Restrictions

1. Effects due to the breakdown of Born approximation at low energies are ignored (but the Coulomb correction is now included):
2. as suggested by Ford and Nelson [], for very low energy photons ($E\le 2.1\; MeV$ ) the electron energy is approximated by sampling from a uniform distribution over the interval $m\to 1/(2E)$ . The reason for this suggestion is that the sampling method used in EGS and in the earlier GEANT versions becomes progressively more inefficient as the pair threshold is approached. This is not true for the sampling method outlined above (the efficiency of the method practically does not depend on the photon energy), but we have chosen to keep this approximation;
3. target materials composed of compounds or mixtures are treated identically to chemical elements (this is not the case when computing the mean free path!) using the effective atomic number computed in the routine GSMIXT. It can be shown that the error of this type of treatment is small and can be neglected;
4. the differential cross-section implicitly accounts for pair production in both nuclear and atomic electron fields. However, triplet production is not generated, and the recoil momentum of the target nucleus/electron is assumed to be zero.

PHYS220