Impacts of Aerosol Chemical Composition on Cloud Condensation Nuclei (CCN) Activity during Wintertime in Beijing, China

Abstract

The cloud condensation nuclei (CCN) activity and aerosol chemical composition were concurrently measured via a scanning mobility CCN analyzer (SMCA) and an Aerodyne Time-of-Flight Aerosol Chemical Speciation Monitor (ACSM), respectively, during wintertime 2022 in Beijing, China. During the observation period, the mean CCN number concentrations ranged from 1345 ± 1270 cm⁻³ at SS = 0.1% to 3267 ± 2325 cm⁻³ at SS = 0.3%. The mean critical activation diameters (D₅₀) at SS = 0.1%, 0.2%, and 0.3% were 172 ± 13 nm, 102 ± 8 nm, and 84 ± 7 nm, corresponding to the average hygroscopicity parameters (κ_CCN) of 0.34, 0.33, and 0.26, respectively. The diurnal variations in D₅₀ suggested that the local primary emissions significantly enhanced D₅₀ at SS = 0.2% and 0.3%, but had less influence on D₅₀ at SS = 0.1% due to the limited size (<150 nm) of particles emitted from primary sources. As PM_2.5 concentration increases, the dominant driver of CCN activity transitions from sulfate to nitrate. At a specific SS, D₅₀ decreased with increases in the degree of internal mixing, implying that the elevated internal mixing degree during atmospheric aging was beneficial to CCN activation. In this study, the commonly used f44 (or O:C) was weakly correlated with κ_org and failed to describe the variations in κ_org. Instead, the variations in κ_org can be well parameterized with the Org/BC ratio. The correlation between κ derived from bulk chemical compositions and CCN measurements was substantially improved when this κ_org scheme was adopted, emphasizing the importance of considering κ_org variations on deriving κ_chem from aerosol chemical composition.

1. Introduction

Aerosol particles can serve as cloud condensation nuclei (CCN) under supersaturation (SS) conditions [1,2], participating in cloud processes and altering the microphysical and optical properties of clouds, which indirectly affect the global climate [3,4]. The Sixth Assessment Report from IPCC (2021) highlighted that aerosol indirect effects remain the most considerable uncertainty in evaluating the anthropogenic contribution on current and future climate change [5]. Remote sensing has been widely used to study aerosol–cloud–climate interactions and provide extensive datasets. However, it is difficult to identify the proportion of CCN from retrieved total aerosol concentration (CCN activation ratio), which would introduce uncertainties in estimating aerosol-related effects on cloud microphysics based on remote sensing. Furthermore, accurate retrievals of cloud droplet number concentrations (N_d) using remote sensing measurements is challenging. Based on the ground-level measurements of aerosol activation properties and parcel theory, the N_d for non-precipitating Planetary Boundary Layer clouds can be derived. Then, the parameters of the retrieval can be optimized through the closure analysis on N_d obtained from remote sensing data and in situ measurement at ground level [6].

The ability of aerosol particles to participate in cloud processes, known as CCN activity, is one of critical parameters in studying the indirect radiation effects and environmental impacts of aerosols. At a given SS condition, the CCN activity is mainly determined by their size distribution, chemical composition, and mixing state [7,8,9]. Previous studies suggested that the size distribution of aerosols is a more critical factor than their chemical composition in governing CCN activity [2,10]. However, the impact of chemical composition on CCN activity significantly increases in situations with low supersaturation [11,12]. The classical Köhler theory [1], combining the Raoult’s law effect with the Kelvin effect, illustrates that the aerosol particle activation depends on particle size, chemical composition, and SS. Petters and Kreidenweis [13] developed a single hygroscopicity parameter κ in parameterizing the Raoult term, which enabled the quantification of CCN activity without requiring information of the dissolved compounds. However, ambient aerosols are complex mixtures, and their sizes and compositions vary at different stages of atmospheric aging process [14,15,16], which include numerous primary and secondary organic components as well. These impose various effects on particle activation behavior and reduce the applicability of the Köhler theory in describing the actual atmospheric aerosols, especially for organic aerosols (OA). Earlier studies specifically focused on parametrizing the empirical correlations between the OA oxidation level and organic hygroscopicity parameter (κ_org) [17,18,19,20,21], suggesting a linear correlation between κ_org and the oxygen-to-carbon atomic ratio (O:C). However, the systematic variability in parametrizations between κ_org and the O:C ratio determined from the different studies remains large [22]. Kuang et al. (2021) recently pointed out that secondary OA (SOA) from different precursors and formation processes may result in SOA with different chemical composition, functional properties, and microphysical structure, consequently exerting distinct influences on κ_org and rendering single oxidation level parameters (such as O:C) unable to capture those differences [23].

To obtain regional parameters for aerosol particle activation, numerous field campaigns have been conducted globally. Alongside the considerable spatial heterogeneity, temporal variability in aerosol activation characteristics is also essential for accurate CCN prediction [7]. The CCN activity can be characterized by two important parameters: activation critical diameter (D₅₀) and activation ratio (AR). These two parameters can be directly measured through combining cloud condensation nuclei counter (CCNc) and scanning mobility particle sizer (SMPS) [24,25], and can also be determined by aerosol chemical composition and particle number size distribution (PNSD) measurements using the Köhler equation and internal mixture hypothesis [20,26,27,28]. The CCN activity closure analysis, based on the above two methods, is a key means of investigating the impacts of aerosol properties on activation processes. Although CCN measurements were conducted under different environments in the North China Plain, only a few studies established a linkage between CCN activity and chemical composition with a high time resolution [27,29,30,31]. Additionally, due to the Clean Air Action of the Chinese government initiated in 2013, stringent emission controls have been imposed to decrease fine particulate matter pollution (PM_2.5, particles smaller than 2.5 μm in diameter) [32]. In the Beijing region, notable reductions in PM_2.5 concentration over the past decade have been accompanied by significant changes in its chemical composition [33,34,35]. Instead of organics and sulfate, organics and nitrate are now the dominant components of Beijing winter haze pollution. Therefore, more investigations are needed to further understand the impacts of these variations on CCN activity in urban Beijing.

In this study, the CCN activity and aerosol chemical composition were simultaneously measured in Beijing wintertime, using a scanning mobility CCN analyzer (SMCA) and an Aerodyne aerosol chemical species monitor (ACSM). An overview of the observation data is presented in Section Overview the Observation Data. The impacts of aerosol chemical composition and mixing state on CCN activity are discussed in Section 4.1. Finally, in Section 4.2, through closure analysis of the aerosol hygroscopicity derived from CCN activity and chemical composition, we estimate κ_org and parameterize it with the concentration ratio of organics to BC (Org/BC).

2. Materials and Methods

2.1. Measurement Site

Measurements of size-resolved CCN activity, aerosol size distribution and composition (organics, nitrate, sulfate, ammonium, chloride, and black carbon) were conducted on the rooftop of the Chinese Academy of Meteorological Sciences (CAMS) building during the period of 15 January–3 March 2022. All aerosol measurements were performed downstream of a PM₁₀ impactor. The ambient air passes through the impactor and then enters the automatic regenerating adsorption aerosol dryer (TROPOS, Leipzig, Germany), which can dry the sample air to a low RH (RH < 30%) to ensure comparability of measurements [36]. The measurement site (116°19′E, 39°57′N, approximately 53 m above the ground) is located in a traffic-intensive area of the west 3rd ring road in Beijing, China, which represents the urban environment influenced by intense anthropogenic emissions [37,38].

2.2. Instrumentation

2.2.1. Size-Resolved CCN Activity and PNSD Measurements

Size-resolved CCN spectra and activation ratios were measured using the SMCA initially proposed by Moore et al. (2010) [39]. In this work, the SMCA consisted of a CCNc-100 (DMT Inc., Blouder, CO, USA), a differential mobility analyzer (DMA, model 3081, TSI Inc., Shoreview, MN, USA), and a condensation particle counter (CPC, model 3772, TSI Inc.). In the SMCA system, the combined DMA and CPC were used as a SMPS during the measurements. The dry particles after the diffusion dryer were neutralized by the X-ray neutralizer and were subsequently classified by the DMA in a scanning mode. Then, the particles were split into two streams: one to the CPC for measurement of total particle number concentration (N_CN) and another to the CCNc-100 for measurements of the CCN number concentration (N_CCN). The flow rates of the CPC and CCNc-100 were 1.0 L min⁻¹ and 0.5 L min⁻¹, respectively. With the total sample flow of 1.5 L min⁻¹ and sheath flow of 7.5 L min⁻¹, the SMCA was protocolled to measure particles at a mobility diameter range of 8–360 nm. The DMA was operated in scanning mode with a time resolution of 180 s, including a scan time of 165 s and a retrace time of 15 s. The supersaturation in the CCNc-100 was set to be 0.1%, 0.2%, and 0.3%, respectively, with a time interval of 9 min (corresponding to three diameter scans of the DMA) for each SS during the experiment. Only the second scan for each SS is used as the CCNc needs time for SS stabilization. Therefore, a complete SS cycle lasted ~0.5 h. The CCNc-100 was regularly calibrated with ammonium sulfate particles at the four SSs (0.1%, 0.2%, 0.4%, and 0.7%).

2.2.2. Aerosol Chemical Composition Measurement

Non-refractory particulate matter (NR-PM) including sulfate, nitrate, ammonium, and organics with dry aerodynamic diameters below 2.5 μm was measured by an ACSM equipped with a PM_2.5 aerodynamic lens at 10 min time resolution. The ACSM data were analyzed using standard data analysis software (Tofware v2.5.13). The collection efficiency (CE), which accounts for particle loss due to bouncing off the standard vaporizer, was derived from composition-dependent calculation following the algorithms proposed by Middlebrook et al. (2012) [40]. The calibrations of ionization efficiency (IE) and relative ionization efficiencies (RIEs) followed the standard procedures described in previous studies by using 300–350 nm pure ammonium nitrate and ammonium sulfate particles [41,42]. The RIE values used for organics, sulfate, nitrate, ammonium, and chloride were 1.4, 1.2, 1.1, 3.4, and 1.3, respectively.

Equivalent black carbon (eBC) mass was measured with an aethalometer (AE33, MAGEE Scientific, Berkeley, CA, USA) at 1 min. The instruments used dual-dots configuration to auto-correct for the loading effect. The measured absorption was converted to BC mass using an apparent mass absorption cross section (MAC) of 7.7 m² g⁻¹ at a wavelength of 880 nm [43]. The λ = 880 nm is chosen to avoid the potential interference of brown carbon at shorter wavelength. The multi-scattering enhancement factor (C value) of 2.88 at 880nm wavelength was used to exclude the multiple light scattering effects.

In addition, PM_2.5 mass concentration data from air quality monitoring station (Wanliu) operated by the Beijing Municipal Ecology and Environment Bureau were used, which is located in the northwest and northeast direction of the CAMS observation site, with a distance of approximately 3 km. The concurrent meteorological parameters (including wind speed, wind direction, relative humidity, and air temperature) are from the national surface meteorological observation stations of the China Meteorological Administration.

2.3. Data Processing

2.3.1. CCN Activation

The N_CN and N_CCN data were used to calculate the size-resolved CCN activation ratio (AR), which defined as the ratio of N_CCN to N_CN at each particle size (N_CCN/N_CN vs. D_p). The influence of multiple-charge particles and DMA transfer function on AR were corrected basing on a modified algorithm developed by Hagen and Alofs (1983) [44] and Deng et al. (2011, 2013) [25,45]. Figure S1 gives a typical example of N_CCN/N_CN vs. D_p during the experiment, and the AR curve can be obtained by fitting the N_CCN/N_CN ratio with the sigmoidal function:

$$\mathrm{A}\mathrm{R}=a+b/(1+\exp (\frac{D_{50}-D_p}{{\sigma }_s}))$$

(1)

where a is the offset from 0 in the y direction and b is the height of the upper plateau of the sigmoid function, D₅₀ is the critical diameter, and σ_s is a measure for the width of the sigmoid function. D₅₀ is also called the critical diameter or the activation diameter, that is, the diameter at which 50% of the particles are activated at a specific SS. The variation in D₅₀ at a given SS can be attributed to the variation in the number fraction of hygroscopic particles or its hygroscopicity, which can be indicated the CCN activity of a particle population. Then, the hygroscopicity parameter κ_CCN, derived from CCN measurements, is calculated from the critical supersaturation ratio (S_c) and D₅₀ from the following equation [13]:

$${\kappa }_{\mathrm{C}\mathrm{C}\mathrm{N}}=\frac{{4\mathrm{A}}^3}{27D_{50}^3{\left(ln{S}_c\right)}^2},\hspace{1em}\hspace{1em}A=\frac{4{\sigma }_{s/a}{M}_w}{RT{\rho }_w}$$

(2)

where σ_s/a is the surface tension of the solution/air interface and the solution is temporarily assumed to be pure water (0.0728 N m⁻¹ at 298.15 K), M_w is the molecular weight of water, R is the universal gas constant (8.314 J mol⁻¹ K⁻¹), T is absolute temperature, ρ_w is the density of water.

The steepness of the AR curve represents the degree of external mixture or the heterogeneity of aerosol chemical composition, which can be quantified by the ratio of (D₇₅ − D₂₅)/D₅₀ [46]. D₇₅ and D₂₅ correspond to the diameters at which 75% and 25% of the particles are activated on the AR curve. Internal mixture implies that particles at any given dry size have equal κ with (D₇₅ − D₂₅)/D₅₀ = 0. Alternatively, various κ at a given particle size indicate externally mixed aerosol compositions with higher (D₇₅ − D₂₅)/D_c values. Although the sigma term from sigmoid fit has been used in previous studies to estimate aerosol mixing state [19,30], the (D₇₅ − D₂₅) value can provide a more accurate representation of aerosol mixing state. This is because the CCN measurement uncertainties would be larger due to the extremely low concentration of activated particles at the low AR range. In addition, the relationship between activation ratio and particle size exhibits nonlinear behavior when AR < 25% or >75% (Figure S1). In contrast, within the particle size range of D₂₅ to D₇₅, the AR increases rapidly and exhibits a relatively linear relationship with the particle size. The sigma term represents the width of entire particle diameter range used in sigmoid fit. Instead of the sigma term, using the (D₇₅ − D₂₅)/D₅₀ value to describe the steepness of the AR curve can reduce the uncertainties in estimating aerosol mixing state.

2.3.2. Aerosol Hygroscopicity Derived from Aerosol Chemical Composition

The hygroscopic parameter κ is solely determined by composition and reflects the Raoult term in Köhler theory [13]. By assuming internal mixing, the aerosol hygroscopicity (κ_chem) can be derived from aerosol chemical composition measurements. The κ_chem of aerosol particles with multiple compositions is contributed by the κ of each volume-weighted composition, following the Zdanovskii–Stokes–Robinson (ZSR) mixing rule [28], expressed as:

$${\kappa }_{\mathrm{c}\mathrm{h}\mathrm{e}\mathrm{m}}={\sum }_i{\epsilon }_i{\kappa }_i$$

(3)

where i represents the ith composition, ε_i is the volume fraction of each composition in the bulk, and k_i is the hygroscopic parameter for each composition present in a multiple-composition particle. For the compositions determined by ACSM the AE33 measurements, a simple ion-pairing scheme [9] was used in this study, expressed as:

$$\begin{array}{rrr}{n}_{{NH}_4{NO}_3}& =& n_{{NO}_3^{-}}\hfill \\ n_{H_2{SO}_4}& =& \max \left(0,n_{{SO}_4^{2-}}-n_{{NH}_4^{+}}+n_{{NO}_3^{-}}\right)\hfill \\ n_{{NH}_4{HSO}_4}& =& \mathrm{m}\mathrm{i}\mathrm{n}(2n_{{SO}_4^{2-}}-n_{{NH}_4^{+}}+n_{{NO}_3^{-}},n_{{NH}_4^{+}}-n_{{NO}_3^{-}})\hfill \\ n_{{\left({NH}_4\right)}_2{SO}_4}& =& \mathrm{m}\mathrm{a}\mathrm{x}(0,n_{{NH}_4^{+}}-n_{{NO}_3^{-}})\hfill \\ n_{{HNO}_3}& =& 0\hfill \end{array}$$

(4)

All species are then converted to volume by assuming a density. The density (ρ_i) and κ_i values for all species used in this study are listed in

Table 1. Density (ρ) and hygroscopicity parameter (κ) of each pure composition used in this study.

Species	Density (kg m−3)	κ
NH4NO3	1725	0.68
(NH4)2SO4	1769	0.52
NH4HSO4	1780	0.56
H2SO4	1830	0.92
Organics	1400	0.1
BC	1800	0

[15]. Here, a constant κ of 0.1 is used for particulate organics [47,48].

3. Results

Overview the Observation Data

The time series of meteorological parameters (such as wind speed, wind direction, ambient air temperature, and relative humidity (RH)), PM_2.5 mass concentration, N_CN, and N_CCN at SS of 0.1%, 0.2%, and 0.3% are shown in Figure 1. During the observation periods, the wind speed was mainly less than 3 m s⁻¹, with an average of 1.3 m s⁻¹. The relatively low wind speeds (<2 m s⁻¹) were generally associated with southerly winds, whereas higher wind speeds (>3 m s⁻¹) corresponded to northerly winds. When persistent southerly winds prevail (such as January 17–24th), the daily average PM_2.5 concentration and RH significantly increased from 19 to 156 μg m⁻³, and from 31% to 77%, respectively. The cumulative enhancement of PM_2.5 concentrations was jointly influenced by local emissions, regional transport, and liquid/heterogeneous reactions at different stages of this persistent pollution episode [49]. The RH was below 60% for most of the time, with an average of 38 ± 18%. The temperature ranged from −9 to 17 °C, with an average of 0.2 ± 4.7 °C. These meteorological parameters indicated relatively dry and cold weather conditions during this campaign.

The temporal variations in both aerosol number and mass concentration are shown in Figure 1c. The range of aerosol number concentration varied from ~1000 to 17,000 cm⁻³, with an average of 6714 ± 3117 cm⁻³. The PM_2.5 mass concentration ranged from 2 μg m⁻³ to 197 μg m⁻³, with an average of 34 ± 39 μg m⁻³. During the most increasing stages of PM_2.5 mass concentration, the aerosol number concentration consistently varied with the mass concentration. However, when an intermittent fog event occurred from January 23rd to 24th, the number concentration decreased with the increasing mass concentration. This suggests that the wet scavenging effect of fog droplets substantially reduced aerosol number concentration, but enhanced liquid-phase reactions alternatively resulting in the increases in mass concentration. The CCN concentration at low supersaturations (0.1%) was significantly lower than the two relatively higher supersaturations (0.2% and 0.3%) (Figure 1d), with average concentrations of 1345 ± 1270 cm⁻³, 2752 ± 2114 cm⁻³, and 3267 ± 2325 cm⁻³, respectively.

The time series of PM_2.5 chemical composition, dry particle number size distribution (PNSD), and D₅₀ are given in Figure 2. The particle chemical composition is closely related to its hygroscopicity and activation characteristics. The hygroscopic properties of aerosols are mainly determined by composition, with inorganics having higher hygroscopicity [9,50] than less water-soluble substances, such as BC [51,52] or primary organics [53]. At a given SS, particles with higher hygroscopicity are corresponding to smaller D₅₀, indicating such particles are more prone to activation and form cloud droplets. During the observation period, organic matter and nitrate were the most important components, with average percentages of 42.4% ± 11.9% and 21.5% ± 12.1%, respectively, followed by ammonium (18.4% ± 5.0%), sulfate (10.3% ± 7.3%), BC (5.0% ± 2.2%), and chloride (2.5% ± 1.8%), respectively. Aerosol chemical composition had a remarkable variation with PM_2.5 concentration levels, characterized by organics dominating at low PM_2.5 concentration levels (<25 μg m⁻³) and nitrate gradually becoming the dominant composition with increases in PM_2.5 concentration. These characteristics of aerosol composition are consistent with recent studies, which reported that nitrate is now the principal component of Beijing winter haze pollution [33,34]. As a hygroscopic component, the increasing fraction of nitrate during pollution episodes facilitates the enhancement of aerosol hygroscopicity and activity. The time series of PNSD in Figure 2b shows that aerosol particles mainly concentrate in nucleation mode (D_p < 100 nm) during the clean periods. However, as the PM_2.5 mass concentration increases, there is a significant increase in the proportion of particle numbers in the accumulation mode, accompanied by a decrease in D₅₀. The average D₅₀ at SS = 0.1% (172 ± 13 nm) was significantly larger than that at SS = 0.2% (102 ± 8 nm) and SS = 0.3% (84 ± 7 nm), indicating that the SS variation was more sensitive to D₅₀ at a lower SS range.

4. Discussion

4.1. Impacts of Aerosol Physicochemical Properties on CCN Activity

4.1.1. Aerosol Chemical Properties

The diurnal variations in D₅₀ at different SS, the mass fraction of m/z 44 in organic mass spectrum (f44), aerosol chemical composition, and PNSD are given in Figure 3. The sources and properties of aerosol particles vary with different size ranges, leading to a diverse diurnal variation in D₅₀ at different SS (Figure 3a–c). As a typical tracer for SOA formation, the f44 can be associated with OA hygroscopicity owing to their positive correlations reported in previous studies [54,55]. Both f44 and the sum fraction of hygroscopic inorganic components (nitrate, sulfate, and ammonium) had a noticeable peak in the afternoon (Figure 3d,e), implying they were significantly influenced by photochemical oxidation in the daytime. Meanwhile, the minimum of D₅₀ for all three SS conditions occurred around 15:00, corresponding to the diurnal peak time of f44 and the fraction of inorganic components. This suggests photochemical reactions on daily basis played an important role in enhancing aerosol hygroscopicity and CCN activity.

Note that the diurnal cycles of D₅₀ at SS = 0.2% and 0.3% exhibited a remarkable maximum around 18:00, with additional peaks occurring at 7:00 and 12:00. In contrast, no prominent diurnal peak occurred in the diurnal cycle of D₅₀ at SS = 0.1%. Traffic and cooking emissions are the main sources of primary OA (POA) in Beijing, and these fresh emissions could lead to a decrease in f44 [56,57]. Therefore, the reduction time in the f44 diurnal cycle matched to the periods of daily source emissions, such as morning and evening traffic peaks, as well as lunch and dinner times. Consequently, the hydrophobic POA was not favor for CCN activation, leading to an increase in D₅₀ at SS = 0.2% and 0.3%. In addition, the diurnal variation in aerosol number concentration within the size range of 60–150 nm was consistent with that of D₅₀ (SS = 0.2% and 0.3%) (Figure 3f), indicating that particles within this size range were significantly influenced by traffic and cooking emissions. For particles larger than 150 nm, their number concentration showed no significant changes during the periods of concentrated emissions, suggesting that local primary emissions had few influences on particle properties with this size range. Therefore, there were no significant peaks observed in the diurnal variation in D₅₀ at SS = 0.1%, which was generally larger than 150 nm during this campaign (Figure 3a).

The mean value of PM_2.5 mass concentration during this campaign is 25 μg m⁻³. Once the PM_2.5 mass concentration exceeded this mean value, a period of pollution accumulation typically occurred for one to several days. Therefore, in order to further illustrate the impacts of aerosol chemical composition on CCN activity at the different pollution levels, we defined the PM_2.5 mass concentrations ≤25 μg m⁻³ and >25 μg m⁻³ as clean and polluted conditions, respectively. Note that the bulk aerosol composition measured by the ASCM is sensitive to larger particles, which contribute the majority of the total mass. Thus, the bulk composition tends to mainly reflect the composition information for larger particles. Here, only the relationships between the fraction of each chemical component and D₅₀ at SS = 0.1% are presented in Figure 4. Under the polluted conditions, the fraction of nitrate increased with PM_2.5 concentrations and exhibited a negative correlation with D₅₀ (r = −0.68), but no correlation between them (r = 0.02) under the clean conditions. In contrast, the fraction of sulfate only showed a correlation with D₅₀ under the clean conditions (r = −0.47). These relationships indicated that as the concentration of PM_2.5 increases, the dominant driver of CCN activity transitioned from sulfate to nitrate. Compared with the secondary inorganic components, both the fractions of organics and BC showed a positive correlation with D₅₀ due to their limited hygroscopicity.

4.1.2. Mixing State and Its Impacts on CCN Activity

As introduced in Section 2.3.1, the ratio of (D₇₅ − D₂₅)/D₅₀ can be used to describe the external-mixing degree of aerosol components. The relationships between (D₇₅ − D₂₅)/D₅₀ and D₅₀ under different SS conditions are shown in Figure 5. Jurányi et al. (2013) pointed out that the (D₇₅ − D₂₅)/D₅₀ varied from 0.08 to 0.12 for ammonium sulfate calibration measurements at SS = 0.1–1.0%, which indicated an internal mixture within measurement accuracy [46]. For our measurements, the value of (D₇₅ − D₂₅)/D₅₀ distributed in a range of 0.14–0.59, with a median value of 0.31, 0.31, and 0.33 at SS = 0.1%, 0.2% and 0.3%, respectively. The higher (D₇₅ − D₂₅)/D₅₀ at SS = 0.3% suggested the smaller particles (D_p < 100 nm) tended to be more external mixing. The (D₇₅ − D₂₅)/D₅₀ in this study is significantly higher than that in a rural background site [31], where the (D₇₅ − D₂₅)/D₅₀ ranged from 0.15 to 0.20 at SS = 0.1 to 0.7%. Compared with the rural background site, aerosol particles in urban Beijing represent a more complex mixture originating from local anthropogenic emissions and regional transport.

For a specific SS, D₅₀ decreased with an increase in the degree of internal mixing, implying that enhancing the internal mixing level facilitated the CCN activity. Particularly, the variation in the mixing state exhibits high sensitivity to D₅₀ when the ratio drops from 0.5 to 0.3. This can be attributed to the increases in the internal mixing level during the aerosol aging process, leading to an enhancement in aerosol chemical homogeneity and hygroscopicity. The relationship between (D₇₅ − D₂₅)/D₅₀ and D₅₀ can be well fitted with an exponential function under the conditions of SS = 0.2% and 0.3%. However, under the condition of SS = 0.1%, their relationship is hard to be adequately fitted. Further analysis of the relationship (SS = 0.1%) revealed that the relatively high values of (D₇₅ − D₂₅)/D₅₀ (0.3–0.4) with small D₅₀ (<165 nm) occurred under high PM_2.5 concentration (>60 μg m⁻³) and high RH (>60%) conditions (Figure S2), leading to the relationship deviated from satisfactory fitting by using an exponential function. Under these conditions, the hygroscopicity and size of aerosol particles were enlarged due to the increasing fraction of nitrate and hygroscopic growth, respectively. Especially for the fog period, the droplets had a more diverse population of κ values, as separated by some of them already condensing droplets. However, after passing through the sampling line and being dried, these enlarged particles experienced a reduction in liquid water content and particle size. As a result, the measured dry particles represent a mixture of ambient particles (or droplets) from a wider size range with various κ values, thereby exhibiting a special kind of chemical inhomogeneity.

4.2. Estimating Organic Aerosol Hygroscopicity

The size-resolved κ_ccn can be derived from size-resolved CCN measurements by using the D₅₀ at various SS. Figure 6 shows the relationships between the κ_ccn and D_p at various environmental sites in China, including urban, rural, coast, and mountain sites. As can be seen, κ_ccn exhibited a positive correlation with D_p at most of those sites except for mountain site, which had a higher κ_ccn for the smaller particles. In these results, κ_ccn rapidly increased with particle size at the range of ~40 to 100 nm, whereas κ_ccn almost stays constant at D_p of 100 to 200 nm. This is consistent with previously reported results on the size-resolved aerosol chemical compositions [27], which presented a substantially varying composition within small particle size range and relatively stable compositions for larger particles. In addition, the κ_ccn observed at urban sites were significantly lower than that at rural sites, which can be attributed to the enhanced aerosol aging state from urban to rural environment [14]. The κ_chem derived from bulk chemical composition were notably lower than κ_ccn during the observation periods of this study, especially for larger particle size.

As the ACSM is sensitive to particle mass rather than number concentration, the bulk composition is dominated by the contribution of the larger particles. Therefore, a better correlation between κ_chem and κ_ccn was exhibited at SS of 0.1% compared with the other two SS conditions (Figure 7a and Figure S3), suggesting the κ_ccn measured at SS = 0.1% was more suitable for closure analysis with κ_chem derived from bulk composition. Here, we assume the discrepancy between κ_chem and κ_ccn was mainly induced by the utilization of constant OA hygroscopicity parameter (κ_org). Then, the κ_org can be derived by subtracting the contribution of the inorganic salts (κ_inorg) from the κ_ccn at SS = 0.1%, expressed as:

$${\kappa }_{\mathrm{o}\mathrm{r}\mathrm{g}}=\frac{{\kappa }_{\mathrm{c}\mathrm{c}\mathrm{n}}-({\kappa }_{\mathrm{i}\mathrm{n}\mathrm{o}\mathrm{r}\mathrm{g}}{\epsilon }_{\mathrm{i}\mathrm{n}\mathrm{o}\mathrm{r}\mathrm{g}}+{\kappa }_{\mathrm{B}\mathrm{C}}{\epsilon }_{\mathrm{B}\mathrm{C}})}{{\epsilon }_{\mathrm{o}\mathrm{r}\mathrm{g}}}$$

(5)

Duplissy et al. (2011) pointed out that the uncertainty in the estimation of κ_org decreases with increasing organic fraction; thereby, only data featuring organic fractions larger than 50% were used in this calculation [17].

As is shown in Figure 7b, the κ_org shows a positive relationship with the concentration ratio of organics to BC (Org/BC), which can be used to reflect the contribution of SOA to total OA [60,61]. Several previous studies suggested κ_org as a function of the f44 or O:C [54,55]; however, κ_org was not correlated to the f44 during the observation periods of this study. This is similar to a recent study conducted by Kuang et al. (2021), which pointed out that κ_org was weakly correlated with O:C, but well correlated with the sum mass fraction of aged biomass burning OA and more oxygenated OA in total OA [23]. In order to parameterize the relationship between κ_org and Org/BC ratios, κ_org values were binned by Org/BC in increments of 5. Then, a power fitting function κ_org = 0.05×(Org/BC)^0.42 was obtained. This fitting function was adopted to reconstruct the hygroscopicity parameter (κ_{chem_new}). As is shown in Figure 7a, a better correlation between κ_chem and κ_ccn was achieved after using the parameterized κ_org instead of constant value (0.1), indicating suitable κ_org scheme play an important role in deriving κ_chem from aerosol chemical composition.

5. Conclusions

The in situ measurements of CCN activity at ground level can be applied to optimize the parameters of retrieving cloud microphysics with remote sensing data. Notable reductions in PM_2.5 concentration over the past decade have been accompanied by significant changes in its chemical composition in the North China Plain. Therefore, more investigations are needed to further understand the impacts of these variations on CCN activity. In this study, aerosol size distribution, chemical composition, and CCN number concentration were measured concurrently to investigate the size-resolved CCN activity during wintertime 2022 in Beijing, China. The critical activation dimeter D₅₀ derived from size-resolved activation ratio can be negatively associated with CCN activity. During the observation period, the average CCN concentrations at SS = 0.1%, 0.2%, and 0.3% were 1340 cm⁻³, 2750 cm⁻³, and 3270 cm⁻³, corresponding to the average D₅₀ of 172 nm, 102 nm, and 84 nm, respectively. The impacts of aerosol size distribution, chemical composition, and mixing state on CCN activity were investigated. The results show that photochemical oxidation on a daily basis played an important role in enhancing aerosol hygroscopicity and CCN activity. Conversely, the local emissions (such as traffic and cooking) significantly enhanced D₅₀ at SS = 0.2% and 0.3%, but had few influences on D₅₀ at SS = 0.1% due to the limitations of particle size from emission sources. As PM_2.5 concentration increases, the dominant driver of CCN activity transitions from sulfate to nitrate. For carbonaceous components, both the fractions of organics and BC showed a positive correlation with D₅₀ due to their weak hygroscopicity. The mixing state of particles was also estimated based on the size-resolved AR curve fitting. At a specific SS, D₅₀ decreased with increasing the degree of internal mixing, implying that the elevated internal-mixing degree resulting from atmospheric aging facilitated to the enhancement of CCN activity.

Aerosol hygroscopicity parameter κ_ccn derived from the SMCA method was size-dependent, varying from 0.26 to 0.34 as D_p increased from 84 to 172 nm. Based on the bulk aerosol chemical composition measured by the ACSM, the κ_chem values were calculated using the ZSR method. The OA hygroscopicity κ_org was estimated by the closure analysis between κ_chem and κ_ccn. However, the commonly used f44 (or O:C) was weakly correlated with κ_org and failed to describe the variations in κ_org during the observation periods of this study. Instead, the variations in κ_org can be well parameterized using the Org/BC ratio, which is usually used as a tracer for the contribution of SOA to total OA. The parameterization of κ_org substantially improved the correlation between κ_chem and κ_ccn, emphasizing the importance of considering κ_org variations when deriving κ_chem from aerosol chemical composition.