A Different Kind of Complex: The Future of U.S. Nuclear Weapons and the Nuclear Weapons Enterprise

Richard L. Garwin

In an October 28 speech to the Carnegie Endowment for International Peace entitled "Nuclear Weapons and Deterrence in the 21st Century," Secretary of Defense Robert Gates noted the continued importance of U.S. nuclear weapons for deterring possible opponents and for reassuring allies that they do not need to develop their own weapons. He argued that, to carry out these responsibilities, a Reliable Replacement Warhead (RRW) as well as a modernized complex for nuclear weapons that would allow the building of new weapons without nuclear explosion testing are needed.

I have great admiration for Secretary Gates and suggested months ago that he should be urged to continue to serve in an Obama administration. His dismissal of the Air Force secretary and chief of staff over negligence in management of nuclear weapons (see page 44) was a rare and appropriate action. In regard to the RRW program and other nuclear weapons questions, however, the leadership of the Department of Defense is ill served by its advisers on nuclear warheads, who appear not to be conveying to the secretary the judgment of the nuclear weapons labs that the plutonium pit of each U.S. nuclear weapon is expected to last at least 85 years.

Gates' recent remarks echo a 2007 report of the Departments of Defense, Energy, and State that suggested that delays in the replacement warhead program would "raise the prospect of having to return to underground nuclear testing to certify existing weapons."[1]

The chairman and ranking member on the House Appropriations energy and water subcommittee wrote in August 2007 that, "It is irresponsible for the administration to make such an assertion." They correctly noted that that "there is no record of congressional testimony or reports sent to Congress by the Administration claiming...that a resumption of testing to verify the performance of warheads would be a necessity."[2]

Congress ultimately rejected the administration's proposal for an ambitious, multidecade plan to build replacement warheads and a nuclear weapons infrastructure to carry out the program. Not only were some key legislators unconvinced of the technical need for replacement warheads, but they questioned the administration's assumptions about the future role and size of the U.S. nuclear arsenal. The 2008 House report on the fiscal year 2008 energy and water appropriations bill requires the Energy Department to provide a "comprehensive nuclear defense and security plan," a "translation into a specific stockpile," and "a comprehensive long-term expenditure plan."

Indeed, there is great uncertainty over the future of the U.S. nuclear weapon stockpile, despite planning within the government and the Energy Department for a considerably lower number to correspond with the 2002 Strategic Offensive Reductions Treaty-perhaps 2,100 "operationally deployed" strategic nuclear weapons on December 31, 2012. This might represent a total stockpile of 4,000-6,000 bombs and warheads.

The United States should move rapidly to a stockpile of 1,000 weapons of the current type. Vigorous and challenging work in the weapons labs and a smaller and more efficient support complex can maintain the safety and reliability of these weapons.


The RRW program emerged in response to the failed attempt by the Energy Department's National Nuclear Security Administration (NNSA) to pursue another nuclear warhead, the Robust Nuclear Energy Penetrator (RNEP). That effort fell short on Capitol Hill amid first confusion and then later concern about RNEP's capabilities and effects.[3] The RNEP fiasco did little to maintain the reputation of the NNSA for integrity and technical competence.

It also seemed to convince lawmakers that they needed to find another way to provide challenging nuclear weapons design and development work at the three weapons laboratories-Los Alamos National Laboratory, Lawrence Livermore National Laboratory, and Sandia Corporation.

With RNEP gone, attention moved to the RRW program, for which competitive design studies were done at Los Alamos and Livermore. Livermore was selected by the NNSA to move to the next phase of development, and its design was renamed WR1. According to Livermore currently,

The goal of the RRW approach is to replace aging warheads with ones manufactured from materials that are more readily available and more environmentally benign than those used in current designs. RRWs can include advanced safety and security technologies, and they are designed to provide large performance margins for all key potential failure modes. Large margins enhance weapons reliability and help to ensure that underground nuclear testing will not be required for design certification.

After NNSA's selection of the Livermore/Sandia-California design, NNSA and the U.S. Navy began to develop a detailed WR1 project plan and cost estimate. The effort has since been halted. While seeking clarification on a number of related policy and technical issues, Congress stopped funding for RRW work in [fiscal year] 2008.[4]

These are very modest goals in contrast to the frequently heard need to replace warheads about whose reliability and safety there is "increasing concern." This concern is usually expanded to argue that, with the accumulation of small modifications to existing warheads in the Life Extension Programs (LEPs), we move farther from the nuclear explosion test base, and at some point, the warheads will no longer be certifiable, as asserted by Gates:

Our nuclear weapons were designed on the assumption of a limited shelf life and that the weapons themselves would eventually be replaced. Sensitive parts do not last forever. We can and do reengineer our current stockpile to extend its life span. However, the weapons were developed with narrow technical margins. With every adjustment, we move farther away from the original design that was successfully tested when the weapon was first fielded. Add to this that no weapons in our arsenal have been tested since 1992. So the information on which we base our annual certification of stockpile grows increasingly dated and incomplete. At a certain point, it will become impossible to keep extending the life of our arsenal, especially in light of our testing moratorium.[5]

I disagree. The NNSA's $5-billion-per-year science-based Stockpile Stewardship Program (SSP) may be essential in providing the foundation for the labs to design an RRW that might indeed be certifiable without nuclear testing, but that same program has provided the basis over time for increasing, not decreasing, confidence in the performance of these legacy weapons.

A key milestone in this regard occurred in late 2006. Until then, the Bush administration had based its case for the RRW program in large measure on the argument that the United States was incapable of remanufacturing plutonium pits, the core of the primary nuclear explosive in U.S. thermonuclear weapons. The NNSA argued that it would be better to start anew with something that could reasonably be traced to a nuclear test explosion but that would give expanded freedom of design in view of a post-Cold War assumption of relaxed requirements on warhead weight and yield.

Yet, in late 2006, the SSP led to the judgment by Livermore and Los Alamos that the plutonium pit in each of our stockpile nuclear weapons has a life exceeding 85 years, perhaps 100 years. This conclusion was endorsed by a technical study by JASON and was published by the NNSA.[6]

Moreover, any modifications to legacy weapons are not added willy-nilly. They can be simulated and evaluated more confidently than can the total redesign that is an RRW. Gates adds, "As I say, we've been re-engineering our stockpile now for essentially 16 years, and we are okay today. It is the longer-term prospect that concerns me." The secretary should consider the proposition that our confidence in legacy weapons can and should grow rather than diminish. What is more, his advisers should open their eyes to progress over the past 14 years.

The fact that the NNSA can now certify that pits have a lifetime of more than 85 years has removed any urgency to engineer and manufacture the RRW. Proper assessment of the accumulation of small modifications in the LEPs can be done with more certainty on the basis of the SSP than can the certification of a new, untested RRW. It remains to be determined whether an RRW can be certified, but the continued performance of legacy weapons can be more reliably certified than an RRW.

Furthermore, no analysis has been offered to show the security or cost benefits over time that would allow a reasoned decision on the RRW program versus other approaches to further reduce the possibility of theft or misuse of existing warheads, such as enhanced security features in the shipping containers for existing warheads.

It cannot be disputed that an RRW could include additional surety measures not present in the legacy weapons, but no analysis has been provided of the benefits of such measures over the many years before the RRW-1, RRW-2, and perhaps more, have fully replaced legacy weapons in the U.S. nuclear forces. Even if the RRW were perfectly secure against misuse, terrorists could concentrate on the non-RRW weapons so that surety of the entire system would not increase much until the RRW took over entirely. Of course, a U.S. RRW does nothing to increase the surety of Russian or Pakistani weapons. Encouraging other nations to develop RRWs or their equivalent is not something that should be advocated, although they should take steps to increase the surety of existing weapons.

Yet, the work done so far on the RRW program has re-energized the nuclear laboratories and their involvement in the nuclear weapons complex. Such a major effort should be undertaken every five years or so. I know firsthand from my involvement with this program that new insights have arisen from the new focus on simulation and computation. According to the Livermore annual report, one portion of the work on tantalum at high pressures "would have taken more than 20,000 years to run using the largest computer that was available 20 years ago."[7]

Quantification of Margins and Uncertainties (QMU)

The National Academy of Sciences (NAS) was asked by the NNSA, as directed by Congress, to study the application of quantification of margins and uncertainties (QMU) in the national labs.[8] I was a member of the authoring committee. The QMU was introduced by the labs themselves around 1995 in order to objectify the process of analysis and decision by which nuclear weapons are designed and certified, either initially or in the ongoing annual certification process. The QMU has been implemented rather differently at Los Alamos and Livermore and is less formally used at Sandia. It attempts to integrate the underground nuclear testing experience together with simulation and designer judgment.

In general, although the QMU has not been formally defined, it has played a constructive role. One important criterion for which there is margin and uncertainty is whether the explosive yield of the primary nuclear component exceeds that required to drive the secondary explosive to full yield. Using the best estimates of the explosive output of the primary and the required drive for the secondary, one defines in this way the margin between the two by which the primary yield exceeds the requirement. There are uncertainties in the yield of the primary and the need of the secondary, and differences arise as to how to combine these uncertainties. Clearly, even with less margin than desired, in only one-quarter of the cases (by probability) is there a significant compromise in the performance of the weapon itself. This occurs when the uncertainty in primary performance leads to a smaller yield and uncertainty in the secondary requirement leads to the need for a larger primary yield to drive the secondary explosive. This additional factor four reduction in probability would mitigate the impact of an apparently less-than-robust margin.

The QMU is clearly a stand-in for an enormous number of "button-to-boom" detailed simulations, with the accumulation of statistics. If the performance gates for which M and U are estimated are appropriately chosen, the QMU approach is apt to be conservative in estimation of weapon reliability. The QMU provides a language for communication among the weapons laboratories and the NNSA that can estimate the benefit of improved boost-gas supply [9] procedures to meet concerns about weapons yield in adverse conditions.

Nuclear Weapons Infrastructure

For many years, the NNSA has been putting forth proposals for modernizing the nuclear weapons infrastructure. In October 2008, the NNSA published its analysis of options to "transform the nation's Nuclear Weapons Complex to make it smaller, safer, more secure, and more cost-effective." This Complex-Transformation Supplementary Programmatic Environmental Impact Statement (SPEIS) puts forward a preferred option for realigning the complex, including a capability to produce 125 plutonium pits per year in Los Alamos on a single-shift, five-day-per-week basis. With even the highly conservative assumption of an 85-year pit lifetime, that target (without an additional expected surge capability to 200 pits per year) could support a stockpile of almost 11,000 nuclear weapons.

If one assumes a target stockpile of 4,000 nuclear weapons from which to field about 2,000 operationally deployed strategic nuclear weapons, the replacement of even one-half the number of current ("legacy") weapons with RRWs would require about 16 years (2,000 warheads divided by 125 warheads per year) after the transformed complex is fully operational in the year 2018 or thereabouts. That is part of the rub with the RRW program. Although the RRW program was supposed to be the basis for complex simplification and downsizing, infrastructure to support the legacy weapons would obviously be required until there were sufficient RRWs in stockpile that all legacy weapons could be dismantled.

It is also clear that the complex cannot be defined or optimized unless a decision is made as to whether 8,000, 4,000, or 999, or 300 nuclear weapons are to constitute the future total stockpile. A commitment to an RRW does not in any way promise to ease the problem of definition and operation of the complex until a quantitative plan is provided and evaluated for building RRWs and replacing legacy weapons.

Improving security of stocks of highly enriched uranium and plutonium against theft and detonation by terrorists should play a far greater role in the complex modification than it has so far, although it is one of the drivers toward consolidation of the complex. If the United States takes so long to reduce the hazard of diversion of its weapons-usable materials, how can it expect an expanded Nunn-Lugar program [10] to significantly reduce the hazard from the stocks of other countries that may be less well protected? Improved surety features in an RRW are far less effective in reducing the overall hazard of terrorist use of nuclear explosives than would be an enhanced program in security of nuclear weapons and materials worldwide, together with massive reductions in nuclear weaponry.

Figure 1 (see print edition) is a recent estimate by the National Resources Defense Council of the U.S. stockpile of nuclear weapons from 1945 to 2008.

The nuclear weapons stockpiles of China, France, and the United Kingdom are estimated to be in the few hundreds and would make up an almost invisible portion of the graph compared to U.S. or Russian holdings. Yet, U.S. national security is imperiled by states with only a few weapons and by the prospect of terrorists acquiring even a single one. Hence the importance of evaluating potential U.S. nuclear weapons activities in terms of their influence on proliferation, access of terrorists to nuclear weapons, and the reduction in the potential nuclear threat to the United States.

Implications for the U.S. Test Moratorium and CTBT

Replying to a question following his speech on October 28, 2008, Gates said, "I think that if there are adequate verification measures, [the United States] probably should" ratify the Comprehensive Test Ban Treaty (CTBT). Although the CTBT is a larger topic than can be fully addressed here, a 2002 NAS study on this topic, [11] of which I was an author, provided satisfactory answers to the question of detection of militarily significant explosive tests in violation of the CTBT as well as maintaining the safety and reliability of the U.S. nuclear stockpile under the CTBT.
The International Monitoring System (IMS), on-site inspections, and transparency measures provided for under the CTBT, combined with U.S. intelligence capabilities, are adequate to detect and deter militarily significant cheating. As the NAS report concluded,

The capabilities to detect and identify nuclear explosions without special efforts at evasion are considerably better than the "one kiloton worldwide" characterization that has often been stated for the IMS. If deemed necessary, these capabilities could be further improved by increasing the number of stations in networks whose data streams are continuously searched for signals.

Underground explosions can be reliably detected and can be identified as explosions, using IMS data, down to a yield of 0.1 [kiloton] (100 tons) in hard rock if conducted anywhere in Europe, Asia, North Africa, and North America. In some locations of interest such as Novaya Zemlya, this capability extends down to 0.01 [kiloton] (10 tons) or less.

In addition, the United States benefits from monitoring capabilities that are currently only available through the CTBT's IMS, including monitoring stations in China, Russia, and other sensitive locations that the United States would otherwise not be able to access. [12]

The NAS panel concluded that the United States "has the technical capabilities to maintain confidence in the safety and reliability of its existing nuclear-weapon stockpile under [a test ban], provided that adequate resources are made available to the Department of Energy's nuclear-weapons complex and are properly focused on this task."

According to the NAS panel, which included three former lab directors, age-related defects mainly related to non-nuclear components can be expected, but nuclear test explosions "are not needed to discover these problems and...not likely to be needed to address them."
Rather, the panel said the key to the stewardship of the arsenal is a rigorous stockpile surveillance program, the ability to remanufacture nuclear components to original specifications, the minimization of changes to existing warheads, and non-explosive testing and repair of non-nuclear components.

Since the publication of the NAS panel's report, confidence in existing warheads has increased over time. In March 2007, Thomas D'Agostino, then acting NNSA administrator, said that the SSP is "working. This program has proven its ability to successfully sustain the safety, security and reliability of the stockpile without the need to conduct an underground test for well over a decade."[13]

Given that it is well-established U.S. policy to maintain our current nuclear test moratorium, it is also clearly in the U.S. interest to solidify the global norm against testing and enhance our capabilities to detect and deter surreptitious testing that could improve the nuclear weapons capabilities of other states.


Substantial nuclear design and capability should be maintained at the national labs: The SSP should focus on the existing advanced computing capability and the modernization and expansion of the computer codes and simulations. In addition, the system ought to be challenged every five years with a competition for the design of simplified nuclear warheads, including a much broader range of options, such as the total elimination of plutonium from U.S. nuclear weapons.

The nuclear weapons infrastructure can be defined only after the target number of nuclear weapons in a particular year is selected. Since 1988, I have argued for an essentially immediate reduction to 1,000 nuclear warheads for the United States and the Soviet Union (now Russia) followed by urgent efforts to negotiate lower caps on the inventories of China, France, and the United Kingdom. The 1,000-nuclear-warhead figure would ultimately include not only deployed warheads but also those in transit, refurbishment, and reserve. Indeed, France and the United Kingdom have limited their deployed warheads, although China has made no quantitative statement about its limits. The Reykjavik-2 initiative, with a long-term goal of an appropriate security structure for eliminating nuclear weapons, argues for massive reductions on this scale as well. Henry Kissinger, Sam Nunn, Bill Perry, and George Shultz are among the leaders of this initiative.[14]

Only by a serious program to lead the way in the massive reduction of stocks of nuclear weapons and of weapon-usable materials, and by an absolute commitment not to have nuclear explosive tests can the United States play a leadership role in eliminating proliferation of nuclear weapons other states, such as Iran and North Korea.


Richard L. Garwin is an IBM fellow emeritus at the IBM Research Center, Yorktown Heights, New York; adjunct professor of physics at Columbia University; and a longtime consultant to the U.S. government on nuclear weapons and military technology.


1. U.S. Secretary of Energy, Secretary of Defense, and Secretary of State, “National Security and Nuclear Weapons: Maintaining Deterrence in the 21st Century,” July 2007, http://nnsa.energy.gov/news/1238.htm.
2. Representatives Peter Visclosky (D-Ind.) and David Hobson (R-Ohio), letter to the Bush administration on the Reliable Replacement Warhead, August 1, 2007.
3. Then-NNSA administrator Ambassador Linton Brooks later apologized for not being clearer that RNEP was neither required nor expected to penetrate many tens of meters into rock, agreeing with outside analysts such as Jonathan Medalia of the Congressional Research Service, Rob Nelson of the Union of Concerned Scientists, and an NNSA-sponsored National Academy of Sciences (NAS) study. They found that it was impossible for a RNEP to penetrate more than a few times its length and that all of the benefit from penetration would come from having the nuclear explosion one or two meters below ground level. This would provide a ground shock equivalent to a nuclear yield 20 times as large as one exploded on the surface of the ground.


There was also no pushback against the NAS conclusion that the radioactive fallout from such an underground explosion would not be contained and could kill between a few people and many hundreds of thousands, depending on the winds and the location of the explosion with respect to population centers. I was a member of the NAS committee and had also analyzed RNEP independently. Although it was outside the scope of the committee’s charge of assessment, I presented arguments why if one really wanted an RNEP, one should not take the approach of extreme “hardening” of the B61-Mod 11 so that it would penetrate into rock or concrete as well as into the earth for which B61-Mod 11 was designed. Rather, one should fit the B61-Mod 11 with a large conventional explosive “shaped charge” adjunct, so that the B61 would not have to penetrate earth at all but could travel in the cylindrical cavity excavated a few milliseconds before by the shaped charge. A trivial modification would need to be made for the weapon to detonate in flight rather than having come to rest.

No interest has been shown in acquisition of this capability, probably because RNEP was intended to provide challenging weapons design and development work rather than fulfill compelling military needs.
4. Lawrence Livermore National Laboratory, “Annual Report 2007,” June 9, 2008, www.llnl.gov/annual07/pdfs/wci.pdf . For example, the excess of predicted yield of the primary explosive of a two-stage weapon over the yield demanded for proper performance of the thermonuclear secondary. Ibid., pp. 7-8.

The report notes that Congressional funding of the RRW program is in abeyance, “while seeking clarification on a number of related policy and technical issues.” Related information from the fiscal year 2008 Omnibus Appropriations Legislation and fiscal year 2009 House Appropriations Committee report is available online here.
5. Robert Gates, speech to the Carnegie Endowment for International Peace, October 28, 2008.
6. “The classified studies looked at pits in each nuclear weapon type and gave specific information on plutonium properties, aging and other information. Overall, the weapons laboratories studies assessed that the majority of plutonium pits for most nuclear weapons have minimum lifetimes of at least 85 years.” NNSA Public Affairs, “Studies Show Plutonium Degradation in U.S. Nuclear Weapons Will Not Affect Reliability Soon,” November 29, 2006, http://nnsa.energy.gov/news/999.htm.
7. Lawrence Livermore National Laboratory, “Annual Report 2007,” p.9.
8. National Research Council of the National Academies, “Evaluation of Quantification of Margins and Uncertainties Methodology for Assessing and Certifying the Reliability of the Nuclear Stockpile,” November 2008.
9. The mixture of deuterium and tritium gases that are supplied to the hollow pit shortly before the high-explosive implosion.
10. The Cooperative Threat Reduction Program initiated in 1992 by Sens. Sam Nunn (D-Ga.) and Richard Lugar (R-Ind.) and implemented in the Defense Threat Reduction Agency.
11. National Academy of Sciences, “Technical Issues Related to the Comprehensive Nuclear Test Ban Treaty,” 2002.
12. David Hafemeister’s recent article brings the CTBT verification debate up to date. See David Hafemeister, “The Comprehensive Test Ban Treaty: Effectively Verifiable,” Arms Control Today, October 2008, p. 12.
13. Thomas D’Agostino, Testimony Before the House Appropriations Subcommittee on Energy and Water Development, March 29, 2007 (hearing on Energy Department’s fiscal year 2008 budget).
14. George P. Shultz, William J. Perry, Henry A. Kissinger, and Sam Nunn, “A World Free of Nuclear Weapons,” The Wall Street Journal, January 4, 2007, p. A15; George P. Shultz, William J. Perry, Henry A. Kissinger, and Sam Nunn, “Toward a Nuclear-Free World,” The Wall Street Journal, January 15, 2008, p. A13.