Abstract
Many countries have long pursued fast neutron breeder reactors, which create more fissile fuel than they consume, because of the expectation that the world will run out of the low-cost uranium used for fuel in most commercial nuclear power reactors. Despite billions of dollars spent on research and development, however, fast neutron reactors have proven unreliable and are frequently shut down due to leaks of their sodium coolant. Different reactors around the world have experienced this leakage problem, indicating that it may have fundamental causes, rather than being a symptom of poor manufacture or operational deficiencies. One possible cause is a series of chemical interactions between the carbon contained in the metallic components of these reactors and the sodium used to cool reactors; these interactions can cause a system’s metal parts to corrode, eventually leading to leaks. If true, this mechanism carries significant negative implications for the safety of these reactors and their economic viability.
In 1972, a Soviet nuclear engineer, V. V. Orlov, told an international conference on the peaceful uses of atomic energy that the Soviet Union was using “the most reliable type of steam generator” for its new BN-350 breeder reactor (IAEA, 1972: 449). The new reactor commenced commercial operations in July of the following year. Four months later, US intelligence sources revealed that a US spy satellite had photographed what appeared to be an explosion or a fire at the nuclear plant (Patterson, 1984). Confirmation came a year later when, after a long silence, Soviet participants admitted at a conference in London that there had been serious sodium leaks in two of the BN-350’s steam generators—the very ones described as “most reliable.”
The 1973 leaks have been attributed to “inadequate quality in manufacturing of heat exchangers” (Raj and Kumar, 2011), but despite much effort, these leaks never stopped, and the reactor experienced 15 leaks during its lifetime. Consequently, the BN-350 never generated above three-quarters of its designed 1,000 megawatts of heat (IAEA, 2007).
A later Russian breeder reactor, the BN-600, reported 27 sodium leaks in a 17-year period, 14 of which led to sodium fires (Saraev, 1998). Some of these were major leaks, involving up to 1,000 kilograms (kg) of sodium (IAEA, 2004). The frequent sodium leaks in the steam generator suggest poor compatibility of sodium and steel.
Breeder reactors: A leaky legacy
Experience with multiple breeders does not seem to have helped. Until the 1990s, France was considered to be one of the leaders in fast breeder technology, because it was the only one to construct a pilot-scale reactor (called the Rapsodie), a demonstration prototype (Phénix), and a commercial-scale sodium-cooled reactor (Superphénix). Even then, France’s experience with sodium leaks does not inspire confidence. Rapsodie was shut down permanently in 1982 after a sodium leak. The Phénix reactor holds the record for the number of sodium leaks, which started even during the testing phase, as soon as the sodium was first put into the secondary cooling circuit (Sauvage, 2004). In the Superphénix, about 30 tons of sodium leaked from the used fuel storage drum and entered the space between the storage vessel and the building containing it.
Nor has the expenditure of vast sums of money helped. Japan spent over $17 billion on fast breeder reactor research, only to have the Monju—its prototype breeder reactor—experience a leak and fire in the secondary sodium heat transport system even during the pre-operational testing phase. It had to be shut down. The accident involved over 640 kg of sodium; a survey by French and Indian nuclear engineers described it as the “worst sodium leak, having a major impact on reactor restart-up” (Guidez et al., 2008: 214).
Germany, for all its famed manufacturing prowess, was not spared either. One of the steam generators of its KNK-II reactor experienced a sodium leak even before actual commissioning (IAEA, 2012).
Other breeder reactors, not listed in the table above, have also experienced multiple leaks. Fermi-I, the United States’ first commercial breeder reactor, suffered leaks in its steam generators during its extended testing period, and such leaks persisted throughout its operational life, between 1963 and 1972 (Kouts, 1983). Both of the United Kingdom’s fast reactors, the Dounreay Fast Reactor and the Prototype Fast Reactor, suffered numerous leaks. And India’s only operating fast breeder reactor (FBTR) experienced multiple sodium leaks (Sasikumar et al., 2003).
In part as a result of such leaks, practically no fast reactor has operated efficiently, at least if one goes by the load factor—a measure of how well a reactor generates electricity, as reported by the International Atomic Energy Agency’s Power Reactor Information System database. Only the BN-600 reactor had a lifetime load factor reach the somewhat respectable figure of 74.1 percent. The British Prototype Fast Reactor and the French Phénix had load factors of 26.9 percent and 40.5 percent, respectively. In contrast, the first and only commercial-scale fast reactor anywhere, the French Superphénix, could manage only 7.9 percent.
Getting to the bottom of things
Why these problems with sodium leaks? In the early years, blame fell on the seemingly inevitable component failures and teething problems involved any time a new technology is created from scratch. However, the consistent pattern of sodium leakages—even after years of operation—suggests that there might be a more fundamental problem at work. If the leaks were only due to faults during construction, for example, then they should not recur after repairs were carried out and the leaking component replaced.
But the leaks’ stubborn persistence suggests other causes. Because the problems occur most often in items made of steel, such as steam generators and certain reactor components—especially those containing ferrous alloys of austenitic and ferritic steel grades 2 —it is possible that something in the metals themselves enables corrosion.
This line of reasoning is supported by several studies indicating that the culprit may be a form of corrosion that occurs in some materials when they are exposed to environments with carbon, a process referred to as “metal dusting.” The corrosion manifests itself as a breakup of bulk metal into a dust-like metal powder—hence the term (Chun and Ramanarayanan, 2005; Grabke et al., 1993, 1996; Lin et al., 2004; Zeng et al., 2002, 2008). Metals like iron and nickel, as well as alloys based on these metals, are susceptible.
This mechanism might be operating in sodium-cooled breeder reactors (Pillai, 2011). Underlying the phenomenon is one of the chemical properties of carbon, which is present in the stainless steel alloys used in various reactor components. On contact with the molten sodium coolant, it tends to dissolve out into one of two forms: as pure carbon or as acetylide (Na2C2), depending on temperature (Ainsley et al., 1974; Thompson, 1979). Because the temperature of the reactor and its coolant system comes down whenever the reactor is periodically shut down for routine maintenance, this temperature change encourages carbon in the sodium coolant to turn to acetylide. This conversion promotes the formation of an unstable and transient but relatively long-lived chemical complex, iron carbide (Fe3C), on the surface of the steel and in that part of the steel’s matrix that is especially close to the surface. When the reactor is brought back to normal operation and the temperature goes back up, this “metastable” substance undergoes decomposition—a form of metal dusting (Pillai, 2011).
The net effect of this thermal cycling behavior is that the alloy matrix gradually disintegrates into a mass of metal and carbon powder, which degrades the mechanical integrity of the components made of metallic alloys. In addition, metal dusting could be further accelerated if carbon activity in the sodium system is increased, for example through the entry of oil from pumps.
The relationship of metal dusting—well understood in non-nuclear systems—and the failures in breeder reactors is not yet empirically verified. Corrosion is seldom the sole cause of failure. Yet, most mechanical failures are preceded by it. Unfortunately, failure analyses tend to pay inadequate attention to the corrosion process that preceded the mechanical failure.
Why use sodium at all?
Broadly speaking, reactors can be classified into two categories: thermal and fast. In the far more common thermal reactors, the neutrons released during fission are slowed by a moderator such as water or heavy water (or graphite in the core). When neutrons react with these materials, they lose energy and become slower, a process called thermalization. In fast reactors, as the name implies, the neutrons remain fast and energetic, because there is no such moderator to slow them down.
In some reactors, those neutrons that escape the core are captured by a blanket made of “fertile materials,” which then get transformed into a new element that itself can be used as fuel. For example, uranium 238—the most common isotope of uranium, constituting about 99.3 percent of naturally available uranium—gets converted into the fissile isotope plutonium 239.
If designed properly, more fissile material can be produced in the blanket than is consumed in the reactor core. Such reactors are called “breeder reactors,” indirectly fueled by uranium 238. Because uranium 238 is much more abundant than the isotope that fuels most thermal reactors, breeder reactors allow more nuclear power to be generated from the same amount of uranium ore.
One key difference between these fast reactors and the more common thermal reactors is the choice of coolant. Because fast reactors do not have any moderator to slow down neutrons, their cores are smaller compared with thermal reactors. And since most fission, and thus energy production, occurs in the core, their power density is much higher. Liquid metals such as molten sodium are more efficient at carrying this heat away than more commonplace coolants such as water. Using sodium as a coolant offers a key benefit: Because sodium melts at a relatively low temperature and boils at a relatively high 883 degrees Celsius—much higher than breeder reactors’ operating temperature—it doesn’t need to be pressurized to stay in the liquid phase.
Although sodium has some safety advantages, it does react violently with water and burns if exposed to air. Consequently, while the sodium coolant itself is not pressurized, its open surfaces have to be covered by an inert gas such as argon, which in turn tends to get swept into the flowing sodium and cause unwanted bubbles. (When the bubbles pass through the core, any argon trapped in the sodium would adversely affect the heat transfer characteristics of the system, and reactivity would fluctuate.)
What’s more, while sodium does not readily absorb the fast neutrons generated in a breeder reactor—one of the reasons it’s used as coolant—when sodium does absorb neutrons, it becomes an intensely radioactive gamma emitter known as sodium 24, with a half-life of 14.95 hours. Consequently, the primary sodium coolant must be confined entirely within the core’s biological shield.
Therefore, many breeder reactor designs use a secondary sodium loop, with a heat exchanger inside the shielding to pick up heat from the radioactive primary sodium and carry it out to the second heat exchanger, where steam is generated. To exchange heat while safely turning water into steam, these secondary exchangers must contain both molten sodium and water, while simultaneously keeping the two components separated. This is accomplished by using thin tube walls, much like the radiator of an everyday automobile. Such walls must be fabricated to very high tolerances and, as detailed above, have proven to be one of breeder reactors’ most troublesome features.
Thin walls, sodium, water, and violent chemical reactions can be a recipe for problems and potential accidents. Add metal dusting, and the situation worsens.
Implications for the future of fast reactors
Assuming that the metal dusting mechanism described is true, what does it entail for fast reactors and their operations?
The simplest way to reduce the rate of metal dusting would be to lower the maximum temperature to which the sodium in a fast reactor is subjected. 3 How low should the temperature be? Experiments involving the exposure of austenitic stainless steel specimens to flowing sodium for long periods of time show that significant corrosion occurs at temperatures above 450 degrees Celsius (Pillai et al., 1997). 4 Metal dusting may occur even at temperatures as low as 350 degrees Celsius.
However, lowering the operating temperature of a reactor results in lower thermodynamic efficiency, and thus less electrical power being generated. These lower efficiency and power levels make breeder reactors less economically attractive because they involve a higher capital expenditure on a per-kilowatt basis than a standard nuclear power plant, as well as requiring more fuel. Experience bears this out. For example, to avoid corrosion problems, the power of the French Phénix reactor was brought down from 255 megawatts of electricity to 145 megawatts (IAEA, 2007). This meant that the capital cost of components per unit of capacity in dollars per kilowatt was effectively increased by 43 percent.
There is a long history of problems confronting fast breeder reactors, including higher costs, safety challenges, and linkages to proliferation (Cochran et al., 2010; Dickson and Abbott, 1992; Von Hippel and Jones, 1997). In addition, there is the persistent problem of a poor operating record caused, for the most part, by sodium leaks, often in the steam generators. Almost all fast reactors have experienced one or more sodium leaks. Developing leak-free steam generators has proven to be difficult, and some authors have termed these components the Achilles’ heel of sodium-cooled systems (Kouts, 1983: 402). There is also the risk that any of these leaks or fires might spin out of control into a catastrophic accident, an ever-present concern with nuclear reactors.
There may be fundamental reasons for these leaks, explained by materials science and corrosion chemistry. As a result, breeder reactors may have to be operated at lower temperatures, even if it would lower their economic viability. If this is not done, there is a price to be paid in terms of operational problems and accident risks.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.