Envlron. sci. Technol. 1984, 18, 827-833
Fogwater Collector Design and Characterization Daniel J. Jacob, Rueen-Fang T. Wang, and Richard C. Flagan" Environmental Engineering Science, Keck Engineering Laboratories, California Instltute of Technology, Pasadena, Callfornia 9 1125
The detailed characterization of a rotating arm collector to sample ambient fog droplets for chemical analysis is presented. Because of the large volume of sample required, and because fog droplets are of supermicron size and are sensitive to local thermodynamic disturbances, conventional methods for atmospheric aerosol sampling are not suitable for fogwater sampling. Design criteria for fogwater samplers are outlined. Devices used in previous investigations are evaluated in light of these criteria. The design of a rotating arm collector is discussed, and it is shown that this instrument performs adequately in preserving the physical and chemical integrity of the sample at all stages of collection. Limitations in the design due to mechanical constraints are discussed. Results of an in situ calibration experiment using a chemically tagged monodisperse aerosol indicate a size cut of 20-pm diameter.
Introduction Supermicron particles contribute significantly to the total mass of a dry aerosol (1). This contribution increases considerably when the aerosol is wetted, especially under supersaturated conditions in which activated condensation nuclei grow rapidly to form cloud or fog droplets. Whereas the mass loading of an urban aerosol under nonsaturated conditions is of the order of lo4 g m-3, the liquid water content in a cloud or fog ranges from 0.01 to 1g m-3, with supermicron droplets constituting the bulk of the aerosol mass. In urban environments fogwater has been found to contain extremely high pollutant concentrations, often associated with high acidities (2,3). Supermicron particles are difficult to collect efficiently because their inertia may prevent them from following the air streamlines converging toward the inlet of the sampler (4). Furthermore, water droplets in the atmosphere are in a fragile thermodynamic balance with the ambient humidity which is very sensitive to perturbations by a sampling device. Conventional methods for collecting samples of total particulate matter may, therefore, lead to sampling biases due to anisokinetic sampling conditions or changes in temperature or pressure. Fog sampling is further complicated by the relatively large sample volume required for the study of the detailed aquatic chemistry. At least 10 mL of sample is needed for the standard inorganic analysis routinely carried out in our fog program (3). If the variation in chemical composition throughout the fog event is to be studied, sampling intervals should be short. For a typical liquid water content of 0.1 g m-3, a sampling rate of over 1.7 m3 min-l is required to collect 10 mL for analysis in an hour. Larger sampling rates are required for the study of light fogs. A rotating arm collector based on the principle of inertial impaction has been developed and used in our intensive fog sampling program. Major advances in the understanding of fog chemistry and the role of fogs in acid deposition have been made by using this device (3, 5). In a recent field intercomparison study of fogwater collectors (6),samples were collected simultaneously with instruments from five different research groups and analyzed for major ions. Ionic concentrations in samples collected by the rotating arm collector and a jet impactor (7)agreed 0013-936X/84/0918-0827$01.50/0
within 5%; other collectors gave systematically either higher or lower concentrations. The rotating arm collector was found to collect water efficiently in both light and heavy fogs. To date, the sampling characteristics of the rotating arm collector have only been qualitatively explored by measuring the change in the droplet size distribution in a cloud chamber which results from its operation. These measurements indicated that the minimum size of particles collected was at least 8 pm but did not provide sufficient resolution to determine the size-dependent collection efficiency (8). In this paper we fiist elaborate on the design criteria relevant to fogwater collection and then present a detailed examination of the design and operation of our rotating arm collector. Constraints on the design due to power requirements and possible sample biases due to aerodynamic heating are explored. Measurements of the collection efficiency as a function of particle size are presented.
Design Criteria for Fogwater Collectors Size Cut. Fog droplets form by activation of atmospheric particles (condensation nuclei) under supersaturated conditions. At the levels of supersaturation found in the atmosphere, the lower size limit for particles to be activated is of the order of 0.1 pm (9). Figure 1 shows how fog formation can shift the size distribution of an urban aerosol; particles in the first mode (below 0.1 pm) are rather unaffected by the condensation process, but most particles in the two higher modes grow by condensation to much larger sizes. Therefore, two types of particles coexist in a fog: (1) supermicron fog droplets and (2) nonactivated, primarily submicron, particles. Being dilute aqueous solutions, fog droplets do not interact with their environment in the same way as the solid or concentrated submicron particles (10,11). I t is, therefore, important that a fogwater sampler differentiate between the two types of particles. Fog droplets range in size from 1to 100 pm, with a mass median diameter usually in the range 10-40 pm (12-15). The dependence of the fog droplet chemical composition on droplet size has not been rigorously investigated to date; general predictions from droplet growth theory (16)are difficult to make because humidities in fogs fluctuate rapidly in a manner that is still poorly understood (17,18). Large droplets are not necessarily more dilute than smaller droplets because they generally result from condensation on larger nuclei. If the total pollutant burden associated with fogs is to be determined, droplets of all sizes should be collected with the same efficiency. Nonactivated submicron particles represent a very small fraction of the total aerosol mass but they could, if collected, contribute a sizable amount of solutes to the sample and result in a serious bias. A sharp lower size cut in the range 1-1O-pm diameter is, therefore, desired. Furthermore, since most of the fog mass is associated with large droplets, droplets up to about 100-pm diameter must be collected without bias. Three methods are available to collect the large particles while excluding the smaller particles: sedimentation, inertial separation, and removal of smaller particles by
0 1984 American Chemical Society
Envlron. Sci. Technol., Vol. 18,
No. 11, 1984 827
Table I. Fogwater Collectors Reported in the Literature
reference passive Mrose (20) Okita (21) Lazrus et al. (22) Sadasivan (23) Falconer and Falconer (24) active Houghton and Radford (12) May (25) May (25) Okita (21) Mack and Pilie (26) Katz (7) Brewer et al. (27) this paper
type
impaction velocity, cm 8-l
sampling rate, m3 min-'
cloth surface grid screen screen grid
ambient wind ambient wind ambient wind ambient wind ambient wind
variable variable variable variable variable
screen
600
102
grid jet impactor screen rotating arm
450 1700 94 1500-5000
11 0.05 1 7
jet impactor screen rotating arm
2000 320 3800-5600
1.2 1.5 5
characteristic width" inlet Stokes inlet, impaction no. for 100-wm cm surface, cm droplets
inlet Stokes no. for 100-wm droplets
? 0.02
variable variable variable variable variable
30
0.01
3c
0.6
10 0.85 7.5
0.05 0.4 ? 0.45
7c ? ?
1.4 13 0.3
0.2 0.026 0.48
5d
?
5c 20
2.0
?f
0.01 ?
? 5
?
"Characteristic width of inlet: radius of circular inlet (12,21,25b,27), half-width of square inlet (25a). Characteristic width of impaction surfaces: radius of wires ( 1 2 , 2 l , 24,25a, 27), half-width of jet (7,25b),radius of rod (26, this paper). "Diameter of droplets collected with 50% efficiency. Calculated from impaction theory for cylinders (19). Calculated by Katz from theory and confirmed by experiment. e Obtained by personal communication from R. L. Brewer. f ? = not reported, or cannot be computed from available data.
-4
6ot-
!
D (pm)
Figure 1. (Top) Typical urban aerosol slze distributlon proflle (7). (Bottom) Expected shift In the slze distribution profile as a result of fog formation.
diffusion. Of these, inertial separation, particularly impaction, most 'eadily achieves sharp size cuts of a few microns and hL 8 been the method favored by past investigators. The
