MSC with limited step size in thin layers

I was testing multiple scattering (MSC) in several material layers on the wall of a tube. I was interested in comparison of the overall MSC effect in the following two cases:

  1. the step size is limited by material boundary - material is crossed in one step
  2. the step size is limited in the code to ~1/5 of the layer width - material is crossed in several steps.
    For this test I turned off all magnetic fields, all processes except for MSC, Coulomb scattering and Transportation. I used a test variable theta defined as angle between the momentum before the first step in a given material and momentum after the last step in a given material (in case 1 it is the same step). The angle is taken in the plane given by the two momentum vectors. I also compared the x component of the direction of particle before the first step and after the last step in a material.
    The material thicknesses and step sizes used in this test are:
  • Mylar - thickness 36um - step limit 6um
  • Copper - thickness 50nm - step limit 10nm
  • Gold - thickness 20nm - step limit 4nm
  • Gas mixture - radius 4.86mm - step limit 0.6mm

The results can be seen in the attached pdf. The effect of MSC in the two cases is the same for Gas mixture, Copper and Gold. However, there is a large difference in Mylar - the overall effect of MSC is smaller when doing multiple short steps wrt one long step.
I tried several step limits in the Mylar. The shorter the step size, the larger the difference wrt long step.
I tried to change the material to Plexiglass or Carbon instead of Mylar, but there is still a difference. However, when changing Mylar to another gas (at high pressure to obtain comparable MSC effect), there seem to be no difference between the case with one long step and multiple short steps.
Finally, I tried using the Mylar in a thinner layer - 3um and compared the MSC effect with one long step and with step size limited to 0.5um. They seem to be in a reasonable agreement.

I used Geant4 version 10.5 on CC7 machine. Particle type is positive pion coming from the decay of 75GeV K+. The momentum of pi+ is mostly between 5 GeV and 75 GeV. We use custom made Physics list, but for MSC the default in EM Opt0 is used (WentzelVI model).

Has anyone come across something similar? Do you have an explanation for the observed differences?
Thank you.

MSC_stepLimit.pdf (1.8 MB)


it is not easy to comment, you need todescribe conditions of simulation: beam energy, particle type, and also Genat4 version and Physics List. You may get similar results also with $G4INSTALL/examples/extended/electromagnetic/TestEm5

The results may be expected or not depending on this information.


Thank you, I updated the post.