Overview

Noblet et al. studied the primary and secondary emissions of one gasoline vehicle and one diesel vehicle. They applied high-resolution mass spectrometry and three multivariate statistical analysis techniques to the dataset, consisting of filter samples from the vehicles’ tailpipe emissions. They found some potential markers for distinguishing emissions from diesel and gasoline vehicles for both primary and secondary organic aerosols. In addition, they have measured several variables continuously during the driving cycles driven.

Overall, the manuscript is well written and provides a detailed description of the research performed. However, as I’m not an expert in filter sampling, in my review I focused on online analysis and the results obtained from the statistical analysis. I feel that analysis and results have been described well and the methods used are well-suited. Overall, the manuscript fits well into the scope of the Aerosol research journal.

I don’t have any major concerns related to the manuscript. I have some minor comments and technical corrections regarding the manuscript that I think should be assessed before publication.

Specific comments

L299: Was the number of experiments constant for each cycle, i.e. two times for WLTC and four times for MW and Urban? That could be mentioned either in the figure caption or in the Figure itself, to help the reader.

L303: “… notable increase in POA emissions …” Was the increase indeed in POA emission mass, or is SOA formation included as well? I.e. should it be just “notable increase in emissions”?

Conclusions section: Generalizability of the results. As the results are based on two vehicles, one might think that the variability in vehicles, fuel, and lubricants might affect the distinctiveness of factors in larger dataset. I would appreciate it if the authors could discuss the generalizability of the results in the paper a little bit more than what they have already done in the Conclusions section. Besides the things authors mentioned, my main concerns are related to differences in e.g. lubricant oils and fuels.

Figures S10-S12: Especially the bottom-left subplot might be problematic for colorblind people. Could the line styles (e.g. solid, dashed, dotted) also be different for the lines in the same subplot?

Technical corrections

L168: EFOM ->EF_OM

L216: Both EIS and IIS are introduced second time, these introductions are unnecessary. Probably you’ve just forgotten to delete these after the earlier introduction of terms has been added to subsection 2.4.

L245: What is the meaning of X or Y in R²X? In the text, a letter connected to R² is X or Y and for Q²  it is constantly Y. In the supplement, the markings are R²X and Q²X.

L301: SOA is also introduced already in the introduction (L44).

L322: “reproducib ility” (extra space in the text)

Table S9: Rightmost column in QToF section of the Table is dropped down by ½ row, probably because the text of the cell is aligned to the middle. Now it is not completely clear that for which rows the values in the rightmost column are referring.

The article titled "Discrimination of Euro 5 gasoline vs. Diesel light-duty engine primary and secondary particle emissions using multivariate statistical analysis of high-resolution mass spectrometry (HRMS) fingerprint" investigated the chemical composition of primary and secondary organic aerosol emissions from gasoline and diesel motor vehicles and attempted to identify unique chemical fingerprints that could help isolate the ambient contributions of these sources. The study focuses on a highly relevant and contemporary topic. Differentiating between the ambient contributions of gasoline vs diesel vehicles is challenging, yet need of the hour for many regions of the world where these sources dominate air pollution problems. The work employed a broad range of analytical instrumentation to capture the emissions composition in detail, which supports the objectives of this study. 

Still, in my view the presented analyses somehow does not make sufficient use of the rich molecular information obtained from this diverse set of instrumentation. I consider this aspect a major shortcoming of this work in its present form. Most of the main text graphs are largely statistical in nature, mandate referring the SI to understand fully and are more suitable for the SI in general. It is difficult for a reader to make use of the main text graphs as standalone visuals to inform their own work, capture the gest of the study or assess it against other work. Yet, given the significance of the research topic and detailed measurements, I think the study can be considered for publication after this and other major concerns detailed below are resolved. 

- I find it quite curious that no mass spectra are presented in this study even though an ACSM and QTOF are employed. It would be good to see how the POA and SOA mass spectra looked for different test conditions pre and post PAM oxidation. With comprehensive measurements performed, I think there is value in also showing how similar were these emissions at the very molecular level. This can inform future source apportionment studies working with soft ionization-based measurements (e.g. EESI-TOF).  

- The marker species shown as chromatogram peaks in figures S23-31 are not very useful, though the snapshots of the mass spectra are. The chromatogram peak shapes and retention times can vary based on instrument settings and column type used for separation. These figures should be revised to make the mass spectral peaks more prominent. 

- In figures S10-S12, it is very difficult to assess how different variables correlate. Scatter plots alongside different timeseries would be helpful. 

- Writing can be improved at several instances in the manuscript and better proofread. 

The research objectives should be made sharper. For example, in line 250, the authors state that marker identification was not the primary aim of this work. However, line 85 contradicts this statement in goal-setting where revealing the differentiating markers for POA and SOA is noted as the objective of this work. The word "fingerprint" is literally in the title of this manuscript. 

Similarly, I had to sift through the manuscript text continuously to link the discussion of results with the order in which the experiments were conducted. I think having a few timeseries (e.g. ACSM organics and a few others) systematically presenting the different phases (test conditions) of the experiments in the main text would have helped. 

Line 387: What does "less marked" mean? Does it mean there are less differentiating markers? 

Another example of a confusing sentence is lines 325 - 327: I am not sure what the authors mean by "(trans-)formation state of the OA" that distinguishes between primary and secondary emissions. 

The use of acronyms/identifiers e.g. LC NEG SOA G-1 is very inconvenient, and forces the reader to oscillate between the main text and the SI. 

Line 390: "The high NOx levels may have also triggered distinct atmospheric reactions, leading to the formation of nitrogen-containing

compounds." I reckon the authors are referring to the formation of organo-nitrates but they should consider the branching ratios before suggesting this to be important and that the ratios vary by the tertiary, secondary or primary nature of carbon in precursor molecules. To a large extent, high NOx conditions direct reactions toward specific pathways by converting peroxy- to alkoxy- radicals resulting in dominant production of oxygenated species. If a speculation is mandatory here, I recommend reinforcing it more effectively. 

- The panels are not labeled in figures 2 and 5. The axes are not labeled for some panels in figure 5. 

- Figure 4: The hierarchical structure is not discernable. I wonder how useful is this figure in the main text.  

- Lines 424 - 426: I am not sure how the authors declared some of these points (e.g. red circle on the left) to be an outlier/erroneous. More supporting evidence should be provided in this regard. An erroneous status cannot be accorded to a datapoint simply by the virtue of its positioning on the VK plot. There are multiple points on the graph where POA and SOA nearly overlap and these are treated as valid data. 

Minor comments: 

- The word "Discrimination" in the title is not accurate. Discrimination represents preference or bias toward one over the other. I think what the authors are looking for is along the lines of differentiation or distinction. I suggest revising the title. 

- Table S3 does not include offline instrumentation even though the table caption says "all". 

- The word "Diesel" should have a small "d", i.e. diesel, everywhere in the text. 

- Line 43: consider replacing "photochemical" with "oxidation". 

- Line 142: closing parenthesis is missing. 

- Line 302: The ranges overlap. Thus the sentence needs to be revised. Stating the ratio between averages of the two ranges makes more sense with ranges (xy - yx vs. xx - yy) in brackets. 

- Line 322: remove the space from the word "reproducibility". 

- Figure 3: should the legend have filled circles with a black outline?