Study of x-ray fluorescence : Development in Geant4 of new models of cross sections for simulation PIXE. Biological and archaeological applications

The National Center of Nuclear Sciences and Technologies (CNSTN) operates an x-ray fluorescence analysis system. Although this system appears relatively simple - it does not comprise a particle accelerator - its use requires additional means of adjustment and calibration. The objective of work is to develop a protocol of optimization of the conditions of x-ray fluorescence analysis. This protocol consists in working out a code able to simulate the system in question in order to reproduce its response (output) to various inputs (inputs). The advantage of simulating such a system lies in the possibility of retorting its evolution as many once as necessary under independent conditions. Consequently, this makes possible the estimation of average values, interpreted like indicators of performance of the studied system. From there, we will be able to act on this same system at ends of improvement of its performances. The simulation which aims at calculating a numerical value by using random processes, i.e. probabilistic techniques, is called Monte Carlo simulation. The Monte Carlo expression was introduced in 1777 by (Le Comte de Buffon) (Throats Louis Leclerc) then reformulated in 1812 per Pierre-Simon Laplace. It is today associated with mathematical methods consisting in observing random numbers chosen to simulate a mathematical or physical problem and to obtain the required solution of the control of these numbers. Knowing that, for the simulation of x-ray fluorescence and the interactions particles matter in general, no one cannot predict where, when and how will interact a given particle, the Monte Carlo method is well adapted to reproduce the random behavior of the process of x-emission. It consists on following each particle, interaction after interaction, by randomly drawing the physical parameters from the phenomena according to functions probability distributions which describe the theoretical laws of the interaction particle-matter. If the process is repeated a great number of times, that makes it possible to reproduce the macroscopic effects and to calculate sizes like, in our case, x-ray fluorescence. Among several tools available for Monte Carlo simulation of the interactions particle-matter, the tool for simulation Monte Carlo Geant4 is characterized by a particularly flexible architecture based on directed technology object. This is why, to build our code of simulation, our choice was fixed on the use of this tool. We treat in the first chapter the guiding principle of x-ray fluorescence, which leads us to present the various phenomena which limit its sensitivity of analysis. We devote the second chapter to the presentation of the Geant4 tool, and our tests of validation of its electromagnetic processes. We describe the performances and the limits of this tool as for the simulation of the cross sections. We will expose in this same chapter our development, in the Geant4 tool, of new models of calculation of the cross sections of ionization of the atoms by the protons and the particles alpha. We discuss also our method of validation of the models in question. This work was object of publication. It was taken into account by the Geant4 project: our development currently replaces the data base EEDL of Geant4 in the process of determination of the cross sections of ionization by charged particles, and makes thus functional the Geant4 tool in the simulation of the x-ray fluorescence induced by charged particles. We describe in the third chapter our use of the code worked out for the simulation of the effectiveness of absorption of the standard Si(Li) detector which constitutes the essential piece of our device of analysis by x-ray fluorescence. We expose the adopted approach which enabled us to optimize the geometrical parameters of the studied detector. This work was presented at the second international conference of spectroscopy. We present, in the fourth chapter, some examples of concrete applications on the study of the potassium content in reference samples of the types mineralogical and biological. We show on this occasion the utility of our simulation program like effective means of adjustment and validation. Lastly, since we will have in the next years at CNSTN the proton-induced x-ray emission (PIXE) technique, we describe in the final chapter, another application in analysis of archaeological samples (coins of medieval currencies) by PIXE technique. This study initially brings us to the experimental control of this alternative of the method of analysis by x-ray fluorescence to be able then to describe, by Monte Carlo simulation, the experimental device which it includes and the spectral answer that it produces. The extension of the capacities of our Monte Carlo simulation code for the adjustment of PIXE spectra is to us of a great utility for our next in situ tests of development of PIXE technique.