This work describes a new method for measuring the viscosity of aerosol particles as a function of temperature. The authors combine a hot-stage microscopy method with fluid dynamics simulations and validate their method with a known standard. When applying the method to lab-generated SOA, viscosity values that were 1-2 order of magnitude higher than predictions were observed. This highlights the need for viscosity measurements of complex systems, and motivates the publication of this research. 

While the method appears to work as applied to one validation system, it is hard to see the generality of this method as a tool for viscosity characterization without further validation. The authors report just 2 data points and conclude the method is valid, and go on to show orders of magnitude differences in a sample system (compared to estimates based on models). In my opinion, more validation systems need to be explored before the applicability of this method can be established. 

In addition to my general critique above, I have some specific comments:

• Can this method work for substances at room temperature or below? What is the viscosity range that is accessible to this technique? 
• It seems that the technique could equally measure the melting temperature - can this be related to Tg and then correlated with a viscosity? This would bypass the need for fluid dynamics and may provide a simpler method for determining Tg than the traditional DSC methods. Is there an advantage to obtaining viscosity, which is then used to determine Tg anyway, in order to derive viscosity across the range of temperatures?
• The authors refer to this as a method for measuring aerosol particle viscosity, whereas in fact it is measuring the viscosity of aerosol material (they are no longer individually resolved particles). 
• It is unclear to me if the sample is on a needle or a slide? From the schematic, the images are looking from below, so is the particle on the side of the needle? What is the effect of gravity on the shape of this system?

Kiland et al. show how a microscope interfaced with a hotplate  can be used to measure the temperature dependence of viscosity for samples where less than 1 mg of material is available. The technique is compared against one substance with known temperature dependence of viscosity. It is further shown that the technique can be applied to secondary organic material collected from an environmental chamber experiment. These data are discussed in the context of parameterization-to-data comparison.

The presented technique is a new addition to the toolbox of scientists who are interested in measuring the viscosity of organic aerosols. The underlying techniques are sound. I am excited to see this work being pursued and I recommend the paper for publication. However I have some suggestions on how the data interpretation might be improved. 

Comments:

The technique focuses on bulk samples. Therefore secondary organic material should be used throughout to distinguish it from aerosol sampling. 

TαNB was heated above the melting point of ~185°C and then cooled to generate glassy material. This step is not mentioned for the SOM. Presumably,  the particles turned glassy in the chamber or the -20°C freezer. Was heating of the SOM avoided due to the possibility of evaporation of more volatile components, thermal decomposition, and thermally induced chemical reactions? Furthermore, not all crystalline substances can be heated above the melting point without decomposition. Presumably this is the reason why sucrose - a seemingly more logical choice to validate the system due to its widespread use in the community - was not used. It would be helpful to add some discussion that highlights the assumptions and limitations associated with the sample preparation requirements. 

The abstract and manuscript puts a lot of emphasis on the comparison with the DeRieux et al. parameterization, showing that the observed values were “significantly higher (1–2 orders of magnitude)” than values predicted by a parameterization. I have some concerns regarding this framing. Foremost, the emphasis in the abstract distracts from the fact that this is a methods paper that has the main purpose to show that viscosity can be measured for small samples, as shown by the comparison with the TαNB bulk data. The SOM experiment nicely demonstrates that the measurement on collected samples from smog chambers is feasible. The apparent discrepancy to a parameterization is probably neither significant nor particularly important in the context of the known experimental and modeling uncertainties. 

The model in this manuscript heavily relies on the assumption of D = 10 from DeRieux et al.and zero error in the Tg prediction. The pure component data underlying the D = 10 estimate has significant scatter. We actually report D = 7 for a-pinene SOA (Petters and Kasparoglu, 2020), which is an almost direct measurement. The overestimated viscosity by 1-2 orders of magnitude reported in the abstract is based on the assumed D and trusting that Tg is known perfectly. I do not believe that better closure should be expected given the uncertainty in composition, aerosol measurement, and uncertainties in the parameterization. 

Figure 7 explores the temperature dependence of viscosity for the SOA in the context of D, showing that no single D value can reproduce the measurements. As discussed in DeRieux et al. and further explored in Petters and Kasparoglu, Tg current viscosity data for a-pinene SOA imply a scatter in the estimated Tg of +/- 25K or with constrained D parameter and hygroscopicity +/-10K (see their Figure 3 and the associated text). Tethering Figure 7 in Kiland to a single Tg value is thus not quite appropriate. Furthermore, Tg can take on a range of values depending on the cooling rate of the substance, or in the case of the chamber the way that the SOA was formed. It would seem more logical to fit the temperature dependent viscosity data using a constant D value and extrapolate the fit to to 10^12 Pa s to find the Tg. The procedure could be repeated for a range of D values that the authors believe might be plausible. This is similar to the procedure we used in Rothfuss and Petters (2017) and Marsh et al. (2018), where we find that extrapolated Tg from a VFT fit compares to +/- 10K with literature values for select substances.  Comparing the extrapolated Tg from the measurements to the predicted Tg likely results in closure within a reasonable interval.

Regardless, given the uncertainty in aerosol chemical composition, the possibility of source fragmentation of farnesene SOA during HRMS measurement, the possibility of sample aging via partial evaporation during storage or due to chemical reactions, and the possibility that the parameterization is simply insufficient to model Tg precisely for all compositions, agreement or disagreement with the DeRieux parameterization is too unconstrained to reach all but the broadest conclusion that the data presented here are reasonable. The subtleties associated with the experiment/parameterization comparison, however, should be more thoroughly discussed in the manuscript. 

Other comments

“Rothfuss and Petters used a dimer relaxation technique to measure viscosities of SOA in the range of 5 × 10^5 to 2 × 10^7 Pa s at temperatures ranging from −15 to 80 °C (Rothfuss and Petters, 2016)”

Rothfuss and Petters did not apply this technique to SOA. Application to SOA is reported in S.Petters et al. (2019) and Champion et al. (2019) 

“On the other hand, this means samples cannot be collected and measured at a different time and different location using this technique.”

I do not understand the value of location and time in this context. The fundamental limitation of the dimer relaxation technique as described in Rothfuss and Petters  is the availability of sufficient number concentration, which makes it difficult to apply the technique to aged aerosol in environmental chambers. However, absent this limitation, it would be preferable to not have to separate the particles from the gas-phase and measure it “then and there”. 

“The area and L  were determined from the sequence of binary images using a MATLAB script.”

The programming language is probably not important here. However, it would be helpful to state the type of image analysis that was applied as well as the potential limitations. For example, does the analysis use edge detection? Is it sensitive to lighting conditions? Does the analysis require adjustment of parameters to get it right  or does it work fully automated out of the box? 

References Cited

Champion, W. M.; Rothfuss, N. E.; Petters, M. D.; Grieshop, A. P. Volatility and Viscosity Are Correlated in Terpene Secondary Organic Aerosol Formed in a Flow Reactor. Environ. Sci. Technol. Lett. 2019, 6 (9), 513–519. https://doi.org/10.1021/acs.estlett.9b00412.

Marsh, A.; Petters, S. S.; Rothfuss, N. E.; Rovelli, G.; Song, Y. C.; Reid, J. P.; Petters, M. D. Amorphous Phase State Diagrams and Viscosity of Ternary Aqueous Organic/Organic and Inorganic/Organic Mixtures. Phys Chem Chem Phys 2018, 20 (22), 15086–15097. https://doi.org/10.1039/C8CP00760H.

Petters, S.; Kreidenweis, S. M.; Grieshop, A. P.; Ziemann, P. J.; Petters, M. D. Temperature- and Humidity-Dependent Phase States of Secondary Organic Aerosols. Geophys. Res. Lett. 2019, 46 (2), 1005–1013. https://doi.org/10.1029/2018GL080563

Petters, M. D.; Kasparoglu, S. Predicting the Influence of Particle Size on the Glass Transition Temperature and Viscosity of Secondary Organic Material. Sci. Rep. 2020, 10 (1), 15170. https://doi.org/10.1038/s41598-020-71490-0.

Rothfuss, N. E.; Petters, M. D. Characterization of the Temperature and Humidity-Dependent Phase Diagram of Amorphous Nanoscale Organic Aerosols. Phys Chem Chem Phys 2017, 19 (9), 6532–6545. https://doi.org/10.1039/C6CP08593H.