In interpreting X-ray intensity data from electron probe microanalyzers, the simplest approximation that can be made is that the X-ray intensity is a linear function of concentration. This linear relationship has been found to hold only in very limited cases. Where this simple relationship does not hold, it is generally assumed that increased concentration of an element is reflected in increased intensity of its characteristic X-ray lines. This assumption, which is always valid for binary samples, has been extremely useful, particularly for the interpretation of X-ray scanning pictures. However, in analyses based solely on observed X-ray intensities, large deviations from the linear relationship can make semiquantitative results misleading.

For multielement systems, the deviations from linearity result in ranges of concentration for each element within which increased concentration of the element does not result in an increase of its characteristic X-ray intensity. The magnitude of these deviations, and. hence the ranges of concentration, is related to the X-ray absorption and secondary fluorescence in the sample, which are in turn dependent on instrumental parameters such as X-ray take-off angle. Such inversions in the normal intensity-concentration relationship can result in gross misinterpretation of X-ray scanning pictures on even a qualitative basis. The various factors that lead to deviations from linearity are evaluated, and methods are given for correctly interpreting X-ray intensity data on a qualitative and semiquantitative basis.