This study explored the sources of aerosol oxidative potential – quantified through the DTT assay – at two sites in Beijing. Daily PM2.5 filters were collected at both sites for a period of one month.  In addition to DTT, water-soluble organics, water-soluble and total metal concentrations, and certain organic markers were measured.  PMF was applied in an attempt to apportion the DTT response to different sources.  Overall, the manuscript topic is certainly relevant for ACP and some of the results are novel and insightful.  However, there are several places where key conclusions are not robustly supported by the data.  There are many places where more nuanced analysis is needed.

Specific Comments:

• There are significant limitations with the present study that need to be discussed, and it is definitely not “comprehensive” as the study claims. The limitations are (1) that measurements were only conducted for a period of one month, and (2) water-insoluble species were excluded from the analysis.  Therefore, the discussion associated with Fig. 2 needs to be qualified.  Similarly, the authors are advised against using descriptors like “exposure-relevant toxicity” and “PM₅ intrinsic toxicity”. Water-insoluble components can contribute to OP (and thus the exposure and toxicity of people in Beijing), yet they were not quantified in this study. The authors should being more accurate/specific with their description of results here and the implications of their findings.
• In several cases, the authors are cautioned against making the well-known mistake of confusing correlation with causation. Section 3.2 concludes that nitroaromatic compounds “could be important contributors to DTT consumption,” on the basis of their correlation with DTTv. To my knowledge, the response of NACs in the DTT assay has not been assessed. Therefore, without this direct knowledge, it might be other components, including those not measured, that correlate with NACs that are driving the correlation. A similar comment applies to discussion surrounding Figure 4, including Line 261-262: “the consumption of DTT from elements depend primarily on its soluble fraction instead of their total content.”  However, it was only the water-soluble fraction that was added to the DTT assay, so it makes sense that DTTv would have much stronger correlations with the water-soluble species.  The authors are referred to the work of C. Sioutas, who has examined the response of soluble and insoluble PM fractions in the DTT assay.
• The PMF results need to be analyzed more critically, and with much more detail. For example, traffic is identified as the most significant contributor to the DTT activity in the south (39.1%). And yet, aside from hopanes, other elements attributed to traffic emissions (Ba, Sr, and Cu) seem to have very low correlation coefficients with DTT? The discussion in Section 3.3 is also confusing: the point of PMF is that it is much more sophisticated than simple linear correlations, yet the discussion here uses correlations with individual species to draw conclusions about the sources contributing to DDTv. Then the discussion moves to the PMF output, but the connection between these is not apparent.
• Many, many method details (blanks, calibration procedures, QA/QC procedures) are missing. The UV-Vis spectrophotometer model is not given. These can be included in the SI, but they are not present at all. Measurement uncertainties were an input into the PMF model (Line 174) yet these were not given for any species, nor the methodology to quantify the uncertainties.
• Also, I acknowledge that the methods the authors have used are widely applied in aerosol studies, however, two potential measurement artifacts need to be acknowledged and discussed. The first relates to potential compositional changes that may occur when the filters are sonicated. Sonication produces hydroxyl radicals and this can change the organic composition (e.g., Miljevic et al., 2014, and references therein).  The second potential artifact relates to metal precipitation during the DTT assay.  This can cause complex responses in the DTT assay that are not straightforward to interpret (Yalamanchili et al., 2023). Again, the authors have applied established methods here, but these potential effects can (and should) still be discussed.
• I don’t follow the explanation in Lines 270-272: why wouldn’t the non-linear response also apply in the north?

Technical Corrections:

• Should all WSOC units be μg-C m^-3 (instead of μg m^-3)?
• Does Fig. S3 and Table S1 together indicate that only ~1-2% of Fe was soluble?
• Figure S1 needs a scale so the distance between the sites can be estimated.
• Line 17-18: sentence needs clarification
• Line 65: “organic” should be “organics”
• Line 116: “foils” should be “foil”
• Line 134: change “were” to “was”
• Line 141: edit sentence for clarity
• Line 207: edit sentence for grammar
• Line 244: suggest changing “are coincide” to “qualitatively agree” or similar
• Line 314: change “wither” to “winter”
• Paragraph beginning on Line 337: is it accurate to qualify these as “regional” differences?

References

Miljevic, B., et al., To Sonicate or Not to Sonicate PM Filters: Reactive Oxygen Species Generation Upon Ultrasonic Irradiation, Aerosol Science and Technology, 48: 1276-1284, 2014.

Yalamanchili, J., et al., Measurement artifacts in the dithiothreitol (DTT) oxidative potential assay caused by interactions between aqueous metals and phosphate buffer, Journal of Hazardous Materials, 465, 131693, 2023.

Filters were collected at two sites in Beijing, one in what is referred to as south (39.61°N) and the other the north site (39.99°N). Based on comparing measurements of water soluble DTTv and PM2.5 mass concentration and various chemical species, this research finds practically all measured parameters were substantially higher in the south vs north. However, DTTv was similar, thus DTTm was much higher in the north; that is the water-soluble components of the particles were concluded to be more toxic at the north site (noted in lines 209-213). 

The authors perform a correlation and source apportionment analysis and find differences.  However, no possible explanation is given for the observed differences in toxicity other than it is likely due to different species.  Another possibility is raised is that the nonlinearity in the DTT assay may be having an effect (noted in the manuscript line 271-273).  This seems to be a viable explanation since it is known that DTTv response decreases with increasing concentrations of species in the extraction vial such that at high metals concentrations the assay becomes much less responsive to differences in metals concentrations. Since the metals concentrations are very high in this study, this could explain the similar DTTv at the two sites and the higher DTTm at the north site with lower PM2.5 mass concentration. Since the difference in DTTm between the sites is a key finding of this paper and is claimed to indicate a more toxic aerosol in the north site, the possibility that it is instead driven by an artifact should be investigated in detail. There are a number of things that could be done. Redo the analysis at a constant aerosol particles mass in the extraction vial at both sites, as suggested by other investigators (Charrier, J. G., A. S. McFall, K. K. T. Vu, J. Baroi, C. Olea, A. Hasson, and C. Anastasio (2016), A bias in the "mass-normalized" DTT response - An effect of non-linear concentration-response curves for copper and manganese, Atm Env, 325-334). Redo the analysis at different particle masses for both sites and see how much that affects DTTm. 

In contrast to this possible limitation with the DTT assay affecting DTTm, the results of Fig 3 showing higher correlations with Abs365 in the north and suggesting more influence from NACs, is a possible cause for the higher DTTm in the north. Maybe this idea could be explored more, eg, although Abs365 is smaller in the north what are the MACs (Abs365/PM2.5 mass)? Maybe a similar analysis could be done for NACs? From a rough calculation based on Table S1, the Abs365/mass (MAC) at the north site is 0.26 vs 0.16 at the south site. For the sum of NACs/mass, the ratio is about 2.3 at the north site and 2 at the south site, both of these suggesting that the DTTm could be higher at the north sites due to these organic species, which could maybe be due to a higher proportion of vehicle emissions.

Overall, this the results of this paper are interesting, but more analysis is needed.

Specific comments.

In Section 2.1 please state: What is the actual distance between sites (I get 42 km). How many samples were collected at each site? State this

In Fig 2, were the protocols for the DTT analysis for the studies shown the same across all these studies?

Estimate the mass of PM2.5 in the extract that was used for the DTT analysis, show a summary comparing the two sets of data, S and North.  (This relates to the comment about possible artifacts related to non-linear response of the assay to metals).

In Figure 2, is this comparison just WS DTT for all data shown. Be clear on what is being compared.  Does the China data in Figure 2 support the findings of a difference in this paper.

PMF analysis. Is n=31 (approximately) sufficient for a robust analysis? Justify.

The correlation analysis is interesting, but a more mechanistic analysis, eg maybe doing more experiments with the filter samples, as noted above, would add strength to the conclusions.  Example, the sentence on line 363-366 in the Conclusions (“The results indicate that in the north trace elements and water-soluble organic compounds, especially BrC…”) is a strong statement simply based on correlations.