EMPA St Gallen, Swiss Federal Laboratories, Materials Science and Technology, Lerchenfeldstrasse 5, CH-9014 St Gallen, Switzerland
Abstract:
The Horwitz curve states a relationship between the analyte concentration in various matrices and the interlaboratory reproducibility as found in collaborative studies. The reproducibility becomes worse (larger) with lowering of the analyte concentration. We found that the Horwitz assumption is too optimistic for analyses in plastics. Based on 15 different proficiency tests with 68 reproducibility-concentration relationships, it becomes clear that the precision can be worse by a factor of ten or more than predicted by the Horwitz curve. Organic analytes seem to be more critical than inorganic ones. One of the reasons is assumed to be the poor homogeneity of many samples. Another problem could be the lack of standard methods for plastics analyses. However, the great observed variability cannot be explained by these two features alone.