Cold-stage system for Impact-initiated Contact freezing Experiments (C-ICE)

Chen, Jingchuan; Rai, Upasna B.; McMillan, Kyle A.; Davis, Ryan D.; Tolbert, Margaret A.; Brooks, Sarah D.

doi:10.5194/egusphere-2026-3946

Preprints

https://doi.org/10.5194/egusphere-2026-3946

Preprints

28 Jul 2026

| 28 Jul 2026

Cold-stage system for Impact-initiated Contact freezing Experiments (C-ICE)

Jingchuan Chen, Upasna B. Rai, Kyle A. McMillan, Ryan D. Davis, Margaret A. Tolbert, and Sarah D. Brooks

Abstract. Ice nucleation by atmospheric aerosols plays a central role in cloud microphysical processes, exerting strong influences on Earth's radiation balance and precipitation formation. Among heterogeneous ice nucleation pathways, contact freezing induced by aerosol-droplet collisions remains one of the least quantitatively constrained, largely due to the scarcity of experimental approaches capable of directly probing impact-initiated freezing under controlled and reproducible conditions. Here we present the Cold-stage system for Impact-initiated Contact freezing Experiments (C-ICE), a laboratory system developed to enable controlled, repeatable, and size-resolved investigation of contact freezing triggered by particle-droplet collisions. C-ICE combines a fixed supercooled droplet with a precisely conditioned aerosol jet, allowing particle size, number concentration, collision geometry, and thermodynamic state to be independently characterized and experimentally constrained. At the single-droplet level, freezing onset is identified using complementary optical criteria based on breath figure formation and grayscale analysis. Application of C-ICE to silver iodide aerosols reveals sharp frozen fraction transitions over narrow temperature intervals (≈2 °C), with median freezing temperatures shifting systematically toward higher values from -12.6 °C for 200 nm particles to -12.0 °C for 800 nm particles, and to -10.1 °C for the polydisperse aerosol. By integrating a theoretical collision model with experimentally determined collision efficiencies, we constrain the collision efficiency of the C-ICE system, enabling measured aerosol concentrations to be converted into effective particle collision rates onto the droplet. This novel C-ICE system provides a robust method for quantifying impact-initiated contact freezing and advances process-level understanding of contact freezing in clouds.

Received: 03 Jul 2026 – Discussion started: 28 Jul 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Preprint (PDF, 1429 KB)

Supplement (849 KB)

Download & links

Jingchuan Chen, Upasna B. Rai, Kyle A. McMillan, Ryan D. Davis, Margaret A. Tolbert, and Sarah D. Brooks

Status: final response (author comments only)

RC1:
'Comment on egusphere-2026-3946', Anonymous Referee #1, 31 Aug 2026
In this manuscript, Chen et al. introduce the C-ICE system, a novel cold-stage setup to measure impact-initiated contact freezing. This instrument is designed to enable the controlled, repeatable, and size-resolved investigation of contact freezing triggered by particle-droplet collisions.
The setup consists of a single 1 ml droplet deposited on a hydrophobic glass substrate mounted on a temperature-controlled cold stage that is cooled at a rate of 1 K/min. Particles are provided by a cooled and dehumidified particle-laden airflow that impinges on the droplet at a jet velocity of 4.0 m/s. Freezing is detected optically using a CCD camera through breath figure and grayscale analysis.
To constrain the collision efficiency (CE), the authors combine theoretical formulations of Brownian motion, inertial impact, interception, thermophoresis, diffusiophoresis, and electric forces with experimental estimates based on the assumption that a single collision triggers ice nucleation. To test C-ICE, silver iodide was used.
Overall, the manuscript is well-written, logically structured, and falls within the scope of AMT. However, different aspects of the setup, data analysis and interpretation are problematic and require clarification. These issues need to be addressed and revised before publication in AMT can be considered.
Major Comments
The experimental setup should be explained in more detail. Specifically, the following points need to be addressed:
How is the relative humidity within the stage controlled during a cooling ramp? There is likely water vapor condensation onto the droplet due to the supersaturation that is generated by cooling. Vapor condensation would heat the droplet surface and hamper the temperature accuracy of the experiment.

The role of the aerosol flow for the humidity within the chamber remains unclear. The manuscript describes the pre-cooling to -15°C within the dehumidifier in detail, but it does not become clear whether the temperature of the dehumidifier is kept at a constant temperature of -15°C or whether its temperature decreases together with the one of the stage. Moreover, it remains unclear at which temperature the aerosol stream is turned on. In the experiment that is displayed in Fig. 2, it was turned on at -11.5°C, i.e. shortly before the droplet froze. Was this always the case or was this especially done to illustrate that the breath figure is due to the aerosol stream? When it was usually turned on while the cooling ramp was already running: how did you decide when to turn it on? How long before freezing was it turned on usually?

What is the distance between the nozzle and the droplet? An air jet of 4.0 m/s should generate a vertical stagnation pressure and surface shear, which again might induce internal circulation within the droplet if the nozzle is placed close to the droplet surface. This needs to be discussed.

Also, the data interpretation requires reconsideration:
The explanation that the breath figure is generated through aerosol-droplet collisions is highly unlikely or even impossible as the impact of 200 nm particles is not sufficient. A more likely explanation is that the breath figure is generated by the humidity within the chamber and/or in the aerosol flow. This again raises questions regarding the water vapor equilibration within the chamber and the aerosol flow.

The competition between contact and immersion freezing is not discussed at all in the manuscript. Considering, that the droplet constantly accumulates particles as soon as the aerosol flow is turned on, the immersion freezing probability increases as the AgI concentration in the droplet becomes higher. This needs to be discussed.

Considering the recent literature, immersion mode freezing is more efficient for AgI than contact freezing (Nagare et al. 2016; Marcolli et al., 2016). Marcolli et al. (2016) reported median immersion freezing temperatures of -9°C for 200 nm AgI particles for residence times of 10 s and -11°C for 3 s. Both temperatures are above the median freezing onset temperature for contact freezing (-12.7°C) in C-ICE. A higher efficiency for immersion compared with contact freezing was also observed for 200 nm AgI particles by Nagare et al. (2016). These authors state that in their setup freezing was induced by single AgI particles collected by the droplets within the chamber. Considering that their freezing temperatures are well below the ones measured with C-ICE, immersion freezing on previous collected AgI particles is the more likely mechanism for the experiments performed in this study. This competition between immersion and impact-initiated contact freezing needs to be discussed adequately.

The authors write at several instances in the manuscript (e.g., lines 417–425 and 505–509) that relying on the assumption that each freezing event corresponds to an impact, a lower limit for the collision efficiency can be obtained. Yet, considering that freezing likely occurs in immersion mode on previously accumulated droplets, the collision efficiency might even be lower. Thus, the freezing data can neither provide an upper nor a lower constraint.

Taking all these issues together, running C-ICE in continuous cooling mode leads to uncontrolled conditions with respect to droplet surface temperature and precludes the accurate determination of the freezing temperature when it is initiated at the particle surface. Moreover, an unambiguous identification of the freezing process as impact-initiated contact freezing is not possible when there is competition with immersion freezing.
Discrimination between immersion and impact-initiated contact freezing might be possible when experiments were performed at constant temperature. The range of temperatures of interest could first be narrowed down in a cooling cycle. Subsequently, the two freezing modes could be discriminated with an off/on sequence of the aerosol flow at constant temperature. When the freezing probability is independent of the presence/absence of the aerosol flow, impact-initiated contact freezing can be excluded.
Specific comments
Line 33 and throughout the manuscript: “freezing” involves per definition liquid water. So, “deposition freezing” should be revised to “deposition nucleation”. If the authors want to stress the involvement of liquid water, they should directly refer to “pore condensation and freezing”.
Line 156: what is meant by “grid points” in this context?
Lines 219–223: “Under the controlled laboratory conditions of C-ICE, such particle-droplet collisions are theoretically predictable controllable and repeatable, as demonstrated in the Results section below. This approach provides a foundation for interpreting impact-initiated contact freezing measurements and for guiding future efforts to parameterize contact freezing in atmospheric models.”: This statement is in harsh contrast to lines 505–509, where it is stated: “It is important to emphasize that the effective particle collision rates shown in Fig. 7b represent conservative lower-bound estimates, because 𝐶𝐸_{𝑏𝑒𝑠𝑡} is constructed as a lower envelope constrained by experimental measurements at 200 nm and 800 nm (Sect. 3.3).” Such contradictions need to be removed. The capabilities of the new setup should be assessed realistically throughout the manuscript.
Lines 290–293: “In immersion or deposition freezing experiments, grayscale-based analyses typically identify the temperature of complete freezing, which reflects bulk solidification of the droplets and is appropriate when nucleation occurs throughout the liquid or vapor phase (Kiselev et al., 2017; Thornton et al., 2023; Thompson et al., 2025)”: None of the indicated references describes a grayscale-based analysis in detail. A reference that would be better suited is Miller et al. (2021). Moreover, I am not aware of any technique that uses a grayscale-based analysis for deposition nucleation. Please give a suitable reference for such an instrument or remove “deposition freezing”.
Lines 344–370: In this paragraph, the performance of the dehumidifier at -15°C is discussed. Is the dehumidifier kept at this temperature throughout a cooling ramp? If yes, what are the consequences when the stage is at higher or lower temperature? Is there still time for the particle flow to equilibrate with the stage temperature and humidity before it reaches the droplet? Please clarify.
Lines 438–452 (adjustment of collision efficiency to experimental values): the experimentally derived collision efficiencies for 200 nm and 800 nm particles are considered here a lower limit (a “lower envelope” as stated on line 441) assuming that one collision triggers freezing. Yet, the collision efficiency might also be much higher with an upper limit of 1, or even lower if freezing occurred via immersion mode on previously collected particles. Therefore, the uncertainty is even larger than discussed. This uncertainty also needs to be acknowledged in the conclusion and in the abstract of the paper.
Lines 491–492: what is the definition of the median effective particle nucleation rate?
References
Miller, A. J., Brennan, K. P., Mignani, C., Wieder, J., David, R. O., and Borduas-Dedekind, N.: Development of the drop Freezing Ice Nuclei Counter (FINC), intercomparison of droplet freezing techniques, and use of soluble lignin as an atmospheric ice nucleation standard, Atmos. Meas. Tech., 14, 3131–3151, https://doi.org/10.5194/amt-14-3131-2021, 2021.
Citation: https://doi.org/10.5194/egusphere-2026-3946-RC1
RC2: 'Comment on egusphere-2026-3946', Anonymous Referee #2, 31 Aug 2026

The manuscript introduces a new experimental approach to quantify the temperature- dependent contact freezing probability for well defined aerosol particles in the laboratory. This is still an under researched field and therefore of interest for the cloud science community.

There are several critical points however, that should be resolved before publication.
My main issues are:
The manuscript lacks a thorough description of the experimental setup and the measurement procedure and a detailed discussion of possible sources of errors. Questions that need to be answered:
How was the aerosol prepared (especially dry or wet dispersion?) Are particles allowed to collide with the droplet from the beginning (i.e. T=ambient) or when (or at which temperature) is the aerosol flow admitted?
From this moment on, aerosol particles might be scavenged into the droplet and might cause immersion freezing at a later time. Please specify the rate of collisions between particles and the droplet, If I judge from Fig. 7b, ~10 particles collide with the droplet every second, so several hundred particles have been scavenged by the droplet before freezing is observed. How can contact freezing be disentangled from immersion freezing under these conditions?
Can deposition ice formation on the particles in the precooler or in the high humidity environment of the experiment chamber be excluded? The precooler might include regions with a temperature gradient normal to the flow direction and thereby may create regions of supersaturation similar to a diffusion chamber.
On collision efficiency (CE): impact CE should depend on the distance between nozzle and droplet (like in an impactor). What was the distance between nozzle and droplet? Why is this not considered when calculating impact CE? What was assumed for the charges on droplet and aerosol particle? Even tiny charges on the droplet may play a role here. Were polarization (image charge) effects taken into account?

Citation: https://doi.org/10.5194/egusphere-2026-3946-RC2
RC3: 'Comment on egusphere-2026-3946', Anonymous Referee #3, 31 Aug 2026

The manuscript describes a device/technique that is designed to measure impact initiated contact freezing. This is an interesting topic, and data is needed. I don’t think we can currently assess just how important contact freezing may be in the atmosphere because we do not have the data needed for such an assessment.
There are some points that need to be addressed in this manuscript though.
Contact freezing is so hard to measure because of the “contact” not the freezing. There are a lot of ways to determine whether a droplet freezes, and I think the method that the authors describe works well for this purpose. I am not convinced that the authors have adequately constrained the “contact” portion of contact freezing.
The assumption is that AgI aerosol will cause freezing of the droplet upon contact, but that remains an assumption. The collision rate can be calculated, as the authors have done, but it isn’t experimentally verified. The particles may be deposited onto the surface of the droplet, then catalyze freezing at some later time. Is that still contact freezing? The stream of air that is used for the aerosol flow is undoubtedly setting up some circulation within the droplet, which could entrain particles from the surface into the interior, which means that you now have a convolution of contact freezing and immersion freezing. Has there been an attempt to show that this is not happening?
This issue is one that almost every attempt to measure contact freezing must confront. Davis et al used the direct approach for contact efflorescence, which isn’t always possible for freezing and submicron particles. Hoffmann et al. studied contact freezing in an electrodynamic trap, and they were not able to “see” the particles colliding with the suspended droplets, but they were able to draw some conclusions about contact vs immersion freezing by examining the functional form of the frozen fraction vs time curves (see their Fig. 2). (They also verified collision rates by looking at particle residuals in droplets with an SEM.) Niehaus et al studied contact freezing, using droplets on a cold stage. They attempted to constrain contact vs immersion freezing by recording the first freezing event as “contact”, then melting the droplet and subsequently cooling to the original temperature. Presumably the particle at the surface would have been subsumed into the bulk, where it would catalyze freezing in the immersion mode. They did not see such events, leading to the conclusion that they were measuring contact freezing.
Considering the emphasis that the authors place on the “impact” in this manuscript, I think some discussion of the context and theory of contact freezing is warranted. I’m not advocating for a review paper, but outlining why impact might matter is important for this discussion. The two papers I would compare and contrast are Cooper and Fukuta.
I would not stress “impact” in this manuscript. I don’t think “contact” is meaningfully measured, and impact even less so. What kind of impact will a half micron diameter particle have as it settles through the viscous boundary layer next to the droplet? The carrier gas may be moving at 4 m/s, but the particles are relaxing to a much, much lower speed as they approach the surface.
Other points:
lines 135-136: the assumption that contact freezing is promoted by a finite number of active sites is questionable. Contact freezing has been shown for soluble salts, where the existence of active sites is almost certainly not applicable. See Niehaus and Cantrell for the contact freezing results, but see the Davis paper already cited, where the phase transition is efflorescence, not freezing.
I think the data in Figure 3 also cast some doubt on the singular hypothesis. The range of freezing temperatures for the background is pretty broad. If the freezing were singular, wouldn’t that be a lot narrower?
Section 3.1, breath figures: The authors present a compelling case that the breath figures are indicative of freezing. The case is not compelling that this is the direct result of an impact. The freezing event might have been catalyzed in the immersion mode as outlined above.
Line 488: “...DMA-selected aerosols…” Is there an effort to correct for the fact that there will be doubly charged aerosol, which are approximately twice as big?
Lines 499-502: The authors assert that the broader range of freezing T seen when they use polydisperse aerosol is a result of the range of particle sizes. The shift to higher T is apparent, but is the slope of the curve appreciably different?

References:
Cooper, W.A., 1974. A possible mechanism for contact nucleation. Journal of Atmospheric Sciences, 31(7), pp.1832-1837.
Davis, R.D., Lance, S., Gordon, J.A. and Tolbert, M.A., 2015. Long working-distance optical trap for in situ analysis of contact-induced phase transformations. Analytical chemistry, 87(12), pp.6186-6194.
Fukuta, N., 1975. A study of the mechanism of contact ice nucleation. Journal of Atmospheric Sciences, 32(8), pp.1597-1603.
Hoffmann, N., Duft, D., Kiselev, A. and Leisner, T., 2013. Contact freezing efficiency of mineral dust aerosols studied in an electrodynamic balance: quantitative size and temperature dependence for illite particles. Faraday discussions, 165, pp.383-390.
Niehaus, J., Bunker, K.W., China, S., Kostinski, A., Mazzoleni, C. and Cantrell, W., 2014. A technique to measure ice nuclei in the contact mode. Journal of Atmospheric and Oceanic Technology, 31(4), pp.913-922.
Niehaus, J. and Cantrell, W., 2015. Contact freezing of water by salts. The journal of physical chemistry letters, 6(17), pp.3490-3495.

Citation: https://doi.org/10.5194/egusphere-2026-3946-RC3

Jingchuan Chen, Upasna B. Rai, Kyle A. McMillan, Ryan D. Davis, Margaret A. Tolbert, and Sarah D. Brooks

Supplement

https://doi.org/10.5194/egusphere-2026-3946-supplement

Data sets

Cold-stage system for Impact-initiated Contact freezing Experiments (C-ICE) Jingchuan Chen, Upasna B. Rai, Kyle A. McMillan, Ryan D. Davis, Margaret A. Tolbert, and Sarah D. Brooks https://doi.org/10.18738/T8/4LM3BB

Video supplement

Jingchuan Chen, Upasna B. Rai, Kyle A. McMillan, Ryan D. Davis, Margaret A. Tolbert, and Sarah D. Brooks

Viewed

Total article views: 268 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
180	61	27	268	38	16	22

HTML: 180
PDF: 61
XML: 27
Total: 268
Supplement: 38
BibTeX: 16
EndNote: 22

Views and downloads (calculated since 28 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	75	21	6	102
Aug 2026	75	24	16	115
Sep 2026	30	16	5	51

Cumulative views and downloads (calculated since 28 Jul 2026)

Month	HTML	PDF	XML	Total
Jul 2026	75	21	6	102
Aug 2026	75	24	16	115
Sep 2026	30	16	5	51

Viewed (geographical distribution)

Total article views: 240 (including HTML, PDF, and XML) Thereof 240 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 08 Sep 2026

Short summary

Ice nucleating particles influence cloud ice formation, affecting precipitation and climate. Yet freezing triggered by particle-droplet collision remains difficult to quantify. We developed a cold-stage method that directs controlled aerosol flows onto single water droplets. Validations with silver iodide show reliable measurements, clear separation from background freezing, and particle-size effects. This work can support better representation of cloud ice formation in models.


Total:	0
HTML:	0
PDF:	0
XML:	0