Impacts of organic aerosols and its oxidation level on CCN activity from measurement at a suburban site in China

1 State Key Laboratory of Earth Surface Processes and Resource Ecology, College of 5 Global Change and Earth System Science, Beijing Normal University, Beijing 100875, China 2 Joint Center for Global Change Studies, Beijing 100875, China 3 Earth System Science Interdisciplinary Center and Department of Atmospheric and Oceanic Science, University of Maryland, College Park, Maryland, USA. 10 4 State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China 5 Key Laboratory of Atmospheric Composition and Optical Radiation, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Hefei 230031, China 15 6 Liaoning Weather Modification Office, Shenyang, 112000, China


Introduction
To reduce the uncertainty of aerosol indirect effects on the radiative balance of the atmosphere, it is important to gain a good knowledge of the ability of aerosol particles to form cloud condensation nuclei (CCN) at the typical supersaturations found in the atmosphere.The CCN activity of aerosol particles is governed by the Köhler theory (Köhler, 1936).This theory determines CCN from aerosol particle size and physicochemical properties, which include the molar volume, activity coefficient, and effect on surface tension (Mc-Figgans et al., 2006).These properties, however, are difficult to measure.
Researchers have proposed single-parameter models to parameterize the CCN activation and hygroscopicity of multicomponent aerosols (Hudson and Da, 1996;Rissler et al., 2006;Petters and Kreidenweis, 2007;Wex et al., 2007).Field experiments have been conducted with the aim of better characterizing particle physicochemical parameters influencing cloud CCN activation.Due to the large spatial variability of aerosol types and compositions, the CCN activation efficiency varies greatly over different regions.CCN number concentrations (N CCN ) can often be better predicted in the background atmosphere (Chuang et al., 2000;Dusek et al., 2003;VanReken et al., 2003;Snider et al., 2003;Rissler et al., 2004;Gasparini et al., 2006;Stroud et al., 2007;Bougiatioti et al., 2009).
The largest errors are associated with urban emissions (Sotiropoulou et al., 2007).This is likely due to the organics component of aerosol particles, which have the largest uncertainty and are not fully understood.Biomass burning aerosols and secondary organics formed from the oxidation of common biogenic emissions are often more difficult to activate (Mircea et al., 2005;VanReken et al., 2005;Lee et al., 2006;Varutbangkul et al., 2006;Clarke et al., 2007;Rose et al., 2010;Engelhart et al., 2012;Paramonov et al., 2013;Lathem et al., 2013;Mei et al., 2013b;Zhang et al., 2014).Particles with aged/oxidized secondary organic components (e.g., organic acids) have been shown to be more hygroscopic (Raymond and Pandis, 2002;Hartz et al., 2006;Bougiatioti et al., 2011) but still much less hygroscopic than inorganic species.The sensitivity of estimated N CCN to organics has been examined in a number of recent studies (Wang et al., 2008;Reutter et al., 2009;Ervens et al., 2010;Kammermann et al., 2010;Ward et al., 2010;Zhang et al., 2012;Mei et al., 2013a).It is widely known that the estimated N CCN is sensitive to changes in organics due to the latter's complex components.The amounts and hygroscopicity parameter of organics (κ org ) vary substantially and lead to significant biases in estimating CCN concentrations and aerosol indirect forcing (Sotiropoulou et al., 2007;Hings et al., 2008;Liu and Wang, 2010).Therefore, field investigations regarding CCN activity and organics impacts, especially in heavily polluted regions, are pivotal to better parameterize CCN in climate models.
Northern China is a fast developing and densely populated region of China, where aerosol loading is high (Li et al., 2007(Li et al., , 2011)), the particle composition is complex, and severe haze pollution episodes are common (Guo et al., 2014).In recent years, CCN measurements have been collected during field campaigns carried out in the region (Wiedensohler et al., 2009;Gunthe et al., 2011;Yue et al., 2011;Deng et al., 2011Deng et al., , 2013;;Zhang et al., 2014).These studies have presented different perspectives on the influence of particle size and composition on CCN activity.For example, Deng et al. (2013) evaluated various schemes for CCN parameterization and recommended that the particle number size distribution (PSD) together with inferred mean size-resolved activation ratios can be used to estimate CCN number concentrations without considering the impact of particle composition.However, Zhang et al. (2014) demonstrated that the 30-40 % uncertainties in N CCN are mainly associated with changes in particle composition.None of the above-mentioned studies have investigated the impact of organics on estimating N CCN in Northern China.Zhang et al. (2012) noted a more significant influence of organics on CCN activity but without concerning the influences of particles oxidation or aging on CCN activity; in addition, the campaign average mass fraction of organics in their study was < 20 %.
The aim of this paper is to examine the sensitivity of CCN activity to aerosol physicochemical properties (especially aerosols containing large amounts of organics, as well as the oxidation level) and also to see how much uncertainty is incurred by applying the CCN efficiency spectra measured at one site to another site in a heavily polluted region.The instrumentation and data used in the study are described in Sect. 2. The method for calculating the hygroscopicity parameter (κ chem ) is introduced in Sect.3. The applicability of the CCN activation spectrum obtained at Xinzhou to the Xianghe site was also investigated in this section, with the aim of further examining the sensitivity of N CCN to aerosol type.Conclusions from the study are given in Sect. 5.

Measurements and data
An intensive observation period field campaign similar to the Aerosol-CCN-Cloud Closure Experiment (AC 3 Exp) (Zhang et al., 2014), called the Atmosphere, Aerosol, Cloud, and CCN (A 2 C 2 ) experiment, was conducted from 22 July to 26 August of 2014 at Xinzhou (38.24 • N, 112.43 • E; 1500 m above sea level), a city with a population of 0.51 million in Northern China.The site is located about 360 km southwest of the metropolitan Beijing area and about 10 km south of the local town center.The site is surrounded by agricultural land (e.g., corn) with little local pollution plums from motor vehicles and industrial activities.Sitting between two mountains (Taihang Mountain to the east and Lüliang Mountain to the west), the site also experiences air masses from Xinzhou to the north and from Taiyuan to the south, the cap-ital of Shanxi Province.Air masses from the northeast and southwest dominate over the site during summer.Depending on the wind direction, measurements at the Xinzhou site can detect air parcels of urban, rural, or mixed origins, including both fresh biogenic emissions around the site and aged aerosols from advection.

Instruments and measurements
During the field campaign, a scanning mobility particle sizer (SMPS), combined with a Droplet Measurement Technologies cloud condensation nuclei counter (DMT-CCNc; Lance et al., 2006), was used for size-resolved CCN measurements as well as PSD measurements.The measured aerosol PSD is within the size range of 14-600 nm.Aerosol chemical composition was measured simultaneously by an Aerodyne aerosol chemical speciation monitor (ACSM; Sun et al., 2012).
The aerosol inlet for the size distribution measurements was equipped with a TSI Environmental Sampling System (model 3031200), which consists of a sharp-cut PM 1 cyclone and a bundled nafion dryer.The size-resolved CCN efficiency spectra were measured by coupling the DMT-CCN C used with the SMPS (Rose et al., 2008).In this step, the particles are rapidly dried with RH < 30 % upon entering the differential mobility analyzer (DMA).Thus, size selection is effectively performed under dry conditions.The nafion dryer and the sheath air inside the DMA are sufficient to remove residual water associated with the ambient particles.Relative deviations in particle diameter should be < 1 %.The sample flow exiting the DMA was split into two parts: 0.3 lpm for the CPC and 0.5 lpm for the CCN counter (CCN C ).The DMA, controlled by TSI-AIM software, scanned one size distribution every 5 min.The CCN C was operated at a total flow rate of 0.5 lpm with a sheath-to-aerosol flow ratio of 10.The inlet RH for CCNc was < 30 %.During the field campaign, the mean sample temperature and pressure measured by CCN C sensors was (24.3 ± 1.4) • C and (898.4 ± 11.7) hPa.The supersaturations levels of CCN C were calibrated with ammonium sulfate before and after the field campaign, following the procedures outlined in Rose et al. (2008).During each CCN measurement cycle, calibrated effective supersaturations were set at 0.075, 0.13, 0.17, 0.39, and 0.75 %.The overall relative error (1σ ) for the supersaturation levels was estimated to be < 3.5 %.The completion of a full measurement cycle took 50 min (10 min for each supersaturation level).
The measurement of non-refractory submicron aerosol species including organics, sulfate, nitrate, ammonium, and chloride were made with an ACSM.During the field campaign, ambient aerosols were drawn inside through a 1/2 inch (outer diameter, the inner diameter is 0.38 inch) stainless steel tube at a flow rate of ∼ 3 L min −1 , of which ∼ 84 cc min −1 was sub-sampled into the ACSM.An URG cyclone (model URG-2000-30ED) was also positioned in front of the sampling inlet to remove coarse particles with a cut-off size of 2.5 µm.Before sampling into the ACSM, aerosol particles were dried using a silica gel desiccant.The residence time in the sampling tube was ∼ 5 s.The ACSM was operated at a time resolution of ∼ 15 min with a scan rate of mass spectrometer at 500 ms amu −1 from m/z 10 to 150.Regarding the calibration of the ACSM, mono-dispersed, size-selected 300 nm ammonium nitrate particles within a range of concentrations were sampled into both the ACSM and a condensation particle counter (CPC).The ionization efficiency (IE) was then determined by comparing the response factors of the ACSM to the mass calculated with known particle size and number concentrations from the CPC.More detailed descriptions of the operation and calibration of the ACSM are given in Sun et al. (2012) and Ng et al. (2011).The campaign averaged mass concentration of PM 1 is 31.6 µg m −3 .
In addition to the ACSM, the black carbon (BC) in PM 2.5 was simultaneously measured at a time resolution of 5 min by a seven-wavelength aethalometer (model AE31, Magee Scientific Corporation).The campaign averaged mass concentration of BC is ∼ 2.5 µg m −3 .During the experiment, the campaign area was generally hot and dry, with an average temperature of 21.6 • C and an average ambient RH of 69.5 %.

Data
The raw CCN data were first filtered according to instrument recorded parameters (e.g., temperature and flow).For example, if the relative difference between the actual and preset sample flows was larger than 4 %, the data were flagged as invalid.The data were also excluded if the "temperature stability" was flagged as "0".Here, the "temperature stability" refers to the T 1 , T 2 , and T 3 in the cloud chamber of the CCNc, which is set to obtain the target supersaturations.If the average differences between preset T 1 , T 2 , and T 3 and the measured values are larger than 0.4 C, the "temperature stability" is flagged as "0".Thus, the data are invalid and will be removed.A multiple charge correction and transfer function (Deng et al., 2011) is applied to each PSD as well as to the CCN efficiency spectrum.Size-resolved CCN and PSD data, measured with a DMT-CCNc and a SMPS (with a particle size range of 10-700 nm) on 7-21 July 2013 at Xianghe (Zhang et al., 2014), are used in this study for comparisons with CCN activity at the Xinzhou site.Aerosol mass concentrations were processed using the ACSM standard data analysis software (version 1.5.3.0).Detailed procedures for the data analysis have been described by Ng et al. (2011) and Sun et al. (2012).The size-resolved CCN activation ratio (size-resolved AR) is defined as the dN CCN / dlogD p divided by the dN CN / dlogD p .These values were measured by SMPS/DMT-CCNc with particle size selection in the DMA.The bulk activation ratio (bulk AR) is defined as the total CCN concentration divided by the total CN concentration.The total CCN and CN number concentrations are integrated by the measured CCN and CN size distribution, respectively, over the whole size range.

Derivation of κ chem
In this study, we calculate κ chem based on bulk chemical composition observations made during the field campaign.The method is very similar to that used by Zhang et al. (2014).As proposed by Petters and Kreidenweis (2007), κ chem can be predicted using a simple mixing rule based on chemical volume fractions for a given internal mixture: where κ i and ε i are the hygroscopicity parameter and volume fraction, respectively, for the individual (dry) components in the mixture and i is the number of components in the mixture.
Measurements from the ACSM in Xinzhou show that the composition of submicron particles was dominated by organics, followed by sulfate, ammonium, and nitrate.The contribution of chloride was negligible (volume fraction of about < 2 %).The analysis of the anion and cation balance suggests that anionic species (NO − 3 , SO 2− 4 ) were essentially neutralized by NH + 4 over the relevant size range.For refractory species, BC represented a negligible fraction of the total submicron aerosol volume (< 3 %).Sea salt and dust are usually coarse mode particles with particle sizes > 1 µm (Whitby, 1978).The contribution of such types of aerosols is thus expected to be negligible for sizes < 1 µm.Therefore, the submicron particles measured by the ACSM mainly consisted of organics, (NH 4 ) 2 SO 4 , and NH 4 NO 3 .The particle hygroscopicity is thus the volume average of the three participating species: (2) Here, the values of κ for (NH 4 ) 2 SO 4 and NH 4 NO 3 are 0.61 and 0.67, respectively.The following linear function derived by Mei et al. (2013a) was used to estimate κ org in this study: κ org = 2.10 × f 44 − 0.11, where f 44 is the fraction of m/z 44 in total organics.The mean value of κ org during the field campaign is 0.115 ± 0.019.

CCN efficiency spectra
During the field campaign at the Xinzhou site, ∼ 790 sizeresolved CCN efficiency spectra at five supersaturation levels ranging from 0.075 to 0.76 % were measured.Figure 1 shows campaign averaged spectra of the measured CCN efficiency at Xinzhou for supersaturation levels of 0.075, 0.13, 0.17, 0.39, and 0.76 %.The observed averaged CCN efficiency spectra during Xianghe campaign in summer 2013 are also shown.Note that the maximum activation fraction (MAF) for Xianghe site showed in Fig. 1 is slight lower than that  we plotted in Zhang et al. (2014).Because some data points with MAF value > 1 were also included previously, larger mean MAF resulted.However, in this paper the data points with MAF > 1.0 were forced to 1 when D p > 300 nm, which could be completely activated at even lower supersaturations, but the MAF would never be larger than 1.0.In Fig. 1, the right panels show the mass concentration fraction of particle chemical compositions at Xinzhou (top panel) and Xianghe (bottom panel) during their respective observation periods.Significant differences in size-resolved CCN efficiency spectra at the two sites are seen.Aerosol particles at Xinzhou activate more efficiently (higher values of AR) at a given particle diameter (D p ) for the same supersaturation level.In other words, a larger D p was required to reach the same activation efficiency at Xianghe.This suggests that aerosol properties at each site differ.
The slope of AR with respect to diameters near D p when AR = 50 % (defined here as the cut-off diameter, D cut ) provides information about the heterogeneity of the composition for size-resolved particles.For an ideal case when all CCN-active particles have the same composition and size, a steep change in AR from 0 to 1 would be observed as D p reached D cut .A gradual increase in size-resolved AR with D p suggests that aerosol particles have different hygroscopicities.The steeper slopes of AR around D cut observed at Xinzhou suggest that the particle composition was less heterogeneous with more hygroscopicity than particles at the Xianghe site.This can be partially explained by the magnitudes of the mean κ chem at the two sites (0.42 at Xinzhou and 0.38 at Xianghe).Also, the f 44 is greater at Xinzhou than at Xianghe.The m/z 44 signal is mostly due to acids (Takegawa et al., 2007;Duplissy et al., 2011) or acid-derived species, such as esters.f 44 is closely related to the organic oxidation level (Aiken et al., 2008).Oxidized/aged acids are generally more hygroscopic and easily activated.Moreover, the primary inorganic particles at the Xinzhou site are sulfates, with a mass fraction that is twice greater than that measured at Xianghe.Therefore, particles at the Xinzhou site consist of more hygroscopic sulfate-dominant inorganics and aged/oxidized secondary organics and can thus be more efficiently activated at a given D p , as shown in Fig. 1.

Air mass influences on CCN activity: a case study
Because air mass back trajectories combined with ambient air measurements can be used for analyzing large-scale air pollutant transport and source identification at a receptor site (Stohl, 1996;Rousseau et al., 2004), in this study we calculated 5-day (120 h) back trajectories using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Draxler and Hess, 1998) with National Centers for Environmental Prediction (NCEP) reanalysis data.TrajStat software (Wang et al., 2009) has been used to calculate trajectories.The arrival height of the trajectories at the Xinzhou site was at the surface.
In Case 1, air trajectories (red line in Fig. 3) originated from the southwest and passed through northern Shaanxi Province and northwestern Shanxi Province, then rounded back to the site from the north/northeast.So, aerosols in this case are closely associated with air parcels north/northeast of the site.The trajectories were very short, suggesting that the air flow was slow during the observational period.Under these circumstances, aerosol loading would be largely impacted by local sources around the site.A high mass fraction of organics (> 60 %) with low f 44 (∼ 10 %) and κ chem (< 0.3) values was measured during the observational period.Furthermore, the PSD showed one peak mode with D p = 56 nm and a high N CN (∼ 1.7 × 10 4 cm −3 ), but low mass concentration of PM 1 (28.36 µg m −3 ).This suggests that particles may be composed of freshly emitted primary aerosols (the biogenic emissions from the plants and trees around the site).This type of aerosol is usually less hygroscopic with a single peak mode primarily composed of fine particles (Whitby, 1978;Hussein et al., 2005).These aerosols cannot activate efficiently.The MAF shown in the top right panel of Fig. 2 is less than 0.6 at all supersaturation levels for particles with D p > 300 nm, indicating that the particles should be largely externally mixed aerosols.
In Case 2 (blue line in Fig. 3), air parcels moved rapidly from the west to the site.The site should then be influenced by the large-scale transport of air masses.For this case, aerosols contain a small amount of organics (< 30 %) but have high f 44 (∼ 14 %) and κ chem values (∼ 0.5).The PSD showed a double peak mode with an N CN of Figure 5.The sensitivity of N CCN to both organics volume fraction (x org ) and oxidation level (using f 44 , the fraction of m/z 44 in aerosol organic material) of organics at supersaturation levels of (a) 0.075, (b) 0.13, (c) 0.17, and (d) 0.76 % for cases when x org = 35 % (blue circles), 52 % (green circles), and 66 % (red circles).The size-resolved CCN data were sorted when the x org > 60 %, 50 % < x org < 60 %, and x org < 40 %, respectively, to do the sensitivity examination.Linear best-fit lines through each group of points are shown.Slopes and R 2 values are given in parentheses.
∼ 1.3 × 10 4 cm −3 and a relatively high mass concentration of PM 1 (81.45 µg m −3 ).The double peak mode suggests that aerosols in this case are a mixture of aerosols from local sources and from other regions (Whitby, 1978;Dal Maso et al., 2007).Because aerosols are aged and oxidized during long-distance transport, these particles are usually composed of secondary organic and inorganic components with more hygroscopicity (Weber et al., 1999;Verver et al., 2000).These aerosols can activate efficiently.The MAF is close to 1 and the slopes of AR around D cut are steep at all supersaturation levels (middle right panel of Fig. 2).This CCN efficiency spectrum is similar to the ideal spectrum of pure ammonium sulfate.
In Case 3 (green line in Fig. 3), air parcels traveled from the northwest to the site.Air masses arriving at the site in this case had passed over densely populated regions with more heavy pollution.A gradual increase in size-resolved AR with D p is seen (bottom right panel of Fig. 2).This is attributed to the diversity in aerosol hygroscopicity because of the complex nature of the chemical composition of aerosol particles.

Correlation of N CN and N CCN
Figure 4 shows N CN as a function of N CCN for different supersaturation levels at the Xinzhou and Xianghe sites.They showed high or moderate correlations at high supersaturation levels (e.g., R 2 = 0.57 at Xinzhou and R 2 = 0.85 at Xianghe at a supersaturation level of ∼ 0.8 %) but quite poor correlations at low supersaturation levels.Although Andreae (2009) proposed using the relationship of CCN and CN, or even aerosol optical depth, to parameterize CCN in models, it would lead to large uncertainties especially when the supersaturations are low.It was noticed that there was an apparent Figure 6.Slopes of the linear fit of estimated and observed N CCN dependence on volume fraction of organics (x org ) at f 44 < 11 and f 44 > 15 % for different supersaturation levels.Mean values of the hygroscopic parameter κ chem at f 44 < 11 % when x org > 60 %, 50 % < x org < 60 %, 40 % < x org < 50 %, and x org < 40 % are 0.27, 0.34, 0.40, and 0.46, respectively, while atf 44 > 15 % the value increased to 0.36, 0.42, 0.46, and 0.50, respectively.higher degree of correlation at Xianghe site for each supersaturation than that derived at Xinzhou site.In light of the similar regimes from which the data are taken and the same instruments by which they have been collected, the discrepancy between Xianghe and Xinzhou should be caused largely by the spatial variations of aerosols types.These variations are primarily attributed to variations in aerosol particle size, i.e., the shape of the PSD as well as particle composition.As presented by Zhang et al. (2014), the relationship between bulk activation ratios and N CN was complex under polluted conditions and was heavily dependent on the physicochemical properties of atmospheric aerosols.

Impact of x org on N CCN
Precise quantification of N CCN is crucial for understanding aerosol indirect effects and characterizing these effects in models.A CCN closure study is useful to examine the controlling physical and chemical factors and to help verify experimental results.N CCN is usually derived from measured aerosol properties, such as PSD and composition or hygroscopicity based on the Köhler theory.Achieving such closure under heavily polluted conditions is more challenging, especially due to the complex effects of organics on CCN activity.In this section, we examine the sensitivity of N CCN to both volume fraction of organics (x org ) and oxidation or aging of organics based on measurement at Xinzhou site.During the observed period, aerosols at the Xinzhou site were dominated by organics, with 12, 23, 39, and 25 % of the data points corresponding to x org > 60 %, 50 % < x org < 60 %, 40 % < x org < 50, and x org < 40 %, respectively.For the purpose of examining the sensitivity of estimated N CCN to x org and oxidation/aging level, we sorted the size-resolved CCN data when x org > 60 %, 50 % < x org < 60 %, 40 % < x org < 50, and x org < 40 %.Furthermore, for each level of x org we tested the impacts on N CCN estimation both from the most-oxidized (with f 44 of higher than 15 %) and least-oxidized (those primary organic aerosols with f 44 of lower than 11 %) organic particles.For example, the size-resolved CCN data points during the period when x org > 60 and also f 44 > 15 % were averaged to generate the averaged CCN efficiency spectra at the five supersaturations.Then we used the produced averaged CCN efficiency spectra to estimate N CCN .
Estimated CCN size distributions at the five supersaturations were firstly calculated by multiplying the averaged CCN efficiency spectrum (by using the averaged CCN efficiency spectra, the aerosol particles were assumed with uniform chemical composition without considering the effects of the temporal variations of the activation curves on CCN activity) with the actually measured PSD.Then, we integrated the estimated CCN size distribution over the whole size range to generate estimated N CCN .The measured CCN size distributions are integrated to produce the observed N CCN .
Observed and estimated N CCN at four supersaturation levels (0.075, 0.13, 0.17, and 0.76 %) were showed in Fig. 5.The data points presented more disperse and weaker correlations at lower supersaturations.The sensitivity of N CCN to volume fraction of organics increased with increasing x org .This is especially true for the case of these primary organic particles with f 44 < 11 %: the slopes obtained from a linear fit of estimated and measured N CCN in Fig. 6 decreased rapidly (almost with a decrease of ∼ 50 %) when the x org varied from < 40 to > 60 % at supersaturations of 0.075 %, while it did not exhibit a lot of reduction (merely ∼ 10 %) along with the increasing of x org at the supersaturation of 0.76 %.N CCN was estimated most accurately at higher supersaturation levels.This is likely because a large fraction of particles was already CCN-active.Also, particle composition has relatively less influence on CCN activation at high supersaturations (Twohy and Anderson, 2008).For the oxidized or aged particles with f 44 > 15 %, the slopes still follow the similar tendency with the variations of x org at low and high supersaturations but changed more smoothly to the x org attributing to the oxidized/aged organic particles being more hygroscopic.
However, the impacts of the aerosol particles oxidization level on estimating N CCN were also very significant.For example, when the particles were composed by large amounts of organics (x org > 60 %), the N CCN at the supersaturation of 0.075 and 0.13 % was underestimated by 46 and 44 %, respectively, at f 44 < 11 %, while the underestimation decreased to 32 and 23 % at the corresponding supersaturation level when f 44 > 15 %.The N CCN at ss = 0.76 was underestimated by 11 and 4 %, respectively, at f 44 < 11 and f 44 > 15 %.The results showed that the estimation of N CCN would be largely improved if the aerosol particles were aged with high oxidation level, especially when the chemical composition of the particles is dominated by organics.However, for the particles with relative low organics (x org < 40 %), the effect caused by the f 44 was quite insignificant both for high and low supersaturations.In Fig. 6, the slopes were all around 1.0 at the two cases of f 44 < 11 and f 44 > 15 %.This can be easily explained.When x org is less than 40 %, the overall hygroscopicity of the particles is dominated by inorganic species such as sulfate and nitrate, which are more hygroscopic (κ inorg usually larger than 0.6) than organic compounds (κ org usually smaller than 0.2).As a result, a larger fraction of particles can be activated.According to the simple mixing rule based on chemical volume fractions proposed by Petters and Kreidenweis (2007), the contribution from organics is quite small.If x org is greater than 60 %, organics will dominate the overall particle hygroscopicity.Particles with a large f 44 are much more hygroscopic and thus more CCN-active.Our results indicated that increasing the proportion of secondary organics would decrease the uncertainties in estimating N CCN and lead to a more accurate estimation of N CCN .

Applicability of CCN efficiency spectra
As a means of testing the applicability of the CCN activation spectra, campaign mean CCN efficiency spectra at different supersaturations observed at the Xinzhou site are used to estimate N CCN at the Xinzhou and Xianghe sites, respectively, which helps to further examine the sensitivity of N CCN to aerosol type.Data from the two sites were measured during the warm season so that the effect of temporal varia-tions in aerosols on CCN levels is reduced.Fitted campaign mean CCN efficiency spectra at the five supersaturations at Xinzhou (corresponding to spectra in Fig. 1) are multiplied by dry PSDs actually measured at the Xinzhou and Xianghe sites, respectively.This generates estimated CCN size distributions at the two sites.They are then integrated over the whole size range (14-600 and 10-700 nm at the Xinzhou and Xianghe sites, respectively) to obtain the estimated N CCN .The measured CCN size distributions at each site are integrated to produce the observed N CCN .
Figure 7 shows estimated N CCN as a function of measured N CCN for different supersaturation levels at the two sites.N CCN at Xinzhou was underestimated by 4-5 % at supersaturation levels of 0.39 and 0.76 % and was slightly overestimated (∼ 2 %) at Xianghe for the same supersaturation levels.Good agreement is seen at the 0.39 and 0.76 % supersaturation levels for data from both sites (R 2 > 0.92).N CCN at Xinzhou was underestimated by ∼ 7 % at supersaturation levels < 0.1 % (R 2 = 0.87).At Xianghe, however, N CCN was overestimated by 19-23 % at supersaturation levels < 0.1 % although the correlation between calculated and measured N CCN was good.Because size-resolved CCN efficiency spectra were applied here, excluding the impact of particle size, the influence of chemical composition on CCN activation can be investigated.The poor estimates of CCN at low supersaturation levels could be attributed to the high sensitivity of N CCN to chemical composition.Because the mass fractions of inorganics and organics measured at the two sites are similar (Fig. 1) and the hygroscopicity for inorganic components is fixed, this overestimation is attributed to the smaller proportion of aged and oxidized organic aerosols at Xianghe compared with aerosols at Xinzhou (f 44 = 17 and 11 % at Xinzhou and Xianghe, respectively).
In this study, we have investigated the impacts of particle physicochemical properties on CCN activity based on field measurements obtained from 22 July to 26 August of 2014 in the suburb of Xinzhou, China.Five-day back trajectories combined with measurements were analyzed to examine air mass influences on CCN activity.CCN efficiency was largely reduced by local air masses, and the MAF was low to < 60 %, suggesting externally mixed and heterogeneous particle composition for local emitted aerosols.The CCN activation efficiency was enhanced significantly when the site was under the influence of air transported from far away, during which aerosols could be mixed well with more hygroscopic secondary organic and inorganic components.The relationship between N CN and N CCN was generally poor.Large errors would arise if using the former to estimate the latter, especially under low supersaturation conditions.
The sensitivity of N CCN estimation to both x org and f 44 has also been examined.A strong dependence of N CCN on the both two parameters was noted.The sensitivity of N CCN to volume fraction and particles oxidization or aging level of organics increased with increase of x org .This dependence also weakens as the supersaturation level increases.When the particles were mostly composed of organics (x org > 60 %), the N CCN at the supersaturation of 0.075 and 0.13 % was underestimated by 46 and 44 %, respectively, if aerosol particles were freshly emitted with primary organics (f 44 < 11 %), while the underestimation decreased to 32 and 23 % at the corresponding supersaturations if the particles were with more hygroscopic secondary organics (f 44 > 15 %).The N CCN at the supersaturation of 0.76 % was underestimated by 11 and 4 %, respectively, at f 44 < 11 and f 44 > 15 %.However, for the particles composed of low organics (e.g., x org < 40 %), the effect caused by the f 44 was quite insignificant both at high and low supersaturations.This is due to that the overall hygroscopicity of the particles is dominated by inorganics such as sulfate and nitrate, which are more hygroscopic than organic compounds.Our results indicated that it would lead to a more accurate estimation of N CCN to increase the proportion of secondary organics, especially when the composition of the aerosols is dominated by organics.
The applicability of the CCN efficiency spectrum measured at the Xinzhou site to the Xianghe site was examined and a good agreement was found when the supersaturation level was > 0.2 %.However, N CCN at the Xianghe site was overestimated by 19-23 % when the supersaturation level was < 0.1 %.Because of the similar mass fractions of inorganics and organics measured at the two sites, we conclude that this overestimation was mainly caused by the smaller proportion of aged and oxidized organic aerosols at Xianghe compared to aerosols at Xinzhou.

Figure 1 .
Figure 1.Mean CCN efficiency spectra at the Xinzhou site (black lines with asterisks) measured from 22 July to 26 August 2014 and at the Xianghe site (red lines with circles) site measured from 7 to 21 July 2013 for different supersaturation levels.Error bars representing 1 standard deviation are shown.Right panels show particle chemical composition in terms of mass concentration fractions at Xinzhou (top panel) and Xianghe (bottom panel) during their respective observation periods.The campaign average mass concentration of PM 1 is 31.6 and 72.4 µg m −3 at Xinzhou and Xianghe, respectively.Note that the preset supersaturation levels were 0.07, 0.1, 0.2, 0.4, and 0.8 % at both sites, but effective supersaturation levels showed slightly different after calibration.

Figure 3 .
Figure 3. Five-day back trajectories for Case 1 (in red), Case 2 (in blue), and Case 3 (in green) calculated using the Hybrid Single-Particle Lagrangian Integrated Trajectory model with National Centers for Environmental Prediction reanalysis data.The arrival height of the trajectories at the Xinzhou site was at the surface.

Figure 4 .
Figure 4. Measured N CN as a function of measured N CCN for different supersaturation levels at the Xinzhou (left panel) and Xianghe (right panel) sites.The scatter plot between CCN_Obs and CN_Obs were fitted with a linear function (in colored lines) and R 2 refer to the correlations of them.

Figure 7 .
Figure 7.Estimated N CCN as a function of observed N CCN for different supersaturation levels at (a) Xinzhou and (b) Xianghe.Note that the campaign mean CCN efficiency spectra at Xinzhou are used for estimating N CCN at Xianghe.Linear best-fit lines through each group of points are shown.Slopes and R 2 values are given in parentheses.