the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 20 Aug 2026
Temperature uncertainty characterisation of the FrESH droplet-freezing assay
Abstract. Reliable offline measurements of ice-nucleating particles (INPs) with droplet-freezing assays require accurate assignment of droplet temperature. We present FrESH (Freezing Experiment Setup Helsinki), a dual-plate immersion-freezing instrument developed for high-throughput analysis of filter-collected aerosol samples. FrESH uses standard 96-well polymerase chain reaction plates cooled in an ethanol bath with optical detection of freezing.
The relationship between bath temperature (TBT) and droplet temperature (Twell) was characterised using five PT100 sensors at fixed temperature and during cooling ramps for two chiller models. Wells were consistently warmer than the bath, and the bath-to-well relationship was approximately linear over the range from 0 to -35 °C, with plate-averaged relations of the form Twell≈0.97 TBT + (0.2–0.4) °C. Root-mean-square residuals of the linear fits were generally around 0.1 °C and remained below 0.2 °C in the sensor-resolved summaries. Analysis of routine FrESH measurement data revealed persistent spatial freezing patterns across the plates, consistent with position-dependent temperature variability. Combining the identified sources of uncertainty gives an estimated droplet-temperature uncertainty of 0.43–0.44 °C (1σ) over the operational range. This temperature characterisation is used to assign droplet temperatures and associated uncertainties in FrESH-derived NINP(T) spectra.
- Preprint
(3909 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-3971', Anonymous Referee #1, 07 Sep 2026
-
RC2: 'Comment on egusphere-2026-3971', Anonymous Referee #2, 22 Sep 2026
Perez Fogwill & Welti (2026) present a description of the droplet-freezing assay FrESH with a focus on the systematic temperature-uncertainty characterization. They decompose the total temperature uncertainty into the contributions from temporal resolution, bath-probe uncertainty, the static bath-well relationship, position-dependent variability, and dynamic effects during cooling ramps. This is done in a methodologically sound way, but without any innovation. The mechanical design of FrESH, with the floating PCR plate holder frame, is however a small but novel design choice, which mitigates fill-level issues that similar freezing assays might have. Overall, the manuscript does not have many weak points and is a solid characterization of the droplet-freezing assay, but it also does not bring innovation and novelty to the table of droplet-freezing assays. Since the scope of AMT does not explicitly demand novelty, I see no reason to reject the manuscript and hence recommend it for publication after revisions.
Comments
Section 3.2.
At the end of 3.2. you mention the biggest weak point of your applied approach (thermal and non-thermal contributions cannot be separated).
It is not clear to me why more PT100s were not simply used. If there is a valid reason it should be mentioned.
Also, if you are limited to five sensors for whatever reason, could you not have randomized the PT100 positions across several runs? That could provide a full picture of the temperature distribution with actual temperature measurements instead of the more indirect approach using freezing statistics. Since the spatial uncertainty was identified as the largest or second-largest source of uncertainty (depending on chiller model), the authors should try to address and characterize this uncertainty as best as possible.What samples were used to derive the freezing statistics? Did you, and if so how, ensure that the sample suspension was homogeneous? Did you fill the PCR plates in an ordered manner or randomly? Depending on your procedure, it might be possible, for example, that certain particles separate in the pipette through gravitational settling, so that the first drops contain more of certain particles, which in turn affects the spatial freezing pattern if the first drops are always placed, e.g., in the top left corner of a PCR plate.
Section 4.1.
Fig. 2
At bath temperatures of -25°C and below, there seems to be a larger step up in the 1σ range compared to the higher temperatures. Do the PT100 sensor readings get less accurate at ≤ -25°C, or is it another effect?
Section 4.3
You combine the uncertainty terms in quadrature, which implies that you see them as uncorrelated signals, but the dynamic and spatial terms, for example, likely share a common cause (bath circulation/heat transfer), which could mean some double-counting or under-counting of correlated error. How do you justify combining the terms in quadrature?
Conclusions section
The authors show that there are consistent temperature bias patterns within one PCR plate and also from plate to plate. However, in the end, they seem to plan on using chiller-model-specific values for the uncertainty and for the correction of the bath-well offset. Can you explain your reasoning behind this?
Further recommendation
As mentioned in my introduction, I am missing innovative and new aspects in the presented manuscript. One aspect that could be explored is different cooling rates and the effects of chiller-specific parameters like cooling capacity and pump flow. For cold-stage-based freezing assays, the effect of different freezing rates was investigated by, e.g., Jakobsson et al. (2022); for freezing assays using PCR plates, different cooling rates are largely unexplored.
References
Jakobsson, J. K. F., Waman, D. B., Phillips, V. T. J., and Bjerring Kristensen, T.: Time dependence of heterogeneous ice nucleation by ambient aerosols: laboratory observations and a formulation for models, Atmos. Chem. Phys., 22, 6717–6748, https://doi.org/10.5194/acp-22-6717-2022, 2022.
Citation: https://doi.org/10.5194/egusphere-2026-3971-RC2
Data sets
Data for "Temperature uncertainty characterisation of the FrESH droplet-freezing assay" Germán Perez Fogwill https://doi.org/10.57707/fmi-b2share.zf2e1-9nd11
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 135 | 54 | 62 | 251 | 58 | 62 |
- HTML: 135
- PDF: 54
- XML: 62
- Total: 251
- BibTeX: 58
- EndNote: 62
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents a temperature uncertainty characterization of the FrESH droplet-freezing assay. Accurate temperature assignment is undoubtedly important for offline measurements of ice-nucleating particles. The authors have also performed a careful set of static and dynamic temperature measurements and constructed an uncertainty budget for the instrument.
However, I have a major concern regarding the novelty and scientific contribution of the present manuscript. In its current form, the work is largely limited to temperature characterization of a specific droplet-freezing instrument. This is primarily an instrumental and engineering characterization problem. I do not see a sufficiently substantial technical innovation or scientific advance that would justify publication as a standalone study. A substantially stronger manuscript would require additional experimental content. In particular, an intercomparison with an established instrument and/or a scientific application demonstrating the benefit of the improved temperature characterization would considerably strengthen the work. Therefore, I cannot recommend publication in its present form.
Major Comments: