According to [1], when a pi- stops in a hydrocarbon, it should be captured on hydrogen only 1-2% of the time. In our simulation of the NOvA detector, which is mostly hydrocarbons, pi- are instead absorbed by hydrogen 20% of the time. In any such capture process, hydrogen is a special case since the, e.g., pionic hydrogen atom is a neutral system whereas other pionic atoms are charged. For mu- capture, Geant4 correctly simulates that the muon is never (or nearly never) captured by hydrogen, so I suspect that the pion capture and muon capture processes don't use the same code to handle attaching the particle to an atom. Please let me know what further detail would be useful. [1] D. F. Measday. The nuclear physics of muon capture. Phys.Rept., 354:243409, 2001.
In Geant4 no pionic atom is formed, as it is for muons. The pion is directly absorbed into the nucleus without an atomic cascade. This is likely the cause of the difference you see. It would be rather straightforward to develop such a pion capture model based on the muon case, but there is currently no manpower to do it.
