Components with extraordinarily high analyte contents, for example copper metal from wires or plastics stabilized with heavy metal compounds, are presumed to be a crucial source of errors in refuse-derived fuel (RDF) analysis. In order to study the error generation of those 'analyte carrier components', synthetic samples spiked with defined amounts of carrier materials were mixed, milled in a high speed rotor mill to particle sizes < 1 mm, < 0.5 mm and < 0.2 mm, respectively, and analyzed repeatedly. Copper (Cu) metal and brass were used as Cu carriers, three kinds of polyvinylchloride (PVC) materials as lead (Pb) and cadmium (Cd) carriers, and paper and polyethylene as bulk components. In most cases, samples < 0.2 mm delivered good recovery rates (rec), and low or moderate relative standard deviations (rsd), i.e. metallic Cu 87-91% rec, 14-35% rsd, Cd from flexible PVC yellow 90-92% rec, 8-10% rsd and Pb from rigid PVC 92-96% rec, 3-4% rsd. Cu from brass was overestimated (138-150% rec, 13-42% rsd), Cd from flexible PVC grey underestimated (72-75% rec, 4-7% rsd) in < 0.2 mm samples. Samples < 0.5 mm and < 1 mm spiked with Cu or brass produced errors of up to 220% rsd (< 0.5 mm) and 370% rsd (< 1 mm). In the case of Pb from rigid PVC, poor recoveries (54-75%) were observed in spite of moderate variations (rsd 11-29%). In conclusion, time-consuming milling to < 0.2 mm can reduce variation to acceptable levels, even given the presence of analyte carrier materials. Yet, the sources of systematic errors observed (likely segregation effects) remain uncertain.
