Review of Keller, Specht, Steigmeier, and Weingartner: a novel measurement system for unattended, in situ characterization of carbonaceous aerosols.

Keller et al. present a first description and characterization of a novel instrument, the FATCAT. Relative to other first reports, I found this manuscript to contain far more than the bare minimum necessary to demonstrate the value of the FATCAT. The data presented are excellent and the system could be an extremely valuable contribution to the field. 

I have several suggestions to clarify the discussion. I recommend publication after addressing these minor comments. 

1.

I did not understand why the authors claim that FATCAT must be operated with an analytical synthetic air gas. The CO2 background of ambient air cannot possibly change fast enough to cause baseline problems. Why not just fit and subtract a baseline? The Magee TCA08, a similar instrument, does this. 

2.

The useful range of the instrument is "above baseline CO2" and "below saturation of the CO2 detector". This means that any lower detection limit corresponds also to an upper loading limit. It may be clearer to state this in the abstract? Then the reader can understand the "operating range" available at a given flow rate and sampling duration. Also, Section 3.3 discusses the upper limit in a confusing way, mentioning only the CO2 sensor's range and not the loading parameters (flow and sampling duration). This could be clarified.

3a.

Line 43: "per definition, only organic aerosol contains carbon-hydrogen bonds" but a few lines later "even soot...contains hydrogen and other elements". The first statement should be removed, there is no definition of organic aerosol like this. The second statement should be updated to "even soot ... contains oxygen and hydrogen" because the oxygen is a much more substantial mass fraction. Then at line 259 the authors state, "by definition, elemental carbon consists exclusively of carbon atoms" which is in contradiction to their earlier acknowledgement that elemental carbon always contains hydrogen (and oxygen). 

It seems like this is a minor error that crept in during revisions of the manuscript. Please remove all statements implying that soot or EC only contains carbon, and ideally add further citations showing that soot contains oxygen and hydrogen.

3b.

In fact, the FATCAT measures TC and the TEOM measures PM. Therefore, the comparison of FATCAT and TEOM has provided a measurement of the mass fraction of carbon in the sample, as mentioned in #3a above. The authors' slope of Figure 4a, 0.94, means that 94% of the sampled mass was carbon and 6% was oxygen plus hydrogen. This 94% compares well with the values of 90% and 93% reported by Corbin et al. (2020), and the values of 90% to 98% quoted in that paper for other literature studies. Therefore, FATCAT/TEOM appears to be an accurate technique for measuring this important quantity, which is required for converting TC measurements to PM mass! 

The authors mention that the 94% carbon fraction is close to the EC/TC of 0.91. That is true, but this is purely a coincidence. Both EC and OC = TC - EC contain carbon and would be measured by FATCAT. This is also explained in Corbin et al. (2020) Equation 6.

By the way, I recommend changing from "91%" to "0.91" for the EC/TC here and in the tables, because it may confuse readers. The sample contained 94% carbon, and that carbon was divided into operational parameters EC and OC. The EC/TC definition is not a physical one and not even thermodenuded soot is "100% EC".

4. 

It is only the Magee TCA08 that would define OC as complementary to eBC. The eBC definition of Petzold should be quoted here, which is a consensus definition. The danger of defining TC = eBC + OC is clear: coatings can cause eBC to be up to 2x larger due to "lensing" or absorption enhancement, and then the definition breaks down. I recommend quoting Petzold et al. (2013) and avoiding partial definitions to avoid confusion.

5.

I was surprised that the authors did not emphasize the improved sensitivity of their device in the introduction. The device has a much lower limit of detection than the thermal-optical analysis to which the authors compare it. This is a significant benefit for e.g. instrument calibration. The FATCAT is so sensitive that it probably cannot sample in parallel to thermal-optical filter samplers.

6.

I believe that it is not the use of an oxidizing atmosphere that prevents pyrolysis (line 368), but the rapid heating protocol. Pyrolysis reactions occur in competition with evaporation and oxidation, and rapid heating means that pyrolysis "loses" the competition. The authors have stated the opposite. I request that the authors add a citation here (or earlier in the manuscript) supporting their opinion, or remove this statement in the absence of clear evidence.

MINOR COMMENTS

Line 68, "Problems like..." were these issues shown to be the cause of problems, or were they simply listed speculatively? I do not believe that the listed issues are demonstrated problems in TOA. Please cite clear evidence if I am wrong, or remove the list of speculative statements.

line 51 onwards only defines thermal refractivity methods in the atmospheric sciences. Materials and other (e.g. soil) scientists also use thermal refractivity methods. Please add "In atmospheric science" before this paragraph.

Line 74 states that "Light absorption methods for measuring eBC are prone to systematic errors". This is not true, light absorption can be measured accurately and without error (note that "lensing" effects are not errors, but physical phenomena). The authors only cite studies on filter-based attenuation photometers after this statement. Perhaps the authors meant to state that filter-based attenuation is prone to systematic errors? 

Line 99, "all OC" should be "all TC". For example, carbon monoxide is not OC.

Line 128, give pressure of sampling site, not just m.a.s.l.? 

Line 174 a comment about the setup for another experiment does not seem to belong here.

Line 193 C1 C3 M1, are not yet defined and the heading doesn't match the text.

Line 211 please expand on "to increase the oxygen content". I understand that FATCAT needs it for combustion, but how does the user know how much dilution is needed?

On line 268 the authors speculate that the TEOM adsorbed OC3.  My feeling is that this is unlikely. See Subramanian et al. (2004, https://doi.org/10.1080/02786820390229354). The speculation here should be removed or made more quantitative, for example, it currently compares the 50 C TEOM with the 250 C OC3 -- totally different temperatures? In fact the results appear quite accurate (see #3b above) but the importance of the problems discussed by Subramanian should indeed be assessed. Ideally, in a separate section.

Line 275: there is no need to refer to thermal-optical methods for the term thermograms, the term is used in other techniques as well. 

Line 288: consider citing other authors in addition to Kelesidis et al., e.g. Maricq (2014, doi:10.1080/02786826.2014.904961).

Line 300 and related: I would recommend calculating the OM/OC ratio, a common metric, to place this discussion in better context.

Line 306: "This is not evident"? Meaning "the collapse of the soot core did not affect the thermogram"?

Figure 6 is discussed by reference to Figure 5's thermograms. Therefore, it would be helpful to overlay one Figure 5 thermogram over Figure 6.

Line 337: Thermograms of homogeneous samples are also not well separated. Here, reviews of thermo-optical analysis could be cited.

Line 370: I would recommend that the authors consider building a library of source signatures, in order to assist with the future interpretation of FATCAT thermograms for source apportionment.

This MS presents a new method to measure carbonaceous aerosol on-line long-term with the added bonus of quick thermograms. The idea is excellent and probably will help obtain data on carb. aerosol in a variety of locations additional to the established sites of monitoring networks. The method is fairly straight forward and does not need complicated data processing procedures (at least not when TC is the desired analyte), which is a very good thing.

As the paper describes first lab tests, it is in essence a proof-of-concept paper, and I am looking forward to see the companion paper currently in preparation where data from actual atmospheric measurements will be shown. I do agree that the work is better described in two separate papers, as it is always good when a new instrument or method is described in detail. A combined paper would either be too lengthy or would not contain enough info on the method. The current MS gives an excellent, in-depth description.

Generally speaking, the part dealing with the thermograms of coated/uncoated CAST soot and biomass smoke is most interesting, and the authors are a bit too modest in their claim that “it is still not clear what information can be extracted from the fast thermograms. But they seem to present a reliable and cost-effective opportunity to gather more information from real world carbonaceous samples" (p 11, lines 340 – 342) – just one look at the thermograms shows that carbon from different samples evolves at different times (i.e. temperatures) and gives thermograms of different shapes, which can be used to derive signatures of e.g. Diesel soot (not tested here, but could easily be done), aged particles and biomass smoke. The current use of thermo-optical analyzers does not utilize the info contained in the highly complex thermograms obtained during sample analysis; only  numbers on the different OC and EC fractions are given. The new instrument gives quick thermograms that can be analysed fairly quickly (or even automatically – AI could help ….)

I therefore recommend acceptance after the small points raised below have been dealt with

Points:

A few pieces of info are missing.

• What is the size of the exposed filter surface area? The LOD’s are given in µg C, translated to µg-C*m^-­³, but it would be interesting to see how this compares to the LOD of thermo-optical methods (usually given in µg-C*cm^-²).
• Which types of firewood were used in the test of biomass smoke? Could have an influence on the thermograms. The carbon emissions are different for different fuels (see Sun et al. 2021, Atmos. Chem. Phys., 21, 2329–2341 or Priestley et al. 2023, Environ. Sci.: Atmos., 2023, 3, 717)
• How long does the cleaning of the system with zero air take?
• Which quartz fibre filters were used for the TOT analyses?
• Was the same type of zero air used in all experiments?

Just a comment: it is really good to see that the effect of filter temperature on pore size is taken into account

A general comment: The English is very good, but please check the whole MS again for correct use of singular/plural “s” and “this” vs. “these”…..

Further comments are given in the order of appearance in the text.

Line 34: the words “sufficient accuracy” can refer only to TC – the whole question about accuracy of determination of EC (or any other component of carbonaceous aerosols) is still open – unless one considers method-specific definitions as “accurate”.

Line 55: the different protocols vary not only in the number of temperature steps, but also in the temperatures at these steps

Line 58: The EUSAAR2 protocol takes 17 minutes for running through a heating cycle, but the necessary cool-off phase takes another 5-10 minutes. I suspect that the time you give for analyzing samples with the IMPROVE protocol also does not include cooling off times, while sample analysis times with your method include cooling off.

Line 92: The statement "... show that these thermograms contain information only about the composition of the aerosol"  is too comprehensive – the thermograms contain info only about the composition of the _carbonaceous_ aerosol

Line 108: explain acronym FHNW (the authors’ address gives the English name of their institution)

Section 2.3: obviously (see Table 1) both a CAST and a miniCAST were used, but only the CAST is mentioned in the text?

Line 190: a nice typo: “tampered oscillating micro balance”

Line 243: explain acronym OEM

Line 268: change “... gas phase EC form the C3" to “…. from …”

Line 300: “ ... most of the organic material corresponds to OC1...." – this is not supported by the histograms shown in Figure 5c

Figure 5: contrary to figure 4, no regression line is given. Drawing a line through the three points for the coated sample seems to give a nice zero intercept – only the point for the uncoated sample does not lie on this line. Any ideas why?  The slope of the line through the points for the coated samples gives an indication of the mass fraction of carbon in the coated particles. Of course one could argue that four data points are too few to draw conclusions, but conclusions are drawn in the discussion of Figure 4, which also has only four points for each of the C1 – C3

Line 366: change “analogue” to “analogous”

Author contributions: EW is not mentioned?

References: Szopa et al refers to a chapter in the new IPCC report – which chapter?