Review of “A comparative study of K-rich and Na/Ca-rich feldspar ice nucleating particles in a nanoliter droplet freezing assay” by A. Peckhaus and Co-authors
The revised version of the manuscript under review here has improved a lot. The language now is fluent, most mistakes have been corrected (the few I stumbled across are given at the end), and that is also true for other technical issues I raised in the first round.
There are, however, still two main topics on which I want to comment, denoted 1) and 2), and the few technical remarks below. After this will have been addressed, this work definitely merits publication in ACP.
1) Concerning the fitting procedures and related conclusions
I respect your choice to not use a single set of values derived for your fitting parameters and test all obtained data against that. But there are still points I want to raise related to the fitting procedure and conclusions you draw from that. This might in parts be similar to some of my original remarks in those cases where I feel there is still room for improvement.
- p3, line 7-8: “This behavior was interpreted in terms of a specific average number of ice nucleating sites per particle reaching unity inside the temperature range where the freezing curve starts to level off.” - This is not correctly said here - this NUMBER (!) (of sites) does not reach a value of 1 here (which is what the text says). This needs to be reformulated.
- p3, line 4-5: “… we show that the observed temperature dependence of the INAS surface density is an inherent feature of the experimental method.” - Later in your work you argue that the fit values you obtain are more widely usable and you even use them to draw conclusions for your own work (see below, the two last quotes I took from your paper, before my point 2) starts). This is inconsistent with what you say here, where the impression arises that obtained results will differ for different experimental methods used. As I will discuss below, there is a difference in what you did and how things were done in Niedermeier et al. (2015) which might explain some of your results concerning n_s* and n_site, and your statement here might not be valid if the fit procedure was done as in Niedermeier et al. (2015). Check for that and then be consistent throughout the text when you argue for or against using the values derived from fitting.
-p13, line 25-26:”… SBM … cannot be effectively used to constrain the fitting routine. [new paragraph] The allowed variability of fit parameters can be reduced if we consider that the same IN material has been used in CR experiments with different weight concentrations W.” - The first here mentioned sentence needs to be reformulated, as it is not the SBM that cannot be effectively used, but instead it is the fact that you did not use as much of the available information as possible, as you then explain in the second sentence quoted here. Instead of “SBM” it would rather be “the subset of input parameters used here”, or something along that line.
-p14, line 11-14 and p14, line 23-24: “This observation, however, hints that n_site should not be treated blindly as a number of active sites activated during the cooling ramp or isothermal freezing, but rather as a number of active sites required by the numerical algorithm to reproduce the freezing curve. Thus, caution should be exercised when interpreting the fit results, as numerical features can be mistaken for physical relationships.” and “… fitting the freezing curves with a three adjustable parameter fit without providing additional constraint does not necessarily lead to a better understanding of IN nature.” - To my understanding, there is a big difference in your data set and that by Niedermeier et al. (2015), a paper you relate to quite a bit. While frozen fractions you measure always reach 1, this is not the case in Niedermeier et al. (2015). There, the amount of material (or better surface area) per droplet is so small, that (statistically) droplets exist that contain a particle that does not have an ice active site at all. This leads to a temperature region where measured frozen fractions do not increase with decreasing temperature and a plateau forms at values for frozen fractions <1. From this region and the assumption of a Poisson distribution, an average number of ice nucleating sites per droplet (lambda) is derived. This is not a fit factor but a value that could be determined, but it cannot be determined from your data set, as your droplets contained so much material that a plateau in frozen fractions < 1 was not seen. Niedermeier et al. (2015) then relate lambda to n_s* (see their equation 7), and n_s* therewith is a particle property, the way it was defined. You vary n_s* (maybe because you cannot constrain it from a plateau?) and then, based on that, obtain varying n_site values which you then judge non-meaningful in a physical sense. To my understanding, this originates in the fact that you cannot see a plateau in frozen fractions < 1, and hence you have one constrain less than Niedermeier et al. (2015). So your judgements I quoted above and related remarks in your manuscript are valid for your data set (and similar ones which do not observe the plateau), but not in general. This needs to be rephrased throughout your text, including the two sentences I quoted here.
-p16, line 13ff: Again, as I understand, n_s* at least as defined in Niedermeier et al. (2015) is a particle property, which, however, you cannot obtain from your data set and need to use as an additional free parameter. n_s* in Niedermeier et al. (2015) was obtained for the lowest temperatures at which frozen fractions were observed to not increase further upon further cooling. But you say here “further increase of the IN active site efficiency … would not result in the further increase of the freezing probability”. This makes sense as the freezing probability is already 1, as you nicely explain in the beginning of this paragraph. So I do not see how you can judge from that that n_s* is a suspension property, as suspensions with higher concentrations will cause all droplets to freeze above the temperature range where the plateau is observed. However, there is an effect of time (as you observed yourself), and the short time for LACIS (1.6 seconds is what you used) is much faster than your own cooling rate, so would this not explain some of the discrepancy observed in Fig. 9?
-p17, line 24ff: Starting with “Multiplying the n_s* …” - here you do use the plateau seen for the highly active fraction to derive that ~ 30% of all droplets contained one of the high temperature active sites, using this to draw further conclusions. I totally agree with you doing it this way, but to do this, you have to trust the derived n_s* value, which, it seems, you repeatedly recommend not to do (see quotes from your text above).
- p21, line 23-24: “The asymptotic active site density n_s*, achieved by n_s(T) as the freezing probability of every droplet in the ensemble approaches unity, can be interpreted as a method independent property, inherent to the suspension only. Together with the mean value of contact angle, this asymptotic value provides a basis for the parametrization of IN properties that is required within the atmospheric modeling.” - Again, I agree with this, but this does not fit to what you said on that topic in your manuscript (see the quotes from your text above).
2) Concerning the presence of organic contaminations
First of all, I wonder about the number you give for the mass concentration of active sites you estimate for you sample: if 1/3 or all droplets contain such a site, and the mass of feldspar per droplet is 1.2*10^-8 g, wouldn’t it be that there is one site in ~ 1/3*1.2*10^-8 g = 7 * 10^-8 g, the reciprocal of which is 1.4 *10^7 /g. This latter value, to my understanding, is the mass concentration of active sites, or am I wrong? Your value is a factor of 50 higher, and I wonder where the discrepancy comes from.
Also, when you then turn to polysaccharides and argue that your sample would have had to be comprised of 10% polysaccharides, you are still “comparing apples with oranges”. Of the ~10% of biological material that was estimated to be present in the new publication you consult here now (Augustin-Bauditz et al., 2016), only a VERY minor fraction of all organic material makes up the ice active polysaccharides (and that fraction might strongly depend on the sample). To cite Augustin-Bauditz et al., 2016: “The size of a single INM was estimated to be 10 nm”. From this (assuming a spherical shape and a density of 2 g/cm^3), one ice active polysaccharide has a mass of ~ 10^-18 g. This, together with the concentration of ice active molecules you derived (n_m of 5*10^7 /g) leads to ~ 5*10^-11 g polysaccharides per gram of your sample. This is clearly not 10%, and I would even argue that such a contamination is not visibly detectable.
Admittedly, these ice nucleating polysaccharides might not occur on their own but be related to some additional (organic) material, but there is room for quite some more organic material that could be there before it would become visible.
To summarize this: please check/ correct the values you give in the text and rewrite the text accordingly (some of your arguments won’t hold, unless I miscalculated something above). And I understand that you do not want to rule out the possibility that there might be a VERY active mineral dust site on the feldspar, but I still recommend to tune down the statements concerning the possibility of biological contamination. (E.g.: Abstract, page 2, line 1: replace “has been ruled out” by “might be unlikely”; page 22, line 10: Replace “found to be too high to be explained by surface contamination …” with something more suitable (see my estimates above); and check for other occurrences and rewrite where necessary).
Technical comments
Abstract line 20-21: This sentence seems to have a “copy-paste-error”: “…, the parameter space can be constrained the unique sets of model parameters for specific feldspar suspensions can be derived.” - Rephrase
page 4, line 15: Replace “In this framework, of this” with “In the framework of this”
page 11, line 8: Add “a” between “occurred at” and “lower temperature”.
page 11, line 25: Replace “… suspension. The …” by “suspensions, the …”
page 15, line 4: I wonder to what the curves were identical. I assume you mean “The fit parameters that provided the best fit of THE DIFFERENT liquid fraction decay curves were identical.”? If yes, please add “the different”.
page 18, line 3: You say “… these sites will be detectable only in concentrated suspensions and setups.” - This is in contradiction with the fact that multiple types of sites were seen for size segregated measurements for some samples in the past (you cited some respective papers in the sentence before), so your statement is only true if the sites occur only very infrequently. I suggest starting this sentence with: ”When some types of sites occur only very infrequently, their presence will be detectable only … .” |
- Printer-friendly version
- Supplement
