Questions about the discrete process

In order to get more detailed dose results, I tried to write a simulation of nano metal oxide based on the AuNP example from the dna folder. I used the discrete Livermore model in this example, but I found that the energy deposit in the nano-particle was much less than the one calculted using the emLivermorePhysics constructor. The correct logical volume is set to the G4Region as I checked and the e_livermoreIoni was detected inside the nano-particle. What could cause this difference? It seems the probability of ionization using discrete Livermore is much lower than the Livermore constructor. I wonder if there is any solutions or whether I have made a mistake.
The code of setting physics model in G4Region is below.