In article EGUSPHERE-2023-1241, "Absorption of VOCs by polymer tubing: implications for indoor air and use as a simple gas-phase volatility separation technique," the authors present an in-depth discussion and model of uptake of gases into tubing materials widely used by the atmospheric measurement community, as well as a technique and apparatus for exploiting this process to provide potential resolution for direct-sampling instruments such as chemical ionization mass spectrometers. This work is a sort of continuation of a number of useful and important papers on this topic from this research group and advances the ideas by providing a model for use. It is generally very well written and presents the data clearly and usefully. I have a few core questions that I would like to see the authors respond to, but I believe only relatively minor revisions are necessary prior to publication.

General comments:

1) In this and previous work, the authors use only a series of ketones. I would like to see a mention or discussion of what impact this might have on the results. Do the authors believe there would be any effect due to e.g., polar functional groups or other chemical classes?

2) It is not completely clear how the MLM and SLM are related, for instance in Sections 2.2.1 and 2.2.2. Are they wholly different? It seems to me that the multi-layer model should collapse to the single layer model for the case the the fraction of polymer available for bulk uptake goes to essentially 0, is that the case? Presumably there is at least some capability for bulk uptake even in the single-layer materials. Is there a downside to using the multi-layer model for all tubing? Does it better represent the time series of the slow diffusing materials (e.g., nylon, Figure 3)? If so, what qualifies as "requir[ing] the multilayer model" to be deemed fast-diffusion (line 230). Similarly, if the MLM were used for slow diffusers, couldn't you extract the partitioning depth directly and directly compare Df between slow and fast diffusers, instead of relying solely on the literature values in Figure 4 (lines 239-247)?

Also, I'm not totally clear, how many layers are in the multi-layer model? Is it just a surface and a bulk, or are there varying depths? Or am I just thinking about it wrong?

3) A result that is implicit through this work is that that nylon delays are similar to PFA, and there is no discussion of outgassing. Should we all just be using nylon for gas sampling, since it is way cheaper? Where is nylon on Figure 8b? Was any outgassing observed? None of this is discussed, and it seems to me could be a radical shift of the atmospheric if I am understanding the results correctly.

4) The applications of the GSV are conceptually interesting, but I'm a bit skeptical of its utility. For instance, in the example given using a long length of teflon as a separator (lines 369-374) - couldn't you just sample through a shorter length of chromatography column and get the same effect? Similarly, in the application of using the GSV as a volatility pass filter, and in many of the applications of the GVS I can think of, seem to me to include signficant complications to interpretation. Because the system is constantly moving toward an equilibrium, the fraction transmitting is constantly changing, so it is not a fixed separator. It is certainly an interesting idea, but strikes me as difficult to draw quantitative meaning from.

Technical comments:

Line 109. The definition of delay time is a little confusing, please rephrase

Line 125. The Cole-Parmer 3-way solenoids have a fairly tortured flow path that can introduce delays and losses, even in PFA. Do the authors see a difference in delay times between PFA lines with and without these in line? An issue that has come up in many applications is the impact of these and similar valves on the sample flow stream, this seems like a good opportunity to examine the issue, which would be helpful for the gas sampling community. The authors allude to this on line 356.

Lines 201-209. I hope you at least automated the brute forcing! If not that sounds exhausting (and more prone to researcher error).

Table 2. Values for C* here are an order of magnitude larger for the same materials tested by Deming et al. Is that a function just of different tubing geometries, and if so, can they be made more directly comparable? If not, what is the reason for that?

Line 291. Typo, should be "cPFA"

Figure 7. It does not look obvious to me that silconert has a much slower response time than cPFA, though Figure 5 (and the text) says so. Am I just unable to see it clearly, or am I missing something. Also, the darkest blue line seems to be missing from cPFA during desorption, or maybe the colors are just different.

Line 324. Was cPFA or cPTFE similarly tested for losses? Perhaps that was shown in a previous publication? That would be helpful, since a conclusion of Figure 5  is that cPFA is a better option than the coated metal options (and easier to work with).

Line 332-336. It is somewhat unorthodox to include in the SI a whole subject/experiment.discussion that is not really a part of the main work or main text. I understand that these experiements may not constitute a separate paper, so perhaps this is a appropriate, but it is certainly unusual, and I would encourage the authors to consider whether this manuscript is actually the right place for this information given that the whole section is relegated to the SI and no other discussion or use of OFRs is included in the main text.

Line 335. "Supplementary information" or something like that instead of "supplemental"

Line 343-345. There is no counterfactual shown about what happens when a PFA period does not follow cSI - does it increase less quickly? The following statement about depassivation is reasonable, but there is no real evidence presented to demonstrate it.

Line 388-392. It is beyond the scope of this work for sure, but I'm curious if the authors how any thoughts on how to incorporate these findings into a better understanding of indoor air? How would one go about modeling a carpet, which has very different geometry and surface area ratios than a tube? Painted surfaces are fairly uniform, but many of the surfaces described here have complex fractal-like geometries.

Review for manuscript entitled, “Absorption of VOCs by polymer tubing: implications for indoor air and use as a simple gas-phase volatility separation technique” by Morris et al

The manuscript describes a new measurement technique to separate gas-phase organic compounds by volatility using different types of sampling tubing. This is an innovative approach that can be leveraged by the broader atmospheric research community to address major research challenges related to, for example, gas and aerosol chemistry of volatile organics in complex mixtures. It is also incredibly valuable that the research group publishes all the models they have developed on their website that is open access to everyone. The introduction is very clear and nicely places the work in context with prior literature. The figures are publication quality and clearly demonstrate major findings from the systematic testing of partitioning delays of different tubing material across a range of compound volatilities. However, the paper seems to end prematurely without demonstrating the use of the gas volatility separator technique in a real measurement application with a mixture of organics. How would one leverage this technique to develop plots of the relative contribution of different volatility classes in the original sample? That still isn’t entirely clear. I also think the manuscript could be strengthened by focusing the tables and figures in the main manuscript on the most critical findings and leaving some of the details for the SI. It was a little challenging to pull the main points out of all the details. I recommend the paper for publication after some (mostly) minor revisions that address the concerns summarized here and described below.

• One of the key findings is the tubing categorization as slow vs fast vs adsorption only. Figure 2 is a nice illustration of the differences between the tubing types for the range of compound volatilities studied. Figure 3 is more challenging to read. Particularly 3b. There are too many lines with too many different colors and dashes. If the main point is just to illustrate that the SLM does not work for polypropylene, I think it would be more effective to show the time-series example for 1 of the ketones (not 2) and then also provide a more quantitative metric for the model prediction for all the compounds in a table next to the graph. What was your metric for deciding when the SML model worked or it didn’t? You provide all the details about the fit parameters for the model but no clear quantitative description to evaluate how well the models did at predicting the measurements. Even just a simple linear regression comparing modeled vs measured values at each time point could possibly work to draw a more objective line between a “good fit” and a “poor fit.”
• Section 3.1.3 doesn’t appear to be providing critical information about the key results. The only figures cited are in the SI, not in the main text. Perhaps this entire section could be moved to the SI? It reads as a minor technical point.
• Section 3.2. It was surprising to read that the absorption delay times are often larger than the desorption delay times for the slow-diffusion materials. This point appears to be glossed over fairly quickly, but could the authors describe the implications of this for using the GVS technique to classify volatility distributions from an ambient measurement? How long does slow-diffusion tubing need to be conditioned before reaching equilibrium? Why would there be more variability in absorption delays than desorption delays?
• Does it make sense for Section 3.2 to be it’s own section? I think the description of key findings in Figure 5 would be more clear if they were discussed in the same section instead of separated between sections 3.2 and 3.3.1 and 3.3.2.
• Figure 8b is the most important figure of the entire paper and it is buried at the end. I think it would be more effective to move this before some of the technical details about humidity effects on desorption of metal tubing or contaminant peaks in conductive tubing (new vs aged).
• The abstract states, “we demonstrate how to use a combination of slow- and fast-diffusion tubing to separate a mixture of VOCs into volatility classes,” but the paper actually stops just short of actually demonstrating that. Can you take the data shown in Figure 8a and develop a synthesized volatility distribution of the starting mixture from that data? Figure 8b shows the distribution being transmitted through each of the different tubes, but the demonstration is missing that final step of actually reverse-engineering the original volatility distribution. I think it would have been even more compelling to demonstrate how this would work with two different mixtures composed of contrasting contributions from SVOC/IVOC compounds.

MINOR COMMENTS

• Lines 252-255: It wasn’t entirely clear what the authors meant by “short times” and “long times” in this description. Do you mean earlier in the absorption phase and later in the absorption phase of the sampling cycle?
• Line 260: unclear what is meant by having to “split the difference” to model the whole time series.
• Lines 329-331: I think this sentence is supposed to be the final sentence of the preceding paragraph. Otherwise it is unclear what “these” materials refers to.
• Line 355: typo in the sentence, “We believe this due to the…”