New Developments in Proton Radiography at the Los Alamos Neutron Science Center (LANSCE)

An application of nuclear physics, a facility for using protons for flash radiography, has been developed at the Los Alamos Neutron Science Center (LANSCE). Protons have proven far superior to high energy x-rays for flash radiography because of their long mean free path, good position resolution, and low scatter background. Although this facility is primarily used for studying very fast phenomena such as high explosive driven experiments, it is finding increasing application to other fields, such as tomography of static objects, phase changes in materials and the dynamics of chemical reactions. The advantages of protons are discussed, data from some recent experiments will be reviewed and concepts for new techniques are introduced.


Introduction
In the mid-1990's a new tool to aid in the mission of stewarding the US nuclear stockpile was invented at Los Alamos, namely proton radiography (pRad) [1]. The concept uses the attenuation due to the nuclear scattering of very short pulses of energetic protons as they transit high explosive driven experiments to provide contrast for flash radiography. The long mean free path of intermediate energy (10's of GeV) protons mitigates many of the difficulties encountered over the previous 5 decades of flash radiograph with high energy X-rays, such as large scatter backgrounds, low dynamic range, and poor position and temporal resolution [2,3].
Experiments preformed using 24 GeV protons provided by the Alternating Gradient Synchrotron (AGS) accelerator at Brookhaven Nation laboratory demonstrated all of the expected gains from pRad when compared to X-rays with a set of unclassified [4] and classified static objects [5]. The use of pRad for dynamic testing of surrogate nuclear weapon primaries is being pursued by both the Russian [3,6] and Chinese [7][8][9][10][11] weapons program but not by the United States.
However, a proton flash-radiography facility that uses the 800 MeV beam from the Los Alamos Neutron Science Center (LANSCE) at Los Alamos National Laboratory has become a work horse for the US stockpile stewardship program for smaller experiments studying the science of explosively driven systems. A recent review describes the techniques of proton radiography and presents some experimental results [2].
In this paper we briefly describe the technique, examine some of the current capabilities of the LANSCE proton radiography facility, illustrate them with some resent results, and discuss possibilities for the future.

Proton Radiography
The interaction of energetic protons with matter is governed to high precision by nuclear and Coulomb interactions. Protons lose energy to the matter because of the Coulomb scattering of the protons from the atomic electrons, and they scatter from the nuclei both because of the strong interaction and the Coulomb interaction with the proton. For the purposes of understanding proton radiography these interactions can be factorized and treated independently. The energy loss is given by the Bethe-Bloch model [12], Coulomb scattering is given by the Moliere theory [13,14], and nuclear scattering can be described by the black disk or optical model of nuclear scattering [15].
Proton radiography is performed by illuminating a target object with a beam of protons and then by focusing the transmitted protons onto a scintillator screen [16,17] using a quadrupole magnetic lens [18,19]. A collimator located at the Fourier plane of the lens is used to control the amount of contrast produced by Coulomb scattering. Nuclear interactions scatter or absorb beam particles generally to angles far outside the multiple scattering cone, which is on the order of 10 mrad in typical LANSCE experiments. The formalism used for analyzing pRad data is described in reference [2] and a summary of the physics listed above can be found in the particle data review [20].
The transmission, t, through an object of thickness, z, for proton radiography through a lens with a collimator acceptance angle, θ C , is the product of the nuclear attenuation and the Coulomb attenuation, the proton radiography transmission equation [2]: The absorption cross section, σ A , for hadrons on a nucleus with mass number A is often approximated by σ A = πr A 2 (accurate to ≈20 %), the geometric cross section of the nucleus, where r A ≈1:2A 1 3 fm. Here p is the proton momentum, β is its velocity relative to the speed of light, and X 0 is the radiation length [21]. For monolithic materials the proton radiography transmission equation can be inverted to obtain the thickness of an object. For objects where z<<λ, one can obtain radiation weighted thicknesses by solving equation (1). Because of generally small backgrounds precise thicknesses (on the order of one percent) can be obtained from proton radiography in these cases [22].

LANSCE Capabilities
The radiography facility at LANSCE is located in the former High Hesolution Spectrometry beam area (Experimental Area C) of Los Alamos Meson Physics Facility, LAMPF. It consists of an achromatic beam transport line into the area, a diffuser and matching section to control the size of the incident beam and reduce chromatic aberrations, a set of proton lenses, a containment vessel for dynamic experiments, and an optical imaging system. The LANSCE accelerator produces an 800 MeV H-proton beam of macro pulses with widths of up to 1 ms and a spacing of 8.33 ms (120 Hz).
Each macro pulse consist of a string of micro pulses spaced by 5 ns with a widths of 100 ps. The time control system for the accelerator allows a set of pulses to be constructed from any combination of micropulses within a macropulse for proton imaging.
The protons are delivered though a beam line that consists of a set of phosphor screens viewed by CCD cameras for beam monitoring and alignment, a strip line detector for measuring the time structure of the beam, a fast transformer for measuring the beam pulse intensity, a set of upstream quadrupole magnets for adjusting the angle position correlation of the beam on the sample to control chromatic aberrations, and a mechanical assembly that can insert any one of a set of tantalum diffusers to control beam spot size on the target.
Three lenses with different magnification can be used for radiography with magnifications of ×1 [19], ×3 [18], and ×7 [23]. The × 1 lens is constructed from four 30 cm diameter bore, 60 cm long quadrupole electromagnets and the ×3 from six 10 cm diameter bore 20 cm long permanent quadrupole magnets. The magnifying lenses (×3 and ×7) use Neodymium Iron Boron quadrupole magnets that need to be periodically remagnetized to repair radiation induced changes in the remnant magnetic field [24]. The ×7 lens is being upgraded to use Samarium Cobalt magnets to mitigate this problem.
We have measured the spatial resolution of the three lens systems at LANSCE by placing 3 mm thick targets of tungsten at the object location and characterizing the edge width. (The thickness of the target can lead to alignment issues so the data are not entirely consistent from year to year). Some results are shown in Fig. 1, along with estimates of the resolution made by fitting the edge with an error function. The resolution is given as the width.
There are many contributions that limit the resolution. These include chromatic aberrations in the magnetic lens system, proton scattering in the exit window and radiation to light converter. Of these, all but the last decrease approximately as 1/M where M is the magnification of the lens system. The spatial resolution measurements, displayed in Fig

Dynamic Experiments
Proton radiography has enabled new classes of experiments that are not effectively performed with flash x-rays. These have included studies of detonation propagation in high explosives, armor penetration experiments, and instability growth in shocked and accelerated interfaces. Here we present a few examples.

Detonation Propagation and Reflected Shock Interactions with the ×1 Lens
Proton radiography has been used to investigate the spatiotemporal evolution of detonation fronts and the associated reflected shocks on a PBX-9502 high explosive charge between an outer cylindrical steel liner and an inner elliptical tin liner (shown in Fig. 3) [25,26]. The charge was initiated with a PBX-9501 booster and a line wave generator (LWG) at 30°from the major axis of the ellipse. This configuration provided a large region where the high explosive is not within the line of sight of the detonation line and thus the data can be used to test various burn models and equation of state formulations. A single frame of the 21 frame movie is shown in Fig. 3.

Richtmyer-Meshkov Instability with the ×3 Lens
A number of experiments have been performed to study the spike and bubble growth in plane wave driven Richtmyer-Meshkov instability growth in both solid and liquid metals [27][28][29][30][31]. These data have been used to develop new models of the dynamics of instability growth and ejecta formation with unsupported shocks. A very recent development is a two wave driver for studying Bsecond shock^effects on instability growth [27]. Proton radiography data taken of this system will allow for a more detailed quantitative understanding of two-shock driven instability and the resulting ejecta formation. Some initial data from Buttler et al. [27] are shown in Fig. 4.

Quasi-Static Experiments
The penetration, resolution, flexible time format, and long standoff make proton radiography useful for a number of static or quasi-static applications. Some examples demonstrating the breadth of this work are given in the following subsections.

Casting Experiment with the ×1 Lens
The × 1 lens provides a 12 × 12 cm 2 field of view useful for imaging large scale phenomena. Clarke et al. [32] and Gibbs et al. [33] have reported data spanning the micro-scale to the mesoscale (obtained with proton radiography and synchrotron X-ray radiography) and the macroscopic scale (obtained with proton radiography). Such data can be used to study metal solidification dynamics across length scales to validate and improve models of important manufacturing processes such as casting.
As an example at the macroscopic scale, a tin-bismuth alloy melt flowing into a 3 mm thick graphite mold is shown in Fig. 6. Here darker regions correspond to higher densities. The fill process was controlled by a plug between the reservoir and the mold made of pure bismuth which had a higher melting temperature than the alloy melt. As the reservoir was heated and the plug melted, the fill was initiated. Proton images were made using a 350 μm thick columnar CsI(Tl) screen, and recoded on a DIMAX CMOS 2 k×2 k pixel [34] camera. Each frame was formed with a beam pulse of~6 × 10 9 protons, distributed as a two dimensional Gaussian across the 12 × 12 cm 2 field of view at the object location.
When the temperatures approached the melting temperature of the plug, a 1000 frame movie was started at 5 Hz. The frames below represent about 3 s of the 200 s movie that covered the filling of the mold. The flexible timing of the linear accelerator for proton radiography was useful to enable capture of the region of interest, without the complication of a trigger to synchronize the beam to the experiment.

Radiography of Surrogate Fuel Rods with the ×3 Lens
There are several advantages to proton radiography when compared to other radiographic probes. One is the long standoff between the object and the image location. This has obvious advantages for explosively driven experiments in which it is necessary to protect Considerable effort is aimed at developing models that can predict structural changes to nuclear fuel pellets as a function of burnup in a reactor. Extending the lifetime has obvious economic benefits. Proton radiography provides data that can provide <50 μm position resolution for objects of the scale of 5 cm, with areal densities of 10 g/cm 2 of uranium oxide or thorium oxide, and in a radiation field that can be larger than 200 Gy/h at 30 cm from the fuel rod, because of its short exposure times and long standoff (the cameras and scintillator are 12 m from the object location).
Here we show data taken on a set of objects with the ×3 magnifier aimed at determining the capabilities of proton radiography for evaluating the damage of activated fuel rods. A previous subset of this data were used to compare X-ray and proton radiography [35] and to demonstrate the isotopic sensitivity of resonance neutron imaging [36]. Plausible methods for using these other probes in the radiation fields of near the activated fuel rods have not been described. With protons the long standoff and fast imaging provide a solution to this problem.
Tomographic proton data were taken on several surrogate fuel assemblies. Proton images were made using a 350 μm thick columnar CsI(Tl) screen, and recoded on a set of 5 three frame CMOS imaging cameras [37].
Seven hundred and twenty frames of data were taken covering 180°by rotating the object by 0.25 degrees between frames. Each frame was formed from an average of 15 camera pictures taken using 3 beam pulses of 6 × 10 9 protons distributed as a two dimensional Gaussian across the 4 × 4 cm 2 field of view at the object location with a widths of 1.3 cm in each direction. The proton data were taken at a rate of 0.5 Hz. The rate was limited by the speed at which the goniometer could reliably rotate the object.
The proton radiography equation was inverted to obtain areal densities. Volume densities were calculated using filtered back projection. In Fig. 7 the results obtained from four different test objects are shown: a set of 10 mm diameter thoria pellets, a set of 5 mm diameter urania fuel pellets compressed to at a range of densities, and two sets of 5 mm diameter urania pellets with various defects emplaced before the sintering process.
The ensemble of data demonstrates a range capability: the ability to quantitatively measure density, Fig. 7(c); the observation of texture in the thick thoria pellets Fig. 7(a); the wide dynamic range of protons in thickness; and the sensitivity of protons to~60 μm features Fig. 7(b) and (d). The texture in the thoria was first characterized in the proton radiography results. The density determinations of the pellets in Fig. 7(c) were within a few percent of the measured densities of the  Fig. 6 Casting mold filling of a 3-mm-thick plate using the ×1 lens (tin 27 atomic % bismuth alloy). Darker regions correspond to higher density. Spacing between frames is 0.2 s sintered samples. The urania samples were interesting because the proton data could be used to observe the migration and changes that occurred in the embedded defects during the sintering.

Radiography of Self-Propagating High Temperature Chemical Reactions with the ×3 Lens
The 120 Hz macro structure of the LANSCE accelerator allows continuous radiography of experiments with dynamic structure, but where the initiation time is difficult to predict. Here we present an example of a movie of a self-propagating high temperature reaction [38][39][40][41][42][43][44][45] in a stoichiometric mixture of titanium and silicon. In these reactions the material changes from a pressed powder to a ceramic pellet. Proton radiography allows study of the dynamics, perhaps aimed at process control for producing near net shaped ceramic parts dynamically.
Data from one test from Bernert et al. [46] are shown in Fig. 8. Figure 8(a) shows a frame just after initiation. Near the top, the tungsten filament that was used to initiate the reaction can be observed. The sequence of radiographs, shown in Fig. 8b, was cut out of a movie that was approximately 100 s long. This allowed the collection of dynamic data even through the initiation time was quite uncertain. The data were taken with the × 3 magnifier, with a position resolution of 50 μm. The exposure times were~180 ns, ensuring there is negligible motion blur. The frame rate was 10 Hz. The data allowed precise measurements of the location and shape of the reaction zone shown in Fig. 8(b) as a function of time and quantitative measurements of the time dependence of the density, Fig. 8(c). The average flame front velocity was determined to be 0.7 cm/s.

Tomography of a Meteorite Sample with the ×7 Lens
The ×7 magnifier proved to be difficult to commission. Although it worked well in a commissioning run, it failed to perform as expected in later runs. Work with the ×3 magnifier has definitively demonstrated that radiation damage leads to demagnetization of the permanent magnet material, particularly in regions of high dose and large differences between the field direction and the magnetization directions [24]. This leads to non-uniform reduction in field strength, which only weakens the focusing slightly, but introduces large optical aberrations in the lens, which results in significant resolution degradation. This problem is especially difficult with the ×7 magnifier because of the small bore which places the neodymium iron boron very close to the intense proton beam.
With this knowledge the ×7 magnifier was recently recommissioned after remagnetization and it was used to perform several of experiments. One of these was tomography on a small sample of the meteorite the exploded over Chelyabinsk Russia in 2013.
After tuning the lens and performing some of experiments that needed only a few radiographs, the meteorite sample was radiographed at 721 angles spanning 360°. The degradation of the resolution across this set of radiographs is apparent in Fig. 9.
Two representative slices from the tomograph are shown in Fig. 10. One can clearly observe higher-Z regions that are likely to be metal inclusions in a silicate base. Many of the small inclusions seem to have a low density region at their center. The line plot shows one of these features.
A new design has been completed for this lens that has shown that the neodymium iron boron magnets can be replaced with much more radiation resistant samarium-cobalt magnets. The new system has been procured and will be commissioned over the next year.

Scanning Proton Microscope
Two new ideas can enable major advances in proton radiography for the sort of quasi-static experiments described above. The first is the realization that by scanning a pencil beam the limiting position resolution in proton radiography is determined by the size of the incident beam and scattering in the target, which can be much smaller than the resolution obtained with lens focusing because the energy and angle spread leading to chromatic blur are no longer an issue.  Fig. 11. For a target of thickness t, the blur from scattering in the target is [21] Δy ¼ As an example, for a 3 mm thick sample of uranium the above gives Δy = 5 μm with an 800 MeV proton beam. This is much better resolution than lens focused radiography can provide. With a collimated beam, a measurement of transmission though an angle collimator measures the object thickness. An object can be radiographed by measuring the attenuation as a function of position with either a scanned beam or a scanned object. A collimated beam can be produced by imaging the beam transmitted through a pinhole. A further reduction in the beam size can be obtained by using a demagnifying lens to focus the beam onto the object. The beam from the LANSCE accelerator is produced with 201 MHz microstructure. Individual beam pulses are about 100 ps long spaced by 5 ns. New developments in both detectors and recording technology make 10 ps measurements now possible. By using a fast detector one can measure the time of flight of the protons through the system and determine their energy as well as the intensity. Thus one can determine the energy loss of protons in a target. The energy loss, dE dy , of protons traversing material is given by the Bethe-Bloch formula [12,21]. The energy loss only varies slightly across through the periodic table from 2 MeV/(g/cm 2 ) in low-Z targets to 1.5 MeV/g/cm 2 in high-Z targets at LANCE proton energies. The energy of transmitted protons can be determined by their time of flight through the lens. One finds that the energy loss, ΔE, is related to the change in velocity Δβ by: ΔE = Δββγ 2 E, where E is the total energy, β is the velocity relative to the speed of light, and γ ¼ 1= . With 10 4 transmitted protons in each 100 ps long proton pulse, it is possible to obtain time resolution, Δt ¼ 100ps= ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 10 4 protons p ¼ 1 ps, when the measurement is limited by proton counting statistics. A 24 m flight path gives sufficient energy resolution to measure the thickness of the 3 mm thick uranium sample to about 0.5 %, about the same precision obtained with multiple scattering radiography with the current system. Combining energy loss information:

Preliminary TOF Radiography Studies
We have taken some preliminary data using a 20 GHz analog bandwidth oscilloscope and a fast multi-channel plate detector. The beam was collimated to a root mean squared spot size of~150 μm for these measurements. TOF data between an upstream strip line detector and the downstream channel plate detector are shown in Fig. 12. Five measurements were made with trains of 80 micropulses spaced by 5 ns with 10 5 protons per micropulse. Time drifts on the order of 100 ps (corresponding to about 5 MeV of energy shift) both across the pulse train and between pulses are evident. The beam energy shifted within a macro-pulse and from pulse to pulse. These shifts are likely to be due to changes in the beam energy because of RF loading in the linera accelerator (LINAC). In future measurements the beam energy can be measured by comparing the outputs from two upstream detectors separated by >10 m to measure the beam energy and correct for these shifts.

Summary
An overview of the LANCE proton radiography system and the experimental program has been given. The position resolution of the ×1, ×3, and the ×7, lenses has been measured and found to depend linearly on inverse magnification over this range of magnifications. This suggests that that further improvements can be obtained at increased magnification. Design studies of higher magnification systems are underway.
Examples of static or quasi-static experiments have been presented for each of the lenses. These experiments cover a   10 Representative slices from the tomographic reconstructions. X-y slices showing the volume density are shown above line plots of the density corresponding to the white lines. An assortment of features can be observed throughout the sample. The bright patches, marked with an arrow on the left, are interpreted to be higher-Z inclusions and on the right a smaller one that appears to have some structure Fig. 11 Quantities used to describe Coulomb multiple scattering in the plane of the figure wide sampling of applications and suggest that as the capabilities of pRad improve, the number of applications for basic science and engineering will grow, especially in areas where the standoff, wide dynamic range and long pulse sequences and flexible time format provided by proton radiography are an advantage.
The possibility of improving the position resolution while providing material identification by using pencil beams and by measuring time of flight as well as attenuation has been described. The addition of these new measurement capabilities would significantly increase the utility of proton radiography for both quasi-static and some dynamic experiments.