i. Quality Assurance Quality assurance (QA) procedures followed at NIOSH and NIOSH contract laboratories have included repeat analyses, analysis of OC standards, and analysis of media and field blanks. Results for repeat analyses of over 200 filter samples at a contract laboratory (DataChem Laboratories, Inc., Salt Lake City, UT) gave a pooled RSD (95% confidence) of about 4% for EC and TC, and 5% for OC [68]. The samples were collected in mines, and some contained other carbonaceous matter such as limestone. In the analysis of OC standards, recovery was typically within 6% of the expected value [68]. For example, the mean recovery for more than 400 sucrose standards analyzed at DataChem over a one-year period was 99.97 % "6.07%. Nearly identical results (100.33 % " 5.15%, n = 462) were obtained by Clayton Laboratory Services (Novi, MI). Different sets of field blanks (41 at DataChem and 129 at Clayton) analyzed by the two laboratories also gave comparable results [68]. Mean blank results (:g carbon per filter) were as follows: DataChem OC = 9.96 " 7.05, Clayton OC = 12.80 "5.76; DataChem EC = -0.54 "1.89, Clayton EC = -0.08 "2.14; DataChem TC = 9.26 "7.30, Clayton TC = 12.70 "6.18. In lieu of an OC-EC reference material, a limited confirmation of results by a second laboratory is advisable. Our laboratory (NIOSH, Cincinnati, OH) has routinely performed repeat analyses on samples previously analyzed at a contract laboratory. As an added check, subsets of representative field samples also have been analyzed by a third laboratory. In the analysis of fifty samples (analyzed at DataChem, NIOSH, and Clayton) from mines, the pooled RSD (95% confidence) was about 6% for TC, 10% for EC, and 12% for OC [68]. These results are consistent with those found in a previous round robin [35]. EPA. The EPA recently conducted a special study [75] as part of the QA oversight for the PM2.5 Speciation Trends Network (STN). Samples collected as part of this network were removed from refrigerated storage eight months after they were analyzed at Research Triangle Institute (RTI, Research Triangle Park, NC). The samples were then shipped to the EPA New England Regional Laboratory (NERL, Lexington, MA) for re-analysis. Good interlaboratory agreement was reported. The acceptance criteria for the archived samples were based on filter loading (:g/cm2 ). The three acceptance criteria are the following: a difference # 1 :g/cm2 for loadings less than 5 :g/cm2 , a relative percent difference (RPD) # 20% for loadings from 5 to 10 :g/cm2 , and an RPD # 15% for loadings above 10 :g/cm2 . The OC, EC, and TC data had 85%, 96%, and 93% of the inter-laboratory results, respectively, within the acceptance criteria. The lower percentages for OC and TC are thought to reflect contamination of lightly loaded samples from adsorbed vapors and handling. An OC blank (passive) correction was applied, but it most likely was not representative of the entire sample set. Additional interlaboratory comparison results are discussed in a following section (see Section 3).
3/15/03
244
NIOSH Manuual of Analytical Methods
