I am making a simulation of energy deposited of some radioactive sources in scintillation detector. I have some results, but I don't have the same distribution energy between experimental spectra and simulated spectra. I do not know why.
For instance, the spectra of Cs-137 is shown. However, it does not have the same resolution.
Someone who can help me how to improve the simulation.
The scintillation detector is of NaI:Tl.
Hi Sebastián Alejandro Sarasti,
The most significant difference is the width of the photoelectric peak at 662 keV. This width occurs after gamma-ray energy is transferred to the scintillator crystal. Your simulation may not include this part.
The number of scintillation photons generated, when 662 keV is totally absorbed by the NaI:Tl crystal, is statistically distributed. It is also a stochastic process whether each scintillation photon is successfully detected by a photodetector (photomultiplier tube or semiconductor device) or lost.
There are two ways to improve the distribution. One simple way is to assume the energy resolution of the detector (crystal + photodetector) and perform a convolution with your results. You can find the energy resolutions for NaI:Tl in literature: typically ca. 7% at 662 keV [*].
* https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=&ved=2ahUKEwi-vdXluuvuAhVCG6YKHXfEDKYQFjAGegQIBRAC&url=https%3A%2F%2Fwww.crystals.saint-gobain.com%2Fsites%2Fimdf.crystals.com%2Ffiles%2Fdocuments%2Flabr-performance-summary-2019.pdf&usg=AOvVaw0ZxE0oS0xXg3PVP8-Qok0g
Another detailed method, if necessary, is to add simulation to track the behavior of the scintillation photons.
The other differences seem to be related to Compton scattering; since the NaI:Tl crystal is sealed in a package, sometimes the energy of the gamma-ray may be partly lost by Compton scattering in the package before it enters the crystal. Then, the gamma-ray energy is not 662 keV but less than 475 keV for the detector. Similarly, Compton scattering from surrounding desks, walls, and other objects may enter the crystal and increase the proportion of that component of the experimental data.
Dear Sebastián Alejandro Sarasti
In addition to the detailed answer of Kengo Shibuya I would like to add that also the optical properties of your scintillator (wave length distribution that you are giving to Geant4 as parameter to simulate scintillations) could also impact you final spectrum.
Kengo Shibuya thanks for your advices. I think that it is neccesary to make a convolution of results after Monte-Carlo simulation with Python.
I was not really sure about making that.
Thanks a lot!
Zakaria Aboulbanine thanks for your advice, but I am simulating energy deposited in the crystal. I understand that scintillation properties is only necessary when you want to simulate emission of low energy photons (Scintillation process and Cherenkov process).
My aim in GEANT is recording the hits in crystal emitted by radioactive sources, but not recording the photons emitted by the crystal.
The simplest way to solve your problem is usage of the gaussian blur with parameters according experimentally determined ones. E.g. use G4RandGauss::shoot(EnergyDeposition,sigma); where sigma=a*sqrt(EnergyDeposition), EnergyDeposition-amount of energy deposed in the sensitive volume while histogram filing in the end of each G4Event
- Badges
- Science topic
- Similar topics
- Computer Communications (Networks)
- Simulators