ACA’s journal, Arms Control Today, remains the best in the market. Well focused. Solidly researched. Prudent.

– Hans Blix,
former IAEA Director-General

Why China Won't Build U.S. Warheads
Share this

Richard L. Garwin

Back to Cox Report Main Page

Despite the uncertainty as to how much information on U.S. thermonuclear weapons has actually been compromised, the charges in the Cox Report are sufficiently serious that one should examine the extent to which the acquisition of such information by China would impair U.S. security. The report focuses primarily on the alleged compromise of design information on the W-88 thermonuclear warhead for the Trident-II (D-5) submarine-launched ballistic missile, and to a lesser extent the W-70 enhanced radiation warhead. The important questions are: What can China do with this information that it could not have done without it, and does it matter to the United States?

A great deal of unclassified information is available on the W-88. It is a modern two-stage thermonuclear weapon that is packaged in a conical reentry vehicle (RV) less than six feet tall with a base diameter of only 22 inches. The explosive yield of the W-88 warhead is about 500 kilotons—some 30 times that of the weapons that destroyed Hiroshima and Nagasaki.

According to the Cox Report, the hard information about China's knowledge of the W-88 comes from a document provided by a "walk-in" who provided the CIA in 1995 with a classified Chinese document containing some still-classified information about the W-88. As noted in the intelligence community's Damage Assessment, it is not known whether China acquired "any weapon design documentation or blueprints."

In order to put in perspective the value of any such information to China, one should review the history and probable status of its nuclear weapons program. China tested its first fission weapon in 1964 (an implosion device fueled by uranium-235) and its first two-stage thermonuclear weapon a little over two years later. This was remarkable progress. All told, China conducted 45 nuclear tests (the same number as Britain) during its 33 years of testing, which ended in July 1996.

The Damage Assessment states: "China has had the technical capability to develop a multiple independently targetable reentry vehicle (MIRV) system for its large, currently deployed ICBM for many years, but has not done so. U.S. information acquired by the Chinese could help them develop a MIRV for a future mobile missile." In other words, while still-classified information on the W-88 might be helpful, there is no reason to believe that China could not have built perfectly adequate warheads for its mobile missiles or a MIRVed missile from nuclear technology that it developed itself, supplemented by facts long declassified about U.S. and other nuclear warheads.

But there are reasons why Beijing would not have sought to build MIRVed missiles in the first place. As stated by the Cox Committee, China has under development a mobile ICBM, the DF-31, which is a smaller missile than its current ICBM, the DF-5A. But, if the motivation for this mobile missile is simply (as suggested by both the Cox Report and the Damage Assessment) a desire to have a secure second-strike capability, multiple warheads may not be necessary or even desirable, since this would increase the value of a missile as a target in a foreign pre-emptive strike.

China's current nuclear doctrine does not require the deployment of large numbers of MIRVed missiles. The United States and Russia deployed thousands of highly accurate RVs in order to be able to destroy with some confidence hardened targets such as missile silos. China's deterrent doctrine requires simply the ability to destroy in a retaliatory strike a modest fraction of the population and industry of a potential enemy. Were China intent on developing a counterforce capability, it could long ago have increased its ICBM force beyond the 20 or so silo-based missiles that can now deliver warheads to the United States. It is likely therefore that the impetus behind the mobile ICBMs is (as the Cox Report implies) to make China's strategic nuclear force more survivable against nuclear attack by the United States.

Moreover, MIRVs are also not the optimal weapons if China anticipates encountering a U.S. national missile defense (NMD) system such as that currently proposed to protect all U.S. territory with hit-to-kill exo-atmospheric interceptors. Instead, China is far more likely to use effective countermeasures (such as light-weight decoy balloons) rather than multiple RVs on its future missiles.

The advanced features of the W-88 come at a price. Its narrow conical RV (the Mk-5) poses serious constraints on the design of the nuclear warhead that it carries. Either substantial payload capacity must be devoted to tungsten or uranium ballast in the nose of the RV, or increased development costs and compromises must be made so that the warhead itself is of small enough diameter to fit well down within the cone. The primary reason for such a narrow cone is to achieve very high accuracy on reentry, in the face of winds. Unless China's strategic nuclear force were to grow greatly in warhead numbers to a point where it could threaten the survival of the U.S. ICBM silos, Beijing would have imposed a significant penalty on itself by deploying such sleek RVs. Very high accuracy is irrelevant to maintaining a countervalue ("city-busting") deterrent.

In addition, as a practical matter, for China to replicate the W-88/Mk-5 combination would require not only major advances in nuclear weapons design and fabrication, but also in much of the ancillary equipment such as firing and fuzing, electronics and tritium gas storage. The impediments to small, modern warheads may lie more in these technologies than in the primary and secondary explosive packages themselves. For all of the above reasons, my own judgment is that, even if China were confident that it had every detail of the W-88 and its Mk-5 reentry vehicle, it would not reproduce the weapon.

In a modern two-stage thermonuclear weapon, the energy released from a boosted primary nuclear explosive (mostly in the form of soft X-rays) is briefly contained within a radiation case. The energy and pressure of this radiation compresses, heats and ignites the secondary explosive. This process, from the implosion of the plutonium in the primary to the compression and ignition of the secondary, is the subject of calculation and experiment. In fact, much of the work on primaries is performed with "hydrodynamic experiments," in which high explosives are used to compress plutonium or simulant materials (with no release of fission energy), with the implosion imaged by intense pulses of penetrating X-rays. The process of compression and ignition of the secondary is also the subject of intense computational effort, but the speed and violence of this process is such that high-explosive experiments have no relevance.

Over the decades, computer codes have been used to gain an understanding of this process. The design of a thermonuclear weapon is an interplay between invention and experiment, but because full-scale nuclear explosions provide only limited diagnostic opportunities, computer models that are believed to incorporate the essential phenomena have been used to develop intuition and to design something that is verified (or contradicted) by test. Quite distinct computational codes are suitable for the implosion of the primary before there is any significant fission energy, for the explosion of the primary, for the transfer of the radiation from the primary to the region of the secondary, for the implosion of the secondary, and for the explosion of the secondary.

Calculations for the early implosion weapons could be "one dimensional" because everything (temperature, pressure, density, neutron numbers, flow of radiation) depended only on the radius, while for a two-stage thermonuclear weapon two-dimensional calculations are both necessary and sufficient because there is an axis of symmetry in a normal two-stage device. With a two-stage weapon, the very complex calculations involved are only approximate, and comparison with results from actual testing is necessary to fix some parameters in the calculation. For the United States, our knowledge of secondaries comes from complex computer codes and from the experience gained from a good fraction of the 1,030 U.S. tests conducted since 1945.

Following the release of the declassified Cox Report, it was revealed that a Los Alamos National Laboratory scientist with access to the computer codes for the simulation of thermonuclear warhead explosions downloaded some of these codes from a classified computer at the lab to an unclassified computer, where they might have been accessed by unauthorized users including China. It has not been made public, and may not be known, whether any detailed design information accompanied these codes. However, even if the information that was transferred comprised the so-called "legacy codes" for U.S. weapons and if that included the programs that were actually used to simulate explosions in connection with the design of the W-88, it is not at all clear that such a transfer in 1998, for instance, would greatly enhance China's future design capability.

In a letter published in The Wall Street Journal on May 17, 1998, Harold Agnew states: "The design of the W-88...is actually quite old. The basic test was done by the Los Alamos Scientific Laboratory...when I was director, and I retired 20 years ago. It is a 'delicate' and neat package." He goes on: "No nation would ever stockpile any device based on another nation's computer codes." In my view, even the United States, with its extensive experience in designing and building advanced thermonuclear weapons, would place little confidence in a boosted primary unless it had been demonstrated in a full-yield nuclear explosion.

A great deal of computer code development has taken place totally outside the nuclear powers' weapons establishments in conjunction with, for example, astrophysics and inertial confinement fusion research, which is also conducted in non-nuclear-weapon states such as Germany and Japan. U.S. nuclear weapons designers benefit greatly from these unclassified efforts elsewhere, and feel it essential to participate in similar activities in the United States such as at the National Ignition Facility at the Lawrence Livermore National Laboratory. They can then apply these and other classified computational techniques and experimental experience to the U.S. stockpile stewardship program. Chinese designers also benefit from this foreign, unclassified, and open research and publication, which has substantially advanced the state of the computational art beyond what it was in the 1970s.

In discussing China's acquisition in the 1980s of design information on the W-70, which is described as "an enhanced radiation nuclear warhead or 'neutron bomb,'" the Cox Report asserts: "The U.S. has never deployed a neutron weapon." In fact, the United States deployed 80 SPRINT ballistic missile defense (BMD) interceptors with W-66 enhanced radiation warheads for use with the Safeguard ABM system at Grand Forks, North Dakota, in the mid-1970s. As stated by Cox Committee member Representative John Spratt (D-SC) in a May 25 statement on the panel's report, the United States also manufactured and stored in Army depots hundreds of such warheads: the W-70 on the Lance tactical missile and the W-79 on an 8-inch artillery round. Because China has no BMD system and would hardly plan intercepts within the atmosphere (to defend hardened targets), that application of the enhanced radiation weapon is irrelevant to its needs. Similarly, it is most unlikely that China will fight troops in combat with neutron bombs rather than with cheaper weapons of higher yield and greater destructive power.

Despite these reservations about the utility to China of the information regarding the W-88 and W-70, one cannot condone the fact that the Department of Energy (DOE) made such minimal investments and such little use of information technology to protect classified information. In particular, elementary cyber-security, such as implementing an access log and a transaction log of users on classified systems, would help to deter and to impede the transfer of classified material to unclassified computers. It would also help to catch the perpetrator if such actions were not deterred. It seems clear that insufficient attention was paid to preventing intentional or unintentional transmittal of classified information to unauthorized persons in ways that would not significantly interfere with efficient operation of the laboratory. Counterintelligence was clearly inadequate as well.

Interestingly, there is no claim that foreign visitors to the U.S. nuclear weapons laboratories acquired classified information. Rather, the instances cited in the Cox Report deal with U.S. citizens who are said to have intentionally provided classified information to China. But cutting off all contact with scientists from Russia, China or certain "sensitive countries," as some in Congress have proposed, is not the solution. The United States gains a great deal of information through these and reciprocal unclassified visits and interchanges, and it is impossible to further some of the highest priority goals of the United States—that of non-proliferation of nuclear weapons, for instance—without substantive visits of technical personnel back and forth. Paying more attention to the avoidance of inadvertent disclosures and to the deterrence of intentional but unauthorized transfer of information is essential. We need rapid and effective action to prevent such transfers in the future, but it should not be action that is faster than thought.

If the Cox Report results in more effective security against unauthorized transfer of classified information from DOE, the Department of Defense and the intelligence community, it will have had a salutary effect. However, the alleged acquisition by the Chinese of the particular nuclear weapon information in regard to the W-88 and W-70 would not appear to directly impair U.S. security. To build nuclear weapons on the basis of this information, China would need to make massive investments and would acquire a capability not particularly helpful to them.

Most importantly, China would need to test such weapons before deploying them—partial design information and computer codes alone are not sufficient. It would be a mistake to imagine that leakage to China of classified nuclear weapons information could be compensated by further U.S. nuclear weapon development, which would also require testing. A newly nuclear state acquiring 1950s or 1960s nuclear weapon technology and design information from either the U.S. or China (or from Russia) is probably the greater danger. But these implosion weapons would also require tests to merit any confidence. The most effective U.S. response to the threat posed by the possible leakage of sensitive nuclear weapon design information to either other nuclear-weapon states or countries with nuclear ambitions is to prohibit their nuclear testing by ratifying the Comprehensive Test Ban Treaty and exerting every effort to bring it into force as soon as possible.


Richard L. Garwin is IBM fellow emeritus at the Thomas J. Watson Research Center in Yorktown Heights, New York.

Posted: April 1, 1999