Forgive me if I am wrong, but are you trying to generate a detector response function, i.e. the pulse height spectra you expect from your scintillator for a given source? As I understand it the common method of generating a detector response function is to tally the energy deposition of the primary particle, and then apply a statistical broadening function based on measured data of your existing system. If I was not wrong, this topic comes up every now and then here is someone who might have had a similar question as you, Detector Response Function .
This 2013 paper by Hartwig and Gumplinger might be relevant, the authors used Geant4 to simulate the response of a liquid scintillator detector Simulating response functions and pulse shape discrimination for organic scintillation detectors with Geant4 - ScienceDirect .