Interlaboratory comparison exercise for micro-aerosol size measurement by cascade impactor

Dougniaux, Grégoire; Bodiot, Cécile; Dhieux Lestaevel, Bernadette; Nuboer, Amandine; Wahl, Robin; Kort, Amel

doi:https://doi.org/10.5194/ar-2024-30

Preprints

https://doi.org/10.5194/ar-2024-30

Preprints

11 Nov 2024

| 11 Nov 2024

Status: this preprint was under review for the journal AR but the revision was not accepted.

Interlaboratory comparison exercise for micro-aerosol size measurement by cascade impactor

Grégoire Dougniaux, Cécile Bodiot, Bernadette Dhieux Lestaevel, Amandine Nuboer, Robin Wahl, and Amel Kort

Abstract. This study presents an interlaboratory comparison (ILC) exercise focused on measuring micro-aerosol size distributions using cascade impactors. The aerodynamic particle size distribution (APSD) is a critical parameter for understanding aerosol behaviour, particularly for health-related applications. The ILC conducted at the Institut de Radioprotection et de Sûreté Nucléaire (IRSN) aims to assess the performances of participating instruments measuring aerodynamic diameter, cascade impactors and an Aerodynamic Particle Sizer (APS) for real time monitoring. The experiments were performed in a custom test bench able to generate aerosols in a size range from 0.2 to 4 µm within a controlled environment. Performance evaluations of the participating instruments considering five distinct aerosol size distributions were assessed, and two methods – Henry's method and lognormal adjustment – were used to calculate the mass median aerodynamic diameter (MMAD) and the geometric standard deviation (σ_g). Statistical analysis using ζ-score and Z'-score ensured the reliability of the results across participating instruments.

The findings demonstrates that most instruments performed within acceptable limits, though variations observed in some cases, particularly for smaller particle sizes. This work highlights the feasibility of standardized ILCs for APSD measurement and offers a framework for improving accuracy and consistency in aerosol size distribution assessments.

Received: 21 Oct 2024 – Discussion started: 11 Nov 2024

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 preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Grégoire Dougniaux, Cécile Bodiot, Bernadette Dhieux Lestaevel, Amandine Nuboer, Robin Wahl, and Amel Kort

Status: closed

RC1:
'Comment on ar-2024-30', Anonymous Referee #1, 21 Nov 2024
The purpose of this article is to present the results obtained during an exercise, which the authors describe as an interlaboratory comparison, aimed at comparing the results produced by cascade impactors used to determine the mass median aerodynamic diameter (MMAD) of aerosols denoting different size.
While the article's style is pleasant, it's very difficult to identify its real purpose.
Thus, it would appear that the aim is not to compare the analytical capabilities of a number of different laboratories (in this case, only two laboratories from the same entity), but rather to evaluate the performance of 3 instruments using the same technology (atmospheric or low-pressure impactors) operated by a quite limited number of participants. While the real ambition of this article is to demonstrate that cascaded impactors (in this case from two IRSN laboratories) are capable of determining the mass median aerodynamic diameter of an aerosol, this does not represent a major scientific contribution.
Beyond this lack of clarity in the presentation of the article's objective, the analytical methodology presented raises a large number of questions, which you can find listed below, leading me not to recommend such a study for publication in the journal “Aerosol Research”.

Specific comments
C1. Lines 13, 59 and 139: the authors mention a range between 0.2 and 4 µm, but Table 1 only covers a range between 0.3 and 4 µm

Introduction
C2. Line 33: the authors state that the cascade impactor is considered a “gold standard”. On what factual basis (ASTM/ISO standards, OECD guidelines or at least scientific articles) is this statement backed up? What's more, if the cascade impactor is considered a “gold standard”, what's the purpose of this article, which aims to demonstrate its performance?

C3. Lines 43 to 47: the objective of the article is not clearly explained, and the rest of the article adds to the confusion by mentioning participants instead of instruments.

Experimental setup
The experimental configuration presented raises several questions:
C4. Why was quiet air sampling chosen?

C5. The authors mention Grishpun's work on sampling probes, but this work proposes an approach for assessing the aspiration efficiency of probes. What types of probes were considered in this study? Have the results been corrected for aspiration efficiency, which could lead to a change in the expected particle size distribution?

C6. Please specify whether each instrument was attached to a different probe or to a single probe:
If several probes were used, how were they chosen according to the sampling rates of the different instruments?

If a single probe was used, how were differences in sampling rates taken into account?

C7. Has homogeneity, in terms of particle size distribution, been assessed within the test chamber? If so, how, and were any disparities noted (which may justify the use of one or more sampling probes, see my last question)?

C8. Was the relative humidity in the measuring chamber measured during the tests?

C9. Did the authors make sure that the particles produced by nebulization were perfectly solid particles and not residual droplets?

C10. KCl has been used as a source of particles by adding sodium fluorescein: what is the nature of the particles generated in the test chamber? Is it a heterogeneous mixture of KCl and sodium fluorescein particles, or KCl particles “tagged” by sodium fluorescein?

C11. Have electron microscope images been taken, together with elemental analyses, to determine the shape and composition of the particles produced (see my last question)?

C12. Table 2 - ILC participants: the title and content of this table are confusing. Indeed, we might have expected to see different laboratories (and therefore different participants) operating similar instruments. Here, the participants are actually instruments operated by two laboratories belonging to the same entity.

Analysis method
C13. Lines 82 et 85: please explain how the uncertainties associated with Henry's methods and lognormal smoothing are obtained.

C14. Please provide details of the methodology and software used for Henry's method and lognormal adjustment.

C15. Beyond the methods for exploiting the data produced using cascade impactors (or the APS), the uncertainties inherent in the instruments themselves are not discussed or presented. How are the uncertainties inherent in the cut-off diameters of the impactor stages, and in quantifying the mass deposited on each stage, taken into account in the uncertainty budget?

C16. Lines 88-89: the link between fluorescein concentrations and KCl masses on each stage is not clearly explained.

C17. Has weighing the mass of each stage, which is the most conventional approach for cascade impactors, been considered (if possible)?

C18. If the masses available on each stage could not be quantified by weighing, why not consider an ICP-MS analysis to directly obtain the mass of K constituent element of the particles generated?

C19. The term “reference value” suggests an additional objective of the article, namely to propose a consensus MMAD value for the experimental conditions produced on this test bench. Is it really necessary to use this term for an average value derived from measurements performed by only 3 different instruments?

C20. Doesn't the choice of two indicators seem useful? Isn't just one enough (e.g. the Z' score or even the Z score)? If not, please explain why both indicators are needed and useful?

C21. The article by Amarouche (2015), cited by the authors, specifies a different range for the normalized deviation E with satisfactory suitability for a value less than or equal to 1, in which case why are the intervals identical for the two indicators (lines 103-105)

C22. Table 3: Are 3 or 4 significant digits after the decimal point really relevant? This point ties in with my questions about the assessment of measurement uncertainties (inherent in the instruments) and MMAD determination uncertainties (according to Henry's method or lognormal adjustment). The standard uncertainties presented in this table appear very low (5 nm for case E)

C23. Discussion of the results is almost non-existent, and is limited to lines 130 to 135. As it stands, this does not justify a scientific publication.

C24. Lines 134-135: “to underestimate the experimental uncertainties” is not clear to me. What do the authors mean?

Conclusion: C25. as the article is not sufficiently clear on its objectives, it is difficult to identify the real contribution of this article through the conclusion.

Technical corrections
C26. Lines 32 and 33: I don't understand the use of “...”.
Citation: https://doi.org/10.5194/ar-2024-30-RC1
- AC1: 'Reply on RC1', Grégoire Dougniaux, 03 Feb 2025
  
  Response to Reviewer #1.
  We would like to thank Reviewer #1 for his attentive review. Despite his conclusion, he has read the whole document carefully and questioned everything. We have improved our paper according to all his comments/suggestions/questions.
  
  C1. We firstly mentioned a range 0.2 - 4 µm. This have been restrained to 0.3 - 4 µm.
  Introduction
  C2. Gravimetric impactors are arguably the most widely utilised instruments for measuring particle size distribution, and their operating principle is well-documented. Gold standard is a bit exaggerated.
  C3.The aim of this article is to demonstrate the participating instrument performance regarding different particle size-distributions using fluorometric analysis for routine tests of HEPA filters in nuclear facilities.
  For these experiments we proceed as an ILC. Thus, even if few instruments were taken from two laboratories, we identified each couple instrument/lab as a participant. Table 3 shows all the participants of this ILC, with their instruments and characteristics.
  C4. The instruments are connected to sampling probes, with thin edges and a diameter adapted to their nominal air flow. The probes are oriented 90° against the ETNA flux. At nominal ETNA air flux, the air velocity is 0.31 ± 0.02 m/s at the level of the probes. Thus, this sampling is in the calm air condition.
  C5. We add an additional experiment to assess the validity of the homogeneity of the sampling probes, with four APS. There is no size-distribution alteration from 0.4 µm to 10 µm on these instruments. The results are not corrected according to Grinshpun equations.
  C6. Each instrument has been individually connected to an adapted probe. There diameters and sampling velocity are now given in the article.
  C7. The homogeneity has been assessed with a supplementary test with four APS. The results are given in the article.
  C8. The relative humidity in the test bench was not carefully followed. However, the control instrument gives values under 40% RH.
  C9. We do not check the composition of the aerosol, if it his perfectly dry or not. However, it has no influence on the sampling efficiency, thus all instrument sample the same aerosol. Under 40 % of hygrometry, we expect dry particles.
  C10. The fluorescein is use as a tracer, thus the KCl particles are tagged with. With there respective quantities, the fluorescein represents 0 % of the particles mass, except for test E, for which it represents 13 % of the particles mass.
  C11. We do not proceed to electron microscope images for this study. However, it would have been a nice addition to this article.
  C12. Table 2 - ILC participants: the title and content of this table has been clarified in the text.
  Analysis method
  C13. We add a supplementary material to detail the Henry and fit analysis, with the uncertainty propagation.
  C14. We provided details of the methodology and software used for Henry's method and lognormal adjustment in a supplementary material.
  C15. The cut-off uncertainties are taken into account in the global uncertainty budget. From the Dekati calibration certificate, the stokes number are calculated for all stages, then used in the actual operation condition (pressure and temperature). At all calculation steps the uncertainties are propagated according to the GUM. Finally, the uncertainties are taken into account, with the measured mass uncertainties from fluorometric analysis, for the Henry method and log-normal fit to obtain the median and geometric standard deviation.
  C16. The text has been modified to explain clearly the link. To obtain the aerosol mass on each impactor filter, thus the APSD, a fluorometric analysis is performed. “For each impactor, all the filters are taken and prepared for mass measurement. The filters are then placed in new petri dishes and identified on the lid (test number, floor number, etc.). An ammonia solution (NH₃ in H₂O) was made. This solution is prepared by diluting 100 mL of ammonia (28%) in 10 L of ultrapure water. The pH of this solution is greater than 9. The filters are placed in plastic beakers containing 10 mL of the ammonia solution for at least 1 hour. This allows the fluorescein deposited on the filter to pass 100% into solution. An aliquot of the solution, now containing all the fluorescein in the filter, is placed in the measuring cell of a calibrated fluorometer. The measurement range is 5.10^-11 to 1.10^-6 g/ml. The intensity of the fluorescence signal measured gives the concentration of fluorescein in the solution. Because the concentration of KCl and fluorescein is determined in the initial solution (Table 2), the KCl mass is proportional to the fluorescein concentration. This process allows to establish the aerosol mass sampled on each filter”.
  C17. The direct weighing is indeed the most conventional method. However, the expected aerosol mass is below the quantification limit of a balance, thus a fluorometric analysis is far more sensible and far more precise than direct mass weighing. The mass of the deposited particle is roughly of 10^-10 g, and the balance capacity is 10^-5 g.
  C18. An ICP-MS analysis would be indeed a reference measurement, particularly useful to seek specific atoms as K. However, We do not access to this kind of equipment, and a fluorometric analysis, is way more simpler even for the sample preparation, cost less and is precise enough and is used routinely for the HEPA filters tests in Nuclear facilities.
  C19. The term “reference value” has been changed to consensual value.
  C20. The standard NF ISO 13528 lists heigh indicators for ILC analysis, with each case use, each specificity.
  We choose the Z’-score, which use a consensual value as reference and the uncertainty of each participant. It indicates the degree of compatibility of each participant value with the consensual. A large absolute value indicates that the result is statistically incompatible with the consensual value.
  We also choose the ζ-score, which use a consensual value as reference and the uncertainty of each participant. A large absolute value indicates an underestimated experimental uncertainty.
  C21. We do not use a normalised deviation E_N(for which the criterium is indeed 1). This indicator needs a reference value (and its uncertainty) independent of the participants. Concerning the Z’ and ζ scores, the criteria are higher (2). Indeed, the uncertainties taken in E_N are expanded (k=2), and those taken for Z’ or ζ are not.
  C22. All the values are rounded thanks to the uncertainties, with two significative numbers.
  The Henry's method and log-normal adjustment to calculate the Mass Median Aerodynamic Diameter (MMAD) and geometric standard deviation (σg) are appropriate. However, in the table 4, the uncertainties appear to be quite low, even 5 nm for a diameter of 320 nm, so 1,5%. They are also quite below of the standard deviation. It could be expected that the standard deviation is an estimator of the uncertainty, thus it should be in the same order of magnitude. Despite the good performance of all participants according to the ζ-score, the instrumental uncertainty should eventually be requalified, especially concerning the Henry's method that appears to underestimate experimental uncertainties, which may require improvement. The log-normal adjustment provides more consistent results.
  C23. We significantly improved the last section of our paper to discuss the results.
  C24. A given uncertainty can be compared to other uncertainties for the same kind of instrument in the same kind measurement. We assume that in this case the uncertainties must be equivalent. Thus, if there are discrepancies between the uncertainty evaluations, some components of the uncertainty budget may be underestimated.
  Conclusion
  C25. We have clarified the objectives, the discussion and the conclusion.
  Technical corrections
  C26. The suspension points have been deleted.
  
  Citation: https://doi.org/10.5194/ar-2024-30-AC1
RC2:
'Comment on ar-2024-30', Anonymous Referee #2, 30 Dec 2024

Dear Editor,
The scientific quality of the article is low. I do not recommend it for publication in this journal.
Best regards,
Anikó

Citation: https://doi.org/10.5194/ar-2024-30-RC2
- AC2: 'Reply on RC2', Grégoire Dougniaux, 03 Feb 2025
  
  Thank you for your comment. I regret your conclusion and the lack of comments/suggestions.
  
  Citation: https://doi.org/10.5194/ar-2024-30-AC2

Status: closed

RC1:
'Comment on ar-2024-30', Anonymous Referee #1, 21 Nov 2024
The purpose of this article is to present the results obtained during an exercise, which the authors describe as an interlaboratory comparison, aimed at comparing the results produced by cascade impactors used to determine the mass median aerodynamic diameter (MMAD) of aerosols denoting different size.
While the article's style is pleasant, it's very difficult to identify its real purpose.
Thus, it would appear that the aim is not to compare the analytical capabilities of a number of different laboratories (in this case, only two laboratories from the same entity), but rather to evaluate the performance of 3 instruments using the same technology (atmospheric or low-pressure impactors) operated by a quite limited number of participants. While the real ambition of this article is to demonstrate that cascaded impactors (in this case from two IRSN laboratories) are capable of determining the mass median aerodynamic diameter of an aerosol, this does not represent a major scientific contribution.
Beyond this lack of clarity in the presentation of the article's objective, the analytical methodology presented raises a large number of questions, which you can find listed below, leading me not to recommend such a study for publication in the journal “Aerosol Research”.

Specific comments
C1. Lines 13, 59 and 139: the authors mention a range between 0.2 and 4 µm, but Table 1 only covers a range between 0.3 and 4 µm

Introduction
C2. Line 33: the authors state that the cascade impactor is considered a “gold standard”. On what factual basis (ASTM/ISO standards, OECD guidelines or at least scientific articles) is this statement backed up? What's more, if the cascade impactor is considered a “gold standard”, what's the purpose of this article, which aims to demonstrate its performance?

C3. Lines 43 to 47: the objective of the article is not clearly explained, and the rest of the article adds to the confusion by mentioning participants instead of instruments.

Experimental setup
The experimental configuration presented raises several questions:
C4. Why was quiet air sampling chosen?

C5. The authors mention Grishpun's work on sampling probes, but this work proposes an approach for assessing the aspiration efficiency of probes. What types of probes were considered in this study? Have the results been corrected for aspiration efficiency, which could lead to a change in the expected particle size distribution?

C6. Please specify whether each instrument was attached to a different probe or to a single probe:
If several probes were used, how were they chosen according to the sampling rates of the different instruments?

If a single probe was used, how were differences in sampling rates taken into account?

C7. Has homogeneity, in terms of particle size distribution, been assessed within the test chamber? If so, how, and were any disparities noted (which may justify the use of one or more sampling probes, see my last question)?

C8. Was the relative humidity in the measuring chamber measured during the tests?

C9. Did the authors make sure that the particles produced by nebulization were perfectly solid particles and not residual droplets?

C10. KCl has been used as a source of particles by adding sodium fluorescein: what is the nature of the particles generated in the test chamber? Is it a heterogeneous mixture of KCl and sodium fluorescein particles, or KCl particles “tagged” by sodium fluorescein?

C11. Have electron microscope images been taken, together with elemental analyses, to determine the shape and composition of the particles produced (see my last question)?

C12. Table 2 - ILC participants: the title and content of this table are confusing. Indeed, we might have expected to see different laboratories (and therefore different participants) operating similar instruments. Here, the participants are actually instruments operated by two laboratories belonging to the same entity.

Analysis method
C13. Lines 82 et 85: please explain how the uncertainties associated with Henry's methods and lognormal smoothing are obtained.

C14. Please provide details of the methodology and software used for Henry's method and lognormal adjustment.

C15. Beyond the methods for exploiting the data produced using cascade impactors (or the APS), the uncertainties inherent in the instruments themselves are not discussed or presented. How are the uncertainties inherent in the cut-off diameters of the impactor stages, and in quantifying the mass deposited on each stage, taken into account in the uncertainty budget?

C16. Lines 88-89: the link between fluorescein concentrations and KCl masses on each stage is not clearly explained.

C17. Has weighing the mass of each stage, which is the most conventional approach for cascade impactors, been considered (if possible)?

C18. If the masses available on each stage could not be quantified by weighing, why not consider an ICP-MS analysis to directly obtain the mass of K constituent element of the particles generated?

C19. The term “reference value” suggests an additional objective of the article, namely to propose a consensus MMAD value for the experimental conditions produced on this test bench. Is it really necessary to use this term for an average value derived from measurements performed by only 3 different instruments?

C20. Doesn't the choice of two indicators seem useful? Isn't just one enough (e.g. the Z' score or even the Z score)? If not, please explain why both indicators are needed and useful?

C21. The article by Amarouche (2015), cited by the authors, specifies a different range for the normalized deviation E with satisfactory suitability for a value less than or equal to 1, in which case why are the intervals identical for the two indicators (lines 103-105)

C22. Table 3: Are 3 or 4 significant digits after the decimal point really relevant? This point ties in with my questions about the assessment of measurement uncertainties (inherent in the instruments) and MMAD determination uncertainties (according to Henry's method or lognormal adjustment). The standard uncertainties presented in this table appear very low (5 nm for case E)

C23. Discussion of the results is almost non-existent, and is limited to lines 130 to 135. As it stands, this does not justify a scientific publication.

C24. Lines 134-135: “to underestimate the experimental uncertainties” is not clear to me. What do the authors mean?

Conclusion: C25. as the article is not sufficiently clear on its objectives, it is difficult to identify the real contribution of this article through the conclusion.

Technical corrections
C26. Lines 32 and 33: I don't understand the use of “...”.
Citation: https://doi.org/10.5194/ar-2024-30-RC1
- AC1: 'Reply on RC1', Grégoire Dougniaux, 03 Feb 2025
  
  Response to Reviewer #1.
  We would like to thank Reviewer #1 for his attentive review. Despite his conclusion, he has read the whole document carefully and questioned everything. We have improved our paper according to all his comments/suggestions/questions.
  
  C1. We firstly mentioned a range 0.2 - 4 µm. This have been restrained to 0.3 - 4 µm.
  Introduction
  C2. Gravimetric impactors are arguably the most widely utilised instruments for measuring particle size distribution, and their operating principle is well-documented. Gold standard is a bit exaggerated.
  C3.The aim of this article is to demonstrate the participating instrument performance regarding different particle size-distributions using fluorometric analysis for routine tests of HEPA filters in nuclear facilities.
  For these experiments we proceed as an ILC. Thus, even if few instruments were taken from two laboratories, we identified each couple instrument/lab as a participant. Table 3 shows all the participants of this ILC, with their instruments and characteristics.
  C4. The instruments are connected to sampling probes, with thin edges and a diameter adapted to their nominal air flow. The probes are oriented 90° against the ETNA flux. At nominal ETNA air flux, the air velocity is 0.31 ± 0.02 m/s at the level of the probes. Thus, this sampling is in the calm air condition.
  C5. We add an additional experiment to assess the validity of the homogeneity of the sampling probes, with four APS. There is no size-distribution alteration from 0.4 µm to 10 µm on these instruments. The results are not corrected according to Grinshpun equations.
  C6. Each instrument has been individually connected to an adapted probe. There diameters and sampling velocity are now given in the article.
  C7. The homogeneity has been assessed with a supplementary test with four APS. The results are given in the article.
  C8. The relative humidity in the test bench was not carefully followed. However, the control instrument gives values under 40% RH.
  C9. We do not check the composition of the aerosol, if it his perfectly dry or not. However, it has no influence on the sampling efficiency, thus all instrument sample the same aerosol. Under 40 % of hygrometry, we expect dry particles.
  C10. The fluorescein is use as a tracer, thus the KCl particles are tagged with. With there respective quantities, the fluorescein represents 0 % of the particles mass, except for test E, for which it represents 13 % of the particles mass.
  C11. We do not proceed to electron microscope images for this study. However, it would have been a nice addition to this article.
  C12. Table 2 - ILC participants: the title and content of this table has been clarified in the text.
  Analysis method
  C13. We add a supplementary material to detail the Henry and fit analysis, with the uncertainty propagation.
  C14. We provided details of the methodology and software used for Henry's method and lognormal adjustment in a supplementary material.
  C15. The cut-off uncertainties are taken into account in the global uncertainty budget. From the Dekati calibration certificate, the stokes number are calculated for all stages, then used in the actual operation condition (pressure and temperature). At all calculation steps the uncertainties are propagated according to the GUM. Finally, the uncertainties are taken into account, with the measured mass uncertainties from fluorometric analysis, for the Henry method and log-normal fit to obtain the median and geometric standard deviation.
  C16. The text has been modified to explain clearly the link. To obtain the aerosol mass on each impactor filter, thus the APSD, a fluorometric analysis is performed. “For each impactor, all the filters are taken and prepared for mass measurement. The filters are then placed in new petri dishes and identified on the lid (test number, floor number, etc.). An ammonia solution (NH₃ in H₂O) was made. This solution is prepared by diluting 100 mL of ammonia (28%) in 10 L of ultrapure water. The pH of this solution is greater than 9. The filters are placed in plastic beakers containing 10 mL of the ammonia solution for at least 1 hour. This allows the fluorescein deposited on the filter to pass 100% into solution. An aliquot of the solution, now containing all the fluorescein in the filter, is placed in the measuring cell of a calibrated fluorometer. The measurement range is 5.10^-11 to 1.10^-6 g/ml. The intensity of the fluorescence signal measured gives the concentration of fluorescein in the solution. Because the concentration of KCl and fluorescein is determined in the initial solution (Table 2), the KCl mass is proportional to the fluorescein concentration. This process allows to establish the aerosol mass sampled on each filter”.
  C17. The direct weighing is indeed the most conventional method. However, the expected aerosol mass is below the quantification limit of a balance, thus a fluorometric analysis is far more sensible and far more precise than direct mass weighing. The mass of the deposited particle is roughly of 10^-10 g, and the balance capacity is 10^-5 g.
  C18. An ICP-MS analysis would be indeed a reference measurement, particularly useful to seek specific atoms as K. However, We do not access to this kind of equipment, and a fluorometric analysis, is way more simpler even for the sample preparation, cost less and is precise enough and is used routinely for the HEPA filters tests in Nuclear facilities.
  C19. The term “reference value” has been changed to consensual value.
  C20. The standard NF ISO 13528 lists heigh indicators for ILC analysis, with each case use, each specificity.
  We choose the Z’-score, which use a consensual value as reference and the uncertainty of each participant. It indicates the degree of compatibility of each participant value with the consensual. A large absolute value indicates that the result is statistically incompatible with the consensual value.
  We also choose the ζ-score, which use a consensual value as reference and the uncertainty of each participant. A large absolute value indicates an underestimated experimental uncertainty.
  C21. We do not use a normalised deviation E_N(for which the criterium is indeed 1). This indicator needs a reference value (and its uncertainty) independent of the participants. Concerning the Z’ and ζ scores, the criteria are higher (2). Indeed, the uncertainties taken in E_N are expanded (k=2), and those taken for Z’ or ζ are not.
  C22. All the values are rounded thanks to the uncertainties, with two significative numbers.
  The Henry's method and log-normal adjustment to calculate the Mass Median Aerodynamic Diameter (MMAD) and geometric standard deviation (σg) are appropriate. However, in the table 4, the uncertainties appear to be quite low, even 5 nm for a diameter of 320 nm, so 1,5%. They are also quite below of the standard deviation. It could be expected that the standard deviation is an estimator of the uncertainty, thus it should be in the same order of magnitude. Despite the good performance of all participants according to the ζ-score, the instrumental uncertainty should eventually be requalified, especially concerning the Henry's method that appears to underestimate experimental uncertainties, which may require improvement. The log-normal adjustment provides more consistent results.
  C23. We significantly improved the last section of our paper to discuss the results.
  C24. A given uncertainty can be compared to other uncertainties for the same kind of instrument in the same kind measurement. We assume that in this case the uncertainties must be equivalent. Thus, if there are discrepancies between the uncertainty evaluations, some components of the uncertainty budget may be underestimated.
  Conclusion
  C25. We have clarified the objectives, the discussion and the conclusion.
  Technical corrections
  C26. The suspension points have been deleted.
  
  Citation: https://doi.org/10.5194/ar-2024-30-AC1
RC2:
'Comment on ar-2024-30', Anonymous Referee #2, 30 Dec 2024

Dear Editor,
The scientific quality of the article is low. I do not recommend it for publication in this journal.
Best regards,
Anikó

Citation: https://doi.org/10.5194/ar-2024-30-RC2
- AC2: 'Reply on RC2', Grégoire Dougniaux, 03 Feb 2025
  
  Thank you for your comment. I regret your conclusion and the lack of comments/suggestions.
  
  Citation: https://doi.org/10.5194/ar-2024-30-AC2

Grégoire Dougniaux, Cécile Bodiot, Bernadette Dhieux Lestaevel, Amandine Nuboer, Robin Wahl, and Amel Kort

Viewed

Total article views: 166 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
130	27	9	166	10	10

HTML: 130
PDF: 27
XML: 9
Total: 166
BibTeX: 10
EndNote: 10

Views and downloads (calculated since 11 Nov 2024)

Month	HTML	PDF	XML	Total
Nov 2024	70	10	3	83
Dec 2024	19	9	2	30
Jan 2025	21	4	2	27
Feb 2025	20	4	2	26

Cumulative views and downloads (calculated since 11 Nov 2024)

Month	HTML	PDF	XML	Total
Nov 2024	70	10	3	83
Dec 2024	19	9	2	30
Jan 2025	21	4	2	27
Feb 2025	20	4	2	26

Viewed (geographical distribution)

Total article views: 196 (including HTML, PDF, and XML) Thereof 196 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 21 Feb 2025

Short summary

Ensuring that all measuring instruments deliver reliable results is an important task. For aerosol measurement, a health-related parameter is not yet well defined worldwide. We have therefore developed a test bench to evaluate and compare instruments on this key parameter. Based on the success of an initial comparison with our instruments, we underline the feasibility of a standardized test and offer a framework for improving accuracy and consistency of aerosol size distribution assessments.


Total:	0
HTML:	0
PDF:	0
XML:	0