Stage Response Calibration of the Mark III and Marple Personal Cascade Impactors

"Department of Mechanical Engineering, University of Minnesota, Minneapolis, MNExperimental and correlated stage responses (the fraction of particles entering an impactor that are collected on a stage) are presented for the Andersen Mark III and Marple personal cascade impactors. The impactors were operated upright and fully assembled so that interstage interference and wall losses could be properly studied. The observed stage responses showed maxima that fell significantly short of unity, meaning that a monodisperse aerosol is never collected exclusuvely on one stage, but is distributed among several stages and internal losses. Correlations for the stage responses are presented so that the experimental results can be used to determine size distributions with available data-inversion algorithms. Simulations with log-normal distributions show significant differences between dpa50 histograms and the more accurate distributions that result by taking the response functions into account.Cascade impactors are widely used for the measurement of aerosol size distributions; their ruggedness and simplicity of operation make them an attractive choice for a variety of environments. In one common application, the aerosol of interest is drawn through the impactor, and the amount deposited on each stage is determined gravimetrically. The primary objective is to extract as much information as possible about the initial size distribution from the resulting set of weights. Unfortunately, the reconstruction of contin-uous size distributions from finite data sets is an ill-posed problem. Not surprisingly, a variety of such reconstruction techniques are presently in use; the success of each method depends on the particular nature of the aerosol sample and the instrument response.Inversion of impactor data by any of these techniques requires that the response functions for each stage be known. The stage response function gives the fraction of particles entering the impactor that deposits on that stage. Values of the response function range from zero to one, and depend on impactor operating conditions (e.g., geometry, flow rate, and gas properties) and particle properties (e.g., diameter, density, and shape factor). As interstage interferences and particle losses will affect each stage's response, experimental determination of stage responses should be made for fully assembled impactors. Unfortunately, impactor stage response functions are rarely reported; rather, single-stage efficiency curves are given that provide an incomplete description of impactor performance.This paper presents experimental stage responses for two fully assembled cascade impactors: the Andersen Mark III and Marple personal impactors (model 298, Andersen Samplers, Inc., Atlanta, Ga.). These calibrations were performed with monodisperse liquid droplets for both impactors and also with monodisperse solid spheres for the Marple personal impactor. The calibration results for the Mark 111 impactor presented here have been previously reported in the response-function format (Rader et al., 1989). The calibration results for the Marple personal impactor, however, were originally presented in the single-stage efficiency for¬mat (Rubow et al., 1987) and have been reanalyzed in this work. Losses within both impactors are considered, and their treat¬ment within data inversion techniques is dis¬cussed. In addition, correlations for each impactor's response functions are presented so that these calibration results can be easily included in data-inversion algorithms. Size distribution inversions using these response correlations are compared with the standard cut-point method for two lognormal test dis¬tributions."