IMEP-29, total arsenic, cadmium, lead, mercury and tin, and extractable cadmium and lead in feed of plant origin : interlaboratory comparison report

The Institute for Reference Materials and Measurements (IRMM) of the Joint Research Centre (JRC), a Directorate-General of the European Commission, operates the International Measurement Evaluation Programme IMEP. It organises interlaboratory comparisons (ILC's) in support to EU policies. This report presents the results of an ILC which focussed on the determination of total As, Cd, Pb, Hg and Sn, and extractable Cd and Pb in feed of plant origin following Directive 2002/32/EC of the European Parliament and of the Council on undesirable substances in animal feed. The test material used in this exercise was a candidate reference material, the matrix being rye grass. The material was relabelled and dispatched to the participants in the second half of October 2009. Each participant received one bottle containing approximately 10 g of test material. Sixty-two participants from 23 countries registered to the exercise of which 59 reported results. The assigned values (Xref) for total Cd, Pb, As, Hg and Sn were the reference values as obtained during the certification campaign taking place simultaneously to the ILC. The assigned values for extractable Cd and Pb were provided by IRMM using isotope dilution-inductively coupled plasma-mass spectrometry (ID-ICP-MS). Participants were invited to report the uncertainty of their measurements, which was done by around 90 % of them. The laboratory performance was evaluated using z- and ζ-scores in accordance with ISO 13528. The standard deviation for proficiency assessment (also called target standard deviation), σ, was fixed at 15 % for all measurands on the basis on the outcome of previous ILCs. The outcome of the exercise was altogether positive, with 68 % or more of the participants reaching satisfactory z-scores for all measurands except for total As and Hg, which appeared to be problematic in this exercise, showing a non-normal results distribution and tendency to very high means. The ζ-scores were not as good as the z-scores, which indicates a persisting problem of appropriate uncertainty estimation. Finally, total Sn was included for the first time in an ILC. Results were better than expected, but can certainly be improved.