Calibrated x-ray intensities of arsenic and gallium, implanted with various ion-energies and doses into silicon, were measured over a wide range of electron-beam energies and angles of incidence. For the first time Monte-Carlo (MC) simulation was applied to evaluate the EPMA-data with respect to depth-profiles parameters, particularly at non-normal electron incidence. Accurate agreement between MC-simulated and measured k-ratios was obtained in the whole range of excitation conditions applied. The resulting range parameters determined from the EPMA data agree closely with those obtained by other authors from RBS or NAA but like these they show systematic deviations from theoretical predictions. Actual measurements on samples implanted with 1014 ions cm-' clearly prove the limits of detectability for As and Ga in Si to be below this number. Similar sensitivity was found for elements with atomic numbers Z > 10 but for light elements (2 < 10) the limit of detectability is increased by about one order of magnitude. Provided the concentrations are appropriate, EPMA combined with MC-simulation is a promising technique for the accurate quantitative, non-destructive, and spacially resolved analysis of depth-profiles.In the last decade electron probe microanalysis (EPMA) has experienced a rapid development to a quantitative microbeam technique for the analysis of layers or stratified systems on bulk Even measurements on implanted elements were reported EPMA is not normally applied for depth profile or thin film analysis, but some features suggest its application: it is a non-destructive quantitative method which is able to detect all elements with Z > 3 and EPMA instruments are available in many laboratories. But neither the signal variation caused by different depth profiles nor important quantities like the limits of detectability, or the accuracy of EPMA-results have yet been assessed. Thus, up to now the EPMA technique can hardly be valued in comparison with more commonly used methods like SIMS, AES, RBS or nuclear activation analysis (NAA). This paper investigates first of all the x-ray intensity variations caused by different depth distributions. Secondly the reproducibility and limits of detectability of the measurement and the accuracy of the quantitative results are assessed.The quantitative interpretation of the measured x-ray intensities requires the theoretical modelling of the electron -sample interaction.' Usually, analytical models are used for this purpose. These models give reliable results for homogeneous samples,16 but for the moment they provide realistic descriptions neither at interfaces nor in the case of non-normal electron incidence in general.4 While the first point really represents a problem for the EPMA-analysis of heterostructures, the second has been avoided up to now by simply not applying non-normal electron incidence. To overcome these * Author to whom correspondence should be addressed restrictions the theoretical calculations in this work were performed by Monte-Carlo (MC) simulation. T...
