Overall Comments 

The Authors present the results of measuring amines by an established liquid chromatography (LC) fluorescence derivatization (FLD) technique that they’ve optimized for their needs in-house. The performance of the LC-FLD is compared to an LC-mass spectrometry (MS) determination following additions of high quantities of amines to real samples. The Authors conclude that the method is validated, but the analytical approaches are not consistent with validation of a technique. It is likely that they have the required data to properly validate their technique, pending revision, based on the experiments they’ve stated were conducted in the methodology.  

The results of the amine determinations are then paired with a number of air quality and meteorological measurements, then apply principal component analysis to perform a basic assessment of associations between these. They survey a small number of other reports from the literature, which could be improved. The initial emphasis in the article regarding the potential role of tanneries as sources of amines is not presented in the results and discussion, nor is a compelling set of measurements with respect to this potential amine source shown. It seems that the manuscript, overall, could use quite a bit of attention to detail in the narrative the authors wish to present alongside a more data-driven validation of their optimized analytical techniques. Specific comments provided by line number below should be addressed before this manuscript should be considered acceptable for publication in Aerosol Research. 

Specific Comments 

Page 2, Lines 43-44: They can also compete with NH3 in these reactions. The work of Bzdek should be mentioned here, as the authors draw on (or sometimes miss) the displacement concept elsewhere. There also needs to be a better job done in describing how gas phase sources of amines translates to their presence in the condensed phase. Reaction schemes that include phase notation could be helpful for this. 

Page 2, Line 45: See work from Denkenberger that performed some of the earliest work on SOA formation from amines.  

Page 2, Line 48: Toxicity depends on the dose, so this statement should be revised, or moved to a more relevant part of the introduction that discusses point sources where the levels of amines may be toxic. 

Page 2, Line 52: Just cite the paper. No need for added commentary. 

Page 2, Line 53: ‘biomass burning and degradation’ are the main continental sources, yes, but the Authors have missed the enormous source of amines in marine environments. Literature survey is not complete and providing correct atmospheric context. 

Page 2, Line 56: Should ‘consumption’ be ‘combustion’ here? 

Page 2, Line 57: ‘Tanneries may also represent a significant source’. Be more explicit, since this is a target industry in your work. What particular aspect of the industrial processes here are thought to be (or known to be) using substantial quantities of amines? In the results and discussion, this idea is not revisited, so maybe it is not relevant at all and can be removed from the introduction? 

Page 3, Lines 67-70: Reading the literature selectively. For example, IC methods have shown substantial improvement. See further work from VandenBoer on this topic, led by Authors Place and Salehpoor, where the noted issues have been overcome. The former work is 6 years old already. 

Page 3, Line 75: This paragraph on atmospheric measurement techniques is missing atmospheric gas sampling of amines by mass spectrometers like the PTR-MS? Not mentioned, but fairly prevalent in the literature. Overall, the other instrumental approaches to measuring amines is not particularly well-synthesize with respect to the advantages and disadvantages each one presents to the atmospheric analytical chemist. 

Page 3, Line 80: ‘simple and fast derivatization step’. Be more specific. Are analytes added directly to aqueous extracts and then analyzed? I.e. one pot? How fast? And what is 'mild' in respect to? Temperature, pH, something else? And contrast this against the general properties of the other derivatization methods. Give specific examples. 

Page 3, Lines 83-84: ‘UV-vis absorption and fluorescence’. Need to state specifically why these are advantageous analytical techniques over something like a mass-spec or conductivity detector. 

Page 4, Line 111: BET measures mass loading. Be specific in how this measurement was used alongside your PM10 samples (or if your sample filters were taken directly from the BET sample impaction setup). 

Page 4, Lines 117-118: Are the artifacts in the gas or particle phase? For the Shen reference regarding DEA here: If this is not clear and decisive, then it may be worth doing a more general description of the potential for analytical bias. If there are different reasons and none are particularly conclusive, then Shen et al could not have made a well described argument. 

Page 4, Line 131: ‘0.1 M HCl’ Give rationale. To keep analytes protonated and non-volatile? 

Page 4, Line 133: The extraction volume of 3 mL is larger than the size of this 1.5 mL test tube? Please revise the procedural details for accuracy. 

Page 5, Lines 135-140: Molar concentration is more important to communicate. For example, it would help here to communicate that the derivatization reagent is present in excess. Similarly for the buffer capacity and reaction condition stability molar concentration and pH of the borate buffer should be given. Revise. 

Page 5, Line 140: How was the temperature control achieved? Heating mantle? Heating block? Please add. 

Page 5, Line 145: Autosampler details should accompany the injection volume at the start of the sentence. 

Page 5, Line 146: Are absorption/emission profiles for all of the analyte derivatives possible to present in the SI? 

Page 5, Line 151: Concentration of B was increased linearly? After a hold at 40% for 1 minute? The gradient program could be written more clearly or placed in a supporting table. 

Page 5, Lines 151-152: ‘elution mixture’ should be ‘mobile phase composition’ 

Page 5, Line 153: In atmospheric samples, the moles per volume of gas sample are much more useful for the community of researchers. As such, it would be most helpful to have calibration data also presented using units of moles instead of mass, since the ranges are not actually the same for each analyte on this basis. Or at least give the mol/L concentrations for this first standard, so people can get an idea of what this mass per solvent volume value is equivalent to. 

Page 6, Lines 198-199: The reaction scheme/equation for amines, generically, should be presented somewhere, alongside the necessary reaction conditions, as stated above. 

Page 7, Line 205: ‘in about 20 minutes’. Performance is being oversold here. The run is nearly 30 minutes long as per the x-axis in Figure 1. 

Page 7, Line 206: ‘of its slightly higher efficiency’. Be quantitative in what you mean by this. Better peak shape? Better N? both? Would be valuable to show the results of the three different separations on a single figure in the SI. A table of performance metrics would also be valuable. 

Page 7, Lines 210-211: Were these by-products identified?? Is the derivatization of NH4+ an asset of this method that should be presented more explicitly? Can these be labeled in Figure 1, alongside the presentation of the chromatogram for a reagent blank and method blank? This would be useful in communicating how ‘clean’ the derivatization is versus how important a high-efficiency separation needs to be when implemented alongside this derivatization. 

Page 7, Line 222: Volume comparison is incorrect to use here. The derivatization depends on the stoichiometry, and here, it seems that you are trying to make sure this is present in greater excess than prior work. That decision needs to be justified using molarity of derivatization agent, and expected range of molarity for analytes in sample extracts at a minimum. 

Page 7, Line 224: The extraction of amines into water has been long established as a quantitative technique. What was actually done here and where are the analytical results of the different tests for time, temperature, etc using control solutions of known composition? These should be tabulated at a minimum, but could probably use some figures (see further comments below). 

Page 7, Line 230: How was this extraction comparison done? Data to prove this? What analytes were assessed? 

Page 7, Line 231: What was the criterion to conclude that the reaction was complete? Where is the data of that criterion versus reaction time, and with a trace for each temperature, to clearly depict this? 

Page 8, Figure 1: Known byproducts should be tagged and described in the manuscript, as well as in this figure. That way, the presence and abundance of otherwise unrecognized amines in the real sample trace can be better communicated through the chemical selectivity of this method. Can panels a and b combined with the traces offset? The axes are also not the same line colour or width, with matched font as the data trace or the labels. Suggest revising for consistency. 

Last, why are the details of the one depicted sample relevant for this plot?  

Page 8, Line 239: ‘that may represent a serious chromatographic interference’. Another reason to present a trace of FLD Intensity versus retention time for an analytical blank to show this. Can be added to the offset traces in Figure 1. 

Page 8, Line 245: ‘faster’ is not really analytically descriptive. What about fewer sample handling steps, meaning higher method throughput? Derivatization completion is not effectively demonstrated to support this statement, nor is the quantitative capability (i.e. recoveries need to be properly assessed, and/or internal standards used to track derivatization, as per further comments below). 

Page 9, Line 248: ‘were estimated’ Estimation is not appropriate for method validation. They should be (and appear to be) measured experimentally. Suggest revising here. 

Page 9, Line 253: ‘six amines at 10 mg/L’ Is this relevant? The final concentration in the analyzed sample is better to present here. Why couldn’t the actual sample and its amine concentrations be used? This would be more reliable in representing reproducibility, as the matrix effects would be properly captured. Understandably, if the entire amine suite is not in the sample, spiking could be a way to evaluate those in the real matrix, but it needs to be clearly presented that the spike levels are reasonably approximating the range expected in real atmospheric aerosol extracts. 

Page 9. Line 253: ‘the recoveries’. This is a test for matrix effects, not recovery. Recovery refers to spiking a known quantity into a sample and then assessing the amount using your typical calibration approach, then calculating the relative error. Revise. 

Page 9, Line 264: ‘also by better accuracy’. Better accuracy has not been demonstrated by any of the presented metrics. Recovery is not an assessment of accuracy. Revise. 

Page 9, Line 266: This discussion is incomplete. Where are the upper and lower range assessments of RSD? These were supposedly done with two different concentration standards, to represent upper and lower limits. How relevant is the calibration range to the quantities found in actual samples? Recovery and matrix effects need to be better explained. Last, the discussion here also needs to indicate the limitations of this approach more fully (complex chromatogram, cannot detect triply substituted analytes, etc.) in the context of the literature, not just other methods reporting use of this derivatization agent. 

Page 10, Line 277: Was butylamine used as an internal standard in the fluorescence detection as well? Why not use mass-labeled IS here? Is an IS even required in the FLD approach? Could that fact be used as a point of analytical strength versus a much more expensive technique like LC-MS? The use of a common residual fragment here from the derivatization reagent seems risky and the selected qualifier ions for each amine are not presented in a table of the SI to demonstrate that the choice of quantitation ion has been done well. 

Page 10, Line 281: ‘measured concentrations of the amine derivatives were statistically equivalent’ This does not represent a true orthogonal assessment of method performance, and therefore validation. Spiking the analytes into your samples at high concentrations instead of at their actual environmental levels is a weak evaluation. 

Page 10, Line 292: This is a self-contradicting statement. Simply say that butylamine 'was not detected' 

Pages 10-11, Lines 299-313: Amine mass loadings have been reported in a large number of studies in recent decades. Why are the specific studies cited here relevant for comparison? Suggest revising the writing here to be more specific or more appropriately referencing the literature (i.e. a review paper that broadly discusses atmospheric mass loadings, like Ge et al). Suggest reducing this discussion of each paper separately since the comparisons that have been selected are arbitrary. It would be more appropriate to conduct a thorough review of comparable and contrastable measurements of atmospheric amines in aerosol samples, supported by a carefully constructed table to present the relevant parameters and amines. 

Page 11, Line 319: Agriculture has been well-reported in the Po Valley to be a driver of aerosol composition. Suggest reviewing more of the prior literature from this location, including potential amine sources. 

Page 12, Lines 335-338: This argument is incorrect. Sulphate must first be neutralized by reduced nitrogen before nitrates can be incorporated significantly into secondary aerosol. This has been established in textbooks for a very long time. Since literature is not being cited, the Reviewer is speculating that the Authors are referring to some recent high impact reports about new particle formation coupling reduced nitrogen species and nitric acid, which was not shown to partition a substantial mass to the condensed phase. 

Page 12, Line 344: Need to reference the literature here. 

Page 12, Lines 345-353: This paragraph starts off alright, but then becomes disorganized and difficult to follow, with lots of speculative statements. Suggest revising to make this more concise and contextually broad to lead into the use of PCA for other amines other than DMA (e.g. cut the statement about calcium, also the discussion of the PCA results, since that is the subject of the next section). Also, take some space to communicate the strengths and limitations of source apportionment techniques, so that your results can be contextualized. 

Page 12, Lines 354-364: Missing here is that the meteorology (T, RH) and concentration of the direct precursors (e.g. NH3 and HNO3) dictates their partitioning to the condensed phase on timescales much shorter than those at which the observations were made (i.e. minutes to hours, versus days). This discussion is rather superficial and does not do a good job relating the PCA results to known phase-transfer mechanisms that underpin aerosol production and growth. 

Page 13, Line 371: Conclusions and abstract should be entirely revised based on the revised contents of the manuscript. 

Table S2: Need to indicate in the caption how measurements below the LOD were treated and presented here. 

Figure S2: Why is no time series of the amine measurements for the three different sites being shown alongside other important components that dictate their presence in particles? 

Table S5: These are spiked values. The sample concentrations should have been determined by using the standard addition approach, not by comparing the concentrations obtained after adding an easily detected concentration to the samples. This is not realistic. 

Table S6: Table formatting is inappropriate for a manuscript, even in the SI. Revise for proper high quality presentation of results.