STATISTICAL TREATMENT OF ETA-AAS (ELECTROTHERMAL ATOMIZATION - ATOMIC-ABSORPTION SPECTROMETRY) SOLID SAMPLING DATA OF HETEROGENEOUS SAMPLES

If pulverised samples contain rare particles of high analyte content ("nuggets"), skewed distributions of analytical results may be found. These distributions can be described by the Poisson probability function depending from the average number of nuggets in a subsample. For the solid sampling technique with electrothermal atomisation (graphite furnace) AAS skewed distributions must be accepted to get accurate results. However, if only a few replicates are performed, extreme values caused by several nuggets in a subsample must be rejected. The mean therefore often is somewhat too low. Compared to the errors of other methods the accuracy of the solid sampling method is sufficient even with samples which show micro-heterogeneity. The nugget effect is documented and discussed for a bovine muscle sample (BCR CRM-184).