It has been found that sampling constants vary greatly not only from major elements to trace elements, but also between individual trace elements. A comprehensive investigation of a potential reference material therefore requires the determination of sampling constants for all elements to be certified, and other analytical methods therefore have to be included. For methods in statistical control the described strategy can be applied.
The -emitter32P was used to determine total phosphorus by INAA in Skim Milk Powder RM 63, a material now certified by the EEC Bureau of Reference (BCR). Samples and comparator were irradiated in the Danish reactor DR 3. One month later the samples were dissolved in water and aliquots counted with a GM end-window counter using absorber thicknesses of zero to 400 mg/cm2. The Synthesis of Precision was introduced to find the absorber best suited for discriminating against other -emitting isotopes and at the same time giving maximum precision.
The use of loss-free counting systems makes possible the exact correction for pile-up and dead-time losses during counting of a mixture of short-lived radionuclides even at very high count rates. However, counting statistics cannot be calculated by taking into account only the Poisson distribution of the incoming -quanta, such as is done in existing computer programs for -spectrometry. At moderate count rates Müller statistic was found to account for the observed variability between duplicate countings; however, at higher count rates the variability of weighing factors was found to be significant in comparison with the Müller statistic. While counting statistics could not be correctly estimated for short-lived species, experiments showed excellent accuracy for initial dead times up to 90%.
Multielement certification analysis by instrumental neutron activation analysis requires simultaneous irradiation of several elemental comparator standards in order to ascertain traceability. Internal consistency of different comparators may be checked by calculation ofko-ratios, which show large deviations from unity in case of stoichiometric or other gross errors. Quality assurance based on the Analysis of Precision ofko-ratios from replicate analyses detects unexpected variability associated with inaccurate comparator standards. In two actual cases of cerification lack of statistical control amongko-ratios led to biased analytical results.
Three botanical reference materials prepared by the National Bureau of Standards have been studied by neutron activation analysis
to evaluate their suitability with respect to the determination of vanadium in biological samples. Various decomposition methods
were applied in connection with chemical or radiochemical separations, and results for vanadium were compared with those found
by purely instrumental neutron activation analysis. Significantly lower results indicate losses or incomplete dissolution,
which makes SRM 1575 Pine Needles and SRM 1573 Tomato Leaves less satisfactory than SRM 1570 Spinach. A reference value of
1.15 mg/kg of this material is recommended, based on results from 3 different methods. All three materials are preferable
to SRM 1571 Orchard Leaves, while Bowen's Kale remains the material of choice because of its lower concentration.
Authors:K. Heydorn, Z. Alfassi, E. Damsgaard, B. Rietz and P. Solgaard
In a European study of fish death in mixing zones of rivers with different acidities a reliable determination of Al and other elements in gills from freshwater fish was required, and both INAA and ICP-MS were studied as candidate reference methods. INAA requires minimum sample handling with a correspondingly small risk of contamination and no blank value; however, a careful study was needed of both nuclear interference from P and the increased detection limit caused by other major elements in the sample, before reliable results for Al could be ascertained. ICP-MS requires dissolution of the sample with a resulting risk of contamination and a significant reagent blank; while sensitivity was good, the interference from N created problems for sample decomposition in the microwave oven. Our experience with actual samples indicates that both methods suffer from considerable contamination problems, requiring that samples be handled in a clean bench with superpure reagents. Nuclear interference was determined experimentally by irradiating stoichiometric P-compounds with and without a Cd-shield; the observed interference of 1 g Al from 50 mg of P was found to require no correction in almost all cases. The accuracy of results was ascertained by analyzing SRM 1577 Bovine Liver.