Improved spectrum simulation for validating SEM-EDS analysis

Abstract

X-ray microanalysis by SEM-EDS requires corrections for the many physical processes that affect emitted intensity for elements present in the material. These corrections will only be accurate provided a number of conditions are satisfied and it is essential that the correct elements are identified. As analysis is pushed to achieve results on smaller features and more challenging samples it becomes increasingly difficult to determine if all conditions are upheld and whether the analysis results are valid. If a theoretical simulated spectrum based on the measured analysis result is compared with the measured spectrum, any marked differences will indicate problems with the analysis and can prevent serious mistakes in interpretation. To achieve the necessary accuracy a previous theoretical model has been enhanced to incorporate new line intensity measurements, differential absorption and excitation of emission lines, including the effect of Coster-Kronig transitions and an improved treatment of bremsstrahlung for compounds. The efficiency characteristic has been measured for a large area SDD detector and data acquired from an extensive set of standard materials at both 5 kV and 20 kV. The parameterized model has been adjusted to fit measured characteristic intensities and both background shape and intensity at the same beam current. Examples are given to demonstrate how an overlay of an accurate theoretical simulation can expose some non-obvious mistakes and provide some expert guidance towards a valid analysis result. A new formula for calculating the effective mean atomic number for compounds has also been derived that is appropriate and should help improve accuracy in techniques that calculate the bremsstrahlung or use a bremsstrahlung measurement for calibration.