PERFORMANCE OF TXRF and GI-XRF ANALYSES - Total-Reflection X-Ray Fluorescence Analysis and Related Methods - page 240

more analyte elements and from the internal standard element. Besides, the
sensitivity values of these elements should be known from preparatory mea-
surements. However, this work has to be carried out only once as a kind of
calibration. Each subsequent quantification is generally easy and reliable, and
consequently TXRF has great advantages over conventional XRF.
Of course, there are some conditions for the applicability of internal
standardization, but the limitations can easily be observed. They apply to
the sample amount used for micro- and trace analyses. On the other hand, a real
surface or thin-layer analysis should be carried out in accord with its own
regulations, which are treated in Section 4.5.
4.4.1PrerequisitesforQuantification
The net intensities of the principal peaks have to be determined in each
spectrum. This can be done following the common practice in EDS. The
sensitivity values for these peaks of the respective elements, however, need
to be determined quite seldom. This is done for any particular excitation mode
and is only repeated after a repair or replacement of instrumental components.
4.4.1.1DeterminationofNetIntensities
An X-ray spectrum generally shows several peaks on a structured background,
some of which may partly overlap. These peaks have already been identified
and assigned to certain elements by means of a previous qualitative analysis.
Each element is consequently represented by two or three principal peaks, or at
least by one. The net intensity N x of the strongest peak or possibly of some
additional peaks, must be determined, for each analyte x . This intensity is
defined by the area under the peaks but should be corrected for the spectral
background and for the overlapping of neighboring peaks. It may be deter-
mined by summation of all the counts being accumulated in the individual
channels representing the peaks and by a subsequent subtraction of the
background and overlapping peaks. This can be done semiautomatically after
setting of“windows”that limit each peak to a lower and upper border of the
photon energy. The correction is made by subtraction of a trapezoidal area
under the peak and between the borders. This method is a rough approxima-
tion that is only suitable for strong peaks on a low background and without
strong overlapping. In this case, the window width can be chosen to be two or
three times the peak width (FWHM = full width at half maximum) so that the
correction is appropriate. In most cases, however, a severe error will result from
use of this method, which is unacceptable.
For that reason, other methods have been developed that can also be
applied automatically. The first corrects for the background and is based on
a Fourier transformation of the spectrum. This mathematical procedure results
in three transformed but significantly different parts of the spectrum. The low-
frequency part represents the background, the medium-frequency portion