Yes, @paulromano is right. It is also clearly written: " [For G4EmDNAPhysicsList], Gamma interactions are based on the Geant4 Livermore/EADL97 models and they are included by default in all G4EmDNAPhysics constructors.*", so based on the old Cullen EPICS data.
In the same subject, I would like to add a note on this which may be useful in the future. One can perfectly modify the cuts in his Geant-4 application, e.g. I did the test with these modifications in my input files, so for secondaries,
SetCutValue(0.02 * eV, “gamma”);
SetCutValue(0.02 * eV, “e-”);
SetCutValue(0.02 * eV, “e+”);
and these modifications for transport cuts,
G4double lowLimit = 0.02 * eV;
G4double highLimit = 100.0 * MeV;
I tried something close to @Shimwell question; a 3eV photon beam irradiating a Uranium slab. I used the G4EmLivermorePhysicsList (so EPICS data for all particles), and it works (the output file for 10 events).
Even if the data (or Geant-4) seems to allow it, this is strongly discouraged by Cullen and Salvat, here is a reminder of what Cullen said (page 5-6) about it for the EPICS-2017 data:
The electron and photon data used in EPICS are limited to ATOMIC DATA, elemental, cold, neutral, isolated atoms; this is in line with its intent for use in engineering applications. This limits the data to be used ONLY at higher energies. I recommend that it should not be used in applications below 100 eV; Francesc Salvat recommends not below 1 keV . Users should be aware that the EPICS data extends to low energy (eV range) ONLY to allow data such as anomalous and coherent scattering factors to be calculated (this involves an integral over the entire energy range of the photoelectric cross sections). At lower energies atomic effects become progressively more important and invalidate the designed features, again: EPICS data are limited to ATOMIC elemental, cold, neutral, isolated atoms. This does not include NUCLEAR data, which at high energies can be substantial. WARNING: CAVEAT EMPTOR: Do not try to use this data at lower energies (eV range) – if you do, your results can be very inaccurate, and you will have nobody to blame except yourself.
In conclusion, I find safer the OpenMC cuts.
Hope this helps @Shimwell.