Review of “Insights into secondary organic aerosol formation from the day- and nighttime oxidation of PAHs and furans in an oxidation flow reactor” by Mais et al.

The authors presented experimental results of secondary organic aerosols (SOA) formation from oxidation of three furans and four polycyclic aromatic hydrocarbons (PAHs) in an oxidative flow reactor (OFR). Experiments were conducted with OH and NO3 oxidation, representing day- and night-time chemistry. Results include SOA yields, effective density, absorption Angstrom exponent, and mass absorption coefficient. The authors concluded from the results that PAHs had higher SOA yields and PAH SOA had higher absorption capability compared to furans. A particularly interesting finding is that SOA density generally increased with particle size. Another important result is that NO3 oxidation of these precursors did not result in higher absorption compared to OH oxidation, because NO2 might be needed to form light-absorbing species. The experiments were well planned and conducted, and the manuscript is well written. There are, however, a few things I would suggest the authors to clarify. I therefore recommend Major Revision, with comments shown below.

Main:

1. My main concern for this study is that the precursor concentrations were super high (mg per cubic meter level), resulting in super high SOA loading (hundreds or even over 1000 of microgram per cubic meter). The authors might need to better justify this because it is quite far away from ambient conditions. There are some previous studies showing that SOA properties (including mass yield) might be loading dependent. One particularly important aspect is that higher loadings might favor partitioning of volatile (and less oxygenated) species into the particle phase, thus affecting the density measured.
2. If the authors want to compare the SOA from seven precursors, how come the level of oxidants and equivalent aging time were so much different among experiments (Tables 1 and 2). External OH reactivity might affect those conditions. But it is not very straightforward to compare SOA for different precursors as well as different aging extents.
3. The finding that SOA density increased with particle size is interesting. But it is a little bit counterintuitive. Let’s say, smaller particles are dominated by very non-volatile species that can potentially nucleate, and easily condense (partition) into the particles because of the low volatility; those species should be very oxygenated and have high density. Large particles, on the other hand, are formed from condensation of lots of less volatile species, which should be less oxygenated, thus possess lower density. The net result, from a wild guess, would be a decreasing trend of density as SOA particle size increases. While my wild guess might not be true, I would suggest the authors elaborate a bit more than what is shown in P12.

Technical:

1. P3/L63: ozone is not a radical?
2. P3/L74: if the OA components are non-absorbing, they will not be classified as BrC, right?
3. P7/L2: just “blanks”? It is not a field study.
4. P8/L205: a unit of either molecules per cubic centimeter or ppt for NO3. Otherwise, it is not straightforward to make a direct comparison.
5. P15/L318: Joo et al. (2019a)? There are a few other places that Author (Year) format should be used.
6. Did the authors have NO2 concentration data from either KinSim model or measurements to rule out the involvement of NO2 to form light-absorbing products?

This manuscript describes studies of SOA formation by PAH and furan compounds present in smoke plumes by long-term oxidation with OH and NO3 radicals in an oxidation flow reactor.  The density, absorption, and yields of the SOA formed are probed by a variety of instruments.  The authors find SOA yields and absorption to be much lower for NO3 radical reactions than OH radical reactions, and also for furan reactions than PAH reactions.  They also find that density depends on particle diameter for particles below 100 nm, an especially interesting result.  The modeling of direct photolysis of precursors, and its comparison to the OH and NO3 reactions being studied, is very helpful.  For many of the precursors studied, this work is the first one to measure SOA yields using NO3 radicals.  This work will clearly be of interest to atmospheric chemists but needs minor revision to address the comments below.

Overall, the introduction is very clear.  The introduction would be enhanced, however, by some discussion of the emission levels of PAHs and furans in biomass burning plumes to establish the environmental significance of this project.  I consulted the reference{Oros, 2001 #2951} which quantified substantial PAH emissions from pine wood burning, but furan emissions were not reported.

The idea that differences in morphology might explain the dependence of particle density on particle diameter seems implausible without some further rationalization.  While the authors are careful to use words like “possibly” and “probably” for this claim (please note that the meaning of these two words is quite different!), it seems much less likely to explain the observed aerosol densities than differences in chemical composition, since the density rises with diameter:  this is the opposite of what one might expect for particles formed by agglomeration of smaller, solid particles.  Furthermore, if the particles are liquid, a non-spherical shape is unlikely.  Some way of rationalizing how particle morphology might cause the observed density trend is needed in the discussion.  Citing other studies of size-dependent SOA properties, if the authors know of any, could also be helpful.

Th authors conclude that PAH SOA has significant UV-vis absorption, but also note that MAC does not increase when OH or NO3 exposures are increased.  How do these two claims relate to each other?  They might seem contradictory to readers without further explanation. 

Section 2.3:  The verbal description of the aerosol sampling instruments is inconsistent with Figure 1.  Specifically, the use of terms “in series” and “in parallel” in this section is very confusing.  Based on Figure 1, the only instruments that are sampling in series are the density instruments (DMA + CPMA + CPC) in the bottom row.  The instruments in the middle row are sharing a common sampling line, but they are sampling from it in parallel.

Line 201:  It seems problematic to run an OFR experiment where the precursor is completely destroyed before exiting the reactor, and the oxidant concentrations therefore rise rapidly.  Why was this done?

Equation 4:  The discussion of this equation centers on how the multiple scattering correction needed to be raised beyond the one provided by the instrument manufacturer.  While this is not a concern for aethalometry – it is standard practice – equation 4 does not contain this term, and so it is unclear from the presentation how the larger C value used in this work influences the results.

Figure 3:  It is unclear which vertical axis goes with the furan + OH oxidation size distribution data.  Is it shared with 2,5-DMF or with the other five compounds (the leftmost axis)?

Section 3.2:  This discussion omits any mention of the 10x larger numbers of aerosol particles produced by OH oxidation compared to NO3 oxidation for four of the precursors.  It makes sense that the aerosol particles from NO3 oxidation grow substantially larger, since there are fewer of them available for gas-phase product molecules to condense onto (or dissolve into).  For 2,5-DMF, where the aerosol size trend seems the opposite of the other precursors (OH oxidation produces larger particles), it can also be explained by an opposite trend in aerosol numbers (OH oxidation produced fewer particles).

Table 3:  It would be helpful to compare the densities to those of the precursor species.  Could the variability in precursor densities help to explain the variability in SOA density?

Line 365:  this claim appears to be a logical error or at least an oversimplification.  Contribution to biomass burning BrC is also determined by the level of emissions of the precursor gases – if the emissions are small, they may not contribute significantly to biomass burning BrC even if they produce highly absorbing SOA in single-precursor studies like these.  This is another reason to reference emission studies in the introduction section.  In addition, POA absorption in biomass burning plumes is significant.  The absorbance of a single-precursor oxidation SOA cannot be directly compared to the absorbance of biomass burning aerosol without taking these other issues into account. 

Line 372:  the reasoning here is confusing.  Once aerosol particles have been formed, couldn’t these experiments be largely observing heterogeneous oxidation processes, even though initially gas-phase reactions are the only thing happening?

Technical corrections

Line 350:  “quite limited at 370 nm” could be more clearly expressed as “limited to 370 nm”

Abd El Rahman El Mais et al. investigated secondary organic aerosol (SOA) formation from the oxidation of furanoids and polycyclic aromatic hydrocarbons (PAHs) using an oxidation flow reactor (OFR) with two different oxidants: OH and NO3 radical. The authors explored the size-dependent effective density of SOA, SOA yield, and SOA light absorption and found that PAH SOA shows higher SOA yield and stronger light absorption than furanoid SOA. Understanding these parameters from such less-studied volatile precursor compounds would improve our understanding of biomass burning impact on air quality and climate change. Overall, the manuscript is well-written and well-displayed. However, I have major comments that need to be addressed before publication.

1. The quantitative analysis requires a more thorough evaluation. Authors are reporting SOA yield, but the wall loss effect is evaluated based on the size range where the mode of number concentration is observed, instead of that of the volume concentration. Authors should diagnose and report if the particle loss at the size range of volume concentration mode was also little as what they have observed from the number concentration loss. Also, SOA yield is a function of organic mass concentration formed during oxidation. The comparison among precursors or with previous studies should consider the organic mass concentration comparison as well. Lastly, authors should address the reason why they studied under atmospheric irrelevant conditions (i.e., high precursor VOCs concentration, no-NOx condition, absence of pre-existing particle).
2. Precursor VOCs concentration varies depending on the experiment while the level of oxidant injection is relatively consistent. This is affecting the VOC reaction with the oxidants: a substantial fraction of 2-methlyfuran and 2,5-methylfuran reacting with O3. Such reaction conditions would affect the RO2 reaction channel (RO2+RO2, RO2+HO2, RO2+NO3), inducing each experiment to have respective SOA-forming  RO2 reaction conditions. Thus, authors need to discuss thoroughly how the SOA yield and other parameters are different among the precursors and from previous studies depending on the VOC-to-NOx ratio or VOC-to-oxidant ratio (also, the presence of seed aerosol).
3. The formation of light-absorbing compounds in SOA is generally associated with nitrogen-containing organics, such as nitro-aromatics, amines, imine, etc. However, experiments here are performed under the NOx-free conditions for OH radical reactions. It would be informative if the authors can provide a more detailed discussion on which composition (or chemical functionalities) of SOA could potentially contribute to the light absorption of SOAs and how such light-absorbing compounds are formed during the oxidation of PAHs & furanoids. In addition, MAC values indeed seem high, reaching the level of ambient biomass burning studies, but the Absorption Angstrom Exponent seems low compared to biomass burning studies. The authors should add a discussion on this as well.

Technical comments:

Line 80: Since the authors use the symbol α for the absorption angstrom exponent, the acronym "AAE" seems unnecessary.

Line 293: It would be better to comment that using the density obtained from individual experiments can also reduce potential biases for quantitative analysis.

Line 322: Please clarify the conditions: e.g., presence of NOx (or [NOx]), type (or concentration) of seed particles.

Line 349: brown carbon comparison between AE 33 & filter-extracted measurements can be relocated after describing absorption of SOA generated via NO3 radical reactions as both OH and NO3 radical experiments show good agreement with literature values.

Line 386, Table 4, etc.: please keep consistency when referring to the range throughout the manuscript. It is mixed up with various formats (e.g., 5-6 times, 5 – 6 times, or 5 - 6 times).

The authors Mais et al. describe a series of OFR experiments oxidizing PAH and furan compounds and characterize the density and optical absorption of the formed SOA. The manuscript is well-written, generally clear, and thoroughly references prior literature. The parent VOCs and questions investigated in this work are important and of general interest to the atmospheric community. I believe the manuscript will be suitable for publication after the major and minor comments below are addressed.

Major comment

I believe more discussion needs to be added on the extent to which the generated SOA is representative of SOA that’s likely to form in the atmosphere.

• The precursor concentrations used in this work are quite high compared to what’s often used in laboratory work and observed during field measurements. High precursor concentrations have been tied to increased SOA formation, an increase in measured SOA yield, and changes in aerosol composition (via AMS f44/f43 fractions) in prior OFR experiments (Kang et al., 2011). It seems possible that the SOA yields reported in this work are generally higher than in past works because of this phenomenon. Could associated SOA compositional changes (for example, increased condensation of SVOCs) bias the measured particle density or absorption properties compared to typical ambient SOA?
• The authors state that the NO2 concentration in the OFR was below their detection limit (line 133). What was the detection limit? Are such low levels typical of what’s expected in regions where NO3 radical is expected to form at night? The authors briefly reference that NO2 addition to PAHs during oxidation can form chromophores (~line 371), and I ask that the authors discuss whether the lack of NO2 during NO3 radical oxidation in the present work leads to SOA that would likely have different absorption properties compared to what would likely form from NO3 radical oxidation in an ambient environment.
• Line 276: can the authors cite any prior literature that observed size-dependent SOA composition differences during oxidation/formation within the same chemical system? Does the observed SOA density size dependence have any implications for SOA density when formed in a system with seed aerosol (such as a typical ambient environment) as opposed to the present work where particles nucleate?

Line 195: the usage of the KinSim model is appropriate, but I believe either the results need to be discussed further or the model re-run. The authors observe both OH and NO3 concentrations to sharply increase after precursor VOC is consumed and state that this leads to high estimated exposures. However, in the actual OFR experiments, the initial VOC oxidation will produce first- and later-generation products that will also be available to react with generated radicals, providing additional reaction partners while also delaying consumption of the precursor VOC. This would, presumably, influence the time-dependent estimation of radical concentrations and overall radical exposure as well as estimates of the precursor consumed.

Minor comments

Tables 1 and 2: also report precursor concentrations in ppbv to make general comparison a little more straightforward.

Line 145: please provide a sample calculation for each species in the SI.

Line 158: if the 6-methylchrysene extraction efficiency is higher than 100%, what does this mean for the extraction efficiency of the sampled precursors and calculating precursor concentration? Is the error in this extraction efficiency incorporated into the concentration error bars reported in Tables 1 and 2?

Line 169: add manufacturer details for listed instruments.

Line 192: is the OH exposure listed here the one used to calculate precursor VOC fate (Figure 2) and therefore SOA yield? Or are the KinSim results used?

Line 217: the ACSM-derived SOA masses appear to be quite high based on Figure S1. Is there an upper limit of quantification for the Q-ACSM? Did vaporizer performance change over time or was any substantial buildup of material on the vaporizer observed during experiments?

Line 221: are similar loss rates expected for larger particle sizes? Many of the experiments produce sizable portions of SOA (both by number and presumably volume) that are in particles with mobility diameter >200 nm (Figure 3).

Line 270: add x-axes to the upper panels.

Line 372: are heterogeneous oxidation reactions competitive under these OFR conditions?

Line 377: why are nitroaromatics not associated with SOA? Keyte et al. (2013) include a discussion of gas-particle partitioning but do not appear to make general conclusions on the fate of nitronaphthalenes or similar compounds.

References

Kang, E., Toohey, D. W., and Brune, W. H.: Dependence of SOA oxidation on organic aerosol mass concentration and OH exposure: Experimental PAM chamber studies, Atmos Chem Phys, 11, 1837–1852, https://doi.org/10.5194/acp-11-1837-2011, 2011.