Laboratory setup to study the polarization properties of size-resolved aerosol particles in exact back-scattering direction

Semwal, Esha; Althausen, Dietrich; Engelmann, Ronny; Hartmann, Markus; Wex, Heike; Oppermann, Thomas; Haarig, Moritz

doi:10.5194/egusphere-2026-4446

Preprints

https://doi.org/10.5194/egusphere-2026-4446

Preprints

17 Aug 2026

| 17 Aug 2026

Status: this preprint is open for discussion and under review for Atmospheric Measurement Techniques (AMT).

Laboratory setup to study the polarization properties of size-resolved aerosol particles in exact back-scattering direction

Esha Semwal, Dietrich Althausen, Ronny Engelmann, Markus Hartmann, Heike Wex, Thomas Oppermann, and Moritz Haarig

Abstract. In this study, a new laboratory-based experimental setup, the Optical Lab for Lidar Applications (OLALA), was developed and characterized to measure the polarization properties of size-resolved aerosol particles in the exact backscattering direction. The linear depolarization ratio was measured at 532 nm wavelength at a scattering angle of 180 ± 0.2° to be comparable to lidar observations. To realize exact backscattering measurements, a 50:50 beam splitter was used to establish a coaxial transmit-receive optical configuration. Aerosols were generated using an atomizer and size-selected with a differential mobility analyzer, which allowed to measure particle size distributions in the fine mode (mostly below 1 μm). Polystyrene latex (PSL) beads, ammonium sulfate, sodium chloride, Arizona Test Dust (ATD), and Saxonian soil dust were used as test aerosols to evaluate the performance of the system. PSL particles of diameter 1 μm exhibited the lowest depolarization ratio of 0.017 ± 0.002, which agrees with theoretical values for spherical particles. Arizona Test Dust, which is characterized by an irregular particle shape, produced the highest depolarization ratio of 0.168 ± 0.048 for particles having a median diameter of 766 nm. For the non-spherical particles like sodium chloride, ATD, and Saxonian soil dust, an increase in the depolarization ratio with increasing size was observed.

Received: 23 Jul 2026 – Discussion started: 17 Aug 2026

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.

Download & links

Esha Semwal, Dietrich Althausen, Ronny Engelmann, Markus Hartmann, Heike Wex, Thomas Oppermann, and Moritz Haarig

Status: open (until 20 Oct 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-4446', Anonymous Referee #1, 29 Sep 2026 reply
The manuscript by Semwal et al. deals with a new “laboratory setup to study the polarization properties of size-resolved aerosol particles in exact back-scattering direction”. The topic is interesting and could be of broad interest to the scientific community. However, in its current form, the manuscript lacks precision and justifications, as the authors do not provide sufficient arguments, justifications and explanations. The motivations of this work in regards to the existing literature remain unclear. No justification is given whether the 180° is observed, and the methodology to evaluate the linear depolarization ratio and its uncertainty are unclear. In turn, the reported values of depolarization ratio and precision appear difficult to consider reliable without further justification, especially as the comparison with the literature is insufficiently discussed. I would therefore encourage the authors to question the validity of their approach and the limitations of their work. The manuscript is sometimes hard to read with too many figures and unuseful quantities like percentiles and surface distribution. I detailed below my major and minor comments and questions. I therefore recommend that these aspects be addressed explicitly in the revised manuscript.
Major comments
While the literature sounds to be clearly quoted, the motivations of this work in regards to the existing literature are not given. As a potential reader, I would like to know the specific motivations of the authors for this work and these are never given. This question sounds important as precise values of depolarization ratios have been published. So, what is the added value of their approach compared to existing ones? Size-selection might be one but when reading the manuscript, the reader understands that the aerosol samples are not size-selected (cf. comment 6), so what are the motivations of the authors? Please specify as potential readers need this information.

The setup itself is questionable, already the aerosol chamber as it is a very small chamber: 1 m long, 1.7 cm inner diameter as confirmed by the picture in Figure 3. It is rather an aerosol flow tube. Aerosol deposition must then be a huge concern preventing measurements especially over 6 or 7 hours as proposed here. Please provide the Allan variance of the concentration measurement over 7 hours. This is important in regards to the use of a lock-in detection over such a large time duration. Also, how are wall effects characterized? How is the flowrate of 1 L/min chosen? The generation process is not either discussed while it may change the morphology of the particles. What about the formation or reorganization of aggregates during atomization and drying? Could the authors provide microscopic images of their samples? These questions are essential because the linear depolarization ratio can be very different.

The optical part of the setup, which is the main novelty, is highly questionable. The reader has no justification and no way to discuss on the ability of this optical set-up to address the 180° angle. The manuscript hence lacks justifications. The only given information is the use of a 50:50 beam splitter. This does not provide any guarantee that back-scattering is reached. You may even have zero depolarization for PSL without being at 180°. What is the reference for the beam splitter? How is it aligned? How is the 180° angle reached and checked? These questions are important for publishing a new setup in AMT. Also, the following statements should be justified: “the diameter of the illuminated volume in the aerosol tube is 2 mm” or the origin of the uncertainty of 0.2° on the 180° angle. Same for the diameter of the second pinhole of 0.5 mm. Please justify how you arrive to these conclusions, as this information is essential for observing back-scattering in laboratory. Stray light is also a major concern, what the reader immediately understands by looking at Figures 1 and 3 where scattering from the green laser is easy to see and dramatically blurs the set-up. The “housing to minimize stray light” then sounds inefficient and stray light sounds too huge to see the back-scattering from aerosols which I suppose to be very small compared to that. So, please explain in detail how you proceeded.

The polarization measurements also need to be justified. For example, the authors proceed exactly as if the 50:50 beam splitter and the mirror they introduce did not modify the light polarization. I doubt it seriously: Fresnel’s formulas tell us the contrary. The modification of the polarization they introduce has then to be quantified to provide precise values of the linear depolarization ratio. The introduction of the halfwave plate for calibration can also introduce biases as the optical path is then changed compared to the actual measurements. As well, what is meant by polarization properties in the title? This is very general while AMT requires to be specific: I suppose you mean the linear depolarization ratio, please use this in this case.

Figure 5 providing the polarization lidar signals must be explained and interpreted. The reader sees really large oscillations but these are not at all commented nor interpreted. This is not in AMT standards as the reader cannot follow your methodology. Please explain the origin of these oscillations and also the origin of the observed huge increase at 15.00 hr which remains elusive. At present, the precision given in Table 2 then sounds incredible (three digits!) and unconceivable. Please clearly detail and explain how the depolarization and uncertainty are retrieved from the raw data in Figure 5. The uncertainty and the limitations of this new setup must then be quantitatively discussed.

The authors often speak about “size-selected” aerosol samples but this domination is false. Since a DMA is used, the reader expects size-resolved measurements but the resulting size distributions shown in Figure 2 are extremely large and are hence not at all size-resolved. The reader hence gets lost. Please correct for this confusing denomination of “size-selected” in the manuscript and in the title, as the reader immediately thinks about size-resolved measurements, which is not the case. Several questions then arise: why using a DMA then if it is to get so large distributions? What are the differences in between samples A to D where all large particles (that contribute most to scattering) are almost the same? The description of the size distributions by its median, quartiles, then surface distribution as ‘descriptors” of the aerosols size populations is also questionable. Please provide the geometric diameter and standard deviation after a log normal fit, performed in linear scale in dN/dLog D. Also, why introducing the surface size distribution then not using it? What is then the goal of Table 1? It seems that the manuscript has been written in a hurry. Please explain or modify.

Figure 4 only shows the particle concentration variation over 2 min. The reader cannot follow how the switching command translates into a concentration cycle, and, ultimately, into an optical signal processed by the lock-in amplifier. Please add to the graph of the particle concentration over 2 min present in Figure 4, the corresponding TTL reference signal and the corresponding raw optical backscattered signals for both polarizations, as usually done in a paper introducing a new lock-in detection. Then, the acquisition spans overs 6 or 7 hours, which raises concerns about drifts over such a large timescale. Due to the transit time of aerosols in the 1 m tube, any drift in the transit time or valve response relative to the fixed 0.05 Hz TTL reference should shift the phase or / and form of the concentration cycle. Is the concentration cycle stable over 7 hours? Please show the Allan variance. Also, the generator and the DMA are prone to drift over so many hours, in size and number concentration. Please discuss and quantify the importance of these drifts on your evaluation of the depolarization.

The comparison with the literature needs to be further discussed. For example, the PSL depolarization ratio and uncertainty are not compatible with zero and this difference has to be explained. It is important as your setup is new and you use PSL to validate your new setup. The depolarization of atd is found comparable to that found by Järvinen et al. (2016) without any comment. This is surprising as both results should not actually compare since Järvinen et al. considered the volume median diameter, while the present manuscript refers to the number median number diameter. Both medians can be very different, especially with so large size distributions. Also, the diameter you get is with the POPS instrument is the PSL-optical diameter. Please discuss and explain this discrepancy with Järvinen et al. (2016). I have a similar comment on the observed discrepancy with Miffre et al. (2016), which is also seldom discussed. In (Tesche et al., AMT, 2016), which you co-authored, the laboratory measurements of atd found in (Miffre et al., 2016) were found to be consistent field measurements of depolarization ratio at 532 nm. I therefore encourage the authors to discuss the limitations of their new setup that could help explain this difference. This is important for future readers.

The comparison with model calculations sounds interesting but the above comments need first to be raised before making such further developments. It then sounds too ambitious for the moment.
Minor comments
The setup is difficult to follow: too many figures, too many schemes and pictures, please gather the information. This is too complex.
Line 54: “the extension of the optical setup to include the wavelengths of 1064 nm and 355 nm is in preparation”. Put that in outlooks, then. I am a bit surprised to recall such simple things.
Caption of Figure 2: The use of the POPS sizer is also questionable as it provides a PSL-optical diameter based on the Mie theory, not applicable to mineral dust then. The authors should at least clearly state this important limitation of their work in the caption of Figure 2.
Line 330 : “Again, ….” : why repeating things ? Please remove to make things clearer.

Reply
Citation: https://doi.org/10.5194/egusphere-2026-4446-RC1
RC2:
'Comment on egusphere-2026-4446', Anonymous Referee #2, 04 Oct 2026 reply
Congratulations to the authors on this interesting work. The new setup addresses a key need of the lidar community, and the experimental results are very promising. However, some aspects of the setup, the size characterization and the uncertainty analysis still need to be addressed. My comments are given below.
Major comments:

The introduction does not clearly show why this setup is needed compared with previous ones. Lines 42–48 read as a list of studies and lack cohesion, and the closely related work of Gómez-Martín et al. (2021) should be discussed. The manuscript is also difficult to follow because Section 2 refers many times to later sections. I suggest describing the setup fully in Section 2 and moving the test samples to Section 3, combined with their results to avoid repetition. The generation of NaCl and ATD aerosols (lines 95 and 101) is vague and repeated, it should be explained once in the methodology part.

Several refractive indices are not given at the working wavelength of 532 nm (PSL at 495 nm, NaCl at 405 nm), and their use is not justified. Note that “495 nm ± 5 nm” is also given as the diameter of the PSL (lines 80 and 151), so this may be a mix-up. Is the solid-phase value for ammonium sulfate valid for particles “dried” from solution? Please give the complex refractive index in all cases, or state that only the real part is known, and give it for the soil dust. PSL are also quasi-monodisperse rather than “uniform and narrow”.

Only one PSL size (1 µm) is shown, without its size distribution. The value 0.017 ± 0.002 differs from the Mie value of 0 (the error bars do not reach zero), so it indicates a systematic offset rather than a “very good agreement”. Are the particles perfectly spherical, or is the setup not capturing the depolarization correctly? Measuring several PSL sizes with the same setup would help to quantify this offset, which is comparable to the values of small ammonium sulfate particles (0.026 ± 0.012).

Wet generation can change the particle properties depending on hygroscopicity, and dust may form aggregates or salt coatings after drying. The dryer in Fig. 1 seems to be a tube with silica beads, which lose efficiency as they saturate. This matters for the long experimental times. How is the relative humidity controlled during an experiment of about 7 h? The generation line is long: how is it cleaned between experiments, and how is the deposition inside the metallic tubes avoided? Was the same concentration used for NaCl and ammonium sulfate? Tests with different concentrations and generation flows for the same aerosol are recommended.

Is the DMA really needed? At these concentrations there is no real size selection. It mainly removes small particles. Fig. 2 shows that the four fractions of ATD have the same supermicron tail and the surface-weighted median diameters are almost identical (1.1–1.2 µm, Table 1). Differences in depolarization cannot be clearly attributed to size. Therefore, “size-resolved”, in the title and abstract, seems too strong. Number distributions also emphasize small particles, which contribute little to the optical signal. Since the surface is related to the cross section, surface-equivalent distributions should be used in the discussion and in Fig. 6, which now shows only number medians.

The parameters used (D25%, Dmed, D75%, IQR) are uncommon and mixed in the discussion. More standard descriptors (e.g. the effective radius and variance of Hansen and Travis) would help. The IQR is a width, not an uncertainty, so it should not be used as an error bar in Fig. 6. What are the size limits of the POPS? Does it provide sizes by analyzing the forward scattering?

Is the illuminated diameter of 2 mm the beam width? How do you control, reduce or account for stray light inside the tube? This is not trivial. A measurement with particle-free air in both valve states would show the detection limit. Have you checked the effects of the half-wave plate on the setup? The beam splitter and mirror M1 are placed before it, so their polarization-dependent transmission and reflection are not included in the Δ90° calibration. These effects should be quantified.

The measurement times are very long, and the calibration is as long as a measurement. How do you deal with small amounts of aerosol in this case? Are the ambient conditions constant or controlled? Why not use a single calibration and a background measurement procedure?

What are the systematic criteria to consider the signal stable, and how are the data subsets selected? Fig. 5 does not show clear stability for ATD. The iterative 2σ clipping and the assumption Neff = N/100 should be justified. The uncertainties seem to be only statistical. The contributions of the calibration constant, the systematic effects mentioned above and the reproducibility between repeated experiments should be included.

In lines 300–304, how do you know that the increase is an effect of shape and not of size? Were the studies mentioned later on the shape of ammonium sulfate made under conditions like yours? Why are no optical microscopy or SEM images of your samples shown? Ammonium sulfate D (0.122) approaches the dust values, which suggests a common cause (e.g. the coarse tail or aggregates), and within the uncertainties the increase with size is not significant for the soil dust. The abstract should be softened accordingly. In Fig. 6, how many experiments were made for each case, and what are the error bars of the red triangles? I recommend showing only the red series of all aerosols in one plot to ease the comparison.

Lines 337–338 (Miffre et al., 2016) report much higher values and a decrease with size, which is opposite to this manuscript, and their error bars clearly separate the two cases. How do you explain this discrepancy? The NaCl values of Sakai et al. (2010) and Järvinen et al. (2016) are also higher. A comparison using consistent size definitions would be very useful.

Lines 359–362 need a better explanation. The cubic model gives values that are too low and the irregular hexahedra values that are too high, so the models only bracket the data, and the “good agreement” in lines 327–328 and 399–401 is not supported. The hypothesis in lines 385–386 can be tested with SEM imaging.

Overall, the main points to solve before moving to new wavelengths are:
Better characterization of the optical setup and refinement of the experimental procedure.

Reconsidering the aerosol generation and the size characterization.

Images of the samples showing their morphology.

Tests with different concentrations and generation flows of the same aerosol.

Better descriptors of the size distributions and a better treatment of uncertainties.

More cohesion in the text and less repetition.

Minor comments:
Lines 27–28: This database collects the results of the CODULAB instrument, so it is that instrument that covers this angular range.

Lines 53–54: Future work should be mentioned only in the Conclusions and future work section.

Line 90: This sentence is redundant, since lines 91–92 are enough, although they should be better explained.

Acronyms should be defined only once (e.g. DMA in lines 86, 117 and 141).

Fig. 2: Please also show the surface- and volume-equivalent size distributions, as these concepts are used later.

Lines 166–167: These lines should be moved to the receiver part.

Lines 185–187: Please rewrite.

Lines 208–211: Please shorten or split.

Lines 203–204, 207 and 294: Please delete.

Lines 289–291: Please delete or rewrite.

Fig. 5: The shaded areas are in some cases shorter or longer than the plot boundaries, and the calibration texts hide part of the plots.

Line 266: The first sentence is not needed.

Lines 282–283 should be moved to lines 266–267.

Table 2: Please display the sizes of groups A, B, C and D to ease reading.

Fig. 6: The caption uses both “Table 2” and “Tab. 1”. Please unify the format throughout the text.

Line 324: A few hundred nanometers of diameter or radius?

Abstract (lines 8–9), lines 395–396 and 408–409: These conclusions are too optimistic. Please soften them.

Reference Volten et al. (2001): There is a typo in the second author’s name (Muñoz).

Reply
Citation: https://doi.org/10.5194/egusphere-2026-4446-RC2

Esha Semwal, Dietrich Althausen, Ronny Engelmann, Markus Hartmann, Heike Wex, Thomas Oppermann, and Moritz Haarig

Viewed

Total article views: 360 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
214	97	49	360	45	36

HTML: 214
PDF: 97
XML: 49
Total: 360
BibTeX: 45
EndNote: 36

Views and downloads (calculated since 17 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	56	19	8	83
Sep 2026	93	53	26	172
Oct 2026	65	25	15	105

Cumulative views and downloads (calculated since 17 Aug 2026)

Month	HTML	PDF	XML	Total
Aug 2026	56	19	8	83
Sep 2026	93	53	26	172
Oct 2026	65	25	15	105

Viewed (geographical distribution)

Total article views: 305 (including HTML, PDF, and XML) Thereof 305 with geography defined and 0 with unknown origin.

Country	#	Views	%

Cited

Latest update: 18 Oct 2026

Short summary

This study presents a new laboratory setup to measure the depolarization ratio of size-resolved aerosol particles in the exact backscattering direction. Measurements with different aerosol types showed that the system can distinguish spherical and non-spherical particles and captures the size dependence of the depolarization ratio. The developed setup is therefore well suited to provide reliable experimental data for improving light-scattering models and the interpretation of lidar observations.


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

Laboratory setup to study the polarization properties of size-resolved aerosol particles in exact back-scattering direction

Viewed

Viewed (geographical distribution)

Cited

1 citations as recorded by crossref.