Unexpected anisotropy in elastic scattering of low-energy neutrons at finite temperature

Physics Processes, Models and Cross Sections
1

Dear Geant4 experts,

I am currently simulating elastic scattering (hadElastic) of low-energy neutrons (in the meV range), and I have encountered unexpected behavior related to thermal vibrations.
I would appreciate your insights on this issue.

The simulation is based on a modified version of the B4 example.
I am using Geant4 version 11.3.2.
In the DetectorConstruction class, I define a 1 cm × 1 cm × 1 cm hydrogen sample using G4Material.
The world volume is set to Galactic, and the hydrogen sample is placed at the origin.
The physics list is set to FTFP_BERT_HP.
In the PrimaryGeneratorAction, I define an incident neutron with a kinetic energy of 25 meV, originating from the position (0, 0, -100 mm) and directed along the momentum vector (0, 0, 1).
In the SteppingAction, I record the final momentum vector of the primary neutron (trackID = 1) after interactions.
I store both the raw (non-normalized) and normalized vectors.
The total number of scattering events** that occurred within the sample is also recorded.
These data are passed through EventAction and RunAction and written into a ROOT file.

I ran the simulation at three different sample temperatures: 0 K, 300 K, and 10000 K, to study the temperature dependence of neutron energy transfer and scattering angles.
For simplicity, I analyzed only events in which the neutron scattered exactly once.
As shown in the attached plots, I observed unexpected behavior.
The figures show the x (left), y (center), and z (right) components of the final momentum vector (non-nomarized) of neutrons that scattered once.
Black lines indicate 0 K, red for 300 K, and green for 10000 K.
At 0 K, where the sample has no thermal motion, neutrons do not gain energy from the sample.
As expected, the distributions of momentum components are nearly flat around 0, with the z-component boosted due to the neutron’s initial direction along +z.
However, at finite temperatures (300 K and 10000 K), I found that the x- and y-components of the momentum vector exhibit different distributions.
This is surprising because the simulation should be symmetric around the z-axis, and such anisotropy is not expected.
Could anyone advise me on what might be causing this asymmetry and how to resolve it?

Thank you very much in advance.

Geant4 Version: 11.3.2
Operating System: macOS 12.7.4
Compiler/Version: Apple Clang 14.0.0
CMake Version: 3.26.4

unnamed
unnamed1798×575 29.6 KB

1 Like
3

Dear @Shusuke_Takada

Welcome to the Geant4 forum!

Maybe @T-loic could help on this matter?

Best,
Alvaro

4

Thank you for looping in @T-loic.

I’d really appreciate any guidance or suggestions on this matter.
Please let me know if I can provide any additional information to help clarify the situation.

5

Dear Geant4 experts,

I would like to follow up on this report regarding an unexpected anisotropy in the elastic scattering (hadElastic) of low-energy neutrons.

I am using Geant4 11.3.2 with FTFP_BERT_HP and simulating 25 meV neutrons scattering on a hydrogen target.
The geometry and beam configuration are fully symmetric with respect to the z-axis.

In my SteppingAction, I directly record the neutron momentum using:
track->GetMomentum();
and fill histograms with the raw values, without any additional manipulation on the user side.

However, at finite temperatures (e.g. 300 K and 10000 K), I observe that the distributions of the x- and y-components of the final momentum are different.
At 0 K, the x and y distributions agree as expected.
This anisotropy is therefore very surprising.

Since I am only plotting the values directly provided by Geant4, I am wondering whether this could be caused by:

  • a bias in the random number sampling for thermal motion,
  • or a dependence on the computing environment.
    Could someone please check whether the same behavior appears on another machine or setup?

We are currently performing neutron scattering experiments on hydrogen and aim to use this simulation for experimental data correction, so resolving this issue is very important for our work.

Thank you very much for your time and support.