Dear Junfeng Liu,

Thank you for reviewing our manuscript and your positive evaluation. Please find our response to your comments below.

1.    In the revised manuscript, we slightly adjusted the color scheme of Figures 2 and 4 to make the graphs more distinguishable.

2.    Indeed, the downstream ozone concentration starts to increase during the first two hours of flushing with air (60% rH, 50 ppb O3) at 80 mL/min. This is in contrast to the KI-loaded filters, which showed stable ozone concentrations at baseline level over the course of 10 hours. Nevertheless, the overall increase of the ozone concentration from around 5 ppb to around 9 ppb over the course of 10 hours is still relatively low, given the input concentration of 50 ppb.

3.    This experiment was already repeated to ensure the correctness. In both cases, a decrease of the Linalool concentration with increasing ozone concentration was observed, both with and without the use of one of the ozone filters. The comparably high variations in the triplicates was also observed, showing that both ozone filters are not reliably depleting ozone under these conditions. 

Dear Reviewer #2, 

thank you for carefully reviewing our manuscript, the very positive evaluation and your constructive feedback. Please find below our response to your comments.

1.    Both samplers are identical MTS-32 multi-tube sequential samplers (Markes International Ltd., UK), set up with the same parameters (sampling time, pump flow rate) and were operated in parallel next to each other. Additionally, the flow through all tube positions of both samplers was verified to be equal (±5%) before the experiment using a flow meter (7000 GC flowmeter, Ellutia Ltd., UK). This description was added to the revised manuscript.

2.    To evaluate the ozone filters, we used two scrubber materials (KI, Na2S2O3), two approaches (load-and-flush, permeation), each at four relative humidity levels (<7%, 30%, 60%, 90%) with three ozone concentrations (baseline, 25 ppb, 30 ppb). For each experiment, three samples and one control sample (without filter) were generated, resulting in a total number of 192 samples.

3.    Thank you for pointing out the missing information. The figure capture was modified according to your suggestion.

4.    Recovery rates exceeding 100% in the load-and-flush approach are likely attributable to inaccuracies in the preparation of the custom-made stock solution used for the experiments and the application of the diluted standard solution using a microliter syringe. In contrast, the calibration curves used for quantification were based on a commercially available, certified stock solution, offering greater accuracy.

In the evaluation of the permeation experiments, recovery rates were normalized to those of the unfiltered samples at baseline ozone concentrations at the corresponding relative humidity levels. As a result, values exceeding 100% are artifacts of the normalization procedure. Without normalization, the recovery rates were below 100%, except for α Pinene at <7% rH and <25 ppb O3, as detailed in lines 382f and illustrated Figure S5.

Dear Reviewer #3, 

thank you for carefully reviewing our manuscript and your constructive comments, which helped to improve our work. Please find below our response to your comments.

1.    We adjusted the schematic layout of the ozone scrubber design according to your suggestion and added markers for the individual parts. The figure caption was adjusted accordingly.

2.    The glass filters were placed in a filter holder and rinsed with the respective salt solution until complete wetting, as described in lines 126f. We expanded the experimental description for loading the ozone filters and added the preparation of the salt solutions.

3.    The flow requirements of the ozone monitor (>600 mL/min) were added in the revised manuscript.

4.    The desired flow rate through the single glass filters (80 mL/min) was added in the revised manuscript.

5.    An explanation of the experimental conditions was added to the “Experiment” columns in Tables S3 and S4. Additionally, the experiments at low relative humidity were renamed to “<7 – 0”, “<7 – 25” and “<7 – 50” to be consistent with the conditions named throughout the manuscript.

6.    For the purpose of the manuscript, a detailed description of the TD-GC-MS results, aside from the measured concentrations, is not the objective and would be beyond the scope of this paper. We therefore prefer to not include this detailed information in the manuscript. Nevertheless, the used TD-GC-MS system was calibrated using a terpenes standard solution (Terpenes MegaMix Standard #1, Restek GmbH, Germany), showing a sufficient chromatographic separation. The consistency of retention times and mass spectra for the mixture of α Pinene, Myrcene, Limonene and Linalool used in the experiments was also verified prior to conducting the experiments. 

7.    Looking at the reactions of the ozone depleting reagents with ozone (Eq. 1 and 2), both reactions require water to take place. While water is a reactant in the ozone depletion reaction using KI, the reaction of Na2S2O3 takes place in an aqueous medium. When water is present in the sampled air stream, both reactions can occur, with higher efficiency at increasing relative humidity levels. At very low relative humidity levels, both reactions are inhibited due to the lack of a reactant for KI and a suitable reaction medium for Na2S2O3. An explanation of the results was added to the revised manuscript.

8.    Yes, the KI-loaded ozone filters still have the same ozone removal potential after 10 days of exposure to ambient air, as shown in Figure 4. Please note that exposure does not mean they are actively flushed with sampled air. Using the same ozone filter twice was not tested, but could be possible based on the breakthrough experiments (10 hours of flushing). Nevertheless, based on our experience from field experiments, we would advise against reusing the filters for a second sampling without applying new reagent, since visual changes (i.e. yellowing) of the filters are observed regularly. 

9.    We added a simplified reaction of copper with ozone in the revised manuscript. The detailed reaction mechanism would go beyond the scope of our work. More detailed information can be found in the cited publications by Kim et al. (2023), Lin and Frankel (2013), and Ma et al. (2024).