I’m curious to hear what folks think about modeling X-ray detectors at high fidelity. Here’s some background:
- 4x quasi-identical scintillation detectors operating 10-~200 keV
- Tungsten collimation and shielding (30 degree FoV)
- Aluminum enclosure, SiPM readout
- Balloon flight ~40 km altitude
My question is: what is a sensible stopping point for this kind of setup to get an accurate instrument response matrix for converting incident photons to counts? I’m fine with adding some info ad-hoc or after a simulation run—I’d rather decouple the effect of atmospheric scattering anyway.
I’m thinking a zeroth-order model using mass attenuation coefficients would be a good start as a sanity check. Then first-order would be one scintillator at a time with everything in its optical path. Second order would be including things like steel screws and copper nearby to account for scattering.
Does it make sense to go to the detail of including steel screws? What sorts of averaging techniques do folks use to speed up these simulations, or make them more robust?
Thanks,
William
I’m thinking a zeroth-order model using mass attenuation coefficients would be a good start as a sanity check.
Sanity check for what? Are you optimizing anything or just need a response now? You have specified an energy range, altitude, enclosure, shielding, and FoV.
Mostly this is because the typical workflow to start with is to modify an existing example that is closest to your application. There are several x-ray specific geant4 examples including three that are x-ray detectors in space environments, xray_telescope, xray_SiliconPoreOptics, and xray_TESdetector. These are published and validated examples so you can be somewhat confident in the physics lists and environments. So look through those and pick one. The challenging part will be getting your specific detector geometry correct. Some of those examples use gdml files but you probably can get pretty close with approximations using simple geant4 solids.
I’d adapt one of those examples and then have runs across the celestial sphere (theta and phi) that are “evenly spaced” in that space (non-trivial problem so I recommend HealPix or the like) with your geometry and physics with the exception of scintillation physics for the SiPM, just get the raw energy deposited.
Enabling the scintillation and light coupling physics could be done separately since those simulations tend to be extremely runtime hungry. The examples there would be in the extended/optical directory. The bigger concern with SiPMs in space will be accumulated radiation damage causing an increase in dark counts over time. That is likely more important to account for than the scintillation physics directly. advanced/microelectronics can give you some ideas but ultimately you will likely have to analytically model that or just look up rad effect studies with real data.