Overestimation of electron yield in gas ionization by fission fragments of U-235

Physics Processes, Models and Cross Sections
1

Geant4 Version: 11.3.2
Operating System: Debian GNU/Linux 13 (trixie)
Compiler/Version: gcc 14.2.2
CMake Version: 3.31.6

Dear Geant4 experts,

I am simulating the ionization of a gas detector by fission fragments from neutron‑induced fission of U‑235. My goal is to obtain the charge spectrum (number of ion‑electron pairs created) in a segmented gas volume. However, I consistently observe a shift towards a higher number of electrons compared to reference expectations. I have tried many standard adjustments, but the overestimation persists.

Geometry:

A thin layer of U₃O₈ (fission source) enclosed in steel tubes. Surrounding it, 20 concentric gas‑filled cylindrical layers (detectors) with Ar + 4% N₂ at 5 atm (density ~8.04 mg/cm³). Neutrons are generated from a primary source (thermal‑like spectrum) and fission occurs in the U₃O₈.

Scoring:
I score the energy deposited by fission fragments (and their secondaries) inside each gas layer. The number of electrons is then derived as N_e = E_dep / W, with W ≈ 26.4 eV for Ar/N₂.

The reconstructed electron spectrum is systematically shifted towards higher electron multiplicities than expected.
Physics lists:
FTFP_BERT_HP (with and without replacing EM physics with G4EmStandardPhysics_option4 ). I also added G4IonPhysics and G4IonPhysicsPHP.
Energy deposit treatment:
I used GetTotalEnergyDeposit() directly. And I also subtracted GetNonIonizingEnergyDeposit() – this did not improve the spectrum.

Below are the spectra that I obtain for two configurations of ionization chambers and the spectra given in publications.
Fission chambers CFUR and CF4 (my and reference spectrum and characteristics of this chambers):

image
image1225×820 64.5 KB

Additionally, in my previous simulation of the CF8 fission chamber, the charge spectrum matched the reference from the article, with only a minor shift - far less pronounced than the one observed in the present configuration.

I would be grateful to hear your thoughts on these issues.

2

I have some naive questions first since I haven’t worked with fission fragments. Fission fragments have enormous LET especially near the end of their range. I could easily see that there would sufficiently high ionization density that you could get plasma screening effects i.e. so many ions in a local region that the electric field is effectively “reduced” leading to undesired recombination losses. This is an effect related to the interactions of ions with each other which Geant4 out of the box does not model.

To first order the signal height from your readout will be proportional to the number of collected ions. The effect I would describe would lead to the simulation consistently overestimating the signal height since it “misses” the losses. I imagine this effect is tied strongly to the gas used, pressure, applied electric field, etc.

I briefly scanned online and found this cern paper which cites having to run this kind of correction for AR + CF4 fission fragment chamber.

This might mean you either perform these kinds of corrections afterward or use the real data to calibrate the simulations ad hoc provided that the shape is generally correct (just shifted or stretched relative to the real case).