Narrow Results

A new x-ray-spectrum-analysis program, which is capable of fitting with response functions including a tail function, an escape peak and sub-lines, has been developed. In this code, the tail function is expressed by combination of two or three Gaussian functions. A tail function, an escape and sub- or satellite-lines are regarded as functions belonging to the main peak and are included in it. A small shift of peak position depending on measuring conditions can be easily corrected in the program. As a result of fitting to practical spectra with the response function thus prepared, it becomes possible to draw a smooth background over a wide x-ray-energy range and to analyze a whole spectrum simultaneously. Thus, accuracy and reproducibility of a spectrum analysis are much improved. By means of this code, correct values of peak yield of Co-Kα, which overlaps with the tail of Fe-Kβ and is quite difficult to be accurately separated by fitting with Gaussians, have been obtained. Furthermore, accuracy of peak separation of a small peak, which overlaps with the escape peak belonging to a huge peak, has been improved. Accuracy of quantitative analysis for high-Z elements by means of Kβ yields has also been improved by using the response function including sub-lines, and it became possible to accurately separate small Kα lines from Kβ lines of the other elements.

A few attempts have been done to reduce ambiguous factors in the analysis of complex x-ray spectra. Uncertainty in peak identification and separation due to overlapping of x-ray peaks can be reduced by taking into account the response function of x-ray peaks and by using peak functions including all the x-ray sub-lines in a x-ray series. Ambiguous factors due to discontinuity of backgrounds which is sometimes followed by huge peaks can be reduced by considering contributions from the pile-up, escape and tail functions. The problem of uncertainty in the background fitting in a region of low x-ray energy, where many peaks overlap each other and determination of the background function contains much difficulties and ambiguities, can be solved by using peak functions involving tails in place of Gaussians.