Pixel clustering

Hello,
I am simulating a 7x7 pixel matrix of a Timepix3 detector in Allpix square where the events are only primary particles (meaning that they are created only outside of the sensor). From the deposited charge in the sensor I noticed that there are some double pixel clusters where there is a distance of one (or even two!) pixels between the pixels that belong in this double pixel cluster. For example, a double pixel cluster could be consisted of an event in pixel coordinates (1 5) and (3 4). How is this possible? As I mentioned before I am only considering the primary particles (photoelectric) so it is not a Compton event or a fluorescence that is occurring inside the sensor. Thank you very much in advance.
Best wishes,
Pinelopi

_Geant4 Version:Geant4-v11.1.2
_Operating System: Linux Fedora

after compton/rayleigh scattering, a primary/incident gamma is still a primary. could that be the case here?
see this or that