Review report – AMT-2023-149

The manuscript proposed by Dandan LI et al. entitled “Ammonium CI-Orbitrap: a tool for characterizing the reactivity of oxygenated organic molecules“ presents the application and potential of a new instruments for the characterisation of a large range of gas phase oxygenated organic molecules (OOMs). OOMs are essential compounds involved in SOA formation and new particle formation processes, their characterization being one of the main challenges in atmospheric chemistry. In that sense, the paper is of great interest for the international scientific community. The paper is clear, and well-structured and contains valuable information. The methodologies regarding the experiments and state of art instruments used are well described, even if some precisions could be added to some extent. The interest of the NH₄⁺.Orbitrap is evidenced; but the results could be discussed more. As a consequence, I recommend the publication of the paper after the authors address the following points:

Main comments

1. 155: Figure S1 does not support the stability of amines, as it does vary over ca. 1 order of magnitude during the period the period shown on Figure S1. Also I do not understand why a time series of 15 days is presented on Figure S1 while the paper focuses on 2 experiments. It is clear from figure 1 that humidity and to a lesser extent T are affecting amines signal, and it might be discussed in the section 3.6 about RH dependence. The same time series corresponding to runs 2211 and 2213 would be more appropriated than a 15 days time series to support authors statement.

2 Experimental approach and product analysis: A brief comment on how instruments other than NH4+ orbitrap have been calibrated or how quantification estimates were performed is necessary. Even if some well-established methodologies exist. This can be part of Supplementary material if the authors do not want to make the manuscript longer.

2.3 How do the authors differentiate a peak from the background? In online-MS studied; it is commonly assumed that a peak is detected when its area as 3 times higher the standard deviation of noise. Is it what has been done using ORBITOOL?

Section 3.1: I am not sure this part is necessary, because this is an illustration that an instrument with a high mass resolving power separates more easily isobaric compounds compared to instruments with a lower mass resolution. Any scientist able to understand what a mass resolution of 160 000 compared to 10 000 means is convinced that the first one is far better for separating isobaric compounds (without any demonstration needed). The interest of the paper is not the mass resolution of the orbitrap but its association to NH₄⁺ as CI. Finally, if the authors find a justification to keep this section, I recommend them to normalize to 1 the Y scale each plot of figure 2.

Section 3.2 must be improved based on comments below:

L.287 288: Does it make sense to compare NH₄⁺.orbitrap to another instrument (PTR-3) that is not optimised to compare OOMs? The authors showed the NH₄⁺.orbitrap is more suited for OOMs detection, but the comparison is not on an equal foot with the PTR-3. Maybe the latter should be excluded from this study?

1. 288-289: As mentioned, the quantification limit of the I-.CIMS is higher than gas phase OOMs concentration. Giving a detection limit for each compound detected by each instrument is probably unrealistic, but the information about the range of limit of detection/quantification for each instrument would be helpful for a reader not expert with all these instruments.

L.302-303: A R² > 0.5 alone is not a good criterion for “high correlation”, as it depends on the number of points associated to each sample, etc. In addition, the good correlation with other instruments could be explained by similar biases, for example. Please temper statements, or strengthen the statistical analysis.

L.353: It is not clear if the raw signal (i.e. counts) or concentrations have been used here? If raw signals are used, can the authors justify their choice? And would Figure 8 be different if the concentrations are used instead of signals?

Section 3.4: a simple comparison on couple of common compounds detected by both PTR-3 and NO3-CIMS would be nice to validate their quantification, showing there. Both instruments are used as reference to “calibrate” NH₄⁺.orbitrap, but are PTR-3 and NO₃-CIMS consistent when measuring the same compound?  In addition, the NH₄⁺.orbitrap falls in a factor 2 comparing with other instruments, which is satisfying and reasonable considering all the uncertainties associated with quantification on online-MS, but cannot be qualified as “good”, which is subjective term.

Section 3.6: the discussion is interesting here, but the results should be more detailed. For example, it is not discussed that intensity of C₈ compounds increased whatever the number of O atoms, while it is more contrasted for other compounds. In addition, this increase can be up to ca. 20 for C₈H₁₂O_2-4 compounds, while it is limited to 1.6 for C₁₀, C₁₉ and C₂₀. Is there an explanation here?  In addition, based on Figure S6, it seems the effect of RH is very important at n_O<8, but what about the effect of n_C? As I just mentioned, the effect of RH seemed to be stronger for C₈ compounds. The increased in polarity or O/C with decreasing C number might be an explanation? This must be investigated. Figure S6 also evidenced that I-.FIGAERO-CIMS sensitivity is only decreasing with increasing RH, while it is not the case for other instruments. The authors should comment this result. Finally, as the authors cannot distinguish the effect of increasing RH on chemical and physical processes (based on experiments presented in the present paper), it is evident that RH influences NH₄⁺.Orbitrap sensitivity, that can be different for each OOM, but this specific effect requires more attention and dedicated studies before the NH₄⁺.Orbitrap can be used in field studies (for example, injection of pure or mixture of standards in atmospheric chamber at varying RH). From what is presented here, the understanding of RH effect on the NH₄⁺.Orbitrap capabilities is too scarce to be able to understand the time series evolution of OOMs that would be obtained in the real atmosphere.

Minor comments:

Section 3.1: OVOCs should be replaced by OOMs, as most of detected compounds are not volatiles.

Figure 5: caption should be more explicit.

Figure 7: here the PTR-3 is probably limited compared its real potential to OOMs, because it has been tuned for efficient detection of NH4+…

1. 409: is.
2. 757: concentrations of OOMs

This work studied CI-NH4-Orbitrap as a powerful tool for characterizing oxygenated organic molecules (OOMs) from atmospheric oxidation of VOCs. The manuscript compares the performance of CI-NH4-Orbitrap with a few other chemical ionization based mass spectrometers with a range of ionization methods and resolving power. The comparison showed that CI-NH4-Orbitrap is a promising instrument to more comprehensively characterize and even quantify a near-complete range of OOMs from oxidation. The work is solid and well written. It will likely deserve publication at AMT. But some sections of the manuscript need to be better clarified and some in-depth discussion is needed.

Detailed comments:

1. Line 39 in Abstract. Change “highly oxidized volatile organic compounds (HOM)” to “highly oxidized molecules (HOM)”.

2. Line 56-58. OOMs can also be generated through bimolecular RO2 pathways not involving autoxidation. Autoxidation is important, but review OOM formation more compressively, other pathways should also be mentioned here.

3. Line 76-87. In the negative ion-based MS, it would be helpful to also mention iodide-CIMS, as it is compared later in the text. One sentence to set up the context would be a good idea, also because iodide-CIMS measures a wide range of OOMs.

4. Line 150. Change the sentence to “The NH4+ reagent ion cannot be directly detected due to…”

5. Line 156. In prior studies using the same NH4+ ionization but with a Tofwerk LTOF, is there evidence regarding the ratio of the sum of these “surrogates” over the reagent ion abundant? If this data is available, it would be useful to mention here.

6. Line 159-166. The uncertainties regarding this semi-quantification method needs to be discussed somewhere. For example, this method assumes that the C10H14,16Ox formulas measured by the three instruments are the same species without artifacts? Do dimers in PTR3 decompose to monomers? How can you obtain the calibration factor from correlation analysis alone (cps vs. cps between different instruments)? Do you need response factors (e.g., ppt/cps) from either the PTR3 or NO3-LTOF to get concentration results for Orbitrap?

By the way, NO3-LTOF is termed in this paragraph, but termed “CI-NO3-LTOF” or “CI-NO3-APi-LTOF” in other places. The terminology needs to be consistent throughout the manuscript.

7. Line 258-267. The dm threshold is also dependent on the level of knowledge of the possible chemical formulas at the normal m/z. In case the chemical formulas are known at high confidence (in the example of known VOC precursors), the threshold may be smaller. But in real cases where more than two peaks are present, the threshold can be larger. This is a very complex issue. The simplified illustration here is certainly useful, but some more in-depth discussion is warranted in real cases.

8. Figure 1. The last plot was not described in the caption.

9. Line 280. In the comparison between CI-NH4-Orbitrap and I-FIGAERO-CIMS, it is unclear that the large difference is number is mainly due to the less selectivity of NH4+ ionization or the higher resolving power of the Orbitrap. Some clarification is needed. The range of oxygen number seems comparable based on previous studies of iodide-CIMS (from nO=2 to HOMs). The detection limit issue mentioned in Line 289 seems to suggest that this difference is largely due to instrument sensitivity tuning issue for iodide-CIMS? If this is the case, the comparison does not really speak for the advantages of CI-NH4-Orbitrap in ionization method and resolving power.

10. Line 284. How was the O/C ratio estimated? 0.4+/- 0.2 seems to be a very large uncertainty. Is this due to the variation between the two experiments? Or uncertainties in the semi-quantification method? With the accurate formula detection, I would expect smaller uncertainties in O/C ratios.

11. Line 294. This number suggests a large fraction of the chemical formulas detected by CI-NH4-Orbitrap are not seen by any of the other three instruments. What are the characteristics (e.g., number of C, H, O, O/C, etc.) of the chemical formulas co-detected vs. only detected by CI-NH4-Orbitrap? Combining PTR3-TOF, Iodide-CIMS, and NO3-LTOF, it appears to me that the overall selectivity is comparable to CI-NH4-Orbitrap. If the difference boils down to the discrepancies regarding sensitivity (e.g., Orbitrap much better than the others and most of the formulas only detected by CI-NH4-Orbitrap are relatively small peaks) and resolving power (e.g., CI-NH4-Orbitrap detects formulas at high confidence, but the other instruments do not), I think it is worth discussing this difference to highlight the superior performance of CI-NH4-Orbitrap.

12. Figure 3. It would be helpful to draw a few lines indicating the major chemical formula series detected by the different instruments in the KMD plots.

13. Section 3.3. It is useful to describe a few major chemical formulas which have the worst correlations. Although it is not a chemistry paper, but providing such information can help others think about the chemical reasons behind these correlations.

14. Figure 5. The figure legend needs to be explained in figure caption. LTOF means NO3-LTOF? Orbitrap-LTOF means Orbitrap-derived concentrations using NO3-LTOF calibration factors? Does the NO3-LTOF actually measure oxygen number down to 2? Do the concentrations in Orbitrap depend on quantification by PTR3 (proton transfer kinetics) or NO3-LTOF (H2SO4 as the sole standard)? If so, they need to be mentioned. I believe that much of the unclarity stems from Eq. (3), as mentioned in my above comment #6. What are the units of c and [X]? If c is unitless (i.e., cps/cps from correlation analysis), and [X] is in concentration (ppt or molecules cm-3), the equation does not make sense because the second term in the right side of the equation is unitless.

15. Line 327 and Figure 6. What are the fractions of the reacted carbon measured by these instruments? Table S1 does not show the steady-state a-pinene concentrations, so it is not possible to estimate these fractions by audience. It is also unclear how the remaining formulas only detected by Orbitrap is treated here. Are they also quantified by the same calibration factors?

16. Line 372-376. Should this be due to increased partitioning of water-soluble compounds to the aerosol liquid water? If increased RH leads to partitioning of SVOCs, why did nO<5 signals increase? Presumably these C8H12O<5 species are SVOC and with the enhanced partitioning, their signals in the gas phase should decrease. The explanation is in contrary to the observation. The changing ionization efficiency and multiphase chemistry described in the following paragraph could be the main reasons. I suggest revising this section. The way it is written is confusing.