4

Lawrence Livermore National Laboratory

A powerful new imaging

technique will complement

the x raying of nuclear

warheads.

5

Lawrence Livermore National Laboratory

ROM the dentist’s office to theaircraft hangar, the use of x rays to

reveal the internal structure of objects isa time-honored practice. However, duringthe past few decades, several industrieshave begun to use thermal, or low-energy,neutron imaging as a complementarytechnique to x-ray imaging for inspectingobjects without taking them apart. NowLawrence Livermore researchers havedemonstrated the power of using high-energy neutrons as a nondestructiveinspection tool for evaluating the integrityof thick objects such as nuclear warheadsand their components.

Experiments conducted over thepast four years at Ohio University by a Lawrence Livermore team havedemonstrated high-energy neutronimaging’s considerable promise inprobing the internal structure of thick

objects composed of materials that areessentially opaque to x rays. Indeed, theresults have proven more successfulthan computer models first indicated orthan Livermore physicists had expected.

The neutron imaging project is fundedthrough the Enhanced SurveillanceCampaign, a key element of the nation’sStockpile Stewardship Program, whichis managed by the National NuclearSecurity Administration (NNSA) withinthe Department of Energy. Nondestructivesurveillance—the search for anomaliesfrom cracks to corrosion in agingstockpiled nuclear weapons systems to assure their continuing safety andreliability—is much more cost-effectivethan disassembling a warhead. Hence,the development of improvednondestructive surveillance techniquesis crucial to the success of stockpile

stewardship in the absence of nucleartesting and to the nation’s defense.

Nondestructive surveillance relieson a range of techniques, including x-rayimaging. X rays are adequate forinspecting the condition of partscomposed of what scientists call high-Z(high-atomic-number) materials such aslead, tungsten, and uranium. However,x-ray imaging is not always effective inrevealing voids, cracks, or other defectsin so-called low-Z (low-atomic-number)materials such as plastics, ceramics,lubricants, and explosives when thesematerials are heavily shielded by thick,high-Z parts. (See the box on p. 6.)

Neutrons Complement X RaysClearly, what is needed is a way to

image shielded low-Z parts as a meansto complement standard x-ray imaging

F

500

300

100

–100

–300

–500

(a) Neutron image of an object with defects taken at the Los Alamos NeutronScience Center (LANSCE). (b) Computer simulations usingLawrence Livermore’s COGMonte Carlo radiation transportcode. The simulations showthat neutron images taken atenergy ranges between 10 and15 megaelectronvolts couldreveal defects in thicklyshielded targets as well asLANSCE images, which weretaken at much higher energies.

500 300 100 –100 –300 –500

(a) (b)

Pixels

Pix

els

S&TR May 2001

of nuclear warhead components forstockpile surveillance. The answer seemsto lie with high-energy neutrons, whichare able to easily penetrate high-Zmaterials to interact with low-Z materials,

yielding clear, detailed images that aredifficult to duplicate with x rays.

According to Lawrence Livermorematerials scientist Jim LeMay, deputyprogram leader for Enhanced Surveillance,

6

Lawrence Livermore National Laboratory

Neutron Radiography

neutron imaging will be valuable tostockpile stewards on a number of fronts.He notes that weapons are randomlyselected from the nation’s nuclearstockpile for inspection. Neutronradiographs could be used as a meansto screen these weapons and select oneor more devices for complete disassemblyand visual inspection. Also, neutronradiography could serve as a valuableinspection tool for identifying thewarheads that actually need refurbishingas well as a valuable quality controltool for inspecting refurbished warheadsbefore they are returned to the stockpile.Finally, neutron imaging of a statisticallysignificant number of units could serveas a baseline assessment of the currentstate of a particular warhead.

Livermore physicist James Hall, theneutron imaging project leader, notesthat imaging systems using thermalneutrons (average energy of about0.025 electronvolt) are well establishedas nondestructive inspection tools inresearch and industry. However, thesesystems are generally limited to

A Neutron Primer

All forms of radiation are attenuated (weakened) by a combinationof slowing, scattering, and absorption processes as they pass throughmaterials. The variation in attenuation through different parts of anobject forms the basis for radiation imaging. The most widely usedand commonly known form of radiation imaging is the x-radiographin which an object is exposed to x rays and an image of the object(essentially a shadow) is recorded on photographic film or with asolid-state camera. Discovered more than 100 years ago, x raystoday have a wide range of industrial and medical applications.

Neutrons, discovered in 1932, are electrically neutral particlessimilar in mass to a proton and present in the nuclei of all elementsexcept hydrogen. Neutron imaging (conceptually similar to x-rayimaging) is commonly done today using neutrons that have anaverage energy of about 0.025 electronvolts. These neutrons aregenerated from fission neutrons produced in a nuclear reactor orfrom the decay of a radioisotope and then passed through thick layersof a hydrogen-rich material such as polyethylene to reduce theirenergy to thermal levels.

Most imaging applications using thermal neutrons exploit theirstrong interaction with hydrogen. For example, thermal neutrons

can be used to inspect or detect explosives inside brass shellcasings and search for corrosion in the aluminum skin of aircraft.

High-energy neutron imaging (for example, in the 10- to15-megaelectronvolt range) is a relatively new technique thatoffers unique advantages over conventional x-ray and thermalneutron imaging, particularly for inspecting light (low-Z, or low-atomic-number) elements that are shielded by heavy (high-Z, orhigh-atomic-number) elements. These advantages are due in partto their greater penetrating power (that is, lower attenuation)through high-Z materials and, compared to x rays, their muchstronger interaction (that is, higher attenuation) in low-Z materials.

Lawrence Livermore physicist James Hall emphasizes thatneutron imaging yields different (and complementary) informationto that obtained with x rays. “The use of one does not necessarilyeliminate the need for the other,” he says. Hall notes that althoughthe ultimate spatial resolution attainable with high-energy neutronimaging—about 1 millimeter—is about 10 times less than the spatialresolution of x-ray imaging done with the most penetrating x-rayspectrum, it may be the only way that researchers can learn anythingabout the internal structure of some extremely thick objects.

Although larger in size than the proposed Lawrence Livermore neutron imaging system, the layout ofthe facility at the Ohio University Accelerator Laboratory in Athens, Ohio, is similar in configuration.The large orange vessel in the background is a Van de Graaff accelerator. It is used to acceleratedeuterium ions into a cell containing deuterium gas to produce high-energy neutrons.

7

Lawrence Livermore National Laboratory

Neutron RadiographyS&TR May 2001

inspecting objects only a few centimetersthick. In the early 1990s, scientists atLawrence Livermore and Los Alamosnational laboratories speculated thathigher-energy neutrons could be usedto image much thicker objects such asnuclear warhead components.

Proof-of-principle tests began in 1994at the Los Alamos Neutron ScienceCenter (LANSCE), a facility thatproduces neutron beams with energies ofup to 600 megaelectronvolts (MeV), fargreater than those used by industry. Thetest object consisted of a 2.54-centimeter-thick lithium deuteride (low-Z) disk thatwas sandwiched between two 5.08-centimeter-thick uranium (high-Z)slabs. Small holes ranging from 4 to12 millimeters in diameter were drilledall or part way through the lithiumdeuteride to simulate defects. A detectorrecorded images of the neutronstransmitted through the object from theLANSCE source with a spatial resolutionof about 1 millimeter, revealing thepresence of all of the holes.

Simulations Bolstered ConfidenceEncouraged by the success of these

initial tests, Hall decided to model theLANSCE experiments using Livermore’sthree-dimensional Monte Carlo radiationtransport computer code called COG.His computer simulations, however,focused on a lower energy range (10 to15 MeV) because neutrons with theseenergies are known to penetrate high-Zmaterials effectively and yet interactmore strongly with low-Z materials thanthe much higher-energy neutrons usedat LANSCE. The COG simulationsshowed that neutron imaging in the10- to 15-MeV energy range should becapable of revealing millimeter-sizecracks, voids, and other defects in thick,shielded targets similar to the one testedat LANSCE.

Hall was also drawn to two otheradvantages of 10- to 15-MeV neutrons.The first is that neutrons in this energyrange are much less expensive togenerate than higher-energy neutrons

such as those produced at LANSCE.Second, lower-energy neutrons areeasier to detect because they allow theuse of plastic scintillators, which aresome 20 times more efficient than theconversion-type detectors required formuch higher-energy neutrons.

One disadvantage of the lowerenergy range is the somewhat reducedpenetrability of high-Z materials, whichmeans exposure times of a few hours andsometimes longer are required for typicalradiographs. However, says Hall, thegreater detection efficiency and loweroverall imaging costs more than makeup for the longer exposure times.

Following the computer simulations,Hall joined forces with colleagues FrankDietrich, Clint Logan, and Brian Rusnakto design and develop a full-scale neutronimaging system for stockpile surveillancethat would be capable of acquiring bothradiographic (single-view) and fulltomographic (three-dimensional) images.The system has to be relatively compact(about 15 meters long), both as a prototypesuitable for installation and use atLivermore and in its fully developedform for eventual installation at otherNNSA weapons complex facilities.

The resulting design features threeprimary components: an accelerator-driven neutron source generating an

intense beam of 10-MeV neutrons, aremotely controlled staging system tosupport and manipulate objects beingimaged, and a detector system withrelatively high efficiency (about 20 to25 percent) that can resolve defects ofabout 1 millimeter in diameter. Toexpedite the system’s development and minimize technical risks, the teamdecided to use commercially availablecomponents and proven neutron imagingtechniques wherever possible.

Ohio University Test BedThe team chose the Ohio University

Accelerator Laboratory (OUAL) inAthens, Ohio, to evaluate theperformance of a prototype imagingdetector beginning in 1997. Althoughthe accelerator facility at OUAL ismuch larger than that proposed in theLivermore design, its layout andconfiguration are similar. In addition, theOUAL staff has extensive experience inthe production of accelerator-driven,high-energy neutron beams.

For the Lawrence Livermoreexperiments at OUAL, a 10-MeVneutron beam is generated by focusingdeuterium ions into a cylindrical 1-centimeter-diameter by 8-centimeter-long deuterium gas cell attached to theend of the beam line. The gas cell is

Lawrence Livermore physicistJames Hall assembles a testobject called a sandwichassembly for imaging at the OhioUniversity Accelerator Laboratory.Behind Hall is a prototypemultiaxis staging system thatsecures and manipulates the testobject. On its way to the detector,the neutron beam passes throughthe test object and immediatelythrough a tapered polyethylenecollimator set into a 1.5-meter-thick concrete and steel wall.

8

Lawrence Livermore National Laboratory

Neutron Radiography S&TR May 2001

(a) A lead cylinder with a 10.16-centimeter outside diameter, a 5.08-centimeter inside diameter, and a polyethylene core was imaged. (b) The polyethylenecore was split into two half-cylinders. One served as a blank, and the other had a series of holes that were 10-, 8-, 6-, 4-, and 2-millimeter-diameter by1.27-centimeter-deep machined into its outer surface. (c) The resulting tomographic reconstructions clearly showed the core’s structure, including theslight gap between the two halves.

10.16 centimeters

10.1

6 ce

ntim

eter

s

5.08 centimeters

1.27 centimeters

Polyethylene Lead

(a) (b)

(c)

capped with thin entrance and exitwindows and maintained at a pressureof about 3 atmospheres to limit the spreadin energy of the resulting neutrons. Thetypical deuterium ion beam current

arriving at the gas cell is on the orderof 10 microamperes, which correspondsto about 60 trillion ions per second. In comparison, Lawrence Livermore’sproposed design will feature a

300-microampere accelerator with a4-centimeter-long deuterium gas cell.

The result is a neutron beam flux only15 times less intense than the intensitycalled for in the full-scale system. As a

Sideview

Topview

12.7 centimeters

12.7

cen

timet

ers

1.27 centimeters

(a) Nine step wedges fabricated from lead, Lucite, mock high explosive, aluminum, beryllium, graphite, brass, polyethylene, and stainless steel wereimaged. Each step wedge has 10 steps ranging in thickness from 1.27 centimeters to 12.7 centimeters. (b) The nine wedges were imaged as asingle unit. (c) The radiographs clearly differentiated the various materials and steps.

(a) (b) (c)

9

Lawrence Livermore National Laboratory

Neutron RadiographyS&TR May 2001

result, images take about 15 times longerto complete at OUAL than they will atLivermore. Nevertheless, the flux issufficient to evaluate the performance of prototype detectors and for LawrenceLivermore researchers to gain valuableexperience in neutron imaging. In manyways, says Hall, the Ohio Universityaccelerator lab has been a “perfect testfacility.”

The experiments conducted thus farat OUAL have focused primarily onradiographic imaging of step wedges madeof different materials and slab or sandwichassemblies, most with holes or otherfeatures machined into them to test thesystem’s resolving power. The sandwichassemblies are typically composed ofblocks of low-Z materials, such aspolyethylene, that are shielded by variousthicknesses of high-Z materials, such as lead or depleted uranium (D-38).Tomographic images of several cylindricaltest objects composed of nested shells ofhigh- and low-Z materials, with machinedfeatures, have also been obtained.

The test objects are mounted on amultiaxis staging system, which islocated on the beam axis about 2 metersdownstream from the neutron source andabout 2 meters in front of the prototypeimaging detector. The detector is housedin a shielded area behind a 1.5-meter-thickconcrete and steel wall with a taperedpolyethylene collimator to help minimizebackground radiation.

Sandwiches, Steps, and CylindersOne of the first experiments

conducted at OUAL involved imaging a 12.7-centimeter-thick lead andpolyethylene sandwich (with featuresmachined into the polyethylene) and aset of 9 step wedges (see top figure, p. 8)fabricated from lead, Lucite, mock highexplosive, aluminum, beryllium, graphite,brass, polyethylene, and stainless steel.Each step wedge had 10 steps rangingin thickness from 1.27 centimeters to12.7 centimeters. The nine wedges weregrouped together and radiographed as asingle unit (looking up the steps from

thick to thin) in a series of two 1-hourexposures. The radiographs clearlydifferentiated the different materialsand step thicknesses.

Another series of experiments involvedimaging a 7.62-centimeter-thick D-38and lithium deuteride sandwich (similarin design to the lead and polyethyleneassembly previously described) andtomographic imaging of a lead cylinderwith a 10.16-centimeter outside diameter,a 5.08-centimeter inside diameter, and apolyethylene core (see bottom figure, p. 8).

The polyethylene core was split intotwo half-cylinders. One served as a blankand the other had a series of holesmachined into its outer (curved) surfacethat were 10, 8, 6, 4, or 2 millimeters indiameter by 1.27 millimeters deep. Aseries of sixty-four 10-minute exposureswas taken of the cylinder at angles evenlydistributed over 180 degrees. Resultingtomographic reconstructions clearlyshowed the core’s structure. Althoughnot well resolved, the narrow (less than0.25-millimeter-wide) gap between thetwo halves of the polyethylene core wasalso visible in the reconstructed images.

Additional experiments at OUAL havefocused on imaging objects made of othermaterials with a variety of machined

features. One object consisted of a10.16-centimeter by 5.08-centimeter by2.54-centimeter-thick slab of ceramicset atop a polyethylene slab of similarsize and shielded by 2.54 centimeters ofD-38. The ceramic piece featured twosets of 4- and 2-millimeter-diameterholes machined to depths of 4, 2, and 1 millimeters (the smallest holecorresponded to a defect with a volumeof about 3 cubic millimeters). The ceramicwas carefully cracked along its centerlineand then reassembled so that the fracturewas barely visible to the naked eye. Thepolyethylene piece featured the same setof 4- and 2-millimeter-diameter holes butno crack.

The object was imaged in a series offorty-eight 30-minute exposures. The finalprocessed image and associated lineoutsclearly showed the crack in the ceramicslab and all of the machined features,including the smallest 2-millimeter-diameter, 1-millimeter-deep hole.

Hall says the contact gap between thetwo ceramic pieces was probably less than0.01 centimeter wide, far less than thedesigned resolution of the imagingsystem. Yet, the gap can still be resolved.“We’re very pleased we can see this kindof detail through more than 2 centimeters

This neutron radiograph of a fractured ceramic and polyethylene test object shielded by2.54 centimeters of depleted uranium shows the crack separating the two ceramic halves aswell as a series of 4-millimeter-diameter (top) and 2-millimeter-diameter (bottom) holesmachined into the ceramic. (A narrow slot was cut in the top of the ceramic to a depth of2.54 centimeters to facilitate cracking the piece along its centerline.)

Ceramic

Polyethylene

Crack

3-cubic-millimeter void

10

Lawrence Livermore National Laboratory

Neutron Radiography S&TR May 2001

The Making of a Neutron Imaging System

penetrate the cell without letting substantial amounts ofdeuterium gas leak out.

An alternative to the rotating aperture design is also beingpursued by the Lawrence Livermore–MIT team. This approach,developed at Brookhaven National Laboratory, uses an intenseplasma discharge to effectively plug the opening of the gas cellby rapidly heating and ionizing any deuterium leaking out.Similar “plasma windows” are being developed for use inelectron-beam welding applications.

The object under inspection will be secured to a stagingsystem that was originally designed at DOE’s Y-12 Plant inTennessee for x-ray imaging. The unit goes up and down andback and forth and rotates a full 360 degrees to permit bothradiographic and tomographic imaging. Calculations and testsconducted at the Ohio University Accelerator Laboratory byLivermore researchers indicate that placing the staging systemhalfway between the source and the image plane of the detectorwill minimize the neutron scattering that can fog the image.

Imaging Detector Has Nevada HeritageThe design of the imaging detector will be based on technology

originally developed by Lawrence Livermore’s Nuclear TestProgram for use at DOE’s Nevada Test Site. The full-scaledetector will consist of a 60-centimeter-diameter transparentplastic scintillator viewed indirectly by a camera with a high-resolution (2,048- by 2,048-pixel) charge-coupled device (CCD)imaging chip.

A thin turning mirror made of aluminized glass will be used toreflect the brief flashes of light generated by neutrons interactingin the scintillator into the CCD camera, which will itself be locatedin a shielded enclosure well out of the neutron beampath. Thecamera will be fit with a fast (f/1.00 or better) lens to enhance itssensitivity and cooled with liquid nitrogen gas to –120°C tominimize thermal electronic noise.

Neutronsource

Object underinspection

Shielding wall

Turning mirror

Imaging scintillator

CCD cameraimagingsystem

~2.5 meters~2.5 meters

The design of Livermore’s neutron imaging system consists ofa high-energy neutron source, a multiaxis staging platform tohold and manipulate an object, and an efficient imaging detector.The development of these components has proceeded in parallelover the past several years.

Neutrons can be produced using accelerators, radionuclides, ornuclear reactors. To achieve a high-energy neutron flux sufficientto image thick objects of interest within reasonable imaging times(a few hours), an accelerator-driven source appears to be the mostpractical option for stockpile surveillance purposes.

The accelerator, based on a commercially available design,will be built to Livermore specifications. The unit will focus anarrow (1.25-millimeter-diameter), pulsed (75-hertz),300-microampere beam of deuterium ions into a 4-centimeter-long cell containing deuterium gas. (Deuterium is an isotope ofhydrogen containing one proton and one neutron in its nucleus.)The collision of the deuterium ions with deuterium gas in the cellwill produce an intense, forward-directed beam of neutrons withan energy of about 10 megaelectronvolts.

Collaborating with MITThe combined requirements of a high deuterium-ion current

and small beam diameter preclude the use of typical thin-walled(“windowed”) deuterium gas cell designs. At an average power ofabout 170 kilowatts per square centimeter, the incident deuteriumion beam would generate far too much heat for any windowmaterial to withstand.

As a result, Lawrence Livermore researchers have teamed withnuclear engineering professor Richard Lanza at the MassachusettsInstitute of Technology (MIT) to develop a “windowless”deuterium gas cell that can be efficiently coupled to a high-current, pulsed, deuterium accelerator. One design underconsideration features a high-pressure (3-atmosphere) gas cellmounted at the exit port of a vacuum system. The cell’s severalpumping stages are isolated from each other by a series ofrotating disks with small holes synchronized to the pulsefrequency of the accelerator. In this way, the holes in the rotatingdisks line up about 75 times a second to allow the ion beam to

The Lawrence Livermore design for a high-energy neutron imaging systemconsists of a powerful neutron source, a multiaxis staging platform to holdand manipulate an object, and an efficient imaging detector.

S&TR May 2001 11

Lawrence Livermore National Laboratory

Neutron Radiography

of uranium, even though we can’t reallyquantify the gap,” he says, adding, “we’reseeing more than we ever expected.”

A number of images have also beentaken of the British Test Object (BTO),on loan from the Atomic WeaponsEstablishment in Aldermaston, UnitedKingdom. The BTO consists of a set of six nested cylinders of graphite,polyethylene, aluminum, tungsten,polyethylene, and tungsten (respectively)with a solid polyethylene core. Twelve30-minute exposures were taken of oneside of the assembly and then reconstructedinto a mock tomographic image. Thereconstruction clearly shows the grossstructure of the object as well as thedetailed joint structure in the outer shells.

Despite the experimental successenjoyed thus far, much work remains to bedone to meet the goal of having a full-scaleneutron imaging system in operation atLivermore by late 2003 or early 2004.Vendors need to be selected to build theaccelerator, the detector’s optics system,and the multiaxis staging system.Meanwhile, plans are under way tomodify an existing Lawrence Livermorelaboratory to house the system.

Once the system’s performance isvalidated at Livermore, it will betransferred to other DOE facilities such

as the Pantex Plant in Texas or the Y-12Plant in Tennessee by late 2005 or early2006. The continuing success of theOhio University experiments makes it likely that neutron imaging will beserving the nation’s stockpile stewardshipneeds within a few short years.

—Arnie Heller

Key Words: Brookhaven NationalLaboratory, COG Monte Carlo radiationtransport code, deuterium, Enhanced

Surveillance Campaign, lithium deuteride, LosAlamos Neutron Science Center (LANSCE),Massachusetts Institute of Technology (MIT),neutron radiography and tomography, NevadaTest Site, Ohio University AcceleratorLaboratory (OUAL), Pantex Plant, scintillator,stockpile stewardship, x-ray imaging, x-rayradiography, Y-12 Plant.

For further information contact James Hall (925) 422-4468(jmhall@llnl.gov).

A number of imageshave been taken of(a) the British TestObject (BTO), whichconsists of (b) a set ofsix nested cylindricalshells made ofgraphite, polyethylene,aluminum, andtungsten. A series of exposures wasreconstructed intomock tomographicimages, which showthe BTO viewed (c) from the top and (d) through its side.

Tungsten GraphitePolyethelene

JAMES HALL received his B.S. in physics and mathematics fromthe University of Southern Colorado in 1974 and his M.S. andPh.D. in physics from Kansas State University in 1977 and 1981,respectively. He joined Lawrence Livermore in 1987 as a physicistcharged with the design and execution of a variety of nucleardevice diagnostic experiments, primarily neutron and gamma-raymeasurements, for the underground nuclear test program at the

Nevada Test Site. With the end of underground testing in 1992, Hall refocused hisefforts on the development of detailed computer simulations of inertial confinementfusion diagnostics, flash x-ray systems, and a variety of nonintrusive inspection systemsproposed for use in stockpile stewardship, cargo and luggage inspection, and nuclearcounterterrorism schemes. As an outgrowth of this work, in 1994 he was selected toserve as the DOE representative and chief science advisor to the 8th Joint Complianceand Inspection Commission meetings associated with the Strategic Arms ReductionTreaty. Hall is currently a principal investigator for the development of high-energyneutron imaging techniques in support of nuclear stockpile stewardship applications.

About the Scientist

(a) (b)

(c) (d)

Uncovering Hidden Defects with NeutronsNeutrons Complement X RaysSimulations Bolstered ConfidenceOhio University Test BedSandwiches, Steps, and CylindersFigure 1Figure 2Figure 3Figure 4Figure 5Figure 6Figure 7Box 1: A Neutron PrimerBox 2: The Making of a Neutron Imaging SystemKey WordsAbout the ScientistContact