By Chris Flores | Jan 1, 2003
months after the press conference to unveil the Project Sapphire operation and
declare "mission accomplished" questions arose that still linger.
Was any of the material stolen or lost?
Did any of the material fall into the hands of a rogue nation?
Was the exact amount ever really known?
By mid-1996 BWXT had sorted through the materials shipped to its Mt. Athos' plant. There appeared to be a shortfall of approximately 55 kilograms of highly enriched uranium.
That is enough to make several atomic bombs, which The Washington Times reported in its Oct. 24, 1996, edition.
The story also quoted a Department of Energy source saying government officials were worried the material might have been stolen or sold to Iran, or diverted between the time the U.S. delegation inspected the plant and the time the U.S. brought the team over for the recovery.
For several days the story spun through Washington.
Pentagon spokesman Ken Bacon deferred most questions to DOE, which was looking into the likelihood that it was an "accounting problem."
Bacon was asked if anyone was worried that the material was diverted in Kazakhstan before the U.S. planes arrived to pick it up.
"We are quite confident that there was adequate security," said Bacon, "but we'll obviously look at all aspects of this."
The Project Sapphire team's chief interpreter, Mike Dosier, said there was no uncertainty in Ulba in Kazakhstan about what was packed. They got it all, he said.
A memo from the Nuclear Regulatory Commission director to the NRC commissioners over "recent media articles" on the discrepancy said measurements in Kazakhstan were "performed on a best effort basis." But NRC documents also indicated there was no definitive accounting of the material at the beginning of the mission.
Differing amounts emerged almost immediately.
Martin Marietta employees at the Y-12 plant said in a trade journal that the material received in October 1994 included 580 kg of highly enriched uranium.
The NRC said in October 1995 there was 582.5 kilograms.
BWXT's inventory data indicated closer to 515 kilograms.
A CIA assessment is 1997 said 565 kilograms.
The discrepancy could lie in Sapphire's scrap heap - material that was swept up, repackaged and shipped. It included highly enriched uranium.
The scrap material was mostly contaminated junk swept off of the floor at Ulba, and would be the least attractive material to steal for a terrorist or country seeking highly-enriched uranium, said Tom Pham, an official at the NRC's material control and accounting division of nuclear security.
Pham said the equipment in Kazakhstan wasn't precise enough to provide accurate measurements of how enriched the uranium was in the scrap material.An NRC memo also said determining the exact amount of uranium in the scrap materials was impossible. The measurement uncertainty for radiation levels of the uranium/beryllium scrap were "on the order of plus or minus 100 percent," said the memo.
The NRC now believes the problem originated in Ulba because was it was impossible to measure exactly how much of the Kazakhstan material was highly enriched. The NRC also said there is no way to know the exact amount of HEU scrap that was repacked in Kazakhstan.
The "scrap" is described on one DOE document as uranium-beryllium oxide powder that is "machining scrap," and has radioactivity at about the same levels of the other forms of Sapphire material.
Measuring uranium enrichment levels is not as simple as weighing the material. It requires tests that measure subatomic particles radiating from the material. Measuring scrap, where radioactive material is mixed with other types of material, is harder than measuring pure nuclear material.
During repackaging in Kazakhstan, data was compiled on the mass of uranium in each can. The U-235 enrichment level was also determined, with two separate tests for each can.
After BWXT received the material, it verified the tamper-safe seals and identified the weight and contents of the containers before securing them in a vault.
NRC records show the uranium/beryllium scrap was removed from sealed containers, tested for dissolution rate, bakedand returned to containers in tamper-proof seals. But one problem was noted.
"During the same period, a security vulnerability existed relative to searches of personnel exiting the processing area under emergency conditions," said the memo.
"Through a review of licensee's records of alarms on emergency exits and discussions with security personnel, the staff concluded that there were no emergency exits from the processing area that could have exploited this vulnerability to commit a theft or diversion."
In short, no one stole any material from Mt. Athos and in all likelihood all the material found in Kazakhstan was shipped out.
Material accounting problems continued to garner attention at Mt. Athos.
BWXT had dissolved half of the 230 kilograms of material by mid-November 1996, and continued to come up about 23 percent short of the DOE's measurements. A loss as small as 50 grams of Strategic Special Nuclear Material such as highly enriched uranium or plutonium during processing is considered significant.
BWXT Sapphire documents indicate the calculated Sapphire loss of was 1,000 times higher than 50 gramshigher than that. Again, this indicated a material shortfall before it arrived at Mt. Athos.
The shortfall remained a constant target of scrutiny. In early 1998, an NRC inspector reviewed BWXT's records of how the Sapphire uranium/beryllium scrap was repacked and stored because of the "high difference" of 24 percent.
Concerns over the discrepancy may have led BWXT to make some license changes before it even received the material.
Its license regarding its "fundamental nuclear control plan" was changed before and during the Sapphire period. While the specifics are classified, the NRC said changes included "a new material balance area" and "new process monitoring control units" for Project Sapphire downblending operations.
BWXT made procedural changes in 1999 for Sapphire materials because of a "lack of confidence" in the measured enrichment levels.
Details of the control plans remain sketchy. A section dealing with how well nuclear material is stored, guarded and accounted for was deleted from an NRC evaluation of BWXT operations obtained through the Freedom of Information Act.
Secrecy continued throughout the process.
Specifics of the BWXT recovery process where the uranium was separated from the impurities were kept secret so BWXT's competitors couldn't have them. So were the nuclear criticality safety evaluations.
A monitoring system for the International Atomic Energy Agency - the U.N. weapons inspection and nuclear cooperation agency - showed BWXT fulfilled, on behalf of the U.S. government, its commitment to Kazakhstan throughout the downblending process.
Those reports are classified, including correspondence with Hans Blix, who is now the chief weapons inspector in Iraq.
Will the questions about the amount of materials ever be answered? Probably not with a 100 percent degree of certainty.
Project Sapphire and references to its "missing bomb materials" will continue to routinely show up on conspiracy web sites. But all indications are the loss was on paper only.
BWXT's original plan to transform the Sapphire material from weapons-grade uranium to fuel was to dissolve it in hot nitric acid and remove any impurities, combine it with uranium enriched to 1 percent and dry it to form low-enriched uranium crystals for fuel.
The downblending was expected to be completed within a year. But when processing began in January 1996, technical problems left the schedule in disarray.
BWXT discovered some of the uranium wouldn't dissolve because it also contained hydrogen. The plan was to remove the hydrogen by baking the uranium in a calciner, a special high temperature furnace that operated in a glove box and vented the gas byproduct through a duct. BWXT told the DEQ it would filter out 99.97 of the contaminants.
Some of the uranium also contained beryllium. It too wouldn't dissolve.
In March BWXT found a Sapphire container that didn't accurately describe the material, sparking concerns that other containers may have been mis-labeled. Processing was put on hold because BWXT wanted more information from DOE.
There were also bureaucratic challenges.
The Nuclear Regulatory Commission said BWXT had to amend its license to bake the material. BWXT argued this wasn't necessary because the process was "extremely benign." BWXT was not happy about the decision because it meant pushing the project further behind schedule. The license amendment was approved in May 1996.
Work resumed in late spring and BWXT began baking the tainted material. It finished baking and dissolving all of the beryllium-tainted uranium by November 1996 except for the uranium/beryllium scrap.
The scrap was proving to be a project stopper.
In August, a BWXT worker got an unwelcome surprise. When scrap powder containing weapons-grade uranium was added to a tray with water and traces of acid to be dissolved, it caught fire.
A report on the incident indicated that small yellow flames appear randomly on spots on the surface of the powder in the tray and then self-extinguished. The flames went out in 10 seconds, but gas bubbles continued for 45 minutes.
The fire, in nuclear terms, did not reach criticality - an uncontrolled nuclear chain reaction. There was no radiation burst or contamination, but BWXT immediately stopped working with the scrap.
David Ayers, a top NRC official in charge of BWXT's region, said there wasn't large enough amount of highly-enriched uranium in the Sapphire fire "to cause a criticality."
The movement of neutrons in nuclear material creates a criticality risk. If there had been more highly enriched uranium in the scrap, the results could have been disastrous.
Sorting out the problem took three months.
BWXT laboratory tests determined the scrap material had not been fully baked and carbides in the material might have caused the flammable gases by reacting with the water. BWXT said baking the material again at high temperatures before further dissolving should prevent a future fire.
Again, more time and added costs.
The NRC also blamed Y-12 contractor Martin Marietta for not sampling the Sapphire material before shipping it to BWXT. Sampling "likely would have identified the materials that caused the flame up," said the NRC.
BWXT was cited for NRC violations in connection with how it stored the material that flamed up, but not for the fire itself. The material included a total of 315 grams of U-235, higher than the allowable limit of 275 grams for one storage area within the vault.
That was a violation of the nuclear criticality safety mass limit. The limit exists because too much mass within one location can increase the potential for a criticality, which is also prevented by limits on the amount, spacing and shape of nuclear materials.
The NRC's Ayers said there was never any threat. "They would have to double or triple the amount for it to be real dangerous."
BWXT continued to process other types of metal and install new downblending equipment.
Between March 1996 and April 1997 at least six spills occurred in the recovery area used for Sapphire work, according to NRC documents.
In March 1996, about 40 liters of liquid waste with low concentrations of enriched uranium spilled, with about 10 liters leaking outside the building through floor-wall joints.
Seven months later, about 80 liters of uranyl nitrate solution was spilled in the uranium recovery area when a plastic PVC drain pipe broke. The solution was cleaned up without any exposures.
Later in that same October about 40 liters of "waste solution containing beryllium and uranium" was spilled in the uranium recovery facility. The solution was being transferred from recovery to a waste treatment facility when PVC piping broke at a glued connection.
No workers were contaminated and "no solution was released to the environment," said the NRC. "The solution was recovered and decontamination activities are in progress."
Beryllium has a much lower likelihood of becoming airborne if it is in a liquid, said the NRC's Ayers, but it's not impossible.
Twice in April 1997, a worker in uranium recovery was sprayed with a uranium-bearing acid solution on and around his face. The worker was taken to Lynchburg General Hospital both times. There was radiation on the person's skin, but at a very low level, said Ayers of the NRC.
The first incident happened because a clogged filter caused a pressure buildup in a pipe and the gasket around the filter ruptured. The second occurred when PVC piping in Sapphire downblending equipment was heated and softened by backed-up uranyl nitrate. The problem was poor system design, said the NRC.
But a March 1997 spill required BWXT to declare an alert and activate its emergency operations center.
The incident occurred when 47 liters of material containing uranium was spilled after a solution of nitric acid and 93 kilograms of uranium enriched to 1 percent overflowed into a collection basin and the floor.
Because of nitrous oxide fumes and clouds in the building, all the uranium recovery area was evacuated. According to the NRC, "there was no release to the environment outside the facility and no personal injuries."
BWXT collected the spilled material into containers and reduced the nitrous oxide in the building. The alert was terminated four hours later.
Again, all Sapphire downblending was stopped pending approval of BWXT's preliminary investigation results. BWXT later discovered its hazards analysis had not considered such an overflow.
But the biggest problems remained with the scrap material.
The scrap process was on hold for all of 1997 and most of 1998 as BWXT installed a new calciner and experienced problems with it.
BWXT wanted to restart processing the scrap in July of 1998, but the furnace wasn't ready for testing until January 1999. It malfunctioned completely on the test run. A chain of events caused scrubber solution and water to spill and ash and smoke to fill the glovebox, which then leaked.
The test did not involve radioactive material.
BWXT had to reengineer the furnace and develop computer programs to make sure there weren't other potential domino scenarios.
"Equipment problems and recent mandatory safety reviews are pushing the restart of the blending operations beyond August," BWXT said in a report to the NRC.
By the fall of 1999, 60 percent of the scrap still hadn't been processed.