contamination can be ruled out, the early EC loss is most likely due to oxidants in the sample. In our laboratory, EC loss in helium has been observed with samples from a few mines [70]. With these samples, the filter transmittance exceeded its baseline (initial) value during the last temperature step in helium. No sample pyrolysis was apparent, so the increase in transmittance was not caused by char loss. Moreover, analysis of sucrose standards revealed only a minor increase in transmittance in helium, and no increase was seen with samples from other mines. For these reasons, oxygen contamination was ruled out. On average, about 10% of the EC was removed at 870 °C in helium, but the carbon removed was included in the EC result because the OC-EC split was assigned when the initial transmittance was reached (in helium). Reducing the maximum temperature to 750 °C was recommended [59] for these types of samples (i.e., no sample charring and early splits in helium). Early EC loss was not seen at this temperature. A lower (< 750 °C) temperature may be required, depending on the type and amount of internal oxidant present. Our analytical result and precision were not affected by the early splits, but a reduced temperature was recommended in the interest of interlaboratory precision, which might be adversely affected. Interlaboratory testing was not conducted to determine if this is indeed the case, and, if so, whether a lower temperature would improve precision.
h. Reference Materials Unfortunately, a suitable reference material for OC and EC is not yet available. A new Certified Value (17.68% g/g) for the TC mass fraction of an urban dust standard (SRM 1649a, formerly SRM 1649) was recently reported by the National Institute of Standards and Technology (NIST), but only Information (not Certified) Values are provided for the EC content. The EC content (as EC:TC fractions) was determined by a variety of methods. As expected from past studies [35, 71–74], the results were quite variable. The EC:TC fractions found by 13 methods ranged from about 7% to 52%, and the data were distributed in three clusters. Method bias was not evaluated. The reported range is obviously too broad to use this material as an analytical standard. According to the Certificate of Analysis (1649a), the reported values may be useful for comparison with results obtained by similar methods, but this may not be the case for methods with optical corrections because filter samples are not available. Depending on its placement, bulk dust on a filter can present different optical properties, which may increase variability. Preparation of a reference material deposited on quartz filters is being investigated at NIST. Sample composition is an important consideration in the production of an analytical reference material. As is true with many standards, no single OC-EC material can be representative of all samples because there are many sources of particulate carbon and many different monitoring sites—both occupational and environmental. Production of an OC-EC reference material is further complicated by the variability between methods (SRM 1649a provides only Information Values for different methods), and the need of a filter-based material for thermaloptical methods. Because sample components that char pose a potential interference in the determination of EC, and interlaboratory variability is greater when samples contain them, analysis of organic materials that char provides an important quality assurance check. Sucrose serves as both an analytical standard and a check of the method’s char correction, but sucrose is a simple
3/15/03
241
NIOSH Manuual of Analytical Methods
