Linear least-squares fit evaluation of series of analytical spectra from planar defects: extension and possible implementations in scanning transmission electron microscopy


T. Walther. Tel: +49 228 9656 198; fax: +49 228 9656 116; e-mail:


In a previous paper, a new technique was introduced to determine the chemistry of crystallographically well-defined planar defects (such as straight interfaces, grain boundaries, twins, inversion or antiphase domain boundaries) in the presence of homogeneous solute segregation or selective doping. The technique is based on a linear least-squares fit using series of analytical (electron energy-loss or energy-dispersive X-ray) spectra acquired in a transmission electron microscope that is operated in nano-probe mode with the planar defect centred edge-on. First, additional notes on the use of proper k-factors and determination of Gibbsian excess segregation are given in this note. Using simulated data sets, it is shown that the linear least-squares fit improves both the accuracy and the robustness to noise beyond that obtainable by independently repeated measurements. It is then shown how the method originally developed for a stationary nano-probe mode in transmission electron microscopy can be extended to a focused electron beam that scans a square region in scanning transmission electron microscopy. The necessary modifications to scan geometry and corresponding numerical evaluation are described, and three different practical implementations are proposed.