Methylmercury in tuna: demonstrating measurement capabilities and evaluating comparability of results worldwide from the CCQM P-39 comparison

Six metrology institutes (NMIs) representing at the Comité International des Poids et Mesures (CIPM) 4 Member States of the Metre Convention and 2 international organisations, and 8 "expert" laboratories selected outside CIPM have compared their capabilities to quantitatively measure methylmercury (MeHg) in a prepared tuna material containing approximately 4.3 mg kg-1 Hg. This comparison was the object of the CIPM-Comité Consultatif pour la Quantité de Matière (CCQM) Pilot Study 39, organised by the Institute for Reference Materials and Measurements (IRMM), from the European Commission - Joint Research Centre. Beside the test material itself, a bottle of the BCR-464 tuna Certified Reference Material (CRM) and an ampoule of IRMM-670, a 202Hg isotope enriched MeHg candidate isotopic CRM, were distributed to all participants, who were free to apply the measurement strategy of their choice. Four, including 1 NMI, relied on external calibration or the method of standard additions, whereas the other 10 implemented an isotope dilution mass spectrometry (IDMS) approach and chose to use the IRMM-670 for their measurements. Alkaline digestion at room temperature (with manual shaking) or high temperature (under sonication, oven or hot plate conditions) was employed by most participants, with hydrochloric acid leaching the second most popular choice. Alkylation (4 phenylations, 4 ethylations and 3 propylations) in the aqueous phase was preferred by a large majority over butylation by the Grignard reaction. All participants were requested to estimate the uncertainty associated with their results and 9 out of 14 stated relative combined uncertainties below 6% (k = 2). Despite this apparent consensus, the perception of which factor caused the largest contribution to this estimation differed among participants because of the differences in the analytical methodologies deployed but also because of wide differences of the concepts of uncertainty estimation. The 'mixture mode' (MM) median, calculated also from the measurement uncertainties stated by the participants, was 1.967 ± 0.204 × 10-5 mol kg -1 (95% confidence). Twelve of the results were re-grouped within a range of less than 0.3 × 10-5 mol kg-1 (MM median = 1.967 ± 0.162 × 10-5 mol kg-1, 95% confidence): they nearly all (1 exception) overlapped with each other within k = 2 stated uncertainties. For the other 2 results the uncertainty seemed to have been particularly underestimated as they lay, respectively, at more