laboratories using an aerodynamic sizing instrument to test the same cyclone agreed within a range of 11% [37]. Further work on the testing protocol is needed to improve interlaboratory agreement. Many current classifiers do not match the shape of the respirable convention exactly and produce biases that depend on size distribution. Two comparisons of several respirable samplers have been performed using the aerodynamic sizing technique [38,39]. The introduction of the thoracic fraction in the ACGIH/ISO conventions has spurred interest in thoracic classifiers for certain types of aerosols, e.g., cotton dust, asbestos and sulfuric acid [19]. The performance characteristics of the vertical elutriator (operated at 7.4 L/min) used for cotton dust approximately meets the thoracic definition [40]. Laboratories in many countries perform asbestos fiber measurement using the technique of counting only fibers with diameters of 3 :m or less; this size selection was shown to be approximately equivalent to thoracic sampling [41]. Further tests indicate that several thoracic samplers may be appropriate for asbestos sampling [42,43]. Thoracic sampling is also recommended for sulfuric acid [44] and metal working fluids [45]. Several samplers based on inertial, cyclone and foam separators have been specifically developed to meet the thoracic definition [46-49]. The CIP-10 sampler has been used for thoracic sampling in Europe [50], but is not applicable to aerosols with a significant submicrometer fraction [46]. Several of these samplers have been tested to compare with the thoracic convention [42,43]. The GK2.69 cyclone (Figure 2i) has been used for metal working fluids [45]. The other samplers referenced are in the development stage and need further testing. The PM-10 standard for environmental sampling is very similar to the thoracic convention and impactors with a 10 :m cutoff size have been used for personal PM-10 sampling [51]. A cascade impactor, e.g., the Andersen personal cascade impactor (Figure 2j), can be used to calculate the thoracic fraction of an aerosol. Although a thoracic sampler is commercially available (Figure 2i), further work is needed to determine its applicability for specific types of aerosol. For example, a thoracic sampler for fibers must result in a uniform deposit of the particles on the filter for accurate analysis results. The overall accuracy of a classifier with respect to sampling in accordance with the one of the sampling conventions can be estimated using a bias map (Figure 3). The bias map displays the percent difference between the predicted mass collected by the sampler and the mass expected according to the convention as a function of the parameters of a lognormal particle diameter distribution for a range of likely workplace distributions. Such a bias map can be used for selecting a sampler for a workplace having a certain range of particle sizes or for developing samplers that agree more closely with the sampling conventions. The bias map in Figure 3 was created by: (a) fitting the penetration curve for the 10-mm nylon cyclone (Figure 2g [35]) at 1.7 L/min with a lognormal curve (a logistic curve also can be used), (b) calculating the bias between the respirable convention and the curve from the previous step for a range of lognormal size distributions, and (c) plotting the bias contour lines as a function of the size distribution mass median aerodynamic diameter (MMAD) and geometric standard deviation (GSD). The 10-mm nylon cyclone shows significant negative biases, especially at large MMADs and small GSDs because the cyclone penetration curve drops off more rapidly with size than the curve for the respirable convention. The “best” flow rate to use in a workplace when sampling according to one of the conventions becomes a matter of judgment, depending on the size distribution typically encountered in that workplace. A cyclone that fits
3/15/03
188
NIOSH Manual of Analytical Methods
