The Maintenance of Nuclear Weapon Stockpiles Without Nuclear Explosion Testing by Richard L. Garwin IBM Research Division Thomas J. Watson Research Center P.O. Box 218 Yorktown Heights, NY 10598 (914) 945-2555 (also Adjunct Professor of Physics, Columbia University) September 22-24, 1995 24th Pugwash Workshop on Nuclear Forces "Nuclear Forces in Europe" London ENGLAND _______ ABSTRACT. Sub-kiloton tests that might add to confidence in continued weapon reliability contribute much less than enhanced surveillance and an intensive stewardship program; so-called hydronuclear tests contribute even less. The primary remedy is remanufacture to within the original tolerances, using the same materials; either on schedule or on signal from detailed inspection and non-nuclear testing of the weapons. Committed to a Science-Based Stewardship Program to maintain skills and capabilities, President Clinton announced 11 August 1995 that the US would seek to negotiate a "true zero yield comprehensive test ban", and President Chirac seems to have adopted a similar position, but following the final series of 8 tests in the Pacific. U242PUGW 083095PUGW 09/26/95 INTRODUCTION. Among the many issues regarding nuclear explosion testing, one has become both popular and timely-- the degree to which tests at various yield levels contribute to confidence in maintaining a stockpile of nuclear weapons of existing type.(1) Following the indefinite extension of the NPT at the conference in New York May, 1995, there was some surprise expressed at President Chirac's decision to conduct a final series of eight nuclear explosion tests in the Pacific, and additional surprise(2) with the revelation that a "Comprehensive Test Ban" was considered by at least some of the nuclear weapon states to permit not only so-called hydronuclear tests of yields traditionally up to 2 kg of high explosive equivalent,(3) but 100 tons.(4) The level of concern increased when it appeared that some in the United States government and Congress argued that continued testing with nuclear explosion yields of several hundred tons or more would be required to maintain confidence in the U.S. nuclear weapons. Some observed that an explosion yield of several hundred tons in a weapon weighing a few tens of kilograms would constitute a very significant military capability, especially in view of the fact that for a decade or so there has been a substantial minority view in the United States urging the development of just such weapons, of yield of 5 tons, 50 tons, or 500 tons. In the United States, the Department of Energy contracted with the MITRE Corporation, requesting the JASON(5) group of ---------------- 1 Some of the other issues are those of understanding of weapons effects, the potential for acquisition of fission or thermonuclear weapons without testing, the degree to which nuclear testing seriously undermines the commitment to non-nuclear weapon status, the degree to which international security is impaired by preventing the acquisition of safer nuclear weapons or those with more advanced characteristics. 2 In technical parlance, some were "shocked! shocked!" 3 One kg of HE contributes about four megajoules (MJ). 4 I use "tons" to mean 1000 kg (more properly, "tonnes"). 5 The JASON Group of consultants to the U.S. government was formed in 1960 on the initiative of the number of senior scientists and advisors to the U.S. government, U242PUGW 083095PUGW 09/26/95 PAGE 2 consultants to the U.S. government to do a classified study of exactly this question of the utility of testing at various yield levels in order to maintain confidence in the safety and reliability of the enduring U.S. nuclear stockpile. THE JASON STUDY. Sidney D. Drell(6) agreed to chair this study, as he did a major study in 1994.(7) In the case of the 1995 JASON Nuclear Testing Study, Sidney Drell enlisted as full-time members of the study group, and as authors, four individuals with long careers in the U.S. nuclear weapon laboratories. Seymour Sack of Livermore and John Richter of Los Alamos ---------------- with the purpose of involving highly capable younger people, largely physicists, in national security affairs. It has been attached administratively to the Institute for Defense Analyses, SRI International, and now the MITRE Corporation. As suits largely academic scientists, most of the JASON work is done during a six-seven week summer study, plus two 2-3 day sessions in Washington DC in Spring and Fall. There are now about 50 JASON members, with the activities overseen by a Steering Committee of the membership. JASON does work for the various departments and agencies of the U.S. government, including the Department of Defense, Department of Energy, U.S. Navy, and the like. About half of the JASON reports are unclassified, and especially in recent years an effort is made to have an unclassified version of classified reports, as well. Of the order of ten substantial reports are prepared during a typical JASON Summer Study, and with a given JASON member contributing to and listed as author of one or more studies. Typically the sponsoring agency provides some background material (often classified), and meetings ("briefings") are held to fully explore the problem or the technical field. 6 Professor of Physics at the Stanford Linear Accelerator Center, a member of JASON since the beginning in 1960, former member and Vice Chairman of the President's Science Advisory Committee, and a person with long experience in national security affairs. 7 "Science Based Stockpile Stewardship," JASON Report JSR-94-345 by S.D. Drell, Chairman, C. Callan, M. Cornwall, D. Eardley, J. Goodman, D. Hammer, W. Happer, J. Kimble, S. Koonin, R. LeLevier, C. Max, W. Panofsky, M. Rosenbluth, J. Sullivan, P. Weinberger, H. York, and F. Zachariasen (110094SBSS). (November 1994, 107 pages. Unclassified, distribution unlimited). U242PUGW 083095PUGW 09/26/95 PAGE 3 have long experience in the design of the "primaries" of thermonuclear weapons; John Kammerdiener of Los Alamos is a major designer of the "secondaries" of thermonuclear weapons, and Robert Peurifoy, retired Vice President of Sandia National Laboratories, was in charge there of all of the non-physics aspects of stockpile nuclear weapons-- especially the Arming, Firing, and Fuzing, as well as packaging, and the like. Drell and Peurifoy published a recent authoritative article on stockpile maintenance.(8) By July 16, 1995 (the 50th anniversary of the Trinity explosion), there was consensus on the details of the Summary and Conclusions of the Study, and over the next few weeks Drell briefed top US Government officials, civilian and military, including the leaders of the 3 nuclear weapon laboratories. Copies of the Summary and Conclusions were made available to all of these briefing recipients. That four-page document was unclassified, but it was still "owned" by the Department of Energy and had not been formally cleared for release. However, on Friday morning, August 4, during a debate on the floor of the U.S. Senate, one Senator read from a copy of the Summary and Conclusions, citing an excerpt (out of context) to buttress his support of a $50 M allocation for hydronuclear testing of weapons at the National Test Site in the state of Nevada. The Department of Energy then made the full text available to Senator Exon and so the JASON Nuclear Testing Report was in this way officially released in record time on August 4, 1995. A copy is attached to this document.(9) The JASON Study was by far the most exhaustive, independent look at the question thus far.(10) ---------------- 8 "Technical Issues of a Nuclear Test Ban," by Sidney Drell and Bob Peurifoy in Annu. Rev. Nucl. Part. Sci __________________________ 1994, 44:285-327 (000094TINT). 9 I am most pleased to be an author of this document. I worked on fission weapons, thermonuclear weapons, and nuclear weapon testing at Los Alamos for 3-5 months each summer for a decade or so, beginning in 1950. I had a lot to do with the design (physics, cryogenic, mechanical) of the 10-MT Mike test of November 1, 1952, and of the five flyable copies (also fueled with liquid deuterium) that were kept in readiness to be delivered by B-36 aircraft. 10 Important earlier assessments had been made, for instance, by Raymond E. Kidder in "Maintaining the US Stockpile of Nuclear Weapons During a Low Threshold or Comprehensive Test Ban," October 1987 (SECRET RD), and earlier statements of position and fact may also be familiar: Letter by Richard L. Garwin, Norris E. U242PUGW 083095PUGW 09/26/95 PAGE 4 POSITIONS OF THE NUCLEAR WEAPON COMPLEXES IN THE VARIOUS NUCLEAR WEAPON STATES. There has been evolution in this field. For instance, in December 1993 the Assemblee Nationale(11) published a report that argued that "simulation facilities" would be required for stockpile maintenance in the absence of nuclear tests, and that because of the decade or so required for France to obtain these facilities, nuclear tests would need to occur after that date in order to calibrate the simulation facilities. However, responding to concerns evoked by French preparations for the series of eight nuclear explosions to take place September 1995-April 1996, an Elysee communique of 16 August referring to President Chirac's letter to leaders of Pacific states describes the position of France as "l'interdiction de tout essai nucleaire quelqu'en soit le niveau."(12) On 10 August, the French representative to the Conference on disarmament in Geneva stated that France has adopted the formulation "to prohibit any nuclear weapon test explosion or any other nuclear explosion" under the CTB. If this was intended to specify "quelqu'en soit le niveau" it was truly a leadership position for France. The following day, 11 August, President Clinton stated "Today I am announcing my decision to negotiate a true zero yield comprehensive test ban. ... The United States will now insist on a test ban that prohibits any nuclear weapons test explosion, or any other nuclear explosion." The technical debate within the United States government has thus resulted in the decision that "a true zero yield comprehensive test ban" achieved in 1996 is compatible with U.S. nuclear weapons remaining safe and reliable for as long as we need them. But standing alone, the formulation "to prohibit any nuclear weapons test explosion, or any other nuclear explosion" seems previously to have been interpreted by the nuclear powers as capable of allowing "permitted experiments" up to ---------------- Bradbury, and J. Carson Mark to President Carter regarding nuclear weapons stockpile reliability under a Comprehensive Test Ban Treaty (published in the Congressional Record 08/16?/78) (081578LGBC). 11 "La simulation des essais nucleaires" ("Simulation of Nuclear Tests") by Rene Galy-Dejean, Report No. 847, 15 December 1993, (121593.RGD). 12 I.e., banning any nuclear test of whatever energy release. U242PUGW 083095PUGW 09/26/95 PAGE 5 2 kg of equivalent high explosive yield, or 10 tons, or even 100 tons or more-- to my mind impairing substantially the non-proliferation benefits of a test ban. In November 1994, Ray Kidder, Christopher Paine, and I spent several days in Paris discussing precisely this question of stockpile maintenance and nuclear testing with individuals in the Assemblee, Senate, Ministry of Defense, staffs of the President and Prime Minister, and the CEA. Soon after, we provided a detailed report on these discussions(13) which was distributed to our interlocutors as well as to substantial numbers of people in the United States and elsewhere. On July 14, 1995, the French newspaper Liberation published a three-page article quoting ___________ extensively from this "FAS-NRDC" report. In consequence, Jacques Bouchard, Director of Military Applications for CEA, published a long reply August 3. It is noteworthy that Bouchard justifies the test series as follows: "To ensure the future of our current system of weapons, in remanufacturing the out-of-date weapons at the end of 20 years, it will be necessary to use more robust designs, less sensitive to variations in the production parameters. The prime objective of our test series is to qualify these robust designs for the future. Let us note in passing that these robust designs go in an opposite direction from the improvement of performance and sophistication. We are far from seeking miniaturization, new weapon effects, or other gimmicks claimed by some to be the only justification for the coming test series-- a claim made in ignorance of the arguments set out above." However, he then used about 40% of his article to attack the expertise and motivation of myself, Kidder, and Paine. The following excerpts from my reply(14) show my technical and policy position: My own advice to the U.S. Government (and to France), published since 1978, takes into account that atoms do not age, and to the extent we desire to do so, we can ---------------- 13 "FAS-NRDC Discussions in Paris Regarding the Necessity of Nuclear Tests for Maintaining a Reliable French Nuclear Force Under a Comprehensive Test Ban" (12/28/94 FINAL), by R.L. Garwin, C.E. Paine, and R.E. Kidder, 11/02-07/94, Paris, FRANCE (122894RCTB). This Report of 22 pages is supplemented by Appendix A1-A4 (Garwin), B1-B6 (Paine-- detailed comments on the Galy-Dejean report), and C1-C12 (Kidder-- comments and background information concerning hydronuclear tests, etc.) 14 Which may or may not be published in Liberation by the ___________ time of the London Pugwash Workshop. U242PUGW 083095PUGW 09/26/95 PAGE 6 make and assemble nuclear warheads as we have in the past. Of course, over the years we have made and will continue to make major changes in the electronics and other systems outside the "primary" nuclear explosive of the warhead and the "secondary" thermonuclear charge. But we should use new manufacturing techniques only to remain more assuredly within the historically allowable range of dimensions and other parameters, and we should resist the lure of new materials and processes for the interior of the nuclear warheads. Our own report of our discussions in Paris states (p.15): "We replied that personally we could appreciate the rationale for such a program, limited in both time and the number of tests. ... we indicated that it would be important in announcing any test series that the French government state its purpose was not to improve the performance or broaden the range of effectiveness of its nuclear weapons, but simply to modify them so that confidence in their performance and eventual remanufacture could be sustained without nuclear explosive testing." On Australian national radio 13 February 1995, I said: "So if the French go ahead with these tests, and they say 'look we're not trying to make our weapons more usable, we're not trying to expand their utility or their range of yields, we are just trying to make sure that we can get along under a test ban treaty with safe and secure weapons,' then I think that even the U.S. will be able to resist the temptation to test. ... Australia is a participant in the Non-Proliferation Treaty. They are involved in the negotiations for a Comprehensive Test Ban and that's where the main push should still be. Now I wish the French would see it my way and decide that they are competent enough to maintain their nuclear stockpile without tests, but it may go the other way." In summary, if existing French nuclear weapons are "parfaitement garanties" neither the promised test series nor continuing low-yield nuclear explosions are necessary to ensure that replacement weapons themselves are "parfaitement garanties." But if officials in France will accept that judgement only after this series of 8 explosions, then this test series is not only acceptable but desirable, in permitting a new era in which there are no nuclear explosions of any kind, worldwide, borrowing a phrase from President Chirac "quel qu'en soit le niveau". U242PUGW 083095PUGW 09/26/95 PAGE 7 COMMENTS ON THE JASON REPORT RELEASED AUGUST 4, 1995. Conclusion 1 says that the United States can have high confidence right now in safety, reliability, and performance margins of those weapons that are designated to remain in the stockpile. This is the same position expressed by Jacques Bouchard in regard to current French weapons, but he states the need to modify the weapons so that they are more robust against manufacturing variations; he does not state unambiguously that after testing these modified weapons no ongoing nuclear tests would be needed. Noteworthy are the key assumptions underlying the JASON Report-- maintain credible nuclear deterrent; commitment to supporting world-wide non-proliferation efforts; no new military or political circumstances that will force the abandonment of the policy not to develop any new nuclear weapon designs. Conclusion 2 emphasizes that the U.S. will need to intervene to maintain confidence in its stockpile, and this intervention will need to take the form of enhanced surveillance; inspection by disassembly and destruction of components; replacing age-affected components, and evaluating the resulting replacement. Conclusion 3 states that opportunities have been identified to further enhance "performance margins" without nuclear test, for instance by modifying the composition of the "boost gas" or replacing it more frequently. The text of the Summary and Conclusions discusses in some details the utility of testing held below a kiloton yield, which is apparently substantially less than the full yield of a weapon primary, in order to "allow study of boost gas ignition and initial burn." A number of uncertainties would need to be answered in the affirmative for such partial-yield tests to be a useful tool for stockpile maintenance. First, it would need to be established through a statistically valid set of such nuclear weapon explosions, that substantially similar weapons in the current stockpile (in which one has "high confidence") produce similar diagnostic signals (yield, etc.) in this test of "boost gas ignition and initial burn." For instance, if the initial burn process were less ____ sensitive to variation of parameters than the full primary yield, it would not be a leading indicator of problems. And if the initial burn was more sensitive than full primary ____ yield, to the extent that it essentially vanished in some perfectly good existing primaries, it would very likely provoke a redesign that would need full primary-yield testing. U242PUGW 083095PUGW 09/26/95 PAGE 8 Assuming that initial tests and analyses actually would show "initial burn" explosion testing to be a valuable diagnostic, it is logically compelling to note that these tests would need to be performed not only on the aging weapons, but on the remanufactured weapons. Thus the CTBT would be transformed into a threshold test ban treaty, and at a yield that would constitute usable military capability (even if 500 tons is much less than 50 kilotons). Conclusion 4 makes this observation and then explicitly compares sub-kiloton testing with the detailed information that would come from enhanced surveillance and remanufacturing and with the enhancement of performance margins previously discussed. Conclusion 4 then observes that information from sub-kiloton testing has some value, which must be weighed against the required expenditures and also the political impact on non-proliferation goals. Conclusion 5 addresses the question of tests at yields too low to initiate boosting. Logically, to hold the yield of a boosted primary below the level required to initiate boosting demands modification of the primary for purposes of test, beyond simply removing the boost gas. Typically, such experiments fall in the class of "hydronuclear" tests, which might be loosely defined as those in which the explosive yield due to fission energy release is less than the explosive energy of the high explosive. In the United States, the only such tests were done during the moratorium 1958-1961. Of these, 35 were done at Los Alamos in shallow holes, with the yield limited to less than 2 kg of high explosive.(15) Major modifications of the implosion are required to reduce the yield to this level. Possibilities are: o Precise sub-scale implosion systems. o Reduced explosive amounts driving standard "pits". o Standard explosive assemblies with less-reactive pits. o Extraneous materials to reduce criticality. These modifications are referred to in the Report as "... changes that drastically alter the primary implosion." ---------------- 15 In fact, the maximum yield observed was about ten times less. U242PUGW 083095PUGW 09/26/95 PAGE 9 The 1958-61 series of hydronuclear tests(16) do have undisputed utility in the development process of a nuclear ___________ warhead. If a nuclear warhead is required to be inherently __ "one-point safe", then normal firing to obtain full design yield must require the detonation of the explosive at two or more separated points. The qualification process then must ensure that detonation at the "worst" single point will not produce a nuclear yield.(17) The final hydronuclear proof test is conducted essentially with a real nuclear weapon, with the detonation ensured to be initiated at only a single point. However, particularly in the 1960 era, 3-D computational capabilities were not adequate definitively to establish that a given design was one-point safe, hence the need for demonstration. To conduct such tests in shallow holes in the backyard requires very high confidence that the yield limit will not be exceeded, and this can be obtained by a "creep up" process, in which more and more explosive is added (or more and more Pu), firing each model at the "worst point" and obtaining confidence that the next step will stay below the test limit. Of course, inherent one-point safety is not the only way to obtain a nuclear stockpile proof against accidental detonation; in some designs, the fissile material can be kept separate from the explosive until the missile has been launched or the bomb dropped from the airplane, or a solid obstruction can be put into the hollow pit, removed just before detonation, etc. Conclusion 6 refers to detailed study of all of the U.S. nuclear weapons that had problems and which involved nuclear tests to resolve. These "were primarily the result of incomplete or inadequate design activities" and this detailed review is the basis of the judgement that the "weapon types in the enduring stockpile are safe and reliable in the context of explicit military requirements." Conclusion 7 takes into account the possibility of adverse miracles, noting that all modern treaties include a "supreme national interest" clause that could be invoked to permit even full-scale nuclear tests (by withdrawal from the CTB) in case of severe "unanticipated technical problems in the enduring stockpile" that could not be handled in any other way. ---------------- 16 And about 20 additional ones were conducted during that same period by the Livermore Laboratory at the National test Site. 17 More precisely, in the United States the requirement is that it have less than one chance in a million of producing a nuclear energy release greater than 2 kg of He. U242PUGW 083095PUGW 09/26/95 PAGE 10 DEFINITION OF "ZERO". An Indian text for the CTBT would prohibit "any nuclear weapon explosion or any other nuclear test explosion or any release of nuclear energy caused by the assembly or compression of fissile or fusion material by chemical explosive or other means."(18) I personally would leave out the "or fusion material" simply because of the large effort around the world toward Inertial Confinement Fusion (ICF) which is largely directed toward civil nuclear energy. Whether or not this will succeed to be an economic source of energy is highly debatable, but it shows no promise of being a weapon of any kind. It is not possible to limit to strictly zero the "fission energy release" during any handling of fissile material, including sub-critical implosion testing. Three kg of Pu-240, for instance, has a spontaneous fission rate of almost one million per second, or about 20 microwatts and pulsed reactors might deliver 50 kJ per kg in a fraction of a microsecond. Evidently, monitoring such undertakings with the purpose of verifying adherence to a CTBT will depend on societal verification.(19) The typical 1000 MWe nuclear power reactor delivers 1 GJ of electrical energy each second, from more than 3 GJ each second of fission heat. At 4 GJ/ton of high explosive, this is some 0.8 ton per second. It would take me too far afield to recommend here suitable treaty language, But now that President Clinton has announced the US decision to "negotiate a true zero yield comprehensive test ban" and President Chirac seems to have communicated the same decision, whatever the definition of zero, it should ban explosively driven or other rapidly assembled systems of fission yield sufficient to begin to melt the fissile material-- on the order of 100 grams of high explosive equivalent. ---------------- 18 Rebecca Johnson, Test Ban Treaty Report, The Bulletin of _______________ Atomic Scientists September/October 1995 (pp. 11-12). _________________ 19 "Societal Verification," by J. Rotblat, Chapter 6 in the 1993 Pugwash book A Nuclear-Weapon-Free World: ________________________________ Desirable? Feasible? edited by J. Rotblat, ______________________ J. Steinberger, and B. Udgaonkar, pp. 103-118 (000093.JR1). U242PUGW 083095PUGW 09/26/95 PAGE 11 ACCEPTABILITY OF A TRUE ZERO-YIELD CTBT. It might be argued, as did the 1993 report of the Assemblee Nationale, that while the United States could maintain confidence in the safety and reliability of its nuclear weapon stockpile under a CTBT, France could not, lacking the "simulation facilities." Many in the public took this to mean that the simulation facilities would actually provide the equivalent of nuclear explosions, and thus replace nuclear explosion testing. U242PUGW 083095PUGW 09/26/95 PAGE 12 In fact, French and U.S. nuclear weapon experts agree that the "simulation" that is important could more properly be called "modelling" of the nuclear weapons on the computers used to design them, and that is the first resort if questions arise as to the necessity of remanufacture. It is clear that intensive surveillance of the stockpile, including nondestructive and destructive testing is required to detect age-related deficiencies. Many of the parts of an intact nuclear weapon can be removed and actually tested-- the radar, the firing set, the boost gas system, and the like, and these can similarly be upgraded (and fully tested) without reducing confidence in the performance of the deployed weapon. Indeed, much of the ancillary equipment is not used during an underground nuclear test, such as that enforcing the "environmental sensing", etc. But will China, Russia, and Britain be able to match the United States, without the Science-Based Stewardship Program? Will they be able to accept a zero-yield CTBT? Clearly the answer is yes. No nation has ever done enough nuclear tests for the test itself to have served as a reasonable statistical indicator of weapon reliability. Every nuclear weapon state must have depended upon non-nuclear test firings for development of its nuclear weapons, and on surveillance, disassembly, and remanufacture for maintaining weapons in usable condition. The difference under a CTBT is that "remanufacture" replaces "manufacture" of weapons of new type. One often hears an argument like the following:(20) "The technology of a modern nuclear warhead is complex and very advanced, and the sophistication achieved in the most recent weapons has led to a strict optimization, leaving little room for modifications. If one attempted to reproduce these sophisticated weapons in 15 or 20 years, one would inevitably introduce modifications because the technical means of production will have changed. For example, the microprocessors that now permit us to fabricate very precisely certain components did not exist 20 years ago. They will be replaced in a few years by systems of a future generation, and it is impossible to go backwards. Certainly, we cannot predict the extent of these modifications, but we also cannot be sure that they will be sufficiently small that the guarantee of reliability will continue to be absolute, as it is for our current warheads." ---------------- 20 J. Bouchard, Liberation, 3 August 1995. ____________ U242PUGW 083095PUGW 09/26/95 PAGE 13 It is true that we can machine things much more conveniently and more accurately than we could 20 years ago, but it is also true that these new machine capabilities can be programmed to produce variations in topography, dimension, etc., like those in the original weapons. To take another example, to do the central development problem on a nuclear weapon that is rotationally symmetric about an axis requires 2-D computation. To compute that same weapon imploded by detonation of the HE at a single point requires in general 3-D computation. And if an age-related defect is discovered, the assessment of its importance would require in general 3-D computation. Some nations might actually carry out such computation and decide (as a result) that the age-related defect is of tolerable magnitude. Lacking that level of computational capability, another nation might decide to remanufacture the affected component. It is a matter of taste, prudence, and resources. In general, personally, I tend toward the conservative remanufacturing solution, which may be less exciting but is available to all. U242PUGW 083095PUGW 09/26/95 U242PUGW 083095PUGW 09/26/95