General Comments

In this manuscript the authors report a combined experimental and modelling study of the formation and fate of acylperoxy radicals formed from the reaction of a-pinene with ozone in a flow reactor. The alkyl and acyl RO2 radicals and highly oxidized molecules (HOMs) were monitored using a chemical ionization mass spectrometer with nitrate ion ionization. RO2 radicals and HOMS were assigned based on elemental formulas and acyl RO2 radicals were distinguished from alkyl RO2 radicals by addition of NO2, which forms RC(O)OONO2 (acyl peroxy nitrates) that are relatively stable under the conditions of the experiments, thus removing acyl RO2 signal. Because the changes in acyl RO2 concentrations can also impact other aspects of the chemistry, a detailed F0AM model employing a modified Master Chemical Mechanism was employed to interpret the results.

Overall, the experiments and modelling were well done and the approach seems to have yielded quite useful and interesting results. The authors provide a very thorough and thoughtful discussion of the results, which is clearly written and easy to follow. Considering the high technical quality of the study and the importance of these reactions to the formation of HOMs and ROOR dimers, both of which are currently of much interest because of their potential role in secondary organic aerosol (SOA) formation, I think the paper is well suited for ACP. I have only a few minor comments.

Specific Comments

1. Line 137: I don’t understand the point of converting signals to “concentrations” using sulfuric acid since the actual concentrations will be highly sensitive to the structure of the RO2 radical and HOM. Presenting the results this way is misleading. Since the “concentrations” are only used to calculate contributions of various species relative to each other, normalized signals will give the same results and be a more honest presentation of the data.
2. Line 149: Have the authors considered partitioning of RO2 radicals to particles and what influence that could have on the results? The vapor pressures of the radicals should be similar to those of HOMs, so I don’t see any reason that they would not form SOA, and they likely undergo different reactions in the particles since isomerization would be restricted.
3. Line 342: In this section it is not clear to me what conclusions are based on measurements, modelling, or a combination of the two. Please make that more clear.
4. Line 530: Considering that RO2 + NO2 rate constants have been measured for a variety of alkyl and acyl RO2 radicals and are pretty consistently ~1E–11 (Orlando & Tyndall 2012), it seems unlikely that the value is as low as suggested here. Any explanation based on RO2 structure would imply that the same effects apply to the RO2 + NO rate constant, which is essentially identical to the NO2 value (Orlando & Tyndall 2012). This would have significant consequences for predictions of conditions under which autoxidation reactions are important in the atmosphere, since this usually depends on the competition between RO2 isomerization and the RO2 + NO reaction. What are other possible explanations for the apparent discrepancy?

Technical Comments

None.

General comments:

In this study, the authors investigated the fraction of acyl peroxy radicals (RO₂) formed from alpha-pinene ozonolysis. Acyl-RO₂s are of crucial atmospheric importance due to their higher reactivity and their role in the formation of aerosol precursors. In flow reactor alpha-pinene ozonolysis experiments, NO₂ was used as an acyl-RO₂ scavenger, and the reduction in RO₂ signals was used to probe the fraction of acyl-RO₂s produced. The paper is well written and makes a significant contribution towards the better understanding of a key aerosol forming system in the atmosphere. I recommend the publication of the manuscript in ACP after the authors address my minor comments below:

• In addition to dimer formation and producing RO, alkyl-RO₂ cross reactions also lead to ROH + R=O products, and if the inital peroxy radical group is on a primary carbon atom, the R=O can be a source of acyl-RO₂ following a secondary OH reaction. Alkyl RO₂s can have significant yields for this reaction. Was this accounted for in the model and in the analysis of the experiments?
• The more functionalized acyl-RO₂s with an -OOH group elsewhere in the molecule are known to undergo H-scrambling reactions to form peroxy acids (R-C(O)OOH) at rates of 1E3 – 1E5 s^-1 (Knap et al. 2017, J. Phys.  Chem. A, 121(7), pp.1470-1479). Can the authors comment on the possible role of this reaction in their experiments? For example, the ring-opened acyl C₁₀H₁₅O₈-RO₂ that they report has a 1,6 H-scramble available that leads to a peroxy acid and an alkyl RO₂. For their model system, Knap et al. estimate a rate coefficient for the 1,6 H-shift of 1.5E5 s^-1. If the rate coefficient is comparable for the alpha-pinene derived C₁₀H₁₅O₈ acyl-RO₂ above, this could to an extent explain the low reduction in signal upon NO₂ addition.
• Line 468: Regarding the speculation of the formation of alkyl C₉H₁₅O₃-RO₂ to explain the discrepancy between experiments and simulations, this would compete with the formation of the ring-opened and ring-retaining C₁₀H₁₅O₄-RO₂. How did the sensitivity analysis of including the C₉H₁₅O₃-RO₂ in the model affect the yield of the C₁₀H₁₅O₄-RO₂s  and the subsequent acyl-RO₂s derived from them?
• Are any of the acyl peroxy nitrates detected by the NO₃-CIMS? Alpha-pinene derived APNs with 8 oxygen atoms or more should have at least 2 -OOH functional groups and will presumably cluster well with NO₃^-. Do the decrease in e.g. C₇ and C₈ RO₂ signals when NO₂ is added show an increase in the corresponding APN signals? I think a spectrum figure maybe in the supplementary showing the acyl-RO₂ and acyl-ROONO₂ peaks would be useful.
• Figure 4. In pathway 3, the final H-shift of the acyl-oxy is unlikely to compete with CO₂ See reaction r12 and description therein in Vereecken et al. 2009, Phys. Chem. Chem. Phys. 11(40), pp.9062-9074.