There is some good natured rivalry, and perhaps some not so good natured rivalry between the backers of the Liquid Fluoride Thorium Reactor (LFTR) and the backers of the Integral Fast Reactor (IFR). Both are so called Generation IV technologies. Each technology is associated with an American national laboratory. The LFTR is the linear descendant of the Oak Ridge National Laboratory Molten Salt Reactor research. The IFR is an advanced Liquid Metal Fast Reactor. IFR technology was developed at Argonne National Laboratory. Readers who are interested in the backgrounds and relative merits of LMFBR and MSR technologies can consult an essay titled, Nuclear Power and Energy Security: A Revised Strategy for Japan, or my review of that essay. The essay authors, Lawrence M. Lidsky and Marvin M. Miller of the Massachusetts Institute of Technology, do an excellent job of laying out the issues, and offer some very real insights onto some of the problems of the current nuclear industry, and also, issues related to the major technological options available to break out of those problems. For example, Lidsky and Miller,
The cost and complexity of the systems needed to deal with the danger of severe accident makes the LWR a poor choice for large central station power plants. Ironically, it is the LWR’s high power density, the very reason it was chosen for submarine use, that is its Achilles heel. Even a 10-second interruption in the supply of cooling water at the surface of a fuel rod can lead to local overheating and irrevocable, cascading damage to the reactor core. As a result, the LWR must rely on defense-in-depth, a system of diverse and redundant backup devices, to guard against such an event. This is a widely used technique, but defense-in depth can not, by itself, guarantee absolute safety; it can only reduce the probability of a serious accident. All nuclear power plants,because of their cost and potential for off-site hazards, have a very low “acceptable” probability of failure. The larger the plant, the lower the acceptable probability of failure. Because the consequence of failure is so large in gigawatt-scale plants, LWR’s have been forced to employ engineered safety systems that promise unprecedentedly, and perhaps unattainably, low probability of failure.
The safety related requirements of the LWR drives up its cost. Lidsky and Miller, who were undoubtedly aware of the IFR and would have included it in their LMFBR class, wrote:
The sodium-cooled LMFBR was the device that was intended to replace the LWR when mined uranium supplies became prohibitively expensive. The LMFBR was chosen over other breeder reactor designs because it was, in theory, capable of very short fuel doubling times, shorter than that of any competing reactor design. The doubling time is the time required to produce an excess of fuel equal to the amount originally required to fuel the reactor itself. In other words, in one doubling time there would be enough fuel available to start up another reactor. In the absence of mined uranium, only a short doubling time would, it was believed, allow nuclear power to grow fast enough to compete with alternative sources of power. Unfortunately, the theoretical advantages of the LMFBR could not be achieved in practice. A successful commercial breeder reactor must have three attributes; it must breed, it must be economical, and it must be safe. Although any one or two of these attributes can be achieved in isolation by proper design, the laws of physics apparently make it impossible to achieve all three simultaneously, no matter how clever the design. The fundamental problem originates in the very properties of sodium that make the short doubling time possible. The physical characteristics of sodium and plutonium are such that a loss of sodium coolant in the center of the core of a breeding reactor (caused, for example, by overheating) would tend to increase the power of the reactor, thus driving more sodium from the core, further increasing the power in a continuous feedback loop. The resulting rapid, literally uncontrollable, rise in reactor power is clearly unacceptable from a safety standpoint. This effect, the so-called “positive void coefficient” can be mitigated by, for example, changing the shape of the core so that more neutrons leak out of the core, but this immediately compromises the reactor’s breeding potential. Safety and breeding are thus mutually antagonistic. This situation can be alleviated to some extent by making radical design changes, but these changes lead to greatly increased costs, and make the reactor prohibitively expensive. Even if the LMFBR could meet its original, highly optimistic, operating goals and the LWR/FBR power cycle were put into operation, it is unclear that the goal of energy security would be achieved. As discussed in the following sections, the measures that would have to be put in place to protect all parts of the fuel cycle against terrorism would have very high social costs. Equally important is the increased risk of accidental or maliciously-induced technological failure. Compared to light water reactors operating on the once-through fuel cycle, the breeder fuel cycle is much more complex and error-prone. This implies a higher probability that the entire nuclear system or a significant fraction thereof might need to be shutdown because of a generic problem, e.g., with sodium containment, in the reactors or an accident in one of the reprocessing or fuel fabrication plants that serve the system.
They conclude,
Strong support for plutonium recycle, with its associated technical risks and societal costs, in the face of increasing evidence that alternative strategies are superior, is clearly counterproductive.
There was an alternative to the plutonium breeding LMFRB as Lidsky and Miller note:
Thorium fuel cycles have also been promoted on the basis of lower long-term waste toxicity and greater proliferation resistance, . . . The initial rationale for introduction of the thorium cycle was the perception that it was more abundant than uranium, and that it could be used to breed U-233, an isotope with superior properties for use in thermal reactors. However, Its terrestrial abundance is not germane to Japan’s energy security concerns because Japan has no indigenous source of thorium and it is hard to imagine a scenario in which uranium is cut off but thorium is available. Conceivably, the use of U-233 in an advanced reactor could reduce the possibility of a common mode failure of a reactor fleet consisting of LEU-fueled LWRs and HTGRs. The Molten Salt Reactor would be a strong candidate for consideration for this role, with a solid research base and an international support group, . . .
IFR backers would counter that they have greatly improved LMFBR safety. I would agree with the claims that the IFR has a number of natural (passive) safety features, which coupled a system of barriers, makes the IFR reasonably if not totally safe. But the totality of IFR safety features would tend to make IFR construction more expensive, and would tend to increase the IFR cost. In addition one IFR safety feature, a large pool of liquid sodium which surrounds the IFR core, appears to be a double edged sword as far as safety is concerned. Although the pool contributes to core temperature stability in case of a coolant system failure, the presence of such a large amount of highly flammable sodium in close proximity to the IFR core, could lead to a huge disaster in the event of a core rupture. Core ruptures would not happen often, but never is a very strong word. Only offensive nuclear safety coupled with passive safety, can come close to guaranteeing "never," and if you put sodium into a solid core reactor, offensive nuclear safety is clearly impossible.

Of course, there are added barriers that could be erected to prevent disaster in the event of an IFR accident. The IFR can be placed in a metal egg, with an internal atmosphere that would not burn when it comes into contact with Sodium. This would add to the expense and complexity of the IFR, and while diminishing the danger, it will not make it go away completely.

One of the technological advantages of the LFTR is its relatively small size. Small size means relative small containment area. That means a lower construction cost. The IFR includes a tub full of molten sodium that is many times larger than the core. In addition machinery to defuel and refuel the TFR core must be included in the inner containment area. The same is true for the primary heat exchange, and the control rod mechanism, and some arrangement for shifting the "spent fuel" out of the core area. All in all we have got quite a lot going on in the containment parameter, which means that the outer containment structure would have to be large.

Further, even if IFR components could be factory built, a lot of onsite assembly would have to go on. Thus the IFR would quite possibly end up being as expensive or more expensive than light water reactors.

There are other issues that point to higher prices, if not higher than LWRs, at the very least, higher than LFTRs. Fast reactors require higher inventories of fissionable materials. I addressed this issue in a previous post, S-PRISM Scalability a Repose for Steven Kirsch.
In that post I noted:
In S PRISM related study "S-PRISM Fuel Cycle Study: Future Deployment Programs and Issues," suggested that as of the year 2000, four hundred tons of plutonium could be recovered from spent nuclear fuel. This in turn would provide enough plutonium to supply start up charges for twenty-two, 1520 MWe S-PRISM facilities with ab output of 33,440 MWe. That is about 12 tons per 1 GWe of reactor capacity.

Clearly then neutron speed has an adverse effect on reactor scalability.

On the other hand neutron speed also influences the fission rate per neutron absorption, this in turn influences neutron production. Pu-239 fissions 25% more often in a fast reactor than in a thermal reactor. On the other hand it still take more Pu-239 to maintain a chain reaction in a fast reactor than in a thermal reactor. Reactor physics tricks and fuel cycle also seem to influence start up charge size.

A recent discussion on the EfT form produced quite a lot of useful information. "Jagdish" reported that
Indian 500MW PFBR is designed to use only two tons of plutonium.

"Honzik" pointed to French research of epithermal/fast Thorium Molten Saalt Reactors. The French, modeling the use of transuranium materials from spent nuclear fuel, in a 1 GB reactor had calculated a need for 7.3 tons of fissile elements (87.5% of Pu (238Pu 2.7%, 239Pu 45.9% , 240Pu 21.5%, 241Pu 10.7%, and 242Pu 6.7%), 6.3% of Np, 5.3% of Am and 0.9% of Cm). Alternatively the reactior would require a start uo charge of 4.6 tons of U-2330.

Lars reported that
The minimum for unity breeding from the French group is 1.5 tonnes u233 / GWe.
Alex P noted:
the french design has an only radial, not axial, blanket, so for comparison I'd think that the fissile start-up in a LFTR with a fully encompassing blanket can be at least one tonn of u-233 per GWe, or even lower
David LeBlanc noted:
The French TMSR design running without graphite moderator needs upwards of 5 tonnes of U233 or 8 or more tonnes of fissile Pu. They could drop this somewhat if they just wanted to barely break even but not very much since they'll start losing too many neutrons that would migrate into the axial reflectors. In designs in which the blanket is nearly fully encompassing you can get by with much lower fissile concentrations. It is only speculation for now but based on early Oak Ridge studies using sphere within sphere designs I think we could probably get things down to 500 Kg of u233 or maybe even lower but 1000 kg is a fine for a conservative estimate. These designs with lower fissile concentration would also be fairly soft spectrums since the salt itself can do a modest job at moderating the neutrons.
The problem of plutonium in nuclear breeding should be noted. In thermal breeders plutonium suffers from poor neutron economy, while in fast neutron reactors plutonium neutron economy improves but does not compensate for the added requirement for fissile material. Radial and axial thorium blankets in a breeder appears to lower fissile demand by as much as 300%, but this principle appears to have been applied in S-PRISIM design.

The S-PRISM design would appear far less scalable than Epithermal or thermal MSRs. David LeBlanc's estimates are based on the use of blankets with Epithermal MSRs. If we estimate that 2 kgs of reactor grade plutonium from spent nuclear fuel about 1 kg of U233, 500 kgs of U-233 would be a similar startup charge to a ton of RGP. Thus the same amount of RGP that will start 33 GWe worth of S-Prism FBRs will also start 400 GWe worth of LFTRs. Clearly the LFTR offers scalability advantages over the IFR/S-PRISM.
The issue of core inventory places the IFR at a double disadvantage. First, it greatly limits the the number of IFR that can be started from existing inventories of spent nuclear fuel. Secondly, It increases core size. Larger core size means a larger containment structure, and, of course, higher costs. What of the potential for a small IFR?

I briefly reviewed the recent Argonne plan, Advanced Burner Test Reactor Preconceptual Design Report. This turned out to be something of a surprise, because it was something that IFR backers are telling us is not needed, a "proof of concept" project. It seems to me that we have been told that the IFR is ready for a commercial prototype. If the IFR is at the proof of concept stage, the argument that LFTR technology is less advanced, because the MSR technology is at a stage where a proof of concept reactor is possible, There are some significant differences. A LFTR of similar output is possible, but it might be considered to be not only a proof of concept, but also a commercial prototype. The ABTR appears to be a burner, designed to use primarily military grade Pu-239, with a little RGP thrown in. Further more it appears to be an old design that has been dusted off. Although the ABTR is capable of breeding with substantial core modifications, Argonne's primary intent is to run it within the conversion range. The highest breeding ration the ABTR is capable of is 1.07, no better than is anticipated for the LFTR.

Thus while IFR advocates such as Y. I. Chang, speak of high breeding ratios, in fact the first IFR prototype will likely not even break even in its typical operations, will not use typical Reactor Grand Plutonium (RGP) as fuel, and can only breed at best in a similar range to that of the LFTR. It does have a far lower inventory of fissionable materials than would be expected in a pure breeder. Don't get me wrong, the ABTR is an impressive reactor with ingenious safety features,. The core is placed underground, and the tub of "cold" liquid sodium is above the core, giving the ABTR a small foot print. I was frankly quite impressed with the ABTR safety features, and I would feel almost as comfortable living next door to it as I would living next door to a LFTR. But the ABTR is simply not a high breeding ratio, LMFBR. If the ABTR is the IFR prototype, a commercial prototype would have to follow, because a high breeding ratio RGP burner would be sufficiently different from the ABTR to require significant tests before the design is released for commercial operation. Of course an IFR R&D program could skip the ABTR entirely and build a high breeding ratio, RGP burning prototype. Y. I. Chang claims both high breeding ratio and a low inventory IFRs are possible, but would require larger cores than the ABTR. No doubt, but larger cores mean larger buildings, less factory based as opposed to field based manufacturing, and thus higher costs. The high breeding ratio IFR will experience some trade offs, and one of them might be proliferation concerns. I might be easier to convince about safety, than other potential IFR critics.

Thus it seems unlikely that the IFR could compete with the LFTR in terms of manufacturing costs or in terms of scaleability. The claim that the IFR represents a more mature technology than the LFTR, appears vastly over blown. Indeed, the LMFBR appears to represent a significantly greater technological challenge than MSR technology, and given the effort that went into the development of the ABTR, a proof of concept reactor, the very large American and world wide investment inLMFBR technology, has yielded far fewer results than a similar investment in MSR technology would have.

What is problematic about the ABTR is its complexity. Complexity makes for expensive manufacture. Because the ABTR study assumes a high percentage of Pu-239 in its fuel, and a not very impressive conversion breeding ratio, it has been impossible for me to determine the actinide inventory for a high breeding ratio, RGP burning IFR. I might review other Y. I. Chang reactor studies, for further information, but I would note that IFR advocates are starting to claim that the LFTR cannot breed, or at least a that a breeding commercial LFTR is impossible. These root of these claims appears to be concern because the LFTR can be scaled to large numbers just as fast as it can be churned out of factories. I am unaware of any documented rational to back up the LFTR can't breed claim.

The IFR backer attitude is understandable. Just two years ago, the MSR was considered dead, and suddenly, out of left field comes a grass roots movement calling for LFTR development, a movement which at the moment appears to rapidly be gaining traction. Suddenly at the very moment when the energy situation appears to be demanding the IFR, the LFTR appears as a candidate energy black swan. LFTR advocates are pointing to a relatively mature energy technology that is more sustainable than the IFR, can be built in large numbers at a low price, and holds the potential to revolutionize the world energy picture. A bunch of unknown upstarts are threatening to take the keys to the candy store away from the IFR community. There must be something wrong. If the LFTR cannot breed, the the keys stay in the hands of the IFR community. Well maybe. IFR costs and scaleability are still open questions.