This study by Harb et al. examines the gas-phase oxidation of phellandrene isomers by NO3 radicals. They conducted gas-phase measurements via FTIR, a combination of PTR/NO+ TOF-MS and NO3-CI-TOF-MS, and particle measurements via offline Orbitrap, ACSM and SMPS. Overall, their work reports the SOA yields for these reactions, along with qualitative particle phase composition measurements (orbitrap). From their online measurements of the gas-phase components they propose in-depth oxidation mechanisms. 

Overall comment: With the manuscript in its current form, I do not recommend this work for publication. This is primarily due to the over-use of the gas-phase data to create detailed mechanisms, when no speciated structural information or quantitation was obtained. Setting the lack of structural information aside, beyond the obvious limitation of the method used (PTR/NO+ TOF-MS) that isomers cannot be resolved, at best this method offers complex detection of these multifunctional species with high LODs. In general, traditional PTR/NO+ methods are not appropriate for the detection of multi-functional (especially organic nitrate) compounds due to fragmentation, unless a pre-separation method is used to quantitatively interpret the complex product ion distributions. I read the predecessor paper (Duncianu et al., 2017), and while the authors were able to demonstrate the formation of molecular ion adducts - the spectra of these synthesized standards were still complex (3 - 5 ions formed per compound), and generally dominated by fragmentation. I do not think it is appropriate to apply this method to a complex system like the oxidation of phellandrenes without supporting, speciated, methods for product identification and quantitation - especially if the aim is to create a detailed oxidation mechanism. 

Specific comments: 

1. In section 2.4 the gas- and particle-phase concentrations of organic nitrates (ONs) are discussed. Was FTIR used for both the gas- and particle-phase measurement? The authors mention collecting particles followed by extraction, but the detection method is not clear. Below they also mention online pON from the ACSM. Also, in line 223-224 referring to these as "molecular concentrations" is mis-leading since this is a bulk measurement. 

2. In section 2.5, the authors mention that the phellandrene oxidation mechanism is not included in the MCM and so they used the limonene mechanism as a proxy. Considering the authors claim that small changes in structure can create large differences in SOA yield/chemistry (e.g., large difference in measured SOA yield between alpha- and beta-phellandrene reported here) the authors need to provide more justification that the use of the limonene mechanism is valid here. For example, does the temporal profile of the decay of precursor and growth of first-generation oxidation products' align? Are the rate constants comparable? Furthermore, has the limonene + NO3 mechanism currently included in the MCM been validated?  

3. In Table 1, the "Date" column can be eliminated and replaced with "experiment 1" etc. This format should be carried out to other figures/discussion (e.g., Figure 2, legend). 

4. Section 3.3, line 355 - 357: The off-set at the origin (Figure 3) between measured ON and reacted MT, doesn't this indicate a high background? I'm not sure I understand the statement this the "non-zero slope at the origin suggests these are primary products."

5. Discussion, lines 395 - 410: Throughout the discussion the authors mention that in their experiments the RO2 fate was dominated by RO2 + RO2 reactions (> 95 %). However, they say that the generation of the closed shell products from the RO2 + RO2 reaction (either self or cross) to form a hydroxy and carbonyl product pair (Russell Mechanism) was negligible. This disagrees with past measurements which have shown branching ratios of averaging around 50/50 between closed shell products and alkoxy radicals, while variable depending on structure the molecular channel is not negligible for these types of RO2 structures (see review by Orlando and Tyndall, 2012). For other systems that also form first generation tertiary RO2 radicals, the closed shell hydroxy nitrate still form through cross reactions in non-negligible quantities (e.g., Claflin and Ziemann, 2018). In this work, the lack of measurement of the hydroxy nitrate, in tandem with the lack of speciated/specific measurements and the complex spectra generated from the method utilized, more so point to the measurement itself not being appropriate to detect these compounds. 

6. Figure 5. Throughout the mechanism, the authors have the alkoxy radicals reacting with O2 to form carbonyls. Justification should be given that this reaction could compete with decomposition under these conditions. The reaction of alkoxy radicals with O2 is typically negligible compared to decomposition or isomerization at room temperature (see Ziemann and Atkinson, 2012 and the references therein, also Vereecken and Peeters, 2009 and 2010). 

7. Line 465: The definitive assignment here of carbonyl, hydroxyl, or hydroperoxide compounds is not appropriate from the methods used. For example, the molecular formula of a "carbonyl" could also correspond to an epoxide (or others) as proposed by the authors previously in the paper. 

8. Lines 488 - 492: Without measured sensitivities for these ions (which also likely consist of multiple compounds, with varying sensitivities), the detection of one molecular formula being "higher" than the other should not be used to conclude that one type of product is formed more than another, or make conclusions about specific reaction channels. Also, definitively stating that these are carbonyl / hydroxy / hydroperoxy nitrates without supplemental measurement to confirm structure or the presence of these functional groups is inappropriate. 

9. Line 530 - 545: The qualitative detection of particle phase compounds containing > 3 nitrogen containing (I assume ON) groups, is very interesting. Further exploring the formation of these compounds (multi-phase or particle-phase reactions?) would be very nice. 

10. Lines 536 - 539: Is this a valid assumption (same sensitivity for each compound)? What is the error? The authors need to provide some justification for the use of this, can they show the same sensitivity using proxy compounds? Otherwise the quantitative nature of this discussion should be eliminated.