Abstract:
The implementation of standardless quantitative X-ray fluorescence analysis to a conventional scanning electron microscope equipped with an X-ray spectrometer is subject of the present work. For this purpose, an optimised sample holder was designed, constructed, successfully characterised and tested, which transfers the operation principle of a transmission-type end-window X-ray tube into the specimen chamber of a scanning electron microscope. The device allows a fast and easy exchange of target, filter, and sample and therefore offers flexible excitation conditions and a high sample throughput. As modifications of the microscope hardware are not necessary, switching between electron microprobe analysis and X-ray fluorescence analysis is easily accomplished. X-ray fluorescence analysis inside the scanning electron microscope offers significantly improved detection limits compared to electron excitation of the X-ray emission spectrum. The analytical results show that in common alloys composed of first row transition metals a two to seven fold decrease of detection limits is achieved. Standardless quantitative trace analysis of heavy elements in a light element matrix is even shown to be possible down to mass concentrations of approximately 3 ppm lead in aluminium corresponding to an atom fraction of only 400 ppb.
A Monte Carlo procedure to predict the spectral response of X-ray excited samples is described. An expansion of this procedure to simulate subsequent electron-photon interactions is presented, which advantageously enables the simulation of electron excited X-ray emission spectra including the Bremsstrahlung background. Standardless unified Monte Carlo quantification of X-ray emission spectra acquired in a scanning electron microscope is thus possible with high accuracy and precision. As Monte Carlo simulations do not distinguish between characteristic X-rays and continuous background, numerical processing of spectra prior to analysis, such as background removal, peak fitting, and overlap correction, can be entirely abandoned.
Unlike fundamental parameter methods, Monte Carlo simulations are solely based on atomic properties. Therefore, valuable additional information such as size and shape of the electron diffusion volume, X-ray depth distribution functions, or many analytical signals such as the spectral distribution of backscattered or transmitted electrons are simulated at the same time. Monte Carlo techniques are easily adapted to suit special requirements, such as more complex sample geometries. In this context, additional applications of the proposed Monte Carlo techniques to the metrology of thin samples by X-ray scattering and electron backscattering are reported exemplarically. These also show excellent agreement between experimental and simulated data in this field.