[Resolution of overlapped spectra in polarization X-ray fluorescence spectrometry by genetic algorithm]

Overlapped spectra occur often in energy dispersive X-ray fluorescence spectrometry. Without a suitable curve fitting process, no correct qualitative and quantitative results can be obtained. The most difficult problems in curve fitting include that elements and their lines are unknown, background is uncertain sometimes, and initial model parameters may be in correct. In order to solve these problems, several available algorithms may be used. Among them are genetic algorithms. In the present work, polarization energy dispersive X-ray spectrometry was used in the determination of raw materials for permanent magnet. A genetic algorithm was successfully applied to the deconvolution of the overlapped spectra in magnetic materials. With its global searching capability, the genetic algorithm beatures higher resolution than the standard Marquardt-Levenberg method in resolving the overlapped X-ray spectra. Because of the powerful capability of genetic algorithms to deconvolute overlapped spectra, the algorithms are useful especially in energy dispersive X-ray spectrometry and complex material analysis.