3.
INTERLABORATORY COMPARISONS a. NMAM 5040 When results of the initial method evaluation were published [34], an interlaboratory comparison was not possible because the thermal-optical instrument was available in only one laboratory. Interlaboratory comparisons are especially important in this type of analysis because variable results have been obtained by different OC-EC methods [35, 71–74], and a reference material is not yet available. After additional laboratories acquired thermal-optical instruments, a round robin comparison [35] was conducted. Matched sets of filter samples containing different types of complex carbonaceous aerosols were distributed to eleven laboratories. Six of the eleven analyzed the samples according to NMAM 5040, while five used purely thermal (i.e., no char correction) methods. Good interlaboratory agreement was obtained among the six laboratories that used NMAM 5040. In the analysis of samples containing diesel particulate matter, the variability (RSD) for the EC results ranged from 6% to 9%. Only low EC fractions were found in wood and cigarette smokes. Thus, these materials pose minimal interference in the analysis of diesel-source EC. In addition, only minor amounts of EC were found in two OC standards that char: about 1% and 0.1% for sucrose and the disodium salt of ethylenediaminetetraacetic acid (EDTA), respectively. Two aqueous solutions of OC standards were included in the comparison as a check on the validity of the char correction and accuracy of the TC results. Variability (RSD) of the TC results for the two standard solutions and five filter samples ranged from 3% to 6%. Another interlaboratory comparison using NMAM 5040 was conducted more recently [77]. Seven environmental aerosol samples were analyzed in duplicate by eight laboratories. Four samples were collected in U.S. cities, and three were collected in Asia. Variability of the EC results ranged from 6% to 21% for six samples having EC loadings from 0.7 to 8.4 :g/cm2 . Four of the six had low EC loadings (0.7 :g/cm2 to 1.4 :g/cm2 ). The variability of the OC results ranged from 4% to 13% (OC loadings ranged from about 1 to 25 :g/cm2 ). Results for TC were not reported, but the variability reported for the OC results should be representative of that for TC because the samples were mostly OC (75% to 92%).
b. Other Methods Thermal methods. Different thermal methods have given consistent agreement for TC, but OC-EC results have been quite variable [35, 71–74]. The degree of variability depends on the sample type. In general, there is greater disagreement among methods when samples contain materials that char [35, 71, 72] (e.g., wood and cigarette smokes). Methods that employ a lower maximum temperature and/or do not correct for char obtain results that are positively biased relative to NMAM 5040 [35]. When a lower temperature (typically 550 °C) is used, less thermal breakdown of refractory organic components (and possibly carbonates) may occur. This, as well as lack of char correction, can positively bias the EC results. In the round robin study [35] discussed above, three laboratories employed four purely thermal methods that specified a maximum in nitrogen (not helium) of about 550 °C. Unlike the
3/15/03
245
NIOSH Manuual of Analytical Methods
