Impact of muffle furnace preparation on the results of crystalline silica analysis.

A prior report demonstrated an unacceptably low level of accuracy in silica analytical testing, with a general negative bias (i.e., underreporting) although other inaccuracies included false-positive results when analyzing blank filters. The possible bias may have been due to the loss of sample during shipping and or sample preparation. We report on a follow-up study that was designed to mimic the original study, but in which sources of variability were evaluated. We found no effect on silica recoveries due to shipping and confirmed the prior study results that the muffle furnace ashing process led to low overall recoveries (49-104%), depending on the adherence to the recommended preparation method. Plasma ashing recoveries ranged from 89 to 108%. Our results suggest that muffle-furnace ashing using a crucible should be restricted. More broadly, however, muffle-furnace ashing is only one source of analytical error that contributes to the relatively poor overall performance revealed by Cox et al. Whatever the case, OSHA should ensure that its proposed requirements to improve laboratory performance will actually lead to the discovery and correction of all major sources of error by participating laboratories. This is particularly important in light of OSHA's proposed reduction in the PEL and action level proposed by OSHA.

How reliable are crystalline silica dust concentration measurements?

To determine how reliably commercial laboratories measure crystalline silica concentrations corresponding to OSHA's proposed limits, 105 filters were prepared with known masses of 20, 40, and 80 µg of respirable quartz corresponding to airborne silica concentrations of 25, 50, and 100 µg/m(3) and were submitted, in a blind test, to qualified commercial laboratories over a nine month period. Under these test conditions, the reported results indicated a lack of accuracy and precision needed to reliably inform regulatory compliance decisions. This was true even for filters containing only silica, without an interfering matrix. For 36 filters loaded with 20 or more micrograms of silica, the laboratories reported non-detected levels of silica. Inter-laboratory variability in this performance test program was so high that the reported results could not be used to reliably discriminate among filters prepared to reflect 8-h exposures to respirable quartz concentrations of 25, 50 and 100 µg/m(3). Moreover, even in intra-laboratory performance, there was so much variability in the reported results that 2-fold variations in exposure concentrations could not be reliably distinguished. Part of the variability and underreporting may result from the sample preparation process. The results of this study suggest that current laboratory methods and practices cannot necessarily be depended on, with high confidence, to support proposed regulatory standards with reliable data.