| 23 Jan 2026
Sodium Thiosulfate-Coated Ceramic Denuders for Ozone Removal in Ultrafine Particle Sampling
Abstract. Ozone (O₃) remaining in sampling air can artefactually alter the chemical composition of collected ultrafine particles (UFP), biasing quantitative analysis of the chemical composition. In this study, we developed and evaluated a sodium-thiosulfate O₃ denuder (TSOD) specifically tailored for UFP sampling and assessed its O₃ scrubbing efficiency, particle losses, and chemical selectivity. In laboratory tests under controlled relative humidity and inlet O₃ levels up to 200 ppbV, the outlet concentration remained consistently between 0 and 0.3 ppbV, demonstrating the O₃ removal efficiency of the TSOD. During an urban field deployment over 7 days O₃ downstream of the TSOD consistently remained at 0 ppbV while ambient O₃ varied between 0 and 65 ppbV. Moreover, for particles with mobility diameters ranging from 10 to 1000 nm, we did not observe any significant losses in particle number concentrations. Using a parallel two-channel UFP sampler (with vs. without upstream TSOD), we quantified O₃-driven sampling artefacts in UFP mass focussing on three types of organic markers. (1) Firstly, we targeted polycyclic aromatic hydrocarbons (PAH), particularly chrysene (Chry), benz[a]anthracene (BaA), benzo[a]pyrene (BaP), indeno[1,2,3-cd]pyrene (IcdP), benzo[k]fluoranthene (BkF), and benzo[b]fluoranthene (BbF). Without upstream O₃ removal, the individual concentration of the PAH were 15–46 % lower. (2) Secondly, for the tire and road wear marker, the antioxidant N-(1,3-dimethylbutyl)-N′-phenyl-p-phenylenediamine (6PPD) and its oxidation product 6PPD-quinone (6PPDq), we observed in-situ ozonation of 6PPD to 6PPDq with transformation yields of about 13 to 20 %. (3) In contrast, biogenic organic acids (bOAs) did not show differences when sampled with or without O₃, as their O₃ reactivity is much lower than the one of the PAH. Moreover, this test indicated that the TSOD did not perturb the gas–particle partitioning of these semi-volatile species. Our results demonstrate that the TSOD (i) efficiently scrubs atmospheric O₃ at relevant mixing ratios, (ii) does not introduce measurable particle losses across 10–1000 nm, and (iii) preserves semi-volatile partitioning.
General Comments:
This article reports on the design and application of a thiosulfate-impregnated honeycomb denuder to mitigate chemical losses from reaction with ozone in ultrafine particle sampling. This is important work, drawing attention to the unfortunately often overlooked interference from ozone in the sampling of atmospheric organic matter. I very much appreciate this study and its reporting and recommend the publication after consideration of a number my specific comments given below.
It would be valuable to also have these experimental details provided: What fraction of ozone makes it through the sampling system without the sodium thiosulfate coating? Given that ozone is a rather reactive gas, a fair amount gets probably removed just by contact with the impactor and plumbing system?
Given the multiple sampling and analysis steps, the chemical quantifications need a thorough experimental determination of the analytical reproducibility of the different chemical classes for the UFP determination. This should be done by deploying complete parallel samplers and not just by parallel sub-sampling of the fractionated aerosol or multiple chemical analyses of the extracts.
Specific Comments:
Line 21, 22: Instead of writing “ 0 ppb”, it would be more accurate to give the threshold of the ozone determination sensitivity, e.g. < 0.3 ppb, or whatever the detection limit of the utilized ozone sensor is.
Line 28: Please give the uncertainty margins of the 15-46% determination.
Line 30: Same here.
Line 49: Possibly also cite the WMO GAW Measurement Guidelines, GAW Report No. 281, Guidelines for Measurements of Non-Methane Hydrocarbons in the Troposphere, that emphasize the need to remove ozone in the sampling of volatile organic compounds.
Line 74. One could also consider [Helmig and Greenberg, 1995].
Line 144: Please mention that these are honeycomb structure channels and that CPSI stands for cells per square inch. How are these connected to your system plumbing? It might be nice to show a photograph of the denuder.
Line 148: I have a hard time believing that shaking of ta 400 CPSI honeycomb denuder will remove all the water? What is the actual diameter of the individual capillaries? This could possibly be checked by weighing a dry denuder and one that has been “shaken”. If residual water remains in the denuder, then that may actually affect the coating efficiency when the denuder is subjected to the sodium thiosulfate solution?
Line 148. “5,6 mol L-1”? Please be consistent with using a decimal point for decimals.
Line 155: Please give complete details about the drying method.
Line 187: Round site coordinates to a reasonable number.
Line 254: Are the error margins 1-sigma standard deviations?
Line 264: Give numeric results and stated uncertainty comparison.
Line 287: How do you conclude that the sodium thiosulfate efficiency is dependent on relative humidity rather than specific humidity?
Line 298: The ozone monitor that was used really isn’t suited for measuring levels below 1 ppb, and the lowest detectable ozone will be quite sensitive to the zero offset that is applied in the monitor. Therefore, the reliably measurable lowest ozone level needs to be carefully determined in a set of zero ozone measurements and only numerical values above the determined ozone detection limit should be reported as numerical data. All other recordings need to be reported as below the detection limit (< x.x ppb) of the measurement.
Figure 4 caption: Replace ‘concentration’ with ‘mixing ratio’ here and elsewhere.
Table 1: Explain in table caption what the reference data are.
Line 458: Most journals these days do not accept this data availability statement. Data should be shared with readers readily within the Supplemental Materials or through a public archive.
Helmig, D., and J. Greenberg (1995), Artifact formation from the use of potassium-iodide based ozone traps during atmospheric sampling of trace organic gases, Journal of High Resolution Chromatography, 18, 15-18.