Is there limit on how small fusion/fission reactor can be?? in Science & Spaceflight Posted November 29, 2016 · Edited November 29, 2016 by RCgothic Nuclear engineer here. No, no fission reactor design can detonate like a nuclear explosion. When the reaction runs away out of control the fissionable material heats up, and there are several mechanisms by which that makes the fuel less reactive, bringing the reaction to a new equilibrium state. The fuel may melt, but if there's an explosion it's going to be as a result of other materials present in the core (hydrogen, steam, molten salt explosions). Designing a bomb to detonate (even to get it to fizzle) is very difficult. You need to convince most of the fuel to react before heat increase brings the reactivity down. This requires both a very dense concentration of fissile material (else neutrons won't propagate through the entire core fast enough) and a very low concentration of non-fissile nuclei (which absorb valuable neutrons). Additionally, the need to exclude non-fissile nuclei also generally excludes the use of a moderator. A moderator is a material that slows down the 'fast' neutrons emitted by fission events to a 'thermal' level which more readily react with fissile nuclei. Without a moderator the fissile material is less reactive, so yet greater density of fissile nuclei is required. This all add up to very high enrichment, typically 95-98%. Even fast reactors don't normally get this high. Most reactor fuel is uranium oxide (UO2) enriched to about 5%, although the presence of the oxygen atoms makes the effective reactivity even lower. Finally, if you attempt to make critical assembly casually, it will just heat up as portions go critical before the full mass. Therefore a very rapid change of geometry is required, either compression or gun type in order to set off the final detonation. Reactors on the other hand are designed not to explode! Not only do they lack any means to effect the final geometry change, sufficient fuel enrichment, and also have far too many foreign nuclei in the way, they are carefully designed not to operate in dangerous reaction regimes. They do this by manipulating several types of criticality: In a sub-critical assembly the reaction is not self sustaining, and if the reaction was previously critical or supercritical the reaction power will be reducing. A critical assembly is one in which the number of neutrons released is precisely as many as is required for the reaction to be self-sustaining at its current power level. A supercritical assembly is one in which each reaction increases the neutron flux. The reaction thus grows exponentially. A power plant must operate in all these regimes. A plant that could not go supercritical could not start up. By adjusting the number of neutrons absorbed in the reactor the power level is controlled. However there are two further types of criticality which are extremely important to the design of reactors, referring more to the response time than whether the power level is changing: In a prompt critical (supercritical) reaction, enough neutrons are immediately released in each fission reaction to sustain further reactions. The timescale of this process is on the order of the travel time of the neutron between reactions (milliseconds). This is the type of reaction required for a bomb, although for reasons discussed above it would still not cause a nuclear detonation in s power plant. The fuel gets (potentially extremely, damagingly) hot and the reaction slows/stops. The speed with which prompt criticality changes the power level of a reactor makes it impossible to control, and reactors are always* designed so that they cannot go prompt critical. Reactors will always absorb too many neutrons, even with all control methods withdrawn. The other type of criticality is delayed critical. A quirk of the fission reaction is that whilst each fission event creates neutrons, so too do the fission products a couple of seconds later as they decay. (It is for this reason a prompt critical reaction cannot be simply critical - if fission neutrons are enough to be self-sustaining, the delayed neutrons will later make it supercritical). If the reactor is operated such that on fissile neutrons alone the assembly is subcritical and delayed neutrons make up the difference to critical or supercritical as required, then the exponential coefficient is on the order of seconds and minutes rather than milliseconds. In conservative designs, reactors can take hours to build up to full power, leaving plenty of time for manual and automated control systems. And all that is why the worst that can happen given total control/coolant failure is a meltdown and not a mushroom cloud.** *The Soviet RMBK design can in certain situations, which is why Chernobyl had a prompt critical excursion when it was messed about with by people who didn't know what they were doing. The heat build up caused a steam explosion and graphite fire. Annihilation of the cooling systems caused the core to melt. **Ok, conventional explosions can also cause mushroom clouds, but you know what I mean.