The most important calibration factor is the use of prescribed reference materials (i.e., NIST-certified SRMs for analyses in the U.S.). In addition, preparation of precise working standards at appropriate concentrations for calibration graphs is crucial. An optimal calibration approach is the preparation of multiple working standards from suspensions of SRMs, with care taken that the particles are deposited onto the analytical filter in an even thin layer. Five or more standards should be used to construct calibration curves. NIOSH methods require the inclusion of several low level calibration standards (20-50 µg). Calibration should be checked each day that samples are analyzed via measurement of laboratory QC samples. The NIST-certified "calibration sets" are intended to be a resource for the preparation of working standards and laboratory QC samples. In studies of PAT laboratory performance, it was found that laboratories that did not use five or more standards, did not include low level standards (<50 µg) or did not check calibration each day that samples were analyzed were found to have poor agreement with other laboratories participating in proficiency testing [47,74]. For directon-filter methods, working standards are prepared by air generation in a chamber [60]. It can be difficult to generate low-level standards (< 50 µg) with precision; it is advisable to use standardized equipment to ensure that deposition of standard materials across the filter matches the deposition of the field samples being analyzed.
Internal Quality Assurance data can be used to track the precision of analyses to demonstrate repeatability of measurements. Data from proficiency testing programs, such as AIHA's Proficiency Analytical Testing (PAT) Program (U.S.) and the Health and Safety Executive's Workplace Analysis Scheme for Proficiency (WASP) Program (U.K.), can be used by the laboratory to demonstrate reproducible results when compared to other laboratories participating in that program. Proficiency testing data should not be used for assessing performance of analytical methods [75].
6. FEASIBILITY OF SILICA ASSESSMENT AT VARIOUS CONCENTRATIONS
The efficacy of sampling and analytical methods for measuring concentrations of hazardous materials may be established using the NIOSH accuracy criterion, requiring better than 25% accuracy at concentrations within the working range of the method [76]. Accuracy, as a percentage of true concentration values, is defined in terms of an interval expected to contain 95% of (future) measurements. To account for method uncertainty, the upper 95%-confidence limit on the accuracy is measured and used in the criterion. Generally, the accuracy of a method is measured over a range of concentrations bracketing the permissible exposure limit (PEL). Use of a range of measurements means that accuracy is assured both at levels below the PEL for possible use in action level determinations and, more significantly, at the PEL itself, where method results must be legally defensible.
NIOSH has evaluated both an XRD silica method (NIOSH Method P&CAM 259, the forerunner to NIOSH Method 7500) and an IR silica method (MSHA Method P-7) in a collaborative test among several laboratories [77]. Experimental conditions and results are summarized in Tables 5 and 6. As indicated, both XRD and IR methods fulfill the NIOSH accuracy criterion over the range of filter loadings measured.
