Friday, March 29, 2013

Burning Metal, Part 2

Those Magnificent Men and their Atomic Machines

Burning Metal: The Los Alamos Molten Plutonium Reactor Experiment and the History of the Fast Breeder

Part II

With special thanks to Prof. R. M. Kiehn

Note on Notation
Money is going to be talked about a lot, but the value of the dollar has been different from year to year. In each case, unless otherwise specified, values will be given in the amount for the year in question, followed in parentheses by the equivalent value in 2011 dollars.

Change of Plans
But, as Los Alamos was preparing the third LAMPRE fuel loading, the nuclear energy landscape was changing.

In 1958 the Atomic Energy Commission (AEC) had set a target of making nuclear electricity cost-effective in regions with high fuel costs. In 1962, they decided they had just about reached that goal. Four private nuclear reactors and two joint public-private projects were producing power by 1962, and another ten were under construction, most of them Light Water Reactors (LWRs). In a landmark report to the president on civilian nuclear power in 1962, the AEC recommended that light water technology be handed over to the private sector. There were obviously further improvements to be made, but they would be evolutionary and incremental, and thus the domain of private enterprise rather than the AEC labs. Properly encouraged by the government, the light water reactor would be cost-competitive with coal and gas by the 1970s except in very low fuel cost areas. The AEC would focus instead on advanced reactor concepts, particularly breeders.

The report forecast a dramatic expansion of nuclear power. Assuming electrical consumption continued growing at the same rate as the past few decades, there was simply no other way to meet demand.

Figure 11: Estimated US Energy Demand, 1962
1 Q/yr = 33,000 GW; Curves A, B, and C denote different growth scenarios
Note: graph shows energy demand, not electricity demand

US electricity demand had been doubling every decade since World War I, and total energy use had been doubling every 30 years since 1900. Nuclear power had arrived just in time: by the year 2000, half of the United States' total energy supply would have to be nuclear – there simply wasn't enough coal. And a growing percentage of those reactors would have to be breeders – there wasn't enough uranium to fuel that many burner reactors. The report estimated that, by 1980, American light water reactors would be producing 40 GWe, equal to 10% of national electrical capacity, and breeders would begin to be introduced into commercial operation by the mid-80s.

The report was agnostic as to what type of breeder should be developed, but the AEC had long favored the Liquid Metal-cooled Fast Breeder Reactor (LMFBR). The technology was further developed with one prototype operating and two more under construction in the US, and the need for a massive buildout made fast breeders' short fuel doubling times particularly attractive. Besides providing the initial fuel loads for more fast breeders, they would also supply older burner reactors for which U-235 was no longer available.

The Experimental Breeder Reactor-2, designed by the AEC's Argonne National Laboratory, and Enrico Fermi Unit 1, built by the Power Reactor Development Corporation with AEC assistance, were both coming online in 1964. The PRDC and ANL program dwarfed the LASL molten plutonium effort, in both funding and facilities. Outside the United States, both the United Kingdom and the Soviet Union were experimenting with LMFBRs of similar design, including the Dounreay Fast Reactor in the United Kingdom and the BR-5 in Russia, both of which went online in 1959.

In 1964, Milton Shaw was named the director of the AEC Division of Reactor Development. Shaw was a protege of Admiral Hyman Rickover, the father of the naval nuclear propulsion program that had birthed the LWRs now rewriting America's energy landscape, and he believed in the same management principles: pick a design and build it. Alvin Weinberg, head of Oak Ridge National Laboratory and a veteran of the Manhattan Project, later recalled, “Milt was like a bull. Extremely hard working, terribly demanding, singleminded. He enjoyed the confidence of [AEC] Commissioner Jim Ramey, and of Rep. Chet Holifield [of the Joint Committee on Atomic Energy] – so his position within the AEC was unassailable.” Dixy Lee Ray, chairwoman of the AEC in the 70s under Nixon, said that Shaw “had very strong opinions, and he was sure that if he were simply stubborn enough and refused to compromise long enough, he could have his way on any issue.”

In the view of Shaw and his supporters in the AEC, the AEC should focus its resources on the solid-fueled LMFBR, the breeder design closest to commercialization. Other reactor programs, for advanced burners or other breeder designs, were distractions. And, with the backing of Jim Ramey and Chet Holifield, Shaw could get what he wanted.

In 1964, the AEC cancelled the planned third LAMPRE run. LAMPRE was decommissioned over the next year, although much of the hardware was left in place. Officially, the LAMPRE was considered unnecessary, with the funds and talent better spent on its planned successor; however, right up until cancellation Los Alamos had been planning to use the LAMPRE for testing materials for the second reactor and training its operators, suggesting this may have been an excuse.

For the moment, however, the Los Alamos Molten Plutonium Program remained alive, with funding stable at about $4.2 million ($30.1 million). However, in what was probably a concession to the changing political climate within the AEC, the project's fuel alloy development group started work on advanced plutonium nitride and carbide ceramic fuels in 1962 – a technology that, while potentially quite useful in solid fuel breeders, has little obvious relevance to the molten plutonium program.

The Fast Reactor Core Test Facility
While LAMPRE-1 was under construction, plans for the LAMPRE-2 had changed. By 1959, LAMPRE-2 had become the Fast Reactor Core Test Facility (FRCTF). The FRCTF was to be a test stand for liquid fuel reactor cores; initial plans called for it to be online by 1963. The facility would provide primary and secondary sodium coolant loops, steam generator, cover gas system, breeding blanket, control system, and remote-operated hot cells. The core capsules, containing the actual reactor cores as well as fuel handling and reprocessing systems, would be replaceable, allowing different concepts to be tested in the same facility. Where LAMPRE had been limited to testing container and fuel materials, the FRCTF could be used for a variety of purposes, including testing techniques for purging gaseous fission products and in-pile fuel reprocessing, developing breeding blankets and control mechanisms, testing new container materials under high power operation, and determining breeding ratios.

The facility would contain two pits, into which prototype reactor cores would be inserted. The pits would be surrounded by a thick outer breeding blanket of uranium-238, designed to remain in place for the life time of the pit. The FRCTF would provide six times the heat flux of LAMPRE and twelve times the neutron flux, allowing LASL to test ideas in an environment not dissimilar from what a molten plutonium power breeder would be like. At this neutron intensity, 3% of the tantalum container material would be transmuted into tungsten for every year of full-power operation.

By 1961, the initial plans had been delayed and scaled back. The steam generator was deleted; instead, the FRCTF would dump heat to air. The outer breeding blanket was also removed, although breeding could still be tested with blankets contained within the cores.

Figure 12: Fast Reactor Core Test Facility

The FRCTF reactor pits would be 33-foot-long stainless steel vessels in concrete-lined holes, about 6½ feet wide at the bottom with a 3½-foot-wide opening at the top. The first pit would be used to test reactors, while the second would be held in reserve – it was anticipated that, after several core loadings, the first pit would become unusable. Besides the test pits, a third pit was also provided for storing core vessels while waiting for their radioactivity to die down, as well as four smaller pits for holding radioactive components from dismantled cores. A 50-ton bridge crane would move the core capsules into and out of the reactor pits.

Each pit was lined with a shield of borated graphite and concrete, inside of which was a sleeve containing movable reflector control elements. At the center of the reflector sleeve would be an 18-inch-wide core thimble, which would contain the actual reactor cores under test. A sodium coolant loop would remove up to 15 MWth of heat from the core, and a second, completely separate loop up to 5 MWth from the reflector; the sodium loops would transfer the heat to separate secondary loops which would dump it to air through a heat exchanger. The pumping system would include two electromagnetic pumps that had been dockside spares for the USS Seawolf's S2G reactor, an abortive attempt to produce a high-power sodium-cooled reactor for submarine propulsion.

Figure 13: FRCTF Reactor Tank and Capsule

If a breeding blanket was to be tested, it would be added to the reflector section, probably using solid uranium alloy pins similar to the design used in the Experimental Breeder Reactor-2. In the long run the lab intended to develop breeding blankets using molten uranium alloy or pastes of uranium dioxide in sodium, but that would have to wait until the core was better developed. Just as in the LAMPRE, coarse control would be by moving the reflector, with fine control from separate reflector rods.

Construction of the FRCTF began in 1963 at Technical Area 35 on the Los Alamos grounds, with the aim of finishing work by May of 1965.

The Direct Contact Reactor
Looking past the FRCTF, the ultimate goal was a power breeder. Although stability had been the initial attraction for LAMPRE, liquid fuel also promised economic advantages. Fuel rods in a solid-fuel breeder are damaged over time by the intense radiation and buildup of fission products – the radioactive “ash” of the nuclear reaction. Also, many fission products are greedy neutron absorbers, stealing neutrons that could be used to produce more plutonium. As a result, the fuel rods have to be periodically removed, dissolved or melted down, processed to remove fission products, mixed with new fissiles from the breeding blanket, fabricated into new fuel rods, and returned to the core. Because of the intense radioactivity in the fuel, all of this must be done remotely, with operators behind thick shielding. This is an expensive process, especially with the ceramic fuel elements used in the type of LMFBRs the AEC was trying to develop in the late 60s and 70s.

In a liquid fuel breeder, by contrast, most of the gaseous fission products would simply bubble out of the fuel. And, without a crystal structure, liquid fuels are not damaged by radiation. The fuel would still have to be periodically reprocessed to remove solid fission products and lingering fission gases, but this could be done by pumping the fuel in and out of the core, without having to shut down for refueling.

Any power breeder based on the LAMPRE design could offer these advantages, but molten plutonium fuel also made possible a radically different method of cooling the reactor: directly mixing the coolant and fuel.

The concept originated at Oak Ridge, where scientists on the molten salt reactor project briefly considered a similar system using liquid bismuth coolant and molten salt fuel. In Los Alamos's version of the Direct Contact Reactor (DCR), the core would consist of a vat of molten plutonium-cerium-cobalt (Pu-Ce-Co) alloy. Fuel would be pumped into the core, where some of it would fission, generating heat.   It would then be pumped out of the core and thoroughly mixed with molten sodium, transferring heat to the coolant. The sodium and plutonium would be pumped into a separator that would remove plutonium entrained with the coolant; from there the sodium would go to a heat exchanger, and the plutonium be pumped back into the core. Several core volumes' worth of sodium would be pumped through every second. By directly mixing coolant and fuel, the system could be roughly ten to a hundred times as efficient in transferring heat as a conventional reactor that kept the two separated. This would allow a DCR core to be extremely compact. A more compact core is a more efficient breeder: the higher the percentage of the atoms in the core that are fuel, the higher the probability that a neutron will be used productively rather than lost to parasitic absorption in the structural material or coolant.

Figure 14: Direct Contact Reactor Core

The sodium would also carry with it many of the neutron-absorbing fission products, removing them from the core. The non-soluble products would be removed elsewhere, while the sodium soluble isotopes would be allowed to remain in the coolant, since they would amount to only 0.1% of the coolant after a full year of operation. Unfortunately, the coolant would also remove 33% to 90% of the delayed neutron progenitors – fission products that release a neutron as they decay – which would cancel some of the efficiency gains and shrink the generation time, the average time between fissions, making the reactor harder to control. But neither problem would be enough to cancel out the advantages of greater core compactness and highly negative temperature coefficient.

The AEC patented the DCR concept in December of 1961. A study that year by Atomic Power Development Associates for LASL proposed a power plant consisting of three compact direct contact cores, each using a Pu-Ce-25Co fuel alloy. Each core, along with its pumps, intermediate heat exchanger, and fuel-coolant separator would be contained underground in a 4-foot-wide, 60-foot-long vertical core capsule, which would be removable for replacement or maintenance. The core capsule would in turn be contained in a stainless steel thimble and surrounded by an inner breeding blanket of uranium oxide – sodium paste. Outside the inner blanket would be a similar outer breeding blanket surrounding all three cores. Sodium would exit the cores at 663 C and heat an intermediate sodium loop, which would then be pumped through a common steam generator to produce 524 C steam to turn a turbine. Each core would generate 227 MWth in the core and another 40 MWth in the inner breeding blankets, for a total power of 800 MWth, 300 MWe, enough to power 250,000 American homes.

Figure 15: DCR Capsule (Left) and Core (Right)

The use of several, smaller cores rather than one big core was necessary to ensure high neutron flux in the breeding blanket and because there were limits to how large the jet pumps and centrifugal separators could be made. It would also allow the plant to shut one core down for maintenance or replacement while still producing power with the other two.

System safety would be ensured by the high negative temperature coefficient of reactivity, and by eliminating excess reactivity from the control system – that is, the system would ordinarily operate “full throttle”, with control rods at their absolute minimum position, so that control rod errors could not accidentally increase reactor power. The control rods would only be used to shut the reactor down, while the core's normal operation would be controlled by varying the plutonium content of the fuel alloy.   In addition, since the core would be spherical in shape, it would be close to its most reactive possible geometry, so damage from external forces – such as an earthquake – would be unlikely to disarrange it into a more reactive state. Finally, the core's very hard neutron spectrum, besides improving the breeding ratio, would also eliminate the positive void coefficient from sodium boiling.

In short, the core would be designed so that any change in state – in control rod position, temperature, geometry – would cause power output to decrease, providing a very high degree of intrinsic safety. And, just in case something had been overlooked, the core structure was designed to survive power surges up to 2,800 C in temperature without failure.

If additional safety was desired, the core fuel and coolant sodium could be continuously reprocessed to reduce the amount of radioactive fission products in the reactor. This would both reduce the amount of radioactive material that could escape from the core in the event of an accident, and reduce the amount of heat generated by radioactive decay that would have to be dealt with in a loss-of-cooling accident. The process could be simplified by adding 0.1% sodium chloride to the coolant, which would cause it to absorb many of the other fission products that would otherwise remain in the fuel, minimizing the need to reprocess the core fuel.

Fuel reprocessing would be done in shielded cells fitted with remote-controlled equipment. Each cell would be four stories tall and fitted with a double set of all equipment, as spares, and retractable horizontal partitions that could isolate any one level. The first cell would contain tanks for storing fuel and blanket paste; the second a continuous still for core sodium and batch retort for blanket paste to prepare the material for reprocessing; the third dissolvers and crystallizers to separate the residue from the second cell into uranium nitrate, plutonium nitrate, and fission product waste. The fourth cell would contain equipment to purify plutonium nitrate to be sold or added to the core and to convert uranium nitrate into uranium oxide for return to the breeding blanket. The fifth cell would hold equipment for removing radioactive contaminants from the helium cover gas. The remaining cells would have machines to service and repair radioactive equipment from the reactor and other cells.

In normal operations, 12 gallons of paste would be drawn from the inner blanket assemblies each day, mixed with 2.5 gallons of fission product-contaminated sodium from the bottom of the core, and heated until the sodium boiled. The sodium gas would be drawn off and returned to the coolant loop, while the remaining solids would be processed by fractional crystallization into uranium oxide, plutonium oxide, and fission products. The first would be returned to the breeding blanket, the second mixed with cerium-cobalt powder to make fuel alloy, and the last disposed of. On paper, the core would be able to produce enough plutonium to fuel a second, identical plant in 15.7 years.

Figure 16: Blueprint of DCR Power Plant

The DCR's greater complexity would mean higher initial capital costs compared to solid-fuel LMFBRs. The study concluded the three-core, 300 MWe DCR would cost $64,071,000 to build in 1961 ($475 million), versus $56,170,000 ($417 million) for an advanced solid-fuel breeder extrapolated from the Enrico Fermi unit 1 reactor. However, the reduced cost of reprocessing would make up for the higher construction cost, yielding a net savings of 0.13 cents per kWh (0.96 cts). Given that the cost of building the solid-fuel breeder was projected at only $1,390 per kWe of capacity in 2011 dollars – compared to more than $2,000 per kWe for modern light water reactors, often much more – both figures must be considered highly optimistic. Similarly, the study's final electricity cost estimate of 0.51 cents per kWh (3.8 cts) from the DCR, and 0.64 cents (4.7 cts) from the advanced solid-fuel breeder, are both significantly below current costs for nuclear energy (figures vary but are generally more than 6 cts per kWh).

The DCR was not the only concept examined by Los Alamos. Other possibilities studied included cores of sealed or vented capsules; a spiral core design; and a paste core design. The capsule cores would be similar to the LAMPRE, scaled up in power by a factor of 100, and possibly with vents in the fuel capsules to allow gaseous fission products to escape, either into a takeoff line or into the coolant. Capsule core studies envisioned a similarly modular setup with 110 MWth cores joined together to form 500 – 1,500 MWe generating stations.

Figure 17: Vented Capsule Core

The spiral core consisted of alternating thin layers of static fuel alloy and flowing sodium coolant separated by tantalum. The capsule and spiral cores offered simpler, easier to build designs, but lacked the advantages of the DCR in efficient heat transfer.

Figure 18: Spiral Core

The paste core was similar to the DCR but used a paste of uranium oxide and plutonium oxide in liquid sodium as fuel instead of molten plutonium alloy. The paste core retained most of the advantages of the DCR and greatly reduced corrosion – a paste core reactor could be made of stainless steel instead of tantalum – but introduced new problems of fuel transport, gas purging, fuel density control, and reprocessing. The paste approach was ultimately abandoned when, in experiments, the paste proved to be so thick it was impossible to pump.

By August of 1960, LASL had already operated a mockup of the Direct Contact Reactor, using subcritical amounts of plutonium in a Pu-23Ce-29Co alloy. The mockup ran successfully for several days, and although some sodium was caught in the core, reportedly no fuel alloy got past the separator. Other mockups using mercury to substitute for the fuel and water for the sodium were also extensively tested to examine core designs.

Figure 19: DCR Pumping Experiment

LASL planned to build a critical version of the experiment, called the Pint Bottle Experiment or PBX, by May 1962. PBX would generate 20 kW of heat using the Pu-9.5Fe alloy from LAMPRE, and would be used to develop direct contact cores to test in the FRCTF. Three months later, the PBX had been delayed in favor of a non-reacting mockup, the PTA, to explore the nuclear and physical parameters of the system, although component procurement was underway for the PBX.

Figure 20: Gamma-Ray Camera Image of PTA Experiment

Unfortunately, after May 1962 no further mention is made of either the PBX or the PTA in the available documents, suggesting the PBX was never built, but no explanation is given as to why. However, interest continued in testing direct contact cores in the FRCTF. As of 1961 the plan was to test a direct contact core in the FRCTF by 1964, spend 4-5 years developing the technology, then move to a true prototype power plant. By the 80s LAMPRE's descendants would be providing safe, clean, inexhaustible energy at a price cheaper than coal.

Tantalum Again
With the Fast Reactor Core Test Facility on the horizon, Los Alamos interest was increasingly focused on Pu-Ce-Co fuels. The materials group also examined many other possible fuel alloys – in particular, Pu-Ce-Co expands like water when it freezes by 1% to 3%, which can damage its container. Considerable effort was spent looking for a quaternary alloy that would reduce this expansion, but without any significant success, and they ultimately returned to Pu-Ce-Co as offering the best combination of low melting point, variable plutonium content, and corrosiveness. By 1962 research into most other plutonium alloys had been abandoned; the capsules would just have to be made strong enough to withstand the expansion.

An exception, however, was made for ternary and quaternary alloys of uranium and plutonium, on which research continued. If a useable uranium-plutonium alloy could be found, new fuel could be bred in the core itself, eliminating the complexity and expense of a separate breeding blanket. But the best that was found was an alloy of uranium, plutonium, and manganese, which melted at 738 C, still too high for use in the FRCTF, and was far too corrosive to be useable. L. D. Kirkbride, part of the team working on fuel alloys, described it as “really a universal solvent. It will dissolve almost anything...” As a result, by 1965, work on ternary uranium-plutonium alloys had also largely ended.

Although still corrosive, the lower plutonium contents of the Pu-Ce-Co alloy were much more forgiving of their tantalum containers than the Pu-9.5Fe alloy or the plutonium-uranium alloys. However, the cobalt in the Pu-Ce-Co fuel alloy tended to react with the tantalum in the weld regions, forming a compound of tantalum and cobalt. (The Pu-9.5Fe fuel formed a similar tantalum-iron compound, but this had been less of an issue.) This layer would ordinarily remain in place, but if disturbed it might flake off, exposing fresh tantalum to react with the cobalt and gradually penetrate the container wall. It also tended to exacerbate mass transfer corrosion.

Figure 21: Tantalum-Cobalt Intermetallic Layer

Adding a carburized layer along the inside of the container wall reduced this effect, as well as further improving resistance to intergranular penetration. The carburization approach was found by accident: a while examining a fuel capsule that had exhibited abnormally strong performance, it was discovered that oil from the vacuum pump in the test rig had leaked onto it, diffusing into the tantalum and forming tantalum carbides that increased the capsule's corrosion resistance. When further experiments confirmed the effect, carburized layers were made a standard feature for fuel capsules.

Some work was also done on capsules made of alloys of niobium with zirconium or tungsten. The niobium capsules were less resistant to corrosion, but niobium absorbs only a fourth as many neutrons as tantalum, so a niobium-walled reactor could operate with a lower plutonium density. With less plutonium the fuel alloy was less corrosive, making up for the lower resistance of the niobium. However, the main effort remained firmly focused on tantalum and tantalum-tungsten alloys.

In fact, by 1965, between the lower corrosiveness of the Pu-Co-Ce and the improved resistance of the new carburized capsules, it took too long for capsules to fail for conventional corrosion tests to be practical. LASL found it necessary to build a special test rig, using resistance heaters to hold capsules in a vacuum chamber at temperatures far above the planned operating temperature of the FRCTF, up to 1160 C, greatly accelerating corrosion; the behavior of the capsule at lower temperatures was then extrapolated mathematically from its performance at high temperature.

Figure 22: Accelerated Corrosion Test Rig
Resistance Heater Not Shown

By 1966, in the view of the lab, the corrosion problem had been essentially solved. Extrapolations from the high-temperature experiments indicated that it would take about 20 years for the fuel to penetrate the new carburized tantalum capsules. Core lifetime was now primarily limited by damage from freeze expansion and from radiation: very intense ionizing radiation – such as experienced in the core of a nuclear reactor – will literally knock atoms out of position in a crystalline material like tantalum. This causes the material to become harder and more brittle, and ultimately to fail. The heavy neutron flux would also transmute tantalum in the container wall into tungsten. Fortunately, tungsten had so far proved to be quite resistant to plutonium corrosion. There was also concern that fission products, bombarding the capsule walls at high speed with the energy imparted by fission, might damage either the carborized layer or the Ta-Co reaction layer on the capsule wall.

With LAMPRE shut down and the FRCTF not yet operational, new fuel capsules were tested in Los Alamos's Omega West Reactor (OWR). The OWR was a light water moderated research reactor used by Los Alamos to provide neutrons for physics experiments. Two environmental cells were installed in the reactor in 1965, consisting of a double-walled stainless steel container holding liquid sodium, held at high temperatures by electrical heaters, to simulate the FRCTF environment. A container within the sodium could hold a fuel capsule for irradiation tests, with the space between the capsule and the container wall instrumented to detect leaks.

Figure 23: OWR Capsule Test Cell

After initial tests using stainless steel dummies and solid fuel slugs to measure gamma ray heat deposition and fission rates, capsules of unalloyed tantalum were tested, fueled with Pu-Ce-Co. At least five capsules were tested, in one case achieving a 2% plutonium burnup. Two capsules were made of unalloyed tantalum. Two others probably used tantalum alloyed with 5% tungsten, one with and one without a carburized layer, but the available documentation does not cover the final round of testing.

The Molten Plutonium Burnup Experiment
By April of 1965, the FRCTF was about 70% complete, and work was under way on the first core to be tested, the Molten Plutonium Burnup Experiment (MPBE).

The MPBE core would be similar to the LAMPRE, made up of 73 subassemblies, each consisting of a stainless steel can holding seven fuel capsules loaded with Pu-Ce-Co fuel alloy. Above and below the fuel capsules would be nickel reflector rods. 18 additional subassemblies would hold solid nickel reflector rods instead of fuel capsules. Sodium would enter the core at 480 C, and exit at 600 C.

Besides being closer to an already tested design, by this point, interest was shifting to capsule cores for power breeders in place of direct contact designs. The reason is not recorded in the available documents, but it may have been felt that capsule cores held lower technical risk compared to the more exotic options.

Figure 24: MPBE Core Cross-Section

The capsules would be 0.4 inches wide and 24 inches long, made of tantalum alloyed with 5% tungsten, with a carburized layer. Unlike LAMPRE, the capsules would have to be welded shut at both ends. The tantalum wasn't strong enough to be formed into a cup that long and thin, and would have to be fabricated as a tube with both ends open. Since the LAMPRE capsules' welds had had serious corrosion problems even when the fuel had only able to reach them by climbing the container walls, the MPBE capsules would use a clever design to keep fuel from reaching the weld. Plugs were designed to fit tightly into the ends of the capsule. After being inserted into the tube, they would be swaged to seal the tube onto the plug, and the ends welded by electron beam. Although some fuel would still reach the weld area through the very thin space between the plug and the tube, this would keep the amount to small quantities. The MPBE would be fueled with Pu-Ce-Co alloy, at a density of 6.2 to 8 grams of plutonium per cubic centimeter, compared to 16 grams per cm3 in the LAMPRE. This was still higher than would be used in an actual power breeder, which might be as little as 3 grams per cm3.

Figure 25: MPBE Fuel Capsule
Darkened lines denote welds

The MPBE's primary aim was to reach a very high burnup. Burnup is the percentage of fuel atoms that have undergone fission; the LAMPRE had reached only 0.5% in its second loading, while the test capsules in the Omega West Reactor reached 2%. The goal of the MPBE was to explore how the physical and chemical properties of the alloy changed as plutonium was replaced with fission products, how the fission products would effect the container material, how different power levels would effect the formation of fission gas bubbles, whether pressure from fission gas buildup would damage the container, and how the reactivity of the fuel would change over time. Assuming all had gone well, MPBE would have been only the first of many cores tested at the FRCTF.

The End of the Road
But the MPBE never entered operation. In 1966 Milton Shaw testified before the Joint Committee on Atomic Energy that, even though the FRCTF was almost finished, the MPBE would not be ready to be loaded until 1969. Construction on the FRCTF was ended, and the LASL molten plutonium power program redirected to focus on plutonium ceramic fuels for solid-fuel LMFBRs, at about half the former level of funding. The Los Alamos Molten Plutonium program was over.

The cancellation of LAMPRE was part of a broader sea change in the AEC's Division of Reactor Development. The organic-moderated program had ended in 1964. The heavy water reactor program was shut down in 1969. The gas-cooled reactor program shrank as projects like the Experimental Beryllium Oxide Reactor and the Ultra-High Temperature Reactor EXperiment were shuttered. The LMFBR program, however, was ballooning, taking up more and more of the development budget over time – and as its budget expanded, the money available for other projects shrank, and cancellations ensued. In some cases these cancellations were clearly justified, as development had shown the designs were unworkable or did not offer significant advantages over other technologies. In other cases, perhaps not.

Figure 26: AEC Power Reactor R&D Operations Funding in 2011 Dollars
Does not include capital funding, or propulsion or auxiliary power reactor programs
Light Water: Includes light water breeder and nuclear superheat
Heavy Water: Includes D2O-moderated, organic-cooled
Metal-Cooled: Early period includes sodium-cooled thermal reactors and LAMPRE
Cooperative: AEC public-private partnerships and projects with Euratom
Other: Includes projects applicable to multiple reactor types or otherwise unclassified

Even within the LMFBR project, almost all funding was directed specifically towards one form: a reactor with a loop-type coolant system using oxide ceramic fuel rods. Shaw directed the project from Washington, insisting on strict compliance to directives and cutting back any research straying into other designs. Even Argonne National Lab, the birthplace of the American LMFBR and designers of the EBR-I and -II, was not immune, with Shaw forcing changes in the EBR-II management in 1968, eliminating Argonne's plutonium reprocessing and fuel metallurgy programs, and transferring key development tasks – including development of the Fast Flux Test Facility (FFTF), the next major step in the breeder program – to Hanford and other groups. In 1968 the AEC planned the FFTF would go critical in 1974 at a cost of $87.5 million ($558 million), including some parts from the abandoned FRCTF, and would be used to test components for the first commercial fast breeder demonstration plant.

The change was not just driven by Shaw and his allies. The light water reactor was succeeding beyond the AEC's wildest dreams. Starting in 1966, based on positive reports from the first few commercial facilities, the utility industry began to invest massively in nuclear energy. A 1969 report revised the AEC's estimates for future nuclear generating capacity from 40 GWe by 1980 to 150 GWe, 25% of all American electricity. By 2000, fully half of American electricity would be nuclear. The great bandwagon market in nuclear reactors had begun.

Without the breeder, how could this vast fleet be fueled? Exploration was continuing to find new uranium deposits, but projected needs were growing faster than supply. The Edison Electric Institute estimated all low-cost American uranium ore would be used up by 1995. The breeder was the only way to meet this vast demand. In 1970, it was decided to move forward on a demonstration plant, the Clinch River Breeder Reactor (CRBR), in Tennessee near Oak Ridge. Construction was expected to cost $200 million ($1.1 billion), jointly funded by the federal government and a coalition of private companies. The first commercial breeder would enter service in 1984.

At least, that was supposed to happen. In 1972, the CRBR's price tag was reevaluated at $700 million ($3.7 billion), with construction planned to begin in 1979. The industrial partners refused to commit any additional funds to Clinch River after 1971, leaving the government paying for 85% of the cost. Elsewhere, the FFTF finally went online in 1982, after costing many times the original estimate. Milton Shaw was long gone by then, forced out by Dixy Lee Ray, Nixon's chairwoman of the AEC, in 1973. The AEC itself followed, reorganized into the Nuclear Regulatory Commission and the Department of Energy. But the Department and the national labs remained committed to the fast breeder project.

By 1978 the projected cost of the CRBR had reached $2.2 billion ($7.5 billion). At the same time, President Jimmy Carter had issued an executive order prohibiting plutonium separation in the US, in a bid to hold back nuclear weapons proliferation. But the whole point of the breeder was to be a plutonium-producing machine. Nonetheless, support in Congress and the Department of Energy kept Clinch River alive as costs spiraled and the administration looked for a way to kill it.

The enormous expense of the LMFBR program in general and Clinch River in specific ended most of what remained of the AEC's other reactor development programs. The Molten Salt Reactor program was terminated in 1976, and the Gas-Cooled Fast Breeder project in 1981.

In 1979, as the CRBR was stumbling towards cancellation, Los Alamos labs began dismantling what remained of LAMPRE. Technicians drilled holes in the remaining sodium coolant lines and injected polyurethane foam into them to fix the leftover sodium and plutonium in place. Then, remote handlers cut the pipes apart, sealing the ends with metal caps and roofing tar. The reactor vessel was lifted out and placed in a shielded cask, and the cask filled with concrete and buried on the Los Alamos grounds.

The election of President Reagan might have meant a change in course for the breeder. Reagan was an advocate for the fast breeder project and rescinded Carter's prohibition of plutonium separation. Construction on CRBR finally began in 1982. But by 1983 the estimated cost had officially reached $4 billion ($8.9 billion), and the Government Accounting Office calculated it would likely hit $8 billion ($18.4 billion) before it was actually finished. Congress finally refused to allocate any more money to Clinch River in 1984, zeroing out its budget. Most of the rest of the Department of Energy's advanced reactor R&D budget leaked away over the next decade. Fast breeder research survived at a lower level at Argonne National Laboratory as the Integral Fast Reactor program until 1994, when President Clinton ended that too.

Looking Back
What happened?

Part of the problem was technological optimism blinding engineers to the true costs of what they were trying to build. Part of the problem was increasing public fear over nuclear safety even in light water reactors, much less fast breeders with their flammable sodium coolant. But, more fundamentally, the fast breeder, solid or liquid fueled, just wasn't needed.

Interest in the breeder in the 40s had been driven by concerns over uranium shortages. AEC purchases for weapons funded a vast wave of exploration for ore, and by 1960 we knew there was more uranium than we had realized – enough to fuel the great bandwagon market in light water reactors. The need for the breeder was pushed back until the 80s or 90s, when the LWRs' appetite for uranium would use up available supplies.

But the demand for fuel for the LWRs had funded further uranium exploration. And it turned out there was even more uranium in the world than they had realized. A lot more. Today, known uranium reserves amount to 7.1 million tons, enough to power the present global reactor fleet for a century, and there is reason to believe much, much more may exist.

At the same time as uranium supplies were expanding, the great bandwagon market in nuclear power was petering out. The reasons for this were complex, but fundamentally, the reactors were just more expensive than expected. Cost overruns for the first generation of American nuclear reactors averaged 200%. It was later learned that the positive results from the commercial plants of the early 60s had been illusory; Westinghouse and GE had sold the plants below cost, assuming that they would ultimately make up the loss with economies of scale. Even before Three Mile Island, orders were falling off. The last American order for a nuclear reactor was placed in 1978, and it was cancelled before it was completed. In fact, no American reactor ordered after 1974 was finished.

Besides cost overruns, the demand for electricity itself had fallen off. The planners of 1962 had assumed the coming decades would see growth similar to the last ones, with electricity consumption doubling every decade. In fact, American use of electricity grew by only 50% in the 70s and 29% in the 80s. America today consumes about 472 GWe of electricity, of which only 101 GWe derives from nuclear fission, none of it from breeder reactors.

Other countries' breeder programs fared little better. Although the UK, France, Germany, Russia, Japan, and India all built fast breeder prototypes, none have so far progressed beyond demonstration plants. With the world awash in uranium and projected demand shrinking, profitability for the breeder was pushed back further and further. By the 80s projections were that it might be profitable by the 2020s. Although the dream of inexhaustible energy still found adherents among the scientists and engineers, it found fewer takers in Washington. The United States government has not built a new power reactor prototype since the Fast Flux Test Facility in 1978.

Looking Forward
But that may not be the case forever. The danger of climate change, evolutionary improvements in light water reactor technology, and the titanic energy appetites of the emerging economies have changed the energy landscape again.

The first new reactor orders in America in thirty years were placed in 2008, for a pair of Westinghouse AP1000 light water reactors. Today four new LWRs are under construction in the United States, construction has resumed on a fifth LWR originally ordered in the 70s, and the Department of Energy is sponsoring a joint public-private program to build up to three Small Modular Reactors at the Clinch River site – the first new government-sponsored prototypes in three decades.

Although this is a promising start, nuclear momentum in the US has slowed for the moment due to the fall in the price of natural gas from fracking, and it is unlikely any more reactor builds will start construction in the near future. But other countries are more than taking up the slack: in China, 29 reactors are under construction, and 51 more are planned. The Chinese government is investing aggressively in nuclear energy, and planning to produce 1,400 GWe using fast breeders alone by the end of this century, almost three times the United States' entire electrical output today. Saudi Arabia, South Korea, Russia, the UAE, and India, among others, are building reactors. And cheap uranium will not last forever.

Other energy technologies may render nuclear fission obsolete before these plans can come to fruition. Or, perhaps, they may not. If the much-delayed atomic age does finally arrive, the breeder probably will not look much like the LAMPRE. But it will serve the same goals and the same dream: clean, plentiful energy, for centuries to come.


History of Reactor Development:

Bunker, Merle E. “Early Reactors: From Fermi's Water Boiler to Novel Power Prototypes.” Los Alamos Science, Winter/Spring 1983, pp. 124-131. Link (pdf).
Bupp, Irvin C., and Derlan, Jean-Claude. Light Water: How the Nuclear Dream Dissolved. Basic Books, 1978.
Civilian Nuclear Power – A Report to the President – 1962. US Atomic Energy Commission, 1962. Link (pdf).
Hewlett, Richard G., and Holl, Jack M. A History of the United States Atomic Energy Commission, Vol. 3: 1952-1960 – Atoms for Peace and War. Link (pdf).
Seaborg, Glenn T., and Loeb, Benjam S. The Atomic Energy Commission Under Nixon. St. Martin's Press, 1993.
Till, Charles, and Yoon, Il Chang. Plentiful Energy: The Story of the Integral Fast Reactor. CreateSpace, 2011.
Weinberg, Alvin. The First Nuclear Era: The Life and Times of a Technological Fixer. AIP Press, 1994.

Direct Contact Reactors and Other Power Reactors:

Hammond, R. Phillip, Russel E. L. Stanford, and J. Ross Humphreys, Jr. “Mobile Fuel Plutonium Breeders: A Study of Economic Potential.” Los Alamos Scientific Laboratory, 1961. LA-2644. Link (pdf).
Kirkbride, L. D. “Molten Plutonium Alloys as Fast Reactor Fuels.” Proceedings of the Conference on Safety, Fuels, and Core Design in Large Fast Power Reactors, October 11-14 1965.
Wykoff, W. R. “Los Alamos Molten Plutonium Reactor Experiment Development Work.” Proceedings of the Symposium on Sodium Reactors Technology, May 24-25 1961. Link (pdf).

Tantalum and the Omega West Reactor Experiments

Andelin, R. L., Kirkbride, L. D., and Perkins, R. H. “High-Temperature Environmental Testing of Liquid Plutonium Fuels.” Los Alamos Scientific Laboratory, 1967. LA-3631. Link (pdf).
Cubitt, Richard L., Ragan, George L. and Kirkpatrick, Donald C. “The Irradiation of Liquid Plutonium Fuels in a Thermal Reactor.” Los Alamos Scientific Laboratory, 1966. LA-3557-MS. Link (pdf).

Fast Reactor Core Test Facility and Molten Plutonium Burnup Experiment

Fast Reactor Core Test Facility Safety Analysis Report. Los Alamos Scientific Laboratory, 1961. LA-2735.
Hall, D. B. A Preliminary Study of a Fast Reactor Core Test Facility. Los Alamos Scientific Laboratory, 1959. LA-2332.
Hannum, W. H., and Kirkbride, L. D. “The Molten Plutonium Burnup Experiment.” Los Alamos Scientific Laboratory, 1965. LA-3384-MS. Link (pdf).
Hannum, W. H. “Nuclear Design of MPBE.” Proceedings of the Conference on Safety, Fuels, and Core Design in Large Fast Power Reactors, October 11-14 1965.
Quarterly Status Reports on LAMPRE Program. Link.


Figure 11: From Civilian Nuclear Power. US Government.
Figure 12: From Fast Reactor Core Test Facility Safety Analysis Report. US Government.
Figure 13: From Fast Reactor Core Test Facility Safety Analysis Report. US Government.
Figure 14: From “Mobile Fuel Plutonium Breeders.” US Government.
Figure 15: From “Mobile Fuel Plutonium Breeders.” US Government.
Figure 16: From “Mobile Fuel Plutonium Breeders.” US Government.
Figure 17: From “Nuclear Design of MPBE.” US Government.
Figure 18: From Preliminary Study of a FRCTF. US Government.
Figure 19: From “LAMPRE Development Work.” US Government.
Figure 20: From Quarterly Status Report on LAMPRE Program, May 1962. US Government.
Figure 21: From “Molten Plutonium Alloys as Reactor Fuels.” US Government.
Figure 22: From Andelin, R. L., Hawkins, A. R., Wilson, N. G. “A New Technique for Testing Molten Plutonium Container Materials at High Temperature.” Transactions of the American Nuclear Society, Vol. 8, pp. 19-20, 1965. US Government.
Figure 23: From “Irradiation of Liquid Plutonium Fuels in a Thermal Reactor.” US Government.
Figure 24: From “Molten Plutonium Burnup Experiment.” US Government.
Figure 25: From “Molten Plutonium Burnup Experiment.” US Government.
Figure 26: Author. Assembled from JCAE Hearings on AEC Authorization Bills. Inflation data from The Inflation Calculator.


  1. Well, I would say that the paste core was similar to the DCR but used a paste of uranium oxide and plutonium oxide in liquid sodium as fuel instead of molten plutonium alloy and that was really a better choice.

  2. I was very pleased to find this site while looking for scaling photographs of the LAMPRE core.

    With the advance of materials science, this is a type of reactor with many outstanding properties that deserves another look. It is also worth considering higher temperature versions.

    I've been collecting literature on this reactor for several years now, and this wonderful link addresses many questions I've asked myself.

    Thanks for this. Most excellent!

    1. Thanks for your comment!

      As fond as I am of the LAMPRE, I'm skeptical it holds any real advantages compared to molten salt designs. MSRs seem to basically have all of the advantages of the LAMPRE, but without some of its disadvantages (such as the use of molten sodium).

      By the way, if you're still looking for LAMPRE documents, send me an email at I've got about 800 MB of LAMPRE technical reports, many of which were not available online at the time, and I'd be happy to share.

  3. Thanks, I disagree with your opening sentence, and look forward to discussing it with you off line.

    I personally believe that the LAMPRE concept, with advances in materials science that should at least be investigated has many features that overcome profound issues with the MSR that have been overlooked.

    Sodium is just a coolant, there's no reason to require it over other possible coolants, in the LAMPRE concept.

    I may have some materials that you might be interested in, and I'm sure you have others in which I'm interested and I therefore appreciate your generous offer.

    We'll be in touch.

    My email is