the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Jul 2026
Measurement report: Seasonal dynamics and driving factors of aqueous-phase photooxidants in atmospheric particles: Implications for wintertime SOA formation
Abstract. Aqueous-phase oxidation processes significantly promote secondary organic aerosol (SOA) formation, driven primarily by photooxidants including hydroxyl radical (·OH), singlet oxygen (1O2*), and organic triplet excited states (3C*). However, seasonal variations and driving factors of these oxidants in atmospheric aqueous phases remain poorly understood. In this study, we quantified the steady-state concentrations of ·OH, 1O2*, and 3C* in PM2.5 extracts under simulated solar irradiation, and estimated their ranges in ambient aerosol water. The results show that [·OH] exhibited no significant seasonal variation, whereas [1O2*] and [3C*] displayed distinct seasonal variations of winter > autumn > summer. All three oxidants correlated strongly with water-soluble organic compounds (WSOC), especially biomass burning-derived WSOC. By extrapolating the fitted relationships between oxidant concentrations ([·OH], [1O2*], and [3C*]) and extract concentrations to ambient conditions, their ranges in ambient aerosol water were estimated as [·OH] = (1.0–5.4) × 10⁻14 M, [1O2*] = (2.3–61.9) × 10⁻11 M, and [3C*] = (1.6–35.8) × 10⁻12 M. The relative contributions of the three photooxidants to aqueous-phase oxidation of typical organic precursors revealed the dominant role of 3C*-mediated reactions in ambient aerosol water, even at low temperatures. This work thus resolves the winter SOA underestimation in current model studies by demonstrating the critical yet overlooked role of ³C*-mediated aqueous-phase oxidation.
- Preprint
(1541 KB) - Metadata XML
-
Supplement
(511 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2792', Anonymous Referee #1, 17 Aug 2026
-
RC2: 'Comment on egusphere-2026-2792', Anonymous Referee #2, 05 Sep 2026
The authors examine the formation of three oxidants – singlet molecular oxygen (1O2*), triplet excited states (3C*), and hydroxyl radical (OH) – in illuminated extracts of fine particles collected from Beijing over three seasons. They studied a very large number of samples, which is commendable as there aren’t many measurements of these oxidants in the literature. They also examine oxidant concentrations in a dilution series and use these results to estimate oxidant concentrations under highly concentrated aerosol liquid water (ALW) conditions. There are, however, several significant problems with these estimates, as described below.
In general, the core of this manuscript is good, but a number of significant issues need be addressed and/or corrected. These are described below. This manuscript should be reconsidered after these major (and minor) revisions.
Main Manuscript Comments
Line 63. The text lists OH concentrations of 1E-13 to 1E-12 M in aerosol liquid water (ALW), but measurements (such as the two papers cited in this sentence) show values of 1E-15 to 1E-14 M. Models show much higher values (up to 1E-12 M, roughly), but these appear to be incorrect, as described by Arakaki et al. (2013).
Line 89. This sentence references the estimate of [1O2*] in ALW of 1E-10 M by Kaur and Anastasio, but this value was incorrectly high because it failed to account for the plateau in triplet concentrations at high dissolved organic carbon (DOC) concentrations. Ma et al. (2023) revisited this issue and came up with extrapolated concentrations in ALW of 1E-12 to 1E-11 M. The authors should update the singlet oxygen concentration on this line.
Line 110. The authors give a reference for the PM collection, but they should repeat the basic information here, such as the sampler type, flow rate, and type of filters used for PM collection. Also, what were the average PM concentrations in the three collection periods?
Line 114. How were filters "peeled"? What total area of filter was extracted into 10 mL water?
Line 118. What were the factors that made the current extracts more concentrated than values in past work?
Line 116. Does this sentence mean that the same portions of filter were sequentially extracted twice or that two different portions from each filter were extracted?
Line 122 and throughout the paper. It appears that no actinometry was performed to determine the photon fluxes in the illumination solutions. This is a problem for two main reasons. First, without actinometry the authors don't know if the photon flux varies across the different positions in the illumination system. If the flux does vary with position (which is likely), there will be systemic problems interpreting the results. Second, without actinometry the authors do not know the typical photon flux in their experiments, they cannot normalize their oxidant measurements across days, and it is difficult to compare results with other studies.
Unfortunately, since the experiments are complete, the authors cannot know the photon fluxes on the days of their measurements. However, they need to at least make actinometry measurements now, report them, and consider normalizing results to a standard sunlight condition. Also, they need to report the spatial variation of photon flux across the four positions in their illumination system by measuring the rate constant for actinometer loss in each position over the course of several days. Assuming there is significant variation (> 10%) between the “brightest” and “darkest” samples positions, they should correct their results by normalizing oxidant measurements for a given sample to a standard photon flux.
Lines 147-156. Add references for (a) the 4-OHBA yield, (b) the second-order rate constants for OH with BA and BA-, and (c) the pKa of BA. Also, in Eq 4 benzoic acid is denoted as BAH, but elsewhere it's abbreviated as BA.
Line 160. (a) Give a reference for the screening factor equation. (b) Also, alpha(lambda) isn't an absorbance but rather an absorption coefficient.
Section 2 broadly.
(a) There is no mention of how much light the PM extracts absorb, which is a large missing piece: since photochemistry starts with the absorption of a photon, it’s important to characterize light absorption by samples. This should include the rate of sunlight absorption under a standard illumination condition (i.e., day/time/location) for each sample. They should also show the average or median UV/Vis light absorption spectrum for each season in the SI.
(b) There is no mention of dark controls. These controls and their results should be described.
Line 172. The methods used to determination j(NO2- to OH) and j(NO3- to OH), and the results, should be shown and discussed in the SI.
Line 177. Samples were diluted 1:1?
Line 191. Give a reference for this rate constant.
Lines 204-205. Give the triplet probe concentrations in micromolar units to be consistent with the OH and 1O2* descriptions.
Lines 221-222. Give references for these rate constants.
Line 224. Show the determination of the rate constants for direct photolysis of the probes in the SI.
Line 244. Give the reference for these ratios.
Line 252. (a) Either here or in the SI, give the literature values for the second-order rate constants of probes with the four model triplets and the accompanying reference(s). (b) The authors should report their calculated mole fractions of the four model triplets for each sample in the SI.
Line 276. The authors should show the ALWC values, T, and RH, for each sample in the SI.
Lines 289-298 and elsewhere.
(a) The oxidant production rates and steady-state concentrations all depend upon the photon flux of the illumination system, which is not specified. How does the photon flux of the system compare to ambient sunlight? Or, alternatively, how does the actinometer rate constant in the system compare to an ambient value?
(b) As described earlier, oxidant production rates and steady-state concentrations should be normalized to a standard photon flux (or actinometer value). Unfortunately, since the authors didn't perform actinometry on each day, they can't properly normalize each result. But based on post-experiment actinometry they should roughly normalize their results to a standard photon flux.
(c) Oxidant production rates and concentrations will vary with season because of seasonal variations in photon flux. But there is no discussion of this and all results are shown under the lab photon flux condition. While they should all be normalized to some environmental photon flux for the main text, it would be helpful to also show a figure of production rates and steady-state concentrations where each season is normalized to that season's photon flux (e.g., picking a midday value in the middle of each sampling period).
Figure 1.
(a) At 7.5 ug PM/mg H2O, water should be the dominant sink for 1O2*. Because of this, the authors can estimate P(1O2*). This should be plotted on Figure 1b.
(b) To help interpret the figure, it would be useful to include a new Panel (a) showing the rate of light absorption under the lab illumination conditions. Variations in light absorption between samples and seasons probably drives much of the sample and seasonal variability in oxidants.
(c) The authors should calculate and show quantum yields for each oxidant. This will reveal the intrinsic photochemical reactivity of the light-absorbing species, i.e., the efficiency with which they convert absorbed light into oxidant.
Lines 316-318. The authors are comparing dilute solution concentrations of 1O2* and OH to make statements about these oxidants under concentrated ALW conditions. But this isn't valid since the two oxidants have different behaviors as a function of solution concentration. Instead, they should move this idea to after the presentation of their extrapolated ALW oxidant concentrations.
Line 345. Of the four model triplets reported by Kaur, which is/are closest to the mean SYR/MeJA rate constant ratios?
Figure 2.
(a) The authors should include the rate of light absorption (or some other measure of sample absorbance) as an independent variable in this comparison of regression correlation coefficients. It seems likely that light absorption is a major factor determining oxidant production rates and concentrations.
(b) Define MH-WSOC and SH-WSOC in the caption.
Lines 385-392. The authors are making too much of the correlations between the OH rate constant for loss (k'OH) and individual organic species. k'OH is driven by the concentration of WSOC and, to a smaller extent, the chemical reactivity of these organics. I suspect that the negative correlation of k'OH with biogenic organic species is simply a result of lower WSOC in the summer samples, when biogenics are more important and biomass burning organics are less important. Conversely, winter samples - which have more biomass burning and less biogenic organics - are more polluted and thus have higher WSOC and higher k'OH. Does this explain the correlation of k'OH with individual organics? On line 392 the authors suggest biogenic organics might inhibit the loss of OH, but this is extremely unlikely: higher concentrations of organics, whether biogenic or not, will give higher rate constants for OH loss.
Lines 429-432. Again, the authors are reading too much into their correlations. The anti-correlation of 3C* and 1O2* with T and RH are most likely because biomass burning is the major source of BrC and there is more biomass burning in the winter, when T and RH are low.
Line 431. Why would higher T and O3 quench aqueous 3C* and 1O2*? This doesn't make sense.
Line 451. The ALWC is probably very different in the three different seasons. If so, I suggest giving season-average values, along with the corresponding PM/H2O ratios. Certainly, the winter values for ALWC are likely much higher than the other two seasons. Also, you should express the PM/H2O values in the same units throughout the paper, not switching between ug/ug and ug/mg, a distinction that is easily missed by a reader.
Line 456. Only one sample was used for the serial dilution oxidant measurements? Was it the same sample for all three oxidants? Which sample was it? Values of the composition and oxidant measurements for the serial dilutions should be included in the SI.
Figure 4.
(a) The lettering scheme for the panels is odd, i.e., why is the first panel lettered (d)? Adjust this so that the panels are lettered in order. Alternatively, move figures (a) - (c) to the left of figure (d).
(b) The OH results do not include the contribution from mass transport of gas-phase OH, which is the major source under dilute conditions and an important source across most of the PM/H2O range. This needs to be included.
(c) The linear extrapolations (in log-log space) from the dilute extracts in panel (d) ignore the increasingly important roles of DOC as the concentration of solutes increase. Above the range of experimental extract concentrations, DOC becomes the dominant sink for 1O2* and 3C*. By not including this effect, the authors are overestimating the concentrations of these oxidants. Similarly, at high solute concentrations, the production rate of OH probably does not outpace the OH sink from organics, leading to a plateauing of [OH]. Thus the extrapolated concentrations of the three oxidants under ALW conditions are too high. The authors should use their measurements of oxidant concentration as a function of dilution to estimate the second-order rate constants for DOC with 1O2* and 3C*, then use these values to construct a kinetic model of oxidant formation and loss (and, therefore, steady-state concentrations) across the entire range of extract concentration. This approach is described in Ma et al. (2023) and Ma et al. (2024). Note that the Kaur et al. (2019) extrapolation for [1O2*], which is similar to what the authors show in Fig. 4, is incorrectly high because it does not account for the decline in [3C*] under more concentrated conditions.
(d) As described on line 494, the horizontal lines on panel (d) assume that there is no further increase in BrC concentration because of solubility limitations above an arbitrary cut-off of 40 ug PM2.5/1 mg H2O. I agree that BrC solubility might be important in photooxidant formation under ALW conditions, but we don’t know enough to assess this. One can imagine there are factors at play in ambient particles that alter the solubility of organics, including BrC, so that it isn't similar to what is determined in dilute solution. For example, increasing organic concentrations will help solubilize organics (at least into organic aerosol), while higher salt concentrations in ALW will decrease (salt out) some organics. The bottom line is that any estimate of the solubility limit of unknown ambient organics is just a guess. Adding the horizontal dashed lines to Figure 4d gives an incorrect sense that this is a reasonably well known estimate of BrC behavior.
Section 4. The conclusions are too long and somewhat repetitive.
Lines 560-561. These references are incorrect: they don't report anything about the temperature dependence of triplet reactions. Please update with the correct references.
Figure 5.
(a) Add the PM/H2O ratios and oxidant concentrations for the two different conditions shown here to the caption.
(b) What pH value do the authors use for these calculations? The rate constants for phenols with triplet DMB (which seems to be their dominant triplet surrogate) are pH sensitive.
(c) Hydroquinone and vanillin both undergo direct photodegradation. These contributions should be added.
(d) Some nomenclature issues: (i) Methyl Jasmonate is missing the "J". (ii) Foliate acetate is an uncommon name for what is typically called cis-3-hexenyl acetate. I suggest using the latter. (iii) 2-methyl-3-buton-2-ol should instead be "2-methyl-3-butene-2-ol"?
(e) The figure is somewhat misleading because it shows only the aqueous pathways, even though gas-phase reactions will dominate for most of the compounds under both conditions. The authors need to modify the figure to show the relative importance of gas and aqueous pathways for the two liquid water content conditions. This could be done, for example, by adding two panels that show the percent of loss due to gas- and aqueous-phase reactions, one for the cloud/fog conditions and one for aerosol water, considering at least OH and O3 in the gas phase. I believe these will show that aqueous reactions are minor except perhaps for some of the larger phenols under cloud/fog conditions.
(e) The ALW results should be updated to reflect the new estimates of oxidant concentrations based on better extrapolations described above.
Supplemental Information Comments
General comments.
(a) The SI is generally sparse and not well explained (e.g., captions are minimal).
(b) There is a large amount of information missing from the SI, including sample collection information (dates, times, average ambient PM2.5 mass concentrations, etc.), the composition of every extract (ions, metals, pH, WSOC, organic tracers, etc.), and light absorption characteristics (e.g., rate of light absorption under lab condition and screening factor) of every extract.
(c) There are no figures showing oxidant measurements. It would be good to include some sample kinetic plots of probe decay during illumination, including dark controls.
(d) The authors should show plots of some of the key regressions that are reported in Figure 2, with points color-coded for the three seasons.
(e) As described earlier in the review, the authors should also add information and results from (i) their dilution series and (ii) their estimates of the triplet pool composition based on their triplet probe reactivity results.
Line 36. (a) What were the dimensions of the glass cylinders used for illumination? (b) What volume of extract was used in each cylinder? (c) What was the pathlength for the standard solution volume? (d) Were four containers illuminated simultaneously, as shown in Fig. S1? If so, on a given day what was the illumination plan, e.g., was it two different samples, each with a duplicate? Or one sample and two replicates and one sample with no replicates? Was there a consistent pattern?
Figure S1. (a) Some sense of the size of the system would be helpful. For example, what are the dimensions of the illumination chamber? (b) It's not clear where the dark chamber sits.
Figure S2. (a) How was WSOC measured? (b) It would be good to add panels that show some of the other important sample extract information, such as light absorption.
Table S3. The rate constant for vanillin with OH is certainly too low and should be at least 10 times higher. It's possible that the value reported by Li et al. (2014) is incorrect.
Citation: https://doi.org/10.5194/egusphere-2026-2792-RC2
Data sets
Seasonal characteristics, influencing factors, and steady-state concentration estimation of photooxidants generated from water-soluble organic compounds in illuminated particle extracts Min Cai https://doi.org/10.5281/zenodo.20805514
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 202 | 80 | 33 | 315 | 38 | 21 | 22 |
- HTML: 202
- PDF: 80
- XML: 33
- Total: 315
- Supplement: 38
- BibTeX: 21
- EndNote: 22
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This study investigates the aqueous-phase photooxidants steady-state concentrations in seasonal PM2.5 samples and explores their underlying controlling factors, and certain novelty can be identified. Nevertheless, several critical issues remain to be addressed as listed below:
(1) The authors have measured photooxidant concentrations across a wide gradient of PM2.5 extract concentrations, covering both concentrated aerosol liquid water and dilute cloud/fog water matrices with high water content. When discussing the role of relative humidity (RH), gas-liquid partitioning of ·OH governed by Henry’s law must be incorporated into the analysis. This partitioning flux exerts a non-negligible contribution to aqueous ·OH pools in high-water-content cloud/fog systems with low WSOC loading, which has not been addressed in the current discussion.
(2) Severe multicollinearity exists between temperature, ozone concentration and seasonal emission regimes (notably reduced biomass burning inputs in summertime). Simple correlation analysis cannot independently validate direct chemical impacts of temperature and O3 on photooxidant budgets. It should be clarified that elevated temperature alone cannot drive brown carbon (BrC) photobleaching; significant photobleaching only occurs when high temperature is coupled with sufficient solar irradiance. Moreover, a mechanistic breakdown of temperature effects on triplet excited states should be fully articulated, including two dimensions: