Aqueous SOA formation from photosensitized guaiacol oxidation: Comparison between non-phenolic and phenolic methoxybenzaldehydes as photosensitizers in the absence and presence of ammonium nitrate
- 1School of Energy and Environment, City University of Hong Kong, Hong Kong, China
- 2City University of Hong Kong Shenzhen Research Institute, Shenzhen, China
- 3Department of Civil and Environmental Engineering, and Centre for Regional Ocean, Faculty of Science and Technology, University of Macau, Macau, China
- 4Shanghai Academy of Environmental Sciences, Shanghai 200233, China
- 1School of Energy and Environment, City University of Hong Kong, Hong Kong, China
- 2City University of Hong Kong Shenzhen Research Institute, Shenzhen, China
- 3Department of Civil and Environmental Engineering, and Centre for Regional Ocean, Faculty of Science and Technology, University of Macau, Macau, China
- 4Shanghai Academy of Environmental Sciences, Shanghai 200233, China
Abstract. Aromatic carbonyls (e.g., methoxybenzaldehydes), an important class of photosensitizers, are abundant in the atmosphere. This study compared non-phenolic (3,4-dimethoxybenzaldehyde, DMB) and phenolic (vanillin, VL) methoxybenzaldehydes as photosensitizers for aqueous secondary organic aerosol (aqSOA) formation via guaiacol (GUA) oxidation under atmospherically relevant cloud and fog conditions. The effects of ammonium nitrate (AN) on these reactions were also explored. GUA oxidation by triplet excited states of DMB (3DMB*) (GUA+DMB) was ~4 times faster and exhibited greater light absorption than oxidation by 3VL* (GUA+VL). Both GUA+DMB and GUA+VL formed aqSOA composed of oligomers, functionalized monomers, oxygenated ring-opening species, and N-containing products in the presence of AN. The observation of N-heterocycles such as imidazoles indicates the participation of ammonium in the reactions. The majority of generated aqSOA are potential brown carbon (BrC) chromophores. Oligomerization and functionalization dominated in GUA+DMB and GUA+VL, but functionalization appeared to be more important in GUA+VL due to contributions from VL itself. AN did not significantly affect the oxidation kinetics, but it had distinct effects on the product distributions, likely due to differences in the photosensitizing abilities and structural features of DMB and VL. In particular, the more extensive fragmentation in GUA+DMB than in GUA+VL likely generated more N-containing products in GUA+DMB+AN. In GUA+VL+AN, the increased oligomers may be due to VL-derived phenoxy radicals induced by •OH or •NO2 from nitrate photolysis. Furthermore, increased nitrated products observed in the presence of both DMB or VL and AN than in AN alone implies that photosensitized reactions may promote nitration. This work demonstrates how the structural features of photosensitizers affect aqSOA formation via non-carbonyl phenol oxidation. Potential interactions between photosensitization and AN photolysis were also elucidated. These findings facilitate a better understanding of photosensitized aqSOA formation and highlight the importance of ammonium nitrate photolysis in these reactions.
- Preprint
(1301 KB) -
Supplement
(2925 KB) - BibTeX
- EndNote
Beatrix Rosette Go Mabato et al.
Status: final response (author comments only)
-
RC1: 'Comment on acp-2022-595', Anonymous Referee #1, 15 Sep 2022
This study analyzes the characteristic of the DMB and VL as photosensitizers reacting with GUA for aqSOA formation, including kinetic analysis, product distributions and chemical characteristics, as well as optical properties. Meanwhile, the effect of AN for aqSOA formation was analyzed. In general, the paper is well written and conclusions are convincing in terms of rational and rigorous experimental design and analyses. I just have several minor comments on it.
- In terms of abundance of products, for GUA+DMB, the abundance of products in the presence of AN is less than that in the absence of AN, however, for GUA+VL, the results are the opposite. Please elaborate.
- For Fig.4, why choose an absorbance wavelength of 180 min for the study? What is the change of absorbance during the whole reaction process, and is the effect of AN obvious on the change of absorbance?
- Why the entire reaction time of this study was 180 minutes, and did the precursors get reacted completely?
- AC1: 'Reply on RC1', Beatrix Rosette Go Mabato, 22 Nov 2022
-
RC2: 'Comment on acp-2022-595', Anonymous Referee #2, 16 Sep 2022
This paper systematically investigated the physicochemical properties of aqueous SOA formed from the photosensitized guaiacol oxidation by using DMB and VL as photosensitizers in the presence and absence of AN. In general, this paper is well written, readable and logical. Before accepted for publication, some revisions should be made. The specific comments are listed as follows:
1. Why did not show aqSOA yields in this work? In my opinion, these data are very useful for readers to understand the importance of these oxidation processes. When these data were described in the paper, the comparison between these date and other similarly reported results should be made.
2. As mentioned in section 2.1, the samples were collected every 30 min for 180 min for offline analyses. Therefore, authors can provide more information about the changes of signal-weighted distributions and visible light absorption of aqSOA formed under different conditions during the whole reaction processes. In addition, the concentration changes of small organic acids during the whole reaction processes should be also supplemented.
3. Please provide the reason why selected the whole reaction times as 180 min.
4. There are still some language mistakes, please carefully check.
- AC2: 'Reply on RC2', Beatrix Rosette Go Mabato, 22 Nov 2022
-
RC3: 'Reviewer Comment on acp-2022-595', Anonymous Referee #3, 20 Sep 2022
This manuscript describes a comparative study of the photosensitization by phenolic and non-phenolic methoxybenzaldehydes in reactions of guaiacol (another phenolic compound, but without an aldehyde functional group), with and without the presence of ammonium nitrate salts. The experiments were conducted in bulk aqueous phase samples in a solar simulator.
The combination of photosensitizing reactions of methoxybenzaldehyde species with ammonium nitrate photochemistry in a series of experiments is especially interesting. The primary conclusion is that the non-phenolic species DNB is approximately 4 times more effective as a photosensitizer than the phenolic species vanillin, and produces slightly more brown carbon. The manuscript includes a great number of qualitative comparisons, but the authors highlight the most important ones in the abstract and conclusion. It will be of interest to atmospheric scientists studying mechanisms of formation of brown carbon and aqueous secondary organic aerosol.
My first concern is that the authors may have oversimplified the complex task of comparing the photosensitizing abilities of VL and DNB, when VL is reacting away at ~20x the rate of DNB (a factor of 8 x 2.4). The reactivity of VL is so great that it successfully competes with GUA in the reaction with the VL triplet (3VL*), reacting with it 24% of the time over the course of the reaction even though the VL concentration is 10x less than GUA. (I estimated this reaction fraction from the stated 2.4x faster decay rate of VL times the VL / GUA concentration ratio of 0.01mM/0.1mM, resulting in a relative loss rate for VL of 0.24 if GUA loss rate = 1.) If one takes into account 3VL* reactions with both VL and GUA, DNB would be at most only 3 times faster than VL at promoting photosensitization reactions in general. A more nuanced kinetics analysis would thus be helpful for GUA + VL and GUA + VL + AN reactions. Furthermore, it could allow some qualitative statements in the paper, such as those in line 204 and 207, to become quantitative: when integrated over the full course of the reaction, what is the impact of the loss of the reactant VL on the total amount of products generated?
Other comments:
Line 95: How are products counted if they appear in both positive and negative modes of ionization?
Line 335: What could highly oxidized species decompose into, that would not be detected and therefore not contribute to the measured O/C ratio? Is this statement alluding to CO2 production?
Figure 3: at the top right, C12 and C11 products are referred to as functionalized monomers. How is this different from a ring-opened dimer? How exactly do the authors distinguish functionalization from dimerization?
Figure 4: In this graph, does 1 or zero = no change in integrated absorbance? In other words, is it normalized somehow?
Technical corrections
Line 94: “represent” should be “characterize”
Line 328: should this say “likely has a furanone group”? Otherwise, how do the authors know this is the correct structure from the many possibilities?
- AC3: 'Reply on RC3', Beatrix Rosette Go Mabato, 22 Nov 2022
-
RC4: 'Comment on acp-2022-595', Anonymous Referee #4, 27 Sep 2022
Overview
Mabato and co-authors studied the aqueous photochemical reactions of guaiacol (GUA), a methoxyphenol from wood burning, in the presence of vanillin (VL; a phenolic carbonyl), dimethoxybenzaldehyde (DMB; a non-phenolic carbonyl), and/or ammonium nitrate (AN). They examined photochemistry in five different reaction solutions: (1) GUA + AN, (2) GUA + DMB, (3) GUA + DMB + AN, (4) GUA + VL, and (5) GUA + VL + AN. For each system, they give the kinetics of loss, information about the products formed, and some very cursory information about light absorption from the resulting reaction mixture.
There are a few interesting pieces in the manuscript, most notably the suggested interaction of AN photoproducts with triplet photoproducts, which I wish the authors had explored more. But, otherwise, the research seems to largely repeat ideas and experiments that have been reported previously. I don’t see new, interesting questions that are driving the current work. In addition, I see several other important weaknesses, as described below.
Major Comments
1. It is not clear what is novel enough about this work that it deserves to be published in ACP. A number of the systems or parts of the manuscript have been reported previously, both by this group and other groups. For example, Mabato et al. (2022) reported results for GUA + VL, while Smith et al. and Yu et al. reported results for GUA + DMB and for another phenol (SYR) with VL. The addition of ammonium nitrate in the manuscript has no significant impact on kinetics or normalized product amounts, but does lead to incorporation of N into the aqSOA, a point made for VL + AN by Mabato et al. (2022) and for general carbonyl + ammonium systems by several past authors. One result is that there is significant repetition of past work. Some examples: (1) the first two paragraphs on page 8 largely repeat what has been shown in previous work, (2) there’s nothing new in Figure 1, as all of these molar absorptivities have been shown by previous groups, and (3) most of the pieces of Figure 3 have been shown in Mabato et al. (2022) or in past work by the Zhang group.
2. The results aren't quantitative in a way that they could be used to model aqSOA formation from 3VL*. For example, the rate constants for decay of GUA given in Table 1 are probably specific for the experimental conditions used here. The same is true for the quantum yields given in the text - these are almost certainly a function of GUA concentration. It would be much more useful to measure fundamental quantities (e.g., second-order rate constants for 3VL* + GUA) that can be used across a wide range of conditions. Can the current set of data be used to determine some fundamental quantities that are widely applicable? If not, how will people use these data to quantitatively understand these reaction systems?
3. I don’t see the utility of [P], the normalized product abundance. In the big picture, what does [P] indicate about a certain reaction system and what do differences in [P] between reaction systems indicate? The authors present it as an equation without any in-depth discussion of its strengths and weaknesses. Since [GUA]t/[GUA]0 is the inverse of the fraction of initial GUA that is present at time t, Equation 2 for [P] could be simplified as A(P,t)/A(GUA,0). Thus [P] depends on at least three variables: (1) the extent of reaction, since A(P,t) probably rises initially and later falls, (2) the fraction of products that give a signal in the HPLC-Orbitrap (e.g., small organic acids probably do not), and (3) the ionization efficiency of each product in the Orbitrap. These issues need to be described in the manuscript; as part of this, the authors should say something about the IE values for the different classes of products that they see. [P] also depends on what is used to normalize peak areas, e.g., VL in Mabato et al. (2022) and GUA here (making it very difficult to compare across the papers), and probably the initial concentration of the normalizing species. Given all of these variables, what do we learn from the Table 1 data of [P] after 180 min of illumination? If they authors want to use [P] to describe products, they need to state what they think this parameter indicates, give us some experimental evidence that it's useful, and say something about its strengths and weaknesses. As it currently stands, the reported values of [P] seem to have no real utility.
4. In each of the five systems, products were measured after 180 min of illumination. But this approach ignores the fact that the systems have different reactivities and so a fixed time of analysis is looking at different generations of products in the different systems (as shown by Yu et al.). This is important because the relative amounts of products are a function of oxidation time in any given solution. So it is difficult to meaningfully compare across different solutions unless the GUA fraction reacted is very similar. This is further complicated by the much faster decay of VL compared to DMB.
5. The presentation of light absorption data for the reaction products is insufficient. There is one figure (Fig. 4) that sums absorbance values across 350 – 550 nm. This is interesting in that it shows the presence of ammonium nitrate doesn’t affect overall absorbance, but this is too coarse a tool to describe the brown carbon products by itself. It would be helpful to show spectra for each solution at 180 min in the supplement. Also, Fig. 4 should be improved by weighting each absorbance spectrum by the spectral actinic flux to properly describe light absorption. This could also be done by calculating the rate of sunlight absorption by each aqSOA for some standard sunlight condition. This matters because the number of solar photons increases enormously from 350 to 550 nm.
Other Comments
Line 125. Just a note: you don't need to bubble air through a solution to make it air saturated. Shaking the solution with air in the headspace, then opening the container, and repeating this several times is sufficient. The downside to bubbling synthetic air is that you can introduce water-soluble contaminants from the air into the solution.
Line 126. What was the flow rate of air through the solutions during illumination. Is it fast enough to be a significant loss mechanism for volatile compounds (e.g,. NOx, small organics, etc.)?
Line 129. What was the pathlength of the photoreactor? What were the corresponding light screening factors for each solution? Are corrections for light screening required to correct the rate constants?
Line 141. Were the decays of GUA, VL, and DMB always first-order? It seems unlikely given that the reactions proceeded for many half-lives of some of the compounds (e.g., VL). It would be helpful to show both examples of good (first-order) and not so good kinetics in the supplement.
Line 234. These two sentences seem contradictory: GUA+DMB had more compounds with higher O:C, but GUA+VL had a higher average OS(C). How to reconcile this apparent discrepancy?
Line 254. Is this statement based only on the higher amounts of oligomers and functionalized monomers in the GUA+DMB case compared to the GUA+VL case? If so, this is weak evidence and really not a "correlation".
Table 1. (a) Are these rate constants normalized to a specific j(2NB) value? Line 144 indicates that rate constants were normalized by dividing by j(2NB), but this does not appear to have been done to the Table 1 rate constants based on their units. If the authors aren’t going to normalize the rate constants, they should discuss the variation in j(2NB) across their samples and give average j(2NB) values for each reaction condition. (b) In the presence of AN, GUA has a rate constant of 8.1E-3 min-1, which is appreciable. But the addition of AN to the DMB or VL solutions has no apparent impact on the rate constant for GUA loss, with a difference much less than the addition increment expected of 8E-3 min-1. How to explain this discrepancy in the kinetics? Is light screening an issue? (c) Experiment #4 is labeled as a second #3.
Minor Comments
Line 87. The Henry’s law constant listed for DMB (5.4E1 M/atm) is far too low. This is either a typo or a problem in the source reference.
Line 108. “…AN generated more N-containing products…” More than what condition?
Line 129. To better simulate sunlight, I recommend the authors add an airmass filter to their illumination system for future work.
Line 191. The authors posit that the OH + VL rate constant is larger than the OH + DMB rate constant, but it seems unlikely that the difference is large given that they are probably both very fast. Have the authors looked for these rate constants?
Line 377. Should clarify that this is referring to stronger light absorption by the products.
Recommendation
This is a difficult manuscript to rate, as it has a few interesting points but some major issues. However, given that there is not a lot that is novel about the work, I am sorry to recommend that it be rejected.
- AC4: 'Reply on RC4', Beatrix Rosette Go Mabato, 22 Nov 2022
Beatrix Rosette Go Mabato et al.
Beatrix Rosette Go Mabato et al.
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
417 | 132 | 13 | 562 | 51 | 0 | 2 |
- HTML: 417
- PDF: 132
- XML: 13
- Total: 562
- Supplement: 51
- BibTeX: 0
- EndNote: 2
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1