the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 May 2026
Development of an offline Aerosol Mass Spectrometry method for organic aerosol characterization in a globally distributed Surface Particulate Matter Network
Abstract. Global measurements of organic aerosol (OA) concentrations and chemical composition remain limited and unevenly distributed. While monitoring networks, including the globally distributed Surface Particulate Matter Network (SPARTAN), provide an established framework for measurements, their current methodologies do not fully support comprehensive OA characterization. Aerosol Mass Spectrometry (AMS) is widely used for real-time OA composition measurements, but its cost, complexity, and logistical requirements limit long-term, multi-site online deployment, particularly at the global scale. Here we develop and evaluate an offline AMS methodology to characterize OA in particulate matter collected on Teflon filters routinely used by monitoring networks. Using a commercial ultrasonic nebulizer coupled with a syringe pump, this offline method is highly reproducible, requires small extract volumes (2 mL), offers low detection limits (1.7 µg OA and 0.43 µg sulfate per filter), and achieves higher nebulization efficiency than previous methods. We evaluate this offline AMS method by co-located online AMS observations. We find that oxygenated OA is effectively recovered (64 ± 28 %) while the recovery is lower for hydrocarbon-like OA due to its limited water solubility. This approach offers new capability for SPARTAN and is readily adaptable to other monitoring networks. Its application across networks will broaden the spatiotemporal coverage of AMS-based OA measurements and improve methodological and instrumental consistency to support ongoing efforts to build a long-term, globally consistent OA dataset.
- Preprint
(1289 KB) - Metadata XML
-
Supplement
(6127 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-2663', Anonymous Referee #1, 22 Jun 2026
-
RC2: 'Comment on egusphere-2026-2663', Anonymous Referee #2, 26 Jun 2026
Ren et al. present a filter-based offline AMS framework to characterize OA chemically and use those fingerprints for source apportionment. Such analyses are becoming more popular, thus this study is timely. Especially, extending current analytical frameworks relying on Quartz Fiber filters to Teflon filters is valuable - as networks typically only use one or the other.After addressing the following points, I recommend the study for publication.
Overall, I have two concerns regarding the presented results:
1. I agree with the authors that future studies will not have collocated online AMS/ACSM data available, especially given the scope to apply the method to a global scale. I am afraid that the presented source apportionment data flow in the main body of the results would almost never happen like the one presented here: 1. fully constraining all factor profiles with an online AMS PMF is unrealistic, 2. PMF would likely be used instead of another NNMF implementation. I appreciate the sensitivity tests that the authors present, but would wish for including more details on those results at least in the SI (e.g. factor time series comparisons with e.g. BClf and secondary inorganic aerosol as well as recoveries obtained for each test). In addition, it is crucial to also perform an additional test that follows best practices of constrained source apportionment analyses (e.g. Canonaco, 2021, https://amt.copernicus.org/articles/14/923/2021/), i.e. only to constrain primary OA compnenents with literature spectra - in your case HOA (not OOA). In my personal opinion, these latter results would likely be the best fit to be presented in the main body of source apportionment comparisons.2. The authors rely on SO4 for internal standard-based quantification. However, SO4 forms readily refractory salts. Thus, other studies use NO3 as internal standard (isotopically labelled or not), despite the potential evaporative losses of NO3. This should be explored here too, since the approach is intended to be used globally, thus dusty regions too. Further validation of OA, based on the same fraction (i.e. soluble fraction) with a reference method (e.g. TOC analyzer) would be highly recommended to further strengthen the quantification power of the approach.
Detailed comments:
Line 28: I assume the detection limits refer to a specific filter portion extracted? If yes, please specify this surface. Are the DLs accounting for incomplete solubility?Lines 65-70: How is OC on Teflon filters measured? I thought this is not possible with a Sunset OCEC Analyzer. Are also Quartz filters available? If available the measurements should also be be compared to this OC.
Line 114+121: This statement raises the expectation of a substantially superior performance compared to previous studies and needs to be substantiated later in the paper (e.g. Cui et al., https://pubs.acs.org/doi/10.1021/acsestair.4c00051#_i28, Fig. S1). The authors advertise the method as a fast efficient approach. How many samples can in the new approach be measured per day (including time on extractions etc.)?
Line 181: Commonly, samples are directly measured quickly measured after extraction to avoid storage artifacts. Given that the extracts are frozen, a comparison of a selection of samples with and without freezing is needed.
Section 3.1: What is the AMS m/z measurement range?
Line 361: Why is the laboratory blank subtracted before quantifying, i.e. before normalizing to the internal standard on a run-by-run basis?
line 384: it would be best to also report detection limits in ug/ml extract.
Line 392: Was this exercise performed based on teflon filters collected in the same way as for SPARTAN?
Line 400: Given that the AMS was equipped with a PM1 lense, was PM1 sampled on the filters in the end? As this point is a bit unclear to me, a clarification would be helpful.
Fig 4: Given that OA is quite oxygenated, an overall good agreement can be expected for the extracts. With the global scope of the method, how well does the method perform for locations with substantially higher contributions of hydrocarbons?
Line 480: Some further validation and discussion of the PMF solutions is needed in the SI to evaluate it (e.g. comparison of HOA vs BClf or NOx and OOA vs secondary inorganic aerosols).
Line 482: In addition, this study and the cited ones assume that OA factors have the same AMS mass spectrum in online and offline measurements. This should be stated as an assumption explicitly.
Citation: https://doi.org/10.5194/egusphere-2026-2663-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 157 | 50 | 11 | 218 | 26 | 10 | 12 |
- HTML: 157
- PDF: 50
- XML: 11
- Total: 218
- Supplement: 26
- BibTeX: 10
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
General comments:
This paper presents results from a novel offline AMS methodology, expanding on existing techniques, making offline AMS analyses more feasible on a larger scale. The study offers a thorough characterization of the ultrasonic atomization system employed and an impressive comparison to real-time AMS analysis of ambient PM. A notable advantage of the offline AMS methodology proposed here is the ability to more easily fold into existing PM sampling networks. This makes it a particularly attractive work to labs and organizations looking to improve our understanding of PM chemistry on a global scale. The manuscript is clear and well written. I have largely only minor questions/needs for clarification. Once these concerns are resolved, I recommend publication in AMT.
Specific comments:
Line 408-409: Were the blanks filter blanks, or was filtered air sampled by the real-time AMS (or both)?
Figure 3: Does the OOA fraction represent the fraction of total NR-PM mass? E.g. fraction of OOA = OOA/total NR-PM? Or is this a fraction of just the organic portion of the sample? This is not clear from the main text or figure caption.
Figure S4/organic formula assignments: From the main text, it seems like the organic ion family assignments are based on the HR-AMS/PIKA ion assignments although this is not fully clear. Was the FTIR analysis used here as well? The differentiation between saturated and unsaturated hydrocarbons (ions) is not terribly convincing based on AMS data alone given the additional fragmentation induced by the vaporizer. To the author’s credit, however, the data provided does make chemical sense with that differentiation in mind.
Line 473/PMF analysis: I would like to see a more thorough explanation of the PMF analysis, specifically why the two factor solution was chosen. From the reviewer’s perspective, there is an overreliance on the chemical interpretability factor (from mass spectral features alone) for this solution, which is not often the limiting factor in a PMF analysis. The Q/Qexp vs P plot is atypical, as is the choice of factor solution with the highest Q/Qexp. Neither of those are exclusionary as-is, but certainly they warrant a more thorough discussion as to why this was deemed the appropriate solution. For example, a time series analysis for tracer ions for OOA/HOA factor profiles including any correlations with external events (e.g. traffic patterns) would help to support this PMF solution and factor assignments.
Figure 7: The generally strong agreement between the reconstructed and online OA mass concentration is impressive. For the extreme outlier samples (e.g. with ratios closer to 2 or 0.5), I would be interested to see if there are any outstanding chemical features of those samples that might offer some rationale for the determined high/low reconstructed mass.
Lines 567-569: The purpose of equating IC-measured sulfate with offline AMS-measured sulfate is not clear and is rather confusing from a reader’s perspective. They two measurements of sulfate are certainly “intrinsically linked” as they are coming from the same source, but the two instruments are not equivalent. They measure sulfate in completely different ways and have distinct strengths and limitations (some of which are mentioned in the main text). This is true even with online AMS measurements of sulfate. Showing the correlation of IC-measured and offline AMS-measured sulfate would be critical here, but that comparison is not shown. There are no chemical or instrumental reasons to assume those two measurement schemes would provide identical results. There is plenty of published work showing that online and offline AMS should correlate well with IC measurements, but none that suggest they are equivalent in any way.
As an example of the confusion introduced here, in Figure S8a, the y-axis is labeled “IC (offline AMS) sulfate (μg m-3)”. Are those measurements from the IC analysis or from the offline AMS analysis? The units imply it is offline AMS-derived measurements, although one could arrive at those units from a typical IC analysis and the quantification method outlined in section 3.2. The comparison of online AMS-sulfate to both offline AMS-sulfate and IC-sulfate is of interest, and it’s not clear what is being offered here.
Beyond that, it is not clear what is gained by equating those two measurements as done in Figure S8. Either this equivalency needs to be further explained in the main text (likely with the addition of the IC-measured and offline-AMS measured sulfate comparison if possible) or the language used should be reworked to avoid this confusion altogether.