Reactors built to produce weapons plutonium in Russia.

Given the destruction that can be wrought with a few kilograms of plutonium or a few times that amount of highly enriched uranium, if made into a nuclear weapon, it is fortunate that these materials are rather difficult to produce in their weapon-usable forms. Indeed, lack of capacity to produce them has long been considered to be the principal technical barrier against the spread of nuclear-weapon capability to additional nations and to subnational groups.[1]

This has been regarded as the principal barrier not only because the technologies for producing these materials are demanding and costly (about which more in a moment), but also because the steps needed to make a weapon once the material is in hand are not as difficult as producing the material is,[2] and because the plutonium and highly enriched uranium that have been produced until now by those possessing the requisite technologies have mostly been well guarded or have resided in forms awkward to steal and difficult to use in weapons (such as plutonium in spent reactor fuel[3]).

So how difficult is the production of these materials? The subsequent subsections elaborate on this question.[4]

Producing HEU

The process of producing HEU begins with acquiring natural uranium. This element is quite widespread in the Earth’s crust. It is mined today principally from sandstone ores in which the uranium occurs at concentrations of 0.03% to 0.2% by weight, but it is also found at concentrations on the order of 10 times lower in rather widely distributed shales, and at concentrations a few times lower still in even more widely occurring granites. Depending on the characteristics of the particular geologic formation, such uranium-bearing rocks may be extracted from underground mines or from open pits.[5]

Once the ore is in hand, the rock is crushed and exposed to an acid bath which leaches the uranium into solution. It is then extracted as an oxide, U₃O₈. These steps by which uranium is extracted from its ore are collectively called uranium milling, and the quite voluminous, mildly radioactive, sand-like residues of the process are referred to as uranium-mill tailings.

Uranium mining and milling at a scale adequate to support a nuclear-weapon program—even a small one—entail rather sizable operations, with distinctive observable characteristics. For example, to obtain enough raw uranium to feed the enrichment operation for a single "gun-type" nuclear weapon (about which more below) would require the mining of more than 13,000 metric tons of sandstone ore contain 0.1% U, as well as the disposal of about 13,000 metric tons of uranium-mill tailings.[6] This is not an operation likely to be manageable by terrorists. Even when conducted by a nation it is the sort of thing that would require some effort to successfully conceal.

The actual process of enriching the U-235 concentration above its value of 0.7% in natural uranium is even more demanding. Because different isotopes of uranium behave almost identically chemically, most separation methods have relied on physical rather than chemical means of separation, based principally on the 1.3 percent difference in mass between U-235 and U-238 atoms.

The main approaches to this task that have been used to date involve first converting the natural uranium to uranium hexafluoride gas (UF₆), followed by physical separation of the lighter U²³⁵F₆ molecules from the slightly heavier U²³⁸F₆ molecules. The best-known technologies for accomplishing this separation are:

• gaseous diffusion plants, which exploit the difference in the diffusion rates of the lighter and heavier molecules through a "cascade" of thousands of porous barriers; or

• centrifuge plants, which use sets of hundreds or thousands of sophisticated, ultra-high-speed, gas-centrifuge machines to separate the molecules based on their differing inertial masses.

Gaseous diffusion plants involve large, complex piping arrangements, esoteric membranes (the characteristics of which remain classified in all countries that have developed them), and immense electric-power requirements for running the compressors that force the uranium hexafluoride gas through the membranes. Centrifuge plants have electric-power requirements 20-30 times smaller than those of gaseous diffusion plants, but the technology for the centrifuges is extremely demanding and closely controlled.

Gaseous-diffusion plants are the size of factories, not of laboratories, and this plus their large electric-power requirements makes them difficult to conceal. Gas-centrifuge plants can be somewhat smaller, and they need less electric power; but they still need room for hundreds of centrifuges so concealment poses some challenges. Neither technology could be mastered by terrorists. Indeed, they are challenging even for countries other than the major industrial nations.

Countries operating gaseous diffusion or gas-centrifuge enrichment plants for the support of commercial nuclear-power operations include the United States, the United Kingdom, France, Russia, China, Japan, Germany, and the Netherlands. These commercial uranium-enrichment plants are being used to enrich uranium only to the 3 to 5 percent U-235 level suitable for use in commercial nuclear power reactors of today’s dominant types. This material cannot sustain a nuclear explosion and so cannot be the driver of a nuclear weapon. In terms of the "enrichment work" needed to separate isotopes, however, it is half way or more toward the 90%+ enrichment levels desirable for nuclear weapons. (See Box 3: Uranium Enrichment: Inputs and Outputs.)

In principle, any of the commercial enrichment plants could be operated in a manner to do the remaining work needed to bring this low-enriched reactor fuel up to weapon-usable levels. Commercial enrichment facilities in countries other than the "authorized" nuclear-weapon states[7] are subject to International Atomic Energy Agency safeguards designed to detect such activity if it occurs. It is not something likely to be accomplishable by terrorists or by agents of a proliferation-inclined country by taking over someone else’s enrichment plant for a few hours.

All five of the "authorized" nuclear-weapon states used gaseous-diffusion and/or gas-centrifuge enrichment plants to produce HEU for their weapons. (None of these countries is producing HEU for weapons at this time.) In the past, these countries produced uranium at a range of enrichments above 20 percent not only for nuclear weapons but also for use in nuclear reactors for propulsion of submarines, other warships, and icebreakers; in research reactors; and in experimental power reactors of a variety of kinds.

Smaller gaseous-diffusion and gas-centrifuge enrichment plants were operated in the past by Argentina and Brazil in connection with nuclear-weapon programs that have since been abandoned, and such plants are operating today in the "de facto" nuclear-weapon states India, Pakistan, and North Korea. (North Korea appears to have obtained the centrifuge technology from Pakistan.)

Other approaches to uranium enrichment besides gaseous diffusion and centrifuges have been explored from time to time but have their own drawbacks. Some have even larger power requirements than those of gaseous diffusion; the aerodynamic-nozzle technology used by South Africa,[8] and the electromagnetic separation technology developed by the United States in World War II and subsequently tried by Iraq in the nuclear-weapon program unveiled by the Gulf War, both fall into this category. Others have very low separation factors and thus need a huge number of stages to reach high enrichment; chemically based processes that have been pursued by France and Japan – and also by Iraq – fall into this category.

Technologies exploiting the capability of precisely tuned lasers to selectively excite uranium-235 atoms, allowing their separation from the uranium-238 items by electromagnetic or other means, appear to have the potential for low energy requirements and high separation factors, and they have been under investigation in a number of countries for decades. They have not yet been developed as practical options, however, and it remains unclear whether the worries of non-proliferation analysts about laser enrichment – that it might finally make possible the production of HEU with modest resources and easy concealment – will ever be realized.

Producing plutonium

As noted above, any nuclear reactor that contains U-238 in its fuel produces Pu-239 in the course of operation, as a result of the absorption of some of the fission neutrons by this uranium isotope. Some of the Pu-239 that is produced is invariably fissioned itself in the course of the continuing chain reaction, and some undergoes successive absorption of further neutrons to become the heavier plutonium isotopes—Pu-240, Pu-241, and Pu-242. Some of these also fission in the course of continuing reactor operation.

The rate at which plutonium accumulates in a reactor’s fuel depends on many factors, including the type and thermal power output of reactor and the characteristics of its fuel, its coolant, and its moderator. (See Box 4: Reactor Types and Terminology). The quantity and isotopic composition of the accumulated plutonium depend also on how the reactor is operated, particularly on how much fission occurs in each kilogram of fuel up until the time it is removed from the reactor, a parameter called the irradiation or burnup of the fuel. (See Box 5: Reactor Size and Performance).

High burnup is desirable for electricity production because it means more saleable energy from the fuel one has paid for fabricating, as well as less "down time" for refueling (in the case of reactor types that need to be shut down and partly disassembled in order to remove their fuel). But high burnup is undesirable for production of weapon plutonium, both because it leads to greater accumulation of the less-desirable even-numbered plutonium isotopes and because higher burnup means the spent fuel contains larger amounts of radioactive fission products in relation to the plutonium quantities present, making it more dangerous and difficult to separate out the plutonium.

The reactors that countries determined to produce plutonium for weapons have built for this purpose have nearly all been fueled with natural (un-enriched) uranium and moderated by graphite or by heavy water; they have ranged in rated thermal power output from 20 to more than 4,000 megawatts.[9] Many of these reactors were designed to be continuously refuelable, meaning that irradiated fuel can be removed from the reactor core and fresh fuel can be inserted while the chain reaction is generating neutrons and power.

This feature enables such reactors to operate at the low burnups needed to make weapon-grade plutonium without needing to be shut down frequently to remove and replace the slightly irradiated fuel.[10] Reactors that must be shut down and opened up for refueling—called "batch refuelable"—can lose considerable operating time in this process. Thus they cannot make as much plutonium in a year (and, in dual-purpose reactors, neither as much plutonium nor as much electricity) as a continuously refuelable reactor of the same rated thermal power.

Regardless of the details, when operated at the very low burnup levels associated with production of weapon-grade plutonium (see Box 4), all the graphite-moderated and heavy-water-moderated production reactors deliver a net rate of plutonium production in the range of 0.9-1.0 grams per megawatt-day of reactor operation. Thus, a very small production reactor with rated thermal capacity of 25 megawatts (the size range of the North Korean graphite-moderated production reactor at Yongbyon) can produce in a year, if it achieves the equivalent of 250 full-power days of operation, about 5.5 kilograms of weapon-grade plutonium – about one bomb’s worth.[11] Clearly, a production reactor l00 times larger, typical of those the United States operated at Hanford and Savannah River, could produce 100 bombs’ worth of plutonium per year.

Building a plutonium-production reactor is a demanding task even for a nation of some technical capability. A number have achieved this, of course, but some of those who have done so have made use of help from more advanced nations. The problem of building a production reactor is not just a matter of having the needed knowledge plus the trained personnel and industrial equipment needed to translate the knowledge into hardware. It is also a matter of being able to acquire the unusual materials that are required: natural uranium, which while widespread in ore is not easy to get in the refined form needed for a reactor; and either heavy water or extremely pure graphite, to serve as the moderator. Producing any of these materials is itself a high-technology operation, and sales are subject to restrictions and tracking.

The challenges of using a production reactor to acquire weapon plutonium are even greater if those trying to do so wish to conceal this activity. The process of building the reactor is not so easy to hide, all the less if help is being provided by another country. The reactor itself can be put underground to hide it from satellites, but this increases the cost and there still must be access points and ventilation shafts whose construction or use might be detected. Most problematic of all for concealment, it is in the nature of a reactor that the 20 or 200 or 2,000 megawatts of energy flow that it produces while operating must be discharged to the environment as heat. (In a dual-purpose reactor, some of the energy is converted to electricity and transmitted elsewhere, but two thirds or more of the energy still ends up in the reactor’s immediate environment as heat.) Heat sources of this magnitude are extremely difficult to hide from infrared sensors on satellites.

Of course, as noted earlier even reactors designed for electricity production will automatically make significant quantities of plutonium, as long as their fuel contains substantial quantities of U-238. In a typical "light-water reactor"—a batch-refuelable reactor type designed for electricity generation—the net rate of plutonium production if the reactor is operated at the high burnups optimum for the electric-generating role is 0.22-0.27 grams of plutonium per megawatt-day. This means that a 3,000 thermal-megawatt light-water reactor that operates at full power for 330 days per year (hence a million megawatt-days per year) will discharge 220-270 kilograms of plutonium per year in its spent fuel[12]—enough plutonium to make something like 40 nuclear weapons.[13]

If such a reactor were operated instead, for purposes of optimum production of weapon plutonium, at a burnup of 1 megawatt-day per kilogram of heavy metal in its fuel rather than the 30-50 megawatt-day per kilogram levels characteristic of commercial operation in reactors of this type (Box 4), the net plutonium production per megawatt-day would rise to about 0.5 grams of plutonium per megawatt-day. But in this mode the reactor would need to be shut down much more frequently for refueling, which means its capacity factor would fall substantially, perhaps to 60%, corresponding to 220 days per year of full-power operation per year. Then the output of plutonium would be 3,000 megawatts x 220 days per year x 0.5 gram per megawatt-day = 330 kilograms per year, and this plutonium would be of higher quality for weapon purposes than in the higher burnup case.[14]

In order to use the plutonium produced in a nuclear reactor in a nuclear weapon, it must be chemically separated from the fission products produced along with it, and from the residual U-238, by reprocessing the nuclear fuel. Reprocessing, like uranium enrichment, is a technically demanding and costly operation; and because of the intense gamma-radioactivity of the fission products, and the health risks posed by the alpha-activity of plutonium if inhaled or otherwise taken into the body, reprocessing is also much more hazardous than enrichment from the standpoint of health and safety.

The approach to reprocessing that has been used virtually universally for military and civilian purposes alike—called the "Purex" process—was worked out in the United States in the Manhattan Project of World War II. It consists of chopping up the radioactive spent fuel into pieces, dissolving these in nitric acid, and then performing a set of solvent extractions on the resulting solution to separate the plutonium, the uranium, and the fission products into three output streams. The uranium or plutonium may emerge finally as nitrates or as oxides. Ultimately, for weapon use, the plutonium would be transformed into the metal.

The set of operations associated with reprocessing is made greatly more difficult than it would otherwise be by the intense radiation emanating from the fission products, much of it in the form of highly penetrating gamma rays. Standard practice is to allow the spent fuel to "cool" for a period months to years before subjecting it to reprocessing, so that some of the shorter-half-life radionuclides will have decayed away.

Even after such cooling, the radiation hazards from spent fuel to those trying to work with it remain high. The dose rate at the surface of a spent fuel assembly from a modern light-water reactor, at typical commercial burn-up and after ten years' cooling time, is around 20,000 rem per hour, and at distance of a meter it is around 2,500 rem per hour. (Recall from Box 2 that a whole body dose of 1,000 rem delivered in a period of less than a week or so is certain to be fatal within weeks.)

At the far lower burnups associated with production reactors operated to make weapon-grade plutonium, the dose rate and any given time after discharge from the reactor is lower, but impatience to get the plutonium out in order to get on with making weapons from it tends to reduce the length of time the fuel is allowed to cool. A fuel assembly from a light-water reactor that had a experienced a burnup level appropriate to weapon-plutonium production and then been allowed to cool for just two years would deliver a dose rate at its surface of nearly 40,000 rem per hour.

In all cases, then, extensive shielding and equipment for remote handling of the materials are required in all stages of reprocessing up to the point where the fission products have been separated from the uranium and plutonium. The equipment must be designed to avoid the possibility that a critical mass of plutonium in a liquid form or as a precipitate could form at any point in the system. And pipes, valves, and vessels that break or clog must be repairable by remote control, because they will be too radioactive to approach even with protective suits. The technology for this is so demanding and difficult that even major industrial nations have ended up building some reprocessing plants that failed almost immediately and were deemed so expensive to repair that they were abandoned.[15]

None of the five "authorized" nuclear-weapon states are any longer carrying out reprocessing for production of weapon plutonium, except for a small amount of reprocessing continuing in Russia because the fuel from two dual-purpose reactors whose energy output is still needed was not designed for long storage without reprocessing. Large reprocessing plants for commercial power-reactor fuel are in operation at La Hague in France, at Sellafield in England, and at Chelyabinsk in Russia. A considerably smaller commercial reprocessing plant is operating at Tokai-Mura in Japan. France has a very small plant for reprocessing breeder-reactor fuel at Marcoule. Two other countries that operated pilot-scale reprocessing plants in the past motivated by commercial possibilities are Belgium and Germany.[16]

India operates small reprocessing plants for its weapon program at Tarapur and Kalpakkam. Israel has a reprocessing plant supplied by France in the same underground complex at Dimona that houses Israel’s plutonium production reactor. North Korea had a very small reprocessing plant at Yongbyon, which it shut down as part of the "Framework Agreement" with the United States in 1994.

Could additional countries build reprocessing plants as part of a nuclear-weapon program? Yes, at considerable cost and some chance of failure. In addition, these facilities are not terribly easy to conceal either in construction or in operation; detectability in operation is aided both by the continuing heat release from the radioactive fission products and by emissions of tell-tale radionuclides, particularly the noble gas krypton-85.

As for terrorists, building a production reactor is well beyond the reach of any terrorist group known to date, and building a reprocessing plant remotely approaching the sophistication of what countries have built would similarly be out of the question. A somewhat controversial memorandum out of the Oak Ridge National Laboratory 25 years ago suggested that spent fuel might be more easily reprocessed, perhaps with technology even terrorists could acquire, if traditional sensibilities about how much of the plutonium needs to be extracted from the fuel, and about protecting workers from highly harmful doses of radiation, were discarded (as terrorists might well be prepared to do).

Many experts remain skeptical about this contention. Even if it is right, terrorists still face the problem of how to obtain and transport the highly radioactive, plutonium-bearing spent fuel to feed such a reprocessing operation. If subnational groups can’t build production reactors, they certainly are not in a position to build a far more complex and costly electric-power reactor so as to be able to skim off plutonium-bearing spent fuel on the side. But the question of whether protection for spent fuel should be upgraded ought to be looked at closely.

Nor, finally, is it very plausible that a terrorist group trying to acquire plutonium for nuclear weapons could succeed in doing so by taking over an already operating plutonium-production reactor or power reactor. The only case in which a subnational group might be able to hold the reactor for long enough to remove any fuel from it, which is not a particularly simple operation, is if the reactor were on the territory of a secessionist faction along the lines of the Chechens in Russia.

Concluding observations on producing nuclear-explosive materials

One of the two routes to producing nuclear-explosive materials requires mastering or controlling uranium-enrichment technology. The other route requires mastering or controlling both a reactor of some size and a spent-fuel reprocessing plant. Terrorists do not seem at all likely to be follow either of these very difficult paths.

Even for some of the less capable countries that could aspire to nuclear weapons, it might seem easier to steal the needed nuclear-explosive material (or, better, buy it in a black market after someone else has stolen it) than to try to make it. That is why protecting such materials wherever they exist, and learning to detect and interdict them in transit if and when protection fails, are so immensely important.

back: weapons effects

next: detection and protection

FOOTNOTES

[1] There are, of course, important political barriers to such proliferation, above all the international norm against proliferation that is embodied in the Non-Proliferation Treaty—which in 1995 was extended indefinitely with overwhelming support from its now 185 parties—and the International Atomic Energy Agency safeguards that monitor compliance with this and related agreements.

[2] There is, alas, no longer any fundamental "secret" about how fission bombs can be made. As discussed earlier, the knowledge and skills to do so, once the materials are in hand, are available to virtually any country and quite plausibly could be mustered by a well organized terrorist group. Fortunately, the more powerful thermonuclear weapons are far more difficult to design and construct.

[3] In the spent fuel from a typical nuclear power reactor, each kilogram of the plutonium that has been produced as a result of absorption of fission neutrons in U-238 is intimately mixed with about four kilograms of intensely radioactive fission products and about 100 kilograms of unreacted U-238. It cannot be used in a nuclear weapon unless it is separated from these diluting and contaminating substances. In a plutonium production reactor using natural uranium with heavy water or graphite as the moderator, a kilogram of plutonium in the discharged fuel is mixed with about one kilogram of fission products in the same 100 kilograms of U-238.

[4] For a useful introduction to these technologies, see Office of Technology Assessment, Technologies Underlying Weapons of Mass Destruction (Washington, D.C.: OTA, December 1993). The indispensable reference on the history and (reasonably) current status of production of nuclear-explosive materials is David Albright, Frans Berkhout, and William Walker, Plutonium and Highly Enriched Uranium 1996: World Inventories, Capabilities, and Policies (Oxford, UK: Oxford University Press for the Stockholm International Peace Research Institute, 1997).

[5] Uranium also exists in seawater, at the extremely low concentration of about 3 parts per billion by weight. This means that a technology capable of extracting half the uranium from a given volume of seawater would need to process about 650,000 cubic meters of water to extract one kilogram of uranium. Means of doing this by selective absorption have been demonstrated on a small scale, but it is not clear whether or when such approaches will become even close to economically competitive with using terrestrial ores.

[6] It has been assumed in arriving at this figure that the U-2325 concentration in the HEU product will be 93% and that the U-235 concentration in the depleted-uranium waste stream will be 0.3%. (See Box 3: Uranium Enrichment: Inputs and Outputs.)

[7] The "authorized" nuclear-weapon states are those certified as such by the Nuclear Non-Proliferation Treaty of 1968—namely the United States, Russia, the United Kingdom, France, and China. "De facto" nuclear-weapon states whose capabilities have been developed and in some cases admitted since that time are Israel, India, Pakistan, and possibly North Korea.

[8] South Africa clandestinely constructed half a dozen gun-type nuclear weapons using HEU it produced with the nozzle technology—these weapons contained about 55 kilograms of weapon-grade HEU each—then publicly renounced the weapons, dismantled them, and joined the Non-Proliferation Treaty as a Non-Nuclear Weapon State.

[9] These have included: in the United States, the 9 graphite-moderated, light-water cooled production reactors deployed at the Hanford site and the 5 heavy-water moderated production reactors deployed at Savannah River (none any longer operating); in the former Soviet Union, the 13 graphite-moderated, light-water-moderated production reactors at Chelyabinsk, Tomsk, and Krasnoyarsk (of which 3 are still operating because these dual-purpose reactors continue to supply heat and electricity to regions that currently lack alternative options); in the United Kingdom, a total of 10 graphite-moderated, gas-cooled production reactors at Windscale, Calder Hall, and Chapel Cross (none still operating): in France, a total of 9 graphite-moderated, gas-cooled reactors at Marcoule and two heavy-water moderated reactors at Celestin (none still operating); in China, one production reactor at Jinquan and one at Guangyuan; in Israel, a heavy-water moderated, air- and heavy-water-cooled production reactor at Dimona; in India, two heavy-water moderated production reactors near Bombay; and in North Korea a graphite-moderated, gas-cooled production reactor at Yongbyon. (France also used its prototype liquid-metal cooled, fast-neutron breeder reactor, the Phenix, to produce some weapon-grade plutonium.) A heavy-water-moderated, light-water-cooled production reactor under construction in Iraq at the beginning of the 1980s was destroyed by an Israeli bombing raid in 1981. See Albright, Berkhout, and Walker, Plutonium and Highly Enriched Uranium 1996: World Inventories, Capabilities, and Policies, op. cit.

[10] Power reactors that have this feature, such as the CANDU (see Box 4), are more difficult to safeguard because of it. That is, the reactor must be monitored continuously to ensure that fuel is not being switched out for purposes of extracting its plutonium for weapons, rather than only needing to be monitored during the periods when the reactor has been shut down and opened up for "batch" refueling.

[11] 25 megawatts x 250 days x 0.9 grams per megawatt-day = 5,625 grams = circa 5.6 kilograms.

[12] This would correspond to a capacity factor of 330/365 = 0.904 or 90.4% (see Box 4), a level of performance now quite commonly achieved in commercial light-water reactors.

[13] The figure of 40 nuclear weapons follows from a round number of 6 kilograms of reactor-grade plutonium per weapon. In practice, the processing of spent fuel to extract the plutonium and the steps then required to fabricate the plutonium will result in some degree of unrecoverable loss from the stock of plutonium one starts out with in the spent fuel. Depending on the size of this loss (which depends on the skill and sophistication of the group running the operation, and also on how hard they are trying to minimize plutonium loss), the number of 6-kilogram weapons that could be made from the stated annual output might be considerably smaller.

[14] The lost electricity output associated with reducing the annual capacity factor of our 3,000 megawatt thermal (1,000 megawatt electrical) reactor from 90% to 60% would be some 2.6 billion kilowatt-hours per year. Thus the effective cost, in electricity terms, of obtaining the higher quality plutonium is this way is about 8 million kilowatt-hours per kilogram. A bomb made from 5 kilograms of this plutonium would therefore represent an effective "investment" of 40 million kilowatt-hours of electricity, even more than the 30 million kilowatt-hours shown in Box 3 to be needed for uranium enrichment for a gun-type bomb containing 60 kilograms of weapon-grade HEU.

[15] A reprocessing plant built by the General Electric company in Morris, Illinois, met this fate.

[16] The large commercial plants have capacities between 600 and 800 metric tons of heavy metal per year, measured as the total uranium and plutonium content of the spent fuel supplied to them. An 800-metric-ton-per-year plant could reprocess the fuel from 30 to 40 large light-water electric-power reactors and in so doing would separate 8-10 metric tons of plutonium per year. The Japanese plant at Tokai-mura has a capacity of 100 metric tons of heavy metal per year; an 800-metric-ton-per-year plant at Rokkasho-mura is under construction, but whether it will ever operate is unclear.