Optimized acquisition parameters and statistical detection limit in quantitative EELS

Imaging and characterization of fine γ′ precipitates in a commercial nickel-base superalloy

Empirical datasets of volume fractions and size distributions of small gamma' precipitates from "real" multi-component engineering nickel-based superalloys are vital to calibrate and validate the computer models which predict high sensitivities of mechanical properties to size and volume fraction of these fine gamma precipitates, and in order to accelerate microstructure and alloy development. Consequently, we investigated a number of imaging techniques available in a Tecnai F-20 FEG/TEM and selected the technique which best enabled rapid and extensive acquisition of these datasets using the engineering alloy, René 88'DT. The EFTEM technique was found to be the most appropriate method for imaging fine gamma' precipitates while further investigation showed that the Cr-M-edge, in comparison with other ionization-edges provided the best images based largely on contrast-to-noise ratio. Imaging of the Cr-M-edge elemental maps were further improved by investigating the effects of microscope parameters, imaging filter parameters and analysis of the experimental electron energy loss spectra obtained from this alloy. In addition, a novel technique to determine the volume fraction of the fine gamma' precipitates without the need to determine the absolute thickness of the TEM foil is proposed.

Model based quantification of EELS spectra

Recent advances in model based quantification of electron energy loss spectra (EELS) are reported. The maximum likelihood method for the estimation of physical parameters describing an EELS spectrum, the validation of the model used in this estimation procedure, and the computation of the attainable precision, that is, the theoretical lower bound on the variance of these estimates, are discussed. Experimental examples on Au and GaAs samples show the power of the maximum likelihood method and show that the theoretical prediction of the attainable precision can be closely approached even for spectra with overlapping edges where conventional EELS quantification fails. To provide end-users with a low threshold alternative to conventional quantification, a user friendly program was developed which is freely available under a GNU public license.

Quantitative energy-filtered image analysis in cytochemistry. II. Morphometric analysis of element-distribution images

A combination of energy-filtered electron microscopy (EFEM) and an image-analysis system (IBAS/2000) is used for a morphometric analysis of chemical reaction products in cells. Electron energy-loss spectroscopic element-distribution images are acquired from cytochemical reaction products in a variety of cellular objects: (1) colloidal thorium particles in extra-cellular coat material, (2) iron-containing ferritin particles in liver parenchymal cells, (3) barium-containing reaction products in endoplasmic reticulum stacks, (4) elements present in lysosomal cerium- and barium-containing precipitates connected with acid phosphatase (AcPase) or aryl sulphatase (AS) enzyme activity. Areas or area fractions are determined from such element-distribution images by application of an objective image segmentation method. By superposition of two or more element-distribution images, mutual element relations are qualitatively established in lysosomal cerium- and barium-containing precipitates connected with acid phosphatase (AcPase) or aryl sulphatase (AS) enzyme activity. By comparing electron spectroscopic images (ESI) with element-distribution images, the mutual contrast per element relations are quantitatively investigated. The obtained gain in resolution in such electron energy-loss spectroscopic element-distribution images will be explained and discussed.

Weighted least squares estimation of background in EELS imaging

In quantitative Electron Energy Loss Spectrometry, a weighted least squares estimation should theoretically be used to estimate the background law below core edge energy, since the variances of the data vary. However, it is found that proper weighting makes the above edge signal-to-noise ratio decrease rather than increase. This result is discussed, and the influence of the bias introduced by the logarithmic transformation of the data is quantified.