| 26 Aug 2026
Black carbon mixing-state heterogeneity influences absorption enhancement: Results from a field study across three seasons
Abstract. Black carbon (BC) is the major light-absorbing aerosol and an important contributor to radiative forcing. Atmospheric aging often coats BC with non-absorbing material, enhancing its mass absorption cross-section (MAC). However, this absorption enhancement depends not only on the bulk amount of coating material but also on how the coatings are distributed across individual BC particles. Here, we represent BC mixing-state heterogeneity using particle-resolved two-dimensional mixing-state distributions. This representation provides a more complete explanation of MAC variability than a single bulk mixing-state parameter. During summer, mixing-state heterogeneity changed little, and MAC remained nearly constant despite a substantial increase in the bulk coating-to-core mass ratio. By contrast, larger changes in mixing-state heterogeneity observed during autumn and winter coincided with MAC increases of up to 35 % and 30 %, respectively. MAC was most sensitive when initially homogeneous BC populations became more heterogeneous, whereas further changes within already heterogeneous populations produced comparatively small responses. A higher fraction of large-core BC particles (>300 nm), associated with biomass and coal combustion, further reduced MAC because of their lower absorbing efficiency. Mie calculations further showed that MAC differed by ~24 % on average under homogeneous and heterogeneous mixing-state assumptions. Our results highlight the importance of explicitly representing BC mixing-state heterogeneity for a better understanding in aerosol optical property variations and improving their simulation in models, thereby reducing uncertainties in estimates of BC radiative forcing.
Yifan Yang and coauthors (https://doi.org/10.5194/egusphere-2026-5074) present an analysis of the rBC mass absorption cross-section (MAC rBC) obtained by normalizing single particle soot photometer (SP2) data to a scattering-corrected filter-based absorption instrument (MAAP) in the field. The field site was a rural site in Germany (Melpitz) for a relatively long period, August 2021 to January 2022.
These type of measurements are rare and the manuscript is relatively high quality. Overall, this represents a valuable contribution to the literature. However, I have some comments that should be addressed before publication.
MAJOR COMMENTS
### "MAC" context:
i. The abstract does not mention the MAC wavelength nor instruments used to derive it. Please add both pieces of information. Please add wavelength to MAC in *all* figures. Please change figure axis labels from "MAC" to "MAC_rBC,637nm"
ii. Please add context in Figure 10 for the "reference MAC" at this wavelength. Consider adding line (with uncertainties) or shading for the MAC of laboratory soot.
### Data averaging period:
Line 138-140 state, "Although extrapolating a lognormal fit to the measured rBC mass size distribution can partially compensate for the detection limit (Kipling et al., 2013; Pileci et al., 2021), the low rBC concentrations at Melpitz reduce the statistical robustness of the size distribution. As a result, uncertainties in SP2 measurements persist."
This statement only applies to the hourly data the authors analyzed for MAC. Yet, the manuscript ultimately discusses seasonal data. Please simply average the SP2 data over longer periods and apply the lognormal fit correction. Even with daily or weekly fits, the authors will still have a large number of data points.
Pileci et al. showed that the fit correction may be >20% in summer at Melpitz and 2% in winter. This correction should be applied, especially as the authors already chose to apply a smaller +5% correction to the MAAP data.
### Figure 3 / Figure S2: MAC vs MRbulk
After increasing the averaging interval in Figure S2 (see above), please move it to Figure 1 in the main text, or a panel in Figure 3. Or, please change Figure 3 from means (the SMA-fitted slope is a type of mean) to box plots, to show distributions in the data. At present, neither figure adequately represents the data set. Figure S2 has too much overlap, the density of points cannot be observed. Figure 3 shows biases that are not in Figure S2 -- the spike in MAC at high MR, for example.
### Figure 4 bias
The text states that "size-resolved MR data are only partially available for rBC cores smaller than 160 nm". This is not a fair statement. Instead, the SP2 only detects rBC cores smaller than 160 nm if they are coated, so any measurement below 160 nm is biased. Please add lines to this figure showing the limits of detection of the SP2, similar to Dahlkotter et al. 2014 http://www.atmos-chem-phys.net/14/6111/2014/ Figure 7b.
### Figure 4: 2D distributions of MR to core size:
Scientifically, Figure 4 shows a very interesting result. The figure shows a subsets of the broader data set, divided into summer/autumn/winter rows, and MR_bulk columns. This is an excellent diagnostic plot, but it seems to show an issue which is not discussed in enough detail. Specifically, there are 3 groups of particles, according to the MR (mass ratio):
Group S: a cloud of particles at Dc ~ 200 nm, MR ~ 0.5 to 1, always visible.
Group TB: a cloud of particles at Dc ~ 400 nm, Mr ~ 2 to 4. more common for MR > 2 and more common in winter.
Group AS: a cloud of particles at Dc ~ 200 nm, MR ~ 10, not common in summer.
If these particles were "typical aged soot" from regional sources then there might be a continuous variation of particles across all Mr, for a fixed Dc. (Of course, this would not be true for a long-range transport source.) One might hypothesize that AS is "aged soot" and S is "fresh soot". Since S is always present, perhaps it is from nearby on-road diesel vehicles.
Group TB: more common in winter and similar to the "tar BrC" described by Corbin and Gysel-Beer 2019 (www.atmos-chem-phys.net/19/15673/2019/). This hypothesis is strengthened by the fact that the large Dc is never observed without coatings! It is fully consistent with the solid fuel combustion discussion in Section 3.6.
### Figure 6, Mie plot
The Mie plot shows "wiggles" in Figure 6a. Consider plotting lognormal distributions with reasonable GSDs instead, to smooth this out and avoid distracting the reader.
### MAC noise:
Related to the previous point, the authors attribute scatter in the MAC (Figure S2) to noise in the SP2 or MAC. Please demonstrate this point by plotting MAC vs rBC, and MAC vs Babs. If the noise does not increase at lower rBC or Babs, then the scatter may be due to physical variability.
Context for this comment: MAC_rBC of up to 30 m2/g are not intrinsically physically implausible. Mason et al. 2018 also reported MAC_rBC of up to 20 m2/g at 660 nm. Adler et al. 2019 reported a similar result. Both studies were influenced by tarballs, which do not generate strong signals in the SP2 (Corbin and Gysel-Beer, 2019), so bias the rBC low and the MAC (Babs/rBC) high. Nevertheless, there are also low values of MAC_rBC in Figure S2 that indicate clear noise issues.
References
Adler, G., N. L. Wagner, K. D. Lamb, K. M. Manfred, J. P. Schwarz, A. Franchin, A. M. Middlebrook, R. A. Washenfelder, C. C. Womack, R. J. Yokelson, D. M. Murphy. 2019. Evidence in biomass burning smoke for a light-absorbing aerosol with properties intermediate between brown and black carbon. Aerosol Science and Technology 53:976-989. doi: 10.1080/02786826.2019.1617832.
Mason, B., N. L. Wagner, G. Adler, E. Andrews, C. A. Brock, T. D. Gordon, D. A. Lack, A. E. Perring, M. S. Richardson, J. P. Schwarz. 2018. An intercomparison of aerosol absorption measurements conducted during the seac4rs campaign. Aerosol Science and Technology 52:1012-1027. doi:10.1080/02786826.2018.1500012
Corbin, J. C. and M. Gysel-Beer. 2019. Detection of tar brown carbon with a single particle soot photometer (sp2). Atmospheric Chemistry and Physics 19:15673-15690. doi: 10.5194/acp-19-15673-2019.
MINOR COMMENTS
Abstract -- Please mention the techniques used, SP2 and MAAP.
Abstract -- "because of their lower absorbing efficiency". Where is this proven in the paper? Does it mean "Mie calculations showed a lower absorbing efficiency for these particles"? Please reword to be more closely related to the underlying evidence.
Figure S3: There appears to be an error, there is virtually no information in the figure. Figure S2 says "high noise" and Figure S3 says "no noise". What exactly is being plotted here? Only fit slopes and not raw data? Please include raw data as a heat map or bar chart.
120, the statement "a wavelength correction factor of 1.05 was applied to adjust [the MAAP wavelength from 660 nm to 637 nm]" does not really explain what the adjustment is, and the reader is obliged to consult the reference. The reference (Mueller et al, 2011) explains that the is factor 1.05 is retained because of an intercomparison between MAAP and a reference photoacoustic instrument, where the slope between the two instruments was 1.05. This reviewer will comment that a 5% uncertainty is unsurprising, and that it seems unnecessary to propagate this correction.
139, "low rBC concentrations at Melpitz reduce the statistical robustness of the size distribution" as a reason to avoid performing lognormal fits is very surprising. Could the authors not integrate over, say, 1 day?
143, MAC_rBC of up to 30 m2/g are not intrinsically physically implausible. Mason et al. 2018 also reported MAC_rBC of up to 20 m2/g at 660 nm. Adler et al. 2019 reported a similar result. Corbin and Gysel-Beer 2019 showed that the SP2 does not respond strongly to "tarballs", which can explain the high MAC_rBC, as discussed in detail by Mason et al.
Can the authors please show a plot of MAC_rBC vs rBC, to support their hypothesis that the unusual MAC_rBC values are due only to noise?
165, which PyMieScatt version was used?