This is an interesting article and provides some rare insights into secondary aerosols during the covid lockdown. However, authors need to do a better job to provide more robust statistical backing to their dataset. Also, some of the conclusions made seem farfetched without strong evidence and rely solely on the previous literature. A major revision is suggested, and comments are provided in the attached document.

The paper entitled “Measurement Report: Enhanced contribution of photooxidation to dicarboxylic acids in urban aerosols during the COVID-19 lockdown in Jinan, East China” by Meng et al. is generally well written and falls within the scope of ACP. However, there are may small issues that overall sum up to the need of major revisions for this manuscript.

Major scientific concerns:

Analysis of delta13C:

• If I understand the procedure correctly then the yield during derivatization ranges from 80% to 85%. Have the authors considered the possibility that the derivatization reaction itself may introduce an isotope fractionation and that there could be a Rayleigh fractionation effect. At yield of 80% to 85%, the isotopic composition of the pooled product is not yet identical to the isotopic composition of the reactant before the start of the reaction. Depending on the prescence and magnitude of the fractionation there could be a significant effect .Have the authors investigated the possibility of a fractionation for this reaction (if yes can they report the fractionation constant for the individual reactant-product pairs reported in this study) and factored the Rayleigh effect into their mass balance calculations in section 2.2.2. These crucial details are completely omitted.
• Normally, when isotope ratios are measured with an accuracy of 0.2 permil then it is necessary that some type of standard is measured alongside the sample in the same mass spectrometer. Can the authors please elaborate on the standard they used (which is hopefully traceable to VPDB )and provide additional details. Did they spike the sample with a standard?
• The GC part of their GC-isotope ratio MS would have some type of injector, and a column temperature program as well as carrier gas flow rates. The authors need to provide these details.

Figure 5: This figure is far too confusing. The authors need to settle on one set of organic compounds/compound ratios and influencing factors for which they want to compare before and after lockdown. They need to keep the arrangement of columns and rows identical both for the panels. Parameters in columns, parameters in rows and sequence in both should be same in the panel that shows before lockdown and the panel that shows after lockdown. Having some unique parameters that are only there in the “before” or the “after” and having parameters that are common but swapped around between the columns and rows makes the visual comparison very difficult for the reader.

Figure 8 panel f: I am not sure that fitting two different lines for day and night here is warranted. This could be the Simpson paradox in action. All day samples except one outlier fall onto the same line as the night samples just within a narrower range of that line.

Statistical analysis: There are no details on the MLR analysis and random forest machine learning analysis in the methods section. Line 163 states “The results showed that in Jinan city only 18.2% of the enhanced O3 concentration was resulted from the meteorological variations, and the other 81.8% was ascribed to the reductions of anthropogenic emissions (Fig. S1)”. Firstly, the supplement does not appear to contain what it is supposed to according to the main text. Fig. S1 contains only the Back trajectory cluster analysis and the PSCF of the trajectories, Figure S2 pertains to the MLR/Random forest results. Secondly Figure. S2, which does appear to show the results of the poorly described Random forest? or MLR? analysis shows the highest importance for temperature and solar radiation. So, I have several problems with the sentence above and the figure:

• Figure numbering in the supplement and main manuscript inconsistent. Not only for this case but overall.
• Figure S2 directly contradicts the statement in the text. Temperature and solar radiation appear to jointly explain more than 60% of the ozone variability. When one throws in RH and WS > 80% appears to be explained by meteorology and <20% by emission changes. Hence it appears this sentence, the abstract and conclusions need to be revised
• There is only one histogram which shows either the results of the MLR analysis or the results of the random forest analysis or some average in the supplement. I can’t imagine both these methods would 100% agree on the relative importance of different factors. Both should be shown separately and contrasted if both techniques were used.
• The implementation of the MLR and random forest analysis is not described in the methods section.

The points raised above have implications for the abstract and also for the conclusions line 440 “However, the O3 concentration increased by 1.3 times synchronously during the LCD, which was largely launched by the reduction of anthropogenic emissions whose contribution reached up to 81.8%.”

PMF analysis: The details of the PMF analysis are completely missing in the method section. Which species were weak and which ones were strong. Why exactly did the authors decide to keep NH4+ out of their PMF model?  Why did they keep K+ (a species that should be a decent biomass burning tracer) out of their model? Same question for why did they keep about half their measured acids out of the model? What was the rational for keeping O3 and LWC in the model instead of using them as independent tracers to verify factor profiles. How did the authors decide on 5 factors as the right number of factors. The PMF mathematically solves the equation “explain all my data under the assumption that there are X number of sources impacting my site” The choice of X as 4,5,6, …n is the largest assumption that modifies the PMF outcome. Second largest is inclusion and exclusion of species and third the choice of weak and strong species and the percent uncertainty of each individual species. Unequal uncertainties can heavily drive the PMF outcome. So, all these details of PMF implementation need to be provided. Particularity, why 5 why not 4 or 6 factors needs a justification. Also, what about the error analysis for the PMF? Was bootstrapping done.

Statistical data analysis in section 3.2 & 3.3.

There are a lot of statements that the concentration of some measured species increased/decreased from before the LCD to during the LCD or from day to night. Most of the parameters have substantial ambient variability as can be seen in Figure 3a & b. Maybe the differences will become more visibly clear if the authors chose to show a box and whisker plot instead of a bar graph with the standard deviation. However, even if it looks less as if the concentration before and during the LCD overlaps within the variability, the authors would still need to use an appropriate statistical test to assess whether the differences between the two periods are significant for each of these cases or not. If the variables follow a normal distribution ANOVA may be appropriate. If they do not follow a normal distribution then a more robust test e.g. Kruskal-Wallis test may be required.

Examples of statements without significance level of the difference include:

Line 187: “upward trend from 437 ± 117 µg m-3 before the LCD to 486 ± 144 µg m-3 during the LCD.”

Line 188 “concentrations of diacids and oxoacids during the LCD increased by 1.1 and 2.1 times”

Line 191 “The daytime concentration of diacids before the LCD was 17% lower than that at night”

Line 193 “C2 concentration increased from 181 ± 47.5 m-3 before the LCD to 239 ± 108 µg m-3 during the LCD”

Line 197 “Both ratios of C2/diacids and C2/TDOCs during the LCD were higher than those before the LCD”

Line 201 “daytime concentration of C2 and the ratios of C2/TDOCs and C2/diacids were lower than those at night before the LCD but an opposite trends were found during the LCD”

Line 209 “Both C2/C4 (8.4 ± 3.4) and C3/C4 (1.6 ± 0.4) ratios during the LCD  were higher than those (3.9 ± 1.5, 0.3 ± 0.1) before the LCD (Fig. 3b),”

Line 214 “It is noteworthy that C9 concentration (12.0 4.0) before the LCD was 2.0 times higher 215 than that (5.9 ± 4.8) during the LCD (Table 2),” apart from significance test the plusminus is missing in the first bracket

Line 220 “its concentration (8.8 ± 11.0 µg m-3 220 ) and relative abundance (0.03 ± 0.01) during the LCD were lower than those (11.0 6.1 µg m-3 , 0.02 ± 0.01) before the LCD”

Line 257 “Both ratios of C2/levoglucosan (1.7 ± 0.6) and C2/K+ (0.2 ± 0.02) at night before the LCD exhibited larger values than those (1.3 ± 0.5, 0.16 ± 0.02) in the day, which was mainly ascribed to the accelerated aqueous formation of C2 at night. “

Minor scientific comments:

Line 102: it appears that the online air quality monitoring data for PM2.5, PM10, CO, SO2, NO2, is from an air quality monitoring station associated with some network. Since the referenced weblink is in Chinese I could not follow it. Could you please briefly elaborate in this paper on

• the location of the monitoring station used and how it relates to your PM offline sampling location (collocated with your PM sampling equipment or somewhere nearby?).
• Inlet height of the air quality monitoring station
• Measurement equipment used to measure those parameters and Q/A Q/C protocol followed by the primary data generator

Table 1: Hope the authors have not taken the simple average and standard deviation of the pH. The high standard deviation suggests they may have. The pH scale is logarithmic and not a ratio variable. Values need to be converted back to the {H+] before averaging for the result to be mathematically correct.

Line 230 “> 0.5, Fig. S2).” Wrong figure number

Line 243 Please look into the following sentence “However, the increase in the ratios of C2/diacids and C2/TDOCs at night indicates that the lowered nighttime PBL height was not the case, which could be supported by the insignificant diurnal differences of primary pollutant markers such as Na+ , Ca2+ , and Mg2+ (t test, p > 0.01) between the daytime and 245 nighttime.”  You can’t state that the PBL height was not more shallow than daytime PBL based on this evidence. You can, however, make a case that the change in PBL height was not the only cause of the change in the observed concentration. Also, before you get into these details a statistical test is required to establish that the night time concentration is significantly higher than the daytime concentration. Since the text may get messy with too many numbers, you may have to put the appropriate tables in the supplement.

Line 269 (Fig. S2) wrong figure number

Line 293 “Therefore, pHis exhibited pronounced negative relationships with C2 and its precursors such as Gly and mGly (R2 ≥ 0.45, Fig. 3a)” which figure is this referring to. Figure 2 does have a correlation analysis but not with pHis included. Fig. 3a has a histogram of concentrations

Minor language related comments:

Line 30: “photochemical oxidation lunched by the higher O3 concentration” please revise lunched is not the right word. It’s an ancient past tense of the verb to eat. “triggered by higher O3 mixing ratios” or “promoted by higher O3 mixing ratios” may be a better wording.

Line 109: “The water extracts were concentrated to near dryness and then reacted with 14% BF3/n-butanol at 100°C for 1 hour to convert butyl esters or dibutoxy acetals” This sentence is not clear and grammatically incomplete. Please clearly state which compound(s) you are converted to which compound(s).

Line 128: “Levoglucosan in the field blank samples is 4% less than the ambient samples.” Please check the language here. Do you really mean to say that your field blank is so high that the levoglucosan in the blank is only 4% less than the levoglucosan in your ambient samples. If that is what you meant to say then consider this comment under major scientific concerns.

Line 159: “A rent study” the authors probably mean to say “A recent study”