First order correction of EELS measurements based on experimental ADF imaging

Abstract

In Physical Failure TEM (Transmission Electron Microscopy) analysis with EELS (Electron Energy Loss Spectroscopy) technique, it is common to directly evaluate the composition of certain elements side-by-side and sample-to-sample, based on the extracted intensity of the corresponding core-loss signals after background removal. However, there can be non-uniformity of TEM sample thickness, difference of sandwiched structures, and/or mixing of different elements between the compared regions. Such variations of local thickness and/or structures can cause incorrect signal extraction in EELS measurement due to the nature of interaction of incident electrons with matter. This short communication provides the first order correction based on ADF (Annular Dark Field) intensity, collected by (High Angle Annular Dark Field), in combination with theoretical simulation of electron interaction of incident electrons with materials, and the direct measurements of the wedged-TEM samples and transmitted beam current of incident electrons, although it does not aim to address all artificial effects from EELS quantification. The electron elastic and inelastic scatterings depends on both the sample thickness and effective atomic number. Their correlation in corporation with multiple scattering are computed with LenzPlus simulation proposed by R. F. Egerton. This paper theoretically and experimentally discusses the cause of EELS inaccuracy and then proposes a first order correction technique which derives a more accurate elemental quantification in EELS measurements. This is especially useful in semiconductor PFA (Physical Failure Analysis) when elemental quantification is need site-by-site and sample-by-sample when thickness variation and/or structural variation exist. Finally, an example of quantification improvement with a structure of Si active area in DRAM (Dynamic Random-Access Memory) is presented.