This manuscript describes new secondary organic aerosol (SOA) yields from decamethylcyclopentasiloxane (D5) oxidation by OH at a range of OH exposures/concentrations and compares the new results to previous publications. The authors find that the SOA yield varies significantly based on the experimental conditions and argue that differences in OH concentration is the cause of this variation. They stress the importance of understanding the effects of OH concentration versus OH exposure in conducting SOA yield experiments. D5 is a volatile chemical product (VCP), a class of compounds that is of emerging interest and importance for air quality. The yields reported here at lower OH concentrations are much smaller than previously published yields and thus previous estimates of SOA from D5 in urban areas may be overestimated. I believe these results merit publication in this journal; however, I think revisions and consideration of additional points are necessary before I can recommend publication.

Main comments

• For me, the main take-home of this work is that the SOA yield for D5 + OH may be substantially lower than previous reports. I think this is an important point and it signifies D5 is a chemical system requiring further investigation. Indeed, there is emerging evidence suggesting that the chemistry of siloxanes differs from more traditional VOCs (e.g., Ren and da Silva, 2020; Fu et al., 2020). As to the cause of the SOA yield varying based on OH exposure/concentration, I think better support as to why/how it varies with OH concentration is necessary if the authors wish to make this a main point of the paper. Personally, I think the extended discussion on the role of OH concentration versus exposure detracts from the take-home point of the paper. I think the new yield measurements merit publication even without a detailed investigation of why the yield differs between experiments. In my opinion, either the focus on the role of OH concentration versus exposure should be deemphasized or a more complete consideration of the chemistry should be included.
• I agree with the point that it is important to understand how the experimental conditions impact SOA yields so that yields can be appropriately extrapolated to the ambient atmosphere. However, I think that the conclusions regarding the experimental issues of high OH versus concentration (lines 266-269) are oversimplified. The current presentation points to high OH concentration as a fatal flaw for this chemical system. However, the OH exposure versus concentration issue is a concern about the relative role of processes that scale with OH and are atmospherically relevant versus those that do not scale with OH (e.g., peroxy radical (RO2) isomerization) and/or that might scale with OH but are not typically important for most systems (e.g. RO2 + OH). Consideration and analysis of the chemistry within the experiment are important for determining if high OH experiments will always be difficult for this chemical system (for instance if photolysis or isomerization are important) or if high OH experiments are possible with careful planning (e.g. by running under conditions to limit RO2 + OH; for instance see Peng and Jimenez (2020) and references therein). This distinction between high OH experiments being possible but requiring careful experimental consideration versus difficult and unlikely to provide useful information is an important consideration for how the community plans, performs, interprets, and extrapolates chamber and flow reactor experiments. I ask the authors to consider these points when making a recommendation about future experimental conditions.
• In my opinion, the thinking of how OH exposure versus concentration affects this chemical system is poorly articulated. The reasoning outlined in lines 165-167 is confusing to me, at least in part because it seems like first-, rather than second-, generation product is a more appropriate term to use. Please clarify the mechanisms that may be impacting this system. To do this, I believe that further information on the radical chemistry in the chamber and CPOT experiments. For instance, one possibility at high OH is that RO2 +OH becomes important. Does the estimated HO2/OH ratio for the experiments support this idea? Would it be possible to adjust the HO2/OH ratio in the experiments to avoid this condition while still maintaining high OH concentrations? Overall, information on how the RO2 lifetime and fate (isomerization or reaction with NO, HO2, RO2, OH) varies across the experimental conditions is necessary for the reader to judge if high OH is the fatal flaw it is made out to be. I recognize that investigating this chemistry for D5 is difficult since little is known about the gas-phase chemistry of D5, however, educated guesses are possible and necessary for an exposure versus concentration argument.
• The idea that concentration and exposure are not necessarily interchangeable is well-known from the heterogeneous chemistry literature (e.g., Liu et al., 2011; Renbaum and Smith, 2011; McNeill et al., 2008). For SOA, Lambe et al., (2015) found only small differences between chambers and flow reactors for many systems. Additionally, Peng and Jimenez (2020 and references therein) have investigated this using models and provided recommendations for operation. At least the work comparing SOA between chambers and flow reactors should be discussed.
• I am not convinced that later generation products can be disregarded (lines 235-236). The contribution of later-generation products to the SOA yield is discarded based on a lack of correlation between the yield and the OH exposure normalized to D5 reacted and the finding that for experiments 16-17 where all the D5 does not react has a higher yield than experiments 18-19 where all the D5 does react. However, experiments 16 & 17 have a higher absolute concentration of D5 compared to experiments 18 & 19. Could the results be influenced by RO2 + RO2 reactions leading to lower volatility products and hence more aerosol in 16 & 17 compared to 18 & 19? Although RO2 + RO2 is typically slow, it can be fast for some RO2. Additionally it has been shown that dimers and products containing more than 5 Si are important in D5 generated aerosol (Wu and Johnston, 2016, 2017). While it is unclear if the dimers are formed via gas- or condensed-phase chemistry, those results do suggest that there may be a D5 concentration dependence. Overall, I think a more detailed characterization of the RO2 chemistry and D5 concentration dependence is necessary before higher generation oxidation products are deemed to not matter (lines 235-236). While D5 RO2 + RO2 chemistry generating dimers may be unlikely to occur in the ambient atmosphere (thus reinforcing the point that there needs to be careful consideration of how experimental conditions relate to the atmosphere), this is different from the OH exposures versus concentration argument.

Minor Points

• Lines 114-120: I think the discussion on 5% of the oxidation products being lost to the walls is somewhat misleading. While I agree that first-generation oxidation products such as the ester will have minimal wall-loss, they will also contribute minimally to aerosol if absorptive partitioning dominates. Later generation products may have higher wall-loss.

Technical

• Figures 1-3: Colors and shapes are hard to distinguish, particularly in the legend and for the red squares and blue circles in Fig. 1a. Perhaps removing the black outline (or making it thinner) and/or making the points bigger would help.
• Please include the RH for the experiments in Table 1.

References

Fu, Z., Xie, H.-B., Elm, J., Guo, X., Fu, Z., and Chen, J.: Formation of Low-Volatile Products and Unexpected High Formaldehyde Yield from the Atmospheric Oxidation of Methylsiloxanes, Environ. Sci. Technol., 54, 7136–7145, https://doi.org/10.1021/acs.est.0c01090, 2020.

Lambe, A. T., Chhabra, P. S., Onasch, T. B., Brune, W. H., Hunter, J. F., Kroll, J. H., Cummings, M. J., Brogan, J. F., Parmar, Y., Worsnop, D. R., Kolb, C. E., and Davidovits, P.: Effect of oxidant concentration, exposure time, and seed particles on secondary organic aerosol chemical composition and yield, Atmospheric Chem. Phys., 15, 3063–3075, https://doi.org/10.5194/acp-15-3063-2015, 2015.

Liu, C.-L., Smith, J. D., Che, D. L., Ahmed, M., Leone, S. R., and Wilson, K. R.: The direct observation of secondary radical chain chemistry in the heterogeneous reaction of chlorine atoms with submicron squalane droplets, Phys. Chem. Chem. Phys., 13, 8993–9007, https://doi.org/10.1039/C1CP20236G, 2011.

McNeill, V. F., Yatavelli, R. L. N., Thornton, J. A., Stipe, C. B., and Landgrebe, O.: Heterogeneous OH oxidation of palmitic acid in single component and internally mixed aerosol particles: vaporization and the role of particle phase, Atmospheric Chem. Phys., 8, 5465–5476, https://doi.org/10.5194/acp-8-5465-2008, 2008.

Peng, Z. and Jimenez, J. L.: Radical chemistry in oxidation flow reactors for atmospheric chemistry research, Chem. Soc. Rev., 49, 2570–2616, https://doi.org/10.1039/C9CS00766K, 2020.

Ren, Z. and da Silva, G.: Auto-Oxidation of a Volatile Silicon Compound: A Theoretical Study of the Atmospheric Chemistry of Tetramethylsilane, J. Phys. Chem. A, 124, 6544–6551, https://doi.org/10.1021/acs.jpca.0c02922, 2020.

Renbaum, L. H. and Smith, G. D.: Artifacts in measuring aerosol uptake kinetics: the roles of time, concentration and adsorption, Atmospheric Chem. Phys., 11, 6881–6893, https://doi.org/10.5194/acp-11-6881-2011, 2011.

Wu, Y. and Johnston, M. V.: Molecular Characterization of Secondary Aerosol from Oxidation of Cyclic Methylsiloxanes, J. Am. Soc. Mass Spectrom., 27, 402–409, https://doi.org/10.1007/s13361-015-1300-1, 2016.

Wu, Y. and Johnston, M. V.: Aerosol Formation from OH Oxidation of the Volatile Cyclic Methyl Siloxane (cVMS) Decamethylcyclopentasiloxane, Environ. Sci. Technol., 51, 4445–4451, https://doi.org/10.1021/acs.est.7b00655, 2017.

This paper describes chamber and flow-reactor measurements of secondary organic aerosol (SOA) yields from the OH oxidation of decamethylcyclopentasiloxane (D5), an important organic compound in indoor environments. There are two major results from this work: (1) chamber yields are much lower than previous measurements, whereas flow-reactor yields are quite high, and (2) this difference is attributed to the use of atmospherically unrepresentative OH levels in the flow-reactor experiments. Both of these are important topics, and certainly will be of interest to the readership of ACP. However the manuscript focuses mostly on the 2^nd result; I think the paper would be stronger if it spent more time discussing the 1^st one (the yields themselves) as well. This could include providing SOA yield parameterizations for use in modeling, and/or describing implications for indoor/urban air quality. In addition, I had a number of questions and comments about the 2^nd conclusion (OH concentration and exposure); these are described below.

1) Much of the manuscript is focused on examining the role of OH concentrations vs. OH exposure; however this discussion (or the possible underlying mechanism) was not always clear:

- 166-169: I had a hard time following this explanation. This is in part because the language is very general, mentioning “intermediates” and “fragments” but not providing concrete examples. Are these intermediates/products radicals or molecules? And why is the focus here on only second-generation (but not first-generation) products? Some more discussion of specific pathways (e.g., oxidation of products, RO₂ self-reactions, oligomerization reactions, etc.) would be helpful.

- Similarly, a major conclusion of the work (line 265) is that “It is the OH concentration, and not the OH exposure, that affects the SOA yield”; but little explanation for why this might be the case. Can the authors suggest some potential mechanisms? (The CIMS data might be of use here.)

- My initial interpretation of the data (especially Fig 2) was that the later-generation condensable products are what lead to SOA formation; since these may not be formed until most D5 reacted, yields will be higher when deltaD5 / initialD5 approaches 1. The implication of this would be that yields are highly dependent on extent of reaction. Lines 226-236 argues against this interpretation, but I don’t follow the argument. Differences in fractions of D5 reacted between experiments 16-17 and 18-19 are described, but these differences are quite small (97-98% vs 100%). A much bigger difference is the absolute amount of D5 reacted (~240 ppb vs ~80 ppb) – this could lead to large differences in aerosol loadings, which in turn could affect yields due to differences in semivolatile partitioning. The oxidation chemistry might be different as well. Given such large differences, it doesn’t seem straightforward to make any conclusions about SOA-formation mechanisms just from comparisons of yields or fraction of D5 reacted.

(Also, the statement “the fraction of D5 reacted correlates with the [OH]” (line 235) seems self-evident, given the role of OH concentration in reacting away D5, as shown in Equation 1.)

- There’s a good deal of literature on the possible nonlinearities associated with OH exposure (that [OH] cannot always be ramped up to accurately simulate long atmospheric timescales), (examples include Renbaum and Smith, Atmos. Chem. Phys. 2011, 11, 6881–6893, Liu et al. 2011, PCCP, 13, 8993-9007, and Palm et al. 2016, Atmos. Chem. Phys. 16, 2943–2970), but these aren’t discussed in this paper. It’s probably worth discussing these treatment in the context of the present results.

- The authors argue for a “the necessity of OH concentrations similar to the ambient environment when extrapolating SOA yield data to the outdoor atmosphere.” What are the implications of this work for other laboratory studies? The last several years has seen a huge increase in the use of oxidation flow reactors for studying SOA chemistry (e.g., Chem. Soc. Rev., 2020, 49, 2570-2616). Is the argument then that these results are flawed, and should not be used in models?

2) A central point of the paper is the relative importance of OH exposure and OH concentration; I have a number of comments related to the calculation of these quantities:

- 144-146: I’m unclear on how OH exposure was estimated using SO₂. At first I assumed SO₂ was added to the reactor (as is sometimes done in OFR experiments), but instead from the text it appears that these SO₂ levels were taken from a different set of experiments from a different laboratory (Janechek et al. 2019); this should be described in greater detail. If that is indeed the case, I think it’s unlikely that the experiments were “identical” – they presumably used a different reactor (with different flows, concentrations, SA/V ratios for wall loss, etc.). And if SO₂ is high enough (relative to the D5) it can affect (decrease) the OH exposure. Therefore such an estimate may well lead to large errors in estimated OH exposure, and probably shouldn’t be used in a quantitative way (at least without large error bars).

- 149-150: I don’t think these are possible explanations, since Equations 1 and 2 should hold regardless of the sources and sinks of OH.

- For calculating OH exposure from Equations 1 and 2, were these single point calculations (estimating exposure from one time point, t), or were they fit to a curve? This latter approach (which is possible at least for the chamber experiments) would give a more precise value.

- 142: the value of k_OH+D5 used was from one study, but as noted in line 126 other studies have found values than span a factor of ~2 (Atkinson et al: 1.55e-12 cm3 molec-1 s-1, Safron et al 2.6e-12 cm3 molec-1 s-1, etc). This in itself adds a factor of ~2 uncertainty to the OH exposure estimate.

- Based on the above comments, I believe there’s substantial uncertainty in the estimated OH exposures in this study. These should be included as error bars in Fig 1 and 2, and incorporated throughout the text in the discussion of results.

Minor comments:

29: calculate -> determine?

76: why were the concentrations used to calibrate the FID so much higher than those used in the experiment? What sort of error in SOA yields might this lead to?

Table 1: it would be helpful to also give the RH and fraction of D5 reacted for each experiment.

121-122: While I agree that wall-loss corrections would not change the conclusions of the paper substantially, a steady-state reactor can still have vapor-phase losses. This can result from extremely low-volatility species, which can sorb to walls essentially irreversibly; also see Krechmer et al. Environ. Sci. Technol. 2020, 54, 12890–12897. (In addition, the text on line 202 implies that vapor wall loss in the flow reactor could indeed lead to a suppression of SOA yields.)

204-205: The comparison should probably be in terms of the dependent variable (SOA yield) not the independent one ([OH]).

219-221: I’m unclear on how to interpret the “OH exposure divided by reacted D5” metric; this is a nonstandard metric so some more explanation would be useful. The explanation on the following sentence (“the number of OH radicals available per reacted D5 molecule”) isn’t quite right since that should be unitless, whereas exposure/deltaD5 has units of time

251-260: The case is made that the yield differences are not from differences in [NO] or RH. Might the oxidation conditions (O₃ vs H₂O₂) play a role?

294: I had to look up the “ICARUS” database (no URL was given); it appears not to have been updated since 2019 (with most data being from 2016-2017) and so apparently is no longer active.

This is an important contribution to the literature on the oxidation of volatile methyl siloxanes, to the literature on aerosol formation from volatile chemical products, and on the interpretation of experiments on SOA formation done at high concentration. It should be published in ACP after addressing the following points. 

Major points.

1.    The abstract should be revisited, particularly by the senior authors with comprehensive experience across all the aspects of the study (modeling implications, flow tube experiments, chamber experiments). I think the major contributions of the work is that (a) both chamber and flow tube experiments were conducted, (b) a wide range of OH concentrations were looked at experimentally, and (c) these were analyzed in a comprehensive manner with two other datasets from literature.  None of these aspects really come out in the abstract clearly. 

 

2.    The abstract is confusing chemical transport models and emission inventories: “SOA yields used in emission and particulate matter inventories.”

3.    The results start very abruptly and the text at line 156 is not doing anything other than referring the reader to the table.  The reader has to do all the work themselves.  Could the authors help out and provide the important context? 

4.    Line 166 is important. I quote the manuscript here: “If a chemical process occurs in which the reaction of D5 and OH forms an intermediate or a second-generation product that then either reacts with OH or fragments, then the competition between the two outcomes is moderated by the relative time required for self-reaction or reaction with OH.”  

(a)    I think this could be written more clearly.

(b)    Line 166 says fragments but 167 self-reaction. Are these same process? 

(c)    the OFR literature shows that the community is aware of this issue.  The concern, however, is usually on the other side – that high OH will lead to fragmentation reactions at the gas-particle interface, or that high OH will lead to fragmentation reactions in the gas phase on similar timescales as condensation of low volatility gas phase products … not that high OH will lead to particle formation while NOT reacting with OH will lead to fragmentation and/or high volatility compounds.  This is an interesting route and a valuable contribution to bring it up.

(to clarify, a and b are reviewer comments where I am looking for a response and/or change to the MS.  c is just a reviewer comment)

5.    Comment: I think the paper has a critical point on the possible role of fragmentation reactions (or reactions that simply lead to high volatility products) that are zero order in OH and in reaction partners that are correlated with OH. Thus, if first generation oxidation products have such pathways (e.g. rearrangement or autooxidation, for example, leading to high volatility products or to conformations that prevent subsequent formation of low volatility products), and these do not involve OH (but the formation of low volatility products does involve multigenerational oxidation with OH), then there will be a strong OH concentration dependence on yield.  For example, if the first generation product of OH attack rearranges to a volatile species with a characteristic time of n seconds, and to form low volatility multi-generational products, OH needs to make a 2nd attack prior to those n seconds, then yields will be dependent on OH and independent of OH exposure. However, there are other mechanisms that could lead to similar dependences, such as concentration effects in the gas-phase, the particle phase, and the gas-particle interface leading to higher order effects not seen at low concentrations.  

Reviewer request: acknowledge that there are more mechanism options that could lead to the same functional dependence on concentration of OH.  

6.    Regarding fates of the first generation oxidation products that are zero order in OH and in reaction partners that are correlated with OH, but that lead to high volatility products or to conformations that prevent subsequent formation of low volatility products (I believe this is what is required for the strong OH dependence) … is there evidence, perhaps from the gas-phase mass spec, of such reactions? If so, please report hypothetical reactions and compounds.

7.    Comment: The paper implies that reactions (or SOA formation scenarios) can be divided into a class where OH exposure matters, and another where OH concentration is dominant.  Not sure this is a useful designation.  It is more complicated than that, and it is the fate of the species in the reactor (fixed or flow) that matters, not strictly the OH concentration. I believe the atmospheric chemistry communities that use both types of tools are increasingly aware of the strength, weaknesses, and factors to look out for in data interpretation and extension to models (Lambe et al. 2012, Lambe et al. 2015, Palm et al. 2016, Peng et al. 2016).  

Reviewer request: reconsider the classification, or provide support for the binary classification. 

8.    Abstract line 10, “necessity of OH concentrations similar to the ambient” is probably overstated. There is a literature on the necessary conditions for atmospheric relevance of OFR / PAMS / flow tube type experiments, and this is taking an end run around that literature and oversimplifying the requirements for atmospheric relevance.  

9.    The introductory material on flow reactors versus chambers is underdeveloped and would benefit from a rewrite and citation to the literature on the topic (e.g., line 40, “researchers use both flow reactors and chambers”).

  

Minor points

1.    The paraphrase of Coggon et al. (2018) at line 19 is not quite accurate and should be revisited. 

2.    Not sure “Cl is negligible” in the correct interpretation of Alton et al. (2020). The global value was 4.6% for D5, but up to 25-30% for Toronto and Boulder under certain conditions.

3.    As an update to Hobson et al. (1997) on deposition sensitivity, one could refer to the work of Janechek et al. (2017), where dry deposition was quantified and dry deposition modeling parameters were discussed in detail. 

4.    The megacity vs. non-megacity distinction at line 38 is not clear. 

5.    The foundational work in the UNC chamber on organosilicon oxidation should probably be mentioned in the introduction (Latimer et al. 1998, Chandramouli et al. 2001).

6.    Explain “batch mode” line 45

7.    Introduction should make it clear that the paper presents new results from both a flow reactor and chamber were used, and give a roadmap to the different sections of the paper.

8.    Line 104 – perhaps state (list a paper or report) where a figure can be found with that static mixer shown. 

9.    Line 113 – could a time series of a typical chamber experiment be included in SI, showing the steady state stability of the precursor implied at line 113 (negligible wall loss). 

10.    I respect and value the important work on uncertainty quantification, but placing it in the caption of the Table is not ideal.  Relocate to methods and/or results and/or SI as appropriate.

11.    Having explicit labeling of chamber vs. CPOT would be helpful, so that one does not have to know which device is associated with which experiment number.

Ref Cited

Alton, M. W. and E. C. Browne (2020). "Atmospheric Chemistry of Volatile Methyl Siloxanes: Kinetics and Products of Oxidation by OH Radicals and Cl Atoms." Environ Sci Technol 54(10): 5992-5999.

Chandramouli, B. and R. M. Kamens (2001). "The photochemical formation and gas-particle partitioning of oxidation products of decamethyl cyclopentasiloxane and decamethyl tetrasiloxane in the atmosphere." Atmos Environ 35(1): 87-95.

Coggon, M. M., et al. (2018). "Diurnal Variability and Emission Pattern of Decamethylcyclopentasiloxane (D5) from the Application of Personal Care Products in Two North American Cities." Environ Sci Technol 52(10): 5610-5618.

Hobson, J. F., et al. (1997). Part H. Organosilicon Materials. G. Chandra. Berlin, Springer-Verlag. 3.

Janechek, N. J., et al. (2017). "Comprehensive atmospheric modeling of reactive cyclic siloxanes and their oxidation products." Atmos Chem Phys 17(13).

Lambe, A., et al. (2015). "Effect of oxidant concentration, exposure time, and seed particles on secondary organic aerosol chemical composition and yield." Abstracts of Papers of the American Chemical Society 250.

Lambe, A. T., et al. (2012). "Transitions from Functionalization to Fragmentation Reactions of Laboratory Secondary Organic Aerosol (SOA) Generated from the OH Oxidation of Alkane Precursors." Environ Sci Technol 46(10): 5430-5437.

Latimer, H. K., et al. (1998). "The Atmospheric Partitioning of Decamethylcylcopentasiloxane (D5) and 1-Hydroxynonomethylcyclopentasiloxane (D4TOH) on Different Types of Atmospheric Particles." Chemosphere 36(10): 2401-2414.

Palm, B. B., et al. (2016). "In situ secondary organic aerosol formation from ambient pine forest air using an oxidation flow reactor." Atmos Chem Phys 16(5): 2943-2970.

Peng, Z., et al. (2016). "Non-OH chemistry in oxidation flow reactors for the study of atmospheric chemistry systematically examined by modeling." Atmos Chem Phys 16(7): 4283-4305.