The authors tried to develop a new method to retrieve the effective density of internally mixed BC (In-BC) based on Köhler theory combining with field observations. However, when they retrieve the effective density of BC using the new method, they assumed too much. This would lead great uncertainties in their results, particularly the mass loading of internally mixed BC. Therefore, the retrieved effective density of internally mixed BC is unbelievable. Although the authors evaluated the uncertainties in their results, the uncertainties are largely underestimated due to that some important effect factors are not considered. More details are listed as follows.
(1) In this work, the authors assumed the same effective density of externally mixed BC (Ex-BC) with different sizes. However, the effective density of externally mixed BC decreases with increasing BC mobility diameters. For Ex-BC with mobility diameter smaller than 80 nm, the effective density can be more than 1 g cm-3, but it would be lower than 0.3 g cm-3 for Ex-BC with mobility diameter larger than 300 nm. Because the authors did not consider the substantial difference in effective densities among Ex-BC with different mobility diameters, the mass size distribution of Ex-BC shown in Fig. S4 would have a large uncertainty. This would affect quantification of Ex-BC mass loading, and further lead to a large uncertainty in In-BC mass loading.
(2) The authors used the AE33 measurement to quantify the total refractory BC (Ex-BC+In-BC), which would largely overestimate the mass loading of total refractory BC. The BC concentrations measured by AE33 based on aerosol light absorption reflect an equivalent BC mass loading, because the aerosol light absorption measured by AE33 not only include the absorption of refractory BC itself, but also the light absorption enhancement caused by lensing effect of BC mixing state. When the light absorption of BC is significantly enhanced by other species coated on BC surface during atmospheric aging, the mass concentrations of atmospheric BC measured by AE33 can be more than twice the refractory BC. The overestimate of the mass loading of total refractory BC would lead to an overestimate of In-BC mass loading. This would strongly affect the retrieved effective density of In-BC.
(3) Because the authors used a narrow range of the variabilities in the effect factors to make sensitive analysis, the uncertainty (±30%) in In-BC density is underestimated. The authors highlighted that the In-BC density was sensitive to the number fraction of nearly hydrophobic POA (NH-POA), but they selected a narrow range of the number fraction of NH-POA (50-90%) to estimate the uncertainty in In-BC density (Figure 1). For source emission, the number fraction of NH-POA can lower than 50%. To take an extreme example, when all NH-POA is attached to BC particles by coagulation process during source emission, the number fraction of NH-POA can be nearly 0. When the sensitive analysis is made with number fraction of NH-POA in the range of 0-100%, the uncertainty in In-BC density will be much higher than ±30%, which is determined by a narrower range of the number fraction of NH-POA (50-90%). Similarly, they also use a narrow range of POA (0.85-1 g cm-3) and SOA (1.2 -1.4 g cm-3) densities to make sensitive analysis.
(4) The large variability in the retrieved BC density (0.14-2.1 g cm-3) based the new method developed in this work can not exclude the effect of the great uncertainties in their results. Therefore, the range of the retrieved BC density in this work is meaningless.
(5) The CCN closure study can not demonstrate that the prediction CCN number concentrations are improved by using the retrieved BC density. A good agreement between the predicted and observed CCN number concentrations (Fig.7) is due to prediction CCN number using the size-resolved κ based on hygroscopic growth factor measurement (κgf). In this work, the BC density is retrieved by assuming that κgf is equal to κ derived from aerosol chemical composition (κchem), shown in Equation (8). Compared with κchem, the measured κgf can improve prediction CCN number concentrations. When the authors made the sensitivity analysis of predicted CCN number concentrations to change of BC density with prescribed valued of 0.14 and 2.1 g cm-3 (Fig. 6), the κ was calculated based on aerosol chemical composition (i.e., κchem). For these cases, the uncertainties in the predicted CCN number concentrations are dominated by κchem. Therefore, the differences in the predicted CCN number concentrations shown in Figure 6 and Figure 7 is mainly due to using κchem and κgf, respectively.
Considering a large uncertainty in the retrieved BC density, I do not think the new method developed in this work will be used in the future studies. |