APPLICATION OF MULTIVARIATE STATISTICAL-ANALYSIS TO TIME-DEPENDENT SPECTROSCOPY

Time dependent spectroscopy is a promising method for the study of dynamic events by microanalysis. Since it produces a large amount of data, specialized data analysis procedures must be developed. Multivariate Statistical Analysis appears to be a good candidate since it allows data analysis without any a priori model of the evolution. It is also able to perform data processing (noise filtering for instance) and data extrapolation, such as the extrapolation to zero-dose. Some potential uses of this method are explored in several simulations as well as in an experimental example in the field of EDX spectroscopy.