Science, Technology, Fission, and the Future

Keynote Speech at the American Nuclear Society Banquet
November 19, 2002

Richard L. Garwin

I am humbled by the thought of addressing this assembly with a meeting theme, "Building the World Nuclear Community-- Strategies for the deployment of new nuclear technologies." Particularly a meeting which commemorates the 60th anniversary of the first controlled nuclear chain reaction. Indeed, this afternoon Alvin Weinberg spoke at the President's special session, with the distinction of having witnessed the birth of the first neutron chain reaction in a couple of billion years.

I also am a physicist, having worked in nuclear and particle physics, as well as in the field of condensed matter including superconductors and liquid and solid helium. I am an experimenter with an aptitude for instrumentation. I worked 40 years at IBM, some of the time as Director of Applied Research, and since 1950 I have spent about half of my time in U.S. government activities, having chaired many committees on technology and national security, and the like.

From 1948 to 1952 I was a close associate of Enrico Fermi, my thesis advisor for my 1949 Ph.D. in nuclear physics. I then became his colleague on the physics faculty at the University of Chicago and worked as a consultant to the Los Alamos nuclear weapons activity where I made several contributions in nuclear weapons design and testing. It was a great experience to share an office with Fermi the summer of 1950 as I became involved in the nuclear weapons program.

I am not the world's expert on the most modern nuclear weapons, although I continue to be involved in both the technical and policy side, but I do have a good feel for first-generation and improvised nuclear explosives-- particularly relevant in this time of concern about proliferation and terrorism.

On the nuclear power side, I worked at Los Alamos on hydrogen-heating graphite and metal reactors for rocketry, and in 1975 on the American Physical Society Study Group on Light-Water Reactor Safety. Nuclear fission is one of the few currently available options for greatly expanding the world's access to energy, but the very nature of the fission process imposes constraints and responsibilities.

Every session of this Meeting would be of interest to me, but I am not here to compete with the technical content and expertise of those sessions. I greatly appreciate the talent and effort-- the physics and engineering embodied in power reactors are tremendous technical accomplishments.

From the very beginning in the U.S. program, theoretical analysis played an essential role in fission systems. This began with the determination of the linkage between infinite-medium criticality and the exponential spatial fall-off in piles a couple of meters across. It was evident in the lattice design essential for the production reactors, and the thermo-hydraulic design of the Hanford piles.

The field came into its own with the design of propulsion reactors using enriched uranium, and the commercial reactors that followed. In contrast, and unfortunately, the technology of nuclear weapons is much simpler than that of commercial reactors. In the course of the explosive fission reaction, there is no need for cooling; no control employed or possible (other than design); and little need for accident analysis.

Of course, nuclear weapons must be safe and secure before they are commanded to explode, and an element of that is a tricky and perhaps 3-D calculation-- for one-point safety, for instance. For the rest, weapon safety and security results from the components and systems supplied by Sandia National Laboratories to ensure that conceivable accidents could result at most in single-point detonation of the explosive, with assurance by the weapon design labs that no significant nuclear yield would result-- to be precise, less than the energy equivalent of 2 kg of high explosive.

Improvised nuclear explosives in terrorist hands are likely to have few of these features. Of course, early U.S. nuclear weapons kept the fissionable core apart from the weapon, and that might be the option chosen for an improvised nuclear explosive.

The world's 300 GWe-class power reactors provide about 20% of the electricity and 7% of the world's energy. Considering desirable and feasible improvements in efficiency of energy use, the world 50 years from now might consume twice as much energy as it does now. If 50% of that were to be supplied by nuclear power, some 4000 GWe reactors would be needed and at 200 tons of natural uranium per year would burn about 0.8 million tons per year.

Current assured reserves would last for four years, but with higher cost terrestrial ore (of perhaps 200 MT) exhaustion could occur in 300 years. And if two billion tons of uranium were extracted from the ocean, one could operate such a large industry for 2000 years. Estimates of cost of uranium from seawater, by groups working in the field, range from $100 to $260/kg; estimates of the cost of saving uranium by reprocessing of LWR spent fuel range from $750/kg of uranium saved. So it is truly important to learn the assured cost ceiling on uranium from seawater, even if it is assuredly uneconomical right now.

A change to breeder reactors when it became economically desirable could extend the supply to 200,000 years. As I emphasized in 1977, there is no hurry about the breeder, until high-cost but affordable resources are perhaps 50% exhausted. One could then enrich some of the remaining uranium and use the exhausted (depleted) uranium in a U-235 seeded Pu/U breeding cycle to extend the supply by a factor 100-- hence the "200,000 years."

Because a breeder uses only one percent as much uranium as does a burner, natural uranium at $3000/kg would be as affordable for a breeder as uranium at $30/kg is for an LWR-- i.e., in cost per kWh.

A rosy future for nuclear power requires:

Fuel availability should be no problem until the industry is much larger.

Spent fuel can predictably be held for many decades in interim repositories, above ground or perhaps underground. Ultimate disposal should be in commercial, competitive mined geological repositories. These could be certified by the IAEA and secured by international status and forces. Similarly, the spent fuel and other forms of radioactive material for deposit in the mined geologic repository should be certified by IAEA. Beyond that, it should be a commercial, moneymaking proposition, providing competitive choices for the reactor operator or fuel supplier.

Risks of normal operation are at present acceptable. Routine reprocessing provides substantially more radiation exposure than does the operation of the reactor, but a primary contribution comes from the mill tailings which have not been properly handled, for the most part. Yet they can be, and should be, once lives anticipated to be lost to low-level radiation exposure are taken into account. Although reprocessing of LWR spent fuel does not solve any of the problems facing the nuclear industry, it would be essential for a breeder economy and should be studied for that role.

Not related to nuclear power-- indeed a hazard that nuclear power is helping to reduce-- is that of proliferation or terrorist use of excess weapon materials. Although excess weapon plutonium is a problem, excess high-enriched uranium is far more hazardous, since it is much easier for a nation or even a terrorist group to build a first-generation nuclear weapon from HEU metal. Such a weapon of full yield need weigh no more than 500-1000 kg, and, unfortunately, it is not difficult to manufacture. For both HEU and excess weapon Pu, consolidation of stocks with improved security is the most urgent and practical step.

Beyond that, the ongoing 20-year purchase of 500 tons of Russian HEU blended down in Russia for use in U.S. reactors as 4.4% LEU is a real contribution to nonproliferation and to international security. Two improvements are urgent and feasible. First, much more HEU exists and should be contracted. But since there is insufficient market in the United States for the resulting LEU, all HEU surplus to its use in nuclear weapons or naval propulsion should be blended down forthwith to 19.9% (where it is difficult to use in nuclear weapons) and from which level it could be further blended for use in LWRs or in breeder reactors, for that matter. The blending to 19.9% requires handling only one-fifth as much UF6 as the blending to 4.4%-- even less, because blending to 4.4% is done with 1.9% U-235 and not with depleted uranium.

Further, the industry ought to loosen the standards on U-234, because the LEU from HEU contains about twice the acceptable 1000 ppm of U-234. But this is an arbitrary level, and the industry and the workers should be able to agree on loosening these standards and maintaining adequate protection for the workers.

Cost is tremendously important; it would be helped by a carbon tax, as the costs of accepting or of countering global warming induced by greenhouse gases are recognized and taken into account. A more highly competitive industry, which could happen with modular reactors would help to reduce costs, if only by a normal learning curve. But greater reactor sales will inspire more innovation as well.

One of the greatest impediments to the expansion of nuclear power is the risk of catastrophic accident. It is not just the possibility of irrevocable loss of the plant (as was the case with Three-Mile Island) but the denial of large amounts of territory and the substantial population dose given to the citizenry in case of an accident resulting in the release of much of the radioactive inventory in the reactor. According to the ICRP coefficient of 0.04 cancer deaths per Sv, the 600,000 person-Sv contributed by Chernobyl is expected to result in some 24,000 deaths from cancer. This mortality is evidently undetectable among the 50 million deaths from cancer expected from natural causes in a population of 250 million; but it is real nevertheless.

I know there is substantial argument in the industry as to whether low level radiation contributes any mortality at all, or whether such cancer mortality is compensated by other improvements in health due to low-level radiation. While such possibilities are not excluded by human experiments or statistics, my advice to the industry is to accept this estimate of 0.04 deaths per Sv, and to show how normal operation and even the occasional disaster still leave nuclear power with a health benefit over competing sources such as coal.

I am open-minded on the breeder reactor and also on near-breeders in which the neutron deficit of the Th/U-233 cycle is made up in a fast reactor by neutrons from accelerated protons. The neutron gap could be closed by feeding a modest amount of excess weapon Pu, HEU, or Pu and lesser actinides from reprocessed LWR fuel. It is largely a matter of cost. But such systems hardly reduce the inventory of fission products, which are the primary concern in a reactor accident and potential melt-down. Of course, even designs which reduce considerably the apparent probability of melt-down should not be adopted if they, instead, present a substantial possibility of accidental criticality, as might occur in a fast reactor if portions of the core could increase in mean density by the collapse of cooling channels, for instance.

But how will the industry obtain the research and development results which can lead to new generations of cheaper, safer reactors? Expanded government support is warranted-- not only in nuclear power but in many other aspects of modern life. Such is surely the case in countering terrorism, and terrorism can be a substantial problem for the nuclear industry.

I served on the National Academies' Committee on Science and Technology for Countering Terrorism and on its Nuclear and Radiological Threat Panel, but I had long been involved in considering terrorist threats. Indeed, the Panel judged that some of these threats are real, and they must be countered. Such threats include sabotage by insiders; larger terrorist teams than are involved in the NRC requirements; and explosives delivered by light aircraft, for instance.

I served two four-year terms on the President's Science Advisory Committee in 1962-65 and 1969-72, as well as a term on the Defense Science Board. The marvels of the capitalist system are to be admired, and especially the result of dizzying incremental improvements in consumer products, such as personal computer technology. But advances such as the Global Positioning System-- GPS-- and reconnaissance and some communication satellites depended on government support. Yet recent recommendations by the President's Council of Advisors on Science and Technology have resulted in minimal increases of funds for the nuclear power sector. But many technologies are too far from the market to be funded by the commercial sector.e.g., seawater uranium; but also superconducting power lines in the 100 GW range (which I published in 1962 and again in 1965).

Furthermore, compensation for individuals going into engineering and science pales beside that for the law, business consulting, and the like.

Perhaps the collapse of the bubble of the last decade will help get things right, and allow society to value properly the real contributions of scientists and engineers to real progress. The game is not beating the other guy, as in sports, for which the top dog receives enormous compensation. It should not be about elevating the annual compensation of a CEO to hundreds of millions of dollars, when there is no evidence that he or she is better than a dozen other people in the company.

Not only our talented foreign students from Asia should be involved in US scientific and technical advances, but Americans as well. More attention to valuing and recognizing not only scientific achievements but engineering innovation and excellence will help to achieve this, but providing real reward for real contributions would be a major step.