In their revised version and in the reply to the reviewers, the authors have extensively reacted on the reviewer’s comments. While most of my comments from the first round of reviews have been satisfactorily addressed, the answers on several important and central comments are rather weak.
As I stated in the first review, this manuscript leaves the impression of work that is published in small incremental steps. A fraction of the measurements in this paper has already been published in the previous paper and several of the relevant issues I raised in my review were postponed to the next paper. The title of the paper advertises quantitative analysis of particulate sodium and potassium salts. This is a quite narrow focus and I think, in order to justify a publication on such a narrow focus in a prestigious journal like AMT, this kind of application of their rTDMS needs to be characterized very comprehensively. Several of the not-so-well answered points of my first review leave the impression that this has not been done yet, and is postponed to later submissions.
Do not get me wrong: I do not claim that the work is erroneous. I just wonder whether it is adequate to publish this good and correct, but to a certain degree incomplete work in AMT. Since this is a decision on the required standards of AMT, I completely leave the decision with the editor and do not suggest whether it should be published as it is, or whether the below-mentioned points should be addressed first.
Remaining comments from first review:
(1) P1L10-11: “refractory” is defined as material that keeps its structural properties at very high temperatures. Examples are oxides or carbides of metals like aluminum or magnesium. In this manuscript, materials are analyzed with the rTDMS that have bulk decomposition temperatures from 142 up to 850°C - with the exception of K2SO4 - and at temperatures of the graphite collector up to 930°C. These are temperatures of which most are within the accessible range of e.g. the vaporizer of the Aerodyne AMS (typical 550-600°C, 800°C can be reached), an instrument which claims to measure “non-refractory” aerosol components. I wonder whether the name “refractory TDMS” is adequate for an instrument with these features; most of the really refractory materials could probably not be measured with this instrument.
Authors reply: The definition of “non-refractory” and “refractory” compounds in atmospheric aerosols is rather empirical and depends on the analysis method. The temperature of the graphite collector are mostly within the vaporizer temperature of an Aerodyne AMS (~600°C), as the referee pointed out. However, the temperature of particles that hit the vaporizer of an Aerodyne AMS may not reach the vaporizer temperature because of the particle bounce and latent heat effects (Saleh et al., 2017). To our understanding, non-refractory sulfate, nitrate, and chloride aerosols measured by Aerodyne AMSs are not strictly defined (Drewnick et al., 2015) and nearly equivalent with AS, AN, and AC in most cases.
The terminology “refractory” does not have a strict definition in the rTDMS. Following the definition by Kobayashi et al. (2021), chemical compounds with a bulk thermal desorption temperature lower than ~673 K are referred to as non-refractory compounds, and the others are referred to as refractory compounds. Although the rTDMS may not comprehensively measure refractory aerosols, we consider that the terminology “refractory” is appropriate to represent the general characteristics of our instrument.
Reviewer reply: I disagree with the line of reasoning of the authors. Indeed, the definition of “non-refractory” and “refractory” in aerosol analysis is based on the analysis method. It is operationally defined: substances that vaporize sufficiently quickly at the operating temperature (e.g. of the vaporizer) to be properly measured with the respective instrument are called “non-refractory” and those which do not vaporize quickly enough or not at all are called “refractory”. This is in good agreement with the general definition of “refractory”. For an instrument called “refractory aerosol TDMS”, I would expect an instrument that measures all substances up to the highest vaporization temperatures and not only of those with a vaporization temperature in a certain temperature range up to the temperature of the vaporizer. In agreement with the typical definition in aerosol analysis, this would again be “non-refractory” material – also operationally defined on the basis of the vaporizer temperature of this instrument. Therefore, I still think that the name is misleading.
(7) P6L140: Some of the ions of Table 2 require some more information. For SS m/z 23 (Na+) and 48 (SO+) are listed in the table. According to Figure 2, the SO2+ signal (m/z 64) is larger than the m/z 48 signal. Why is it not included in the table? Furthermore, for PN, PC, and PS the C3H3+ ion is listed in the table. What is the origin of this ion?
Authors reply: We have added ion signals at m/z 64 for SS and PS in Table 2. We have also added ion signals at m/z 64 for PS in Figure 2. We consider that the origin of C3H3+ ions (m/z 39) was contamination of organic compounds on the graphite collector.
Reviewer reply: I wonder whether the signal at m/z 39 could also be from a contamination of the vaporizer with potassium – from earlier experiments with potassium salts.
(9) P7L161-167: How was the background signal, i.e. the signal outside the peak integration area, accounted for? For some of the m/z signals in Figure 2 it does not return to zero after the peak, how is this handled?
Authors reply: We have added the description about insufficient vaporization of PN and PC particles in Section 3.1. The ion signals at m/z 39 from PN particles did not reach the background level after the second peak, suggesting that PN particles were not fully vaporized by the current laser power settings. This may lead to underestimation of the sensitivity at m/z 39 for PN particles. Furthermore, a small increase in the ion signals at m/z 39 from PC particles was observed after 40 s, indicating that PC particles were not fully vaporized by the first laser power setting (7.5 W for 40 s). We estimated the effect of the small peak to be ~20% of that of the main peak by comparing the ion signals after 40 s with those before 40 s.
and
(14) P10 Figure4: In Figure 4(b) the m/z 39 signal (K+) is strongly enhanced after the first group of peaks and even more after the second peak (40-50s). What causes this enhanced background signal? How do you deal with it when calculating the total signal area? Is this slowly vaporizing potassium?
Authors reply: The enhancement after the first group of peaks would be caused by the tailing of ion signals from PC particles and the onset of the thermal decomposition of PS particles. The enhancement after the second peak may be caused by the incomplete vaporization of PN particles. We have not investigated the effects of the enhanced background signals in the current study. This issue will be addressed in future studies.
Reviewer reply: Both comments, i.e. (9) and (14), were intended to point towards the same direction and both have not been addressed in the reply. The point of the comments was: how was the shifted baseline after these peaks accounted for, when quantifying the signal intensity? This was rather asking for the peak integration details (which baseline was subtracted from the peak? How was the baseline assumed below the peak? …) than for the reason for the shifted baseline.
(12) P7L178: Is the relative peak area in the measurements proportional to the relative composition of the particles, i.e. can the composition of a multi-component particle be reliably calculated from single-component calibrations? This should be stated clearly.
Authors reply: Kobayashi et al. (2021) suggested that the ion signals for multi-component sulfate particles could be approximated as the linear combination of ion signals originating from single-component sulfate particles based on mass closure tests. We did not perform detailed mass closure tests in the current study because of significant uncertainties in determining the mass of multi-component particles (especially for the particles generated from the seawater samples). Alternatively, we compared the SN/SC and SS/SC ratios estimated from the QMS ion signals with those predicted from the ionic concentrations in the solutions. This point was added in the last part of Section 2.3.
and
(13) P8 Figure 2: Why is the NO+ signal in Figure 2(a) and (d) about 10 times more intense than the Na+/K+ signals (similar intensity after multiplication with 0.1)? Are Na and K incompletely vaporized?
Authors reply: This is related with the question (11). The sensitivities at m/z 30 for nitrate particles were larger than those for the other m/z peaks (Figs. 2 and 3). The difference in the sensitivities cannot be explained by the difference in the electron ionization cross sections. We have not identified the mechanisms that caused the variability in the sensitivity values. The difference in the divergence angle of evolved gas molecules after the thermal desorption might be a possible mechanism (Uchida et al., 2019; Ide et al., 2019). We have added this point in the last part of Section 3.1.
and
(20) P14L292: What could these “matrix effects” be? What could cause these differences in signal intensities? How would mixtures of K- and Na-salts behave? These results show that Na cannot be quantitatively measured with this method. This should be stated clearly in Abstract and Conclusions.
Authors reply: We have added the description “The SS/SC ratios estimated from the ion signals at m/z 23, 36 (H35Cl+), and 48 (SO+) agreed well with those predicted from the solution concentrations to within ~10%. The SN/SC ratios estimated from the ion signals at m/z 30 (NO+) and 36 also agreed with those predicted from the solution concentrations to within ~15%, whereas the SN/SC ratios estimated from m/z 23 were significantly lower than the predicted values.” in the abstract and conclusions.
The matrix effects mean that the sensitivities depend on coexisting compounds. We have added this point only in the conclusions because the descriptions in the abstract become too lengthy. We have not tested the mixtures of K- and Na-salt particles. The quantification of K-salt particles would be investigated in future studies.
Reviewer reply: According to your answers, you did not perform experiments to determine whether the mass of individual components in multi-component particles is correctly measured (12). At the same time, you have no strong, experimentally based answer to the question why some of the substances are measured with much higher (10 times) sensitivity than others (13, but also 11). Finally, you clearly state that there are matrix effects, which affect the sensitivities of substances depending on coexisting compounds (20). All these issues (as well as the assumptions on collection efficiency) affect quantification of the various aerosol components. For a paper in a prestigious journal as AMT, that advertises the instrument and the study to “quantify” refractory aerosol components, and which focuses only on a small subgroup of aerosol components, I would expect a more in-depth and comprehensive characterization of the method. These answers suggest that the instrument is not yet well characterized for the analysis of sodium and potassium salts. To justify this title, I think that such issues need to be resolved.
(17) P13L243-249: These decomposition equations do explain the occurrence of the m/z 30 (NO+) signal, however, they do not explain the occurrence of the m/z 23 and 39 signals. Is the final product of these equations (Na2O2, K2O2) vaporized or is it further decomposed into Na/K and O2? Are these ions (Na2O2+, K2O2+) observed in the mass spectra?
(18) P13L250-251: I do not understand how the sequential thermal decomposition of the Nitrate salts causes the bimodal peaks at m/z 23 and 39. This would only be the case if the intermediate Na- and K-containing products would vaporize to form the Na+/K+ peaks. However, if this would be the case, why is not all the material vaporized during the first peak? Furthermore, if e.g. NaNO2 is vaporized, is the respective ion observed in the mass spectra?
Authors reply: Because the questions (17) and (18) are related with each other, we will collectively answer them. We have revised the explanation in Section 4.1. Our experimental data indicate that the thermal decomposition of SN particles yielded gas-phase Na and NO, and that of PN particles yielded gas-phase K and NO. However, the temporal evolution of the ion signals at m/z 23 and 39 suggests that the thermal decomposition processes of SN and PN particles were not represented by single-step reactions. Tagawa (1987) proposed the following reactions for thermal decomposition of bulk SN in dry air:
NaNO3 → NaNO2 + 1/2 O2 (~491–750°C)
NaNO2 → 1/2 Na2O2 + NO (~750–850°C)
Na2O2 → Na2O + 1/2 O2 (> 850°C)
and those for PN in dry air:
KNO3 → KNO2 + 1/2 O2 (~526–750°C)
KNO2 → 1/2 K2O2 + NO (~750–900°C)
K2O2 → K2O + 1/2 O2 (> 900°C)
We speculate that NaOx and KOx produced via the first step reactions underwent further thermal decomposition reactions to yield gas-phase Na and K in the rTDMS.
Reviewer reply: Thank you for this detailed information on the suggested thermal decomposition process. Nevertheless, it does not answer the question, whether the intermediate products (e.g. NaNO2, Na2O2, KNO2, K2O2) are vaporized (which would be necessary to form one of the Na+ and K+ peaks, each) and whether the respective ions (NaNO2+, Na2O2+, KNO2+, K2O2+) were detected, supporting these assumptions. |
