Page 1 of 1

Dose rate vs. Dose (Geant4-Serpent2)

Posted: Sat Jan 30, 2021 10:29 pm
by Ahmed_Naceur
Dear Community,

I have a very simple problem, an isotropic source (slab source) in front of a multi-slab geometry (cuboid here), and I compared Serpent's "x" dose profile to that of Geant-4.

With dr -200, Serpent gives, after normalization by the volume of the detector, and converting hour to seconds, a dose rate in Gy/s.

Geant gives an "x" dose profile in Gy/particle, after normalizing by the number of events.

I think to bring back the equivalent comparisons, it is a simple problem of normalization. In Serpent, I used set srcrate 1, which stands for one particle emitted per second. If I divide the Serpent dose by 1 particle/s, I should get Gy/particle. This result does not correspond at all to the dose profile or to the dose order of magnitude given by Geant-4. By the way, I tried using the flux distribution or even the speed of the photon in the middle (introducing medium refractive index), but it didn't work.

Do you have any suggestions? or a systematic way to have Gy/particle instead of Gy/s in Serpent.

In advance, thank you.

Re: Dose rate vs. Dose (Geant4-Serpent2)

Posted: Sun Jan 31, 2021 5:40 pm
by Ana Jambrina
The results for the dose rate should be in units of Gy*cm3/h (integrated over the volume), so dividing by the detector volume should yield the dose rate in standard units Gy/h (per source particle), provided that the source is appropriately normalized.
So, if you convert Gy/h to Gy/s, having set the source rate to 1 particles/s ('srcrate' = 1), it should result in Gy/particle.

Having said that, it seems not to be a problem related with units and/or normalization. The fact that the gamma spectra seems to be different suggest that something else is going on.

I’m not familiar with Geant4, but I can say that Serpent and MCNP show a good agreement in regards of photon transport. Even if it sound obvious I would suggest to verify the compatibility of both simulations. Some checks:
- Photon physics interaction models
- Photon cross-section data
- Simulation model definitions: source, material compositions and volumes, detectors, etc.

Regarding the photon dose rate detector:
- do not misinterpret the material linked to the response detector (dr <response> <material>) with the material detector itself (dm <material>); i.e. by setting dr -200 void, the response is not pre-assigned with a specific material (when the detector scores in a collision, the cross-section is taken from the material at the collision point - allowing calculating integral reaction rates over regions composed of multiple materials). In order to limit the spatial domain where the detector scores, add ‘dm <mat>’ option to the detector card (something like: det dose dr -200 void dm <mat>).
- for better statistics, if you are looking for dose rate at a discrete location (small region), define the detector as a super-imposed track-length detector.
- the attenuation coefficients are calculated based on nuclide composition, density and energy (check 'photon_attenuation.h' module looking for correspondence with Geant4 - with reference Hubbell, J. H. and Seltzer, S.M. (2004), "Tables of X-Ray Mass Attenuation Coefficients and Mass Energy-Absorption Coefficients." (version 1.4),

Re: Dose rate vs. Dose (Geant4-Serpent2)

Posted: Fri Jul 23, 2021 4:07 pm
by Ahmed_Naceur
Thank you so much Ana for your reply. And so sorry for the delayed response.

It is just a normalization error made in my post-treatment (sorry). We report an excellent agreement between Geant-4 and Serpent-2 in a multi-slab 3D reactor benchmark (MOX/UOX/Gd) and in medical physics (Rogers and Mohan multi-slabs benchmark for external radiotherapy and brachytherapy). Studied energies cover all clinical spectrum beam energies. Geant-4 calculations are fully-coupled, i.e. electron-positron-photon transport. Serpent-2 calculations are, of course, photons only + secondary models (TTB, annihilation and relaxation). For high energies, agreements are reported within material volume, excluding electron buildup effects and interface retrodiffusions. Thus, we are able to highlight Serpent's impact. Here, attached, an example of the excellent agreement for 3 MeV photon beam incident on homogeneous slabs.

Also, this study includes Geant-4 gamma only transport.

Other details will follow.

Thanks Ana.

Re: Dose rate vs. Dose (Geant4-Serpent2)

Posted: Mon Aug 02, 2021 1:09 pm
by Jaakko Leppänen
Also note that Serpent has an analog photon heating detector (mt -12) that calculates the exact energy deposition from collisions. This should be physically more rigorous compared to using a response function (mt -200) that converts flux to dose. The limitation with the analog detector is that it doesn't account for energy carried away from the region by secondary electrons (this would require an electron transport mode). All energy is instead deposited locally, so it may not be very accurate near the boundaries.

Re: Dose rate vs. Dose (Geant4-Serpent2)

Posted: Mon Aug 02, 2021 1:18 pm
by Ahmed_Naceur
This is exactly what was used in this study. I finally used mt -12, after a massive tests of other available options.