Interlaboratory comparison of dioxin analyses in extracts of flue gas samples

Abstract

On behalf of the Coordinating Committee on Dioxin Research an interlaboratory comparison on the compliance of measurement quality of dioxins (PCDDs and PCDFs) has been organized. The interlaboratory comparison concerns clean-up and instrumental analysis and was intended as a demonstration of comparability of results. After a successful demonstration the participants are considered to be competent to perform analysis for regulatory purposes. Eight samples were collected at two different incinerators with levels ranging from 0.008 - 1.3 ng TEQ/m03. Four of these samples were distributed in duplicate. Therefore, each participant analyzed 12 ampullae containing sample material. The target for comparability was 11 out of 12 results within 25% of the assigned value. The comparability of results for the six participants over a large range of concentrations (0.02-2 ng/m03) is satisfactory: the relative standard deviation within this range is approximately 15%. Three of the six laboratories met the criterion described above. For the other laboratories the cause of exceeding the criterion could be determined and be eliminated after adaption of the analytical procedure. These participants were given the opportunity to prove their measurement quality in a confirmation study, based on the analysis of five spiked mixtures of the samples from the first investigation. The analytical results of the confirmation study were consistent with those of the first round and 14 out of 15 results were within 25% of the assigned values. It is concluded that in the concentration range of 0.02-2 ng TEQ/m03 the accuracy of dioxin analyses of sample extracts, expressed as a 95% confidence interval, is approximately 25%. However, it should be noted that demonstrations of measurement quality, like this one, are only representative of a certain period making repetition necessary. Moreover, the evaluation of comparability of overall results should include other sources of variation like sampling and 12C-standards of different origin.<br>