation rate of gases, the PNSD of the dry and wet particles (obtained by applying the HGF estimated from H-TDMA-derived liquid water method) were used.
Where, r represents radius of the particles, Dg represents the binary diffusion coefficient evaluated following Maitland (1981) (1.18*e -5 m 2 /s). Cg is the kinetic velocity of the gas molecules, calculated in . Here, R and M are the ideal gas constant (8.314 kg.m 2 /mol/K/s 2 ) and molar mass of the gas, respectively while T represents the ambient temperature. dN/dlogr is the number size distribution and γ is the uptake coefficient of the gas. 200 The uptake coefficient of N2O5 was estimated following the method proposed in Chen et al. (2018b) and Chang et al. (2016) and references therein. The uptake suppression effect of N2O5 due to the presence of secondary organic aerosol (SOA) was considered following the method in Anttila et al. (2006). Based on our source apportionment using Positive matrix factorization (SoFi tool, ME2, Francesco Canonaco, PSI), two oxygenated organic aerosol factors (OOA), usually interpreted as SOA, and three primary 205 organic aerosol factors (POA) were determined. The fraction of SOA in the total organic aerosol (OA) was 60%~90% during the observed period, which is quite consistent with the results of a previous study in Beijing (Huang et al., 2014). Hence, 75% was used as the ratio of SOA/OA in our model to calculate uptake coefficient of the N2O5, where the suppression effect of SOA on the uptake of N2O5 was estimated following the work of Anttila et al. (2006). Additionally, the reaction of chloride with N2O5 210 was not considered in this study due to its limited mass concentration (on average 5% of the PM1 mass concentration during the marked haze period), which might cause uncertainty in the k_N2O5 calculation.
Therefore, γHNO3=0.01 and γHNO3=0.5 are selected to calculate the lower and upper limit of condensation rate of HNO3 in the atmosphere.

Equilibrium of NH4NO3
The equilibrium dissociation constant of NH4NO3 (Kp) under dry conditions was calculated as a 220 function of ambient temperature (Seinfeld. and Pandis., 2006) in the following equation [5].
= 84.6 − − 6.1 [5] Taking into account the associated liquid water, the equilibrium vapor pressure of HNO3 was calculated by employing the Extended-Aerosol Inorganic Model (E-AIM) Model Ⅱ H + -NH4 + -SO4 2--NO3 --H2O (Clegg et al., 1998) using HR-ToF-AMS data, NH3 from GAC-IC, and meteorological parameters (RH, 225 T). In this calculation, a simplified ion pairing scheme was performed to ensure the ion balance of the input chemical composition following the method in Gysel et al. (2007).

Light extinction coefficient and visibility calculation
Size-resolved chemical composition of the NR-PM1 from HR-ToF-AMS, mass concentration of equivalent black carbon from MAAP, PNSD data and the H-TDMA-derived liquid water were used to 230 calculate light extinction coefficient (including light absorption and scattering) and visibility degradation of size-resolved particles by the Mie scattering theory described in Barnard et al. (2010).
Here, size-resolved equivalent black carbon mass concentration was inferred by the particle mass size distribution measurement from single particle soot photometer in PKUERS. The method of redistribution of liquid water and HR-ToF-AMS data has been described in the supporting information 235 (Text S1, HR-ToF-AMS introduction section). Thus, with the re-distributed datasets as the input of the Mie scattering theory, the light extinction coefficient for atmospheric particles in the absence and presence of liquid water with a size range of 100~2500 nm in stokes diameter can be derived. Due to lack of measurements on aerosol particle morphology and mixing state, we assume particles are spherical as described in Barnard et al. (2010). To perform Mie calculation, the complex reflective 240 index of each component is given in Table 1 of Barnard et al. (2010) and references therein. This method shows good agreement with measurements in Mexico City and is consistent as the regional atmospheric chemistry model WRF-Chem. Here, Ext_550nm_wet and Ext_550nm_dry represent the  Figure S6) and elevated ambient RH ( Figure 1a). As marked 'haze period' in Figure 1, an obvious increase of NR-PM1 was observed. The secondary inorganic components (sulfate, nitrate and ammonium) were dominant the major contributor of the secondary inorganic components and accounted for up to ~44% of NR-PM1 mass, while sulfate contributed for ~12% on average.
In the recent decade, severe haze events with high aerosol mass loading occurred frequently in Beijing during wintertime Hu et al., 2017;Sun et al., 2014;Sun et al., 2015).

Mutual promotion effects between liquid water and nitrate formation
In the following discussion, the high fraction of particulate nitrate during the 'haze period' is elucidated by theoretical calculations considering the uptake of N2O5 and HNO3, and the thermodynamic equilibrium of NH4NO3. In particular, the role of aerosol water uptake in particulate nitrate formation is 280 comprehensively investigated.
N2O5 is an important gaseous precursor for nitrate formation via its hydrolysis to form HNO3 during nighttime (Brown et al., 2006). Liquid water can enhance aerosol surface areas and volumes, thereby increasing the available heterogeneous reacting medium. Across the development of 'haze period', the estimated liquid water increased from ~1 μg/m 3 at the beginning (2 th March, 14:00~18:00 p.m.) to ~75 285 μg/m 3 when the haze was fully developed (4 th March, 4:00~8:00 a.m.). The total surface area and volume concentrations of particles were increased by the liquid water by 2~3% at the beginning and by about 25~40% in the fully-developed haze compared to the 'dry' values, respectively (see Figure S7 and S8). Additionally, from the beginning to the fully-developed haze, the uptake coefficient of N2O5 was enhanced by a factor of 9 from 0.002 to 0.018, and the k_N2O5 increased by a factor of 20 (dry 290 particles); while, considering the increased particle surface area and volume due to water uptake, the respective value of enhanced k_N2O5 was 25 (Figure 2a). Apart from providing extra reacting medium, the abundant liquid water can liquefy the aerosol particles and may reduce any kinetic limitation of mass transfer for reactive gases Shiraiwa et al., 2011) and impact thermodynamic equilibrium of semi-volatile compounds (Kulmala et al., 1993;Topping et al., 2013)  was greater than the equilibrium Kp nearly all the time (Figure 3). In this case, gaseous NH3 and HNO3 in the atmosphere were supersaturated and would tend to partition into the dry particle phase gradually even in the absence of liquid water. Schwartz (1986) used below. As shown in Figure 4, the partitioning ratio (molar ratio between particulate and total nitrate) increasing with RH was observed during the development of haze, and 98% 320 of nitrate was present as particle phase when the haze was fully developed with liquid water increasing from 1 µg/m 3 to ~75 µg/m 3 .
Furthermore, the presence of aerosol associated water was substantially enhanced by the uptake rate of HNO3, which could dominate the gaseous HNO3 partitioning into particle phase throughout the haze developing. Because the negligible equilibrium vapor pressure suggests that HNO3 condensation loss 325 was not limited by thermodynamic equilibrium but limited by its uptake rate. The condensation (or uptake) rate of HNO3 (k_HNO3) can be calculated using equations [3][4]. Here, the lower and upper limit of k_HNO3 were calculated assuming the uptake coefficient (γ) of HNO3 in the range of 0.01 to 0.5 (Fenter et al., 1994;Leu et al., 1995;Beichert and Finlayson-Pitts, 1996;Abbatt and Waschewsky, 1998;Guimbaud et al., 2002). As shown in Figure 2b and 2c, the lower (upper) limit of k_HNO3 330 increased by a factor of 2.9 (2.7) for dry PNSD and 3.5 (3.1) for wet PNSD from the beginning to fullydeveloped haze period. As one can see, the liquid water facilitated the rate of HNO3 uptake and hence the particulate nitrate formation.
The above analyses quantify the effect of the increased aerosol surface area and volume concentrations resulting from the water uptake on the particulate nitrate formation through increased uptake of N2O5 335 and HNO3. Such an effect becomes more pronounced with the increasing pollution throughout the haze event owing to the simultaneously increasing ambient RH. Owing to its hygroscopicity, the increased ammonium nitrate mass fraction led to a further increase in aerosol surface area and volume concentrations through additional increase in liquid water, further enhancing uptake of condensable vapors.

340
It is worth noting that a similar co-condensation effect between water vapor and semi-volatile organic components (Topping and McFiggans, 2012;Topping et al., 2013;Hu et al., 2018) could promote the haze formation as well, for which there may be some evidence in the current case. Such a cocondensation effect will lead to the enhancement of semi-volatile organic and inorganic (e.g., nitrate) material with the increasing RH in a developing haze. The associated water will favor partitioning of 345 both particulate nitrate and semi-volatile organic materials to the particle phase depending on the organic solubility, providing a linkage between the development of increasing organic and inorganic particle mass.

The key role of liquid water on visibility degradation
Aerosol particles grow up in size as ambient RH increases, further enhances their extinction coefficient 350 and impacts visibility (Zhao et al., 2019;Kuang et al., 2016). In this section, size-resolved extinction coefficient of aerosol particles was estimated, and the influences of liquid water on the extinction coefficient and visibility were quantitatively evaluated. As shown in Figure 5a, the total light extinction coefficient of dry and wet aerosol particles enhanced by a factor of 4.3 and 5.4, respectively, from the beginning to a fully-developed haze. Correspondingly, the calculated visibility without considering 355 liquid water degraded significantly from ~10 km to less than 2 km within 48 hours during the marked 'haze period'. The contribution of aerosol associated water to visibility impairment was negligible in the beginning (2%), while it was significant (up to 24%) in the fully-developed haze (Figure 5b). This indicates that liquid water facilitated visibility degradation during haze development.
The influences of liquid water on visibility degradation varied with aerosol particle size. The size-360 resolved chemical composition data showed that the inorganic species, mainly nitrate, were dominant components in the aerosol particles within the size range of 300~700 nm ( Figure S3). Correspondingly, the particles in this size range contained most of the liquid water (50~80% of the total aerosol liquid water content of PM1). According to discussion in Sec. 3.2, the mutual promotion effect between liquid water and particulate nitrate can promote their formation. Aerosol particles in this size range 365 experienced the most significant enhancement of light extinction due to water uptake (Figure 6a and 6b) and contributed 70~88% of the total extinction coefficient of the total NR-PM1 ( Figure S9). In conclude, the rapid nitrate formation enhanced the aerosol extinction coefficient during haze developing, while the aerosol water uptake further enhanced the visibility degradation by increasing extinction coefficient and promoting nitrate formation. 370 It is worth noting that the enhanced dimming effect will further shallower the planetary boundary layer (PBL), which, in turn, depresses the dilution of water vapor and particulate matter in the atmosphere, hence leads to a higher RH and aerosol particle mass loading (Tie et al., 2017). Such effect is beyond the scope of this study.

375
In this study, we observed a nitrate-dominated (up to 44% of non-refractory PM1 mass concentration) particulate matter pollution episode, which is typical during winter haze in Beijing, China. A clear coincrease of aerosol particle liquid water and particulate nitrate was observed, demonstrating the mutual promotion effect between them via observation-based theoretical calculations.
As shown in Figure 7, the water uptake by hygroscopic aerosols increased the aerosol surface area and 380 volume, favoring the thermodynamic equilibrium of ammonium nitrate and enhancing the condensational loss of N2O5 and HNO3 over particles. The enhanced particulate nitrate formation from the above pathways increased the mass fraction of particulate nitrate, which had a lower deliquescence RH than sulfate and resulted in more water uptake at lower ambient RH (Kreidenweis and Asa-Awuku, 2014). Hence, the increased aerosol particle surface area and volume concentrations due to water 385 uptake, in turn facilitates particulate nitrate formation. Hence, a feedback loop between liquid water and particulate nitrate is built up. Therefore the enhanced particulate nitrate components can accelerate the feedback compared with sulfate-rich pollution over the NCP region in the past . This self-amplification can rapidly degrade air quality and halve visibility within one day. Our results highlight the importance of reducing the particulate nitrate and its precursors (e.g. NOx) for mitigation 390 of haze episodes in NCP region.

Data availability
The observational dataset of the BEST-ONE campaign can be accessed through the corresponding author Z. Wu (zhijunwu@pku.edu.cn).

Figure 1: The time series of (a) NR-PM1 chemical composition measured by the HR-ToF-AMS
and ambient RH (red solid line), (b) size-segregated aerosol particle liquid water and the total mass concentration of liquid water with smaller than 1 µm in aerodynamic diameter (red solid line), (c) size-segregated aerosol particle surface area and total aerosol particle surface area without considering particle hygroscopic growth.