30 Sep 2021
30 Sep 2021
Cloud Condensation Nuclei (CCN) Activity Analysis of Lowhygroscopicity Aerosols Using the Aerodynamic Aerosol Classiﬁer (AAC)
 ^{1}Department of Chemical and Biomolecular Engineering, University of Maryland, College Park, MD 20742, United States
 ^{2}Department of Chemistry and Biochemistry, University of Maryland, College Park, MD 20742, United States
 ^{1}Department of Chemical and Biomolecular Engineering, University of Maryland, College Park, MD 20742, United States
 ^{2}Department of Chemistry and Biochemistry, University of Maryland, College Park, MD 20742, United States
Abstract. The Aerodynamic Aerosol Classifier (AAC) is a novel instrument that sizeselects aerosol particles based on their mechanical mobility. So far, the application of an AAC for Cloud Condensation Nuclei (CCN) activity analysis of aerosols has yet to be explored. Traditionally, a Differential Mobility Analyzer (DMA) is used for aerosol classification in a CCN experimental setup. A DMA classifies particles based on their electrical mobility. Substituting the DMA with an AAC can eliminate multiple charging artifacts as classification using an AAC does not require particle charging. In this work, we describe an AACbased CCN experimental setup and CCN analysis method. We also discuss and develop equations to quantify the uncertainties associated with aerosol particle sizing. To do so, we extend the AAC transfer function analysis and calculate the measurement uncertainties of the aerodynamic diameter from the resolution of the AAC. The analyses framework has been packaged into a Pythonbased CCN Analysis Tool (PyCAT 1.0) opensource code, which is available on GitHub for public use. Results show that the AAC sizeselects robustly (AAC resolution is 10.1, diffusion losses are minimal and particle transmission is high) at larger aerodynamic diameters (≥∼85 nm). The sizeresolved activation ratio is ideally sigmoidal since no charge corrections are required. Moreover, the uncertainties in the critical particle aerodynamic diameter at a given supersaturation canpropagate through droplet activation and the subsequent uncertainties with respect to the singlehygroscopicity parameter (κ) are reported. For a known aerosol such as sucrose, theκderived from the critical dry aerodynamic diameter can be up to ∼50 % different from the theoretical κ. In this work, we do additional measurements to obtain dynamic shape factor information and convert the sucrose aerodynamic to volume equivalent diameter. The volume equivalent diameter applied to κ Köhler theory improves the agreement between measured and theoretical κ. Given the limitations of the coupled AACCCN experimental setup, this setup is best used for low hygroscopicity aerosol (κ ≤ 0.2) CCN measurements.
 Preprint
(3105 KB) 
Supplement
(383 KB)  BibTeX
 EndNote
Kanishk Gohil and Akua AsaAwuku
Status: closed

RC1: 'Comment on amt2021258', Anonymous Referee #2, 11 Nov 2021
Review of Gohil and AsaAwuku, AMT 2021
GENERAL COMMENTS
This study describes the use of a relatively new system to sizeselect particles known as the Aerodynamic Aerosol Classifier (AAC) manufactured by Cambustion to measure cloud condensation nuclei (CCN) activity of aerosols. This combination of AAC + CCN counter has not yet been characterized in the literature and the study uses a single model compound, sucrose, to verify the theoretical calculations and uncertainties calculated from the transfer function of the AAC. This work offers a nice extension of the work by Moore et al. for the differential mobility analyzer + CCN counter system and should provide a useful tool for future users as the system becomes more popular. As part of the study, a Python package is available through GitHub for the activation diameter and uncertainty calculations. Although I checked that it was available, I did not run the package myself and cannot comment on its capabilities. In addition, I was only able to follow the equations at a high level and have some specific comments listed below.
I see four minor issues associated with this study. The foremost is the use of the aerodynamic diameter to initially calculate the hygroscopicity parameter. The main reason that the diameter is in the original Kohler equation is to determine the number of soluble moles in the particle. It would therefore make sense to present the calculations using the volumeequivalent diameter first, especially since the point of this study is to demonstrate the usefulness of the AAC and to present the results in the best possible light. I could see keeping the discussion about the use of aerodynamic diameter when the particle density and shape factor are unknown since this will be the case for some applications. I realize that this point is entirely stylistic and I leave it to the editor and the authors to determine whether this suggestion should be implemented. Related to this point, the tone of the article would be improved if lines 328330 of the conclusions were reframed to say that it is important to use the volumeequivalent diameter when possible to calculate kappa, since that is more representative of the terms in the Kohler equation, instead of framing it around the aerodynamic diameter.
A more important issue is the relative uncertainties presented in equations 12, 15 and 16. Errors are never subtracted, otherwise a large uncertainty in the terms in the denominator could reduce the overall relative uncertainties, which is unreasonable. There should be an absolute value around the terms, leading to the relative errors being summed (i.e. all the subtraction signs should be changed to addition). Can the authors explain why they are adding the relative errors instead of adding them in quadrature? Once corrected, the new uncertainties should be propagated through the rest of the manuscript in the revised version.
Another concern is that the uncertainty in the calibrated supersaturation of the CCN counter is not included in the overall uncertainty calculations. The authors should include a discussion of the uncertainties in the diameters measured by the DMA used in the calibrations, the uncertainty of the fitted critical diameter (as shown in the Supplement), and their effect on the calibrated supersaturation. This should then be included in the uncertainty in kappa.
A final concern is that the final uncertainty for the measured kappa value presented on line 290 is 0.006. Was this calculated from Equation 16? This value is comparable to the standard deviation of the kappas presented in Table 1 (0.007) and suggests that the repeatability of the measurement, over a range of supersaturations, is worse than the instrumental uncertainties and all the analysis presented. Was this true of repeated measurements at the same supersaturation? Please provide some perspective on this.
SPECIFIC COMMENTS
Equation 13  Can the authors provide some background about the origins of this equation?
Line 226  Dp50 in Equation 16 refers specifically to the volume equivalent diameter. The other diameters should not be used.
TECHNICAL COMMENTS
Fig. 4 caption – “The measurements WERE performed…”
 AC1: 'Reply on RC1', Akua AsaAwuku, 10 Dec 2021

RC2: 'Comment on amt2021258', Anonymous Referee #1, 18 Nov 2021
In this manuscript the authors are characterizing a measurement system combining the Aerodynamic Aerosol Classifier (AAC) and the DMT CCN counter in order to measure the CCN efficiency of size selected aerosol particles. The purpose is to determine the uncertainty of the AAC classification and propagate the error to the kappavalues determined from the 50% activated fraction of aerosol in the CCN counter. The authors also compare the obtained kappavalues to those measured using a conventional DMACCNC system.
The manuscript is well within the scope of AMT, and the measurements appear sound, but the uncertainty analysis could be improved. I have a few comments/questions related mostly to the activated fraction, the sigmoid function, and their impact on the uncertainty of the obtained kappa.
First, the sigmoid curves of the sizeresolved activated fractions measured using the AACCCNC are much wider than those measured with the DMACCNC. This can be seen e.g. by comparing the curve of Fig. 4(b) to the green (?) curve of Fig. S3 (note that there are 7 curves in the latter figure but only 6 rows in Table S2 so it is not completely clear which curve corresponds to which row). What are the reasons for the wider sigmoid? Does it follow directly from wider transfer functions of AAC compared to DMA? Would not a wide sigmoid in itself impact the uncertainty of kappa via making the diameter of 50% activated fraction more uncertain?
Secondly, the fitted sigmoid curves do not appear to reach unity but seem to approach a constant value of something like 0.95. Is this true? (Please provide the sigmoid fitting functions in the supplement.) If it is true, does it mean that about 5% of the particles are lost in the CCN counter? Wouldn’t it then be logical to determine the critical diameter from 50% of the maximum activated fraction of the fitted sigmoid curve and not from 50% of the input aerosol concentration? (This obviously applies to the DMACCNC measurements as well).
Finally, there obviously is some statistical uncertainty in the fitted sigmoid curves, For example, at high activated fractions, the blue datapoints in Fig. 6 are rather scattered and mostly below the sigmoid. Can you determine what is the error of the critical diameter associated with the statistical uncertainty of the fitting function, and how it further impacts the error estimate of the resulting kappa value? Or is the statistical uncertainty perhaps within the error limits caused by the uncertrainties of the measured aerodynamic diameters?
 AC2: 'Reply on RC2', Akua AsaAwuku, 10 Dec 2021
Status: closed

RC1: 'Comment on amt2021258', Anonymous Referee #2, 11 Nov 2021
Review of Gohil and AsaAwuku, AMT 2021
GENERAL COMMENTS
This study describes the use of a relatively new system to sizeselect particles known as the Aerodynamic Aerosol Classifier (AAC) manufactured by Cambustion to measure cloud condensation nuclei (CCN) activity of aerosols. This combination of AAC + CCN counter has not yet been characterized in the literature and the study uses a single model compound, sucrose, to verify the theoretical calculations and uncertainties calculated from the transfer function of the AAC. This work offers a nice extension of the work by Moore et al. for the differential mobility analyzer + CCN counter system and should provide a useful tool for future users as the system becomes more popular. As part of the study, a Python package is available through GitHub for the activation diameter and uncertainty calculations. Although I checked that it was available, I did not run the package myself and cannot comment on its capabilities. In addition, I was only able to follow the equations at a high level and have some specific comments listed below.
I see four minor issues associated with this study. The foremost is the use of the aerodynamic diameter to initially calculate the hygroscopicity parameter. The main reason that the diameter is in the original Kohler equation is to determine the number of soluble moles in the particle. It would therefore make sense to present the calculations using the volumeequivalent diameter first, especially since the point of this study is to demonstrate the usefulness of the AAC and to present the results in the best possible light. I could see keeping the discussion about the use of aerodynamic diameter when the particle density and shape factor are unknown since this will be the case for some applications. I realize that this point is entirely stylistic and I leave it to the editor and the authors to determine whether this suggestion should be implemented. Related to this point, the tone of the article would be improved if lines 328330 of the conclusions were reframed to say that it is important to use the volumeequivalent diameter when possible to calculate kappa, since that is more representative of the terms in the Kohler equation, instead of framing it around the aerodynamic diameter.
A more important issue is the relative uncertainties presented in equations 12, 15 and 16. Errors are never subtracted, otherwise a large uncertainty in the terms in the denominator could reduce the overall relative uncertainties, which is unreasonable. There should be an absolute value around the terms, leading to the relative errors being summed (i.e. all the subtraction signs should be changed to addition). Can the authors explain why they are adding the relative errors instead of adding them in quadrature? Once corrected, the new uncertainties should be propagated through the rest of the manuscript in the revised version.
Another concern is that the uncertainty in the calibrated supersaturation of the CCN counter is not included in the overall uncertainty calculations. The authors should include a discussion of the uncertainties in the diameters measured by the DMA used in the calibrations, the uncertainty of the fitted critical diameter (as shown in the Supplement), and their effect on the calibrated supersaturation. This should then be included in the uncertainty in kappa.
A final concern is that the final uncertainty for the measured kappa value presented on line 290 is 0.006. Was this calculated from Equation 16? This value is comparable to the standard deviation of the kappas presented in Table 1 (0.007) and suggests that the repeatability of the measurement, over a range of supersaturations, is worse than the instrumental uncertainties and all the analysis presented. Was this true of repeated measurements at the same supersaturation? Please provide some perspective on this.
SPECIFIC COMMENTS
Equation 13  Can the authors provide some background about the origins of this equation?
Line 226  Dp50 in Equation 16 refers specifically to the volume equivalent diameter. The other diameters should not be used.
TECHNICAL COMMENTS
Fig. 4 caption – “The measurements WERE performed…”
 AC1: 'Reply on RC1', Akua AsaAwuku, 10 Dec 2021

RC2: 'Comment on amt2021258', Anonymous Referee #1, 18 Nov 2021
In this manuscript the authors are characterizing a measurement system combining the Aerodynamic Aerosol Classifier (AAC) and the DMT CCN counter in order to measure the CCN efficiency of size selected aerosol particles. The purpose is to determine the uncertainty of the AAC classification and propagate the error to the kappavalues determined from the 50% activated fraction of aerosol in the CCN counter. The authors also compare the obtained kappavalues to those measured using a conventional DMACCNC system.
The manuscript is well within the scope of AMT, and the measurements appear sound, but the uncertainty analysis could be improved. I have a few comments/questions related mostly to the activated fraction, the sigmoid function, and their impact on the uncertainty of the obtained kappa.
First, the sigmoid curves of the sizeresolved activated fractions measured using the AACCCNC are much wider than those measured with the DMACCNC. This can be seen e.g. by comparing the curve of Fig. 4(b) to the green (?) curve of Fig. S3 (note that there are 7 curves in the latter figure but only 6 rows in Table S2 so it is not completely clear which curve corresponds to which row). What are the reasons for the wider sigmoid? Does it follow directly from wider transfer functions of AAC compared to DMA? Would not a wide sigmoid in itself impact the uncertainty of kappa via making the diameter of 50% activated fraction more uncertain?
Secondly, the fitted sigmoid curves do not appear to reach unity but seem to approach a constant value of something like 0.95. Is this true? (Please provide the sigmoid fitting functions in the supplement.) If it is true, does it mean that about 5% of the particles are lost in the CCN counter? Wouldn’t it then be logical to determine the critical diameter from 50% of the maximum activated fraction of the fitted sigmoid curve and not from 50% of the input aerosol concentration? (This obviously applies to the DMACCNC measurements as well).
Finally, there obviously is some statistical uncertainty in the fitted sigmoid curves, For example, at high activated fractions, the blue datapoints in Fig. 6 are rather scattered and mostly below the sigmoid. Can you determine what is the error of the critical diameter associated with the statistical uncertainty of the fitting function, and how it further impacts the error estimate of the resulting kappa value? Or is the statistical uncertainty perhaps within the error limits caused by the uncertrainties of the measured aerodynamic diameters?
 AC2: 'Reply on RC2', Akua AsaAwuku, 10 Dec 2021
Kanishk Gohil and Akua AsaAwuku
Kanishk Gohil and Akua AsaAwuku
Viewed
HTML  XML  Total  Supplement  BibTeX  EndNote  

522  102  11  635  29  3  6 
 HTML: 522
 PDF: 102
 XML: 11
 Total: 635
 Supplement: 29
 BibTeX: 3
 EndNote: 6
Viewed (geographical distribution)
Country  #  Views  % 

Total:  0 
HTML:  0 
PDF:  0 
XML:  0 
 1