Group 5. Local hygroscopic closure (Swietlicki)

1. Introduction

The Working Group involved people dealing with hygroscopic behaviour of aerosol particles within ACE-1. The local hygroscopic closure study at water vapour subsaturation is closely linked to the Working Groups on Mass Closure, Scattering Closure and CCN Closure. Understanding the hygroscopic behaviour of the aerosol particles is vital for the assessment of both the direct and indirect effects of aerosols on global climate.

2. Objective

The objective of the hygroscopic closure study at water vapour sub-saturation is to determine whether the observed hygroscopic behaviour can be derived from the size-resolved chemical composition data using an appropriate model on hygroscopic growth.

It was suggested by the WG that all studies within ACE-1 dealing with hygroscopic growth at water vapour sub-saturation should have at least one hygroscopic growth model in common. The suggested model is based on empirical data on the relationship between water activities and molalities for a number of relevant substances and is described in more detail below.

3. Properties affecting hygroscopic growth

The following properties affect the hygroscopic growth (at a given relative humidity; RH):

  1. Composition of the hygroscopically active particle fraction
  2. Water-soluble (active) volume fraction on the individual particles
  3. Number fraction of particles in each externally mixed hygroscopic group (more- and less- hygroscopic groups, sea spray group)
  4. Surface tension
  5. Shape factor of the dry particles

All properties should be known as a function of size. The Kelvin curvature correction factor should be included in the calculations for Dp<ca 200 nm. In the following, it will be assumed that the dry particles are spheres (unity shape factor) and that the surface tension is that of pure water. The validity of these assumptions can be estimated with a sensitivity analysis.

4. Data used for local hygroscopic closure studies at water vapour sub-saturation

The data marked with a "greater than sign" (>) are the data which will actually be used in the following hygroscopic closure studies.

  1. Hygroscopic growth of individual particles
  2. Aerosol Size Distribution
  3. Size-resolved chemical composition
  4. Relative humidity at which the impactor was operating

5. Data availability

Data Platform PI Status
H-TDMA Discoverer Swietlicki In CODIAC
DMPS Discoverer Covert/Quinn In CODIAC
Impactor Discoverer Quinn Preliminary (major ions)
Impactor Discoverer Sievering Leg-1 ready (major ions)
RH Discoverer Quinn Available
H-TDMA Cape Grim Covert Not available
DMPS Cape Grim Wiedensohler In CODIAC
Impactor Cape Grim Cainey In CODIAC
Impactor Cape Grim Sievering In CODIAC
RH Cape Grim ? ?

6. Hygroscopic growth model for water vapour sub-saturation conditions

Empirical data on molalities for single electrolytes (salts and acids) as a function of water activity are available from several published measurements with electrodynamic balances. (Molality: moles of solute per kg of solvent (water); Water activity aw: relative humidity (in %) divided by 100*the Kelvin correction factor). A list of which water activity data should be used will be given later.

Water activities for multiple electrolytes can be obtained from mixing rules. The mixing rule recommended by the WG is the ZSR (Zdanovskii, Stokes, Robinson) mixing rule which states that the sum (taken over all single electrolytes i) of the ratios between the molality of electrolyte i in the mixed aqueous solution and the molality of the same electrolyte in the single-electrolyte solution (for a given water activity) should equal one (1). This rule combined with information regarding the molar fractions of the various salts (or acids) obtained with the cascade impactors makes it possible to calculate the molalities for a given water activity. Knowing the particle dry size (i.e. the moles of the various ions), the amount of water absorbed by the particle and thus the resulting wet size can be calculated.

This last step is dependent on the density of the aqueous solution. Density data for supersaturated solutions are scarce, but can be found e.g. in Tang and Munkelwitz JGR99(1994)18801-18808. The density of the aerosol particles at the RH at which the cascade impactors operated must also be know to convert from aerodynamic diameter (in impactor) to mobility diameter (in DMA). A list of which density data should be used will be given later.

The impactor data give the moles of the various cations and anions found in the (dilute) aqueous solution. However, the data on water activity is given for various salts (and acids) e.g. ammonium sulphate, not for the ions themselves. There was some discussion within the WG on how to determine which solid molecular forms should be used in the ZSR mixing rule. It was suggested that a thermodynamic equilibrium model could be used for this purpose. Such models are based on the assumption that equilibrium between all phases (gas, liquid and solid) exists. This means that they do not recognise the presence of supersaturated salt solutions for suspended particles, since a solid phase is postulated to form as soon as the water activity drops below the mutual point of deliquescence. There is some evidence from H-TDMA observations on the Discoverer that supersaturated particles can be found in a marine background environment. Using the water activity data for single electrolytes combined with the ZSR mixing rule allows for the presence of such supersaturated particles in calculations of hygroscopic growth. The actual thermodynamic equilibrium model to be used will be determined later. A sensitivity analysis should be performed to determine whether the choice of solid molecular forms is significant for the calculation of solution molalities.

The final step in the calculations of hygroscopic growth is to compare the modelled growth with the observed growth to give the soluble volume fraction for each individual particle.

7. Questions to be addressed by the local hygroscopic closure study

The starting hypothesis for the local hygroscopic closure study at water vapour sub-saturation conditions is that only the inorganic ions (and MSA) are needed to account for the hygroscopic growth. It is thus assumed that any organic material (other than MSA) is hygroscopically inactive and that no surface active material is present. The question to be answered is then:

  1. Can the inorganic (+ MSA) particle fraction fully explain the hygroscopic behaviour of the sub-micrometer aerosol particles?

    Another proposed study addresses the state of mixture of sea spray particles:

  2. To what extent are the sea spray derived ions externally mixed?

7.1 Local (ionic) hygroscopic closure study

The first question can be addressed in two steps (a) and (b).

(a) Can the integrated particle volumes obtained from the DMPS sub-micrometer aerosol particle size distribution data account for the inorganic (+ MSA) particle masses collected with the cascade impactors?

This is essentially a mass closure since it does not explicitly involve a model of hygroscopic growth. Nevertheless, a slight correction for hygroscopic growth needs to be included in order to correct the impactor cut-off sizes from humid to dry conditions. The H-TDMA on the Discoverer measured hygroscopic growth also at 50 % RH (on the upper side of the RH-hysteresis loop), which was the RH at which the impactors were meant to be operated. However, excessive heating of the inlet resulted in a somewhat lower impactor RH (ca. 35-45% RH). The H-TDMA growth data at 50 % RH need therefore be slightly corrected (decreased) based on the assumption that no crystallisation occurred. The diameter growth factors at 50 % RH were normally between 1.1 and 1.2 (for externally mixed sea spray group: 1.4-1.5). This study also assumes that the soluble volume fraction is unity.

The integration of DMPS integrated particle volume can be compared with the cascade impactor particle ion mass assuming a (dry particle) density. Closure is obtained if the resulting density matches the estimated (dry particle) density within the estimated errors.

(b) Can the number of major ions observed with the impactor be derived from observations of hygroscopic growth and the particle size distribution using a model of hygroscopic growth?

This closure study proposes to focus on the number of major ions. The reasons for this are: (1) Size-resolved gravimetric and organic data is largely lacking which means that soluble volume fractions can not be determined from impactor data and; (2) The number of ions in solution is a property which is directly related to hygroscopic growth for ideal aqueous solution obeying Raoult's law. It is a hygroscopic closure study since it explicitly makes use of a model of hygroscopic growth.

The hygroscopic closure study can be performed as follows:

  1. Estimate the density of the particles at the RH at which the impactors were operating (to convert cut-off sizes from aerodynamic to mobility diameter).
  2. Estimate the chemical (ionic) composition for each H-TDMA dry size from the impactor data.
  3. Calculate the growth at 90 % RH (using the ZSR mixing rule).
  4. Compare calculated growth with observed to derive the soluble volume fraction.
  5. Convert the mobility diameter impactor cut-off sizes at humid (impactor) conditions to dry diameters, taking into account the soluble volume fractions of the externally mixed aerosol (more- and less-hygroscopic groups, sea spray group) as observed by the H-TDMA.
  6. For each impactor interval, integrate the dry DMPS spectrum within the size ranges calculated in the previous step to get an active (ionic) and an inactive volume, taking into account the number fraction of particles in each hygroscopic group as observed by the H-TDMA.
  7. Estimate the density of the active (dry) volume to calculate the ionic mass and thus the number of major ions on each impactor stage.
  8. Compare the actual number of ions on each impactor stage with that derived from DMPS/H-TDMA data.

Closure is obtained if the observed and calculated number of ions agree within the estimated errors. It should be noted that the model of hygroscopic growth is used to calculate the soluble volume fractions as well as the corrections of impactor cut-off sizes from humid to dry conditions.

7.2 Extent of external/internal mixture for the sea spray derived ions

The extent of external/internal mixture for the sea spray derived ions is important if one wishes to parameterize the relationship between the aerosol light scattering properties and the mass concentrations of sea spray components, since externally mixed sea spray particles have a much larger hygroscopic growth than particles for which a large fraction of the mass is due to gas-to-particle conversion processes. This also means that externally mixed particles are more active as CCN than if the sea spray derived ions are internally mixed with nss-sulphate. Some insight of the transformation rates for sea spray particles can be obtained by studying to what extent the sea spray derived ions are externally mixed. This is done by calculating to what extent the externally mixed sea spray particles (observed with the H-TDMA) can account for the mass of sea spray derived ions (especially sodium and chlorine) collected in the impactors.

The sea spray mixture closure study can be performed as follows:

  1. Estimate the density of the externally mixed sea spray particles at the RH at which the impactors were operating (to convert cut-off sizes from aerodynamic to mobility diameter).
  2. Assume that particles with growth factors larger than those of the more- hygroscopic particle group are externally mixed sea spray particles.
  3. Combine DMPS/H-TDMA data to calculate the (dry) number size distribution of externally mixed sea spray particles. Compare with the lower end of the DMPS/APS "sea spray" mode.
  4. Assume that the externally mixed sea spray particles have average sea spray composition (no surface film enrichment?)
  5. For the externally mixed sea spray particles, calculate the growth at 90 % RH using the ZSR mixing rule.
  6. For these particles, compare calculated growth with observed growth to derive the soluble volume fraction.
  7. Convert the mobility diameter impactor cut-off sizes at humid (impactor) conditions to dry diameters, taking into account the soluble volume fraction of the externally mixed sea spray particles as observed by the H-TDMA.
  8. For each impactor interval, integrate the dry DMPS spectrum within the size ranges calculated in the previous step to get a sea spray derived ionic volume, taking into account the number fraction of externally mixed sea spray particles as observed by the H-TDMA.
  9. Estimate the (dry) density of this volume to calculate the number of sea spray derived ions.
  10. Compare the actual number of sea spray derived ions (e.g. Na+) collected on each impactor stage with that derived from DMPS/H-TDMA data.

The result of these calculations is the fraction of sea spray derived ions present in an external mixture. This is not strictly a closure study since there is no independently determined value with which to compare. There is only an upper bound in the sense that the integrated DMPS/H-TDMA data should not result in more sea spray derived ions than was observed in the impactor. The fraction of externally mixed sea spray derived ions should be related to meteorological conditions (e.g. wind speed, height and cloudiness along the air mass back trajectory ending at sea level) in order to obtain a parametrisation of the sea spray particle transformation rate as a function of particle size.

8. Time schedule

The hygroscopic closure study will not be ready for the 1 June 1997 dead-line. The main reason for this is that some essential data are still lacking or have only recently been made available. A paper describing the observations of particle hygroscopic behaviour made with the H-TDMA instruments on the Discoverer and hopefully also at Cape Grim will be ready by 1 June 1997, in time for the JGR special ACE-1 issue.

Recommendations of which water activity data, density data and thermodynamic equilibrium model to be used for the calculations will be given by September 1997. The paper on hygroscopic closure should be ready in December 1997.