Nursanto et al. present methodological development to separate ammonium nitrate and organic nitrate signal in the time-of-flight aerosol chemical speciation monitor (TOF-ACSM) with capture vaporizer (CV). This has been a challenge due to the unit mass resolution (UMR) of the instrument, limiting the ability to separate different ions at the same nominal m/z as the NO (30) and NO2 (46) signal. A further challenge is the inclusion of a CV, which is used to improve quantification of aerosol concentration but also induces more thermal fragmentation of the ions, leading to most of the nitrate signal occurring at m/z 30 (NO), limiting the ability to use the methods previously published about using the calibrated NO to NO2 ratio from ammonium nitrate and the derived average ratio of pure organic nitrate aerosol, a.k.a. the "ratio-of-ratio" (RoR) method .

Using data previously collected from different aerosol mass spectrometers (AMS) and ACSMs with CV, the authors first investigated improving the fragmentation table, a tool used to separate ions at the same nominal m/z to differentiate the signal. As discussed in prior publications, a revised fragmentation table was necessary for the CV TOF-ACSM that they apply for the paper and recommend for future users. Next, they investigate the limits of quantification of the CV TOF-ACSM due to the low signal of NO2, and what nominal RoR to utilize for the TOF-ACSM (which is different than what is used for an AMS with standard vaporizer). After determining the limits of quantification and error propagation, the authors provide initial results from measurements conducted at an long-term monitoring site and from a chamber experiment.

This paper is of use for the TOF-ACSM community, as there are many TOF-ACSM with CV collecting long-term measurements. As emissions change (and thus aerosol chemistry), being able to differentiate ammonium nitrate from organic nitrates is of great value, as these two different NOx reservoirs have different properties for the aerosol, and provide insight into the chemistry controlling the pollution. After the authors address the following comments, the paper fits into AMT.

1) There is concern about frag_org[46] vs frag_org[45], as the R^2 is very weak. What is the general fractional contribution of frag_org[46] and frag_org[45] to the total signal (e.g., does it need to be corrected if this signal is low, especially in regards to NO2)?  Further, the correction of frag_org[46], as the authors conduct throughout the paper, is dependent on the aerosol being observed. As ambient aerosol is difficult to a priori know what is the origin, how much further uncertainty is introduced into this correction. E.g., looking at Fig. 1d,  for less oxidized organic aerosol (LO-OOA) and aerosol influenced by isoprene and a-pinene, which would all be scenarios expected to generally have high contribution of signal towards organic nitrate aerosol instead of ammonium nitrate, it appears the correction over corrects the signal at m/z 46. Wouldn't this then lead to a too low contribution of signal to NO2 and thus under reporting organic nitrates?

2) Section 4.2:  It is not clear why geometric mean was used to derive the ratio of pure organic nitrate (RpON). Not being a statistician, I do not understand the full reasoning behind using geometric mean, and why it makes more sense than arithmetic mean. If the authors could provide more details and references why geometric mean between two extreme values was used would strengthen the selection and section.  

3) Section 6.1: Co-located measurements of pON is extremely challenging and rarely possible, which is understood. If there was anyway to have a co-located measurement, from a chamber study or somewhere else where there was another ACSM with CV and another pON, would strengthen this section. Currently, the results shown in Fig. 5 and 6 are hard to judge if the trends and mass concentrations make sense.

4) Section 6.2: This section is not very convincing in that the TOF-ACSM CV is sensitive towards pON. Combination that the fraction of pON reported by AMS and ACSM diverge, indicating that a single correction value for the fragmentation table may not be applicable, and that the scatter plot (Fig. 8) is really driven by two points (e.g., the values before limonene was injected, which is ~0, and the values after limonene was injected, which could be averaged into one point). Thus, the analysis from this one chamber experiment is suggestive that the CV TOF-ACSM may not be able to quantify pON and would potentially over attribute nitrate signal to pON instead of ammonium nitrate (by ~50-60%). Further analysis of this one experiment, or analysis of another chamber experiment with different chemistry, if possible, is needed to better understand the uncertainty and whether is is a precursor dependency and/or uncertainty with a constant fragmentation table correction.

Please find my comments in the attached pdf file. 

In their manuscript “Development and validation of a NOx+ ratio method for the quantitative separation of inorganic and organic nitrate aerosol using CV-UMR-ToF-ACSM”, Nursanto and co-authors present and evaluate a method to extract quantitative information on the fractions of organic and inorganic particulate nitrate from unit-mass resolution data from a ToF-ACSM, equipped with a capture vaporizer. For this purpose, they analyze the ratio of the fragments at m/z 30 (NO+) and m/z 46 (NO2+), which is different for inorganic and organic nitrates. Since the capture vaporizer generates stronger fragmentation and consequently less m/z 46 signal, compared to that at m/z 30 and since this brings the m/z 46 signal closer to the limit of detection and requires an improved correction for “other” contributions to the nitrate-related m/z, the method to extract inorganic and organic nitrates from AMS mass spectra needs improvements and extensions. The method, developed by the authors, is clearly described in their manuscript and several validation experiments are presented.

The manuscript is clearly written and the developed method is clearly described with sufficient detail. The validation experiments and analyses are also clearly described and good to follow. The presented method is valuable for the growing ACSM aerosol monitoring community, providing a method to separate organic and inorganic nitrates in their data sets. The description and validation of this method fits well into the scope of Atmospheric Measurement Techniques.

The manuscript shows a number of (minor) technical issues, which should be addressed before publication. In addition, my major concern is, whether the limitations and uncertainties of the method are fairly addressed. With uncertainty ranges of frequently 100% and above, the method does not necessarily provide robust and always meaningful information on the quantitative contributions of organic and inorganic nitrate to total nitrate. I do not think that this limitation is adequately addressed in the manuscript. Please see my detailed comments regarding this issue.  I think, after addressing these issues, the manuscript should be published in AMT.

Detailed comments:

L9: Providing a number for the concentration limit (0.6 µg/m3) only makes sense when the associated averaging time is also provided here (same comment for line 65/66).

L12: I doubt that “good” is the right word to describe this correlation. While the correlation is tight (high r2), the CV finds almost 60% higher pON fractions, compared to the SV. It would be informative to the reader not only to provide lower nitrate concentration limits but also information on accuracy and uncertainty of this method in the abstract.

L13: This sounds like that the presented method is universally usable for these instruments. It would be desirable that a statement is included in the abstract, stating that for each instrument (and potentially even tuning-specific conditions of the instrument like aerosol beam alignment or vaporizer temperature) the method has to be adapted.

L38-39: I suggest rephrasing this sentence to make clearer that the combination of the AMS vaporizer and ionizer interactions with the analytes results in different fragmentation patterns, i.e., stress the process that leads to the fragmentation pattern instead of presenting it as an inherent feature of the different nitrates.

Generally, the NOx+ ratio is not just different between inorganic and organic nitrates, but between nitrates with different volatility, e.g., between ammonium nitrate and other, less volatile, inorganic nitrates (e.g., KNO3), which also show a larger NOx+ ratio.

L42 (Eq. 1): Also introduce “C” in the text.

L50: ACSM means aerosol chemical speciation monitor (see Aerodyne website).

L58-59: While the CV is actually intended to improve quantification, the IPL is intended to transmit particles up to 2.5 µm into the instrument, not to improve quantification.

L78: What do you mean with “variation of empirical NOx+ ratio for pAmN”?

L100: I assume, 525 °C is the vaporizer temperature, right?

L101: The vaporizer is centered on the vaporizer? Reword.

L112: According to the URG website, this cyclone has a PM2.5 cut-off at 3 lpm flow rate.

L129-131: Why are most LODs lower for this instrument at 2 min averaging time, compared to the other one with 10 min averaging time? Do the instruments generally behave differently due to different measurement history?

L151: I suggest rewording to “… assumed to be exclusively of organic origin.”

L152: The “mass concentration of organic fragment at m/z 30 and m/z 46” does not sound correct. There is nothing like “a mass concentration” of individual ions (even though it is clear, what you really mean). I suggest rewording to something like “… the signal contribution of organic fragments at m/z 30 and m/z 46”.

The same comment holds for Line 157.

L160: “A multiplyer … is added” sounds odd. Better “is included”.

L163: The a_Org[x] multiplier in Table 1 is potentially different for every instrument, depending on e.g. particle beam alignment or vaporizer temperature or instrument history and potentially also dependent on the type of organic aerosol measured (which might affect fragmentation patterns of the organics). This should be made clear. As it is written right now, it sounds that a general multiplier can be used.

What is a “CV inlet”? In my understanding the CV is part of the analysis section of the ACSM and not part of the inlet system. Same: line 164.

L165/166: This sentence is not correct. It sounds like that a small fraction of the signal at m/z 29 has a relationship to m/z 30. This is not what the frag table means. Furthermore, the frag table does not deal with correlations but with relationships between m/z-related signals.

L166: may be better “larger contributions of organic fragments at …”.

L167-169: I would also argue that because of the greater NO3 fragmentation in CV (and consequently smaller remaining NO2 (m/z 46) signal fraction) the correction of m/z 46 for organics contributions is much more relevant.

L171: Why SOA? Does POA not contribute to m/z 30 and 46?

L179: Table S1, third and fourth column and Table 1, fourth and fifth column.

L179/180: Are really the whole mass spectra correlated - or not rather only the signals at those m/z which are under investigation here (e.g., m/z 29 and 30 as well as 42/43/45 and 46)?

L186/187: This is not true. Fig. 1a and 1b show the correlations between m/z 30 and m/z 29 and between m/z 46 and m/z 45, which apparently are the best correlations of all correlations that were tested. This Figure does not, however, show that these are the best correlations since the other ones are not shown here. (Same comment line 191/192).

L205ff (Section 3.2): Are all these results in this section generated also with a CV-ACSM or was a different instrument used in these studies?

The massive differences, especially for the a_Org[46],[45], which span almost over an order of magnitude and which probably would directly translate in NOx-ratios that span over a similar range, are a massive limitation for the presented method.

This must be discussed and the resulting limitations of the method must be assessed. Is there a potential way out of this issue or does this mean that this method will not provide results better than the order of magnitude of NO3_Org and NO3_AmN for an unknown aerosol?

L220: The NOx ratios for these two (nominally identical) instruments are very different from each other (more than a factor of two) while having very small individual uncertainties. What causes these huge differences? Different histories of the vaporizers? Different particle beam alignment? Different vaporizer temperatures? Different tuning of the instruments? All these influences have the potential to change over time.

I think it is crucial to know what causes such large changes in NOx ratio in order to have a robust method to calculate the different nitrate fractions.

L230: This sounds like there is a strong dependence of the NOx ratio on vaporizer temperature - is that the case?

L268, Figure 2: It would be helpful if the nomenclature used in the Figure and in the Figure caption would be the same. Furthermore, partially the description of the Figure does not agree with the Figure itself - e.g., that the "NOx+ ratio" is shown in panels c and d.

L271-272: Strictly, the range of ROR or the lower range of R_pON should include values where R_pON is zero (values below zero are physically not reasonable). Then there is an extremely large uncertainty in the determination of R_pON with this method.

L280ff: I am not very convinced about the robustness of the determined R_pON, calculated from an upper limit that is 35 and 72 times as large as the lower limit, which, on the other hand, is arbitrarily set to the same, very small, value. The consequences of this approach are two R_pON values which are very similar to each other, suggesting a good agreement, while upper limits as well as R_pAmN differ by a factor of two and there are 1.5 to 2 orders of magnitude in the range, determined for R_pON.

At least a reasonable analysis of the uncertainty for this approach is needed, that includes the uncertainty of this approach but also uncertainties due to the reasons which lie behind the differences (factor of 2) between the individual instruments. I would not be surprised if this results in an overall uncertainty in the order of a couple of hundred percent for the separation of AmN and ON.

L292-294: These two sentences seem to contradict each other. How are the DL for the NO2+ and NO+ provided? Since there is probably nothing like 0.044 µg/m3 of NO2+ ions anywhere, these DL only make sense if they are given with relation to ambient NO3 concentrations. This, however, would only make sense if the different magnitudes of the two related signals (as mentioned in the following sentence) are already accounted for in the DL calculation.

I suggest rewording this paragraph.

L302: As mentioned before, it seems not reasonable to use the term "NO+ concentration" for the NO3-related signal at m/z 30. There is nothing like an NO+ ion concentration and definitely nothing in the order of a few ng/m3. Better use "NO+ signal intensity"

L305-307: I agree that with this criterion it is possible to obtain reliable NO3 concentrations, however, if it is not possible to determine the ON fraction because the signal at m/z 46 is too low to reliably determine the ON contribution to it, how is it possible to calculate the fractional AmN contribution to it?

L332, Figure 3, caption: I wonder how this apportionment can be called "reliable" if for all concentrations the uncertainty range starts at a fraction of 0 and for almost all concentrations it ends at a fraction of 1 for f_pON. I would say that this means that the ON fraction of the nitrate is largely unknown.

L359: There is not “proportionality” observable in the respective plots. The f_pON just increases with increasing Org fraction and the f_p_AmN increases with increasing NH4 fraction, however the values are not proportional to each other.

L370-371: How do we observe that this method is able to separate the ON and AmN contributions to total nitrate? Just because it produces results which are only during a fraction of the time chemically impossible? Is there any evidence that this separation reflects reality? E.g., why is for the autumn event (the only one with higher Organics than Nitrate concentrations) the average fractional ON contribution the largest of all four examples? Why does R_obs not seem to reflect the ratio of ammonium to organics?

L374, Figure 5: In the Figures it looks like a third up to half of the data points show fractional contributions below zero or above 1, which is chemically impossible. The uncertainty range covers the full possible range from 0 to 1 for most of the data. This does not seem like a robust information.

L383: How do you know that this RoR is correct for the type of ON, generated under these conditions. This might explain the differences in Figure 7c.

L385, Figure 7: Why was not the same averaging method used for the CV- and the SV- instruments?

L410: A “successful separation” was shown. For ambient concentration levels, there is no indication that the calculated separation reflects the actual ambient separation of ammonium and organic nitrates. This, e.g., could be done using co-located HR-AMS measurements with SV.

L426: As this lower limit (for R_pON) was set arbitrarily, can you determine how the results would change if this lower limit would have been selected differently, e.g., lower by an order of magnitude?

This would provide the reader with an idea how crucial this (arbitrary) selection is.

L444: It would be desirable if a statement about uncertainty and accuracy of this method would be clearly given: How well does it work for separation of AmN and ON in ambient measurements?