the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 04 Nov 2024
Preparation of hybrid calibrated absorption cross sections for a compact UV-DOAS measurement
Abstract. A low-cost differential optical absorption spectroscopy (DOAS) spectrometer has been calibrated using a hybrid approach in which multiple absorption cross sections (ACSs) are measured and compared with previously obtained ACSs. The ACSs obtained from the measurement in this study are referred to as precise measurement (PM) ACSs, while those obtained from the simulation are referred to as hybrid simulation (HS) ACSs. The compact UV grating spectrometer was used to measure PM ACSs at different pressures and temperatures ranging from 295.75 K to 298.15 K. The spectral range was 230–320 nm with a spectral resolution of 0.28–0.39 nm. Uncertainties were evaluated and traceable to SI units. The relative standard measurement uncertainty of the PM ACS for o-xylene, p-xylene, SO2, benzene, styrene and toluene were 3.1 %, 3.1 %, 3.1 %, 3.0 %, 3.0 % and 2.9 %. respectively. This combined relative standard uncertainty includes contributions from optical path length (1.4 %), CRM concentration (≤0.74 %), absorbance repeatability (≤1.5 %), pressure (0.78 %) and temperature (2.1 %). ACSs for reactive trace gases without CRMs were derived using the HS method. This involved convolving literature spectra with measured instrumental functions (IFs) and subsequently applying spectral fitting to achieve ACSs that are well-aligned with those that would be obtained if measured with this spectrometer. The uncertainty of the HS ACSs results from the referenced ACSs, the determination of the IFs and the fitting procedures. The uncertainty of the referenced ACSs is taken directly from the literature and ranges from 0.05 % for benzaldehyde and formaldehyde to 5 % for oxygen. The uncertainty of the IFs is 0.67 %, 0.74 % and 0.65 % in regions I, II and III respectively. The total uncertainty for HS ACS is estimated to be 0.25 % for benzaldehyde, 1.01 % for NO2, 0.95 % for formaldehyde, 0.75 % for p-cresol, 0.74 % for m-xylene, 4.13 % for phenol, 4.19 % for ethylbenzene, 2.83 % for ozone in region I, 3.51 % for ozone in region II, 5.24 % for oxygen in region I and 5.31 % for oxygen in region II, taking into account the above uncertainty components. In a laboratory measurement of a gas mixture (benzene, toluene and ortho-xylene), the difference between the measurement values of the certified reference material (CRM) produced by the gravimetric method and the DOAS measurement is 8.5 % for benzene, 1.6 % for toluene and 4.9 % for ortho-xylene. Therefore, the uncertainty of the DOAS system was estimated to be 15.2 % for benzene, 8.2 % for toluene and 11.6 % for o-xylene, taking into account the uncertainties of the ACS (≤3.1 %), the fitting procedure (≤0.6 %), the difference with the CRM value (≤8.5 %) and the path length (1.4 %). In general, the system is ready for use in field measurements using the long-path DOAS technique.
- Preprint
(1395 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on amt-2024-153', Anonymous Referee #3, 25 Nov 2024
The manuscript of Trisna and co-authors applies the well-established active DOAS (Differential Optical Absorption Spectroscopy) technique in the deep UV spectral range between 230 and 320nm. The authors set up an own custom build active DOAS system for the measurements of aromatic compounds like benzaldehyde, formaldehyde, NO2, p-cressol, m-xylene, phenol, ozone and other in a measurement cell. The instrument setup was very basic with a low-quality compact spectrometer and without fulfilling quality criteria for such DOAS instruments like described in Platt and Stutz (2008).
The authors observed a difference between modeled absorption spectra and measured absorption spectra of the system with reference gases. The manuscript describes the development of a hybrid calibration of the UV DOAS system combining modeled and measured absorption spectra. From the topic this fit in the scope of AMT. While the development of this hybrid calibration seems to be reasonable and may be scientifically valuable, I cannot recommend this manuscript for publication due to the following severe issues:
- The measurement setup does not fulfill basic quality criteria for such active DOAS systems (details on the obvious quality issues listed below). With a poor setup any measurement technique likely derives wrong measurement results. It is no surprise that a high difference between modeled and measured SO2 absorption spectra of up to 62% (p. 4 l. 93) is observed. This is not due to the DOAS measurement technique but due to the setup and data processing. The authors did not try to improve the instrument and correct instrument effects. This is a lack of thoroughly scientific work. I would expect that the differences between modeled and measured absorption cross section will almost disappear with appropriate setup and data correction.
- The setup feature basic lack of instrument requirements:
- The use of a xenon arc lamp (p. 5 l. 117) will cause a high rate of unused light which will act as stray light, as the main emission range of this lamp is in the range of 350 to 600 nm. This will cause a high stray-light amount. No reduction and characterization of stray light is applied which will be relevant for these measurements. Measurements in this spectral range would better use deuterium lamps or LEDs.
- The used spectrometer AVASpec-ULS2048LTEC (p.5 l. 131) is not well suited for this spectral range. It features a very low quantum efficiency of ~ 10% in the UV spectral range, causing not only a low signal, but also a high stray light. A spectrometer with a back-thinned or CMOS detector are much better suited for this deep UV spectral range due to a higher quantum efficiency.
- Stray light in the spectrometer is not characterized and corrected. The system will have significant stray light which requires correction.
- The spectrometer configuration with an entrance slit of 50 µm width (p. 5 l. 133) would cause with this spectrometer a spectral under-sampling of the instrument line function. The full width half maximum should be minimum 5 to 6 pixels (Platt and Stutz, 2008), but here it will be ~ 4 pixel. Modeling absorption spectra can thus be wrong.
- It looks like non-linearity of the spectrometer detector are not considered. The applied detector has a very strong non-linear response influencing the measurements.
- The instrument line function of the spectrometer (Figure 2) is not acceptable. Compact spectrometers have a limited quality, but this is much worse than normal. Avantes AVASpec-ULS spectrometers have a much better instrument line function similar to Gaussian shape which do not vary much over the spectral range (when correctly adjusted). Also, the baseline around the instrument line function is not zero and varying over the spectral range. This may indicate a stray light issue or defect spectrometer grating. Something is clearly wrong with this spectrometer or setup.
- The quality of the wavelength calibration is not described which may have here a significant contribution to the difference.
- The measurement with calibration bottles ignores losses and reaction of the gas in the bottle, on cell walls, inlets etc.. The gas concentration in the measurement cell may be different than expected. This is a big disadvantage of the reference gas method in comparison to the modeled absorption cross section method and can have a significant influence on the comparison.
The authors need first to quantify if the differences are due to the low-quality setup and what cause the problems. Second the effects like stray light need to be corrected as good as possible. Afterwards I can see a scientific relevance for the manuscript if the hybrid calibration is developed as a tool for low quality UV-DOAS systems (which do not fulfill typical DOAS quality criteria) to allow such systems still to derive representative measurement data. However, the manuscript than need a complete rewriting as the focus will change. A misunderstanding that the correction is needed due to the DOAS technique or due to the use of compact spectrometers must be avoided.
Minor technical points:
- p. 4 l. 88: The convolution not necessarily need to assume a constant instrument line function. A broadening can be simulated and included in the modeling.
- p. 4 l. 101: The temperature dependence of ACSs are in IR large, but almost ignore-able in the UV.
- p. 5 l. 133: What grating type? What is the blaze wavelength?
- p. 10 l. 283: The DOAS fit procedure can also include shift and squeeze of the modeled ACS. This is here required due to an insufficient wavelength calibration accuracy. A correction for the hybrid simulation is understandable, but it is not a general issue for the DOAS fit procedure.
- Figure 1: What is the fiber optic type and gas cell type?
- Figure 4: The measured and modeled ACSs (upper panel) seem to agree better than the comparison between the measured and the hybrid simulated ACSs (lower panel). So hybrid is worse than modeled? This seem to be wrong.
- Figure 5: The large difference of the modeled ACSs depending of the instrument line function indicate a spectrometer or setup issue. They should be very similar.
Literature:
Platt, U. and Stutz, J. (2008) Differential Optical Absorption Spectroscopy Principles and Applications. Springer-Verlag. http://www.springer.com/environment/environmental+engineering+and+physics/book/978-3-540-21193-8
Citation: https://doi.org/10.5194/amt-2024-153-RC1 - AC1: 'Reply on RC1', Jeongsoon Lee, 08 Dec 2024
-
RC2: 'Comment on amt-2024-153', Anonymous Referee #1, 25 Nov 2024
The manuscript provides a detailed discussion of absorption cross-sections using DOAS technology, and the experimental procedures are described comprehensively. However, in its current form, the manuscript requires substantial revision to improve clarity and consistency.
Major Concerns
- The manuscript’s writing and structure are often confusing, making it difficult to understand the main arguments and methods. A notable example is the inconsistent definition of the hybrid simulation (HS) method for absorption cross-sections (ACSs). For instance, in Lines 163–166:
Hybrid simulation (HS) of ACSs involves converting the reference absorption spectrum from the literature using the instrument function (IF) to simulate how the absorption spectrum would appear if measured by the specific spectrometer.
However, after Section 3.3, the HS method appears to change its definition to include convolution and nonlinear least-squares (NLS) fitting. Throughout the manuscript, the term "HS method" is used multiple times, and it is unclear whether it refers to a standard methodology or a novel approach proposed by the authors, named the hybrid calibration method. More clarity and differentiation are necessary. - Similarly, there is ambiguity regarding the blue line in Figure 5—does it represent the high-resolution cross-section or the convoluted cross-section? The manuscript appears to present conflicting descriptions, which require resolution for accurate interpretation.
- As stated by the authors, commonly used cross-sections are either convoluted or experimentally measured. My understanding is that the authors have proposed an innovative method that not only convolutes the high-resolution cross-sections but also applies NLS fitting. This method could potentially be extended to other gases within the same spectral range where experimental cross-sections are unavailable (e.g., Line 295). While this seems a promising innovation, I have reservations about its broader applicability. Absorption cross-sections reported by different studies often exhibit inherent differences, raising doubts about the validity of applying the same correction factor across various studies. To address these concerns, I suggest that the authors provide additional experimental evidence, such as simultaneous measurements of benzene and toluene cross-sections within the same spectral range, comparing the correction factors obtained through separate NLS fittings.
- The correction factors presented in Table 5 are extremely small. Would these corrections not already be manageable within the DOAS fitting process without requiring the proposed methodology? Additional clarification is needed to justify the necessity of this approach.
Minor Concerns
- Figure 5 appears before Figure 3 in the manuscript, which disrupts the logical flow. Please reorder the figures appropriately.
- Line 303: The subscripts for (O_2) and (O_3) are incorrect and require revision.
- Table 4: The detection limits reported for Trost et al. seem unusually low—three orders of magnitude smaller than other devices. Please verify these values.
- Table 5: The drift for benzene is reported as (10^{-14}). Such an exceedingly small number raises questions about its significance—please clarify.
Citation: https://doi.org/10.5194/amt-2024-153-RC2 - AC3: 'Reply on RC2', Jeongsoon Lee, 08 Dec 2024
- The manuscript’s writing and structure are often confusing, making it difficult to understand the main arguments and methods. A notable example is the inconsistent definition of the hybrid simulation (HS) method for absorption cross-sections (ACSs). For instance, in Lines 163–166:
-
RC3: 'Comment on amt-2024-153', Anonymous Referee #2, 26 Nov 2024
The manuscript by Trisna and co-authors employs the widely-used active DOAS (Differential Optical Absorption Spectroscopy) technique in the deep UV range of 230–320 nm. The authors developed a custom-built active DOAS system to measure aromatic compounds, including benzaldehyde, formaldehyde, NO₂, p-cresol, m-xylene, phenol, ozone, and others within a measurement cell. The instrument setup was relatively simple, utilizing a low-quality compact spectrometer.
As Anonymous Referee #3 pointed out, since literature absorption cross-sections are typically measured with various types of spectrometers, it is essential to more thoroughly discuss the properties of the spectrometer used in this study.
To begin, I strongly encourage you to incorporate the valuable feedback from both Anonymous Referee #3 and Anonymous Referee #1. Furthermore, I would like to remind you of a few suggestions from the access review process that have only partially been addressed:
Consider splitting the manuscript into two separate papers: One focusing on the method for applying both measured and simulated literature absorption cross-sections, and another dedicated to the actual measured absorption cross-sections. The current manuscript lacks clarity regarding the method combining measured and convolute cross-sections, which is challenging to follow.
A separate analysis of the measured absorption cross-sections could be highly valuable for the scientific community, particularly if it includes more detailed comparisons with existing literature data, where not a lot of published laboratory measurements are currently available. For example, see the plots provided in Serdyuchenko et al. (2014).Further investigate the instrument function on the technical side: This type of instrument generally produces higher-quality spectra. Ensure the slit and/or fibers are well and uniformly illuminated. Additionally, if using a xenon arc lamp in this wavelength range, the instrument may experience notable impacts from stray light within the spectrometer. These effects should be measured and discussed. Refer to comments from Anonymous Referee #3 for additional insights.
The data analysis and explanation of processes and variables remain somewhat challenging to follow. Thank you for adding a list of acronyms following the access review, which has improved clarity (a bit).
Lastly, please address the issue in Figure 6: Applying a convolution should not alter the absorption cross-section itself, as the broadband contribution should be uniform across all instrument line functions. This discrepancy requires further discussion, if not revision. If this data represents fitted concentrations, it does not reflect the absolute absorption cross-section and may be misleading.
https://amt.copernicus.org/articles/7/609/2014/amt-7-609-2014.htmlCitation: https://doi.org/10.5194/amt-2024-153-RC3 - AC2: 'Reply on RC3', Jeongsoon Lee, 08 Dec 2024
Data sets
Preparation of hybrid calibrated absorption cross sections for a compact UV-DOAS measurement (data set) Beni Adi Trisna, Sang Woo Kim, Yong-Doo Kim, Miyeon Park, Seung-Nam Park, and Jeongsoon Lee https://doi.org/10.5281/zenodo.13677342
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 140 | 47 | 7 | 194 | 5 | 4 |
- HTML: 140
- PDF: 47
- XML: 7
- Total: 194
- BibTeX: 5
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1