The study investigates the photochemistry happening in atmospheric liquid water in Fairbanks, Alaska in winter. The authors evaluated the indirect photochemistry happening in the condensed phase under rain/fog conditions and under more concentrated aerosol liquid water conditions. I do appreciate the works that the authors have put in the manuscript and would support its publication. My expertise is in photochemistry and ozone chemistry in the condensed phase. I will focus my review on these topics and let other referees address the gas phase side of the manuscript.

Main comment

Aqueous-phase hydroxyl radical production. I think that the authors should mention the potential for ozone to generate hydroxyl radical in the condensed phase. Hydroxyl radicals can be generated in the condensed phase by ozone from mostly two reactions:

1) ozone reaction with superoxide:

 O₃ + O₂^•- → O₃^•- + O₂ k=1.6E⁹ M^-1 s^-1

O3^•- → O₂ + O^•-

Assuming that all the hydrogen peroxide formed during irradiation comes from superoxide dismutation, the authors should be able to estimate the superoxide steady-state concentration and the hydroxyl radical production rate from this reaction. This chemistry is discussed in details in Ma et al.¹ and Guo et al.²

2) Ozone reaction with DOC. It is known that ozone reaction with Dissolved organic matter (DOM) produces hydroxyl radical with a yield between 10-20% and that the reaction is dominated by DOM phenolic moities.^{3; 4} Using the ozone concentration of 3E^-9 M for the 2/4 sample and the carbon concentration of 22M for the same sample, the  2/4 sample phenolic moieties concentration can be estimated to be ≈0.54M using the relationship presented in Onnby et al.⁴ (note that this relationship was developed for DOM and that a value for atmospheric PM would be more appropriate). Using the apparent phenol, O₃ second-order rate constant at pH 5 (1.4E⁴) one can calculate a reaction rate constant of 2.3E^-5 M s^-1 and using a hydroxyl radical yield of 15% a hydroxyl radical production rate of 3.4E^-6 M s^-1. This is relatively higher than the hydroxyl radical production transfer rate presented on Table S6 for the same sample (1.9E^-7 M s^-1).

It should be noted that both pathways 1 and 2 are pH dependent and faster at higher pH due to superoxide pKa and to the apparent phenol, O₃ second-order rate constant being pH dependent.

Small comments/corrections

L.65, I think that the authors should also mention long-lived phenoxyl radicals as they play an important role in surface waters for the phototransformation of phenols.⁵

L.80 “In addition, Kapur et al. (2024) reports •OH in dark samples from the ALPACA campaign using EPR, noting their environmental persistence in the dark at low concentrations.” Production of Hydroxyl radical in the dark is known for DOM, see Page et al.⁶ I would suggest switching persistence for production.

L. 180 (equation 1). It is known that the non-linearized form of the equation is slightly better for fitting the data.

L. 413 ”Altogether, we attribute the wide range of Φ^1O2* values to minor differences in BrC sources or small changes in the degree of chemical aging”. Ozone exposure is also known to increase DOM Φ^1O2* dramatically.⁷

L. 460. “the average predicted lifetime of ³C* is 0.9(±0.6) nanoseconds”. This number is getting close to the lifetime of the singlet excited state. Boyle et al. fitted the decrease in DOM fluorescence using a 3 component exponential, the faster exponential had lifetime below 150ps while the medium exponential was between 0.3-1.3 ns and the longer exponential 2-5.5 ns.⁸ this suggest that singlet excited state may also play a role at ALW concentrations.

References

1. Ma J, Nie J, Zhou H, Wang H, Lian L, Yan S, Song W. 2020. Kinetic consideration of photochemical formation and decay of superoxide radical in dissolved organic matter solutions. Environ Sci Technol. 54(6):3199-3208.
2. Guo Y, Yu G, von Gunten U, Wang Y. 2023. Evaluation of the role of superoxide radical as chain carrier for the formation of hydroxyl radical during ozonation. Water Research. 242.
3. Sonntag C, Gunten U. 2012. Chemistry of ozone in water and wastewater treatment: From basic principles to applications.
4. Onnby L, Salhi E, McKay G, Rosario-Ortiz FL, von Gunten U. 2018. Ozone and chlorine reactions with dissolved organic matter - assessment of oxidant-reactive moieties by optical measurements and the electron donating capacities. Water Research. (144):64-75.
5. Remke SC, Burgin TH, Ludvikova L, Heger D, Wenger OS, von Gunten U, Canonica S. 2022. Photochemical oxidation of phenols and anilines mediated by phenoxyl radicals in aqueous solution. Water Research. 213.
6. Page SE, Sander M, Arnold WA, McNeill K. 2012. Hydroxyl radical formation upon oxidation of reduced humic acids by oxygen in the dark. Environ Sci Technol. 46(3):1590-1597.
7. Leresche F, McKay G, Kurtz T, von Gunten U, Canonica S, Rosario-Ortiz FL. 2019. Effects of ozone on the photochemical and photophysical properties of dissolved organic matter. Environmental Science & Technology. 53(10):5622-5632.
8. Boyle ES, Guerriero N, Thiallet A, Del Vecchio R, Blough NV. 2009. Optical properties of humic substances and CDOM: Relation to structure. Environ Sci Technol. 43(7):2262-2268.

Note on reviewers: Nadine Borduas-Dedekind at UBC worked with two graduate students, Keighan Gemmell and Claudia Sardena to compile this review.

Overview:

In this article, the authors quantify the concentrations of three photo-produced oxidants from PM_2.5 filters in Fairbanks, Alaska. These filters were extracted in the lab and then illuminated with a xenon lamp to measure the production of OH radicals, triplet excited state BrC, and singlet oxygen. The measurements were then used to model their concentration as a function of aerosol liquid water content. The authors found that laboratory-irradiation of wintertime extracts can generate substantial amounts of these oxidants. They conclude, like in their title, that once modeled they are surprised at how many oxidants could be produced.

This manuscript is nicely suitable for ACP and makes an important contribution to the field by working with field-collected samples, in the laboratory and using these measurements to predict concentrations. There is a large quantify of data presented (which at times had us confused and other times impressed, see comments below), and the presentation could also be improved for clarity and conciseness, but overall, congratulations to the authors.

General comments:

L220-225 & 271: We wondered how the presence of H₂SO₄ could impact the lifetime of the excited state oxidants? A major sink of excited state oxidants, specifically singlet oxygen, is the deactivation with the solvent. Is the deactivation of these oxidants the same in H₂O and H₂SO₄(aq)? What might be the additional role of ionic strength of the solution on singlet oxygen lifetime?

Dark OH radical chemistry: line 80-81, the authors cite (Kapur et al., 2025) about the dark OH source. When estimating OH radical concentrations in ambient PM, how would these concentrations be affected by the dark OH source? Was this dark source taken into account? If so how? (We didn’t see these calculations in Table S18.)

Liang et al. 2024 showed that triplets oxidize SO2 (not accounted for in this study’s model as per line 561) and could be further mentioned in the introduction on line 92.

It might be worth adding some clarity over the filter description. Lines 120-123 explains that these filters are 24h, but Table S1 suggests different times, including the footnotes. If different numbers of filters were mixed together, then their mass would be different, and we’re wondering how to interpret this inconsistency. Why did they authors extract different number of filters for their extracts? Also, Table S1 footnote mentions different collection times and that the timing was corrected, but we were confused at how this value was corrected. We think this information could benefit from additional clarity in the main text and on Figure 2. How did mixing the filters together affect the interpretation of the results (how to compare such different masses)?

Specific comments:

Lines 302-303 & 318: “Moreover, BrC and HONO both have significant UV-A absorbance and are not impacted by the suppression of UV-B photons in winter sunlight…”. BrC absorbs in the UV-B range, and can be impacted by the suppression of UV-B photons in the winter. We agree that there is significant BrC absorbance of UV-A light, so that BrC excitation can still occur in the winter, but wouldn’t it be impacted by UV-B photons? This point is reiterated in line 318 which the authors might want to consider clarifying.

Line 390: “The lack of correlation might be caused by the specificity of the syringol probe, which only quantifies the oxidizing subset of the 3C* population, while 1O2* is produced from the total 3C* population”. We agree, and there might be additional references to consider to help support this point like: (Bodesheim, Schutz and Schmidt, Chem. Phys. Lett., 1994 ; Schweitzer et al., J. Phys. Chem. A, 2003). These authors show that low E^o* triplet states give higher ¹O₂ yields. Therefore, the subset of triplets that are producing the most singlet oxygen are likely not reacting with syringol.

Figure 1: an arrow connecting BrC to OH would be important as the premise of this study is to show how BrC produces OH.

Figure 3: It could be worth comparing a high pollution day with a low pollution day to indicate the spread of the absorbance of the BrC?

Figures 5,6,7: What do the error bars represent? This information could be included in the figure caption to help the reader understand.

Figure 4: CTC has less data points plotted compared to House data points. But from Table S1 it seems like this CTC data exists. We weren’t sure why some data were omitted? Could the authors clarify?

Figure 9: we wondered how the concept of Arctic haze might be impacting the seasonality obtained in this figure. Could the authors comment and reference Arctic haze literature?

Figure 9:We were also excited to see the similarity of singlet oxygen seasonality compared to what we found in Bogler et al. 2022.

L405-407: Could the authors help explain the uncertainty for ³C^*?

SI: We wondered if the authors could include their FFA and syringol decays to look at the linearity of the data? Our group would also be interested in seeing the 2NB data for comparison. (It can be a multi panel figure to help with clarity.)

SI: The SI is certainly extensive. However, we didn’t find the filter mass data included. We think it would be useful to Figure 2 as a comparison of the mass collected and the ambient mass measured. This filter mass could also be added directly to Figure 2.

Figure S1 would benefit from having a legend on the figure to explain the colour coding.

Figure S4 is likely overfitted in some panels. We might recommend not adding a linear fit. For instance, some outliers are overfitted (like panel f which incorrectly has the highest r2).

Figure S8: we thought this figure was quite valuable for understanding predictions. Would the authors consider adding it to the main text?

Small comment on the writing - it might be worth adding more headings and sub-headings to better help find the information. For example section 4.2 goes on for 3 pages. It would help us in our reading if the info was more broken down to specific results.

Overall, great work on putting all this data together!

Heinlein et al. investigated the production of aqueous photooxidants by light absorbing PM collected in Alaska during the 2022 ALPACA field campaign. The authors reported that BrC photochemistry is significant, with daytime 3C* and 1O2* concentrations in PM predicted to be up to 2×10^-12 M and 3×10^-11 M, respectively, and moderate OH radical concentrations of up to 5×10^-15 M. Using a photochemical model, the authors showed that PM photochemistry cumulatively dominates, generating 76% of daytime secondary sulfate formation largely due to in-particle hydrogen peroxide, which contributes 25-54%. Lastly, the authors reported year-round 3C* production, with the highest rates occurring in late winter when Fairbanks experiences severe pollution and in summer when wildfires generate BrC.

Overall, the results are exciting, this paper is well-written, and the subject matter fits the scope of ACP. I have a few comments that the authors should address:

Line 114: Why was benzoic acid chosen as the OH probe in this study? Why not use benzene or terephthalate as has been previously used by the authors and other studies?

Line 123: How would the collection of a different PM size range affect the results? How much BrC is expected from the missing 0.7 to 2.5 um range in the CTC PM samples? Also, do the authors expect drastically different BrC composition in the PM collected from the “House” site vs. CTC site? I suppose if the BrC composition is very different, this will impact the apparent yields and steady-state concentrations of the photooxidants.

Line 136: By adjusting the PM extracts to pH 1, won’t the authors be solubilizing the transition metals (e.g., Fe which solubilizes at pH ~2) during their extraction process? Won’t this cause an over estimation of the concentrations of water-soluble transition metals in the extracts?

Line 157: What are the purposes of using a water and AM1.0 air mass filter in their photochemistry experimental setup?

Line 175: Why set the experimental temperature to 10 C?

Section 2.6.2. Why do this dilution series only for pH 1? Why not for pH 5?

Figure 5: I have several comments regarding this figure

• It looks like the x axis of panel (a) may have a different scale from those of panels (b) and (c)?
• It looks like the pH may have an effect on 1O2* steady-state concentrations from panel (c) judging from the data points between 1/13 and 1/27? Can the authors comment on that?
• How did pH affect 3C* steady-state concentrations?

Line 372: Which “past studies”?

Line 383: Should state that the Fe referred to here is water-soluble since water-insoluble Fe cannot drive Fenton chemistry.

Figure 6: There should be some discussion on how and why pH affect the quantum yields of the photooxidants.

Figure 7: I suggest adding inserts of magnified views of the traces near the x axis. Currently, it is difficult to see the data trends.

Section 4.2: Why was only pH 1 chosen for the modeling component? From my understanding, Campbell et al. (Sci Adv 2024) showed that there are two clusters of pH during the campaign: “high” (pH 3 to 5.5) and “low” (pH -1 to 1). How will pH affect the modeling results? Also, I assume that the authors needed to assume a ALWC value for modeling study. What ALWC was used? The campaign average of 8 ug/m3 reported by Campbell et al.?

General comment: It is not very clear to me why two sites were used for the study. There was very little discussion on the difference in BrC composition and photooxidant production in PM collected at the “House” site vs. CTC site in the paper. I suggest more discussion on this in the revised manuscript.