IRAQ'S PROGRAMS TO MAKE HIGHLY ENRICHED
URANIUM AND PLUTONIUM FOR NUCLEAR
WEAPONS PRIOR TO THE GULF WAR

By David Albright

The Institute for Science
and International Security

I. Introduction

Because of repeated violations of its international nonproliferation obligations, Iraq is a special case. It is the only country forbidden to possess separated plutonium and HEU, commonly called "fissile materials." Under UN Security Council Resolution 687, adopted by the Security Council on 3 April 1991, Iraq was obliged to pledge not to acquire or develop nuclear weapons, nuclear weapon-usable material or the means to make them. In accordance with this resolution, Iraq's nuclear capability was systematically destroyed. However, Iraq retains a formidable amount of 'know-how' concerning how to make nuclear weapons with its original cadre of scientists and technicians and, perhaps, small quantities of equipment and materials. A major concern is that Iraq will resurrect its program if given the chance, especially if Saddam Hussein remains in power.

Uncovering the major aspects of Iraq's pre-1991 nuclear weapon program took many years, involving many reassessments as new information was uncovered by international inspectors or declared by Iraq. Major revelations about the Iraqi nuclear weapons effort followed the defection in August 1995 of General Hussein Kamel, Saddam Hussein's son-in-law and the former head of the Ministry of Industry and Military Industrialization (MIMI), which in the late 1980s supervised Iraq's weapons of mass destruction programs. One of the most startling pieces of new information was that Iraq had launched a 'crash program' after its invasion of Kuwait in August 1990, that was intended to turn its safeguarded highly enriched uranium (HEU) into a nuclear weapon.

These revelations about Iraq's program resulted from inspection rights granted by the United Nations Security Council to the UN Special Commission (UNSCOM) and the International Atomic Energy Agency Action Team.² Under this mandate, inspectors systemically uncovered large portions of Iraq's nuclear weapon program, including an enrichment program that was expected to produce large amounts of HEU by the mid-1990s.³

The IAEA's major findings are presented in a series of reports available from the United Nations. This assessment is based on these reports, as well as on Iraqi documents and declarations, the author's own assessments when he participated in the nuclear inspection effort, and interviews with Iraqis, Germans who provided significant assistance to Iraq's nuclear efforts, and other Action Team inspectors. This report first briefly discusses Iraq's former nuclear weapon program. Iraq's fissile material production program up to the 1991 Persian Gulf War is then summarized. The remainder of the chapter focuses on the specific uranium enrichment methods and plutonium production strategies that Iraq pursued, including the crash program.

II. The Iraqi nuclear weapon program

Iraq stated that it formally decided to build nuclear weapons in 1988, although many necessary and related activities had taken place earlier. Under the 1988 plan, Iraq intended to have its first weapon by the summer of 1991. Iraq had worked on developing the capability to make fissile material over a decade prior to this date, and has explained that its decision in 1988 reflected the expectation that indigenously produced HEU would become available within a few years.

Iraq intended to put its nuclear weapons on ballistic missiles, but the conceptual nuclear weapon design of mid-1988 was too heavy for delivery by Iraqi missiles. Subsequently, the group in charge of the nuclear weapon program was ordered to reduce the weight of the design to about one tonne or less.

Questions remain about the status of Iraq's nuclear weapon program at the time of the Allied bombing campaign in 1991, when most activities were halted. Nevertheless, the Action Team inspectors have concluded that, with the accelerated effort under the crash program (described in section VII), Iraq could have finished a nuclear explosive design by the middle or end of 1991 if certain technical problems were overcome. However, it would have needed significantly longer to develop and test a warhead design of the Al Hussein missile. Such a design would have required a warhead with a diameter of 70-80 cm, much smaller than the approximately 120-cm diameter of the design nearing completion in early 1991.

Iraq was also developing a larger missile, based on the Al Abid satellite launcher, able to carry a 1 tonne warhead with a diameter of 125 cm a distance of 1200 km. According to the IAEA Action Team, however, Iraq said that this missile would not have been completed until 1993.

Iraq was also planning to build a nuclear testing site. At the time of the Allied bombing campaign, Iraq had picked candidate sites in the southwest of the country, but it had not performed a site investigation. In addition, Dr. Jaffer Dhia Jaffer, the head of the nuclear weapons program during the 1980s, told inspectors that Iraq did not plan to conduct a test before it had accumulated a few nuclear weapons. His statement implies that an underground test, if it were to occur, would have been mainly intended to demonstrate Iraq's nuclear capability. Iraq has stated that it planned to develop confidence in its weapon designs through an extensive experimental testing program that stopped short of a full-scale nuclear test.

The risk

Despite international sanctions, the Persian Gulf War and the ensuing Action Team inspections, Iraq retains many capabilities and possesses considerable nuclear know-how. This knowledge would allow it to restart its nuclear weapon program at fairly short notice. The US CIA testified to Congress in 1993 that it believes that Iraq 'probably still has more than 7000 nuclear scientists and technicians and may harbor weapons-related equipment and technology.'⁴ Although the number of key individuals would probably number only a few hundred, this cadre retains a formidable amount of information about creating a nuclear weapon program. If both sanctions and international inspections end, the CIA estimated that Iraq could produce enough fissile material for an atomic bomb in five to seven years.⁵ Access to fissile material obtained illicitly overseas could shorten this time frame considerably, to less than one year. If Iraq finished extensive preparations and manufacturing of other nuclear weapon components, it could finish a nuclear weapon within a few weeks after receiving sufficient HEU.

Immediately following the defection of General Kamel, Iraq handed to the IAEA more than 500,000 pages of previously hidden documents, 17 tonnes of high strength 'maraging steel' and a stock of carbon fiber sufficient for more than 1000 gas centrifuges. Iraq subsequently stated that many of its nuclear weapon-related teams of experts had remained together, working in what one inspector referred to as 'unreal career fields.' Although Iraq says that these teams worked on non-proscribed activities, suspicion remained that the teams were kept together to facilitate restarting the weapon program.

The extent of Iraq's work on nuclear weapons remains unknown in the period between the end of the Persian Gulf War and late 1998 when the inspectors left Iraq. Some information implies that Iraq was reconstituting its weapon program in the period right after the war. In addition, Iraq's well-documented efforts to hide its past activities from inspectors, its apparent decision to keep its technical teams together, and its long refusal to provide complete declarations of past activities have created intense suspicion of its weapon-related activities since 1991, and of its intentions once inspectors left.

To ensure greater transparency of Iraqi industrial activities, the UN Security Council decided in the early 1990s that the IAEA Action Team will conduct long-term monitoring of Iraq to assure that it complies with its obligations not to acquire or develop nuclear weapons or nuclear weapon-usable materials. The IAEA's ongoing monitoring and verification (OMV) program was based on extensive and detailed Iraqi declarations of its nuclear and nuclear-related activities, the monitoring of declared activities and the ability to search for covert activities.⁶ It ended in December 1998 when the inspectors left Iraq.

When functioning, the goal of the OMV program was to detect an Iraqi attempt to acquire nuclear materials or other essential nuclear weapon components before a nuclear weapon could be developed. While not designed to discover prohibited theoretical studies, small-scale laboratory research or prototype machines, such as single centrifuges, the OMV program represented the most stringent international nuclear inspection program ever instituted and served as a powerful deterrent to Iraq when it was in place.

It is difficult to know how Iraq would acquire fissile materials for nuclear weapons if it decided to do so. Scenarios vary from reconstituting one of its enrichment efforts to acquiring fissile material in the former USSR. Because Iraq's nuclear weapon program before the Persian Gulf War was so vast, it could probably produce fissile material and build a nuclear weapon considerably faster than many other developing states. As a result, the absence of inspectors since late 1998 raises deep concern that Iraq is making steady progress toward nuclear weapons. Perhaps it has covertly acquired sufficient material overseas for a nuclear weapon, but no evidence for such acquisition has emerged.

III. Summary of Iraqi fissile material production plans before 1991

Initial efforts to obtain plutonium

Iraq's efforts to obtain plutonium are believed to date from the 1970s.⁷ At that time Iraq concentrated on acquiring nuclear facilities overseas that would have been safeguarded, since Iraq had signed the NPT in 1968. Nevertheless, the current understanding is that these facilities were intended for secret use to produce unsafeguarded plutonium for nuclear weapons or duplicated in secret for plutonium production for such weapons.

In 1976, Iraq succeeded in buying from France a 40-MWth materials test reactor called the Tammuz-1, or Osirak, reactor. However, in June 1981, just prior to the reactor's initial operation, Israel bombed the reactor because it was convinced that it would be used to produce plutonium for nuclear weapons. Although Iraq continues to deny it intended to make unsafeguarded plutonium in this reactor, it has declared that in the second half of 1979 Iraqi scientists estimated maximum production in this reactor at 2 kg of plutonium per year. Unofficial French estimates are reported to be about double this quantity, and Israeli estimates are about four times as high.

Following the bombing of the Tammuz-1 reactor, Iraq decided to pursue the following two paths: (a) to replace the Tammuz-1 reactor or to develop a heavy water or enriched uranium reactor and associated plutonium separation capability; and (b) to develop uranium enrichment production capacity. It also decided that any unsafeguarded efforts would be pursued in utmost secrecy.

The post-1981 plutonium path

At first, Iraq concentrated on trying to replace the Tammuz-1 reactor but soon realized that it could not buy a replacement. A secret project to build a 40-WMth heavy water, natural uranium reactor was therefore launched. Iraq pursued this project until the late 1980s when it de-emphasized plutonium in favor of uranium enrichment, which at that time appeared more promising. However, Iraq continued its efforts to learn how to separate plutonium from irradiated fuel and to make heavy water, two indicators that it held fast to at least its ambitions for a nuclear reactor.

The HEU path

Even before the Israeli bombing of the Osirak reactor, Iraqi scientists had been evaluating the development of uranium enrichment technologies. However, Iraq has declared that a decision by the Iraqi leadership to pursue these options came after the June 1981 bombing. Dr. Jaffar, also the father of the enrichment program, told inspectors that the attack on the reactor was a good catalyst for the enrichment program. In fact, he stated that without the attack the program could not have been implemented.

An Iraqi evaluation finished in the second half of 1981 concluded that electromagnetic isotope separation (EMIS) was the most appropriate technology for Iraq and that gaseous diffusion was the next most appropriate option. Gas-centrifuge technology was viewed as too difficult to accomplish at that time.

Iraq has stated in its declaration that in late 1981 it formally decided to establish an enrichment technology to produce HEU. It established the Office of Studies and Development (OSD) headed by Jaffar.

EMIS was adopted as the primary option, with the goal of building production units each able to achieve 15 kg per year of 93 percent enriched uranium. Gaseous diffusion was selected as the second option, with a goal of producing LEU that could be used as a feedstock for EMIS, dramatically increasing overall HEU production.

Khidhir Hamza, a senior Iraqi nuclear official who in the early 1980s headed the theoretical group in the gaseous diffusion program, said in an interview that the Iraqi leadership tasked the gaseous diffusion program with the goal of making HEU directly. If EMIS succeeded, the gaseous diffusion program would produce LEU for the EMIS program.

EMIS

Progress on EMIS was slow. Only in 1987 did Iraq contract with a Yugoslavian firm to build its first EMIS production facility at Al Tarmiya, north of Baghdad. This contract occurred just before Iraq formally decided to start work on building a nuclear weapon. Goal quantities of weapon-grade uranium for this site were set at 15 kg per year using natural uranium feed.

In late 1987, Iraq decided to build a replica of Al Tarmiya at Ash Sharqat, about 200 km northwest of Baghdad. This facility, which was built by Iraqis only, was originally viewed as the second production site that would come into operation roughly at the same time as Al Tarmiya. In the late 1980s, this plan was modified to one in which Ash Sharqat would come into operation after Al Tarmiya was fully operational.

Iraq recognized the importance of LEU feed in raising the output of its EMIS plants. Iraqi estimates of the output using LEU feed (2.5 percent) vary between roughly 40 and 50 kg of weapon-grade uranium per year at each facility. The variation reflects different plant design and performance uncertainties. Towards this end, Iraq was seeking unsafeguarded LEU on the international market during the late 1980s, although Iraq declared its search was unsuccessful. In particular, it was trying to obtain unsafeguarded LEU produced by the Brazilian nuclear program or acquired by Brazil from China and routed to Iraq. At the time, Brazil had several nuclear sites and activities outside IAEA safeguards and had imported unsafeguarded enriched uranium from China. Iraq's relationship with China was strained, however, and Brazil denies it supplied any LEU either directly or indirectly. Inspectors found no evidence to refute this claim. In consultation with key member state analysts, the Action Team concluded that Iraq failed in its quest to obtain unsafeguarded from Brazil.

Al Tarmiya faced repeated delays and technical problems, and by the time of Iraq's invasion of Kuwait, it was at least a year behind schedule. At the time, Al Tarmiya was not expected to produce its first goal quantity of weapon-grade uranium or 15 kg, until at least 1992, assuming that the plant would function well and that a stock of low-enriched uranium would be used (see below). If natural uranium were used, the date for the production of the first goal quantity would have been 1993 or later. Civil construction at Ash Sharqat was to be completed in late 1990, after which the plant would have been kept on standby awaiting the installation of separators in the mid-1990s.

In total, both plants could have produced 25-100 kg of weapon-grade uranium if they had operated successfully. However, problems in the separators made such a possibility unlikely for many years.

Gaseous diffusion and chemical enrichment

In 1987 or 1988, the Iraqi leadership realized that the gaseous-diffusion program was not progressing well, and Iraq decided to de-emphasize this effort and instead concentrate on chemical enrichment as a source of LEU feedstock for the EMIS program. By 1990, Iraq hoped to build a chemical enrichment plant to produce about 5 tonnes per year of 4 to 8 percent enriched uranium. A time schedule for the major milestones in this approach is unavailable.

Gas-centrifuge enrichment

In August 1987, a split developed in the enrichment program, and the research group in charge of developing gaseous-diffusion technologies was transferred from the Al Tuwaitha Nuclear Research Center to a new site on the northern edge of Baghdad near Rashidiya, later named the Engineering Design Center (EDC). The reasons for this transfer vary. Among inspectors, this change primarily reflected personality conflicts between Jaffar and Mahdi Ghai al Ubeidy, the leader of this group, and this group's belief that the EMIS program was progressing too slowly and would fail to overcome certain technical problems. Hamza said that the transfer reflected Ubeidy's failure to demonstrate that his gaseous diffusion program could separate uranium on a laboratory scale. Ubeidy had bragged to Kamel that he could do so, but Kamel called his bluff. According to Hamza, after failing to enrich uranium, Ubeidy blamed Jaffar for not supporting his gaseous diffusion program sufficiently. Consistent with this version, Ubeidy and other leaders of this group stated to inspectors that Jaffar was primarily committed to the EMIS program. As a result, the President moved this group from the Iraqi Atomic Energy Commission and placed it under Kamel's supervision. At the time, Kamel had no authority over Jaffar who was based at the Iraq Atomic Energy Commission.⁸

Later, in late 1988, virtually the entire nuclear weapons program was placed under Kamel who had become head of the powerful Ministry of Industry and Military Industrialization (MIMI), which was officially established in May 1988. Jaffar was made Deputy Minister of MIMI and the heads of the major programs were made Director generals of MIMI.

When the diffusion group was transferred to Rashdiya, it was assigned to responsibility for developing gas centrifuges. It concentrated on centrifuges when diffusion was de-emphasized.

This group managed to acquire extensive overseas cooperation in designing and building gas centrifuges. After being placed under Kamel, this group gained access to extensive contacts with German businessmen and experts involved in supplying Iraq with conventional armaments and ballistic missile manufacturing capabilities. These businessmen provided Ubeidy's group with access to German gas centrifuge experts. Their assistance was so extensive from 1988 to 1990 that it was key to Iraq's steady progress in developing and manufacturing centrifuges. Despite such help, however, at the time of the Persian Gulf War, Iraq was still a few years from an operating plant able to produce goal quantities of weapon-grade uranium, declared by the centrifuge program as 1000 centrifuges producing 10 kg of weapon-grade uranium per year.

The crash program,

By the time Iraq invaded Kuwait, Iraq still lacked an indigenous source of fissile material; its enrichment plants were still far away from producing HEU. In mid-August 1990, the Iraqi leadership ordered the diversion of its stock of safeguarded HEU fuel. Iraq's initial plan was to extract the HEU from the fuel, further enrich a portion of it, and build a nuclear weapon. Hamza said that the weaponization group was also considering the direct use of HEU without further enrichment. The goal was to execute this plan and build a nuclear weapon within six months, although by the time of the Allied bombing campaign in mid-January 1991, which stopped the effort, Iraq had fallen several months behind and was unlikely to finish a nuclear explosive device until at least the following summer or the end of the year. A nuclear warhead for a ballistic missile would have taken significantly longer, according to Action Team assessments supported by member states.

Whether Iraq planned to obtain HEU for additional devices remains a matter for speculation. However, Iraq is suspected to have developed such plans.

Iraq is believed to have continued certain aspects of the crash program after the Gulf War. Information remains sketchy. Iraq denies such efforts, but is now believed to have tried to reconstitute a portion of its nuclear weapon program until it was stopped by Action Team inspections in May 1991.

IV. The Iraqi Enrichment Program

Iraq's enrichment program was both vast and clandestine. It included at least five enrichment approaches that were pursued at a cost of more than $1 billion.

As mentioned above, the first efforts concentrated on gaseous diffusion and EMIS. In the 1970s, Iraq had started a rudimentary laser enrichment program, but the leaders of the program, such as Jaffer, did not view this program very seriously. Later, both gas centrifuges and chemical enrichment were added to the development effort. At the time of the Persian Gulf War, Iraq's two most successful programs involved EMIS and gas centrifuges. Laser enrichment and gaseous diffusion had been de-emphasized, but chemical enrichment was approaching the pilot stage.

The EMIS Program

Iraq's most developed enrichment technology was electromagnetic isotope separation. The EMIS machines developed by the United States during World War II were called calutrons and produced the HEU that destroyed Hiroshima. After the war, the United States declassified much of the EMIS technology, much of which was used by Iraq. Some aspects of Iraq's separators, particularly the magnet design, were more advanced than the Manhattan Project design. Therefore, to preserve this distinction the Iraqi machines are not called calutrons. Iraq refers to them as 'Baghdadtrons.'

Iraq's 1981 evaluation of uranium enrichment technologies saw many advantages in EMIS and few disadvantages. The advantages included that (a) EMIS was well documented in the open literature; (b) the basic scientific and technical problems associated with the operation of EMIS separators were straightforward to master; (c) the computational software and main equipment were not on international export control lists, making procurement easy; (d) the design and manufacture of the main equipment for prototypes could be accomplished indigenously; (e) the feed material would be relatively easy to produce and handle; (f) final enrichment would be accomplished in two stages in machines that act independently of each other, so one or more separator outages do not affect the operation of other separators; and (g) LEU feed could be used for a substantial increase in productivity.

A special advantage was that EMIS involves large and static equipment rather than high-speed moving parts. This type of technology was better suited to the relatively inexperienced group of Iraqi engineers and scientists who had to design and manufacture the components in the early 1980s.

According to Hamza and Iraqi documents, Iraq started work on EMIS related topics in the 1970s. By the early 1980s, Iraq had already acquired know-how and experience in vacuum technology, magnet design, and ion sources.

Iraq also determined several disadvantages in choosing EMIS. An extensive research and development phase was necessary before significant production could start. The process also has high costs, is labour intensive and uses uranium inefficiently.

In an attempt to lessen the process's dependence on armies of people to produce enriched uranium, Iraq sought to increase automation in its production scale facility using on-line computers.

Development of the EMIS program

Iraq carried out a large research and development effort on all aspects of EMIS at both Al Tuwaitha and Al Tarmiya and was implementing its EMIS program in three overlapping phases.

Phase One. The first phase, which lasted from 1982 to 1987, involved basic R&D of all aspects of EMIS. During the phase, Iraq built its first separator, the R40 separator, in which ion beams traveled a circular path with a radius of 400 mm and had a maximum current of 1 milliampere. The first separation uranium occurred in January 1986. Table 11.1 describes the major separators.

Phase Two. The second phase, which started in 1983, reached an experimental stage in 1987 and continued until the Allied bombing campaign in January 1991. The original design idea, later abandoned for two stages of separators (see below), was to construct a 'unit cell' of two single-ion source separators, the first with a 1000-mm radius (R 100) and the second with a radius of 500 mm (R 50), that could produce HEU from natural uranium feed. Another R100 separator, with two ion sources, was also constructed to test operations with multiple sources. During this phase, other types of experimental separators were developed in order to test new magnet designs and multiple ion-sources for the production-scale separators slated for phase three.

A pilot plant to treat uranium solutions for liners and collectors was completed at Al Tuwaitha in the late 1980s. Al Tuwaitha also initially produced the uranium tetrachloride feedstock. Later, Iraq established an industrial-scale plant at Al Jesira, near Mosul, in 1990 to make and recycle feed material from Al Tarmiya separators.

The R100-1, R100-2, R100-3 and R50 separators were put into operation in 1987 at Al Tuwaitha in building 80 and began enriching uranium in the spring of 1988. Separators R100-1 and R100-2 had one source each and R100-3 had two sources. These sources achieved average collected beam currents of about 85-120 milliamps. Although designed to reach enrichments of about 12 percent, they typically produced less than 8 percent enriched material. Separator R50 had one source and was designed to reach enrichments of 18 percent using natural uranium feed, or about 80 percent enriched uranium, using feed of 12 percent enriched from the R100 separators.

Based on the author's analysis of monthly Iraqi progress reports for these separators, which were seized by inspectors in September 1991 in Baghdad, these machines operated through 1990, but never obtained their design capacities. Iraqi declarations in 1991 that the R100 separators achieved an availability of 35 percent were misinterpreted initially by the inspectors as the fraction of time these machines operated at design values. However, 'availability', as used by Iraq, referred to the amount of time a separator spent collecting enriched and depleted uranium during a 'run', not the fraction of a year nominal output was achieved. Each run was divided into four phases under normal conditions: preparation of operation, elementary operations, continuous operations and halting of operations. Typically, the third phase started when the collector pockets were opened and enriched material collected. The availability refers to the fraction of time in a run that a separator spends in phase three, regardless of the amount of enriched uranium produced.

Table 11.1 -- Selected EMIS separator design specifications

For example, according to the progress report for November 1990, which was one of the best months for the Al Tuwaitha machines, the three R100 separators operated throughout the month, accomplishing a total of 26 runs, of which 14 operated in phase three. The other runs encountered problems, such as short circuits in the ion source, which curtailed operation before phase three was reached. The other runs also encountered numerous breakdowns and shutdowns, but operation could be resumed without starting over. The average availability, or the fraction of time each spent in phase three, was about 35 percent. But when the separators reached phase three, they were not operating at design values. In November, the three separators produced only about 44 grams of material with an average enrichment of 6.5 percent. This amount corresponds to maintaining design current levels an average of only 19 percent of the month, ignoring the additional fact that design enrichments were not achieved.

A similar phenomenon occurred throughout 1990, during the third year of operations. Analysis of the monthly progress reports for these separators indicates that during 12 months of operation the three R100 separators produced about 360 grams of enriched uranium with an average enrichment of about 7 percent. If each separator operates at full output an average of 35 percent of the time, these three separators should product about 75 grams of 7 percent enriched uranium per month, or about 900 grams per year. Since the separators produced only about 40 percent of this amount, they were effectively operating at design currents an average of only 15 percent of the time during 1990.

As noted above, these separators were beginning to work somewhat better by late 1990, operating at design currents an average of 20 percent of the month. The reason was that improvements continued to be made in basic aspects of machine operation. For example, during November 1990, following an experiment to measure the characteristics of the ion beam, the operators discovered a better position for the collectors in the R110-3 separator that dramatically improved its separation of the uranium. However, this progress report does not suggest that the separators were on the verge of achieving design currents significantly longer than the 1990 average of 15 percent. All the separators continued to require extensive maintenance and to encounter technical malfunctions in the ion sources and other components.

In total, Iraq said that the Al Tuwaitha machines produced about 640 grams of enriched uranium with an average enrichment of 7.2 percent from the spring of 1988 through 1990. Of this quantity, 7.3 grams comprised HEU enriched to over 20 percent, with an average enrichment of 24 percent. The highest enrichment level reached was slightly over 40 percent, although only about 0.06 grams of this material was declared to have been produced in this way. All uranium enriched above 20 percent was produced in 1989 in the R50 separator from natural uranium feed by drastically increasing the machine's separation factor. The penalty was that only minute quantities of material could be produced.

According to Hamza, these higher enrichments were achieved to demonstrate that the EMIS program was making progress and as a trick to thwart growing criticism of Jaffar's program, which was way behind schedule. Hamza's explanation is consistent with detailed Iraqi documents on the production of enriched uranium in the EMIS program that show this sudden production of HEU in this one time period and no further HEU production until the program was curtailed at the time of the Persian Gulf War.

Phase Three: Al Tarmiya and Ash Sharqat. The last phase of the EMIS program aimed at reaching industrial-scale production. According to an Iraqi document seized by inspectors in 1991, Iraq signed a contract with a Yugoslavian firm in April 1987 to build a few tens of buildings at Al Tarmiya for a cost of about $110 million. Another $55 million were set aside to build the chemical recovery buildings and the electrical power supply station. At the time the contract was signed, the construction project without the chemical recovery buildings was expected to take 37 months, or until May 1990. Finishing the recovery buildings would add about four months to the project. This part of the project was delayed, and the recovery buildings were unfinished as of January 1991.

In total, the site covered about 800,000 square meters. The EMIS separators received electrical power from a sub-station about 0.5 km from Al Tarmiya. The power lines were buried to conceal the amount of electricity consumed by Al Tarmiya.

Al Tarmiya was to house 90 separators, 70 with an ion radius of 1200 mm (R120), and 20 with a radius of 600 mm (R60). The R120 separators had four ion sources with total design currents of 600 milliamps and pole pieces with diameters of 4.5 meters. The R60 machines had two ion sources with total design currents of 100-200 milliamps.

Installation of the separators at Al Tarmiya was to occur in phases. Table 11.2 shows an Iraqi plan dated late 1988 or early 1989 for the installation of the R120 separators. Iraq has stated that it did not prepare a revision of this schedule, despite falling behind.

The first eight R120 separators were installed between February 1990 and September 1990. In July 1990, work started on installing the second line of 17 separators, but by January 1991, none had been installed (see table 11.3).

Table 11.2 -- An R120 separator deployment schedule for 70 machines, declared by Iraq (but not achieved)

The R60 separators were to be installed in parallel in four phases of five separators each. The installation of the first five started in November 1990, but none were installed by January 1991. Iraq has stated that not all of the major components had been manufactured by this date.

The separators at Al Tarmiya did not perform well. Monthly progress reports for Al Tarmiya show that the R120 machines were experiencing serious operating problems.

Although the Iraqis told the Action Team in the summer of 1991 that these machines should have an availability of 55 percent, they were not even close to this level at the time of the Persian Gulf War. The major priority of the EMIS program was in fact improving the separators' availability.

According to Iraqi declarations in 1991, peak production in the eight R120 separators occurred in November 1990 when they produced a total of about 150 grams of enriched uranium with an average enrichment of about 4 percent, far below design levels for this separator. Operating at design values for an average of 55 percent of time, these eight separators (each with four sources) should produce about 115 grams of uranium-235 in about 640 grams of 18 percent enriched uranium (or equivalently, 2900 grams of 4 percent enriched uranium). Thus, in November, the separators produced at about 5 percent of design value at an enrichment level of 4 percent.⁹

In total, Iraq declared in 1991 that Al Tarmiya produced about 685 grams of enriched uranium with an average enrichment level of 4 percent. All of this material was declared as less than 10 percent enriched. The chemical processing was done both at Al Tarmiya and Al Tuwaitha. Once the new chemical facilities at Al Tarmiya would have been finished, all chemical operations would have been conducted at Al Tarmiya.

Table 11.3 -- Actual R120 separator deployment schedule at al Tarmiya

Jaffar expressed pride in his program to the inspectors, telling them in 1991 that Iraq's system was, or would soon be superior to the US calutron.¹⁰ Whether Iraq's EMIS program would have succeeded remains unclear to this day. In any case, the Al Tarmiya facility was functioning as an advanced research and development facility for separators when the Coalition forces bombed it in January 1991.

Phase three: Ash Sharqat. Iraq was building a twin facility at Ash Sharqat, 200 km north-west of Baghdad, and construction was about 80-90 percent complete when the site was bombed. Iraq had expected to finish civil construction and 'check-out' the buildings by the end of 1990. It declared after the war that no separators had been installed and no uranium had been used in the chemical buildings.

Iraq's original plan (dated to the end of 1988 or early 1989) called for installation of the first half of the separators at Al Tarmiya, followed by the installation of the first half of those at Ash Sharqat. Then the second half would be installed at Al Tarmiya before completing installation at Ash Sharqat. In the beginning of 1990, the plan was changed to full installation of separators at Tarmiya and then full installation at Ash Sharqat. In the meantime, Ash Sharqat was to be kept running at a 'low pace.'

In explaining the change in their plans, the Iraqis stated that the logistics involved in the deployment at two separate sites would have been too difficult. Housing was in short supply at Ash Sharqat, which is an isolated site far from population centers. Training personnel for two sites simultaneously was also beyond Iraq's means at that time; in fact, Al Tarmiya was already encountering a shortage of trained personnel.

The motivation for bringing two sites into near simultaneous operation is simple to understand. If one were to be identified and destroyed the other might survive undetected. Adding to the chance that Ash Sharqat would remain undetected was the fact that it was built exclusively by Iraqis, unlike Al Tarmiya. To make indigenous construction at Ash Sharqat easier, the Iraqi contractor used Al Tarmiya's plan with some simple adjustments in the buildings and layouts.

Phase three: 1987 document. Iraq's strategy for the production, installation, operation and material flow for Al Tarmiya was laid out in a September 1987 document entitled 'New procedures for setting up and operating the third phase of a separation system.'¹¹ This document evaluated various options for producing the first goal quantity of enriched uranium, set at 15 kg of weapon-grade uranium. The report contains a plan for deploying and bringing into operation all 90 separators at Al Tarmiya over a roughly 40-month period starting in August 1989 and ending about January 1993. (For the R120 separators, the schedule is similar to that in table 11.2, although the number in each phase is somewhat different.)¹² By the end of this period, the separators would have produced the first goal quantity of weapon-grade uranium, using natural uranium feed. After this date, the study estimated production at about 12.7 kg of weapon-grade uranium per year.

The study estimated that if a stock of 1.7 tonnes of 2.5 percent enriched uranium feed was used in the R120 separators and then further enriched in the R60 machines, Al Tarmiya could produce the first goal quantity after about 24 months or by August 1991. This date corresponds roughly to when Iraq hoped to finish its first nuclear weapon under its 1988 plans. The study did not provide the potential source of this LEU. Its author has denied knowing where the LEU would have originated; he said that the size of enrichment level of the LEU stock was given to him to use in this study. He said in particular that he did not know that Iraq was in possession of a stock of safeguarded LEU obtained from Italy (1.77 t of 2.6 percent enriched material), which is very close in both quantity and enrichment to the stock mentioned in his study. In any case, it appears, based on Iraqi statements, that in 1987 Iraq intended to acquire a stock of LEU, probably unsafeguarded, to speed up the production of HEU at Al Tarmiya.

In this case, the study planned to first feed the R120 separators with this LEU material. Once this material was exhausted, natural uranium would be used as feedstock.

The author of the 1987 report assumed that annual production of HEU using LEU feed would reach about 27 kg of 93 percent enriched uranium per year once all 90 separators were installed. This annual value could have been raised to about 40-50 kg per year if the R60 currents were successfully raised to 100 milliamps.

The schedules in this report were not met in practice (see below), and Iraq did not divert its stock of safeguarded LEU as this report strongly implies it planned to do. The delays at Al Tarmiya and the question of whether it could have achieved nominal outputs are analyzed below.

Annual output at Al Tarmiya

As mentioned above, the design capacity of Al Tarmiya was to be about 13 kg of weapon-grade uranium a year, using natural uranium feed. Western experts differ on the level of output that would actually have been achieved. The transition from prototype machine operation to industrial production can be difficult and extrapolations are at best fraught with uncertainty. Al Tarmiya's actual production rate is even more difficult to estimate because Iraq had been unable to produce HEU, operate any of its machines regularly because of frequent breakdowns, or achieve design currents when machines did operate.

Soon after the inspections began in 1991, IAEA inspectors estimated Al Tarmiya's output at about 12-15 kg of weapon-grade uranium a year.¹³ However, a few inspectors believed that Al Tarmiya could have produced considerably more, perhaps over 20 kg of weapon-grade uranium a year.

All of these estimates were based on initial inspections of equipment and information supplied during initial declarations. These estimates assume consistent separator operation at design currents, successful computer automation of machine operation and highly efficient changing of liners and collectors. Jaffar also led the IAEA inspectors to believe that Iraq had solved many of the problems in starting and operating separators.

The rationale for these estimates is clear. At the design levels for the ion currents and the machine availability stated in table 11.1, the four collectors of a R120 separator would receive about 2000 grams per month of enriched and depleted uranium. This uranium would contain in total about 14 grams of ²³⁵U, if natural uranium were used as feed. For seventy R120 machines, the yearly collection would total almost 12 kg of ²³⁵U. This ²³⁵U would be contained in about 65 kg of 18 percent enriched material, assuming complete separation into light and heavy fractions of uranium. If we ignore losses in the tails and assume complete recovery of this material and further enrichment of all of this material to weapon-grade in R60 machines, one arrives at about 13 kg of weapon-grade uranium per year.

Existing estimates above 20 kg per year assume that the R120 separators would have achieved availability factors approaching 100 percent. Inspectors said that Iraqis also told them in the summer of 1991 that they planned to change the sources, collectors and vacuum liners considerably faster than originally planned. More recent information shows that the program's major effort was to achieve the design values listed in table 11.1, including an availability of only 55 percent. However, Jaffar stated in 1996 to this author that if these design levels were achieved, he planned to seek improvements in ion sources, higher currents and collection efficiency rather than availability.

An important part of the availability factor is the reliability of the machines themselves, an aspect of separator operation that was causing serious problems. Based on the frequent breakdowns of the separators at Al Tarmiya throughout 1990, Iraq could expect difficulty in achieving design values at Al Tarmiya, let alone exceeding them.

Unless Iraq improved the performance of its separators, Al Tarmiya might have produced as little as a few kilograms of weapon-grade uranium per year. Achieving design levels would have required Iraq to overcome the problems plaguing its ion source, collectors, and other components and to assemble sufficient trained personnel. Iraq is believed to have been capable of eventually overcoming these problems, but doing so may have caused serious delays in achieving full operation at Al Tarmiya.

Minimum time to first goal quantity

The problems in the EMIS program would have most directly affected the time that Iraq needed to acquire its first goal quantity of HEU. One way to determine the minimum time to produce a goal quantity is to assume that design values could have been achieved and to base the estimate on the expected deployment dates of the separators in table 11.2, after accounting for known delays.

By the time of the Allied bombing campaign, the deployment plans were at least 6-12 months behind schedule. Thus, under this estimate, all 90 separators would have been deployed no sooner than between June/July 1993 and January 1994. Using natural uranium feedstock, this date corresponds to the production of the first goal quantity of HEU. If LEU feed were used, this date would be in the first half of 1992.

The deployment schedule would probably have been further delayed because the separators encountered so many unexpected start-up problems. Iraqis have stated that phase one of the separator installation was actually serving as an R&D stage, and not a production phase as originally planned.

An analysis of Iraqi documents and interviews of Iraqis raise serious questions about whether Iraq could have met the schedule assumed above. In particular, the monthly progress reports from 1989 and 1990 on machine operation at Al Tarmiya and Al Tuwaitha list persistent problems in operating the separators. Throughout 1990 at Al Tarmiya, currents at the collectors remained significantly below design levels of 150 milliamps, often not exceeding 100 milliamps at a collector. Total current for all four sources in a separator rarely exceeded 400 milliamps, and was often considerably less. The most serious problems included unreliable ion sources and inadequately manufactured components for the vacuum liners and chambers. Until these problems were overcome, enriched uranium production would remain low, no matter how many separators were installed.

Despite years of development, the program was facing serious problems. In total, the EMIS program had produced only about 1300 grams of low-enriched uranium and virtually no HEU.

However, EMIS technology is relatively straightforward, and Iraq is judged to have been able to overcome these problems with sufficient time and expense. Thus, there is no reason why Al Tarmiya and later Ash Sharqat could not have achieved near full operation, if Iraq had not invaded Kuwait. But it is highly unlikely that its first goal quantity could have been produced before 1992 in the case of LEU feed or before mid-1993 using natural uranium feed. Because of the problems in the program, these dates would have probably been significantly later, by at least a year or two.

Iraq could reconstitute its EMIS program. If it acquired secretly or diverted its existing stock of LEU, it would need a facility with about one-third the number of separators planned for Al Tarmiya to make enough HEU for about one nuclear weapon per year. Such a facility, if planned carefully, would be difficult to detect without an intrusive inspection regime in Iraq.

The gaseous-diffusion program

Iraq decided in late 1981 to launch an effort to build a gaseous-diffusion plant. It saw several advantages; for example, the technology was commercially proven. However, Iraqi scientists lacked the basic 'know-how.' Furthermore, gaseous diffusion involves uranium hexafluoride, which is difficult to produce and handle; and it requires a large number of items which were internationally controlled and could not be built indigenously. Despite these drawbacks, Iraq was to construct a cascade of several hundred stages to produce 4-5 tonnes per year of 3-4 percent enriched uranium as feedstock for the EMIS program. Hamza said that the program was ordered to develop a cascade able to make HEU directly.

Gaseous-diffusion work started at Al Tuwaitha in 1981 or 1982, focusing initially on basic scientific and technical questions. A key part of this initial effort was the investigation of a suitable porous barrier tube, which is the basic enrichment element and considered highly classified by countries that have mastered gaseous diffusion. In 1985, as work on barrier progressed, Iraq increased its work on other components of the separation stage, such as compressors, heat exchangers and control systems.

The first milestone of the gaseous-diffusion program was to operate a single working stage model with uranium hexafluoride gas and all auxiliary systems and components. Iraq commissioned one separation stage that used a surrogate material for uranium hexafluoride.

In mid-1987, the team, named Group 1, moved to an industrial site near Rashdiya, north of Baghdad, where it created a large enrichment research and development center. After its transfer, Group 1 also received a mandate to develop gas centrifuges (see below).

The gaseous-diffusion program was also facing difficulties. Iraq has declared that it cut back its work on gaseous diffusion in 1987. According to Iraq's declaration, the plan to introduce uranium hexafluoride into the experimental stage was never implemented, and the project involving the single stage was cancelled in 1987. According to Hamza, the plan to use uranium hexafluoride in an experimental stage failed, because the barrier clogged. As discussed above, this episode led to the transfer of this group to Kamel's organization. Iraq subsequently focused on the development of small-size cascades utilizing small compressors that could be manufactured domestically or bought overseas in sufficient quantities.

Although Iraq attempted to make many types of barrier tube, it succeeded only with one type made of an anodized aluminum. In 1988 Iraq successfully separated uranium hexafluoride in a laboratory set-up with this type of barrier. However, an anodized aluminum barrier is usually considered too fragile for use in an industrial enrichment plant. Nevertheless, Iraq made several hundred such tubes and conducted several successful experiments that demonstrated the corrosion resistance of the tube to uranium hexafluoride. Although Iraq achieved measurable uranium isotopic separation, the amounts of enriched uranium involved in such experiments would have been very small.

Iraq's work on compressors focused mainly on the foreign procurement of suitable machines for experimental facilities. Iraq has stated that it did not carry out its plans to develop the capability to produce compressors indigenously from reverse engineering procured items or to secure large numbers of them from overseas suppliers for a cascade.

Little progress was made in the gaseous-diffusion effort. Iraq has declared that the scale of the technological and manufacturing requirements for this effort led to its cancellation in 1989. It has reported that it dismantled the last of the equipment in 1991, although Hamza said that the program continued and made progress in the 1990s. Iraq could try to build a gaseous diffusion facility in secret, and this facility would be difficult to detect without inspectors on the ground.

The gas-centrifuge program

Iraq has said that it launched its gas-centrifuge program in August 1987 with the creation of the Rashdiya center. The prior experience with gaseous-diffusion technology was useful, particularly work relevant to the corrosion properties of uranium hexafluoride gas, which is also the material used in gas centrifuges. A gas-centrifuge effort had not been envisioned under the 1981 plan. The Iraqi planners viewed gas centrifuges as even more difficult than gaseous diffusion because of the special materials involved and the high rotational speed of the key parts.

Iraq was able to make significant progress on this technology only after it obtained extensive foreign assistance from German gas-centrifuge experts. Iraqis have stated that in the spring of 1988, Iraqi centrifuge experts approached a German firm, H+H Metalform GmbH, for help in making gas centrifuges. This firm was already assisting Iraq's conventional armaments and ballistic-missile manufacturing programs and was a leader in producing maraging-steel tubes for missiles and rockets. German gas centrifuge experts hired by H+H recommended using maraging steel rotors for gas centrifuges, and H+H was providing the equipment and expertise to make high precision tubes for centrifuges.

H+H quickly became Iraq's most important contact for assistance in obtaining centrifuge expertise and manufacturing equipment from abroad. In essence, it acted as a 'funnel' for a wide variety of important components, materials, 'know-how', design information and manufacturing equipment for Iraq's gas centrifuge program.

Iraq's declared goal for the gas-centrifuge project was a 1000-machine centrifuge plant able to produce 10kg of weapon-grade uranium per year. The initial plan was to build first a workable centrifuge, then assemble experimental cascades, and finally construct a facility holding 1000 operating centrifuges. The timetable for this effort at the time of the Persian Gulf War is not certain. However, Iraq has stated that it planned to commission the 1000-machine facility in 1994.

Centrifuge development

The Iraqi effort focused on overcoming a weak technological base and creating the technical infrastructure to research, develop and manufacture gas centrifuges. Iraq worked on two types of centrifuge. The first type, the 'Beams', or 'oil' centrifuge, was dropped in 1989 as a result of its inefficiency and of steady progress on the 'Zippe-type' or 'magnetic' centrifuge.

However, Iraq appears to have planned to build both types of centrifuge at the start of its effort and also appears to have realized that the Zippe-type was the more effective and practical one to develop. Nevertheless, the Beams-type centrifuge was more accessible to Iraq, and it provided valuable experience in building and operating centrifuges. One Iraqi stated that without the experience with the oil centrifuge, the team may not have been able to properly understand the Zippe-type centrifuges, or the highly specialized design assistance provided by German centrifuge experts in 1988, facilitated by H+H Metalform.

Beams or 'oil' centrifuge. The centrifuge program started working on an unclassified oil centrifuge that was developed by Jesse Beams in the United States in 1930s and 1940s. According to Iraqis, valuable mechanical design information for this centrifuge was available in the open literature. By the end of 1987, the team had built its first machine, called the GS-1 (gaseous separator-1). During the next two years, Iraq made many modifications to this original subcritical design as it struggled to overcome problems of excess vibration, inadequate seals and excessive power consumption.

Most of this work was carried out in building 22 at Rashdiya, which was finished in the first quarter of 1988. Inside this building, the team built two 'pits.' Each pit was about 6.5 meters deep, extending 2.5 meters below ground level and about 4 meters above the floor. Forty-centimeter thick concrete walls, to which the test centrifuges were attached, surrounded the above ground portions of the pit. The Iraqis had calculated that this wall thickness was required in case a heavy centrifuge jacket (60-80 cm long) was to break away from its fixtures while spinning at high speed and crash into a wall. One pit was designed to conduct mechanical tests, and the other one was designed to use process gas in the centrifuges.

Because of known weaknesses in Iraqi manufacturing capabilities, the centrifuge group decided to supervise the production of high-precision components at indigenous manufacturing installations and to check the quality of the components jointly with the manufacturer. Nevertheless, the reject rate for complements remained high because of low machine quality, inexperienced staff and the shortage of suitable cutting tools and fixtures.

Raw materials, such as duralumin for the centrifuge cylinders, vacuum oil and stainless steel, were imported in sufficient quantities for research and development purposes. Motors and frequency converters were also imported. European firms were Iraq's main suppliers.

The centrifuge program was initially plagued by poor diagnostic instrumentation and a lack of theoretical understanding of rapidly rotating machinery. Towards the end of this effort in 1989, Iraq had overcome many of these problems, mainly through importing better bearings and more sophisticated computer codes and balancing machines. For example, until the program obtained a sophisticated balancing machine from the German company Ruetlinger, it did not realize that the centrifuge's inability to spin at design values was the result of improperly machined rotating components which were unbalanced. After correcting these problems, the centrifuge achieved machine speeds in excess of 50,000 revolutions per minute (rpm).

Iraq had declared that it designed, but did not build, 10- and 50-machine cascades. It also said that in 1988 it envisioned building a 4000-machine cascade of oil centrifuges at a site south of Al Taji. This site was also originally envisioned to be the location of an oil-centrifuge manufacturing facility. With a goal quantity of 10 kg per year of weapon-grade uranium, each machine would have needed to achieve an output of roughly 0.5 separative work units (SWU) per year.

The process and mechanical difficulties inherent to this type of machine were never completely overcome. Nevertheless, by mid-1989 Iraq was conducting separation tests using a gas composed of freon and carbon dioxide. This mixture, unlike uranium hexafluoride, does not react chemically with the oil in this type of centrifuge. Although the mixing of the process gas and the oil remained a problem, Iraq has stated that this problem was manageable. However, it said that uranium hexafluoride gas was not used mainly because the operating speed of these centrifuges while filled with process gas was too slow (about 21,000 to 25,000 rpm) for a noticeable separation of uranium isotopes.

Parallel to the Beams centrifuge effort, the research group was making steady progress on the magnetic centrifuge, which was more efficient and practical than the oil centrifuge. By mid-1989, Iraq decided to cancel the oil centrifuge project, but allowed it to continue until late 1989 in order to provide training on centrifuge operation.

Subcritical Zippe-type or 'magnetic' centrifuge. Work on this machine started formally in the second quarter of 1988. A major break occurred in August 1988, when the centrifuge expert Bruno Stemmler, accompanied by Walter Busse and H+H personnel, traveled to Baghdad and gave the Iraqis two secret assembly drawings of a subcritical centrifuge and a supercritical centrifuge with two coupled sections. These machines had been developed in Germany, particularly at MAN Technology AG in Munich in the late 1960s and early 1970s. Both Stemmler and Busse had started working at MAN during this period. Stemmler remained an employee until about 1990; Busse retired a few years earlier.

During this meeting and subsequent meetings in Baghdad and Germany, Stemmler provided many detailed drawings of centrifuge components from several different models. He also gave the Iraqis about 70 classified technical reports, actual components, production data and procurement information for early-generation Zippe-type centrifuges. These technical reports contain many details for the production and operation of early-generation Urenco centrifuges.

Until Iraq invaded Kuwait, Iraq continued to receive a steady stream of outside assistance, mostly facilitated by H+H. Starting in the spring of 1989, a third German expert, Karl Heinz Schaab, started providing important centrifuge assistance. According to the Iraqis, he provided components, such as carbon-fibers tubes and samples of bellows and baffles, carbon-fiber rotor manufacturing assistance, about 90 secret technical reports, and components and equipment important to the mechanical testing of subcritical machines. The Iraqis valued Schaab, because, unlike the other foreign experts, he was 'very good with his hands.' He also provided or assembled centrifuge test equipment.

Iraq also participated in specialized training courses in Europe. The Iraqis have stated that the goal of these courses was to acquire specific information relevant to centrifuges, although the training companies did not know the true Iraqi purpose for attending.

The most important course was held at INTERATOM in Germany in 1989. INTERATOM made centrifuge cascades for Urenco. The training course involved about 20 Iraqis, each of whom was trained for periods between 4 days and 10 weeks in the second half of 1989. During these courses, the Iraqis learned about 'dual-use' vacuum technology, advanced welding, piping design and the materials science of maraging steel. In secret, without the knowledge of INTERATOM, Iraqis stated that they entered classified areas of the company where gas-centrifuge cascade piping was located and copied exact dimensions and arrangements of pipes, valves and tri-flanges. INTERATOM cancelled a second series of training courses scheduled for 1990 because of growing suspicions in the West German Government about Iraqi intentions.

After unsuccessful attempts to manufacture oil-centrifuge components, Iraq realized it would have difficulty manufacturing components for magnetic centrifuges. It therefore decided in late 1988 to place orders for the manufacture of many key centrifuge components with European firms in sufficient quantity for 50 machines, which were needed for the Iraqi prototype development program.

Some of the components were procured directly from European suppliers; others were provided as demonstration tests during negotiations for the purchase of machine tools in Europe. Although tighter export controls prevented Iraq from obtaining components for all 50 machines, it acquired enough of them to run its research program.

With all this inadvertent and deliberate assistance, Iraq created several reliable designs of subcritical centrifuges during 1989 and 1990 that were based on both maraging-steel and carbon-fiber rotors. In this effort, the available designs were critical, but not sufficient. The designs did not specify all the parts, requiring Iraq to work out the detailed specification of components without foreign assistance. In addition, Iraq had to create a testing program and a small-scale manufacturing program for many centrifuge components.

In 1990, the centrifuge team may have been starting to opt for carbon-fiber rotor designs rather than maraging-steel ones. Schaab said that he had already supplied about 40 finished carbon-fiber rotors and was helping Iraq to develop an indigenous capability to make them. Meanwhile, Iraq had not yet mastered the manufacture of maraging-steel rotors on an H+H-supplied 'flow-forming' machine, nor had they received an adequate number of rotors from H+H.

By March 1990, Iraq had received from overseas and domestic sources most of the components for its first prototype centrifuge. Within about a month, the first mechanical test stand was operating in building 10 at Rashdiya using a carbon-fiber rotor provided by Schaab. The Iraqis have said that Schaab personally assisted them in assembling this first prototype.

After a few more months the test stands' rotor was spinning at up to 60 000 rpm, or at a wall speed of about 450 meters per second (m/s). In July, Iraq also started a test stand utilizing uranium hexafluoride gas.

These two test stands continued to operate until the end of 1990, when they were disassembled and hidden. It is known that the production of enriched uranium was not a goal at this point, because the slightly enriched product from the test stand was mixed with the tails and reused, rather than kept separate.

Iraq has said that it did not build additional test stands using uranium hexafluoride. It has stated that it planned to operate two optimized centrifuges in parallel or in series, but it had not optimized the centrifuges as of the end of 1990 and thus had not yet conducted this particular experiment.

The capacity of Iraq's test machine during one test run at the end of 1990 reached 1.9 SWU per year. Results in previous months were less successful. With additional machine optimization, inspectors believed that this design could have achieved a separative output as high as 2.7 SWU per year. Because the rotor was made of carbon-fiber, it could have spun faster that 450 m/s, which is near the limit for maraging-steel rotors. Thus in theory, it could have reached a higher separative work output. Iraq, however, has stated that it expected an output of only 2 SWU per year, even when using carbon-fiber motors. One centrifuge expert said that he expected the rotor to spin at only 400 m/s, which is consistent with the declared separative output.

Centrifuge manufacturing

In late 1988, Iraq decided to build a dedicated centrifuge-manufacturing site and obtain the necessary manufacturing equipment and raw materials abroad. This decision was controversial because at the time Iraq lacked a proven centrifuge prototype. As a result, Iraq had to order raw materials and equipment with the knowledge that the design would change and thus render some of the orders obsolete. But this approach permitted the program to obtain many key items before the international community understood Iraq's true intentions and tightened export control policies.

Although the first site considered for the plant was near Al Taji, the team selected Al Furat, about 30 km south of Baghdad. This site had existing buildings, workshops and utilities and thus could start operations more quickly than the one at Al Taji. The plan was to build 'clean environment' workshops where the temperature and cleanliness would be carefully controlled, a requirement for manufacturing and assembling magnetic centrifuges.

The civil works at the site were nearing completion at the time of the war. More importantly, as of August 1990, Iraq was on the verge of acquiring enough manufacturing machine tools and associated textures to complete its capability to make centrifuges. Only the UN Security Council embargo imposed after Iraq invaded Kuwait kept it from receiving the rest of manufacturing equipment.

Iraq declared in 1991 that Al Furat would be capable of making 200 centrifuges a year, although IAEA inspectors at the time believed that annual production would reach 600 machines a year and under certain circumstances production could reach 2000 centrifuges a year.

In 1996, Iraq changed its story and said that it had envisioned Al Furat's production starting at 1000 centrifuges a year in 1992, reaching 4000 machines a year after a future expansion in 1994 and 1995. The Iraqis have said that they expected a very high failure rate, so the actual number of centrifuges that would have been produced remains an open question.

Because the Iraqi construction industry was unfamiliar with making clean room facilities, Iraq contracted with German and British firms to make two clean-room buildings. The contract to construct Building B01, which was to house dimensional quality control, mechanical testing of single centrifuges, final assembling of centrifuges and a hall for a 100-machine cascade, went to INTERATOM and was signed in October 1989. The original time schedule called for the building to be completed in October 1990. Iraq decided to build the civil works of building B02 itself, which was to produce maraging-steel rotors and lower bearings, but it awarded a contract to construct its mechanical and electrical utilities to ITSC, a British company. This contract was signed in late September 1989, and the building was expected to be finished in December 1990.

Meanwhile, by late 1989 the centrifuge program was advancing. Rashdiya was making components that had not been ordered abroad, such as the lower bearings and the motor. The program was starting to receive key components, testing equipment and manufacturing items for the prototype centrifuge from abroad. Because buildings B01 and B02 were many months from completion, the program leaders decided to modify building 10 at Rashkiya for centrifuge assembling and testing. This building held the centrifuge test stands mentioned above.

In early 1990, building B03 at Al Furat, which was closer to completion than either B01 or B02, was outfitted with machines to make molecular pumps and the outer casings. In addition, B03 housed research and development efforts for maraging-steel rotor production.

Iraq apparently planned to finished Al Furat, although its expected completion date remains unclear. The centrifuge program may have also turned over the operation of Al Furat to other Iraqi industrial organization while maintaining staff participation for supervision. In addition, Rashdiya would have probably continued to manufacture certain components.

Iraq did not reach the point at which it could produce a sizeable number of centrifuges. Moreover, the embargo significantly impaired its ability to make centrifuges. Nevertheless, Iraq had made great progress in creating an indigenous centrifuge manufacturing capability in just a few short years.

When completed, Iraq's manufacturing facilities may have generated far more reject parts than adequate ones. Iraqi centrifuge experts in fact expected a significantly high reject rate. They planned to compensate for this weakness by trying to make thousands of centrifuges a year. If Iraq could have continued to procure sufficient raw materials and other necessary items, it may have succeeded in producing adequate centrifuges at a rate of many hundreds per year by the early 1990s.

Carbon-fiber manufacturing

Because Iraq was encountering problems in making maraging-steel rotors, it decided in 1989 to develop the capability to make carbon-fiber rotors. Schaab had supplied about 40 carbon-fiber tubes. He made these tubes on a manual winding machine at his company ROSCH in Kaufbeuren, German, southwest of Munich.

In late 1989, Iraq said it signed a contract with the Swiss firm ALWO for an automated carbon-fiber winding machine using designs and some equipment obtained from Schaab. The final construction of this machine was delayed by months, because of problems in obtaining a German export license for a computer-controlled subcomponent of the winding machine from Siemens. According to Iraqis, the embargo following Iraq's invasion of Kuwait disrupted the planned shipment of this machine. In September 1990, Iraq arranged to have this machine sent to Iraq through an intermediary company in Singapore (see section VII on the crash program). However, the machine made it only to Jordan, where it was embargoed. It was subsequently located and inspected by the Action Team.

The 1989 contract with ALWO and Schaab included a provision for a two-week training course and the production of 50 finished rotor tubes. The training was originally scheduled to be held at ROSCH in Germany or at an unspecified place in Austria. The expected date of the training was the autumn of 1990. After the imposition of the embargo, Iraq stated that it planned to hold the training session in either Austria or Latin America. The Iraqis and Schaab stated that the training sessions did not occur.

Based on the available information Iraq knows how to make carbon-fiber rotors, at least in theory. Iraq possessed a plentiful supply of the basic building blocks for rotors, namely carbon fiber and resin, until after General Kamel's defection when at least the carbon fiber was handed over to the inspectors. If it somehow obtained a winding machine, it would have had years to master the production of the rotor.

Cascades for subcritical Zippe-type centrifuges

In 1989, Iraq originally intended to house a 1000-machine cascade (or two 500-machine cascades) along with a uranium hexafluoride production plant at a site south of Al Taji. After conducting initial studies, Iraq maintains that it did not do any construction work at this site, although it apparently intended to do so.

Building B01 at Al Furat was selected to house an experimental cascade of 36 machines able to produce 3 percent enriched uranium. This plan was modified to 120 centrifuges in a cascade, and the cascade hall was enlarged accordingly. This larger cascade would have produced bout 1kg per year of weapon-grade uranium.

However, because construction at Al Furat was delayed, the centrifuge team decided to build a new building at Rashdiya to house the experimental 120-machine cascade. This facility, called building 21, was unfinished at the time of the Allied bombing campaign. An important benefit of this change would have been the elimination of the need to move key personnel between Rashdiya and Al Furat, allowing personnel to be concentrated at Rashdiya. For a similar reason, a decision was made to locate uranium hexafluoride production at Rashdiya and not at the Al Taji site.

The Rashdiya cascade hall was designed to be larger than the similar one at Al Furat, leading to suspicions that Rashdiya would contain a significantly larger cascade. Iraq denied this allegation.

Supercritical design

Iraq has declared that it did not build any supercritical designs, which are characterized by having longer rotors than subcritical machines. European supercritical centrifuges have rotors made of two or more roughly 50-cm long rotor tubes connected together by a flexible maraging-steel joint (called a 'bellows') that acts like a spring. The bellows can be very difficult to make. The supercritical centrifuge itself is complicated to assemble and balance.

What Iraq obtained. As mentioned above, Stemmler had provided Iraq with a general assembly drawing of a two-rotor, maraging-steel machine. Iraq, however, says it did not intend to build one. A major surprise following Kamel's defection was Iraq's declaration that in August 1989 Iraq obtained from Schaab a MAN design for a 3-meter long supercritical centrifuge and some detailed drawings of some of the key components. Such a machine would have had a separative capacity of about 20-30 SWU per year, depending on the rotor's speed of rotation, and represents a sophisticated, mid-1980s Urenco design. Iraq did not obtain a complete set of the 3-meter machine designs; some components designs were missing.

The Iraqis said that Schaab also supplied three samples of bellows, a number of carbon-fiber baffles and many classified technical reports about mid-1980s centrifuges. Combined with its acquired drawings and knowledge of materials, these technical reports would greatly expedite the production and operation of modern subcritical and supercritical centrifuges.

Iraqi declarations and resulting questions. Iraq has consistently downplayed the level of its work on supercritical centrifuges. It has declared that it concentrated on subcritical design and its work on the supercritical 3-meter centrifuge design was progressing more slowly. It said it barely worked on the two-rotor design at all.

Questions remain, however, about the scope of Iraq's work on supercritical centrifuges. For example, why would Iraq ignore a G2 design, which is considerably easier to construct than a 3-meter design? Other questions result from apparently contradictory information supplied by Iraq.

Iraq has admitted compiling a 3-meter design and doing some simple theoretical analyses of it. Iraq also enlarged the doors and raised the roof height of the cascade hall in building 21 at Rashdiya to permit the installation of the 3-meter machines. It also said that it enlarged the size of the doors to a building at Al Furat where machine testing would have occurred.

Questions also remain about Iraq's efforts to procure bellows manufacturing equipment. Iraq has stated that in October 1989 Schaab offered to arrange a contract for the supply of bellows-production equipment and know-how. Schaab told the Iraqis that he could also provide a 'friend' who could assist in providing specific bellows know-how, but Iraq says it did not take Schaab up on this offer. The Iraqis said they judged that they could make the bellows themselves at Rashdiya. Failing in that, they believed Schaab's offer would still be on the table.

Another contradictory piece of information is that Iraq ordered bellows manufacturing equipment from a French firm that would have been suitable for making maraging-steel bellows, although the order was for producing stainless steel bellows. The embargo prevented the delivery of this equipment.

Annual output of the centrifuge program

Iraq has stated that its goal was the operation of 1000-centrifuge cascade in 1994 that would produce 10 kg per year of weapon-grade uranium. Few, however, believe that this number of centrifuges or this annual quantity was the final goal of Iraq's centrifuge program.

Iraqi centrifuge experts appear genuine in their belief that building and operating 1000 machines would represent a tremendous accomplishment. They have stated that they did not think they could meet this goal by 1994, despite the official declaration. The head of centrifuge program told this author that this goal was unlikely to be reached before 1997 or 1998.

The Iraqi centrifuge experts have not provided any documents that could substantiate their pessimistic view of their own progress. In addition, other information contradicts their claims. For example, their view is belied by the sheer amount of equipment and know-how that was being supplied by foreign experts and companies just before the war and the ability of the Iraqis to establish new procurement routes through the Far East. In addition, if Iraq succeeded in assembling its centrifuge production facilities on the schedule it has recently declared, it expected to produce 1000 centrifuges per year by the end of 1992. It planned to expand annual centrifuge production several-fold by the end of 1995.

As in the case of EMIS, the centrifuge program may have faced serious problems as it moved from laboratory testing of gas-centrifuge equipment to industrial production of significant quantities of enriched uranium. Iraq might have encountered difficulties in procuring all the items it would have needed and in achieving the high-quality manufacturing standards necessary for the production of reliable centrifuges. However, inspectors believe the Iraqi centrifuge team was close to the point at which it could build and operate a centrifuge with confidence, although the total economic cost might have been high and some delays might have continued to occur.

A maximal estimate of the Iraqi centrifuge capacity can be derived by using a modification of Iraq's planned manufacturing schedule and assuming adequate procurement of raw materials. This estimate also conservatively assumes that 50-75 percent of the centrifuges produced will be rejected. The output of each centrifuge is taken as 2.7 SWU per year, which means that Iraq would have optimized its subcritical design prior to manufacturing. Table 11.4 shows that, by the year 2000, Iraq could have an estimated annual capacity of roughly 14 000 - 28 000 SWU per year. At this rate, and assuming a tails assay of 0.5 percent, Iraq could make about 87-175 kg of WGU per year. At 15 kg per nuclear weapon on average, this is enough HEU for 5-11 weapons a year.

Table 11.4 -- Maximal estimated centrifuge production^a

Another upper-bound estimate is based on Iraq's procurement efforts. Iraq has admitted, and the IAEA has verified, that it had obtained enough specialized metals and components to build thousands of machines. In some cases, Iraq acquired enough items for up to 10 000 machines. If it is assumed that Iraq aimed to produce a total of 10 000 machines, and again assuming a 50-75 percent reject rate, Iraq might have succeeded by 2000 in building 2500-5000 machines although the upper bound would have likely required the procurement of additional materials. Assuming that each machine would have had a capacity of 2.7 SWU per year, the total capacity of all these machines would have been 6750- 13 500 SWU per year. Assuming that the tails assay was again high (about 0.5 percent), Iraq could have produced about 40-80 kg of weapon-grade uranium per year. Assuming 15 kg per weapon, Iraq could build two to five nuclear weapons per year with this amount of HEU.

The available information tends to discount Iraq's ability to make a supercritical centrifuge, at least for several years. As a result, the effect of supercritical machines on these estimates is ignored. However, with time, Iraq could have mastered the production and operation of supercritical machines, if it decided to do so.

If the centrifuge program had succeeded, even marginally, Iraq would have had a significant supply of HEU. The supply could have been large enough to reduce dramatically the need for the EMIS program, while providing for a large number of nuclear weapons (see table 11.5). Because of the flexibility of centrifuges compared to gaseous diffusion, there would have been little value in using the centrifuges to produce LEU for further enrichment to weapon-grade in the EMIS separators. Thus, the centrifuge program appears to have been Iraq's most successful effort aimed at making fissile material.

In any case, the war and the ensuing inspections crippled Iraq's centrifuge manufacturing program and ended the program before HEU was produced. Inspectors believed in 1998 before they left that Iraq did not have the capability to make a significant number of centrifuges. However, Iraq has accumulated a large body of information and considerable expertise about Urenco-type centrifuges. If given the opportunity, Iraq could reconstitute its centrifuge program considerably faster than other states could develop them.

Chemical enrichment

In the spring of 1988, the gaseous-diffusion program was encountering problems and the gas centrifuge effort was in its infancy. In order to find another way to produce LEU for the EMIS separators, Iraq decided to initiate the development of a chemical enrichment facility.

The Iraqi program essentially copied the French CHEMEX solvent-extraction process and the Japanese ASAHI ion-exchange technique. These chemical enrichment approaches depend on solvent extraction technologies, that is, reprocessing and ion-exchange technologies. According to the Iraqis, both of these technologies were relatively accessible to Iraq.

It is unclear when Iraq first started research into chemical enrichment or how much design information Iraq obtained from abroad. Iraq has stated that in 1980 France proposed to an Iraqi team that it participate in the development of its CHEMEX enrichment process. According to Hamza, France wanted in exchange that Iraq would accept "caramel" fuel instead of 93 percent enriched uranium fuel for the Osirak reactor, which is the level of enriched uranium France contracted to sell Iraq. Iraq rejected this offer, but it is unknown whether at that time Iraq somehow obtained design information for the CHEMEX process.

At the time of the Persian Gulf War, Iraqi scientists had accomplished significant laboratory work in both techniques. They had finished designs for pilot plants to produce small amounts of enriched uranium, but they had not implemented construction at Al Tuwaitha.

Iraq procured key equipment, such as distillation units, mixer-settler batteries, pulsed columns and pumps for its laboratory-scale work from France, Sweden and Germany. Much of this equipment was destroyed in the bombing of Al Tuwaitha, but some was salvaged.

Iraqi success in chemical enrichment would have probably depended on additional, larger-scale foreign procurement. At the time of the Persian Gulf War, however, little of this procurement had occurred, although Iraq has said that it had placed an order with a French company for glass columns, mixer-settlers and other items for its solvent extraction pilot plant. The embargo following Iraq's invasion of Kuwait ended this transaction.

Production-scale plants were designed for the solvent extraction and the ion-exchange techniques and were essentially larger versions of the pilot plants. The design studies called for the production of 4000 kg of 3 percent enriched uranium (120 kg ²³⁵U) per year. In the process of choosing which approach to pursue further, Iraq has said that it conducted paper studies of a hybrid that would have been able to enrich uranium to 8 percent. In both cases, Iraq has stated that no construction occurred.

There is insufficient information to estimate when or if Iraq could have completed a chemical enrichment plant for uranium isotope separation. Indications are that the program was encountering serious difficulties as it moved out of the laboratory towards the pilot-stage. Similarly, it is difficult to determine if Iraq would reconstitute and make progress on this technology in the absence of inspectors.

The laser enrichment program

In May 1994, the IAEA learned from member states that Iraq had pursued uranium enrichment through laser isotope separation (LIS) at the Al Tuwaitha site. According to member states, Iraq had studied both molecular (MLIS) and atomic vapor (AVLIS) technologies.

In 1991, Iraqi officials denied that Iraq had undertaken any laser enrichment activities. During the 26^th IAEA inspection in the summer of 1994, after continuing to deny the existence of the program for four days, Iraqi officials finally admitted that the existing Laser Section 6240 had 'received an objective [in 1981] from the Atomic Energy Commission to work in Laser Isotope Separation.'

Following Hussein Kamel's defection, Iraq provided more information about its laser separation program. However, the basic story has remained unchanged. In 1987, with little progress to show, the program was downgraded to a 'watching brief' and a number of key personnel were transferred to other programs, particularly EMIS. The remaining personnel in the Laser Section worked in basic research and in improving Iraq's scientific expertise.

The IAEA concluded that Iraq's LIS program was loosely coordinated, largely empirical and had not made much progress. The inspectors found no indications that this program had reached the point of an integrated experiment that separated any uranium, although a rudimentary laser excitation experiment related to MLIS was carried out on 1 gram of uranium hexafluoride.

According to statements made by Iraqis to the inspectors, export controls and voluntary refusals by several suppliers had prevented the Iraqi program from importing critical pieces of equipment, such as advanced laser instrumentation and related accessories.

Table 11.5 -- Projected Iraqi weapon-grade uranium inventories.

V. Projected indigenously produced weapon-grade uranium inventory for Iraq

Although Iraq's nuclear weapon program has been destroyed, questions remain about when it would have acquired its first nuclear weapon and how many weapons it could have built. Based on the above discussion, an attempt is made below to answer these questions at least speculatively. We will never know the true answers. The figures in table 11.5 are the author's estimates of the amounts of weapon-grade uranium Iraq could have produced based on the known successes and failures of the program before the Persian Gulf War. These estimates are based on the following assumptions:

1. Al Tarmiya would have been operated successfully at an average output of 10 kg of weapon-grade uranium (93 percent) per year starting in 1995. This value assumes that the R120 separators would have sustained currents totaling about 400 milliamps at the collectors, and a machine availability of 55 percent. Ash Sharqat would not have been finished.
2. Iraq could have operated a 120-machine centrifuge cascade by early 1993, added a 500-machine cascade in early 1995, added another 500 machines by 1996, and reached 2500 operating machines by early 1997. The analysis assumes that after 1997 the number of machines would have remained at 2500 for several years. The separative output of each machine is assumed to be 2.7 SWU per year. The tails assay is taken as 0.5 percent. Iraq would be expected to achieve a significant expansion in the number of centrifuges or any increase in the separative output per machine soon after the turn of the century. The increase in separative output could include the construction of supercritical machines.
3. Iraq would not have succeeded in building a LIS plant or chemical enrichment plant able to produce significant quantities of LEU or the EMIS program; it would not have acquired a significant stock of unsafeguarded LEU or diverted its safeguarded stock.
4. Iraq would have succeeded in designing a nuclear weapon requiring 15 kg of weapon-grade uranium and developing recycling capabilities to maximize its supply of weapon-grade uranium.