Dear @chenchcc
Welcome to the forum!
Without knowing the details of your simulation and your objectives, I will try to address some points that might be relevant.
First of all, if your aim is to perform a reactor criticality study, then already the choice of code is questionable, since FLUKA does not run criticality problems. This is further compounded by the multi-group treatment of neutrons <20 MeV, although point-wise transport will become available with the next major release later this year. It is up to you to judge whether your objectives are compatible with the above limitations.
About 252Cf, indeed FLUKA does not generate the prompt fission neutrons, as far as I am aware. This could be addressed by directly generating neutrons from an appropriate spectrum via a SOURCE routine. The 252Cf neutron spectrum is well-documented and could be implemented without excessive effort. (As a side note, the A<247 limit in HI-PROPE is applied by Flair as a check on the input, but it is not actually enforced by FLUKA.)
Now to the Sb-Be source. I would suggest that also in this case the best way to go would be to directly generate an appropriate neutron spectrum via BEAM card or SOURCE routine, since the spectrum is almost monoenergetic around 23 keV, with a small component around 378 keV. As a second option, you could generate the principal 124Sb photons (1691 and 2091 keV, or just the 1691 keV), activating photonuclear reactions with a PHOTONUC card and biasing the photon interaction length in the BeO via a LAM-BIAS card. My guess is that you would not get enough neutrons inside the reactor for a study of reasonable duration, and little would be gained anyway, but you can try. Things will be even more dire if you use a 124Sb ISOTOPE source as primary (always with PHOTONUC and LAM-BIAS as above), given the low probabilities of the interactions (i.e. the appropriate decay gammas of 124Sb inducing a (γ,n) reaction in 9Be).
Finally, keep in mind that in order for any scoring to provide results when running a decay calculation (RADDECAY card in semi-analogue mode) you need one or more DCYSCORE cards properly associated to your scoring cards. Otherwise getting all zeros is the expected outcome.