Elimination of analytical error due to the use of catch weights with loss-eliminated alphas in X-ray fluorescence spectrometric analysis of limestone using borate fusions

In X-ray florescence (XRF) spectrometry, the weight of a sample can be partially lost during the high temperature fusion necessary for the casting of borate glass disks. This weight loss is due to the decomposition of compounds such as carbonates and nitrates, and the subsequent loss of loss-on-fusion (LOF) compounds such as CO2, NO2 and O2, etc. Without a prior experimental determination of LOF, its effects on XRF analysis can be eliminated using special matrix correction coefficients called loss-eliminated alphas (LEAs) or αije. Using this method, samples are usually analyzed ‘as-received’ and the difference between 100% and the measured sum of elemental oxides gives the LOF. The LEAs are dependent on the nominal weights of sample and flux specified for the calculation of LEAs. The effect of the use of catch-weights (CWs; weights other than the nominal weights) on the analysis is generally corrected using a simple proportional correction factor. In contrary to the popular belief that CWs can be satisfactorily used for all sample matrices, the results presented here indicate that the error can be significant (>typical precision) and systematic as in the analysis of limestone where the LOF is approximately 40–45%. However, CWs could be used with a modified LEA value to eliminate the error on the analysis of major elements associated with the lost compounds. The results for the analysis of Standard Reference Material, 1C Argillaceous Limestone, are in good agreement with the certified values.