In this work, the authors provide a kinetic and mechanistic study of the OH-initiated oxidation of 3-penten-2-one (3P2), and its by-product 2-hydroxy-propanal (2HPr), chemistry that is of potential importance in biomass burning plumes.  Major outcomes of the work include a first-time determination of the products of the OH-initiated oxidation of the two species, and a determination of the rate coefficients for their OH reaction (each of which was the subject of one previous study).

It is my opinion that this is likely to be a publishable body of work, although I think there are a few quantitative issues that need to be resolved before publication is recommended.

My major comment has to do with the modeling work that is done. I think much more needs to be said about the details of this modeling (even though this is outlined in a recent paper by the same authors), to lend confidence in the key parameters that are derived from it (e.g., k(OH+2HPr), branching ratios for this reaction, ‘corrected’ yields of various products).  First, the details of what is included in the model, how it is run, what parameters are varied, etc., needs to be included.  Further, the sensitivity of the model output to the varied parameters needs to be shown (presumably graphically).  As examples, how sensitive is the model to the OH/2HPr rate coefficient, to the branching ratios to the two channels in this reaction, etc.  What does an 80:20 branching in the OH/2HPr reaction look like compared to 70:30 or 90:10? Etc…

More minor remarks:

How is the 3P2 introduced into the chamber and how is its initial concentration determined? In particular, if this calibration is volumetric or similar, how is the 85% purity of the purchased sample dealt with?

Page 6 – Did the authors determine a wall loss for the reference species used in the k(OH) determinations? If so, were they negligible?  What fraction of the 3P2 loss is due to the walls compared to the loss due to OH reaction in a typical experiment?

Page 7 – The formation of acetaldehyde from the large Criegee radical formed in the 3B2OL / O3 system is a bit surprising to me. A standard pathway for decomposition might be to form OH and methylglyoxal, rather than acetaldehyde?  Can the authors provide any further ideas or speculation on its formation mechanism? (Any chance of impurities existing in the alcohol?).

Minor corrections, suggestions, etc.:

I don’t know what an Euler-Cauchy model is. Please explain briefly.

Page 2, line 33, I suggest “Unsaturated ketones are of increasing interest as …”

Page 2, line 43, I suggest “is believed to be a …”

Page 2, line 48, I suggest “Some field measurements of BB plumes…”

Page 2, line 59, I suggest changing “Besides” to “In particular” (or something similar).

Page 2, line 59, Change “expand” to “expanding”.

Page 6, line 167:  Maybe “all of the experimental results” instead of “the whole…”

Page 7, line 184: ‘coexist’ instead of ‘coexists’.

Page 12, line 290: Specify that this is from the OH + 3P2 reaction.

Page 14, line 360, also Page 16, line 392:  The equivalent yields of the two products does not prove that they come from the same pathway.  I suggest using different wording here. ‘suggests’ or ‘consistent with’?

Page 17, top:  Nitrates are quite strong absorbers in the IR.  Was there any sign of their formation? Can any limit on their yield be obtained, using nitrate IR cross sections (which are reasonably transferable) from other works?

Page 17, line 413:  Specify that it is 3P2 that you are referring to.

Page 17, line 419:  While I agree that 3P2 (and by-products) are a source of PAN, the overall contribution to NOx reservoirs will depend on the concentrations of 3P2 relative to other PAN sources.  Perhaps ‘soften’ the language here?

This paper presents the results of a kinetic and product study of the reactions of 3-penten-2-one (3P2) with OH, which relevant to models of biomass burning plume chemistry. The results gave reasonably well characterized rate constants and product yield data, and should eventually be published. They also investigate the kinetics and mechanisms of its oxidation product 2-hydroxy propanal (2HPr) by modeling irradiations of reaction mixtures after it is formed in the experiments with 3P2. The results were used primarily as a means to correct the yield data in the 3P2 experiments for secondary reactions, but also provide a useful addition to the literature, though the results for 2HPr are more uncertain because they are obtained from modeling a complex system.

The use of modeling with a simplified mechanism to correct the yield data presents some uncertainties, most of which are discussed in the text. However, the approach seems reasonable though not well described (see below). Fortunately, the corrections do not have an excessively large effect on the reported yields (see Table 2), and the general conclusion that the 2HPr + acetyl peroxy route is about twice as important as the methyl glyoxal + acetaldehyde route is not affected. This is an interesting result in the context of SAR development, as well as for improving models for 3P2 reactions.

Although this paper makes a contribution to the literature and should eventually be published, it does have some problems that need to be addressed. Most of the problems with this paper were noted in the posted review of Anonymous Reviewer #1, which I have read prior to writing my comments (but after reading the paper), along with the response by the authors. I agree with the comments of this reviewer, and believe that for the most part (but see below) the authors propose changes that should adequately address these comments.

The major problem noted by Reviewer #1 is the inadequate description of the modeling method used to correct the data and obtain the yield parameters. As the reviewer noted, the reference to the "Euler-Cauchy" model to derive parameters or corrections to the data is unclear, since this refers to a general solver method that could be applied to any system. In response, the authors improved the text around line 120 to better describe how the corrected yields were derived. The new Table 2 is a valuable addition to the text.

However, the proposed changes to the reference of "Euler-Cauchy" in the abstract is not totally adequate. In responding to the reviewer, they changed the sentence in the abstract from "Employing an Euler-Cauchy model to describe the temporal behaviour of the experiments, the further oxidation of 2HPr was shown …" to "Employing a simple modeling tool describe the temporal behaviour of the experiments, the further oxidation of 2HPr was shown …". However, the modeling tool did not describe the temporal behavior of the chemical system, their assumed mechanism (given on the new Table 2) did. A better change may be "Employing a simple chemical mechanism to analyze the temporal behaviour of the experiments, the further oxidation of 2HPr was shown …". The modeling methodology need not be given in the abstract, since it is not newly developed in this study, and is presumably adequate.

Reviewer #1 also noted that reaction of the hydroxy-substitued Criegee intermediate shown on Figure 2, forming acetaldehyde. is not the expected pathway. In any case, Figure 2 doesn't show the whole mechanism for the CI decomposition, since not all atoms are accounted for. In response, the authors wrote that acetaldehyde formation was unexpected and may be due to experimental impurities, and that speculation of the CI mechanism was beyond the scope of this work. But at least they should give a complete proposed mechanism on Figure 2, or show a "?" and "+ other products" on the figure to indicate that they don't know the mechanism.

However, there is quite a reasonable explanation for acetaldehyde formation from this CI on Figure 2. The H on the -OH could move to the outer O on the intermediate via a 6-member ring transition state, which could then rearrange rapidly to form acetaldehyde + formaldehyde + OH. This may not be the only route (the other route forming methyl glyoxal as the reviewer noted could also occur), but it is possible acetaldehyde formation may occur at least part of the time (and maybe dominate). Perhaps add both routes to Figure 2 and state in the discussion that relative importance of each is uncertain and not further investigated in this work.

Reviewer #1 noted other problems with the paper, but it appears that the authors' responses were adequate, so I will not discuss them here. However, I have a few additional comments and suggestions. Other than these, I did not see major additional problems with the paper.

It was unclear whether the product concentrations plotted on Figure 4 have been corrected for secondary reactions when that figure was first introduced. Later (around line 314) it is implied that Figure 4 shows uncorrected data. It might be a good idea to show both corrected and uncorrected data on Figure 4, so one can get a feel of the size and effects of the corrections.

Around line 294 it was stated that yield plots of 2HPr showed "small but precise" curvature. However, the 2HPr data on Figure 4 don't look particularly curved. Are they referring to corrected data? If so, this is another reason to include corrected data on Figure 4.

The failure to observe the PAN analogue CH3-CH(OH)-CO-OONO2 shown in Figure 7 might be due to another rapid 1,4 H-shift reaction where the H on the -OH moves to the peroxy O next to -NO2, forming HNO3 and (ultimately) CO2 and acetaldehyde. This could be potential source of acetaldehyde in the system that is not accounted for in their model (new Table 2). Would this affect the acetaldehyde yield from 3P2 that fit the data?

General comments:

This study by Illmann et al. describes kinetics, products study, and reaction mechanism for the OH-initiated oxidation reaction of 3-penten-2-one which had been detected in biomass burning plumes. To understand biomass burning chemistry and its role on tropospheric ozone and SOA formation, the chemistry of individual chemical components of a biomass burning plume should be known. This study presents a comprehensive understanding of the atmospheric chemistry of 3-penten-2-one which could be important from biomass burning perspective. The authors also discussed OH reaction for 2-hydroxypropanal which is an oxidation product of 3-penten-2-one.

This work is of high scientific quality and is suitable for readers of Atmospheric Chemistry and Physics (ACP). However, in many cases, the detail about the measurements and the analysis is missing in the manuscript which could bring questions to the mind of the readers. Therefore, I recommend publishing this manuscript in ACP after addressing the issues mentioned below. Necessary modification to the manuscript should be done accordingly.

Specific comments:

1. Page 4, line 117: A polynomial function was used for CO₂ calibration. Any possible reason why a linear correlation was not observed?

2. Page 4, line 121: Describe Euler-Cauchy approach. At least some details are required for the readers of this manuscript who are not familiar with this method.

3. Page 5, line 147: The sample purity is low. Even a sample purity of 85% is not good in this kind of precise measurement and it could severely impact the product studies. It creates a lot of uncertainty if we don’t know what is that 15% impurity and what chemistry it can bring here.

Did you try to distill the sample?

Given the fact that the IR cross-section of 3-penten-2-one was determined in the present study, it is not possible to check impurities using FTIR if you don’t have reference spectra of the sample from other independent studies that used more pure samples.

FTIR itself is not sensitive enough to identify small impurities and a more sensitive method e.g., GCMS could have been used.

4. Page 6, line 154: Why is the loss rate of the sample about one order of magnitude larger in the chamber having higher volume? I expected an opposite trend as the S/V is expected to be less for the larger cell.

Please show representative first-order decay plots for wall loss in Supplement.

What are the S/V values for the two chambers?

What are the wall loss rates for the reference compounds?

5. Page 6: 3-penten-2-one + OH kinetic data looks satisfactory. For presentation purposes, the data points for individual experimental runs could be shown using different symbols/colors to highlight data quality for individual experimental runs.

Typically, how much corrections (in percentage) were made on individual data points to account for the 3-penten-2-one wall loss?

It was noted that a previous study from the same laboratory reported a rate coefficient that is slightly higher than the number obtained in the present study. It was noted that they agree within 20%, to my opinion this difference is rather large. Mesityl oxide was an impurity in the previous study, but this should not bias kinetic results in relative rate measurements if it does not interfere with your spectral subtraction. In fact, when I see kinetic data for Isoprene reference (Figure 1), there are some points (probably correspond to the last row of Table 1 i.e., Expt 3P2#6) that are bias low making the whole kinetic data scatter and the rate coefficient bias low. If only E2-butene data are considered, then the correspondence is much better ~ 10%.

6. Page 7: In situ generation of 2-hydroxypropanal – Determination of 2-hydroxypropanal (2HPr) infrared cross-section, which was used later for its yield determination, is based on the assumption Yield (2HPr) = 1 – Yield (HCHO). This assumption is valid if HCHO is formed only by pathway a (Figure 2) and 2HPr is formed only by pathway b and there are no other pathways for POZ decomposition. The authors did not discuss the total fate of corresponding CIs (from pathways a and b) which can complicate the analysis. For example, the CI for pathway b for 3B2OL is CH₂OO i.e., the simplest Criegee Intermediate. Bimolecular self-reaction of CH₂OO (whose rate is very fast) could form additional HCHO, then the above assumption would be invalid. Similarly, it was shown that the branching ratio for acetaldehyde formation from the CI that is produced from pathway a is 0.36, what is the fate of the rest of CI? If there is a bimolecular self-reaction, then it would produce additional 2HPr. Give proper evidence that these bimolecular self-reactions of CIs are not happening in your experimental condition.

The quoted uncertainty for OH absorption band cross-section is very high which would make the product yield data unreliable. The high uncertainly was chosen for the error on the HCHO yield and wall loss of 2HPr. Give more details on that.

Does the wall loss follow the first-order decay?

Did you also observe wall loss for 3B2OL and corrected it?

7. Page 11, line 266: If the loss of acetaldehyde and methylglyoxal by OH reaction is significant, should not you observe nonlinearity in their yield data (Figure 4) at longer times?

8. Figure 5: The fate of CH₃CO radical following reaction with O₂ and NO is shown to be the formation of CO₂ and HCHO. If HCHO is formed by CH₃OO reaction, then other products such as methanol are also expected.

9. Page 13, line 312: Describe the model that was used to correct 2HPr yield due to the 2HPr + OH reaction. Input/output could be provided as a Supplement.

10. Page 13, line 330: A rate coefficient for 2HPr + OH reaction is quoted but the details on how this number was obtained are not presented. Please provide the necessary details. The rate coefficient differs from a previously measured value by 30% which is not an excellent agreement.

11. Page 14, line 353: Again, the details of the model that was used to find the branching ratios, was not given. Provide necessary details here and Input/Output as a supplement.

Did you do any sensitivity analysis?

12. Page 17, line 416: The authors have concluded that OH reaction is the dominant degradation process since they did not observe measurable photolysis in their experiments. This is not true because UV light intensities and OH concentration in the chamber and in the real atmosphere are not the same. In general, aldehydes have photolysis lifetimes that are comparable with OH reaction lifetimes. A conclusion on this can only be made after evaluating its atmospheric photolysis lifetime from UV absorption cross-section and quantum yield values.

Minor technical correction:

1. Title: The word “main” could be omitted as there are other major oxidation products.

2. Page 9, line 229: Also add 2HPr.