Dear Authors,

I concur with the assessment of Referee #2, that the Authors have not satisfactorily addressed in the actual revised manuscript the justifiable and reasonable questions raised during peer review. While you provided a detailed rebuttal to the referee comments and short comments from Heike Wex and Ben Murray in your "author's response" file, you made few changes to the actual revised manuscript to reflect this. Please carefully take the comments from the second review from Referee #2 to further substantially revise your manuscript to properly address the major concerns raised during peer review. The Referee has suggested several paths that you can take to properly address these issues in your revised manuscript. I look forward to reading the next version of your manuscript that clearly addresses these issues.

Best regards,
Ryan Sullivan

Thank you for submitting a further revised version of your manuscript. This has largely addressed and clarified the questions raised in the last round of reviews.

The causes of the discrepancies and disagreements found using different ice nucleation methods is an important topic that will likely continue to see active discussion for many years to come. This paper contributes to this discussion, by focusing on the possible effects of particle coagulation and sedimentation on the ice nucleation properties derived from cold plate droplet freezing methods where particle suspensions in water are studied. This is in contrast to the “dry” methods where particles are dispersed in air as an aerosol before they are nucleated into droplets and subject to ice nucleation conditions.

This manuscript is now almost acceptable for publication in ACP. I do feel the clarity of the manuscript could be significantly improved however in some places, and I ask that the authors consider the following changes and suggestions as they revise their manuscript for final publication in ACP. There are also some errors and typos I noticed that should be corrected.


It strikes me that it is not made clear that the MICC experiments were focused on condensation/immersion mode freezing. This should be made very clear early on, so that the applicability of the comparison to the cold plate and other immersion freezing experiments is very clear. You might also briefly mention of any depositional ice nucleation below water saturation is ever observed, and if so do you simply ignore this data and just focus on the total/maximum ice crystal concentration produced in the chamber after the expansion and cooling?

Similarly, I found it hard to fully understand how the MICC expansion chamber probe data was analyzed to determine the ice crystal concentrations. It seems that when possible the 3V-CPI measurements of particle shape are used to distinguish ice from droplets. If that is correct please make this clear. In some cases the ice crystals were too small/round to distinguish from their shape, and so instead ice is determined by particle size from the CDP. If that is also accurate please make this clear, and also state what particle size was used as the split between cloud droplet and ice crystal, and a justification for this cut size. In the few experiments where size and not shape was used to measure the ice crystal concentration, you could draw a line on the CDP figure indicating this cut size.
Also, how is the aerosol particle concentration (used to derive the aerosol surface area) measured, using only the UHSAS data?
The size range of the WELAS probe should also be mentioned on page 5 for completeness. How is the WELAS data used?
I’m not sure that all the plots of the different expansion experiments are necessary or add much to the main paper. It is probably hard for a non-expert to really glean much from these plots, and they take up a lot of space in the paper. It is good to document them, and the discussion of each where different features of each expansion are discussed. I think most of these could be moved to Supporting Information instead, as there are already 17 figures in the paper, which is quite a lot. A few exemplary expansion figures could remain in the main paper. You could describe the general features and the data analysis more generally, instead of discussing each expansion separately. Any important unique features of particular expansions could still be discussed in the main paper, or the supplemental.

The expansion experiment figures (Figs. 2-9) could be annotated to indicate the specific features (cloud droplet formation/evaporation, ice formation) mentioned in the text. Color coded vertical lines could be used to indicate these main events, for example.

The typical range of particle concentrations used in the cold plate experiments should be mentioned, such as on the top of page 4. Right now these numbers are missing. Then it will be more clear why you probed the range of particle concentrations you used in your experiments and models, such as on page 14, line 12.

This is a rather long and confusing sentence in the Abstract: “Results from applying a coagulation model to the size distribution of mineral particles present in the suspensions, as used in the cold-stage-derived parameterisations, were used to investigate the idea that the mineral particles coagulate in suspension, which either removes the particles from the drops by sedimentation, whilst in the bulk suspension, or reduces the total particle surface area available for ice nucleation to take place.”


Section 3 – At the beginning, should reiterate that condensation/immersion mode freezing was studied in these experiments.

Paul DeMott recently published an empirical parameterization for mineral dust INP, based on laboratory and field studies. As this is highly relevant to this work you should discuss the salient points from that paper that relate closely to your work.

DeMott, P. J.; Prenni, A. J.; McMeeking, G. R.; Sullivan, R. C.; Petters, M. D.; Tobo, Y.; Niemand, M.; Möhler, O.; Snider, J. R.; Wang, Z.; Kreidenweis, S. M. Integrating laboratory and field data to quantify the immersion freezing ice nucleation activity of mineral dust particles. Atmospheric Chemistry and Physics 2015, 15, 393–409, doi:10.5194/acp-15-393-2015.


Page 4, line 17: Do you mean “> 5 nm”?

The fact that you do multiple expansions on the same mineral aerosol is important, as the most efficient ice nucleants are removed in the first expansion, thus changing the derived n_s values. While you do discuss this on page 10, it seems very odd to me that you averaged the n_s values from the successive expansions together. This does not seem like an accurate reflection of the aerosol’s actual ice nucleation properties. Why not create an additional figure similar to Fig. 10 where you plot the n_s values for the different expansions as a different colour, for the different aerosol types. This would be a very interesting plot as it would show just how much each expansion changes the remaining aerosol’s ice nucleation properties. Just averaging the different expansions together seems inappropriate as it is not the same aerosol in each experiment – the nest ice nucleants have sedimented out.

Section 4.2.1: Cite some key references to the ACPIM model.

Page 12, line 1: “as observed”. Do you mean depositional ice nucleation was or was not observed here?

Page 12, line 8: In discussing these parameterizations it would help if the parameters that they are based on was briefly summarized. Are they all based on particle surface area, as is used here?

Section 4.2.2: Switched from past to present tense. Should stick with past tense.


Page 14, line 21: Regarding the initial mineral dust size, I am quite sure these is a large distribution of particle sizes, extending into the supermicron range for these mineral powders. That has certainly been my experience doing experiments with many types of mineral dust powders. So while I see the point that the mean (?) particle size increased with time, I don’t think one can assume there were no initial supermicron particles.

Page 14, line 26: Are these the Dynamic Light Scattering experiments? DLS is often referred to but never stated what these experiments are. Is that what the Malvern Zetasizer Nano ZS measures? Much more details are required here, specially as most of the atmospheric science community is not familiar with DLS as it is used for colloids not aerosols.

Why is only one particle size listed for the DLS measurements in Table 2? Is this the mean or median value? Surely a distribution of particle sizes is measured. Would be great to include the variance of the measured size distribution, as that can indicate if the size distribution narrowed or widened with time in the water. That can provide further information about the physical processes causes the change in particle size.

Page 16, line 30: should state that an /increase/ in wt % results in an /decrease/ in fractional surface area.

As the particles in droplets on a cold plate are subjected to a temperature gradient, are thermophoretic effects possibly important in causing particle aggregation/coagulation? Should discuss if you can or can’t rule out the role of thermophoresis.

Page 18, line 18: Should spell out what “it” is.

Page 20, line 11: Typo, “it” instead of “in”.

Page 20, first bullet point: Should reiterate the differences in the Feldspar versus other mineral data that lead to this conclusion. If I recall the Feldspar data agreed even when high particle concentrations were used.



Appendix: I found several typos, please proof read carefully.

Page 20, line 2: solved, not solve

Line 5: “When by the”

Line 7: delete “the”

Page 23, line 16: “(source)” – missing reference?

Page 24, line 20: “like changed”


Table 1: Are the two values listed in [ ] for the two lognormal modes of the size distribution? Please specify in the table caption.

Table 2: See comment above about why only one (average?) value for the particle size from DLS measurements is given.

Figure 1 caption: Typo, “monitor”

Figure 2, and similar figs 3-9: I found it hard to know which y-axis to use for the lower panel. The red text of the right y-axis suggests that only the 3V-CPI data is on the right y-axis, but the legend says the blue traces are also on the right y-axis. Perhaps you can see the confusion with the colour coding here. Perhaps blue could be used for the left y-axis and the droplet concentration only. And/or you can add left/right arrows beside the traces or legend entries indicating which axis to follow for each.

See earlier comment about moving most of the chamber expansion plots to supporting information.

Figures 13 & 14 should be combined into one 2 panel figure.

figure 15: Specify in caption if this model included Brownian only or Brownian and gravitational forces.

Figure 16 should have a more explicit caption, hard to understand what data is used here. What system is the Atkinson data for? What does “shifted to Niemand line for natural dust” mean?



Ryan Sullivan