Je. Fernandez et A. Tartari, EDXRF PROCEDURE FOR QUANTITATIVE-ANALYSIS MATCHING THEORETICALLY GENERATED REFERENCE SPECTRA TO MEASURED SPECTRA, X-ray spectrometry, 24(6), 1995, pp. 277-282
The diversity of processes that contribute to an x-ray fluorescence sp
ectrum make it difficult to obtain an accurate estimate of the line in
tensities needed to determine the sample composition in XRF analysis,
The radiative contributions forming the spectrum are produced by the i
nteractions that x-ray photons undergo with the atoms of the sample. M
any multiple scattering terms contribute to the continuous portions of
spectrum that overlap the lines of interest. The lines themselves are
formed by the overlap of many narrow (discrete) contributions and som
e continuous ones: one 'primary intensity' plus several enhancement te
rms, The lines can also interfere with other neighbouring lines produc
ed in the sample, The resulting spectrum, sufficiently complicated at
this stage, is still modified in a complex way by the influence of the
detector response and successive digitalization by the multi-channel
analyser, The difficulties that such spectral complexity introduces in
the process of extracting the 'primary fluorescence intensity' from t
he experimental spectrum can be overcome by using a theoretical spectr
um of reference that, in this paper, was calculated using the code SHA
PE. Such a theoretical spectrum has the property that all the single c
ontributions are well known, and therefore can be used to estimate cor
rectly the fraction of 'primary fluorescence intensity' under any expe
rimental line of interest in the real spectrum enclosed between two gi
ven energy channels, Since the theoretical spectrum can be computed on
ly for a known composition, the calculated fraction can be initially o
btained by feeding SHAPE with an arbitrary composition, The fraction e
stimated with the theoretical spectrum is used to correct the correspo
nding peak area in the experimental measured spectrum, which in turn i
s used to obtain a new calculated composition, This result is used to
feed a new spectrum computation, defining an iterative method that con
verges to the unknown composition, The theoretical details of this ite
rative method are described and some experimental examples of EDXRF an
alysis using this technique are given. The advantages of the method fo
r treating some paradigmatic cases of analysis are also illustrated.