A Vocus CI-ToF with a NH₄⁺ reagent ion source was applied to sample the roof air in Upper Manhattan in wintertime. A range of VOCs to I/SVOCs including acetates, glycols, glycol ethers, alcohols, etc., which have uses in personal care products, fragrances, solvents, and/or other volatile consumer products, have been measured. Concentrations, dependences of meteorology conditions, and the relative enhancement ratios with typical tracers of the targeted compounds were discussed. The application of ammonium as the positive-ion reagent gas provides another angle of understanding of the compounds with a diverse range of chemical functionalities, showing the advantage of NH₄⁺ ionization, which would enhance the knowledge of VCPs-related emissions in the megacity region. The concentration ratios of the targeted compounds to several common tracers, such as acetone, CO, and benzene, would be helpful to understand the emission structure. Overall, I believe this paper is worth publication after some minor revision.

As said in Line 151-152, the measured concentration of functionalized compounds could be emitted from diverse, distributed sources around New York City. And the additional contribution from other sources (e.g., biomass burning) would bias calculations of urban emission ratios in this study, as described in Line 225-226. Both local and regional sources can contribute to the concentration of functionalized compounds measured in the roof, but the authors seemed to attribute all the emissions to local VCP sources without any discussion about long-range transport contribution or local non-VCP sources. I am confused that the emission from vehicles, for example, was completely missed in any of the discussions in this paper. Authors should add words evaluating the impact of regional and local non-VCP sources on the targeted compounds.

Below are several additional comments

1. Line 33: It is not accurate to say “online measurements of oxygenated VCPs in a…” because the 10 days measurement was not the direct and exclusive measurement of the VCPs sources. The ambient concentration of these compounds may come from other sources. Using oxygenated organic compounds might be appropriate.
2. Line 86-87: “These health impacts will be modulated by the air change rates at which indoor emissions of ROC are transferred outdoors.”
3. It might be better to put Line 98-114 before Line 83-96 for a smooth transition of background introduction.
4. There is a lack of summary of the advantage and current research state of using NH₄⁺ as the chemical reagent ion in CI-TOF instruments in the Introduction part. Maybe several sentences right after Line 120 describing the ability to measure I/SVOCs and other chemical functionalities, as the follow-up of Line 83-96.
5. Line 138: The citation of Warneke et al. needs to be corrected.
6. Line 185-186 and Line 335-338: I agree that the buffering effect of water clusters in the reactor can remove the humidity dependence but several more compounds should be added in figure 1 to reinforce your statement. Especially the ones that were discussed in the later sections. Could you also add some discussion on why the sensitivity of MEK increased by 10% with humid calibration?
7. Line 221-228: It is fine to filter the data but there was a lack of reference to support the application of benzene-to-toluene ratio>2 as the threshold for identifying biomass burning events. I recommend using the enhancement ratio of acetonitrile to CO as a more exclusive tracer for labeling biomass burning.
8. Line 327-331: Examples are needed to show the bound of “slight over-or under-estimation”. Especially the ones that showed deviated results when compared with the VCP inventory. If the calculated concentrations could have a factor of 0.5 to 8 differences depending on the relative abundance of isomers (described in Line 323-325 and figure S5), this impact should be evaluated.
9. Line 357-359: Please provide more evidence to support the statement or I am not convinced that C₂H₅OH signal was more like dimethyl ether instead of ethanol.
10. Line 590-592: Maybe I missed something, but this statement was not clear. How to translate the emission inventory to concentration? Or where did the factor of 2 come from?
11. Line 595-597: I don’t agree with this statement. If D5 has large indoor sinks, either chemically or physically, the reduced ventilation would have less impact on its contribution to ambient levels than a compound that has small indoor sinks.
12. Line 600-603: “…, benzyl alcohol showed its potential as an additional VCP compound…”
13. Line 644: Authors should be more serious about the determination of background concentrations. More evidence is needed to support using 5^th percentile of the data as the regional background. The background can be vital in the discussion in section 3.5. Please add more references as well in Line 774.
14. Line 709-711: Since the acetonitrile and CO data were available, the authors might check the enhancement ratio. According to Line 221-228, the data that was influenced by biomass burning had been removed based on benzene-to-toluene ratio, but if the peak of C₂H₇NO was caused by biomass burning, authors should reconsider how to filter the biomass burning event.

The authors present measurements from New York City, USA, to evaluate the emission strength of oxygenated volatile organic compounds (OVOCs) that originate from volatile chemical products (VCPs). They use the VOCUS in NH₄⁺ mode to detect and quantify for the first time OVOCs related to VCP emissions including glycols, glycol ethers, acetates, alcohols, and others. They evaluate the instrument performance for these compound classes based on laboratory calibrations and provide field enhancement ratios with typical tracers including benzene, CO, and acetone. Furthermore, they perform a comparison of the observations to two VCP emission inventories to find good agreement for many compounds but also differences that are further discussed. This paper is great and fits well within the scope of ACP after the following minor comments are answered.

Main comment:

Throughout the paper, the authors have no comments on the influence of other pollution sources. There is still traffic, cooking, biomass burning, and industrial activities that could influence the observed concentrations in NYC. To my knowledge, how much these sources contribute to OVOC emissions is currently unknown; therefore, more discussion is essential here. I understand that performing detailed statistical analysis to apportion the different pollution sources (e.g., PMF) is not the scope of this paper, however, discussing the limitations of the approach followed here will be important especially given that OVOC emissions from the above sources (e.g., cooking and biomass burning) are also not well constrained. Furthermore, OVOCs can be a product of secondary chemistry and although the measurements are during the winter, increased concentration of these OVOCs due to atmospheric chemistry cannot be excluded and should be at least discussed. 

I appreciate Figure 2 where the authors highlight the performance of the NH₄⁺ VOCUS. It would be great if the authors could comment further on the influence of protonation, charge transfer, and fragmentation on the calibrated target VCP as well as other compounds they calibrated. I think that a table in the supplement showing the percent per compound that was ionized by NH4+, vs. other ionizations at different RH would be valuable given that these are the first measurements and detailed calibrations of these OVOCs. For example, are there specific compound classes that may be influenced more than others by different ionizations when operating the VOCUS in NH₄⁺ that the community should be aware of? I consider that highlighting the limitations of this technique is as valuable.

Minor comments:

Line 137-138: I would delete this line given that this campaign has not happened yet.

Line 225-230: The authors discuss how they exclude the influence of biomass burning. However, this is not the most precise approach to guaranteeing the full separation of this influence. Furthermore, there are more urban sources emitted in NYC than VCPs. Traffic, cooking, and industrial emissions from the outer regions could affect the measurements and therefore also the inventories. Oxygenated compounds that have high background due to long-range transport could also influence the concentration of the OVOCs studied here.

Line 328: I would suggest deleting “slight”.

Line 412-438: The authors nicely discuss the influence of chemistry on the chemical degradation of VCP emissions. Given that these are oxygenated compounds I would also expect the influence of secondary production of these pollutants in this region even during the winter. This would be an important point especially when the observations exceed the expected emissions based on the inventories.

Line 487-489: I think it will be important to discuss what the influence of other sources is. For example, although it is winter, chemistry can contribute to these signals but also other sources such as cooking emissions could influence the observed concentrations.

Figure 4-5: It would be interesting to discuss the diurnal variability of these compounds and whether this could tell us something regarding their emission source.

Line 520-524: Acetone also originates from other primary emissions such as traffic, cooking, etc. that the authors should discuss here.

Line 766: It would be good if the authors defend that acetone is predominantly from VCPs vs. other primary or secondary sources with references.

Line 774: Prior work? Reference?

Line 933-934: I would recommend changing “deviated” to a more detailed statement and expanding on the results of this study further here.

General comments

Khare and coauthors show the utility of ammonium-adduct chemical ionization in measuring a range of species from VCPs in the NYC metro area. They give a detailed explanation of the method, identify compound classes measured in the area with a generalized estimate in sourcing, and compare these measurements to VCP inventories. The measurements provide constraints on the inventories of many VCPs and shows that these inventories need to be modified for some species in the region.  This analysis gives a preliminary survey of the region in the wintertime and sets the stage for future, more detailed measurements in this area to come in a different season.

My main comment is that there were collocated PTRMS and GCMS measurements that should be leveraged more to help with conclusions made on magnitudes in concentration and for isomer identification. This should be addressed to some degree since it could be used in the evaluation of inventories. This does not have to be done for every species listed but should be for the few species that overlap in each spectrum and considerably deviate from inventory estimates (e.g. acetone, ethanol, ethylene glycol). These species and suggestions for comparisons are listed in the specific comments below. The limitations of the GCMS are not clear so if isomer identification cannot be performed it should be explained why.

I also believe there should be more discussion of figures of merit and uncertainties in both the ammonium-adduct CIMS measurement and inventories. For some species the concentrations are listed at <10 ppt and there should be some statement of limits of detection as well as signal to noise filtering. After these two main comments and other minor comments are addressed, I believe this manuscript is suitable for publication.

Specific comments

Introduction: this is a small comment but there is a little confusion as to how VCP is defined. Are VCPs the actual material (ink, paint, etc.) from which emissions come from or are they the emitted chemical species? Your definition on line 59-60 make it sound like the former but the rest of the introduction sounds like the latter. I think there should be consistency or more definition for the reader since this is propagated throughout.

Line 120: a lot of the species you list that have previous “measurement challenges” can be measured with an iodide adduct CIMS. It seems like one of the largest advantages is the low water dependence on signal but it is not clear which species your method can measure that the ICIMS cannot or the differences in signal and sensitivity. Can you list either in the introduction or methods these advantages for someone who might be considering this method? Since  this manuscript is a balance of a technical discussion of the utility of this novel method as well as meausurements focused it would be good to highlight how the method stands out.

Line 138: you should either explain what AEROMMA is (and define the acronym) to show why it is important that there would be preliminary VCP measurements here or remove this line. I would suggest removing the line since you reference AEROMMA and GOTHAAM later (which should also be explained with a line if still referencing).

Line 146-147: state in general what factor higher this sensitivity is for relevant molecules to support this.

Line 261-267: there are many species listed in this manuscript with concentrations that cannot be predicted with VCPy which is fine because it’s an inventory. I am wondering if the uncertainty in product-specific indoor emission fractions is large enough to account for these discrepancies. Either listing uncertainties for those fractions here or giving an example of why it would or wouldn’t change results (e.g., the difference in acetone and ethylene glycol) would reduce some reader assumptions.

Line 283-284: what was your criteria for high signal to noise?

Line 287-290: you should put a y axis on figure 2a even if it is just a relative amount just to show that it is linear and support your statement that there is a low parent ion to fragment ratio. I am assuming it’s not a log plot like 2b but I can’t make an assumption on no axis. Also if you are going to state that the PTRMS has high parent ion to fragment ratio for similar classes of molecules you should either show a comparative spectrum like in 2b or list some species ratios across the two instruments.

Figure 2: are the high peaks >= m/z 300 internal standards? If so list that in the figure caption.

Line 327-331: Since you refer to over and under estimations due to the ensemble of isomers detected in ambient air, can you reduce this uncertainty when using the isomers detected onsite with the GC?

Line 349-350: Can you get closure in the differences in C3H6O signal using the contributions of the GC? It looks like from table 1 you were monitoring C3H6O.

Line 355-357: without any calibration time series your statement as written does not convince me that this was not ethanol. The PTRMS also has a low response to ethanol but should still provide a measurement to compare this against. The PTRMS also shouldn’t measure dimethyl ether well so it would be more ethanol signal for that instrument. I suggest including this comparison.

Line 366-368: this statement of low collisional energy for higher masses seems to be supported from your figure 1 temperature dependence of alpha pinene relative to the other smaller compounds, although this is a hydrocarbon. In fact, the sensitivity goes up for alpha pinene. Would you say that it is the thermal stability of the adduct that drives this temperature dependence or the increased frequency of lower energy collisions of larger compounds? Or is it the relative amount of NH4H2O+ to NH4+? You should try adding another higher MW compound to figure 1 to support or disprove this or add a line about ion adduct strengths near figure 1.

Figure 4: list where these uncertainties come from. Is it just the standard deviation used in table S2?

Figure S8: is there a reason alpha cedrene and alpha pinene are chosen for OH oxidation? Is that what is prominent in an inventory or was it chosen based on the GC?

Line 552-555: this underestimation in ethylene glycol is interesting. Does this molecule correlate with any other in the spectrum in a meaningful way or is there just a general enhancement in many molecules pre-01/25? This molecule should also be detected by a PTRMS. Was there a strong corresponding signal in the PTRMS? It would be helpful to have a calibrated comparison for this since it’s such a large unpredicted concentration and currently there is no explanation other than there must be a higher emission (which is fine).

Line 568: 10 ppt of C4H10O2 and 5 ppt of C6H12O3 is great. Are these concentrations listed averaged over 1 Hz or higher? It would be helpful to list some general limits of detection for this method and if they’re already presented in another paper you should refer to that paper with some numbers. It would support how useful this method could be for the reader.

Line 708-711: can you show that this was from a biomass burning source from a backward trajectory analysis or a CO measurement? If these species tracked with others that you would assume were from VCPs this could be a helpful way to distinguish them and reduce uncertainty in your model.

Figure 7: is there any product-specific uncertainty you can place on these models that would modify the emissions ratios or is the uncertainty just general that applies to all species? I think including some model uncertainty in either the model descriptions or here would be helpful.

Line 863-865: some of these species could be measured with a PTRMS. Can you provide a comparison to show that there is a strong deviation from the model for this ratio for the PTRMS too?

Technical corrections

Line 152: use consistency in ammonium or NH4+.

Figure S9: the species need to legible.

Line 428: define “k value”

Figure 5: the x axis ticks need to be consistent for each figure.

Figure S9: the axes and error bars are illegible even if zoomed all the way in. This needs to be corrected.

Figure S10: a suggestion: making the colorbar linear (e.g., viridis in python) would make this a lot easier to interpret. This could be a really neat plot if it was easier to track the inventory emissions.

Figure S13: same comment as Fig S9. The axes and error bars are illegible even if zoomed all the way in. There are a lot of species but they can’t be read.