In order to complete the simulation work of PGNAA, I simulated the (n,γ) reaction of neutrons and Ti, and the resulting γ spectrum. Unfortunately, fluka’s simulated results don’t match the real ones. The simulation results of mcnp can be consistent with the real results.
Clearly, there are a lot of peaks here that shouldn’t be there. At 1381.7(5.18b), 6418.4(1.96b) and 6760.1(2.97b)keV, the intensity ratio of the peaks was also wrong.
I really don’t understand why the gamma spectrum calculated by fluka is wrong when both MCNP and fluka use the ENDF database? Ti.flair (7.7 KB) PGNAA cross section.pdf (20.2 KB)
As the FLUKA manual mentions in the 10.2.3.3. Radiative Capture section, the radiative capture does not use the γ-information from the ENDF database(s) but rather it invokes FLUKA own γ-de-excitation model, in order to keep the correlations of the emitted lines and energy/momentum conservation.
The FLUKA γ-de-excitation model is using the level-scheme of the residual nucleus e.g. 48Ti(n,γ)49Τi, which explains the many more lines you see in the plot (which are physical).
On the other hand, the ENDF database for 48-Ti(n,γ)49Ti contains the inclusive spectra of 57 discrete prompt gamma lines up to 3.99MeV, together with a continuum spectrum in a form of a histogram with a resolution of 20keV up to 8.142MeV, smearing out all other lines.
In the FLUKA γ-de-excitation model, if the excitation energy does not fall on a known energy level, FLUKA is using a statistical approach to create an initial random transition to a known level. This initial transition is creating a spurious 8.142MeV gamma and alters the relative ratio of the 6.4 MeV and 6.76MeV gammas with respect to the 1.38MeV gamma.
Thanks to your report, we have identified a possible future solution so as to keep the benefits of the FLUKA γ-de-excitation (correlations, E,p conservations, and the full reproduction of the level scheme), while ensuring relative branching ratios for known isotopes.
If I didn’t use the ENDF database, the gamma spectrum would be even more outrageous.
On the contrary, the gamma spectra obtained using the ENDF database are more reliable. However, the emission intensity cannot match the standard cross-section.
How should I use FLUKA to obtain the correct gamma spectrum?
The second plot you produced reflects the structural limitation of the groupe-wise treatment, where the gamma energy is discretized in 42 groups (as discussed in the manual). Note that nobody advised you to switch to it: the point of @vasilis was that, even with the use of the point-wise treatment and the ENDF library, the gamma generation is not constrained to reproduce the gamma inclusive spectrum of the latter, in order to preserve correlations and conserve energy on a event-by-event basis. This is the reason for the discrepancy of the line intensities in the first plot, whose cure - that shall still respect the aforementioned conservation - is not yet finalized/available.
I’m desperate to get an accurate prompt gamma spectrum. Is there a way to do this?
I personally understand whether it is possible to turn off the “standard-de-excitation model of FLUKA” and just use the gamma rays given in the ENDF database. How do I close this model?
