Overview
Table 2 summarizes the observed CCN activity during the campaign. Overall,
the average NCCN at 0.1 %, 0.2 %, 0.4 %, and 0.7 % SS were about 3100,
5100, 6500, and 7900 cm-3, respectively. The average ARs at the above four SSs were 0.26, 0.41, 0.53, and 0.64, respectively. The
average D50 values at the above four SSs were 156, 107, 78, and 58 nm,
respectively. The NCCN at 0.7 % SS was, respectively, lower than those
of the previous measurements (10 731 cm-3 at 0.67 % SS) in July 2006
in Guangzhou (Rose et al., 2010) but much higher than those measured (2085 cm-3
at 0.7 % SS) in May 2011 in Hong Kong (Meng et al., 2014), while
the NCCN/NCN,tot
was lower than that from the previous
measurements (at 0.67 % SS) in Guangzhou (0.59; Rose et al., 2010) and
similar to that from the measurements (at 0.7 % SS) in Hong Kong (0.64;
Meng et al., 2014). The D50 was larger than that in the previous
measurements in Guangzhou (49 nm) and in Hong Kong (47 nm), due to the lower
particle hygroscopicity in Guangzhou. The differences of the κCCN values between the two measurements (0.21 in this winter campaign
vs. 0.28 during the summer season in Guangzhou both at 0.7 % SS) suggested
that the particles in the summer were in general more hygroscopic and hence
were more readily activated than those in the winter, implying different
chemical composition of the particles between the two distinct seasons.
Mochida et al. (2010) measured the size-resolved CCN activity in Cape Hedo,
a remote marine site rarely affected by anthropogenic emissions. The results
showed that the D50 at 0.1 % SS in Cape Hedo was about 130 nm, much
smaller than that in Guangzhou, leading to higher hygroscopicity of
atmospheric particles in Cape Hedo than that in Guangzhou.
The mass fraction of the bulk NR-PM1 composition (a) and the
mass fraction of the size-resolved composition (b).
![]()
Summary of the measured CCN concentration, activation ratio, and
D50 at the four supersaturations during the campaign.
SS
0.1 %
0.2 %
0.4 %
0.7 %
NCCN (no. cm-3)
Max
15 165
19 989
25 964
26 208
Min
258
361
408
502
Mean ± SD
3103±1913
5095±2972
6524±3783
7913±4234
NCCN/NCN,tot
Max
0.68
0.75
0.89
0.94
Min
0.06
0.10
0.19
0.28
Mean ± SD
0.26±0.10
0.41±0.14
0.53±0.15
0.64±0.13
D50 (nm)
Max
268.90
194.04
145.28
97.17
Min
112.47
76.60
43.50
24.21
Mean ± SD
156.02±19.48
106.66±16.99
77.96±14.86
58.45±10.68
Figure 2 shows the average mass fraction of NR-PM1 bulk composition and
size-resolved (64–731 nm) composition. The organics were dominant in the bulk
NR-PM1 (50 %), followed by sulfate (26 %) and nitrate (12 %)
(Fig. 2a). The mass fraction of the organics decreased with the size (Fig. 2b)
from 73 % at 64 nm to 42 % at 397 nm. The mass fraction of organics
at 397 nm was close to that of NR-PM1 bulk, due to the fact that the
PM1 mass is dominated by particles in a diameter range of
200–500 nm (Tan et al., 2016). In comparison, the dominant
NR-PM1 species observed in Hong Kong were sulfate (51.0 %) and
organics (28.2 %) (Lee et al., 2013), which is significantly different from our
measurements, due probably to different origins of the dominant air masses
between the two seasons. The measurement site in Guangzhou was impacted
predominantly by the air mass from the north, where straw burning contributes to
a high mass fraction of organic matter (Cao et al., 2008).
The median and interquartile PNSD, κ obtained from HTDMA,
AMS, and CCN measurements during the campaign. The κ was pointed
against their corresponding median D50 (SMCA and AMS) or measured
diameter (HTDMA). Dots represent the median value and the bars
represent the interquartile range. The blue, red, and green colors represent
κCCN, κAMS, and κHTDMA, respectively.
![]()
Figure 3 shows the κ values based, respectively, on the CCN (κCCN), AMS (κAMS), and HTDMA (κHTDMA)
measurements, along with the measured PNSD (10–400 nm) during the campaign. The shadow area represents the
interquartile range of the PNSD. A distinct peak at around 90 nm was
observed from the PNSD (Fig. 3). The κAMS was calculated based
on the size-resolved chemical composition, assuming the particles are
internally mixed. At 0.7 % SS, the D50 was about 58 nm. Hence, no
κAMS was reported at this SS since we only measured particle
composition above 63 nm using the AMS in this study. The κ values
were shown in the interquartile range, with the largest variation from the
CCN measurements (Fig. 3). As shown in Fig. 3, the difference between
κCCN and κHTDMA is statistically insignificant
at all employed diameters, while the one between κAMS and
κCCN became statistically significant at larger sizes of the
particles. The p value between the κAMS and the κccn was close to 0, indicating that the correlation between them is
significant. The κHTDMA values were lower than those of the
corresponding κCCN at most of the particle sizes, consistent
with the previous observation (Pajunoja et al., 2015). This was probably due
to the fact that particles contain a certain fraction of low-solubility
composition, such as SOA, contributing
differently to hygroscopic growth and CCN activation. The available AMS data
(Fig. 3) show that the κAMS values were lower than the
corresponding κCCN and κHTDMA values at all size
ranges, and the differences become larger with increasing particle sizes.
This was probably due to underestimated hygroscopicity in the organic
composition when using the AMS data, since we assumed a κ value of
0.1 for all organics at all particle sizes. The hygroscopicity increased
with particle diameters due to aerosol aging which increased the hygroscopic
organic contents. Previous studies showed that the κorg values
of larger particles are lower than those for smaller particles (Lance et
al., 2013; Zhao et al., 2015), and hygroscopicity of organics is often found
to be related to its chemical composition (f44 or O/C) in both field and
laboratory studies (Chang et al., 2010; Massoli et al., 2010; Lambe et al.,
2011; Mei et al., 2013, and other references therein). We showed that the
f44 increased with the particle size from the AMS data (Fig. S1 in the Supplement). Note that
the f44 for particle diameters smaller than 100 nm was discarded due to the
data quality. The results indicate that the degree of oxidation of the
organics was higher for larger size particles and the hygroscopicity for
larger particles is higher (Chang et al., 2010). The measured κmean values fall in a range of 0.22–0.30 for the particle sizes of
40–200 nm measured by the HTDMA in this study. The other aerosol hygroscopicity
measurement in PRD (Jiang et al., 2016) reported the κmean
values ranging from 0.18 to 0.22 in the 2012 winter season and 0.17 to 0.21 in
the 2013 summer season, suggesting an increase of the aerosol hygroscopicity,
which might result from an increasing mass fraction of nitrate in recent
years (Zhang et al., 2015; Itahashi et al., 2018), although the fraction
decrease of less hygroscopic compounds is not as significant as the fraction
increase of the nitrate. However, the fraction of the non-hydroscopic
compounds (i.e., EC) decreases more rapidly than the organic compounds.
The size-resolved activation ratios measured by the SMCA at four
different supersaturations. Note that the curves were fitted according to
the SMCA measurements.
![]()
Figure 4 shows the ARs measured by SMCA at four
supersaturation ratios (0.1 %, 0.2 %, 0.4 %, 0.7 %) for particles below 300 nm.
The activation curves obtained in this study were segmented into three
sections: a steady rise at low ARs, a middle sharp increase, and a plateau
at almost 100 % AR. We defined the steepness as the rate at which the AR
increased with the particle sizes. Figure 4 shows the steepness increased
with the SS, indicating that the curves became steeper with the SS and a
larger variation of the D50 was expected. In addition, the CCN activity
was more sensitive to particle diameters at higher SS, which can be seen
from partial derivative of κcritical by ∂D50 (Eq. 15):
∂κcritical∂D50=-4A39D504(lnSc)2.
For a certain SS, the κcritical value became more sensitive to
D50 with decrease of the D50. Meanwhile, a high SS usually led to
a low D50. Therefore, the AR would vary with Dp more readily at higher
SS and the curve would become steeper. A higher SS allowed a smaller
particle to be activated and the activation curve became steeper, and vice
versa for a lower SS.
The steepness of activation curve was also associated with the heterogeneity
of aerosol chemical composition; that is, a steeper activation curve meant
that aerosol particles had higher similarity in hygroscopicity. A bimodal
distribution (peaks at about 1–1.1 and 1.5–1.7 Gf) of the Gf-PDFs was
observed along the Gf coordinate at all five sizes of the particles
measured by the HTDMA in this study (Fig. 5), corresponding, respectively, to the
less- and more-hygroscopic modes. Larger size particles contain higher
fractions of more-hygroscopic inorganic matter which leads to the increase
of Gf of the more-hygroscopic mode. The less-hygroscopic mode usually represents
externally mixed black carbon or fresh organics. Thus, the less-hygroscopic
mode for larger size particles more likely represents the external mixing
non-hygroscopic black carbon with a Gf value of 0.8–1.1, indicating that the
Gf of the less-hygroscopic mode decreased and that of the more-hygroscopic mode
increased with the particle diameter (Fig. 5). Since less-hygroscopic
particles were usually associated with externally mixed black carbon (BC) or
fresh organics and more-hygroscopic particles usually represent the
inorganic matter or BC coated with inorganic matter (internally mixed).
The decrease of peak area of less-hygroscopic mode and the increase of
more-hygroscopic mode indicate that the number fraction of less-hygroscopic
particles decreased while the more-hygroscopic particles fraction increased.
Thus, the particles became more internally mixed. Here, a parameter σ
is introduced to illustrate the deviation of Gf-PDF (Gysel et al.,
2009):
Gfmean=∫0∞Gfc(Gf)dGfσ=(∫0∞(Gf-Gfmean)2cGfdGf)12,
where the c(Gf) denotes Gf-PDF and Gfmean denotes number weighted
mean Gf. The σ was employed as a measure of the spread of Gf-PDF
which represents the heterogeneity of aerosol chemical composition (Sjogren
et al., 2008; Liu et al., 2011). A small σ indicated that the
heterogeneity of aerosol chemical composition was low and aerosol particles
had higher similarity in hygroscopicity. The parameter C (in Eq. 5) determined the
shape of the activation curve which was segmented into steep and smooth
parts. A small C value means a steep activation curve, and vice versa. Here,
an activation curve was assumed to be steep when the C values were lower
than the lower quartile of all the C values, while the activation curve
was considered to be smooth when the C values were higher than the upper
quartile of all the C values. Table 3 summarizes the σ values of
Gf-PDF for the corresponding steep and smooth activation curve at the four
supersaturations. In general, the σ increased with the diameter,
indicating that larger particles had higher heterogeneity of aerosol
chemical composition. Meanwhile, the σ values for smooth curve were
generally higher than the σ values for steep curve. The results
implied that the shapes of activation curves were related to the
heterogeneity of aerosol chemical composition. Cai et al. (2017) compared
the Gf-PDF between Guangzhou and Cape Hedo and the results showed that only
more-hygroscopic particles were observed in Cape Hedo, indicating that
atmospheric particles tend to be more internally mixed in Cape Hedo than in
Guangzhou. Meanwhile, atmospheric particles in Guangzhou have a higher
degree of external mixing affected by more anthropogenic emissions, which in
turn affect the CCN activity.
The average of σ values of Gf-PDF measured by the HTDMA for
five diameters for the steep and smooth activation ratio at four
supersaturations.
SS (%)
0.1
0.2
0.4
0.7
Dp (nm)
Steep
Smooth
Steep
Smooth
Steep
Smooth
Steep
Smooth
40
0.13
0.17
0.12
0.17
0.11
0.19
0.11
0.19
80
0.16
0.20
0.14
0.20
0.14
0.21
0.14
0.20
110
0.17
0.21
0.15
0.21
0.15
0.21
0.16
0.20
150
0.19
0.22
0.17
0.23
0.17
0.22
0.18
0.21
200
0.20
0.23
0.19
0.24
0.19
0.24
0.19
0.23
The Gf-PDF as a function of Gf measured by the HTDMA for the five
diameters.
![]()
Impact of organics on CCN activity
Figure 6 shows the relationship between the D50 obtained from the SMCA
measurements and the size-resolved mass fractions of organics (forg) at
three supersaturation ratios (0.1 %, 0.2 %, and 0.4 % SS). In general,
the D50 increased with forg at the three SSs, with slopes of 127,
66, and 21, and fitting coefficients (R2) of 0.47, 0.31, and
0.1 at 0.1 %, 0.2 %, and 0.4 % SS, respectively. The particles usually became less
hygroscopic with increase of the organic fractions (forg), which then
required larger particles to be activated. At lower SS, better correlations
were found between the forg and the D50 because the D50 was
more sensitive to hygroscopicity (the activation ratios increase slower
with particle sizes at lower SS as shown in Fig. 4). It was hence more
obvious at lower SS that the modification of the particle hygroscopicity
caused by the change of the mass fraction of organic matter could greatly
modify the D50 which might further affect the CCN activity. At higher
SS, according to Eq. (6), particles were more easily activated as CCN and the
change of particle hygroscopicity would not significantly alter the CCN
activity.
The relationship between size-resolved mass fraction of organics and
D50 at three supersaturations. The red, blue, and green dots and line
represent 0.1 %, 0.2 %, and 0.4 % SSs, respectively.
![]()
Organics can affect the CCN activity via two opposite ways: they can
decrease the CCN activity by increasing the less hygroscopic organic
fraction of the particles and thus increase the D50 as shown in Fig. 6;
they can also increase the CCN activity by decreasing the surface tension of
the particles. The latter effect has been demonstrated experimentally. For
example, an increase of CCN activity was observed when organics were added
to sulfate ammonium (Engelhart et al., 2008). In this study, we investigated
the impacts of organics on CCN activity through adjusting the value of
surface tension until the calculated AR values based on HTDMA measurements
agreed with those obtained from SMCA measurements (measured AR). The
calculated AR values were systematically lower than the corresponding
measured ones because here the surface tension of bulk pure water
(0.072 N m-1) was assumed when calculating the AR from the HTDMA measurements
(Fig. 7). Note that the surface tension is not the only factor that
determines the AR and other factors such as the sparingly soluble compounds
in the particles may contribute to the AR, although they are currently not
understood. Previous studies found that the hygroscopicity of the particles
measured by the HTDMA could be lower than that measured by the CCNc (Chan
et al., 2008; Pajunoja et al., 2015; Petters et al., 2009; Hansen et al.,
2015; Hong et al., 2014) which might be attributed to low soluble compounds
in the particles. The deviation of the calculated AR from the measured AR is
probably dependent on the degree of dissolution of particles and the
oxidative state of the organics in the particles.
The predicted activation ratio based on HTDMA measurement vs. the
measured activation ratio at 0.1 %, 0.2 %, 0.4 %, and 0.7 % SS for
40, 80, 110, 150, and 200 nm particles.
![]()
The surface tension of a nanoparticle was substantially different from that
of its bulk solution due to the curvature effect (Ahn et al., 1975; Bogdan,
1997). The effects of size and composition on the surface tension were
currently not well understood. Here, we proposed an approach to evaluate the
impact of organics on the surface tension (σs/a) based on the
fraction change of the calculated AR to the measured AR. We defined this
fraction change (δAR) as a function of surface tension,
diameter, and supersaturation:
δARσs/a,Dp,SS=ARmDp,SS-ARcσs/a,Dp,SSARm(Dp,SS)×100%,
where ARmDp,SS is the measured AR for a certain
diameter and SS, and ARc(σs/a,Dp,SS) is the calculated AR for a
certain diameter, SS, and σs/a. We excluded particles at the
size of 200 nm because they were easily activated even at 0.1 % SS and the
δAR was expected to be independent of σs/a.
Here, the σs/a value varied between 0.03 and 0.072 N m-1
(surface tension of pure water). Figure 8 shows the δAR as a
function of σs/a for the four particle diameters (40, 80, 110,
150 nm). The δAR decreased with increase of the σs/a for all given particle sizes, changing more rapidly for smaller
particles (i.e., from 200 % to -100 % for 40 nm) than bigger particles
(i.e., from 20 % to -10 % for 150 nm). The R2 values between measured AR
and predicted AR for a certain diameter and four supersaturations at σs/a=0.072 N m-1 were 0.35, 0.93, 0.95, and 0.91, respectively.
The δAR values reached zero when the σs/a was
set to be about 0.054 N m-1 for 40, 80, and 110 nm particles, and 0.062 N m-1
for 150 nm particles, with a R2 of 0.88, 0.94, 0.94, and 0.88,
respectively. As a compromise, here we adopt a σs/a value of
0.058 N m-1 (denoted as σs/a∗) to predict AR.
This σs/a∗ value significantly increased the R2
compared to that based on the pure water assumption (0.072 N m-1) for 40 nm
particles, while it was reasonably well for other sizes of particles (80,
110, 150 nm). The AR was then recalculated using the σs/a∗ value and the prediction was significantly improved
(Fig. 9). We calculated the p value between the measured and predicted AR
and the results showed that the p value is close to 0, indicating a
significant correlation between the two variables. The results demonstrated
that partitioning of organics into aerosol particles would decrease their
surface tension. Therefore, the pure water assumption for surface tension
would lead to high uncertainties when it was applied to predict the activation
ratios of the aerosol particles at a certain size. Note that we did not
consider the effects of individual organics due to the limited data from the
chemical composition measurements. How chemical composition affects the
surface tension of the particles is yet to be investigated.
The relative deviation between predicted AR and measured AR at
different assumed σs/a; the color code represents R2
between calculated AR and measured AR.
![]()
The predicted activation ratio using the new surface tension assumption
(σs/a∗) based on HTDMA measurement vs. the measured
activation ratio at 0.1 %, 0.2 %, 0.4 %, and 0.7 % SS for 40, 80,
110, 150, and 200 nm particles.
![]()
The <inline-formula><mml:math id="M347" display="inline"><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi mathvariant="normal">CCN</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> prediction
The <inline-formula><mml:math id="M348" display="inline"><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi mathvariant="normal">CCN</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> prediction based on the HTDMA
measurements
In this study, we used several approaches to predict the NCCN based on
the HTDMA measurements, from either the activation curve or the D50.
Table 4 summarizes the methods that were used to predict the NCCN,
along with the slope and R2 between the predicted and the measured
values. The mixing state of the aerosol particles is an important parameter
in determining the NCCN. The prediction of NCCN using the activation
curve means the NCCN was calculated based on Eq. (7). Meanwhile, the
prediction of NCCN using the D50 means that the NCCN was
calculated based on Eq. (8) and the D50 was determined from fitting the
size-resolved activation ratio by Eq. (5). The activation curve represented
actual mixing state, while the D50 approach assumed that all particles
were internally mixed. Scheme 5 in Table 4 was the method based on the
activation curve with the new σs/a∗ (0.058 N m-1). Equations (7)
and (8) were, respectively, used to calculate the NCCN
following schemes 1, 2, 5, and the rest of the schemes. Scheme 5 (real-time
activation curve using σs/a∗) provided the best
NCCN predicted value (closest to the measured one), followed by scheme
3 (real-time D50) > scheme 4 (average D50) > scheme 1
(real-time activation curve) > scheme 2 (average
activation curve). The R2 values for all the approaches were in general
high (around 0.93). The CCN prediction based on scheme 2 led to the largest
underestimation over the measured values. In general, the real-time data
(schemes 1 and 3) gave better predicted NCCN than the corresponding
average data (schemes 2 and 4).
The overview of different schemes used in the NCCN prediction
based on HTDMA measurements.
Scheme
Method
Slope
R2
1
Real-time activation curve
0.8275
0.93
2
Average activation curve
0.8183
0.93
3
Real-time D50
0.8869
0.93
4
Average D50
0.8738
0.93
5
Real-time activation curve using σs/a∗
0.9377
0.93
Figure 10 shows the correlation between the measured NCCN and the
predicted NCCN from schemes 1–5 at the four SSs. For schemes 1–4, the
predicted NCCN values were found to be the largest deviation from the
corresponding measured ones at 0.1 % SS among all the approaches, probably
due to the pure water assumption for surface tension (σs/a=0.072 N m-1). Meanwhile, because the CCN activity was sensitive to
hygroscopicity of the particles at low SS, the uncertainties of
hygroscopicity data would lead to large errors in the prediction of CCN. As
discussed in the previous section, the D50 was more sensitive to the
σs/a at lower supersaturations, leading to a large deviation of
the NCCN from the measured value. The best agreement between the
calculated AR and the measured AR was seen using scheme 5, as the slopes at
the four SSs were close to 1 (Fig. 10q–t).
The relationship between measured NCCN and predicted NCCN
based on schemes 1, 2, 3, 4, and 5. The black lines represent 1:1 lines.
![]()
The <inline-formula><mml:math id="M389" display="inline"><mml:mrow><mml:msub><mml:mi>N</mml:mi><mml:mi mathvariant="normal">CCN</mml:mi></mml:msub></mml:mrow></mml:math></inline-formula> prediction based on AMS measurements
We proposed five approaches based on HTDMA measurements to predict the
NCCN in the previous section. Alternatively, we can calculate the
NCCN based on AMS measurements. Here, we proposed four methods based on
either size-resolved chemical composition or bulk PM1 chemical
composition from the AMS measurements (Table 5). Here, we assumed that the
particles were internally mixed and the median κAMS
obtained from bulk composition was 0.28, higher than those from
size-resolved composition (0.24–0.26 in Fig. 3), probably due to a higher
mass fraction of inorganic matter in bulk NR-PM1 (Fig. 2). We excluded
the size-resolved data at 0.7 % SS due to their poor quality. Note that
the impact on the calculated κAMS values and the
predicted NCCN was minor using the κ value (0.53) of
ammonium sulfate from Petters and Kreidenweis (2007). For example, the
κAMS values slightly increased from 0.27 to 0.28 at
0.1 % SS; the slopes for schemes 6, 8, and 9 in Table 5 slightly increased
from 0.9859 to 0.9898, 0.9721 to 0.9834, and 0.9742 to 0.9973, respectively,
while the one for scheme 7 did not change. Figure 11 shows the correlation
between the measured and predicted NCCN from schemes 6–9. The NCCN
was underpredicted at 0.1 % SS and was overpredicted at 0.7 % SS. We
proposed three potential factors that might impact NCCN prediction
based on AMS measurements. (1) The assumed κorg values were
probably underestimated for particles larger than 100 nm, leading to the
underestimated NCCN at low SS. As shown in Fig. 3, the predicted
κ shows a larger deviation from the measured value for a larger
particle. The D50 values were more sensitive to particle hygroscopicity
at lower SS, as discussed in the previous section. (2) Pure water was
assumed for surface tension. As we have shown in the previous section,
the σs/a values for the aerosol particles were found to be much
smaller than the σs/a for pure water (0.072 N m-1). As a
result, the pure water assumption for surface tension led to the NCCN
underestimation. In addition, again the D50 was more sensitive to
σs/a at the low SS. (3) Black carbon (BC) particles were not
included and the particles were assumed to be internally mixed. The BC particles were known to be
non-hygroscopic and had a low CCN activity. During the campaign period, the
average BC concentration was about 5.91 µg m-3,
which accounts for 7 % in PM2.5. The assumption of no BC particles would lead to the
overestimation of NCCN. Here, the particles were assumed to be
internally mixed in the AMS measurements. This would lead to an
overestimation of the NCCN when the ambient particles tend to be
externally mixed (Wang et al., 2010; Sánchez Gácita et al., 2017).
However, the internal mixing assumption seems to play a minor role in
predicting the NCCN at 0.1 % SS since the particles at about
140–180 nm tend to be internally mixed, as shown in Fig. 5. In this case, the κorg assumption and the pure water assumption played more important
roles than the mixing state assumption at low SS (i.e., 0.1 % SS). Figure 11
shows significant underestimation of NCCN at 0.1 % SS (Fig. 11a,
e, i, m), while more or less comparable to the measured NCCN at
higher SS (i.e., 0.2 %, 0.4 %, 0.7 %). The difference between the
κAMS and κCCN became smaller and the corresponding
D50 value decreased with the increase of the SS so that the impacts of
the κorg assumption and the pure water assumption became minor
with the increase of the SS. Instead, the internal mixing state assumption
would play a more important role in the prediction (Meng et al., 2014). As
shown in Fig. 5, the peak height and area of the less-hygroscopic mode
became larger for the smaller size particles (i.e., 40 nm particles),
implying that small particles were likely to be externally mixed; that is,
the non-hygroscopic or less hygroscopic species including BC and insoluble organics were
less likely coated with inorganic salts. Hence, the internal mixing
assumption could lead to an overestimated NCCN.
The overview of methods used in the NCCN prediction based on
AMS measurements.
Scheme
Method
Slope
R2
6
Real-time bulk composition
0.9859
0.91
7
Average bulk composition
1.0108
0.91
8
Real-time size-resolved composition
0.9721
0.87
9
Average size-resolved composition
0.9742
0.86
The relationship between measured NCCN and predicted NCCN
based on schemes 6, 7, 8, and 9. The black lines represent 1:1 lines.
![]()
As discussed above, the two important parameters (κorg and
σs/a) had significant impacts on the NCCN prediction. We
denoted κorg∗ and σs/a∗ as
important representations, respectively, for hygroscopicity and surface
tension contributed from organics. We also pointed out that the κorg was dependent on the particle size, and hence here we further
assumed the κorg∗ values to be 0.15 and 0.1,
respectively, for particles larger and smaller than 100 nm. Note that we
previously assumed the κorg to be 0.1 for all particle sizes.
Here, we gave an example of the improvements at 0.1 % SS when the κorg and σs/a values were, respectively, replaced with the
κorg∗ and σs/a∗ ones (Fig. 12).
The κAMS value calculated at 0.1 % SS based on κorg∗ was 0.288, very close to the corresponding
κCCN value (0.30), indicating that an improvement was made
for the NCCN prediction when including the κorg∗
value. The NCCN prediction could be greatly improved when including both
σs/a∗ and κorg∗ in the
calculation. For example, the underestimate of NCCN decreased from
44 % (Fig. 11a) to 4 % (Fig. 12b). In addition, we also investigated the
effects of the σs/a values in a range of 0.054 to 0.062 N m-1
as discussed in Sect. 3.2. The shadow area in Fig. 12b
represents the variation of linear fit between the measured and predicted
NCCN. An under- and overestimated value of 16 % ( slope of 0.84) and
8 % (slope of 1.08) was obtained for the predicted NCCN to the
measured NCCN using a σs/a value of 0.054 and
0.062 N m-1, respectively, indicating that the predicted NCCN agreed
reasonably with the measured ones when the σs/a values between
0.054 and 0.062 N m-1 were used in this study. We conclude that the
predicted NCCN can agree better with the measured one when including
both σs/a∗ and κorg∗ in the
calculation at low SS.
The relationship between measured NCCN and predicted NCCN
at SS 0.1 % based on size-resolved chemical composition using κorg∗ (a) and κorg∗ and σs/a∗ (b). The shadow area represents the variation of the
linear fit using the σs/a values from 0.054 to 0.062 N m-1.
![]()