Probably a stupid car question...

[Lissa]

I agree with Lissa on this point. One can say "choose a better nuclear reaction", but nature really doesn't offer one. You can say "choose better shielding materials" but even the best possible still take pretty good thicknesses to work.

The Catch-22 is that if you use low energy reactions, you don't need as much shielding but then the pile has to be much larger to get a sustained reaction. If you use high energy reactions, you can make the pile smaller, but you then need more shielding. Optimizing just for weight, you still end up with something more suitable for railroad engines than cars or trucks. And that optimization involves using some nasty materials and insufficient containment for purely mechanical accidents.

There has been recent work showing that fission product of 235U induced by thermal neutrons show points where fission results in almost stable isotopes. What we do now is a mostly uncontrolled reaction where we need to deal with a potpourri of fractured unstable byproducts, where most of the dangerous stuff (highly energetic) decays within days to weeks. The challenge is to create a fission reactor that results in controlled more stable "waste", and you will greatly reduce the need for shielding.

Your fission fragments are a crap shoot on what you get from the reaction, likewise, you are going to get neutrons and gammas that are lost to the system (we always look at getting ~2.3 neutrons per fission, you need 1 neutron to continue the chain reaction at steady state, that other 1.3 is typically "lost"). And a thermal neutron is a neutron moving with a velocity around 2 km/s (2200 m/s, but close enough for this). A fast neutron (which is what breeders and weapons use), is typically moving at slightly relativistic speed, several 10s Mm/s. In order to get neutrons from fast (cause they're born fast from the fission) to thermal, they have to bounce off a variety of other, typically lightweight atoms - best being Hydrogen preferably the Protium isotope, until they drop from those near relativistic speeds to around 2 km/s (and best lose is going to 1/2 per interaction, so it's going to take a number of interactions to finally drop by a factor of 1000). In order to get that number of interactions, you have to have quite a lot of moderating materials (this is why we typically use water or carbon in burner style reactors, the typical power generating kind). So, size is going to be an issue no matter what, the difference is how much of a neutron flux you're going to need and at what power level your reaction is going to be at.