According to recent articles like this one, Bill Gates is v-e-r-y enthused about his investment in TerraPower, which hopes to put a breakthrough reactor on the market sometime around 2020 ... or maybe later.
Indeed Terrapower's proposal (and that's all it is at this point, with no reactor in operation) appears worth investigating.
A "traveling wave reactor" sounds like something out of the seventh dimension, but the general idea has been discussed since 1958. TerraPower worked out further details and filed for a patent. The company might be more easily understood by the general public if they had called it CandlePower.
About candles: When you buy a candle, the wax embedded in the new wick has so much surface area that a kitchen match can volatilize the wax to a flammable gas, which ignites. The wick's flame melts a pool of wax at the top of the candle; this liquid moves up into the wick's fibers by capillary action (aka "wicking") and the cycle continues smoothly and slowly. The wick carbonizes at the top, and the candle shortens until the last of the wax is smeared across the top of your candleholder. A candle works because a wave of heat travels across the body of the fuel and makes it available, a little bit of time, for the reaction to continue. A candle is a stable system: yes, a candle can start a house fire if it tips over, but it can't explode like a gasoline lantern.
A TerraPower wave reactor wouldn't get shorter over the years, but as with a candle, raw material would be converted to fuel slowly and continuously, as a wave of radiation moved very slowly through the reactor's volume. Imagine a long container with fuel rods filled with "depleted uranium," which is nearly all U-238 rather than fissionable U-235 (which is also A-bombable in high concentrations, if a critical mass can be brought together very quickly).
In between the fuel rods are coolant channels; a high-performance coolant is vital since the wave reactor would produce a lot of heat.
But it can't get going without help. That requires mounting a small conventional nuclear reactor at one end. This is necessary to get the fuel production going. The startup reactor's neutron radiation converts the nearby U-238 to plutonium-239 through a chain of reactions. Pu-239 is plenty fissionable, so much so that bomb designers can't use the simple methods that are suitable for U-235 bombs. The new plutonium fissions and that produces more neutrons, which converts more U-238 into plutonium fuel. According to TerraPower a wave reactor could turn out heat for a hundred years or more, depending on size, and most of the energy would come from relatively cheap U-238. About 99.3 percent of natural uranium mined from the earth is U-238, little of which is being used today.
Present-day reactors can do the same transmutation trick with U-238, if someone takes the trouble to chemically separate plutonium out of spent fuel rods. Once separated, the plutonium can be machined into fuel rods to go back in reactors. But reprocessing is highly controversial for the wastes produced, and some miscreants might be tempted to divert the separated plutonium for nuclear weapons rather than reactor fuel.
A principal talking point for wave-reactor advocates is that everything happens inside a big box, without the need to pull out extremely radioactive fuel rods and send them off to a robot-operated refinery. Another attraction is that a wave reactor will burn natural uranium, which is plentiful. Here's a video and animation with more detail.
These sound great! Building a working prototype is a good idea, along with tests of other promising new-generation reactors. But there will be special hurdles for the TerraPower machine to cross. One is handling the coolant safely. This is projected to be liquid sodium because of the high heat. This molten element reacts readily, even violently, with other chemicals like water. It might be possible to immerse the wave reactor in a bath of liquid sodium, avoiding problems with pipe malfunctions.
Will the extremely long-duration of the reactor cause unmanageable radiation damage to the metals used? At question here is what happens to metal atoms after getting whacked by neutrons, hundreds of times per atom. It might be okay, but there is little evidence one way or the other past 30 years of constant exposure. Will off-gassing of xenon (a reactor byproduct) be a problem? Uneven heat transfer?
Here's a link to a good Bulletin of the Atomic Scientists article summarizing the work that will be needed to test out the safety of the traveling-wave and other off-beat reactors.
As one longtime observer of the nuclear field told me, "The wave reactor is revolutionary. Its new materials and features may all be faultless, leading to a reactor that operates more safely and economically than today's reactors." But if significant shortcomings appear along the way, he cautions, "the wave reactor can actually drop below the performance levels of today's reactors."
Finally, this bit of really ancient history to give perspective: two billion years ago, a dozen or so natural nuclear reactors ran for hundreds of thousands of years in present-day Gabon, Central Africa. The Oklo reactors could operate because uranium back then was four times richer in U-235 than today's uranium ore (some of today's reactors run at this low level of enrichment, 3 percent). Second reason it was possible: the rock formation was soaked in groundwater, which acted as a neutron moderator.
Using evidence from the unusual mix of isotopes in the rock, nuclear physicists surmise that a given reactor ran for about two and half hours, then stopped because the groundwater boiled away (probably making all kinds of weird underground noises if anybody had been around to listen). Then after everything cooled, water returned, and short-lived fission decay products broke down, each reactor cranked up for another short run. This went on for umpteen thousands of years. Heat output: somewhere around 100 kW, thermal. That's a tiny fraction of the thermal output of today's commercial reactors, but rather impressive for a reactor that started and ran, hands-off.