Though hardly transparent, Director of National Intelligence John Negroponte’s testimony on Iran before the Senate Intelligence Committee on February 2 was clearly cautious. The U.S. intelligence community judges that Iran probably has neither a nuclear weapon nor the necessary fissile material for a weapon, he stated. If Iran continues on its current path, it “will likely have the capability to produce a nuclear weapon within the next decade,” he added. The basis for this estimate remains classified, although Iran’s lack of knowledge and experience in building and running large numbers of centrifuges for uranium enrichment was reportedly an important consideration. When pressed, U.S. officials have said that they interpret Negroponte’s remark to mean that Iran will need roughly 5-10 years before it possesses nuclear weapons.
Despite this caution, a handful of U.S. officials have since attempted to overstate Iran’s nuclear progress, contradicting even this latest estimate. It appears that in the ongoing crisis between Iran and the United States, the crucial struggle for public perception of the Iranian nuclear threat is well under way.
Following an International Atomic Energy Agency (IAEA) briefing of U.N. Security Council permanent members and Germany in mid-March about a group of 164 centrifuges at Iran’s Natanz uranium enrichment site, U.S. officials began to distort what the IAEA had said. Under the cloak of anonymity, these officials told journalists that Iran’s actions represented a significant acceleration of its enrichment program. The IAEA was “shocked,” “astonished,” and “blown away” by Iran’s progress on gas centrifuges, according to these U.S. officials, leading the United States to revise its own timeline for when Iran will get the bomb. In reality, IAEA officials said they were not surprised by Iran’s actions. These U.S. statements, a senior IAEA official told the Associated Press, came “from people who are seeking a crisis, not a solution.” 
Some outside experts and officials, including Defense Secretary Donald Rumsfeld, may be trying to undermine U.S. intelligence assessments on Iran’s timeline to the bomb by highlighting the intelligence community’s failure to correctly assess Iraq’s weapons of mass destruction efforts.  Although the intelligence community deserves strong criticism for its analysis of Iraq’s weapons programs, the more recent Iranian analysis has been subject to more thorough review and is more consensual than the Iraqi assessments. For example, centrifuge experts at Oak Ridge National Laboratory, who challenged faulty CIA conclusions that Iraqi aluminum tubes were for a reconstituted nuclear weapons program long before the war, have been central in assessing Iran’s gas centrifuge program for the intelligence community, according to a U.S. intelligence official.
Iran is indeed on the verge of mastering a critical step in building and operating a gas centrifuge plant that would be able to produce enriched uranium for either peaceful or military purposes. However, it can be expected to face serious technical hurdles before it can reliably produce large quantities of enriched uranium.
Many details about Iran’s technical nuclear capabilities and plans are unknown, and the IAEA has neither been able to verify that Iran has declared its nuclear activities in full nor to establish conclusively that Iran does not have hidden nuclear enrichment sites. Western governments view with skepticism Iranian denials of intentions to produce highly enriched uranium (HEU) or to build nuclear weapons. Yet there is no evidence of an Iranian decision to build a nuclear arsenal, let alone any knowledge of an official Iranian schedule for acquiring nuclear weapons.
During the past three years of IAEA inspections, the international community has learned a great deal of information about the Iranian program that can be used to estimate the minimum amount of time Iran would need to produce enough HEU for a nuclear bomb. According to several possible scenarios, Iran appears to need at least three years before it could have enough HEU to make a nuclear weapon. Given the technical difficulty of the task, it could take Iran much longer.
With political rhetoric likely to intensify during the coming months, it is essential to have as clear an evaluation as possible of Iranian nuclear capabilities. It is also essential to avoid repeating the mistakes that were made prior to the Iraq War, when senior Bush administration officials and their allies outside government hyped the Iraqi nuclear threat to gain support in confronting Iraq.
Out of the gate
Iran’s recent actions appear aimed at rapidly installing and running gas centrifuges, which can be used to separate uranium 235 from uranium 238–the process known as enrichment. In early January 2006, Iran removed 52 IAEA seals that verified the suspension of Iran’s P-1 centrifuge uranium enrichment program that had been in effect since October 2003. (The P-1 centrifuge is a design that Iran developed from plans acquired through the nuclear smuggling network of Pakistani scientist A. Q. Khan.) The seals were located at the Natanz, Pars Trash, and Farayand Technique sites, Iran’s main centrifuge facilities. On February 11, Iran started to enrich uranium in a small number of centrifuges at Natanz.
After removing the seals, Iran also started to substantially renovate key portions of its main centrifuge research and development facility, the Pilot Fuel Enrichment Plant at Natanz. Iran secretly began construction on the pilot plant in 2001, and it installed about 200 centrifuges in 2002 and 2003. The pilot plant is designed to hold six 164-machine cascades, groups of centrifuges connected by pipes that work together to enrich greater amounts of uranium to higher enrichment levels than a group of individual centrifuges. The plant has space for additional, smaller test cascades, for a total of about 1,000 centrifuges.
At Natanz and Farayand Technique, Iran quickly restarted testing and checking centrifuge components to determine if they were manufactured precisely enough to use in a centrifuge. By early March, Iran had restarted enriching uranium at the pilot plant in 10- and 20-centrifuge cascades.
Iran also moved processing tanks and an autoclave–used to heat centrifuge feed material known as uranium hexafluoride into a gas prior to insertion into a centrifuge cascade–into its main production facility, the underground Fuel Enrichment Plant (FEP) at Natanz. This plant is designed to eventually hold 50,000-60,000 centrifuges. Iran told the IAEA that it intends to start installing the first 3,000 P-1 centrifuges at the FEP in the fourth quarter of 2006. A key outstanding question is whether Iran has procured from abroad or domestically manufactured all the equipment and materials it needs to finish the first module of 3,000 centrifuges.
Iran’s Uranium Conversion Facility at Isfahan, which converts natural uranium into uranium hexafluoride, has continued to operate since restarting in August 2005, following the beginning of the breakdown in the suspension. By May 2006, Iran had produced 110 metric tons of uranium hexafluoride.  Assuming that roughly 5 metric tons of uranium hexafluoride are needed to make enough HEU for a nuclear weapon, this stock represents enough natural uranium hexafluoride for more than 20 nuclear weapons. Although this uranium hexafluoride contains impurities that can interfere with the operation of centrifuges and reduce their output or cause them to fail, most IAEA experts believe that Iran can overcome this problem and that the issue of hexafluoride impurity has been overblown in the media. Iran is known to be working to improve the purity of its uranium hexafluoride. If necessary, Iran could use its existing stock of impure material, either further purifying this uranium hexafluoride or settling for reduced output and a higher centrifuge failure rate.
A key part of the development of Iran’s gas centrifuge program is the operation of the 164-machine test cascades at the Natanz pilot plant, which will be the workhorses of any future centrifuge plant. Iran finished installing its first test cascade in the fall of 2003, but the cascade never operated with uranium hexafluoride prior to the October 2003 suspension. On April 13, 2006, Iran announced that it had produced low-enriched uranium (LEU) in its 164-machine cascade. Soon afterward, it announced that it had enriched uranium up to a level of almost 5 percent.
Restarting the cascade took several months because Iran had to repair damaged centrifuges. According to IAEA reports, many centrifuges crashed or broke when the cascade was shut down at the start of the suspension in 2003. Before introducing uranium hexafluoride, Iran had to reconnect all the pipes, establish a vacuum inside the cascade, and prepare the cascade for operation with uranium hexafluoride.
The initial performance of the P-1 centrifuges in this cascade has been lower than expected. Based on the April 12 statements of Gholam Reza Aghazadeh, head of the Atomic Energy Organization of Iran, the average annualized output of the centrifuges in this cascade is relatively low.  In the same interview, Aghazadeh implied that he expects the average output of each P-1 centrifuge to almost double in the main plant.
In addition, Iran has not yet run this cascade continuously to produce enriched uranium. According to a Vienna diplomat, the cascade operated with uranium hexafluoride only about half of its first month of operation, although it continued to operate under vacuum the rest of the time. During this period, according to a May 19 Agence France Presse report, the cascade produced only “dozens of grams” of enriched uranium, far below the more than 2,000 grams Aghazadeh predicted the cascade would produce running continuously for that length of time. The Iranian centrifuge operators do not yet have sufficient understanding of cascade operation and must conduct a series of longer tests to develop a deeper understanding.
The IAEA reported in April that Iran is building the second and third cascades at the pilot plant. A senior diplomat in Vienna said in a late-April interview that the second and third cascades could start by early summer. This schedule would allow Iran to test multiple cascades running in parallel, a necessary step before building a centrifuge plant composed of such cascades. The diplomat speculated that Iran could continue with this pattern, installing the fourth and fifth in July and August, respectively. The space for the sixth cascade is currently occupied by the 10- and 20-machine cascades, he said.
Iran would likely want to run its cascades individually and in parallel for several months to ensure that no significant problems develop and to gain confidence that it can reliably enrich uranium in the cascades. Problems could include excessive vibration of the centrifuges, motor or power failures, pressure and temperature instabilities, or breakdown of the vacuum. Iran may also want to test any emergency systems designed to shut down the cascade without losing many centrifuges in the event of a major failure. Absent major problems, Iran is expected to need until the fall or later to demonstrate successful operation of its cascades and their associated emergency and control systems.
Once Iran overcomes the technical hurdle of operating its demonstration cascades, it can duplicate them and even create larger cascades. Iran would then be ready to build a centrifuge plant able to produce significant amounts of enriched uranium either for peaceful purposes or for nuclear weapons.
The underground path
Answering the question of how soon Iran could produce enough HEU for a nuclear weapon is complicated and fraught with uncertainty. Beyond the technical uncertainties, several other important factors are unknown. Will Iran develop an enrichment capability but produce only LEU for use in nuclear power reactors and not any HEU for use in a nuclear weapon? Will Iran withdraw from the Nuclear Non-Proliferation Treaty (NPT), expel IAEA inspectors, and concentrate on building secret nuclear facilities? How does the Iranian regime perceive the political risks of a particular action, such as trying to make HEU in the pilot plant? What resources will Iran apply to finishing its uranium enrichment facilities? Will there be preemptive military strikes against Iranian nuclear sites?
For the purposes of these estimates, a crude fission nuclear weapon is estimated to require 15-20 kilograms of weapon-grade uranium (HEU containing more than 90 percent uranium 235).  Iran’s most direct path to obtaining HEU for nuclear weapons is to build a relatively small gas centrifuge plant that can make weapon-grade uranium directly.  If Iran built such a plant openly, it would be an acknowledgement that it seeks nuclear weapons and would invite a harsh response from the West and the IAEA.
As a result, Iran would likely pursue such a path in utmost secrecy, without declaring to the IAEA the facility and any associated uranium hexafluoride production facilities. Because Iran announced earlier this year that it was ending its implementation of the Additional Protocol–an advanced safeguards agreement created in the 1990s to fix traditional safeguards’ inability to provide adequate assurance that a country does not have undeclared nuclear facilities or materials–the IAEA would face a difficult challenge discovering such a clandestine facility. The IAEA has already reported that it can no longer effectively monitor centrifuge components, unless they are at Natanz and within areas subject to IAEA containment and surveillance.
A centrifuge plant containing about 1,500-1,800 P-1 centrifuges is sufficient to make more than enough HEU for one nuclear weapon per year. (Each P-1 centrifuge is assumed to have an output of about 2.5-3 separative work units [swu] per year.  With a capacity of 4,500 swu per year, this facility could produce as much as 28 kilograms of weapon-grade uranium a year. )
Iran has enough components to build up to 5,000 centrifuges, according to some senior diplomats in Vienna. Other senior diplomats, however, have said that Iran may not have 5,000 of all components, and that many components are not expected to pass quality control. In total, Iran is estimated to have in hand enough decent components for at least 1,000 to 2,000 centrifuges, in addition to the roughly 800 centrifuges already slated for the pilot plant. Iran could also build new centrifuge components, and, in fact, may have already started to do so.
If Iran had started to build a clandestine plant with 1,500-1,800 centrifuges in early 2006, it could assemble enough additional usable machines in about 15-18 months, or by about mid-2007. It would need to assemble centrifuges at the upper limit of its past rate, about 70-100 centrifuges per month, to accomplish this goal. In the meantime, Iran would need to identify a new facility where it could install the centrifuge cascades, since it is unlikely to choose Natanz as the location of a secret plant. It would also need to install control and emergency equipment, feed and withdrawal systems, and other peripheral equipment. It would then need to integrate all of these systems, test them, and commission the plant. Iran could start immediately to accomplish these steps, even before the final testing of the 164-machine cascades at Natanz, but final completion of a clandestine plant would be highly unlikely before the end of 2007.
Given another year to make enough HEU for a nuclear weapon, and a few more months to convert the uranium into weapon components, Iran could have its first nuclear weapon in 2009. By this time, Iran could have had sufficient time to prepare the other components of a nuclear weapon, although the weapon may not be small enough to be deliverable by a ballistic missile.
This result reflects a worst-case assessment for arms control. Iran can be expected to take longer, as it is likely to encounter technical difficulties that would delay bringing a centrifuge plant into operation. Factors causing delay could include difficulty assembling and installing so many centrifuges in such a short time period, inability to achieve the relatively high separative work output used in these estimates, difficulty acquiring sufficient dual-use equipment overseas, taking longer than expected to overcome difficulties in operating the cascades as a single production unit, or a holdup in commissioning the secret centrifuge plant.
Iranian officials have recently announced that they are also working on developing the more advanced P-2 centrifuge, the designs for which were also obtained from the Khan network. Iran’s progress on this centrifuge appears to lag behind that of the P-1 centrifuge, as evidenced by a lack of procurement records for P-2 parts. The IAEA has been unable to determine the exact status of the P-2 program, but what is known appears to exclude the existence of undeclared P-2 facilities sufficiently advanced to significantly shift projections of the amount of time Iran would need to produce nuclear weapons.
Readying a “breakout”
Another way that Iran could produce HEU for nuclear weapons would be to use its Natanz production facility, even though the centrifuge module is being designed to produce LEU for use in nuclear reactors. Iran has said it intends to start installing its first module of 3,000 centrifuges in the production facility’s underground halls in late 2006, though it doesn’t presently have enough centrifuge parts to complete the module. Since the pilot plant would likely have already produced a relatively large amount of LEU, the time to produce enough HEU for a nuclear weapon in this facility could be dramatically shortened.
At the above rates of centrifuge assembly, and assuming that it has, can produce, or acquire abroad enough P-1 centrifuges and associated equipment, Iran could finish assembling the module’s 3,000 centrifuges sometime in 2008. Although Iran would likely build and operate some cascades before all the centrifuges are assembled, it will probably need at least another year to finish this module, placing the completion date in 2009 or 2010. Unexpected complications could delay the commissioning date. Alternatively, Iran could accelerate the pace by manufacturing, assembling, and installing centrifuges more quickly. Given all the difficult tasks that must be accomplished, however, Iran is unlikely to commission this module much before the start of 2009.
If Iran decided to make HEU in this module, it would have several alternatives. Because of the small throughput and great operational flexibility of centrifuges, HEU for nuclear weapons could be produced by reconfiguring the cascades in the module or by batch recycling, which entails feeding the cascade product back into the same cascade for subsequent cycles of enrichment.
Reconfiguration could be as straightforward as connecting separate cascades in series and carefully selecting the places where new pipes interconnect the cascades. Iran’s 3,000-centrifuge module is slated to be composed of almost 20 164-centrifuge cascades, operating together under one common control system. With such a setup, reconfiguration would not require the disassembly of the individual cascades and could be accomplished within days. Such a setup could lessen by 10 percent the enrichment output, and the HEU’s final enrichment level may reach only 80 percent, which is still sufficient for use in an existing implosion design, albeit with a lower explosive yield.
With a reconfigured plant, and starting with natural uranium, 20 kilograms of HEU could be produced within four to six months. If Iran waited until it had produced a stock of LEU before reconfiguring and then used this stock as the initial feedstock in the reconfigured plant, it could produce 20 kilograms of HEU in about one to two months.
Batch recycling would entail putting the cascade product back through the cascade several times, without changing the cascade’s basic setup. Starting with natural uranium, cascades of the type expected at Natanz could produce weapon-grade uranium after four to five recycles. Twenty kilograms of weapon-grade uranium could be produced in about six to twelve months. If the batch operation started with an existing stock of LEU, the time to produce 20 kilograms of weapon-grade uranium would drop to about one to two months.
Whether using batch recycling or reconfiguration, Iran would likely operate the module to make LEU so that any production of HEU would be expected to happen quickly. Still, using either of these breakout approaches, Iran is not likely to have enough HEU for a nuclear weapon until 2009, and technical obstacles may further delay the operation of the module in the production facility.
Looking at a timeline of at least three years before Iran could have a nuclear weapons capability means that there is still time to pursue aggressive diplomatic options and time for measures such as sanctions to have an effect, if they become necessary.
In the short term, it is imperative for the international community to intensify its efforts to disrupt or slow Iran’s ongoing overseas acquisition of dual-use items for its centrifuge program. Iran has encountered greater difficulty acquiring these items because of the increased scrutiny by key supplier states and companies, forcing Iranian smugglers to look elsewhere. As Iran applies more devious methods or seeks these items in other countries, greater efforts will be required to thwart it from succeeding.
It is vital to continue to understand what Iran has accomplished, what it still has to learn, and when it will reach a point when a plan to pursue nuclear weapons covertly or openly could succeed more quickly than the international community can react. Although these estimates include significant uncertainties, they reinforce the view that Iran must foreswear any deployed enrichment capability and accept adequate inspections. Otherwise, we risk a seismic shift in the balance of power in the region.
1. George Jahn, “U.N. to Inspect Iran Enrichment Program,” Associated Press, March 25, 2006.
2. In an April 18, 2006 interview on the Laura Ingraham Show, Rumsfeld said he was “not confident” that the U.S. intelligence community’s estimate of Iran’s nuclear timeline was accurate (transcript available at www.defenselink.mil/transcripts/2006/tr20060418-12862.html). At a May 9, 2006 press conference, he said that the “wrong” intelligence used to justify the U.S. invasion of Iraq should “give one pause” when evaluating the credibility of intelligence regarding Iran (“Rumsfeld: Iraq Errors Affect Assessment of Iran,” CNN, May 9, 2006).
3. This quantity refers to the amount of uranium mass in the uranium hexafluoride.
4. The annualized average output of each centrifuge was about 1.4 separative work units (swu) per machine per year, based on Aghazadeh’s statement of a maximum feed rate of 70 grams per hour and the production of 7 grams per hour of 3.5 percent enriched uranium. The feed and product rates imply a tails assay (the fraction of fissionable uranium 235 in the waste stream) of 0.4 percent. This relatively low output could mean that the aluminum centrifuge rotors are spinning at a lower speed than possible. For the main plant, Aghazadeh said that 48,000 centrifuges would produce 30 metric tons of low-enriched uranium per year. Assuming a tails assay of 0.4 percent and a product of 3.5 percent enriched uranium, the estimated average output of each machine would be about 2.3 swu/year. With an assumed tails assay of 0.3 percent, the estimated output rises to 2.7 swu/year, high for a Pakistani P-1 design but possible if the centrifuge is further optimized.
5. Iran could be expected to initially build a crude, implosion-type fission weapon similar to known designs. In 1990, Iraq initially planned to use 15 kilograms of weapon-grade uranium in its implosion design. An unclassified design using almost 20 kilograms was calculated in a study by the author and Theodore Taylor in about 1990. A larger quantity of HEU is needed than the exact amount placed into the weapon because of inevitable losses during processing, but such losses can be kept to less than 20 percent with care.
6. Alternatively, Iran could secretly build a “topping plant” of about 500 centrifuges and use a stock of low-enriched uranium produced in the pilot plant as feed to produce HEU. However, the estimated timeline for this alternative route is not significantly different from the one outlined in this scenario.
7. These values for separative work are at the high end of the possible output of Iran’s P-1 centrifuge. Actual values may be less.
8. This calculation assumes a relatively high tails assay of 0.5 percent. As a centrifuge program matures and grows, it typically reduces the tails assay to conserve uranium supplies.