the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Bridging temporal resolution and molecular specificity with a compact GC-PTR for the measurement of volatile organic compounds
Abstract. Volatile organic compounds (VOCs) play important roles in atmospheric chemistry and air quality, but measurements often require a tradeoff between temporal resolution, molecular specificity, and instrument size and complexity. Online chemical-ionization mass spectrometry provides sensitive, rapid measurements but can be affected by isomeric contributions, fragmentation, and other ion-chemistry interferences, whereas gas chromatography (GC) provides molecular separation and specificity at lower temporal resolution. Here, we present a compact, low-power VOC measurement system that combines in-situ gas chromatography and thermal desorption (TD) preconcentration with a newly developed ultracompact proton-transfer-reaction time-of-flight mass spectrometer (PTR-TOF-MS) equipped with a novel vacuum-ultraviolet (VUV) ion source. The complete system occupies a 50 x 65 x 55 cm space, weighs < 70 kg, and consumes < 600 W of power during typical operation. The instrument combines high-time resolution, direct PTR measurements with periodic chromatographic measurements for regular molecular speciation. The system was operated continuously for 30 days in Thun, Switzerland with stable calibrant response (average variation of 3.5 %) during the ambient measurement period, with demonstrated direct-PTR limits of detection (LODs) of 13 – 136 ppt (1-min) and GC-PTR LODs of 0.1 – 4.0 ppt (1 L sample) across 11 calibrants. Together, these results demonstrate a complementary approach to instrument miniaturization in which reduced detector performance can be offset by upstream analytical separation and preconcentration. In this configuration, the direct PTR measurement provides high-time resolution observations, while periodic chromatographic measurements provide molecular specificity, sensitivity enhancement, and information on analyte-specific ion chemistry and interferences. By reducing the size, power requirements, and operational complexity, this system provides a pathway toward broader deployment of chemically specific VOC measurements at long-term monitoring sites, on mobile platforms, and in locations where infrastructure or reliable power is limited, to investigate emission sources, atmospheric processing, and the formation of secondary pollutants.
Competing interests: The authors declare the following competing financial interest(s): MSC, FLH, UR, VY, BML, DW, MC, and JJ are current or former employees of either Aerodyne Research or Tofwerk AG which manufacture and supply both the TD-GC systems and the compact chemical ionization mass spectrometer described in this work.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(1340 KB) - Metadata XML
-
Supplement
(1227 KB) - BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-6002', Anonymous Referee #1, 02 Nov 2026
-
RC2: 'Comment on egusphere-2026-6002', Anonymous Referee #2, 07 Nov 2026
This paper presents a compact and low-power VOC measurement system that combines gas chromatography with thermal desorption with a proton-transfer-reaction time-of-flight mass spectrometer (PTR-TOF-MS) using a vacuum-ultraviolet ion source. The system provides both direct-PTR and GC-PTR analysis methodologies, with the direct-PTR providing 1-min average limits of detection between 13 – 136 ppt and the GC-PTR 1 L sample limits of detection of 0.1-4 ppt for the 11 calibrants in their standard. The authors describe different measurement timing strategies, with both single-channel and dual-channel GC options, as well as parallel and series sampling options for the GC-PTR samples. The authors use data from a 30-day measurement period in Thun, Switzerland, at their facility, but do not make the data publicly available due to proprietary reasons. A calibration system is described, and subsets of ambient observations and using the from the direct- and GC-PTR data are
The paper is somewhat lacking in several ways. By using data from only a single location from only this instrument, the authors don’t present a clear picture of the applicability of the instrument for different chemical regimes or source types. There are no associated or concurrent measurements presented to explain differences between the direct-PTR and GC-PTR methodologies, either of non-VOC trace gases, including ozone, or other VOC observations. Further, the paper does not address the limitations of PTR-MS detection for analysis of a significant subset of VOCs, including saturated hydrocarbons and halocarbons. It also doesn’t explore many of the C1-C4 OVOCs, with a rather limited subset of OVOCs in the calibration mixture addressed (acetaldehyde, acetone, and MEK).
Finally, the purpose of the paper is not clear. To demonstrate the full value of the instrument, the full HR TOF spectrum should be utilized to extract all available information from both the direct-PTR and the GC-PTR methodologies, in particular for times when there are discrepancies between the two data streams. These details are just glossed over with “further analysis would be warranted” (line 392) or that if the entire dataset is considered instead of looking at only a 2-day period, that the agreement improves (lines 418-420). An attempt, at least, should be made to resolve or explore these differences to provide some guidance to potential future users on how to manage measurement discrepancies.
Specific Comments
Lines 46-48 – I agree that many online systems have a lower time resolution, but there are examples of online GCs with a much faster time resolution than an hour – e.g., TOGA instruments. Perhaps this could be mentioned as a caveat.
Lines 154-155 – It has been demonstrated that the inclusion of PFA tubing in the analyte stream can increase outgassing/bleed contamination, create molecular absorbance issues, and/or lead to higher backgrounds or ghost peaks. Please discuss how you have ensured that the use of PFAs or Teflon fittings in the analyte sampling line is not negatively impacting your analyte sensitivity, accuracy, or the CCl4 normalization. In light of some of the measurement discrepancies, this might be a serious issue.
Regarding dual-channel characterization (~ Lines 165-168), were the samples collected in parallel tested to ensure consistent sensitivity? Presumably, collecting samples in parallel would mean that one sample is kept in a sorbent trap for a longer duration of time, and any differences would point to potential interferences from chemical instability in the trap. (Based on Figure 4, which shows very different focus trap cooling times.)
Line 180 – What qualifies as a high level of reactant gases (e.g., ozone), and what would the threshold be? Ambient levels of ozone don’t typically vary by orders of magnitude, so would typical summertime urban levels require an oxidant trap and other levels not? How is this distinction quantified?
Line 189 – What is the material of the frits?
Line 223 – It would be beneficial to see an example of the 2-channel sampling method for multiple samples (series) to round out the examples shown in Figure 4 and Figure S3. Further it would be helpful to show Figure 3 with different flow paths highlighted to help visualize the different flow paths during the different preconcentration steps.
Section 2.3 Calibrations – I would like to see far more information about the calibrations. From what I can tell, the sensitivity calculations are based on a 10-day period in which 2 back-to-back calibrations were performed on the TD-GC system every ~ 26.5 h, so approximately 20 calibration runs, so there should be statistics on the sensitivities. Further, it isn’t clear why only a 10-day period was used for the sensitivity assessment, and/or why it would matter if there were power outages or maintenance between the sensitivities calculated from the calibrations. Also regarding the calibrations, please discuss possible humidity or reagent impacts using UHP N2 for the dilution rather than scrubbed ambient air or even zero air including oxygen, and how you ensure that there aren’t differences in the PTR results introduced due to differing humidity regimes.
Line 244 – I’m perplexed by why only the unit mass resolution data were used? Why not take full advantage of the high-resolution data? Especially for the species that have significant measurement discrepancies (isoprene, benzene, acetaldehyde, MEK etc.) it seems that the HR data or even full unit mass spectra would provide valuable information to understand these differences.
Table 1 – Some of the GC-PTR limits of detection seem very precise in comparison to the others. Should these really have 3 significant figures? Also, shouldn’t the normalized standard deviation be associated specifically with the GC-PTR? Were the blanks considered during the LOD determination? It seems that there might be an LOD issue with the GC-PTR benzene, based on Figure S7. Either that or there is a background correction issue for the direct-PTR benzene. Please address.
Figure 5 – While I’m not a huge fan of this figure, specifically for plots C-L (I would like to see these plots on a much more expanded y-axis resolution) I suppose there is some value in seeing all this data together. I do think that Figure S7 needs to be in the main manuscript, however, because it presents very clearly the limitations of the direct-PTR method and the value of the combined direct-PTR and GC-PTR. Also, regarding the C9 aromatics – what about ethyltoluenes?
Regarding Figure S7 (and Line 393 in the main manuscript) – the levels of acrylonitrile seem relatively high considering that the levels of acetonitrile are so low (and not necessarily associated with the elevated acetonitrile per Figure 5 – please comment on that, too). Normal Northern Hemisphere background levels of acetonitrile are typically closer to 100-200 ppt depending on the time of year and the last ~ 6 months of hemispheric wildfires, but 50 ppt or even lower seems very low. Is there a possible zero or background correction issue? Is there any other data that would support such low CH3CN values?
Regarding acetaldehyde (lines 385-392) - The diurnal discrepancy points to a potential ozone interference, especially considering the presence of non-Sulfinert types of surfaces that are in contact with the samples. It would be worthwhile to look at the discrepancy between the direct-PTR and GC-PTR acetaldehyde and see if there is any relationship with ozone (I would hope there are ozone observations in the vicinity even if they are not associated directly with the study?)
Figure S5 – This plot should have error bars demonstrating the known uncertainties. It would also be helpful to label the outlier, which appears to be acetaldehyde, or for that matter, all the species. Can the authors comment on why acetaldehyde doesn’t fall on the line? Lastly, these are somewhat arbitrary plots and relationships based on the species that are included in the calibration standard, and that should be considered and discussed.
Table S1 – Why only use the period between March 3 and March 5?
Technical comments:
Throughout both the manuscript and SI – “exemplary” is used several times when “example” is what is meant.
Line 11 – chemical ionization mass spectrometry is not usually hyphenated (see also line 58)
Lines 14, 62, 78, 84, and 490 – be consistent with hyphenating in situ or not.
Lines 18, 439 and in Figure S4’s caption x2 – there should be a comma after Switzerland.
Line 51 – for consistency, use the same hyphenation on proton-transfer-reaction mass spectrometry throughout – e.g., lines 15, 51, 55, 98, 531, etc.
Line 55 – usually, rate constants “k” are italicized.
Line 92 – remove the space between PTR and -based.
Line 102 – 30 kg is fine, no need for “kgs”.
Line 103 – after m/z 100, either change the comma to a semicolon or start a new sentence.
Line 144 – define sccm
Line 148 – add a comma after “e.g.”
Line 215 – define SLPM
Line 232 – N2 should have a subscripted 2.
Line 236 – “… consecutive calibrations with preceding and the succeeding system zeros.” I think there’s an unneeded “the” in there?
Line 261 – the “plus” should be superscripted in CCl3+
Line 268 – Section 3a doesn’t exist?
Line 274 – the style guide recommends DD Month YYYY for dates.
Line 306 – Fig. S5 uses “cm3”, not “mL”
Line 347 – subset shouldn’t have a hyphen.
Line 357 – ethylbenzene is one word (x2) (also in Fig. 5 caption)
Lines 359-360 – trimethylbenzene is also one word (x3), as is propylbenzene (x2) (also in Fig. 5 caption)
Figure 5 caption – Time series is two words, and subset is one word.
Figure S4 – “location in Thun, Switzerland,” would be more appropriate
Figure S6 – please define CT and MH
Citation: https://doi.org/10.5194/egusphere-2026-6002-RC2 -
RC3: 'Comment on egusphere-2026-6002', Anonymous Referee #3, 12 Nov 2026
Claflin et al., describe the design and field deployment of a compact instrument combining thermal desorption gas chromatography with an ultracompact PTR-ToF-MS equipped with a novel VUV ion source. The system combines higher time resolution direct-PTR measurements with periodic chromatographic separation to provide improved molecular specificity and lower detection limits. The authors demonstrate the instrument’s performance during a 30-day ambient measurement campaign in Thun, Switzerland, including comparisons between direct-PTR and GC-PTR observations, although the quantitative comparison focuses on a limited subset of the measurements.
Overall, I think the development of this instrument is extremely valuable. The reduced size and power requirements are promising for expanding the accessibility of VOC measurements, and the ambient observations effectively demonstrate the importance of chromatography for resolving isomer contributions and interferences in PTR measurements. The manuscript is generally well organized, and I support its publication. However, I believe there are some major items that need to be addressed in order to better characterize and communicate several aspects of the instrument’s performance. In particular, I would like to see a more comprehensive comparison between the direct-PTR and GC-PTR observations across the measurement campaign, as well as further discussion of the instrument’s operational stability and sampling methodology. Addressing the following points would strengthen the manuscript and better establish the capabilities and limits of the system.
Major comments:
1) Section 2.4 and Section 3.2, lines 244-245 and 362-424: Characterization of mass spectral interferences and agreement between direct- and GC-PTR measurements
The comparison between direct-PTR and GC-PTR observations is a central component of this manuscript and their disagreement warrants further investigation. In section 2.4 lines 244-245, the authors state that the PTR mass spectral data were integrated at unit mass resolution without post-collection mass spectral fitting. Given that the instrument has a reported resolving power of ~1200m/dm at m/z 100, why was the available mass resolution not utilized in an attempt to further quantify potential interferences in this specific study? Although the authors acknowledge that certain isobaric interferences can be important for specific species and may not be separated with the ToF’s resolution, such as those affecting isoprene at m/z 69, it’s not clear whether they attempted to do so. I would strongly recommend that, whenever possible, a high-resolution analysis be used to evaluate or constrain the magnitude of interferences in this specific dataset, even if complete resolution is not always possible.
Along the same lines, the authors frequently attribute disagreement between the two methods to isomeric contributions or fragmentation, often supported by previous literature. While these explanations are reasonable, there is relatively little quantitative evaluation of these contributions using the measurements presented here. Could these be investigated using the specific mass spectral and chromatographic data from this study?
Additionally, the interpretation of these discrepancies appears to assume that the GC-PTR measurements are quantitatively accurate. Could differences in GC trapping efficiency or other sampling-related losses also contribute to the disagreement? I would recommend further discussion of how these possibilities were assessed and/or determined to be negligible.
This comparison between direct- and GC-PTR is quantitatively demonstrated in Figure S7, but given the weight of this discussion in the manuscript I recommend moving this figure to the main text. I also suggest that the correlation statistics for the 30-day measurement period be reported rather than the two-day subset presented in Table S1. Further, given the authors’ statement that agreement between the two modes varies with time and environment, additional consideration of the temporal variability in these relationships would also strengthen the discussion.
2) Sections 2.3 and Sections 3.1 lines 279-314: Calibration procedures, reported sensitivities
The authors report an average variation in calibration response of 3.5%, which is highlighted in the abstract as evidence of stable operation during the 30-day campaign. However, Section 3.1 and Table 1 indicate that the reported standard deviations were calculated using only a 10-day period that excluded external disturbances. Why were the calibration statistics limited to this period when approx. daily calibrations were conducted throughout the campaign? I recommend presenting the calibration response over the full 30-day deployment and discussing how representative the selected 10-day period is of the instrument’s overall performance.
Section 2.3 describes PTR multipoint calibrations conducted at the beginning and end of the measurement period, but the results of these calibrations are not presented or clearly discussed. Were the sensitivities obtained from the initial and final calibrations consistent with the daily calibrations and how, if at all, were they incorporated into the reported sensitivities and uncertainties?
3) Section 3.2, lines 398-423: Temporal variability of interferences and proposed corrections to direct PTR measurements
The authors suggest that periodic GC measurements could provide a means of correcting interferences in high-time-resolution direct-PTR observations. While this is a potentially valuable application, I question whether the 90-minute GC duty cycle provides sufficient temporal information to implement such corrections reliably. The contributions of interfering species to a particular m/z may vary substantially with changes in emissions and trace gas composition between GC measurements. How would these changes be accounted for when applying a correction to the direct sampling data? I recommend that the authors either demonstrate a potential correction approach using the available observations or more carefully qualify this as a future application that has not been demonstrated. The ability to resolve and quantify individual VOCs at a lower sampling frequency is already a significant advantage of the combined instrument, particularly for long-term ground-based observations, and does not necessarily require demonstrating that all direct-PTR measurements can be corrected.
4) Section 2.1, Lines 180-185: Characterization of the oxidant trap
The authors describe the use of an oxidant trap when sampling air containing consistently high concentrations of reactive gases, like ozone. However, it is unclear how these conditions were defined and whether the oxidant trap was used throughout the measurement period. I also think an example demonstrating the reduction of sampling artifacts with and without the trap may be useful for characterizing its performance in this configuration. The material used in this system should also be checked, sodium sulfite should be written as Na2SO3 rather than NaSO3.
5) Sections 2.1 lines 110-117 and 3.1 lines 324-339: Description of the ion sources
The introduction of the novel VUV ion source is an important aspect of this work, but its description is relatively brief. I recommend expanding its discussion in section 2.1. Additionally, the comparison between VUV and plasma ionization in 3.1 and Figure S6 would benefit from a more detailed description of both ionization methods in the Methods section to provide readers with a clearer basis for interpreting the observed similarities and differences in product-ion distribution.
Minor comments:
- Introduction, lines 42-50: The discussion of the tradeoff between chromatographic specificity and temporal resolution should briefly acknowledge fast GC-MS systems, such as NCAR’s TOGA. While these systems can achieve substantially faster measurement cycles than the instrument presented here, they generally have a larger footprint and greater operational requirements, which would provide useful context for the advantage of this compact system.
- Section 2.1, lines 100-104: “kgs” can be “kg”, “exemplary” should be “example”
- Section 2.1 line 144: Define sccm and ensure that flow rates expressed in sccm and cm3/min are used appropriately and consistently throughout the manuscript.
- Section 2.1, line 153: Missing “the” in the sentence beginning “it should be noted that [the] preconcentration system…”
- Section 2.3, line 236: The phrase “with preceding and the succeeding system zeros” could be revised to “preceded and followed by system zeros”
- Section 2.4, lines 260-268: Please explicitly state what the reported normalized counts are normalized to. Additionally, remove the duplicated “that” in “It should be noted that that the sensitivities…” and replace the reference to “Section 3a” with the appropriate section number.
- Section 3.1, Table 1, and lines 288-302: Please verify the reported sensitivity ranges against Table 1. The maximum direct-PTR sensitivity is reported as 1769 ncps ppb-1 in the text, whereas Table 1 lists 1784 ncps ppb-1 for acrylonitrile. Similarly, the GC-PTR sensitivity range reported in the text differs from the values in Table 1.
- Section 3.1, line 305: “sensitives” should be “sensitivities”
- Section 3.1, Table 1: Change “calibrate response” to “calibrant response” and include m/z in the quantification-ion column heading.
- Section 3.1, lines 306; Figure S5: Please use consistent volume units in the discussion of the sensitivity comparison. Additionally, both panels of Figure S5 report identical R2 Could the authors clarify whether uncertainties in the sampling flow rates propagate into this comparison of volume normalized sensitivities?
- Section 3.2, lines 347-360 and Figure 5: Please correct “sub-set” to “subset” and GC resolved” to “GC-resolved”. Compound names should also be corrected, ethylbenzene, trimethylbenzene, and propylbenzene are usually one word. In line 358, change “Similar for the C9 aromatics” to “Similarly, for the C9 aromatics”
- Section 3.2, Figure 5: The caption again uses “exemplary” instead of “example”.
- Section 3.2, Line 395: “GC resolved” should be “GC-resolved”
- Section 3.2, line 403: The authors report that the direct-PTR measurement of m/z 69 is, on average, 28.9 times higher than GC-resolved isoprene. Including the standard deviation or interquartile range would help characterize the variability in this comparison.
- Section 3.2, line 446: “percent variabilities are as follows” should be “…were as follows”
- Section 3.2, Lines 487: “each isomers contribution” should be “each isomer’s contribution”
- Conclusion and Abstract: The discussion of “sensitivity enhancement” should distinguish improvements resulting from GC preconcentration from improvements in detector sensitivity, consider rewording.
- Supporting information, Figures S7 and Table S1: Both captions reference Figure 4 when they likely should refer to Figure 5
- Table S1: Again, m/z should be specified in column heading
- Figures S8 and S9: Replace “exemplary” with “example”
- Throughout the manuscript: Ensure consistency in terminology and formatting, particularly “in situ” versus “in-situ”, “high time resolution” versus “high-time-resolution”, “time series” versus “timeseries”, and the use of chemical subscripts and superscripts.
Citation: https://doi.org/10.5194/egusphere-2026-6002-RC3
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 394 | 187 | 36 | 617 | 97 | 34 | 37 |
- HTML: 394
- PDF: 187
- XML: 36
- Total: 617
- Supplement: 97
- BibTeX: 34
- EndNote: 37
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Claflin et al. describe a compact VOC measurement system that couples a dual-channel thermal desorption gas chromatograph with a new ultracompact PTR-TOF-MS fitted with a VUV photoelectron ion source. The integrated system alternates between direct PTR sampling, acquired at 1 Hz and averaged to 1 min, and GC-PTR analysis on a 90 min cycle. The authors report direct-mode LODs of 13 to 136 ppt (1 min) and GC-mode LODs of 0.07 to 4.01 ppt (1 L sample) for an 11-component standard, show that GC and direct sensitivities agree once sample flow and collection time are accounted for, and use ambient CCl4 resolved by the GC to normalise data streams for detector drift. A 30-day deployment from the side of a building in Thun, Switzerland, in late winter and early spring 2025 is used to compare direct ions with GC-resolved compounds. Agreement is good for acetonitrile, acetone, toluene, the C8 and C9 aromatic sums and summed monoterpenes, but direct mass 79 overestimates benzene by a factor of two, direct mass 69 averages 28.9 times the resolved isoprene, and masses 45 and 73 carry substantial signal not attributed to acetaldehyde or MEK. The paper closes by speciating eleven monoterpenes and discussing their diurnal and wind-direction behaviour.
Major Comments
The hardware and the GC-PTR coupling are useful and fall clearly within the scope of AMT. A compact, low-power instrument that delivers both fast and speciated VOC data would be of real value to the community. In its current form, however, the manuscript reads more as a demonstration that the system works than as a full method characterisation. It establishes that the system runs and that GC and direct data agree for well-behaved ions, but it addresses interferences only for a few selected masses and not in the quantitative, full-spectrum way its own dataset would allow, does not yet demonstrate detection capability in the environments it claims to serve, only partly characterises the ion source that is its main hardware novelty, and does not make its data available. None of these points is fatal, and the dataset can support a strong paper, but I think each deserves attention before the conclusions can be fully defended. My comments follow, then specific comments by section, then technical points.
It would also be useful to see the following. (a) An uncertainty budget and limits of quantification for both modes, not only LODs. (b) A limitations paragraph on platform suitability, covering pinhole-controlled sample flow (L122-124) under changing inlet pressure, humidity, oxidant levels and low-temperature operation, as well as consumables and ancillary hardware, namely the helium carrier, UHP nitrogen, calibration cylinders and the two supplemental inlet pumps. It would be helpful to state whether these are included in the quoted footprint, mass and powe. (c) Tests of sample volumes above 1 L with breakthrough checks for the lightest species, since larger volumes are the obvious route to lower GC LODs in remote air.
Specific Comments (by section)
Methods
Results and discussion
Technical and Editorial