The Development of Flash Radiography

76 Los Alamos Science Number 28 2003

The Development of

Flash Radiography

Gregory S. Cunningham and Christopher Morris

The Manhattan Project PHERMEX and DARHT

Proton Radiography at LANSCE

In many ways, flash radiography isto nuclear weapons what medicalx-radiography is to the human

body: It allows one to see inside acomplex structure without disturbingit. Starting in the Manhattan Projectand continuing to this day, flash radi-ography has been used to take stop-action pictures of dynamic events:from the detonation of high explosivesto the implosion of a mock weaponassembly containing a surrogate mate-rial for the nuclear core.

In this article, we will explain thebasic principles of flash radiographyand trace decades of progress inimproving image quality. A major goalis to follow the hydrodynamic implo-sion, or hydrotest, to the point atwhich the surrogate core is maximallycompressed. Los Alamos state-of-the-art x-ray hydrotests at the Dual-AxisRadiographic Hydrodynamic Test(DARHT) facility and facilities envi-sioned for the future will producetime-sequenced images of the implo-sion dynamics and provide views ofthe implosion along multiple lines ofsight. Images from hydrotests atDARHT are already playing a crucialrole in solving stockpile issues, includ-ing those related to the certification ofremanufactured parts. A recent LosAlamos invention, proton radiography,is now fielded at the Los AlamosNeutron Science Center (LANSCE),the Laboratory’s medium-energyaccelerator facility. Proton radiographyis also providing important data for thenuclear weapons program. A higher-energy proton radiography machinehas the potential for providing a newlevel of quantitative precision andquality to the data from hydrotests.

Historical Origins

Quite remarkably, the best analog toour current experimental program in sci-ence-based stockpile stewardship isfound in the program to develop the

plutonium implosion bomb during theManhattan Project. It was known fromthe start that enough fissionable materialfor a single bomb, either uranium-235or plutonium-239, would not becomeavailable until late in the project.Consequently, the gun device—whichpropels one subcritical piece of fission-able material into another at highspeed—was the favored method forassembling a supercritical mass.Because it was straightforward, thisapproach had the highest probability ofsuccess, whereas the spherical implo-sion of a subcritical configuration wouldpresent major technical challenges.

At that time, the nuclear propertiesof plutonium-239, the new manmadeisotope, had been only crudely deter-mined. When a barely visible speck ofplutonium produced at Ernest O.Lawrence’s cyclotron at the BerkeleyRadiation Laboratory arrived at LosAlamos, scientists from the PhysicsDivision used the material to measuremore definitively the neutron numberper fission and the cross sections for fis-sion neutron capture and scattering.These were the first nuclear experimentscompleted at Los Alamos, and theresults were encouraging. Because theneutron number for plutonium wasindeed higher than that for uranium-235,plutonium would likely yield a moreefficient nuclear explosion.

These measurements were fed intocomputational bomb design models assoon as they became available. Later,however, when reactor-produced pluto-nium arrived, the scientists detected ahigh neutron background, which, EnricoFermi quickly showed, derived from thespontaneous fission of the isotope pluto-nium-240, a reactor byproduct present inthe sample. Gun assembly of plutoniumpieces containing plutonium-240 wouldbe too slow to prevent premature initia-tion of the chain reaction by the neutronsfrom spontaneous fission, and thereforethe likely outcome would be a fizzlerather than an efficient nuclear explo-sion. This finding forced the project to

switch goals and aim for an implosiondevice.

In the implosion device, high explo-sives surrounding a spherical assemblywould be detonated at many points,and the resulting converging sphericaldetonation wave would compress thenuclear material to a supercritical con-figuration. One had to measure thevelocity of the implosion and the stateof the metal during assembly in orderto determine the optimal timing of theneutron initiators needed to achievesuccessful device performance.Director J. Robert Oppenheimer calledthis endeavor “one of the most urgentof the project’s outstanding problems.”Its solution occupied a talented groupof physicists, chemists, and electricaland mechanical engineers. Details ofthe dynamic response of materials sub-jected to high-explosive drive werestudied in small-scale experiments inGMX Division (predecessor of thepresent Dynamic ExperimentationDivision). To test the overall perform-ance of the implosion device, theimplosion group performed so-called“integral” experiments on mock assem-blies, which had the correct geometryand components except for a nonfis-sionable, surrogate core in place of theplutonium pit.

As we discuss below, radiographywith x-rays was a key diagnostic forthose small-scale and integral tests andhas remained so through the decades.In fact, many of the experimental toolsdeveloped in the 1940s to study theevolution of high-explosive-drivensystems—electrically charged metalpins, optical framing cameras, andflash radiography—are still the stan-dard diagnostics for monitoringweapons implosion. Today, throughthe use of vastly improved equipmentand modern data acquisition and analy-sis methods, these tools are still help-ing us quantify important phenomena(such as the details of material failureand high-explosive detonation) anddevelop accurate, predictive physical

Number 28 2003 Los Alamos Science 77

Flash Radiography

models to describe them. Before wetrace the history of flash radiographyfor studying weapon implosion, weexplain the basic principles of thistechnique, as well as some limitationsthat we hope to circumvent withadvanced techniques.

Attenuation Radiography forStockpile Stewardship

In modern hydrotests, we replacethe fissionable pit of a nuclear primarywith a mock pit made of a surrogatematerial such as natural uranium, lead,or tantalum. This nonnuclear system isimploded and its dynamics studied toprovide constraints for the physicsmodels used in numerical simulationsof weapons performance. The primarydiagnostic of hydrotest experiments ispoint-projection flash radiography. Atthe times of most interest, the experi-ment is illuminated with a short pulseof x-rays, and the transmitted flux(number of x-ray photons per unitarea) is recorded on a suitably shieldeddetector.

Figure 1 is a diagram showing thegeometry of a static experiment on theFrench test object (FTO), which wasdesigned to allow French and U.S.experimenters to collaborate on flashradiography methods and analysis.High-energy x-rays are producedthrough the bremsstrahlung interactionof energetic—10 to 30 millionelectron volts (MeV)—electron beamswith high-Z targets (that is, targetsmade of materials with high atomicnumbers). Interaction with a positivelycharged nucleus causes an electron tobend (accelerate) and therefore radiate,or emit, photons. The loss of energybrakes its speed, hence the termbremsstrahlung (or braking radiationin German) for both the process andthe emitted radiation. Although theemitted photons have a continuousenergy spectrum, most of the photonsthat are transmitted through a

hydrotest assembly have an energynear 4 MeV. That energy is near theminimum in the absorption cross sec-tion for the materials in the assembly.

A crucial performance parameterthat we would like to be able to meas-ure with flash radiography is the den-sity of the surrogate material atnuclear time—the time a “real” systemwould start producing a significantamount of nuclear energy. As we willdiscuss below, by measuring the atten-uation of the x-ray flux transmittedthrough the center of the assembly, wecan determine the integrated quantityρΑ, the areal density, or line-of-sightmass, of the implosion system:

(1)

where ρ(x, y, z) is the volume densityof the hydrotest object and z is thelongitudinal coordinate through theassembly. For an object with constantdensity, ρ0, the areal density is ρ0 × L,where L is the longitudinal thicknessof the object.

Because the areal density of the sys-tem at nuclear times is very large, weneed an enormous dose (that is, inten-sity × time) of x-rays to make thetransmission measurement. The dosescurrently available, even from the latest

ρ ρAL

x y z dz= ( )∫ , ,0

,


Flash Radiography

Foam

Copper

Electron

High-Ztarget

Roughcolimator

Gradedcolimator

Knock-offelectron

Detector

X-rays

BlurringfromX-raysource

Tungsten

High-Z nucleus (+)

BremsstrahlungX-ray production

Electron(–)

Electronpath X-ray

photon

Figure 1. Basics of Flash X-RadiographyThis schematic shows (from left to right) the production of x-ray pulses, their trans-mission through the so-called French test object (FTO), and their detection. At left,high-energy electrons (red) hit a high-Z target and interact with the heavy nuclei toproduce x-rays (blue). The x-rays are attenuated by a rough collimator, whichdefines the field of view and then by a graded collimator that flattens the transmis-sion profile to reduce scattering into the center of the image. The FTO consists ofan inner spherical shell of tungsten (inner radius = 1 cm and outer radius = 4.5 cm)and an outer shell of copper (outer radius = 6.5 cm) surrounded by a shell of foam(outer radius = 22.5 cm). The location of material interfaces recorded at the detectorare blurred because the x-ray source has a finite extent and because electronsknocked into motion by arriving x-rays have a finite range in the detector.

flash x-ray machines, are insufficient toprovide the quality of data thatweapons scientists will require to ade-quately constrain their calculations forfuture certification. In pursuit of betterdata quality, flash x-ray machines arebeing constantly upgraded andimproved; at the same time, new dataanalysis technologies are being devel-oped and implemented at Los Alamosand other weapons laboratories.

The most basic physics of transmis-sion radiography is contained in theBeer-Lambert law, the solution to thedifferential equation that describes thenumber of particles surviving transportthrough a medium without interaction.The Beer-Lambert law can be derivedfrom investigating transmissionthrough an infinitesimally thin piece ofmaterial with constant density ρ0 andthickness l. In this case, the probabilitythat a particle goes through with nointeraction is (1 – σρ0l/A), where σ isthe cross section for interaction in cen-timeters squared and A is the atomicmass in grams. For a material withfinite thickness L, the probability of nointeraction is

(2)

where λ = A/σ is called the interactionlength in grams per centimeter squared(gm/cm2). Thus, if N0 photonsimpinge on the material, then, on aver-age, the number N that will make itthrough without interaction is given by

(3)

This equation can be generalized tomaterials with varying density as

(4)

where ρA is the areal density defined

in Equation (1).Equation (4) tells us that we can

determine the areal density of theobject in units of the interactionlength λ of the incident radiation bymeasuring the ratio of incident to sur-viving particles:

.(5)

This simple analysis leaves outimportant details. For example, λdepends on energy, and the x-raysource is not monoenergetic. Also,scattered x-rays produce background“fog” in the image. Nevertheless, thissimple analysis provides an importantguide for evaluating and developingradiographic tools.

For example, we can calculate theuncertainty in the measured value ofareal density, ∆ρA, under the assump-tion that the only source of noise isthe Poisson (counting) statistics of thetransmitted beam. That uncertainty isgiven by

(6)

The optimal interaction lengthλoptimal would be one that minimizes∆ρA for a given object. Setting to zerothe derivative of the uncertainty withrespect to λ and solving for λ, we findthat λoptimal = ρA/2, or the optimumequals half the thickness (or areal den-sity) of the object. For the FTO, ρA =182.5 gm/cm2, so the optimum inter-action length is 91.75 gm/cm2. Canwe achieve such a long interactionlength?

In the case of x-rays, λ variesstrongly with x-ray energy. The inter-action length reaches a maximumvalue at the energy at which the crosssection for producing electron-positron pairs (that cross sectionincreases with increasing energy)becomes comparable to Compton scat-tering (which decreases with increas-

ing energy). The maximum interactionlength of x-rays is weakly dependenton atomic number, Z, and thereforefor all high-Z materials, λ is maxi-mum at about the same x-ray energy,namely, near 4 MeV. The interactionlength λ in uranium for 4-MeV x-raysis about 22 gm/cm2, or a little over acentimeter in natural uranium, muchsmaller than the thickness of ahydrotest assembly. The relativelyshort interaction length of x-rays inheavy elements implies a large uncer-tainty in the areal density measure-ments, even when extremely highdoses are used.

The First Radiographs of Implosion

Let us now go back to the origin ofweapons radiography. In the spring of1944, the Manhattan Project shifted itsmain focus to implosion after the dis-covery that reactor-produced plutoniumhad a high neutron background that wasdue to the presence of plutonium-240.The previous fall, John von Neumannhad suggested that, with enough highexplosive driving an implosion of a fis-sionable metal core, one could ignorethe strength of the material and assumethat the solid material behaved like afluid. In this case, partial differentialequations could be written and solved ina numerical program on IBM machinesto determine the velocity of implosion.However, the input to the equations, thatis, the high-explosive drive and theequation of state (EOS) of the metal andexplosives needed to be determined.Flash radiography was one of theimportant diagnostic techniques used inquantifying the spherically converginghigh-explosive drive.

In the initial experiments conductedin 1943, Seth Neddermeyer’s grouptried surrounding a small metal spherewith weak explosive charges and deto-nating it at many points. The scientistsexpected the diverging spherical waves

∆ρλ

A

N= 1 .

ρλA N

N= −

ln

0

.

N N

A

e=−

0

ρλ ,

N N

L

e=−

0

0ρλ .

limn

nL

L

ne

→∞

−−

=1 00

ρλ

ρλ

,


Flash Radiography

from each detonation point to canceleach other out upon interaction, creat-ing the desired converging detonationwave. The results were disappointing.The recovered ball of metal had beencompressed but showed many asym-

metries. Later experiments with multi-ple points of detonation around a cylin-drical shell showed that high pressuredevelops where the detonation wavescollide, which could result in the for-mation of “jets.” The group then modi-

fied commercial x-ray machines toachieve precision timing of the x-rayflashes to about 1 microsecond so thatthe x-ray flashes could be coordinatedwith the explosive shots. Indeed, theresulting images confirmed that jets didform at the interaction between multi-ple detonation waves. That diagnosisled to the design of explosive “lenses,”which shape the individual detonationwaves so that such high-pressure areasdo not form. Flash radiography ofsmall imploding metallic spheresrevealed two other reasons for nonidealimplosion: density variations in theexplosives and asynchrony among theindividual detonators. These discover-ies led to improvements in the manu-facture of explosives and to “electricdetonation” for more reliable timing.

To interrogate implosion of a full-scale device (with a surrogate pitmaterial), they would need more pen-etrating radiation—x-ray photons withenergies near 4.0 MeV. Oppenheimerdecided to acquire the University ofIllinois betatron, an electron accelera-tor that produced 1-microsecond-longpulses of 15-MeV electrons. Asalready discussed, the high-energyelectrons produced bremsstrahlungradiation as they passed through ahigh-Z target. The high-energy pho-tons produced by the betatron pene-trated the high explosive of afull-scale device but were stopped bythe large areal density of the pit itself;therefore, a “shadow” of the pit’souter contour could be observed bydetection of the photons that made itthrough the device. These photonswere detected when a sheet of leadglass was placed on the other side ofthe device. Interactions of x-rays withthe atomic electrons in the lead glassproduced energetic recoil electrons,and those recoil electrons made visi-ble tracks in a vertical cloud chamber.That system provided the first flashradiograph of an “integrated” test (seeFigure 2). The work on flash x-rayimaging of full-scale devices was said


Flash Radiography

Figure 2. Radiographs of an Explosively Driven Implosion ExperimentThese radiographs were taken in 1944 to image the outside edge of the surrogatepit during a hydrotest. The x-rays used to image the implosion were generated withthe 15-MeV electron beam from the betatron borrowed from the University of Illinois.The detector consisted of a lead glass converter and a Wilson cloud chamber todetect the recoil electrons. The dark area in each image is the shadow cast by thepit. The radius of the pit is smaller in the right image.

Figure 3. The Radiolanthanum Experiments This photo taken at Los Alamos during the Manhattan Project shows the “remote han-dling” of a kilocurie source of radiolanthanum located inside a lead container. Thisstrong gamma-ray source would be placed at the center of a hydrotest assembly tomeasure the areal density of the pit (from the center outward) as a function of time.

to be “among the most impressive ofseveral such achievements at LosAlamos” (Hawkins 1961).

After developing techniques to diag-nose symmetric implosions, theManhattan project pioneers wanted tomeasure what was happening insidethe pit. For that purpose, they placed asmall capsule of the radioactive isotopelanthanum-140 at the center of the pit.As the high explosive compressed thepit, the 1.46-MeV photons from thedecay of the lanthanum-140 penetratedthrough to the outside of the pit, andtheir intensity was measured as a func-tion of time. Those data provided ameasure of the areal density of the pitas a function of time. Although theexperiments were effective, the envi-ronmental hazards (see Figure 3) ofboth their production and their after-math resulted in abandonment of theprogram in 1962.

PHERMEX

The Los Alamos facility known asPHERMEX (for pulsed high-energyradiographic machine emitting x-rays)was commissioned in 1963 (seeFigure 4). It was the first of a new gener-ation of flash x-ray machines designedto produce enough flux to penetrate thecenter of a hydrotest experiment at“nuclear” time. The design of this high-energy pulsed x-ray machine, includingtechniques for recording the images, wasthe culmination of extensive Los Alamosstudies led by Doug Venable and com-pleted in the early 1950s.

Those studies defined the doseneeded to penetrate the center of ahydrotest assembly and the feasibilityof getting good images of the less-dense parts of the assembly. It wasshown that systems up to 4 λ in thick-ness could be imaged on film detectors.For objects up to 10 λ in thickness, thelarge background of scattered radiationwould obscure the film images.

The studies were performed on static

test objects that had been stretched inthe beam direction to have areal densi-ties commensurate with the high vol-ume densities reached during hydrotestimplosions. The doses needed to meas-ure the internal densities of those testobjects were determined as a functionof scale. The idea was that, althoughPHERMEX could not see through afull-scale device, the hydrodynamicscould be studied at quarter or half scalewith surrogate materials. To minimizethe obscuring effects of scattered x-raysin the thickest regions, the objects wereradiographed through a graded collima-tor designed to be an approximateinverse of the object. Graded collima-tion dramatically reduced the scatteredbackground while still allowing thethinner parts of the object to be seen.This technique has been crucial inenabling radiography across the fullrange of configurations reached duringhydrotests.

At PHERMEX, three very large50-megahertz radio-frequency (rf) res-

onators provide the energy needed toaccelerate short pulses of electronstotaling 9 microcoulombs of electriccharge to 30 MeV. Those pulses arethen directed at a high-Z target to pro-duce x-rays. When PHERMEX wascommissioned, it produced200-nanosecond-long pulses of x-raysexceeding 9 roentgens at 1 meter fromthe production target.

Thousands of experiments havebeen performed at the facility, includ-ing small-scale experiments to developthe physics of high-explosive-drivensystems (see Figures 5 and 6), and alarge number of major hydrotests.Much of this work has been compiledand presented in a marvelous summaryof shock physics (Mader et al. 1980).PHERMEX has also undergone manyupgrades during its lifetime. Currently,it can produce a single 200-nanosecond-long pulse with a spot size of 3 millime-ters and a dose of 400 roentgens at adistance of 1 meter from the target,more than 40 times the dose provided


Flash Radiography

Figure 4. A PHERMEX ShotThis photo shows an explosive shot at PHERMEX. The high-energy pulsed x-raymachine has served the weapons program for 40 years.

when it was commissioned.For most of PHERMEX’s history,

a technique known as screen-enhanced film has been used to recordimages of the experiments performedin front of the facility. That is, thetransmitted x-rays are convertedthrough Compton scattering into mov-ing electrons in a millimeter-thicksheet of lead. Then, as the electronsslow down in the sensitive photo-graphic film located behind the leadsheet, their tracks are recorded (seeFigure 1). Recently, active cameras

have been developed and fielded.They have higher sensitivity, higherquantum efficiency, and wider lineardynamic range than film. The latestversion of these active cameras cantake up to four sequential images andthereby tap the two-pulse capability ofPHERMEX. For the first time, we canget high-quality images for each indi-vidual pulse (See the article “TheDARHT Camera” on page 92.)

Ultimately, the amount of charge(number of electrons) delivered byPHERMEX, and in turn, the x-ray

dose, are limited by the stored energyin the rf resonators. The voltage in theresonators decreases as energy istransferred to the beam; the resultingspread in beam energy at high elec-tron current leads to an unacceptablylarge spot size on the x-ray productiontarget. Also, high-explosive-drivenexperiments generate material veloci-ties in the range of several millimetersper microsecond (mm/µs). Thus, thematerial moves appreciably during thelength of the typical 200-nanosecond-long x-ray pulse. Reducing this“motion blur” to levels that do notinterfere with the interpretation of theexperiment requires pulse lengths ofabout 100 nanoseconds or less. Inspite of the limitations in dose andpulse length, PHERMEX has been aworkhorse for the weapons programduring the 40 years of its operation.

The DARHT Facility

A recent review (Ekdahl 2002) ofthe current state of electron accelera-tors for flash radiography reports thatthe United States, the UnitedKingdom, and France are all develop-ing new flash-radiography capabilitiesto meet the challenges of maintainingnuclear weapons stockpiles under therestrictions of a moratorium on nucleartests. This new generation of machinesis designed to provide higher doses,better position resolution, shorterpulse lengths, and in some cases, datafrom a single experiment taken atmultiple times and along multipleaxes, so that the time-dependence andthree-dimensional (3-D) aspects of animplosion can be elucidated.

As early as 1968, Doug Venablehad proposed adding a second x-rayaxis to PHERMEX, perpendicular tothe first, to allow orthogonal tomogra-phy. And in 1981, long before themoratorium on testing, John Hopsonand Tim Neal presented the first con-cept for the DARHT facility at Los


Flash Radiography

Figure 5. PHERMEX Radiograph of Hydrodynamic Flow Dating from the late 1960s, this radiograph shows colliding shock waves in alu-minum. The horizontal lines are thin foils of high-density metal interspersed in thealuminum to indicate the material flow. A Mach stem has formed at the intersectionof the shock waves.

Figure 6. Double-Pulse PHERMEX Radiograph of Spall in IronAn iron plate is driven by an explosive initiated at a number of individual points.Theiron has spalled into a series of layers, and the effects of the initiation points areapparent. Successive transmission of two short x-ray pulses produces two sequentialimages on one film.

Incident shocks

Reflected shocks

Mach stem

Alamos. It would be built at a new fir-ing site and use two high-dose pulsed-power machines to provide orthogonalx-ray views of a single hydrotest, acapability similar to that at the AtomicWeapons Establishment (AWE) inAldermaston, England.

The linear induction accelerator(LIA), rather than the pulsed-powerdiode machines originally proposed, isthe technology being used for the twoflash-radiography machines atDARHT (see Figure 7). LIAs werepioneered for flash radiography atLawrence Livermore NationalLaboratory in the 1980s. Their operat-ing principle is similar to the betatrontechnology used for the first high-energy flash radiography in LosAlamos in the sense that stored energyfrom a pulsed-power system is induc-tively coupled to a high-current elec-tron beam. But in the LIA, the cou-pling is accomplished by a row ofmany induction cells rather than a sin-gle transformer. Because there aremany transformers, each coupled toits own external energy-storagedevice, the electron currents can bemuch larger than the limiting currentsat PHERMEX. In fact, kiloampereelectron currents can be readily accel-erated in an LIA.

One performance feature to helprank machines that produce differentx-ray spot sizes and doses is the root-square-mean (rms) error with whichthe radiographs from each machinecan be used to determine the positionof a material interface. This rms erroris inversely proportional to the squareroot of the dose d and proportional tothe radiographic position resolution∆x. Charlie Martin of AWE proposedthe simple radiographic figure of meritFOM = d/∆x2. The position resolutionincludes contribution from the x-rayspot size, the pulse length (which pro-duces motion blur), and the detectorresolution. The machine design deter-mines the first two of these.

The first axis of DARHT, which hasalready been used for hydrotests, hasdelivered a dose of 500 roentgens in a60-nanosecond-long pulse over a2-milllimeter spot. For the same pulselength, PHERMEX can provide a doseof only 120 roentgens in a spot of3 millimeters. These performanceparameters indicate that DARHTachieves about an order-of-magnitudeimprovement over PHERMEX in termsof Martin’s radiographic figure ofmerit. The second axis of DARHTaccelerated its first beam to full energyat the end of 2002 and will soon pro-

vide pulses with dose and spot sizesimilar to those produced by the firstaxis. The new feature of this secondaxis is the production of four suchpulses in 2 microseconds. That capabil-ity should be available to interrogatehydrotests by the end of 2005.

The first axis of DARHT is alreadyproviding weapons scientists with theclearest views ever seen of the insideof a hydrotest. The hydrodynamic datafrom those tests are used to validatenew physics models that are beingincorporated into weapons codes.Once the second axis becomes avail-able, scientists will take four sequen-tial radiographs along one axis andone radiograph along the perpendicu-lar axis, thus providing the very first3-D data from a single U.S. hydrotest.

Figure 8 compares radiographs ofthe unclassified FTO (refer to Figure 1)taken at PHERMEX in the late 1980sand at the first axis of DARHT. TheFTO was designed to compare the per-formance of flash x-ray machines. TheDARHT system—the combination ofx-ray source and detectors—clearlydemonstrates a very dramatic increasein performance. This facility is expectedto be the centerpiece of the nation’shydrotest program for at least a decade.


Flash Radiography

Figure 7. The DARHT Facility(a) The DARHT facility houses two linear induction accelerators set at right angles to each other and focused on a single firingpoint. (b) Each accelerator consists of a row of induction cells, each coupled to its own energy-storage device. The pulsed-powermachine accelerates very large (kiloampere) electron currents, which produce very intense x-ray pulses that are 60 ns long.

(a) (b)

Detectors, Collimators,and Data Analysis

The x-ray dose and spot size pro-duced in state-of-the-art multipulse x-raymachines are limited by interactionsbetween the electron beam and the high-Z target. When an electron beam carry-ing thousands of amperes interacts withan x-ray production target, it creates ahigh-density plasma of ionized targetmaterial and surface impurities. Theelectrons then interact with the plasma,dynamically changing the effective focalpoint and increasing the spot size. Inaddition, the beam causes material at thesurface of the target to spall, whichreduces the target thickness and, in turn,reduces the dose as a function of time.Both the destruction of the target and thebeam-plasma interactions make the goalof multipulse high-dose x-ray radiogra-phy difficult to attain. Some progress isbeing made in techniques to mitigatethese problems, but so far, improve-ments have been only incremental.

Difficulties inherent in increasingthe dose have led researchers to searchfor optimal ways to extract the maxi-mum information from the availabledoses. They have adopted and extendedtechniques first investigated at PHER-MEX: graded collimation, advanceddata analysis, image plate detectors,and multiframe active cameras. Moresensitive detectors allow measurementsto be made with less incident dose.Large-scale Monte Carlo calculationshave led to improved experimentaldesigns that increase signal-to-noiseratio by reducing scattered background.New data-analysis techniques have ledto optimally estimating features ofinterest (refer to Figure 9). Theseadvances in experimental design, detec-tors, and data analysis have been asimportant as the increased power andresolution of the x-ray beams forincreasing the information thatweapons scientists obtain from flashradiography.


Flash Radiography

Figure 9. Improvements in Data Analysis (a) The radiograph of a steel cylinder with a diameter of 12 cm was taken with acobalt-60 source. The cylinder has a conical section machined out of the top. Squaregrooves (2 mm by 2 mm) were machined at the bottom of the cylinder and on theinner surface of the conical section to assess the quality of the radiograph and sub-sequent image processing. (b) The reconstruction process, which first extracts theareal density and eventually the volume density of the cylinder, makes readily appar-ent many of the grooves that were almost invisible in the original radiograph. Theenhanced noise at the center of the cylinder is an unavoidable consequence of thisprocess. The reconstruction process was first implemented in the 1980s.

Figure 8. A Comparison of X-Radiographs from DARHT and PHERMEXA photograph of half of the FTO is shown in (a), a radiograph of the FTO from PHER-MEX is shown in (b), and another radiograph of the FTO from DARHT is shown in (c).The DARHT image reveals a dramatic improvement in quality caused by an improvedsource and better detector. The PHERMEX radiograph was taken with a graded colli-mator, whereas the DARHT radiograph was taken with a small, field-of-view, rough col-limator, imaging less of the object with lower background. The boxes outline theapproximate field of view of each of the radiographs. Comparing these radiographswith the early radiograph in Figure 2 reveals the progress made in flash x-ray radiog-raphy since the Manhattan Project .

(a)

(b)

(a) (b)

(c)

Proton Radiography

In spite of numerous improvementsin high-energy flash radiography overthe past 50 years, the dose limitations,position resolution, and backgroundsstill limit the utility of the technologyfor obtaining adequate quantitativeinformation from hydrotests for stock-pile certification. Recently, a newidea, lens-focused proton radiography,has provided a potential solution tothese problems.

As described earlier, electromagnet-ic scattering processes limit the maxi-mum interaction length of an x-ray toa value far from the optimum forhydrotest experiments. Hadronicprobes provide an alternative. Hadronsare fundamental particles, such as neu-trons and protons, that interact withmatter through the strong (nuclear)force. The absorption cross section,σA, for the strong interaction ofhadrons with a nucleus with massnumber A can be approximated as

(7)

which is the geometric cross section ofthe nucleus, where rA ≈ 1.3A1/3 fem-tometers. (A Fermi, or femtometer, is10–15 meter.) This absorption crosssection implies that the mean freepath, λ* = 1/nσ (where n is the num-ber density of atoms), for a hadron inuranium of nominal density is a lengthof about 10 centimeters, or an interac-tion length of 200 gm/cm2, an order ofmagnitude larger than that of high-energy x-rays. This interaction lengthis almost perfectly matched tohydrotest radiography. Consequently, amuch lower incident flux of hadronswill produce the same statistical infor-mation now obtained from a higherflux of high-energy x-rays.

Of course, protons are charged, andtherefore, when they interact with mat-ter, the Coulomb force between theprotons and the charge of the electronsand the nuclei in the material causes

the protons to continuously slow downand scatter into other directions.However, for a proton with highenough energy, electromagnetic scat-tering processes produce only smallchanges in its direction and energy,even when it traverses a significantthickness of material. Thus, nuclearinelastic scattering remains the domi-nant mechanism removing protonsfrom an incident high-energy protonbeam. Consequently, high-energy pro-tons have a large interaction lengthand are interesting as a radiographicprobe. Moreover, current acceleratorsroutinely produce high-intensity, shortpulses of high-energy protons.

Focusing Protons. The charge onthe proton allows using magnetic lensesto focus a proton beam of a selectedenergy or momentum (the two arealmost equivalent at high energies, hun-dreds of times above the rest massenergy of the proton, which is 938.272MeV). This focusing capability givesgreat flexibility to proton radiographyand leads to many advantages over thestandard point-source x-radiographyshown in Figure 1.

Figure 10 shows the proton radiog-raphy line at LANSCE, which has fourseparate lenses: an angle matchinglens (not shown), consisting of threequadrupole magnets, and three imag-ing lenses, each consisting of fourquadrupole magnets. The initial beamgoes through the angle-matching lens,which is tuned to focus the protons ofa selected initial momentum Pi ontoimage plane 0 such that the protons(black rays) are spread over an areaequal to that of the object and eachproton’s directional angles relative tothe beam axis along the z-direction (θiin the x-z plane and φi in the y-z plane)are proportional to its distance fromthe beam axis, that is, θi = Axi and φi= –Ayi. To calibrate our experiment,we measure the beam intensity atselected points on image plane 0. Lens0 then refocuses the protons to have

the same position–angle correlation atthe object plane that they have atimage plane 0, just inverted fromplane 0 to the object plane. Protonstransmitted through the object arescattered by Coulomb forces into acone of angles about their initialdirections (represented by the red andblue rays). Even though the transmit-ted protons now have a spread inmomentum, lens 1 can refocus thebeam onto image plane 1 because thelens is designed to cancel the leadingchromatic aberrations (the changes inimage position caused by variations inmomentum). A collimator in lens 1,like the f-stop of a conventional cam-era, selects the range of proton anglesthat can be transmitted to image plane1. The contrast of the image can beincreased by selection of a collimatorthat cuts into the Coulomb scatteringcone.

Correcting Chromatic Aberrations.The largest aberration in the lens systemis chromatic, or momentum dependent.That is, protons whose momentumvaries from that for which the magneticlens is tuned are the leading cause ofimage blurring. The angle-matchinglens has been designed so that the pro-ton trajectories incident at the objectplane have a position–angle correlationthat minimizes the chromatic aberrationin the imaging lenses. In particular, thatposition–angle correlation at the objectplane is such that the largest chromaticaberrations in image position canceleach other—the aberration proportionalto θi cancels the one proportional to xi,and the aberration proportional to φicancels the one proportional to yi.

High-Efficiency Detection.Protons are detected with high effi-ciency when a thin sheet of scintilla-tor is placed at the image plane. Asthe protons pass through, the scintilla-tor emits enough light for an image tobe stored in a gated charge-coupleddevice camera, but it does not perturb

σ πA Ar= 2,


Flash Radiography


Flash Radiography

Lens 2

Lens 3

Lens 2Lens 1Lens 0

Lens 1

Collimator 1Collimator 0 Object plane

Calibrationplane

Collimator 2cuts into

Coulomb cone

Image plane 1

Image plane 2

Coulomb cone

(Mat

chin

g le

ns)

Electromagnetic

Collimator

Seven-camerasystem

Containmentvessel

x

zθ

φyBeam direction

(d)

(a)

(b)

(c)

Figure 10.The Proton Radiography “Microscope” in Line C at LANSCE(a) The containment vessel and imaging lenses are shown. (b) A schematic ofthree of the four magnetic lenses traces the paths of 800-MeV/c protons from thecalibration plane through the object plane to the two image planes. The matching

lens (not shown) produces a specific correlation between the positions (xiand yi) and angles (θi and φi) of the protons incident on the calibration

plane, measured relative to the beam axis along z. The imaginglenses 1 and 2 can refocus protons that have lost some energy

through Coulomb interactions with the object, providedthose protons have the correct angle–position correla-

tions at the object plane. Note that collimator 2cuts farther into the Coulomb cone than

does collimator 1. (c) A 3-D drawing ofthe system includes the contain-

ment vessel and the camerasystem in line C. (d) This

photo shows the cam-era system.

the proton beam significantly. In fact,so few protons are absorbed that thesame pulse can be reimaged by lens 3.Lens 3 has a smaller collimator thanlens 2 to cut farther into the Coulombscattering cone (note that the bluerays are blocked). Since Coulombscattering depends on the Z-numberof the material, this double imagingprocess enables material identificationfrom proton radiographs.

The ability to refocus protons meansthat the closest detector (image plane 1)can be at a long standoff distance fromthe blast. That characteristic, combinedwith the momentum selectivity of thelens, results in much lower back-grounds in proton radiography than inx-ray radiography. As a result, the needfor background mitigation techniquessuch as graded collimation, an essentialpart of x-ray experiments, is eliminated.The combination of lower backgroundsand energy-independent cross sectionsallows a new level of precision notavailable in x-radiography.

Proton radiography is being studiedwith the 800-MeV/c proton beam pro-vided by the LANSCE accelerator andwith a 24,000-MeV beam provided bythe alternating gradient synchrotron(AGS) at Brookhaven NationalLaboratory.

Studies at LANSCE

On average, about 40 small-scaledynamic experiments are being per-formed each year at the proton “micro-scope” in line C at LANSCE (refer toFigure 10). Each experiment (orphysics package) uses up to 10 poundsof high explosive. The physics pack-age is contained in a vessel shown inFigure 10(c). These experiments aredesigned to study high-explosive deto-nation and dynamic material failureand to perform small integral tests tovalidate the models used in weaponscodes. At the time that the first experi-ments were being planned, designed,

and fielded, we were also developingand demonstrating a suite of tech-niques to image dynamic events safely,rapidly, and reliably. We now havedetectors that can record 21 timeframes of a single experiment. Wehave also demonstrated effective con-tainment techniques, beam monitoring,the use of multiple proton lenses andimage planes to distinguish differentmaterials, and analysis techniques forextracting quantitative informationabout the material densities after thematerials have been shocked.


Flash Radiography

Figure 11. Detonation Waves TurningCorners—Experiment vs Simulation (a) The photograph shows two experimentsdesigned to study corner turning of detonationwaves in insensitive high explosives. The detonationtravels from the thin donor region to the thickeracceptor region. (b) A time sequence (1 to 8) of pro-ton radiographs of the experiment (left of centerline)and corresponding simulations (right of centerline)show the volume densities in the explosive as a det-onation wave progresses from the donor to theacceptor region. Calculations using the DSD modelcorrectly predict many features of the detonationbut do not predict the dead region.

Experiment Simulation

(8)

(7)

(6)

(5)

(4)

(3)

(2)

(1)

(b)

(a)

As an example, Figure 11 illus-trates the experimental setup, the data,and the detonation shock dynamics(DSD) simulations of a set of experi-ments to study the corner turning of adetonation front in the high explosivePBX 9502. This insensitive, plastic-bonded material has been incorporatedin some systems in the nuclearweapons stockpile to increase safetybecause it was specifically designed tobe hard to detonate. The downside isthat detonation waves do not propa-gate as well in PBX 9502 as they doin conventional high explosives. As aresult, when the geometry of theexplosive forces the detonation waveto turn a corner, dead zones (regionsthat do not detonate) appear.

During the experiment, a detonationwave is launched in a cylindrical stalkof PBX 9502 by a booster, and asequence of proton radiographs is takenas the detonation wave propagates intoa cylinder with a larger diameter.Figure 11(b) shows a frame-by-framecomparison of radiographs (left) andthe corresponding simulations (right)for that sequence. The proton radi-ographs show that a dead region (dark)of unburnt explosive in the shape of adoughnut remains after the detonationwave expands in the larger cylinder.This phenomenon is not yet captured inthe DSD model (the DSD model is pre-sented in the article “High ExplosivesPerformance” on page 96). The experi-mental data guide the development ofmodels aimed at better predictions ofthis phenomenon.

The 800-MeV/c proton beam atline C has also been used to radi-ograph materials as they break apart,or spall, under the influence of highstrain rates. Figure 12 shows protonradiographs of a half cylinder of tita-nium driven by the detonation of anembedded half cylinder of highexplosive. The rapid propagation ofthe high-pressure detonation wavedown the high-explosive half cylin-der produces very high strain rates in

the titanium. As a result, the metalexpands much more before failingthan it would if it had been slowlystretched. The cracks and voids thatdevelop as the metal expands arereadily apparent in the inset. Thenew models needed to describe thisbehavior are being developed, andproton radiography provides datathat help in this effort.


Flash Radiography

Figure 12. Studying the Dynamic Failure of aTitanium Half Cylinder(a) The experimental setup shows a half cylinder oftitanium around a half cylinder of high explosive,both contained in an aluminum tube. The protonbeam will be directed through the middle region ofthe cylinder, where the aluminum has been cut away.(b) A sequence of proton radiographs, timed inmicroseconds, shows a detonation wave in the highexplosive moving down the cylinder and driving ashock wave through the titanium. A blowup from thefirst frame shows the locations of the shock front,titanium, and the detonation front in the high explo-sive. A blowup from the last frame shows the frac-ture pattern that develops in the titanium underthese high strain rates.

Fracture pattern

at 20.76 µs

Detonation frontAluminum

tube

Titanium

(a) Experimental setup (b) Results timed in microseconds

Proton beam

High explosive

Titanium

High explosive

High-Energy ProtonRadiography

The experiments with the 800-MeV/cproton beam at LANSCE have beenimmensely valuable in demonstratingand developing proton radiography. Forradiographs of the much higher arealdensities involved in full-scalehydrotests, a higher-energy proton beamis required. We have conducted studiesusing high-energy proton beams thathave exactly the format (beam size andintensity) needed to perform flash radi-ography for full-scale hydrotests.

Our first higher-energy experimentsat AGS used secondary protons (whoseenergy is lower than that of the mainbeam), which are produced at very lowrates. Although the exposures lasted forseveral hours, these first radiographs ofthe FTO, made with a quadrupole lens,showed great potential for the tech-nique. In more recent experiments, weused a fast-extracted high-energy beam(30-nanosecond-long pulses) of up to1011 protons from the accelerator toradiograph various static test objectsand to develop techniques for quantita-tive analysis of dynamic experiments.

Figure 13 shows the dramaticimprovement in FTO radiographsobtained at the AGS. These experimentshave demonstrated low backgrounds,multiple views, good statistics, and quan-titative precision. We have also com-pared proton and x-ray radiography byradiographing the same thick, classifiedtest object with the first axis of DARHTand with the high-energy proton beam atAGS. The results of this classified exper-iment demonstrate the dramaticimprovement in the quality of radiogra-phy expected from this new probe.

Material identification has alsobeen demonstrated for static experi-ments at the AGS in experiment 933.Material identification would be valu-able in studying the properties ofmaterial interfaces in hydrotests, butmoving objects present some new dif-ficulties that must be studied. We are

developing Monte Carlo simulationcodes to study the properties of theentire beam line, including the effectsof test objects with complicatedgeometries.

Very recently, we conducted anotherseries of experiments at the AGS onstatic test objects. These experimentswere designed to allow assessing thequantitative accuracy of proton radiogra-phy for the study of criticality and mix,effects that are of the highest possibleimportance to stockpile stewardship.The test objects were carefully crafted toprovide unprecedented fidelity toweapon design calculations. Some ofthe test objects constituted a weaponimplosion “time-series,” reflecting veryprecisely the microsecond-timescalechanges in device configuration that thedesign calculations predict.

Another aim of the recent experi-ments was to demonstrate a capability

to study mix and other stewardship-relevant phenomena in dynamicexperiments at the AGS. Although therecent experiments used only statictest objects, they were designed topave the way for future experimentsthat would incorporate high explo-sives and be fielded in containmentvessels. Even though a far cry fromweapons hydrotests, these dynamicexperiments would both provideunique data on hydrodynamic per-formance needed for the stewardshipprogram and demonstrate that thistype of imaging is fully compatiblewith the technology and infrastructureneeded in the hydrotest regime—including, of course, processes andprocedures required for the protectionof the environment, personal safety,and health.

The success of proton radiographyhas led Los Alamos to propose a new


Flash Radiography

Figure 13. Proton Radiographs of the FTO The radiograph on the left was recorded in a 4-h exposure in a secondary beam withabout 109 protons with an energy of 10 GeV in energy.The radiograph on the right wasmade in a 40-ns exposure with about 2 × 1010 protons with an energy of 24 GeV.

facility for hydrodynamic testing. Thenew proton-radiography facility willbe used to make detailed quantitativemovies of hydrotests that capture withunprecedented precision the timedevelopment of an implosion.

Summary

Penetrating flash radiography hasprovided critical information toweapons designers since the incep-tion of the Manhattan Project.Radiographic machines used duringthat time provided images of theouter pit surface to calibrate numeri-cal models of the hydrodynamic per-formance of the device. Radiographywas also used to identify severalimportant early problems with theimplosion device—for example,interaction of the high-explosivewaves that caused jetting of theheavy metal. PHERMEX, commis-sioned nearly 40 years ago, providedthe first radiographic machine capa-ble of obtaining detailed data on thedensity distributions at the center of aprimary in a radiographic hydrotest.In the intervening 40 years, there hasbeen tremendous progress in x-raymachine and detector performance,scatter reduction, and quantitativeanalysis of flash x-ray radiographyfor stockpile stewardship. The secondaxis of DARHT, soon to be commis-sioned, will complete a state-of-the-art facility that will provide weaponsdesigners with their clearest views ofthe inside of a hydrotest everobtained.

Both PHERMEX and DARHTrepresented quantum leaps forward inour ability to peer inside implosionsof mock nuclear weapon systems.However, the ultimate goal to pro-duce highly quantitative, 3-D, time-evolving density maps, needed forcertification without testing, has stillnot been reached. In the future, it islikely that new radiographic machines

with improved performance will beneeded to certify the enduring stock-pile. The recent invention of protonradiography at Los Alamos has thepotential to meet this future need.Experiments have shown that protonradiography can provide high-qualityradiographic information at manytimes during dynamic experimentsusing the 800-MeV/c proton beamfrom LANSCE. Experiments per-formed with the higher energy24,000-MeV/c proton beam from theAGS accelerator have demonstratedlow backgrounds, small statisticalerrors, and well-controlled systematicuncertainties. The combination ofhigh-quality radiography, small-scaleexperimentation, and predictive mod-eling can form the foundation of arobust stockpile stewardship programwithout underground testing.Interestingly, more primitive ver-sions of the same tools formed thedesign basis for the successfulTrinity test in 1945. �

Acknowledgments

We are grateful for the time DougVenable, John Hopson, and Tim Nealspent with us describing the history ofPHERMEX and DARHT. SusanSeestrom’s early editing was a greathelp in allowing us to get to a finalmanuscript. The Reports Library,Copy Center, and photo archive wereinvaluable resources for delving intothe early history of flash radiographyat the Laboratory. We would also liketo thank Tim Neal, John Zumbro,Stephen Sterbenz, and Ken Hansonfor contributing materials and provid-ing helpful advice.

Further Reading

Anthouard, P., J. Bardy, C. Bonnafond, P.Delsart, A. Devin, P. Eyharts et al. 1998.AIRIX Prototype Technological Results atCESTA. In Proceedings of the 1997 ParticleAccelerator Conference. (Vancouver, B. C.,Canada, May 12–16, 1997). Vol. 1, p. 1254.Edited by M. Comyn, M. K. Craddock, M.Reiserr, and J. Thomson. Piscataway, NewJersey: IEEE.

Alrick, K. R., K. L. Buescher, D. J. Clark, C. J.Espinoza, J. J. Gomez, N. T. Gray et al.2000. “Some Preliminary Results FromExperiment 933.” Los Alamos NationalLaboratory document LA-UR-00-4796.

Beer, A. 1852. Bestimmung der Absorption desRothen Lichts in Farbigen F1üssigkeiten.Ann. Phys. Chem. 86: 78.

Boyd, T. J. 1967. “PHERMEX: A PulsedHigh-Energy Radiographic MachineEmitting X-Rays.” Los Alamos ScientificLaboratory report LA-3241.

Burns, M. J., B. E. Carlston, T. J. T. Kwan, D.C. Moir, D. S. Prono, S. A. Watson et al.1999. DARHT Accelerators Update andPlans for Initial Operation. In Proceedingsof the IEEE Particle AcceleratorConference. (The 18th Biennial ParticleAccelerator Conference, New York, March27–April 2, 1999). Vol. 1, p. 617.Piscataway, New Jersey: IEEE.

Christofilos, N. C., R. E. Hester, W. A. S.Lamb, D. D. Reagan, W. A. Sherwood, andR. E. Wright. 1964. High-Current Linear-Induction Accelerator for Electrons. Rev.Sci. Instrum. 35 (7): 886.

Ekdahl, C. 2002. Modern Electron Acceleratorsfor Radiography. IEEE Trans. Plasma Sci.30 (1): 254.

Eyharts, P., P. Anthouard, J. Bardy, C.Bonnafond, P. Delsart, A. Devin et al. 1998.Beam Transport and Characterization onAIRIX Protototype at CESTA. InProceedings of the 1997 ParticleAccelerator Conference. (Vancouver, B. C.,Canada, May 12–16, 1997). Vol. 1, p. 1257.Edited by M. Comyn, M. K. Craddock, M.Reiserr, and J. Thomson. Piscataway, NewJersey: IEEE.

Ferm, E. N., C. L. Morris, J. P. Quintana, P.Pazuchanics, H.L. Stacy, J. D. Zumbro et al.2002. Proton Radiography Examination ofUnburned Regions in PBX 9502 CornerTurning Experiments. AIP ConferenceProceedings 620 (1): 966.

Goldsack, T. J., T. F. Bryant, P. F. Beech, S. G.Clough, G. M. Cooper, R. Davitt, and R. D.Edwards. 2002. Multimegavolt MultiaxisHigh-Resolution Flash X-Ray SourceDevelopment for a New HydrodynamicsResearch Facility at AWE Aldermaston.IEEE Trans. Plasma Sci. 30 (1): 239.


Flash Radiography

Hawkins, D. 1961 (written 1946). “Los AlamosScientific Laboratory—Manhattan DistrictHistory, Project Y, the Los Alamos Project.”Los Alamos Scientific Laboratory reportLAMS-2532 (Vol. I).

Hoddeson, L., P. W. Henriksen, R. A. Meade,and C. Westfall. 1993. Critical Assembly—A Technical History of Los Alamos duringthe Oppenheimer Years, 1943–1945.Cambridge: Cambridge University Press.

Kwan, T. J. T., C. M. Snell, and P. J.Christenson. 2000. Electron Beam–TargetInteraction and Spot Size Stabilization inFlash X-Ray Radiography. Phys. Plasmas 7(5): 2215.

Mader, C. L., T. R. Neal, and R. D. Dick. 1980.LASL PHERMEX Data, Vol. I. Berkeley:University of California Press.

Mueller, K. H. 1985. Collimation Techniquesfor Dense Object Flash Radiography. Proc.Soc. of Photo-Optical InstrumentationEngineers 491 (1) :130.

Morris, C. L. et al. Defense Review LACP-02-89 (to be published 2003).

Neddermeyer, S., and A. R. Sayer. 1946.“Betatron Cloud-Chamber Studies withLevitated Assemblies.” Los AlamosScientific Laboratory report LA-483.

Röntgen, W. C. 1896. On a New Kind of Rays.Nature 53: 274.

Venable, D. 1964. PHERMEX. Phys. Today 17(12):19.

Watson, S., T. Kauppila, L. Morrison, and C.Vecere. 1997. The Pulsed High-EnergyRadiographic Machine Emitting X-Rays(PHERMEX) Flash Radiographic Camera.In SPIE 22nd International Congress onHigh-Speed Photography and Photonics.(International Congress on High SpeedPhotography and Photonics, Santa Fe, NewMexico, October 27–November 1, 1996).Edited by D. L. Paisley, and A. M. Frank.Vol. 2869, p. 920. Bellingham, Washington:SPIE.

Gregory Cunningham graduated with a bach-elor’s degree in electrical engineering from theUniversity of Michiganin 1986. He earned hisPh.D. in electrical engi-neering from the sameuniversity in 1992. Gregwas hired as a technicalstaff member at LosAlamos in 1992 and hasbeen in the DynamicExperimentationDivision since. In 1997,he became leader of the Image Analysis Teamand the Advanced Radiography Simulation andAnalysis Project, positions he held until 2002.During this time, Greg led development of theBayes Inference Engine, a sophisticated soft-ware tool now routinely used at Los Alamosand AWE for quantitative analysis of hydrotestradiographs. Currently, Greg is collaboratingwith the Applied Physics Division to help elu-cidate and quantify the role of existing and pro-posed radiographic experiments and facilitiesas they apply to certification of the nation’snuclear stockpile.


Flash Radiography

Christopher Morris received a B.S. inphysics from the Lehigh University in 1969and a Ph.D. in nuclearphysics from theUniversity of Virginiain 1974. Chris workedas a postdoctoralresearcher at theUniversity of Virginiafrom 1973 to 1976. Hethen became a technicalstaff member in theMedium EnergyPhysics Division at LosAlamos. Currently, he is in the PhysicsDivision. In addition to leading the develop-ment of a variety of charged-particle radi-ographic techniques at Los Alamos, usingprotons, electrons, and muons, Chris has alsostudied pion-nucleus interactions and is nowworking with a collaboration to build an ultra-cold neutron source for measuring neutronbeta decay. In 1996, Chris became aLaboratory fellow.

Documents

The Development of Flash Radiography