Quantification of boundary segregation in the electron microscope

Abstract

When studying equilibrium grain-boundary segregation using the small (~1 nm) electron probe of the scanning transmission electron microscope and X-ray energy dispersive spectroscopy, the assumptions made about the size and shape of the beam-specimen interaction volume may introduce errors in quantification of up to a factor of five. Comparisons between experimental segregation profiles and different theoretical models have shown that a Gaussian model for the electron distribution will provide the best description for the interaction volume. Calculations of minimum detectable segregation levels have shown that optimum sensitivity is not achieved in the thinnest samples or even with the smallest probe sizes. In addition, operating at 300 keV (rather than 100 keV) will halve the minimum detectable segregation level, assuming all other experimental conditions are equal. Rastering the electron probe over a fixed area while acquiring spectra improves the accuracy of quantification but at the price of reducing sensitivity by at least a factor of three.