ABSTRACT
A fundamental parameters programme has been developed to generate artificial EDX spectra modelling grain boundary segregation and variations in foil thickness. Series were subjected to linear multivariate statistical analysis (MSA) to explore the form of the information components. Segregation information was confined to one component, confirming recent experiments. Self-absorption generated a series of components but suggested that just a single component would be observed in typical experiments, validating the use of MSA for EDX in many practical situations.
