This work utilized FIGAERO-LToF-CIMS for offline organic aerosol volatility characterization. The authors identified a series of CHO and CHON compounds from ambient samples and developed empirical volatility-molecular formula functions making use of the desorption thermograms that can be obtained by FIGAERO. This study also compared two different methods for laboratory standard compound calibration, which is useful information for FIGAERO users. This paper is overall well written and organized. One suggestion is that, both CHO and CHON were characterized for ambient samples, but it seems that they were treated the same in subsequent analyses. It would be helpful to label CHO and CHON differently in the figures. In addition, O/C ratio was used to distinguish compounds, which should also be discussed differently for CHO and CHON compounds. More specific comments are described below.

1. Line 14: “Because most standard particulate organic compounds are scarce…” This sentence is incomplete.
2. Line 86: the success of the methods depends on many factors, not only the standards’ thermogram characterization. Please clarify here.
3. Line 112: what is the heating temperature ramping rate? The ramping rate can have an influence on the thermal desorption/decomposition process (Yang et al. 2021), and more details here would be helpful.
4. Line 138: can the authors explain here how they determine the particle density?
5. Line 149: “It is assumed that the atomized particles were internally mixed with the same mass ratio as that in the solution.” Ammonium sulfate is much less volatile than organic compounds mixed with it, and it’s highly possible the AS/Org ratio in the particles is higher than that in the solution. More evidence or discussion of the potential bias is needed.
6. Figure 1: Besides fitted lines, can the author also add the raw data points from each measurement? In addition, can the authors add to the legend the corresponding experiment sets No. (as in Table 1)?
7. Line 235-240: sample NO.4 and NO.5 are different mixtures of different masses, making it hard to compare them. I suggest adding another set of experiments atomizing 500 ng AS + 500 ng Organics (each).
8. Line 285: 181 out of 1448 measured species were included in further analysis. I wonder how much of the total signals can be accounted for by the 181 compounds. Are they the most dominant compounds?
9. Line 306: “The data points for the higher-temperature ones in double-peak thermograms that in fact, do not correspond to a Tmax are removed.” How many, if not all, of them are removed?

References

Yang, L.H., Takeuchi, M., Chen, Y., Ng, N.L., 2021. Characterization of thermal decomposition of oxygenated organic compounds in FIGAERO-CIMS. Aerosol Science and Technology 55, 1321-1342.

General:

The authors present the calibrations of Tmax obtained from FIGAERO thermograms using mixed organic and inorganic calibrants, and investigate the effect of ammonium sulfate on the Tmax of several organic standards. Then they use the calibration result from the mixture of ammonium sulfate and five organic standards to derive a formula-based parameterization for the volatility estimation of the organic compounds measured in a rural area in China, and also compare this parameterization with previous parameterizations. Studies on the effect of inorganic species on the thermogram and Tmax behavior of organic compounds are important but scarce. From this point of view, the paper provides new input on this. However, the paper is for now a bit more technical sound, since the scientific discussion or application of this derived parameterization is not enough and feels unfinished. I would therefore recommend that this paper be published on ACP only after major revisions, with more scientific input into the paper.

Major:

1. Since the calibration results were the basis for the volatility parameterization for the ambient organic species and substantial discussions on Tmax changes of organic standards were stated to be due to the addition of ammonium sulfate, it would be important to separate the mass loading effects and matrix effects (caused by ammonium sulfate or within these organic standards). For example, Tmax of pure PEG-6 standard increased from 40 degC to ~50 degC with an increasing mass loading from 100 ng to 500 ng (for No2 and No3 experiment set, see Figure S3-S4), which could be due to the mass loading effect. The Tmax of PEG-6 mixed with ammonium sulfate (with a mass loading of 200 ng) didn’t change compared to that of pure PEG-6 standard (with a mass loading of 100 ng, for No3 and No4 experiment set, see Figure S4), which may exclude the matrix effect due to the addition of ammonium sulfate. But the Tmax of PEG-6 mixed with ammonium sulfate and other organic standards (with a mass loading of 1000 ng) increased to ~60 degC compared to the Tmax (~50 degC) of PEG-6 mixed with ammonium sulfate (with a mass loading of 200 ng, for No4 and No5 experiment set, see Figure S4). This increase could be due to several reasons such as the matrix effects within these organic standards, higher mass loadings of ammonium sulfate, as well as higher total mass loadings on the filter. Since the mass loading for the calibration experiments varied between 100-1000 ng, discussions on the effect of different mass loadings and potential different matrix effects on the comparison of thermograms and Tmax for this study as well as how this would influence the derived parameterization from ambient observation is necessary.
2. The derived parameterization behaves similar as Li et al (2016) and Stolzenburg et al (2018) for the 15 HOMs (O/C:0.25-1), but worse than Li et al (2016) and Stolzenburg et al (2018) for the 132 CHO (O/C:0-0.25). If this is true, I don’t see why we should use this parameterization from this study instead of Li et al (2016) or Stolzenburg et al (2018). More scientific discussions on this parameterization method is needed.

Specific:

Line 41 – What do the authors mean for the “particle-associated phase”? Is it different from particle phase?

Line 49 –SIMPOL is more a structure-based estimation method of vapor pressure instead of formula-based estimation method. Check a recent work by Isaacman-vanwertz and Aumont, ACP, 2021 (https://acp.copernicus.org/articles/21/6541/2021/). In their work, they also mentioned about Daumit et al (2013) method for formula-based estimation method and modified version of Li et al (2016) method. Please replenish/revise this paragraph.

Line 112-113 – Please add the flow for the UHP N2 to a 0.1 mCi radioactive Am-241 source, and the ramping rate for the heating.

Line 141 – It’s not very clear whether the authentic organic standards are mixed together within each experiment sets of No. 1-3, or they are injected/atomized one by one? For No. 4 and No.5, it’s more clear. Please specify in the texts or Table 1.

Line 156 – For Table 1, the mass loading for different experiments varies. According to Huang et al (2018; https://acp.copernicus.org/articles/18/2883/2018/) and Wang and Ruiz (2018; https://acp.copernicus.org/articles/18/15535/2018/), different mass loading on the filter could influence the thermogram shape and Tmax of organic compounds. It seems the author also observed similar behavior, if one compare the Tmax of PEG-6,7,8 from No.2 and No.3 experiment sets in Figure S3 and S4. Since both No.2 and No.3 experiment sets are done by atomization method but with different mass loadings on the filter, Tmax values are found to differ. For example, with 100 ng of mass loading Tmax of PEG-6 was 40 degC (Figure S4), but with 500 ng of mass loading its Tmax increased to ~50 degC (Figure S3). Since the study is based on the Tmax calibration, could the authors comment on the effect of different mass loadings on the comparison of thermograms and Tmax for this study as well as how this would influence the derived parameterization from ambient observation?

Line 163-166 – What’s the aerosol mass loading on the filter? Please add this information.

Line 170-190 – Could the authors comment on the uncertainty for the C* calculation as well as the uncertainty using this parameterization method?

Line 226-229 – From Figure S4 caption, it seems the No.3 set was done for each standard one by one using atomization method. If that is the case, the figure shows the PEGs 6-8 Tmax for No.4 set (AS+each standard) was similar to those for No. 3 set (Each standard). It thus would indicate the increase of PEGs 6-8 Tmax for No. 5 set (AS+five standards) is probably not (only) due to the matrix effects caused by the addition of ammonium sulfate, but due to the matrix effects within these organic standards. Besides, as the authors mentioned in Line 237-240, the Tmax increase could also be due to higher mass loadings for No. 5 set. It would be important to separate these different reasons, i.e., matrix effects within organic standards or due to the addition of ammonium sulfate, or mass loading effects.

Line 229-231 – Clear connection between more partitioning of organic acids and lower volatility of particulate organic compounds is missing. For example, partitioning of SVOC would increase the volatility of particles. Please explain/clarify a bit more.

Line 245-247 – As for the calibration line is similar to that of Nah et al (2019), do the authors mean the lines are quite close to each other? But the slope is a bit different from that of Ylisirniö et al (2021) and Nah et al (2019). Could it be due to Nah et al (2019) used acids with O/C <0.25?

Line 259-261 – It would be informative to add the fitted equation (or the fitted parameters a and b of Eq. (3)) of No.5 experiment set here or in Figure 1.

Line 264-267 – Please add the contribution of each group (CHO, CHON, other, unidentified) to the total signal in brackets after each group.

Line 280-281 –What are these dominated compounds and their potential sources? C6H10O5 could be levoglucosan from biomass burning. How about the others?

Line 324-330 – Would be nice to mark C6H10O5, C16H32O2, C17H34O2, C18H32O2, and C18H34O2 in Figure S7.

Line 338-341 – The authors mentioned in Line 280-281 about the dominated compounds including C13H25NO2, C9H17NO2 etc. It seems these CHON with nO<=2 are quite important. Would the Equation (4) still be applicable to them since the equation subtract 3nN for each nO? Besides, the Equation 4 is a modified parameterization in Mohr et al (2019) specified for HOM with big nO. Maybe the modified Li et al (2016) parameterization equation by Isaacman-vanwertz and Aumont, ACP, 2021 would be a better option considering the fits in Table 2 are mainly based on SVOC and LVOC?

Line 377-379 – Since Isaacman-vanwertz and Aumont (2021) found that the vapor pressures of CHON compounds estimated by Li et al. (2016) significantly biased with an increase of the number of nitrogen atoms, it would be important to add the comparison of the logC* (Formula) vs. logC* (SIMPOL) for CHON compounds as an additional panel in Figure 5.

Line 397-404 – In general Eq. (4-1) and (4-2) parameterization behaves better than Donahue et al (2011) and Mohr et al (2019). Based on Figure 5, Eq. (4-1) behaves similar as Li et al (2016) and Stolzenburg et al (2018) for the 15 HOMs (O/C:0.25-1), but Eq. (4-2) behaves worse than Li et al (2016) and Stolzenburg et al (2018) for the 132 CHO (O/C:0-0.25). If this is the case, could the authors comment on why we should use this parameterization from this study instead of Li et al (2016) or Stolzenburg et al (2018)?

Technical:

Line 17 – Change to “formulae” throughout the manuscript.

Line 30 – Change to “for organic compounds with different O/C ratios”.

Line 33 – Either say “a significant mass fraction” or “total submicron particulate mass”.

Line 110 – A typo for “UHP N2”.

Line 197 – Refer to Figure S3 when describing the results.  

Line 224 and 226 – Refer to Figure S4 when describing the results.

Line 350 – Either “O:C” or “O/C”. Make it consistent throughout the text.

Line 399-404 – Too long sentence. Please reformulate.

In the manuscript "Volatility Parameterization of Ambient Organic Aerosols at a rural site of the Northern China Plain", Ren et al. report on the results of thermal desorption mass spectrometry measurements of the organic aerosol component of filter-collected ambient aerosol. The work also features a convincing effort in carefully "calibrating" the desorption method so that effective volatilities (or vapor pressures, saturation concentrations C*) can be inferred for individual organic compositions. The calibration experiments did not turn out as good as they maybe could have, and much of the ambient data (section 3.2) at least appears to have been discarded in favor of focusing on easier-to-analyze signals. An additional data selection criteria, however, was the continued prevalence of the respective compositions, which makes the selection particularly useful despite its potential narrowness. Importantly, the calibrations allow a thorough analysis of the observations. The authors explore how the resulting C* values could be parameterized based on compositions. They compare their findings to previous attempts in the literature, but which have to a large part been relying on calculations using group contribution theory to extrapolate to compositions observed in organic aerosol. Thereby, this study's analysis of thermal desorption mass spec data goes deeper than what is often provided by other studies using similar datasets.

For these reasons, I believe the manuscript is of considerable interest and deserves publication in Atmospheric Chemistry & Physics, although I do suggest substantial "polishing" before that, considering my comments below.

I also stumbled upon an apparent contradiction in the analysis, e.g. when comparing Figs. 3-4 with Fig. 5. I am elaborating on that in my comments as well. Some clarification is at least warranted, or possibly some semi-major revision.

General notes:

I noted that only one week's worth of ambient was being used in this study. Was there particular reasons for that?

An additional result that might be worth looking into, similar to the analysis using retrieved T_max values, would be the widths of the fitted peaks, which I understand were largely unconstrained.

For example, could those widths tend to change with increasing/decreasing molecular weights, O/C or T_max? And most importantly: could they help identifying decomposition processes?

Technical comments:

Check that peer-reviewed papers are being cited rather than their discussions papers (e.g. ACPD and AMTD), where available.

There are many (individually minor) grammatical and semantical mistakes here and there. The manuscript is readable overall, but they do impede comprehensive reading at some places. Some of the semantical mistakes at least will also confuse readers, in particular if not familiar with the used methodology. I suggest appropriate proof-reading/language checks.

Major specific comments:

Lines 367-369: I am not following here. Aren't the compounds used for the "Eq. (4-1)" parametrization by definition containing (exclusively) OOA species?

In any case, however, the fair agreement with SIMPOL predictions is remarkable (even though only shown for 15 selected compounds) and worth pointing it, as it follows directly from calibrations and measurements, without using SIMPOL calculations. Whereas, if I remember correctly, the other cited works (Li, Tröstl, Stolzenburg) indeed based their parametrizations on SIMPOL calculations (so their agreement is expected).

Beyond Fig. 5a, however, it would be interesting to see how the authors' parametrization, i.e., Eq. (4-1), worked out for the other compounds in this group? I.e., do measurement-derived C* agree with SIMPOL also for other compositions in the high-O/C group, besides the 15 examples shown? That would however require making assumptions on molecular structures and new SIMPOL calculations. But the authors did that for members of the low-O/C group (Fig. 5b), so my suggestion might be relatively straight forward to implement. (But not sure which compounds were used in Fig. 5b see comment below also ("Moreover,....").)

Lines 391-399: I am lost a bit again. I had to read the first sentence (lines 390-393) several times, as I failed to understand for a while which 42 compounds were being referred to.

More generally, the paragraph here spends a lot of time arguing why the "Eq. (4-2)" paraemtrization agrees poorly with NIST data, whereas SIMPOL-based parametrizations perform better (Fig. S9). The point raised about this study's parametrization being based on observations of compounds with generally lower volatility (at least as inferred by T_max) may be part of the reason. But the main point, which may be lost, is that the agreement with SIMPOL is also poor for the other 132 (=observed?) compounds (Fig. 5b). The questions I would then have are: (1) is there a reason for SIMPOL to be less accurate for compounds with lower O/C? or (2) is there a reason for the "Eq. 4-2" parametrization to be less accurate, i.e., for the inference of C* based on T_max to be not or less valid?

In other words, the results shown in Fig. S9, although interesting, may distract/confuse the more important points of discussion in the main text.

Moreover, it remains unclear which those 132 selected compounds for Fig. 5b are. Is there substantial overlap with observed compositions, or possibly even hardly any?

And a key message of Fig. 5b is that parametrization Eq4-2 gives TOO LOW C* values (compared to SIMPOL). On the other hand, Figs. 3 and 4 suggest that parametrization Eq4-2 would yield overall HIGHER C* values than parametrization Eq4-1 (which agrees well with SIMPOL, Fig. 5a).

What am I missing?

As a consequence, I would be more careful in the final summary of atmospheric implications (lines 410+). I agree that Eq. 4-1 is doing quite well (for higher-O/C lower-MW compounds), but I am not convinced about Eq. 4-2 (for lower-O/C higher-MW compounds). At least I would not go as far as to claim that it is "more accurate" (and more than what?) for ambient aerosol. The authors hypothesize that interactions with inorganic aerosol components play a more important role for low-O/C compounds, thus lowering their effective C*. That hypothesis is plausible in principle, and I agree that those interactions are insufficiently studied. But here, it remains rather poorly supported by some discrepancies during calibration. (In the minor comments below, I also suggest an alternative hypothesis for those discrepancies.) Alternative hypotheses would be warranted too. For example, could those higher-MW compounds be structurally different in some fundamentally difference for SIMPOL to stop working? Is there a possibility for the FIGAERO method to be less reliable for those compounds?

And besides that, I still don't quite understand the discrepancy between the following (see also comment above, "Moreover,..."):

- Eq. 4-2 gives LOWER C* than expected e.g. using Li et al. (2016)

- Eq. 4-1 gives C* about as expected by Li et al. (2016)

- Compounds used for establishing Eq. 4-2 have HIGHER C* than expected from the C* of the compounds used for establishing Eq. 4-1.

Minor specific comments:

Abstract:

I suggest making clearer that grouping into two different O/C regimes was also supported by systematically different thermal desorption behavior (Figs. 3-4).

Indeed, I believe this is also a key result that is missing (or unclear) in the abstract.

Main text:

General: should briefly go into the difference between a compound's saturation vapor pressure and a compound's effective saturation vapor pressure (or concentration) in regards to partitioning in/out of aerosol particles.

Lines 56-60 ... If I remember correctly, Tröstl et al. did not know the molecular structures of the observed compositions classified as HOM, nor their saturation concentrations, but guessed the former and correspondingly modeled (using SIMPOL?) the latter. 

(Subsequently, it also remains unclear here what Stolzenburg et al. were "fitting" to.)

Lines 71-74: An important missing piece of information on the FIGAERO procedure is that the desorption temperature is ramped up linearly. 

Lines 77-80: A weakness of the "second method" is potential measurement artifacts that  obscure the true composition of detected species, which thermal desorption methods are prone to. But I would not conclude that the "third method" is generally and necessarily superior, as these lines now seem to suggest.

Lines 94-95: I believe some text is redundant here (explained twice what volatility is important for).

Line 100: I would explicitly mention how (by which of the 3 methods) was "C* measured".

Line 110: Unclear how thermal desorption was performed. There was a flow of 2.3 lpm and one of 1.0 lpm. If only 1.0 lpm went through the filter and into the IMR, what happened to the remainder 1.3 lpm flow, and what was it for?

I would also clarify in this paragraph how the filter was heated -- or rather that (presumably) it was the UHP N2 that would pass through the filter that was heated. In this regard: where and how was the nominal desorption temperature measured?

Lines 113-115: Why was 134 °C chosen as the highest temperature? Typically, FIGAERO is operated using desorption temperatures up to 200 °C. A shorter ramp could make sense if going primarily for T_max, as most "nice" peaks would probably occur before 134 °C. On the other hand, I would be worried about accumulating organic material on the filter, which might cause measurement artifacts... 

Line 116: How was the "blank filter" used for obtaining backgrounds?

Also: What desorption temperature ramp rate was used? Thornton et al. (2020) and Ylisirniö et al. (2021), e.g., have suggested that that ramp rate may affect the thermograms and hence T_max.

Section 2.2: Was the desorption procedure (as described in 2.1) for the calibration experiments the same as for the field measurements?

Fig. 1: I believe the caption could be considerably shortened, using reference to Table 1. Table 1 plus the legend of Fig. 1 contain most of the information given in the caption.

Lines 224-228: I would opt to disagree with the conclusion that the reason that T_max for calibration set #5 were higher than for set #4 was due to the ammonium sulfate. I would instead rather argue the effect was due to increased filter loading: 1000 ng for set #5 vs. 200 ng for set #4. 1000 ng is clearly in the range of filter loadings that previous studies have seen increased T_max for that were argued to arise from matrix effects, specifically, I believe, loss of relative surface area available for desorption (Huang et al. 2018; Thornton et al., 2020).

Both effects might play a role, but I do not think they can be separated here.

Oh, reading on, I see that this reason is in fact brought up towards the end of the paragraph. I would consider reformulating or restructuring the paragraph to improve clarity.

However, I still disagree with the wording in Line 232 ("In other words,..."). Enhanced interactions between aerosol components could lower the effective C* also without increasing viscosity. Again, reading on, I believe the authors would agree with that, but the way viscosity is brought up may be confusing...

Lines 244-247: I am not sure I am following. Also, judging from Ylisirniö et al. (2021), more information on the calibration procedure used by Nah et al. (2019b) would be needed to ascertain that the the cyan dash-dot and solid black lines in Fig. 1 should actually be compared.

Lines 254-261 (last paragraph in 3.1): I think the argument for using calibration #5 for the analyzing the ambient measurements is sound, in principle, and would deliver the most accurate estimates for C* here.

However, it could not be only the deposited mass on the filter that affects T_max, but also the particle (mass) size distribution of the deposited aerosol. I wonder therefore also how those distributions compared between calibration experiments and ambient samples?

Lines 274-276: For clarity, I would re-iterate, which peak was chosen for obtaining T_max in cases where there was more than one fitted peak. I.e., was it always the cooler one, as indicated in Section 2? (Fig. S5d shows an interesting example where of a wide lower cooler peak, and a sharper higher hotter peak. Was it still the cooler one that was chosen, despite being lower?)

Oh, peaking forward, I see that T_max of both or all peaks were actually considered, at least at first. So this comment may be moot. Still, these lines could already be clearer in respect to how (and which) T_max were obtained and used for further analysis.

Line 307 (and in general): It seems to be suggested, as also suggested in previous instances in the text, that the higher T_max obtained from thermograms exhibiting double peaks have rigorously been attributed to arising thermal decomposition, rendering them unusable for further analysis for assessing C*. That may be correct reasoning in many instances, but conceivably not always (e.g., for the case of isomers with substantially, but not necessarily unreasonably, different C*). It may be of interest to at least initially keep those hotter peaks "in the game", and discard them at a later point, as the authors' further analysis would probably be able to make a much stronger argument pro/contra decomposition being involved.

Fig. 4: I would clarify in the caption: Are the whiskers ultimately due to variability in the measured T_max (and hence variability in derived C*)?

Line 328-329: Please clarify, are those correlations for the two given compounds average correlations (plus standard deviations), while averaging over the individual correlations with each compound in the respective groups? That's how I understood it, but I could see how I could also be misunderstanding...

Line 338: Please remind the reader on which volatility range the fits/parametrizations in Mohr et al. are based on?

Table 2: Please include also values that have been used in/suggested by the cited literature!

Line 353: Was there any specific kind of OOA that Donahue et al. (2011) referred to corresponding to the yellow-dashed box in Fig. S7?

Line 362: "accuracy" in which respect? Please clarify.

Line 366: What is "therefore" referring to?

The study conducted by Ren et al. is looking into FIGAERO-I-CIMS thermograms, more precisely the maximum temperatures of the first peaks of the thermograms (Tmax). The study starts with a suite of laboratory experiments with PEG samples of different volatilities which they either inject on the FIGAERO filter using a syringe or nebulize, dry, dilute and collect onto the FIGAERO filter. They attain similar results as Ylisirniö et al. (2021), reproducing a quantitatively similar relationship between saturation vapor pressure (Psat) and Tmax. Ren et al. investigate this relationship further by making mixtures of various PEG, citric acid and erythritol with ammonium sulfate (AS), either as a mixture between one organic component and AS or as one mixture between all organic components and AS. These were then deposited on the FIGAERO filter following the atomizer method. A discrepancy could be observed between the Psat -Tmax relationships derived from these two experiment types. They attribute the mismatch to either AS-derived effects such as organic salt formation, viscosity limitations in evaporation or matrix effects. Finally, the authors utilize the Psat -Tmax relationship derived from the mixture calibration involving all organic compounds compounds to calculate saturation vapor concentrations (C*) for their field data. They further noticed that these Tmax-derived C* displayed as a function of molecular weight showed two groups/clusters with characterized with different O:C-ratios. The authors finally derive two molecular formulae-based C* parameterizations for these groups, respectively. The results are compared to other molecular formulae-based C* parameterizations.

The manuscript could be potentially very useful for the FIGAERO community if the calibration results were investigated further and the reasons behind the mismatch could be narrowed down and the potential influence of inorganic salts or matrix effects on Tmax and therefore C* in the field could be assessed. It would be useful if the authors could carefully evaluate how reliable the C* derived from Tmax are under environments with high mass loading and inorganic salt concentrations. I find this necessary before parameterizations are being derived from these Tmax – C* relationships. I recommend publication after major revisions.

Main comments:

I think it would be crucial to understand/narrow down what actually caused the change in the Psat -Tmax relationship when comparing the line derived from the single PEG+AS mixtures vs one solution. I would suggest you to perform more laboratory experiments such as: 1. A single organic mixture (all organic compounds included) without AS, 1000 μg deposited on filter with the atomizer method; 2. Replicating No.5 experiments and probing the effect of mass loading.

The authors should also provide information of the size distribution and temperature for the different experiments, especially for No.4 and No.5. It should be noted that Ylisirniö et al. (2021) mention their significant role in causing discrepancies in the Psat -Tmax relationship.

The authors should at least provide the 95% credible intervals along the fitted lines in Figure 1 if showing all the data points (replicates) decreases the readability of the graph. It would be useful to see how much scatter there is between replicates. This scatter could even hold some information about possible loading effects.

The authors should think about providing a schematic of their laboratory setup. They mention the possibility of organic salt formation in No.5 mixture that could cause the increase in the observed Tmax values when compared to No.4 experiments. If the sample is dried immediately after the nebulizer there is not much time for any organic salt formation under favorable conditions (high humidity and aerosol liquid water content). Do the authors think there is time for such reactions to actually happen?

After understanding the significance of the matrix effects on the Psat -Tmax relationship in the calibration data – do the authors still recommend deriving Psat from Tmax?  Do the authors observe variability in Tmax in the ambient samples that covary with mass loading? How much variability was there in the mass deposited on the 30 filters analyzed and how does it compare to the 1000 μg calibration reference? How would the matrix effects from ambient samples affect the predicted C*?

It is unclear to me whether the FIGAERO was measuring in real time during the field campaign or whether the filters collected (described in Sect. 2.3) were measured with the FIGAERO offline. Could you please clarify.

Minor/technical comments:

• L34: is the Nizkorodov et al. (2011) the best reference for this statement?
• L40-: The descriptions of the past methodologies are incomplete. The description is for example missing classic work with thermodenuders (TD) without CIMS that have been mounted as part of tandem volatility differential mobility analyzers (TDMA) or coupled with an AMS (TD-AMS). The introduction also lacks description of the way the thermograms measured by the TDMA or TD-AMS are being modelled to gain information of C* or VBS (see for example Cappa, 2010, in Atmos. Meas. Tech or Cappa and Jimenez, 2010, in Atmos. Chem. Phys.). In addition, dilution experiments among many others should not be forgotten. The authors seem to be citing more the work for predicting equilibrium partitioning coefficients than actual C* measurements. If the authors wish the provide a list of methods used previously, they should cite more relevant literature and make sure to include a complete description. Alternatively, they could focus on describing just the methods relevant for CIMS.
• L322-L323 Figure 4 does not contain a red or blue dashed circles, maybe ellipse would be a better word. I found this confusing at first.