There is a great deviation between energy deposition and MCNP calculation

sphere 5um-water.flair (2.1 KB)
My calculation model is as follows:
For a 5um spherical shell, the resulting energy deposition is 130% off:

I adjusted the spherical shell to 5cm and the error was reduced to 45%. Please have a look at the program. What is the reason?

Did you divide the USRBIN result by the region volume, as it has to be done in the special case of a region mesh (see the manual, note 13 of the USRBIN card)?

The volume is 62.8322cm^3, 8.7661E-11/62.8322 = 1.39516E-12, obviously smaller
The Dose given by USRBIN card and THE F6 unit of MCNP are both GeV/g, I don’t think it is necessary. At the same time, the difference of 5cm software decreased significantly

You are missing two fundamental points.
First, in case of a region mesh (as the one you’ve asked for), the result given by the USRBIN card is not in GeV/g, rather GeV cm^3 /g, because FLUKA does not know the value of your region volume and so cannot normalize by it. This is what I recalled you above and is indicated in the manual. So, in order to get the average dose in GeV/g over your water layer, it’s definitely necessary to divide your FLUKA result by the volume of the latter.
Second, especially when simulating the 5 micron layer case, the transport thresholds play quite a significant role, and in principle should be lowered as much as possible, not to artificially stop in water secondary electrons that may leave and not deposit all their energy there (in reality, the presence of other materials also matters, since water cannot stay suspended in vacuum). You took the default thresholds of PRECISIO, i.e. 100 keV for electrons, and so got 1.4 E-12 GeV/g as you just calculated above. Making use of the EMFCUT card and setting 1 keV for electrons, which is their minimum limit in FLUKA, one gets 4 to 5 E-13 GeV/g, namely about three times less, due to the increased number of escaping electrons. If you set the electron threshold at 2 MeV, i.e. you do not allow them to leave water (which is an artificial assumption), you will get 4.1 E-10 GeV/g (three orders of magnitude more!), which matches well your MCNP result.

Thanks for the teacher’s answers and the satisfactory results were obtained according to the suggestions. From your answer also harvest knowledge, thank the teacher!