General Comments:

This manuscript presents a thorough comparison of multiple NTA workflows for analysis of water-soluble brown carbon aerosol samples. This is a much-needed contribution to the field, and additionally, the best workflow was used to identify new potential precursors to aqueous SOA. My scientific comments are minor, and I have a few comments to address clarity and technical issues with the manuscript. Overall, this manuscript is worthy of publication in ACP.

Specific Comments:

It is not clear how model compounds were selected. Were these compounds chosen from studies focused on aqueous brown carbon, or on biomass burning in general? Or perhaps they were chosen because of commercial availability? Since workflow performance was assessed on the ability to correctly identify these compounds, it seems important to communicate how exactly these standards were chosen and cite references that informed the choice, if applicable.

I would welcome the addition of a comment from the authors about how well their combustion apparatus simulates real BB conditions, but do not feel it is absolutely necessary prior to publication. I am always a bit skeptical of these small-scale reactors in terms of their applicability to real conditions, and I am even more skeptical given this is a brand-new combustion system that has not been previously characterized. That being said, I don’t think the very nice NTA work done here is invalidated by the choice of combustor.

I found Figure 9 to be confusing – if the unlabeled areas correspond to unidentified molecules in each group, how are the authors determining that those molecules belong to that group? The information shown in Table S4 and S5 only added to my confusion. It seems like a reference to Tables S6 and S7 here would be more appropriate.

Technical Corrections:

I found Table 1 to be confusing in the sense that it implies no identifications were made at the highest level of confidence (e.g. Level 1). I suggest the authors add an example in the first row, since the SI shows that the method successfully reached Level 1 for a variety of compounds.

Several times throughout the manuscript, something to the effect of “level ≥ 4” is used. The context of these sentence conflicts with the literal meaning. In other words, what I think the authors are attempting to communicate is ‘a level of confidence higher than 4,’ which would correspond to a level with a numerical value ≤ 4. This occurs at lines 279, 344, 367, and 373 (and perhaps elsewhere). While a reader familiar with the Schymanski confidence level scheme will likely understand what is being implied, I suggest these be reworded to be more clear for those less familiar with NTA.

Line 290: missing a space between the words “abundant” and “in”

Line 293: I suggest spelling out molecular weights and organic aerosols to be more clear.

Commas in strange places in lines 299 and 301

Remove “identified” before the comma in line 303

Line 307: “characteristics” should be “characteristic”

Line 311: By the time I got here, I had totally forgotten what “STs” were. Consider not using an acronym all to refer to the surrogate standards.

Overall, the manuscript is timely and provides an important comparison of multiple NTA workflows. This work will provide a benchmark for others to work towards when undertaking NTA and highlights the type of optimising and testing which should take place. The link with toxicity is a great step forward, especially highlighting the lack of correlation with general aerosol composition (O:C, H:C etc). 

However, while I agree with the approach of NTA optimisation and semi-quantification, I have some major comments regarding the LC-MS and semi-quantification methods used. Overall, the manuscript should be published in ACP once comments are addressed. 

Specific comments:



L125- What mobile phase (A or B) does the 5% apply to? 

L135 – How were the methods optimised? Was it based on a mix of all 59 standards and/or spiked into the sample matrix? 

L137 – How were the 59 organic molecules determined? Based on previous literature and/or commercially availability? How representative of the overall composition are the standards? Are you biasing towards this functionality? 

L155/6 – Not entirely sure what this means. How was the structure annotated just based on MS1? 

Section 2.5 

You have run 59 surrogate standards for library building/identification. When authentic standards were identified (level 1) in your BB samples, did you use the authentic standard for quantification? If so, why not? 

Could you have assessed the error of the semi-quantification technique based upon comparisons between authentic and surrogate standard quantification?

How were the 5 surrogate standards chosen?

Why does tetradodecanoic acid cover the tr range of above 35 minutes with a retention time of 26 minutes?

Why weren’t all 59 standards used for semi-quantification? Your tr range used by each standard is large, including more standards would have narrowed the windows.

Did you spike the standards into the sample matrix to determine matrix effects on concentrations given the sample complexity?

Section 3- Many of the analytes will ionise in both the negative and positive mode. How were these species dealt with in terms of final concentrations and composition? 

L235 – What sample did you use to determine this? I.e a mixture of standards/spiked sample 

L239 The use of “identifies” here makes it seem like this is a known compound, maybe change to annotated?

Fig 6 – How representative are your standards to the overall detected composition?