What is a fast reactor?
For terminology, see the NRCs nuclear glossary.
To understand why we should or should not build fast reactors, it is necessary to understand what they do and how they differ from other technologies.
The difference between fast reactors and ordinary ones is that the fast reactor uses the neutrons released by nuclear fission in uranium directly, without slowing them down first. This makes the fast reactor more difficult to build, in the sense that more concentrated fissile material is needed in the core to keep it running, and extra steps must be taken to avoid excessive neutron leakage. This has nothing to do with safety, only how to keep the reaction going. The advantage is that a fast reactor can "burn" all of the heaviest elements (the actinides), instead of mostly U-235, some Pu-239 and a tiny bit of U-238.
This feature of the fast reactor has multiple very interesting consequences, in fact a carefully designed fast reactor can eliminate every one of the traditional drawbacks of nuclear energy: Nuclear waste, nuclear proliferation, safety and the lack of renewability.
A fast reactor in combination with a small fuel recycling plant can burn all of the uranium fed to it, instead of just 1%, as current reactors do. This reduces the volume of the waste by a factor of 20, or 100 if depleted uranium is defined as waste. In addition, reprocessing plants for fast reactors have a much easier job than a normal reprocessing plant - only the true ashes of the fission process, the fission products, must be removed to keep the reactor running. The radioactivity of the fission products decreases to below the level of the natural uranium ore in 300 years, as opposed to the 300000 years which regular spent nuclear fuel needs to reach the same level. Safe storage of a volume of 1/20 the size of the already small volume of normal spent nuclear fuel for a period of just 300 years is trivially simple.
Nuclear weapons proliferation:
Fast reactors can be very resistant to nuclear proliferation.
First, no nation has ever first aquired nuclear weapons by using plutonium from power reactors. All of them have chosen either enrichment of uranium or dedicated plutonium production facilities camouflaged as research reactors. India and the US have demonstrated that plutonium from power reactors can be used in bombs, but that was only after having aquired a significant nuclear weapons expertise. Bombs made from plutonium from power reactors are very radioactive, which complicates production and necessitates heavy uranium radiation shielding and active cooling of the bomb core to keep it from cooking its conventional explosives trigger. Also, such a bomb has an unpredictable explosive yield, and the unavoidable gamma radiation it would emit makes long distance detection of the weapon possible. No military would want a bomb like that when they can more easily build more reliable bombs that are easier to handle.
Second, in a reprocessing plant for fast reactors, plutonium is never separated from all of the other actinides and the recycled fuel also always contains so high levels of radioactive impurities that it must be handled remotely. Trying to go near it without heavy shielding would mean instant death. Even after a successful theft, which would require a remote manipulation device and a large truck containing many tons of radiation shielding, further processing would be required to separate the plutonium. The high irradiation levels of a fast reactor also makes the plutonium produced in such a reactor even less useful for weapons than regular reactor grade plutonium.
Third, the volume of recycled fuel is much smaller than in the thermal reactor counterpart, so there is less material to steal or divert to a weapons program.
Fourth, the small size of the reprocessing plant means that it is practical to co-locate reactor and reprocessing facility. Spent fuel will be reprocessed on-site, which means that no material is supposed to leave the premises for reprocessing. The facility also produces its own fuel. This means that transportation of dangerous material is almost eliminated, the only exception is newly created fuel for other fast reactors. Any sudden increase of truck traffic would be extremely suspicious.
Even if someone despite all of these obstacles managed to obtain sufficient plutonium to build a bomb, the result would be a very impractical and unreliable weapon.
In addition to the above, fast reactors remove the need for uranium enrichment. This closes the easiest path to nuclear weapons. Finally, stockpiled civilian and military plutonium, plus weapons grade uranium can be quickly destroyed in a fast reactor. The destroyed weapons material can then be reprocessed into more fuel.
Fast reactors can be completely passively safe. This was repeatedly demonstrated in the EBR-II test reactor. The reactor was subjected to both the Three Mile Island and the Chernobyl accident scenarios without any mishap. The reactor simply shut down by itself without operator intervention, and the passive decay heat removal systems performed as they should.
Most fast reactors are cooled by liquid metal, and therefore the coolant is not under pressure. This simplifies construction and enhances safety. Injecting water into the high-pressure, boiling Fukushima reactors turned out to be rather challenging - such a scenario is impossible in a liquid metal cooled reactor. All pumps, valves, heat exchangers and other equipment exposed to radioactive primary coolant is contained inside of the reactor pool. This, together with the lack of pressure, eliminates the possibility of leaks of radioactive coolant.
Fast reactors make uranium renewable:
It sounds implausible, but it is in fact true. The energy density of uranium is so enormous, and the fuel efficiency of carefully designed fast reactors so high, that in effect, uranium is a renewable resource. This is because so little fuel is required that the price of electricity is almost completely independent from the price of fuel even when the uranium is extracted from sea water. There is enough uranium in the oceans to provide all the fuel we might ever want until the sun engulfs the earth in some 4 billion years. Erosion continually sends more dissolved uranium into the sea than we can possibly burn, so we would not even be able to change the uranium concentration in the oceans.
No new uranium would be needed for a long time, we already have enough spent fuel and depleted uranium stockpiled to keep us going for several hundred years.
One implementation of the IFR, the successor of EBR-II, is GE Hitachi's S-PRISM reactor. This design is ready for building, and provided it is coupled with a suitable reprocessing facility, it has all of the above mentioned properties.