A general Monte Carlo simulation of ED-XRF spectrometers. New developments Tom Schoonjans 1 , V. Armando Solé 2 , Manuel Sanchez del Rio 2 , Claudio Ferrero 2 and Laszlo Vincze 1 1-X-ray Microspectroscopy and Imaging group (XMI), Ghent University, Belgium 2-European Synchrotron Radiation Facility (ESRF), Grenoble, France Introduction Since their inception during the Manhattan project, Monte-Carlo simulations have been used for a variety of applications in physics, mathematics, engineering, finance and more. When applied to X-ray fluorescence spectroscopy, it is well known that Monte-Carlo simulations are useful for predicting the spectral response of samples irradiated with an X-ray beam of given characteristics. This requires the development of a dedicated computer code that simulates the histories of a large number of individual photons, whose trajectories in the system are modelled as a number of straight steps. At the end of each step, an interaction occurs, leading to a change in direction, energy and polarization state. A photon’s trajectory is terminated when it leaves the system, or when the detector captures it. If detection occurs, one count is added to the appropriate channel of a virtual multichannel analyzer. The algorithm is greatly optimized through the introduction of variance reduction techniques. XMI MSIM •Software package dedicated to the Monte-Carlo based simulation of ED-XRF spectrometers • Based on the work of Vincze et al. •Update of the physical data (cross sections, line energies etc.) through linking with xraylib •Simulation of M-lines and cascade effects (radiative and non-radiative) •Improved detector response function: pulse pile-up and fluorescence/Compton escape peaks •Support for multithreading (OpenMP) and multiprocessing (MPI) •Callable as a plugin from PyMca: quantification of XRF data • Written in ANSI-C and Fortran 2003 •Makes use of GNU Scientific Library, libxml2, HDF5, GTK+... • Distributed under the GNU General Public Licence: www.github.com/tschoonj/xmimsim References [1] L. Vincze, K. Janssens, and F. Adams. Spectrochim. Acta, 48B(4):553–573, 1993. [2] A. Brunetti, M. Sanchez del Rio, B. Golosio, A. Simionovici, and A. Somogyi. Spectrochim. Acta, 59B(10-11):1725–1731, 2004. [3] V. A. Sole, E. Papillon, M. Cotte, P. Walter, and J. Susini. Spectrochim. Acta, 62B(1):63–68, 2007 Experimental All measurements were performed at beamline L of HASYLAB, Hamburg. The beam was unfocused and monochromatized using a multilayer monochromator at varying energies, depending on the sample. The intensity of the beam was reduced using exit slits and a 1mm All absorber. The X-ray fluorescence and photons were collected using a Vortex Si drift detector. Conclusions In order to validate the code, a comparison was made between simulated and experimental spectra of standard reference materials (NIST and Goodfellow), revealing excellent agreement for K- and L-lines and good agreement for M-lines. The importance of a correct simulation of the cascade effect is demonstrated. Future work will focus on improving the detector response function, as well as on the incorporation of electron- matter interactions (Bremsstrahlung and impact ionization).

A general Monte Carlo simulation of ED-XRF spectrometers. New developmentslvserver.ugent.be/xmi-msim/poster_ICXOM2011.pdf · 2011. 10. 3. · A general Monte Carlo simulation of ED-XRF

Download PDF Report

Upload
others
View
22
Download
0

Embed Size (px)

Citation preview

A general Monte Carlo simulation ofED-XRF spectrometers. New developments

Tom Schoonjans1, V. Armando Solé2, Manuel Sanchez del Rio2,Claudio Ferrero2 and Laszlo Vincze1

1-X-ray Microspectroscopy and Imaging group (XMI), Ghent University, Belgium2-European Synchrotron Radiation Facility (ESRF), Grenoble, France

IntroductionSince their inception during the Manhattan project, Monte-Carlo simulations have been used for a variety of applications in physics, mathematics, engineering, finance and more. When applied to X-ray fluorescence spectroscopy, it is well known that Monte-Carlo simulations are useful for predicting the spectral response of samples irradiated with an X-ray beam of given characteristics. This requires the development of a dedicated computer code that simulates the histories of a large number of individual photons, whose trajectories in the system are modelled as a number of straight steps. At the end of each step, an interaction occurs, leading to a change in direction, energy and polarization state. A photon’s trajectory is terminated when it leaves the system, or when the detector captures it. If detection occurs, one count is added to the appropriate channel of a virtual multichannel analyzer. The algorithm is greatly optimized through the introduction of variance reduction techniques.

XMI MSIM•Software package dedicated to the Monte-Carlo based simulation of ED-XRF spectrometers•Based on the work of Vincze et al.•Update of the physical data (cross sections, line energies etc.) through linking with xraylib•Simulation of M-lines and cascade effects (radiative and non-radiative)• Improved detector response function: pulse pile-up and fluorescence/Compton escape peaks•Support for multithreading (OpenMP) and multiprocessing (MPI)•Callable as a plugin from PyMca: quantification of XRF data•Written in ANSI-C and Fortran 2003•Makes use of GNU Scientific Library, libxml2, HDF5, GTK+...•Distributed under the GNU General Public Licence: www.github.com/tschoonj/xmimsim

References[1] L. Vincze, K. Janssens, and F. Adams. Spectrochim. Acta, 48B(4):553–573, 1993.[2] A. Brunetti, M. Sanchez del Rio, B. Golosio, A. Simionovici, and A. Somogyi.Spectrochim. Acta, 59B(10-11):1725–1731, 2004.[3] V. A. Sole, E. Papillon, M. Cotte, P. Walter, and J. Susini. Spectrochim. Acta, 62B(1):63–68, 2007

ExperimentalAll measurements were performed at beamline L of HASYLAB, Hamburg. The beam was unfocused and monochromatized using a multilayer monochromator at varying energies, depending on the sample. The intensity of the beam was reduced using exit slits and a 1mm All absorber. The X-ray fluorescence and photons were collected using a Vortex Si drift detector.

ConclusionsIn order to validate the code, a comparison was made between simulated and experimental spectra of standard reference materials (NIST and Goodfellow), revealing excellent agreement for K- and L-lines and good agreement for M-lines. The importance of a correct simulation of the cascade effect is demonstrated. Future work will focus on improving the detector response function, as well as on the incorporation of electron-matter interactions (Bremsstrahlung and impact ionization).

http://www.github.com/tschoonj/xmimsim