LINE-SHAPE EFFECTS ON THE DETERMINATION OF COSTER-KRONIG PROBABILITIES USING SI(LI) X-RAY-DETECTORS

The great majority of measured values of the Coster-Kronig probability f(23) derive from x-ray coincidence experiments using radionuclide sources or from synchrotron-radiation excitation of thin foils. Very little attention has been paid in published work to the role of line-shape effects and satellites in the x-ray spectra that are obtained. It is shown that this neglect can influence f(23) values derived from such measurements. This may explain in part the well-established trends for measured f(23) values to fall below the theoretical predictions for elements with atomic number Z < 90 and to fall above when Z > 92. In elemental analysis techniques based upon x-ray emission spectroscopy, where a database of atomic parameters such as Coster-Kronig probabilities is used, the same approximations are made in treating the spectra. Until a more sophisticated approach to spectrum reduction is developed, it may be preferable in this context to use ''nominal'' Coster-Kronig probabilities derived from experiment in preference to theoretical values.