the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 10 Jun 2026
Traceable calibration of in-situ aerosol absorption instruments with monodisperse nigrosin
Abstract. The absorption coefficient of light absorbing aerosols is difficult to measure with low uncertainty and improvements of measurement procedures based on traceable calibration are needed. Reducing measurement artefacts can be achieved by using in-situ direct measurement methods such as photo-acoustic spectroscopy and photo-thermal interferometry. We developed a traceable calibration method based on monodisperse nigrosin particles. Nebulized nigrosin forms nearly spherical particles for which it is possible to calculate the absorption coefficient in the UV-NIR range using Mie theory. In the presented study, we have experimentally tested size- and mass-selection techniques using a differential mobility analyzer (DMA), a centrifugal particle mass analyzer (CPMA) and a tandem of both in series to compare Mie-calculated absorption coefficient with the measured one using the photo-thermal aerosol absorption monitor PTAAM, traceably calibrated with NO2. We observed that the nigrosin particle density changes with particle size. Because the absorption coefficient depends mainly on particle mass it is preferrable to base the Mie calculation on the measured particle mass instead of the mobility diameter. Calculated versus measured absorption coefficients differed by +5 % to +11 % for the DMA, -2 % to -3 % for the CPMA and +2 % to +8 % for tandem of the CPMA and the DMA. Combined standard uncertainties (coverage factor k=1) for PTAAM calibrated with monodisperse nigrosin particles selected by the DMA, the CPMA, and the CPMA and the DMA tandem are 6.9 %, 5.8 % and 5.2 %, respectively. The optimal classification setup is a tandem of the CPMA and the DMA which avoids the systematic bias of both neutral (in the CPMA) and multiply-charged (in the DMA) particles and provides a high enough absorption signal. Experimentally the simplest selection method is based on the CPMA.
Competing interests: LD, GM and JYD are employed by Haze Instruments d.o.o., the PTAAM manufacturer. GM is an editor for AMT.
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
(1290 KB) - Metadata XML
-
Supplement
(941 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-1501', Anonymous Referee #1, 19 Jul 2026
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-1501/egusphere-2026-1501-RC1-supplement.pdfCitation: https://doi.org/
10.5194/egusphere-2026-1501-RC1 -
RC2: 'Comment on egusphere-2026-1501', Anonymous Referee #2, 20 Jul 2026
Review of "Traceable calibration of in-situ aerosol absorption instruments with monodisperse nigrosin" by Drinovec et al. (egusphere-2026-1501)
General comments
This is a careful, well-organised and timely metrological study on a problem of real importance to the AMT readership. Traceable calibration of in-situ absorption instruments is a recognised bottleneck, and the systematic side-by-side evaluation of DMA, CPMA and tandem classification, together with an explicit uncertainty budget, is a valuable and practically useful contribution. The identification and quantification of the two competing biases (multiply-charged particles in the DMA, neutral particles in the CPMA) is a genuinely useful result, and the recommendation of the tandem configuration follows logically from it. The experimental work appears sound and the manuscript is clearly written. I recommend publication after the following points have been addressed. Several of them are substantive and, in my view, amount to a major revision, but all should be tractable with the data already in hand.
1. The size dependence of the measured nigrosin density deserves a deeper analysis (Table 1, p. 6; Fig. 3B, p. 5; L188-191, p. 5).
This is, in my opinion, the most interesting and the least developed result of the paper. If Dp and Dm from Table 1 are recast as a difference rather than a ratio, one obtains Dp - Dm = 3.8, 4.2, 4.3, 4.4 and 3.8 nm for mp = 0.1, 0.3, 1, 3 and 10 fg. In other words, the discrepancy is an essentially constant additive offset of ~4 nm in diameter over two decades in particle mass, and the apparent "size dependence of the density" is simply the geometrical consequence of that constant offset. This is a strong constraint on the physical explanation, and it is not consistent with the explanation currently offered in the text (fusion of incompletely dried particles, L188-191), which would produce a size-dependent void fraction rather than a constant diameter offset. A constant offset of this kind - equivalent to a shell about 2 nm thick - is much more suggestive of, e.g., residual water on these hygroscopic particles at the operating RH, a residual solute or surfactant layer, or a systematic offset between mobility and volume-equivalent diameter for this material. I would encourage the authors to present the data in this form and to discuss the candidate explanations explicitly - it would considerably strengthen the paper. Note also that the polystyrene bead validation (0.8 % in density, i.e. ~0.25 nm in diameter, Table S1) rules out a simple DMA sizing offset and therefore points to a material-specific effect. This matters beyond bookkeeping: if the particles carry a thin low-index shell, the homogeneous-sphere Mie model is not strictly appropriate, and the sign of the residual bias in Figs. 6B and 10B may be related.2. The uncertainty budget omits the contribution of the assumed material density (Table 2, p. 10).
The manuscript states (L194-195, p. 5) that basing the Mie calculation on Dp rather than Dm changes the calculated absorption coefficient by 5-20 %, which shows that the result is strongly sensitive to the assumed particle density. Yet Dm is obtained from the CPMA-selected mass together with an assumed rho = 1.6 g cm-3, and the paper itself demonstrates that the measured effective density deviates from this value by up to 20 %. Table 2 contains a 2 % component for CPMA mass and a 2 % component for the Mie calculation and refractive index, but no component for the density / mass-equivalent-diameter conversion. Please add this component explicitly, state the sensitivity coefficient (d ln babs / d ln rho), and justify the value adopted. Depending on the answer to point 1, this could be one of the larger terms in the budget, and it would affect the ranking of the three configurations.3. The particle-based method is validated only at 450 nm, although its main motivation is the near-IR.
The Introduction motivates particle-based calibration by the fact that NO2 absorption is too weak for sub-ppm calibration in the near-IR (L79-80, p. 2). The PTAAM P02 has an 808 nm channel, and the 808 nm calibration is stated to be transferred from 450 nm using the Mie-calculated babs(808)/babs(450) ratio (L136-138, p. 4). However, all measured-versus-calculated comparisons presented (Figs. 4B, 6B, 8B, 10B) are at 450 nm. The central practical claim of the paper - that monodisperse nigrosin enables calibration where NO2 cannot be used - is therefore not directly tested. If 808 nm data were recorded, please show the same comparison at 808 nm; if not, please state this limitation explicitly in the Discussion and Conclusions and temper the corresponding claims. The wavelength dependence of the multiple-charge and neutral-particle biases (both of which act through the size dependence of the mass absorption cross-section) would also be worth a sentence.4. Please clarify the traceability claim and the logic of the comparison.
The title and abstract present the nigrosin method as a "traceable calibration". Strictly, the traceable chain demonstrated here runs NO2 permeation standard -> CAPS -> PTAAM (450 nm); the nigrosin comparison then relies on Mie theory plus a literature refractive index, neither of which is traceable to SI in the metrological sense. What the experiments demonstrate is a validation of the nigrosin-based method against an NO2-traceable calibration, with the residual differences (+5 to +11 %, -2 to -3 %, +2 to +8 %) setting a floor on the achievable agreement. For a paper of this metrological character I think this distinction should be made explicitly, and the wording in the title, abstract and conclusions adjusted accordingly. Relatedly, please state whether the residual differences are statistically consistent with the combined uncertainty of the comparison (which must combine the calibration uncertainty and the Mie-calculation uncertainty), or whether an unexplained bias remains.5. Refractive index provenance and batch dependence.
The manuscript itself notes that nigrosin optical properties vary between batches (L93-94, p. 3, citing Foster et al., 2019), and then adopts refractive index values determined previously (L171-173, p. 5). Please state explicitly whether the nigrosin batch used here is the same physical batch as in Drinovec et al. (2022), and if not, how the transfer was justified. If nigrosin is to be recommended as a practical transfer standard, the procedure should specify whether the refractive index must be re-determined for each batch, and what that adds to the uncertainty. A short, explicit step-by-step recommended protocol (classification settings, Rm, solution concentration, required checks) would substantially increase the practical value of the paper for AMT readers.6. Generality beyond the PTAAM.
The title refers to "in-situ aerosol absorption instruments" in general, but only the PTAAM is tested, and the conflict of interest statement notes that three authors are employed by the manufacturer (which is properly declared and is not in itself a problem). To support the general claim it would be valuable to apply the method to at least one independent instrument, e.g. a photo-acoustic spectrometer. If that is not possible, please discuss explicitly what would change for PAS - for instance microphone/resonator response, sensitivity to gas-phase absorbers and to relative humidity - and consider narrowing the title.7. Repeatability and the interpretation of the error bars.
Figures 4B, 6B, 8B and 10B carry error bars on both the measured and the calculated values, but the captions do not state what these represent - the standard deviation of the raw signal, the standard error, or the combined uncertainty of Table 2 - nor how many independent repetitions underlie each point. For a calibration method, repeatability is a key figure of merit. Please define the error bars in the captions, state the number of repetitions and the period over which they were acquired, and, if the bars represent within-run scatter only, add a statement of day-to-day reproducibility. The latter would be particularly convincing for a method intended as a transfer standard.Specific comments
p. 1, L33-34 (Abstract): The phrase "Calculated versus measured absorption coefficients differed by ..." would be clearer if the sign convention were stated, i.e. that positive values mean the Mie calculation exceeds the measurement. Please also state that all these values refer to 450 nm.p. 2, L43-58 (List of Abbreviations): SMPS is used in Sect. 3.1 (L180, p. 5) and rho / Rm are used throughout but none appears in the list. Please add SMPS, Rm and the density symbols. Also, Dp is defined as "electrical mobility diameter" here but referred to as "mobility diameter" and "particle diameter" elsewhere; please use one term consistently.
p. 3, L124-127 (PTAAM version): P02 uses laser diodes at 450 and 808 nm, whereas P01 used DPSS lasers at 532 and 1064 nm. Laser diodes have a considerably broader and more temperature-sensitive emission spectrum. Given that the NO2 absorption cross-section is strongly structured near 450 nm, the effective wavelength, the spectral width (FWHM) and its stability are critical for the calibration. The text mentions that the pump spectrum was measured with a CCS100/M spectrometer (L133-135), which is good practice, but please report the measured centre wavelength and FWHM, describe how the convolution with the NO2 cross-section was performed, and include the resulting uncertainty contribution in Table 2.
p. 4, L136-138: Since the refractive index values are quoted "at 450 nm and 808 nm" from a study performed at 532 and 1064 nm, please state whether these were measured directly at 450/808 nm or interpolated from a measured spectrum, and give the associated uncertainty.
p. 4, L140-146 and Fig. 1: The 15 min Allan deviation is 0.03 Mm-1, but the "actual measurement uncertainty for 15 min experiments" is stated as 0.3 Mm-1, a factor of ten larger. Please explain the origin of this factor more quantitatively. At the signal levels used for the monodisperse comparisons (babs = 14.5-38 Mm-1, Table 3) this corresponds to 0.8-2 %, which is not negligible relative to the 5.2-6.9 % combined uncertainty; please state whether it is already covered by component K.
p. 4, L152-154: "at sample flow of 3 L min-1 sample flow" - the phrase "sample flow" is duplicated. Please also state the RH after the diffusion drier more precisely than "below 30 %", since residual water is directly relevant to the density discussion (General comment 1).
p. 5, L183-185 and Fig. 3: The sentence "Effective particle density and particle volume were calculated based on the selected particle mass (Fig. 3A) and the mobility diameter, respectively" is hard to follow on first reading. The reference to Fig. 3A is presumably to the selected masses shown in the legend, whereas the resulting densities are plotted in Fig. 3B, which is where a reader would expect the callout to point. Please rephrase so that it is clear which quantity is taken from which panel.
p. 5, Fig. 3 caption, L204: "coverage factor_k= 1" - formatting error; should read "coverage factor k = 1".
p. 5, L188-191: "The low experimental values in the particle density for nigrosin might be due to fusion of particles before being completely dry." Please see General comment 1. In addition, if fused (doublet or higher) particles are present in significant numbers, this also undermines the homogeneous-sphere assumption in the Mie calculation. Please quantify the fraction of fused particles seen in the SEM images (Fig. S5) and bound the resulting error, e.g. by comparing with a dynamic shape factor estimate.
p. 6, Table 1 caption: "mobility dimeter" -> "mobility diameter". Please also add a column for Dp - Dm (see General comment 1), and state the measured effective density uncertainty (9.2 %) alongside the tabulated values rather than only in the caption.
p. 6, L215-218: The volume size distribution shows a non-negligible contribution above 350 nm that is attributed to an artefact of the inversion algorithm, and particles above 400 nm are simply excluded from the Mie calculation. This is a fairly consequential ad hoc choice. Please quantify how much the calculated polydisperse absorption changes if the cut-off is moved to, say, 300 or 500 nm, so that the reader can judge the sensitivity of the +10 % polydisperse result to this choice.
p. 6, Fig. 4 caption, L226: "nigrosine" -> "nigrosin" (used elsewhere throughout).
p. 7, L253-256: "the absorption was then calculated as the sum of the absorption of singly-charged particles with mass m0, doubly-charged particles with mass 3.5*m0 and triply-charged particles with mass 6.6*m0". Please state how the factors 3.5 and 6.6 were obtained (presumably from the measured mass-mobility relation) and give their uncertainty, since these directly determine the 15-25 % correction. Also, "multiply-charged particles accounted for 15 % and 25 % of the total light absorption" gives two numbers, whereas three particle masses (0.7, 2.4 and 5.7 fg) are compared in Fig. 6B; please give the value for each of the three settings.
p. 7, L257-259: A resolution of Rm = 1.5 is very broad. Please state the resulting width of the selected mass distribution and quantify the error introduced by treating such a broad distribution with a single mass-equivalent diameter in the Mie calculation (the mass absorption cross-section varies with size, cf. Fig. S3).
p. 8, L296-301 and Fig. 8A: The neutral-particle contribution is quantified as 2.6 % of total absorption at Rm = 3 and then described as "negligible" at higher Rm. Fig. 8A shows data at Rm = 3, 7 and 15, so please give the measured numerical values for Rm = 7 and 15 rather than the qualitative statement, and state the detection limit of the electrostatic precipitator method.
p. 8, L276-277: "Because the fraction of the multiply charged particles is several times smaller compared to the singly charged particles the error is smaller than 1 % and the correction was not applied." Please give the measured fraction and show the arithmetic leading to <1 %.
p. 9, L321-325: In the tandem configuration the CPMA is run at Rm = 5 and the DMA at Rm = 3. Please state explicitly the resulting mass and mobility transmission windows, and confirm quantitatively that doubly-charged particles (Dp = 93 nm for m0 = 1 fg, as stated at L312-313) are indeed outside the DMA transfer function at Rm = 3, since 93 nm and 110.4 nm differ by only about 16 %.
p. 10, L336-338: "The combined uncertainty is calculated as the square-root of the linear sum of squared standard uncertainty components" - the wording is confusing. Please rephrase as "the root sum of squares of the standard uncertainty components", and state explicitly the assumption that the components are uncorrelated (which is stated in the Table 2 caption but should also appear in the text).
p. 10, Table 2: (i) Please add the density / mass-equivalent-diameter component (General comment 2) and the particle-loss-correction component (comment on L138-139). (ii) Component C, "Mie calculation & nigrosin refractive index", is given as 2 % from 2 % and 3 % uncertainties in the real and imaginary refractive index; please show the sensitivity coefficients, since these are size- and wavelength-dependent. (iii) Please indicate for each component whether it is Type A or Type B, as is conventional in a metrological uncertainty budget.
p. 10, L349-350: "We accessed 5 % additional uncertainty" -> "We assessed". Please also state how the 5 % was arrived at.
p. 10, L357-358: The CPC uncertainty is taken as 3 % from the literature. Since both a CPC and an FCAE were available in this study, a direct CPC-versus-FCAE intercomparison would substantiate this value at essentially no extra experimental cost, and would strengthen the traceability argument for the DMA and tandem configurations. Was such a comparison performed?
p. 11, L395-397: The AAC comparison is a useful addition. Since the AAC selects by aerodynamic diameter, converting to mass again requires a density assumption; given the present results this caveat is worth one sentence.
p. 12, Table 3: The entry "Resolution parameter Rm = 0.1" for the polydisperse case is not meaningful, as a polydisperse distribution has no resolution parameter. Please replace it with "n/a" or with the geometric standard deviation of the distribution.
p. 12, L424-426 (Conclusions): The tandem configuration is declared "optimal" on the basis of a combined uncertainty of 5.2 % versus 5.8 % for the CPMA alone. Given the size of the individual budget components, this difference is not significant. I would suggest presenting the two as essentially equivalent in uncertainty and distinguishing them on the practical grounds the authors already give (signal level versus experimental simplicity), which is the more defensible and more useful message.
p. 12, L410-412 (Conclusions) and Abstract: "preferrable" -> "preferable" (occurs in both the Abstract, L32, and the Conclusions, L411).
Figures 2, 5, 7 and 9: These four schematic diagrams each occupy a separate figure but consist of only four to six boxes. Combining them into a single four-panel figure would save considerable space and, more importantly, would let the reader compare the four configurations at a glance, which is the central comparison of the paper.Figures 3, 4, 6, 8 and 10: The plots are presented in a spreadsheet-style default format. Axis label and tick font sizes are small and inconsistent between panels, and the panel identifiers (A)/(B) sit below the panels rather than within them. Please redraw to a consistent style and at publication resolution, following the AMT figure guidelines. In the bar charts (Figs. 4B, 6B, 8B and 10B) a difference plot (calculated minus measured, in %) as a second panel would convey the key result far more directly than the side-by-side bars.
References: The reference list contains an entry "Drinovec, L., Jagodic, U., Pirker, L., et al.: A dual-wavelength photothermal aerosol absorption monitor ..., 2022" and a second entry "Supplement of: A dual-wavelength photothermal aerosol absorption monitor ..., 2022b", but there is no entry labelled "2022a". The in-text citations therefore mix "Drinovec et al. (2022)" and "Drinovec et al. (2022b)", leaving the "b" suffix orphaned. Please relabel the two entries as 2022a and 2022b consistently, or cite the supplement as part of the main reference. Please also check "Romshoo et al, 2022" (L71, p. 2), which is missing a period after "al", and the inconsistent rendering of "Yus-Diez et al., 2021" versus "Yus- Diez et al., 2025" (also L71, p. 2), the latter carrying a stray space.
Citation: https://doi.org/10.5194/egusphere-2026-1501-RC2
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 143 | 56 | 23 | 222 | 30 | 17 | 16 |
- HTML: 143
- PDF: 56
- XML: 23
- Total: 222
- Supplement: 30
- BibTeX: 17
- EndNote: 16
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1