Science 1663

DARHT—A New Hydrotest Facility for Stockpile Stewardship

The Dual Axis Radiographic Hydrotest Facility will house the world's first machine capable of taking x-ray mini-movies of a mock nuclear weapon implosion.

David Honaberger examining one of the refurbished accelerator cells for DARHT's second-axis accelerator.

Through a heroic effort, Los Alamos scientists are building the world's most powerful x-ray machine for analyzing nuclear weapons. It will be the first to generate a sequence of pictures showing the dynamic events that trigger a nuclear detonation.

The Idea behind DARHT

In 1992, when the United States declared a moratorium on nuclear weapons tests, Los Alamos scientists were asked to keep the stockpiled weapons in top condition without ever trying them out. The maintenance program was called Stockpile Stewardship.

Scientists knew that weapons in storage would be damaged over time by their own radioactivity and would need replacement components. But would the replacements function as required? And how could that be checked under the testing moratorium?

The answer was to perform the next best thing to a real nuclear test—a full-scale mockup of the events that trigger the nuclear detonation.

DARHT was originally designed to produce two simultaneous x-ray images taken in perpendicular directions. The facility’s intense x-ray flashes (the green rays shown here) will be generated when high-energy electron pulses from each accelerator axis slam into tungsten targets (red).
Illustration by Donald Montoya, IRM Communication Arts and Services.

During a weapon's crucial triggering phase, explosive charges that surround the nuclear fuel are detonated at multiple points. The result is a shock wave that moves inward (implosion) at supersonic speeds, compressing the fuel to higher and higher density. Implosion ends when the fuel reaches a supercritical density, the density at which nuclear reactions in the fuel build up an uncontainable amount of energy, which is then released in a massive explosion.

To make the mockup non-nuclear, a heavy metal surrogate (such as depleted uranium or lead) stands in for the nuclear fuel, but all other components can be exact replicas of the real thing and their behavior tested under implosive conditions.

During the test the surrogate fuel and other components become hot enough to melt and flow like water, so this mock implosion is called a hydrodynamic test, or hydrotest.

Standard practice is to take a single stop-action snapshot of the weapon mockup's interior as the molten components rush inward at thousands of meters per second.

A series of snapshots would be even better. Scientists could follow the implosion's progress for as long as possible and compare the pictured component positions with the predictions of computer simulations.

No one has yet made such a set of ultra-high-speed images. But it will finally be possible when the second arm (axis) of DARHT, the Dual Axis Radiographic Hydrotest facility, finally comes online.

The Beginning at Los Alamos

Planning for DARHT began in the early 1980s. The idea was to have a pair of separately housed giant x-ray machines pointing at right angles toward a test object between them. During a hydrotest, a single short pulse of x-rays from each machine would simultaneously penetrate the imploding test object, affording scientists instantaneous front and side views of the implosion. From those would come the first accurate three-dimensional picture of implosion dynamics.

At the onset of the test moratorium in the early 1990s, DARHT's planned capability was seen as the perfect match for the new challenges of Stockpile Stewardship. Hydrotests at DARHT, in combination with accurate computer simulations, would let scientists guarantee, without testing, that stockpiled nuclear weapons would perform as specified if they were ever needed.

Approval for the two DARHT axes came in stages, with the first axis approved for construction in 1992 and the second axis (initially to be a twin of the first) in 1997. But by then, the U.S. Department of Energy (DOE) had made a different decision. It wanted the second axis to deliver not one view of the implosion, but the never-before-captured series of views.

The change in scope was to have unexpected consequences.

DARHT's First Axis—Sharpening the X-Ray Image

The challenge for the first stage, the first axis, was to design a much more powerful and precise x-ray source that would yield significantly higher-quality images than ever before.

X-rays that can penetrate the heavy metal in a weapon mockup are typically made at an electron accelerator. An electron beam moving at near the speed of light is smashed into a tungsten target. The electrons are yanked off course by the strong electrostatic pull of the positively charged nuclei in the tungsten atoms, and their sudden change in direction causes them to give off energy in the form of high-energy x-rays.

The 3-foot-diameter induction cells of DARHT’s first axis.

Scientists already knew how to use a short burst (pulse) of high-energy electrons (rather than a continuous beam) to make a short pulse of high-energy x-rays. The new challenge was for the accelerator to deliver a very large number of electrons in a single pulse—several thousand amperes of electric current (household circuit breakers blow at 20 amps)—to generate a super-intense x-ray flash that could penetrate the mockup late in the implosion. That's when the heavy metal surrogate comes close to the density at which nuclear reactions in a real weapon start to build up in the fuel.

Furthermore, to increase the image quality, the electron beam-pulse would have to be ultrashort and focused to a very small spot on the tungsten target. As with the hole in a pinhole camera, the smaller the beam spot, the more point-like the area producing x-rays, and the sharper the resulting image. Also, to achieve stop-action shots of materials barreling inward at thousands of meters per second, the electron pulse (and resulting x-ray flash) needed to be shorter than 100 billionths of a second, about a million times shorter than exposures achieved with a high-end conventional camera.

A static test object placed between the DARHT first-axis x-ray source (cone-shaped projection at right) and a camera system (left). The sphere is used to test the strength of the x-ray pulse.
Photo by LeRoy Sanchez, IRM-CAS Media Services .

That combination was a very tall order. But Lawrence Livermore National Laboratory in California had already developed an advanced electron accelerator for its own x-ray hydrotest facility, and that machine, known as a linear induction accelerator, met many of DARHT's requirements. In 1987 Los Alamos chose the same type of accelerator for its facility, but with more stringent requirements (see "How It Works" on the next page).

When completed in 1999, the first-axis accelerator could readily produce one short electron pulse (60 billionths of a second), of extreme intensity (2,000 amps) and with an energy of 20 million electronvolts. And it could focus the beam to a 2-millimeter-diameter spot on the target. It was the smallest spot size and shortest pulse length ever achieved at that intensity.

As a result, the overall image quality was 10 times higher than ever before achieved at a Los Alamos facility and about 3 times higher than was possible at Livermore's x-ray facility.

Beginning in December 1999, Los Alamos weapons designers were privy to the clearest single views ever made of the inside of a hydrotest object. The views helped validate new descriptions of implosion physics used in computer simulations of weapons performance.

The Drive for Multiple Pulses

By the time Los Alamos knew that the second-axis acclerator needed to produce multiple x-ray pulses, the environmental impact statements for the entire DARHT facility had already been approved, and construction of the twin buildings was complete. The long narrow hall for the second-axis accelerator was empty and waiting. But now it was the wrong size.

Making multiple x-ray pulses from a single-pulse induction accelerator would require creating a single electron pulse about 2 millionths of a second long—33 times longer than the pulse in the first axis. It would then be chopped into four shorter pieces that would reach the target sequentially.

But boosting that longer-lasting pulse to high energy would require an accelerator four to five times longer than the space planned for it!

With the second-axis building already complete, the scientists had to find a way to squeeze a long accelerator into the much shorter space—a kind of "square peg in a round hole" problem. It could be done, but only by leaving behind some well-honed accelerator design principles.

A Bold New Design

A single-pulse linear induction accelerator like that planned for both DARHT axes consists of a long row of doughnut-shaped magnetic induction cells, each connected to a high-voltage generator. . At the instant of firing, each generator discharges its power, creating a pulse of electric current through its induction cell, which in turn creates a large voltage difference across the gap separating that cell from its neighbor.

How It Works: Linear Induction Accelerator

The single-pulse linear induction accelerator in each DARHT axis consists of a long row of doughnut-shaped induction cells (only three are shown here in this two-dimensional view) with a large accelerating voltage difference, –200 kilovolts (kV), across the gap between each pair of neighboring cells. The electron beam-pulse travels through the central bore of the cells, receiving a 200-kiloelectronvolt energy kick each time it passes though a gap.

To create the accelerating voltage across the cell gap, a negative voltage pulse from a generator enters each cell (red) and travels down the high-voltage plate, which connects to the inner cylindrical surface of each cell. Together, the high-voltage plate, the cell end plate, and the inner and outer cylindrical surfaces form a conducting cavity. The voltage pulse returning to ground (zero volts) generates a current (red-to-blue transition) around the magnetic cores and an increasing (inductive) magnetic field (not shown) within them.

If the cavity were empty, it would act like a short circuit, drawing too much current from the generator and reducing the voltage pulse length to a few billionths of a second. When filled with the annular magnetic “cores” surrounding the central bore, the cavity acts like an inductance, resisting the flow of current from the generator.

Simultaneously, electrons injected into the beam line—the aligned "holes" in the induction-cell "doughnuts"—speed through the vacuum at the center of the cells, getting an energy kick at each gap.

Clearly, the voltage pulses across the cell gaps would have to persist for as long as the electron pulse did, so the scientists had to greatly slow the rate at which the power was discharged from each generator. But the voltage pulse quits when the induced magnetic field in the core of each induction cell builds up to saturation (the induced magnetic field can increase no more).

Since the time to saturation grows in proportion to the cross-sectional area of the magnetic core, the only way to prolong the voltage across the cell gaps for a relatively long time—2 millionths of a second—was to add more magnetic material to the cores. But the accelerator could not be lengthened along the beam line, so the only choice was to increase the diameter of the magnetic cores.

A design team from California's Lawrence Berkeley National Laboratory and Los Alamos ventured boldly into this unexplored design territory, designing new magnetic cores that were twice the diameter of the first-axis cores.

The team kept the increase to a minimum by replacing the material (ferrite) used in the first axis cores with metglas—paper-thin ribbons of amorphous iron tape. The maximum magnetic field strength (saturation point) in metglas is five times higher than in ferrite.

The magnetic tape was insulated by thin (less than a thousandth of an inch) layers of mylar and wound up into a roll of 20,000 turns to make mammoth six-foot-diameter cores, each four inches wide and weighing more than one and a half tons. Four cores went into each induction cell.

Sparks Fly—What's the Problem?

In early 2003, after fully assembling the accelerator, with its 78 induction cells, and successfully testing it with the cells at a lower voltage, the DARHT team was ready to crank the machine up to full power.

And then the unthinkable happened. Electrical breakdown! As the cells fired in sequence, they began to spark.

This three-dimensional cutaway view shows the two regions of a single induction cell—(1) the oil-filled region containing the high-voltage plate and the magnetic cores and (2) the inset region where vacuum, metal, and high-voltage insulator meet. Electrical breakdowns were observed in both regions. Each induction cell weighs about 15,000 pounds and contains four narrow magnetic cores saturated with oil.

After recovering from the initial shock, Los Alamos gathered a team of the best accelerator and pulsed-power scientists and engineers from Los Alamos, Lawrence Berkeley, and Lawrence Livermore National Laboratories, as well as industry experts. This team launched an all-out effort to identify the problems and find solutions that would allow the machine to be re-engineered without adding new materials and components.

The origin of the high-voltage electrical breakdown turned out to be unexpectedly high electric fields between the high-voltage plate and the oil-insulated magnetic cores and at sites where metal, high-voltage insulator, and vacuum meet in the vacuum side of the cell.

Why wasn't the danger of high-voltage breakdown identified in the original design of the cells or detected during the original fabrication and testing phase? Because unbeknownst to the developers, the equipment for calibrating the test voltage was faulty.

Kurt Nielsen, the lead pulsed-power scientist for refurbishment of the cells, describes how the team began to turn things around. "Tens of fixes, both high-voltage and mechanical, were identified, thoroughly tested, and implemented."

Images of four electron-beam spots produced by the Scaled Down Accelerator and (above left) the corresponding electron pulses (orange) chopped out of a single pulse (magenta).

The most dramatic change was lengthening each cell by one inch, which redistributed the electric fields in the oil-filled magnetic core region and reduced their magnitude—but also reduced by four the number of cells that would fit in the building. Another was modifying the high-voltage vacuum insulator that separates the magnetic core region of each cell from the vacuum beam line to ensure that it prevents the high voltage from leaking across the cell accelerating gap.

The Proof Is in the Testing

To check that the fixes would work, the team went through a series of rebuilds, called pre-prototypes, solving technical problems along the way and then testing the final configurations at high voltage.

Six prototype cells were then built and tested by being fired hundreds of thousands of times at 20 percent over the required voltage of 200 kilovolts. Not a single breakdown!

At the same time, the team launched an experimental campaign using the original cells but operating them at lower voltage (100 kilovolts) to produce electron pulses with lower energy (about 7–million electronvolts). The experiments tested the stability of a 2-millionths-of-a-second, 1-thousand-amp electron pulse, the longest pulse ever at such intensity. The results of the experiments put to rest lingering questions about stability, showing definitively that the long pulse did not break up or start corkscrewing as it traveled down the accelerator.

By July 2005 the team had entered the second phase of the project: to develop and implement a plan to disassemble and refurbish all the cells. As Juan Barraza, the lead mechanical engineer, says, "We introduced lean manufacturing, a production line approach, and it worked so well it's become a model for other Lab projects."

Final proof of the design came in the form of the Scaled Accelerator, a scaled-down version (test stand) of the full-energy machine. Twenty-six of the 74 refurbished cells, along with the "kicker," the component that would chop the long pulse into four short ones, were put together with the target system. The Scaled Accelerator was then ready for a first-time test at an energy of 8-million electronvolts.

The moment arrived in October 2006. The team fired up the Scaled Accelerator and watched in jubilation in the control room as four ultrashort, super-intense, rapid-fire x-ray bursts, made from a single electron pulse, flashed on the monitor panels. It was the first time ever anywhere in the world.

This first success was followed by a series of further tests that were just completed in February 2007. All the evidence suggests that x-ray intensity, pulse length, and spot size will easily scale to meet specifications when operating at full energy.

The Future

DAHRT Project Director Ray Scarpetti (left) and Deputy Director Subrata Nath, standing proudly beside the long rows of refurbished cells now being installed in the second axis.

"A group of talented folks worked tirelessly in the face of skepticism. The result is a cell whose performance exceeded the original specifications," says Project Director Ray Scarpetti.

Scarpetti's deputy, Subrata Nath, sums up the significance of their work by saying, "The ability to produce multiple pulses with varied intensities in a pre-set time sequence means that the weapon designers will get to specify what they want to see, and DARHT will be able to deliver."

In June the DAHRT team will start full-energy commissioning of the entire second-axis accelerator.